24888
April 2002
E THE WORLD BANK
Government at Risk
Government
at Risk
CONTINGENT LIABILITIES
AND FISCAL RISK
Hana Polackova Brixi
Allen Schick
editors
A COPUBLICATION OF
THE WORLD BANK AND
OXFORD UNIVERSITY PRESS
C) 2002 The International Bank for Reconstruction and Development /
The World Bank
1818 H Street, NW
Washington, DC 20433
All rights reserved.
1 2 3 4 04 03 02 01
A copublication of the World Bank and Oxford University Press.
Oxford University Press
198 Madison Avenue
New York, NY 10016
The findings, interpretations, and conclusions expressed here are those of the
author(s) and do not necessarily reflect the views of the Board of Executive Direc-
tors of the World Bank or the governments they represent.
The World Bank cannot guarantee the accuracy of the data included in this work.
The boundaries, colors, denominations, and other information shown on any map in
this work do not imply on the part of the World Bank any judgment of the legal
status of any territory or the endorsement or acceptance of such boundaries.
Rights and Permissions
The material in this work is copyrighted. No part of this work may be reproduced
or transmitted in any form or by any means, electronic or mechanical, including
photocopying, recording, or inclusion in any information storage and retrieval
system, without the prior written permission of the World Bank. The World Bank
encourages dissemination of its work and will normally grant permission promptly.
For permission to photocopy or reprint, please send a request with complete
information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers,
MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com.
All other queries on rights and licenses, including subsidiary rights, should be
addressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington,
DC 20433, fax 202-522-2422, e-mail pubrights@worldbank.org.
Library of Congress Cataloging-in-Publication Data has been applied for.
Contents
Foreword ix
Nicholas H. Stern, Senior Vice President and Chief
Economist, World Bank, and
Gobind T Nankani, Vice President, World Bank
Acknowledgments xi
Introduction 1
Hana Polackova Brixi, World Bank,
and Allen Schick, University of Maryland
PART I LEARNING TO DEAL WITH
FISCAL RISKS IN GOVERNMENT PORTFOLIOS
POSSIBLE ANALYTICAL AND INSTITUTIONAL FRAMEWORKS
1. Dealing with Government Fiscal Risk: An Overview 21
Hana Polackova Brixi, World Bank,
and Ashoka Mody, IMF
2. Accounting and Financial Accountability to Capture Risk 59
Murray Petrie, The Economics and Strategy Group
3. Budgeting for Fiscal Risk 79
Allen Schick, University of Maryland
4. Institutional and Analytical Framework for Measuring
and Managing Government's Contingent Liabilities 99
Suresh M. Sundaresan, Columbia University
V
vi CONTENTS
5. Analytical Techniques Applicable to Government
Management of Fiscal Risk 123
Krishna Ramaswamy, the Wharton School
6. Fiscal Sustainability and a Contingency Trust Fund 143
Daniel Cohen, Universite de Paris,
Ecole Normale Sup&rieure
7. A Framework for Assessing Fiscal Vulnerability 159
Richard Hemming, IMF, and Murray Petrie,
The Economics and Strategy Group
COUNTRY EXAMPLES
8. Evaluating Government Net Worth in
Colombia and R6publica Bolivariana de Venezuela 181
William Easterly and David Yuravlivker, World Bank
9. The Challenges of Fiscal Risks in Transition:
Czech Republic, Hungary, and Bulgaria 203
Hana Polackova Brixi, World Bank, Allen Schick,
University of Maryland, and Leila Zlaoui, World Bank
10. Analyzing Government Fiscal Risk Exposure in China 235
Kathie L. Krumm, World Bank, and Christine P. Wong,
University of Washington
11. Dealing with Contingent Liabilities in
Indonesia and Thailand 251
Hana Polackova Brixi, World Bank,
and Sudarshan Gooptu, World Bank
12. Dealing with Contingent Liabilities in Colombia 269
Juan Carlos Echeverry, Ver6nica Navas, Juan Camilo
Gutierrez, and Jorge Enrique Cardona,
Government of Colombia
PART II DEALING WITH SPECIFIC
SOURCES OF GOVERNMENT FISCAL RISK
ANALYTICAL AND MANAGERIAL TOOLS
13. Pension Guarantees: A Methodology for Assessing
Fiscal Risk 283
George G. Pennacchi, University of Illinois
CONTENTS vii
14. Measuring and Managing Government Contingent
Liabilities in the Banking Sector 311
Stijn Claessens, University of Amsterdam,
and Daniela Klingebiel, World Bank
15. Government Insurance Programs:
Risks and Risk Management 335
Ron Feldman, Federal Reserve Bank of Minneapolis
1 6. Contingent Liabilities of the Central Bank:
Analyzing a Possible Fiscal Risk for Government 355
Mario Blejer, Central Bank of Argentina,
and Liliana Schumacher, IMF
PRACTICE
17. Contingent Liabilities in Infrastructure:
Lessons from the East Asian Financial Crisis 373
Ashoka Mody, IMF
18. Monitoring Fiscal Risks of Subnational Governments:
Selected Country Experiences 393
Jun Ma, Deutsche Bank
19. Guarantees as Options: An Evaluation of Foreign
Debt Restructuring Agreements 419
Sweder van Wiinbergen, University of Amsterdam
and the Centre for Economic Policy Research,
and Nina Budina, World Bank
20. The Fiscal Risk of Floods: Lessons of Argentina 451
Alcira Kreimer, World Bank
Conclusion: Toward a Code of Good Practice
on Managing Fiscal Risk 461
Allen Schick, University of Maryland
Foreword
IN PUBLIC FINANCE, it no longer suffices for analysts and institutions
to focus solely on budget revenues and expenditures. Recent history
demonstrates that fiscal performance and, in turn, economic develop-
ment can be seriously disrupted by the sudden, unexpected costs of
hidden contingent liabilities and other unanticipated fiscal risks.
During the second half of the 1990s, unreported contingent liabili-
ties and related fiscal risks contributed to economic crises and disrupted
growth in a number of developing countries, motivating stepped-up
efforts at the World Bank to devise new concepts and tools for analyz-
ing and managing public finance. With the aim of improving the analy-
sis of fiscal risks and supporting policy advice in this area, the Economic
l'olicy Unit of the Bank's Poverty Reduction and Economic Manage-
ment Network established the Quality of Fiscal Adjustment Thematic
Group. This book was produced as part of the effort by this Thematic
Group to promote new thinking about public finance.
We now know that conventional frameworks for fiscal analysis that
concentrate on direct, explicit liabilities fail to address contingent fiscal
risks. For example, fiscal sustainability analysis that focuses, as is typi-
cally the case, on the officially reported budget deficits fails to detect
possible future increases in government debt and payments that may
emerge from both explicit and implicit government guarantees on en-
terprise credit, state insurance schemes, exchange rate guarantees, and
commitments to assist failed banks. Similarly, the government budget
process and documentation generally fail to scrutinize the substantial
claims on public resources that are associated with contingent liabili-
ties, realized and potential.
This book is a notable step forward in filling gaps in our under-
standing of fiscal risks and in developing suitable frameworks for man-
aging them. Through country cases and advances in conceptual design,
the book provides a menu of practical ideas for policymakers and
scholars for bringing fiscal risk within the ambit of public finance. It
demonstrates that government fiscal analysis needs to cover the entire
ix
x FOREWORD
portfolio of direct and contingent liabilities, as well as assets and the
revenue base. This in turn requires the government to identify, classify,
and assess its fiscal risks so that it can provide reliable estimates of future
payments that may ensue from past and pending liabilities. Only by iden-
tifying and measuring its exposure can a government bring its risks un-
der effective control. This task has been facilitated by the availability of
new methodologies, such as value-at-risk analysis and options pricing.
But despite these advances, governments still face technical challenges in
dealing with risk. However, the greater challenges are political and infor-
mational. Governments must care enough about fiscal and economic per-
formance beyond the short term to impose limits on future risk-taking
and to invest resources in identifying and controlling fiscal risks.
The essential challenge for governments is to launch a full and forth-
right effort to avoid excessive risk-taking and to prudently manage the
risks that they do take. Doing so typically requires governments to com-
mit themselves to greater transparency and broader fiscal discipline
than they have had in the past.
The benefits for governments that make the effort are enormous,
not only with respect to their future fiscal stability, but also with re-
spect to their capacity to achieve broader policy objectives. The ideas
and cases presented in this book should prompt governments to under-
take this effort.
At the World Bank, the analysis and control of fiscal risks has now
become an integral part of its assistance to member countries. Working
closely with policymakers, the Bank tailors its analytic support and
policy advice to country-specific circumstances, taking into account the
technical challenges (few countries have a reasonably complete inven-
tory of government contingent liabilities and other fiscal risks), the in-
stitutional set-up (fiscal and quasi-fiscal institutions and relationships),
and the political sensitivity of the issues (a full accounting of fiscal risks
often shows a government to be in a less favorable financial condition
than is reported in official statements).
There is no single approach to dealing with contingent liabilities and
fiscal risk. As the studies in this book indicate, governments that have
sought to control their risk exposure have taken several different ap-
proaches. This book points the way ahead by setting out general prin-
ciples of sound fiscal management and by providing specific examples
of innovative country practices.
Nicholas H. Stern Gobind T Nankani
Senior Vice President Vice President and Head of Network
Development Economics Poverty Reduction and Economic
and Chief Economist Management Network
The World Bank The World Bank
Acknowledgments
WVE WOULD LIKE TO EXPRESS our sincere gratitude to all the distin-
guished authors for their valuable contributions to this book. We also
thank our many colleagues and friends who provided helpful advice
and suggestions. We acknowledge in particular the support and feed-
back on the papers that we received from the Quality of Fiscal Adjust-
ment Thematic Group (QFA TG) of the Economic Policy division of
World Bank's Poverty Reduction and Economic Management (PREM)
Network. Assistance from other World Bank staff, staff of other inter-
national institutions, government officials, academics, and researchers,
who provided valuable feedback on earlier drafts of the chapters, is
also gratefully acknowledged. Finally, we would like to thank the anony-
mous reviewers of this book whose challenging comments and sugges-
tions have greatly improved this text.
We acknowledge with thanks the financial support we received for
this book and related research and dissemination from the Economic
Policy division of World Bank's PREM Network through the QFA TG.
Support received from the Public Expenditure Thematic Group of the
Public Sector Division is also gratefully acknowledged.
The views expressed in this book are those of the authors and should
not be attributed to any particular institution, including institutions
with which individual authors may be associated, including the World
Bank, International Monetary Fund, the Federal Reserve System, and
others.
xi
INTRODUCTION
Government at Risk:
Contingent Liabilities
and Fiscal Risks
Hana Polackova Brixi
World Bank
Allen Schick
University of Maryland
ONE DOES NOT HAVE TO SEARCH far to see evidence of governments at
risk. In some countries, a long spell of seemingly sound fiscal manage-
ment was followed by a fast-moving crash that forced the government
to spend unbudgeted resources to pay obligations that few knew ex-
isted. When economic conditions stabilized, the government was left
with elevated levels of debt and other obligations. Similarly, in other
countries government debt soared much faster than the official deficit
would have suggested, because the government was compelled to make
good on obligations that had been assumed to be outside the budget-
ary framework. Today, in still other countries, the public finances are
facing a difficult future as the off-budget obligations accumulated in
the past come due in the decades immediately ahead. In all these sce-
narios, governments at risk have faced or are facing major fiscal chal-
lenges as a result of their contingent liabilities, which tend to remain
outside the framework of conventional public finance analysis and
institutions.
When risks come due in developed countries, they impose costs on
government budgets and sometimes temporarily reduce economic out-
put. When governments in less-affluent countries take imprudent risks,
the effects of such risks may spread quickly across the economy, cause
1
2 BRIXI AND SCHICK
capital to flow out to safer havens, possibly compel the government to
change economic course, and so retard or even reverse the country's
development. Moreover, if the mechanisms for information disclosure
are weak and market institutions not well developed, the risks assumed
by government generate a bias in the behavior of economic agents and
moral hazard in the markets, and thus work against development even
before they are realized.
In many countries, the reality or prospect of unbudgeted fiscal risks
coming due has been a wakeup call to extend fiscal management be-
yond the budget to all actions and transactions that put the govern-
ment in financial jeopardy. Doing so is difficult, however, because it
requires government to enter uncharted fiscal territory, where analyti-
cal frameworks are sometimes difficult to apply, accounting standards
are underdeveloped or poorly enforced, and the data are inadequate or
hidden from public scrutiny. And, because contingent liabilities often
grow from fiscal opportunism, when policymakers seek to hide the
real fiscal cost of their decisions and to reduce the reported budget
deficit, bringing them under control may become first of all a question
of political will.
This book aims to provide motivation and practical guidance to
governments seeking to improve their management of fiscal risks.
Among other things, it addresses some of the difficult analytical and
institutional challenges that face reformers tooling up to manage gov-
ernment fiscal risks, and it describes the inadequacies of conventional
practices as well as recent advances in dealing with fiscal risk. It also
presents several untested ideas for developing new instruments for regu-
lating and valuing fiscal risks. In so doing, the authors recognize that
some novel schemes are not yet sufficiently developed to warrant im-
mediate application by governments. But pushing forward the frontier
of public finance today, the book aims to enhance the practice of fiscal
analysis and management in the future.
This volume has grown out of World Bank initiatives to assist coun-
tries that are working to understand and manage the fiscal risks facing
their governments. These initiatives, led by the World Bank Quality of
Fiscal Adjustment Thematic Group during 1998-2001, covered over
40 interested countries and involved partnerships with many govern-
ment agencies, leading universities, research institutions, international
agencies, and associations of practitioners around the world.
The Magnitude of the Problem
We define fiscal risk as a source of financial stress that could face a
government in the future. The book focuses particularly on the fiscal
risks that are realized when uncertain events occur-such fiscal risks
INTRODUCTION 3
are often associated with government contingent liabilities. Recent his-
tory has brought with it many examples of contingent liabilities that
challenge government finances. The explicit and implicit government
insurance schemes in the domestic banking sector that emerged from
the 1997 financial crisis in East Asia added some 50 percent of gross
domestic product (GDP) to the stock of government debt in Indonesia,
30 percent in Thailand, and over 20 percent in Japan and Korea. In the
1980s, similar schemes generated a fiscal cost of over 40 percent of
GDP in Chile and around 25 percent of GDP in C6te d'Ivoire, Uru-
guay, and Republica Bolivariana de Venezuela.' In the 1990s, Brazil
and Argentina saw their government debt escalate when the central
government had to bail out commitments made by subnational gov-
ernments. Government debt in Malaysia, Mexico, and Pakistan soared
from unexpected defaults on government guarantees that had been
issued to promote private participation in infrastructure. Several chap-
ters of this book illustrate that contingent liabilities may become the
most critical factor in a country's fiscal performance.
Empirical analysis of past increases in the stock of government debt
confirms that realized government contingent liabilities account for a
large share of those increases. Kharas and Mishra (2001) illustrate
across nearly 50 countries that large increases in the stock of govern-
ment debt cannot be explained by the governments' reported budget
deficits (see Figure 1). Calling the annual increase in government debt
that is in excess of budget deficit a "hidden deficit," Kharas and Mishra
show that hidden deficits have stemmed mainly from the cost of real-
ized contingent liabilities and realized risks in the government debt
portfolio (particularly the currency risk of government foreign debt
instruments). In some developing and transition countries, contingent
liabilities have contributed on average to hidden deficits of more than
2 percent of GDP annually over a period of more than 10 years. The
analysis by Kharas and Mishra also indicates that contingent liabilities
tend to be associated with speculative attacks and currency crises.
In the recent past, several factors have worked to increase govern-
ments' exposure to fiscal risk and their tendency to incur hidden defi-
cits. The rapidly increasing volumes and volatility of international
private capital flows have accelerated the growth of domestic finan-
cial systems but also have made these systems, and thus implicitly the
countries' fiscal authorities, more vulnerable. This condition was clearly
illustrated during the three years following the 1997outflow of foreign
capital from East Asia. Privatization and reduction of the explicit fi-
nancial role of the state allowed many governments to cut their bud-
geted expenditures, but required either explicit or implicit promises
that the government would come to the rescue should the private sec-
tor fail to deliver expected outcomes. Such guarantees and promises,
in turn, have increased the uncertainty of future public financing
4 BRIXI AND SCHICK
Figure 1. Average Annual "Hidden" Deficits
(percent of GDP over different 5- to 20-year periods from 1970s to 1990s)
Developed
Developing and Transition Countries Countries
8
7
6
5
4
3
2
1
-3
-4
U O
Note: Graphs represent average annual increases in government debt unexplained
by reported deficits.
Source: Kharas and Mishra (2001).
requirements. Furthermore, these guarantees have boosted moral haz-
ard in the markets. Loans and investments with a full guarantee suffer
from insufficient analysis and supervision by creditors. Moreover, the
beneficiaries of poorly designed state insurance schemes tend to ex-
pose themselves to excessive risks. For example, in the United States
the generous benefits of the federal flood insurance program have re-
sulted in the excessive construction of homes in flood-prone areas (U.S.
GAO 1998). This market behavior makes it more likely that later the
government will be asked to provide financial support.
The Political Economy of Contingent Liabilities
and Fiscal Risk
Often fiscal risks, particularly those in the form of contingent liabili-
ties, arise from politics and fiscal opportunism rather than economic
policy. Policymakers tend to build up government contingent liabili-
ties to avoid difficult adjustment and painful structural reforms. In
INTRODUCTION s
this process, credit guarantees replace budgetary subsidies; take-or-
pay contracts come in lieu of liberalizing prices and restructuring the
energy, water, and other vital sectors; "letters of comfort" allow insol-
vent enterprises and banks to avoid bankruptcies; and so on.
In some instances, government support in the form of contingent
liabilities may be justified. In Europe and the United States, few would
argue against the immediate provision of a government guarantee to
cover the legal liability of airlines after the September 2001 terrorist
attacks in the United States. Government support was deemed justi-
fied for the political risk negatively affecting the airlines. Government
guarantees, as the form of support, also were appropriate because the
airlines would, in their own interest, try to avoid further terrorism
onboard.2
In many cases, however, governments have assumed contingent li-
abilities either in pursuing low-priority objectives-that is, on pro-
grams that would not have withstood public scrutiny-or in using
off-budget support when other forms would have been more appropri-
ate. Some governments have provided letters of comfort to cover the
commercial risk of foreign investors that have taken a position in do-
mestic financial institutions or enterprises. But with the ensuing moral
hazard, bailouts have often followed, frequently financed directly from
special government funds rather than through the budget, and have
tended not to be a good use of public money.
A common example of off-budget forms of support that are not
appropriate is the provision of credit guarantees to enterprises that
continually incur losses. While the government may have a good rea-
son to support some of such enterprises (for example, the national
railways, if their losses are the result of government fare pricing policy),
budgetary subsidies or direct government loans would sometimes be
more effective and almost always less expensive.
Whether for railways or airlines, often the best response to calls for
government support is to encourage restructuring, privatize enterprises
and financial institutions (and recapitalize them in the process if needed),
break down monopolies, and liberalize prices. But the ease of issuing
government guarantees and other promises of future possible govern-
ment support allows the government to postpone these sometimes dif-
ficult and, in the short term, costly actions.
In most countries, government is able to offer a promise of future
contingent support without seriously considering the future cost to the
taxpayer. Governments doing their accounting and budgeting on a cash
basis have particularly wide scope to behave opportunistically.3 In con-
trast to commercial accounting practices, which in most developed
countries require firms to recognize the future cost of pensions and
other risks on their balance sheets, few governments disclose the pro-
spective costs of their off-budget commitments. At the time of signing
6 BRIXI AND SCHICK
a letter of comfort or a guarantee contract, the government is able to
claim no cost to the budget. In some countries, a variety of govern-
ment entities are able to take on such commitments-sometimes with-
out even informing the ministry of finance or any other authority. Other
countries have centralized the guarantee-issuing authority at the min-
istry of finance or similar entity, but still do not require such an entity
to report the government's off-budget commitments until they fall due.
And even after they fall due, government may just issue more debt or
find alternative ways to cover them, without ever recording the event
in any reports. As a result, guarantees and similar forms of off-budget
support that create contingent liabilities turn out to be a relatively
easy way for government to avoid scrutiny of the risks inherent in
channeling its support.
As these examples of the inappropriate use of off-budget forms of
government support indicate, fiscal opportunism that gives rise to gov-
ernment contingent liabilities tends to grow out of a narrow scope of
conventional fiscal analysis and fiscal management. Scrutiny that fo-
cuses solely on the government's cash-basis budget and deficit invites
policymakers to generate contingent liabilities and other fiscal risks
outside the budgetary framework (see Box 1). Many countries have
learned firsthand that a narrow focus on cash-basis budget, deficit,
and debt compels governments to delay investments and structural
reforms, run down public assets, raise temporary revenues (sometimes
by assuming long-term liabilities in exchange for cash), and distort
spending priorities and the timing as well as the form of government
support (see, for example, Forte 1998, Polackova 1998, and Easterly
1999). As reforms (such as pension reform, the downsizing of public
employment, and enterprise and bank restructuring) that may require
higher deficits in the short term are put on hold, fiscal opportunism
puts economic growth at risk. Whether contingent liabilities are as-
sumed in the effort to maintain the status quo and avoid reforms, or
just to provide government support outside the budget and thus con-
ceal its cost and financing, their existence generates uncertainties about
future public financing requirements and so threatens future fiscal sta-
bility and the country's development. In this context, Selowsky (1998)
has emphasized that reported deficit reduction does not necessarily
imply "quality" of fiscal adjustment, which has the dimension of
sustainability as well as efficiency.
Overall, development tends to be correlated with a shift in risk from
individuals and individual economic agents and sectors to the state. As
governments promote development and economic conditions improve,
policymakers are pressured to take on commitments they may have
avoided earlier. Social security programs, various state insurance
schemes (targeting various beneficiaries, including enterprises, devel-
opers, farmers, and depositors), umbrella guarantees covering agen-
]NTRODUCTION 7
Box 1. Fiscal Risks as a By-product of Deficit Targeting
When loose rules for fiscal management are accompanied by pressure
for fiscal adjustment, vote-seeking politicians and budget-seeking bu-
reaucrats have additional cause for opportunism. Paradoxically, the
incentive to mask the true cost of risks often rises when government
comes under pressure to tighten its budget constraints. When pressure
for adjustment is slack, the government may have little incentive to
hide the costs of its financial commitments. But when stringent fiscal
targets are imposed, wily spenders have incentive to substitute illusory
adjustments for actual ones. And when proper accounting rules and
strong enforcement mechanisms do not accompany the targets, the
spenders have ample opportunity to evade the controls.
Various studies have shown that weak enforcement produces bud-
getary opportunism. In 1985 the United States enacted the Gramm-
Rudman-Hollings law, which promised to progressively reduce the size
of the deficit and to produce a balanced budget by 1991. But, as Schick
(1995) points out, instead of genuine austerity, the law spawned bud-
getary legerdemain that increased the government's exposure to fiscal
risk. The volume of loan guarantees escalated, the government was
slow in responding to a costly crisis in the banking sector, and it adopted
policies (such as asset sales) that weakened its long-term fiscal posture.
In effect, as Rubin (1997) shows, faced with the Gramm-Rudman con-
straint on fiscal deficits, the U.S. Congress has reduced direct lending
by $50 billion and increased loan guarantees by $178 billion, replacing
budgetary outlays by explicit contingent liabilities.
In the process of fulfilling all the criteria for EU membership, the
Maastricht criteria on government deficit and debt were applied by
some countries in ways that escalated fiscal risk. The ploys, well de-
scribed by Forte (1999), included defining government narrowly, so
that the finances of various state-owned or controlled institutions were
not included in the calculation; hiding a portion of the governments
debt in various nongovernmental accounts; substituting guarantees for
loans and grants; recording subsidies as purchases of assets; devising
off-budget expenditures in lieu of direct financing; and deferring ex-
penditures on infrastructure and maintenance. Some European Union
governments received one-time payments from enterprises in exchange
for assuming future pension liabilities; others reduced their reported
public debt by reclassifying certain state enterprises as private entities.
in Italy, the railways have raised funds through the financial markets
to cover their deficits for many years with government agreement and
an explicit guarantee from the treasury. Yet, those operations had no
impact on the measured fiscal deficit or on the measured stock of gov-
ernment liabilities (Glatzel, 1998).
(Box continues on the following page.)
8 BRIXI AND SCHICK
Box 1 (continued)
Creative accounting and budgeting practices have been used also in
developing countries to portray their fiscal condition in a much more
favorable light than is warranted. When pressured by adjustment pro-
grams administered by the World Bank or IMF, some developing coun-
tries have privatized assets, disinvested in infrastructure and other public
goods, and replaced subsidies by directed credit and credit guarantees.
Widespread recourse to illusory adjustments has led Easterly (1999) to
conclude that when outside institutions demand a reduction in the deficit
or debt, the affected government often responds by creating a fiscal
illusion: achieving more favorable deficit and debt figures while di-
vesting assets, accumulating contingent liabilities, and in other ways
eroding the government's net worth.
cies in particular lines of business (for example, agricultural credit and
guarantee funds, housing funds, and export credit funds), and specific
guarantees that cover anything from borrowing by state-owned enter-
prises to commercial risks facing private investors, all tend to grow
significantly as countries progress in their development.
Transition and emerging market economies face particularly large
fiscal risks. Their dependence on foreign private financing, weak regu-
latory and legal enforcement systems, opaque ownership and distorted
incentive structures, inadequate information disclosure, and the weak
disciplinary effects of the international financial markets tend to in-
crease the incidence of failures in the financial and corporate sectors.
Such failures in turn often generate political pressure on governments
to intervene through bailouts. A history of bailouts, particularly if
coupled with a lengthy tradition as a paternalistic state, only contrib-
utes to the spread of moral hazard in the markets.
In addition, transition and the emergence of new markets involve
enormous risks: by entrepreneurs in starting new businesses or ac-
quiring old ones; by investors in providing venture capital; by im-
porters and exporters in building new trade opportunities; by farmers
in facing volatile prices and competition; by state-owned enterprises
in taking on excessive risk pursuing profit or being barred from charg-
ing market prices; by workers in seeking employment free of govern-
ment intervention. Understandably, some economic agents seek to
transfer the risk to government explicitly by obtaining guarantees or
other forms of assurance that government support will be forthcom-
ing. Without extensive guarantees, there is some likelihood that pri-
vate enterprise will be stillborn or stunted, the inflow of capital will
INTRODUCTION 9
be inadequate, investors will be unwilling to acquire state enterprises,
and depositors will be reluctant to place their money in domestic
banking institutions.
Markets that have a short history and offer limited information
restrict the understanding that investors as well as politicians have of
the risks they are taking. Imperfect knowledge induces both investors
and politicians to underestimate the future potential cost of their deci-
sions. Underestimation tends to be greatest when costs are contingent
on future occurrences, such as repayment of loans or the performance
of enterprises, and when the government bears implicit obligations
that depend on future decisions, such as on whether to make good on
uninsured bank deposits. This factor explains in part why in many
transition and emerging market economies creditors have tolerated
excessive risk exposure by domestic financial institutions and enter-
prises before fleeing. During the early years of change in an economic
system, it is tempting for politicians to take the position that risks are
justified because they enable the economy to grow more robustly. Later,
politicians often feel that they have no choice but to assist troubled
enterprises and financial institutions. As economies integrate with the
international markets, more reliable data become available and more
scrutiny is demanded. This tendency enhances the ability of both gov-
ernments and investors to estimate risks with standard methodologies.
Investors are then more likely to become more cautious in buying gov-
ernment debt instruments and thus to subject governments to greater
cliscipline.
Scope for Fiscal Opportunism
Across countries, the main sources of fiscal risk and their underlying
political economy tend to be similar. We now review the most com-
rnon examples and highlight the scope for fiscal opportunism that ex-
ists in the various cases. We do not claim, however, that fiscal
opportunism actually arises in all countries and in all such cases.
The largest scope for fiscal opportunism is traditionally offered by
the financial sector. Governments are accustomed to using financial
institutions, private or state-owned, to finance various projects and
support programs. Development banks, policy banks, and credit and
guarantee funds are authorized by the government to borrow in the
markets to finance its programs. They raise resources to build roads,
power plants, or schools; provide credit to farmers, enterprises, or
health insurance funds-sometimes for new investment projects, other
times to cover operating losses; and offer guarantees to exporters or
developers. Because many such programs, although sometimes justifi-
able on policy grounds, are not profitable financially, financial institu-
tions accumulate liabilities without securing the revenues to pay them
10 BRIXI AND SCHICK
off. For the government, financial institutions appear to be a conve-
nient channel for promoting various agendas without directly burden-
ing the budget. But later, when the financial institution is unable to
roll over its debt, it ultimately needs government resources. If these
resources are provided directly from the proceeds of government bor-
rowing, the budget deficit remains unaffected. In some countries, gov-
ernment exercises substantial influence over the banking sector, and
financial institutions do not pursue the interest of their creditors and
depositors (without relying on government bailout). As noted earlier,
the result is widespread losses and possibly a banking crisis. In resolv-
ing a banking crisis, many governments again use the financial sector
to create a fiscal illusion. Often asset and management companies are
created for the sole purpose of raising revenues to recapitalize banks
outside the budgetary framework of the government (Klingebiel 2000).
In some countries, state-owned enterprises are the vehicle used to
implement programs of a fiscal nature. By giving state-owned enter-
prises the responsibility for providing unemployment benefits, pen-
sions, schooling, housing, and such, the government may again ensure
service delivery without directly burdening its budget. Later on, should
enterprises be short of resources to cover the cost of all these services,
the government would provide support-possibly through a financial
institution. When privatizing an enterprise, the government may have
to take over its obligations, which sometimes can be done in the form
of a guarantee issued by an autonomous privatization fund or credit
and guarantee fund (for example, by an environment fund to cover the
future environmental liabilities of the enterprise). On the other hand,
the government may be able to obtain a cash payment for assuming
some of the enterprise obligations, as was recently done in France for
enterprise pension obligations (Forte 1998). The government also may
require enterprises to charge artificially low prices for "necessities," in
effect avoiding the need to pay family transfers from the budget. To
cover the ensuing losses, the government may then issue a guarantee
on a credit to be taken by the enterprise from a commercial bank.
Ultimately, when the government has to provide financing it may do
so via a guarantee fallen due-again outside the budgetary framework.
Subnational governments are able to devise similar routes for op-
portunistic fiscal behavior. Many create their own financial institu-
tions to raise revenues for off-budget programs, issue guarantees, or
borrow directly in the financial markets. Because their obligations
appear to be backed implicitly by the central government, they find it
possible to raise funds even if the financial sustainability of their ac-
tivities raises doubts. Fiscal risks taken by subnational governments
are complicated by the fact that subnational policymakers may them-
selves rely on an implicit promise of the central government's help.
Depending on the political clout of each individual subnational gov-
INTRODUCTION 11
ernment, it may be untenable for a central government to let a
subnational government go bankrupt. Because most countries do not
have clear regulations or a monitoring system for subnational govern-
ment risk-taking, as financial markets develop, subnational govern-
ments tend to accumulate excessive obligations that eventually may
compel the central government to provide a bailout.4
In recent years, government promotion of private participation in
infrastructure, although often justified on policy grounds, has become a
major source of fiscal risks. The justifiable objective of promoting pri-
vate initiative may be diluted by the lack of political will to establish an
adequate pricing mechanism, unbundle monopolies, and introduce a
risk-sharing mechanism with the private developers and creditors. Ex-
plicitly through build-operate guarantee contracts, or implicitly through
a perceived responsibility for the provision of core services, the govern-
ment may be called on to step in with financing in case of failure.5
Public pension and health schemes are another common, albeit pre-
dictable, source of fiscal risk for governments. Arguably, society is
better off when such risks are pooled and when the pool is expanded
to cover all citizens or residents. But whatever its advantages, pooling
transfers the risk to future rather than current government budgets. In
an aging society, a promise of high pension and health benefits affects
future government finances enormously. Most governments, however,
still do not consider this intergenerational impact.
Reforms
Reforming any of these areas or, more broadly, how government deals
with fiscal risk is likely to be costly politically, because constituencies
from pensioners to bank owners have reasons to oppose such reforms.
The average taxpayer would benefit-but he or she usually lacks lob-
bying power. Furthermore, such reforms imply that policymakers might
have considerably less scope for fiscal maneuvering in the future. Chap-
ters of this book address the various aspects of fiscal opportunism as
they arise in different circumstances and make recommendations for
improvements in the light of their political feasibility.
This book has been prepared in recognition of the probability that
national governments will continue to shoulder various fiscal risks.
The contributors to this volume recognize that risk-taking by the gov-
ernment is justified in some instances. But they take the position that if
risks are here to stay, they should be properly regulated and managed,
with appropriate information and oversight and full accounting for
the costs that may be imposed on government.
It is assumed here that a necessary first step toward fiscally prudent
policies is for policymakers to identify, classify, and understand the
fiscal risks facing the government. Comprehension of the fiscal risks
12 BRIXI AND SCHICK
and their consequences may encourage governments to avoid the risks
that are bound to surface within a politically meaningful time horizon.
For risks that extend beyond that time frame, achievement of fiscally
sound behavior may depend on market discipline. In particular,
policymakers are more likely to gravitate toward fiscally sound deci-
sions if the media, the public, investors, credit-rating agencies, and
multilateral institutions understand the government's fiscal performance
in its entirety and if there are sanctions when politicians expose the
state to excessive risks and then conceal those risks.
The contributors to this volume seek to identify institutional mecha-
nisms that can be applied domestically and internationally to optimize
the amount of risk-taking by government. Domestically, an agency
that is insulated from direct political pressures-for example, a su-
preme audit institution or an autonomous government debt manage-
ment office-can assess and report on the direct and contingent fiscal
risks of each government agency and of government as a whole. Al-
though voters do not necessarily care about government fiscal risk,
public explanation of the fiscal risks by an independent audit office
may encourage the international forces of restraint. To be effective,
international restraint should be used to ensure that the government
applies the international rules for fiscal analysis not only to the budget
and debt, but also to contingent liabilities. Specifically, international
pressure may compel the government to meet certain quality stan-
dards: to define, measure, and monitor its full fiscal performance, us-
ing sound indicators and methods as defined by international authorities
such as the International Monetary Fund, World Bank, European Com-
mission, or sovereign credit rating agencies and investors.
Organization of This Book
This book explores the problem of fiscal risk along two dimensions.
One dimension is that of the entire government portfolio of fiscal risks;
the other is that of selected specific sources of government fiscal risk.
Accordingly, the book is divided into two parts, each offering a con-
ceptual treatment of the issues along with country examples.
Part I begins with an overview of different approaches to dealing
with government fiscal risks (Chapter 1 by Hana Polackova Brixi
and Ashoka Mody). The overview offers a classification of fiscal risks
(the Fiscal Risk Matrix) and, with extensive references to the exist-
ing literature and country practice, summarizes various analytical
and institutional approaches toward government management of fis-
cal risk. In particular, it outlines an approach to managing govern-
ment fiscal risk in the context of the portfolio of government assets,
sources of future revenues, and direct and contingent liabilities; sets
1rNTRODUCTION 13
policy formulation in the context of fiscal risk management; and of-
fers some guidance on structuring guarantees and government pro-
grams to minimize their risk.
Next, the book explores the institutional factors affecting the op-
portunistic behavior of policymakers and suggests corrective measures
writh examples of good practice across countries. In this context, Chap-
ters 2 and 3 by Murray Petrie and Allen Schick, respectively, illustrate
the inadequacies of conventional cash-basis reporting, accounting, and
budgeting, and call for more comprehensive and sounder approaches.
These chapters particularly highlight the usefulness of requiring gov-
ernment to publish a statement of contingent liabilities and fiscal risk.
T hey also outline the benefits of accrual-basis accounting and budget-
ing for government fiscal risk management.6 These chapters provide
many examples of country practice from developed, transition, and
developing countries.
Subsequent chapters explore the practice of risk management in the
private sector and its applicability to government. Chapter 4 by Suresh
Sundaresan provides an overview of the analytical tools and practices
used by financial institutions and corporations to manage their risk
exposures, and then applies the methodology to valuing government
guarantees. Similarly, Chapter 5 by Krishna Ramaswamy applies a
factor model to government risk analysis-particularly for risk-taking
by public sector entities, including state-owned enterprises.
Then, linking the discussion of government and private sector prac-
tice, the chapters that follow focus on approaches to expanding fiscal
analysis to incorporate fiscal risks and to bringing market incentives
into the government's thinking about fiscal sustainability. Chapter 6
by Daniel Cohen integrates contingent liabilities with the traditional
fiscal sustainability analysis and offers an institutional arrangement to
introduce market discipline into government risk-taking. Chapter 7 by
Richard Hemming and Murray Petrie further expands the fiscal
sustainability framework and introduces a framework for assessing
the exposure of a government's future fiscal performance to risks.
The country examples in Part I offer additional conceptual ap-
proaches and illustrate some of the discussion in the earlier chapters.
Reflecting on the ability of policymakers to generate fiscal illusion,
Chapter 8 by William Easterly and David Yuravlivker applies a com-
prehensive approach to fiscal analysis in the form of evaluating the net
worth of the governments of Colombia and Republica Bolivariana de
Venezuela. Chapter 9 by Hana Polackova Brixi, Allen Schick, and Leila
Zlaoui recognizes the special challenges in fiscal risk management fac-
ing transition countries and evaluates various aspects of the quality of
fiscal adjustment and fiscal management related to government con-
tingent liabilities in the Czech Republic, Bulgaria, and Hungary. In Chap-
ter 10, Kathie Krumm and Christine Wong incorporate contingent
14 BRIXI AND SCHICK
liabilities into the fiscal sustainability analysis for China. Looking at
the portfolio of contingent liabilities and risks from the perspective of
government debt, Chapter 11 by Hana Polackova Brixi and Sudarshan
Gooptu presents scenarios for government debt management in Indo-
nesia and Thailand. In a similar spirit, Chapter 12 by Juan Carlos
Echeverry and others discusses reforms in dealing with contingent li-
abilities as implemented under the leadership of the Colombia Treasury.
Part II presents analytical and institutional approaches that gov-
ernments might consider when facing risks in specific government
programs or sectors. Chapter 13 by George Pennacchi focuses on the
risk of guarantees that are often taken on by governments imple-
menting pension reforms and utilizes an option-pricing methodology
to value a government's risk exposure. In Chapter 14, Stijn Claessens
and Daniela Klingebiel analyze the ways in which to measure and
reduce the government's risk exposure in the banking sector. Chapter
15 by Ron Feldman discusses risks arising from government insur-
ance programs. Recognizing the implicit responsibility of the fiscal
authorities (and thus of the government budget) for the central bank's
positive net worth, Chapter 16 by Mario 1. Blejer and Liliana
Schumacher utilizes a value-at-risk approach to assessing the central
bank's own risk exposure.
Looking at country experience and practice, Chapter 17 by Ashoka
Mody analyzes the lessons of the 1997 East Asian financial crisis for
private participation in infrastructure and associated government con-
tingent liabilities. For fiscal risks taken on by subnational governments,
Chapter 18 by Jun Ma offers a framework that would allow the cen-
tral government to monitor and discipline the subnational governments'
risk exposure and thereby reduce the associated exposure by the cen-
tral government. Chapter 19 by Sweder van Wijnbergen and Nina
Budina applies option-pricing methodology to evaluating government
foreign debt restructuring agreements for Bulgaria. Finally, the fiscal
risk of floods, particularly in the case of Argentina, is explored in
Chapter 20 by Alcira Kreimer.
Reflecting on available country experience and on the new con-
cepts presented in the book, the concluding chapter by Allen Schick
draws together a list of policy recommendations for governments seek-
ing to bring their fiscal risks under control.
The descriptions and discussions in this book of the concepts and
practices of dealing with contingent liabilities and other fiscal risk
suggest that a broad range of approaches for governments to use in
analyzing and managing such risks are available. In this respect, the
book illustrates that contemporary practices have yet to be standard-
ized. Under these circumstances, the book seeks to motivate policy-
makers and policy analysts to pay attention to the fiscal risks govern-
ments face, and it provides a rich menu of practices that may be applied
in countries that are serious about confronting such risks.
INTRODUCTION 15
Notes
1. For an overview and analysis of the cost of banking crises, see Honohan
and Klingebiel (2000).
2. In addition, countries argued that they could not afford to let several
big airlines go under simultaneously, because the effect on jobs and confi-
dence could be too great. A temporary, targeted government subsidy to over-
come the temporary financial shock may be appropriate to smooth job
reductions and allow the strongest to survive. It should be acknowledged,
however, that the credit guarantees for which the airlines lobbied may have
the additional effects of delaying the restructuring in the airline industry
that, given the increasing losses of many airlines even before the attacks, was
considered overdue. For analysis of the pros and cons of various scopes and
forms of government support to the airline industry after the terrorist at-
tacks, see, for example, the Economist ("More Pain Ahead," September 22,
2001, and "Uncharted Airspace," September 28, 2001).
3. In cash-basis accounting, expenses and liabilities are accounted not when
the obligation is incurred, but only when the government makes the actual
cash transfer. Thus a government collecting a fee for assuming a liability (for
example, when it issues a guarantee or accepts the pension liability of an
enterprise under privatization) reports the transaction as a net revenue gain.
4. Dillinger (1999) discusses the economics and political economy of cen-
tral government bailouts of subnational governments in South America.
5. Irwin and others (1998) provide examples and analysis of public risk in
private infrastructure.
6. An accrual-basis accounting system without accrual budgeting will not
ensure that governments adequately consider contingent fiscal risks in policy.
Although this system encourages governments to prepare a statement of con-
tingent liabilities and financial risks, it generally does not require that the
liabilities be included in the balance sheet and that the associated risks be
evaluated and quantified. International accrual accounting standards require
that liabilities be accounted only when the obligation is due with certainty.
For a discussion of the rules of probability and risk assessment, see Interna-
tional Accounting Standards Committee (1997).
References
Dillinger, William. 1999. "Fiscal Management in Federal Democracies: Ar-
gentina and Brazil." Policy Research Working Paper 2121. World Bank,
Washington, D.C.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April).
Forte, Francesco. 1998. "Accounting and Financial Practices in Light of Con-
text of the Maastricht Treaty." In European Union Accession: The Chal-
lenges for Public Liability Management in Central Europe. Washington,
D.C.: World Bank.
16 BRIXI AND SCHICK
Honohan, Patrick, and Daniela Klingebiel. 2000. "Controlling Fiscal Cost of
Banking Crises." Policy Research Working Paper 2441. World Bank, Wash-
ington, D.C.
International Accounting Standards Committee. 1997. International Account-
ing Standards 1997. London.
Irwin, Timothy, Michael Klein, Guillermo Perry, and Mateen Thobani, eds.
1998. Dealing with Public Risk in Private Infrastructure. World Bank
Latin American and Caribbean Studies. Washington, D.C.: World Bank.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, E H. Rogers, and L. Squire, eds.
World Bank Economists' Forum. Washington, D.C.: World Bank.
Klingebiel, Daniela. 2000. "The Use of Asset Management Companies in the
Resolution of Banking Crises: Cross-country Experience." Policy Research
Working Paper 2284. World Bank, Washington, D.C.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Rubin, Irene. 1997. The Politics of Public Budgeting: Getting and Spending,
Borrowing and Balancing. Chatham, N.J.: Chatham House.
Schick, Allen. 1995. The Federal Budget: Politics, Policy, and Process. Wash-
ington, D.C.: Brookings Institution.
Selowsky, Marcelo. 1998. "Fiscal Deficits and the Quality of Fiscal Adjust-
ment." In The Challenges for Public Liability Management in Central
Europe. Washington, D.C.: World Bank.
U.S. GAO (General Accounting Office). 1998. Budgeting for Federal Insur-
ance Programs. Washington, D.C.: Government Printing Office.
PART I
Learning to Deal
with Fiscal Risks in
Government Portfolio
Possible Analytical
and Institutional Frameworks
CHAPTER I
Dealing with
Government Fiscal Risk:
An Overview
Hana Polackova Brixi
World Bank
Ashoka Mody*
International Monetary Fund
IMANY GOVERNMENTS HAVE FACED serious macroeconomic instabil-
ity as a result of obligations that were not recorded in any fiscal docu-
inents. Governments may have taken advantage of guarantees and
financial companies to implement their policies outside the budgetary
system, or they may have just been blind to risk spreading in the mar-
lcets. Whether the result of fiscal opportunism to conceal the true fis-
cal cost of government programs, or of an effort to find more efficient
ways to achieve policy objectives (by, for example, offering guarantees
instead of direct loans and cash subsidies), or of lenience toward moral
hazard in the behavior of market agents, such obligations often turn
out to be very costly. At some point in time, many guarantees fall due,
state insurance programs require subsidies, and banks involved in policy
lending or exposed to excessive risk with the hope of government bail-
out eventually file for such a bailout. And, as illustrated in Mexico in
1994, in East Asia in 1997, and, to some extent, around the world
after the September 2001 terrorist attack on the United States, these
hidden obligations tend to surface and require public resources all at
once in times of crises and economic slowdown.'
The author was with the World Bank when he wrote this article; he is currently with
the IMF.
21
22 BRIXI AND MODY
But the conventional approaches to public finance analysis fail to
reveal hidden government obligations and the associated fiscal risks.
Similarly, public finance institutions, including systems for government
budget management, debt management, accounting, financial control,
and public scrutiny, often remain blind to government contingent li-
abilities. Thus a string of years in which government has reported a
balanced budget and low public debt suggests neither that the govern-
ment has been fiscally prudent nor that it will enjoy fiscal stability in
the near future.2
This chapter begins by discussing a simple typology of sources of
fiscal risk-the Fiscal Risk Matrix. It then introduces the Fiscal Hedge
Matrix and expands on the standard government asset and liability
management framework. Using this framework, it outlines principles
for dealing with fiscal risks in their policy and institutional contexts.
The chapter then explores the possibilities for governments to reduce
their risk exposure by enabling market agents to better manage their
own risks, developing risk-sharing mechanisms, hedging, and building
reserves. Finally, the chapter offers a set of questions to assist policy-
makers in learning about fiscal risks in their own country. Overall, this
chapter introduces a number of topics that will be further elaborated
in the rest of the chapters in this volume.
The Fiscal Risk Matrix
The Fiscal Risk Matrix presented in Table 1.1 divides sources of fiscal
risk-that is, sources of future possible financing pressure on the fiscal
authorities of a country-into four groups according to the following
characteristics: direct versus contingent, and explicit versus implicit.3
Direct liabilities are predictable obligations that will arise in any event.
Contingent liabilities are obligations triggered by a discrete but uncer-
tain event. For government policies, the probability of a contingency
occurring and the magnitude of the required public outlay are exog-
enous (for example, the occurrence of a natural disaster) or endog-
enous (for example, the implications of market institutions and of the
design of government programs on moral hazard in markets). Contin-
gent liabilities also rise with weaknesses in the macroeconomic frame-
work, financial sector, and regulatory and supervisory systems, and
with weak information disclosure in the markets. They also emerge
from so-called quasi-fiscal activities-that is, activities of a fiscal na-
ture that the government pursues outside its budgetary framework.4
Explicit liabilities are specific government obligations defined by law
or contract. The government is legally mandated to settle such an ob-
ligation when it becomes due. Implicit liabilities represent a moral
obligation or expected burden for the government not in the legal sense,
but based on public expectations and political pressures.
DEALING WITH GOVERNMENT FISCAL RISK 23
7able 1.1 Government Fiscal Risk Matrix
Direct liabilities Contingent liabilities
Sources of (obligation (obligation if a
obligations in any event) particular event occurs)
Explicit * Sovereign debt (loans * State guarantees for non-
Government contracted and securities sovereign borrowing by and
liability as issued by central other obligations of sub-
recognized government) national governments and
by a law * Expenditure composi- public and private sector
or contract tion (nondiscretionary entities (development banks)
spending) * Umbrella state guarantees
* Expenditures legally for various types of loans
binding in the long term (mortgage loans, student
(civil service salaries loans, agriculture loans,
and pensions) small business loans)
* Trade and exchange rate
guarantees issued by the state
* State guarantees on private
investments
* State insurance schemes
(deposit insurance, income
from private pension funds,
crop insurance, flood in-
surance, war-risk insurance)
Implicit * Future public pensions * Default of a subnational
A moral (as opposed to civil government or public/private
obligation service pensions), entity on nonguaranteed
of govern- * Social security schemes- debt/obligations
ment that * Future health care * Banking failure (support
reflects financing, beyond government in-
public and * Future recurrent costs surance, if any)
interest group of public investment * Cleanup of liabilities of
pressures projects entities being privatized
* Failure of a nonguaranteed
pension fund, employment
fund, or social security fund
(protection of small investors)
* Possibly negative net worth
and/or default of central bank
on its obligations (foreign
exchange contracts, currency
defense, balance of payments)
* Other calls for bailouts (for
example, following a rever-
sal in private capital flows)
* Environmental recovery, dis-
aster relief, military financing
a. In this framework, these services fall in the category of government direct implicit
liabilities if their provision is not mandated by law. If mandated by law, then these services
fall in the category of government direct explicit liabilities.
Source: Polackova (1998).
24 BRIXI AND MODY
Government direct explicit liabilities are legal or contractual obli-
gations of the government that will arise in any event. These obliga-
tions constitute the main subject of conventional fiscal analysis. They
are: the repayment of sovereign debt, expenditures based on budget
law in the current fiscal year, and expenditures in the long term for
legally mandated items such as civil service salaries and pensions and,
in some countries, even the overall social security system. Among these
items, recent literature has focused on risks embedded in the size and
structure of the government debt portfolio (see Nars 1997, World Bank
and IMF 2000, and Dooley 2000 for an overview).
Government direct implicit liabilities will also arise in any event,
but the government will not be legally obliged to act on them. Such
obligations often arise as a presumed consequence of public expendi-
ture policies in the longer term. Given their implicit nature, these ob-
ligations are not captured in government balance sheets. Typically,
they are high for demographically driven expenditures. For example,
future pensions payable in a public pay-as-you-go scheme, unless guar-
anteed by law, constitute a direct implicit liability. Its size reflects the
expected generosity of and eligibility for pensions and future demo-
graphic and economic developments. Among direct implicit liabilities,
recent literature has particularly explored public pension liabilities
(World Bank 1994; IMF 1996; OECD 2000; Bodie and Davis 2000).
Contingent explicit liabilities are government legal obligations to
make a payment only if a particular event occurs. Because the fiscal
cost of contingent explicit liabilities is invisible until they are trig-
gered, they represent a hidden subsidy, blur fiscal analysis, and drain
government finances only later. For that reason, state guarantees and
financing through state-guaranteed institutions look politically more
attractive than budgetary support even if they are more expensive later.
In the markets, contingent government obligations may immediately
create moral hazard, particularly if the government guarantee covers
the whole rather than a part of the underlying assets and all rather
than selected political or commercial risks. State insurance schemes
often cover uninsurable risks of infrequent losses that are enormous in
total magnitude. Thus, rather than financing themselves from fees,
they redistribute wealth and rely on government net financing. To date,
research has focused on issues of measurement and management of
loan guarantees (Mody and Patro 1996; Lewis and Mody 1997), in-
vestment guarantees in infrastructure (Chase Manhattan Bank 1996;
Irwin and others 1998), state development finance institutions (Yaron
1992), pension guarantees (see Chapter 13 by Pennacchi in this vol-
ume), deposit insurance (Leaven 2000; World Bank 2001a), crop in-
surance (Hueth and Furtan 1994), and other state insurance schemes
(U.S. GAO 1998, and Chapter 15 by Feldman in this volume).
DEALING WITH GOVERNMENT FISCAL RISK 25
Contingent implicit liabilities depend on the occurrence of a par-
ticular future event and on government willingness to act on them.
Such obligations are typically not officially recognized until after a
failure occurs. The triggering event, the cost at risk, and the required
size of government outlay are uncertain. In most countries, the finan-
cial system represents the most serious contingent implicit government
liability. Experiences have indicated that markets expect the govern-
ment to help financially far beyond its legal obligation if stability of
the financial system is at risk (for examples, see Chapter 14 by Claessens
and Klingebiel in this volume and World Bank 2001a). Fiscal authori-
ties also are often compelled to cover losses and obligations of the
central bank, subnational governments, state-owned and large private
enterprises, budgetary and extrabudgetary agencies, and any other
agencies of political significance.5
The effect of fiscal risks arising from government direct and contin-
gent liabilities can best be analyzed in the context of an extended gov-
ernment balance sheet that includes future revenues as well as contingent
liabilities, and assets as well as direct liabilities. This approach builds
on the assets and liabilities management literature (Cassard and
Folkerts-Landau 1997; OECD 1999; World Bank and IMF 2000) and
can be referred to as an extended assets and liabilities management
framework. In this approach, a Fiscal Hedge Matrix (Table 1.2) comple-
ments the Fiscal Risk Matrix to illustrate the different sources of po-
tential revenues that can serve to cover government obligations. Sources
of government financial safety also can be divided into direct and con-
tingent, explicit and implicit. Direct explicit sources reflect the
government's legal power to raise income from its existing, tangible
assets. Direct implicit sources also are based on the existing assets, but
these are not under the government's direct control and thus may off-
set fiscal risks to a limited degree only. Contingent explicit sources
relate to the government's legal power to raise finances in the future
from sources other than its own assets. Finally, contingent implicit
sources are not available to the government until a particular situation
occurs and, even then, require the government to make a special case
for their utilization. The two matrixes (Tables 1.1 and 1.2), once filled
with country-specific items, would help a government to identify the
exact scope for its fiscal analysis and management.
The value of government assets and future revenues and the cost of
government obligations are associated with different types of risk. Clearly,
the government's residual, unhedged exposure to fiscal risk is the result
of the correlation among, rather than a simple summation of, the effects
of the different types of risks on the individual items in its extended
balance sheet. The types of risk include: refinancing risk (constraints on
the government's ability to issue debt-exacerbated particularly by short
26 BRIXI AND MODY
Table 1.2 Government Fiscal Hedge Matrix
Contingent sources of safety
Sources of Direct sources of safety (dependent on future events,
financial (based on the stock such as value generated
safety of existing assets) in the future)
Explicit * Assets recovery (work- * Government revenues from
Based on out and sales of non- resource extraction and sales
government performing loans and * Government customs
legal powers sales of equity) revenues
(ownership * Privatization of state- * Tax revenues
and the right owned enterprises and - minus tax expenditures
to raise other public resources (exclusions, exemptions,
revenues) * Recovery of govern- and deductions, which
ment loan assets reduce taxable income)
(resulting from earlier - minus revenue commit-
direct government ment (to subnational
lending) governments)
- minus revenues sold
forward (commodity
forward sales) and pledged
as collateral (partly at risk)
* Hedging instruments and
(re-)insurance policies
purchased by the government
from financial institutions
Implicit * Stabilization and contin- * Profits of state-owned
Based on gency funds' enterprises
government * Positive net worth of * Contingent credit lines and
indirect central bank financing commitments
control from official creditors
* Current account surpluses
across currencies
a. Stabilization and contingency funds may be designated for a general or very specific
purpose and can be under direct or indirect government control. Thus their classification
may be different in each case.
Source: The authors.
maturities of and maturity bunching in government obligations), liquid-
ity risk (risk of having to sell assets at loss-intensified by maturity
mismatch between assets and liabilities and by rigidities in the
government's capacity to raise revenues and cut expenditures), currency
risk (exchange rate risk and cross-currency risk, exposure to short-
term exchange rate volatility-arising from the currency structure of
government debt and exchange rate guarantees, which is partly offset
by the currency structure of government assets and revenues), interest
rate risk (particularly associated with floating interest rates), commodity
DEALING WITH GOVERNMENT FISCAL RISK 27
price risk (swings in the price of oil, rice, and similar commodities),
derivative risk (risk of large losses from the use of derivative instru-
ments), medium- and long-term sustainability risk (risk arising from
adverse trends that underlie government finances), political risk (risk
of policy reversals and political instability), and operational risk (poor
valuation and risk assessment, system errors, poor organizational struc-
tures, corruption, and fraud). The literature has mainly explored the
impact of most of these types of risk on government direct liabilities
(World Bank and IMF 2000) and of selected types of risk on govern-
rnent revenues and expenditures (see, for example, Larson, Varangis,
and Yabuki 1998 on the impact of commodity price). To analyze the
overall government risk exposure and its sensitivity to different risk
types, taking into account possible correlations among risks across the
entire extended government balance sheet, public finance will need to
build on tools developed in finance, such as portfolio optimization
and factor analysis (see Chapter 4 by Sundaresan and Chapter 5 by
Ramaswamy in this volume).
Dealing with Risk in Fiscal Analysis
and Fiscal Management
Because it is impossible for governments in a market environment to
avoid fiscal risk, they need to control and manage their risk exposure.
Dealing with fiscal risk is important not only from the perspective of
future fiscal stability. With respect to allocative efficiency, for example,
only with a view toward its likely full fiscal cost in the future can a
proposed government guarantee be properly scrutinized against other
competing programs. As for operational efficiency, only with a full un-
derstanding of the various types of risk involved can such a guarantee be
structured in a way to provide the desired support without unnecessar-
ily generating moral hazard and exposing the government to risk.
But do governments have the incentives and capacity to reflect fis-
cal risks in their policy choices and to carry out a good fiscal risk
management strategy? It depends on how well they understand the
issues and on the kind of scrutiny and pressures policymakers face in
dealing with fiscal risks and their consequences. There are good ex-
amples to build on. Australia and South Africa use a medium-term
expenditure framework to enhance predictability of fiscal performance
and, particularly in the case of South Africa, to make their govern-
ments accountable for their risk analysis as well as macroeconomic
and demographic assumptions.6 Canada, the Netherlands, and the
United States have incorporated analysis of selected contingent liabili-
ties and tax expenditures into their budgetary frameworks, requiring
budget allocations and reserve funds to reflect the present value of
28 BRIXI AND MODY
future potential outlays and foregone revenues (Congressional Research
Service 1998). Sweden and Colombia have authorized their govern-
ment debt management agencies to track and manage the risk of con-
tingent liabilities, and they require the beneficiaries of government
guarantees to pay the full present value of their expected fiscal cost
up-front into a reserve fund, which also is managed by the debt man-
agement agency (see Calderon Zuleta 1999 and Chapter 12 by
Echeverry and others in this volume). In India, the Federal Reserve
Bank and Ministry of Finance have carefully assessed the government's
exposure to fiscal risk across the entire Fiscal Risk Matrix and, as a
consequence, established a Guarantee Redemption Fund to cushion
the future fiscal cost of central government's guarantees. They also
implemented rules on subnational government guarantees and on
subnational guarantee funds to provide for the future expected cost of
subnational guarantees (Commonwealth Secretariat 2001).
The analysis and management of government exposure to fiscal risks
have three dimensions: the macroeconomic context, specific fiscal risks,
and the institutional framework. The macroeconomic context of the
government's exposure to fiscal risks relates to its capacity to absorb
financial pressures that it may realize in the future. How much room
for maneuver does the government have to absorb fiscal risks? Limits
on the government's absorption capacity for fiscal risks are determined
by possible macroeconomic constraints, such as the existence of a cur-
rency board and fixed exchange rate arrangements, and by the trends,
rigidity, and sensitivities of the general government's expenditures and
revenues, the size and liquidity of its assets, the structure of its obliga-
tions, and its future possible borrowing constraints. For example, does
the government have access to reliable sources of financing, such as
deep domestic bond markets, or is it critically dependent on the confi-
dence of foreign investors?
In this context, fiscal performance relates to developments in the
government's entire extended balance sheet (across the two matrixes
presented above). Thus fiscal sustainability analysis would be replaced
by an analysis of government net worth and of the future financial
pressures and financing options. In this volume, Chapter 8 by William
Easterly and David Yuravlivker applies the net worth approach to
Colombia and Republica Bolivariana de Venezuela. Focusing on obli-
gations, in Chapter 10 Kathie Krumm and Christine Wong integrate
contingent liabilities with fiscal sustainability analysis for China. In
Chapter 9 Hana Polackova Brixi, Allen Schick, and Leila Zlaoui try to
overcome the shortcomings of conventional deficit measurement and
calculate the "true fiscal deficit" and "hidden debt" of the govern-
ment of the Czech Republic, reflecting the cost of contingent liabili-
ties. For Bulgaria, they discuss the impact of the currency board
DEALING WITH GOVERNMENT FISCAL RISK 29
arrangement on the government's vulnerability to fiscal risks. Chapter
7 by Richard Hemming and Murray Petrie outlines a framework for
analyzing fiscal vulnerability, which is broadly defined as the ability of
government to achieve its fiscal policy objectives.
Dealing with specific fiscal risks requires addressing the following
three areas of questions: First, what are the sources of fiscal risk? Par-
ticularly, which budgetary and off-budget government programs may
generate unexpected financial pressure on the government in the fu-
ture? Second, what types of risk is the government exposed to? For
example, are the cost of the government's debt service, the value of its
revenues, or the likelihood that government contingent liabilities will
become reality affected by movements in the exchange rate, domestic
or foreign interest rate, and commodity prices, or by the risks of its
own fiscal mismanagement and operational failure (institutional risks)?
Third, how sensitive is the overall fiscal position to the various sources
and types of risk? What are the possible stress scenarios linking the
realization of different fiscal risks, and what are the possible conse-
quences for the government fiscal position?
The goal of government fiscal risk analysis (and management) is to
ensure that government has the cash available to meet its obligations
and deliver on its policies, under any more or less likely conditions.
Therefore, less likely stress scenarios require as much attention as the
most likely baseline scenario.7 Results of stress testing, reestimated
periodically for changes in the underlying assumptions (to mark the
expected fiscal outcomes to market), will be critical for government
risk management in deciding, for example, on an appropriate level of
reserves and hedging strategy. In Chapter 11 in this volume, Brixi and
Gooptu, taking into account the correlation in the sensitivity of the
different items in the Fiscal Risk and Fiscal Hedge Matrixes to the
different types of risk, build a stress scenario for the government of
Indonesia.
For major contingent liabilities, how does one assess their size, prob-
ability of realization, and possible future fiscal effects? The literature
on valuing credit and project guarantees and state insurance programs
has built on contingent claims analysis (see Mody and Patro 1996,
Lewis and Mody 1997, Arthur Andersen 2000, and Marrison 2001
for an overview). Contingent claim analysis uses two basic concepts:
the expected costs (that is, the most likely or average cost) and the
unexpected costs (that is, the maximum likely cost with a particular
small probability, also referred to as value at risk or cost at risk) of the
contingent liabilities.
To reveal both the average and the maximum likely fiscal cost of
government contingent liabilities, one would generate a large number
of random scenarios (for example, using Monte Carlo simulations, as
30 BRIXI AND MODY
described in Arthur Andersen 2000 and Marrison 2001). In the distri-
bution of outcomes of the different scenarios, scenarios that occurred
most often would indicate the average fiscal cost and scenarios (also
referred to as stress scenarios) that occurred at a predetermined
probability (usually 5 percent or 10 percent) would indicate the cost
at risk. (In this volume, valuation of contingent liabilities using value-
at-risk and option-pricing approaches are discussed particularly in
Chapter 4 by Sundaresan, Chapter 13 by Pennacchi, and Chapter 19
by van Wijnbergen and Budina. For an overview of different approaches
as they have already been applied, see U.S. GAO 1993, 1998.)
As such, the average cost of contingent liabilities is a measure of
government subsidy implied by the issuance of a government guaran-
tee or other program of contingent government support. Thus from a
policy point of view, the average cost estimate can be used to judge
whether the government would be willing to support the project through
an equivalent up-front cash subsidy (in which case, as further discus-
sion will point out, the government also should be willing to budget
for the expected cost of contingent liabilities at the time of assuming
them). The stress scenario implies a fiscal cost that the government has
to consider in its risk management approach and has to be prepared to
pay in the future.
The institutional framework for dealing with fiscal risks mainly
relates to the rules and practice of information disclosure, monitoring,
fiscal planning, and budgeting. The institutional framework affects
the government's incentives and ability to constrain, control, and man-
age its fiscal risks. The framework must be such that it promotes a
risk-awareness culture in government and minimizes the scope for fis-
cal opportunism. Whether or not policymakers extend the focus of
public finance institutions to cover fiscal risks depends on several fac-
tors, including the definition and measurement of internationally rec-
ognized fiscal indicators (that is, the pressure of institutions such as
EUROSTAT, the International Monetary Fund, the World Bank, the
Organisation for Economic Co-operation and Development, and the
United Nations), public pressure (by the media, independent audit in-
stitutions, watchdog agencies, and legislators), investors' demands and
preferences (for example, to what extent sovereign credit rating agen-
cies pay attention to fiscal risks, and investors punish the government
for concealing relevant data and exposing the country to excessive
fiscal risk), and whether or not fiscal risks attract the attention of
reform-minded policymakers.
Fiscal risks are likely to attract more attention if the government is
required to disclose them. Disclosure can be in the form of a simple
statement of contingent liabilities and tax expenditures, or full-fledged
financial statements based on an accrual accounting system. The gov-
ernments of Australia, Canada, the Netherlands, New Zealand, and
DEALING WITH GOVERNMENT FISCAL RISK 31
the United States offer good practices to follow. The fiscal statements
of these countries list the various sources of fiscal risk; discuss their
nature and sensitivities, implications for future fiscal position, and
allocative efficiency (compared with budgetary spending); and, where
applicable, provide their face value or estimated future fiscal cost or
both (see Chapter 2 by Petrie for examples and references). Govern-
ments also should enforce requirements that market agents disclose
information about their risk exposures. Particularly agencies that may
appear to be implicitly guaranteed by the government should be sub-
ject to strict disclosure requirements. In this context, for example,
subnational governments should report on their guarantees and on the
activities of their own financial enterprises; state-owned enterprises
should report on their environmental commitments; and financial in-
stitutions should report on their off-balance sheet items. Such broad
disclosure would allow market agents as well as the government to
conduct proper monitoring and analysis and to react to possible moral
hazard.
Although the organizational setup for fiscal risk management will
be specific to every country, some general principles apply (see Box
1.1). These include: centralize the risk-taking authority (possibly the
Box 1.1 Division of Responsibilities for Fiscal Risk
Management
Large banks, including J. P. Morgan and Deutsche Bank, have divided
the functions of designing and authorizing new transactions, analysis,
and record-keeping among three different offices. The front office serves
as the central point for the design of financial instruments, and it has
the exclusive authority to enter into new derivative and debt transac-
tions. Its objective is to ensure the required levels of available cash and
optimize the overall return-risk ratio. The middle office provides analysis
of future obligations and payoffs and their sensitivities for the entire
portfolio. Finally, the back office is responsible for record-keeping and
maintaining comprehensive databases.
Maintaining these functions independent of each other improves
transparency and the control of portfolio risks and prevents the front
office from exceeding its predetermined risk exposure limits. The gov-
ernments of Ireland and Sweden, for example, have successfully ap-
plied such a division of responsibilities to their debt management.
Many governments have centralized the authority to issue debt,
guarantees, tax exemptions, and other off-budget programs. Now they
should expand the scope of risk management and adjust organizational
structures and responsibilities accordingly. See Nars (1997).
32 BRIXI AND MODY
ministry of finance to oversee any risk-taking in the public sector);
separate risk-monitoring (for example, internally by the debt man-
agement office and externally by the supreme audit institution) from
risk-taking; and connect risk-taking with budgeting and debt man-
agement practice. Obviously, accountability structures are crucial to
ensure the best efforts and reduce the scope for fraud and corruption.
Policymakers and civil servants need to be accountable for the ad-
equacy of their risk analysis and the assumptions underlying deci-
sions that involve fiscal risks and for managing the overall government
risk exposure. Therefore, the role, of the independent audit (and the
supreme audit institution), as it is in the case of the U.S. General
Accounting Office, would extend beyond its conventional limits to
cover all aspects of government risk analysis and risk management
(see Chapter 2 by Petrie).
Budgeting and accounting rules influence the allocation of resources,
affect the timing and recognition of transactions, and may provide
opportunities and incentives to shift costs and risks from one period to
another and from one part of a government to another. Cash flow
budgeting, for example, makes guarantees and other contingent forms
of government support look more attractive than direct loans or cash
subsidies. It treats direct loans and subsidies as outlays, but does not
recognize contingent liabilities until default occurs, at which point the
government has little choice but to make good on past commitments.8
Direct loans and subsidies thus appear expensive and the contingent
form of support cheap. To make matters worse, in cash flow budget-
ing income earned from origination fees on guarantees is booked as
current revenue, making it appear that the government is profiting by
taking these risks.
In contrast, an accrual-based budgeting and accounting system re-
quires that the net present fiscal cost associated with individual gov-
ernment programs, including programs that generate a contingent
liability, be included in budget documents and thus be made visible
from the moment government decides to launch them. Moreover, for
contingent liabilities it encourages the government to set aside resources
up-front at the time of their launching (see Box 1.2 for a private sector
analog and Chapter 3 by Schick for a detailed discussion).9 To strengthen
accountability, the budget would be set in the context of a publicly
announced medium-term fiscal framework, which would later make
any departures from the original risk analysis apparent.
The experience of a number of countries (most of which is dis-
cussed in this book), including Canada, Colombia, Hungary, the Neth-
erlands, South Africa, Sweden, and the United States, has indicated
that a comprehensive shift of the accounting and budgeting systems to
accrual basis is not necessary in order for government to take control
'DEALING WITH GOVERNMENT FISCAL RISK 33
Box 1.2 Budgeting for Risk in the Private Sector
Programs of contingent support are often akin to put options, which
create the obligation, but not the right, to buy an asset at certain pre-
defined strike levels. Charging the full option price when writing (sell-
ing) an option, as is typical in the private sector, amounts to immediately
fully provisioning the expected cost of contingent support. The price
of an option reflects the present value of the future possible loss, which
may be incurred by the underwriting institution. As illustrated by the
Black-Scholes formula, the price increases with the time to expiry (for
example, maturity of the guaranteed loan) and with the volatility of
the underlying asset (for example, share price of the enterprise whose
debt is under the guarantee). Financial institutions charge the full op-
tion price immediately at the time of selling the option. The amount is
then used either to build reserves or to buy a hedge. See Hull (1997).
of its fiscal risks.'0 Tables 1.3 and 1.4 summarize measures to enhance
policymakers' understanding of fiscal risks and their incentives as well
as capacity for dealing with them. These measures can be built upon
any decent public finance management system.
Reducing Government Risk Exposure
A number of commodity-exporting nations have tried to manage their
risk exposure by building reserves in booms and creating stabilization
funds, but the results have been mixed (Davis and others 2001 provide
an overview). With more success, several governments have used de-
rivatives and "exotic" debt instruments (such as debt instruments linking
the amount of debt repayment to commodity prices) to hedge risks in
the government assets and liabilities portfolio." Other governments
also have successfully purchased reinsurance for risks (such as weather
risks) from a large international reinsurer.'2 In recent years, increasing
market integration has made it possible to pool risk across countries
and thus has enabled financial markets to provide insurance against
risks, such as crop risks and disaster risks, that had been considered
uninsurable before. Financial markets have typically welcomed gov-
erinent risk management initiatives (in the early 1990s, for example,
the government of New Zealand witnessed a rapid improvement in the
terms of its sovereign borrowing once it announced and implemented
34 BRIXI AND MODY
Table 1.3 Systemic Measures to Reduce Government
Exposure to Fiscal Risk
Fiscal policy Fiscal management
First- First-
* Identify and reconsider govern- * Assign responsibility for identifying,
ment programs and promises recording, and reporting government
that imply significant risks, no obligations, assessing the future
longer serve a significant social likely fiscal cost, and monitoring
or economic purpose, or can be government exposure to fiscal risk.
replaced by market instruments * Analyze the sensitivity of the
such as private insurance, government's fiscal position to the
derivatives. different types of risks and create
* Identify, classify, and analyze all a list of early warning indicators
major sources of fiscal risk and that signal possible future fiscal
financial safety (build an extended pressures (for example, exchange
government balance sheet). rate movements, rate of credit
* Announce that only those contin- growth in the banking system,
gent liabilities included in a public and demand growth vis-a-vis the
statement of fiscal risk will be existing capacity in infrastructure).
honored. * Establish nominal, nonbudgetary
* As a context to policy decision- control mechanisms for fiscal risks
making, introduce the concept such as information disclosure on
of fiscal vulnerability. government obligations and tax
expenditures, exposure limits (for
example, maximum amount of
guarantees outstanding), and ear-
marking of future funds to cover the
likely costs of contingent liabilities.
Later- Later-
* For policy decisions, analyze full * Recognize and disclose information
fiscal performance and its vulner- about government full fiscal perfor-
ability to risks and expand the mance, including exposure to
medium-term fiscal framework fiscal risk.
to cover the effects of fiscal risks. * Scrutinize fiscal risks in the budget
* Identify the government's risk process before they are taken.
absorption capacity and (accord- * Consider government risk exposure
ing to its risk preference and its in structuring programs.
capacity to absorb and manage * Build capacity for analyzing and
risk) determine the government's managing risk (including mitigation
optimal risk exposure and reserve of the risk at source, transferring
and hedging policy. the risk to parties better able to
* Develop a government risk manage- bear the risk, and monitoring and
ment strategy that would guide managing any residual risk that
policymakers and staff in day-to- cannot be mitigated or transferred)
day decisions involving govern- and for auditing government
ment risk exposure. dealing with fiscal risk.
* Consider private sector coverage * Build a mechanism to enforce the
to replace existing and proposed disclosure, monitoring, and regula-
government programs. tion of risks in both the public and
private sectors.
Source: Polackova (1998) and the authors.
DEALING WITH GOVERNMENT FISCAL RISK 35
Table 1.4 Measures to Control the Fiscal Risks of Individual
Government Programs
Fiscal policy Fiscal management
Before accepting- Before accepting-
* Assess how the program fits with * As part of fiscal planning and
policy objectives. budgeting, evaluate the risks and
* Consider the program's risks and estimate the likely future fiscal
likely future fiscal cost and com- cost.
pare alternative forms of govern- * Design the program well, provid-
ment support, including actions to ing for risk mitigation and risk
enable private sector coverage. transfer, to minimize government
* Outline and announce the limits risk exposure and moral hazard in
of government responsibility with the markets.
respect to the program in order to * Decide on the management of
minimize moral hazard. residual risk (for example, set a
hedge and an additional reserve
requirement).
When accepted- When accepted-
* Stick to the preset limits of the * Report the risk (for example, issue
scope of the program and of the a statement of fiscal risks).
associated government * Budget for the net present value of
responsibility. the expected fiscal cost of the
* Reconsider the program's program.
relevance in the context of the * Continuously monitor the perfor-
evolving needs, changing structure mance under the program, including
of the economy and role of the the program's risk factors, reserve
government, and advances in adequacy, and behavioral effects.
technology and financial markets. * Update the risk analysis and risk
management strategy.
* Audit the validity of the risk analysis
and the quality of risk management.
* Draw consequences if a bias in the
original risk analysis and govern-
ment decision is revealed.
* Prepare contingency plans for
dealing with the program (whether
explicit or implicit) if fiscal risks
are realized.
Upon execution- Upon execution-
* Execute strictly within the * Evaluate performance under the
preset limits of government program, compare and report the
responsibility. actual fiscal cost versus the original
* If implicit, assess the fit with cost estimates, and punish for
policy priorities and possible failures (including a bias in risk
moral hazard effects before analysis and deficiency in risk
executing. management).
Source: Polackova (1998) and the authors.
36 BRIXI AND MODY
its risk management strategy). Since the 1997 crisis in Asia, invest-
ment banks and sovereign credit rating agencies have increasingly dis-
cussed government risk exposure in their country risk analyses (see,
for example, Standard and Poor's 1997).
For governments and for enterprises, the objective of risk manage-
ment is to align the demand for funds with revenues (Froot, Scharfstein,
and Stein 1994). Private companies and financial institutions have
benefited from using enterprise-wide risk management strategies and
financial markets to manage and hedge risk. In addition to learning
from their experience, which is discussed in Chapter 4 by Sundaresan,
governments often also have the option of supporting broader reforms
that reduce risks or make them insurable in the markets.
For governments, the management of risk entails three complemen-
tary tasks: involving the private sector (mitigating the risk at source
and developing financial risk markets), transferring the risk to parties
better able to bear the risk (creating risk-sharing arrangements), and
managing any residual risk that cannot be mitigated or transferred
(monitoring, building reserves, and hedging).
Ultimately, risk mitigation with private sector involvement is the
most desirable long-run strategy, because it not only reduces the
government's exposure to fiscal risk but also reduces the vulnerability
of the economy to shocks. Instead of assuming risk, governments would
enable markets to deal with it. For example, a power sector that is
organized to permit competitive generation and distribution will fos-
ter efficient use of resources while at the same time lowering the risk
arising from excessive installation of capacity (for other examples, see
Table 1.5).
Similarly, by supporting the development of the markets for risk
instruments, the government can effectively withdraw from its direct
role in dealing with many risks. In this regard, policymakers need to
ask: Are the risks for which government coverage is sought truly unin-
surable in the private sector? How can these risks be made insurable?
For example, regulatory changes can encourage large international
insurers to access the local market and pool risks uninsurable in a
small economy and can make derivatives accessible to local market
agents. Private risk markets reduce the need for the traditional govern-
ment programs such as disaster risk insurance, crop insurance, and
minimum price policies. Similarly, new financial instruments may help
domestic financial institutions to better manage risk, thereby reducing
their demand for government guarantees. However, risk mitigation
strategies and markets for risk instruments may require fundamental
sectoral reforms and thus cannot be developed overnight. (For a dis-
cussion linking fiscal risk with sectoral reform, see, for example, Irwin
and others 1998). Therefore, more effective risk-sharing may be the
practical short-term strategy.
DEALING WITH GOVERNMENT FISCAL RISK 37
Table 1.5 Risk, Possible Fiscal Cost, and Private Sector
Solution
Type of risk Coverage Possible fiscal cost Private sector solution
Credit Debt service and Principal plus Credit enhancement
guarantees losses due to interest plus Risk-sharing with
default possible penalties creditors
by the creditor
for default
Guarantee Minimum ab- Guaranteed Sound regulatory frame-
on minimum solute amount amount times work for pension
return of (monetary value) the number of funds and the overall
pension funds pensioners financial markets
Minimum relative Average wage
amount (share share times the
of average wage) number of
pensioners
Project Design and Very large if not Correcting for any
guarantees development capped market failures that
Construction risk reduce access of
Operating risk investors to adequate
(cost overrun, risk protection
delays) mechanisms in the
Demand/revenue markets
risk Risk-sharing with
Financial risk investors and
(exchange rate, creditors
interest rate)
Force majeure
Environmental risk
Political and policy
risk
Disaster Losses due to Very large if Environment allow-
Insurance disasters notcapped ing direct access to
international in-
surers and reinsurers,
catastrophe bonds
and derivatives
Deposit Bank deposits Face value of Regulations allowing
Insurance all deposits if bank access to risk
not capped protection and risk
markets
Disclosure of informa-
tion on bank perfor-
mance and manage-
ment, international
competition, low
limits on the deposit
amounts guaranteed
(Table continues on the following page.)
38 BRIXI AND MODY
Table 1.5 (continued)
Type of risk Coverage Possible fiscal cost Private sector solution
Price Minimum price Guaranteed min- Encourage direct
support of a product/ imum price access to interna-
commodity minus actual tional derivatives
price, multiplied markets
by quantity
Implicit Explicitlv: none Almost unlimited Sound regulatory frame-
guarantee Implicitly: obliga- work for accounting,
on various tions of subnation- information dis-
obligations al governments, closure, and audit
state-owned finan- Credible announce-
cial institutions, ments and actions by
enterprises and the government to
funds, large mar- minimize expecta-
ket agents, etc. tions of a bailout
Source: The authors.
Risk transfer and a good risk-sharing mechanism require clear policy
objectives and understanding of all the underlying risks in a program.
In the private sector in the last 20 years, the possibilities for transfer-
ring risk have been growing rapidly. Derivative products have allowed
firms to subdivide, isolate, and swap various risks, but they also have
created new risk exposures that are not easy to quantify (Garber 1998).
So far, for governments the primary method of transferring risk has
been through risk-sharing provisions in their guarantees and insurance
contracts. Recent practice and literature have suggested that carving
out commercial risk from government coverage significantly reduces
the negative behavioral effects of such government programs in the
markets as well as limits government risk exposure (U.S. GAO 1998;
World Bank 1999c, 2000). For example, in the banking sector, should
the government provide deposit insurance, then a relatively low cap
on government protection would increase the incentives of bankers to
improve due diligence and project selection, lowering risk and the
wasteful use of resources. Similarly, price support programs require
their own risk-sharing arrangement with, for example, farmers (see
Box 1.3).
For implicit contingent liabilities, risk-sharing tends to be applied
ex post. As Honohan (1999) argues, however, fiscal cost is lower (and
government crisis management more efficient) if government has an
ex ante, confidential contingency plan (for example, deciding ex ante
which stakeholders-domestic depositors in local currency, domestic
depositors in foreign currency, foreign depositors, creditors, and share-
holders-to assist and how much before a crisis occurs). Similarly, the
r)EALING WITH GOVERNMENT FISCAL RISK 39
Box 1.3 Government Price Support Programs for Farmers
International markets are not always able to offer adequate instruments,
such as futures, options, or insurance policies, to protect farmers against
the volatility of their particular product. In reality, it may be difficult for a
farmer to sell (short hedge) futures on his or her product in order to make
the final selling price certain and thus protect against losses from a pos-
sible future reduction in its price. Thus the government may still be the
only source of protection. (Even when markets do not offer instruments
of adequate protection to individual farmers, they may offer hedging or
reinsurance instruments on a larger scale and customized basis to govern-
ment-for example, customized derivatives contracts over-the-counter or
a reinsurance policy). In many countries, government protection comes in
the form of a price support program, which guarantees farmers a mini-
mum price for their output. Such programs generate for the government
an obligation to pay farmers the difference between the market price and
the guaranteed minimum price should the price of their product drop
below the minimum price.
If price support turns out to be the preferred choice of government
support, how should the program be designed to minimize both moral
hazard on the side of the farmers and the future fiscal cost on the side of
the government? To deter overproduction, the amount of product guar-
anteed must be limited-by imposing a nominal ceiling and quotas and
by charging the farmers a fee per unit of guaranteed product. (For a par-
ticular minimum price, the Black Scholes options-pricing formula allows
one to determine the fee. As Cox and Rubinstein 1985 explain, the only
required variables for the calculation will be the given minimum price,
the actual price, the volatility of actual prices over past years, and the
time to expiration-that is, the number of days ahead for the minimum
price to apply. Alternatively, the government may sell the limited amount
of per-unit price guarantees in an auction). If the objective is to provide
subsistence to temporarily impoverished farmers and encourage them to
sustain a limited amount of land or number of bushels or livestock (rather
than flood the market with excess production), the program may be effi-
ciently designed as a put spread, setting not only the minimum price but
also the maximum amount of support paid to farmers per unit of product.
For example, if the minimum price is set at 100 and the maximum amount
of government support at 20, government pays 15 if the actual price is 85,
but no more than 20 even if the actual price drops well below 80.
Changing tastes that may cut demand for a particular commodity (such
as lamb and beef in the early 21st century) and cause it to drop in price
erode the rationale of the commodity's strategic importance. Thus the
risk of a continual drop in demand should not belong to the government.
Also, risks that the quality of (and thus the price that can be charged for)
a domestic product drops compared with the quality and price of that
sold by international competitors, or that new technology and fertilizers
drive prices down permanently, should not belong to the government.
These are reasonably well under the farmers' control. Thus a price sup-
port program would become effective only if the reasons for a low price
are temporary and clearly and entirely out of the farmers' control.
40 BRIXI AND MODY
central government can impose limits, either confidentially or pub-
licly, for its responsibilities in the event a subnational government goes
bankrupt (for example, to ensure the water supply but not the mayor's
salary).
As for dealing with any residual risk that cannot be mitigated or
transferred, this also is best done within the expanded assets and li-
abilities framework (for literature on the portfolio approach to risk
management, see Cassard and Folkerts-Landau 1997; World Bank 2000;
World Bank and IMF 2000). Across the two matrixes presented in
Tables 1.1 and 1.2, analysis and stress testing of the impact of various
types of risks help to account for the correlation in the value of gov-
ernment spending pressures and proceeds under different scenarios,
and to identify the government's residual, unhedged risk exposure (the
types of risk that have the strongest overall fiscal effect).
The government then has two basic approaches to dealing with its
risk exposure and to protecting itself against rare events: building up
reserves and using financial hedges. Once a government determines
the level of loss it is capable of and willing to absorb, it can establish
reserves against unexpected losses. Several concerns, though, are asso-
ciated with provisioning. First, setting reserves on the basis of portfo-
lio risk analysis (as opposed to the additive loss exposure of each
government program) is advantageous but it has yet to be fully tested
by governments (Lewis and Mody 1997). Second, reserves have an
opportunity cost-and one that is particularly high in bad times and in
poor countries. The challenge is to acknowledge the opportunity cost
of not having reserves-when government is stuck, unprepared for a
sudden increase in its obligations or a drop in its revenues. For govern-
ments that find themselves unable to obtain favorable credit in such
situations, fiscal crises become a reality with all their negative conse-
quences. Furthermore, how should reserves be invested, and who will
be responsible for investment decisions? How does government ensure
reserve adequacy? And how does government prevent the misuse of
reserves? Arguably, politicians will always find ways to tap reserves
even for purposes other than those originally intended. The experience
in many countries, particularly with stabilization funds, indicates that
neither laws nor rules prevent misuse.
Analysts have been searching for possible approaches to improving
reserve adequacy and reducing possible misuse. In Chapter 6 of this
volume, Daniel Cohen proposes that countries with developed capital
markets create a transparent reserve fund and sell its shares to private
owners. Market mechanisms would serve to discover the share price
of the reserve fund, primarily reflecting on the adequacy of its capitali-
zation and of its use. For countries with underdeveloped capital mar-
kets, a reputable foreign institution may be entrusted with the task of
DEALING WITH GOVERNMENT FISCAL RISK 41
Box 1.4 Margin Calls to Collateralize Risk
Learning from the practice of margin calls applied by investment banks,
government may be able to reduce moral hazard in the markets and its
own risk exposure by requiring beneficiaries of its programs to make
collateral payments when their performance deteriorates. The collat-
eral "penalty" would be calculated as the increase in the mark-to-market
fiscal cost. This practice would encourage the beneficiaries of govern-
ment programs to limit their own risk exposure and generate resources
for the government contingency reserve fund when the government's
risk exposure increases. But it would demand tight monitoring of per-
formance in the real and financial sectors.
Investment banks periodically monitor their credit risk exposure on
clients' portfolios against predetermined, uncollateralized limits and
require clients to make collateral "penalty" payments for the excess
mark-to-market value of a potential loss over the limit. The limit is
defined ex ante, as part of the contractual agreement between the bank
and the client, often in terms of both the most likely loss and value at
risk facing the bank with respect to a specific sector, region, or market
segment. For a specific portfolio, when assumptions underlying its risk
analysis deteriorate, the bank requires the client (a "margin" call) to
immediately make a collateral payment equal to the excess of the mark-
to-market potential loss over the limit.
managing the reserves under predetermined parameters of risk and
returns. A contract would specify permissible claims on reserves and
make other claims subject to a penalty and ex ante public disclosure.
Both approaches could be complemented with margin calls to collaterize
risk (see Box 1.4).
Hedging does not fully substitute for reserves, because all contin-
gencies cannot be foreseen and market hedges are not available for
contingencies that can be visualized.'3 However, where hedging is pos-
sible, it may be superior to building reserves for governments with
good capacities and control mechanisms. The fiscal costs of different
government programs may be negatively correlated,'4 but creating
additional government programs with the sole objective of risk-pool-
ing would be a questionable practice for government. Therefore, hedg-
ing and purchase of reinsurance may be needed to complement pooling.
Take, for example, a government that largely depends on tax rev-
enues from copper sales. When the price of copper goes down, the
government is short on revenues. If its access to borrowing is limited,
it has to cut public expenditures abruptly. Instead of building a revenue
42 BRIXI AND MODY
stabilization fund, the government can look for possibilities to stabi-
lize its fiscal performance by structuring its obligations to reflect cop-
per prices. For example, the government may attempt to link its
liabilities to the source of volatility-that is, to issue bonds that offer a
yield inversely linked to copper price. Structuring liabilities according
to the sources of volatility in the government portfolio of contingent
liabilities and revenues reduces the overall volatility in the future fiscal
outlook and thus offers an alternative to stabilization reserve funds.
When revenues are low, debt service will become less expensive and
vice versa. Private insurance companies have similar experience in is-
suing catastrophe bonds, which offer lower yields when a hurricane
occurs than when it does not.i" The government also may purchase
customized derivatives that will deliver a positive payoff, inversely
related to the copper price, when the price of copper drops below a
specific threshold.
Similarly, governments have found ways to hedge their risks through
derivatives applied to their obligations portfolio. Many governments-
most notably, Belgium, Colombia, Hungary, Ireland, New Zealand,
and Sweden-have experimented with asset and liability management
approaches in order to match currency, interest rate, and maturity risks
in the portfolios of their liabilities and assets. They have utilized inter-
est rate swaps, currency swaps, currency forwards, and other deriva-
tives to achieve the desired risk profile in their debt portfolio.
Possibilities to hedge risks of contingent liabilities have been ex-
plored mainly with respect to minimum price support policies and
crop insurance. To hedge their risk exposure, several governments (for
example, of Colombia and Mexico) have been purchasing futures and
customized forward contracts. (For an example of the use of options
to hedge price support policy, see Box 1.5). Ideally, the price of the
hedge would be passed on to the program beneficiaries (for example,
as a fee to be charged the farmers discussed in Box 1.3). Some finan-
cial instruments, such as catastrophe bonds, which link their yield in-
versely to the occurrence of a particular catastrophe, would help to
hedge risks of associated contingent liabilities and revenue decline.
For more examples, using a combination of the above approaches, see
Table 1.6.
The borderline between hedging and speculation is, however, some-
times difficult to draw. Recent experiences of companies and hedge
funds have reconfirmed that derivatives that provide less than a per-
fect hedge may generate risks on their own. Also, the use of derivatives
is dangerous if there is not a clear strategy on which specific risks to
hedge and to what extent to hedge them. For government as well as
for private companies, the ad hoc availability of new hedging instru-
ments, and the attractions of financial engineering, should never drive
risk management decisions. Potentially risky hedging techniques in-
DEALING WITH GOVERNMENT FISCAL RISK 43
Box 1.5 Securitizing Government Risk
How does a government hedge the risk of price support policies? Sup-
pose the government offers a minimum price guarantee on a commod-
ity. Assuming the floor price is set at $10 per unit, the government pays
the difference between the floor and actual price if the actual price falls
below $10 per unit (see graph).
This payoff exactly illustrates that providing a price support policy
is equal to shorting a put option (selling the right to sell a commodity
at a specified minimum price). To hedge against possible losses, the
replicating strategy suggests buying puts from international financial
intermediaries. The cost to the government is the difference between
the total fees collected by the government from the commodity pro-
ducers (if possible in an auction) and the price paid by the government
for the put. The strategy allows the government to convert its fiscal
cost from the form of an unknown contingent liability to a fixed, up-
front payment.
clude a dynamic hedging strategy, which requires readjusting the hedg-
ing mechanism as its underlying assumptions change (Hull 1997).
Therefore, before a treasurer or debt managers launch sophisticated
risk management techniques, top policymakers need to decide to what
extent the government should be exposed to risk-that is, to what
extent should the government require that its expected fiscal pressures
be matched with its actual resources? Addressing this question, in turn,
will depend on the extent to which government can rely on ad hoc
borrowing and tax increases and on the extent to which the govern-
rnent can afford to restructure or default on its budget programs (are
arrears permissible?) and on its contingent and direct liabilities. And
what is the government's risk management capability?
44 BRIXI AND MODY
Table 1.6 Reducing Government Risk Exposure
Source of risk Reduce risk in design Reduce exposure for risks taken
Guarantees Cover only selected Adjust risk exposures in direct
risks such as political/ liability and assets portfolio (for
policy risks. example, reduce exposure to the
pertinent currency if exchange rate
risk is covered by the guarantee).
Establish a reserve fund for all
guarantees.
Limit total benefits paid to the
amount available in the reserve
fund.
Ensure reserve adequacy by trans-
forming the reserve fund into a
public company with shares
freely traded.,
Disaster Cap maximum benefit. Issue catastrophe bonds (possibly
insurance Insure middle rather than for a basket of likely disasters).
first portion of loss. Purchase reinsurance for risks in
excess of a threshold that is
deemed fiscally bearable.
Deposit Require information dis- Establish separate reserve fund.
insurance closure, sound regula- Limit total benefits paid to the
tion, supervision, and amount available in the reserve
enforcement before in- fund.
troducing deposit Ensure reserve adequacy by
insurance. transforming the reserve fund
Cap maximum benefit. into a public company with
Insure middle rather than shares freely traded.'
first portion of loss.
Price Auction policies. Purchase payoff-replicating
support Cover only selected risks derivatives.
such as political/policy Purchase reinsurance.
risks.
Implicit Make announcements Seek contingent credit line from
guarantees and act to minimize IMF.
to banks and moral hazard.
enterprises
Privatiza- Find a strategic investor Use proceeds to reduce govern-
tion and (future revenue). ment liabilities or future
asset sale Maximize privatization obligations.
revenue.
(Table continues on the following page.)
DEALING WITH GOVERNMENT FISCAL RISK 45
Table 1.6 (continued)
Scurce of risk Reduce risk in design Reduce exposure for risks taken
Purchase Pay only estimated market Sell bad assets quickly.
o0: bad price (a part of recap-
assets italization objectives).
Allow for market mech-
anisms to deal with bad
assets (strengthening the
position of creditor, bank-
ruptcy processes, and so
forth).
Make announcements
and act to minimize
expectations of future
possible repeated
purchases.
Commod- Demand base payment Issue commodity-linked
ity tax independent of com- bonds.
modity price. Purchase commodity-linked
derivatives.
Purchase insurance.
Repay- Require collateral. Purchase default insurance.
inent of
direct
lending
a. The interests of the fund shareholders would contribute to ensuring reserve adequacy-
that is, to charging guarantee beneficiaries, such as banks, adequate premiums. This ar-
rangement loosely imitates the arrangement suggested in Chapter 6 by Daniel Cohen in
this volume.
Source: The authors.
The Fiscal Risk Questionnaire: An Example
For many governments, the objective to understand and manage their
exposure to fiscal risk is likely to require major efforts throughout an
extended period of time. The fiscal risk questionnaire that follows
provides a set of questions that analysts and policymakers may find
useful in dealing with government fiscal risk.
0 Fm 0m
46 BRIXI AND MODY
Fiscal Risk Questionnaire
A. Macroeconomic Context
1. What are the macroeconomic constraints on the government's
future fiscal performance?
Identify constraints such as a currency board or other inflexible
exchange rate arrangement, high public debt levels, low sover-
eign credit rating, or shallow domestic debt markets that risk the
government's future ability to issue debt.
2. What trends are affecting the government's general fiscal position?
Analyze medium-term and long-term trends to provide a con-
solidated picture inclusive of all levels of government, off-budget
funds, public assets, and liabilities.
3. How sensitive are these fiscal trends to the underlying macroeco-
nomic, demographic, and policy assumptions?
Build stress scenarios for the medium-term and long-term fiscal
outlooks with respect to the underlying assumptions.
B. Sources of Fiscal Risk
1. What are the major sources of fiscal risk? What are the largest
and riskiest contingent government liabilities in the country? Are
they explicit or implicit? Fill in the Fiscal Risk Matrix (Table
1.1) with specific items.
2. What are the main types of risks determining the size and ur-
gency of the items in the Fiscal Risk Matrix?
Consider currency risk, interest rate risk, commodity price risk,
refinancing risk, operational risk, political risk, policy risk, and
similar risks.
3. Overall, are the items identified in the Fiscal Risk Matrix likely
to raise significantly the future financing requirement of the gov-
ernment? If so, under which circumstances mainly?
Consider the overall size of contingent liabilities (face values)
and a stress scenario if these obligations are realized. Compare
the overall face value of contingent liabilities with the reported
government debt and with the government's future borrowing
capacity.
C. Analysis of Selected Risks
1. How clearly defined are the public sector and the spheres of gov-
ernment responsibility? Is there a precise legal delineation of the
DE_ALING WITH GOVERNMENT FISCAL RISK 47
public sector (for example, in the form of a full list of public
sector agencies) and of government responsibilities? If not le-
gally defined, is it clearly understood which services are guaran-
teed by the government?
2. State-guaranteed institutions and directed credit
Identify institutions that fulfill orders of the government to ex-
tend financing to enterprises, banks, agencies of any kind, or
households. Review their balance sheets and statements of con-
tingent liabilities.
What type of government support do these institutions receive
(for example, privatization revenues, cheap financing via the cen-
tral bank, state guarantee for borrowings)?
3. Guarantees
Review government guarantees, their issuer (the ministry of fi-
nance or other government agency), beneficiaries, creditors, face
values, the type of risks and their shares covered, currency of
denomination, and risk estimates if any.
4. State-owned enterprises and banks
Review the largest state-owned enterprises and their balance sheets
and risk statements and the largest state-owned banks and their
balance sheets and risk statements.
D. Sources of Government Financial Safety
1. What are the major sources of government financial safety? What
are the largest tradable government assets and other sources of
future revenues? Fill in the Fiscal Hedge Matrix (Table 1.2) with
specific items.
2. How sensitive are these financial sources to the respective types
of risks? What are the likely scenarios for the future government
revenue stream?
3. Taking into account the correlation in the sensitivity of govern-
ment finances to the different types of risks, compare the sce-
narios of the likely future government revenue stream with the
likely financing pressures to emerge from the items identified in
the Fiscal Risk Matrix. What is the worst-case scenario?
E. Recording and Reporting: Transparency
1. For each item in the matrixes, which agencies are responsible for
final approval, recording, monitoring, and data consolidation?
48 BRIXI AND MODY
2. Which agencies and authorities are informed ex ante about con-
tingent liabilities and, overall, about the fiscal risks associated
with programs under government consideration?
* The issuing agency only?
* The related sector ministry?
. The finance ministry?
* The cabinet?
. The parliament?
* Others?
3. Which agencies can instantaneously retrieve up-to-date informa-
tion about the items listed in the matrixes? Which documents
report such information? What is the time lag in reporting?
4. Which sources of fiscal risks are not reported to the:
. Ministry of finance
* Cabinet
* Central bank
. Parliament
* Foreign investors
* Public?
F. Institutional Arrangements: Accountability
1. Do any legal requirements apply to the government with respect
to estimating, accounting, and reporting the future fiscal costs
associated with its direct and contingent liabilities?
No
Yes, in the budget process
* when the government is called on to pay
* when cash is transferred
* other.
2. Which of the obligations identified in the Fiscal Risk Matrix are
not regulated by any law and depend fully on ad hoc government
decisions?
3. What is the legal and regulatory framework for:
* State guarantees: the requirements for their design (the type of
risks that can be covered, the extent of the required risk-
sharing), issuance (is only the ministry of finance authorized?),
government control mechanism (required reports from the
creditor and beneficiary, audit and valuation requirements),
and the realization mechanism if they fall due.
DEALING WITH GOVERNMENT FISCAL RISK 49
* Subnational governments, public sector agencies and enter-
prises, and state-guaranteed institutions: the financial man-
agement and reporting requirements and government control
mechanism.
* Demands on the government to extend ad hoc previously un-
foreseen financial support: the legal requirements and prac-
tice for deliberation in government decisionmaking.
4. Is the government legally required to explain increases in public
liabilities (particularly increases above the levels explained by
budget deficit figures)?
No
Yes,
* to the parliament
* to the public
* other.
5. What is the authority and capacity of the supreme audit institu-
tion with respect to the government's risk exposure and risk
management?
Is the supreme audit institution authorized to review and ca-
pable of reviewing the quality and assumptions of the
government's fiscal risk analysis, fiscal risk management strat-
egy, and fiscal risk management practice?
G. Policy: Practice
1. When the government considers new promises of contingent gov-
ernment support, how much attention does it pay to risk analysis,
to the issues of moral hazard in the markets, and similar things? In
particular, do the ministry of finance, cabinet, central bank, or
parliament quantify the future fiscal cost of alternative options in
a single medium-term fiscal framework? Describe the risks of al-
ternative options (direct versus contingent support program, pos-
sibilities for risk-sharing in guarantee contracts, and others).
2. In which areas and under what circumstances do the public or
interest groups expect the government to provide financial sup-
port beyond the budget?
3. Identify examples of times at which the government withstood
political pressure and did not provide financial support beyond
the budgeted figures (for example, when the government refused
to solicit financial support for a failed enterprise or bank).
50 BRIXI AND MODY
4. Are public enterprises and banks, state-guaranteed institutions,
and creditors and beneficiaries under state guarantees "rewarded"
and "punished" for the quality of their management of risk?
Provide examples.
H. Fiscal Risk Management: Capacities
1. Describe the capacities of the ministry of finance, other govern-
ment agencies, public sector institutions and enterprises, and state-
guaranteed institutions to analyze, monitor, and control the risks
of government programs and contingent liabilities.
What methodologies have been used to analyze the size and risks
of specific contingent liabilities? What have been the challenges
and results in applying these nmethodologies? What has been the
process of monitoring and controlling risk?
2. Describe the process of designing a state guarantee, a state insur-
ance program, or any other program that involves fiscal risk.
Particularly focus on the risk analysis and treatment of moral haz-
ard in the design process. For example, to what extent are risks
shared between the government and the program beneficiaries?
3. What measures have been implemented by the parliament, cabi-
net, ministry of finance, sector ministries, and other public agen-
cies to prevent and reduce fiscal risks arising from the public and
private sectors?
For example, are there any limits on enterprise debt, subnational
government obligations, or central bank obligations? Are any
actions taken if they appear too high? Does the government have
an explicit risk management strategy with respect to its overall
fiscal risk exposure?
4. Does the government build contingency reserves, purchase rein-
surance, or hedge to mitigate its fiscal risk exposure?
Who determines the required size of the contingency reserves
and how? What is the hedging practice?
Conclusion
Dealing with contingent liabilities and other fiscal risks has recently
surfaced as an increasingly important issue in public finance analysis
and public finance management. Learning from country experience
and from new research (both discussed in the chapters of this book),
this chapter has outlined a framework to guide analysts and
policymakers in their attempts to understand and improve the man-
DEALING WITH GOVERNMENT FISCAL RISK 51
agement of fiscal risk. Further research and long-term efforts by gov-
ernments will be needed, however, for public finance analysis and public
finance institutions to truly come to terms with government fiscal risk,
and for policymakers and civil servants to acquire the incentives and
capacity to optimize government exposure to risks.
Notes
1. The literature discussing the ultimate fiscal and economic costs of hid-
(fen government obligations is large. Kharas and Mishra (2001) illustrate the
fiscal cost of contingent liabilities across over 30 countries. Works by Brock
(1992), Kryzanowski and Roberts (1993), Caprio and Klingebiel (1997), and
World Bank (2001a) discuss government bailouts in the banking sector. Free-
man and Mendelowitz (1982) illustrate a bailout in the private corporate
sector. U.S. GAO (1998) draws lessons from the past fiscal cost of credit
guarantees and government insurance programs. Townsend (1977), Salant
(1983), and Bardsley (1994) analyze adverse behavioral effects and the ex
post fiscal cost of price-fixing schemes. Dillinger (1999) and Fornasari, Webb,
and Zou (2000) analyze bailouts of subnational governments by the central
government.
2. The conventional approach to the analysis of fiscal sustainability is
limited in two ways: first, it looks only at the liability side of the public
sector balance sheet, and, second, it considers only direct liabilities, ignoring
contingent liabilities, both explicit and implicit. Under the conventional ap-
proach, the actual deficit is compared with the estimated sustainable deficit
level that will keep the debt-to-GDP (gross domestic product ) ratio constant
for feasible rates of growth, real interest, and inflation. This approach as-
sumes that keeping a constant ratio of public debt to GDP will ensure public
sector solvency and avoid debt crises in the future. Another, less-stringent
requirement is to test for the no-Ponzi-scheme condition for public debt,
followed up by the neoclassical solvency approach. This methodology checks
for public solvency by comparing the annual rate of growth of the govern-
ment debt-to-GDP ratio with the real interest rate. If the debt ratio system-
atically grows faster than the real interest rate, the public sector is considered
insolvent. Conventional fiscal analysis also tends to neglect the sensitivity of
the fiscal position to risks, such as macroeconomic volatility, called contin-
gent liabilities, and unclear expenditure commitments, and the ability of the
government to cope with shocks.
For a critique of standard approaches to fiscal sustainability analysis and
deficit measurement, see Eisner and Pieper (1984), Buiter (1985), Bean and
Buiter (1987), Fischer and Easterly (1990), Blejer and Cheasty (1991), East-
erly, Rodriguez, and Schmidt-Hebbel (1994), Selowsky (1998), Easterly (1999),
and Kharas and Mishra (2001). These studies reflect the fact that, depending
on the definition and measurement methodology, deficit measures may mean
52 BRIXI AND MODY
very different things. They explore the impact of the coverage of a govern-
ment budget, the definition of government, and the accounting and budget-
ing frameworks, as well as the impact of inflation, seasonality, structural
changes in the economy, and the business cycle on the results of fiscal analy-
sis. Doubts have been expressed about the feasibility of incorporating hidden
government obligations in fiscal analysis. For example, Eisner (1984) notes
that the valuation of contingent liabilities of government in the calculation
of deficits is subject to the criticism that the government can legislate actions
that may seriously change their future value.
For a critique of how public finance management frameworks treat con-
tingent liabilities and fiscal risk, see, for example, World Bank (1997), Schick
(1998, 2001), and Schiavo-Campo and Tommasi (1999).
3. This section draws on Polackova (1998).
4. In Chapter 7 of this book, Richard Hemming and Murray Petrie try to
differentiate quasi-fiscal activities from contingent liabilities, using a nar-
rower definition for contingent liabilities.
5. Dillinger (1999) provides examples of the subnational government risk
of Argentina and Brazil. Hutchinson (2000) and World Bank (2001b) illus-
trate the fiscal risks of state-owned enterprises. In the United States, an inter-
esting body of literature has been developed around the so-called government-
sponsored enterprises (GSEs), which provide guarantees in housing and
agriculture markets. The best-known GSEs are Fannie Mae and Freddie Mac.
Creditors have perceived GSEs as beneficiaries of an implicit guarantee of
the U.S. government and thus have been willing to offer cheap financing. For
an overview, see Stanton and Moe (2001) and Van Order (2000).
6. The Medium-Term Expenditure Framework (MTEF), as first applied in
Australia, tackles fiscal opportunism with respect to government expendi-
tures and revenues by requiring policymakers to analyze and disclose the
assumptions and medium-term (three to five years) implications of their budget
proposals and to be accountable for any departures from their targeted levels
in year-by-year decisionmaking. In South Africa, MTEF extends further to
cover consequences of off-budget items such as government guarantees. For
a discussion, see World Bank (1997), Schiavo-Campo and Tommasi (1999),
Kruger (1999), and Schick (2001).
7. Stress scenarios illustrate the sensitivity of expected fiscal outcomes to
normal and abnormal changes in the underlying assumptions. For a specific
program, stress testing will show how normal and extreme changes in the
underlying factors (such as commodity price) over the next 3, 6, and 12
months will affect its fiscal cost. More broadly, applied to the government's
fiscal position stress testing helps detect key fiscal vulnerabilities of govern-
ment. A convenient way to generate less likely scenarios and stress scenarios
is the value-at-risk (cost-at-risk) approach. For an introduction to the value-
at-risk literature, see, for example, Best (1998) and Butler (1999); for an
application to contingent liabilities, see Blejer and Schumacher (1998).
DEALING WITH GOVERNMENT FISCAL RISK 53
A baseline scenario would reflect the expected, most likely outcomes and
utilize actuarial, econometric, or contingent claim analysis methods (see Lewis
and Mody 1997 and Mody and Patro 1996 for a summary and Hull 1997 for
a textbook introduction).
8. Many governments pay the cost of defaulted contingent liabilities di-
rectly from borrowing proceeds. In such cases, contingent liabilities never
affect the budget and budget deficit. After they fall due, they are only re-
flected in an increase in government debt.
9. The use of present value budgeting may or may not affect cash-based
estimates of the government's fiscal deficit. It depends on whether the effect
on the deficit is recorded when money is transferred from the budget to a
contingency fund (then no effect is recorded when a guarantee is called and
paid for from the contingency fund) or only when actual cash payments are
disbursed from the program account. For more discussion on accounting and
budgeting for risk, see also Lewis and Mody (1997), Brixi, Ghanem, and
Islam (1999), Brixi, Papp, and Schick (1999), and World Bank (1999b).
10. A full accrual-based accounting system, though desirable, is neither
necessary nor sufficient to ensure that governments adequately report and
consider contingent fiscal risks in policy. Although an accrual accounting
system encourages governments to prepare a statement of contingent liabili-
ties and financial risks, it generally does not require that contingent liabili-
ties be included in the balance sheet and that the associated risks be evaluated
and quantified. International accounting standards, for example, require only
probable contingent liabilities (contingencies with a relatively high probabil-
ity of realization) to be included in the balance sheet, leaving the others in a
separate statement of contingent liabilities without requiring analysis of the
underlying risks. In the case of budgeting, budgeting for the present value of
the future fiscal cost of contingent liabilities may be combined with cash
budgeting for budgetary spending (cash-based) programs. Countries success-
fully combining reporting on contingent liabilities with cash accounting in-
clude the Czech Republic, Hungary, and South Africa, and those budgeting
for contingent liabilities within a cash-based budgeting system include Canada,
the Netherlands, and the United States.
11. Derivatives can be designed in many different ways to fit the risk
management objectives. Basic derivatives include futures and forwards, swaps,
and options. A future contract is an arrangement between two parties to buy
or sell an asset at a predetermined future time and price. These contracts are
normally traded on an exchange. A similar arrangement, a forward contract,
can be made with a financial intermediary over-the-counter, without involv-
ing an exchange. A swap is an arrangement in which two parties exchange
the stream of payment of two different assets. For background on futures,
forwards, and swaps, see Hull (1997). An option contract gives the holder the
right but not the obligation to buy (call option) or to sell (put option) an asset
at a predetermined future time and price. This predetermined price is known
54 BRIXI AND MODY
as the strike price and the predetermined date is known as the expiration or
maturity date. At the time of purchase, the buyer of an option contract pays an
option price to the option writer. Option-pricing analysis, most often employ-
ing the Black-Scholes formula, serves to set the option price. For background
on options, see Cox and Rubinstein (1985). On the actual and possible use of
derivatives as hedging instruments by governments, see, for example, Claessens
(1993), Nars (1997), World Bank (1999a, 2000), and Patterson (2001).
12. Mody (2000), for example, describes the experience with rainfall in-
surance purchased by the government of Nicaragua from a private reinsurer.
13. Hedging refers to the strategy of protecting oneself against losses aris-
ing from changes in market conditions. It involves relatively safe perfectly
offsetting transactions (that would "perfectly" offset gaines/losses arising
from changes in market conditions) and relatively risky dynamic hedging
strategies (that require continual rebalancing as market conditions evolve).
For an introduction, see Hull (1997).
14. Pooling less than perfectly correlated risks (for example, hurricane,
drought, and fire) would allow the government to reduce the volatility in the
total long-term cost of its insurance programs.
15. Catastrophe bonds are obligations whose interest and principal pay-
ments are linked to a catastrophic event. For example, they could call for a
reduction in the interest or principal, or for extension of maturity if losses
related to the underlying event exceed the trigger level. This arrangement is
called reinsurance protection (see Insurance Services Office 1999).
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Arthur Andersen. 2000. General and Specific Methodologies for Valuing
Contingent Liabilities. Washington, D.C.
Bardsley, Peter. 1994. "The Collapse of the Australian Wool Reserve Price
Schemes." Economic Journal 104 (426): 1087-1105.
Bean, Charles, and William Buiter. 1987. "The Plain Man's Guide to Fiscal
and Financial Policy." Employment Institute, London, October.
Best, Philip. 1998. Implementing Value at Risk. Chichester; New York: John
Wiley.
Blejer, Mario, and Adrienne Cheasty. 1991. "The Measurement of Fiscal
Deficits: Analytical and Methodological Issues." Journal of Economic
Literature 29: 1644-78.
Blejer, Mario, and Liliana Schumacher. 1998. "Central Bank Vulnerability
and the Credibility of Commitments: A Value-at-Risk Approach to Cur-
rency Crises." IMF Working Paper WP/98/65. International Monetary
Fund, Washington, D.C.
EEALING WITH GOVERNMENT FISCAL RISK 55
Bodie, Zvi, and Philip Davis. 2000. The Foundations of Pension Finance. 2
vols. Cheltenham; Northampton: Elgar Reference Collection.
Brixi, Hana Polackova, Hafez Ghanem, and Roumeen Islam. 1999. "Fiscal
Adjustment and Contingent Government Liabilities." Policy Research
Working Paper 2177. World Bank, Washington, D.C.
Brixi, Hana Polackova, Anita Papp, and Allen Schick. 1999. "Fiscal Risks
and the Quality of Fiscal Adjustment in Hungary." Policy Research Working
Paper 2176. World Bank, Washington, D.C.
Brock, Philip. 1992. "External Shocks and Financial Collapse: Foreign Loan
Guarantees and Intertemporal Substitution of Investment in Texas and
Chile." American Economic Review 82 (2): 168-73.
Buiter, Willem. 1985. "A Guide to Public Sector Debt and Deficits." Eco-
nomic Policy (November): 14-79.
Butler, Cormac. 1999. Mastering Value at Risk: A Step-By-Step Guide to
Understanding and Applying Var. London; New York: Financial Times/
Prentice Hall.
Calderon Zuleta, Alberto. 1999. "Valuing and Managing Risk Associated with
Government Contingent Liabilities." In Proceedings of the Second Sovereign
Debt Management Forum. Washington, D.C., November 1-3. World Bank.
Caprio Jr., Gerard, and Daniela Klingebiel. 1997. "Bank Insolvency: Bad
Luck, Bad Policy, or Bad Banking?" In Michael Bruno and Boris Pleskovic,
eds., Annual World Bank Conference on Development Economics 1996.
Washington, D.C.: World Bank.
Cassard, Marcel, and David Folkerts-Landau. 1997. "Risk Management of
Sovereign Assets and Liabilities." IMF Working Paper WP/97/166. Inter-
national Monetary Fund, Washington, D.C.
Chase Manhattan Bank. 1996. Risk Management Handbook, Private Sector
Participation in Urban Infrastructure. Washington, D.C.: U.S. Agency for
International Development.
Claessens, Stijn. 1993. "Risk Management in Developing Countries." Tech-
nical Paper 235. World Bank, Washington, D.C.
Commonwealth Secretariat. 2001. External Sector Contingent Liabilities-
A Study on India. London, January.
Congressional Research Service. 1998. Tax Expenditures. Compendium of
Background Material on Individual Provisions. Committee on the Bud-
get, U.S. Senate, Washington, D.C.
Cox, John, and Mark Rubinstein. 1985. Options Markets. Upper Saddle
River, N.J.: Prentice Hall.
Davis, Jeffrey, Rolando Ossowski, James Daniel, and Steven Barnett.
2001. "Stabilization and Savings Funds for Nonrenewable Resources.
Experience and Fiscal Policy Implications." Occasional Paper 205. Inter-
national Monetary Fund, Washington, D.C.
Dillinger, William. 1999. "Fiscal Management in Federal Democracies: Ar-
gentina and Brazil." Policy Research Working Paper 2121. World Bank,
Washington, D.C.
56 BRIXI AND MODY
Dooley, Michael. 2000. "Debt Management and Crisis in Developing Coun-
tries." Journal of Development Economics 63 (October): 45-58.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April).
Easterly, William, Carlos Alfredo Rodriguez, and Klaus Schmidt-Hebbel. 1994.
Public Sector Deficits and Macroeconomic Performance. Washington,
D.C.: World Bank.
Eisner, Robert. 1984. "Which Budget Deficit? Some Issues of Measurement
and Their Implications." American Economic Review 74 (2):138-43.
Eisner, Robert, and Paul Pieper. 1984. "A New View of the Federal Debt and
Budget Deficits." American Economic Review 74 (2): 11-29.
Fischer, Stanley, and William Easterly. 1990. "The Economics of the Govern-
ment Budget Constraint." World Bank Research Observer 5 (2).
Fornasari, Francesca, Steven Webb, and Hcng-fu Zou. 2000. "The Macro-
economic lImpact of Decentralized Spending and Deficits: International
Evidence." Annals of Economics and Finance 1 (November): 403-33.
Freeman, Brian, and Allan Mendelowitz. 1982. "Program in Search of a
Policy: The Chrysler Loan Guarantee." Journal of Policy Analysis and
Management I (summer): 443-53.
Froot, K. A., D. S. Scharfstein, and J. C. Stein. 1994. "A Framework for Risk
Management." Harvard Business Review (November-December).
Garber, Peter. 1998. "Derivatives in International Capital Flows." Working Pa-
per 6623. National Bureau of Economic Research, Cambridge, Mass., June.
Honohan, Patrick. 1999. "Fiscal Contingency Planning for Banking Crises."
Policy Research Working Paper 2228. World Bank, Washington, D.C.
Hueth, Darrell L., and William H. Furtan. 1994. Economics of Agricultural
Crop Insurance: Theory and Evidence. Norwell, Mass.: Kluwer Academic
Publishers.
Hull, John. 1997. Options, Futures, and Other Derivatives. Upper Saddle
River, N.J.: Prentice Hall.
Hutchinson, Gladstone. 2000. "The Challenge of Fiscal Sustainability: Pub-
lic Enterprise Financial Relationship with Government in Jamaica."
Lafayette College. Processed.
IMF (International Monetary Fund). 1996. "Aging Populations and the Fis-
cal Consequences of Public Pension Schemes." Washington, D.C.
Insurance Services Office. 1999. "Insuring Catastrophic Risk." Processed.
Irwin, Timothy, Michael Klein, Guillermo Perry, and Mateen Thobani. 1998.
Puiblic Risk in Private Infrastructure. Washington, D.C.: World Bank.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, E H. Rogers, and L. Squire, eds.
World Bank Economists' Forum. Washington, D.C.: World Bank.
Kruger, Coen. 1999. "Valuing and Managing Risk Associated with Govern-
ment Contingent Liabilities." Presentation at the Second Sovereign Debt
Management Forum, Washington, D.C., November 1-3. World Bank.
DEALING WITH GOVERNMENT FISCAL RISK 57
Kryzanowski, L., and G. S. Roberts. 1993. "Canadian Banking Solvency."
Journal of Money, Credit and Banking 25 (3): 361-76.
Larson, Donald F., Panos Varangis, and Nanae Yabuki. 1998. "Commodity
Risk Management and Development." Policy Research Working Paper
1963. World Bank, Washington, D.C.
Leaven, Luc. 2000. "Banking Risks around the World: The Implicit Safety
Net Subsidy Approach." World Bank. Processed.
Lewis, Christopher, and Ashoka Mody. 1997. "The Management of Con-
tingent Liabilities: A Risk Management Framework for National Gov-
ernments." In Timothy Irwin, Michael Klein, Guillermo Perry, and
Mateen Thobani, eds., Dealing with Public Risk in Private Infrastruc-
ture. World Bank Latin American and Caribbean Studies. Washington,
D.C.: World Bank.
M\4arrison, Christopher. 2001. "Risk Measurement for Project Finance Guar-
antees." Capital Markets Company. Processed.
Mody, Ashoka. 2000. "Nicaragua: Rainfall Risk Management Project." World
Bank. Processed.
Mody, Ashoka, and Dilip Patro. 1996. "Valuing and Accounting for Loan
Guarantees." World Bank Research Observer 11 (February): 119-42.
Nars, Kari. 1997. Excellence in Debt Management. London: Euromoney.
OECD (Organisation for Economic Co-operation and Development). 1999.
How Should Governments Invest Financial Assets and Manage Debt.
PUMA/SBO/RD (99). Paris: OECD.
. 2000. Reforms for an Aging Society. Paris: OECD.
Patterson, Linda. 2001. "The State of Texas Hedging Program: A Long Posi-
tion and a Long Perspective." Presentation delivered at World Bank, June
25. Patterson and Associates. Processed.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Salant, S. W. 1983. "The Vulnerability of Price Stabilization Schemes to Specu-
lative Attack." Journal of Political Economy 91: 1-38.
Schiavo-Campo, Salvatore, and Daniel Tommasi. 1999. Managing Govern-
ment Expenditure. Manila: Asian Development Bank.
Schick, Allen. 1998. A Contemporary Approach to Public Expenditure Man-
agement. Washington, D.C.: World Bank.
. 2001. "Does Budgeting Have a Future?" Presentation at OECD Se-
nior Budget Officials' Meeting, Paris, May. Organisation for Economic
Co-operation and Development.
Selowsky, Marcelo. 1998. "Fiscal Deficits and the Quality of Fiscal Adjust-
ment." In The Challenges for Public Liability Management in Central
Europe. Washington, D.C.: World Bank.
Standard and Poor's. 1997. "Criteria: Financial System Stress and Sovereign
Credit Risk." Sovereign Ratings Service (December).
58 BRIXI AND MODY
Stanton, Thomas, and Ronald Moe. 2001. "Government Corporations and
Government Sponsored Enterprises." In Lester M. Salamon, ed., The Tools
of Government: A Public Management Handbook for the Era of Third-
Party Government. New York: Oxford University Press.
Townsend, R. M. 1977. "The Eventual Failure of Price Fixing Schemes."
Journal of Economic Theory 14: 190-99.
U.S. GAO (General Accounting Office). 1993. Bank Insurance Fund: Review
of Loss Estimation Methodologies. Washington, D.C.
. 1998. Budgeting for Federal Insurance Programs. Washington, D.C.
Van Order, Robert. 2000. "The Structure and Evolution of American Sec-
ondary Mortgage Markets, with Some Implications for Developing Mar-
kets." Freddie Mac, Washington, D.C. Processed.
World Bank. 1994. Averting the Old Age Crisis. Washington, D.C.
1997. Public Expenditure Management Handbook. Washington, D.C.
1999a. "Dealing with Commodity Price Volatility in Developing
Countries: A Proposal for a Market-based Approach?" Discussion Paper
for the Roundtable on Commodity Risk Management in Developing Coun-
tries, Washington, D.C., September 24.
1999b. Malaysia: Public Expenditure Review. Washington, D.C.
1999c. Toolkils for Private Participation in Water and Sanitation.
Washington, D.C.
- . 2000. Proceedings of the Second Sovereign Debt Management Fo-
rum. Washington, D.C., November 1-3.
- 200 la. Finance for Growth: Policy Choices in a Volatile World. Policy
Research Report, May.
. 2001b. "Parastatal Performance and Strategy in Bangladesh."
Processed.
World Bank and IMF (International Monetary Fund). 2000. "Guidelines for
Public Debt Management" .
Yaron, Jacob. 1992. "Assessing Development Finance Institutions-A Public
Interest Analysis." World Bank Discussion Paper 174. World Bank, Wash-
ington, D.C.
CHAPTER Z
Accounting and Financial
Accountability to Capture Risk
Murray Petrie
The Economics and Strategy Group
THE TRADITIONAL CASH-BASED accounting and reporting systems used
by many governments provide inadequate information within the ex-
ecutive for the management of fiscal risks.' They also do not produce
enough information for the legislature and the public to hold govern-
ments accountable for the management of fiscal risks. The traditional
focus on cash has been consistently associated with fiscal management
practices that are short term and reactive. Poor information has inter-
acted with poor incentives to discourage decisionmakers from taking
a longer-term view of fiscal risks and their management.
In recognition of these shortcomings, a range of initiatives has been
introduced in different countries and by international agencies that,
while not motivated solely by a desire for better risk management, has
important implications for the management of fiscal risk. The initia-
tives taken include: greater overall transparency in fiscal management,
supplementary reporting of noncash information in budget documents
and the final accounts, changes to the basis of accounting, and moves
by governments to tighten control over fiscal risks.
This chapter discusses these initiatives. It does so within a frame-
work that focuses on risk to the aggregate fiscal position-that is, on
sources of variability in the overall level of government spending, rev-
enues, the fiscal balance, and value of assets and liabilities. While these
sources of variability manifest themselves through variability in indi-
vidual spending, revenue, and financing programs, it is the impact on
the aggregate fiscal position that is the chief object of concern.
Risk is defined here as any situation in which there is a range of
possible outcomes around the expected fiscal position. Inherent uncer-
tainty about which of a number of different possible states of nature
59
60 MURRAY PETRIE
will apply in the future creates a range of possible fiscal outcomes.
This approach captures forecasting risk, in addition to uncertain obli-
gations such as guarantees and other contingent liabilities.2 The focus
in this chapter is on the annual budget and the short- to medium-term
fiscal position.
In general, the objective of financial risk management for any entity
is to improve the entity's financial position and performance, while
protecting the entity from unacceptable variance in returns-and in
particular from the risk of unacceptable losses. It is clear that the speci-
fication of what is an unacceptable level of fiscal risk will vary from
country to country. It will depend on the initial fiscal position, the
degree of fiscal flexibility, the nature and extent of fiscal risks, the
quality of the information available, the capacity for risk manage-
ment, and on the perceived "return" to risk-taking. Assessing what a
government's appetite for risk should be is an extremely complex is-
sue, which at present is at the boundary of public finance theory, let
alone practice (for a discussion of these issues, see OECD 1999 and
Polackova 1998). For most governments, determining optimal risk
exposure will not be of practical relevance until major progress has
been made in identifying, analyzing, quantifying, reporting, and man-
aging existing fiscal risks. This is the subject of the rest of this chapter.
The chapter is structured as follows. The next section describes the
weaknesses in traditional cash-based budget management systems with
respect to the management of fiscal risks. That section is followed by
one that outlines a framework for fiscal risk management, comprising
both macro-level and micro-level elements, and provides country ex-
amples. The chapter ends with some concluding remarks.
Weaknesses in the Treatment of Risk under
Traditional Cash-based Accounting Systems
Governments have historically reported their financial performance
using cash-basis accounting.3 The reasons for this include simplicity,
compliance with legal requirements, and the government's borrowing
requirement and the macroeconomic impact of the budget.
Cash-basis accounting has, however, a number of well-recognized
weaknesses. Some relate specifically to inadequacies in the treatment
of fiscal risks. For example:
o The cash basis results in the incomplete or inaccurate measure-
ment of current transactions. In many instances, there is a discrepancy
between when a commitment is entered into, when resources are used
and the economic effects are felt, and when a cash payment is made.
For example, contingent liabilities are not reflected in pure cash-basis
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 61
financial statements until and if they have to be paid.4 The real effects
of government lending and insurance programs and of civil service
pension schemes are distorted by cash-basis reporting.5
* There is a lack of information on stocks of assets and liabilities
and the relationship between flow and stock variables. Although most
governments provide at least some information on public debt, depre-
ciation is not reported, and risks relating to an impending need for
capital replacement may remain hidden.
In addition, there has generally been no reporting of the sensitivity of
the fiscal position to changes in key forecasting assumptions.
More seriously, the deferred recognition of expenditures can hide
an accumulation of problems until they reach massive proportions.
The classic example of this is regulatory forbearance in banking super-
vision. Regulators and politicians may be tempted to defer dealing
with insolvent banks in the hope the banks will recover, or, if not, that
the cost will fall on a future administration.6 With the expectation of
future government support, the managers of distressed banks may take
on even higher levels of risk in the knowledge they are in effect gam-
bling with public money. The cost of financial sector reconstruction in
the last two decades has been massive in many countries and consti-
tutes a major source of fiscal risk.
In general, therefore, pure cash-basis accounting and reporting re-
sult in inadequate information for use by the executive in managing
fiscal risks and for use by the legislature and the public in holding the
government accountable for such management. Moreover, it often
leaves financial markets guessing about the true state of public finances
and the likely (as opposed to the budgeted) fiscal deficit. In recogni-
tion of these shortcomings, some countries have introduced in recent
years a range of initiatives in fiscal risk management. These include
overall transparency in the public accounts7; supplementary reporting
of information on fiscal risks in budget documents, in the final ac-
counts, or in both; a change to the basis of accounting; and a central-
ized approach to implementing fiscal risk management within the
government. These initiatives are discussed in detail in the next section.
Recent Initiatives to Build Accountability
in Fiscal Risk Management
The approach to fiscal risk management outlined below contains what
might be thought of as both macro- and micro-level elements. The
macro-level element covers broad transparency to the public in the
conduct of fiscal policy. The more micro-level elements comprise specific
reporting of information on fiscal risks, the choice of an appropriate
62 MURRAY PETRIE
basis of accounting, and a centralized approach to the implementation
of fiscal risk management.
Greater Overall Fiscal Transparency
Conducting fiscal policy in a transparent and open manner is a high-
level approach that has the potential to reduce significantly fiscal
risk and improve risk management. Making information available
publicly requires institutional capacity and systems within govern-
ment so that the information is available within government in the
first instance.8 Such "internal transparency" is capable on its own of
producing major gains in the management of fiscal risk through im-
proving the information base within the executive for fiscal
decisionmaking and strengthening accountability within government
for risk management.
In the absence of public transparency, however, there is limited ex-
ternal accountability and less assurance that there will be a sustain-
able improvement in the management of fiscal risk. And in the absence
of external accountability, there may be less incentive for the govern-
ment to implement sound internal risk management systems in the
first place.
Construed broadly, fiscal transparency requires much more than
just the timely publication of the budget and final accounts. It also
covers clarity of roles and responsibilities; commitment to the timely
publication of complete information on the past, present, and pro-
jected fiscal position; open budget preparation, execution, and report-
ing; and independent assurances of the integrity of fiscal information.9
Transparency in all these dimensions is a critically important ele-
ment of a medium-term fiscal risk management strategy. While there
are elements of transparency that relate specifically to the reporting of
fiscal risks (see below), there are also broader elements that can sig-
nificantly reduce fiscal risk. For example, the main source of fiscal risk
in many countries is debt servicing. Transparency in the conduct of
debt management is an important element in effectively managing risks
and ensuring the accountability of those responsible. Transparency in
this respect involves clarity of roles and responsibilities and clear ob-
jectives for debt management; open processes for conducting debt
management operations; and regular and timely reporting of a range
of details of government debt.'0
A further problem in many countries has been the lack of clarity of
roles within the public sector. Directives to state-owned enterprises
(SOEs) to conduct a proliferation of quasi-fiscal activities have often
resulted in these institutions accumulating losses and needing to be
bailed out."1 In formal terms, such potential spending can be seen as a
type of implicit contingent liability (as illustrated by the Fiscal Risk
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 63
Matrix in Chapter 1). A transparent, arms-length governance frame-
work for SOEs, with separate identification (and funding from the
government budget) of noncommercial obligations, can reduce what
is in many countries a major source of fiscal risk. A lack of clarity in
the respective roles of the central government and lower-level govern-
ments and ad hoc mechanisms for determining intergovernmental trans-
fers are also sources of fiscal risk in many countries.
A more general point is that greater transparency in the conduct of
fiscal policy reduces the risk of errors. In a situation in which func-
tions such as macroeconomic forecasting, fiscal forecasting, and
macrofiscal analysis are conducted entirely within the government,
there may be no independent check on the quality and accuracy of
fiscal information and analysis.
Supplementary Reporting of Information on Risks
]:n view of the deficiencies in cash-basis reporting, a number of coun-
tries have in recent years initiated supplementary reporting of infor-
mation on fiscal risks. Some international standard-setting agencies
also have been revising or developing standards for fiscal reporting
that bear on the management of fiscal risks.
For example, the Code of Good Practices on Fiscal Transparency
issued by the International Monetary Fund (IMF) contains a require-
ment that countries publish a fiscal risk statement with the annual
budget. The Fund's "Manual on Fiscal Transparency" indicates that
the statement should contain information on specific fiscal risks-such
as contingent liabilities-and on general forecasting risks.
The International Federation of Accountants (IFAC) is undertaking
a medium-term project to develop a core set of standards for financial
reporting by governments. The standards are for the cash and accrual
bases of accounting. As part of this project, IFAC published the study
Governmental Financial Reporting: Accounting Issues and Practices.
Rather than prescribing particular accounting treatments, it describes
the types of additional disclosures that some governments using cash-
basis reporting make on contingent liabilities. It also notes that report-
ing of information on contingencies is required under accrual accounting
(see IFAC 2000: 47-49, 171).
The IMF is also revising its Government Finance Statistics Manual.'2
Some major changes are proposed, including use of an accrual basis of
recording and compilation of information on contingent liabilities in a
supplement to the balance sheet.
There are different ways in which to categorize fiscal risks as a
useful way of organizing supplementary reporting. One of the possi-
bilities would be according to whether they are forecasting risks (these
are related to the types of risk discussed in Chapter 1) or specific fiscal
64 MURRAY PETRIE
risks (discussed as sources of risk in the Fiscal Risk Matrix of Chapter
1). Forecasting risks are the inherent risks involved in forecasting the
fiscal aggregates. Budget forecasts are normally highly sensitive to the
assumptions made about a small number of key parameters. Govern-
ments also are typically exposed to specific fiscal risks. These include
contingent liabilities such as guarantees, indemnities, uncalled capital,
and legal action against the government.
Forecasting Risks. Forecasting risks should be disclosed in the cen-
tral government's annual budget documents. The realism and reliabil-
ity of the budget are generally highly dependent on the quality of the
underlying macroeconomic forecasts on which the budget forecasts
are based. Typically, there also will be a small number of key forecast-
ing assumptions related to particular revenue sources or expenditure
programs (for example, the price of oil or the exchange rate). Variabil-
ity in the annual cost of debt servicing due to factors such as exchange
rate, interest rate, and maturity structure risk can be a major source of
exposure as well.
The budget documents should therefore provide information on the
key assumptions on which the budget forecasts are based, and they
should illustrate the sensitivity of the budget to variations in these key
assumptions. Periodic assessments also should be published of the reli-
ability of budget macroeconomic and fiscal forecasts compared with
outturn, with an analysis of deviations by major category. For example,
information should be provided on the effects on forecast revenues,
expenditures, and the overall balance of, say, a 1 percent decrease in
growth of the gross domestic product (GDP) from that assumed in the
budget. Half of the member countries of the Organisation for Eco-
nomic Co-operation and Development (OECD) publish a fiscal sensi-
tivity analysis."3 U.S. budget documents, for example, contain detailed
information about and discussion of the economic assumptions under-
lying the budget, including comparisons with the assumptions devel-
oped by the Congressional Budget Office and with the assumptions
contained in the administration's budget for the previous year (see
United States 1999: chap. 1, Economic Assumptions).
In addition to sensitivity analysis, it is desirable to publish some
alternative medium-term scenarios in which different economic growth
and aggregate fiscal developments are combined. These can illustrate
the robustness of the budget in the face of broad alternative develop-
ments. An example of scenario analysis reporting by the New Zealand
government is shown in Table 2.1.
In this context, a government might go the further step and discuss
its fiscal strategy in the event that the economic and fiscal outlook
turns out to be less favorable than that contained in the budget fore-
casts. Discussion of broad fiscal contingency plans could help to re-
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 65
Table 2.1 Fiscal Scenario Reporting: Summary of Alternative
Scenarios, New Zealand
1997/98 1998/99 1999/00 2000/01 2001/02
actual forecast projection projection projection
Production GDP (%)
Central forecast 2.0 (0.3) 2.9 3.5 3.0
Export-led recovery 2.0 (0.3) 3.8 4.4 2.7
Wreak recovery 2.0 (0.4) 2.0 1.8 3.7
Nominal expenditure GDP (%)
Central forecast 3.3 0.4 3.8 5.2 4.6
Export-led recovery 3.3 0.4 4.9 6.5 5.1
Wleak recovery 3.3 0.3 2.7 2.6 5.2
Operating balance (billion $)
Central forecast 2.5 2.2 (0.0) 0.8 1.5
Export-led recovery 2.5 2.2 0.3 1.7 2.8
Weak recovery 2.5 2.1 (0.6) (0.7) (0.3)
Source: Based on Table 3.1 in Government of New Zealand (1999: 59).
duce market uncertainty about the possible path of fiscal policy. Pro-
viding markets with a broad indication of what sort of fiscal adjust-
ments will be made in response to possible adverse developments-for
example, spending cuts or deferrals, tax increases, a bigger deficit, or
some combination of these elements-may reduce the risk of abrupt
market reactions to adverse market developments. Such an indication
also may make it more likely that a government will be ready to take
quick action when and if an adverse event does occur.
Specific Fiscal Risks. For contingent liabilities, supplementary re-
porting should take the form of a statement of the outstanding stock of
contingent liabilities of the central government. In addition to its pub-
lication with the final accounts, a statement of contingent liabilities
should be included with the annual budget documents in order to pro-
vide a complete picture of the fiscal position at the time the budget is
presented. To qualify as a contingent liability, the amount of expendi-
ture at risk should be material,'4 and the likelihood that the item will
result in future expenditure should be more than remote. Under ac-
crual accounting, to qualify as a contingent liability a possible future
expenditure must also be less than likely."
For each such contingent liability, or pooled program of contingent
liabilities, information should be presented on its nature and potential
fiscal significance. For example, for a loan guarantee (or a portfolio of
similar loans) reporting should cover the amount of the loan, to whom
66 MURRAY PETRIE
the loan has been advanced, and the duration of the loan. Whether
there have been any changes in the details of the item since the previ-
ous reporting date also should be noted. Where possible, an estimate
should be provided of the expected cost of each contingent liability-
desirably in the form of a range rather than just a point estimate. Some
contingent liabilities are nonquantifiable, however, while an estimate
of expected cost for others would not be sufficiently reliable. An ex-
ample of contingent liability reporting, taken from the first trial bal-
ance sheet of the Japanese government, is shown in Box 2.1.
Other specific, short-term fiscal risks that should be reported with
the budget include the following:
* Where there is an unusual degree of uncertainty about the likely
cost of a material expenditure item in the budget, this should be dis-
closed. For example, perhaps the government has made provision for
meeting the costs of reconstruction following a major disaster. At the
time of finalizing the budget the cost allowed in the budget may be
very uncertain.
* Where items have not been included in the budget at all because
of the extent of uncertainty about their timing, magnitude, or eventu-
ality, these items should be disclosed. For example, the government
may have announced its intention to increase food subsidies or public
pensions, but the details of the decision may not have been finalized
sufficiently for inclusion in the budget. The government of New Zealand
reports these sorts of specific fiscal risks in its Budget Economic and
Fiscal Update report accompanying the annual budget (Government
of New Zealand 1999).
A Change to the Basis of Accounting
The basis of accounting refers to the set of accounting principles for
recording transactions that determine when the effects of transactions
or events are recognized for financial reporting purposes. The accrual
basis of accounting entails recognition of transactions or events at the
time the transaction or event occurs rather than at the time a cash
payment is made. Accrual accounting also entails the production of a
full balance sheet. These two differences between cash and accrual
accounting have important implications for the management of fiscal
risk. In addition, the disclosure of supplementary information on con-
tingent liabilities and commitments is required under IFAC's Gener-
ally Accepted Accounting Practice (as reflected in international
accounting standards)."6
A balance sheet encapsulates a longer-term perspective on the
government's financial position. In principle, it represents the estimated
present value of future cash flows-provided they meet the accounting
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 67
Box 2.1 Reporting of Contingent Liabilities in Japan
The following table is an extract from the Japanese government balance
sheet (preliminary trial), as of March 31, 1999 (Japan, Study Group on
Explanatory Methods of Fiscal Position 2000).
1. Contingent liabilities.
(1) Liabilities for loan guarantees and loss compensation
contracts (million yen)
Amount in Amount
Items foreign currency in yen
Compensation for Nuclear
Energy Business 798,000
Guarantee for principal and
interest payment for bonds
of Japan Finance Corporation
for Small Business 400,350
Total 48,928,627
(2) Claims for damages in pending cases (miUlion yen)
Cases Amount claimed
Claim for injunction in Amagasaki Air
Pollution Case (Kobe District Court, (wa)
No. 2217, 1988; (wa) No. 1766, 1995) 12,168
Appeal case for restitution of unjust enrich-
ment (Endless Money Chain Case)
(Fukuoka High Court, (administrative; ko)
No. 11, 1996) 10,396
Total 88,510
Note 1: All the amounts claimed are mentioned in this table, whether
or not the Government is expected to win the case.
Note 2: In cases where the claimed amount is more than 1 billion yen,
case names are mentioned.
(3) Other major contingent liabilities
Title Outline
Project for When natural disasters occur and prefectural
providing aid authorities provide aid money to the head
money to assist of each household affected by disasters, the
victims' recovery Government shall be liable to share part of
from disasters, the aid money under the provisions of
Cabinet orders.
68 MURRAY PETRIE
definition and recognition criteria and can be reliably measured. A
balance sheet therefore provides significant additional information
about the future implications of current policies. For example, under
accrual accounting information is provided on the full cost of current
civil service pension policies, the accumulating burden on future bud-
gets, and the variability in the value of the liability from year to year.
Such information can help focus debate on the appropriate design of
pension schemes. The choice between defined-contribution and de-
fined-benefit pension schemes, for example, involves significant differ-
ences in the amount of fiscal risk borne by the government.'7
However, accrual accounting on its own does not provide all the
information that a fiscal economist would wish. This can be a source
of confusion because of the sometimes different use of the word ac-
crual by economists and accountants. To some economists, accrued
expenditures mean any real consumption or outflow of resources. To
an accountant, an accrued expenditure must meet the definition of a
liability and the recognition criteria of a liability-chiefly, that it can
be reliably measured. The effect of the definition and recognition cri-
teria in accrual accounting is that there is a significant difference be-
tween the information that would be contained in a "comprehensive
balance sheet"'8 and that contained in a balance sheet produced in
accordance with current internationally recognized accounting prac-
tices. For example, the expected cost of possible obligations such as
one-off guarantees is not generally recognized as a liability and inte-
grated into the budget by governments that have adopted accrual bud-
geting. A guarantee does not meet the recognition criteria for an
expense, unless it is judged more likely than not that the expense will
in fact occur and the expected cost of the guarantee can be estimated
with sufficient reliability. In general, this means that one-off guaran-
tees will not normally be recognized as expenses in the budget.
The desirable approach to accounting (and budgeting) for specific
fiscal risks requires balancing two important principles in public fi-
nance. First, decisionmaking is best informed, and incentives are best
aligned, if governments recognize the cost of commitments at the time
they are made. Second, budget appropriations should be based on re-
liable information compiled on the basis of widely accepted account-
ing policies and supported by credible institutional arrangements."9
The best approach seems likely to depend on the particular circum-
stances in individual countries, including the significance and nature
of specific fiscal risks, the existing accounting system, and the country's
financial management capacity. Introducing an annual contingency fund
and expanding the reporting of information on fiscal risks-both fis-
cal sensitivity and specific fiscal risks-can be readily done in the con-
text of a cash-basis accounting and budgeting system. Malaysia and
Japan are examples of countries using a modified cash basis of ac-
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 69
counting, but which report supplementary information on guarantees.20
Much of the information on fiscal risks is in any case provided through
such supplementary reporting even in those countries that have adopted
full accrual accounting. The adoption of accrual accounting is a large
undertaking, and it should be guided by broader considerations than
the management of fiscal risk alone.
A Centralized Approach to Implementing
Fiscal Risk Management
The nature of the fiscal risk management function lends itself to a
centralized policy setting and oversight. The links to the budget pro-
cess, and the need (at least in principle) to take a portfolio approach to
the analysis of risk, suggest that a completely decentralized approach
would not be effective. Depending on the overall public sector man-
agement framework, however, some combination of central oversight
and decentralized accountabilities will be appropriate.
A further important source of fiscal risk in most countries, as noted,
is the sensitivity of the fiscal position to changes in the economic envi-
ronment. An early step in a risk management strategy should be re-
porting the fiscal sensitivity analysis. This requires close coordination
between those responsible for fiscal and macroeconomic forecasting,
particularly in generating realistic alternative macroeconomic scenarios
to test the sensitivity of the fiscal baseline.
In generating alternative fiscal scenarios, allowance should be made
in one scenario for a combination of adverse events to stress test the
medium-term fiscal baseline. Such a scenario might include a fall in
economic growth, a drop in government revenues, an increase in rou-
tine government spending, a shortening of the maturity structure of
public debt, the calling of some guarantees, and expenditure demands
from implicit contingent liabilities. The likely correlations between
these adverse developments mean that while such a scenario may not
be likely, it will nevertheless be very much within the range of possi-
bilities (as is all too often demonstrated by actual country experiences).
At the same time, information should be aggregated across the cen-
tral government on the specific fiscal risks to which individual govern-
ment agencies are currently exposed. The information should include,
to the extent possible, an estimate of the likely range of expected cost
to the government for each risk or pooled program of risks. This task
might best be achieved as a component of the fiscal reporting indi-
vidual agencies must provide to the ministry of finance.21 This requires
a clear, common framework defining fiscal risks-for example, in the
government's accounting policies (for contingent liabilities) and in the
budget instructions sent to departments. It requires communication of
a clear expectation that agency heads will be accountable for complying
70 MURRAY PETRIE
and of a means to verify compliance. It also requires careful central
monitoring. For example, the maturity profile of the different risks
and guarantees must be analyzed to avoid any undesirable bunching
of exposures. Identification of priorities is required to facilitate choices
between competing proposals for government guarantees. And, once
sound governance arrangements are in place, ongoing central moni-
toring is required of the performance of individual agencies in mitigat-
ing the fiscal risks to which they are exposed. Compliance by individual
agencies will, in turn, mean they must have the necessary accounting
and information systems to capture and report the relevant informa-
tion, and an effective internal control environment.
Kazakhstan provides a good example of a centralized approach to
the ongoing management of existing guarantees (see Box 2.2). Every
entity whose foreign borrowing is guaranteed by government is re-
quired to provide quarterly financial statements to a special monitor-
ing unit in the Ministry of Finance. They also must lodge the funds
required for loan repayment installments into a restricted account one
month before each installment is due and maintain the equivalent of
one installment in the account at all times. This provides scope for the
government to take action in the event of a failure to lodge a payment,
before a call is made on the budget to honor the guarantee.
Another measure that would strengthen the accountability of
policymakers and civil servants for prudent management of risks would
be to incorporate risk-taking and risk management within the scope of
the external audit. This would mean that the supreme audit institution
would audit the information provided by each government agency on
the fiscal risks to which it is exposed. In New Zealand, each govern-
ment department must maintain a Register of Contingent Liabilities in
which the details of all contingent liabilities are recorded. Each minis-
ter responsible for a department is required to sign a certification twice
each year that the department's schedule of contingent liabilities is
complete and accurate to the best of his or her knowledge. These docu-
ments are subject to external audit. The New Zealand controller and
auditor general also has reported to Parliament on the performance of
individual entities in managing specific fiscal risks (see New Zealand
Controller and Auditor General 1999). Similarly, the U.S. General
Accounting Office reports to the U.S. Congress on contingent liabili-
ties and other risks facing the federal government and on the analysis
and management of the risks by the responsible departments.22
There may still be a problem with guarantees "in the bottom drawer"
that never see the light of day until they are presented as a fait accom-
pli to the ministry of finance. This is likely to be a problem in coun-
tries with a history of hidden guarantees and off-budget activities. One
way to try to break out of this low-level equilibrium might be to take
the approach sometimes adopted for expenditure arrears-that is, the
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 71
Box 2.2 Management of Fiscal Risks in Kazakhstan
Kazakhstan provides a good example of a concerted approach to strength-
ening the management of fiscal risks (Kazakhstan 2000). The impetus for
reform came after a proliferation of guarantees culminated in a large call
on the budget during a difficult period of fiscal consolidation. This situa-
tion focused attention on the need to bring guarantees under effective
centralized control. Kazakhstan also faces substantial variability of bud-
get revenues, in part because of its reliance on oil revenues.
Reforms introduced to improve the management of fiscal risks include
the following:
* All guarantees now require the prior approval of the minister of
finance.
* Every entity whose foreign borrowing is guaranteed by government
is required to provide an annual business plan and quarterly financial state-
ments to a special monitoring unit in the Ministry of Finance. They also
must lodge the funds required for loan repayment installments in a restricted
account one month before each installment is due and maintain the equiva-
lent of one installment in the account at all times. The Ministry of Finance
reports quarterly to the government on the financial condition of guaran-
tee recipients. This reporting provides scope for the government to ensure
timely action is taken to prevent a call on its budget for covering such debts
in the event of a failure by the guarantee recipient to lodge a payment.
* The government has introduced an annual ceiling on the stock of
new guarantees that can be issued. This ceiling has resulted in careful
scrutiny of all new guarantee proposals. A number of foreign-financed
new capital projects have not gone ahead because, on closer examina-
tion, it was decided that it was not appropriate for the government to
issue a guarantee.
* An allowance is made within net lending in the central government
budget for the estimated cost of guarantees in the next year. This allow-
ance comprises a mixture of one-off guarantees that are very difficult to
estimate and some large guarantee programs for which some historical
loss information is available. If the funds are not required, the spending
authority lapses at the end of the year.
* A revenue-dependent contingency fund has been established within the
budget. It contains a list of worked-up capital projects that can only pro-
ceed as and when interim budget revenue targets are met during the year.
government announces that any government guarantees that have not
been declared to the minister of finance by a particular date will not
be honored. If combined with a commitment by the government to
publish a statement containing all guarantees, this approach might be
sufficiently credible to flush out the legitimate government guaran-
tees. The recipients of legitimate guarantees would have both a strong
72 MURRAY PETRIE
incentive to ensure that the guarantee's existence was declared and the
means to monitor whether it was declared. The success of such an
approach would, however, be critically dependent on the presence of
sufficient political will to enforce the strategy.
The next step would be the imposition of a control framework for
the management of existing fiscal risks. This step might be taken on a
decentralized basis, in which each agency would be responsible and
accountable for managing its own fiscal risks in a prudent manner.
Central agencies would monitor performance, but might require that
a prior clear and comprehensive risk management strategy be in place
in each agency. Alternatively, or in addition, the ministry of finance
might impose specific risk management policies and practices for ge-
neric or significant risks (for example, a foreign exchange risk man-
agement policy to which each agency with significant foreign exchange
exposure must adhere). Specific responsibilities should be assigned
within government for monitoring fiscal risks across the SOE and fi-
nancial sectors.
Central controls also should be imposed on taking on new specific
fiscal risks. Because of the potential for budget discipline to be circum-
vented, and for a loss of fiscal control through the issuing of guaran-
tees, centralized processes should be put in place to enable the
government to control the issuing of new guarantees and indemnities.
Depending on the country concerned, this control might mean a re-
quirement for prior approval of the minister of finance, the cabinet, or
parliament.23 In Japan, the Parliament approves an annual ceiling on
the face value of new guarantees issued in each fiscal year (Japan,
Ministry of Finance 2001).
Approval of a new guarantee also might depend in part on the cir-
cumstances of the recipient of the guarantee-such as its financial sound-
ness or the quality of its governance. In South Africa, policy restricts the
issuing of guarantees to certain situations (see Box 2.3). For example,
no guarantees are provided to private institutions unless management
decisions can be influenced directly by the government. Processes also
might be put in place to constrain the issuing of new guarantees to the
annual budget round. Such a step would at least allow some comparison
of the merits of individual spending and guarantee proposals in terms of
cost-effectiveness. It also could facilitate broad judgments about the
consistency of the total "spending and guarantee package" with the
government's budget and medium-term fiscal strategy.
Some Concluding Remarks
There is a hierarchy of approaches to improving the management of
fiscal risk. A fundamental first step is developing an understanding of
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 73
Box 2.3 Management of Fiscal Risks in South Africa
South Africa's approach to fiscal risk management comprises both a
macro-level framework and provisions at the micro level (Kruger 1999).
The macro-level framework includes:
* full transparency of fiscal management to ensure accountability (fail-
ure to comply with reporting requirements is a criminal offence);
* a medium-term expenditure framework, which enhances transpar-
ency and predictability;
* intergovernmental fiscal arrangements involving constitutional re-
strictions on the ability of provincial governments to borrow and powers
for the national government to intervene in the event a province incurs an
unauthorized expenditure;
* a coordinating and supervisory role in borrowing by state-owned
enterprises (SOEs); and
* a regulatory environment for the banking and financial sector so
that systemic risks do not pose a threat to planned fiscal outcomes.
The framework for managing fiscal risks at the micro level includes:
* quantification of all financially related assets and liabilities;
* a clear distinction between contingent liabilities and actual liabilities;
* strict guidelines for issuing guarantees-no guarantees to assist pri-
vate institutions unless management decisions can be influenced directly;
guarantees may be provided where there is an obligation in terms of inter-
national treaties, or where foreign loans are considered to be in the na-
tional interest (guarantees for commercial entities are in the process of
being phased out);
* guarantee fees to act as a disincentive to use guarantees and to create
a level playing field where SOEs are competing with the private sector;
* a Public Finance Management Act that establishes full account-
ability, clear reporting responsibilities, and the use of accrual accounting
principles, including the production of a consolidated balance sheet; and
* management of implicit fiscal risks through classifying SOEs on the
basis of the tolerable risk appetite per institutional type-for example,
through restructuring commercial enterprises for privatization; through full
cost recovery in public utility pricing with cross subsidization to be trans-
parent; and through insurance providers charging risk-related premiums.
the main sources of fiscal risk in a particular country. Such an under-
standing requires aggregation and centralization of information across
the central government on the specific fiscal risks to which individual
government agencies and the central government itself are currently
exposed (using, for example, the Fiscal Risk Matrix presented in Chapter
1). This step in itself is a demanding exercise. From this step should
74 MURRAY PETRIE
follow attempts to assess at least the broad order of magnitude of the
most significant risks and how they affect different elements of the
government's revenues, expenditures, assets, and liabilities. This as-
sessment enables ongoing monitoring of risks, identification of prior-
ity areas for reducing exposure to existing risks, and some control over
taking on new risks.
Major gains might be made at this stage in many countries through
reexamining some basic policies from the perspective of fiscal risk
management. For example, the framework for intergovernmental fis-
cal relations, the need to retain government ownership of some SOEs
and banks, the management of public debt, and the quality of pruden-
tial supervision of the financial sector are all areas where good policy
design and transparency can make a major contribution to reducing
fiscal risk.
Reporting detailed information on specific fiscal risks, and on the
sensitivity of the fiscal position, should be an early part of a risk man-
agement strategy. Another objective is to publish information on broad
alternative macrofiscal scenarios. Such reporting requires a support-
ing public management infrastructure. In addition to ensuring better
information for the executive on which to base fiscal policy, it pro-
vides the crucial added discipline of external accountability. Efforts to
improve the management of fiscal risks also might benefit from incor-
porating risk management into the scope of the external audit. At the
same time, improvements should be initiated on key deficiencies in
broader fiscal transparency arrangements.
Central controls over who is authorized to take on new risks, such
as issuing guarantees, should be put in place. The possibility of avoid-
ing the risk altogether, or of shifting it partially to other agents, should
be examined at the outset. Consideration of many guarantees also might
be held over for deliberation alongside other fiscal priorities in the
annual budget round. Specific responsibilities should be assigned within
the government for monitoring fiscal risks across the SOE and finan-
cial sectors and for monitoring the financial position of agencies with
government-guaranteed debt. Charging a fee for the issuance and main-
tenance of a guarantee might be a useful additional means of ensuring
sound scrutiny of proposals for new guarantees. Moreover, some at-
tempt should be made to compare the merits of guarantees against
competing fiscal priorities. This might involve setting a limit on the
value of new guarantees entered into.
A change in the basis of accounting toward accrual accounting can
result in better information and accountability for fiscal risk manage-
ment. From an implementation risk perspective, however, it would in
general seem prudent to move first from pure cash-basis reporting to
supplementary reporting of fiscal risks, and the implementation of a
general budget contingency reserve, before considering the establish-
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 75
ment of dedicated reserve funds for contingent liabilities or introduc-
ing full accrual accounting.
There is great scope for improvements in fiscal performance in many
countries through better management of fiscal risks. The deficiencies
of traditional pure cash reporting have reinforced the tendency of gov-
ernments to take a short-term and reactive approach to fiscal manage-
rnent. New and expanded sources of fiscal risk increase the imperative
for better fiscal control, and new techniques offer the possibility of
improved management and better outcomes.
Notes
1. The author would like to acknowledge helpful comments on earlier
clrafts from David Webber, Ian Ball, Jon Blondal, and Istvan Szekely.
2. Contingencies are defined by the International Accounting Standards
Committee (IASC) as conditions or situations whose ultimate outcome, gain
or loss, will be confirmed only on the occurrence, or nonoccurrence, of one
or more uncertain events. See IFAC (1998: paras. 692-701).
3. See IFAC (1998) for a discussion of the definitions of the different bases
of accounting (cash, modified cash, modified accrual, and accrual accounting).
4. With the exception of origination fees, which, if charged, show as a
favorable cash impact at the time the guarantee is issued, further distorting
the picture.
5. See U.S. GAO (1998a: 5) for a discussion of the distortions of insur-
ance and lending programs under cash accounting. "Cash-based budgeting
for federal insurance programs may provide neither the information nor
incentives necessary to signal emerging problems, make adequate cost com-
parisons, control costs, or ensure the availability of resources to pay future
claims."
6. See U.S. GAO (1998a: 6): "Many analysts believe that the cash-based
budget treatment of deposit insurance exacerbated the savings and loan crisis
by creating a disincentive to close failed institutions. Since costs were not recog-
nized in the budget until cash payments were made, leaving insolvent institu-
tions open avoided recording outlays in the budget and raising the annual
deficit but ultimately increased the total cost to the government."
7. The move to greater overall fiscal transparency, supplementary report-
ing, and the choice of accounting basis are, of course, motivated by many
considerations in addition to risk management objectives.
8. Effective accountability also may require a change in attitudes to embed
a feeling of personal responsibility for fiscal risk management in the public
service culture. Such a change should perhaps be seen in the broader context of
civil service reform. This is beyond the scope of this chapter, but may be a key
element in bringing about an improvement in fiscal risk management.
76 MURRAY PETRIE
9. These are the four general principles of the International Monetary
Fund's Code of Good Practices on Fiscal Transparency. See the IMF website
.
10. See the IMF's "Fiscal Transparency Manual" for a discussion of report-
ing public debt. The manual can be found on the IMF website .
11. Quasi-fiscal activities (QFAs) are activities undertaken under the di-
rection of government by a central bank or state-owned financial or com-
mercial enterprise that are fiscal in character-that is, the effects of the activity
could in principle be duplicated by budgetary measures in the form of a tax,
subsidy, or direct expenditure. Examples are guarantees, subsidized lending,
and financial sector bailouts. These activities have similar economic effects
whether they are undertaken by a central government agency or a central
bank or a public financial enterprise. They can be very large, and they need
to be taken into account in assessing fiscal performance and fiscal risk.
12. See IMF (1996: 85-87). Also available on the Fund's website .
13. Some OECD countries take additional steps to ensure the integrity
and quality of the macroeconomic forecasts. These range from an expert
independent review panel that comments publicly on the forecasts; to a re-
view of the macroeconomic assumptions by the National Audit Office and a
legal requirement to publish the entire Treasury macroeconomic model (United
Kingdom); to basing the government's official forecasts on private sector
consensus forecasts (Canada); to fully contestable fiscal forecasts produced
by a separate entity, the Congressional Budget Office, which reports directly
to the legislature (United States).
14. It would seem desirable, however, to report contingent liabilities where
the likelihood of actual expenditure is very small but the amount potentially at
risk is very large.
15. Where accrual or modified accrual-basis accounting is used, only those
events that are judged less than likely to result in future expenditure are in-
cluded in supplementary reporting as contingent liabilities. Those events judged
likely to result in future expenditure are recognized immediately as a liability-
that is, they are defined as liabilities rather than contingent liabilities.
16. See IFAC (2000: 171). In accounting, commitments are defined as a
government's responsibility for a future liability based on an existing contrac-
tual agreement. Examples include long-term leases or multiyear contracts for
the purchase of capital equipment.
17. Under a defined-benefit scheme, the government bears the risk of a
mismatch between the return on any scheme assets and the defined pension
obligation. Under a defined-contribution scheme, the contributor bears the
risk of uncertain return on pension fund assets.
18. That is, a balance sheet containing all the prospective cash inflows
and outflows compiled using realistic projections based on current policies.
ACCOUNTING AND FINANCIAL ACCOUNTABILITY 77
An example of a comprehensive balance sheet is the Fiscal Risk Matrix and
Fiscal Hedge Matrix shown in Chapter 1.
19. Valuing risk may be a demanding exercise; it requires a modeling
capacity that is not readily available in government offices. And even the
results of the best models need to be treated as rough estimates of future values
rather than predictions. The U.S. General Accounting Office, for example, has
commented critically on the ability of U.S. federal agencies to reasonably esti-
mate subsidy costs in their credit programs: "Until weaknesses are addressed
the credibility of loan program cost information they submit will continue to
be questionable" (U.S. GAO 1998b). Therefore, scenario analysis may be more
useful for policymakers than a single expected cost figure.
20. See IFAC (2000: 51) for a discussion of the additional reporting pre-
pared in Malaysia in the context of a modified cash basis of accounting. A
feature typical of the modified cash basis system is holding the books open for a
specified period after year-end to overcome some of the timing problems of pure
cash-basis accounting.
21. References to the ministry of finance are a generic reference to the
ministry or department with primary responsibility for fiscal policy coordi-
nation and budget management.
22. For examples of such reports, see U.S. GAO (1998a, 1998b, 1998c) and
the U.S. General Accounting Office website .
23. A recent survey of OECD countries found that in three the approval of
only the minister of finance was required for the granting of a guarantee, while
in the great majority the approval of parliament was required (Blondal 1999).
References
Blondal, Jon. 1999. "Management of Fiscal Risk in OECD Member Coun-
tries." Presentation at World Bank Course on Managing Fiscal Risks, Wash-
ington, D.C., June 8-11.
Government of New Zealand. 1999. Budget Economic and Fiscal Update 1999.
Wellington.
IFAC (International Federation of Accountants), Public Sector Committee. 1998.
Draft Guideline for Governmental Financial Reporting. New York.
.2000. Governmental Financial Reporting: Accounting Issues and Prac-
tices. New York.
IMF (International Monetary Fund), Fiscal Affairs Department. 2001. "Code
of Good Practices on Fiscal Transparency" and "Manual on Fiscal
Transparency".
IMF (International Monetary Fund), Statistics Department. 1996. Government
Finance Statistics Manual: An Annotated Outline. Washington, D.C.
Japan, Ministry of Finance, Financial Bureau. 2001. FILP Report 2000.
February.
78 MURRAY PETRIE
Japan, Study Group on Explanatory Methods of Fiscal Position. 2000. "The
Japanese Government Balance Sheet (Preliminary trial)." Tokyo, October.
Kazakhstan. 2000. "Rules for Carrying Out the Monitoring of the Financial
Condition of Legal Entities that Have Received Non-governmental For-
eign Loans under Republic of Kazakhstan Government Guarantees." Astana.
Kruger, Coen. 1999. "Managing Fiscal Risks: The South African Approach."
Presentation at World Bank Course on Managing Fiscal Risks. Washing-
ton, D.C., June 8-11.
New Zealand Controller and Auditor General. 1999. How Are State-Owned
Enterprises Managing Foreign Exchange Risk. Chapter 8, First Report for
1999. Wellington.
OECD (Organisation for Economic Co-operation and Development). 1999. How
Should Governments Invest Financial Assets and Manage Debt? PUMA/SBO/
RD (99). Paris.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington D.C.
U.S. GAO (General Accounting Office). 1998a. Budgeting for Federal Insur-
ance Programs. GAO/T-AIMD-98-147. Washington, D.C.
. 1998b. Credit Reform: Greater Effort Needed to Overcome Persistent
Cost Estimation Problems. GAO/AIMD-98-14. Washington, D.C., March.
.1998c. Report on US Government's Consolidated Financial Statements
for FY97. GAO/AIMD-98-127. Washington, D.C.
United States. 1999. Budget of the United States Government, Analytical Per-
spectives. Fiscal Year 1999. Washington, D.C.: Government Printing Office.
CHAPTER 3
Budgeting for Fiscal Risk
Allen Schick
University of Maryland
As MEASURES OF FISCAL RISK, conventional budgets are deficient on
two counts. First, they have a short time frame-one year in countries
that have only annual budgets, three to five years in countries that
budget within medium-term fiscal frameworks. These time horizons
are too short to account for the downstream risks taken by govern-
rnents when they establish pension systems and other entitlements,
issue or guarantee loans, or promise to make good on shortfalls in
financial performance. Second, conventional budgets record only cash
flows; they do not account for the buildup of liabilities, contingent
obligations, or the future cost of past commitments.
Because budgets measure cash flows rather than liabilities and short-
term payments rather than long-term risks, politicians have an incen-
tive and opportunity to provide benefits to those who seek assistance
from government in ways that mask the true cost, leading to policies
and actions that worsen future fiscal conditions. Governments create
fiscal illusions, beneficiaries behave in morally hazardous ways, and
the upshot is escalation in fiscal jeopardy. This pattern is widespread;
it occurs in developed, transitional, and developing countries. In de-
veloped countries, the failure to properly control and account for risk
may burden future budgets and take a bite out of economic growth. In
transitional and newly developed countries, however, the failure may
retard development and reverse recent economic gains. In developing
countries, failure to deal adequately with risk may diminish already
d1im economic prospects.
The study of fiscal risks held by government is in its infancy. Build-
ing on work done by Hana Polackova Brixi and others, this chapter
discusses means by which national budgets might be transformed into
more effective instruments to control fiscal risk (Polackova 1998a).
79
80 ALLEN SCHICK
The first section briefly outlines alternative approaches for dealing
with fiscal risk, and the two sections that follow examine alternative
approaches for incorporating risk into financial statements and bud-
get decisions. The final sections consider options for controlling fiscal
risks through market-type arrangements.
Approaches to Managing Fiscal Risk
Governments determined to manage the risks they take can choose
from a variety of approaches. Some have been tried by a few govern-
ments, some by none. This section clarifies the various approaches,
which are elaborated later in this chapter.
The first approach is for government to be open about the types of
risks it faces, the volume and possible costs of these liabilities, and the
probability that various commitments will come due. This approach is
in line with the contemporary drive for transparency in fiscal matters,
but it requires distinguishing between explicit and implicit risks.
The second approach is to incorporate decisions on risks into the
ongoing budget process, thereby enabling the government to compare
direct and contingent expenditures without biasing the outcome in
favor of one or another type of transaction. Not all risks can be man-
aged through the budget, however. The more direct and explicit the
risk, the greater is the suitability of budgeting for estimating the cost
to government and setting aside resources for this purpose.
Third, government can manage risk by limiting risks before they
are taken. This approach would entail establishing criteria for deter-
mining whether the government should issue guarantees or enter into
other contingent commitments, assessing the degree of risk in the light
of these criteria, and refusing to take on risks that do not meet the
government's standards.
Finally, government may rely on market-type mechanisms to shift
all or a portion of the risk to private entities. Some of the innovations
in this area have had little or no application in government. Although
they are commonly used by commercial enterprises, the aim here is
not to recommend particular reforms but to stimulate innovative think-
ing on how governments might come to grips with practices and con-
ditions that may jeopardize their future fiscal health. Table 3.1
summarizes features of the four approaches.
The four approaches are differentiated not only in method but in
objective as well. This first is based on the notion that government
should be informed before it takes on new risk. The second empha-
sizes budget neutrality-the rules of budgetary accounting should not
bias policy in favor of any particular instrument. The third approach
rests on the notion that government should be risk-averse and should
BUDGETING FOR FISCAL RISK 81
Table 3.1 Comparison of Four Approaches to Managing
Fiscal Risk
Approach Main objective Limitation(s)
1. Reporting Transparency: fuller Many governments do not have
on financial account of financial accurate, comprehensive financial
statements condition and risks. statements. These statements can-
not cover implicit risks or risks with
low probability. Just publishing
statements does not itself change
risk-taking behavior by government.
2. Cost-based Budget allocations Costing methodology is not well
budgeting reflect prospective developed, and cost estimates may
cost to government; be unreliable. Does not cover
risk-taking competes implicit risks. May be treated as a
with other claims technical exercise rather than
on budget. as a real allocation of resources.
3. Rules for Applies to guarantees Political pressure may override the
taking fis- and other contingent criteria. Very few governments
cal risks liabilities; criteria have rules determining when to
applied before risk enter contingent liabilities.
is taken.
4. Market-type Relies on market to Governments typically take on
arrangements reduce risk held by contingent liabilities because of a
government and to decision not to rely on the market.
more accurately
estimate the cost of
risks taken by it.
Source: The author.
accept new risk only when stringent criteria have been fulfilled. The
fourth approach takes the view that in the best of circumstances gov-
ernment is inherently a poor assessor and regulator of risk, and so it
should turn these tasks over to the market.
Transparency in Reporting on Fiscal Risk
]Liabilities and expectations, which are neither known nor recorded,
cannot be effectively controlled. An essential first step, then, in mak-
ing risks held by government transparent is to inventory the array of
82 ALLEN SCHICK
risks and liabilities borne by it. This is not an easy task, however, and
so the first effort may not yield a comprehensive or completely accu-
rate account. Several governments, including that of South Africa, have
used the Fiscal Risk Matrix presented in Chapter 1 to identify the risks
they are holding and the policy remedies that might be applied. In
South Africa, the official charged with this responsibility has indicated
that striving to fill in each of the boxes in the matrix made government
aware of significant risks that were previously unknown. Table 3.2
shows the results of this work. Note that for most of the entries the
government has been unable to estimate the downstream costs it might
face. Nevertheless, filling in the matrix has been a useful early step in
mapping out the government's exposure and response.
In compiling an inventory, it is necessary to canvass state entities
and programs in order to identify the agencies authorized to enter into
commitments, the transactions or conditions that have been insured,
the contingencies that would trigger government payments, and the
volume of outstanding liabilities. Although a comprehensive account
may be out of reach, government can identify most of its direct and
contingent obligations by concentrating on those sectors and programs
in which these risks typically occur: agriculture, housing finance, bank-
ing, small business, state enterprises, imports and exports, insurance
schemes, and infrastructure development.
There are two schools of thought on whether government should
acknowledge implicit liabilities. One urges transparency; the other
counsels fuzziness. Both aim to discourage moral hazard, but they dif-
fer on how this is best accomplished. The case for openness and disclo-
sure rests on the argument that moral hazard abates when the
government clearly and credibly signals how it will respond to pos-
sible failures and events-for example, whether it will indemnify un-
insured depositors or pay for essential services provided by bankrupt
municipalities. It might declare that depositors will be compensated
for losses only up to a certain amount or that it will finance certain
municipal services but not others. These signals, it is argued, will deter
affected parties from behaving in ways that add to the government's
potential liabilities. Of course, signaling its intentions is effective only
if the government holds a credible position and acts as promised. Wrong
signals provide added incentives for misbehavior.
It would be appropriate for the government to take further action
to discourage moral hazard when it makes implicit liabilities explicit.
For example, if the government steps in to pay for the ongoing services
of insolvent municipalities, it also should enact laws or regulations
that deter local governments from spending beyond their means, as
well as legislation extending bankruptcy rules and procedures to these
entities. If it accepts responsibility for expected (but not legally re-
quired) future pensions, the government should restructure the pension
BUDGETING FOR FISCAL RISK 83
Table 3.2 South African Policy Approach
Type of risks Policy approach
Explicit direct risks
Sovereign borrowing Identify risks and formulate risk-averse strategy.
(ZAR380 billion)
Medical schemes Adjust policy in budget.
Civil pensions Adjust policy in budget.
Explicit contingent risks
Loan guarantees Phase out guarantees.
(ZAR73 billion) Revise authority to borrow and issue guarantees.
Cap borrowing authorities and approve and
coordinate borrowing strategies.
Guarantees on private Share risk (contracts).
investment Establish joint project limits.
Establish country limits.
Cap limits per institution.
State insurance schemes Cap government risk exposure.
Share risks (also offshore).
Implicit direct risks
Socioeconomic expen- Analyze policies.
diture Establish medium-term expenditure framework
also to reflect contingent liabilities.
Better reflect cost in annual budget.
Recurrent expenditure Incorporate in fiscal planning and budgeting.
of public investment Introduce "corporate governance" in projects.
(also state-owned
enterprises)
Implicit contingent risks
Default of subnationals Monitor and introduce ex ante warning signals.
Systemic risks Monitor.
Liabilities and risks of Consider fiscal risks when restructuring.
policy failure from Monitor.
privatization/commer-
cialization
Disaster relief/unavoidable Build contingency reserves.
expenditure Establish contingent credit lines and purchase
reinsurance.
Monetary/exchange Rethink interest rate and exchange rate policy
management to contain government risk exposure.
Monitor central bank reserve management,
derivative use, and risk exposure.
Source: The author.
84 ALLEN SCHICK
system to put it on a sound financial basis. Without corrective mea-
sures, making implicit costs explicit would almost certainly worsen
the fiscal posture.
The counterargument is that the government should not divulge its
intentions on implicit risk because doing so would greatly increase moral
hazard. This is the position taken by the International Monetary Fund
(IMF) in statements elaborating on its new Code of Good Practices on
Fiscal Transparency (IMF 1999). The IMF code vigorously promotes
openness on fiscal matters, but it nevertheless recommends that "im-
plicit guarantees, such as the possibility that a government may in the
future bail out a public enterprise or private sector bank" be excluded
from statements on contingent liabilities (paragraph 67). The statement
adds that implicit guarantees should be excluded "because of the poten-
tial moral hazard to which being transparent about such provisions could
give rise" (emphasis in original).
When a government considers whether to make implicit liabilities
explicit, it must take account of the credibility of its position. More
than one government has announced that it will not pay depositor
claims on failed financial institutions only to be compelled by political
or economic circumstances to provide assistance.
Indeed, there can be no blanket rule for all implicit liabilities. In
some cases, maintaining a fuzzy position will shift a portion of the risk
to market actors if they bet wrong on what the government will do. In
other cases, fuzziness will leave the government with higher costs in
the end when it makes good on implicit commitments. Implicit pen-
sion liabilities should not be treated the same as implicit exchange rate
guarantees. For the former, the government might do well to recognize
the liabilities and to provide for them in the budget; for the latter, the
government might be better off keeping importers and exporters un-
certain about its intentions.
Accounting for Contingent Liabilities
Public accounting systems generally recognize direct liabilities, not
contingent ones. The Draft Guideline for Governmental Financial
Reporting issued by the International Federation of Accountants (IFAC)
in 1998 takes the position in paragraph 443 that "commitments and
contingencies are items which do not meet the definition and recogni-
tion criteria" for incorporation in financial statements (IFAC 1998).
These criteria define a liability as "a present obligation of the enter-
prise arising from past events, the settlement of which is expected to
result in an outflow . . . of resources." According to IFAC, a liability
should be recognized in financial statements when "it is probable that
an outflow of resources . . . will result from the settlement of a present
obligation and the amount at which the settlement will take place can
BUDGETING FOR FISCAL RISK 85
be measured reliably" (paragraph 443). Contingent liabilities do not
satisfy these criteria, because they depend on future rather than past
events and cannot be reliably measured. Nevertheless, the IMF Code
of Good Practices on Fiscal Transparency specifies in Statement 2.1.3.
that "[s]tatements should be published with the annual budget giving
a description of the nature and fiscal significance of contingent liabili-
ties, tax expenditures, and quasi-fiscal activities" (IMF 1999). (For a
more detailed discussion of accounting and reporting of contingent
liabilities and other fiscal risks, see Chapter 2 by Petrie in this volume.)
Measuring Risks
The IMF fiscal transparency code urges that each nation present in its
annual budget a statement of fiscal risks, including those deriving from
guarantees and insurance schemes, and that, where feasible, these risks
be quantified. Doing so is a challenging task, for risk assessment and
measurement are much more advanced in the business sector, where
various statistical tools and hedging strategies are widely used. Few
governments have had much experience in this area, but there is no
reason why they cannot adapt relevant commercial practices to their
needs. As governments gain experience in assessing risks, the quality
of their estimates is likely to improve.
Reporting the estimated costs of contingent liabilities and other fis-
cal risks in budgets or on financial statements may spur governments
to produce point estimates, which specify a definite cost but are al-
most always wrong and misleading. Few (if any) governments have
the capacity to measure accurately the probability that future contin-
gencies will come due and the cost they will incur if they have to make
payments pursuant to these obligations. Less than a decade after it
spent more than US$100 billion resolving widespread insolvency in
the banking sector, the U.S. government issued in fiscal 2000 a provi-
sional balance sheet estimating the present value of future deposit in-
surance liabilities at only $1 billion (United States, OMB 1999: Table
2-1). It may be that measures adopted to regulate the lending practices
of financial institutions and fees paid for deposit insurance will pro-
tect the U.S. government against future losses. But surely one cannot
rule out the possibility of widespread distress in the banking industry
or capital markets that would cause insured losses to escalate.
Ideally, risks should be estimated in terms of a range, with the key
assumptions and probabilities published alongside the estimates. In
addition, given the difficulty of estimating future costs, the following
suggestions may be useful for governments with limited capacity to
rnanage risks. First, concentrate on the riskiest endeavors, the ones
likely to account for most downstream liabilities. Second, make pre-
cise cost estimates only when warranted by experience and when the
86 ALLEN SCHICK
risks are pooled rather than concentrated. And, third, report fiscal risks,
even when it is not possible to quantify costs.
Formulating Financial Statements
Although IMF recommends that a statement on contingent liabilities
be included in a country's budget, there is a modest trend toward pub-
lishing them as notes to financial statements. Three contemporary de-
velopments have given new prominence to government financial
statements: (a) the shift underway from cash-based public accounting
to the accrual basis; (b) the growing reliance on financial statements to
report on a government's financial condition; and (c) the broadened
role of auditors in reviewing these financial statements and in assess-
ing the government's performance.
New Zealand pioneered in accrual accounting and budgeting a
decade ago when its Public Finance Act (1989) mandated that all gov-
ernment entities apply commercial accounting principles. The conver-
sion was swift and relatively painless and has led to the publication of
audited financial statements for all departments as well as a combined
financial statement for the government; a supporting schedule to this
statement lists both quantifiable and nonquantifiable contingent li-
abilities. Over the past decade, approximately a dozen national gov-
ernments have followed suit, and more are likely to join the movement
to the accrual basis under prodding from international organizations.
In commercial practices, contingent risks have an impact on the
balance sheet when provision is made for expected losses. To the ex-
tent, therefore, that losses on contingent liabilities can be measured, it
would be appropriate to make provision for them on the balance sheet.
But at this stage, cost estimates for these liabilities usually lack the
precision associated with recording direct liabilities. The most sen-
sible approach, therefore, may be to list contingent liabilities in the
notes, not in the financial statement.
Budgetary Neutrality
The rationale that has led some governments to introduce accrual ac-
counting also applies to the budget. Cost-based budgeting is designed
to make a government responsible for the resources it uses and strength-
ens the government's capacity to manage risks that are taken in cur-
rent budgets but payment for which emerges in subsequent fiscal years.
Polackova (1998b: 50) correctly argues that an "accrual-based account-
ing system without accrual budgeting is neither necessary nor suffi-
cient to ensure adequate policy consideration for contingent liabilities
and other fiscal risks." But there are several impediments to budgeting
BUDGETING FOR FISCAL RISK 87
for contingencies on a cost basis. One is that the development of ac-
counting standards for government budgets is still in its infancy. An-
other, already noted, is that cost estimates tend to be less reliable for
contingent rather than direct liabilities. As long as cost is based on
budget estimates rather than market prices, combining direct and con-
tingent expenditures in a single cost measure may be highly mislead-
ing as to the resources actually expended or risked by government. A
third issue is that the time horizon of budgeting may be too short for
allocating costs arising from contingent liabilities. Even in countries
that have a medium-term timeframe, budget projections typically ex-
tend out only three to five years. Fiscal risks often spill over well be-
yond this period.
There are four basic approaches to budgeting for contingent liabili-
ties. One is to present background information on contingent liabili-
ties and other financial risks in the budget, but to make budget decisions
only for direct expenditures and for payments pursuant to existing
commitments. Another is to devise a separate budget for contingent
liabilities and risks. The third is to integrate direct and contingent
liabilities on a cash basis. And the fourth option is to integrate the two
types of liabilities on a cost basis. The first three options are discussed
briefly, and the final one is considered in the light of novel U.S. bud-
getary procedures for direct and guaranteed loans.
Presenting Information in the Budget
The first option is to publish information on contingent liabilities and
risks in the supporting schedules, but not in the budget estimates. The
inclusion of supplementary information is a common practice in bud-
geting. Many national budgets provide background data on the
economy, grants to subnational governments, programs and activities,
and other matters. Although the supplementary information is not
voted, it assists parliament, attentive groups, and the public in assess-
ing the government's budgetary intentions. Similarly, background in-
formation could be provided on contingent liabilities, but these data
would not be combined with the estimates. For most governments,
providing such information would be a significant advance, even if the
budget listed only the various risks and liabilities but did not provide
cost estimates.
Parallel Budgeting for Contingent Liabilities
Alternatively, the government would compile a parallel budget for
contingent liabilities and related risks. Although separate from the regu-
lar budget, the "contingent liabilities" budget would be voted by par-
liament. This parallel budget would, like a regular budget, specify all
88 ALLEN SCHICK
the commitments authorized for the fiscal year; it might also limit the
amount of contingent liabilities outstanding and set aside cash resources
for expected calls on contingent liabilities during the year.
During the 1980s, a parallel budget system was introduced by the
U.S. government for direct and guaranteed loans. This parallel budget
set a total limit on the amount of new loan commitments and allo-
cated this total among particular programs and agencies. This separate
system was replaced in the early 1990s by the integrated arrangement
described below. This U.S. system is limited to risk associated with
direct and guaranteed loans; it is not applied to other contingent
liabilities.
Clearly, few governments have sufficient information to compile a
comprehensive budget for all contingent liabilities. It may be appro-
priate, therefore, to budget only for those contingencies for which rea-
sonably reliable data are available. For example, government may have
reasonably accurate and complete information on the loans it guaran-
tees, but know little about the debt incurred or guaranteed by state-
owned enterprises. In such a situation, it would be prudent for the
budget to cover only the former, even though the latter may pose greater
risk. Of course, government should endeavor to progressively improve
its coverage of fiscal risk by investigating areas where information is
meager but the risk of loss is great.
An alternative approach would be to budget for changes in the vol-
ume of known contingent liabilities and to concentrate on those pro-
spective losses that can be reasonably estimated rather than all such
liabilities. To do this, the government might construct a baseline (simi-
lar to the baselines used for expenditure projections in medium-term
frameworks) that would estimate future payouts for previously autho-
rized or outstanding contingent liabilities. Each year the baseline would
be adjusted for changes in projected payouts stemming from new gov-
ernment actions (such as the issuance of additional guarantees, changes
in economic conditions affecting the probability that the government
will have to make future payments, and reestimates of the losses ex-
pected from existing contingent liabilities). The baseline data would
be presented for government decision. Each year the government would
decide on new explicit contingent liabilities by adjusting the baseline.
But the government would not control total contingent liabilities; in-
stead, it would focus on year-to-year changes in estimated payouts for
those risks included in the baseline.
Integrating Contingent Liabilities into the Cash-based Budget
The third option would go further and combine payments on contin-
gent liabilities with the conventional cash-based budget. In this ar-
rangement, the government would set aside resources in the budget to
BUDGETING FOR FISCAL RISK 89
pay for losses expected during the year or over the medium term. It
also would use the budget to regulate the total volume of guarantees
or the amount of new guarantees to be issued during the fiscal year.
A comparison of the approach initially taken by the Netherlands
and that applied by Hungary shoiws various ways of integrating guar-
antees with direct expenditures. When it initially incorporated guar-
antees in the budget, the Netherlands recorded total new guarantees
as expenditures. Thus within a fixed budget constraint, the issuance
(or authorization) of new guarantees crowded out an equivalent amount
of direct expenditure. In effect, the budget made provision for the full
value of guarantees, not for expected payouts. Clearly, this treatment
was intended to discourage the issuance of guarantees in lieu of con-
ventional grants and subsidies. Once this practice was developed, the
Netherlands shifted to a "cost" basis that budgeted for estimated
payouts.
The government of Hungary has adopted seven interlocking budget
controls on guarantees (see Chapter 9 by Brixi, Schick, and Zlaoui in
this volume). First, the volume ol: guarantees authorized in the budget
is limited to a certain percentage of state revenues. Second, the volume
of outstanding guarantees issued by various state entities (such as the
Hungarian Development Bank) is limited by law. Third, the budget
sets aside funds for payments expected to be made during the fiscal
year pursuant to existing guarantees. Fourth, guarantee contracts are
reviewed by the Ministry of Finance, which closely monitors the issu-
ance of guarantees by departments and other entities. Fifth, material
information on new guarantees (such as the amount, conditions, justi-
fication, lender, and borrower) are published in a government resolu-
tion. Sixth, the annual budget reports the probability of default and
expected payments on each guarantee program. Finally, the issuance
of guarantees is reported to the State Audit Office.
Although the system is not perfect-for example, some major guar-
antees are exempted from the limit on the total volume issued each
year-it has greatly reduced the exposure of government to losses from
contingent liabilities. In fact, disbursements for calls on defaulted guar-
antees have been below the amounts set aside in the budget for this
purpose. The main problem for Hungary has been dealing with im-
plicit contingent liabilities. These are not covered by the new budget
control system, and payments to cover these losses have far exceeded
those made for explicit liabilities.
Budgeting for the Cost of Contingent Liabilities
Budgeting for contingent liabilities on a cash basis may result in either
too much or too little control of the government's risk of loss. The
early approach taken by the Netherlands overstated the cost, because
90 ALLEN SCHICK
the government was likely to be liable for only a fraction of the losses
it was guaranteeing. By contrast, the Hungarian method may under-
state cost, because the government will have to make good on its com-
mitments even if the amounts provisioned in the budget are inadequate.
Moreover, when contingent liabilities are budgeted on a volume
and cash basis, politicians may have an incentive to substitute them
for grants and other disbursements. While cash payments are bud-
geted as outlays in the year they are made, payments for guarantees do
not appear as outlays until later years, when default or other events
occur. On a cash basis, guarantees are inexpensive relative to grants,
even though they may cost more in the long run.
To deal with this problem, the U.S. government introduced new
budgetary rules for direct and guaranteed loans in 1992. These rules
are designed to neutralize budgetary incentives and to make politi-
cians indifferent to whether they choose grants, direct loans, or guar-
antees. These three types of transactions are budgeted on a cost basis,
rendering the timing of cash payments less relevant in allocating gov-
ernment resources. The Netherlands now uses a similar system.
The current U.S. system, which has not been altered since its intro-
duction in 1992, shifts the budgetary basis of loans and guarantees
from cash flow to subsidy cost. This cost is defined in law as "the
estimated long-term cost to the Government of a direct loan or a loan
guarantee, calculated on a net present value basis, excluding adminis-
trative costs." Net present value is calculated by discounting estimated
future cash outflows (loan disbursements and payments on defaults)
and inflows (origination fees, repayment of principal and interest on
direct loans, and recoveries), using a discount rate equal to the interest
rate paid by the U.S. government on borrowings of a comparable ma-
turity. A separate appropriation is made for the projected subsidy cost
of each loan program; this appropriation is included in the budget as
an outlay, even though money might not be disbursed until years later.
Subsidy costs are included in the computation of total budget expendi-
tures and in the surplus or deficit.
Budgeting for direct and guaranteed loans entails complex proce-
dures for estimating subsidy costs and new accounting procedures for
recording the cash flows associated with loan transactions. The pro-
cess is designed to differentiate between the subsidized portion of loans
that is budgeted as a cost and the unsubsidized portion that is bud-
geted as a below-the-line transaction. Budget resources are provided
only for the subsidy cost, which almost always is significantly less
than the face value of the loan or guarantee. The subsidy cost of direct
loans is the present value of the amounts not repaid (minus fees and
recoveries) and the difference between the interest rate charged bor-
rowers and the cost of money to government. The subsidy cost of guar-
BUJDGETING FOR FISCAL RISK 91
anteed loans is the present value of the difference between cash pay-
ments for defaults and cash received from fees and recoveries.
The subsidy cost is estimated at the time a direct loan is obligated
or a loan guarantee commitment is made. Actual loan performance,
however, often varies from early estimates, sometimes significantly.
Accordingly, the subsidy cost is annually reestimated during the life-
tirne of loans and guarantees, and an automatic appropriation is pro-
vided to cover overruns. Because this appropriation is automatic,
government agencies are not penalized in their budgets if they under-
estimate subsidy cost. Since fiscal 1992 when the credit system was
first implemented, underestimates h,ave been relatively minor, but it is
important to bear in mind that this entire period has been largely one
of robust economic growth in the UJnited States.
As noted, a key objective of the subsidy approach is to ensure that
budget decisions are neutral-that is, they are not skewed in favor of
or against any particular type of transaction. Budgeting for subsidy
cost puts direct loans, guaranteed loans, and grants on an equal basis.
All are budgeted in terms of cost to government rather than in terms of
cash exchanged. The hypothetical example below contracts the cash
basis and subsidy cost treatment of these transactions (in millions of
U.S. dollars).
Transaction Amount budgeted
Type Amount Cash basis Subsidy cost
Loan $100 $100 $15
Guaranteed loan 100 -2 30
Grant 20 20 20
This hypothetical case compares a direct loan of $100 million, a
guaranteed loan of $100 million, and a grant of $20 million. On a
cash basis, the direct loan would appear in the budget as the most
costly transaction, because the entire amount disbursed is recorded as
an outlay. By contrast, the guaranteed loan would be budgeted as a
revenue gain for the government, because it receives income from origi-
nation fees in the year the guarantee is issued. On a subsidy cost basis,
however, budgeted outlays for the clirect loan would be reduced from
$100 million to $15 million, because projected repayments of princi-
pal and interest would be included in the measurement of subsidy costs.
Budget outlays for the guaranteed loan would rise from -$2 million to
$30 million, because projected defaults in later years would be in-
cluded in the subsidy cost, making it the costliest type of transaction.
The amount recorded for a grant would not change, but the cost basis
92 ALLEN SCHICK
would make it more readily comparable to direct and guaranteed loans.
Note, however, that the hypothetical amount shown here for a grant is
based on disbursements, while the cost reported for a loan and a guar-
antee is based on projections of future discounted cash flows.
Conversion to the subsidy cost basis entails maintaining separate
budgetary accounts for the subsidized and unsubsidized portions of
loans and guarantees. Program accounts receive appropriations for
subsidy costs; financing accounts handle the cash flows associated with
the nonsubsidized portion. Program accounts are included in the bud-
get; financing accounts are recorded as "means of financing" and their
cash flows are not included in budget receipts or outlays.
The subsidy cost basis is currently used in the United States only for
direct and guaranteed loans, not for other contingent liabilities. How-
ever, legislation tabled in the U.S. Congress in 1999 (but not enacted)
would have shifted all U.S. government insurance programs to this
basis (H.R. 853, sec. 604, 106th Cong., 1st sess., 1999). The legisla-
tion provided that beginning with fiscal 2006, insurance commitments
would be made only to the extent that budget resources were appro-
priated to cover their "risk-assumed cost." In the legislation, this cost
is defined as "the net present value of the estimated cash flows to and
from the Government resulting from an insurance commitment or
modification thereof." Inasmuch as the volume of insurance commit-
ments is many times greater than that of loan guarantees, enactment
of this legislation might have an enormous impact on the budgetary
treatment of contingent liabilities. Although no action has been taken,
this type of proposal is likely to be revived in the future.
Criteria for Contingent Liabilities
The two broad approaches discussed thus far-accounting and budget-
ing for fiscal risks-would significantly enhance government's under-
standing of the contingent liabilities it faces. But they do not deal with
the critical issue in risk management: whether government should com-
mit itself to the contingent liability in the first instance. In the business
sector, financial institutions rigorously partition decisions on risk from
risk assessment. One decisionmaking entity is responsible for negotiat-
ing insurance contracts, loans, and other commitments; the other as-
sesses the credit-worthiness of borrowers, the risk inherent in the activity,
and other risk factors. Moreover, the institution would finalize its com-
mitment only if the risk factors were assessed to be within acceptable
parameters. In government, however, risk commitment and risk assess-
ment are often done by the same entity. In the United States, for ex-
ample, the government agency that guarantees loans also estimates the
subsidy cost by assessing the probability of default.
BUDGETING FOR FISCAL RISK 93
C(anadian Principles for Regulating Risk
The best time to manage risk is before the commitment is made; after-
ward, all government can do is to account for the risk it has under-
taken in financial statements or the budget and to manage its portfolio
of liabilities in ways that mitigate losses. The government of Canada
introduced a set of principles in the mid-1980s to regulate the risks it
takes on loans and loan guarantees. For this discussion, the most sa-
lient principles are the following:
* In the case of loans, any concessional terms, such as a below-
market interest rate, are treated as budgetary expenditures. In some
cases, the subsidy is so high that the entire loan is budgeted as an
expenditure.
* Before a loan or guarantee is tendered, the sponsoring depart-
ment must analyze the project and demonstrate that it cannot be fi-
nanced without government assistance, and that cash flow will be
adequate to cover repayment of the debt as well as interest and operat-
ing costs and yield a satisfactory rate of return.
* Risk must be shared with private equity sponsors who supply a
substantial portion of the required funds from their own resources.
Moreover, guarantees must provide that in the event of default, the
government shall recover its losses from private equity sponsors.
* Bankers should share risk by bearing a minimum of 15 percent of
the net loss associated with any clefault. This arrangement would give
them an incentive to undertake a rigorous assessment of their risk
exposure.
* Interest rates on loans should be set to cover the government's
cost of money and estimated future losses on loan guarantees. Fees
should be imposed to recover estimated future losses and to defray
administrative expenses.
* Provision should be made fo:r loans and guarantees at the time they
are issued. The amount provisioned should be based on an assessment
of risk, and sponsoring departments must pay for these provisions out of
fees earned in issuing guarantees or their annual appropriations.
* New loans and loan guarantee programs must be approved by
the minister of finance and authorized by Parliament.
* Departments and Crown corporations are required to report on
their contingent liabilities. These reports are published as notes to the
government's annual financial statement. Moreover, estimates of con-
tingent liabilities and losses are audited by the auditor general who
reports directly to Parliament.
Certain exceptions to these rules exist, but the overall effect has
been to compel government to consider its risk exposure before guar-
anteeing loans.
94 ALLEN SCHICK
Limiting the Government's Liability to the
Amount Provisioned for Losses
The U.S. and Canadian methods for budgeting for fiscal risks rely on
loss estimates made at the time loans are issued or guaranteed. Inasmuch
as future defaults cannot be known perfectly ex ante, it is possible they
will be underestimated. If this occurs, the amount appropriated for sub-
sidy cost in the United States or the amount provisioned for losses in
Canada will be immediate. In fact, politicians may have a strong incen-
tive to underestimate risks, especially in the American system where
an automatic appropriation is available to cover unbudgeted losses.
They do not pay a budget penalty when costs are underestimated.
In Chapter 6 of this volume, Daniel Cohen has proposed a novel
scheme to impel government to disclose the true risk deriving from
guarantees and other contingent liabilities. He would have the gov-
ernment provision for projected losses by setting aside money equal
to its estimated liability in a reserve fund. Once this money is re-
served, the government would have no further liability. All claims
against guarantees would be paid by the reserve fund, not by govern-
ment. Most important, claims would be limited to the resources avail-
able in the reserve fund; if the reserve fund were depleted, no further
claims would be paid.
This arrangement would provide lenders and others seeking gov-
ernment guarantees with a strong incentive to demand that risk be
accurately assessed. An underestimate would devalue guarantees and
shift risk from the government's reserve fund to private parties. Cohen
also has suggested that the reserve fund be privatized, with its shares
traded publicly. The share price would reflect the market's assessment
of whether adequate provision has been made for the risk insured by
the reserve fund.
The Cohen proposal has not been implemented by any government,
though doing so would not be a difficult technical feat. It is not cer-
tain, however, that his mechanism would effectively constrain the
government's liability to the amount provisioned. Ex ante valuation of
risk is inherently inaccurate, and barring government from adjusting
the amount provisioned as it gains additional information may be im-
practical. Firms often adjust the amount provisioned for bad debt and
other losses in response to new information, and there is little reason
to expect the government to behave otherwise. One can foresee enor-
mous political pressure on a government to bail out underfunded re-
serve funds; in fact, it is likely to be blamed for the underfunding. Of
course, if government has the option of replenishing the reserve fund,
it would lose much of the incentive to provision adequately for losses.
BUDGETING FOR FISCAL RISK 95
Even if it were to curtail explicit risk, the Cohen arrangement would
not curtail the government's exposure to implicit contingent liabilities.
In these cases, there may be strong jpressure on the government to com-
pensate for losses even if it were not: legally obligated to do so, and even
if the reserve fund lacked sufficient resources to cover these claims.
Using the Market: to Regulate Risk
An alternative to the Cohen plan would be to allow the market to
assess and allocate risk rather thati the government. The main advan-
tages to bringing the market into play would be less political oppor-
tunism, diversified risk, and lower government losses. Every risk insured
by government can be insured cormmercially at some cost, or not at all
if: the probability of loss were so high that no private insurer would be
willing to take it. When government replaces the market, it shifts the
cost to itself. Understandably, private risk takers actively search for
opportunities to shift the cost to government, and they often have little
difficulty recruiting politicians to go along. Any market-type remedy,
therefore, must return all or a portion of the cost to private risk-takers.
The easiest way to accomplish this would be for the government to
refrain from tendering guarantees: another would be to adopt the rig-
orous screening criteria applied by Canada. But assuming the govern-
rnent was bent on accepting risk, it might take several measures to
reduce its exposure.
Risk-sharing
One simple way for government to reduce risk is to share it with lend-
ers, borrowers, importers/exporters, enterprises, or others seeking guar-
antees. If the government were to enforce a rule that it would assume
no more than 50 percent of the risk, the private parties holding the
other half (or more) of the risk would have to think about their own
exposure before proceeding with the transaction. A bank would have
to consider the credit-worthiness of borrowers; it would no longer
suffice to care only about the quality of the guarantee. Some guaran-
teed transactions might still proceed, but others surely would be aborted
if the government limits its liabiliity.
In a variation on this approach, the government would insure the
[ast rather than the first loss. Risk-sharing would be promoted by the
use of high deductibles, which would require the insured party to pay
for the loss up to a certain monetary value. Government liability would
take effect only after the deductible is satisfied.
96 ALLEN SCHICK
Risk-based Premiums
One of the anomalies of risk-taking is that government often charges
less for its riskiest guarantees. A well-established enterprise might be
charged market interest rates or an initiation fee; startup ventures
might receive concessional interest rates and have the fee waived. As
perverse as this seems, there is a certain political-economic logic to
this behavior. Risky borrowers, the argument runs, need government
assistance because they have no recourse to private markets. They
cannot afford to pay up-front fees or at-market interest rates. There-
fore, the government should assist them by forgoing these charges or
offering concessionary terms. In effect, the guarantee serves as a sub-
sidy. Arguably, guarantees are an inefficient form of subsidy. Assis-
tance might be better provided through grants rather than contingent
liabilities that mask the true cost of government. On this basis, gov-
ernment would do well to charge risk-adjusted premiums for its guar-
antees. Some lenders and borrowers, importers and exporters, and
other risk-takers would be deterred by high premiums, thereby re-
ducing government's exposure.
Reinsurance of Government Risk
One of the most common means used in the private sector to limit
liability is the purchase of reinsurance. This practice is rarely applied
in government, however. There is no technical impediment to govern-
ment purchasing reinsurance when it guarantees loans, agricultural
prices, or any other event or outcome. As in markets, the amount paid
by government would reflect the risk it has taken. The cost of reinsur-
ance not only would give government a powerful signal about the risk
it is holding, but also might dissuade it from taking risks that would
require very high reinsurance premiums. In other words, the practice
of immediately reinsuring itself would deter government from under-
writing transactions adjudged by the market to have the highest prob-
ability of loss.
The reinsurance model can accommodate several variations. For
example, government could reinsure only a portion of risk, or it could
base premiums on the cost of reinsurance. In both cases, reinsurance
would be a means of promoting risk-sharing.
Conclusion
Risk management is still in its infancy in public finance. Governments
have little of the experience and few of the instruments used by firms
and markets to assess and control risk. Some techniques perfected in
BUDGETING FOR FISCAL RISK 97
the private sector may be adapted for government use in the years
ahead, and some of the cutting-edge practices introduced by Canada
and the United States might be tried in other countries. But dealing
with contingent liabilities is not easy in any country and is especially
difficult in developing and some transitional countries where the mar-
kets and insurance sectors are relatively undeveloped and where the
nmargin for error is narrow.
Some critics have urged that the best posture is for governments to
take few risks. Clearly, risk-sharing can be more broadly applied by
insuring only a portion of possible loss, levying risk-adjusting premi-
ums, and purchasing reinsurance. But in modern times, even well-
managed governments in sturdy economies are called on to accept
risks that in an earlier age might have been held by the household or
enterprise. Many positive things, including economic improvement,
have happened because governments have taken fiscal risks. It is pre-
cisely because governments will continue to expose themselves to vari-
ous contingencies that they should be more transparent about the risk
they face, more willing to make provision for these risks in their bud-
gets, and more insistent on sharing the risk with others.
References
IFAC (International Federation of Accountants), Public Sector Committee. 1998.
Draft Guideline for Governmental Financial Reporting. New York.
IMF (International Monetary Fund), Fiscal Affairs Department. 1999. "Manual
on Fiscal Transparency." Washington, D.C.
Polackova, Hana. 1998a. "Contingen-t Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
. 1998b. "Government Contingent Liabilities: A Hidden Risk to Fiscal
Stability-A Consideration for EU Accession." In European Commission,
European Union Accession: The Challenges for Public Liability Manage-
ment in Central Europe. Washington, D.C.: World Bank.
United States, OMB (Office of Management and Budget).1999. Analytical Per-
spectives, Budget of the United States Government, Fiscal Year 2000. Wash-
ington, D.C.: Government Printing Office.
CHAPI'ER 4
Institutional and Analytical
Framework for Measuring and
Managing Government
Contingent Liabilities
Suresh M. Sundaresan
Graduate School of Business, Columbia University
THE MANAGEMENT OF RISK in tthe process of earning an attractive
return on the capital employed is a challenge that confronts many
organizations in the private and public sectors. For private corpora-
tions, which are widely regarded to be profit-maximizing (or value-
maximizing) entities, this task has assumed paramount importance
with the increased level of competition, globalization, and securitization
of the markets where risks are priced and traded. The institutional and
analytical frameworks used to manage and measure the risks of corpo-
rations and set the right incentives for managers have been studied
extensively. As for the public sector, economists have widely investi-
gated the problems associated there with the measurement of deficits
and contingent liabilities, but no operational guidelines have emerged
on how contingent liabilities should be reported in any measure of
government deficit.
The measurement of risk and its accurate reporting in the calcula-
tion of federal, state, and city deficits are important to many organiza-
tions such as the International Monetary Fund (IMF), World Bank,
lenders in the private sector, and ultimately citizens of different gen-
erations. Fiscal liabilities tremendiously influence the deficit and thus
the actual risk and opportunities faced by a country. An example of a
liability that may result in intergenerational transfers is the social
99
100 SURESH M. SUNDARESAN
security system and the aging of the population. The percentage of
retired people in the population is increasing in many countries. This
increase will lead to a greater burden on the working population even
when other factors are held constant.
This chapter reviews the literature on measurement of fiscal defi-
cits, government liabilities, and the analytical tools for measuring gov-
ernment contingent liabilities. It also discusses the merits of measuring
a liability accurately in the context of the fact that governments typi-
cally have much discretion to legislate actions that can drastically af-
fect the future value of its liabilities (which cannot be predicted ahead).
Despite valid concerns about relying on a single measure of deficit, I
argue that an accurate measure of liability is very important to lending
institutions and legislative bodies. In this context, I argue that the valu-
ation of contingent liabilities in reporting the fiscal risk of govern-
ments is qualitatively different from a standard loan guarantee (put
option-pricing) problem. This is rooted in the observation that the
presence of a guarantee (an example of a contingent liability of the
government) alters the stochastic process of the variable on which the
guarantee is provided in the first place.
An example should amplify this point. Suppose the government
decides to guarantee loans of private sector companies in an infra-
structure area such as power. This guarantee covers the repayment of
interest and principal, which may be denominated in a foreign cur-
rency. The presence of the guarantee results in an excess supply of
power, leading to a supply response. In anticipation of this supply
response, the consumer demand for power-intensive goods also goes
up, leading to a demand response. In equilibrium, this response will
affect the price of power at which the demand is equal to supply.
This situation makes the valuation of loan guarantees a challenging
problem.
This chapter is organized as follows. The next section reviews the
economics literature on deficits as well as the operational concerns
of development institutions. It is followed by a description of the
institutional features that monitor the risks taken by corporations.
The next section contrasts the institutional arrangements that gov-
ernments face in addressing the risk management problem. It also
explores the differences between the objectives of a corporation and
of a government and how such differences may affect their propen-
sity toward risk-bearing. A description of the tools of risk manage-
ment used in the private sector follows, along with my argument that
many of these tools need to be recast in a fundamental way to be
relevant to managing the risks faced by governments. The final sec-
tion outlines a model for measuring the exposure of a government's
contingent liabilities. Most of the technical arguments are presented
in the Appendix.
MEASURING AND MANAGING CONTINGENT LIABILITIES 101
Economic Theory on Deficits
Economic theory has much to say about the measurement of deficits,
the relevance of any deficit measures, and the role of deficits in the
welfare of the economy. This section briefly reviews the literature.
Accounting, Reporting, and Policy Flexibilities
The construction of an ideal balance sheet for a country has engaged
the attention of economists for over 20 years. Papers by Buiter (1983)
and Bean and Buiter (1987) have attempted to lay down some concep-
tual groundwork to address this issue. The conceptual and method-
ological issues in the measurement of fiscal deficits, which are materially
affected by the way in which the liabilities are measured and reported,
have been discussed in an extensive review paper by Blejer and Cheasty
(1991). At the heart of measuring liabilities is the issue of measuring
revenues and expenditures of governments, on the one hand, and fi-
nancing, on the other. Blejer and Cheasty (1991) point out that "the
government debt criterion" assumes that if a transaction extinguishes
a liability or creates a liability, then it is considered as financing. Un-
der "the public policy criterion," t[ransactions that further the goals of
policymakers are classified as taxing and spending. Depending on the
doctrine used, then, the deficit measures may mean very different things.
Another item that introduces a variation in the deficit measure is the
choice between cash accounting and accrual accounting. Under strict
cash deficit accounting, only those government outlays for which cash
has been distributed within one year is part of the budget balance.
Likewise, only those actual cash revenues received within the year go
to the budget balance. In the accrual measure of deficits, the actual net
resource preemption is accounted for regardless of the cash flows dur-
ing the year. Depending on the concept used (which may be closer to
either the cash or the accrual concept), the measure of deficit will
represent different things.
The issues surrounding the measurement of deficit and their impli-
cations have been articulated in papers by Eisner (1984) and Eisner
and Pieper (1984). An insight that emerges from this strand of litera-
ture is that a single measure of deficit may never be adequate for pub-
lic policy debate or cross-country comparisons. Moreover, deficits can
display seasonal patterns, may vary with the growth rate of the economy,
and are easily changed through legislative actions and tax policies.
Eisner (1984) notes that the valuation of a government's contingent
liabilities in the calculation of deficits is subject to the criticism that
the government can legislate actions that may seriously change the
future value of its contingent liabilities. Inflation also can significantly
affect the reported budget deficit in a significant manner. Finally, when
102 SURESH M. SUNDARESAN
the economy is growing and undergoing structural changes, any mea-
sure of deficit is likely to be not very informative. Thus the measure of
deficit should control for factors such as seasonality, the stage of the
economy, and inflation levels.
Ricardian View of Deficit Measurement
The standard approach to budget deficits makes the following case: if
the government borrows money (and effectively reduces taxes), it may
promote aggregate demand by the consumers. An implication is that
the private savings will be lower than the implied tax cut. This will
call for an increase in the real rate of interest to restore savings. But
the higher real rate will crowd out investment, leading to a lower stock
of capital. In this sense, a budget deficit induced by borrowing is a
burden, especially for future generations who will face a diminished
stock of capital.
At the heart of any discussions of government liabilities and defi-
cits lies the Ricardian equivalence theorem, which argues for the per-
fect substitutability of tax and debt financing under some simplifying
assumptions. In a seminal contribution, Barro (1974) argues that
intergenerational altruism is a key factor in restoring the neutrality
of fiscal policies. The present value of government expenditures
(broadly defined to include hidden contingent liabilities) must equal
the present value of tax revenues, because the government budget
constraint must hold intertemporally. It then follows that the present
value of taxes cannot change unless the present value of all expendi-
tures also changes. In short, any deficit-induced tax cut must result
in a future tax increase with the same present value. An important
implication of the Ricardian equivalence theorem is that any current
budget deficit leads to an increase in private savings, which exactly
offsets the decrease in the government's savings. This has important
implications for the role of deficits and their measurement. Under
the Ricardian view, an accurate measurement of a deficit may help
forecast future taxation policies. (For discussion, also see Barro 1979,
1989; Bean and Buiter 1987; Bernheim 1987; Bernheim and Bagwell
1988; Abel and Bernheim 1991.)
Corporations have the choice of issuing debt or equity, and the
Modigliani-Miller theorem sets the benchmark for any debate on
whether a specific combination of debt and equity is value-maximizing
from the perspective of the corporation. The government can run a
deficit or cut taxes, and the Ricardian equivalence theorem defines the
circumstances under which fiscal policy choices matter. Because de-
viations from the Ricardian equivalence result may be expected, whether
there is a budget deficit or a tax cut may no longer be a matter of
indifference.
MEASURING AND MANAGING CONTINGENT LIABILITIES 103
Effects of Government Guarantees
The volume of government contingent liabilities is of concern to pri-
vate sector lenders, the IMF, the World Bank, and future generations
of citizens. Loan guarantees, which are an example of a state contin-
gent liability, may lead to moral hazard and excessive risk-taking and
an oversupply of loans. The impact: of government guarantees on for-
eign investments was examined by Wu (1950), who argued that the
threat of expropriation and nonconvertibility of currency earnings were
the biggest risks of concern to the guaranteeing institutions. The possi-
bility that government actions may affect the allocations of resources
and the response by private sector entities has been well recognized in
the literature. Brock (1992) points out that government guarantees on
foreign loans have led to investment booms that have affected the level
and the path of the outputs in the economy whose risks are being
measured. Brock (1992) also notes that "loan guarantees may also
distort an economy's macroeconomic adjustment by permitting a post-
ponement of the liquidation process." He presents case studies from
Chile and Texas to show that sorme delays in the closure of insolvent
financial institutions have had disastrous macroeconomic consequences.
A similar argument has been advanced by Kryzanowski and Roberts
(1993) to suggest that the absence of bank runs in the Canadian bank-
ing system may not be due to the bank branching system but to "the
forbearance of regulators coupled with an implicit guarantee of all
deposits." The fact that loan guarantees and subsidies lead to actions
by entities in the private sector h,as been well articulated by econo-
mists. Insightful papers by Townsend (1977) and Salant (1983) note
how price-fixing schemes are doorned to fail and are prone to specula-
tive attack by agents in the econormy. Bardsley (1994), who examined
the collapse of Australia's Wool Reserve Price Scheme, points out that
price-fixing schemes can fail even when they are not backed by a fixed,
exogenously set financial limit as assumed by Townsend (1977). It is
clear that lenders would like to have accurate information about what
the valuations of contingent liabilities are even if the government is
able to pursue tax and other fiscal policies in the future to dramati-
cally change the future valuation of these liabilities. The point is that
any current valuation of such liabilities must be viewed within the
context of dynamic actions that can be taken by the government to
significantly affect the valuation. With this caveat in mind, we turn to
some important papers in the realm of valuing contingent liabilities.
Valuation of Loan Guarantees
The presence of contingent liabilities in the private sector and their ef-
fects have been studied in a variety of different contexts. In a pioneering
104 SURESH M. SUNDARESAN
paper, Merton (1977) investigated the effect of deposit insurance and
its valuation. His insightful analysis showed that loan guarantees are
basically put options written on the underlying assets backing the loans.
Later Merton (1978) extended his analysis to value deposit insurance
when there are costs of surveillance. In a related contribution,
Borensztein and Pennacchi (1990) examine the valuation of interest
payment guarantees on debt issued by a developing country. They ex-
ploit the market prices of debt in the secondary market to infer an
unobservable state variable whose realizations depend on whether there
will be contractual debt service or default. Borensztein and Pennacchi
assume that the stochastic process followed by the state variable is
unaffected by the presence of a guarantee.
A direct application of the insights of such papers to the measure-
ment of the contingent liabilities of government is precluded for sev-
eral reasons. First, the announcement of a guarantee produces a supply
response. For example, government loan guarantees end up increasing
the stock of debt because of the supply response. This is not modeled
in option pricing. Second, and perhaps more important, because of
this supply response, the level and the path of outputs and prices change,
leading to an endogenous shift in the underlying path of the economy.
This implies that the standard option-pricing paradigm, which assumes
an exogenous stochastic process for the underlying assets that is in-
variant to the presence of the guarantees, cannot be applied in a "cook-
book" style to valuing the government's contingent liability. The moral
hazard problems associated with price subsidies and guarantees make
the valuation much more difficult.
Summary of Insights from the Economics Literature
In summarizing the insights of the economics literature, I argue for the
following priorities in research on fiscal risk measurement in general
and the valuation of contingent liabilities in particular:
* The neutrality of fiscal policy under some conditions implies that
a dollar of debt financing is no different from a dollar raised through
taxes. Although this may not strictly hold, it forms a useful theoretical
basis for examining how one should assess a federal deficit. The abil-
ity of the government to legislate future actions that may significantly
alter the current valuations of contingent liabilities makes any isolated
valuation measure less relevant for policy debates.
* Government actions such as subsidies and loan guarantees lead
to private actions and responses that may affect the level and path of
the outputs of the economy. They may induce moral hazard-for ex-
ample, a government guarantee on debt issued by entities in the pri-
vate sector may reduce the incentive of these entities to stay current in
MEASURING AND MANAGING CONTINGENT LIABILITIES 105
their debt obligations. (But this is mitigated by the fact that private
sector issuers may care about their reputations if they are repeated
participants in credit markets.) Another possibility is that the guaran-
teeing entity is able to extract significant rents from the debtors be-
cause of the dire need to get the scarce capital. These observations
underscore the need to evaluate the costs of government actions in a
general equilibrium or a more inclusive setting than just viewing the
measurement of contingent liability as an isolated problem.
* Some government actions such as loan guarantees and price-fixing
schemes may lead to a harmful response by agents in the economy,
such as speculative attacks on buffer stocks and subsidy programs and
inefficient liquidation of insolvent firms in the presence of guarantees.
This observation leads to the possibility that the conventional mea-
sures of valuing such contingent liabilities may underestimate their
true costs.
* A corollary to this observation is that the valuation of contingent
liabilities cannot be conducted in isolation of their effects on the asset
side, as well as the liability side, of the government's balance sheet.
'rhis is particularly challenging because the present value of benefits
and other costs generated by government actions is often difficult to
quantify but nonetheless very important.
Operational Guidelines for Measuring Risk
Economic theory thus offers important guidelines in the measurement
and management of risks assumed by government's implicit and ex-
plicit guarantees and subsidies. T he focus of this literature is to pro-
vide a conceptual and a methodological basis for understanding and
reporting deficits. However, for policymakers and institutions such as
the World Bank and the IMF, which extend loans and assistance to
developing and underdeveloped economies, the issue of correctly mea-
suring the risks assumed by such economies is of practical and opera-
tional concern because it affects their lending and assistance policies.
A recent contribution by Brixi and Zlaoui (1999) and Chapter 3 by
Allen Schick and Chapter 9 by 1-iana Polackova Brixi, Allen Schick,
and Leila Zlaoui in this volume have explored risk measurement is-
sues applied to specific countries as well as to the general issue of
contingent and hidden government liabilities. The key contributions
in this strand of work can be surnmarized as follows:
* Practical measures of liabilities can be put in the context of a
Fiscal Risk Matrix (see Polackova 1998 and Chapter 1 of this volume).
Four types of risks are recognizecd in this framework: (a) direct explicit
liabilities, (b) direct implicit liabilities, (c) explicit contingent liabili-
ties, and (d) implicit contingent liabilities. Direct explicit liabilities
106 SURESH M. SUNDARESAN
are government obligations such as government employee wages that
are clearly identifiable and fall under the rubric of the national legal
framework. Direct implicit liabilities are future obligations of gov-
ernment that do not necessarily constitute legal obligations but rather
moral obligations. Explicit contingent liabilities cover state guaran-
tees on loans issued by nongovernment institutions. Finally, implicit
contingent liabilities include hurricane damage, earthquake damage,
and so forth.
a Specific proposals for budgetary controls as a means of control-
ling and managing fiscal risk include: (a) improved reporting of the
actual performance of programs under the state guarantees; (b) pro-
viding the right incentives by requiring significant risk-sharing-this is
similar to reinsurance contracts in which the reinsurer requires the
insurance company to take the first layer of damages before insuring a
prespecified second layer; and (c) modifying and supplanting the ac-
crual-based accounting procedures to shed light on hidden liabilities.
* Risk management tools that are already prevalent in the private
sector can be applied to the public sector, such as prudent provision-
ing, stress testing, and organizational allocation of risk management
responsibilities.
The challenge that remains in risk management is to improve the
practical task of risk reporting by more completely tapping into the
insights of economic theory that were summarized earlier. This chap-
ter attempts to do precisely that.
Institutional Framework for
Risk Measurement in the Private Sector
One place to look for some guidance on risk measurement and man-
agement is the private sector. Any firm wishing to actively manage risk
has to pass the hurdle that firms need not manage the risks that are
easily diversified by stockholders. The basis for this hurdle arises from
capital market theory and asset-pricing models in the finance litera-
ture. Unless active risk management results in a higher value to the
stockholders, there is no need for a firm to manage risk. Typically, two
factors make risk management by a firm value-maximizing. The first
is the cost of financial distress: if the firm's risk of financial distress is
increased as a consequence of not hedging market risks and the result-
ing process of financial reorganization or liquidation is costly, then
risk management adds value to the firm. Second is the underinvestment
problem: if the benefits of a project largely accrue to the bondholders,
then the firm may underinvest. This situation can be mitigated by hedg-
MEASURING AND MANAGING CONTINGENT LIABILITIES 107
ing under some circumstances. There also may be tax and regulatory-
related incentives for firms to hedge risk.
A fundamental aspect of corporate organization is that there are
markets in which the claims issued by corporations are actively traded.
Such markets include equity markets such as the New York Stock Ex-
change as well as bond markets in both exchanges and the dealer com-
munity. These claims are exchanged by sets of investors many times in
any day. For this process to operate efficiently, considerable informa-
tion has to be produced daily about the risks assumed by corporations.
Thus corporations operate in an economic environment in which in-
formation is produced by many independent outside groups. In well-
developed economies, the following organizations produce information
about corporate risks.
Stock analysts follow the earnings potential of corporations. For
big corporations such as Microsoft, thousands of analysts assess the
economic future of the company and produce earning forecasts and
generate recommendations about whether the stock price of the com-
pany (which summarizes the risk-return tradeoff to the stockholders)
is "fair."
Credit rating agencies such as Moody's and Standard and Poor's
analyze the economic environment faced by the companies that issue
debt capital. Such agencies often place some faltering companies un-
der "credit watch" and downgrade companies whose economic stand-
ing has deteriorated or is in danger of deteriorating.
Commercial banks monitor the activities of the borrowing corpo-
rations actively. Bank loans are typically senior and secured, and
banks attempt to anticipate any economic difficulty that the bor-
rower may face.
External auditors certify the financial standing of the borrowing
corporation and provide credible reports on the risks faced by the com-
pany. By enforcing high accounting and reporting standards, auditing
firms are able to issue financial statements (including statements about
off-balance sheet liabilities) that provide useful guidelines to stake-
holders in the company. Although this is generally the case, from time
to time auditors appear to perforn their functions poorly.
Finally, external markets, which fall into three categories, apply
here. The first category of markets is those in which the claims issued
by corporations are traded. Perhaps the most important source of in-
formation about the risks taken by corporations is these markets them-
selves. Equity and bond prices respond quickly in an efficient market
to "news" about the riskiness of companies. The second category of
markets is those for corporate control. A fall in equity prices may
reflect in part a company's declining economic fortunes. In such circum-
stances, the company may becomne the target of a takeover attempt,
108 SURESH M. SUNDARESAN
hostile or otherwise, in markets for corporate control. The possibility
that such a takeover might result in a reorganization of the company
gives the right incentives to senior managers of the company to seek
the right balance between risk and returns. Finally, there are markets
in which corporations can participate to manage their exposure. Ex-
amples are those for securitization, derivatives securities, and credit
derivatives.
The private sector therefore has several institutions and markets
that produce, disseminate, and update information about the risks taken
by corporations. Within this rubric of institutional factors, the task of
articulating the risks taken by corporations boils down to a determi-
nation of which risks the corporation decides to bear (based on its core
competence) and which risks it can either hedge or parcel out to play-
ers in the capital markets. The actual reporting of risk takes the form
of quarterly earnings reports and annual audited financial statements,
which provide stakeholders with much information. Independent re-
ports by stock analysts and credit rating agencies also provide a wealth
of information. Finally and perhaps most important, markets provide
almost continual information on the risks of companies.
The risk management tools used in the private sector include the
following:
* Measuring market risk. This category includes the methods used
to assess the exposure of a company to interest rates, foreign curren-
cies, commodity prices, and macroeconomic factors. The concept of
value at risk (VAR) is becoming a popular way to measure and report
the risk of financial institutions and even nonfinancial entities.
* Measuring and provisioning for credit risk. The credit risk of a
borrowing company is summarized in its credit rating reports, the prices
of its loans and bonds, and its stock prices. The financial statements of
the company also report its credit exposure.
* Stress testing. Subjecting the borrowing company to stress testing
by simulating extreme economic environments will reveal how well it
might perform under adverse circumstances.
* Contractual tools: marking position to markets, requiring exten-
sive credit screenings, establishing margins, and arranging for collat-
eral requirements. By requiring that swaps be marked to market on a
daily basis, swap dealers reduce their credit exposure to daily losses.
In the absence of marking to market, such losses may accumulate to a
level that may threaten the survival of one of the counterparties. Of-
ten, contractual provisions stipulate that positions must be settled in
cash if one of the counterparties is downgraded by a rating agency.
Such provisions try to internalize in contractual provisions future in-
creases in credit risk.
MIEASURING AND MANAGING CONTINGENT LIABILITIES 109
Institutional Framework for Governments
In sharp contrast to the private sector, governments have far fewer
independent institutions and markets that monitor and produce infor-
mation about the risk and contingent liabilities assumed by the gov-
ernment. This section begins by identifying the institutions that
articulate the risks of government and contrasting them with those in
the private sector.
Because governments do not issue stock, there are no stock analysts
who produce information about the risks assumed by them. The ab-
sence of this source of information is a major difference between pri-
vate sector entities and governments.
Governments do borrow money in public bond markets. The debt
issues of government are evaluated by the credit rating agencies, such
as Moody's and Standard and Poor's, which analyze the economic and
political environment faced by the governments. This can be a poten-
tially useful source of information. Over the last decade sovereign bond
markets have grown at a much more rapid rate than sovereign loan
markets. This means that there are market rates that reflect informa-
tion about the risks of countries.
Commercial banks and bank syndicates extend loans to countries,
as do organizations such as the ]:MF and World Bank. These institu-
tions monitor the activities of the borrowing countries actively and
produce a wealth of information on the risks of governments.
As for external markets, the category of markets that contains in-
formation about government risk is limited to the sovereign bond and
loan markets. Governments do have access to derivatives markets, and
they use them to manage some of their market exposure.
Broadly, the risks facing a government arise from the political and
economic environments. Such risks are present whether the govern-
ment in question represents a developed economy such as the United
States or a developing economy such as India. The risks that arise
from the political climate facing the country include the possibility of
war or dealing with large-scale immigration arising from instability in
neighboring regions or natural dlisasters. Such risks inevitably create
fiscal problems of considerable significance. Risks that arise from acts
of God such as earthquakes or hurricanes also may contribute to ma-
jor fiscal strains. The rest of this chapter will explore risk management
tools that comprise or address the following:
a The institutional framework for risk management issues that
deal with moral hazard issues--that is, the accountability of govern-
ment actions-and the legal framework for and enforcement of con-
tracts. Any risk management system is dependent on the underlying
110 SURESH M. SUNDARESAN
institutional framework for its efficient implementation. This chapter
will specify the institutional framework needed to effectively measure
and manage government risk.
* The contingent liabilities of government. Examples of
government's many contingent liabilities include various guarantees,
pension liabilities, indexed wage contracts, and health insurance. Us-
ing some of the existing framework for classifying these liabilities, this
chapter will highlight the risk measurement problem in each category.
* The unique risks of government: lender of last resort, disaster
relief, public health, famines, financing the growth of impoverished
sections, and real options in the economy. An example of real options
is the tremendous growth and foreign currency earning possibilities
that arise from information technology for a developing country. Sub-
sidies such as tax exemption may accelerate use of the Internet in re-
mote villages, which can form the basis for promoting primary
education and eventually eradicating illiteracy.
This discussion is prefaced, however, with a brief review of the
methods by which corporations measure and manage risks. This is a
useful starting point, because many of the risk management tools and
techniques have been developed in the context of corporations. The
chapter then articulates the differences between the contingent obliga-
tions and fiscal risks of corporations and governments. This allows a
focus on the key differences between corporate and government risk
management practices. The institutional framework that underlies risk
management in the corporate sector will be contrasted with the mecha-
nisms in place for monitoring the risks of government.
Risk Management Tools in the Private Sector
There are two distinct approaches to managing and measuring risks in
the private sector. The first approach is to identify those risks that the
organization believes it is particularly good at managing and earning
an attractive return on in the process. Such risks are fully borne by the
firm. Risks of the firm that are more efficiently borne by the markets
are securitized and parceled out where such markets are well devel-
oped. In their absence, such risks are laid off in listed or dealer deriva-
tives markets such as bond futures contracts or swaps. An example
should serve to illustrate this point. Mortgage companies extend
fixed-rate mortgages (FRMs) and adjustable-rate mortgages (ARMs).
To fund such a loan portfolio, such firms issue liabilities that typically
are short term in nature. Some of the liabilities may be noncontingent;
others may be contingent. Let us first focus on a simple noncontingent
liability in the context of the simple balance sheet shown below.
MEASURING AND MANAGING CO:NTINGENT LIABILITIES 1ll
Assets Liabilities
Fixed-rate mortgages (FRMs), Six-month certificates
6 percent, 30-year, $500 million of deposit, $500 million
principal amount
Adjustable-rate mortgages (ARMs) One-year certificates
indexed to one-year Treasury bill of deposit, $500 million
yield, $500 million principal amount
Consideration of the risk assumed by the mortgage company re-
quires a look at the exposure of its liabilities in relation to that of its
assets. If the company decides to hold both FRMs and ARMs in its
portfolio, then its risk picture turns out to be quite different: FRMs
have a long maturity in relation to the short-term certificates of de-
posit (CDs) that were issued by the company to help fund the FRM
portfolio. In other words, there is a maturity gap (or a duration gap).
T he precise nature of the gap depends on the risk of prepayments, but
that is not the principal concern now. On the other hand, the ARMs
are reset every year, and their market risk is closer to the market risk of
the liability that was issued to create the ARM portfolio. Hypotheti-
cally, assume that the gap measure is zero for the ARM portfolio. In
this illustration, the mortgage company may decide to securitize the
FRM portfolio, sell it in the mortgage-backed securities (MBS) mar-
ket, and use the proceeds to retire the liabilities and keep the rest as
profits. It may decide to maintain the ARM portfolio in which the
market risk is relatively low; the remaining exposure is the credit risk
that the company may believe it is in a better position to manage and
earn an attractive return on. In such a situation, the risk management
and reporting boil down to two actions. First, securitize and dispose of
the risk that the market is able to bear more effectively. This reduces
the size of the balance sheet and puts a lesser burden on the regulatory
capital requirements. Second, manage the risk of the remaining liabili-
ties in relation to the remaining assets on the balance sheet to earn a
better return.
Although here the balance sheet is illustrated with direct (non-
contingent) liabilities, the same argument applies to contingent liabili-
ties. A government agency such as the Government National Mortgage
Association (GNMA or Ginnie Mae) in the United States securitizes
pools of mortgage loans and sells them to institutional investors. But
at any time it may have a "pipeline" of mortgage loans that are yet to
be securitized. On this pipeline, Ginnie Mae is exposed to serious risk
of prepayments. By issuing callable bonds (a contingent liability), Ginnie
Mae can attempt to align the prepayment risk of its pipeline assets-
that is, if interest rates fall and there is a prepayment, then Ginnie Mae
112 SURESH M. SUNDARESAN
can use the cash flows from prepayment to call the liability back. To
report the value of this contingent liability without simultaneously
recognizing the asset it is hedging would be a mistake. Of course, in
these examples the market values of contingent liabilities and the as-
sets they help generate are easier to measure than in the case of a
government. Herein lies the challenge of government's fiscal risk
measurement.
Regulatory agencies also have a major interest in making sure that
financial institutions have a prudent risk management practice. In the
United States, institutions such as banks are under the guarantee of the
Federal Deposit Insurance Corporation (FDIC), and the government
has a major interest in ensuring that these institutions remain solvent.
In other instances, the government is interested in ensuring that the
risks are managed in a way so that in the event of a "financial conta-
gion," markets and institutions do not collapse. With these in mind,
central banks have coordinated mechanisms for managing market risk,
credit risk, operational risk, and liquidity risk in the world's financial
markets.
To summarize, the risk management actions that can be taken by
firms in the private sector can be placed in distinct categories:
* Customize the assets and liabilities so that the overall market
exposure is kept at a minimum.
* Securitize risks through the use of special-purpose vehicles (SPVs)
and sell the cash flows from the assets in the SPV to institutional inves-
tors. This will be done by firms that believe that the cost of securitization
(which requires credit enhancements, liquidity enhancements, and other
legal expenses) is less than the benefits that arise from securitization.
Such benefits may include regulatory capital relief from a reduced bal-
ance sheet, an ability to manage the firm better because of the reduced
balance sheet, and an increased ability to focus on managing those
assets in which the firm believes it has a comparative advantage.
* Use derivative markets to manage risk when appropriate. There
are two types of markets in which firms hedge their risks. The first is
listed markets, where derivative securities such as options and futures
contracts are traded in open outcry markets. Examples of such mar-
kets in the United States include the Chicago Board of Trade and the
Chicago Options Exchange. Listed markets such as these tend to be
standardized and offer extensive liquidity. Some Central American and
Latin American countries hedge their external floating dollar debt by
trading in the eurodollar futures contracts at the Chicago Mercantile
Exchange, for example. Second, firms also use dealer markets to man-
age their risks. Dealer products tend to be more customized and illiq-
uid. Examples of dealer derivatives include interest rate swaps, foreign
currency swaps, and interest rate caps.
MEASURING AND MANAGING CONTINGENT LIABILITIES 113
* Engage in prudent provisioning for "credit events" that may hap-
pen to counterparties in the future. Besides provisioning capital for
future contingencies, firms may also enter into contractual safeguards.
Contractual provisions may include collateral requirements, marking
positions to market periodically and on a contingent basis, and
optionality to terminate or renegotiate if the credit rating of the
counterparties is in jeopardy.
* Explore the growing credit derivatives market. In this market
firms can obtain full or partial insurance for specified credit events.
Credit default swaps, credit-linked notes, collateralized loan obliga-
tions (CLOs), and collateralized bond obligations (CBOs) are examples
of such derivatives.
Formulating Risk Management
Practices for Governments
In formulating the risk management practices of government, it is im-
portant to first determine the incentives that will encourage the pri-
vate sector to assume the risks that the government is ill-equipped to
bear. In addition, the government should attempt to mitigate any sig-
naling costs that may inhibit optimal provision of liquidity. For ex-
ample, as is often the case, suppose that the central bank, as the lender
of last resort, provides credit under the discount window. But the bor-
rowings under the discount window may fall because of the percep-
tion that other institutions regarcl discount window borrowing to be a
signal of liquidity or credit problems of some significance. Govern-
ments can design mechanisms that may result in a better allocation of
risks. A few examples will illustrate these points. In all these actions,
there are costs and benefits that have to be articulated before such
actions are taken.
Often, governments have to face the risks associated with acts of
God such as earthquakes or hurricanes. Currently, such risks are in-
sured by insurance companies and reinsurance firms. By helping to
create a market for the insurance of such events and securitization of
such risks, government is able to transfer at least part of that risk to
the capital markets. The development of property and casualty in-
surance companies, reinsurance companies, catastrophe-based futures
contracts, bonds, and so forth in certain markets does in fact suggest
that such risks can be managed in the private sector to some degree.
The role of such markets has been discussed in Braun, Todd, and
Wallace (1998). The growth of catastrophe-linked markets also has
complemented the reinsurance markets and helped to change the in-
centives in the reinsurance markets, which are widely perceived to be
noncompetitive.
114 SURESH M. SUNDARESAN
By issuing guarantees and by supporting loan programs through
direct borrowings, the federal government has improved the flow of
credit into the housing sector. The development of mortgage-backed
securities markets where the risk of prepayments is parceled out to
institutional investors such as pension funds and insurance companies
has freed the commercial banks from their exposure to this risk. This
is an example of a government policy in which the risks are reallo-
cated in the economy so that institutions bear only those risks in which
they perceive a comparative advantage. On the other hand, such deci-
sions may prove to be costly to the government. In the 1980s, when
the interest rates shot up and the yield curve became inverted, some
U.S. federal agencies such as the Federal National Mortgage Associa-
tion (FNMA, or Fannie Mae) had to be rescued by the government
through "regulatory forbearance." In the last decade, agencies such as
Fannie Mae have grown so rapidly and have leveraged their equity
capital so much that the risk exposure implicitly facing the govern-
ment could be quite substantial in the event of a failure. Moreover,
given the fact that agencies such as Fannie Mae are privately owned
companies held by stockholders, it is reasonable for the government to
reevaluate whether explicit and implicit subsidies that are currently in
place should be continued. For example, Fannie Mae does not pay
taxes, it has a direct line of credit with the Treasury, and its securities
are exempt from certain regulatory restrictions that make them more
attractive to institutional investors.
In response to the potential demand for liquidity during the Y2K
period, the central bank of the United States came up with a proactive
plan rooted in options pricing. As the lender of last resort, the central
bank has a responsibility to extend credit to vulnerable sectors of the
economy, but by selling liquidity options the central bank signaled
ahead of time that it was ready to extend credit should there be a crisis
related to Y2K. Such an action has several effects. First, it signals that
the central bank is ready to extend credit. Second, by inviting sealed
bids, it protects the identity of the potential buyer of the liquidity op-
tions. This is important because there is increasing evidence that bor-
rowing in the "discount window" has reputational costs for the
borrower. By specifying the securities that are acceptable as collateral
in the options contract, the Federal Reserve Bank lets investors know
the terms under which it is offering liquidity.
In addition, the government can follow the practice of the private
sector in providing more transparent accounting and reporting of its
balance sheet where there is a concerted effort to identify and measure
all liabilities (both contingent and noncontingent) and relate them to
the asset side of its balance sheet. For example, the guarantees ex-
tended by the government to help develop a securitized market for
MEASURING AND MANAGING CONTINGENT LIABILITIES 115
mortgages should show up as a liability. But so must the present value
of the benefits that have accrued to the taxpayers as a result of the
growth of these markets. The cost of obtaining credit for housing might
be considerably higher in the absence of the development of
mortgage-backed securities markets that grew because of the govern-
ment guarantees. This should be stressed along with the cost of gov-
ernment guarantees.
A Framework for 'Valuing Guarantees
A model for valuing government guarantees is sketched out in the
Appendix to this chapter. This section stresses the intuition behind the
general approach. Figure 4.1 provides the demand and supply curves
for an underlying product or service in the absence of a guarantee. The
equilibrium price level is P = 27. Once the government introduces a
guarantee to the suppliers, more firms may enter the market, leading
to an increased supply and possibly an increased level of risk because
some weaker firms may attempt tc take advantage of the guarantee to
enter the market. Figure 4.2 shows this effect.
As noted, the Appendix to this chapter outlines a model in which
one can examine quantitatively the effect of a price guarantee on the
vvelfare of the economy and the allocation.
Figure 4.1. Equilibrium Price
Demand/supply
120
0oo - Demand
80 - ~~~~~~~~~~~~~supply
60 -
40
20 -
20
O I I I I I I I I I I I I -I I I I L . .I .I. .
0 10 20 30 40 50 60 70
Price
Source: The author.
116 SURESH M. SUNDARESAN
Figure 4.2. Equilibrium Price
Demand/supply
120
100o Demand
100 - Demand Supply (under guarantee)
80 -
40Supply (nooguarantees))
40-
20 -
O 1 r_I I I L I I I L_ I I I I I I I Ij
0 10 20 30 40 50 60 70
Price
Source: The author.
Conclusion
This chapter has presented an overview of some risk management
practices in the private sector and the extent to which they may be
useful for managing the fiscal risk of government. Problems that are
unique to the risk management in government were identified. They
included the potential supply responses to government guarantees
and the ability of government to influence significantly the value of
any guarantee by its future actions. In recent times, some innovative
approaches have been used by governments to manage risk. The use
of liquidity options by the central bank of the United States, the se-
lective use of guarantees to promote the development of private mar-
kets for securitization, and incentives for the development of risk
insurance markets are some examples of innovations in which gov-
ernment has helped develop markets where risks are parceled out to
investors who are willing to bear them.
Annex 4.1. A Model of Valuing Government Guarantees
A standard model of valuing contingent liabilities views it as a put
option. Consider a simple closed economy in which the price of the
output follows an exogenous process:
MEASURING AND MANAGING CONTINGENT LIABILITIES 117
dP/P = (xdt + aFdz
where cx and a are positive scalars and {z, t > 01 is a Brownian motion
process describing the uncertain evolution of the price of the good in
the absence of any government contingent liability. Assume that the
supply curve in the economy is elastic to the price of the good, so that
the supply S is described as the constant elasticity of the supply curve
in
S aPb
where
dS
b= S
dP
p
is the elasticity of the supply with respect to the output price. If b = 0,
then S = a and is inelastic. In this economy, with no price guarantees,
one can characterize the welfare of the consumer by deriving the value
function and the optimal consurnption (demand). What happens to
this stylized economy when the government introduces a subsidy or a
guarantee to the price level? There will be a shift in the optimal con-
sumption (demand) and in the value function of the consumer. Study-
ing these questions requires modeling two things. First, identify the
supply response to the guarantees. Second, identify how such supply
responses affect the equilibrium price dynamics through the optimal
demand of agents in the economy. To study these questions formally,
let us investigate the optimal allocation in this economy and measure
the value to the economy in the absence of any contingent liability.
This will serve as a benchmark ifor evaluating the effect of such a li-
ability on the price process through (a) a supply response and (b) opti-
mal demand allocation.
Bencbmark Allocations
Let the economy be described by a representative agent who maxi-
mizes the expected lifetime utility of consumption of the good as
Max,E 1[ eFes u(cs)ds]
The wealth of the consumer who also holds all the stock of the
goods supplied is simply
W =.PS = aPb, I
118 SURESH M. SUNDARESAN
The dynamics of the wealth process is
dW = W [(b + l)ax + I b(b + 1)a2]dt - Pcdt + W(b + 1)adz.
2
The optimal consumption problem of the consumer may be specified
in terms of the Hamilton-Jacobi-Bellman (HJB) equation, where J is the
value function associated with the optimal consumption allocation:
0 u(c) - PJ + Jw ((b + 1)aW + ½12b(b + 1)c22W - Pc) +
| /2jww(b + 1)2W2o2+Jp cP + 1/JppP2 +J WP(b +
The first order condition for oprimality is
us = PJW
This problem cannot be solved in closed form in general unless ad-
ditional restrictions are placed on the problem. To get concrete results,
assume that the utility function is of the constant relative risk aversion
(CRRA) type, so that
u(c) =-
y
For this problem one can solve explicitly for the optimal consump-
tion policy and the value function of the economy. This is stated as a
proposition below.
The optimal consumption (demand) is linear in wealth and nonlin-
ear in the price level
c= [ho4]y- w
where 4D is a function of the price. The function 4 is found by solving
the partial differential equation
0= r - p[y<>] + [yFD]((b + 1)x + ½b2b(b + p)2-yPLY]' )
+ 1/2[yD](b + 1)2a2 + [F)pcP] + I/2DPPo2P2 + D>PYP(b + J)G2
The value function of the economy is
J = W¢I(P).
The function (D(P) is given by
m/ P) = a,p-r
MEASURING AND MANAGING CONTINGENT LIABILITIES 119
and
1 [ p yba - 2b(b + 1)ya2 d (y + 1)yyc2 + y2a2(b + 1)]
Note that combining the expressions for J and 'D(P) yields
J=a[ ] .
Allocation with a Guarantee
Consider a simple guarantee scheme that is specified as follows: the
government will pay the amount P - P whenever P < P. In such a
situation, there is the following supply and consumption response. The
supply function will be
S =aP for P> P
and
S = aP b for P < P.
This supply response is consistent with the expectations of pro-
ducers that the government will compensate them in states of the
world where the actual price of the good is less than the guaranteed
price. As the price goes below the guaranteed level, the supply stays
fixed and the only uncertainty that faces the consumer is the wealth
level. Corresponding to these two regions, the wealth dynamics evolves
as follows:
dW= W[(b + 1)a + ½12b(b + 1)a2]a!t- Pcdt + W(b + 1)adz, when P > P
and
dW = Wadt - Pcdt + Wadz, when P < P.
The optimization problem facing the consumer now can be stated as
[ u(c) - pJ + Jw ((b + 1)(xW + ½12b(b + 1)a2W - Pc) + Jp aPl
0 ax + ½/2JWW(b + 1)2W2a2+ ½/2JPPo2P2 +JpwWP(b + 1)a2]'
when P > P and
0 = Maxc [u(c) - pJ + Jw (aW - Pc) + i/2½WWW2G2], when P < P.
120 SURESH M. SUNDARESAN
It turns out that the value function can be written as J = WYTI(P)
and ] = a2WY.
Enforcing the conditions of value matching and smooth pasting
conditions yields
J (P, W) = J(P) =* ¢(P) = a2
and
JW(P, W) = Jw(P) > 'P(P) = a2.
This framework, using the value function and the optimal consump-
tion rule, provides a guideline for a numerical solution.
References
Abel, Andrew, and Douglas Bernheim. 1991. "Fiscal Policy with Impure
Intergenerational Altruism." Econometrica 59: 1687-1711.
Bardsley, Peter. 1994. "The Collapse of the Australian Wool Reserve Price
Schemes." Economic Journal 104: 1087-1105.
Barro, Robert, J. 1974. "Are Government Bonds Net Wealth?" Journal of Po-
litical Economy 82: 1095-1117.
- 1979. "On the Determination of the Public Debt." Journal of Political
Economy 87: 940-71.
- 1989. "The Ricardian Approach to Budget Deficits." Journal of Eco-
nomic Perspectives 3 (spring): 37-54.
Bean, Charles, and William Buiter. 1987. "The Plain Man's Guide to Fiscal and
Financial Policy." Employment Institute, London, October.
Bernheim, Douglas. 1987. "Ricardian Equivalence: An Evaluation of Theory
and Evidence." In Stanley Fischer, ed., NBER Macroeconomics Annual (Na-
tional Bureau of Economic Research). Cambridge: MIT Press.
Bernheim, Douglas, and Kyle Bagwell. 1988. "Is Everything Neutral?" Journal
of Political Economy 96: 308-38.
Blejer, Mario, and Adrienne Cheasty. 1991. "The Measurement of Fiscal Defi-
cits: Analytical and Methodological Issues." Journal of Economic Literature
29: 1644-78.
Borensztein, E., and G. Pennacchi. 1990. "Valuation of Interest Payment Guar-
antees on Developing Country Debt." IMF Staff Papers 37 (December): 806-
24, International Monetary Fund, Washington, D.C.
Braun, Anton, Richard Todd, and Neil Wallace. 1998. "The Role of
Damage-Contingent Contracts in Allocating the Risks of Natural Catastro-
phes." Research Department, Federal Reserve Bank of Minneapolis, April.
Brixi, Hana Polackova, and Leila Zlaoui. 1999. "Managing Fiscal Risk in
Bulgaria," Draft paper. World Bank, Washington, D.C., November.
MEASURING AND MANAGING CONTINGENT LIABILITIES 121
Brock, Philip. 1992. "External Shocks and Financial Collapse: Foreign Loan
Guarantees and Intertemporal Substitution of Investment in Texas and
Chile." American Economic Review 82 (2): 168-73.
Buiter, Willem. 1983. "Measurement of the Public Sector Deficit and Its Im-
plications for Policy Evaluation and Design." IMF Staff Papers (June):
306-49, International Monetary lFund, Washington, D.C.
Eisner, Robert. 1984. "Which Budget Deficit? Some Issues of Measurement and
Their Implications." American Economic Review 74 (2): 138-43.
Eisner, Robert, and Paul Pieper. 1984. "A New View of the Federal Debt and
Budget Deficits." American Economic Review 74 (2): 11-29.
Kryzanowski, L., and G. S. Roberts. 1993. "Canadian Banking Solvency."
Journal of Money Credit and Banking 25 (3): 361-76.
Merton, Robert C. 1977. "An Analytical Derivation of the Cost of Deposit
Insurance and Loan Guarantees." lournal of Banking and Finance 1: 3-11.
- 1978. "On the Cost of Deposit Insurance When There Are Surveillance
Costs." Journal of Business 51 (July): 439-52.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Salant, S. W 1983. "The Vulnerability of Price Stabilization Schemes to Specu-
lative Attack." Journal of Political Economy 91: 1-38.
Townsend, R. M. 1977. "The Eventual Failure of Price Fixing Schemes." Jour-
nal of Economic Theory 14: 190-9.
Wu, Yuan-Li. 1950. "Government Guarantees and Private Foreign Investment."
American Economic Review 40 (1): 61-73.
CHAPTER 5
Analytical Techniques Applicable
to Government Management
of Fiscal Risk
Krishna Ramaswamy
The Wharton School, University of Pennsylvania
INSTITUTIONS, CONTRACTS BETWEEN institutions, and entire market-
places have developed to facilitate the transfer of risks from those un-
willing or unable to bear them to those who are both willing and able
to do so.
A privately owned corporation typically has a portfolio of projects
or activities, each of which contributes a random amount per period
to its overall profits. It will, in general, finance these projects with
forms of debt or equity capital or with both. In deciding between alter-
native investment projects, the privately owned corporation looks to
add value, net of the cost of the investment. The computation of that
value involves discounting the risky stream of cash flows from the
project at a rate that is adjusted to account for the risk of the project.
In deciding between alternative ways of financing its portfolio of
projects (for example, with short-term or long-term debt, or convert-
ible debt, or preferred stock), the corporation seeks to maximize the
value of its shareholder ownership. The shareholders have limited li-
ability-a feature that permits shareholder-appointed managers to take
greater risks in search of profitable opportunities. The celebrated
]\Vodigliani-Miller theorem offers assurance that, under certain condi-
tions, the total value of the projects undertaken is unaffected by the
way in which the firm's activities are financed. This result ensures that
the manner in which the risk of the aggregate cash flows from all the
projects is partitioned does not affect the firm's current market value.
123
124 KRISHNA RAMASWAMY
Absent bankruptcy costs, the use of derivatives to "hedge" the cash
flows from the activities does not in itself add value.
A government, or state-owned enterprise, also has a portfolio of
activities, although the scale of some of these may be dictated and thus
better recognized as commitments. These activities will most often be
paid for with general tax revenues, and in some cases, the entity might
float a public debt issue with a guarantee from another source. It is in
comparison with its counterpart in the private sector that several im-
portant differences emerge. First, the choice between alternative ac-
tivities as expenditures or "investments" is generally not made
actively-rather, it is set exogenously by law or by a political process.
Second, where the private corporation can avail itself of market prices
in imputing (current) values to lines of activity or even entire divi-
sions, the state-owned entity cannot perform valuations of equal reli-
ability. Indeed, the notion of "valuing" departments or activities within
a government budget-by computing the present value of future net
outlays or receipts, thereby acknowledging the intertemporal nature
of these activities-runs at odds with the conventional cash flow-based
budgeting that forms the basis of decisions in the public sector. Third,
where the private corporation's shareholders can seek protection from
its creditors, the public entity cannot avoid its explicit and implicit
commitments in the event of a crisis. Here I have highlighted the three
major differences that an academic finance theorist would see; readers
with experience in these contexts will recognize other differences.
The purpose of this chapter is to describe the analytical techniques
that can reasonably apply to the measurement of risks within a fiscal
budgeting context and to discuss, in conceptual terms, the analytical
methods that enable a decisionmaker to analyze them.
What are the main analytical tools for measuring fiscal risk that
can be brought to bear from financial theory? A popular paradigm in
finance employs mean-variance analysis, in which undesirable fluc-
tuations in wealth or cash flow, as measured statistically by the vari-
ance (or equivalently by the standard deviation), are balanced against
the desirable characteristic of the mean or expected value. Researchers
have developed models of risk measurement in this context-known
as factor models-in which the randomness in the fluctuations in the
value of an investment is related to some basic economic variables
such as the gross national product (GNP) or industrial production.
This class of models enables computation of the risk (standard devia-
tion) of an asset or a cash flow as it derives from more basic and
fundamental influences; it also enables identification of those influ-
ences that are more important than others. This chapter explores an
application of factor models within a mean-variance framework and
applies it to a budgeting context in which random (cash flow) expen-
ANALYTICAL TECHNIQUES 125
clitures from several distinct units (departments or subnationals, say)
create the fiscal risk for a national body.
What are the main tools for controlling fiscal risk? This question is
riot easily answered for the following reasons. First, fiscal risk is an
outcome of commitments that are either explicit or implicit, and the
immediate control measure, to scale back these commitments or to
"deleverage" them, may not be a realistic alternative in many cases.
Second, the normal prescription-to purchase insurance and effect the
transfer of the risk to those who a:re willing and able to bear the risk-
may require the use of markets that are not yet developed and that in
many cases remain closed for fundamental reasons. That said, forms
of risk transfer in the public sector, such as loan guarantees and insur-
ance coverage, do constitute valicl methods.
For the application of the relevant analytical methods (such as op-
tion pricing), some conditions mlust be met.' In order to provide a
useful treatment, I will
* Stay (as much as an academic can) within the practical context in
which most government decisionmakers find themselves.
* Take the basic situation to be one in which a parent government
unit (a federal or national authority) oversees the budgets of several
state-owned enterprises, or subnational units, or even departments.
* Take as given a common period for which the parent and sub-
units are assessing the fiscal risks.
* Take the typical budget process to be one in which each subunit
prepares its budget for the future time period with a common set of
forecasts-of, for example, the expected rate of growth in per capita
income, an average exchange rate or interest rate, or expected fuel
prices-that are determined by the parent government unit or that are
agreed on in preparing the budget.
* Assume that the parent government takes up the net deficits that
occur at the level of the subunits. This means that surpluses and defi-
cits at the subunit levels are passed through to the parent unit.
It is possible to relax most of these assumptions and employ the
ideas in other contexts as well. The final assumption, however, is more
critical. I have taken as a working assumption (because it simplifies
the analysis considerably) that the parent government receives the sur-
pluses and is responsible for the deficits. Such a symmetrical pass-
through of surpluses and deficits rnight make the analysis more suitable
for a parent unit that oversees several subunits or departments that
have little or no autonomy. In the conclusion of this chapter, I have a
few remarks about the one-sided case in which only the (gross) aggre-
gate deficits (perhaps exceeding some level) are borne by the parent
unit and what methods can be applied in this case.
126 KRISHNA RAMASWAMY
Risk Measurement in a Fiscal Context
Contributors to this book (see Chapter 1 by Brixi and Mody and Chapter
18 by Ma) and others (see, for example, Polackova 1998 and Lewis
and Mody 1997) have discussed the classifications of risk in a fiscal
context and recommendations about its measurement and control. In
this section, I add to this practical discussion by describing the analyti-
cal tools that can be applied from the field of finance. To do so, I
introduce some symbols to simplify the nature of the analysis, but I
surround them with their definitions and explain in words the impor-
tant concept to be retained.
It is useful discipline for every government unit to recognize all its
activities and departments nominally as sources and users of funds.
Because some units may be both, it is recognized that in making a
quantitative tally one might equally rely on the net source of funds to
describe a department's contribution to the overall fiscal surplus or
deficit. Within a government unit i, I denote the net cash flow amounts
in a budget period t as (ft, which can be thought of as the period's net
expenditure. I treat positive numbers as a use of funds (as is typical of
most expenditure-oriented activities) and negative numbers as a source
of funds (as would be normal for a surplus unit or a revenue author-
ity). The squiggle above the letter C serves as a reminder that the
projected cash flow amount is a random variable. Like any random
quantity, it has an anticipated component-the expected value-which
is almost always the basis for the departmental budget allocation. In
what follows I use the terms anticipated budget or expected cash flow
interchangeably to apply to the expected value, denoted by E(C,).
The realized cash flow in period t can take values that are higher or
lower than expected, and the unanticipated component is denoted as
u,, and is the difference between the realized and expected values:
,= C,, - E(C1)
Note that ex ante-at the beginning of period t-the value of u;, is
random, but when it is revealed that the i-th activity has a larger (or
smaller) deficit than had been budgeted, ii,, is positive and is an un-
pleasant (or negative and a pleasant) surprise, ex post.
I characterize the fluctuations in the unanticipated shocks iit as the
fiscal risk contributed by the i-th government unit. By its definition,
the average value of these shocks is zero.
An individual unit might further recognize that there are some domi-
nant sources that drive the fluctuations in the values of u;t. For ex-
ample, the costs of operating a network of schools will involve energy
use and related transportation costs, where both the amount used and
its price are weather-dependent and outside the control of the oversee-
ing department. In other contexts, the demand for government relief
ANALYTICAL TECHNIQUES 127
services might be related inversely to the aggregate output and em-
ployment, or heavily dependent on the international price of a locally
produced and exported commodity. A classification of the underlying
influences that affect the unanticipated component of the i-th unit's
net expenditure is valuable not only in helping design systems to moni-
tor and control those risks, but also in permitting the supranational or
overseeing entity to assess its exposure to those influences.
A risk model of this type is conventionally called a factor model.
For example, in assessing the risk of a portfolio of common stocks, a
financial theorist would assess the stocks' individual risks (standard
deviations) as well as the correlations between every pair of stocks in
the portfolio. A more parsimonious description of the risk structure of
the stocks in this portfolio is a model that assumes there are a few
common factors that affect each stock's returns and that a given stock
has an elasticity or a sensitivity to each of these factors. The fluctua-
tions in a given stock's returns depend on the sum of the influences
contributed by each of these factors, plus a risk that is unique or idio-
syncratic to that stock and uncorrelated to the returns of the other
stocks. Anyone summarizing a factor model for a useful representa-
tion of the risks in stock market portfolios would need for a given
st:ock a list of the elasticities to each of the K common factors and a
measure (standard deviation) of the risk of its unique component. In
practice, such a model can be statistically estimated from historical
data using the technique of factor analysis in which the data reveal
what the common influences are, or it can be conducted with a
prespecified list of factors that are imposed a priori (for a clear and
useful exposition of such a procedure, see Sharpe 1981).
In the context of the application here, the lack of adequate histori-
cal data related to budget expendlitures (especially in a government
context in which departments and activities are being added over time)
will not permit the statistical estimation of such a factor model. How-
ever, the very procedure of constructing a periodic budget will point
up the factors (such as fuel prices or exchange rates) that are influen-
tial and lead to the estimation of the elasticities.
Tbe Measurement of Risk at the Level of the Unit
The probability distribution of the unanticipated shock ii, to the i-th
department's budget in period t associates with each (possible) level of
the random shock an assessment of the likelihood of its occurrence.
The standard measure of risk in this context is the variance of the
related quantity, the standard deviation. When the likely shocks are
symmetrical around their average value of zero, the standard devia-
tion measure reveals that approximately two-thirds of the realized
outcomes will lie in an interval of one standard deviation on either
128 KRISHNA RAMASWAMY
side of zero. The graph that follows shows such a distribution; it en-
compasses extreme values that are more than two standard deviations
away and thus occur with lower probability (likelihood), but the bulk
of the outcomes occur in the middle.
I I I
-3 -2 -1 0 1 2 3
Subunit i's deficit ;,,
Arriving at a forecast of the standard deviation of ii, is a difficult
but necessary step. It can draw on historical experience, or it can be
derived from a subjective forecast of, say, three or more scenarios for
the most significant influences on the department's activity.
One could expect that the larger the department, the larger would
be its expected cash flow E(C;,) and the larger would be the standard
deviation of its unanticipated component. There is no reason to ex-
pect, however, that there is a special relationship-proportional or
linear-between the expected values and the risks of a departmental
cash flow when viewing these quantities across departments. One could
(reasonably) expect that the risk or standard deviation of a particular
department is related positively to the growth rate in the department's
budget-the year-on-year growth in the anticipated budget item E(Ci,).
Note, however, that the level of the anticipated budget E(C) may be
a policy-driven variable that can be taken as exogenously given or as a
forecast, and its growth rate might also reflect an active decision com-
ing from a political process.
The Aggregation of Cash Flows of Constituent Units
At the supradepartmental or governmental level, the aggregation of
the cash flows requires adding up the contribution from the N con-
stituent units:
Aggregate Cash Flow = + C, + C3, + + CN,
Within the period, and working with a cash flow-based budget pro-
cess, one can anticipate the fiscal deficit at the governmental level to
be the expected value of
ANALYTICAL TECHNIQUES 129
E(Aggregate Cash Flow) = E(C,,) + E(C2,) + E(C3,) + * + E(CN,)
which works out to the sum of the anticipated budgets of the N con-
stituent units. If this aggregate nlmber is positive, then the govern-
ment anticipates and plans for a deficit, given that I chose to use positive
numbers to represent net expenditures or uses of funds.
What can be said about the risk at the level of the parental govern-
mnent unit? Typically, it is committed to meeting the needs of unantici-
pated net expenditures at each department level, and I shall assume
that it does. It might have the ability to draw on the unanticipated
surplus from those units that has come in with lower-than-expected
net expenditures, or higher-than-expected net revenues, perhaps with
a lag. In other, perhaps more realistic situations, the ability to use the
cash flow from surplus units might be limited. I proceed here with the
assumption that the government can draw on these resources fully.
This assumption enables me to write that the unanticipated compo-
nent of the government deficit labeled in period t is U,, which is the
sum of the unanticipated components at all the departmental levels
Ut ,+ "21 q I3t + + UN,
If each departmental unit stayed just at its anticipated budget level,
then the realized value of each unanticipated component would be
zero, and, as a result, the government deficit also would be right on
target. However, this is an unlikely outcome; in practice, some units
would show net expenditures higher than targeted and others might
come in at below-target levels.
In this scheme is a diversification effect. The risk (or standard de-
viation) of the aggregate unanticipated deficit is not equal to the sum
of the risks (or standard deviations) of the individual unanticipated
components. The standard textbook analysis would point up that if
the unanticipated components were less than perfectly correlated-
assuming for the sake of argument that some pairs were negatively
correlated-the aggregate component would show a lower level of
fluctuation than the sum of component unit risks would lead one to
believe.
But a more useful analysis can be brought to bear conceptually.
Suppose, as was argued earlier, that for the i-th unit the unantici-
pated net expenditure iu,-the deviation of actual expenditures from
those that were anticipated and budgeted-is weakly and positively
influenced by the realized growth rate in per capita income, but
strongly and positively related to the deviation of fuel prices from
the values that were used in planning the anticipated budget. These
driving "factors" are the sources of risk, and the sensitivities of the
net excess expenditures over the anticipated budget to these factors
130 KRISHNA RAMASWAMY
will be large in relation to fuel prices but small in relation to the
income growth rate.
Other budget units might face exposures to these two "factors"-
the growth in incomes and fuel prices, factor sensitivities of different
magnitudes-and they also might be exposed to an additional and
different group of factors altogether. The exposure of the suprana-
tional government to these factors is the sum of the exposures of the
individual units to each of the separately identified factors. The pres-
ence of common and pervasive factors-including aggregate demand
levels, exchange rates, interest rates, international commodity price
levels, and so on-in each departmental unit means that the variances
and covariances of the unanticipated net expenditures across depart-
ments will display a "factor" structure. The Appendix to this chapter
shows the computation of the aggregate risk in the case in which one
can identify a few common and pervasive factors that affect the net
expenditures of the constituent units; a stylized numerical example
with two common factors is provided.
The computation of the aggregate risk reveals the following proper-
ties, in each of the following conditions:
* Case 1. The individual units are exposed to diverse and unrelated
risk factors, so that there is little commonality among them. In this,
the risk reduction achieved depends on the levels of the relative planned
expenditures (the anticipated budgets) within the units.
* Case 2. The separate units are exposed to the same common fac-
tors in the same way; the sensitivities to these factors across the units
are all of the same sign but of different magnitudes. In this case, the
exposure at the government level to the factors is the sum of the expo-
sures across the units. Here the magnitudes of the budgets remain rel-
evant, but the important point is that the exposure to the factor-oil
prices or interest rates-is averaged across the units.
* Case 3. There is commonality among the factors to which the
individual units are exposed, but the sensitivities to these factors are of
opposing signs, so that, for example, one unit has a positive exposure
to fuel price rises while others have negative exposures. It is in this
case that the maximum reduction in exposures to the factors can be
anticipated, but it would be an atypical occurrence.
What can be said of the risks that are unique to the subunits, the
realizations of Z? If the sums have been done right and all the common
factors have been considered, these risks should be unrelated. How-
ever, it can be anticipated that some departments' expenditures would
be hit by "acts of God"-earthquakes or unforeseen emergencies, and
so on-and that the incidence of such "catastrophic" risks might be
modeled further, perhaps in a Monte Carlo simulation context.
ANALYTICAL TECHNIQUES 131
Is it possible to compute a figure to serve as a barometer of future
fiscal shocks? Given a distribution for the aggregate fiscal risk, it is
natural to ask what is the likelihood of an unanticipated shock of a
given size to the deficit? This calculation would be similar in spirit to
the daily evaluations of the prospective risk exposure of the portfolio
of a large financial institution. In these organizations, there is a daily
calculation of the value at risk (VALR), or of a related number, which
indicates the level of loss one can expect over the next day or week
with a prespecified probability level. For example, a 10-day, 95 percent
VAR of $45 million would indicate that a loss of more than $45 mil-
lion could be expected over the next 10 days with 5 percent probabil-
ity. In the context here such a calculation is possible, not on a short-term
basis-which would not really be relevant-but over the future inter-
val for which the budgets are prepared.2
To summarize, the nature of fiscal risks that derive from an aggre-
gate exposure to several subunit budgets can be assessed quantitatively
by employing a factor structure in which the factors as economic in-
fluences are imposed from prior knowledge of the subunits' activities.
A.n estimation of such risks is a first step to computing the exposures
and the overall liability of the parent unit.
The Control of Fiscal Risks
A shareholder-managed corporation in the private sector can exploit
the diversification of the cash flows from the aggregation of the com-
ponent cash flows from all the projects and divisions on its balance
sheet in precisely the same way as described here for the government.
However, the corporation in the private sector can alter the alloca-
tion of its investments across its divisions, thereby controlling the
risk (and the expected profitability) of the overall enterprise to its
current shareholders as owners. But such changes in the allocation of
budgets to the underlying departmnents are not generally feasible for
the government unit.
This brings us squarely to the issue of how the risk at the aggregate
governmental level can be controlled. It is by now well recognized
(and it has been stressed repeatedly elsewhere in the chapters in this
volume) that budgeting for government fiscal risks is made more diffi-
cult when the subunits under the government's direction exploit the
fact that excess unanticipated expenditures will be financed by the
parent unit. An explicit (or implicit) coverage of this type provides the
wrong incentives for the subunits: they may no longer take pains to
prepare accurate budgets and forecasts, and they may not monitor
expenditures and control risks to the extent that they would if they
were residual claimants as are the shareholders of a private sector
132 KRISHNA RAMASWAMY
corporation. When a subunit has the ability to affect the level of the
net expenditures, any coverage of the excess net expenditures over the
anticipated budget will raise issues related to moral hazard. In this
sense, the assessment of the probability distribution of each unit's i,. is
made doubly difficult.
In an individual's risk asset portfolio, the most immediate and di-
rect way to reduce risk is to allocate part of the portfolio wealth to the
risk-free asset.3 That way, the overall risk is reduced. When an insurer
(the example here is of the government unit) assesses the actuarial or
expected value of any future liabilities arising from unanticipated ex-
penditures, and it incurs and provisions for it, it effectively invests in a
risk-free asset to reduce its risk. In practical terms, this is equivalent to
limiting the size of its activities.
As we shall see in the rest of this chapter, in cases in which the
liability incurred is related to an economic quantity that is effectively
a traded asset, it is possible to actively control the liability.
Market-Based Vehicles for Controlling Risk
What are the other ways in which a privately held corporation can con-
trol its risks? Exchange-traded and over-the-counter methods for risk
control undertaken by corporations take myriad forms, but their diver-
sity (and complexity) can be reduced to one or more of the following:
* Forward and futures contracts. These are methods that enable
the user to lock in a fixed price or fixed rate at a future date.
* Options contracts. These are essentially the same in this context
as acquiring insurance. They enable the user to purchase protection
against unfavorable outcomes. The purchase of put options would in-
sure against unfavorably low prices for an asset or commodity that
one wishes to sell. The purchase of call options would provide insur-
ance against unfavorably high prices for an asset or commodity that
one is preparing to buy.
* Swaps. These are methods that provide protection against risk
over several periods, being in essence a portfolio of forward positions.
Note that these contracts can apply to risks related to economic
quantities such as exchange rates, interest rates, equity values, com-
modity prices, including energy prices, and default premiums. (Other
mechanisms for risk management, such as securitization and sale of
related securities, are discussed elsewhere in this volume.) Often the
risk that a particular company faces might not be identical to a traded
underlying asset-but this "basis risk" poses only occasional difficul-
ties to most risk managers' operations.
ANALYTICAL TECHNIQUES 133
A conventional question in this context posed to the risk manager is
related to the cost of risk reduction or hedging. The fact that no outlay
is needed to engage in forward contracts, or for that matter in closely
related swap contracts, cannot be taken to imply that there is no cost
to hedging. By locking in, via a forward contract, a fixed price for the
acquisition of heating oil in the future, the consumer reduces the fluc-
tuations in expenditures, but knows that the fixed price that was quoted
is adjusted for the risk of the fluctuations that someone else is now
bearing. In the case of options contracts, the up-front fee for the option
is the actuarially fair price for insuring against unfavorable future prices.4
In looking at the risk inherent in its fiscal deficit, a government
typically sees the risk of low growth as the most important factor. It
would be impractical for the government to hedge against this "fac-
tor."5 But there are other factors that drive at least part of the fiscal
(deficit against which the government can hedge, such as fuel prices,
commodity prices, and interest rates.
One contingent liability that is discussed repeatedly in this volume
relates to a loan guarantee, or an insurance guarantee such as that
given for deposits at financial institutions. It is well known that these
guarantees are isomorphic to (typically) put options. The celebrated
Black-Scholes model provides the analytical basis for the valuation of
these guarantees; it is based on the assumption of the tradability of the
underlying risk and the feasibility of a dynamic hedging strategy. When
these assumptions do not hold (as they may not in the context of many
developing countries' guarantee provisions), it does not mean that the
valuation cannot proceed; in these cases, one can reasonably use the
technique of computing the discounted expected value.6
Here I must stress that the recognition of the liability and the com-
putation of its "fair value" take on the usefulness of an amulet, be-
cause these actions in themselves do not limit the liability or even help
to control it. A government thai: provides guarantees is in the same
position as an individual investor who has written a put option-and
that investor typically will not remain sanguine in the knowledge that
he has carefully recorded the opcion's Black-Scholes value as a liabii-
ity. It is only in the constructive act of taking offsetting positions that
the risk is reduced.
The Factor Model and Contingent Liabilities
The factor model described earlier highlighted the fact that a parent
government unit faced the sum of the exposures to the common eco-
nomic influences that affected the subunit budgets. Some of these com-
mon influences may be traded economic quantities, a fact that the
parent unit can exploit in one of two ways.
134 KRISHNA RAMASWAMY
The parent unit can sell explicit guarantees to the subunits against
unfavorable outcomes in these quantities. Such guarantees could be
hedged by the explicit purchase of marketed insurance contracts at the
level of the parent. Under this scheme, the parent serves as a ware-
house that manages a book of these liabilities. Incentives at the sub-
unit level are maintained-they would still need monitoring-because
the economic quantity that is the basis of the guarantee is observable
to both parties, so long as the parent is not also the guarantor of the
deficit from all sources of risk.
An alternative scheme is for the parent to take up the exposures to
the factors from the subunit levels, but purchase insurance against the
resulting aggregate factor exposures to traded underlying risks directly
from the market. Consider the following example: a state government
oversees the expenditures of several units, including the Department
of Highways, the Department of Prisons, the Department of Unem-
ployment Relief, and the Department of Public Education. Each sub-
mits a budget and prepares estimates of the quantity of fuel and heating
oil it would need, using projections for the demands for its services
and using forecasts of forward prices. A severe winter would lead to
higher consumption of heating oil at both the schools and the prisons,
perhaps at higher-than-expected prices. Extremely severe winters would
require increased use of fuel to clear the highways, but savings in heat-
ing oil at the schools. A prolonged bout of bad weather can hurt agri-
cultural output and affect future unemployment levels. Both heating
oil and gasoline are traded commodities, and it is possible to structure
risk management of the unanticipated expenditures in relation to a
fuel-related factor. Weather derivatives are available, as are insurance
contracts against severe weather. An assessment of the exposures of
the separate departments to fuel price fluctuations and to extreme
weather scenarios would permit the parent state government to either
hedge the unanticipated demands on its resources or provide explicit
insurance where it is feasible to do so.
Conclusion
Analytical techniques for risk measurement and risk management that
have been developed in finance are applicable to the management of
government fiscal risks; the most well-known example among these is
the provision of guarantees or insurance in the context of pension ben-
efits, deposit insurance, and protection against default by a public or
private corporation.
In this chapter I have highlighted the role of risk measurement in
the fiscal context by employing a factor model, which forms the basis
of risk measurement in portfolio analysis. The model can be adapted
ANALYTICAL TECHNIQUES 135
to the fiscal context where a parent government oversees the expendi-
tures of several subunits with separate budgets; the risk factors in this
context may be recognized by examining the economic influences within
these budgets. Whereas the typical application of the factor model in a
portfolio context leads to the allocation of investments, in the context
of fiscal risks the application is confined to measuring and monitoring
risks. However, this is an importan't first step in the use of techniques
to control for these risks. The factor model is particularly suited to the
aggregation of risks and the recognition of the net exposure of the
parent unit to latent risks in the departmental or subunit budgets.
Annex 5.1
This Appendix spells out the technical details of the analytical model.
The model corresponds to a K-factor model of the net expenditures of
N separate entities, which were described as departments or subunits
earlier in this chapter. The parent government unit aggregates the cash
flows across these subunits and is assumed to be responsible for the
aggregate net expenditures that are unanticipated, implying that at the
end of the period the parent government will meet the net demands for
aid from each of the subunits.
Let C,, represent the cash flow of the i-th subunit, which is first
written as
C,, = E(C,,) +
This equation says that the ranclom cash flow from the i-th unit in
period t is (tautologically) decomposed into its expected value plus a
mean zero error, iu,. The expected value corresponds to the antici-
pated budget or forecast budget for department i; the error term uit
corresponds to the surprise or what earlier in the chapter I called the
unanticipated net expenditure.
Factor Structure
Now I shall impose a linear factor structure on the unanticipated net
expenditure; K pervasive common factors will affect the i-th depart-
ment's budget expenditures:
K
it= E (1ikfk} + Eji
k = i
This relation says that the K separate factors are underlying sources of
risk that drive the risk at the department's budget level. The sensitivity
of the budget to the particular factcr k-say, the fuel price, representing
136 KRISHNA RAMASWAMY
the deviation of the realized fuel price from the value that was used in
designing the department's budgeted expenditure for fuel-is captured
by bik. Large positive values for b,k would imply that the department's
net expenditures amplify the fluctuations in fuel prices around their
forecast values; small and positive values indicate a lower sensitivity,
while a negative value indicates that the effect on the department's
budget is to promote a surplus (deficit) relative to budget estimates of
total net expenditures when fuel prices are higher (lower) than ex-
pected. Note that some factor sensitivities may be zero-the subunit
may have no exposure to exchange rates or interest rates. The final
term, e,, represents the risk that is unique to the i-th subunit and
unrelated to any of the chosen factors; it is the idiosyncratic risk whose
variation will be the only source of variation in the subunit's realized
budget figures if each of the factors takes on the value zero (which
would mean, for the example of the k-th fuel price-related factor, that
the realized fuel price was equal to its expected value, used in prepar-
ing the anticipated budget).
The following remarks apply to the choice of the factors and the
factor sensitivities:
* The factors should be chosen to represent the major common
influences to which the subunits are exposed. The examples I give: fuel
prices, foreign exchange rates, interest rates, aggregate economic ac-
tivity-related variables such as employment or per capita income, the
international price of a locally produced or imported item or raw
material, and so on.
* The factor sensitivities can actually be traced through the budget
process as a budgeting variance. By tracing the effect of a 1 percent
change in the forecast fuel price, one can ask what the effect on the
overall budget would be, keeping other things constant.
* One assumption I place on the model says that the expectation of
the unique or idiosyncratic shock at the department level, conditional
on the factor realizations, is zero-that is,
E (ii,tif, f2, f3, .. I f) = °.
This relation says that the unique or idiosyncratic term is just that:
unpredictable noise that is unrelated to the factors.
* Note that I do not make the assumption that the factors are
uncorrelated to each other. In fact, I expect that several assumptions
about the factors have been used in devising the budget items, including
assumptions about average fuel prices, unemployment levels, and ex-
change rates. And I expect that these economic quantities are related.
* Now consider the factor structure for the net unanticipated ex-
penditures for two separate subunits, i and j. If I have extracted all the
common factors between the net random expenditures of the subunits,
ANALYTICAL TECHNIQUES 137
then I can conclude that the correlation (equivalently, covariance) be-
tween the residual or idiosyncratic risks, £it and i,,, is zero.
In typical finance applications, the factors are extracted from the
data, and the usefulness of the factor model comes from the properties
of the correlation between these "residual" risk terms. There, it is
usually argued that the correlations between the idiosyncratic risks are
weak, or zero.
In the application that is the fccus here, the factor model is being
employed to capture commonalities in net unanticipated budget ex-
penditures. Here I do not have the luxury of a sample of past data
from which to extract common factors: rather, I will impose such fac-
tors and attempt to capture as many important economic influences
that can make the subunit's actual expenditures depart from budgeted
expenditures. Of course, one outcome of this is that if my specification
cf these factors is partial or incomplete, then the residual risks can be
correlated across the subunits.
The variance of the unanticipated net expenditures at the level of
the subunit is then
Var(Cit) = Var(iu,,) = Var |, bjk4k } + Var(ij,)
Notice that a degree of "diversification" is taking place within the
subunit itself: the net budget shortfall or surplus will be driven by the
aggregate influence of K factors. The ratio of the variance due to the
factors to the total variance captures how much of the variation in
the budget expenditures is captured by the K chosen factors in the
decomposition.
Aggregation to the Government Level
The parent unit serves to absorb the sum of the unanticipated net ex-
penditures at the level of the subunits, so let us look to find the prop-
erties of
N
ut= E1Ujt
Using the factor decomposition that was developed for ii,, yields
XN l XN l N N
138 KRISHNA RAMASWAMY
which shows that the unanticipated demand on the parent's deficit
coverage (a) depends on the same K factors, and (b) is also exposed to
the sum of the idiosyncratic shocks to each department's budget that is
unrelated to the K factors.
Defining the K separate sensitivities at the parent level as
N
Bk=( bikJ k = 1, 2, . . ., K
reveals that the sensitivity to the k-th factor at the aggregate level is
the sum of the sensitivities at all the subunit levels. The aggregate
unanticipated deficit can then be written as
K N
ut XBk fk + ( E£).
k = I i= I
The following observations regarding U,, the unanticipated budget
expenditures at the aggregate level, are now in order:
* Summing the exposures across each of the K factors would reveal
that for some cases the risk exposures are reduced in the aggregate.
* Assuming the residual risks E, at the subunit levels are not corre-
lated across subunits, one finds that the risk unrelated to the factors at
the aggregate level has a smaller effect on the aggregate risk.
* What are the properties of the distribution of the aggregate bud-
get risk? Earlier in the chapter, I define in broad terms a number (by
analogy to value at risk, VAR) that applies to a percentile of the distri-
bution of U,. Unlike the corresponding VAR measures, which are ap-
plied to the potential cash flow gains and losses over the next day or
week, the fiscal risk measures described here apply to the risk levels
faced by the government unit over the next budget period.
Aggregate Budget Deficit (Positive) or Surplus (Negative)
Prob p
-3 -2 -1 0 1 2 C 3
Aggregate deficit ($ billions)
ANALYTICAL TECHNIQUES 139
In the graph, an aggregate deficit figure larger than 4 = $2.5B is
seen with probability p. I have used the distribution of the aggregate
deficit as a bell-shaped curve only as an example.
This measure is best suited to timne periods and situations in which
the risks do not change-that is, the VAR number is likely to be accu-
rate only for short periods and for portfolios whose risk characteristics
do not change during the periods. If the portfolio has options posi-
tions, or nonlinear characteristics, then these assumptions can be vio-
lated. In the context of government budgets and fiscal risks, the budget
periods are typically annual (at best quarterly), and there may be con-
tingent liabilities that have option-like characteristics. For these rea-
sons, the at-risk figure computed as described above will not be accurate
and should only be taken as an approximation.
* Some of the K factors (say the first m factors, k = 1, 2, . . ., m)
may be related to traded assets, and the remaining factors (k = m + 1,
m + 2, . . . , K) may be important economic influences that are not
traded. In that case, what would be the risk levels if the decision were
made to hedge some of the traded risks-perhaps with the use of de-
rivatives for which there is a market?
7'he Factor Model: A Numerical Example
A, simple numerical example will illustrate the workings of a factor
model. Assume that there are four government units or state-owned
enterprises (SOEs), and that two common economic factors affect the
cash flows of these units-GNP growth ( f) and the average level of
energy prices (f2). Each government unit prepares a budget for the
coming year, and each unit's budget process builds in certain baseline
assumptions about GNP growth and the expected energy price level.
The realized GNP growth rate and, the actual energy prices will consti-
tute departures from these baseline assumptions and will therefore lead
to surpluses and deficits relative to the budget amounts in each unit's
planned cash flow. These baseline assumptions are provided by a cen-
tral forecasting service, and, for the sake of argument, suppose that
these assumptions include a forecast of the standard deviation of the
forecast percentage of GNP growth (a, = 0.5%) and the standard de-
viation of forecast error in baseline energy prices (aY2 0.7%), and that
these factors are uncorrelated.
Suppose now that the expected budget amount for Unit 1, say, is
$50 million, in accordance with the baseline assumptions. An analysis
of the budget process of this unit reveals that the effect of a 1 percent
departure of GNP growth rate from the baseline expected growth rate
corresponds to an unanticipated $5 million in expenditures, and a 1
percent departure from assumed baseline energy prices gives rise to an
unanticipated $1.2 million in expenditures. These numbers are derived
140 KRISHNA RAMASWAMY
by tracing their effect through the budget planning process. In the no-
tation of the previous section, b, = 5 and b,2 = 1.2 for Unit i = 1. In
addition, other idiosyncratic factors affect the cash flows of Unit 1
that are unrelated to the two common factors. Assume that the stan-
dard deviation of the idiosyncratic risk in the budget for Unit 1 is $2
million. In the notation established above, the idiosyncratic risk in
Unit l's budget is Var(1, ) =(1,,) = $2M.
These three quantities-the sensitivities b.,, b, to the two factors, plus
the idiosyncratic standard deviation Y(i,, )-constitute the numbers that
must be assessed for each unit, from the information embedded in the
budget process. Listed in the table below are the numbers for the nu-
merical example.
GNP Energy Idiosyncratic Budget
sensitivity sensitivity risk level
Item b.1 b, o(e) E(C,)
Unit 1 5 1.2 2.0 -25
Unit 2 1 0.5 0.5 10
Unit 3 0 1.5 2.5 15
Unit 4 -12 0.1 4.0 30
Total -6 3.3 Not applicable 30
An explanation of the entries in the table follows:
* The last column of the table contains the expected budget level of
the unit, in millions of dollars. Only the first unit is revenue-produc-
ing; the others expect to incur deficits. However, the factor model
applies to the unanticipated component-the risk in the aggregate defi-
cit-which is the departure from the expected levels. The total (aggre-
gate) expected budget deficit is $30 million, as shown in the last row.
* Unit 4 has a negative elasticity to the GNP growth rate; if the
economy grows faster than expected, by 1 percent, then the budget
deficit actually declines in that unit by $12 million. But the forecast
also acknowledges that the idiosyncratic risk in that unit's cash flows
is also highest, while it is least sensitive to energy prices. This unit
might have responsibilities that devolve, for example, on health or
education, or welfare, so that its expenditures might be reduced if the
economy does well.
* Unit 4 has no exposure to GNP; its risk draws entirely from energy-
related and idiosyncratic risks.
* The last row provides the total aggregate exposure to the two
common factors. Note that the aggregate exposure to the energy price
factor simply adds up and that no unit provides a diversifying expo-
sure to energy prices. In contrast, Unit 4 has a large negative exposure
ANALYTICAL TECHNIQUES 141
to GNP, so that the aggregate exposure is actually negative. In the
notation of the section above, B, = -6 and B2 = 3.3.
* Finally, note that the standard deviations of the idiosyncratic risks
do not add up, so it makes no sense to sum them.
To compute the budget risks for each of the units and for the parent
unit, take the numbers corresponding to the standard deviations of the
two factors provided by the central forecasting service, aY = 0.5% and
G2 = 0.7%. Then the variance of Unit l's budget can be computed as
Var(Cit) = b2 I 121 + b12( + (i,
which computes to
Var(C,,) = (52 X 0.52) + (1.22 x 0.72) + 22 = 10.96,
implying that Unit l's budget on a stand-alone basis (its own total
risk) has a standard deviation of ^F10.96 = $3.31M. It can be verified
using the numbers in the table that the stand-alone risks of Units 2, 3,
and 4 are given by the standard deviation levels of $0.79 million, $2.71
million, and $7.21 million, respecrively. As expected, Unit 4 shows the
highest overall risk level.
Thus the expected budget expenditure for Unit 1 is revenue of $25
million, which translates into the statement that the realized revenue is
forecast to fluctuate in the region $25M ± 3.31M = $21.69M to $28.31M
two-thirds of the time, assuming a normal distribution of factor and
idiosyncratic influences. Similarly. Unit 4's budget will show a deficit
ranging from $22.79 million to $37.21 million in two out of three years.
But the common factor exposures at the individual unit level ag-
gregate to provide total exposures as shown in the last row of the
table. The aggregate deficit can now be computed as
i=4
Var(Aggr Def.) = B 2ol + B2C2 + a (E
which when we use the numbers yields
40.84 = ((-6)2 x 0.5 2) + ((3.3)2 x 0.72) + {22 + 0.52 + 2.52 + 42)
The standard deviation at the aggregate level works out to 4084
= $6.39M, implying that the parent will feel a deficit in the range of
$30M ± 6.39M = $23.31M to $36.39M in two out of three years. As
this stylized example shows, an averaging of factor exposures reduces
the exposure to GNP fluctuations, and the diversification of idiosyn-
cratic influences occurs as well. Indeed, the risk at the parent level is
less than the total stand-alone risk felt by the managers at Unit 4.
142 KRISHNA RAMASWAMY
Notes
1. See, for example, Chapter 4 by Suresh M. Sundaresan and Chapter 13
by George Pennacchi in this volume. I will restrict the discussion in this
chapter to analytical methods that follow up on the factor model just intro-
duced.
2. There is a close correspondence between the practice of stress testing
the budgets and this effective VAR number.
3. The most immediate way to increase risk is to leverage the portfolio by
borrowing.
4. In some arrangements, the contract for insuring the risk buries the cost
of hedging in a net payment that occurs later-that is, the cost of insurance is
paid at the end, in favorable outcome states, so no up-front fee is needed. But
in those cases, the insurance is not free and must still be budgeted for in a
fiscal context.
5. It would require the government to take a "short" position in the mar-
ket-to enable it to receive cash when aggregate output is below expecta-
tions-and in any case this would require a reliable positive correlation
between output and market values, which is not typically observed.
6. In the event that the no-arbitrage assumptions are met, one can com-
pute the expectation of the payoff under the guarantee in a risk-neutral
economy and discount it at the risk-free rate.
References
Arrow, K. J. 1971. Essays in the Theory of Risk Bearing. Chicago: Markham.
Arrow, K. J., and R. C. Lind. 1970."Uncertainty and the Evaluation of Pub-
lic Investments." American Economic Review 60: 364-78.
Lewis, Christopher, and Ashoka Mody. 1997. "The Management of Contin-
gent Liabilities: A Risk Management Framework for National Govern-
ments." In Timothy Irwin, Michael Klein, Guillermo Perry, and Mateen
Thobani, eds., Dealing with Public Risk in Private Infrastructure. World
Bank Latin American and Caribbean Studies. Washington, D.C.: World
Bank.
Merton, R. C. 1990. Continuous Time Finance. Cambridge, Mass.: Blackwell.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Sharpe, W. F. 1981. "Some Factors in NYSE Security Returns: 1931-79."
Journal of Portfolio Management 8 (4): 5-19.
CHAPTER 6
Fiscal Sustainability and
a Contingency Trust Fund
Daniel Cohen
Universite de Paris, Ecole normale superieure
Tlins CHAPTER SETS UP A simple framework to analyze the conse-
quences of contingent liabilities for the management of public finances.
From this framework, it suggests the design of an institutional solu-
tion to the problem of fiscal opportunism in government when dealing
with contingent liabilities.'
The idea developed here is that government should handle contin-
gent liabilities by setting up an independent trust fund that itself will
have limited liabilities. The government is expected to be the key, but
not necessarily the only, shareholder of the fund. When the fund is
established, it is endowed with capital that corresponds to the overall
size of the government commitmenit. A government would finance this
endowment by raising correspondingly its debt. This first stage has the
merit of imposing on government decisionmaking a one-to-one rela-
tionship between the government's commitments to the private sector
and its own liabilities to the financial market. As a shareholder of the
trust fund, on the other hand, the government can expect to receive
dividends. The dividends should be set so that they keep constant (say,
in real terms) the resources that are managed by the trust fund-in
effect, maintaining a constant level of guarantees that the fund is en-
titled to offer. Under these circumstances, the government needs only
to raise taxes by an amount that corresponds to the difference between
the debt it has issued and the "market value" of the trust fund. The
market value of the trust fund corresponds in theory to the difference
between the resources with which it has been endowed and the cost it
is expected to cover. Raising taxes to the difference between the initial
143
144 DANIEL COHEN
endowment of the trust fund and what it should be expected, on aver-
age, to pay is exactly what the theory of optimal taxation suggests.
There are two merits to and two critical questions about this scheme.
The first merit is that it forces the government to internalize immedi-
ately the cost of the scheme it wants to implement. The second merit is
that the limited liability nature of the trust fund puts an explicit ceiling
on government involvement. If risks are initially underreported, this
would not represent an additional burden to the government, unless it
willingly accepts the idea of sharing the extra burden with the private
sector. This leads to the first question. How can the government ap-
propriately assess the market value of its commitments? This question
can be partially answered by dragging in the private sector. If shares of
the trust fund were to be marketed, the information revealed by the
market could then be channeled to the government, allowing it to fi-
nance adequately the net burden that this commitment adds to those
of its other commitments. The second question is associated with the
nature of the risks involved. Should the government set up one trust
fund for all contingent liabilities, or one for each category of risk? In
principle, it is optimal to set up one trust fund for each category of
risk. As I will argue, however, pooling risks tends to lower the cost to
the government and is therefore bad from an allocative perspective,
but good from a financial perspective.
The chapter proceeds as follows. I first briefly summarize the frame-
work of the theory of government solvency. I then examine how it
should be extended to encompass contingent liabilities. Finally, I ana-
lyze the merits (and the limits) of the institutional framework I have
sketched.
The Basic Analytical Framework
An Intertemporal Approach to Government Solvency
The simplest framework for analyzing fiscal sustainability is to follow
the law of motion of government debt over time (all technical details
are given in the Appendix to this chapter). The debt buildup of the
government stems from the primary deficit (which is simply the differ-
ence between government expenditures other than interest payments
and government revenues) and the interest bill itself. While the former
can, in principle, be changed at will, the latter is the legacy of the past
and can only be changed with time. This explains how a profligate
government can exhaust the solvency of its successors by leaving out a
stock of debt that might turn out to be "unsustainable" (on these is-
sues, see, for example, Buiter 1985 and Cohen 1991).
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 145
A direct way to assess whether a government is solvent is to write,
in present value terms, all items that will contribute to the buildup or
to the reduction of the debt in the future. (The present value terms
simply weight any item forthcoming at a time in the future by the
inverse of the compounded interest that applies up to that time.) The
increase in the present value of the debt between any initial time and
the future can then be written straightforwardly as the sum of the
present values of forthcoming primary surpluses. Solvency thus can be
defined as follows: a government is solvent if and only if it is able to
generate in the future a stream of primary surpluses that are sufficient
to repay (in present value terms) t:he stock of outstanding debt that it
has inherited from its predecessors. It is a straightforward analytical
exercise to demonstrate that such a condition will hold if and only if
the present discounted value of the debt will itself converge toward
zero in the long run. Indeed, if this is the case this condition implies
that the debt is "repaid" asymptotically and that there will always be
a chain of investors willing to rel'inance the government (so long as
they are certain that the government will continue to implement the
set of primary surpluses that generate such an outcome).
Incorporating Contingent Liabilities
The above principles address the case in which a government can com-
rnit itself to a stream of primary surpluses that can handle the debt it
has inherited. Let us now see how these principles should be amended
to encompass the case of contingent liabilities. In order to do so, let us
assume that the government grants the private sector a guarantee against
a one-off event such as an earthquake or a banking crisis. To simplify
t:he presentation, let us assume here that the guarantee will cost, if
exercised, a fixed value that is known in advance. Clearly, the cost of
the guarantee must take account of the probability that the shock will
occur, but also, and most important, the likely timing of the shock
itself. Assume that the shock is rnost likely to occur in 10 years, al-
though it might also occur, with a lower probability, in 2 or 20 years.
The government therefore should budget the contingent liabilities based
on the present value, mostly for a 10-year horizon, of the expenses
that will be involved on average.
So long as interest rates are positive, this present value will be less
than the expected value of the cost itself. Clearly, there are only two
situations in which the expected cost (guarantee times probability of
exercising it) may not be the right measure of the burden involved:
when the time horizon over which the contingent liability is expected
to be exercised is very short, or when the interest rate is very low. In
each of these two extreme cases, the government should simply budget
146 DANIEL COHEN
the expected value of the cost (cost times probability of occurrence).
In all other intermediate cases, it should budget a lower number.
This simple result is at odds with the conventional idea that any
contingent liability should be incorporated in the budget either at face
value or at a lower number that is measured simply by the likelihood
that the guarantee will be exercised. The basic principles just presented
reveal that the expected timing of the event is crucial. Events that are
certain to occur sometime in the future may cost less than events that
might occur with a probability lower than one, but fairly soon if they
do occur. Take the example of a financial crisis affecting the banking
sector. It is a sure thing in most developing countries that such crises
will occur. No one expects it to happen tomorrow morning, however,
which allows the government to take actions progressively in order to
meet the expected cost (see below). Imagine instead that the govern-
ment extends a guarantee against the risk that the Y2K bug will dis-
rupt the payment system. Here the probability is certainly smaller than
one, but the timing of the event is very sharp: it might cost the govern-
ment more to provision the latter risk than the former.
Practical Implications
How should the government handle the financing of contingent li-
abilities in practical terms? Two different methods come to mind. One
option is to wait until the guarantee is exercised. At that point, the
debt is increased by an amount corresponding to the guarantee to be
exercised. Another option is to raise taxes preventively in order to
match the market value of the expenses that are generated by the con-
tingent liability. If government pays any attention to smoothing taxa-
tion, as is usually the case when taxation is distortionary, this latter
option is bound to dominate the former one. At this point, however,
the complexity of the political process plays a critical role. From the
perspective of a current government that does not internalize the wel-
fare of its successors, the first scheme is likely to appear preferable, at
the expense of social efficiency. (See Polackova 1998 and Kharas and
Mishra 2001 for a more detailed discussion of contingent liabilities
and their fiscal implications.)
Government Risk and Institutional Designs
The Risk of Government Repudiation
In most developing countries, a state of fiscal crisis is almost a fact of
life, making it difficult for the government to guarantee contracts in
the future, when its own debt is quite often at risk. In order to give
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 147
some empirical insights into the extent of such risk, I calculated the
primary surpluses of three European countries-Belgium, Italy, and
Ireland-that faced deep financial problems in the 1980s but were
successfully dragged out of their crises. To summarize, in the 1980s
Belgium's government surplus as a fraction of the tax receipts of the
high debtors of the Organisation for Economic Co-operation and De-
vetopment (OECD) was 10 percent. Italy's corresponding surplus was
12 percent and Ireland's 15 percent.
These numbers, which are not far apart, point to the idea that the
"financial crisis" of a state originates from a relatively low ratio of
government surpluses to tax receipts. Contingent liabilities such as
protecting the banking sector or guaranteeing some major environ-
mental damage often involve costs that are typically larger than the
immediate financing capacity of thie government. In such cases, the
government would not be willing to honor its previous commitment
and the entire scheme would collapse.
Institutional Designs
If the government's signature is at risk, the guarantee that it granted
will not necessarily be honored. Take the extreme case in which the
risk that is protected corresponds to the risk of government default
(such as a "systemic" risk that threatens the banking sector and simul-
taneously endangers public finances). The private sector will then never
be protected ex post and government guarantees will be pointless. In
such a case, there is room for institutional improvements that will
protect the private sector against the risk of government default. One
simple way to address this issue is to set up an independent trust fund
that is endowed by the government by an amount that is explicitly
geared toward protecting the privare risk. Let us now review how such
a scheme should proceed.
In terms of credibility, the easiest way to proceed is to endow the
trust fund with an amount that corresponds to the risk that is pro-
tected and to make the government the shareholder of the fund. This
one-off transfer is financed through an increase in government debt.
As long as the risk does not materialize, the fund makes profits that
can be captured by the government (essentially in the same way that a
central bank's profits are redeemed by the government). Once the risk
materializes, the trust fund extends its guarantee to the private sector,
regardless of the government's financial situation. The fact that the
trust fund capitalizes on its assets as long as the risk has not material-
ized means that the government receives a transfer every period, as
dividends, as long as the risk does not show up. Through such a policy
(assumed to be designed in real terms), the fund's resources are kept to
the initial level. In present value terms, this strategy corresponds to a
148 DANIEL COHEN
net receipt, which can be subtracted from the direct cost originally
involved in the transfer. This leaves the government with a net burden
that obviously corresponds with the number involved in the first evalu-
ation of the cost, and that must be matched by raising taxes by the
corresponding expected amount. Clearly, then, the optimal taxation
scheme that I alluded to earlier in this chapter can be readily delivered
by this scheme.
Bailing in the Private Sector
There is no reason why the government should be the only shareholder
of the trust fund. In particular, when there is uncertainty about the
nature of the risk involved it may be critical that the government build
on the private sector's knowledge.
Assume, for example, that the risk covered by the government ma-
terializes with a probability that is common knowledge among private
agents but not shared with government officials. If shares of the trust
fund are marketable, the government can immediately observe the full
value of the trust fund and needs only to finance the residual. In prac-
tice, obviously, the government needs only to trade a few shares in
order to capture the information on the value of the risk. This infor-
mation can be immediately acknowledged in the government's account.
In that case, the government simply has to raise taxes to take account
of the net burden. Clearly, however, nothing prevents the government
from going further and perhaps selling the entire project to the private
sector. In such a situation, the net cost is readily measured by the one-
off difference between the level of the guarantee and the market value
of the trust fund, and the optimal taxation scheme is easy to imple-
ment. This situation corresponds to many practical examples such as
that of an export credit agency. The government could well help the
private sector to create such an agency that insures exporters against
the risk of foreign default, and then privatize the agency itself. The
government's involvement is limited to a one-off endowment that can
be properly financed. The same might be true of a guarantee on the
banking sector. The government could set up a version of the U.S.
Federal Deposit Insurance Corporation (FDIC) aimed at guaranteeing
deposits and then let the banking community manage and sustain the
agency later on.
Pay as You Go
The trust fund scheme also can be applied to the cases in which the
government is involved in repeated transactions (for example, taking
care of pensions) rather than one-off mechanisms only. If the govern-
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 149
ment rolls over every period new guarantees as the old ones are exer-
cised, the nature of the game becomes different. But one can still rely
on a trust fund to protect goverm-nent finance from the risk that the
scheme is misconceived. Assume that the government is willing to com-
mit itself up to a given amount and is willing to renew its commit-
ments in order to keep that amount constant over the years. Rather
than self-managing this set of guarantees, the government could as
well create a fund, endowed appropriately each period, that would
monitor the risks. This arrangement would have the virtue of protect-
ing government finances from unwvanted shifts in the underlying prob-
abilities. If the projects involved are riskier than initially thought, the
private sector will bear the consequences, not the public finances. This
may not be an optimal risk-sharing method, but when the government's
solvency is threatened by an unwanted increase in public expenditures,
it might be the best way to go. Similarly, one could drag in the private
sector. Think of a trade credit agency that the government wants to
help by granting guarantees against adverse events. The government
could endow the trade agency wit:h a given flow of resources and let
the private sector bring some additional collateral. If the risk is ap-
propriately measured, then the scheme is neutral to the private sector.
If it is underreported, then the collateral brought by the private sector
will be lost (at least in part). This could form the basis of a truthful
exchange of information. At a minimum, the private sector would not
want to underestimate the fundamental risk. In fact, it would want to
overestimate that risk, but this would then immediately raise govern-
mnent involvement and would have a deterrence effect.
Merging Risks
Until now, it was implicitly assumed that the government had to deal
only with one category of risks. Assume instead that there are two
classes of risk (the general case would proceed immediately): good
risks and bad risks. Good risks that fall due later (on average) than
bad risks are less costly to guarantee. In theory, the optimal strategy
for the government would be to pledge these two risks, each with a
different trust fund. There is, however, an interesting paradox here:
not being able to monitor the two classes of risk reduces the cost of
operating the trust fund, in a situation in which each class of risk is
managed separately. The intuition behind this apparently paradoxical
result is that the bad risks disappear more rapidly than the good ones,
so that the process by which the risks are reallocated regularly benefits
the good ones. It is then a safe strategy, from the government perspec-
tive, to pool risk, provided, of course, that the usual adverse selection
biases are taken care of.
150 DANIEL COHEN
Conclusion
Whatever the precise institutional setting that one has in mind, it should
be clear that the proper management of contingent liabilities imposes
two imperative constraints. One is a full reporting of the cost involved.
Another is the proper internalization of these costs by the government.
The specific mechanism proposed here is one in which the full cost is
determined by setting up a limited liability framework that imposes,
by fiat, an upper limit to the scope of government intervention. The
proper internalization by government proceeds from the fact that the
financing should be up-front. Although other, somehow less extreme
methods may be possible, it is nonetheless critical that the government
acknowledge ex ante the fiscal implications of the guarantees involved.
Intervening ex post is not only bad from the point of view of the tax
system (which would suffer from the strains of increased expenses),
but also bad for the sectors that might feel protected by government
guarantees to discover only too late that governments do default re-
peatedly on earlier commitments.
Annex 6.1
Fiscal Sustainability and Contingent Liabilities
The simplest framework for analyzing fiscal sustainability is one in
which all government debt is short term, the interest rate is a constant
r, and the economy is deterministic.2 Before relaxing these assump-
tions, one by one, let us first investigate how fiscal sustainability can
be analyzed in this highly simplified environment.
Call D, the stock of government debt at time t, G, government ex-
penditures, and T, the amount of tax collections. The law of motion of
government debt can then be written as
(6.1) D, = (1 + r)D,, + G,-.
(G,- T ) is simply the primary deficit of the government, and rD,1 +
G,- T, represents the overall deficit. At this stage, equation 6.1 is noth-
ing but an accounting identity that states that the increase in govern-
ment debt amounts to the overall deficit. To see how this equation can
be turned into a constraint imposed on government finances, let us
write (6.1) in present value terms, from the perspective of an initial
(and arbitrary) initial time t = 0. Dividing both sides of the equation by
the discount factor 1/(1 + r)' produces
(6.2) D.T D' X_ I GI - T,
(6.2) ~~(1 +r)t (+ r)'- (1 +r)t
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 151
which has a straightforward interpretation: in present value terms the
increase in debt is worth the present value of the primary deficit. Add-
ing up all such equations 6.2 from ttie initial time t = 0 to an arbitrary
time T in the future yields
DT ~T GI- T
(6-3) ( )T = D_1(1 + r) = E I t)
(1 +r) - =0 (1 +r)'
The increase of the debt, in present value terms, is the sum of the
present values of the primary deficit. When T is allowed to go to infin-
ity, this yields
lim DT + r),
(6.4) ( T = D 1( + r) = E
T-- (1 r)T t = 0 (1 +r)
The asymptotic value of the present value of the debt is worth the
sum of all past primary deficits and the initial debt.
The straightforward definition of solvency that is then imposed on
the government is called the "transversality condition"-namely
( 6.5s ) lim D T O
(6.5) ~~T-- (1 +r)T
This condition states that the government should let the present
value of its debt go to zero in the long run. Once this condition is
granted, one can see from equation 6.4 that the government is solvent
in the following sense that
T GT
(6.6) E i = D_1(i + r) + E -
(6.6) ~t =0 (1 +r)t t= 0 (1 + r)t
The sum of all tax collections 1:o come, when written in present
value terms, must match the initial value of the debt accumulated by
the government and the sum of all present values of future government
expenditures. The beauty of the transversality condition is that a simple
requirement on long-run debt is equivalent to a complex analysis of
the determinant of future tax and expenses.
The Case of a Growing Econonly
Consider the simple case of a growing economy in which the tax capa-
bility of the government is driven by an exogenous law
(6.7) T, = T0(1 + n)'
152 DANIEL COHEN
in which n is the underlying growth rate of the economy. Assume fur-
thermore that the interest rate is larger than the growth rate-that is,
(6.8) r > n.
In order to simplify the analysis, also assume that government ex-
penditures grow at the same exogenous rate n-that is,
(6.9) G, = GO(1 + n)'
One can then write the present value of the primary deficits as
(6.10) Il = (Go - To) 1 +-r
0 (I+ r)t -
so that the government is solvent if and only if
D To - Go
(6.11) D <
Assume that G is lifted up to the limit of what the government can
afford. Then the primary surplus of the government must be exactly
equal to
(6.12) To - Go = (r- n)D .
One can straightforwardly see that debt is exactly growing at the
rate n, so that at any later time one has
(6.13) T,- G, = (r-n)D,
and
(6.14) D, = (1 + n)D,,.
Clearly, this implies that
(6.15) ~lim _ __= Do lim ~ I0
(6.15) to - (1 + )t = Do t -(I + r ) = °
The solvency condition does not then imply that debt should be
stabilized. Indeed, in this critical example, debt is growing to infinity.
It does impose, however, that debt should not be growing faster than
tax receipts. In other words, it imposes that the debt-to-tax ratio should
be bounded.
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 153
Contingent Liabilities
Assume that the guarantee has pledged to insure a risk that is expected
to be exercised according to a Poisson process of parameter k. This
means simply that the probability that the guarantee will be called
upon at time t is ?I(1 + X)'; t 2 1. Call K the value of the government
guarantee. From a market perspect:ive, the value of the grant can then
be simply written as
(6.16) V=ET( + r)T
in which Er represents the expected value with respect to the time T at
wvhich the government will have to extend its guarantee-that is,
(6.17) V = E _- K = K
I (1 +X (+ r)' (+ r)
The market value V of the guarantee coincides with the amount
involved, K, in two limiting cases: when 1 goes to infinity or when r
goes to zero. In the first case, 1 infinitely large, the guarantee is ex-
pected to be delivered very rapidly; in the second case, the fact that it
will be granted sooner or later is irrelevant if the market puts no weight
on present events compared with later ones. In either of these two
extreme cases, the government should fund the guarantee immediately
and include it at face value as part of government expenses.
In less extreme cases, however., the government should set aside a
lower number that takes account of the fact that the guarantee will be
exercised at some time in the future, thus corresponding to a lower
cost than the face value itself.
In that case, taxes must be raised to the point where they solve for
the new intertemporal budget constraint
(6.18) i t + Do +--x KT
1 (1 + r)t (r + X) 1 (+ r)t
where
(6.19) T =T F+ K
(r +)
Repeated Shocks
Let us now investigate the dynamics that are involved when the gov-
ernment extends continually the set of guarantees that it grants to the
154 DANIEL COHEN
private sector. For simplicity, assume here that all the guarantees per-
tain to the same class of risks, parametrized by X (this hypothesis is
relaxed below). Call K, the value, at any given time t, of the outstand-
ing stock of guarantees that are granted by the government. In market
terms, the value V, of these commitments can be written simply as
(6.20) V, = I K, + V,
(1 +r) [(+?.)
the sum of the present value of the payments that the government is
expected to make in the next period and of the remaining commit-
ments one period later. Furthermore, if X, is the flows of new contin-
gent liabilities that the government extends every period, the outstanding
stock of government guarantees can be written as
(6.21) Kt+, - Kt = - Kt + Xt
(I +X)
since a fraction 2I(1 + X) is redeemed each period and a new flow X, is
added to the total.
Referring to (6.20) and (6.21), specifically expressing K,, in the
(6.21) and substituting it for K,, in (6.20), one can write a general
formula that measures the cost (in present value terms) to the budget
of such a government policy:
(6.22) Vo = K 1~+~ Y '
(r + X) t =1(1 + r)t(1 + k)' [s= I (1 + X)]
The intuition behind such a formula can be readily conveyed by
examining a few relatively simple cases. For example, when X, = 0, the
government lets the stock of guarantees be depleted as time passes.
This yields the same result obtained in the previous subsection:
(6.23) V= K
(r + X)
Now consider the case in which the government keeps replenishing
the stock of contingent liabilities as they are exercised in order to main-
tain the stock of long-term liabilities at a given level K. The overall
cost of such a policy can be simply written as
(6.24) V= K
r(l + X)
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 155
which then has a straightforward policy implication: the government
has to raise taxes by the amount
(6.25) AT=rV=- K
(1 + X)
to be solvent. This is a version of the pay-as-you-go system, because
now the government pays the contingent liabilities as they get going.
Two polar cases are thus emerging. When the government is com-
mitted to a one-off guarantee, funding its scheme is essential to avoid
a one-off increase in taxes when the time comes to exercise the guar-
anree. But when the government is continually rolling over the amount
of its contingent liabilities, then a pay-as-you-go system is fine in theory,
provided that an explicit limit is set on the amount of the government's
guarantees.
Merging Risks
Assume that there are two classes of risk: good risks that fall due with
a Poisson process 1,, and bad risks that fall due with a Poisson process
?2 with X2 > , (bad risks come earlier on). If K, is the amount commit-
ted to the good risk and K2 the amount committed to the bad risk, then
the government must endow each fund with
(6.26) Fl = 1 ; F2 = r22
rr
or equivalently pledge to each of the two trust funds annual transfers
z1=rF1=XK1 and Z2 =rF2 =XK2.
Assume, however, that the government cannot differentiate between
the two groups, and furthermore assume that the government is main-
taining a constant level of commitments, as they were exercised, to the
private sector, no matter which one of the two risks is covered. The
value of such commitments can be written as
(627a)=1 + r) IVI 1 + ;K (t + 12) K2 (t + 1)
(.7 t+)= 1-K t : I(+X1)
(6.27b) K, (t + 1) = - __ - K, (t) +XI (t + 1)
(1 + X1)
156 DANIEL COHEN
(6.27c) K, (t + 1)= k- K2(t) +X2(t+1
(6.2 7c) ~~~(1 + K2)I(()X(t+1
(6.27d) X(t) _ Xl(t) + X2(t) = X,K1(t) + X2K2(t).
In this model, the outstanding level of government guarantees K(t)
+ K2(t) = K(t) is maintained at a constant through government deci-
sions, but the internal composition of these guarantees is left to its
own dynamics. Call 0 the proportion of good risk in the economy and
assume that the reallocation of new guarantees is extended accord-
ingly-that is,
(6.28a) X,(t) = OX(t)
(6.28b) X,(t) = (1 - 0)X(t)
in which
(6.29a) X,(t) = 01X,K,(t) + X2K2(t)]
(6.29b) X2(t) = (1 - 0)[?,K,(t) + k2K2(t)].
Asymptotically, the composition of the fund will evolve toward
(6.30a) () - 0) + 2K
(6.30b) K1(= 0)K
The total asymptotic cost of the fund will then be
(6.31) ~r [= l (1 - 3 ) + k2]
while monitoring the two risks separately would involve the cost
(6.32) F' = 1 10?,l + (1 -O)X2]K.
The cost of managing the merged trust fund is strictly below the
cost of monitoring each of the two funds separately. The intuition is
given in the text.
FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND 157
Notes
1. The problem of fiscal opportunism encompasses several aspects. Gov-
ernments whose time horizon is rarely far beyond the next election usually
enjoy extending contingent liabilities because they are not counted as a cur-
rent cost. The short time horizon also implies that governments tend to un-
clerrate the risk of their contingent liabilities and, particularly, make inadequate
provisions for future contingent pavments. Similarly, contingent liabilities
expose the government to a risk of default that is not necessarily a source of
concern for financial markets. A government's repudiation of its contingent
liabilities does not necessarily imply that it will default on other debts. There-
fore, under standard conditions no strong incentives are working against
government risk-taking and toward proper provisioning for risk. In this vol-
uime, issues of fiscal opportunism are discussed in more detail in the Intro-
duction and in Chapter 1 by Hana Polackova Brixi and Ashoka Mody and
Chapter 3 by Allen Schick.
2. I thank Hana Polackova Brixi for very helpful comments on an earlier
draft. Errors are mine.
References
Buiter, Willem. 1985. "A Guide to Private Sector Debt and Deficits." Eco-
nomic Policy (November): 14-79.
Cohen, Daniel. 1991. Private Lending to Sovereign States. Cambridge, Mass.:
MIT Press.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, F. H. Rogers, and L. Squire, eds.
World Bank Economists' Forum. Washington, D.C.: World Bank.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
to Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
CHAPTER 7
A Framework for Assessing
Fiscal Vulnerability
Richard Hemming
International Monetary Fund
Murray Petrie
The Economics and Strategy Group
FISCAL VULNERABILITY GOES BEYOND a situation in which a govern-
ment currently pursues inappropriate fiscal policies, or it lacks the
ability to implement better policies.' While obviously bad policies and
a lack of implementation capacity will almost always signal vulner-
ability, governments that seem well positioned in relation to both may
nonetheless find that they are vulnerable in two respects. First, under-
lying weaknesses may be present that do not affect fiscal outcomes
today but could at some time in the future prevent a government from
achieving its fiscal policy objectives. And, second, such weaknesses
may limit a government's ability to respond to future fiscal policy chal-
lenges, such as the need for fiscal consolidation as part of a coordi-
nated policy response to an external shock.
The focus in this chapter is on fiscal vulnerability from a macro-
economic perspective, and the suggested framework for assessing vul-
nerability highlights four macrofiscal aspects of vulnerability: (a)
incorrect specification of the initial fiscal position; (b) sensitivity of
short-term fiscal outcomes to risk; (c) threats to longer-term fiscal
sustainability; and (d) structural or institutional weaknesses affect-
ing the design and implementation of fiscal policy. The framework is
intended primarily to be a basis for identifying situations in which an
anticipatory response to potentially poor macrofiscal outcomes is
159
160 RICHARD HEMMING AND MURRAY PETRIE
required. But it also goes beyond this point, because a clear link
exists between fiscal vulnerability and economic vulnerability more
generally.2 This being the case, the importance of assessing fiscal
vulnerability also derives from its contribution to effective macro-
economic surveillance. The framework suggested in this chapter can
be viewed as providing guidance that will assist countries in assess-
ing their fiscal vulnerability.
Work on fiscal vulnerability has paralleled work on financial sector
vulnerability (see Downes, Marston, and Otker 1999). Strong links
exist between fiscal and financial sector vulnerability, because fiscal
vulnerability can manifest itself as a financial sector problem (such as
government preemption of bank lending on concessional terms), while
addressing a financial sector problem can be a source of fiscal vulner-
ability (such as government support for bank restructuring). A new
financial architecture should pay attention to both fiscal and financial
sector vulnerability, although the former has so far attracted much less
attention.
Reflecting the fact that a lack of transparency can be a major source
of vulnerability, there is a considerable overlap between assessments
of fiscal vulnerability and assessments of fiscal transparency. For this
reason, many of the issues discussed in this chapter are also covered by
the IMF's Code of Good Practices on Fiscal Transparency (the trans-
parency code) issued by the International Montetary Fund (IMF) and
taken up in more detail in IMF's Manual on Fiscal Transparency (the
transparency manual)-see IMF (2001). Indeed, it is inevitable that
vulnerability assessments will be informed by fiscal transparency as-
sessments included in Reports on the Observance of Standards and
Codes (ROSCs), for those countries where they are available, or un-
dertaken independently of the ROSC process. Transparency is also
important, because a prerequisite for a vulnerability assessment is a
reasonable degree of transparency. It would be difficult to assess the
vulnerability of a totally nontransparent fiscal system. This chapter
covers certain parts of the transparency material that are absolutely
central to vulnerability assessments, both in the sense of identifying
sources of vulnerability and in facilitating assessments.
The chapter is organized as follows. The next section specifies the
fiscal policy objectives that provide a benchmark against which vul-
nerability assessments are made. The third section discusses different
macrofiscal aspects of vulnerability and the methodology for vulner-
ability assessments. That section is followed by one that suggests vul-
nerability indicators and provides guidance on their interpretation,
and then by one that addresses some aspects of vulnerability that are
microstructural, as distinct from macrofiscal, in nature. The last sec-
tion offers concluding comments.
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 161
Fiscal Polic:y Objectives
Fiscal policy can be viewed as operating at three levels. The first is the
aggregate level, where the concern is with total expenditure and taxa-
tion (or the revenue effort more generally), the overall fiscal balance
and the associated deficit financi[g or use of fiscal surpluses, and the
fiscal consequences of accumulated liabilities and assets. The second is
the sectoral level, where there is a strategic focus on the broad struc-
ture of spending (across major programs) and revenue (mainly across
major tax bases). And the third is the program level, where the empha-
sis is on the microeconomic efficiency of individual spending and tax
programs.3
Fiscal vulnerability can manifest itself and can therefore be assessed
at any of these three levels. This chapter focuses on fiscal vulnerability
in a situation in which a government is exposed to the possibility of
failing to achieve its aggregate fiscal policy (or macrofiscal) objec-
tives. These objectives are:4
1. Foremost, a government should seek to avoid excessive fiscal
deficits and debt; they could directly threaten short-term macroeco-
nomic stability and longer-term fiscal sustainability.
2. A government should ensure that fiscal policy contributes to ef-
fective demand management by retaining sufficient flexibility to re-
spond in an appropriate and timely way to domestic and external
macroeconomic imbalances.
3. A government should raise revenue in a manner consistent with
maintaining reasonable and stable tax rates, which contribute to an
environment that encourages economic activity.5
Fiscal vulnerability could reflect a possible inability to meet any or
all of these macrofiscal objectives. Certainly each is important in its
own right. Thus a fiscal vulnerability assessment might suggest that,
given a government's expenditure plans: the money creation necessary
to finance the fiscal deficit may lead to inflation or a debt sustainability
problem may emerge; fiscal policy will have to be undesirably tight,
and maybe pro-cyclical, during an economic downturn; or tax rates
will have to increase over time to levels that are likely to have signifi-
cant disincentive effects. Any onie of these possible macrofiscal out-
comes would be a source of concern. However, in most circumstances
such outcomes will not be independent. So if deficits and debt are a
concern, providing a fiscal stimulus during a recession will usually be
costly given the high interest rate premia that are imposed when the
fiscal position is weak. And if the need instead is to contain a boom,
the scope to do so may be limit-ed if the room for tax increases has
been exhausted before deficits and debt become a concern.
162 RICHARD HEMMING AND MURRAY PETRIE
There are also clear interactions between macrofiscal objectives and
what might be called (to distinguish them from macrofiscal objectives)
the microstructural objectives of fiscal policy-that is, objectives set
at the sectoral and program levels. The causality can run both ways.
Weaknesses in the design and operation of spending and tax programs
can and do contribute directly to poor macrofiscal outcomes. At the
same time, if poor macrofiscal outcomes are possible, then it is likely
that some of a government's microstructural objectives will go unmet
as a consequence. There is also the possibility that the aggregate fiscal
policy approach could indicate that a country is not especially vulner-
able, yet at the same time the country is falling short of meeting, for
example, a high-priority poverty alleviation target. Clearly, such an
outcome can create vulnerability in the obvious sense that a govern-
ment that fails to meet such a critical equity goal may have to respond
in a manner that either compromises its macrofiscal objectives (if it
has to spend more on poverty alleviation programs and raise already
high taxes or borrow excessively to pay for the additional spending),
or, if there is little unproductive spending, it will have to cut some
other high-priority programs or shift some spending off-budget (which
simply shifts the source of vulnerability). If a government does neither
of these things, its political survival may be threatened.
Macrofiscal Aspects of Vulnerability
The starting point for meaningful analysis and discussion of fiscal vul-
nerability from an aggregate fiscal policy perspective is a clear view of
the initial fiscal position, both in terms of whether macrofiscal objec-
tives are initially being met and in terms of the quality of the informa-
tion that is available about the initial fiscal position. In particular, it is
important that the initial fiscal position describes the full range of
fiscal activity in the economy. The next step is to develop an under-
standing of the range of possible short-term macrofiscal outcomes by
assessing their sensitivity to underlying risk. The focus should then
shift to government exposure to medium- and longer-term adverse trends
or influences that may affect fiscal sustainability. Finally, attention
should be paid to vulnerabilities that arise from weaknesses in the
structure of public finances, the institutional capacity for fiscal man-
agement, and the broader effectiveness of government. These four as-
pects of fiscal vulnerability provide the organizational structure of the
framework for assessing vulnerability that is suggested in this chapter.
Because assessing vulnerability is a forward-looking endeavor, it
necessarily requires that a view is formed about future economic de-
velopments in general and future fiscal developments in particular. In
this connection, the position taken in this chapter is that fiscal vulner-
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 163
ability assessments need to be prudent, in the sense that they should be
prepared using a framework that has a bias toward downside risks. A
prudent approach can be justified by asymmetries in the economic
implications of unfavorable versus favorable outcomes, especially given
the weak initial fiscal positions or deficit bias in many countries., It is
also necessary to lean against a systematic tendency toward optimism
arising from elements of the political environment, such as the short
time horizon of elected officials.
*The Initial Fiscal Position
The initial fiscal position is clearly a source of vulnerability to the
extent that the macrofiscal objectives described above are initially not
met. Thus if the starting position is characterized by high deficits and
debt, an inability to respond to macroeconomic imbalances (for ex-
ample, because automatic stabilizers are small and discretionary fiscal
policies take time to formulate and implement), or very high tax rates,
then there would usually be a presumption of vulnerability. However,
the concern here is more with the possibility that the available fiscal
information may not relate to the full range of fiscal activity that is
undertaken by or on behalf of the government.
It would typically be the case that a description of the initial fiscal
position takes the latest central government budget as its starting point,
and this is reasonable if the budget is realistic. If the budget is not
realistic, the estimated outturn for the preceding year should be the
starting point. If a vulnerability assessment is being made some time
after the budget has been formulated, it would be appropriate to begin
with an estimate of the budget outturn rather than the budget or the
preceding year's outturn. This would allow several factors to be taken
into account, including: new data for the outturn for the preceding
year that render a budget based on an earlier estimate unrealistic; re-
vised macroeconomic forecasts affecting the assumptions underlying
the budget; and other fiscal policy developments (new programs, offi-
cial statements about budgetary policy, and so forth).
From such a starting point, it is crucial to go beyond the budget,
because it is most unusual that the budget captures all fiscal activity:
* Where lower levels of government are large, focus on the general
government.7
* Include extrabudgetary activities of the central government and
lower levels of government.
* Cover quasi-fiscal activities undertaken outside government. The
central bank, public financial institutions, and nonfinancial public
enterprises to varying degrees are all involved in such activities, which
tend to generate implicit or explicit obligations for the government. In
164 RICHARD HEMMING AND MURRAY PETRIE
many cases, precise quantification may be difficult. A qualitative state-
ment about quasi-fiscal activities would in the first instance suffice,
but rough orders of magnitude should be provided for the main quasi-
fiscal activities and the need for better information should be empha-
sized. Where quasi-fiscal activities are significant, the focus should
shift to the public sector.
Another problem is that fiscal activity is often measured in an unre-
liable or incomplete manner because of weak accounting and control
systems. With poor fiscal data, there will often be large discrepancies
in the fiscal accounts, for example between above-the-line and below-
the-line fiscal balances. The cash accounting traditionally used by gov-
ernments, while having a number of advantages, also has inherent
weaknesses as a measure of fiscal activity. Most notably, it fails to
reflect activities that give rise to arrears and noncash-based provisions
to clear arrears (for example, netting of expenditure arrears and tax
offsets).
It is important to take into account changes in the stock of govern-
ment liabilities and assets. Knowing a government's gross debt is a
minimum requirement for longer-term sustainability analysis, while
information on the structure of debt (that is, maturity, fixed versus
variable interest rates, and currency composition) is needed to assess
short-term fiscal risk. If a government has sizable financial assets, its
net financial debt is more relevant than its gross debt to longer-term
sustainability. And where available, even partial information on other
asset transactions (most usefully on privatization and government in-
vestment in productive assets) would provide some basis for determin-
ing whether the effects of fiscal policy actions that show up as changes
in aggregate revenue and expenditure, and the resulting changes in the
fiscal deficit, are being accompanied by changes on the government's
balance sheet that might undo their impact.8
Explicit and implicit contingent liabilities also should be covered.
The Fiscal Risk Matrix in Chapter 1 of this volume outlines the classi-
fication and examples of such government obligations. Where there is
provisioning against contingent liabilities-for example, a deposit in-
surance scheme-the level of provisioning should be noted and the
focus should be on the uncovered contingent liability. Again, the quan-
tification of contingent liabilities may be difficult. If so, they should be
handled in the same way as quasi-fiscal activities.
Finally, the difficulty in interpreting fiscal balances can be a source
of vulnerability. This difficulty derives in part from the extent of fiscal
activity to which the fiscal balance refers. But it also relates to the way
in which the fiscal balance is measured. Because assessing vulnerabil-
ity will necessarily become a comparative exercise-assessments for
some countries will inform assessments for other countries-there are
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 165
advantages to using an internationally comparable measure of the fis-
cal balance (and other fiscal aggregates) in all countries. The overall
balance, appropriately adjusted in response to its most obvious short-
comings (for example, to reflect expenditure arrears, the use of privati-
zation proceeds), is best suited to this purpose. However, the overall
balance should be supplemented by other fiscal balance measures where
they provide a better indicator cf the stance of fiscal policy from a
macroeconomic perspective.9
,Short-term Risks
Fiscal outcomes are exposed to variations in key underlying assump-
tions and other factors. (Chapter 5 by Ramaswamy in this volume
addresses this fact by applying a factor model.) To assess the sensitiv-
ity of fiscal outcomes, it is necessary to move beyond the initial fiscal
position, because the one-year time horizon of a budget, and by impli-
cation of the initial fiscal position, does not do justice to the full extent
of the short-term risks to which fiscal outcomes are exposed. For this
reason, the initial fiscal position should be accompanied by a short-
term forecast that looks at least two years ahead. The short-term fore-
cast should be based on unchanged policies, in the sense that policy
intentions that have been announced but not implemented should be
excluded, and it should be purged of temporary measures affecting the
initial fiscal position. This forecast should be fairly detailed, but possi-
bly less so than the description of the initial fiscal position. Both the
initial fiscal position, which typically itself has a forward-looking com-
ponent, and the short-term forecast should then be subjected to sensi-
tivity analysis.
The principal short-term risks that need to be addressed are the
following:10
* The initial fiscal position and the short-term forecast are sensi-
tive to changes in macroeconomic variables and other sources of eco-
nomic risk. Unanticipated changes in growth of the gross domestic
product (GDP), unemployment, inflation, interest rates, external trade,
capital flows, the exchange rate, and other macroeconomic variables
affecting macrofiscal outcomes give rise to forecasting risk affecting
revenue, expenditure, and financing. The structure of debt is impor-
tant in this regard, because it affects the fiscal risk associated with
short-term movements in interest rates and the exchange rate. Rev-
enue and expenditure are also subject to other risks affecting the tax
base (such as corporate profitability), nontax revenue (such as mineral
prices), and spending programs (such as government wage increases).
* It is possible that contingent liabilities will be called when no
budget provision has been made to meet them. The appropriate way
166 RICHARD HEMMING AND MURRAY PETRIE
to provide for contingent liabilities, and the practicalities of doing so,
raise issues that have resulted in limited provisioning (see Chapter 2
by Petrie in this volume for further discussion). Implicit liabilities will
always entail more substantial risk, since provisioning is usually judged
inappropriate because of moral hazard problems.
* There may be a lack of clarity about the size of specific expendi-
ture commitments in that provision may be made in the budget for
spending on an activity (such as bank restructuring), but there is less
than the usual precision about the cost implications of that activity.
* Finally, some fiscal policies may be defined imprecisely. This would
be the case where a government announces a policy intention (such as
providing incentives for saving and investment) but either the details
of the way in which the policy is to be implemented are not well devel-
oped or the implications of an announced method of implementation
are unclear.
Fiscal analysis should therefore involve an examination of the range
of possible short-term macrofiscal outcomes, with a focus on varia-
tions in underlying assumptions and other parameters to which a prob-
ability or likelihood of different events can be attached, albeit in some
cases only approximately. However, while the realization of a typical
risk might signal the need for policy adjustment, it would not neces-
sarily imply vulnerability because the consequences may be easily
accommodated. This being the case, there is merit in subjecting a short-
term forecast to more aggressive stress testing, especially when there are
reasonable grounds to consider that a substantial shock to the economy,
be it global, regional, or country-specific, is more than a remote short-
term possibility. This situation is addressed in the next section in the
context of stress testing a baseline medium-term projection."
Longer-term Fiscal Sustainability
Even if fiscal outcomes are not exposed to significant short-term risks,
running persistent fiscal deficits may result in debt levels that become
a source of fiscal and broader macroeconomic difficulties over the
medium term. The standard debt dynamics analysis is the usual basis
for identifying the impact of deficits on indebtedness, and more gener-
ally for assessing the implications of past and current fiscal policies for
longer-term sustainability. Where available, market-based indicators
(for example, a government debt rating or interest rate premia) can
supplement debt dynamics analysis.'2
The starting point for an assessment of longer-term sustainability is
a baseline, medium-term fiscal projection, which typically would look
at least five years ahead. It should assume a continuation of current
policies, which will often require difficult judgments as to what is and
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 167
what is not a current policy.'3 Alt'hough it encompasses the initial fis-
cal position and the short-term forecast, such a baseline does not need
to be as detailed. To assess vulnerability, the baseline medium-term
projection should be supplemented by alternative scenarios. These sce-
narlos should illustrate the responsiveness of the medium-term fiscal
outlook to different specifications of the initial fiscal position and short-
term forecasts. In particular, it should be recognized that short-term
risks may persist, and some could be greater, in the medium term. For
example, certain contingent liabilities, such as pension guarantees, are
probably a bigger source of medium-term risk than of short-term risk.
Stress testing would then be used to assess the impact of short-
term and medium-term shocks (for example, a global interest rate or
business cycle shock, a regional or country-specific reversal of mar-
ket sentiment, or a sharp deterioration in the terms of trade). Stress
testing should identify how the fiscal outlook would change under
circumstances ranging up to the fairly extreme and shed light on why
the outlook changes in the way it does.'4 It is therefore important to
identify the key transmission mechanisms through which the main
fiscal aggregates are affected. MAoreover, the likely correlation be-
tween different transmission mechanisms should be recognized. For
example, a major macroeconomic shock such as an output collapse
may cause revenue to fall, expertditure to increase (for example, on
the social safety net), and contingent liabilities to come home to roost,
all at the same time. A downside scenario should envisage some of
the worst things that can happen (for example, full-blown crises)."5
In this way, stress testing, in combination with identification of the
policy responses that reduce vulnerability, provides the basis for sys-
tematic fiscal contingency planning, which has to be a key outcome
of effective surveillance.
Long-term projections and scenarios are a natural extension of
medium-term analysis. However, they can be even less detailed given
their more speculative nature. It is particularly important to take into
account long-term expenditure pressures. The impact of demographic
developments on pensions and health spending is an obvious source of
such pressure in many countries, The possible exhaustion of a natural
resource that generates substantial revenue, and any associated envi-
ronmental degradation, will also be relevant to long-term fiscal
sustainability in some countries.
Structural Weaknesses
The composition of expenditure and revenue is important in assessing
vulnerability. A principal source of vulnerability is a high proportion
of nondiscretionary spending to total spending, which limits a
government's flexibility to adjust spending levels downward when it
168 RICHARD HEMMING AND MURRAY PETRIE
needs to do so. Nondiscretionary spending is that for which a govern-
ment is under a legal or other strong obligation to meet. The most
notable examples are interest payments, formula-based transfers to
lower levels of government, and public pensions.16 However, the com-
ponents will vary from country to country, and classifying all spending
as either nondiscretionary or discretionary may require difficult judg-
ments. In many countries, the distinction may boil down to that be-
tween spending on transfers (broadly defined) and spending on goods
and services.
Some large expenditure items are a source of vulnerability, because
they are resistant to adjustment given the powerful interest groups
they serve. Military spending is a case in point, although a large gov-
ernment wage bill may be every bit as entrenched where public sector
unions are strong or the government is an employer of last resort.
There also may be latent expenditure needs that do not manifest them-
selves until triggered by a shock or discontinuity of some kind. Any
significant gap in expenditure, compared with established norms, could
be exposed in this way. For example, the need for a social safety net
may become apparent only after an economic crisis, but once in place
it will almost certainly become permanent.
A good tax structure is one in which revenue derives from a range
of taxes with broad bases, ideally large macroeconomic aggregates
(that is, wages, profits, and consumption, including imports of
consumables). Not only will such a structure tend to result in reason-
able tax rates, but it also will ensure a moderately elastic tax system,
which is desirable from the point of view of facilitating countercyclical
fiscal policy through the operation of automatic stabilizers. A revenue
composition dominated by just one or two taxes, especially if they
have narrow bases, is a source of vulnerability, both in terms of in-
creasing a government's exposure to unexpected fiscal developments
because revenue from just a few taxes is likely to be volatile (for ex-
ample, trade tax revenue is highly sensitive to exchange rate changes),
and in terms of limiting its capacity to respond when the need arises
because tax rates probably have to be very high. Frequent tax law
changes, especially when they result in more exemptions, tax holi-
days, and other relief, as is often the case, can add to vulnerability by
progressively undermining the tax base. A government's capacity to
respond is also constrained by extensive earmarking, which limits the
scope for discretionary tax changes. Finally, a heavy reliance on nontax
revenue, the main sources of which (grants, royalties, privatization
proceeds, central bank profits) may not be stable or particularly re-
sponsive to policy intervention, also can contribute to vulnerability.
The institutional capacity for fiscal management is a major deter-
minant of fiscal vulnerability. Numerous aspects of the institutional
capacity for fiscal management could be relevant to fiscal vulnerabil-
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 169
ity-most notably, the administrative capability for expenditure man-
agement and revenue collection, the roles and responsibilities of gov-
ernment and within government, the public availability of information,
the budget process, and the integrity of fiscal information collection.
The emphasis should be on particular weaknesses that can signal that
poor macrofiscal outcomes are possible (for example, expenditure and
tax arrears, ineffective audit procedures).
Finally, a government that has a general reputation for being inef-
fective will usually be vulnerabi'e. Thus a government that gets in-
volved in too many activities that should be left to the private sector,
whose agents (that is, public servants, public enterprise managers) have
a relationship with politicians that is inconsistent with the arm's-length
principle, or that is characterized by extensive nontransparency and
corruption, cannot be expected to meet its macrofiscal objectives on a
consistent basis.
Clearly, the four aspects of vulnerability just highlighted are closely
related. Obviously, an incorrect specification of the initial fiscal posi-
tion makes it difficult to assess both short-term risks and longer-term
sustainability. Moreover, a chronic misspecification of the initial fiscal
position is likely to be a manifestation of weak institutional capacity
for fiscal management. But some of the interactions are subtler. For
example, weak fiscal institutions can act to amplify rather than dampen
macroeconomic volatility, as would be the case when a country pur-
sues pro-cyclical fiscal policy because an inability to save fiscal re-
sources generated by a buoyant economy-reflecting political pressures
for such savings to be spent-leads to a pattern of tax cuts during
expansions and tax increases during recessions.17
Vulnerability Indicators
There are clear gains in convenience if the results of the analysis of the
initial fiscal position, short-term risks, longer-term sustainability, and
structural weaknesses could be summarized in a few key vulnerability
indicators, and the preceding discussion suggests some obvious candi-
dates. However, selecting a set of vulnerability indicators involves trade-
offs. While the inclusion of a large number of indicators increases the
probability that fiscal vulnerability will be identified, they may make
it difficult to identify the main sources of vulnerability. A large num-
ber of indicators also will tend to increase the information require-
ments of vulnerability assessments, which would create difficulty in
ensuring comparability across countries and generally make assess-
ments less manageable. In the end, then, a set of vulnerability indica-
tors that is fairly parsimonious is needed.
The set of indicators described in general terms in Box 7.1 and in
more detail in the Appendix focuses on the aspects of vulnerability
170 RICHARD HEMMING AND MURRAY PETRIE
Box 7.1 A Possible Set of Fiscal Vulnerability Indicators
* Fiscal position indicators-weak initial fiscal position; incomplete
coverage of government fiscal activity; poor accounting and control; in-
sufficient balance sheet information; sizable uncovered contingent liabili-
ties; and significant quasi-fiscal activities.
* Short-term risk indicators-high sensitivity of short-term fiscal out-
comes to changes in key macroeconomic variables; inappropriate debt
structure; variable revenue sources and expenditure programs; calling of
uncovered contingent liabilities; and other expenditure risks.
* Longer-term sustainability indicators-unfavorable debt dynamics;
low government debt rating or high interest rate premia; adverse demo-
graphic trends; and rapid natural resource depletion or serious environ-
mental degradation.
* Expenditure indicators-large share of nondiscretionary spending
or transfers; excessive military spending; and significant gaps in expendi-
tures (for example, social security, safety net, health and education, infra-
structure).
* Revenue indicators-inelastic revenue system; highly concentrated
tax revenue; frequent tax law changes; extensive earmarking; and reli-
ance on grants and other major nontax revenue sources.
* Fiscal management indicators-large expenditure arrears and use
of netting arrangements; marked deviation between the original budget
and the budget outturn; nonexistent or weak medium-term budget plan-
ning; long delays in preparing and auditing final accounts; large tax ar-
rears and use of tax offsets; a large stock of tax refunds, especially for the
value added tax (VAT); an out-of-date taxpayer register; and an ineffec-
tive tax audit program.
* Government effectiveness indicators-poor results from surveys of
public sector performance, corruption, and so forth.
discussed in this chapter and avoids reference to features of the fiscal
system that are unrelated to vulnerability at the aggregate level. Even
so, for a number of reasons these indicators should still be regarded as
provisional. First, the acid test of their suitability will come only when
they are implemented in the context of actual vulnerability assessments.
Country experience may suggest that there are more useful alterna-
tives to the suggested indicators. Second, country experience may also
suggest a need for additional indicators. Political factors, for example,
are not included, despite the fact that a specific political event, such as
an approaching election, may generate an unusual amount of uncer-
tainty about short-term fiscal policy in many countries. There also
may be a case for looking at characteristics of the fiscal management
system that give rise to vulnerabilities beyond those directly related to
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 171
aggregate fiscal policy, because indirect effects (for example, where
macrofiscal outcomes are affected by the need to meet microstructural
objectives) will often be important. Finally, the scope for greater use of
survey-based indicators that capture important supplementary infor-
mation bearing on fiscal vulnerability should be explored."8
The fiscal vulnerability indicators can be used in a range of ways.
At the least demanding extreme, they can be viewed as no more than a
checklist that someone assessing fiscal vulnerability could use to keep
track of whether he or she has covered all the relevant aspects of vul-
nerability. At the other much more demanding extreme, it is easy to
envisage them as a stepping-stone on the way to the type of systematic
assessment that could ultimately produce an index of vulnerability. In
the first instance, the indicators should probably be used more as a
checklist, albeit an evolving one as the indicators are modified in re-
sponse to experience with their use. But once a final set of indicators
begins to emerge, they can be used in a more ambitious way, perhaps
with the objective of developing a small number of broad categories of
vulnerability. In this connection, however, it will be necessary to as-
sign the indicators priority, to take account of differences in fiscal in-
stitutions across countries, and perhaps to develop indicators relevant
to particular groups of countries (or weight indicators differently across
groups of countries).'9
In whatever form they are used, fiscal vulnerability indicators have
to be interpreted with caution. It has already been noted that indica-
tors of short-term fiscal risk may point to the need for a relatively
small fiscal correction within the context of an otherwise strong un-
derlying fiscal position. It is important to distinguish this from a situ-
ation in which the underlying position is also weak.20 Although it will
often be the case that a high level of exposure to short-term fiscal risk
will be a leading indicator of longer-term vulnerability, it is important
that fiscal vulnerability assessments avoid judgments based only on a
few indicators.
The policy significance of a similar level of fiscal vulnerability also
will vary across countries, depending on the broader economic con-
text in which fiscal policy is placed. For example, evidence of short-
term risk may be a serious concern in a country with a currency board
or a fixed exchange rate, where the scope for discretionary monetary
policy is limited. Government deficits and debt also may be more of a
concern in a country with relatively low national savings and a high
external debt, or with high inflation and an underdeveloped financial
sector. And fiscal vulnerability in a country where an external crisis
could have contagion effects should warrant particular attention. Fi-
nally, a given level of fiscal vulnerability also may be a cause for greater
concern if combined with a low level of adherence to a broader set of
standards related to government performance.
172 RICHARD HEMMING AND MURRAY PETRIE
Microstructural Aspects of Vulnerability
This chapter has focused so far on1 a government's aggregate fiscal
policy objectives. As noted, however, a government has to pursue
other microstructural objectives, and failure to meet such objectives
can create vulnerability just as surely as failure to meet macrofiscal
objectives. A government can have a wide range of microstructural
objectives. Abedian (2000) focuses on those microstructural objec-
tives that are most closely linked to a government's socioeconomic
legitimacy. Equity goals related to poverty alleviation, income redis-
tribution, and improvements in other indicators of human develop-
ment fit into this category. So too do many of the efficiency goals
that traditionally (that is, in a welfare economics sense) justify gov-
ernment intervention in the economy, such as providing public goods
and merit goods, taking account of the external costs and benefits
associated with private decisions, and addressing other aspects of
market failure.
Two microstructural aspects of vulnerability are discussed in Abedian
(2000). The first aspect is those undesirable features of the environ-
ment supporting government activity that can lead to government (or
bureaucratic) failure. This failure in turn manifests itself in the pursuit
of inappropriate equity and efficiency goals. Weaknesses in this area
extend beyond those noted above as relevant to macrofiscal vulner-
ability and cover such things as the political process, the public sector
culture or ethos, and the general approach to governance. The second
aspect is the fiscal management system, where weaknesses that again
extend beyond those noted above result in delivery failure, that is a
large gap between a government's equity and efficiency goals and what
it actually achieves.21
Concluding Comments
The aim of fiscal vulnerability assessments is to identify those features
of a country's fiscal system that compromise the ability of the govern-
ment to meet its aggregate fiscal policy objectives. They also provide a
basis for managing vulnerability in order to limit the government's
exposure to possible adverse outcomes, in particular by enhancing its
ability to respond to fiscal and broader economic developments. Man-
aging vulnerability requires a preemptive effort to address potential
problems revealed by vulnerability assessments.
It is primarily the organizational framework that vulnerability as-
sessments provide that is new. However, the true value added will not
become clear until the framework is implemented. In particular, the
vulnerability indicators will almost definitely need fine-tuning and
possibly more extensive reconsideration.
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 173
Finally, although this chapter has discussed government liabilities
and assets in terms of the importance of having information on their
levels and structure as indicators of vulnerability, it has not discussed a
government strategy for their management. Yet, as Chapter 1 by Hana
Polackova Brixi and Ashoka Mody indicated, government liabilities
and assets can be managed in a way to contain vulnerability by reduc-
ing government risk exposure. The sophistication of a government's
risk management strategy may not be done justice by the indicators
suggested above. More work is rneeded to offer practical guidance on
optimal risk-reduction strategies.22
Annex 7.1:
Measurement of Fiscal Vulnerability Indicators
Indicators Measures
Fiscal position indicators
Weak initial fiscal Overall fiscal balance as a share of GDP
position Other fiscal balance measures as a share of GDP
(where relevant)
Net financial debt as a share of GDP
Size of aatomatic stabilizers (small/average/large)
Average and maximum rates of tax (for each
main tax)
Incomplete coverage Revenue covered in fiscal data as a share of
of government fiscal general government revenue
activity Expenditure covered by fiscal data as a share
of general government expenditure
Poor accounting and Fiscal balance measured from above-the-line
control relative to the fiscal balance measured from
below-the-line
Insufficient balance sheet Gross debt
information Net financial debt
Other balance sheet data
Sizable uncovered Gross contingent liabilities as a share of total
contingent liabilities revenue
Net contingent liabilities as a share of total revenue
Or
Descriptrion of main contingent liabilities and
quantification of largest net contingent liabilities
Significant quasi-fiscal Quasi-fiscal activities as a share of total revenue
activities Or
Description of main quasi-fiscal activities and
quantification of largest quasi-fiscal activities
(Table continues on the following pages.)
174 RICHARD HEMMING AND MURRAY PETRIE
Indicators Measures
Short-term fiscal risk indicators
High sensitivity of short- Impact of variations in forecasted GDP growth,
term fiscal outcomes to inflation, balance of payments, exchange rate,
changes in key macro- and interest rates on the fiscal balance
economic variables
Inappropriate debt Maturity (short, medium, and long term), interest
structure rate structure (fixed versus variable rates), and
currency composition of debt
Variable revenue sources Impact of variations in other economic and
and expenditure noneconomic determinants of revenue and
programs expenditure on the fiscal balance
Calling of uncovered Net contingent liabilities as a share of GDP;
contingent liabilities expected payments in connection with
guarantees and other contingent liabilities
Other expenditure risks Description of programs and policies that give
rise to risks
Longer-term sustainability indicators
Unfavorable debt 5-10 year projection of gross or net debt as a share
dynamics of GDP, and change in the primary balance as a
share of GDP required to stabilize the debt ratio
at the current level or at a specific target level
Low government debt The Bloomberg website
rating and/or high provides information on how to calculate
interest rate premia interest rate premia.
Adverse demographic Long-term projection of retirement age and
trends school-age population relative to total and
working population; impact on expenditure as
a share of GDP and on tax rates
Rapid resource depletion Years of usable reserves at current exploitation
rate; resource-related revenue as a share of total
revenue; resource-related financial assets as a
share of GDP; serious environmental
degradation
Expenditure indicators
Large share of nondiscre- Nondiscretionary spending and transfers as a
tionary spending and/ share of GDP
or transfers
Excessive military spending Military spending as a share of GDP
Significant gaps in Programs for which spending as a share of GDP
expenditure is significantly below the average for compa-
rable countries
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 175
Indicators Measures
Revenue indicators
Inelastic revenue system Tax elasticity or buoyancy
Highly concentrated Revenue composition, particularly trade tax
tax revenue revenue as a share of total tax revenue
Frequent tax law changes Major tax changes, especially new exemptions and
other reliefs, every year or every two years
Extensive earmarking Revenue from earmarked taxes as a share of total
revenue
Reliance on grants and Nontax revenue as a share of total revenue;
other unstable nontax compcsition of nontax revenue
revenue sources
Fiscal management indicators
Large expenditure arrears Expenditure arrears as a share of total revenue;
and use of netting significant netting of arrears; inability
arrangements to repDrt on sizeable arrears
Marked deviation between Expendit:ure outturn relative to original expendi-
the original budget and ture; resort to large supplementary budgets
the budget outturn
Nonexistent or weak Effective medium-term budget planning
medium-term budget
planning
Long delays in preparing Length of time between end of fiscal year and (a)
and auditing final preparation of final accounts and (b) release
accounts of audited accounts
Large tax arrears and Tax arrears as a share of total revenue; sharp
use of tax offsets increase in tax arrears; significant tax
offsets
A large stock of tax Stock ol: tax/VAT refunds as a share of tax/VAT
refunds, especially for revenue
value added tax (VAT)
Out-of-date taxpayer Currentiess of taxpayer register by main tax
register
An ineffective tax audit Coverage of tax audit; targeting of tax audit
program
Government effectiveness indicators
Poor results from surveys Information available from the Institute for
of public sector perfor- Management Development's World Com-
mance, corruption, and petitiveness Yearbook; the Transparency
so forth International's Corruption Perceptions Index
176 RICHARD HEMMING AND MURRAY PETRIE
Notes
1. This chaper was originally issued as IMF Working Paper WP/00/52 by the
International Monetary Fund (IMF), Washington, D.C. It draws on a companion
paper by Iraj Abedian and has been prepared with significant input from William
Allan, John Crotty, and Steve Symansky. It also has benefited from comments
made by many colleagues in the Fiscal Affairs Department at the IMF and by
seminar participants at the European Central Bank and the World Bank.
2. This is illustrated by recent work that points to variables related to fiscal
imbalance-for example, high domestic credit growth and large current account
deficits-as being among the strongest predictors of an external crisis (IMF 1999).
3. The World Bank (1998) refers to these as the level-1, level-2, and level-3
operations of fiscal policy.
4. In addition to these general objectives, a country may have specific fiscal
policy objectives that the government has set for itself (for example, as reflected
in a fiscal rule) or that may have been agreed on with others (for example, as
part of IMF conditionality).
5. Stability of tax rates is also justified by the fact that the distortionary cost
of taxation is reduced by smoothing tax rates over time. Tax smoothing is also
consistent with countercyclical fiscal policy.
6. Prudence is consistent with the usual approach in accounting, where fi-
nancial statements consistently err on the side of caution in recording events or
transactions that are likely to have a favorable impact, while being less cautious
when the results are likely to be unfavorable.
7. Although lower levels of government that set their own objectives and
are subject to market discipline can be viewed as independent of central govern-
ment from an economic perspective, from a vulnerability perspective consolida-
tion is desirable (although vulnerability could be assessed independently for lower
levels of government).
8. Easterly (1999) finds evidence from an analysis of countries borrowing
from the IMF and the World Bank and of European countries covered by the
Maastricht Treaty that fiscal adjustment often takes the form of privatization
and cuts in government investment, so that changes in reported fiscal deficits
represent illusory rather than real adjustment because there is an offsetting bal-
ance sheet transaction.
9. Depending on a country's circumstances, other such measures might in-
clude the structural balance, the operational balance, the primary balance, or
the augmented balance.
10. These risks are the same as those discussed in the fiscal transparency
manual, because a requirement of the fiscal transparency code is that all govern-
ments should publish a fiscal risk statement with the annual budget.
11. However, if the focus of a vulnerability assessment is on short-term fiscal
outcomes alone, stress testing should be part of such an assessment.
12. They cannot, however, substitute for such analysis, or for vulnerability
assessments more generally, because there is little evidence that either debt rat-
ings or interest rate premia adequately reflect fiscal sustainability. They are in-
A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY 177
fluenced more by external sustainabihty, level of development, and the depth of
the market for a country's debt.
13. The difficulty is illustrated by the discussion of fiscal policy in Asia in
IMF's World Economic Outlook, October 1998 (IMF 1998), where, in distin-
guishing discretionary from nondiscretionary fiscal measures, it had to be de-
cided whether holding nominal spending constant when spending had in the
past increased represents changed or unchanged policy.
14. Stress testing clearly goes beyond the usual scenario analysis, which in
the fiscal area tends to involve producing higher-growth and lower-growth sce-
narios to illustrate the benefits of stronger fiscal policies and the costs of weaker
fiscal policies than in the baseline.
15. There is a parallel here with risk assessment in the private sector. Stan-
dard value-at-risk methodologies used in financial analysis show how much a
bank or firm could potentially lose over a specified time period for likely market
movements. Stress testing is used to assess and manage extreme risks.
16. Not all nondiscretionary spending is necessarily a problem. For example,
spending on unemployment compensation is cyclically sensitive. It therefore acts
as an automatic stabilizer during a cyclical downturn, reducing the need for
discretionary fiscal policy.
17. Talvi and Vegh (2000) find that fiscal policy in developing countries is
for such a reason highly pro-cyclical. They suggest that attention be paid to
designing fiscal arrangements (such as stabilization funds) aimed at ensuring
that fiscal savings generated during good times is saved for bad times.
18. The weakness of survey-based indicators is that surveys fail to reflect the
strength with which views are held and thus the weight of opinion. However,
they can incorporate information frcm a wider range of sources.
19. In this connection, Bird and Banta (1999) suggest indicators for transi-
tion economies that take account of their special circumstances.
20. This is analogous to the differentiation often applied to companies to
distinguish those with short-term cash flow problems but positive net worth
from those with negative net worth.
21. These sources of microstructural vulnerability also affect what Tanzi
(2000) refers to as the quality of the public sector.
22. For some discussion of the issues in this area, see Skilling (1999).
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Abedian, Iraj. 2000. "Micro-Structural Aspects of Fiscal Vulnerability." Inter-
national Monetary Fund, Washington, D.C. Processed.
Bird, Richard M., and Susan M. Ba:nta. 1999. "Fiscal Sustainability and Fiscal
Indicators in Transitional Countries." Paper presented at the USAID Confer-
ence on Fiscal Reform and Sustainability, Istanbul, June.
178 RICHARD HEMMING AND MURRAY PETRIE
Downes, Patrick T., Dewitt D. Marston, and Inci Otker. 1999. "Mapping Fi-
nancial Sector Vulnerability in a Non-Crisis Country." IMF Policy Discus-
sion Paper PDP/99/4. International Monetary Fund, Washington, D.C.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy 28 (April): 57-86.
IMF (International Monetary Fund). 1998. World Economic Outlook, Octo-
ber 1998. World Economic and Financial Surveys. Washington, D.C.
. 1999. World Economic Outlook, October 1999. World Economic
and Financial Surveys. Washington, D.C.
IMF (International Monetary Fund), Fiscal Affairs Department. 2001. "Code
of Good Practices on Fiscal Transparency" and "Manual on Fiscal Trans-
parency" .
Skilling, David. 1999. "How Should Governments Invest Financial Assets and
Manage Debt?" Paper presented at the 20th Annual Meeting of Senior Bud-
get Officers, Organisation for Economic Co-operation and Development,
Paris.
Talvi, Ernesto, and Carlos A. Vegh. 2000. "Tax Base Variability and Procyclical
Fiscal Policy." NBER Working Paper 7499. National Bureau of Economic
Research, Cambridge, Mass.
Tanzi, V. 2000. "The Quality of the Public Sector." International Monetary
Fund, Washington, D.C. Processed.
World Bank. 1998. Public Expenditure Management Handbook. Washington,
D.C.
Country Examples
CHAPTER 8
Evaluating Government Net
Worth in Colombia and Republica
Bolivariana de Venezuela
William Easterly
and Davicd Yuravlivker
World Bank
ONE OF THE MOST IMPORTANT questions to ask about public finance
in any given country is: Does the government's balance sheet contain
assets (treasures), such as oil or valuable public enterprises, that can
help the government meet its long-run budget constraints? Or does the
government's balance sheet contain large implicit liabilities (time
bombs), such as a pension system temporarily in surplus but headed
for big future deficits? Even if the government's balance sheet contains
treasures, are those treasures being depleted to finance current con-
sumption, at the expense of futuLre generations?
The conventional approach to assessing the sustainability of a given
fiscal deficit is to compare it with the deficit necessary to keep the
public debt-to-GDP (gross domestic product) ratio constant. This is
justified by the reasoning that if a country is not already in a debt
crisis, then keeping the public debt ratio constant will be sufficient to
avoid a debt crisis in the future.'
The main objectives of this chapter are: (a) to present the assets/
liabilities/net worth approach to evaluating the true financial position
of the public sector in order to achieve a better assessment of longer-
term fiscal sustainability; and (b) to present applications of this ap-
proach to the cases of Colombia and R6publica Bolivariana de Venezuela
in order to illustrate its use and to help their authorities in the process
of fiscal adjustment. In particular, this study could help ensure that the
181
182 EASTERLY AND YURAVLIVKER
fiscal measures that governments are implementing to address present
fiscal gaps are coherent and consistent with achieving longer-term fis-
cal sustainability. In addition, the conceptual framework developed
here and its application to these two country cases could be useful in
analyzing similar problems in other countries.
The next section of this chapter presents the assets/liabilities/net
worth conceptual framework and methodology. The final section sum-
marizes the main findings of the chapter and offers some concluding
remarks.
Assets/Liabilities/Net Worth Approach
Conceptual Framework
The limitations of conventional sustainability analysis call for broad-
ening the sustainability analysis to consider the evolution of the
government's complete balance sheet (Table 8.1).2
Which components of the government's fiscal policy package improve
the government's net financial position and which ones worsen it?
Anyone considering the longer-term sustainability of a government's
finances would find it helpful to think first about the balance sheet
that the government must manage.3 Although one cannot always mea-
sure all of the items in this balance sheet, the balance sheet approach is
a useful way to think of which government actions imply progress or
regress toward sustainability.4 Investing in a project with an economic
rate of return higher than the discount rate, for example, improves the
government's net financial position. (The government has to devise a
way to capture the revenues that accrue to the society from this project,
however.) Forward-looking individuals and firms (and credit rating
agencies) will react more favorably to a fiscal adjustment that pre-
serves high rate of return activities while cutting consumption than to
a myopic package that cuts projects with high returns just to improve
this year's fiscal balance.
Note that in Table 8.1 the present values of government consump-
tion or government tax revenue do not appear in the balance sheet.
The government's intertemporal budget constraint is that the present
value of government consumption minus government revenues be less
than or equal to the government's net worth (more on the daunting
task of implementing this in a moment). If the government's inter-
temporal budget constraint is violated under current revenue and ex-
penditure plans, then fiscal sustainability does not hold. The future
fiscal adjustment required to satisfy the intertemporal budget constraint
is a measure of the distance from sustainability in government finances.
A more sophisticated analysis would refer to a desired net worth level,
EVALUATING GOVERNMENT NET WORTH 183
Table 8.1 Government Balance Sheet for Fiscal Sustainability
Assessment
Assets Liabilities
Government-owned public goods (such as Public external debt
infrastructure, schools, and health clinics
that generate an adequate econoniic rate
of return and an indirect financial rate of
return through tax collection)
Government-owned capital that is finan- Public domestic debt
cially profitable (anything for which gov-
ernment can charge user fees to generate
an adequate financial rate of return)
Value of government-owned naturall re- Contingent liabilities (for
source stocks (such as oil and minerals) example, bank deposit
guarantees, net present
value of pension scheme,
guarantees of private debt)
Expected net present value of loans to pri- Government's net
vate sector and other financial assets worth
Source: The authors.
but that would depend on the social preferences of each particular
society, and it goes beyond the scope of this chapter.
Methodology: Applying the Concept
Sustainability can be evaluated using the balance sheet approach in
two ways. The first is to estimate the stocks in the government's bal-
ance sheet and assess whether the public sector net worth is positive or
negative. If it is negative, then sustainability will require that the present
value of tax revenues minus government consumption be sufficient to
cover the negative net worth.
The second approach would look at sustainability in flow terms.
The criterion for sustainability would then be to maintain a constant
ratio of net worth to GDP. The idea is that if there is no payments
crisis today, then keeping the net worth-to-GDP ratio constant will
avoid a payments crisis in the future. So fiscal sustainability would
require keeping government net worth to GDP constant. If there is a
payments crisis today, then the r ule would imply increasing the ratio
of net worth to GDP.
If one cannot measure government net worth directly, a pragmatic
approach would be to first measure all balance sheet items that can be
184 EASTERLY AND YURAVLIVKER
measured to calculate the net debt of the government and then evalu-
ate individual government actions for their effect on fiscal and exter-
nal sustainability.
Note that analyzing this broader notion of fiscal sustainability re-
quires many assumptions to calculate concepts such as the implicit
pension debt and the value of oil reserves. This kind of exercise should
be thought of as illustrative of how the government's long-run finances
will evolve if certain assumptions hold.
What about balance sheet items that cannot be measured? All we
can say is that sustainability is worsened by any action that reduces
government assets or increases contingent liabilities. For example, a
cut in operations and maintenance (O&M) spending that lowers the
value of government-owned highways will worsen sustainability, even
though this spending cut improves the conventional measure of the
fiscal deficit. (If the proceeds from cutting O&M are used to repay
foreign debt, and if the rate of return on O&M is the same as the
interest rate on foreign debt, the spending cut would have a neutral
effect on fiscal sustainability. However, estimates of returns to O&M
are usually well above the interest rate on foreign debt.) A cut in prof-
itable public investments will worsen fiscal sustainability, even though
this cut improves the conventional measure of the fiscal deficit. Ex-
pansion of deposit insurance increases the government's contingent
liabilities and so worsens fiscal sustainability, even though it does not
show up in conventional deficit measures. Deposit insurance may bring
benefits such as increased deposits in the banking system, but the con-
tingent liability of the government should still be accounted correctly.
Consuming the revenues from extraction of a nonrenewable resource
or from privatization worsens sustainability.
In summary, balance sheet accounting can produce a better long-
run perspective on fiscal sustainability than can be obtained from con-
ventional deficit measures. Although the task of measuring government
assets and liabilities is always difficult and sometimes impossible, the
balance sheet approach still encourages clear thinking about what ac-
tions will improve or worsen the government's long-run finances. That
is particularly important when governments need to implement urgent
stabilization measures, to ensure that the measures are consistent with
longer-term fiscal sustainability.
Results from Colombia and
Republica Bolivariana de Venezuela
This section reviews the application of the assets/liabilities/net worth
approach to Colombia and Republica Bolivariana de Venezuela. These
are summaries of two very thorough studies conducted in both coun-
tries by local teams that had a unique access to the data required to
EVALUATING GOVERNMENT NE'T WORTH 185
carry out such a task. The Colombian study expands the traditional
government balance sheet by calculating other assets as well as contin-
gent and implicit liabilities of the public sector such as pension liabili-
ties and the potential costs of a peace agreement with the guerrillas.
The Venezuelan study, rather than looking at stocks, calculates the
changes in the main public sector assets and liabilities since 1970, and
then looks forward and calculates the net present value of the key
assets and liabilities of the public sector. Both studies conclude that
fiscal adjustment measures are needed to ensure longer-term fiscal
sustainability.
Colombia
Echeverry and others (1999) perform a comprehensive estimate of
public sector stocks, including contingent assets and liabilities (Table
8.2). Because their estimates are fraught with uncertainty and re-
quire many special assumptions, one should think of this exercise as
mainly illustrative.
Table 8.2 estimates that the Colombian public sector has 162 per-
cent of GDP in assets. The most important assets are investments,
plant and equipment, and natural resources. (The estimate does not
include the value of national infrastructure because of the difficulties
of valuation.) However, the Colombian public sector has liabilities
amounting to 232 percent of GD)P, implying a negative net worth of
70 percent of GDP. The most important liability by far is the implicit
pension debt, which amounts to 156 percent of GDP. In contrast, the
total public debt, which is usually the focus of sustainability analyses,
amounts to only 20 percent of GDP. The advantage of the balance
sheet approach is that it identifies a "hole" in the intertemporal public
finances that would not have been apparent using conventional deficit
or debt measures.
How could Colombia's finances be made sustainable if the public
sector's net worth is negative? To cover the negative net worth, the pub-
lic sector could run a perpetual primary surplus (government revenues
minus government noninterest spending). Here is a rough idea of the
amount of surplus necessary: the present value of a perpetual surplus of
x percent of GDP at a discount rate of r and a GDP growth rate of g
would simply be x/(r - g). If r = 0.10 and g = 0.04, for example, this
would imply a required perpetual primary current surplus of 4.2 percent
of GDP to cover negative net worth of 70 percent of GDP.
Given the importance of the implicit pension liability in the overall
picture, it is worthwhile to look at it in more detail. Table 8.3 breaks
down the pension liability into its components.
The most striking finding is that the central government's pension
obligations are more important than the obligations of the general
social security system (although the latter are far from trivial). Another
Table 8.2 Comprehensive Public Sector Balance Sheet, Colombia, 1997
(percent of GDP)
National level Territorial level Total
Decentralized Central Decentralized Central Public sector
Assets 48.4 42.5 25.9 23.6 162.3
Current assets 15.8 5.6 4.5 3.6 29.S
Cash 1.7 1.3 0.5 1.0 4.4
Investments 3.5 2.2 1.6 0.7 8.1
Rents 0.0 0.8 0.0 0.6 1.4
Accounts payable 9.0 1.2 1.9 0.8 12.9
Inventories 0.9 0.0 0.4 0.0 1.3
Other 0.6 0.0 0.3 0.4 1.3
Fixed assets 32.6 36.9 21.4 20.1 132.8
Investments 1.4 17.5 2.0 6.9 27.8
Rents 0.0 0.0 0.0 0.2 0.2
Loans 6.3 1.9 2.6 0.2 11.0
Plant and equipment 13.8 2.4 10.8 5.6 32.5
Public goods 3.3 0.1 0.4 3.3 7.1
Natural resources 1.0 14.4 0.0 1.1 38.4
Other (mostly electromagnetic spectrum) 6.9 0.5 5.6 2.8 15.8
Liabilities 34.9 20.8 15.4 7.1 232.3
Current liabilities 14.4 5.3 3.4 3.1 26.1
Required deposits 5.2 0.0 0.1 0.0 5.3
Public debt 0.1 2.8 0.3 0.4 3.6
Financial obligations 2.4 0.0 0.2 0.3 3.0
Suppliers' credits 3.7 1.1 1.2 1.5 7.6
Labor obligations 1.1 0.2 0.8 0.7 2.8
Bonds 1.2 0.3 0.0 0.0 1.5
Other 0.5 0.9 0.7 0.2 2.4
Long-term liabilities 20.4 15.5 11.9 4.0 205.9
Public debt 1.2 14.2 1.8 2.6 19.8
Financial obligations 3.1 0.0 2.5 0.2 5.7
Suppliers' credits 0.8 1.0 1.2 0.1 3.1
Labor obligations 0.0 0.0 0.1 0.1 0.2
Bonds 2.4 0.1 0.2 0.1 2.7
Other 12.9 0.2 6.1 1.0 20.2
Contingent liabilities 154.1
Pension liability 156.5
Net other contingent liabilities -2.4
Other 0.1 0.0 0.1 0.0 0.2
Public sector net worth 13.5 21.7 10.6 16.5 -69.9
Public Treasury 0.0 23.9 0.2 16.2 -92.0
Institutional 13.0 0.0 9.8 0.0 22.8
Other 0.5 -2.2 0.6 0.3 -0.7
Total liabilities, including net worth 48.4 42.5 25.9 23.6 162.3
Sources: Echeverry and others (1999) and the authors of this chapter.
188 EASTERLY AND YURAVLIVKER
Table 8.3 Estimates of the Implicit Pension Liability,
Colombia, 1997
(percent of GDP)
Entity Pension liability
National central government, 69.76
Teachersb 30.10
Armed forces, 15.92
Rest of central government' 23.74
National decentralized governmentd 5.76
Caprecom' 2.99
Ecopetrol' 2.28
Ecocarb6n' 0.03
Caja Agraria' 0.46
Territorial, 32.54
Social security, 48.48
Total 156.54
a. Liabilities assumed directly by the national government.
b. Based on actuarial estimates done for Echeverry and others (1999).
c. Based on actuarial estimates done in 1997.
d. Liabilities assumed by state enterprises and decentralized agencies.
e. Based on actuarial estimates done in 1998.
Sources: Ministerio de Hacienda y Credito Piblico, Colombia, and calculations of
Echeverry and others (1999).
way to reach sustainability would be to reform the government's pen-
sion system in order to raise contributions, postpone retirement, or
take other measures to lower the net present value of pension obliga-
tions. The same reforms to the general social security system could
also help reach sustainability.
Other contingent liabilities do not place a very heavy strain on pub-
lic sector finances. Table 8.4 shows that nonpension contingent liabili-
ties only sum to 14.5 percent of GDP. Among the most important of
the other contingent liabilities are the fiscal guarantees to the metros
(urban subway systems), the cost of achieving peace with the domestic
guerilla movements, and the expected present value of the bailout of
the financial system.
It is important to note that the public sector also has assets of condi-
tional value. The Colombian government has partial claim to oil, gas,
and coal reserves, whose value is contingent on world prices. (The price
assumptions are US$10 per barrel for oil, $30 per ton for coal, and
$1.30 per cubic foot for gas.) Note that these asset calculations subtract
off the unit costs of exploration and extraction. The public sector also
owns the electromagnetic spectrum, which it sells to cell phone compa-
EVALUATING GOVERNMENT NET WORTH 189
T'able 8.4 Other Contingent Liabilities of Public Sector
(Excluding Pensions), Colombia, 1997
Contingent liabilities Percentage of GDP
Natural disasters 1.10
Earthquakes 1.09
Floods 0.01
Bailing out financial system 2.18
Infrastructure guarantees 5.89
Roads 0.10
Airports 0.70
Electricity 0.50
Telecommunications 0.23
Metros 4.36
External debt guarantees 0.69
Judicial findings against public sector 0.16
Guarantees of municipal and provincial debt 0.48
Cost of reaching peace agreement 4.04
Total 14.54
Sources: Echeverry and others (I1999) andl the authors of this chapter.
nies. As of 1997, these contingent sources of future government rev-
enues were: petroleum reserves, 14.6 percent of GDP; natural gas re-
serves, 9.0 percent; coal reserves, 11.9 percent; and electromagnetic
spectrum, 18.3 percent. They totaled 53.9 percent of GDP.
Note that the value of the stock of natural resources is not that high
relative to GDP, only amounting to 36 percent of GDP. Later it is
revealed that the value of natural resources in Republica Bolivariana
de Venezuela is much larger.
In summary, the Colombia experience illustrates the value of the
balance sheet approach to public sector finances. This approach finds
negative net worth of the public sector, requiring a fiscal adjustment
whose need would not have been apparent from conventional deficit
measures.
Republica Bolivariana de Venezuela
For Republica Bolivariana de Venezuela, Garcia, Balza, and Villasmil
(1999) use a more dynamic approach to look at longer-term fiscal
sustainability. They calculate the annual change in the net worth of
the public sector during the past 30 years. That change was used to
finance public consumption and to subsidize domestic consumption of
190 EASTERLY AND YURAVLIVKER
oil products. They also calculate the implicit and contingent liabilities
of the public sector and project the evolution of public assets, liabili-
ties, and net worth into the future in order to estimate the maximum
nonoil fiscal deficit that would be consistent with longer-term fiscal
sustainability. Furthermore, these projections are done under a range
of assumptions to test for their sensitivity to various domestic and
external conditions. Finally, Garcia, Balza, and Villasmil discuss key
policy measures that would help the fiscal accounts move in the right
direction to achieve a solid financial position on a sustainable basis.
The Evolution of Public Assets, Liabilities, and Net Worth, 1970-
98. Garcia, Balza, and Villasmil (1999) describe five major develop-
ments that affected the financial position of the Venezuelan public
sector in this century:
o 1930-70-the discovery of oil and use of its revenues to finance
investment in infrastructure and the provision of public services;
o 1973-83-the decision to invest heavily in public enterprises by
means of growing external indebtedness;
o 1983-88-the decision to reduce investment in social sectors and
public infrastructure in order to maintain a high level of investment in
public enterprises and to service external debt;
o 1989-95-the use of privatization revenues to finance recurrent
public deficits; and
o 1996-98-the decision to increase investment and production
in the oil sector, financed by cuts in expenditure and investment in
other areas.
The sharp increases in oil prices in 1973-74 and in 1978-79, and
the expectation that prices would continue their upward trend, led to
a growing reliance on external debt to finance public expenditure and
advance ambitious investment projects. At the same time, the avail-
ability of huge oil revenues and foreign credit delayed implementation
of the structural reforms that were necessary to improve the efficiency
and competitiveness of the economy. Eventually, the volatility of oil
prices led to large ups and downs in the Venezuelan economy in gen-
eral and in the fiscal accounts in particular.
Estimating the evolution of public assets and liabilities is a very
difficult exercise. Garcia, Balza, and Villasmil (1999) focus on four
main sources that add liabilities or reduce assets: changes in the exter-
nal debt, changes in the domestic debt, privatization revenues, and the
value of oil production, net of production and investment costs. They
also identify three main uses that add assets or reduce liabilities: in-
vestment in nonfinancial public enterprises, other public investment,
and changes in international reserves. The difference between varia-
tions in uses and sources would then reveal the changes in the net
EVALUATING GOVERNMENT NET WORTH 191
worth of the public sector. Tables 8.5 and 8.6 show the evolution of
these variables from 1970 to 1998.
According to Table 8.5, the net worth of the public sector (in net
present value terms using a discount rate of 5 percent) declined by
nearly US$300 billion (287 percent of 1999 GDP) over the period
1970-98. It should be noted, however, that this estimation does not
include the discovery of 50 billion barrels of new oil reserves during
those years, which brought the total level of proved reserves to 76
billion barrels. Depending on the price of oil, these new discoveries
would represent $60-$180 billion in net present value terms. The de-
cline in the net worth of the public sector shows the drop in assets, or
increase in liabilities, used to: (a) finance current public consumption
(net of nonoil revenues) amounting to $177 billion and (b) subsidize
the domestic consumption of oil products (which were sold not only at
below-border prices but even below their production costs for most of
the period) by $86 billion in net present value terms. The remainder,
nearly $32 billion, is classified as errors and omissions, which reflect
rneasurement errors in public expenditure accounts and transfers to
the private sector that were not properly accounted for.
The loss of public net worth could be overstated, because current
public expenditures include the provision of education and health ser-
vices, which are in fact an investment in human capital. From that
point of view, there was an increase in private assets that compensates
for the decline in public assets. However, the magnitude of the drop in
net public worth is well beyond the fraction of current public expendi-
ture that could be accounted for as investment in human capital. Fur-
thermore, the drop in the quality of public services during the period
as well as the decline in real per capita GDP of over 10 percent suggest
that the public sector "jewelry" was not used in the most efficient
way. That calls for an evaluation of the use of public resources to
ensure that they do not undermine the financial position of the public
sector and that they contribute efficiently to the development of the
physical and human capital base of the country.
The loss of public net worth also could be understated because many
of the public investments turned out to be unprofitable. For example,
the heavy investment in steel production in the 1970s and 1980s, at a
time of overcapacity in the world steel market, almost surely did not
increase the value of public capital very much.
Contingent and Implicit Liabilities. The calculation described above
does not take into account the existence and evolution of contingent
or implicit public sector liabilities. Therefore, after discussing short-
term fiscal stabilization measures for 1999-2000, Garcia, Balza, and
Villasmil (1999) identify and estimate the net present value of the pub-
lic sector liabilities embodied in: (a) the social security system; (b) the
Table 8.5 Net Worth of the Public Sector, R6publica Bolivariana de Venezuela, 1970-98
(millions of U.S. dollars)
Value of
oil prod- Investment
uction (net by non-
Change Change invest"ment financial Change Net Subsidies
in net in net Privati- and oper- public Other in inter- Change in current to domestic Errors
external domestic zation ational enter- public national public expenses oil con- and
debt debt income expenses) prises investment reserves net worth (nonoil) sumption omissions
(8) = (5) + (6) +
(7) - (1) - (2) - (l 1) = -(8)
(1) (2) (3) (4) (5) (6) (7) (3) - (4) (9) (1 0) -(9) - (10)
1970 152 81 940 225 573 85 -289 -16 75 230
1971 270 31 1,569 442 610 444 -375 180 87 108
1972 1,084 43 1,701 867 605 218 -1,138 339 100 699
1973 135 99 2,665 859 782 724 -534 389 114 31
1974 -112 502 9,204 816 1,932 3,842 -3,004 -707 132 3,579
1975 -301 331 7,843 1,534 3,028 2,613 -698 -196 151 742
1976 2,667 342 6,127 1,566 3,702 -286 -4,154 1,436 174 2,544
1977 5,322 1,446 8,256 3,985 3,494 -425 -7,970 2,328 635 5,008
1978 4,779 803 6,668 3,927 3,172 -1,707 -6,858 1,836 253 4,768
1979 6,721 308 11,485 3,457 2,977 1,302 -10,778 3,184 917 6,677
1980 4,656 1 16,083 3,762 3,529 -715 -14,164 5,040 1,976 7,148
1981 2,331 1,535 16,099 4,681 3,460 1,594 -10,231 6,526 2,814 890
1982 -1,281 1,090 11,175 5,564 4,830 1,420 830 5,846 1,984 -8,660
1983 2,167 975 12,245 4,964 4,781 1,110 -4,531 4,418 2,749 -2,635
1984 -1,127 -2,032 16,130 1,971 3,756 1,320 -5,924 3,050 3,363 -488
1985 -1,470 1,704 15,303 2,152 3,933 1,281 -8,171 2,462 3,838 1,871
1986 293 1,565 5,780 3,241 3,880 -3,892 -4,409 2,211 898 1,300
1987 2,313 -3,840 9,864 2,407 2,311 -483 -4,102 3,484 2,586 -1,968
1988 645 128 7,142 2,843 3,933 -2,705 -3,844 4,111 1,748 -2,015
1989 -1,321 -1,960 11,176 2,121 1,499 740 -3,535 4,282 2,883 -3,630
1990 1,746 -323 10 15,018 1,993 2,162 4,348 -7,948 6,455 3,592 -2,100
1991 874 1,497 2,278 11,340 1,515 4,118 2,346 -8,010 5,842 2,447 -280
1992 3,306 -272 30 8,566 1,679 2,261 -1,104 -8,793 5,749 2,315 729
1993 -7,571 1,309 32 8,238 1,068 1,788 -345 503 4,101 2,070 -6,674
1994 174 2,058 18 8,603 644 1,391 -1,149 -9,967 11,770 2,413 -4,216
1995 -2,365 1,299 20 9,400 962 2,390 -1,784 -6,785 5,797 3,002 -2,015
1996 -988 -3,996 1,159 15,120 854 1,731 5,506 -3,205 3,397 3,504 -3,696
1997 212 -843 2,425 12,538 1,154 3,332 2,589 -7,256 4,547 2,896 -187
1998 -666 -1,338 110 271 1,135 3,313 -2,969 3,102 1,672 -1,573 -3,202
Total 22,645 2,543 6,082 266,549 62,388 79,273 13,919 -142,238 99,533 48,143 -5,443
NPVa 71,586 12,948 7,643 546,088 132,279 168,260 35,527 -295,199 177,181 86,365 31,653
a. In 1999 net present value, at a 5 percent discount rate.
Sources: Garcia, Balza, and Villasmil (1999) and the authors of this chapter.
Table 8.6 Net Worth of the Public Sector, R6publica Bolivariana de Venezuela, 1970-98
(percent of GDP)
Value of
oilprod- Investment
uction (net by non-
Change Change investment financial Change Net Subsidies
in net in net Privati- and oper- public Other in inter- Change in current to domestic Errors
external domestic zation ational enter- public national public expenses oil con- and
debt debt income expenses) prises investment reserves net worth (nonoil) sumption omissions
(8) = (5) + (6) +
(7) - (1) - (2) - (I 1) = -(8)
(1) (2) (3) (4) (5) (6) (7) (3) - (4) (9) (10) -(9) - (10)
1970 1.1 0.6 7.0 1.7 4.3 0.6 -2.2 -0.1 0.6 1.7
1971 1.8 0.2 10.3 2.9 4.0 2.9 -2.5 1.2 0.6 0.7
1972 6.5 0.3 10.2 5.2 3.6 1.3 -6.8 2.0 0.6 4.2
1973 0.7 0.5 13.2 4.3 3.9 3.6 -2.6 1.9 0.6 0.1
1974 -0.4 1.7 30.6 2.7 6.4 12.8 -10.0 -2.4 0.4 11.9
1975 -0.9 1.0 24.2 4.7 9.3 8.0 -2.2 -0.6 0.5 2.3
1976 7.1 0.9 16.4 4.2 9.9 -0.8 -11.1 3.8 0.5 6.8
1977 12.2 3.3 18.9 9.1 8.0 -1.0 -18.2 5.3 1.5 11.4
1978 9.9 1.7 13.9 8.2 6.6 -3.6 -14.3 3.8 0.5 9.9
1979 11.7 0.5 19.9 6.0 5.2 2.3 -18.7 5.5 1.6 11.6
1980 6.7 0.0 23.2 5.4 5.1 -1.0 -20.4 7.3 2.8 10.3
1981 3.0 2.0 20.6 6.0 4.4 2.0 -13.1 8.4 3.6 1.1
1982 -1.6 1.4 14.1 7.0 6.1 1.8 1.0 7.4 2.5 -10.9
1983 2.7 1.2 15.1 6.1 5.9 1.4 -5.6 5.4 3.4 -3.2
1984 -1.9 -3.4 26.9 3.3 6.3 2.2 -9.9 5.1 5.6 -0.8
1985 -2.4 2.8 24.7 3.5 6.3 2.1 -13.2 4.0 6.2 3.0
1986 0.5 2.6 9.6 5.4 6.4 -6.4 -7.3 3.7 1.5 2.1
1987 4.8 -8.0 20.5 5.0 4.8 -1.0 -8.5 7.3 5.4 -4.1
1988 1.1 0.2 11.9 4.7 6.5 -4.5 -6.4 6.8 2.9 -3.3
1989 -3.0 -4.5 25.7 4.9 3.4 1.7 -8.1 9.8 6.6 -8.3
1990 3.6 -0.7 30.9 4.1 4.4 8.9 -16.4 13.3 7.4 -4.3
1991 1.6 2.8 4.3 21.2 2.8 7.7 4.4 -15.0 10.9 4.6 -0.5
1992 5.4 -0.4 0.0 14.1 2.8 3.7 -1.8 -14.5 9.5 3.8 1.2
1993 -12.6 2.2 0.1 13.7 1.8 3.0 -0.6 0.8 6.8 3.4 -11.1
1994 0.3 3.5 0.0 14.7 1.1 2.4 -2.0 -17 1 20.2 4.1 -7.2
1995 -3.1 1.7 0.0 12.2 1.2 3.1 -2.3 -8.8 7.5 3.9 -2.6
1996 -1.4 -5.7 1.6 21.5 1.2 2.5 7.8 -4.6 4.8 5.0 -5.3
1997 0.2 -1.0 2.7 14.2 1.3 3.8 2.9 -8.2 5.1 3.3 -0.2
1998 -0.7 -1.4 0.1 0.3 1.2 3.5 -3.1 3.3 1.8 -1.7 -3.4
NPV. 69.7 12.6 7.4 531.3 128.7 163.7 34.6 287.2 172.4 84.0 30.8
a. In 1999 net present value, at a 5 percent discount rate.
Sources: Garcia, Balza, and Villasmil (1999) and the authors of this chapter.
196 EASTERLY AND YURAVLIVKER
guaranteed minimum pension for public sector employees; (c) the labor
liabilities of the public sector; and (d) the public guarantee of bank de-
posits. Those estimations are later used to assess the long-term fiscal
sustainability of the public sector in Republica Bolivariana de Venezuela.
The pay-as-you-go social security system that was in place until
1998 collapsed because of various factors, including a growing infor-
mal sector that currently accounts for 53 percent of employment, eva-
sion of payment of contributions to the system, and the use of social
security funds to cover the deficits of the health care system. Had it
not been reformed, that system would have gone bankrupt, affecting
over a half-million people and costing the public sector the equivalent
of 77 percent of GDP in present value terms. The 1998 reform created
a mixed system of personal accumulation funds and a solidarity pen-
sion fund administered by the state. The retirement age of women was
raised to 60 years (as it is for men), contribution rates were doubled to
12-13 percent, and the minimum contribution period was raised from
15 to 20 years.
The fiscal cost of the reform has three components: (a) the existing
retirees and those retiring in 1999 under the old system; (b) the "rec-
ognition bond" to people who contributed to the old system but who
have not yet reached retirement age; and (c) the solidarity pension
fund. Under certain assumptions, the total fiscal impact of the new
system was estimated, in present value terms, at nearly US$60 billion,
equivalent to 65 percent of GDP at the end of 1998.
There are special pension regimes in the public sector that are not
contributive, including those of the central government, the local gov-
ernments, most public enterprises, autonomous agencies, the judicial
system, the national universities, and the armed forces. These regimes
are financed directly from the budgets of these agencies, whose statis-
tical and financial information base is in many cases quite deficient.
At the end of 1998, these regimes had nearly 250,000 retirees and nearly
1 million current public sector employees under their umbrella. A con-
servative estimate of this public liability is US$12.4 billion in present
value terms, equivalent to 13.5 percent of GDP at the end of 1998.
The labor liabilities of the public sector include the accumulated
benefits of one month's salary per year of work per employee. The
1997 reform of the Organic Labor Law abolished the indexation of
those benefits to the last salary and the doubling of the benefits when
the employee was laid off for "unjustified reasons," but extended the
benefit to two months per year. The new law also called for payment
of the arrears of this benefit, which were accumulated under the old
law, over a period of five years, including an additional bond for the
transfer from the old to the new system. By the end of 1998, the total
public debt on this account amounted to US$7 billion, equivalent to
7.6 percent of GDP.
EVALUATING GOVERNMENT NET WORTH 197
R6publica Bolivariana de Venezuela had two systemic banking crisis
in 1961-63 and 1994-95 and three large bank bankruptcies in between.
In all cases, the central bank (and the deposit insurance guarantee agency,
or FOGADE, in 1994-95) stepped in to cover all deposits, and in the
last crisis it extended that coverage up to a limit of VEB10 million, two
and a half times higher than the VEB4 million established in the original
deposit insurance scheme. At the end of 1998, the total deposits in the
banking system amounted to US$16.3 billion, or 17.7 percent of GDP.
Of those, $7.5 billion consisted of deposits under VEB4 million, which
are guaranteed by FOGADE, and $9.8 billion of deposits under VEB10
million, which was the actual coverage limit after the last banking cri-
sis. FOGADE, on the other hand, had assets of only $760 million. There-
fore, the implicit public guarantee amounts to about $9 billion, equivalent
to nearly 10 percent of GDP. The estimate of the contingent liability
would be lower, depending on the probability of banking losses.
Table 8.7 shows the assets, liabilities, and net worth of the Venezu-
elan public sector in terms of net present value (US$ millions). Assets
include reserves of oil, gas, and coal and international reserves, leaving
out other mineral reserves as well as the value of public utilities and
public enterprises because of estimating difficulties. Liabilities include
external and domestic debt, social security, the labor and pension li-
abilities of the public sector, and bank deposit guarantees. The calcula-
tions assume a growth rate of 4 percent and a discount rate of 9 percent.
As Table 8.7 shows, the difference between assets and liabilities
results in a public sector net worth of US$43.5 billion, equivalent to
4.5 percent of GDP. This net worth could be consumed by allowing the
public sector to have primary noncoil deficits on the order of 2 percent
of GDP on a permanent basis.
lable 8.7 Net Worth of the Public Sector in
R.epublica Bolivariana de Venezuela
(net present value, millions of U.S. dollars)
Assets Liabilities
Oil, 120,000 External debt 23,613
Gas 16,000 Domestic debt 3,980
Coal 2,300 Deposit insurance 10,961
International reserves 14,849 Labor debt and public
pensions 20,032
Seigniorage 10,834 Social security debt 61,921
Total 163,983 Total 120,507
Net worth 43,476
a. Assuming a price of oil of US$16 per barrel.
Sources: Garcia, Balza, and Villasniil ( 19 99) and the authors of this chapter.
198 EASTERLY AND YURAVLIVKER
This analysis has certain limitations, because the implicit assump-
tion is that the public sector has perfect access to international credit
markets and can borrow on the basis of the net present value of its net
worth. However, as recent history has shown, access to international
credit is anything but perfect, and it depends on changing global con-
ditions. Furthermore, the volatility in oil prices introduces a lot of
uncertainty into any calculation of the value of that asset. At the same
time, these calculations show that continuing the trend of the past 29
years, when the nonoil primary deficit averaged 12.8 percent of GDP,
is clearly unsustainable. Even nonoil primary deficits on the order of 7
percent of GDP, the average of the 1990s, would require an average oil
price of nearly US$20 per barrel. Thus Garcia, Balza, and Villasmil
(1999) concluded that, even under relatively optimistic assumptions,
the Venezuelan public sector could not maintain, on a permanent ba-
sis, nonoil primary deficits of over 6 percent of GDP.
Overdependence on future oil revenues is risky, because within 30
years or so there could be a technological change that would bring
about a sharp decline in the demand for oil. Furthermore, exploiting
and consuming an exhaustible resource raise the issue of intergen-
erational appropriation of the national wealth. In that respect, there
are two approaches to oil revenues: one is to treat them as any other
source of revenue; the second is to treat them as an asset being de-
pleted, making sure that the proceeds of oil are invested and generate
other income in the future, resulting in a constant rent over time.
Apart from the principle of equal rights to the national wealth by
all generations of Venezuelans, the second approach also would pre-
vent a major adjustment in the future, whenever the production of oil
goes down. To implement this approach, the government could invest
part of the oil revenues abroad, to cushion the county from permanent
oil shocks (temporary shocks are addressed by the oil stabilization
fund). Keeping these resources abroad also would reduce overvalua-
tion pressures, which have negative effects both on the fiscal accounts
and on the real sector of the economy. However, the country would
have to balance these considerations with the need to expand infra-
structure investment and human capital accumulation within the coun-
try itself, which means that the economic rate of return on oil revenues
could well be, at least in the next few years, higher when invested
internally than when kept abroad.
Finally, it should be noted that the results are sensitive to the as-
sumptions about GDP growth and the discount rate. Robust, sustain-
able growth is essential for long-term fiscal sustainability, and reducing
the volatility of economic activity (by limiting the impact of oil price
shocks, for example) would have a positive effect in terms of a lower
country risk and the risk premium that Republica Bolivariana de Ven-
ezuela pays on its external borrowing. Also, the quality of fiscal ad-
justment is very important in order not to undermine the infrastructure
EVALUATING GOVERNMENT NET WORTH 199
development of the country. Thus, apart from improving the efficiency
and effectiveness of public investment and expenditures, there is a need
to increase nonoil revenues and to reduce the contingent and implicit
liabilities of the public sector. In terrns of levels, the nonoil fiscal deficit
should on average be below 4 percent of GDP in order to ensure longer-
term fiscal sustainability.
Summary and Conclusions
This chapter describes two ways to evaluate sustainability using the
balance sheet approach. The first is to estimate the stocks in the
government's balance sheet and assess whether public sector net worth
is positive or negative. If it is negative, then sustainability will require
that the present value of tax revenues minus government consumption
be sufficient to cover the negative net worth.
The second approach is to look at sustainability in flow terms. The
criterion for sustainability would t:hen be to maintain a constant ratio
of net worth to GDP or, if a desired level of net worth can be deter-
mined, to maintain that ratio above the desired level. The basic idea is
that if there is no payments crisis today, then keeping the net worth-to-
GDP ratio constant will avoid a payments crisis in the future. Thus
fiscal sustainability would require keeping the government net worth-
to-GDP ratio constant, or above a minimum desired level. If there is a
payments crisis today, then the rule would imply increasing the ratio
of net worth to GDP.
This chapter presents the conclusions of two country studies, for
Colombia and Republica Bolivariana de Venezuela. The first study
uses the first approach, calculating stocks of assets and liabilities, and
the second study illustrates the second approach, focusing mostly on
changes of assets and liabilities over time. Both, however, have clear-
cut conclusions about the actions needed to achieve longer-term fiscal
sustainability.
For Colombia, Echeverry and others (1999) estimate that the pub-
lic sector has 162 percent of GDPI in assets. However, the Colombian
public sector has liabilities amounting to 232 percent of GDP (the
most important by far is the implicit pension debt, which amounts to
156 percent of GDP), implying a negative net worth of 70 percent of
GDP. In contrast, the total public debt, which is usually the focus of
sustainability analyses, amounts to only 20 percent of GDP. Given this
estimate of net worth, achieving longer-term fiscal sustainability would
require a perpetual primary current surplus of 4.2 percent of GDP.
For Republica Bolivariana de Venezuela, Garcia, Balza, and Villasmil
(1999) reveal that in 1970-98 thie net worth of the public sector de-
clined by nearly US$300 billion, the equivalent of 287 percent of 1999
GDP. This decline shows the drop in assets, or rise in liabilities, used
200 EASTERLY AND YURAVLIVKER
to: (a) finance current public consumption (net of nonoil revenues)
amounting to $177 billion, and (b) subsidize the domestic consump-
tion of oil products (which were sold not only at below-border prices
but even below their production costs for most of the period) by $86
billion in net present value terms.
The difference between Venezuelan assets and liabilities results in a
public sector net worth of US$43.5 billion, equivalent to 45 percent of
GDP. That would allow the public sector to have nonoil deficits on the
order of 2-6 percent of GDP on a permanent basis, depending on the
assumed future oil price.
These calculations show that continuing the trend of the past 29
years, when the nonoil primary deficit averaged 12.8 percent of GDP,
is clearly unsustainable. Even nonoil primary deficits on the order of 7
percent of GDP, the average of the 1990s, would require an average oil
price of nearly US$20 per barrel. Thus Garcia, Balza, and Villasmil
concluded that, even under relatively optimistic assumptions, the Ven-
ezuelan public sector cannot maintain, on a permanent basis, nonoil
primary deficits of over 6 percent of GDP.
In conclusion, our picture of public finance in two test cases, Co-
lombia and Republica Bolivariana de Venezuela, was dramatically al-
tered by using the balance sheet approach to public finances. We found
both treasures and time bombs in the governments' balance sheets.
Notes
1. This approach is used as well to assess the consistency between a bud-
get deficit and inflation targets, where not only the debt ratio is presumed
constant but also the money-to-GDP ratio. Revenues from money creation
are calculated as the sum of the inflation rate times the money-to-GDP ratio
(the inflation tax) and the growth rate times the money-to-GDP ratio (seignior-
age). Economists have used these calculations on many occasions to assess
the consistency of the fiscal stance with avoiding a debt crisis and excessive
inflation. The classic references are Buiter (1983, 1985) and Anand and van
Wijnbergen (1989). Other examples of references are Marshall and Schmidt-
Hebbel (1994) for Chile, Haque and Montiel (1994) for Pakistan, and
Morande and Schmidt-Hebbel (1994) for Zimbabwe.
2. For an illustration of the shortcomings of conventional sustainability
analysis, see Blejer and Cheasty (1991), Polackova (1998), and Easterly (1999).
3. Buiter (1983, 1985) pioneered the use of the balance sheet approach to
fiscal accounts. The government of New Zealand has pioneered the use of
the balance sheet approach in its fiscal accounting (Scott 1996).
4. Some assets are omitted because they are impossible to measure, such
as the value of the human capital embodied in the government's work force.
EVALUATING GOVERNMENT NET 'VORTH 201
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Anand, Ritu, and Sweder van Wijnbergen. 1989. "Inflation and the Financ-
ing of Government Expenditure: An Introductory Analysis with an Appli-
cation to Turkey." World Bank Economic Review 3 (January): 17-38.
Blejer, Mario I., and Andrienne Cheasty. 1991. "The Measurement of Fiscal
Deficits: Analytical and Methodological Issues." Journal of Economic
Literature 29: 1644-78.
Buiter, W 1983. "Measurement of the Public Sector and Its Implications for
Policy Evaluation and Design." IMIP Staff Papers. (Updated and reprinted
in Mario 1. Blejer and Andrienne Cheasty, eds. 1993. How to Measure the
Public Deficit: Analytical and Methodological Issues. Washington, D.C.:
International Monetary Fund.)
- . 1985. "A Guide to Public Sector Debt and Deficits." Economic Policy
(November): 14-79.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April): 57-86.
Echeverry, Juan Carlos, Maria Victoria Angulo, Gustavo Hernandez, Israel
Fainboim, Cielo Numpaque, Gabriel Piraquive, Carlos Rodriguez, and
Natalia Salazar. 1999. "El Balance del Sector Publico y la Sostenibilidad
Fiscal de Colombia." Archivos de Macroeconomia. No. 115. National
Planning Department, Bogota.
Garcia Osio, Gustavo, Rafael Rodriguez Balza, and Ricardo Villasmil. 1999.
"Ajuste y Sostenibilidad Fiscal de Largo Plazo-El Caso de Venezuela."
Oficina de Asesoria Economica y Financiera del Congreso de la Repub-
lica, Caracas. Processed.
Haque, Nadeem Ul, and Peter Montie:i. 1994. "Pakistan: Fiscal Sustainability
and Macroeconomic Policy." In William Easterly, Carlos Alfredo
Rodriguez, and Klaus Schmidt-Hebbel, eds., Public Sector Deficits and
Macroeconomic Performance. Washington, D.C.: World Bank.
Marshall, Jorge, and Klaus Schmidt.Hebbel. 1994. "Chile: Fiscal Adjust-
ment and Successful Performance." In William Easterly, Carlos Alfredo
Rodriguez, and Klaus Schmidt-Hebbel, eds., Public Sector Deficits and
Macroeconomic Performance. Washington, D.C.: World Bank.
Morande, Felipe, and Klaus Schmidt-Hebbel. 1994. "Zimbabwe: Fiscal Dis-
equilibria and Low Growth." In William Easterly, Carlos Alfredo
Rodriguez, and Klaus Schmidt-Hebbel, eds., Public Sector Deficits and
Macroeconomic Performance. Washington, D.C.: World Bank.
Polackova, H. 1998. "Contingent Liabilities: A Threat to Fiscal Stability."
World Bank PREM Notes. No. 9. Washington, D.C.
Scott, G. 1996. "Government Reform in New Zealand." IMF Occasional
Paper 140. International Monetary Fund, Washington, D.C.
CHAPTER 9
The Challenges of
Fiscal Risks irn Transition:
Czech Republic, Hungary,
and Bulgaria
Hana Polackova Brixi
World Bank
Allen Schick
University of Maryland
Leila Zlaoui
World Bank
AN ESSENTIAL STEP IN controlling the expansion of government con-
tingent liabilities and reducing fiscal risk is being able to identify and
measure them. In this chapter we discuss how this may be done, and
we demonstrate how an assessment of fiscal adjustment may change
substantially when a broader picture of government obligations is in-
cluded. The chapter is based on a 1999 analysis of fiscal adjustment in
the Czech Republic, Hungary, and Bulgaria.
The Czech case provides an example of officially balanced govern-
ment budgets and of the extensive use of guarantees and other forms
of off-budget support. The case study shows how to deal with some
(lifficult conceptual and measurenient issues when estimating govern-
inent contingent liabilities and the unreported portion of fiscal deficit.
In contrast to the Czech Republic, Hungary internalized most fiscal
r isks in the government debt and constrained off-budget fiscal activi-
ties. While the Czech Republic enjoyed "budget balance," Hungary
203
204 BRIXI, SCHICK, AND ZLAOUI
faced high budget deficit and debt levels. Less-visible aspects of fiscal
adjustment, however, pull the comparison of the fiscal performance of
thc two countries in the opposite direction. Somewhat like Hungary,
Bulgaria maintained low, transparent government exposure to fiscal
risk. And somewhat like the Czech Republic, Bulgaria has been com-
mitted to low budget deficits and macroeconomic stability-since 1997
at least, after having introduced a currency board arrangement. In
contrast to both the Czech Republic and Hungary, however, Bulgaria
has been slow in the transition process and has yet to meet the fiscal
challenges of the needed enterprise restructuring and investment in
infrastructure and environment.
Although each of the three countries has taken a different path to
fiscal adjustment, economic realities and opportunities may lead to
more similar behavior in the future. As the Czech Republic's hidden
liabilities came to light in 1998, the government faced pressure to im-
pose discipline in resolving old and taking on new fiscal risks. As for
Hungary, the favorable fiscal performance in the second half of the
1990s emboldened the government to take on greater liabilities than it
had been able to handle during the adjustment period. With the Czech
government taking a tougher stand on contingent liabilities and the
Hungarian government showing signs of loosening its established con-
trols, a comparison of the two countries in the future may show a
different pattern than that evident at the end of the 1990s. Bulgaria's
recent favorable fiscal performance has yet to be tested by the tradeoff
of fiscal prudence versus a more aggressive strategy toward the country's
development and accession to the European Union (EU). All three coun-
tries consider their accession to the European Union a key policy pri-
ority, and their motivation to meet the EU accession requirements
(including requirements on the quality of infrastructure and environ-
ment) is high.
The three sections that follow apply the Fiscal Risk Matrix (pre-
sented in Chapter 1 of this volume) and analyze the "true" fiscal posi-
tion of the Czech Republic, Hungary, and Bulgaria. The main
conclusions and suggestions for future work are summarized in the
final section.
Measuring the True Fiscal Deficit
of the Czech Republic
The Czech Republic has been known for its balanced government bud-
gets. In contrast to most countries, however, fiscal performance in the
Czech Republic encompasses a significant number of government ac-
tivities financed outside the budgetary system. These activities gener-
ate fiscal risks. Recently, these off-budget fiscal risks have become more
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 20s
visible, because state guarantees and agencies that are either explicitly
or implicitly guaranteed by the government have generated significant
claimis on the budget (see Table 9.1).
Given the magnitude of off-budget activities, fiscal analysis in the
Czech Republic should identify all the main activities of a fiscal nature
Table 9.1 Fiscal Risk Matrix, Czech Republic
Contingent liabilities
Souirces of Direct liabilities (obligation if a
obligations (obligation in any event) particular event occurs)
Explicit
Government * Foreign and domestic * State guarantees
liability as sovereign debt * Liabilities and other obligations
recognized * Budget expenditures of Konsolidacni Banka
by a law or * Future legally bind- * Obligations of Ceska Exportni,
contract ing expenditures EGAP (an export guarantee fund),
and Deposit Insurance Fund
Implicit
A "moral" * Future investment ex- * Obligations of National Property
obligation of penditures to meet EU Fund (own debt, guarantees,
government accession requirements and obligations to Ceska Inkasni,
that reflects * Future recurrent ex- Land Fund, and similar entities)
public and penditures related to * Liabilities of Ceska Financni and
interest group public investment Czech National Bank (result of
pressures projects the central bank's nonstandard
* Military expenditures operations)
as required by North * Liabilities of banks (Komercni
Atlantic Treaty Organ- Banka, Ceska Sporitelna)
ization (NATO) * Further losses and defaults of
large enterprises (Czech Railways)
* Obligations of PGRLF (an agri-
cultural credit and guarantee fund)
* Liabilities and other obligations
of subnational governments
* Liabilities of credit unions
(Kampelicka) and private
pension funds
Note: The liabilities listed refer to the fiscal authorities of the central government.
Because the government is legally obliged tc pay future public pensions (a public pay-as-
you-go pension scheme), future pensions constitute a direct (expected with certainty)
explicit (legal) liability. The expected investrnent expenditures that are needed to meet EU
accession requirements are the major direct implicit liability. State guarantees and financ-
irig through state-guaranteed institutions ale key examples of explicit contingent liabili-
ties. And, like that of many other countries, the financial system represents the most
serious source of implicit contingent liabilities for the Czech government.
Source: The authors.
206 BRIXI, SCHICK, AND ZLAOUI
in order to determine the "true fiscal deficit." Excluding the quasi-
fiscal activities of the central bank, the Czech National Bank, the "hid-
den" part of the fiscal deficit comprises two main components: (a) net
spending on programs of a fiscal nature by special off-budget institu-
tions-Konsolidacni Banka (KOB), Ceska Inkasni (CI), Ceska Financni
(CF),' and the National Property Fund (NPF); and (b) implied subsi-
dies extended through state guarantees. For the financial relationships
of the special institutions, see Figure 9.1 (developed by the Ministry of
Finance of the Czech Republic).
For any given year, net public spending by these institutions in-
cludes cash outlays on new programs in the form of directed credits
and asset purchases,2 and interest expenditures. This spending is ad-
justed for debt collection, interest revenue, and other revenue from
programs. Table 9.2 shows the components of the "hidden" deficit. In
the remainder of this section, we explain each row of this table in
detail.
Figure 9.1. Financial Relationships of Special, Off-Budget
Institutions, Czech Republic
- Financial Flows - Guarantees
Czech National Bank Ministry of Finance
Government
Redistribtition credi onal Property Fund
Refinanicing credit 1
Special Ceska Konsolidacni Ceska
institutions Financni e Banka lnkasni
(CF) (KOB) (Cl)
pa cilt for 1 Purchase of
bad assets || bad loans , .
Bsad asscts Cororate
Corporate Banks, enterprises, Refinancing an enoi-
and hospitals, and other ConB ena
financial entities qualified CSOB liabilitie
sectors for support Bank
Borrowing and Bonds J
bonds issued issued
Source: Ministry of Finance.
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 207
Table 9.2 Sources of "Hidden" I)eficit in the Czech Republic,
1993-98
(billions of Czech koruny)
1993 1994 1995 1996 1997 1998
Konsolidacni Banka
(KOB)' (net public
expenditures) 7.7 7.3 4.5 0.9 10.6 28.8
Ceska Inkasni (CI)
(net public expend-
itures) 20.1 6.6 4.9 4.8 3.1 2.7
Ceska Financni (CF)
(net public expend-
itures) n.a. n.a. n.a. n.a. 0.6b i.8b
National Property
Fund (NPF) (net
public expenditures,
excluding KOB
and Cl) 4.2 8.2 4.3 1.9 2.0 2.6
State guarantees
("hidden" subsidy,
r isk-adjusted) 0.1 -0.4 1.3 14.9 51.5 26.7
Total (% of GDP) 3.2 1.5' 1.1 1.5 4.1 3.5
n.a. Not applicable.
a. Activities of KOB include a credit to finance stabilization program of CF. Therefore,
table includes only interest payments by CF (which are reported as interest income of
KOB).
b. These figures are interest payments to KOB on credit taken by CF from KOB to
finance stabilization program. In addition, CF paid interest to CZKO.8 billion and CZK2.8
billion in 1997 and 1998, respectively, to Czech National Bank on its credit from Czech
National Bank to finance consolidation program.
Sources: Ministry of Finance, Konsolidacni Banka, Ceska Financni, National Property
Fund, and calculations by the authors.
Until 1993, off-budget programs dealt mainly with pretransition
problems inherited by the banking sector. These programs had been
financed through Konsolidacni Banka. This bank was capitalized by
the National Property Fund (the privatization agency whose revenues
are derived from asset sales and borrowing on domestic markets) and
borrowing from the Czech National Bank.3 In 1995 the Ministry of
Finance established Ceska Inkasni, a nonbank financial institution with
the mandate of cleaning up the portfolio of a state-owned bank, the
CSOB. Covered by a guarantee issued by the National Property Fund,
Ceska Inkasni obtained a credit from the CSOB and used this credit to
purchase CSOB's bad assets at face value.
208 BRIXI, SCHICK, AND ZLAOUI
During 1996-98, a new bank consolidation and stabilization pro-
gram was launched to deal with newly emerging problems in the bank-
ing sector. In order to implement these programs, the Czech National
Bank established Ceska Financni, another nonbank financial institu-
tion. In 1998 Ceska Financni had in its portfolio nonperforming assets
purchased at face value from small and medium-size banks (in the
amount of about CZK50 billion, or 3 percent of GDP). It financed the
purchase through borrowing (one-third) from Konsolidacni Banka and
(two-thirds) from the Czech National Bank.
The Czech National Bank also financed other bank rescue opera-
tions, which became the source of a further addition (CZK161 billion,
over 9 percent of GDP) to its portfolio of substandard assets in 1998.
Of the total amount of substandard assets held by Czech National
Bank, the government covered the risk for 12 percent of the assets. A
further 22 percent of these assets were in the form of a credit from the
Czech National Bank to Konsolidacni Banka and thus were indirectly
also covered by government.
Aside from the bank rescue operations, Konsolidacni Banka and,
less directly, the National Property Fund financed government pro-
grams to support troubled health insurance companies, public hospi-
tals, and the Czech Railways, to build infrastructure, and to clean up
industrial enterprises for privatization (see Table 9.3). The National
Property Fund financed these programs partly from privatization rev-
enues but also partly from its debt issuance. The contributions to the
"true" fiscal deficit by the National Property Fund exclude principal
repayments and thus do not reflect the ongoing financing of pre-1993
programs by the National Property Fund. In addition, both Konsolidacni
Banka and the National Property Fund accumulated their own contin-
gent liabilities in the form of various guarantees on environmental
liabilities.4
The impact of guarantees on the hidden deficit is estimated as the
net implicit subsidy provided through guarantees in a given year from
the portfolio of guarantees issued in that year. This estimate is the
potential fiscal cost of government obligations, which will emerge from
the guarantees in the future. If the amount of this subsidy had been
transferred to a guarantee reserve fund the same year the guarantee
was issued, it would have covered potential future claims emerging
from the guarantee. The cost of default would be paid from the guar-
antee reserve fund and thus would not affect the budget and the deficit.
Our assessment of each guarantee and its underlying project pro-
vided the basis for estimating their future fiscal costs. Projects were
ranked according to their risk. Accordingly, the default risk of each
guarantee was estimated. The probability of default was determined
by careful consideration of each loan. Table 9.4 shows the amounts of
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 209
Table 9.3 Programs Covered by National Property Fund,
Czech Republic, 1993-98
(billions of Czech koruny)
1993 15994 199s 1996 1997 1998
Financing environment-
al rehabilitation . . 0.1 0.8 1.0 1.4 2.1
Financing development
of railway route n.a. n.a. n.a. 0.1 . . 0.2
Support to state-owned
enterprises 2.1 I5 0.9 0.3 0.2 0.3
Support to agricultural
businesses n.a. 6.1 1.0 n.a. n.a. n.a.
Bond interest 2.1 1.5 1.6 0.S 0.4 0.5
National Property Fund's
"hidden" fiscal deficita 4.2 8.2 4.3 1.9 2.0 2.9
Others, already included
in hidden deficit cal-
culation:
Health insurance
companies
(through KOB)b n.a. n.a. n.a. 0.8 0.4 0.4
Support to aviation
companies
(through KOB)b n.a. n.a. n.a. n.a. 0.1
Provisions to Ceska
Inkasni (Cl)' n.a. n.a. 3.4 10.3 5.5 6.0
Stabilization pro-
gram of CF
(through KOB) b n.a. n.a. n.a. n.a. 0.6 1.8
Others, included in
reported budget
deficit:
Transfers accord-
ing to state
budget law' 9.5 19.4 10.7 n.a. n.a. n.a.
n.a. Not applicable.
. . Negligible.
Note: KOB = Konsolidacni Banka; CF = Ceska Financni.
a. Excluding transfers to KOB, Cl, and CF and transfers according to state budget law.
b. These items are excluded from the "true" deficit calculation. National Property Fund's
expenditures related to KOB, CI, and CF are accounted for as financing items of these
institutions.
Sources: National Property Fund's annual reports and calculations by the authors.
210 BRIXI, SCHICK, AND ZLAOUI
Table 9.4 Risk of Guarantees Issued, Czech Republic,
1993-98
(face values, billions of Czech koruny)
Risk 1993 1994 1995 1996 1997 1998
Very high (90%) 0.0 0.0 0.0 10.8 51.7 31.0
High (30%) 0.0 0.0 0.0 16.2 20.3 0.0
Medium (15%) 5.0 0.0 13.3 3.0 5.8 7.8
Low (5%) 3.7 0.0 1.8 0.0 0.0 87.0
Total 8.7 0.0 15.1 30.0 77.8 125.8
Sources: Ministry of Finance and calculations by the authors.
guarantees issued according to their risk ranking. The implicit subsidy
(risk-adjusted) imbedded in state guarantees is calculated by multiply-
ing the loan amount for which a guarantee was issued by the default
risk. To avoid double counting, the net implicit subsidy, or the net
contribution to the hidden deficit in a given year, is defined as the total
implicit subsidy provided in a given year minus guarantee claims paid
from the budget and reported in the budget that year. Table 9.5 pro-
vides the risk-adjusted amounts of guarantees issued each year and the
claims paid from the budget on guarantee defaults each year.
Estimates of the "true" fiscal deficit in the Czech Republic (Table
9.6) indicate that, contrary to the widely accepted view, the Czech
Republic's fiscal performance has not been noteworthy for its fiscal
restraint. Moreover, demands on new guarantees and programs to be
financed through various off-budget agencies have been growing. If
left to grow as in the past, the off-budget risk to future fiscal stability
Table 9.5 Contribution of Guarantees to "Hidden" Deficit,
Czech Republic, 1993-98
(billions of Czech koruny)
Risk 1993 1994 1995 1996 1997 1998
Very high (90%) 0.0 0.0 0.0 9.7 46.5 27.9
High (30%) 0.0 0.0 0.0 4.9 6.1 0.0
Medium (15%) 0.7 0.0 2.0 0.4 0.9 1.2
Low (5%) 0.2 0.0 0.1 0.0 0.0 4.4
Subtotal 0.9 0.0 2.1 15.0 53.5 33.4
Budget paid out (-) -0.8 -0.4 -0.8 -0.1 -2.0 -6.7
Total 0.1 -0.4 1.3 14.9 51.5 26.7
(as % of GDP) 0.0 -0. 0 0.1 1.0 3.1 1.5
Sources: Ministry of Finance and calculations of the authors.
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 211
Table 9.6 Czech Republic "True" Fiscal Deficit, 1993-98
(percent of GDP)
1993 1 994 1995 1996 1997 1998
Reported fiscal deficit -0.5 -1.3 0.3 0.5 1.1 1.4
"Hidden" fiscal deficit in
the special institutions
(KOB, CI, CF, and NPF) 3.2 1.9 1.0 0.5 1.0 2.0
"Hidden" fiscal deficit in
guarantees (net hidden
subsidy, risk-adjusted) 0 0i 0.1 1.0 3.1 1.5
'True" fiscal deficit 2.7 0.6 1.4 2.0 5.2 4.9
Source: Calculations of the author.
would increase significantly. By mid-1999, the government had only
begun to develop an institutional mechanism to keep a check on its
off-budget obligations and the ensuing fiscal risk.
The sharp increase in the amount and risk of guarantees issued by
the state is troubling. The bulk of tlhe increase has emerged from the
government's support of the banks and the Czech Railways. In 1997
and 1998 the government issued a CZK22 billion (1.4 percent of GDP)
guarantee to Czech National Bank on some of its very risky lending
for bank restructuring and a CZK31 billion (nearly 2 percent of GDP)
guarantee to a bank (CSOB) on its claim against a Slovak financial
institution (Slovenska Inkasni). To support the Czech Railways, the
government issued two guarantees, each over CZK20 billion with a
very high default risk in 1996 and 1997 on railway modernization.
The hidden cost of the guarantees has already started to show as a
growing claim on the budget emerging from guarantee defaults. Claims
on the budget increased from about CZK1 billion annually during
1993-96 to CZK2 billion in 1997 and almost CZK7 billion in 1998.5
Another related, troubling fact is the rapidly increasing level of hid-
den public liabilities. Stocks of these liabilities have been accumulat-
ing outside the budgetary system as a result of the hidden deficits (annual
flows), mainly in the form of borrowing by the special institutions to
finance their government programs.6 Table 9.7 shows the approximate
levels of hidden public liabilities, excluding the nonguaranteed quasi-
fiscal operations of the Czech National Bank. Comparison of the fig-
ures of hidden deficits in Table 9.2 with the resulting hidden liabilities
in Table 9.7 illustrates the extent of cross-financing among the special
institutions and of the use of privatization revenues to partly cover the
cost of off-budget programs.
Off-budget programs, such as guarantees and support extended
through Konsolidacni Banka, the National Property Fund, PGRLF (an
212 BRIXI, SCHICK, AND ZLAOUI
Table 9.7 Hidden Public Liabilities, Czech Republic, 1993-98
(billions of Czech koruny)
1993 1994 1995 1996 1997 1998
Konsolidacni Bankaa 79 81 79 70 86 98
Ceska Inkasnia 20 27 25 17 8 7
National Property Fund' 29 33 40 22 17 15
State guaranteesb 3 3 6 28 74 107
Hidden public liabilities, 131 144 150 137 185 226
Hidden public liabilities
(% of GDP)' 13.1 12.1 10.8 8.7 10.9 12.4
Reported gross government
debt 159 162 154 155 173 194
Reported gross government
debt (% of GDP) 15.8 13.7 11.2 11.2 10.3 10.6
a. Activities of Konsolidacni Banka include financing of the stabilization program of
Ceska Financni. Therefore, the table does not include Ceska Financni as a separate entity.
All figures are nct of provisions and reserves.
b. Guarantees outstanding at the end of each year adjusted for risk.
Soiurces: Konsolidacni Banka, Ceska Inkasni, National Property Fund, and calcula-
tions of the authors.
agricultural credit and guarantee fund), and other, possibly new agen-
cies and guarantee funds, impose costs on taxpayers with a delay but
with no discount. As has already begun to happen, past hidden deficits
and servicing of the hidden government debt outside the budgetary
system generate claims on the government budget.
State guarantees generate significant budget claims. Assuming that
no new state guarantees are issued, the budget may need to cover about
CZK4 billion annually in future years and CZK33 billion in 2002 if
the debt of Slovenska Inkasni to CSOB is not resolved. Table 9.8 builds
on Table 9.5, and, taking into account individual guaranteed debt re-
payment schedules, it shows the expected guarantee claims on future
budgets. Figures in Table 9.8 are obtained by multiplying the default
risk by the annual scheduled payments. More conservative assump-
tions of default risk would increase the estimated claims on budget
resources.
Another source of future claims on the budget is Konsolidacni Banka.
It experienced a loss of about CZK14.4 billion in 1998 that later was
covered by a state bond issue. New programs, however, have required
further borrowing and continue to generate losses. Claims on the state
budget reached CZK80 billion during 1999-2000.
Without further privatization revenues, the National Property Fund
will need to borrow further to meet its commitment for Ceska
Financni, Ceska Inkasni, environmental recovery, and railway devel-
opment, and to cover principal repayments for its obligations.7 Analy-
THE CHALLENGES OF FISCAL RISKS; IN TRANSITION 213
Table 9.8 Estimated Guarantee Claims on the Budget,
Czech Republic, 1999-2030
(billions of Czech koruny)
Guaran-
tees out- Average Expect-
standiing default ed total 1999- 1999-
in 1998 risk claims 1999 2000 2001 2002 2003 2003 2030
284.8 38% 107.4' 3.3 4.9 5.4 33.3 3.7 50.5 97.8
a. The government guarantees both interest and principle repayments. The figure shown
is the net present value of guarantee claims on future budgets.
Sources: Ministry of Finance and calculations of the authors.
sis of the National Property Fund's commitments, excluding those
for Konsolidacni Banka, suggests that, to meet its obligations, the
fund will need about CZK15 billicn a year during 1999-2003 and
about CZK33 billion in 2004.
In the medium to long term, off-budget financing of government
activities, guarantees, and other contingent liabilities surfaces as in-
creases in government debt. In the Czech Republic, the expected in-
crease in the public debt by the arnount of hidden public liabilities
(estimated in 1999 at around 12.4 percent of 1998 GDP as shown in
Table 9.7) is significant but not disastrous. What appears as disastrous
is the dynamic in the rise of the hidden public liabilities. Clearly, the
levels of new guarantees issued and the new government programs
entrusted for financing to Konsolidacni Banka are not sustainable. Their
continued growth at the current pace would in a few years endanger
fiscal stability and thus play against the country's objective of EU ac-
cession. The situation will appear more serious if "implicit" govern-
ment liabilities are included in the deficit and debt calculations.
Finally, it remains questionable to what extent off-budget programs
contributed to achieving the main policy objectives of the government.
In some instances, these programs substituted for structural reforms.
For example, instead of preventing problems in the banking sector from
recurring, bailouts of banks and etiterprises paid for failures likely to
occur again. Sometimes programs that did not qualify for budgetary
support (for example, an additional subsidy to the Czech Railways) did
qualify for assistance outside the budget (such as a very risky guarantee
extended to the railways). Moreover, often these programs implied that
government would help again in the event of future failures, and thus
may have generated moral hazard among market agents, reducing the.r
incentives to improve productivity and competitiveness.
Because EU accession and integration with European markets are
the highest policy priority, the government has a powerful incentive to
enhance its fiscal management. Recent moves in this direction include
214 BRIXI, SCHICK, AND ZLAOUI
periodic reporting on contingent liabilities and fiscal risks, new laws
on budget management and state asset management, consolidation of
the assets and liabilities of Ceska Financni and Ceska Inkasni in a
single portfolio under Konsolidacni Banka, a law restricting the func-
tions of Konsolidacni Banka, privatization of the large state-owned
banks, the planned introduction of a medium-term expenditure frame-
work that would reflect off-budget fiscal risks, and a new capacity in
the Ministry of Finance to analyze and manage fiscal risks. If imple-
mented, these initiatives should significantly narrow the gap between
the reported and true budget deficit and improve the overall effective-
ness and efficiency of the use of public resources.
Transparency and Containment
of Fiscal Risk in Hungary
Compared with the Czech Republic, Hungary emerged in 1990 with a
much larger government debt and a higher expenditure-to-GDP ratio.
But Hungary has been more deternined and successful in managing
fiscal risk. In liquidating old contingent obligations and undertaking
new ones, the government has been guided by two principles: explicit
risks should be identified, and new risks should be undertaken within
the approved budget framework. These principles-transparency and
containment-have not always been applied consistently, but they have
put the government in a better position to withstand economic setbacks.
Transparency has been achieved as the risks and costs of past policies
have been made explicit and, where appropriate, funds have been set
aside to pay for them. The pension system is an example. Restructuring
the pension system compelled the government to recognize costs that
had long been concealed and make provisions in the budget for them.
Containment has proceeded along two tracks: liquidating preexist-
ing risks through pension reform, bank consolidation, and enterprise
consolidation, and regulating the volume of new risks through the
rules and procedures discussed in the previous section of this chapter.
Although these controls are on a cash basis and do not use sophisti-
cated risk assessment techniques, they may nevertheless be effective.
Recent fiscal performance attests to the prudent management of
risk. The amount set aside in the budget for calls on individual and
institutional guarantees is less than 1 percent of total budgeted expen-
ditures. The amount paid out in actual calls has consistently been less
than the budgeted amount. The contingent liabilities of state institu-
tions are barely half of the authorized level. Conservative estimates of
calls from outstanding guarantees of all types are less than 2 percent of
budgeted expenditures. Government and state institutions have estab-
lished several reserve funds to cover fiscal risks, including a deposit
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 215
insurance fund, the private pension guarantee fund, and funds provi-
sioned by various institutions to pay for calls on the risks they hold.
The draw on these funds has been very small thus far. Risk in the
financial sector has been diversified and transferred (in most cases) to
foreign investors. Financial markets have given Hungary a much more
favorable rating than in the past, as reflected by the government's suc-
cess in floating longer-term bonds.
Like in the Czech Republic, early in the transition, fiscal stability in
Hungary was undermined by the bat!king system and its large volume
of rnonperforming loans. To finance bank restructuring and encourage
privatization, the government issued bonds and guarantees (Table 9.9).
Because they bypassed the budget, bonds and guarantees created "hid-
den deficits." Only guarantees, however, in a very small amount, gen-
erated a hidden debt.
Table 9.9 Fiscal Cost of Banking Problems, Hungary, 1991-98
(billions of Hungarian forints)
1991 1992 1993 1994 1995 1996 1997 1998
Bonds issued, 0.0 0.0 285.6 47.0 6.0 9.0 0.0 182.0
Guarantees
issued 10.4 0.0 0.0 0.0 0.0 16.5 27.8 44.0
Total "hidden"
cleficit 10.4 0.0 285.6 47.0 6.0 25.5 27.8 226.0
Amortization
of bonds
from budget 0.0 0.0 0.0 0.0 0.0 0.0 85.9 0.0
Interest paid on
bank consoli-
dation bonds 0.0 0.0 0.0 54.5 96.6 102.7 86.5 50.5
Guarantees
called 0.0 2.3 0.4 0.0 6.8 2.9 11.8 6.2
Guarantees
recovered 0.0 0.0 0.0 0.0 -0.2 -0.1 0.0 -1.1
Total realized
fiscalcost 0.0 2.3 0.4 54.5 103.2 105.5 184.2 55.6
Total realized
fiscal costP
(% of GDP) 0.00 0.08 0.01i 1.25 1.85 1.53 2.16 0.58
a. Bank consolidation was largely financed directly from new debt issues, bypassing the
budget. Bonds issued in 1993 to finance bank consolidation (over 8 percent of GDP)
partly also served debtor (enterprise) consolidation.
b. The total realized fiscal cost, as shown through the government budget, is calculated
on a cash basis as payments on bond amortization and interest plus payments on guaran-
tees called, minus revenues from guarantees recovered. The amounts of bonds and guar-
antees issued suggest how the fiscal cost of bank consolidation will affect future budgets.
Sources: Ministry of Finance and Government Debt Management Agency.
216 BRIXI, SCHICK, AND ZLAOUI
In its first stage, reform of the banking sector did not contain fiscal
risk nor was its fiscal cost transparent. The first stage involved govern-
ment acquisition of a substantial portion of the bad debt held by banks.
The loan strategy transferred risk from the banks to the government,
but the true cost was concealed by treating the transactions as asset
purchases rather than as subsidies. Although some of the acquired loans
were sold or restructured, most were transferred to state agencies (such
as the Hungarian Development Bank) or written off. The second stage
of reforms entailed risk-sharing, with assisted banks required to sign
consolidation agreements spelling out the measures they would take
to improve their financial condition. This strategy was mainly financed
from 20-year bonds. In the 1994-96 period, annual interest payments
on these loans averaged approximately 1.5 percent of GDP. Successful
consolidation set the stage for privatization of most banks,' allowing
the government to transfer most of the risk into foreign hands.
However, the government violated its transparency rule in the case
of Postabanka. A run on the bank in 1997 prompted the government
to provide new equity capital and loan guarantees. The government
channeled most of the assistance through the Hungarian Development
Bank and the State Privatization and Holding Agency, thereby creat-
ing not only a "hidden" fiscal deficit but also a hidden debt (liabilities
accumulated in these two agencies in exchange for low-quality assets).
In late 1998 the government was compelled to recapitalize Postabanka
by injecting HUF150 billion (1.5 percent of GDP) from government
bonds and to compensate the Hungarian Development Bank in the
amount of HUF40 billion for Postabanka-related losses.
Compared with the Czech Republic, Hungary chose a more trans-
parent approach to enterprise privatization and to the use of
privatization revenues. Most of privatization proceeds were used to
repay foreign debt (see Table 9.10). As for the privatization process
itself, Hungary proceeded slowly, company by company, restructuring
and managing them before selling them, and insisted on cash sales
rather than on vouchers and other noncash transactions implemented
in the Czech Republic. The State Property Agency has not warranted
the future financial condition or performance of privatized enterprises,
nor has it indemnified the new owners for the cost of meeting new
environmental standards. It has issued several types of guarantees in
the privatization process, but maintains a reserve fund, which appears
sufficient to cover expected claims.
Hungary's pension reform of 1997 reduced the government's larg-
est implicit liability and made it explicit.9 The reform required new
workers to enroll in a fully funded, privately managed, defined-contri-
bution system and gave most existing workers the option of remaining
in the government-operated, defined-benefit program or migrating to
the new system. The reform significantly reduced the government's
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 217
Table 9.10 Use of Privatization Revenues, Hungary, 1990-98
(billions of Hungarian forints)
1990 1991 1992 1993 1994 1995 1996 1997 1998
Government debt
repayment 0.5 22.4 51.5 57.5 151.7 218.4 22.9 254.8 39.0
Transfers to state
budget 0.0 0.0 0.0 0.0 0.0 150.0 100.0 0.0 0.0
Municipalities 0.0 2.3 4.8 3.4 6.0 6.1 21.6 26.4 0.0
Direct privati-
zation costs 0.0 1.1 6.2 7.6 25.3 33.6 33.0 36.1 46.3
Reorganization
ccsts 0.0 0.0 8.7 49.5 8.0 9.8 16.8 12.1 13.5
Guarantees 0.0 0.0 5.8 7.8 7.0 3.7 33.2 16.0 51.6
Total 0.5 25.8 77.0 125.8 198.0 421.6 227.5 345.4 150.4
Total
(% of GDP) 0.02 1.1 2.6 3.6 4.5 7.5 3.3 4.1 1.5
Sources: Ministry of Finance and State Property Agency.
future fiscal exposure, but it raised the short-term budget deficit by
requiring the government to make up shortfalls in the old pension
fund. The more workers who switched, the greater the reduction in
future liabilities, but the greater the loss of income in the old fund, the
greater the reported budget deficit.
To build public support for pension reform, the government guar-
anteed that each participant in the new, privately managed pension
funds would receive benefits equal to at least 93 percent of the pension
that would have been paid by the old system. Because only workers
below the age of 47 had the option of joining the new pension system,
the government will not face any call on this guarantee for at least 15
years, when the first cohort of participants in the defined-contribution
plan retire. If the guarantees are called in the future, payments will be
made by the new Private Pensions Guarantee Fund, which accumu-
lates reserves for this purpose by charging privately managed pension
funds fees for various services. There is some risk that these guarantees
may spur participants to enroll in funds that promise the highest yields,
without concern about whether the funds will have sufficient resources
to cover their promises. To guard against undue risk-taking, the gov-
ernment has introduced various regulatory measures to oversee the
privately managed funds.
Throughout the transition process, Hungary also has taken new
fiscal risks by guaranteeing loans, indemnifying importers and export-
ers, and taking other steps to promote investment and entrepreneurial
218 BRIXI, SCHICK, AND ZLAOUI
behavior. Moreover, Hungary has authorized a number of quasi-inde-
pendent institutions to issue guarantees, including Eximbank, MEHIB
(Hungarian Export Credit Insurance Ltd.), the Credit Guarantee Com-
pany, the Rural Credit Guarantee Foundation, and the Hungarian
Development Bank, and thus has accumulated contingent liabilities
(see Table 9.11).
Overall, Hungary's government has been prudent in issuing guar-
antees directly or through state-established institutions. Compared with
the Czech Republic, Hungary has maintained a set of strict controls
that limit the volume of guarantees, making provision in the budget,
and require assessment of risks prior to the execution of guarantee
contracts and timely reporting on the risks of outstanding guarantees
(see Box 9.1). The controls are cash-based; they limit the volume of
new guarantees or the total outstanding amount and set aside cash in
Table 9.11 Contingent Liabilities Outstanding and Expected
Claims, Hungary, as of 1998-99
(billions of Hungarian forints)
Out- Average Expected
Type Ceiling standing risk (%) claims
Individual guarantees
Individual guarantees
(within the limit) 28a 120 30 36
Individual guarantees
(beyond the limit) 158b 125 5 6
Guarantees to activities
of specific institutions
Hungarian Development Bank 80 32 5 2
Eximbank 75 52 7 3
MEHIB-Hungarian Export
Credit Insurance Ltd. 185 76 8 6
Credit Guarantee Co. 49 42 5 2
Rural Credit Guarantee
Foundation 23 10 5 1
Total n.a. 457 n.a. 56'
n.a. Not applicable.
Note: The table excludes guaranteed loans from international financial institutions.
a. The 1998 ceiling on the issue of new guarantees.
b. Total amount of individual guarantees beyond the percentage limit.
c. Because the government guarantees both interest and principal repayments, this fig-
ure roughly represents the net present value of future claims. The 1998 budget appropri-
ated HUFI 1.8 billion for guarantee calls and actually paid HUF7.9 billion. The 1999
budget set aside HUF12 billion for calls on individual guarantees and another HUF9.5
billion to cover guarantees issued by state institutions.
Sources: Ministry of Finance and calculations of the authors.
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 219
Box 9.1 Hungarian Public Finance Institutions Designed
to Constrain the Accumulation of Contingent Liabilities
Hungary's Public Finance Act is the legal basis for issuing and budgeting
for guarantees. It provides that the annual budget limit the volume of
guarantees undertaken during the fiscal year. It also requires that funds
be set aside in the budget for calls expected during the year. And it re-
quires the government to publish information on borrowers, the reasons
for issuing specific guarantees and the amounts, and the risk and the con-
ditions pertaining to the guarantees. T he government must report each
guarantee to the State Audit Office, and the closing accounts for each
fiscal year must include all guarantees undertaken and payments made.
The next link in the control process is the annual budget, which limits
the volume of guarantees to be issued (or outstanding) during the year
and sets aside funds for guarantees that may be called. At present, the
budget limits the total of new guarantees to 1 percent of budgeted expen-
ditures. The government generally has stayed within the limit, but for
several reasons it issues guarantees or incurs related obligations well in
excess of the limit. The limit applies only to individual guarantees issued
clirectly by the government; it does not cover guarantees issued by state
institutions. Guarantees for various strategic purposes (such as for oil and
gas imports) and for loans contracted with international finance institu-
tions are exempt from the limit. The limit may be exceeded during the
year by a decision of the parliament.
The budget contains numerous provisions regulating the issuance of
guarantees by the government and state institutions. The following are
among the most important. The government may charge an origination
fee of up to 0.5 percent of the value of its commitment. The budget limits
the amount of guarantees that each state institution may have outstand-
ing at any time during the year. (In sorne cases, the limit is adjusted annu-
ally to accommodate increased credit activity. For example, the 1999
budget raised the ceiling for the Hungarian Export Credit Insurance Ltd.
to HUF250 billion from the HUF1 85 billion designated the previous year.)
The government may reinsure up to a, specified percentage of the obliga-
tions of various state institutions (for example, it reinsures 70 percent of
the obligations of the Credit Guarantee Company and the Rural Credit
Guarantee Foundation). The budget provides for guarantee programs such
as new student loan guarantees, and it appropriates a fixed amount to
cover potential calls during the year.
The third link in the control chain is a decree prescribing the proce-
dures to be used in undertaking individual guarantees. The decree cur-
rently in effect designates the minister of finance as guarantor on behalf
of the state and specifies the form and content in which proposed guaran-
tees should be presented to the government for decision. The proposal
(Box continues on the following page.)
220 BRIXI, SCHICK, AND ZLAOUI
Box 9.1 (continued)
should include an explanation of the reasons for the guarantee, informa-
tioIn on any previous guarantees issued to the same borrower, and an
assessment of the probability that the guarantee will be called. The gov-
ernment is required to publish decisions on individual guarantees in the
official Hungarian Gazette. It maintains an up-to-date database on out-
standing guarantees, including information on the purpose of each trans-
action, the amount guaranteed, the date the guarantee was contracted,
and when the government obligation will expire. During 1998, official
records showed that the government entered into 34 individual guaran-
tee contracts within the 1 percent limit, two guarantees outside the limit,
and several dozen guarantees on foreign loans by banks and enterprises.
A guarantee becomes effective only when a properly executed contract
has been signed. The Ministry of Finance has prepared several standard
contracts for various types of arrangements. Each contract includes pro-
visions protecting the government's interest in case of default and proce-
dures for recovering payments made pursuant to calls.
the budget to cover possible calls during the fiscal year. The controls
are not self-enforcing, however. Their effectiveness depends on the
extent to which the rules are followed.
Despite its generally prudent record, the government still holds sig-
nificant risks. First, the budget constraints are not as constrictive as
they appear to be. Guarantees tendered outside the 1 percent budget
limit exceed those within the limit. Second, state institutions have be-
come active risk-takers, and (as occurred in the use of the Hungarian
Development Bank as a conduit for bailing out Postabanka) some-
times at the request of the government. Third, in view of the country's
high public expenditure and debt ratios the government may have little
margin in financing contingent obligations that come due. Fourth,
Hungary, like other countries, still faces some implicit liabilities in the
financial sector, in privatized enterprises, and in privately financed
infrastructure. Fifth, there is pressure on government to undo some
reforms because previously unrecorded liabilities now burden the bud-
get. So far, the government has not succumbed to the temptation to
show a more favorable budget posture by, for example, reverting to
the old pension system and hiding future pension liabilities.
The favorable fiscal posture may, however, lull the government into
taking risks that it may have avoided during the adjustment years. In
1999, it increased the ceilings on contingent liabilities of various insti-
tutions and has boosted the ceiling on new individual state guarantees
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 221
from 1 percent to 1.5 percent of state revenues. Moreover, it has pro-
posed new, large highway construction programs, to be implemented
through the Hungarian Development Bank. Using such an off-budget
channel to finance a huge public investment program may open the
door to other schemes in the future.
Fiscal Risks of Transition under the
Currency Board Arrangement in Bulgaria
After the introduction of a currency board arrangement in 1997-2000,
Bulgaria's government achieved rnajor success in fiscal adjustment.
T'he tight macroeconomic constraint introduced by the currency board,
however, also reduces the government's capacity to absorb risks and
respond to future possible shocks by increasing its borrowing. Thus
Bulgaria's fiscal performance is more vulnerable to risks than those of
the Czech Republic and Hungary, where government borrowing is less
constrained. Applying the Fiscal Risk Matrix, Bulgaria's government
has identified the main sources of its risk exposure (see Table 9.12).
Sovereign Debt
The size and structure of Bulgaria's public debt are somewhat worri-
some. After the 1992-94 successful restructuring, the government re-
dluced its external debt to 82 percent of GDP and domestic debt to 4
percent of GDP by the end of 1999 (Figure 9.2). It also announced
ifurther reductions for the future. Prudent debt management allowed
Bulgaria to reclaim investors' confidence. Over 1998-2000, the aver-
age maturity of domestic Treasury bills doubled, reaching almost two
years, and the sovereign credit rating on foreign debt stabilized at B2/
B+ with a positive outlook. In spite of continued debt reduction, how-
ever, Bulgaria remains one of the most heavily indebted countries of
Central and Eastern Europe.
The current structure of government debt may give rise to increases
in the cost of future debt service and new borrowing (Figure 9.3). The
reason is that several risks are at play. Refinancing risk, as measured
by the maturity structure of public debt and its volatility, appears lim-
ited, but so are the refinancing options. For external debt, the average
portfolio maturity is long-over 13 years-which limits refinancing
risk. However, neither restructured Paris and London Club obliga-
tions nor the debt to international financial institutions can be rolled
over easily. Similarly, the government will find it difficult to refinance
its long-term foreign currency-denominated domestic debt. Refinanc-
ing risk thus remains important as long as the domestic bond market
222 BRIXI, SCHICK, AND ZLAOUI
Table 9.12 Fiscal Risk Matrix, Bulgaria
Sources of Direct liabilities Contingent liabilities
obligations (obligation in any event) (obligation if a particular event occurs)
Explicit
Government o Foreign and domestic o Individual state guarantees for
liability as sovereign debt (size nonsovereign borrowing and
recognized and structure) [H] obligations [L]
by a law or o Future pension ex- o Obligation to recover past
contract penditures required environmental damages assumed
by law [M] in enterprise privatization and
o Health expenditures other environmental liabilities [Ml
required by law [M] - Obligations of business promo-
tion bank [L]
o Obligations of export insurance
agency (insurance policies to
cover political and medium-term
commercial risks) [L]
o Obligations of state fund for
agriculture [L]
Implicit
A "moral" a Accumulated and o Environmental commitments for
obligation expected public in- still unknown damages and
of govern- vestment needs to nuclear and toxic waste [U]
ment that sustain delivery of o Cleanup of enterprise arrears and
reflects public public services and liabilities IM]
and interest meet key require- o Default of municipalities on own
group ments for accession nonguaranteed debt, own
pressures to the EU IH] guarantees, or own obligations to
o Future recurrent provide critical public services IM]
costs of public in- - Support to the banking sector in
vestment projects case of crisis [L]
Note: Risk level: H = high; M = medium; L = low; U = unknown. Obligations listed refer
to the fiscal authorities, not the central bank.
Source: The authors.
remains shallow and unimpressed by government bonds of three-year
and higher maturities.
Currency risk is less important. The currency board arrangement
pegging the Bulgarian leva to the euro appears credible and thus re-
moves foreign exchange risk. Cross-currency risk remains a fiscal
risk, particularly with respect to the appreciation of the dollar vis-a-
vis the euro.10
Most problematic appears the interest rate risk. At the end of 1999,
three-quarters of government debt had a floating interest rate. During
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 223
Figure 9.2. Bulgaria's Public Debt, 1993-99
US $million
16,000
14,000 =
12,000. _ .
1f)000.
3,000
6,000
4,000
2,000 oilleStic
1993 1994 1995 1996 1997 1998 1999
Source: The authors.
Figure 9.3. Structure of Bulgaria's Public Debt
as of December 31, 1999
Total Public Domestic
G-24 and EU 4.5%'o
Guarantees 6.2%
IMF Ote
World Bank 10.7% 1Othr
7.8% J Leva-donominated
Paris Club - 4.0%
7.7% FX denominated
> / Other ; 4.8%
)4 I I Guarantees (Leva)
4.8%
Guarantees (FX)
0.8%
London Club
47.2%
Note: IMF = International Monetary Fund; FX = foreign exchange.
Source: The authors.
224 BRIXI, SCHICK, AND ZLAOUI
1997-98, the exceptionally low LIBOR (London interbank offered rate)
allowed the government to keep debt service at about 10 percent of
central government revenues. Any increases in the LIBOR, however,
will increase the government's debt burden significantly.
In the short run, given its financing constraints, the government can
do little to alleviate the interest rate risk. As much as possible, how-
ever, it will need to issue new debt at a fixed interest rate.
Pension and Health Care Expenditures
By the late 1990s, the pension and health care systems in Bulgaria had
become highly inefficient and financially unsustainable. The pension
system suffered from adverse demographics, generous entitlements, a
low retirement age, and declining revenues despite high contribution
rates. The universal and nominally free health care system was charac-
terized by low investment, declining quality of health services, and in-
creasing costs to the patients in terms of side payments (such as user fees
and bribes). To strengthen its fiscal outlook and the quality of services
provided, the government launched ambitious pension and health care
reforms in 1999. The pension reform aims to restore the long-term vi-
ability of the traditional pay-as-you-go scheme by reducing entitlements
and completing it with fully funded voluntary and compulsory pillars
(Figure 9.4). Under the health care reform, a Health Insurance Fund
Figure 9.4. Projected Deficit of Pillar I with and without
the Reform, Bulgaria, 1999-2047
Percent of GDP
0.5
0
-0.5
-1.0
-1.5
-2.0
-2.5 - Old System
-New Systemi
-3.0
1999 2009 2019 2029 2039
Source: Government of Bulgaria.
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 225
created in 1999 is expected to gradually contract out health care provi-
sion to competing public and private providers. The reforms tempo-
rarily increase government spending in both sectors. The health care
reform involves up-front costs to build institutional capacity and gradu-
ally recapitalize Bulgaria's health care facilities. Transition to the new,
funded components of the pension system while maintaining a pay-as-
you-go scheme entails up-front revenue losses for the government in
1999-2001 before delivering financial improvement.
The Financial Sector
In the financial sector, the 1996-97 hyperinflation, crisis, and subse-
cluent policy actions by the Bulgarian National Bank effectively cleaned
up the banking sector. The ensuing reform successfully capped both
cxplicit and implicit government obligations for lost deposits and failed
banks. Since then, improved supervision has controlled the exposure
of banks to liquidity and interest rate risks. The quality of banks' loan
portfolios and their foreign exchange exposure are not a significant
fiscal risk so far.
State Guarantees and Guarantee Funds
Bulgaria's explicit contingent liabilities are modest. The government
has applied prudent limits and regulations to state guarantees. The
government entered the year 2000 reporting guaranteed debt (face
value) of about 17 percent of GDP. Of this amount, 9 percent of GDP
was in the form of the central bank's debt to the International Mon-
etary Fund (IMF) and a further 4 percent of GDP was in the form of a
deposit guarantee that expired in April 2000. Calls on the remaining
guarantees (about 4 percent of G]DP outstanding in 2000) would have
limited fiscal cost (Figure 9.5). In addition, the government guarantees
obligations of agencies such as the State Fund for Agriculture, the Ex-
port Insurance Agency, and the Business Promotion Bank. These guar-
antees also are small so far, but, as we discuss in the next section, the
government is likely to face increasing pressure to provide guarantees
and other forms of off-budget support in the near future.
Economic Restructuring and Investment Requirements
Compared with the Czech Republic, Hungary, and several other more
advanced EU accession countries, Bulgaria has delayed most struc-
tural reforms and investment. Only in 1997 did Bulgaria stabilize its
macroeconomic conditions and embark on a comprehensive program
of reforms to lay the foundations of a market-based economy. To sup-
port these reforms, and particularly to introduce financial discipline in
226 BRIXI, SCHICK, AND ZLAOUI
Figure 9.5. Fiscal Impact of Called Guarantees, Bulgaria,
1999-2004
Percent
3.0
2.5 80% fall
2.0-
260% fall
1.5
1 0 ~~~~~~~4
1.0 4%fl
0.5 =
0
1999 2000 2001 2002 2003 2004
Source: The authors.
banks, enterprises, and government agencies, the government estab-
lished the currency board arrangement. In addition, the government
needs to support the reforms financially. The reform program required
expenditure on protecting the poor, modernizing public administra-
tion, reforming the judicial system, renewing infrastructure, and ad-
dressing environmental problems. Low levels of public investment
(averaging about 1-2 percent of GDP during 1993-98) and deferred
maintenance during most of the 1990s only sharpened the challenge of
transition and of meeting the requirements of EU accession in the early
2000s. Under its 2000-2004 public investment program, the govern-
ment plans to spend 3-3.5 percent of GDP annually. The objectives of
economic growth and EU accession, however, call for significantly more
new investment as well as for low budget deficits. Thus the govern-
ment will need to encourage private sector initiatives. As long as com-
mercial banks are opposed to providing long-term credits and foreign
creditors lack confidence, this in turn will require government guaran-
tees through the various existing and possibly new state-guaranteed
agencies. Box 9.2 illustrates the fiscal risks that have arisen in Bulgaria
from private participation in the energy sector.
Fiscal Vulnerability
Bulgaria's current fiscal position appears solid. In the late 1990s, rev-
enue performance was strong and resilient to the economic shocks
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 227
Box 9.2 Fiscal Risks Associated with Bulgaria's
Energy Sector
In the energy sector, fiscal risks arise mainly from (a) explicit state
guarantee contracts for loans to state-owned energy enterprises and
(b) explicit and implicit state guarantees of long-term take-or-pay con-
tracts. Contracts of both types are likely to increase rapidly if invest-
ments to modernize energy infrastructure and to meet European Union
accession requirements are not carefully selected or properly structured.
The years 1999-2000 witnessed a high demand for state guarantees
on energy projects. The government considered credit guarantees on en-
ergy investments in the amount of about US$850 million to be imple-
mented during 2000-2004. These include investments in nuclear plant
safety, waste disposal, and plant upgrades ($380 million); electricity trans-
mission and dispatch ($150 million); district heating ($120 million); and
expansion of gas transit capacity to Turkey ($47 million).
Take-or-pay guarantees also are demanded, and the one such guaran-
tee already provided is expected to be called. In 1998 the government
provided a long-term take-or-pay guarantee on a gas contract between
Bulgargaz and Gazprom. This guarantee requires Bulgaria to take or pay
for predetermined annual volumes of gas until 2010. The contracted vol-
ume of 4 billion cubic meters for 1999 was valued at US$320 million. A
slightly higher volume was contracted for 2000 and onward. In Bulgaria,
however, the demand for gas has been declining and the government is
expected to pay for the contracted volume that is not taken.
Expected power purchase agreements between the state-owned Na-
tional Electricity Company and privatized power producers also call
for long-term take-or-pay state guarantees. A state guarantee is sought
on the electric company's obligation to purchase electricity for about
US$180 million a year for 15 years upon completion of a $400 million,
840-megawatt rehabilitation project, and for about $175 million a year
for 10-15 years upon completion of a new $1.0 billion, 600-megawatt
plant. (Source: Brixi, Shatalov, and Zlaoui 2000.)
caused by, for example, the Russian crisis and the Kosovo war." Gen-
eral government revenues increased from 31.4 percent of GDP in 1997
to 36.8 percent of GDP in 1998 and exceeded 40 percent of GDP in
1999, providing adequate coverage for expenditures that rose from
35.8 percent of GDP in 1998 to 41.2 percent of GDP in 1999, mainly
driven by the social sectors.
Room to accommodate fiscal risks, however, is limited. First, al-
though the currency board arrangement is effective in achieving fiscal
stability, it does by definition reduce the range of options otherwise
228 BRIXI, SCHICK, AND ZLAOUI
available for deficit financing and therefore the scope for fiscal expan-
sion or for accommodating sudden financing pressures. Out of the
four possibilities that are available to most countries in financing their
public sector deficit (printing money, reducing foreign reserves, and
foreign and domestic government borrowing), the currency board ar-
rangement inhibits the former two. Foreign and domestic borrowing,
along with exceptional proceeds such as privatization revenues, thus
remains the only means of deficit financing and of raising money to
face sudden shocks.
Following adoption of the currency board arrangement, the main
source of deficit financing shifted from the domestic banking system
(on a net basis) to privatization revenues"2 (Figure 9.6), followed by
external borrowing from official creditors. As the privatization pro-
cess nears an end, revenues from the sale of state-owned enterprises
will subside. The largest and most profitable state-owned enterprises
were already privatized. Further sizable revenues could be expected
from the privatization of BTC (Bulgaria telecommunications company),
Bulgartabac (tobacco company), Bulbank (the largest state-owned
bank), and several power distribution companies in 2000-01, after
which the scope for raising substantial revenues from privatization
shrinks. This may raise questions about the availability of resources
for debt payment and investment financing.
Second, similar to other EU accession countries, Bulgaria's govern-
ment faces a constraint on both the revenue side and the expenditure
Figure 9.6. General Government Overall Balance
and Its Financing, Bulgaria, 1996-99
Percent
15
= Overall balance = Domestic
10 = Net external = Privatization
5
0 -
-5
-10
-15
1996 1997 1998 1999
Source: The authors.
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 229
side. With revenues at about 40 percent of GDP in 1999 and a bias
toward a payroll tax, a further increase in tax rates would damage
investment and growth. In this respect, the government's main objec-
tive is to broaden the tax base and strengthen tax collection in order to
lower tax rates on labor and income. The structure of government
spending is rigid, with a large share of nondiscretionary expenditures.
In 1999 spending was dominated by social security, debt service, op-
erations and maintenance, and wages. Pensions and other social out-
lays accounted for 12 percent of GDP, wages for over 5 percent of
GDP, and interest payments for 4 percent of GDP. The operations and
rnaintenance expenditure amounted to 8 percent of GDP, while capital
and defense expenditures each accounted for 4 percent of GDP.
Fiscal Risk Management
In line with the requirements of a currency board arrangement, Bul-
garia has maintained comfortable levels of both external and fiscal
reserves. In its fiscal reserves account, which consists of the balances
of all government budgetary and extrabudgetary accounts in the bank-
ing sector, the government maintains a certain floor throughout the
fiscal year. This floor was around 11 percent of GDP at the end of
1999 and is expected to rise to accommodate contingent expenditures
on structural reforms, interest payments above the baseline projec-
tion, or a shortfall of official firnancing relative to program projec-
tions. In addition, the government allocates resources within the budget
for contingent expenditures. About 1 percent of GDP in 1999 and 1.2
percent of GDP in 2000 were allocated for possible calls on guaran-
tees and implementation problems of pension and health care reforms.
Large reserves, however, impose opportunity costs of investment and
growth. Therefore, the government has sought to improve its capacity
to analyze, mitigate, and manage risk; reflect risks in its medium-term
fiscal strategy; and combine its reserve requirement with public-pri-
vate risk-sharing mechanisms ancl with a good hedging strategy.
In building its risk management capacity, the government achieved
some success in managing its debt. After rescheduling the stock of
Paris Club and London Club obligations, the government successfully
bought back debt owed to the former Comecon banks at deep dis-
counts, eliminating over US$1 billion of external debt in nominal terms.
In its new debt management strategy, the government set as its main
goals to minimize the volatility of debt servicing cost, foster the deep-
ening of the domestic capital market, enhance investor confidence by
increasing fiscal transparency, attain an investment grade rating for
sovereign debt instruments, and reduce the debt level below the
Maastricht threshold criteria of 60 percent.
230 BRIXI, SCHICK, AND ZLAOUI
To achieve its debt management goals, the government announced
a plan to actively manage its refinancing, exchange rate, and interest
risk exposures through quantitative benchmarks that would preset the
ratio of domestic versus foreign debt and the ratio of the fixed versus
floating interest rate and would gradually raise the duration of domes-
tic and foreign debt. The government also seeks to establish a special
performance benchmark against which to evaluate the actual results
of debt management and to introduce a cost-at-risk framework to
measure the variability of its future debt service. Finally, the govern-
ment has explored derivative instruments to contain interest risk. Given
Bulgaria's sub-investment grade credit rating, which makes derivatives,
such as the needed fixed-for-floating interest rate swaps, pricey, World
Bank hedging products appear attractive."
Similar to the situation in Hungary, the government's institutional
arrangements for managing fiscal risks have been strong in many as-
pects but not totally reassuring. The government established a simple
framework for dealing with guarantees. In particular, it developed a
comprehensive register of guarantees and introduced regular publica-
tion of the aggregate amounts of guarantees outstanding along with
the government debt figures. The register covers all external and do-
mestic guarantees, indicating the beneficiary, creditor, project title,
amount, currency, and debt repayment schedule. The government also
centralized the issuance of new guarantees and subjected each new
guarantee to the executive and legislative scrutiny associated with the
regular budget process.
In terms of nominal limits, Decree 482 of 1997 set the annual limit
on guarantees (face value) issued at 20 percent of expected budget
revenues. A recent amendment to this decree, however, dropped the
complementary ceiling of 20 percent of GDP on the total amount of
guaranteed debt outstanding, replacing it with a flexible ceiling to be
set in the budget process each year. This change significantly expanded
the legal room for the government to issue new guarantees.
Furthermore, under state guarantee contracts the government al-
ways covers the full amount of the debtor's obligation and all risks,
without analyzing their determinants. Such coverage was largely re-
quired in the 1990s, because official creditors, providing concessional
resources for development projects and balance-of-payments support,
dominated the list of creditors under state guarantees. The govern-
ment understands, however, that, if extended to commercial creditors,
this practice would negatively affect market behavior, creating moral
hazard for both debtors and creditors.
Municipal borrowing is subject to a legal limit of 10 percent of the
annual revenues of the respective municipality. Of this amount, as much
as 10 percent of the preceding month's revenues can be in the form of
a short-term, interest-free credit from the central budget. Municipali-
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 231
ties are allowed to seek the remainiing credit from commercial banks
and from other municipalities and request state guarantees for them.
Subject to the approval of the Securities and Exchange Commission,
municipalities are allowed to issue bonds. In 1999 the city of Sofia
successfully placed EUR50 million in Luxembourg. Bond issues are
not covered by explicit state guarantees, but chances are that the cen-
tral government would face pressures to intervene should a city or
municipality become insolvent. Municipal guarantees and other con-
tingent municipal obligations and municipal ownership of enterprises
and financial institutions are not regulated under any existing law.
The Risk versus Investment Dilemma
Bulgaria finds itself in a complex and challenging situation. Macro-
economic stabilization was more painful and required more drastic
measures than in most other EU accession countries. Because the sta-
bilization succeeded, fiscal management has been prudent and trans-
parent but has yet to be tested by the risks of transition and investment
requirements of recovery and EU accession. With an already high level
of indebtedness, Bulgaria faces a difficult tradeoff. Should the country
adhere to a very conservative stance toward debt and fiscal risk, or
should it accept a slower pace of debt reduction and promote to a
larger extent investment, economic restructuring, and programs to
protect the poor and vulnerable?
Concluding Remarks
Our work in the Czech Republic, -lungary, and Bulgaria demonstrates
the importance of including contingent liabilities when assessing the
magnitude of a true fiscal adjustment and when analyzing fiscal
sustainability. To the extent that explicit expenditures are shifted off-
budget or replaced by the issuance of guarantees, the achieved im-
provement in fiscal balances is overstated. For the Czech Republic, we
find that the adjustment may have been overstated by some 3-4 per-
cent of GDP annually. The accumulation of contingent liabilities today
is a threat to future fiscal stability. Thus a stabilization program that is
accompanied by a buildup of contingent liabilities may not be sustain-
able. In Hungary, on the other hand, transparency and containment of
fiscal risks, rather than low budget deficits, have been the guiding
principles of fiscal policy and management. Although government-
reported debt in Hungary has been about five times higher than in the
Czech Republic, Hungary's future fiscal performance is likely to suffer
less from pressures of guarantee claims and hidden liabilities. Since 1997,
Bulgaria has tried to achieve both transparency and low exposure to
232 BRIXI, SCHICK, AND ZLAOUI
fiscal risks as well as low budget deficits. Many of the challenges of
transition that already have been overcome in the Czech Republic and
Hungary, however, have yet to test the Bulgarian government's com-
mitment to fiscal prudence.
There never is a final ending to a country's management of fiscal
risk. One country overwhelmed with the accumulated cost of past hid-
den liabilities takes steps to constrain new risks and to liquidate old
claims in a transparent manner; another country does a good job of
regulating contingent liabilities, but then is lured into hiding new li-
abilities by the promise that the costs will be covered by the dividends
of future economic growth. In comparing the Czech Republic and
Hungary, countries like Bulgaria can learn much from both the fail-
ures and the successes.
Notes
1. Ceska Financni has financed two blocks of programs geared toward bank
revitalization. One block, in the total amount of approximately CZK35 billion,
is financed and guaranteed by the Czech National Bank. The other, called the
stabilization program, in the amount of about CZK12 billion, is financed through
Konsolidacni Banka and thus is guaranteed by the government. It is only the
latter block that is considered in the "true" deficit calculation. It is included as
an activity of Konsolidacni Banka.
2. The assets purchased through off-budget programs are of extremely low
quality. Therefore, the analysis considers asset purchases as a spending program
rather than as a financial transaction.
3. The debt to the Czech National Bank still constitutes about half of
Konsolidacni Banka's total debt.
4. Risk assessment of guarantees issued by the National Property Fund and
Konsolidacni Banka is not available. Therefore, calculation of the "true" fiscal
deficit includes only the implicit subsidy extended through net spending by the
special institutions and through guarantees issued directly by the state, but not
guarantees issued by special institutions.
5. Because the guarantee claims paid from the budget have contributed to the
reported deficit, the "hidden" deficit that emerges from guarantees only includes
the difference between the hidden subsidy extended by the government through
new guarantees and the claims mostly on guarantees issued in previous years.
Unadjusted for guarantee claims, the hidden subsidy through guarantees actu-
ally reached CZK55 billion and CZK32 billion in 1997 and 1998, respectively.
6. Hidden public liabilities are calculated on a gross basis. The analysis fo-
cuses on gross liabilities, because the quality of directed loans extended and
assets purchased through off-budget programs is so extremely low and their
potential value is on average estimated at about 10 percent (3 percent for CI,
less than 10 percent for CF, and under 20 percent for KOB).
THE CHALLENGES OF FISCAL RISKS IN TRANSITION 233
7. The initial bond issue by the National Property Fund was used mainly
to capitalize Konsolidacni Banka.
8. During 1994-97, state ownership fell from 67 to 22 percent, while for-
eign ownership rose from 15 to 61 percent.
9. Hungary inherited a mandatory retirement scheme that promised future
benefits but did not put aside sufficient money in the pension fund to pay for
them. Because it is on a cash basis, the budget does not disclose unfunded pen-
sion liabilities, including those that are direct, explicit obligations of the govern-
inent. But projections showed that as the population aged over the next
half-century, the cash deficit in the pension fund would widen to 6 percent of
GDP.
10. From the balance-of-payments perspective, this risk is partly neutralized
by euro-denominated foreign reserves and dollar-denominated net exports.
11. Investors' confidence as well as renewed growth and macroeconomic
stability emerged from the introduction of the currency board arrangement and
sound policies. Economic growth recovered in 1998, recording a 3.5 percent
positive growth rate, and continued in 1999 with 2.5 percent growth despite
the unfavorable external environment marked by turmoil in the region and
emerging international markets. Inflation was confined to the single-digit level.
12. In Bulgaria, privatization receipts cannot be used for current budget-
ary expenditures. They enter the fiscal reserve account and can be used for
debt repayments and investment financing. The privatization receipts of the
municipalities can be used for ecological projects, investment debt repay-
ments, or writing off nonperforming loans of municipality-owned enterprises.
13. For example, fixed-spread loans allow borrowers to flexibly fix the inter-
est rate on disbursed amounts at any r:ime during the life of the loan, to create a
cap or collar (a floor and cap simultaneously), to unfix or change the rate on
disbursed amounts, and to adjust the currency and loan repayment terms if
needed. This way, the loan maturity and currency structure may be set to smooth
an uneven and sensitive future debt servicing profile. World Bank hedging prod-
ucts such as interest rate swaps, caps and collars, currency swaps, and commod-
ity swaps allow reductions in the risk exposure arising from old World Bank
loans and, possibly in the future, from the country's overall debt portfolio. By
dealing with some of these instruments, countries also gain experience that is
useful later in accessing the international derivatives markets.
Bibliography
Blejer, Mario, and Adrienne Cheasty. 1991. "The Measurement of Fiscal Defi-
cits: Analytical and Methodological Issues." Journal of Economic Literature
29: 1644-78.
Bokros, Lajos, and Jean-Jacques Dethier, eds, 1998. Public Finance Reform
during the Transition: The Experience of Hungary. New York: Oxford Uni-
versity Press.
234 BRIXI, SCHICK, AND ZLAOUI
Brixi, Hana Polackova, Hafez Ghanem, and Roumeen Islam. 1999. "Fiscal
Adjustment and Contingent Government Liabilities: Case Studies of the Czech
Republic and Macedonia." Policy Research Working Paper 2177. World
Bank, Washington, D.C.
Brixi, Hana Polackova, Anita Papp, and Allen Schick. 1999. "Fiscal Risks and
the Quality of Fiscal Adjustment in Hungary." Policy Research Working Paper
2176. World Bank, Washington, D.C.
Brixi, Hana Polackova, Sergei Shatalov, and Leila Zlaoui. 2000. "Managing
Fiscal Risk in Bulgaria." Policy Research Working Paper 2282. World Bank,
Washington, D.C.
Buiter, Willem. 1983. "Measurement of the Public Sector and Its Implications
for Policy Evaluation and Design." IMF Staff Papers 30: 306-49, Interna-
tional Monetary Fund, Washington, D.C.
- 1985. "A Guide to Public Sector Debt and Deficits." Economic Policy
(November): 14-79.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April): 57-86.
Glatzel, Dieter. 1998. "The Measurement of Deficit and Debt under the
Maastricht Treaty: Some Statistical Considerations." In The Challenges for
Public Liability Management in Central Europe. Washington, D.C.: World
Bank.
Honohan, Patrick. 1999. "Fiscal Contingency Planning for Financial Crises."
Policy Research Working Paper 2228. World Bank, Washington, D.C.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits, and
Currency Crises." In S. Devarajan, F. H. Rogers, and L.. Squire, eds., World
Bank Economists' Forum. Washington, D.C.: World Bank.
Kotlikoff, Laurence. 1993. Generational Accounting. New York: Free Press.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Rubin, Irene. 1997. The Politics of Public Budgeting: Getting and Spending,
Borrowing and Balancing. Chatham, N.J.: Chatham House.
Selowsky, Marcelo. 1998. "Fiscal Deficits and the Quality of Fiscal Adjustment."
In The Challenges for Public Liability Management in Central Europe. Wash-
ington, D.C.: World Bank.
World Bank. 1999. "Czech Republic: Dealing with Contingent Liabilities." In
Czech Republic toward E U Accession, A World Bank Country Study. Wash-
ington, D.C.
CHAPTER 10
Analyzing Govrernment Fiscal
Risk Exposure in China
Kathie L. Krumm
World Bank
Christine P. Wong
University of Washington
THREE RECENT DEVELOPMENTS have brought the fiscal risk of con-
tingent liabilities into the center of policy debates in China.' First, in
response to the economic slowdown in 1997-98, which was com-
pounded by the East Asian financial crisis, the government implemented
a fiscal stimulus program in 19983 that continued through 2001. In a
country that has managed a conservative fiscal stance that has kept
budget deficits within the range of 1.5-2.5 percent of the gross domes-
tic product (GDP) throughout the past two decades of transition to a
market economy, this deficit spencling raised immediate questions about
the sustainability of such stimulus. China's fiscal deficit by official
definitions was 2.8 percent of GDP in 2000.2 Second, as China pre-
pares to implement measures to clean up the portfolios of large state-
owned commercial banks, the design of asset management companies
and other measures for resolving bad debts have raised questions about
the appropriate fiscal contribution, highlighting the large hidden li-
abilities inherited from the planned economy. And, third, to strengthen
the social safety nets in order to support an accelerated program of
restructuring state-owned enterprises (SOEs) and the economic slow-
down, the central government rnade special appropriations in 1998
and 1999 to bail out local social security schemes, focusing attention
on the contingent liabilities associated with these schemes.
The focus on fiscal risks is long overdue in China. Like other tran-
sition economies, the budget in China is not comprehensive, and the
235
236 KRUMM AND WONG
government undertakes substantial activity off-budget, such as direct-
ing bank loans to favored enterprises and investment projects. Some of
these activities have produced significant off-budget liabilities for the
government, to which must be added past contingent liabilities that
generate claims on future budgets. Moreover, even though China's
explicit public debt is relatively modest-with about 22 percent of
GDP in domestic debt and 4 percent of GDP in medium- and long-
term external debt in 20003-the government's revenue capacity is also
small; the budget is only 15 percent of GDP (as of 2000), and the
portion under the central government is only half.
This chapter begins by assessing the fiscal risk of contingent liabili-
ties in China. It then relates them to fiscal stability and identifies ma-
jor measures critical to managing fiscal risks. Finally, it discusses the
institutional context in which contingent liabilities are created and
need to be managed.
An Assessment
Sources of Risk
China is faced with several potential obligations that could pose a
significant fiscal risk: banking and nonbanking financial sector con-
tingent liabilities; pension liabilities; foreign public and publicly guar-
anteed debt; and liabilities from private participation in infrastructure.
Financial Sector. The contingent liabilities in the banking sector rep-
resent implicit liabilities.4 However, within the banking sector the na-
ture of the implicit obligation differs. The largest category applies to
the four large commercial banks owned by the central government.
These big four are major deposit-takers whose likely loan losses far
exceed any provisioning. Because they are state-owned, the implicit
guarantee for depositors and other creditors is strong. Risks include
some exposure to weak real estate markets, but they mainly reflect
quasi-fiscal lending and financial weakness in the domestic corporate
sector. The state-owned enterprise reforms to date have failed to stem
losses sufficiently. Unfavorable global and domestic economic devel-
opments are putting further pressure on the financial performance of
enterprises, including foreign-invested and nonstate firms. To date, the
central government has taken on the task of covering the losses. A
recapitalization exercise for the four large state-owned banks was car-
ried out in 1998 through the injection of the proceeds from the issu-
ance of special Treasury bonds. But this exercise was not based on
loan classification, interest accrual, and loss provisioning standards
that meet international best practices. Further in 1999, the central
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 237
government announced plans to establish asset management compa-
nies to remove part of the bad loarts from these four large banks, with
the first pilot established in April. The state also has an implicit liabil-
ity with respect to the financial performance of the state-owned policy
banks established in 1994. For example, the agricultural banks have
"grain debts," accumulated losses in grain procurement and distribu-
tion as well as diversion to nongrain businesses, which are not backed
by grain stocks or covered by sufficient budgetary allocations. The
China Development Bank's largely infrastructure portfolio is not sub-
ject to adequate risk analysis and provisioning. However, these banks'
portfolios are dwarfed by those of the commercial banks.
In addition to those in the state-owned banking sector, implicit li-
abilities are associated with weaknesses in regional and city commer-
c ial banks. Other parts of the financial system also have a high
proportion of nonperforming assets. Many nonbank financial inter-
mediaries that have developed alongside the traditional banking sec-
tor since the mid-1980s have taken high risks; trust and investment
companies (which account for 3 percent of China's financial assets)
are one example. Two-thirds of rural credit cooperatives (which ac-
count for 10 percent of China's financial assets) are reportedly in very
weak financial positions.
The stated principle is that the central government is fiscally re-
sponsible for state financial institution failures, and local governments
are fiscally responsible for failures of local ones. In practice, though,
this principle has worked only in certain circumstances. In some cases,
the central bank or the state-owned banks take on the obligations to
small depositors in the absence of matching performing assets. In the
closures of small troubled financial institutions over the last couple of
years, the Chinese authorities have de facto protected depositors, even
where deposit-taking was technically illegal. Indirectly, the central
government, which owns the central bank and the state-owned banks,
has incurred fiscal costs. Yet there is clearly a difference in perception
among depositors as to the strength of the implicit central government
guarantee, as reflected in a spate of localized runs on rural credit co-
operatives and a decline in the share of bank deposits held by the non-
state-owned banks. The complex policy question likely to emerge
increasingly will be how to allocate the fiscal burden when there is
significant overlap or the decisions by one part of the government
affect outcomes in the rest of the economy because of financial sector
contagion or social stability concerns. In addition, almost all local,
non-state-owned banks were licensed by either the head or local of-
fices of the central bank (before the reorganization that reduced the
control of local governments over the central bank's branches), raising
the question of the responsibility of the central government, because
central bank branches were party to the licensing and inadequate
238 KRUMM AND WONG
supervision of weak local financial institutions. Thus, in an assess-
ment of fiscal risk, this situation implies that the central government
cannot rule out an additional fiscal burden associated with such local
government financial institutions.
Pension Liabilities. In China, the pension system constitutes an
implicit contingent liability for the government or enterprise running
the pay-as-you-go (PAYG) pension scheme. Pension liabilities are in-
cluded in this analysis of contingent liabilities, however, because there
are risks of additional fiscal costs to the government if it ultimately
must pay the pensions out of other revenue streams, depending on the
nature and extent of the reforms to the pension scheme. Faced with a
rapidly aging population, the government has recognized the current
system as nonviable, has made significant progress in moving from an
enterprise-based system to a pooled system, and is planning to make a
transition from the present PAYG system to a funded, multipillar one.'
In principle, the responsibility for contingencies associated with
pension liabilities is that of local governments (municipal and provin-
cial). A complex issue arises, however, when national concerns about
mitigating inequity or maintaining social stability are threatened. For
example, the central government provided support to the provinces in
both 1998 and 1999 to cover those whose collections were less than
disbursements, and the provinces extended similar support to locali-
ties strapped for financing. At the same time, the central government
rightly is unwilling to extend an explicit guarantee so long as it has no
control over its potential exposure. With administration decentralized
to a municipal and often corporate level, special early retirements dis-
cretionary, and benefit levels still subject to decentralized wage calcula-
tions, the possibilities for abuse are obvious. Strengthening mechanisms
for control will be essential to clarifying the nature of government
responsibility.
Foreign Public and Publicly Guaranteed Debt. Total public and
publicly guaranteed debt as recorded by the State Administration for
Foreign Exchange exceeds the external debts of government. This rep-
resents an explicit contingent liability in addition to the direct liability.
Moreover, additional foreign debt with the explicit or implicit guaran-
tee of local governments but without official registration at the State
Administration for Foreign Exchange has come to light over the past
year, in particular as part of the clampdown on the financial activities
of the troubled international trust and investment corporations. The
authorities have issued several statements in recent years warning in-
vestors that unauthorized external borrowings would not be guaran-
teed by the central government. Nevertheless, the closure and bankruptcy
in late 1998 of the Guangdong International Trust and Investment
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 239
Corporation, the second largest foreign debt-issuing international trust
and investment company in China, sent a strong signal to markets that
the Chinese government was not implicitly guaranteeing all external
obligations of Chinese entities. However, the experience of other coun-
tries indicates that the fiscal authorities may be called upon to honor
short-term debts in the event of a short-term loss of confidence in
order to circumvent a more prolonged crisis.
Private Participation in Infrastructure Projects. China has followed
the global trend of growing priva:e investment in infrastructure, par-
ticularly in power plants and highways. Regulatory constraints, how-
ever, have so far limited private investment in power distribution and
transmission and in water distribution and sanitation systems. The
fact that large projects require approval by the State Development
P'lanning Commission (SDPC) probably has held contingent liabilities
associated with foreign investment to a minimum, because the SDPC
has taken a cautious approach to government assumption of risk. Most
projects with private participation are at the provincial and municipal
levels. Anecdotal evidence suggesits that these projects are being bro-
ken down into units of less than $30 million to avoid the regulatory
gaze of the SDPC. At these local levels, it is more common to have
guaranteed rates of return on investment in power and transport, which
represent contingent liabilities for those levels of government. While a
few of the provincial power projects are 100 percent foreign-owned,
the majority are jointly owned by foreign and Chinese interests, with
the latter holding the majority of ,hares in most cases. The situation is
similar in transport and water.
Size of Risks Associated with Contingent Liabilities
The size of China's contingent liatbilities is likely to be significant. In-
corporating them into calculations of fiscal debt would imply mul-
tiples of government debt as currently narrowly defined and reported.
Illustrative estimates have been carried out for the following signifi-
cant sources of risk: state-owned banks, the pension system, and for-
eign borrowings. Roughly measured more comprehensively, World Bank
estimates of the Chinese government's debt rise from the compara-
tively low level of about 11 percent of GDP to levels comparable with
the narrow government debt ratios of middle- and higher-income coun-
tries. The impact of contingencies on debt levels should give pause to
those evaluating fiscal risk based on the narrow fiscal position.
Illustrative Estimates of Contingent Liabilities Associated with State-
owned Commercial Banks. Regaidless of the measures taken to prevent
future contingent liabilities associated with state-owned commercial
240 KRUMM AND WONG
banks, a major fiscal burden cannot be avoided. For one thing, poli-
cies and institutional measures clearly are critical to reducing the flow
of increased bad loans, but the estimates made here refer to inherited
bad loans. These illustrative estimates assume that any contribution of
fiscal resources is conditioned on thorough and credible financial, op-
erational, and managerial restructuring and efforts to stem the flow of
additional quasi-fiscal loans that are deemed impaired. Otherwise, this
sector could well represent an even greater eventual drain on fiscal
resources. Second, there is probably only limited scope for spreading
the burden of bad loans across other agents. For example, the initial
burden is usually taken by the bank owners through loss of equity, but
in China's case the large banks are state-owned.
An illustrative range for the contingent liabilities is measured by
estimating the magnitude of nonperforming loans in the banking sys-
tem, the recovery rate on loans, and the size of the banking system, as
shown in Table 10.1.
A previous official estimate of nonperforming loans in the banking
system of about 25 percent of GDP was based on the standardized
classification system prevailing in 1995. The actual level is probably
significantly higher than this, not only because of the introduction of a
new classification system that conforms more closely to international
standards but also because other countries' experience indicates that
banking crises reveal higher losses than supervisory estimates.6
An earlier official estimate (in 2000) of 6 percent of loan losses
implies an official assumption of a 70 percent recovery rate, or a 30
percent loan loss rate. However, based on experience in other coun-
tries, a more likely range is 10-40 percent recovery, or a 60-90 per-
cent loan loss rate. Experience with collateral foreclosure and
liquidation indicates that, even in those cases with economies of scale
and a framework for asset sales, such as selling bundles of loans (Reso-
lution Trust Corporation in the United States) or encouraging overseas
Table 10.1 Illustrative Loan Losses, China
(percent of GDP)
Recovery rate 70 40 20 10
Loan loss rate (100 = recovery rate) 30 60 80 90
Nonperforming loan level
25 6 (official) 13 17
40 10 20 27 31
60 15 31 40 46
80 20 41 54
Note: Table is based on the loans of the four large state-owned banks, representing 85
percent of GDP.
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 241
investors willing to take risks in a particular industry or enterprise
(Thai Financial Restructuring Authority's auction of hire-purchase car
loans), recovery rates were about 70 percent and 50 percent, respec-
tively. Experience with recovery on. unsecured corporate lending is more
elusive, let alone in those transition economies where the private sec-
tor is less developed. For example, Mexico's asset management com-
pany, in a very unsuccessful case, had to take assets from banks that
continued to operate, and it has sold less than 1 percent of the assets
transferred. The asset management company in the Philippines, set up
in 1987 to deal with a diverse range of assets, including restructuring
large state enterprises, is still in operation today and has disposed of
only 50 percent of the transferred assets. In Central European banking
reforms, recovery rates have been in the 34-40 percent range.
Finally, the loans of the four large state-owned commercial banks
account for 85 percent of GDP. The loans of all deposit money banks,
including medium-size banks and urban and rural credit cooperatives,
account for 111 percent of GDP.
Although these are speculative estimates that make a number of
assumptions, they indicate the importance of this contingency's fiscal
implications. For example, if 40 percent of state bank commercial loans
were nonperforming and only 40 percent of the original value can be
recovered, the loss would be about 20 percent of GDP. If this same
condition holds for all money center banks, the loss would be about
25 percent of GDP. If 60 percent of the state bank commercial loans
were nonperforming and only 20 percent of the original value can be
recovered, the loss would be about 40 percent of GDP. The loss would
be the same magnitude if, for example, 40 percent were nonperforming
with only 10 percent recovery.
Given the uncertain nature of the central government's liability for
contingent liabilities in other parts of the financial system, there has
been no attempt to provide illustrative estimates for these.
Illustrative Estimates of Implicit Pension Liabilities. These estimates
draw on simulations carried out in an earlier World Bank study (World
Bank 1997a). One estimate is based on a cash flow approach, which
indicates transition costs to the budget of anywhere from zero to 1.5
percent of GDP during the period of time under consideration.
Another estimate is based on the net present value of the fraction of
implicit pension debt (IPD) that is not envisaged under current policy
to be financed through pension sector reform options. The size of the
implicit pension debt-that is, the present value of benefits that have
to be paid to current pensioners plus the present value of pension rights
that current workers have already earned and would have to be paid if
the system were stopped today--was estimated at about 40 percent of
GDP as of 1995.7 If one assumes additional coverage of workers under
242 KRUMM AND WONG
the scheme, a portion of whose contributions would be to the manda-
tory basic scheme, the estimate would be reduced to about 15 percent
of GDP (World Bank 1997a: Table 3.9). Because the analysis was car-
ried out in 1995, these estimates may well underrepresent the liability,
which likely continued to grow faster than GDP during the decade.
This analysis also assumes that the necessary steps are taken in terms
of retirement age, benefit levels, collection enforcement, and returns
on investments of pension surplus funds to ensure the financial viabil-
ity of the first (basic benefit from social pool) and second (supplemen-
tal benefit from funded individual account) pillars going forward. This
option is only illustrative of the kinds of adjustments that would affect
the contingent liability. However, it reflects the importance of the pen-
sion system contingency to longer-term fiscal risk.
Illustrative Estimates of Contingent Liabilities Associated with
Foreign Public and Publicly Guaranteed Debt. At year-end 1997 the
external debts of the Chinese government as reported by the Minis-
try of Finance stood at US$72.5 billion. However, the total public
and publicly guaranteed long-term and short-term debt registered
with the State Administration for Foreign Exchange stood at $117
billion, so that the government has an explicit contingent liability of
$44.5 billion in addition to its direct liability, or about 5 percent of
GDP. There may be unregistered debt that is guaranteed by local
governments, which could turn into implicit liabilities for the central
government (for example, see the discussion of international trust
and investment corporations below).
Illustrative Estimates of Contingent Liabilities Associated with Pri-
vate Participation in Infrastructure Projects. As a reference, total for-
eign investment in power, transport, and water is under US$30 billion,
or well under 5 percent of GDP.8 Because contingent liabilities are
only a small fraction of the investments in those sectors where there
has been some private participation, the aggregate potential liability is
relatively modest at present. However, it may pose concentrated risks
in certain localities. Anecdotal evidence suggests that some problems
are already beginning to surface. For example, slow growth in 1999
caused in some instances a default in guarantees of the minimum off-
take of electricity.
Risk Analysis
To analyze the fiscal risk implied by these contingent liabilities, we
used a fiscal rules approach augmented by inclusion of the liability
estimates. The balance sheet method was considered, because it en-
compasses not only liabilities but also assets. This is attractive in the
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 243
case of China given that-in addition to contingent liabilities-the
government has at its disposal significant real assets that could be liq-
uidated-for example, in the form of divestiture to the private sector
Flowever, in the China case, the balance sheet method was found too
informationally demanding to be applied.
The fiscal rules approach is captured in a debt dynamics analysis.9
The estimates of the size of contingent liabilities are summarized
into one of two variables, stock olf debt or annual cash flow expendi-
ture supplements. Debt dynamics analysis is then carried out based
on narrowly defined fiscal numbers, and again based on more com-
prehensive definitions of stock of debt or annual expenditures incor-
porating contingent liabilities. This enables a comparison, and thus
an analysis, of the implications for fiscal risk of considering contin-
gent liabilities.
Conclusion
This analysis leads to four conclusions about risk management in China.
First, the implied level of public liabilities taking contingent liabilities
into account is multiples of the narrowly defined stock of public debt
without taking these into account, which is the shorter bar at the far
right of Figure 10.1.
Second, these levels of public liabilities taking contingencies into
account are not that out of line with comparators in medium- and
high-income countries because of the low level of explicit debt and
debt servicing. Based on a narrow definition of public debt (thereby
excluding contingent liabilities in those countries), the average debt-
to-GDP ratio for middle- and high-income countries in 1995 was 39.9
percent and 60.7 percent, respectively (World Bank 1997b, 1998).
Third, the central government revenue-generating capacity is out of
line with the comparators. This makes a strong case, from the perspec-
tive of macromanagement, for some recentralization of fiscal manage-
ment in China in contrast to worldwide trends. These points are
illustrated in Figure 10.2.
Fourth, progress on structural reforms is key to managing fiscal
risks. Three major areas are the banking system, the pension system,
and budget management. Banking reform aimed at stemming the_fur-
ther accumulation of bad debts is essential. Reform of the pension
system should aim at financing the implicit pension debt and strength-
ening the financial viability of the system going forward. Strengthen-
ing the capacity of government to manage public resources, including
mobilizing central revenues, making the budget more comprehensive,
as well as rationalizing the balance of central and local responsibilities
and resources, is essential to meeting the government's huge contin-
gent liabilities in the future.
244 KRUMM AND WONG
Figure 10.1. Stock of Debt as Percent of GDP, China,
Illustrative Estimates for Year-End 1997
Percent of GDP
25
20
15
10
5
0
Financial Additional Pension Foreign Public
sector financial liabilities publicly domestic
liabilities sector guaranteed and foreign
(assuming liabilities debt and debt
20% loan (40% loan short-term (narrowly
loss) loss) debt defined)
Source: World Bank staff estimates.
Figure 10.2. Stock of Debt and Central Government Revenue,
China and Middle-Income and High-Income Countries
Percent of GDP
80
70 - Public debt
60 - Central government revenue
40-
30-
20-
10
China Middle-income High-income
Source: World Bank staff estimates.
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 245
Various scenarios using the debt dynamics framework can be used
to illustrate these points. For example, illustrative scenarios suggest
that without reducing the accumulation of bad debts in the banking
system, debt ratios could explode. '0 Three alternative assumptions are:
the continued accumulation of bad loans at the past rate; at the other
extreme, immediate cessation of quasi-fiscal lending through the bank-
ing system and no further accumulation; or a feasible reform option of
gradual adjustment in quasi-fiscal lending with a steady decline and
eventual elimination. Under the first scenario, the stock of debt con-
tinues to grow in an unbounded way. Under the second scenario, the
debt-to-GDP ratios fall off quickly as does the share of central govern-
ment revenue taken to cover interest costs. Under the third "baseline"
scenario assumptions, the stock af debt continues to climb but peaks
in a few years, declining thereafter, with primary balances turning
positive. Note, however, that although declining, the share of interest
in central government revenue would remain high throughout this pe-
riod (35-50 percent). Also, a number of downside factors may be highly
c orrelated. China could face lower growth, less robustness of revenues,
high loan losses in state-owned banks, and additional depositor guar-
antee obligations from other worsening financial institutions. These
possibilities further suggest the urgency of tackling the enterprise per-
formance problems and banking sector transformation underlying fi-
nancial sector weakness."
Institutional Setting for Creating
and Managing Fiscal Risks
Information and Accountability
The minister of finance is responsible for preparing and presenting the
budget to the legislature, the National People's Congress (NPC), each
year. At present, the budget does not include any information on con-
tingent liabilities. However, the government is accountable to the leg-
islature in explaining any sudden increases in debt. Because of historical
experiences, political leaders are acutely aware of the inflationary risks
associated with a large fiscal imbalance, and thus any proposed ad-
justments are heavily debated and scrutinized.
An apt example is related to rhe recapitalization of the four large
state-owned commercial banks, which was a decision that reflected in
part the contingent liabilities embedded in the financial sector. In the
interim between the annual budget exercises, the Ministry of Finance
consulted with the NPC's econornic and finance committee and then
made a special submission to seeks the NPC's approval before issuance
of the special Treasury bonds used to carry out this transaction.
246 KRUMM AND WONG
Awareness of the costs of hidden and contingent liabilities has al-
ready led to some demands for change in China. For example, re-
sponding in part to recent concerns about fiscal sustainabiliry and in
part to the auditor-general's harsh criticisms of budget implementa-
tion, in June 1999 the National People's Congress ordered the Minis-
try of Finance to strengthen budget processes, increase the transparency
and accountability of public spending, and disclose publicly all inter-
governmental transfers by province. The Ministry of Finance has taken
this as an opening to introduce some planned reforms to the budget
system, the first step of which was to require government ministries
and departments to report all resources and spending, including bud-
getary and extrabudgetary accounts. More reforms are planned over
the next few years to improve information flows and strengthen ac-
countability. They include: introducing Treasury management, revis-
ing the system of budget classification, strengthening budget forecasting,
changing the budget cycle and consultation processes, and strengthen-
ing monitoring and audit procedures.
Policy Practice and Risk Management Capabilities
Policy practice and risk management capabilities are still quite limited
in China as in many other developing countries. A further complica-
tion is that the transition process from a centrally planned to a social-
ist market economy results in explicit and implicit involvement of the
government-and the party-throughout the economy, thereby fur-
ther blurring lines of responsibility. However, there is a growing aware-
ness of the importance, and the complexities, of managing certain
components of fiscal risk associated with contingent liabilities.
One recent change in the government's policy practice was a clari-
fication of the rules of the game for sovereign responsibility for semi-
official financial activities. As mentioned above, the cleanup effort
among nonbank financial intermediaries accelerated in 1998 with the
closure of several prominent ones. This move began to establish limits
on sovereign responsibility and challenged previously held assump-
tions about the extent of liabilities of the state. Meanwhile, the central
government had to withstand considerable pressure from local gov-
ernments, foreign financiers, and others. The difficulties of the inter-
national trust and investment corporations have prompted markets to
increasingly differentiate between sovereign and nonsovereign entities
in China. In light of the complexities revealed during this experience,
the government is continuing to examine its policies in this area.
Regarding pension liabilities, in designing measures to address the
immediate and urgent problem of funding pensions for employees of
distressed state-owned enterprises, the government is cognizant of the
moral hazard issues. As noted, the government has recognized the
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 247
irnportance of improving administration in order to enhance internal
controls before extending an explicit guarantee. It consolidated re-
sponsibility for social insurance in March 1998 under the Ministry of
Labor and Social Security.
As for private participation in infrastructure, the central govern-
ment, as noted, has been quite prudent about the risks it will assume
and is working with the World Bank (and others) to modify its prac-
tices in light of experiences emerging from the East Asian financial
crisis. However, the government still faces the challenge of strengthen-
ing the legal and regulatory framework that guides infrastructure de-
velopments at all levels of government, with adoption of competitive
procedures for the award of concessions, so that investors will be less
likely to demand guarantees.
In conclusion, China faces high, fiscal risks associated with contin-
gent liabilities of the financial sector, the fiscal obligations arising
from reform of state-owned enterprises and the social security sys-
tem, and the problematic fiscal relations between the central and
local governments. Clearly. the government's ability to manage these
fiscal risks depends largely on how it addresses these problems. Ex-
panding on conventional fiscal analysis and putting these problems
into a broader fiscal context may help reinforce the importance of
addressing them. One legacy of the planned economy is that the
Ministry of Finance is a relatively weak institution: it is not in charge
of formulating a comprehensive budget; it is not empowered by law
to be the sole manager of public funds and assets; and the budget
process is not well-designed to ensure efficiency of public spending.
A comprehensive reform of the budget system is essential for strength-
ening the institutional arrangements for managing fiscal risks. Given
that the necessary reforms will redistribute power and resources across
ministries and departments, however, the process will likely be pro-
tracted and difficult and will require commitment from the highest
levels of China's political structure.
Notes
1. This chapter represents the views of the authors and not necessarily those
of the World Bank. The authors are grateful to Li Xiaoshi, Xie Ping, Li Keping,
Guo Xiangjun, Peng Longyun, Bill Easterly, Hana Polackova Brixi, and Estelle
James for their invaluable comments during seminars held in Beijing and con-
sultations held in Washington, D.C., and to Wei Ding, E. C. Hwa, Tom
Richardson, Richard Scurfield, and JoAnn Paulson for their contributions to
the work. We also are indebted to Liuo Guangqin and Liu Li-gang for invalu-
able administrative support and research assistance.
248 KRUMM AND WONG
This chapter is based on work carried out by the World Bank China pro-
gram in collaboration with the Ministry of Finance of the People's Republic of
China in late 1998 through mid-1999, as summarized in World Bank (1999).
The International Monetary Fund also participated in the work.
2. In addition to the budgeted fiscal deficit, a supplementary fiscal package
was announced in late 2000 for western region development.
3. See World Bank (2001). The figure for the domestic public sector debt
includes Treasury bonds, as well as policy financial bonds and other financial
bonds (end-period outstanding).
4. In the early 2000s, liquidity has not been the issue. The core of the state-
owned banking sector remains liquid even if nonperforming loans are substan-
tial and growing. Deposits have continued to grow, with substantial excess
reserves held with the central bank. The domestic savings rate is high, and there
are few alternative saving vehicles to banking institutions. In the current envi-
ronment of deflationary concerns, the policy concern is the reverse-namely,
that households are reluctant to consume. Liquidity is also stronger than it oth-
erwise would be because banks have been hesitant to lend.
5. A State Council decision of August 1998 sets out the basic parameters-
namely, (a) provide a mandatory basic public benefit for a redistributive role,
with pooling to be extended from the municipal to the provincial and ultimately
to the national level; (b) establish mandatory individual accounts, though these
accounts are largely unfunded currently; and (c) encourage establishment of
supplementary, voluntary individual accounts.
6. In 2001 the chairman of the Bank of China, one of the four state-owned
commercial banks, revealed a higher estimate of nonperforming loans, and the
level in the other banks likely exceeds that of the Bank of China.
7. The IPD estimates were based on the current features of the system (re-
tirement ages of 55 for women and 60 for men, replacement rates of 80 percent
for retirement and 40 percent for disability, and real wage indexation of 50
percent). Additional assumptions include projected wage increases at an annual
rate of 5 percent during 2000-2010 and 4 percent during 2011-2030; discount
rates at 5 percent during 2000-2010 and 4 percent during 2011-2030. The
IPD is even larger if pension contributions are lost through mismanagement or
fraud, which appears to be a significant problem in the current fragmented,
locally managed systems.
8. Estimates of foreign investment are on the order of US$15 billion in the
power sector, $11.5 billion in transport, and probably less than $0.5 billion in
water treatment facilities.
9. The International Monetary Fund participated in the World Bank work
on this issue. The debt dynamics framework draws substantially from Tsibouris,
1998/99).
10. These scenarios make a number of assumptions, including a two-
percentage-point improvement in revenue capacity; no change in the central-
local government share of those revenues; real GDP growth in the 6-8 percent
GOVERNMENT FISCAL RISK EXPOSURE IN CHINA 249
range; and loan losses in the state banking sector ranging from 20 to 40
percent of GDP.
11. In this regard, the policy guidelines announced by the Party Central Com-
inittee in September 1999 focused heavily on SOE reform.
References
Tsibouris, George. 1998/99. "Fiscal Sustainability in China." IMF Working
Paper. International Monetary Fund, Washington, D.C.
'World Bank. 1997a. China 2020: Old Age Security. Washington, D.C.
1997b. World Development Report. Washington, D.C.
- . 1998. World Development Indicators. Washington, D.C.
- . 1999. "China: Weathering the Storm and Learning the Lessons." Coun-
try Economic Memorandum. Washington, D.C.
--. 2001. "China Update." Beijing, March.
CHAPTEER I I
Dealing witlh Contingent
Liabilities in Indonesia
and Thailand
Hana Polackova Brixi
and Sudarshan Gooptu
World Bank
AMONG THE EAST ASIAN COUNTRIES, Indonesia and Thailand be-
long to those most heavily hit by the 1997 financial crisis. Their expe-
rience has only confirmed how rapidly government contingent liabilities
can turn into actual liabilities, how dearly they have to be paid for,
and what serious consequences they have for the fiscal and overall
economic performance of a country for many years into the future.
In both Indonesia and Thailand, government debt as a share of the
gross domestic product (GDP) more than tripled during 1997-2000
(the situation was similar in Korea and, to a lesser degree, in Malay-
sia). Most of the increase came from implicit government guarantees
to the financial sector (as a cost of recapitalizing the banking sector),
from explicit government guarantees in power and road sectors (in the
form of government outlays to honor "take-or-pay" contracts), and
from government credit guarantees to enterprises that had been called.
Although the governments dealt with obligations that came due
after the 1997 shake-up, new fiscal risks emerged as a consequence.
This chapter focuses on the risks facing the governments of Indonesia
and Thailand at the end of 2000, builds an illustrative stress scenario
for Indonesia, analyzes the largest risks facing the government of Thai-
land, and describes the strategies that have been adopted by these gov-
ernments to prevent and deal with fiscal risks in the future.'
251
252 BRIXI AND GOOPTU
Analyzing the Overall Government
Risk Exposure in Indonesia
As a result of the 1997 crisis, Indonesia's government entered the new
millennium facing two major sources of fiscal risks. One was related
to the large portfolio of government debt; the other was related to
government programs, some in the form of contingent support, whose
cost may suddenly rise.
Government debt increased from US$53 billion (23 percent of GDP)
before the crisis to about $134 billion (83 percent of GDP) in early
2000. Almost three-quarters of the increase in debt is new domestic
debt arising from the financial crisis-$72 billion in bonds issued to
recapitalize banks and to compensate Bank Indonesia for liquidity cred-
its. This rise in debt is the combined result of past policy mistakes and
the economic crisis, not new spending. Debt service obligations (inter-
est and amortization) are projected at over 40 percent of government
revenue for several years. Moreover, the government's debt service is
subject to serious risks, because about one-half of the government debt
portfolio is denominated in foreign currency, and a majority of the
domestic debt had interest payments flexibly linked to the domestic
interest rates and inflation (see Table 11.1). Thus an increase in the
domestic interest rates, depreciation of the rupiah, and inflation, as
well as new debt needed to account for past or new policy mistakes,
would significantly increase the cost of the government's future debt
service.
As for the unexpected spending pressures, these may arise from
possible social or political pressures and from off-budget obligations
associated with financial institutions, state-owned enterprises, inde-
pendent government agencies, and subnational governments. The like-
Table 1 1.1 Domestic Government Debt, Indonesia,
December 1998-June 2000
(cumulative, trillions of rupiah)
Variable- Fixed- Inflation
rate rate indexed Hedge
bonds bonds bonds bonds Total
Dec. 1998 0 0 100 0 20
March 1999 0 0 165 0 165
June 1999 95 9 218 0 322
Sept. 1999 95 9 218 0 322
Dec. 1999 204 51 218 27 500
March 2000 204 53 218 25 500
June 2000 330 59 218-253 29 636-671
Sources: Ministry of Finance and World Bank estimates.
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 253
lihood of these pressures actually resulting in additional government
expenditures would depend heavily on the government's commitment
to prudent and transparent fiscal management.
The Portfolio of Government Fiscal Risks
rhe Fiscal Risk Matrix in Table 11.2 illustrates the sources of future
possible pressures on government finances in Indonesia. The largest
contingent liabilities relate to the banking sector (the government's
explicit guarantee on interbank claims and an implicit commitment to
recapitalize large state-owned banks and maintain a stable financial
system). Sources of fiscal risk also include the power sector (concerns
about excess power generation capacity and the unsolicited, long-term
take-or-pay contracts signed with some of the 26 independent power
producers before the crisis),2 the pension system,3 and credit programs
and guarantees vis-a-vis the private sector.4
The Fiscal Hedge Matrix in Table 11.3 summarizes potential sources
of government financial security--that is, sources of future revenues
that may become available to the government to meet its obligations.
Among these, the largest is the value of assets collected by the Indone-
sian Bank Restructuring Agency (IBRA), an asset management com-
pany established by the government after the 1997 crisis.
Both obligations and sources of financial safety are sensitive to vari-
ous factors. Recovery of IBRA assets, for example, requires the work-
out and sale of nonperforming loans and sale of equity, some of which
was previously owned by powerful individuals. Thus the recovery rate
would depend on the extent of investor confidence and government
commitment.5
Overall, the leading factors affecting the potential fiscal costs aris-
ing from the government's obligations and its capacity to meet them
in the future can be summarized as follows. First, investor confi-
dence is vital for the government: to successfully recover an estimated
IDR207 trillion from asset sales by IBRA and raise revenues from
privatizing state-owned enterprises. It is also critical for renewed
growth, keeping interest rates down, achieving government tax rev-
enue targets, and establishing a successful domestic bond market.
Investor confidence, in turn, will depend on prudent fiscal and mon-
etary policies, sound market institutions, and transparency in gov-
ernment decisionmaking.
Second, domestic interest rates in Indonesia strongly affect the cost
of debt service and the size of likely off-budget losses. A one-percent-
age-point increase in the SBI (Bank Indonesia certificate) rate would
increase the cost of servicing the domestic debt by about 0.3 percent of
GDP annually (IDR4 trillion). An increase in domestic interest rates
would weaken government credit programs6 and increase likely
254 BRIXI AND GOOPTU
Table 11.2 Fiscal Risk Matrix, Indonesia
Sources of Direct liabilities Contingent liabilities
obligations (obligation in any event) (obligation if a particular event occurs)
Explicit
Government * Sovereign debt (dom- * Blanket guarantee on bank
liability as estic and external, depositors (cost IDR600
recognized loans contracted and trillion during 1997-99)
by a law or securities issued by * Guarantee on interbank claims
contract government) * Umbrella government
Expenditures-non- guarantees for nonsovereign
discretionary and borrowing by small and
legally binding in the medium-size enterprises,
long term (salaries and farmers, BULOG (govern-
pensions of civil ser- ment rice procurement
vants, minimum bene- company), and other entities
fits under the pay-as- * Trade and exchange rate
you-go pension guarantees via the Export
scheme Taspen) Bank, INDRA (Indonesian
Debt Restructuring Agency),
and other entities
Implicit
A moral * Future recurrent * Losses associated with take-
obligation of costs of public invest- or-pay contracts of public
government ment projects and utility companies
that reflects other discretionary * Support to enterprises (gov-
public and expenditures ernment possibly covering
interest group losses and assuming nonguar-
pressures anteed obligations of state-
owned or private enterprises)
* Subsidies related to the pricing of
rice and regulated oil products
via BULOG and PERTAMINA
(state oil company)
* Possible need for further recap-
italization of any banks that
fail to reach the 8 percent cap-
ital asset ratio by year-end 2001
* Possible need for further recap-
italization of Bank Indonesia
* Possible spillover of subnation-
al government obligations to
the central government
Note: Estimates of the future potential revenues are reflected in the stress scenario
depicted in Figure 11.1.
Sovrce: World Bank staff.
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 2ss
7able 11.3 Fiscal Hedge Matrix, Indonesia
Sources of Direct sources of safeiy Contingent sources of safety
financial (based on the stock (dependent on future events, such as
safety of existing assets) value generated in the future)
Explicit
Based on . IBRA (asset man- . Government revenues from
government agement company) oil and gas
legal powers assets recovery * Tax revenues less revenue
(ownership (workout and committed to subnational
and the right sales of nonper- governments
to raise forming loans and * Savings from cuts of discre-
revenues) sales of equity) tionary expenditures, such as
* Privatization of state- subsidies
owned enterprises * Hedging instruments and
and other public (re-)insurance policies pur-
resources chased by the government
from financial institutions
. Recovery of loans made by
government to public
enterprises
Implicit
Based on * Bank Indonesia's re- * Future profits of state-owned
government serves to the extent of enterprises and agencies
indirect positive net worth (ad- under government control
control justed for their liquid-
ity and currency risk)
Source: World Bank staff.
corporate and banking sector losses that could in turn increase govern-
ment obligations.
Third, political actions can significantly alter the size of off-budget
obligations and the revenues from asset sales. For example, a soft stance
vis-a-vis large banks or enterprises could result in further bank recapi-
talization costs and reduce IBRA revenues. Indirectly, any perceptions
among market participants that there is a lack of political will in deal-
ing with the complex issues emerging from the ongoing process of
corporate and financial restructuring may tend to undermine the re-
turn of investor confidence.
Furthermore, policy actions directly affect government revenues (for
example, through tax exemptions), current expenditures (such as subsi-
dies on petroleum products, eleciLricity, and rice), and potential future
expenditures (such as potential liabilities arising from credit programs;
the operations of BULOG, the government rice procurement company;
256 BRIXI AND GOOPTU
and power projects). A failure to maintain fiscal discipline, which could
happen with the implementation of fiscal decentralization, would in-
crease government debt.
Fourth, operational risks pervade fiscal and debt management ar-
rangements. A shortage of trained staff in many aspects of fiscal risk
and debt management, together with inadequate information and in-
stitutional arrangements within government, gives rise to the likeli-
hood of making misjudgments and poor implementation decisions.
For example, an audit of Bank Indonesia in 2000 revealed that the
bank did not correctly implement the bank recapitalization scheme
and as a result actually had negative net worth. Subsequently, the gov-
ernment had to recapitalize Bank Indonesia.
Fifth, changes in the rupiah affect external debt service payments,
exchange rate guarantees provided by the Export Bank and the Indo-
nesian Debt Restructuring Agency (INDRA), and liquidity and sol-
vency problems of banks with large foreign debts. This effect on
expenditures is partly offset by changes in oil revenues and the market
value of export companies under IBRA control. Deregulation of do-
mestic fuel prices would increase the offsetting effect that oil price has
on fiscal balances. Cross-currency risk, particularly the risk of yen
appreciation, is significant, because 39 percent of government foreign
debt is denominated in yen and Indonesia has a large (US$1 billion)
deficit in yen-denominated trade. A 10 percent appreciation of the yen
would increase government debt by nearly $2.3 billion.
Sixth, commodity price changes, particularly for oil and rice, affect
government finances. The budget deficit falls by about 0.1 percent of
GDP for every US$1 rise in the oil price (World Bank 1999). An increase
in the price of rice affects the government budget through BULOG losses.7
The baseline scenario for Indonesia for 2000 appeared to be that of
relative stability and gradual overcoming of the government debt bur-
den. This scenario, however, assumed that the government takes the
actions needed to have investor confidence return and to achieve re-
newed economic growth. Should the government fail to take the needed
actions, or should other factors negatively affect, for example, inves-
tor confidence, a different scenario would become reality in conse-
quence. To illustrate such a possible scenario, the interplay of its
underlying factors and its potential fiscal cost, a hypothetical stress
scenario is built below.
A Stress Scenario
By simulating the possible impact of a sudden decline in investor con-
fidence, Figure 11.1 illustrates the importance of containing off-budget
losses and counteracting fiscal risks. This stress scenario assumes that
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 257
Figure 11.1. Fiscal Risk Stress Scenario, Indonesia,
as of May 2000
Government debt/GDP (percent)
[40
120 - --- -.
100
80
60 a sa
40
20
1996 1998 2000 2002 2004 2006 2008 2010
investor confidence declines again in 2003, which is followed by a
sudden decline in demand for rupiah and other Indonesian assets. This
decline would generate a temporary sharp depreciation of the rupiah
(say, 12,000 rupiah per U.S. dollar during 2003-04) and an increase in
the domestic interest rate (up to 40 percent in 2003 and 28 percent in
2004). The decline in the value of the rupiah would make imports
more expensive and thus generate inflation (up to 30 percent in 2003
and 10 percent in 2004). This lOSS of confidence and the associated
rise in uncertainty would negatively affect overall economic perfor-
mance (reducing GDP growth from 6 to -2 percent in 2003 and 0 per-
cent in 2004), lead to renewed primary fiscal deficits (5 percent of GDP
in 2003 and 2004), generate new off-budget losses through bank recapi-
talization and other contingent liabilities (up by about IDR100 trillion
in 2003 and 2004), and erode revenues from government asset sales.
Following this 2003-04 stress, all assumptions are assumed to return to
the baseline. The scenario illustrates that a temporary loss of confidence
not only would slow down the repayment of government debt, but also
could actually result in a further increase in government debt as a share
of GDP and thus create an even more difficult debt burden.
Measuring Selected Fiscal Risks in Thailand
Since the 1997 Asian financial crisis, the government of Thailand has
faced increasing fiscal risks, particularly in the form of contingent
258 BRIXI AND GOOPTU
liabilities emerging from the banking and enterprise sectors (Box 11.1).8
Some of these obligations are of a legal nature (for example, state-
guaranteed debt; others reflect policy commitments (for example, the
nonguaranteed obligations of state-owned enterprises and revolving
funds to sustain price controls). So far, when the state budget has been
unable to cover claims arising from these obligations, state-owned
enterprises (such as the Royal State Railway of Thailand and state
rubber plantation) and extrabudgetary funds (namely, the Sugar Cane
Fund) have borrowed further instead. If enterprises and funds incur
continued losses, however, this practice of allowing their further bor-
rowings (often with explicit government guarantees) only postpones
Box 11.1 Categories of Contingent Liabilities in Thailand
The explicit contingent liabilities of the Thai government are commit-
ments that are based on law and contracts (they primarily include guar-
anteed debt and other liabilities of state-owned enterprises) and the
financing needs under price support programs. The Act Empowering
the Ministry of Finance to Guarantee, B.E. 2510 (1967), provides the
Ministry of Finance with the mandate to issue sovereign guarantees on
domestic and external borrowings by government agencies, limited
companies that are partly owned by the state, and specialized financial
institutions in Thailand, subject to the approval of the cabinet.
Government implicit contingent liabilities are commitments that
are based on political announcements, public expectations, and pos-
sible interest group pressures. In Thailand, these mainly include:
* Obligations of the Financial Institutions Development Fund
(FIDF), an agency established by the Bank of Thailand to issue bonds
and recapitalize financial institutions, provide liquidity support, and
cover deposit insurance claims. (In 2000 the Ministry of Finance asked
the Bank of Thailand to refinance THB20 billion of FIDF bonds. The
first tranche of these new government bonds with maturity of six years
and an 8 percent coupon rate was auctioned in September 2000. This
was followed by another tranche of THB14.8 billion with maturity of
11 years and a 7.5 percent coupon rate.)
* Liabilities of extrabudgetary funds.
* Possible negative net worth of Bank of Thailand.
* Losses, nonguaranteed obligations, arrears, and deferred mainte-
nance of state-owned enterprises (including concession agreements of
state-owned utilities, arrears of State Railways on fuel bills, and de-
ferred railway track rehabilitation).
* Future possible commitments and obligations of subnational gov-
ernments (for example, via losses of provincial enterprises).
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 259
the need for government direct financing and increases the size of gov-
ernment off-budget obligations.
Given the nature of its contingent liabilities, the government of
Thailand has often found it unimportant to distinguish whether the
obligations are explicit or implicit. Most explicit and implicit govern-
ment contingent liabilities directly relate to losses in state banks, spe-
cialized financial institutions, and banks in which the government has
intervened. Because these institutions already have problems of capi-
talization, any additional losses will have to be fiscalized in the future.
The recapitalization needs of banks will rise further if state-owned
enterprises and funds default on their nonguaranteed debt. For ex-
ample, should the Sugar Cane Fund default on its nonguaranteed debt
to the Bank for Agriculture and Agricultural Cooperatives, a new
nonperforming loan in the portfolio of this already undercapitalized
financial institution would only increase its recapitalization needs. With
respect to the financial sector, should the fiscal authorities fail to cover
the obligations of the Financial Institutions Development Fund (FIDF,
an asset management company that issued bonds to pay for the bad
assets of Thai banks) or any future bank recapitalization needs, the
Bank of Thailand would do so ex post, thereby running down its net
worth. But there does not appear to be an adequate cushion in the
Bank of Thailand's net worth to meet this contingency, and eventually
public resources would be required to maintain its solvency and cred-
ibility. In addition, many enterprises and funds are critical for delivery
of public services, such as cheap passenger transport. Therefore, un-
less policy priorities change the government will eventually pay for the
nonguaranteed as well as guaranteed liabilities, arrears, and deferred
maintenance of such enterprises and funds.
Obligations vis-a-vis the Banking Sector
In early 2000 it became clear that the cost of past recapitalization
programs, largely reflected in the liability portfolio of the FIDF, and
the further recapitalization needs of the Thai banking sector would
significantly affect the amounts of future government borrowing. The
FIDF estimated that the fiscal authorities would need to provide about
THB1.1 trillion to cover its liabilities and operating costs. This amount
is expected to be paid from new government borrowing during 2000-
2005. Other available estimates range from THB1.1 trillion to THB1.4
trillion (World Bank 2000d). This cost would increase with lower re-
covery rates on nonperforming loans in state banks and intervened
banks, higher domestic interest rates, and a lower market value for the
four joint venture banks expected to be sold by the government.
On top of the FIDF's financial iinstitution rescue and depositor/credi-
tor protection program, the government is expected to support the
260 BRIXI AND GOOPTU
recapitalization and subsidy needs of the specialized financial institu-
tions, such as the Government Housing Bank and the Bank for Agri-
culture and Agricultural Cooperatives. Furthermore, any future costs
of protecting depositor funds in private banks, net of insurance and
bank fees, would have to be fiscalized. Finally, fiscal risks also have
emanated from the structure of the FIDF debt portfolio and the bond
issuance strategy.
Debts of State Enterprises
The government has provided guarantees on domestic and external
borrowing to most nonfinancial enterprises and to selected financial
institutions. Ar the end of 1999, the stock of guaranteed debt sur-
passed THB1 trillion (see Figure 11.2). The stock of nonguaranteed
debt by nonfinancial state-owned enterprises amounts to nearly
THBS00 billion (World Bank 2000d). In addition to these reported
amounts, several large enterprises have accumulated arrears (for ex-
ample, on the fuel cost and for deferred maintenance on the State
Railway and Bangkok Mass Transit Authority).
In Thailand, the Public Debt Management Office was entrusted
with the task of estimating the likely future demands of state-owned
enterprises on the government budget. Its approach was to focus on
the recent and expected financial and economic performance of enter-
prises that represent an explicit or implicit fiscal risk for the govern-
ment. In its analysis, state-owned enterprises were divided into four
mutually exclusive risk categories: a risk level of 5 percent for best
performers, 25 percent for enterprises likely to make profits, 50 per-
cent for enterprises with fluctuating profit and investment performance,
and 90 percent for enterprises that have been showing continued losses
or have embarked on very risky projects.9 For each category, the out-
standing liabilities of enterprises were adjusted by their respective risk
weights (outstanding liabilities that are either explicitly or implicitly
guaranteed by the government were multiplied by the risk level of the
respective enterprise).
By aggregating the risk-adjusted amounts of guaranteed liabilities
of enterprises, the Public Debt Management Office obtained the total
risk-adjusted level of its explicitly and implicitly guaranteed debt. The
risk-adjusted amount of guaranteed debt can be interpreted as the net
present value of the future expected fiscal cost of government guaran-
tees, or as an amount of reserves the government should have accumu-
lated by the end of 1999 to cover future calls on the budget.'° To assess
likely future increases in the enterprise pressure on the government's
budget, we have assigned the respective risk levels also to the repay-
ment schedule of guaranteed debt. Figure 11.2 compares the guaranteed
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 261
Figure 11.2. Guaranteed Debt Outstanding and
Its Repayment, Thailand, 1999
Guaranteed Debt Outstanding
Billions of Thai baht
:1,250
1,000
750 -
500-
250
07-
Face values Risk-adjusted
Guaranteed Debt Repayment Schedule
Billions of Thai baht
200
180 -
160
140-
120-
100 -
80
60-
40 Risk-adut
20-
0
2000 2001 2002 2003 2004 2005 2006 2007 2008
Fiscal year
Note: Includes scheduled principal repayment on domestic and external guaranteed
debt and interest payments on external guaranteed debt.
Source: Calculations by the authors based on information provided by Thailand's
Public Debt Maintenance Office and Arthur Andersen (2000).
262 BRIXI AND GOOPTU
debt repayment schedule in face values with the risk-adjusted amounts.
Because many enterprises have relied on short-term borrowing and
have held unhedged foreign exposures, government risk exposure rises
with increasing interest rates and with depreciation of the baht."
Fiscal risks from state-owned enterprise operations also may emerge
from government policy actions (such as price controls that may give
rise to operating losses) and their own investment decisions (such as
ambitious and risky projects). The State Railway, for example, was
actually bankrupt in mid-2000 and required over THB60 billion in
public resources just to cover its existing obligations-on top of planned
government subsidies to cover its future operating losses and new in-
vestments. To a large degree, however, its operating losses and the
amount of further possible government obligations depend on the third-
class passenger fare, which is regulated by the government. 12Tables 11.4
and 11.5 illustrate the extent to which the growing bills of the State
Railway have necessitated government subsidies, short-term borrowing
(mainly overdrafts on bank accounts), arrears (in payments for fuel and
payroll), and deferrals (in maintenance and investment). Given the
policy commitment of the government, if the State Railway is not able
to roll over its credits, obtain new credit and supplies for a promise to
pay in the future, and further defer maintenance, government subsi-
dies and credit guarantees will increase.
Table 11.4 Financial Performance and Outlook,
State Railway of Thailand (SRT), 1994-2000
(billions of Thai baht)
1994 1995 1996 1997 1998 1999' 2000O
Operating loss 0.51 1.54 1.49 1.64 2.84 6.56 4.77
Reported liabilities 25.63 18.50 21.69 27.36 29.45 35.48" 41.52
out of which short-
term debt 3.49 2.51 4.66 4.90 4.05 6.03 6.04
Deferred main-
tenance - - - - - 32.54c 20.08
Arrears - - - - 2.02' 2.16d 0
- Not available.
a. Projections.
b. SRT's reported liabilities are almost entirely covered by state guarantees.
c. Track rehabilitation needed to sustain, not expand or upgrade, the existing train
traffic. The 2000 budget committed THB 12.46 billion.
d. Mainly unpaid fuel cost. The cabinet passed a resolution on November 23, 1999, to
pay off the arrears.
Source: Information provided to the authors by SRT and Thailand Development Re-
search Institute.
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 263
T,ble 11.5 Government Cash Flow to State Railway
of Thailand (RT), 1994-2000
(billions of Thai bahts)
1994 1995 1996 1997 1998 1999' 2000'
Operating loss
compensation 1.16 1.26 1.17 1.16 1.36 1.77 2.78
Loan repayment 0.70 0.74 0.96 0.85 1.04 0.75 1.67
Capital investment 1.65 2.18 0.82 1.45 3.70 2.61 4.26
a. Projections.
Source: Information provided to the authors by SRT and Thailand Development Re-
search Institute.
Developing a Government Risk Management Strategy
I)uring 2000-2001 in Indonesia, any progress in dealing with the ex-
isting government fiscal risks was complicated by emerging political
and economic imbalances. Furthermore, a clear authority for dealing
with fiscal risk had yet to be assigned. Because the responsibility for
government debt management was split among several directorates of
the Ministry of Finance and Bank Indonesia, even the task of collect-
ing and synthesizing information about the government debt service
profile proved difficult to accomlplish. Acknowledging the problems,
the government considered establishing a single debt management of-
fice in the Ministry of Finance and entrusting this office with the man-
agement of government direct and contingent liabilities.
To build its strategy and capacity in the management of govern-
ment debt and fiscal risks, the Indonesian government considered the
following three principles. First, the cabinet, in consultation with the
parliament, would set the prioril:ies among competing debt manage-
ment objectives. For example, should the government focus on mini-
mizing the immediate cost of debt service or on minimizing its risk
exposure? The debt management office would focus on implementing
these government priorities. Given the potentially catastrophic conse-
quences of further increases in debt, the government proposed to place
a high priority on reducing the government's risk exposure, ensuring a
steady and predictable debt service profile, and enhancing the gov-
ernment's access to debt markets. Specifically, this meant reducing the
volatility of debt servicing arising from refinancing risk, currency risk,
and interest rate risk and issuing domestic bonds of specific maturities
to set a benchmark for the domestic bond market.
Second, the debt management office would develop a capacity to
analyze debt management options and their likely consequences. The
office was expected to analyze future debt service costs and the debt
264 BRIXI AND GOOPTU
profile under alternative assumptions, the risks emerging from the
government's debt portfolio (refinancing risk, currency risk, interest
rate risk), the risks of government contingent liabilities (off-budget
pressure emerging from the obligations of state-owned enterprises,
banks, independent government agencies, and subnational govern-
ments), and the impact of alternative borrowing strategies on the cost
and volatility of future debt service.
Third, the debt management office would advise policymakers on
the pros and cons of alternative policy choices. As part of this, the
office would build awareness of the expected cost of debt service, ex-
isting contingent liabilities, and proposed budgetary and off-budget
commitments. This would also cover the borrowing and risk exposure
of subnational (provincial- and district-level) governments, indepen-
dent government agencies, and state-owned enterprises.
Progress on this agenda in Indonesia has been slow. Meanwhile,
Thailand has embarked on an ambitious program of building govern-
ment risk management capacity in its Public Debt Management Of-
fice, established at the Ministry of Finance in 1999. New legislation
entrusted the Debt Office with the primary responsibility for analyz-
ing and actively managing the government's obligations, including its
direct and contingent liabilities. Underlying responsibilities included
debt service forecasting, cash management, risk management (to ad-
dress currency, interest rate, funding and refinancing, credit and op-
erational risks, among other things), implementing transactions related
to project finance, and monitoring and advising the government on
how to deal with its contingent liabilities. In close collaboration with
the Bank of Thailand, the Debt Office began to develop a comprehen-
sive, medium-term financial and debt management and risk manage-
ment strategy for the government to approve and, then, for the Debt
Office to implement. The risk management strategy was expected to
specify to what extent the government is prepared to take on risks in the
form of contingent liabilities and similar obligations and to face associ-
ated volatility in its future financing requirement. To take on such new
responsibilities, the Debt Office began to build a whole range of systems
(including a database for direct and contingent liabilities, risk manage-
ment models, and cash management systems) and recruit and train staff.
Beyond the improvements at the Thai Public Debt Management Of-
fice, broader institutional changes have been launched to support gov-
ernment risk management. For example, in dealing with contingent
government liabilities, the Debt Office would need to rely on the close
cooperation of several agencies and units of the government and of the
Bank of Thailand. It also would need to rely on data collected and infor-
mation processed by the comptroller general's department and to coop-
erate with the Fiscal Planning Office and Budget Office. With respect to
government contingent liabilities arising in the financial sector, the Debt
Office would need to rely on analytical inputs from the Bank of Thai-
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 265
land and the FIDF. In this context, the responsibilities of the Debt Of-
fice, Comptroller General's Department, Fiscal Planning Office, and
Budget Office in dealing with contingent liabilities would have to be
clearly specified. Furthermore, to ensure that the Comptroller General's
Department and Debt Office obtain information that is relevant for
analyzing and monitoring the risk of contingent liabilities, the minister
of finance has sought to update the existing reporting requirements for
state-owned enterprises, extrabudgetary funds, and other public sector
entities and government agencies. Similarly, the Bank of Thailand was
asked to consider enhancing reporting requirements for financial insti-
tutions. Finally, the Ministry of Finance called for government guide-
lines for risk exposure and financial and risk management of state-owned
enterprises and other public sector entities. Such guidelines would help
the government to better pursue its shareholder interests in state-owned
enterprises and would be enforced through government representatives
sitting on the boards of directors of enterprises and in the State Audit
Office. Interestingly, these guidelines would be quite an innovative ap-
proach to dealing with risk-taking in the public sector and possibly would
be of interest to other countries.
Notes
1. This chapter draws on several World Bank studies (2000a, 2000b, 2000c,
2000d).
2. Toll roads are mainly financed by domestic banks and, unless generat-
ing expected revenues, they deteriorate bank performance. To the extent that
they were already part of a bank's rtonperforming portfolio during bank re-
capitalization, they do not represent an additional contingent liability.
3. In 1995 just 12 million workers (less than a fifth of the work force)
were enrolled in pension plans. Pension enrollments, however, grew rapidly
after the 1997 financial crisis without a corresponding increase in their fund-
ing and asset base. The government is legally responsible for financing the
civil service employees' pension scheme (Taspen) and does so from its gen-
eral revenues. Government responsibility for the public pay-as-you-go pen-
sion scheme (Jamsostek) is only implicit. Jamsostek has repeatedly reported
zero return on its assets (earlier because of the financial crisis, later mainly
because of its high administrative cost), making the level of future public
pension benefits questionable.
4. For example, in the case of Garuda Indonesia, the state-owned airline,
the government decided to take over the repayment of Garuda Indonesia's
external debt totaling about US$1.8 billion for the lease of 11 Boeing 737s.
This arrangement implied a fiscal cost of $62 million a year, on average,
for eight years. Of particular concern can be the off-balance sheet obliga-
tions of public sector corporations., which ultimately are the obligations of
the government.
266 BRIXI AND GOOPTU
S. This relationship has already surfaced through asset management com-
panies in a number of countries, including Mexico and the Philippines. IBRA
reported a significant drop in debt collection rates following the Bank Bali
scandal in 1999. The book value of IBRA assets is reported to be IDR533
trillion. The average expected recovery rate is about 30-35 percent, which is
still high in the context of international experience (Klingebiel 2000).
6. In 2000-01, the government is expected to face a contingent liability of
about IDR28 trillion on credits outstanding. For most credits, such as KUT
(credit for farming undertakings) and KKPA (credit for primary cooperatives
members), the government explicitly covers default risk. The average matu-
rity of these credits is 12 months and the average default risk is 40 percent.
7. BULOG procures domestically produced rice at a price fixed by the
government and distributes rice to rhe poor and the army at low prices.
8. No major sources of fiscal risk (that is, of volatility in government fi-
nancing and borrowing requirements) have been identified in the portfolio of
government assets and revenues. The government seems to face a downward
trend rather than volatility in the value of its assets and future revenues.
9. See Arthur Andersen (2000) for the details underlying the assignment
of a risk level to each enterprise.
10. In all guarantee contracts the government covers the entire principal
and interest payments against all risks.
11. Sensitivity analysis would be needed to determine the extent of the
change in government risk exposure with respect to the underlying variables.
Unlike in Indonesia, as of 2001 a stress scenario to illustrate the maximum
likely fiscal cost of contingent liabilities had not been developed in Thailand.
12. The Thailand Development Research Institute calculated that a gradual
increase in the third-class passenger fare by 40 percent would yield nearly
THB300 million in 2000 and over THB500 million in 2001. Sixty percent of
operations by the State Railway of Thailand is in passenger transport, the
fare of which is regulated by the state. The third-class fare, which applies to
about 90 percent of passenger transport, has been fixed since 1974. The
railway reports about a THB20 billion positive net worth, thanks mainly to
the reported value of its net properties of THB43 billion. For details, see
Thailand Development Research Institute (1999).
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Arthur Andersen. 2000. "Thailand-The Fiscal Costs of State Enterprises
and Private Participation in Infrastructure." Processed.
Klingebiel, Daniela. 2000. "The Use of Asset Management Companies in the
Resolution of Banking Crises: Cross-Country Experiences." Policy Re-
search Working Paper 2284. World Bank, Washington, D.C.
CONTINGENT LIABILITIES IN INDONESIA AND THAILAND 267
Thailand Development Research Institute. 1999. State Railway of Thailand
Short-term Financial Strategy Development Project. Final Report. Bangkok,
June.
World Bank. 1999. "Republic of Indonesia: Second Policy Reform Support
Loan." Report No. P-7308-IND. Washington, D.C.
--. 2000a. "Public Spending in a Time of Change." Poverty Reduction
and Economic Management Sector Unit, East Asia and Pacific Region.
Washington, D.C., April.
--. 2000b. "Thailand: Social and Structural Review." Report No. 19732-
TH. Washington, D.C., January 2.5.
--. 2000c. "Indonesia: Managing Government Debt and Its Risks." Re-
port No. 20436-IND. Washington, D.C.
- . 2000d. "Thailand: Building Government Capacity to Manage Its
Debt and Fiscal Risks." Washington, D.C. Processed.
CHAPTER IZ
Dealing with Contingent
Liabilities in Colombia
Juan Carlos Echeverry,
Veronica Navas, Juan Camilo Gutierrez,
Jorge Enrique Cardona
Government of Colombia
COMPARED WITH THAT OF most South American countries, Colombia's
fiscal management was prudent until the mid-1950s.' Since then, how-
ever, the decentralization of publlic sector contracts, including those
between the government and labor unions, has loosened budget con-
straints and given rise to large contingent liabilities.
Traditionally, the Colombian government has based its fiscal analysis
on the study of deficit dynamics and has explained an unsustainable
fiscal stance in terms of growing and persistent deficits. As a conse-
quence, the pursuit of short-term goals, such as immediate deficit reduc-
tions, has led to short-sighted fiscal policy decisions, delivering many of
the classic examples of fiscal illusion described by Easterly (1999).
In view of the shortcomings arid negative consequences of the con-
ventional focus on budget deficits and the benefits to be gained from
analyzing a comprehensive balance sheet of the government, the Co-
lombian National Planning Department decided in 1997 to improve
the official government balance sheet produced by the National Ac-
counts Office. In order to present a more accurate picture of the in-
creasing fiscal imbalance, officials called for the new government
balance sheet to include both implicit and contingent, as well as ex-
plicit and direct, assets and liabillities. In the process, the value of all
these items had to be estimated and reestimated (Echeverry and others
1999). Unexpectedly to some, the process of identifying and quantify-
ing government obligations revealed major cracks in the institutional
269
270 ECHEVERRY, NAVAS, GUTIERREZ, AND CARDONA
mechanism in Colombia's public finance management-cracks that
had given rise to large government contingent liabilities, many of which
had been poorly designed and were exposing the government to exces-
sive risk.
Quantifying Government Contingent Liabilities
In its wide-ranging attempt to identify and then quantify all major
sources of government fiscal risk, the government focused on the big-
gest items: pension liabilities, a guerrilla peace "buyout," natural di-
sasters, the financial sector, credit guarantees to the public sector, and
concessions on private financing of infrastructure.
Pension liabilities, mainly in the form of government-guaranteed
pensions to social security beneficiaries and teachers, surfaced as the
biggest source of fiscal risk for the government (for an analysis of
Colombia's comprehensive balance sheet, see Chapter 8 by Easterly
and Yuravlivker in this volume). The net present value of future pen-
sion outlays (an implicit direct liability for the government) was esti-
mated at about 159 percent of the gross domestic product (GDP), of
which almost half was designated for social security beneficiaries and
one-third for teachers.
In its analysis, the government also included seemingly unquantifiable
contingent liabilities such as the cost of possible peace negotiations.
For example, to determine the cost of the contingent liability of deal-
ing with Colombia's guerrillas, analysts evaluated the guerrillas' busi-
ness and calculated the sum required to purchase such a business. Based
on their calculation of the profitability of guerrilla violence, analysts
concluded that guerrilla groups in Colombia expect a profit of US$430
million per year (Echeverry and others 1999:17-20). Therefore, as-
suming that a guerrilla is risk-averse and taking into account a dollar
interest rate of 6 percent a year, they estimated that an investment of
approximately $7 billion was required to eliminate a business with
such high levels of profitability in the long term. Taking into account
efficiency gained from restructuring the army and from acquiring su-
perior military hardware, analysts were able to estimate the cost of
reaching peace with the guerrillas at about 4 percent of GDP (in net
present value terms in 1997).
Natural disasters, particularly earthquakes and floods, constitute
another major implicit contingent liability in Colombia. After disasters,
the government has had to provide resources for the reconstruction and
rehabilitation of affected areas and for future damage prevention. Aided
by evaluations of the cost of the most recent earthquake in Colombia
(January 1999), analysts were able to estimate the value of natural di-
sasters as a government contingent liability. They estimated a net present
CONTINGENT LIABILITIES IN CO:LOMBIA 271
value of about 1.1 percent of GDP for earthquake contingency and about
0.04 percent of GDP for flood contingency.
Government analysts also estimated the implicit contingent liabil-
ity arising from the financial sector. Again, previous experience with a
financial crisis and its fiscal implications were analyzed. Based on an
analysis of historical cost and a rough assessment of the present health
of financial institutions, analysts estimated that the future liquidation
of insolvent institutions and the recapitalization of others represent a
contingent liability of between 1.2 percent and 1.7 percent of GDP in
net present value terms.
Guarantees on public sector debt, forms of hidden subsidies in Co-
lombia, are another major item. These liabilities are subject to subnational
government borrowing regulations. which traditionally have been inef-
ficient, allowing rapid growth of regional debt. Since 1997, borrowing
has been limited to the payment capacity of the regional entity con-
cerned and thus associated with the generation of operational savings.
Ihe bailout cost of current regional debt was calculated to be 0.5 per-
cent of GDP. Another source of contingent liabilities is judicial rulings
against the state, which increased over the past decade from 0.02 per-
cent of GDP in 1990 to 0.3 percent of GDP in 1998.
Significant contingent liabilities in Colombia result from private
sector concessions in infrastructure. The state has transformed its role
from that of direct service providcr to guarantor of minimum sales to
private sector service provider. Initially, contracts involved fixed terms
xvithin which the government gua:ranteed a level of profitability. After
some adjustments, current contracts are flexible in length, which al-
lows low profitability to be compensated over time. Both types of
government guarantees, however, involve uncertainty over whether,
when, and how much public financing will be required in the future.
Across the guarantees issued, the net present value of contingent li-
abilities in the infrastructure sector was estimated at about 6 percent
of GDP.
In the transport sector, guarantees covered traffic volume and road
construction costs. The guarantees on traffic volume are invoked if the
anticipated income falls below an agreed-on minimal level related to
predicted traffic volume. However, if income rises above an allowed
maximum, it becomes a contingent asset and the excess is returned to
the state. The calculated net present value of the contingent liability
associated with traffic volume is 0.1 percent of GDP. On road con-
struction a large guarantee was issued on the second runway at Bogota's
El Dorado International Airport. In addition to the requirement to
maintain both runways, the guarantee also covers minimum income
conditions. The net present value of this contingency for the duration
of the airport project (20 years) was estimated to be about 0.7 percent
of GDP.
272 ECHEVERRY, NAVAS, GUTIERREZ, AND CARDONA
In the energy sector, the contracts for energy providers have been
issued under onerous conditions. The government has had to assume
the difference between the actual market price and the price agreed on
with investors. The net present value of this contingency was esti-
mated to be about 0.5 percent of GDP.
In the telecommunications sector, the government has provided
minimum income guarantees (contained in four-year joint venture con-
tracts), covering a certain number of conversation minutes. The net
present value of these guarantees is about 0.2 percent of GDP.
Other contingent liabilities were expected to emerge from the con-
struction of the Bogota Metro. Here the contingency is related to the
probability of liabilities being larger than predicted, and is partly a
consequence of regulatory ambiguities. The law that authorized the
construction of the Bogota Metro states that if excess costs imputable
to the public sector are generated, the national government is obliged
to cover 70 percent of excess costs. In addition, the government covers
currency risks related to external financing as well as the cost of con-
struction delays and of required public services (such as laying pipes
for the waterworks).
Most of the contingent liabilities just described emerged from the
1991 constitution, which deepened the decentralization process (the
transfer of responsibilities and resources to lower levels of govern-
ment) and promoted private participation in the provision of public
services and infrastructure. On the positive side, private investment
in infrastructure increased rapidly, nearly doubling in the 1990s to
surpass 5 percent of GDP by 1998. Additional benefits were, among
other things, more efficient use of resources and an increase in the
quality of these services. In 1997 the future fiscal cost of these projects
was estimated to be about 1.5 percent of GDP (which was about a
quarter of the underlying investments). The government acknowl-
edged that it could have achieved the same policy outcomes through
direct subsidies that would have been less expensive than guarantees.
Meanwhile, the government was facing increasing pressure to issue
further guarantees.
The Institutional Underpinnings
The 1997 assessment made it clear that often nonpriority or poorly de-
signed projects were undertaken thanks only to the loopholes in govern-
ment fiscal management that made it possible for such projects to obtain
government guarantees. Making matters worse, the government was
assuming most if not all of the risk associated with the programs and
projects benefiting from guarantees. The inefficient risk allocation mainly
reflected the fact that contracting government entities were not clearly
CONTINGENT LIABILITIES IN COLOMBIA 273
liable for their guarantees. Overall, the government had only a few rules
in place to guide the issuance and oversight of guarantees and other
contingent liabilities (until 1998, the government also lacked a legal
framework for private participation in the provision of public services).
Furthermore, accounting rules required that outlays on contingent li-
abilities that fell due be acknowledged as investment overruns. In the
budget, however, investments were already squeezed. Lack of budget-
ary funds thus meant that payments due to the guarantee beneficiaries
grew at high penalty interest rates Eor up to 18 months at a time.
When the size and unpredictability of payments grew beyond the
handling capacity of some public entities, budgetary adjustments (most
notably in the so-called vigencias futuras, or future expenditure com-
mitments) and structuring adjustments (in the mechanisms that insure
liquidity for the project) were made. These adjustments gave the gov-
ernment greater ability to honor its guarantees in a more timely fash-
ion. Fundamental problems continued, however. The government did
riot have the capability to value its obligations, and, more important,
it did not have adequate assets to offset its obligations or the cash to
provide for them.
Handling contingent liabilities through vigencias futuras implied
that future budgetary expenditures could be earmarked, ensuring bud-
geting while explicitly recognizing the obligations. Even though this
mechanism ensured the availability of funds to support the liabilities,
it introduced severe rigidities for future government budgets and was
based on crude valuation methodologies for contingent liabilities. The
liquidity mechanisms have taken the form of trusts or standby loan
facilities, which imply important costs because they call for greater
than necessary budget provisions. In addition, this system has proven
inefficient in providing the liquidity required by private investors.
Therefore, recent projects have relied on the participation of financial
intermediaries that manage the resources allocated to cover these li-
abilities when they emerge.
Under the initiative of the Ministry of Finance and the General Di-
rectorate of Public Credit, the government has sought to correct most
of these difficulties, while improving its fiscal discipline and generat-
ing the appropriate conditions in the real sectors in order to, for ex-
ample, attract private capital into the provision of public services
without the need for new guarantees. Measures were taken from a
normative perspective, such as Law 448 (1998) and Decree 423 (2001),
with the purpose of both providing liquidity as contingencies became
effective and minimizing the inconveniences that resulted from the
reduction in investment resources, or other items, in that event. Law
448 regulated the valuation, budgeting, and control of contingent li-
abilities; created a contingency lund; and defined the resources to be
devoted to financing such liabilities.
274 ECHEVERRY, NAVAS, GUTIERREZ, AND CARDONA
Law 448 (1998) and Decree 423 (2001)
The principal elements of Law 448 are that it:
o establishes the obligation for public entities to budget contingent
liabilities as debt service, implying that they must make the appropri-
ate provisions to face such liabilities (in this way, the government can
assure liquidity for contract guarantees);
o creates the State Entities' Contingency Fund to tackle liabilities
generated under specific conditions that are defined by the government;
o regulates the approval and control of contingent liabilities;
o establishes methodologies and procedures for the valuation, bud-
geting, and control of contingent liabilities;
o grants regulatory powers to the General Directorate of Public
Credit (the debt office of the Ministry of Finance) and to each state
entity's planning office; and
o establishes the requirement that contingent liabilities be explic-
itly included in contracts.
Decree 423
o assigns the National Council of Economic and Social Policy
(CONPES) responsibility for establishing the guidelines for contrac-
tual relationships between public entities and the private sector for the
development of infrastructure-thus CONPES is to direct the state's
contractual risk policy "based upon the principle, that public entities
should assume those risks which are proper to their public character
and their social objective, while the private parties should confront
those risks which are determined by their profitable nature" (Article
16, Decree 423); and
o creates the State Policy for Contractual Risks, which determines
the type of risks each sector is allowed to undertake, while unifying
the economic policy on risk management. A recent document issued
by CONPES (No. 3107) provides the general guidelines on the types
of risks that may be undertaken by the various sectors, and it compiles
the individual strategies that should be developed by the specific infra-
structure sectors to tackle such risks.
Valuation Methodologies
The General Directorate of Public Credit together with international
consultants produced risk quantification methodologies-one gen-
eral methodology and four sector-specific ones associated with the
road infrastructure, energy, basic sanitation, and drinking water sec-
tors. Contingent liabilities deriving from public credit operations also
may be regulated in the near future. The methodologies aim to deter-
CONTINGENT LIABILITIES IN COLOMBIA 275
mine the magnitude of the expected present value of contingent li-
abilities, breaking them down by risk source, and the distribution of
payments over time. This should facilitate the construction of an
expanded balance sheet, an explicit recognition of contingent liabili-
ties, and the incorporation of cost-of-assumption measures in the
contract structuring.
Budgeting for Contingent Liabilities
Government entities are to define their guarantee policy, based on a
valuation of the contingent liabilities using the respective methodol-
ogy and to quantify the financial and budgetary costs of assuming
such obligations. Subsequently, they are to include in their budgets the
contribution to the State Entities' Contingency Fund that corresponds
to the current year as opposed to the full present value of the liability.
Such a contribution is to be budgeted as debt service, not investment.
Therefore, the transition from thc previous system for existing con-
tracts may result, from a budgetary perspective, in the perception of a
reduction in investment.
Overall, the national government policy aims to make contingent
liabilities in infrastructure projects evident through their proper valu-
ation and give their inclusion in the budget the same priority as debt
service.
D)eposit Plan
A deposit plan approved by the General Directorate of Public Credit
cletermines the amount and timing of deposits to be made to the Con-
tingency Fund. These series of payrnents are meant to distribute through
time the building of an asset to offset the generated contingent liabil-
ity. This arrangement should ensure liquidity and thereby guarantee
the availability of resources to cover the contingencies emerging in the
next year or two.
State Entities' Contingency Fund
The Contingency Fund is meant to ensure a close relation between the
value and liquidity of the guarant:ee and the explicitly agreed on obli-
gations in contracts for infrastructure projects. Under this arrange-
ment, the guarantee's credibility is improved, because deposits in the
fund constitute the offsetting asset required to cover the contingency,
and the resources' value is maintained over time as earned interests are
reinvested. The funds will be obtained from contributions made by
those public entities that participate in projects involving guarantees
to the private sector. The fund acts as an account in the sense that it is
276 ECHEVERRY, NAVAS, GUTIERREZ, AND CARDONA
only held responsible for an amount equivalent to the contribution of
the respective entity. Deposits within the fund are broken down by
both project and individual risk level. Therefore, the deposits made by
different entities are not pooled together. In the event that contingent
liabilities do not arise, the entities' payments may be either reimbursed
or transferred to other projects.
Contingent Liability Appraisal
Because the individual risk profiles of contingent liabilities can vary
over time, the General Directorate of Public Credit will produce a
yearly revaluation in order to introduce the appropriate changes in the
entities' deposit plan, releasing resources when risk expires. This pro-
cess increases the probability that the entity's yearly deposits will cover
at least the contingency's current year value, therefore allowing a ho-
mogenization in magnitude and distribution over time of both the li-
ability and the offsetting asset.
Overall, Law 448 and Decree 423 represent important steps to-
ward attaining fiscal discipline through proper budgeting, adequate
and objective accounting through new valuation methodologies that
grant greater transparency to the contract system, and the assurance
of liquidity to offset assumed liabilities through the new fund. All of
the above enhance the achievement of a dynamic fiscal balance through
efficient intertemporal risk allocation.
Structural Adjustment
The analysis of fiscal imbalances in Colombia within an intertemporal
perspective reveals that the public sector does not have a short- or
medium-term liquidity problem, but rather one of long-term insol-
vency, which is stressed mainly due to the high value of pension liabili-
ties. This problem emphasizes the importance of decreasing contingent
liabilities in order to achieve a dynamic fiscal balance. The main prob-
lems faced by authorities are the size of the adjustment needed and the
institutional constraints to be overcome in order to pursue an efficient
allocation of savings to confront flow and stock imbalances.
Among the most important structural reforms directed toward fis-
cal adjustment are those that tackle the unsustainability problem asso-
ciated with pension liabilities and subnational entities:
o Subnational Government Income and Expenditure Reform. The
main reform measures are centered on reducing expenditures. In addi-
tion to a reduction in regional pressures on the central government via
further decentralization, policies are directed toward more efficient
CONTINGENT LIABILITIES IN COLOMBIA 277
expenditures: the rationalization of the size of the payroll and funding
of regional and teacher pension funds. Reductions in the expenditures
oF the central government refer essentially to public investment, with
an emphasis on shifting infrastructure investment to the private sector
through a concession system.
* Reform of Regional Transfers,from the Central Government (Con-
stitutional Amendment). Currently, national government transfers to
the regions are the budget item with the greatest relative importance;
they are equivalent to 39 percent of the total national expenditure.
Additionally, they have recorded a high growth rate, rising from 2.8
percent of GDP in 1990 to 5.3 percent in 1999.
The proposed reform will reduce the pressure that these transfers
place on the central government by allowing the transfers to grow in
real terms at a rate equivalent to that of the population (1.5 percent)
once they reach their maximum value as a percentage of national cur-
rent income.
* Creation of a Regional Pension Fund. Regional pension liabili-
ties have reached a level equivalent to 39 percent of GDP, which al-
ready represents a threat to fiscal stability; the lack of regional
government savings has resulted in serious payment delays. The objec-
tive of the reform policy is therefore to direct resources toward the
creation of reserves with the purpose of covering pension liabilities for
a maximum period of 30 years. The necessary resources will be ob-
tained from the joint participation of the regional and central govern-
ments. Regional funding will be derived from an increasing share of
transfers to municipalities.
* Rationalization of the Social' Security System. One of the most
important factors in the current worsening of Colombia's fiscal situa-
tion is transfers to the social security fund. These transfers represent
30.4 percent of those at the central government level, of which 17
percent is related to severance payments. Not only do these transfers
represent a major burden for the central government, but the design of
the social security system implies the rapid growth of these liabilities.
UJnder the current system, severance payments suffer geometric growth,
vvhich resulted in the doubling of this liability between 1996 and 1999.
This geometric growth is the result of the drawing of severance pay-
rnents discounted by the nominal value at the moment at which the
contribution took place as opposed to the moment of the last contri-
bution. The elimination of the retroactivity of severance payments has
been proposed. In addition, the proposed reform of social security con-
templates a five-year extension of the age of eligibility for receiving
pensions, for both male and female workers, and a 300-week exten-
sion of the minimum contribution period (previously 1,000 weeks). In
the same sense, reformers contemplate a reduction in the rate at which
the pension value increases as a percentage of a determined salary,
278 ECHEVERRY, NAVAS, GUTIERREZ, AND CARDONA
according to the time of contribution. Moreover, calculation of the
pension salary is meant to be transformed so that it is based on a 20-
year income reference period rather than the last 10 working years.
Further measures imply the elimination of special pension regimes,
which entails introducing teachers and armed forces, among others, in
the common regime (Law 100); accelerating the transition to a new
system in terms of the increase in the retirement age and reduction in
the pension salary; and increasing the requirements for the guarantee
of a minimum pension and the restrictions associated with the transfer
within pension funds in order to avoid speculation and savings mis-
management.
Once the fiscal package has been designed, the government has to
confront two independent bodies when seeking approval of the re-
forms: Congress and the Constitutional Court, both of which can ei-
ther hinder or promote government efforts to impose long-term fiscal
adjustments. Therefore, intertemporal budgeting and financing are a
necessary, but not sufficient, condition for long-term sustainability.
The commitment of these two arms of public power to harmonizing
their policies with fiscal adjustment strategies, by assimilating under-
standing and assuming the intertemporal consequences of their deci-
sions with the dynamic structure of budget constraint, has become as
crucial as the executive strategy itself (Echeverry and Navas 2000).
Note
1. This paper benefited from the collaboration of the Macroeconomic
Analysis Unit in the National Planning Department and the Analysis Divi-
sion in the Ministry of Finance.
References
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April): 57-86.
Echeverry, Juan Carlos, and Ver6nica Navas. 2000. "Confronting Fiscal Im-
balances via Intertemporal Economics, Politics and Justice: The Case of
Colombia." Archivos de Macroeconomia. No. 129. National Planning
Department, Bogota. (Forthcoming, Economic Review, Federal Reserve
Bank of Atlanta).
Echeverry, Juan Carlos, Maria Victoria Angulo, Gustavo Hernandez, Israel
Fainboim, Cielo Numpaque, Gabriel Piraquive, Carlos Rodriguez, and
Natalia Salazar. 1999. "El Balance del Sector Publico y la Sostenibilidad
Fiscal de Colombia." Archivos de Macroeconotinia. No. 115. National Plan-
ning Department, Bogota.
PAR'T II
Dealing with
Specific Sources of
Government Fiscal Risk
Analytical and Managerial Tools
CHAPTER 13
Pension Guarantees:
A Methoclology for
Assessing ]Fiscal Risk
George G. Pennacchi
University of Illinois
TRADITIONALLY, MOST COUNTRIES THAT have provided social secu-
rity to the elderly have chosen unfunded, defined-benefit pension sys-
tems. Many of these pay-as-you-go (PAYG) systems have come under
stress because of demographic trends, mismanagement, and economic
downturns.' Some countries have responded by increasing worker con-
tribution rates, cutting retirement benefits, or making other fiscal ad-
justments to improve the financial viability of their systems. However,
these types of reforms have been politically difficult and have moved
pension systems only partway toward long-term solvency.2 Uncertainty
about the sustainability of PAYG systems and their perceived inequi-
ties also can lead to worker dissatisfaction and evasion.3
An approach that holds greater promise of long-term pension re-
form is to allow part of an indiviclual's pension to take the form of a
fully funded, defined-contribution plan. Combined with a defined-ben-
efit component, such a system represents a dual-pillar approach. The
first pillar can take the form of a guaranteed minimum pension for
those who contribute over a sufficiently long period to the defined-
contribution plan, such as in Chile's current pension system. Alterna-
tively, the first pillar can be a separate defined-benefit PAYG program,
but of a more limited size than that of traditional single-pillar systems.
The defined-contribution second pillar would be funded by payroll
deductions and could be privately or publicly managed.
By reducing the defined-benefit component of an individual's pen-
sion, a dual-pillar system decreases the exposure of pension benefits to
283
284 GEORGE PENNACCHI
political manipulation. Potential distortions of an individual's work
and savings behavior are mitigated, because the defined-contribution
component links at least a part of payroll contributions to an individual's
pension benefits. Yet by maintaining a defined-benefit portion, the sys-
tem provides insurance to workers who experience low lifetime wage
incomes or realize low investment returns on their defined-contribu-
tion accounts. In this sense, a dual-pillar system provides diversifica-
tion against political, income, and investment risks.4
Because pension systems that include a defined-contribution com-
ponent expose individuals to investment risks, governments typically
find it necessary to provide some type of guarantee on individuals'
pension returns. Such guarantees, however, create a government con-
tingent liability that can be difficult to value-that is, the government
creates an obligation when it commits to providing the guarantee, but
the actual payments required to fulfill it are often far into the future
and difficult to predict. Also, the guarantee payments may depend not
only on the returns earned on financial market investments but also on
the future levels of workers' wages and other macroeconomic and de-
mographic factors.
Although it can be a complex task, estimating the value of these
government guarantees is important for gauging the complete fiscal
cost of pension reform. Guarantees represent an implicit subsidy to
participants in pension plans that, in principle, should be included in
government budget statistics. Moreover, it may be possible for the
government to recover the cost of its guarantees by charging an initial
fair insurance premium to the beneficiaries of the guarantees. In some
cases, such a risk-based premium, equal to the government's liability,
can reduce the economic distortions associated with the provision of
the guarantee.
This chapter describes a methodology for assessing the cost to a
government of providing pension guarantees. The approach is alter-
natively referred to as contingent claims analysis, arbitrage-pricing
theory, or option-pricing theory. This framework was first applied to
valuing financial options, but it has since been extended to value
different types of guarantees, such as government deposit insurance
and loan guarantees. The chapter also discusses the economic as-
sumptions underlying this valuation technique. And it describes par-
ticular types of pension guarantees provided by different country
governments and how a government's risk associated with these guar-
antees can be managed.
Four main types of government pension guarantees are analyzed.
They share a common characteristic in that their values depend, at
least in part, on the risk from investments in financial securities. The
first two types are associated with a defined-contribution pension plan.
They are a guarantee on the periodic rates of return earned by a de-
PENSION GUARANTEES 285
fined-contribution pension fund and a guarantee of a minimum pen-
sion benefit for an individual participant in a defined-contribution plan.
The second two types of guarantees relate to defined-benefit plans.
Because many countries' pension systems include a voluntary third
pillar in the form of an employer-sponsored pension plan, this chapter
considers a government guarantee of employee benefits promised by
an employer-sponsored defined-benefit plan. It also analyzes a guar-
antee of benefits paid by a mandatory, partially prefunded, single-pil-
lar, defined-benefit pension system in which payroll tax rates are linked
tc) the investment performance of the plan's trust fund. Although it is
not covered in this chapter, Goss (I1999) discusses how the solvency of
an unfunded, defined-benefit PAYG plan can be assessed.5
The chapter is structured as follows. The next section discusses the
basic economic assumptions on which the contingent claims valuation
framework rests. It considers the financial market conditions neces-
sary for such assumptions to be valid. The example of valuing a simple
pension guarantee that follows illustrates the basic technique used
to compute the value of guarantees. The next section then analyzes
the types of guarantees provided by different country governments.
It points out the factors that influence the value and risk of such guar-
antees and how a guarantee might be structured in order to control the
government's exposure and risk. The final section offers concluding
comments.
The Methodology Used to Value Pension Guarantees
Guarantees and insurance are examples of contingent claims or de-
rivatives. Other examples include financial options, futures, and swap
contracts-securities whose payments are tied to the value of an un-
derlying security or asset. Guarant:ees are similar to options.
Option contracts fall into two categories. A call option is a contract
that gives its holder the right to buy a particular underlying asset at a
prespecified price at some future date. A put option gives its holder the
right to sell a particular underlying asset at a prespecified price at
some future date. This prespecified price is referred to as the option's
exercise price or strike price. Of the two types of options, put options
are the most similar to pension guarantees. Consider a government
guarantee of a minimum value for an individual's pension account. In
the future, if the value of the individual's pension account (corresponding
to the value of an underlying asset'l is lower than the prespecified mini-
rnum pension level (corresponding to an exercise price), then the indi-
vidual can "sell" the pension account to the government and receive
the minimum pension. In this case, the government realizes a future
expense equal to the difference between the minimum pension and the
286 GEORGE PENNACCHI
value of the individual's account balance. If, instead, the future value
of the individual's pension account exceeds the value of the minimum
pension, the individual maintains the account (chooses not to exercise
the option to sell it to the government) and the government's realized
expense is zero. Thus as the provider of the guarantee (put option), the
government realizes a future expense equal to the difference between
the guarantee value (exercise price) and the pension account balance
(underlying asset) if and only if this difference is positive.
Fortunately, important advances in valuing contingent claims, in-
cluding government guarantees and insurance, have been made over
the past few decades. Since the theoretical work on option pricing by
Black and Scholes (1973) and Merton (1973), financial economists
have developed and refined a general valuation technique. This re-
search is probably the most significant contribution of modern finance
theory. It has led to rigorous, yet practical, valuation models that have
stimulated the development of new financial securities and markets.
In addition to pricing options, this work has been applied to valuing
many nonstandard contingent claims, including government deposit
insurance, loan guarantees, and pension guarantees.6
This option-pricing framework is attractive because it requires rela-
tively few assumptions and the assumptions are often realistic. The
approach starts by noting that a contingent claim or derivative inherits
the same risk as its underlying security or asset. In particular, because
a pension guarantee provides insurance against low security returns,
its payoff is tied to the value of the underlying securities held by the
pension fund. For example, an annual rate of return guarantee on a
defined-contribution pension fund requires the government to make a
payment to participants if their pension fund's annual return is below
a prespecified level. Another example is a government guarantee of a
minimum pension for a participant in a mandatory, defined-contribu-
tion pension plan. The participant's pension account balance at retire-
ment is partly determined by the returns earned by his or her pension
fund's securities. If this account balance is below the prespecified mini-
mum pension, then the government is obligated to make a payment
equal to the difference between the minimum pension and the account
balance. In both of these cases, the guarantee payment depends on
particular security returns. Thus the guarantee's value inherits a risk
similar to that of the pension fund's securities.
The key insight of option-pricing theory is that, under particular
conditions, the risk from providing a contingent claim can be hedged
by appropriate trading in its underlying securities. The cost of pur-
chasing this hedge portfolio, which eliminates the liability from pro-
viding the contingent claim, must then equal the value of the contingent
claim. In other words, it may be possible to create a hedge portfolio,
involving positions in the underlying securities, whose cash flows ex-
PENSION GUARANTEES 287
actly replicate the cash flows of the contingent claim.7 Such a hedging
portfolio represents an asset whose value perfectly offsets the liability
of providing the contingent claim, so that the net liability of the pro-
vider always equals zero. If it is possible to create this hedging portfo-
lio, then its value must equal that of the contingent claim since their
future cash flows will be identical. If their values differed, then arbi-
trage would exist.
Arbitrage refers to a situation in which an investor can make a
riskless profit without investing any of his or her own wealth. More
informally, arbitrage can be described as a "free lunch." In highly com-
petitive security markets, arbitrage is extremely rare, because it im-
plies that investors can make riskless profits while putting none of
their personal wealth at stake. In the context of a contingent claim and
a hedging portfolio that replicates the claim's payoff, if the prices of
these two assets were not identical., arbitrage would result from selling
the overvalued asset and buying tlhe undervalued one. The difference
in the values of the two assets would then represent a riskless profit.
An assumption often made by the option-pricing approach is that there
are no transactions costs to buying or selling the contingent claim's
underlying securities and that in such a "frictionless" market arbitrage
does not exist.
In the context of a government pension guarantee, suppose a gov-
ernment chooses to hedge its risk of making payments to pension fund
participants if security returns are low. How might this be done? A
successful hedge would be one in which the government makes a profit
if security prices decline, offsetting its losses in the form of its pay-
rnents to pension participants. Such profits could be made if the gov-
ernment had a short position in the underlying securities. This could
be accomplished by short-selling the securities. In a short sale of a
security, the short-seller (government) borrows a share of a security
from its owner and promises to return it at some future date.8 At the
beginning of the short-sale period, the short-seller trades the security
to a third party, thereby obtaining proceeds equal to the security's
rnarket price. These proceeds can then be invested in a riskless secu-
rity, such as a short-term government bond.9 At the end of the short-
sale period, the short-seller repurchases a share of the security at its
end-of-period market price and returns it to the owner. Therefore, the
short-seller's profit at the end of the short-sale period equals the
security's beginning-of-period market price plus accrued interest mi-
Ilus its end-of-period market price. Thus the lower the security's return
over the period, the greater is the short-seller's profit.
An alternative, though similar, method for establishing a short posi-
tion in securities is via a futures market. If futures contracts on the
securities, or on indices of the securities, are traded, then a short posi-
tion in a futures contract would provide essentially the same hedge as
288 GEORGE PENNACCHI
a short sale. Hedging with a short futures position would result in a
profit when the securities underlying the futures contract have a low
return. Moreover, an attraction of hedging with futures on the securi-
ties is that the transactions costs are often quite low compared with
those for hedging with the actual underlying securities.
By either method of hedging, the government would be "privatiz-
ing" its risk in the sense that the risk from low security returns is
transferred to individuals or institutions willing to bear it. In the case
of the short-sale hedge, the risk shifts to the party that purchases the
securities during the short-sale period. In the case of a futures contract
hedge, the risk is transferred to the party taking the opposing long
position in the futures contract. The parties that assume this risk may
be domestic or foreign investors wishing to diversify their portfolios.
More specifically, it can be shown that whether the government
hedges via a short sale or a short futures contract, the appropriate
hedge involves a short position in fewer shares of the underlying pen-
sion fund securities than those actually held by the pension funds.'" In
addition, the hedge involves an additional investment in risk-free secu-
rities-that is, the government would choose to reduce or repurchase
some of its debt.
In practice, governments typically do not attempt to hedge their
exposure to guarantees-that is, the risks of guarantees are not priva-
tized. However, the option-pricing approach values these guarantees
at what would be the theoretical marginal cost that investors would
charge the government to take over its commitments. It is assumed
that such investors would require an expected rate of return on pro-
viding the guarantee that is consistent with the expected rate of return
they require from holding the underlying securities. The Appendix to
this chapter details this relationship between the expected rate of re-
turn on the government's guarantee and the underlying securities.
In cases in which a perfect hedge is not possible, so that the guarantee's
cash flows cannot be replicated exactly, the option-pricing approach
might still be useful in deriving, or deriving bounds on, the guarantee's
value. For example, if the securities underlying the guarantee are costly
to trade, it may be possible to trade using alternative securities that have
no transactions costs. With these alternative securities, a government
might be able to create an imperfect hedge portfolio whose "tracking
error" relative to a perfect hedge represents nonsystematic or diversifiable
risk. In such a case, the guarantee can be valued exactly as if a perfect
hedge could be created (see Merton 1998 for a derivation of this result).
In addition, there may be other situations in which the hedge portfolio's
tracking error risk is not fully diversifiable but upper and lower bounds
on the cost of the contingent claim can be derived (see Cochrane and
Saa-Requejo 2000 and Bernardo and Ledoit 2000). The more closely
PENSION GUARANTEES 289
the guarantee's cash flows can be replicated by the underlying securities,
the tighter will be the bounds on its cost.
In some cases, guarantees depend not only on the values of underly-
ing financial securities but also on other sources of uncertainty. A mini-
rnum pension guarantee for an indlividual participant in a mandatory,
clefined-contribution pension plan is an example. In addition to the
returns earned by securities in the individual's pension fund, the guar-
antee depends on the contribution levels of the individual which, in
turn, depend on his or her wages. Thus this guarantee is affected by
wage risk. However, it may be possible to hedge wage risk using a
portfolio of financial securities in which the portfolio's tracking error
represents only nonsystematic risk. In such a case, the value of the
pension guarantee can be derived (see Pennacchi 1999a for a discus-
sion of the issues involved in modeling wage risk). A similar technique
might be used to handle other types of risks that affect pension guar-
antees, such as mortality risk."'
Valuing a Pension Guarantee: A Simple Example
This section illustrates how option pricing can be used to value a simple
pension guarantee. Suppose that a defined-contribution pension fund
holds a diversified portfolio of common stocks. The value of the pen-
sion fund, and therefore the value of the participants' retirement ben-
e fits, will vary with fluctuations in this stock portfolio. The pension
fund could insure itself against significant declines in the value of its
securities by purchasing a put option on its portfolio. For example,
assume that the value of the pension fund's securities currently equals
$10 million, and it wishes to insujre itself against a drop in the value of
the portfolio to below $9 million during the next year. If options on
such a stock portfolio were available from an options exchange or
from an investment bank derivatives dealer, put options on the portfo-
lio having a one-year maturity and an exercise price of $9 million
could be purchased.
The value of the put option contracts when they mature in one
year's time can be written as max ($9M - S, 0), where S, is the ran-
dom value of the pension fund's stock portfolio at the end of the year
and max ($9M - S, 0) means select the maximum of $9 million minus
S, and 0. Thus the put option, being analogous to a guarantee or insur-
ance contract, has a positive maturity value only if the end-of-year
value of the pension fund's stocks, S,, sinks to less than $9 million. By
owning this "portfolio insurance," the pension fund has a combined
end-of-year value given by S, + max ($9M - S, 0) = max ($9M, S,),
which is never less than $9 million.
290 GEORGE PENNACCHI
Rather than obtaining insurance by purchasing put options from an
exchange or dealer, a government could provide the equivalent pen-
sion insurance. In this case, a government guarantor has an end-of-
year liability given by max ($9M - S,, 0). Figure 13.1 graphs the
end-of-year values of the pension fund if it were uninsured (equaling
S,), of the guarantee (equaling max ($9M - S,, 0)), and of the pension
fund if it were insured (equaling max ($9M, Sj)). Clearly, the guaran-
tee places a lower bound on the end-of-year value of the insured pen-
sion fund.
The value of the government guarantee is easily determined at the
end of the year when S, becomes known, but it is more challenging to
determine its value (cost) at the beginning of the year. If the guarantee
were exactly equivalent to a traded put option contract, then its mar-
ket value could be inferred from the current market price of the op-
tion. However, rarely, if ever, does a pension guarantee have an exact
private market counterpart. Thus another method, such as option pric-
ing, is needed for calculating the guarantee's theoretical market price.
Figure 13.1. End-of-Year Values of Stock Portfolio,
Pension Fund Guarantee, and Pension Fund
(millions of dollars)
End-of-year values
15
14 Uninsured pension fund /
13-
12 -
10 Insured pension fund
7 7
6
5
4
3 / / \P ension fund guarantee
1 \
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
End-of-year stock portfolio value
Source: Based on calculations from the author.
P:ENSION GUARANTEES 291
As detailed in the Appendix, one useful option-pricing method is
the "risk-neutral" valuation technique or, more generally, the martin-
gale-pricing approach. According to this technique, the guarantee's
present value equals the risk-neutral expectation of the guarantee's
end-of-year value discounted at the risk-free interest rate. The risk-
neutral expectation of a random future value is defined as its expected
fuiture value under the assumption that the expected rate of return on
the underlying assets equals the risk-free interest rate."2 Thus in using
this valuation technique, one would compute the expected value of
max ($9M - S1, 0) by setting the expected rate of return on the pension
fund's stock portfolio equal to the risk-free rate, even though the "true"
expected rate of return on the stock portfolio is really different. The
present value of the guarantee then equals this value discounted at the
risk-free interest rate.
It should be emphasized that this risk-neutral valuation or martin-
gale-pricing technique does not assume that the stocks' expected rates
of return truly equal the risk-free rate. The only assumption being
made is that the equilibrium values of the guarantee and the underly-
ing stock portfolio are such that investors price the risk inherent in
these two assets in a consistent manner. It just turns out that, math-
ematically, one can arrive at the correct value by computing the
guarantee's expected payoff as if the stocks' expected rates of return
equaled the risk-free rate and then discounting this expected payoff by
the risk-free rate.'3
Intuitively, the risk-neutral technique gives the correct value for
the guarantee because of two erroneous assumptions whose effects
cancel. One incorrect assumption is that the underlying stock portfo-
lio has an average rate of return equal to the risk-free rate-that is, its
expected rate of return contains no risk premium. This implies a risk-
neutral expectation of the guarantee's future payoff, max ($9M - S,, 0),
that differs from its true expectation, leading to the first error. The
other incorrect assumption is that this risky payment should be dis-
counted at the risk-free rate rather than at a rate reflecting the risk
premium of the underlying stock portfolio, leading to the second error.
Because both the first and second errors fail to account for a risk pre-
inium, the first error overstates the expected value of the guarantee's
payoff by the risk premium, whlile the second error understates the
discount factor by the risk premium. Mathematically, the two errors
cancel, leading to a correct valuation. Importantly, this technique does
not require making a tenuous assumption about the true risk premium
,:f the underlying stocks.
To calculate the guarantee value, one needs to specify the variance
of the stock portfolio's return, which is a measure of the volatility of the
portfolio's value. This variance can be estimated relatively accurately
from data on the stocks' historical returns. Then, one can compute the
292 GEORGE PENNACCHI
risk-neutral expected value of max ($9M - S,, 0) using a Monte Carlo
technique to simulate the end-of-year values of the portfolio, SI. This
calculation involves applying a random number generator to compute
many possible outcomes of S, where the portfolio's expected rate of
return is set equal to the risk-free rate and its variance of return is set
equal to an estimate based on historical returns. Then, for each real-
ized value of Sl, the quantity max ($9M - SI, 0) is computed and these
quantities are averaged over all of the realizations. For a sufficiently
large number of random simulations of S, the average value of max
($9M - S,, 0) will be arbitrarily close to its theoretical risk-neutral
expected value. The last step is to discount this average using the risk-
free rate, which leads to the present value of the guarantee.
Figure 13.2 illustrates this technique in the context of this simple
example. Assuming that the beginning-of-year security portfolio value
equals $10 million, the risk-free interest rate equals 5.00 percent, and
the annual variance of the portfolio's return equals 4.00 percent, the
figure shows 200 random realizations of the risk-neutral path of the
security portfolio over a 52-week (one-year) period. These simulations
Figure 13.2. Risk-Neutral Paths of Stock Portfolio Values
Portfolio value (millions of dollars)
20
19
18
' 7
1 6
15
14
13
12
10
9 ~
8
7
6
0 4 8 12 16 20 24 28 32 36 40 44 48 52
Weeks
Souirce: Based on calculations from the author.
PENSION GUARANTEES 293
generate a frequency distribution of values of S,-that is, a set of real-
izations of the end-of-year (week 52) risk-neutral stock portfolio value.
For a sufficiently large number of simulations, this frequency distribu-
tion takes the shape shown in Figure 13.3.14
The risk-neutral value of the government's payment is found by
computing the quantity ($9M - S,) for each realization of S, less than
$9 million.'5 Summing these quantities and dividing by the total num-
ber of simulations (for example, 10,000) results in the risk-neutral
expected value. Finally, multiplying this expected value by the risk-
free discount factor 1/1.05 results in the present value of the guaran-
tee, which in this example equals $231,000.
This Monte Carlo method for computing the guarantee's value is in
the spirit of scenario testing. Each simulated value of S, can be viewed
as one possible scenario for the pension fund's end-of-year portfolio
value. However, the purpose of the technique is not to reveal the vari-
ous possible ex post future payments the government would be re-
quired to make. Rather, it is to find the present value of the contingent
guarantee payment.
Figure 13.3. Risk-Neutral Distribution of End-of-Year
Portfolio Values
0.20
0.18
0.16-
0.14-
0.12 -
0.10
0.08 -
0.06 -
0.04 -
0.02
0
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
End-of-year portfolio value (millions of dollars)
Source: Based on calculations from tht author.
294 GEORGE PENNACCHI
In practice, government guarantees are more complex than the stan-
dard put option-type guarantee in the example. In addition to risk
from portfolio returns, some guarantees depend on individuals' wage
and mortality risk and risk from changes in market interest rates.
However, the Monte Carlo simulation technique can be extended to
include these other factors (a discussion of how these multiple factors
can be modeled is given in the appendix of Pennacchi 1999a).
Analyzing Specific Types of Pension Guarantees
This section considers various types of pension guarantees offered by
different countries, discusses how these guarantees can be valued, and
analyzes their risk characteristics.
Guarantees on a Pension Fund's Rate of Return
Some countries with pension systems that include a defined-contribu-
tion pillar guarantee that pension funds earn a minimum rate of return
over given periods of time. These guarantees differ in the way that
countries define the minimum return, in the length of the period over
which pension fund returns are calculated, and in the actions taken if
a pension fund fails to earn the minimum return.
In Uruguay, pension funds are known as Asociaciones de Fondos de
Ahorro Previsional (AFAPs) and can be publicly or privately managed
(see Mitchell 1996 for a discussion of Uruguay's reformed pension
system). The government of Uruguay guarantees that public (but not
private) pension fund participants earn a minimum annual real rate of
return of 2 percent. Thus the government is required to make up the
difference between this minimum return and the public pension fund's
actual return if, during any year in the future, the actual return turns
out to be less than this minimum. Pennacchi (1999a) employs an op-
tion-pricing approach to derive an explicit formula for the value of
such a fixed real rate of return guarantee.'6 For the general case of a
fixed minimum return equal to m, the formula shows that the value of
the guarantee increases as r - m declines, where r is the risk-free real
interest rate. Also, the greater the variance of the returns earned by the
pension fund's portfolio of securities, and the greater the growth rate
of participants' net contributions to the pension fund, the higher is the
guarantee. Indeed, if the growth rate of net contributions is positive,
the present value of the government guarantee can become very large
the longer the time horizon over which the government commits to
making this guarantee.
The structure of this fixed real rate of return guarantee is somewhat
worrisome. Even if pension funds are invested in bonds that pay a real,
PENSION GUARANTEES 295
riskless return, such as the index bonds offered by the Uruguayan gov-
ernment, there is no assurance that their real returns will exceed 2
percent. The real yield on an indexed bond can vary and, in principle,
fall below 2 percent or even become negative. This may explain why
rate of return guarantees offered by other countries tend to differ from
Uruguay's fixed rate of return guarantee. Other countries' guarantees
tend to set a minimum rate of return that is linked to a time-varying,
market benchmark rate of return.
Chile's mandatory defined-contribution pension system is based on
private pension funds known as Administradoras de Fondos de
Pensiones (AFPs). The minimum annual real rate of return earned by
an AFP is linked to a benchmark that equals the average annual real
rate of return earned by all Chilean AFPs. Let Ra be the realized aver-
age annual rate of return earned by all other AFPs in a given year.
Then the requirement is that each AFP earn at least min(Ra- a, 13R),
vvhere a = 0.02, 3 = ½/2, and "min(x, y)" means select the minimum of
x and y. This implies that when Ra exceeds 4 percent, each AFP must
earn at least ½/2Ra, and when Ra is below 4 percent, each AFP must
earn at least Ra minus 2 percent. Each AFP must hold capital (a guar-
antee fund) of at least 1 percent of the value of its pension portfolio,
invested in the same security portfolio as that of its pension fund. In
the event that the fund's return is less than min(Ra - a, f3Ra), it must
make up the difference from its capital and replenish its capital within
15 days. Failure to do so would lead to a loss of the AFP's license.
Thus if an AFP maintains a capital ratio of c = 0.01, the government
would be exposed to loss when an AFP earns less than min(R.- a, P3Ra)
-- c = min(Ra- a - c, IRa- c).
As discussed in Mitchell and Barreto (1997), Peru requires the same
minimum return on its second-pillar, private defined-contribution pen-
sion funds (Administradoras Privadas de Fondos de Pensiones) as does
Chile. Argentina has a similar dual-pillar system, but the minimum
rate of return required of its private pension funds equals min(Ra- a,
I3Ra), where a = 0.02, but a = 0.7. As described in Fischer (1998),
Colombia sets a minimum rate of return for each of its private pension
funds, Administradoras de Fondos de Pensiones (AFPs). However, the
minimum is linked to a benchmark return that depends not only on
the average return earned by other AFPs, but also on a Colombian
stock market index and an index of publicly traded debt. Moreover,
the benchmark's weights assigned to the stock and debt indices depend
on the proportion of Colombian stocks that the individual AFP holds
in its portfolio. The stock index's weight in the benchmark increases,
up to a limit, with the AFP's portfolio allocation in stocks.
The Eastern European countries that have instituted dual-pillar pen-
sion reforms also provide rate of return guarantees. As reported by
Palacios and Rocha (1998), Hungary requires that its second-pillar,
296 GEORGE PENNACCHI
defined-contribution pension funds earn a return equal to at least 90
percent of the return on a benchmark index of long-term government
securities. Thus Hungary has gone a step further than Colombia in
that its benchmark depends solely on an index of debt securities. Po-
land provides still another variation on a rate of return guarantee. As
Cholon, G6ra, and Rutkowski (1999) describe, Poland requires that
its private pension funds earn a rate of return linked to the average
return earned by all Polish pension funds, R,. The formula for the
minimum return has a structure similar to those of Chile, Peru, and
Argentina. It equals min(R,- a, ,BR), where cx = 0.04 and 3 = ½/2.
However, rather than calculate returns on an annual basis, a pension
fund's return is compared with this benchmark over a 24-month period.
Both Fischer (1998) and Pennacchi (1999a) use option-pricing ap-
proaches to value relative rate of return guarantees. They note that
these guarantees are similar to options in that they exchange one asset
(the pension fund's portfolio) for another (the benchmark portfolio).
Fischer (1998) calculates the value of Colombia's minimum return
guarantee, while Pennacchi (1999a) values guarantees of the form of-
fered by Chile, Peru, and Argentina. From their findings, some general
conclusions can be drawn about the characteristics of these guaran-
tees. First, as in the case of a fixed minimum return guarantee, the
greater the growth rate in net new contributions to the pension fund,
the greater is the government's liability. Second, the cost of the
government's guarantee increases as the variance in the pension fund's
portfolio return rises and as the correlation between the pension fund's
portfolio returns and the benchmark returns falls. This latter effect has
the potential to create an incentive for pension funds to take excessive
risk by increasing their tracking error vis-a-vis the benchmark.
To control this potential moral hazard behavior, governments have
established additional regulations. For one thing, governments provid-
ing rate of return guarantees regulate the types of securities that a pen-
sion fund can hold. Pension funds are limited in the proportions of their
portfolios that can be invested in domestic stocks, foreign investments,
and high-credit risk debt. This is an obvious attempt to directly limit the
pension fund's portfolio variance. Another regulation that works indi-
rectly through incentives is setting minimum capital requirements, c.
Because managers of a pension fund will lose their capital if the fund's
return falls below its required minimum, but their capital will not in-
crease if the fund posts high returns, the greater a fund's required capi-
tal, the smaller is the fund managers' incentive to increase their portfolio's
risk. In addition to capital requirements, many governments specify that
if the return on a pension fund exceeds a prespecified benchmark, the
excess return must be placed in a special reserve fund.'7 This reserve
fund is then used, in addition to capital, to make up shortfalls should the
fund's return fall below its required minimum in the future. By not re-
I'ENSION GUARANTEES 297
warding the participants in the fund with excessively high returns, this
regulation also reduces risk-taking incentives.
The impact of these regulations appear to have been successful in
reducing moral hazard, but, in some sense, they may have been too
successful. For example, Chilean AFPs appear to have portfolios that
closely resemble each other, giving participants relatively little choice
regarding a risk-expected return tradeoff (see the evidence in Diamond
and Vald6s-Prieto 1994). To avoid such "herding" behavior, Hungary
allows wider minimum and maximum returns; Poland allows a longer
period (24 rather than 12 months) over which a fund's return is com-
pared with the benchmark; and Mexico requires no minimum or maxi-
mum rate of return whatsoever for its private pension funds.
Some countries, such as Colombia, Hungary, and Poland, charge
pension funds insurance premiums for their government rate of return
guarantee. The pension funds' premiums are then deposited into a
guarantee trust fund out of which claims are paid.'8 Although such
premiums lower a government's net cost of providing guarantees, the
government is still obligated to make good on its guarantee if the guar-
antee fund becomes insolvent. UJsing the option-pricing approach,
Pennacchi (1999b) shows how a Monte Carlo simulation technique
can be used to calculate a government's net liability when it charges
premiums and manages a guarantee fund.
Minimum Pension Guarantees
Because a fully funded, defined-contribution plan generally earns risky
returns, almost all countries whose social security systems include such
a plan offer some type of safety net.'9 Typically, this safety net takes one
of three forms: (a) a defined-benefit first pillar; (b) a minimum pension
guarantee; or (c) a defined-benefit first pillar together with a mini-
mum pension guarantee. For the first type of safety net, contributors are
entitled to certain defined benefits that are only partially earnings-related.
As in a single-pillar, defined-benefit system, a minimum pension benefit
is implicit in the formula used to specify the relationship between an
individual's contributions and retirement benefit, and it is unrelated to
the individual's balance in his or her defined-contribution pillar
(Argentina's social security system is an example.). The focus of this
chapter is, then, on valuing a government's minimum pension guarantee
for safety nets of the second and third types, because in these cases the
government's payments for a minimum pension depend on the invest-
ment risk of the individual's defined-contribution pillar.
A government guarantee of a minimum retirement benefit is one
way in which to introduce an element of income redistribution or in-
come insurance into a social security system.20 Governments set a mini-
mum pension level for qualifying retirees that is often indexed to the
298 GEORGE PENNACCHI
country's price level, or its average or minimum wage, or both.2' If at
retirement an individual's accumulated pension fund balance is insuf-
ficient to buy an annuity that would pay at least the minimum pen-
sion, then the government guarantor can respond in one of two ways.
One possibility is that the government makes an immediate payment
on behalf of the retiree equal to the difference between the minimum
pension annuity and the retiree's fund balance, thereby allowing the
individual to purchase the minimum pension annuity.22 An alternative
possibility is that the government allows the retiree to make pension
account withdrawals equal to the minimum pension, and, if the indi-
vidual lives long enough to exhaust the account, the government be-
gins paying the minimum pension directly to the individual. For either
possibility, the present value of the government's liability at the time
of the individual's retirement depends on the level of real interest
rates. The reason is that the level of interest rates affects both the
cost of purchasing a minimum pension annuity and the present value
of the government's payments after withdrawals deplete the individual's
account.
From the preceding discussion, one can identify three random fac-
tors that play a key role in determining a government guarantor's
liability. First, the real value of wages earned by individuals deter-
mines their pension fund contributions, which, in turn, affect the
real values of their pension account balances at retirement and the
level of the minimum pension set by the government. Second, the
real rates of return earned by pension funds affect the workers' pen-
sion account balances at retirement, because these rates of return
determine how quickly workers' contributions grow. Finally, the level
of real interest rates at retirement affects the government's liability
by determining the cost of the minimum pension annuity or the present
value of the government's payments after depletion of an individual's
account balance.
Zarita (1994) uses an option-pricing approach to value Chile's mini-
mum pension guarantee. His model allows pension funds to earn a
random rate of return, making a worker's pension account balance at
retirement random as well. If the worker's savings at retirement is less
than the cost of an annuity providing the minimum pension, the gov-
ernment is assumed to make a payment to cover the difference. The
value of this government payment is calculated using a Monte Carlo
simulation of the worker's risky pension investment assuming a deter-
ministic level of wage contributions each period and a constant real
interest rate.
Pennacchi (1999a) also values Chile's minimum pension guarantee
by extending the framework in Zarita (1994). Along with random
pension fund returns, a worker's real wage, and thus monthly pension
PENSION GUARANTEES 299
contribution, is assumed to follow a random process.23 The evolution
of real wages also influences the government's minimum pension ben-
efit. In addition, real interest rates are assumed to follow a random
process. Another difference from Zarita (1994) is that a retired worker
is assumed to select a scheduled withdrawal of funds from his or her
pension account rather than purchase an annuity. This is arguably more
realistic, because Chile gives retirees the choice between these two
actions only if they have a sufficient account balance to purchase a
rainimum pension annuity.14
Pennacchi (1999a) calculates guarantee values using a Monte Carlo
simulation based on three random, and possibly correlated, processes:
the real rate of return on pension fund assets, the growth in real wages,
and the change in the short-term real interest rate. An additional mi-
nor source of uncertainty is a person's mortality. The probability of
dleath at each age is assumed to be uncorrelated with economic vari-
ables and is taken from Chile's official life table.
The results of Zarita (1994) and Pennacchi (1999a) show that the
value of the government's minimum pension guarantee for a particular
person is a decreasing, convex function of that person's initial wage. In
addition, giving retired workers the option of a scheduled withdraw,
r ather than mandating the purchase of an annuity, raises the cost of
the guarantee. This higher cost stems in part from the Chilean gov-
ernment's withdrawal formula, which allows greater withdrawals for
individuals with higher savings, and this arrangement tends to subdue
the effect that greater retirement savings has on the age at which pen-
Sion fund balances are depleted. Other findings from these studies con-
firm that greater volatility in wage and pension fund returns also
increases the value of the minimum pension guarantee.
The Chilean case is an example of a minimum pension guarantee
provided solely with a defined-contribution pillar. Countries with similar
arrangements include Colombia, F l Salvador, Kazakhstan, and Mexico.
However, the valuation methodology is readily extended to the case of
a minimum pension guarantee provided with both defined-benefit and
defined-contribution pillars. Hungary and Poland are examples of such
social security systems. Hungarians participating in the defined-
contribution second pillar are entitled to a second-pillar retirement
annuity of no less than 25 percent of the value of their first pillar (see
Brixi, Papp, and Schick 1999 for a discussion of Hungary's pension
guarantees). In Poland, the government guarantor incurs an expense
if the value of individuals' first- and second-pillar accounts is less
than the minimum pension, which equals approximately one-third of
the average wage.25 In both of these cases, the government's cost of
providing a minimum pension depends on the retirement balances in
both the defined-contribution and defined-benefit accounts. Valuing
300 GEORGE PENNACCHI
this guarantee is similar to the case of a guarantee that depends only
on a defined-contribution balance, the only difference being that the
guarantee's maturity value now depends on the individual's defined-
benefit level at retirement, which depends, in turn, on the individual's
wage contributions.
Guarantees on Employer-Sponsored,
Defined-Benefit Pensions
In many countries, social security is supplemented by voluntary, em-
ployer-sponsored pensions. Worldwide, the majority of these employer-
sponsored pensions are of the defined-benefit type.26 Absent a guarantee,
employees' promised pension benefits are contingent on their corpo-
rate sponsor's ability to make good on these promises, subjecting the
employees to default risk. To remove this risk exposure, many govern-
ments provide guarantees on these privately sponsored, defined-benefit
pension funds.
In analyzing the cost of defined-benefit pension guarantees, it is
useful to distinguish between pension plans that are collateralized by
pension fund assets and those that are not. In most countries, private,
defined-benefit pensions are partially or fully funded by a separate
portfolio of assets-that is, a pension trust. However, there are impor-
tant exceptions, such as France, Germany, and Japan, which do not
require corporations to segregate particular assets from other corpo-
rate assets for the sole purpose of backing pension benefits. Rather,
pension liabilities are combined with other corporate liabilities and
backed only by the general assets of the corporation. Thus for these
nonsegregated pension plans, pension guarantees can be valued using
option-pricing techniques developed to value default risky corporate
liabilities. A review of these models for valuing corporate debt is given
in Merton (1990: chap. 13.4).
More common are corporate-sponsored, defined-benefit pensions
in which a pension fund, segregated from other corporate assets,
collateralizes pension liabilities. Such plans can be found in Canada,
the Netherlands, the United Kingdom, and the United States. In gen-
eral, pension funds of this type can be "underfunded," meaning the
value of the fund's assets are less than its liabilities, or "overfunded,"
meaning the value of its assets exceeds the present value of employees'
accrued benefits. Should a corporate sponsor become bankrupt at the
same time that its insured pension fund is underfunded, then the gov-
ernment guarantor would face a claim.
Several studies have applied option-pricing techniques to value this
type of pension guarantee.27 Because the government experiences a
loss only when both the corporate sponsor fails and the sponsor's pen-
sion fund is underfunded, the value of such a guarantee must depend
PENSION GUARANTEES 301
on the financial conditions of both, the corporate sponsor and its pen-
sion fund. Marcus (1987) was the first to use an option-pricing ap-
proach to model both financial structures. His model assumes that
both the asset/liability ratio of the sponsoring corporation and that of
its pension fund evolve randomly. Bankruptcy is assumed to occur
when the value of corporate assets first falls below the value of corpo-
rate liabilities. If this bankruptcy occurs, it is assumed that the
government's payment for providing the guarantee equals the value of
pension liabilities less pension asseis. However, the Marcus model does
not restrict this government payment to being positive. In other words,
if the pension fund was overfunded at the time of the bankruptcy, the
model assumes that the government guarantor receives a payment equal
to the difference between pension fund assets and liabilities.
Because in practice a government does not obtain a positive pay-
rnent should a firm with an overfunded pension plan fail, Pennacchi
and Lewis (1994) revised the Marcus analysis to restrict the govern-
rnent guarantor's payment to being positive. The guarantee is then
analogous to a put option on the pension fund's assets, with an exer-
cise price equal to the pension fund's liabilities and a contingent matu-
rity date determined by the sponsoring firm's bankruptcy. Lewis and
IPennacchi (1999) extend the model further to allow for random mar-
ket interest rates. This is an important generalization, because the
present value of employees' promised benefits can be highly sensitive
to interest rate risk. In addition, their study derives the value of the
insurance premiums paid by corporations for pension guarantees as-
suming the structure of insurance premiums charged by the Pension
Benefit Guaranty Corporation (PBGC), the U.S. government guaran-
tor of private pensions. This allows them to calculate the net cost of
pension guarantees (the present value of guarantee payments less the
present value of insurance premiums) for a sample of U.S. corporate
pension funds.
Lewis and Pennacchi's empirical findings indicate that the govern-
inent's cost of pension guarantees can be significant, even for a finan-
cially sound corporation with a currently overfunded pension plan.
The reason is that pension guaraiitees can be very long-term commit-
ments. There is a significant probability that a corporation, especially
one in a high-risk industry, could experience financial distress in the
future, at which time its pension fund could switch from overfunded
to underfunded status. Their results indicate that government regula-
tors should act to help prevent pension underfunding. One policy fol-
lowed by the PBGC is to charge corporations insurance premiums that
rise with the level of pension underfunding. Not only would higher
'premiums help to reduce the government's net cost of insurance if
underfunding rises, but they also would reduce a corporation's incen-
tive to begin pension underfunding.
302 GEORGE PENNACCHI
Guarantees on Social Security Systems Funded
with Risky Investments
The U.S. social security system is a traditional single-pillar, PAYG,
defined-benefit program. At present, contributions from current work-
ing-age individuals exceed payments to currently retired individuals,
leading to an increase in the notional reserves of the system's trust
fund. However, at current contribution rates and benefit levels this
reserve surplus will be dissipated following the retirement of the baby
boom generation. One proposal to reduce the system's insolvency is to
invest a portion of the trust fund's reserves in private equities (com-
mon stocks) rather than continue the current practice of investing all
of the reserves in U.S. Treasury bonds.28 The attraction of such a plan
is that because equities tend to earn a higher return than government
bonds, future trust fund balances, when invested in equities, would
have a higher expected value. It is argued that such a change in trust
fund investment policy would improve the solvency of the system.
Current contributions need not be raised as much nor do future ben-
efits have to be reduced as much in order to cope with the large num-
ber of future retirees. Essentially, this proposal would have the
government take advantage of the positive risk premium earned by
equity securities.
Smetters (1999) uses an option-pricing approach to demonstrate that
such a policy represents no free lunch. He considers policies in which
future payroll taxes and benefit levels are adjusted, depending on the
investment performance, and the resulting size, of the trust fund's future
balances. His analysis shows that once the value of the contingent gov-
ernment spending on benefits in excess of payroll taxes is taken into
account, such a policy change will not, in general, improve the solvency
of the social security system. In other words, when one properly values
the government's guarantee of a particular future tax and benefit struc-
ture, the supposed benefits of the trust fund earning an equity premium
are illusory. In large part, this stems from the greater risk introduced
into future fiscal policy by such a policy change, and accounting for
such risk neutralizes the higher expected net benefits.
Conclusion
Government pension guarantees play a key role in gaining public ac-
ceptance of pension reforms. To avoid political meddling and enhance
the efficiency of work and savings decisions, reformers often strengthen
the link between pension contributions and pension benefits and also
give individuals a choice about their pension fund investments. How-
ever, such a change can increase an individual's exposure to lower-
P:ENSION GUARANTEES 303
than-expected retirement benefits because of poor lifetime earnings or
poor returns earned on pension contributions. Pension guarantees pro-
vide a safety net by reducing an individual's exposure to various sources
of downside risk.
Recent advances in option-pricing theory have provided important
insights for valuing pension guarantees. Perhaps the most attractive
feature of this approach is the relatively few assumptions needed to
calculate guarantee values. This chapter illustrates how a Monte Carlo
simulation technique can be used to calculate the value of a wide vari-
ety of pension guarantees. Such cost calculations, while often com-
plex, are important for gauging the full expense of a pension reform.
T'hey also may be the basis for setting fair (risk-related) insurance pre-
miums that could enable the government to recoup its cost of provid-
ing guarantees. By setting fair premiums for guarantees, the government
may be able to mitigate the morail hazard incentives associated with
the guarantees.
Annex 13.1
The Relationship between the Expected Rates of Return on a Pension
Guarantee and Its Underlying Securities:
Suppose that the return on a risk-free security equals r, but that
investors require an expected rate of return on the underlying securi-
ties equal to r + Ocy,, where Oa, is defined as the securities' risk pre-
mium. This risk premium is defined to equal the product of the
securities' return volatility, a,, and the market price of risk, H. This
market price of risk reflects investors' preferences for bearing the type
of risk that derives from holding the securities.
The pension guarantee, being a contingent claim on the securities,
inherits the same risk as the unclerlying securities. However, in gen-
eral, its sensitivity to this risk differs from that of the underlying secu-
rities. Thus the contingent claim has a return volatility (standard
deviation of return) denoted by c,,, which is some quantity of risk that
differs from c5. But because it is affected by the same source of risk,
the expected return on the pension guarantee reflects the same market
price of this risk, 0. Thus the option-pricing approach implies that the
rate of return that investors expect to earn on providing the guarantee
equals r + Oar. This notion, that the expected returns on the contingent
claim and its underlying securities reflect the same market price of
risk, is also an implication of there being an absence of arbitrage op-
portunities if investors incur no transactions costs when trading in the
underlying securities.29 Therefore, the option-pricing approach is also
referred to as arbitrage-pricing theory.
304 GEORGE PENNACCHI
Computing the Cost of a Guarantee Using Risk-Neutral Valuation:
Let Go denote the present value (cost) of a government guarantee of
a minimum end-of-year pension fund value equal to X. Its value can
be computed as Go = [1/(1 + r)l E-[max(X - S1, 0)], where EJ[] is the
risk-neutral expectations operator that takes the expected value of its
argument subject to the condition that the expected rate of return on
all assets equals the risk-free interest rate, r.
In computing E-[max(X - S, 0)], a Monte Carlo technique can be
used to simulate the end-of-year values of S,. This involves using a
random number generator to compute many possible outcomes of S,
where the portfolio's expected rate of return is set equal to r and its
volatility of return is set equal to an estimate of its true standard devia-
tion, a,. For each realized value of Sl, the quantity max (X - Sl, 0) is
computed, and these quantities are averaged over all of the realiza-
tions. For a sufficiently large number of random simulations of Sl, the
average value of max (X - Sl, 0) will become arbitrarily close to the
theoretical value E-[max(X - S, 0)].
Notes
1. In most of the world, rising life expectancies and declining fertility
rates are increasing the ratio of retirees to working-age people. Along with
this rise in the old-age dependency ratio, the tendency for politicians to promise
generous future pension benefits also often undermines the long-term viabil-
ity of PAYC systems.
2. Examples include reforms enacted in the Czech Republic, France, Ger-
many, Japan, and the United States. Implementing a complete reform is diffi-
cult; current politicians may be unable to commit to long-term reforms,
because changes may be undone by future politicians. Such a lack of commit-
ment can reduce the incentives of competing interest groups to reach a com-
promise.
3. See Holtzman (2000) for a description of the disadvantages of PAYG
systems relative to multipillar pension schemes. A country with a PAYG sys-
tem may experience lower aggregate savings relative to one with a fully funded
retirement system. In addition, because a goal of most pension systems is to
redistribute income from higher-income to lower-income workers, in defined-
benefit systems this redistribution reduces the correspondence between an
individual's contributions and his or her pension benefits. Inefficiencies can
result if individuals work less or underreport income in jobs covered by the
pension system.
4. Mitchell and Zeldes (1996) and Holtzman (2000) discuss the issues sur-
rounding a country's choice of pension reform, including whether the country's
financial markets and financial institutions could support individual defined-
PENSION GUARANTEES 305
contribution accounts, the country's current fiscal situation, and how accu-
rnulated benefits to participants in the old pension system would be treated.
5. Goss (1999) defines solvency for a defined-benefit PAYG system as the
condition in which payroll tax revenues and pension reserves are sufficient to
pay all expected benefits over a given time period without the need for bor-
rowing by the pension administrator.
6. See the Nobel Prize address of Robert C. Merton (1998) for a discus-
sion of the many applications of contingent claims theory.
7. Such a situation is referred to as a complete market-that is, a market
in which it is possible to synthetically construct (replicate the payoffs of) a
particular security or contingent claim from other securities or contingent
claims.
8. Because security brokers often maintain custody of the securities owned
by their clients, short-sellers may borrow securities directly from these bro-
kers rather than the actual owners. For example, the government could bor-
row (short-sell) the same securities owned by the country's pension funds
that are held in the custody of security brokers. Such a transaction would
have no effect on the pension funds' operations. Alternatively, another gov-
ernment agency may be holding the appropriate securities for some other
purpose, say as part of a defined-contribution pension fund organized for
government employees. In this case, the government guarantee provider could
short-sell the securities held by the government employee pension fund, a
transaction requiring only an accounting entry to transfer the securities.
9. This would be accomplished by the government using the proceeds to
buy back some of its debt.
10. For example, if the security holdings of pension funds were $1 billion,
the government's hedge would involve a short position in less than $1 billion
of the same securities.
11. Mortality risk can be estimated using actuarial data. Often, individual
mortality risk is assumed to be fully diversifiable, but that may not be the
case if there are unpredictable changes in aggregate life expectancies.
12. The term risk-neutral describes a hypothetical economy in which all
investors are risk-neutral. In such a situation, the equilibrium rates of return
on all assets would equal the risk-free rate.
13. The proof of this result is beyond the scope of this chapter, but math-
ematical derivations can be found in Duffie (1996) and Kocic (1997).
14. Figure 13.3 gives the theoretical probability (density) of different real-
izations of S,. The frequencies generated by the Monte Carlo simulations
closely approximate those shown ir this graph for a sufficiently large num-
ber of simulations.
15. The fraction of all realization,s for which S, is less than $9 million can
be derived from Figure 13.3. It equals the area under the graph that is to the
left of $9 million divided by the total area under the graph.
16. Because of the simple nature of this particular guarantee, an explicit
formula can be derived by noting the similarity between this guarantee and a
306 GEORGE PENNACCHI
series of "forward start" options. In general, it is not possible to derive a
formula for all guarantees. Therefore, one could resort to numerical valua-
tion using the previously discussed Monte Carlo simulation technique.
17. For example, in Chile returns exceeding 2 percent more than the aver-
age return on all AFPs are channeled to reserves. For Argentina, returns in
excess of max(1.3R., R, + 0.02) become reserves. In Hungary, returns ex-
ceeding 40 percent more than the benchmark are placed in reserves.
18. According to Fischer (1998), Colombian AFPs pay a fixed annual
premium equal to 0.17 percent of their assets. Palacios and Rocha (1998)
report that the current premium paid by Hungarian pension funds equals 0.4
percent of total contributions. Polish pension funds pay a premium that var-
ies with the level of the guarantee fund, because, by law, the total value of the
guarantee fund cannot exceed 0.1 percent of total pension fund assets.
19. The only exception seems to be Peru. See Schwarz and Demirguc-
Kunt (1999) for a comparison of the social security systems of countries that
have made major reforms.
20. A "solidarity tax," used in Colombia, is another method. Queisser
(1995) reports that Colombian pension participants who earn more than
four times the minimum wage are taxed an additional 1 percent of their
income. The revenue from this solidarity tax is matched by government bud-
get transfers and used to subsidize the contributions of targeted poor groups
in order to extend coverage of the formal social security system.
21. In Chile, the minimum pension is set at the discretion of the govern-
ment, but it is typically equal to about 25 percent of the average wage.
Colombia's minimum pension equals its minimum wage, while in Mexico
the minimum pension was set equal to Mexico City's 1997 minimum wage
and then indexed to the Consumer Price Index. Argentina's minimum pen-
sion is approximately 40 percent of the average wage, and in Poland it is
approximately 33 percent, net of contributions.
22. Governments usually require that a new retiree's pension fund bal-
ance be used to purchase, from a regulated insurance company, an annuity
that pays at least the minimum pension.
23. In Chile, an individual's mandatory pension contribution equals 10
percent of his or her wage.
24. As discussed in Turner and Wantanabe (1995), and Smalhout (1996),
a worker who reaches retirement with a pension balance that is slightly above
or at the price of a minimum pension annuity will have an incentive not to
purchase an annuity but to choose the scheduled withdrawal option. By choos-
ing this scheduled withdrawal, the worker will receive free longevity insur-
ance at the government's expense. Should he or she live longer than expected,
the government provides a minimum pension. If, instead, the worker does
not live as long as expected, the worker's heirs will inherit the balance of the
pension account. Thus in some countries the worker receives a government
subsidy that would not have materialized if the worker had purchased an
annuity at retirement. For someone reaching retirement with moderate to
PENSION GUARANTEES 307
small pension savings-that is, the individual most likely to require mini-
mum pension assistance-it is more realistic to assume a scheduled with-
drawal of pension funds.
25. The first pillar in Poland is a PAYG notional account in which contri-
butions grow at a specified interest rate. The second pillar is a privately
managed, defined-contribution account.
26. In the United States, however., the number of employees covered by
defined-contribution plans recently surpassed the number of employees cov-
ered by defined-benefit plans.
27. As noted in Cholon, G6ra and Rutkowski (1999), this type of pension
fund guarantee is quite similar to government guarantees of annuities offered
by private insurance companies. Thus these valuation models of guarantees
for corporate-sponsored, defined-benefit pensions can be applied to annuity
guarantees such as those provided by the government of Poland.
28. For example, in his January 19, 1999, State of the Union speech,
President Bill Clinton proposed to invest some of the trust fund reserves in
IJ.S. common stocks. One proposal analyzed by the U.S. Advisory Council
on Social Security would also call for investing approximately 40 percent of
the trust fund in U.S. equities.
29. The intuition for why an absence of arbitrage implies that the ex-
pected return on the contingent claim reflects the same price of risk as its
underlying security is as follows. Suppose an investor provides the guarantee
and hedges by short-selling the underlying security. When there are no trans-
actions costs, this hedge involves short-selling Ca,IScs, shares of the underly-
ing security for every guarantee offered, where C is the current value of the
guarantee and S is the current share price of the underlying security. Because
they are affected by the same source of risk, when there is an unexpected
movement in the per share value of i:he underlying security of ScF,, the value
of the contingent claim changes by Ccs. Thus this hedge provides a riskless
position for the provider of the guarantee, because when the guarantee's
value increases by Cck it is negated by a change in the short position in the
security of -(Cao1Scs,)ScY = -Ccs. Now, because the investor receives $C when
agreeing to take over the government's guarantee and obtains $(Cas/Scso)S
from short-selling the underlying security, the sum of these proceeds is in-
vested in a riskless security. In the a.bsence of arbitrage, the investor's com-
bined portfolio must earn the risk-free rate. Adding together the return on
this riskless security of [C + (Ccat/S,:;)S]r plus an expected return from the
short position of -(Cca,Sc)S(r - 0q) equals C(r + Oo), the expected return
on the contingent claim, which reflects the same market price of risk.
References
The word processed describes inforrnally reproduced works that may not be
commonly available through libraries.
308 GEORGE PENNACCHI
Bernardo, Antonio E., and Olivier Ledoit. 2000. "Gain, Loss, and Asset Pric-
ing." Journal of Political Economy 108: 144-72.
Black, Fischer, and Myron S. Scholes. 1973. "The Pricing of Options and
Corporate Liabilities." Joturnal of Political Economy 81: 637-54.
Brixi, Hana Polackova, Anita Papp, and Allen Schick. 1999. "Fiscal Risks
and the Quality of Fiscal Adjustment in Hungary." Policy Research Working
Paper 2176. World Bank, Washington, D.C.
Cholon, Agnieszka, Marek G6ra, and Michal Rutkowski. 1999. "Shaping
Pension Reform in Poland: Security through Diversity." World Bank Pen-
sion Reform Primer Working Paper. World Bank, Washington, D.C.
Cochrane, John H, and Jesus Sai-Requejo. 2000. "Beyond Arbitrage: Good-
Deal Asset Price Bounds in Incomplete Markets." Journal of Political
Economy 108: 79-119.
Diamond, Peter, and Salvador Vald6s-Prieto. 1994. "Social Security Reforms."
In Barry P. Boxworth, Rudiger Dornbusch, and Raul Laban, eds., The
Chilean Economy: Policy Lessons and Challenges. Washington, D.C.:
Brookings Institution.
Duffie, Darrell. 1996. Dynamic Asset Pricing Theory. Princeton: Princeton
University Press.
Fischer, Klaus P. 1998. "A Discrete Martingale Model of Pension Fund Guar-
antees in Colombia: Pricing and Market Effects." Working Paper. Universite
Laval, Quebec.
Goss, Stephen C. 1999. "Measuring Solvency in the Social Security System."
In Olivia S. Mitchell, Robert J. Myers, and Howard Young, eds., Pros-
pects for Social Security Reform. Philadelphia: University of Pennsylvania
Press.
Holtzman, Robert. 2000. "The World Bank Approach to Pension Reform."
Processed.
Kocic, Aleksandar. 1997. "Numeraire Invariance and Generalized Risk Neu-
tral Valuation." Advances in Futuires and Options Research 9: 157-73.
Lewis, Christopher M., and George G. Pennacchi. 1999. "Valuing Insurance
for Defined-Benefit Pension Plans." Advances in Futures and Options
Research (10): 135-67.
Marcus, Alan J. 1987. "Corporate Pension Policy and the Value of PBGC
Insurance." In Zvi Bodie, John Shoven, and David Wise, eds., Issues in
Pension Economics. Chicago: University of Chicago Press.
Merton, Robert C. 1973. "Theory of Rational Option Pricing." Bell Journal
of Economics and Management Science 4: 141-83.
- 1990. Continuous-Timne Finance. Cambridge, Mass.: Blackwell Pub-
lishers.
- 1998. "Applications of Option-Pricing Theory: Twenty-Five Years
Later." American Economic Review 88: 323-49.
Mitchell, Olivia S. 1996. "Social Security Reform in Uruguay: An Economic
Assessment." Pension Research Council Working Paper 96-20. Wharton
School, University of Pennsylvania.
PENSION GUARANTEES 309
Mitchell, Olivia S., and Flavio Ataliba Barreto. 1997. "After Chile, What?
Second-Round Pension Reforms in Latin America." Revista de Analisis
Econ6mico 12: 3-36.
Mitchell, Olivia S., and Stephen P. Zelcles. 1996. "Social Security Privatization:
A Structure for Analysis." American Economic Review 86: 363-67.
Palacios, Robert, and Roberto Rocha. 1998. "The Hungarian Pension Sys-
tem in Transition." In Lajos Bokros and Jean-Jacques Dethier, eds., Public
Finance Reform during the Transition: The Experience of Hungary. Wash-
ington, D.C.: World Bank.
Pennacchi, George G. 1999a. "The Value of Guarantees on Pension Fund
Returns." Journal of Risk and Insurance 66: 219-37.
--. 1999b. "The Effects of Setting Deposit Insurance Premiums to Tar-
get Insurance Fund Reserves." Journal of Financial Services Research 16
(2/3): 153-80.
lPennacchi, George G., and Christopher M. Lewis. 1994. "The Value of Pen-
sion Benefit Guaranty Corporation Insurance." Journal of Money, Credit
and Banking 26: 735-53.
Queisser, Monika. 1995. "Chile and. Beyond: The Second-Generation Pen-
sion Reforms in Latin America." International Social Security Review 48:
23-39.
Schwarz, Anita M., and Asli Demirguc-Kunt. 1999. "Taking Stock of Pen-
sion Reforms around the World." Social Protection Discussion Paper No.
SP9917. World Bank, Washington, D.C.
Smalhout, James H. 1996. The Uncertain Retirement. Chicago: Irwin Pub-
lishing.
Smetters, Kent. 1999. "The Equivalence between State Contingent Tax Policy
and Options: An Application to Investing the Social Security Trust Fund in
Equities." Working Paper. Whartcon School, University of Pennsylvania.
Turner, John A., and Noriyasu Wantanabe. 1995. Private Pension Policies in
Industrialized Countries: A Comnparative Analysis. Kalamazoo: W. E.
Upjohn Institute for Employment Research.
Zarita, Salvador. 1994. "Minimum P'ension Insurance in the Chilean Pension
System." Revista de Analisis Econ6mico (9): 105-26.
CHAPTER 14
Measuring and Managing
Government Contingent
Liabilities in the Banking Sector
Stij n C]aessens *
University of Amsterdam
Daniela Klingebiel
World Bank
t.jOVERNMENT CONTINGENT LIABILITIES in the banking sector arise from
explicit government guarantees on banking system liabilities through
a public deposit insurance scheme. Public contingent liabilities also
can emerge when governments try to limit the loss of confidence in the
financial system during periods of financial turbulence by taking over
or guaranteeing liabilities that are not formally protected. This "moral'
obligation of the government to protect depositors and other creditors
reflects public expectations, but it also is influenced by pressures from
interest groups, with governments bailing out unprotected creditors
instead of imposing losses on them.
Recent experience suggests that the fiscal costs of a banking crisis can
be substantial and can, in extreme cases, threaten fiscal stability. As
such, an analysis of the causes of and means of limiting these costs has
to be an essential component of efforts to limit and manage government
,contingent liabilities. This chapter analyzes two questions. First, can
contingent liabilities arising from the banking system be measured accu-
rately? Second, and more important, what can governments do to con-
trol fiscal liabilities arising from the banking sector? The chapter finds
'The author was with the World Bank when he wrote this article; he is currently with the
University of Amsterdam.
311
312 CLAESSENS AND KLINGEBIEL
that existing approaches to measuring contingent liabilities in the bank-
ing sector, while representing a good start, are often inadequate. Most
are based on static models that focus on the typical causes of a banking
crisis, but they fail to recognize the dynamic dimensions of managing
contingent liabilities arising from banking distress. The analysis sug-
gests that a substantial portion of the fiscal costs arises from improper
crisis management. The chapter also outlines how the probability of
banking crises can be reduced through a variety of instruments.
The chapter does not analyze the government's direct or indirect
role in the financial system in noncrisis times, a role that may influ-
ence the fiscal costs and risks of a financial crisis. This role of the
government has at least three aspects. The first concerns the proper
role of the government in setting the framework for a country's finan-
cial development and in providing deposit insurance and other guar-
antees. The second aspect relates to the government's ownership of
financial institutions. And the third aspect is the short- and long-run
costs and benefits of any directed credit programs or other ways in
which the government influences, through the financial sector, the al-
location of resources in an economy.
The first aspect is a very broad topic (for a recent review, see World
Bank 2001a). The consensus is that the general role of the government
should be in setting up the institutional, legal, and accounting infra-
structure to improve financial development. Proper regulation and
supervision are key government roles and crucial to reaping the gains
from the deepening of the financial sector. Whether there is a role for
deposit insurance and guarantees in enhancing financial sector devel-
opment is less clear. Properly used, deposit insurance can assist in im-
proving financial stability, and credit guarantees may help institutions
attract loans at more reasonable rates. But in many countries a high
fiscal cost and an increased risk of financial crises have been associ-
ated with deposit insurance and government guarantees (see Demirguc-
Kunt and Detragiache 2000; World Bank 2001a). These costs have
arisen because the private sector has reduced its monitoring of finan-
cial institutions and moral hazard has increased. When governments
are unable to monitor financial institutions effectively, implicit and
explicit guarantees can have large costs and retard the longer-term
development of a financial sector. Poor financial development can in
turn expose the government to additional fiscal risk.
On the second aspect, the general experience with government own-
ership in the financial sector has been very poor and quite costly in
terms of financial sector development. Government-owned financial
institutions that borrow for the purpose of lending hurt private sector
efforts to do the same and thereby impede financial development. Em-
pirically, government ownership of banks has been found to retard fi-
nancial sector development and increase the risk of financial crises (for
a review, see World Bank 2001a and 2001b). In particular, Barth, Caprio,
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 313
and Levine (2000) show that for a sample of 66 countries greater state
ownership of banks tends to be associated with higher interest rate
spreads, less private credit, and less activity on the stock exchange and
less nonbank credit, after taking account of real gross domestic product
(GDP) per capita, corruption, expropriation risk, bureaucratic efficiency,
and the law and order tradition of the country, all of which could influ-
ence financial development. This evidence is confirmed for a larger sample
of countries in Barth, Caprio, and Levine (2001). A separate study, with
data from private industry sources covering the 10 largest commercial
and development banks for each of 92 countries for 1970 and 1995,
convincingly highlights the causal link (La Porta, Lopez de Silanes, and
Shleifer 2000). It finds that greater state ownership of banks in 1970
was associated with lesser financial sector development, lower growth,
and lower productivity, and that these effects were larger at lower levels
of per capita income, with less inil:ial financial sector development and
w,ith weaker property rights protection.
The evidence on the third aspect, the role of directed credit and
government involvement in the allocation of resources through the
financial sector, is more mixed (for a review, see Caprio and Demirgiiu-
](unt 1997). In East Asia throughout the 1980s, for example, directed
credit programs were generally considered successful (see World Bank
[993). In weak institutional settings, however, attempts to reach
underserviced groups often miss their targets, are captured by special-
interest groups, and have large fiscal costs. Thus an immediate cost of
directed credit programs may be that funds are directed to uses that
might not arise if judgments were based on rational economic criteria.
Directed credit programs also can have an impact on financial sector
development. The U.S. government loan programs to agriculture, for
example, have significantly impeded the development of private sector
agriculture finance. Loan guaranitees reduce the incentive to develop
private credit evaluation systems, which ultimately reduces the
allocative efficiency of the financial system. Even if the goals are laud-
able, such as providing more financial support to small-scale agricul-
ture, better housing for lower-income families, more education benefits
(student loans), and help to small businesses, the long-term costs to
financial sector development may well offset the benefits. The empiri-
cal analysis by Barth, Caprio, and Levine (2001) suggests that, in gen-
eral, fewer regulatory restrictions and limits on commercial bank
activities produce more diverse income streams, encourage greater
competition and therefore efficiency, and allow banks to benefit from
economies of scale and scope. Directed credit programs, as forms of
government restrictions, would thus lead to exactly the opposite re-
sults: less financial sector development, greater risk of banking crisis,
high ultimate fiscal costs, and lower economic growth.
The rest of this chapter is divided into six sections. The first section
details why banking crises matter by reviewing cross-country experience.
314 CLAESSENS AND KLINGEBIEL
The second section, which reviews existing banking crisis prediction
models, is followed by one that assesses currently available approaches
to measuring contingent liabilities arising from banking system distress.
The fourth section analyzes instruments and government tools for con-
trolling fiscal contingencies and the costs of a banking crisis, focusing
on the initial phases of a financial crisis. The fifth section presents policy
tools to prevent or at least reduce the probability of future systemic
banking crises. It is followed by the chapter's conclusions.
Bank Insolvency and Government Liabilities
In recent decades, many developing and developed countries have ex-
perienced banking crises requiring major-and expensive-overhauls
of their banking systems. By one count, 112 episodes of a systemic
banking crisis have occurred in 93 countries since the late 1970s (Caprio
and Klingebiel 1999). Banking crises have been very disruptive for
two reasons: they tend to be expensive in terms of the fiscal costs that
governments, and ultimately the taxpayers, shoulder, and they also
can be costly in terms of forgone economic output. By one measure,
banking crises have been the most important source of government
contingent liabilities (see Kharas and Mishra 2001).
The prevalence of banking system failures has been at least as great
in developing and transition countries as in the industrial world. How-
ever, financial crises have proved substantially more expensive in de-
veloping countries (see Figure 14.1). In a survey of 40 banking crises,
Honohan and Klingebiel (2000) found that governments of develop-
ing countries spent on average 14.3 percent of GDP to clean up their
financial systems, double the amount that governments of industrial
countries paid out. Some of the banking crises in developing countries
were very expensive. Of those for which there are data, Argentina's
crisis in the early 1980s proved to be the most expensive restructuring
exercise-about 55 percent of GDP-followed by Indonesia's with a
price tag to date of about 50 percent of GDP and Chile's at 41.3 per-
cent. Among industrialized countries, Japan's long and drawn-out bank-
ing crisis has been the costliest; authorities already have spent about
20 percent of GDP to restructure the system.
In the past two decades, governments and, thus ultimately tax-
payers, have largely shouldered the costs of restructuring banking
systems. When governments have not been able to mobilize suffi-
cient resources-through additional taxes, spending cuts, or borrow-
ing-to service the additional debt burden, they have resorted to
inflationary financing, passing the losses on to those holding cur-
rency, with adverse consequences for growth and income distribu-
tion. Although the fiscal costs estimates have often been staggering
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 315
Figure 14.1. Fiscal Cost of Banking Crises
United States 1981-91 m
Spain 1977-85
Finland 1991-94
Brazil 1994-96
Japan 1992-present
Venezuela 1994-97
C6te d'lvoire 1988-91
Chile 1981-83
Iidonesia 1997-present
Argentina 1980-82
0 10 20 30 40 50 60
Fiscal costs as a percentage of GDP
Source: Honohan and Klingebiel (2000).
(see Figure 14.1), the true economic costs of banking crises have been
even higher than these estimates suggest. Fiscal costs do not include,
for example, the costs borne in some cases by depositors and other
creditors of failed banks. They also do not take into account that
depositors and borrowers have been often "taxed" in the form of
widened spreads to make up for bad loans left on banks' balance
sheets. In addition, costs do not reflect the distortions introduced by
granting borrowers some monopoly privilege or other means to im-
prove their profits and thereby repay their loans.
The Causes of Banking Crises and Predicting Crises
T he large fiscal costs of banking crises raise the question of whether
these costs can be avoided, or at least reduced, through specific policy
measures. A starting point for answering this question is to identify
the exact causes of banking crises. The literature on financial crises is
growing (see Kaminsky and Reinhiart 1999 and Goldstein, Kaminsky,
and Reinhart 2000 for a review of banking crises prediction models).
Demirguc-Kunt and Detriagiache (1998, 2000) employ an economet-
ric model to explain the incidence of banking crises in a large sample
316 CLAESSENS AND KLINGEBIEL
of countries. They find that macroeconomic variables-GDP growth;
changes in trade, the real interest rate, inflation, growth of credit, fis-
cal surplus, and reserves cover for money stock; and recent financial
liberalization-are important factors in explaining the incidence of
financial crises. As such, their model provides important information
about the factors contributing to and triggering banking crises. Spe-
cifically, they find evidence that large current account and fiscal defi-
cits, rapid credit growth, and financial liberalization in a weak
institutional environment increase the probability of a crisis, confirm-
ing the earlier results of Caprio and Klingebiel (1997). Other, more
recent work has highlighted the role of microeconomic and bank-spe-
cific factors (Gonzales-Hermosillo 1999; Laeven 1999; Bongini,
Claessens, and Ferri 2001).
Most econometric models, however, have not been able to predict
accurately the timing of banking crises. Models typically predict too
many crises-type I error-if the threshold probability is kept low, or
they miss many crisis episodes-type 11 error-when the threshold prob-
ability is raised. For example, Demirguc-Kunt and Detragiache (2000)
reveal that their model would have missed the recent banking crisis in
several of the affected East Asian countries. Indeed, as late as May
1997 the model would not have signaled any problems-for example,
it predicted a probability of a banking crisis in Thailand of only 3.3
percent. Overall, they conclude that their model "found the overall
image. .. Ito have] been a rather reassuring one."
A major limitation of these models as a forecasting tool is that
they rely heavily on macrofinancial indicators such as interest rates
and exchange rates, whose sudden spikes are hard to forecast. With-
out these variables-which themselves are often the consequences of
financial distress rather than the causes-these models provide little
advance warning. Other studies have included longer-term institu-
tional and policy factors that can predispose a country to crises (Hardy
and Pazarbasioglu 1998; Keefer 1999). But these factors are often
difficult to quantify and therefore do not decrease type I or type 11
errors significantly.
Although these models may be of limited use as forecasting tools,
they consistently point toward areas of policy reform that can be
expected to reduce the incidence, and possibly the costs, of financial
crisis. The results suggest that governments that pursue relatively
sound macroeconomic policies-characterized by low budget defi-
cits, low inflation, and an exchange rate policy that avoids the real
appreciation of the exchange rate-can significantly reduce the inci-
dence and costs of banking crises. Preventing rapid credit growth
and phasing in financial liberalization properly in a weak institu-
tional environment are two other areas of policy reform that follow
from most empirical work.
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 317
Approaches to Measuring Contingent
Liabilities in the Banking Sector
The banking crises prediction models reviewed earlier in this chapter
do not aim to quantify the extent of potential government contingen-
cies arising from a crisis, but just their incidence and timing. This
limitation may stem from the fact that market participants are typi-
cally more interested in the occurrence of a crisis than in its final
costs. Also, contingent liabilities arising from banking crises are dif-
ficult to measure. Information on financial institutions' balance sheets
and overall risk exposure is often of poor quality and limited value in
developing countries. The markei: values of banks' assets can be vola-
tile in any environment, but even more so in a financial crisis when
they are affected by large changes in interest rates and exchange rates,
the financial condition of the corporate sector, and economic growth,
among other things. Nevertheless, investigators have attempted to
quantify the fiscal exposure arising from the banking system. Three
major approaches to measuring contingent liabilities in the banking
sector can be distinguished: (a) the gross assets-at-risk approach used
by Standard and Poor's Sovereign Ratings Service (1999); (b) the value-
at-risk (VAR) approach of Blejer and Schumacher (1998), which is
also outlined in Chapter 16 by Mario Blejer and Liliana Schumacher
in this volume; and (c) the accounting approach to bank losses by
Honohan (1999).
Standard and Poor's (S&P) estimates the gross banking assets at
risk from a major economic downturn. The framework relies on a
quality rating of countries based on perceived vulnerability to asset
quality pressure during a recession. The rating, which is subjective, is
provided by S&P staffers who also conduct ratings of countries. The
model then applies a factor-between zero and one, depending on the
rating-to the total assets of the financial system. The figure obtained
through this exercise is called the "gross banking assets at risk" from a
major economic downturn. The risk of a downturn occurring in the
forecast horizon is then assessed in a separate exercise. As the label
"gross banking assets at risk" suggests, the S&P approach tries to es-
timate the gross exposure of the economy to banking system distress.
As such, it does not measure the direct fiscal exposure of the govern-
ment if a banking crisis actually occurs. The method provides an up-
per limit to contingent fiscal liabilities and as such is closer to the
value-at-risk approach, a concept now commonly used in commercial
banks to assess risks.
Blejer and Schumacher (1998) employ a value-at-risk approach to
appraising a central bank's solvency. They examine the factors from
traditional central bank operations and off-balance sheet positions that
may affect solvency, including Foreign exchange and financial sector
318 CLAESSENS AND KLINGEBIEL
guarantees. While a central bank cannot fail commercially, it may be-
have equivalently if it has to forsake a commitment to an announced
nominal regime (such as an exchange or inflation target) because its
solvency has been threatened. If indeed the probability of a central
bank abandoning its commitments relates to its vulnerability to sol-
vency losses, measures such as VAR could be used as forward-looking
indicators of credibility crises. This model might help in identifying
the risk of a financial crisis; it does not, however, provide a methodol-
ogy for estimating the fiscal costs arising from an actual crisis.
Honohan (1999) applies a VAR approach to the whole banking
sector. His model assesses the probability and the extent of fiscal con-
tingencies arising from the banking sector because of changes in the
macroenvironment. Using accounting data on the size and composi-
tion of the balance sheets of the banks, the model simulates several
short-term risk factors that can adversely affect bank balance sheets.
These risk factors are then used to estimate the amount of contingent
fiscal liabilities. The approach is as follows. First, the model specifies
and quantifies possible economic shocks, such as changes in the ex-
change rate, changes in property market values, changes in the terms
of trade, or a general economic downturn. Then, using banks' balance
sheet information, it classifies various balance sheet items according
to whether and to what extent they would be adversely affected by
these economic shocks (see Table 14.1).
Table 14.1 Capturing the Effects of Macroshocks
on Individual Bank Balance Sheets
Primary effects Secondary effects
Exchange Foreign exchange (FX)- Other FX-denominated
rate shocks denominated loans assets (net, affected
(affected both directly mainly by currency
by currency translation translation)
and indirectly by changed
loan loss experience)
Property Loans to real estate develop- Other loans secured on
market pers (net worth of borrow- real estate (value of
shocks ers directly affected) collateral declines)
General Loan portfolio (net worth Marketable investments
economic of borrowers directly (market values affected)
shocks affected)
Problematic Government-related loans Other loans (evergreening,
accounting (unrecognized collection overly optimistic col-
problems due to political lateral valuation, and
influence) so forth)
Source: Honohan (1999).
GOVERNMENT CONTINGENT LIA131LITIES IN BANKING 319
The model then applies a multiplier to each group of balance sheet
items based on the size of the assumed shock to the affected balance
sheet category. In the next step, the capital positions of the various
banks are estimated after the effects of each shock and the systemwide
capital deficiency are calculated. The latter is assumed to be met by
the government. The same exercise is repeated for different sizes of
shocks, such as different changes in the exchange rate, with probabili-
ties assigned to each discrete macroshock. Finally, a matrix is con-
structed for a given macro shock, indicating the potential size of the
losses (Table 14.2). For example, in the hypothetical country the po-
tential fiscal liabilities could amount to US$6.9 million if the exchange
rate change is greater than 40 percent and loan losses greater than 50
percent. As Honohan notes, this methodology provides an estimate of
maximum fiscal liability, because it assumes that banks will be made
whole by the fiscal authority. The model would, however, allow for
loss absorption by other claimants.
VAR-type approaches are commonly used by financial institutions
and are being introduced in estimating government contingent liabili-
ties as well (see Chapters 4 by Sundaresan and 5 by Ramaswamy in
this volume). Applying VAR-type approaches to the banking sector
can be a good starting point, but it has limitations when estimating
potential banking losses and the resulting fiscal costs. VAR-type ap-
proaches are essentially backwarcl looking and static, and the results
clerived from them can be problematic. The fiscal costs of banking
crises are determined not only by the size of the macroshock, the ini-
tial conditions, and the known risk exposures in the banking sector,
but also by government policies in managing the crisis. The cost of a
banking system failure is a function of the policies adopted.
Table 14.2 Potential Fiscal Liability Arising from
a Depreciation and Loan Losses
V=
'US$ million 0% 14.3% 25% 40% 50%
0.1 0 0 0 0 0
0.2 0 0 0 0 0
0.3 0 0 0 0 1.0
?(v)= 0.4 0 0 0.6 2.4 5.8
0.5 0.5 1.2 2.7 6.9 14.7
0.6 2.6 4.5 7.5 16.1 26.9
0.7 6.6 10.2 14.7 28.9 43.2
Note: v is percentage depreciation; ?(v) is percentage loan losses. The entries provide
the total loss of the joint events.
Source: Honohan (1999).
320 CLAESSENS AND KLINGEBIEL
One way to identify the most important policy measures for con-
taining fiscal costs is to review past crises. In a recent paper, Honohan
and Klingebiel (2000) analyzed the determinants of the total fiscal
costs for 40 banking crises during the 1980s and 1990s. They found
that government policies in the initial containment phase of the crisis
are more important explanatory factors of the final fiscal costs, as
measured ex post, than the initial macro- and microvulnerabilities.
Indeed, according to Honohan and Klingebiel, fiscal costs are system-
atically associated with a set of crisis management strategies. They
found that the explanatory variables employed-mainly policy varn-
ables-can explain 60-80 percent of cross-country variation in fiscal
costs. Their empirical findings reveal that unlimited deposit guaran-
tees, open-ended liquidity support, repeated recapitalizations, debtor
bailouts, and regulatory forbearance add significantly and sizably to
costs. Using the regression results to simulate the effects of these poli-
cies, they learned that if countries had not extended all these policies,
the average fiscal costs in their sample could have been limited to
about 1 percent of GDP-that is, a little more than one-tenth of what
was actually experienced. On the other hand, policy could have been
worse: had countries engaged in all of the above policies, fiscal costs in
excess of 60 percent of GDP would have been the result (Figure 14. 2).
As Table 14.3 indicates, liquidity support and forbearance measures
are the costliest ones. Even if deposit guarantees, forbearance, and
repeated recapitalizations are employed, not extending liquidity sup-
port could reduce fiscal costs by almost two-thirds.
These results suggest that the contingent liabilities arising from a
banking crisis are importantly influenced by government policies rather
Figure 14.2. Size Effect of Crisis Resolution Policies
Strict policies
Actual policies
Lax policies
0 20 40 60 80
Percent of GDP
Note: Graph shows the predicted fiscal cost of a hanking crisis if countries adopt
different kinds of crisis resolution policies.
Source: Honohan and Klingebiel (2000).
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 321
Table 14.3 Estimated Individual Impacts of Policy Variables
on Fiscal Costs
(percent of base case fiscal cost)
Estimated saving
(percent of base case fiscal cost)
Best policy by switching one policy tool from:
employed in Strict value Lax value
percentage
of cases A B C
I.IQSUP 0.421 165.1 50.8 -62.3
F:ORB-A 0.763 120.6 82.9 -54.6
FORB-B 0.158 118.6 13.1 -54.3
REPCAP 0.763 112.1 77.5 -52.9
GUAR 0.447 55.7 21.9 -35.8
I'DRP 0.789 50.7 38.2 -33.6
Note: LIQSUP indicates whether unlimited liquidity support was provided to banks.
Two measures of forbearance are: FORB-A, which captures whether insolvent banks were
permitted to continue functioning, and FORB-B, which denotes whether other bank pru-
clential regulations were suspended or not fully applied. REPCAP indicates where banks
were repeatedly recapitalized. GUAR denotes that either governments issued an explicit
guarantee on all bank liabilities or market participants were implicitly protected from any
losses if public banks' market share exceeded 75 percent. PDRP indicates that govern-
inents implemented across-the-board public debt relief programs.
Columns A and B show the effect of a pc,licy relaxation-that is, switching the value of
one policy variable from strict to lax-on the predicted cost of the crisis. Column A as-
sumes all other variables are held at strict values; Column B sets the other policy dummies
in the regression at their sample mean (fractional) values. Column C shows the effect of a
policy tightening, assuming all other variables are left at their lax values. Thus, for ex-
ample, a policy of unlimited liquidity support would increase the average crisis cost by 65
percent of the value that would occur if all variables were held at their strict values.
Source: Honohan and Klingebiel (2000).
than only by ex ante weaknesses. Furthermore, the results indicate
that the fiscal outlays incurred in resolving banking system distress
can to a great extent be attributed to the measures adopted by the
government during the first two years of the crisis. They also suggest
that governments can control fiscal costs by adequately managing the
crisis resolution process. The specific measures most likely to be suc-
cessful are analyzed in the next section.
Approaches to Controlling
the Costs of a Banking Crisis
Cross-country experiences point to several clear processes and prin-
ciples, which if adopted during rhe containment phase of a banking
crisis, will control the fiscal costs of the crisis (see, for example, Sheng
322 CLAESSENS AND KLINGEBIEL
1996 and Claessens 1999). The empirical results highlight that an ad-
equate government strategy should, during the early or containment
phase of a crisis, include the following three main components:
o Governments should not provide liquidity support to weak fi-
nancial institutions to delay problem recognition but rather should
address the problem early on and develop a strategy that deals with
the problem comprehensively and credibly.
o Governments should avoid issuing a blanket government guaran-
tee that distorts incentives for both managers and shareholders of banks
to behave prudently and for their creditors to monitor the institution
closely.
o Governments should not employ forbearance policies-that is,
policies involving deviations from prudential standards aimed at al-
lowing institutions to recapitalize from increased earnings, but only in
certain limited circumstances.
Cross-country evidence also suggests a set of policies that are im-
portant both to control fiscal outlays in the short run and to strengthen
over the medium term the incentive framework in which market par-
ticipants operate.' In particular, the following medium-term policies
are important:
o Government should not bail out banks, nor is it necessary to bail
out depositors in all cases.
o Governments should complement bank restructuring with an
adequate framework for corporate restructuring.
o A proper incentive framework for restructuring should be adopted.
The Short-term Containment Phase
Liquidity Support. To contain fiscal costs arising from a banking
crisis, government should not provide liquidity on an ongoing basis
until it is satisfied that the bank is viable and oversight is adequate.
Governments have often used liquidity support to delay crisis recogni-
tion and avoid intervening in de facto insolvent institutions. But this
strategy is doomed to failure. Managerial and shareholder incentives
quickly shift for a financial institution when it becomes insolvent:
managers have no incentive to run the institution on a viable basis and
their actions often speedily drain away resources-including liquidity
support from the central bank. In 1997 this situation was disastrously
demonstrated in Thailand, where large-scale liquidity support-10
percent of GDP-was extended to finance companies that turned out
to be black holes of insolvency.
Government Guarantees. Governments should not resort to the quick
fix of giving guarantees to depositors and creditors to stem the loss of
GOVERNMENT CONTINGENT LIA]BILITIES IN BANKING 323
confidence. Such guarantees limit their maneuverability in allocating
losses in the future. More important, if guarantees are credible, they
reduce (large) creditors' incentives to monitor financial institutions
and therefore increase the importance of early intervention in weak
institutions to stop managers and shareholders from engaging in the
"gamble for resurrection." As a result, the costs of banking crises in-
crease, especially in weak regulatory environments.2
Forbearance. Many countries have adopted a policy of forbearance
to allow banks to recapitalize from retained earnings. Explicit for-
bearance policies include lower capital adequacy requirements, more
lenient tax treatments, tax breaks, lower reserve or provisioning re-
quirements, more lenient accounting standards and practices, guaran-
tees on bank assets or liabilities (or both), and lower interest rates on
liquidity support. Forbearance also may be implicit-that is, authori-
ties turn a blind eye to violations of laws, standards, and regulations
by either individual banks or the banking system. Cross-country expe-
rience shows that forbearance strategies have rarely been successful.
As the above empirical evidence shows, forbearance often increases
the costs of financial crises, especially in weak microenvironments when
authorities are unable or unwilling to stop the transfer of resources out
of financial institutions that long ago turned insolvent.
Because of dangers of misuse, some experts propose that forbear-
ance be forsworn completely. One could argue that forbearance has a
useful role in special circumstances-in cases of a sharp external shock
faced by otherwise sound financial institutions and a good institu-
tional framework (including proper regulation and supervision and
monitoring by the market). If used, forbearance should be transparent
and accompanied by stringent standards, realistic targets and imple-
mentation periods, and strict enforcement with prompt and transpar-
ent action in case of violation. It should not be adopted in situations
with existing micro weaknesses or where owners are wholly or partly
responsible for the bank's predicament.
The Medium-term Rehabilitation and Restructuring Phase
Allocating Losses and Use of Public Resources. When a financial
institution is insolvent, the claims of shareholders and subordinated
debtholders should be written down entirely before public money is
forthcoming. Restructuring can strengthen financial discipline by allo-
cating losses not only to existing shareholders, but also (at least some
losses) to creditors and depositors who should have monitored the
bank. Allocating losses to creditors or depositors will not necessarily
lead to a run on the bank or end in contraction of aggregate money
and credit, and output. In some past crises, governments imposed losses
on depositors, and there were little (or no) adverse macroeconomic
324 CLAESSENS AND KLINGEBIEL
consequences or flight to currency (Baer and Klingebiel 1995). Eco-
nomic recovery was rapid, and financial intermediation, including
household deposits, was restored within a short time (Figure 14.3).
Financial discipline was further strengthened when management,
deemed to be part of the problem, was changed as well, and banks
were operationally restructured.
This does not mean that government support should always be
withheld. Rather, it can imply the opposite, that countries need (or
must be perceived) to have large enough public war chests to deal
with the large costs. Often a government's instinctive reaction to a
crisis is to allocate too few public resources. Unsure of the amount of
help available, financial institutions tend to hide the true extent of
their problems, and existing and potential shareholders will not put
up new capital. More generally, insufficient government support un-
dermines the confidence of depositors and investors. Yet public sec-
tor capital injections should not be a bailout of existing shareholders.
Rather, the aim is to allocate losses transparently and minimize costs
to the taxpayers, while preserving incentives for the infusion of new
private capital.
If public money is provided, assisted banks should be required to
draw up an acceptable business plan, verified by third parties, that
covers capital restructuring and operational restructuring to reduce
costs and improve profit prospects without taking on additional risks.
Adequate safeguards are needed to ensure that banks do not subse-
quently become undercapitalized, and they should include strict and
regular monitoring and supervision, on-site and off-site. In the past,
many countries failed to follow these principles. They resolved their
financial crises in part through partial or full public bailouts, which
reinforced the perception of an implicit government guarantee on de-
posits and other bank liabilities to the detriment of market discipline.
In some cases, bank management, at least partly responsible for the
problems, was not even changed for the course of the restructuring,
which further undermined incentives for prudent behavior. The linger-
ing effects of such policies contributed to the 1997 banking crisis in
East Asia (Alba, Hernandez, and Klingebiel 1999).
Corporate Restructuring. A large portion of nonperforming loans is
often just a reflection of overleveraged borrowers and corporate distress.
Lowering corporate sector debt is frequently essential to corporate
revival and reducing vulnerability to future shocks. This situation por-
tends debt restructuring along with operational restructuring. Coun-
tries have used different approaches for corporate restructuring: a
centralized, government-led approach that concentrates asset recov-
ery in one public agency or asset management company (AMC), or a
GOVERNMENT CONTINGENT LIA131LITIES IN BANKING 325
Figure 14.3. When Depositors Absorb Losses: Estonia (1992),
Argentina (1980-82), Japan (1946), and United States (1933)
Real Deposits
Index (crisis = 1)
3.5
Argentina*.' --
3.0
20.5 - ..--.,.' '.
2.0
'1.5
United States
1.0
-24 -18 -12 -6 0 6 12 18 24
Month before and after crisis
Industrial Production Index
Index (crisis = 1)
1.8
1.6 - United States
1.4 Argentina Estonia
1.2 --
1.0
0.8
Japan,,
0.6 ,.,,,,,,,,,,, , , .. .. ........ ........... ...........
-24 -18 -12 -6 0 6 12 18 24
Month before and after crisis
Source: Baer and Klingebiel (I1995).
326 CLAESSENS AND KLINGEBIEL
decentralized approach led by banks and other creditors. Both ap-
proaches have advantages and disadvantages.
Centralization of assets may permit a consolidation of skills and
resources and easier monitoring and supervision of workout practices.
As claims are consolidated, more leverage may be obtained over debt-
ors and perverse links between banks and corporations may be bro-
ken, allowing better collection on (connected) loans. Yet an AMC
holding a large portion of corporate claims is difficult to insulate from
political pressures. Moreover, a transfer of loans breaks the links be-
tween banks and corporations, links that may have positive value given
banks' privileged access to corporate information. And if AMC assets
are not actively managed, credit discipline in the whole financial sys-
tem can be undermined, increasing the overall costs of the crisis. Pub-
lic AMCs have had a mixed record (Klingebiel 2000) and often have
led to higher fiscal costs because assets were inadequately managed
and consequently lost value (see Box 14.1).
The decentralized, creditor-led workout approach relies on banks
and other creditors to resolve nonperforming loans. Because banks
have the institutional knowledge of the borrower and because their
own survival depends on asset recovery, they may be better willing
and able to maximize recovery value and avoid future losses. Further-
more, banks can provide new loans in debt restructuring. Successful
decentralized debt workouts require, however, limited or no owner-
ship links between banks and corporations (otherwise, the same party
would be both debtor and creditor), adequately capitalized banks, and
proper incentives for banks and borrowers. The very slow speed of
restructuring in Japan, for example, has stemmed in part from exten-
sive ownership links among banks, other financial intermediaries, and
corporations. As a result, there was a deadlock on claims in Japan,
which was not broken for a long time. Similarly, banks need to be
adequately capitalized to have the loss absorption capacity to engage
in corporate restructuring. If governments allow banks to recapitalize
via increased earnings over a longer time horizon-either through im-
plicit or explicit forbearance-banks' abilities to engage in rapid cor-
porate restructuring are limited.
The Incentive Framework. Regardless of whether banks have the
resources, the degree and sustainability of corporate restructuring, and
therefore the final fiscal costs, will depend on the incentives for proper
restructuring. With weak incentives, creditors may be inclined to roll
over nonperforming loans rather than to restructure them, hoping for
eventual recovery, with the downside risks covered by an implicit gov-
ernment guarantee. The incentives for proper restructuring include ad-
equate accounting and troubled debt restructuring standards that ensure
that standards are applied uniformly across all classes of creditors and
GOVERNMENT CONTINGENT LIABIILITIES IN BANKING 327
Box 14.1 The Use of Asset Management Companies in the
Resolution of Banking Crises--Cross-Country Experience
In the past, asset management companies (AMCs) have been employed
to address the overhang of bad debt in the financial system. Two main
types of AMCs can be distinguished: those set up to help and expedite
corporate restructuring and those established as rapid asset disposition
vehicles. A review of seven AMCs in Finland, Ghana, Mexico, the Phil-
ippines, Spain, Sweden, and the United States reveals that they have a
mixed record. In two out of three cases, corporate-restructuring AMCs
did not achieve their narrow goals of expediting bank or corporate
restructuring. These experiences suggest that AMCs are rarely good
tools for expediting corporate restructuring. Only the Swedish AMC
successfully managed its portfolio, acting in some instances as lead
agent in the restructuring process. It was helped by some special cir-
cumstances, however: the assets acquired were mostly real estate-re-
lated, not manufacturing-related (vihich are harder to restructure), and
were a small fraction of the banking system (which made it easier for
the AMC to maintain its independence from political pressures and to
sell assets back to the private sector). Rapid asset disposition vehicles
fared somewhat better, with two out of four countries-namely, Spain
and the United States-achieving their objectives.
The successful experiences suggest that AMCs can be used effec-
tively, but only for the narrowly defined purposes of resolving insol-
vent and nonviable financial institutions and selling their assets. But
even achieving these objectives required many ingredients: a type of
asset that can be easily liquefied (real estate), mostly professional man-
agement, political independence, a skilled resource base, appropriate
funding, adequate bankruptcy and foreclosure laws, good information
and management systems, and tiansparency in operations and pro-
cesses. In the Philippines and Mexico, the success of the AMCs was
doomed from the start, because governments transferred politically
motivated loans or fraudulent assets to the AMCs. A government agency
susceptible to political pressure and lacking independence finds these
assets difficult to resolve or to se:ll off. Neither of these agencies suc-
ceeded in achieving its narrow objectives. (Source: Klingebiel 2000.)
that classification is based on the borrower's demonstrated ability to
repay. To that effect, bank regulators may need to evaluate all rules,
regulations, and policy statements to ensure that they facilitate, not
hinder, restructuring by allowing partial debt forgiveness; permitting
immediate debt write-off for tax purposes; providing flexibility in valu-
ing payments in-kind and tax relief on such payments; eliminating
328 CLAESSENS AND KLINGEBIEL
artificial ceilings on assets acquired by financial institutions through
restructuring; eliminating unnecessary taxes, duties, and levies on shares
issued as a result of debt-to-equity conversions; and eliminating all
taxes on the issues or exchange of debt instruments used in debt
restructuring.
Other measures that influence bank and corporate incentives for
proper restructuring are liberal foreign investment and foreign entry
rules and liberal merger and acquisition policies. During times of cri-
sis, foreign investment can provide much-needed capital and exper-
tise. Adequate bankruptcy procedures are necessary to ensure that
debtors can be forced to negotiate in good faith and minimize the
transactions costs of resolving financial distress.
Tools for Reducing the Probability of Banking Crises
The probability of banking crises can be reduced by undertaking fi-
nancial restructuring in tandem with other fundamental reforms-
strengthening prudential regulation; adopting internationally accepted
accounting, auditing, and financial reporting standards and practices;
and toughening compliance and regulation to align the incentives of
market participants with prudent banking. The extent to which exces-
sive risk-taking is curbed by regulation and penalized by the supervi-
sory authority as well as by the market greatly influences the behavior
of financial institutions. The overall incentive framework in which
financial institutions operate significantly affects the risks of a future
financial crisis.
Three groups could potentially monitor financial institution man-
agers: owners, the market, and supervisors. The government should
strive to ensure that each exerts pressure on managers to engage in
prudent risk-taking. In industrial economies, authorities erect entry
barriers; they enforce modest capital requirements, usually above the
BIS (Bank for International Settlements) minimum of 8 percent (capi-
tal to risk-weighted assets); and intermediaries face market discipline
in money and capital markets, which usually are weakened by explicit
government guarantees and are supervised by one or more govern-
ment agencies. Industrial country authorities have tended to permit
bank exit, though some individual banks still engage in excessive ex-
pansions that cause systemic difficulties. In developing and transitional
economies, where risks are greater because of the small and often more
concentrated economies, where shocks often are larger and volatility
greater, and where the market's ability to monitor banks is hampered
by poor information, governments need to enhance the ability and
incentives of these three groups.
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 329
Incentive Structure of Owners, Creditors,
and Other Claim Holders
T hose who own equity in a bank in principle have both the ability and
the incentive to monitor the actions of their bank. They tend to pro-
vide effective self-regulation when they have much at risk, in the form
of either capital or expected future profits. Moreover, well-capitalized
banks are usually better monitored by their shareholders. Small share-
holders, however, will tend to free ride, so it is important that govern-
ment makes sure that there are some large stakeholders, or strategic
investors, who will take and bear the responsibility for running the
bank. Inside and outside investors need to face the loss of their invest-
rment, and they and their managers need to see the possibility of bank
failure or exit from the industry to encourage prudent behavior.
Some emerging economies have raised minimum capital ratios above
that for most industrial economies to take into account the riskier
environment in which banks operate and the difficulty in measuring
the economic net worth of a bank using backward-looking accounting
measures.3 Even then, capital adequacy is inherently a backward-look-
ing accounting indicator of the true solvency of the financial institu-
tion. Some banks with high measured capital have become insolvent
in short periods of time,4 even in economies with good accounting
standards and practices.
The increased incentives to engage in excessive risk-taking when
the capital adequacy position is weakened make it all the more impor-
tant not to rely just on accounting capital adequacy alone. Countries
have applied one or more of the following measures: limiting entry or
otherwise raising franchise value (future profitability), which can be
collected only by banks that remain open; enhancing the liability of
directors and shareholders, which New Zealand authorities have un-
dertaken; and requiring the issuance of subordinated debt. Some coun-
tries also have enhanced liability beyond current capital levels by
applying stiff penalties when bankers violate regulations or agreements
with supervisors about how they will take and monitor risks.5 Devel-
oping country authorities need to choose (at least) one of these addi-
tional methods for improving the incentives of bank owners to behave
prudently. Although some of these methods may be relatively blunt,
the costs of not using them can be quite high.
Incentive Structure of Market Participants
Market participants, principally those who enter into a creditor rela-
tionship with a bank, monitor and discipline the relationship if they
have the ability and the incentives. The ability to monitor banks depends
330 CLAESSENS AND KLINGEBIEL
on the reliability and range of information available. The starting point
therefore is adequate accounting standards and practices. Authorities
in some countries recently put into place extensive disclosure require-
ments backed up by enhanced liability (New Zealand), mandatory
ratings by at least two private rating agencies (Chile), and an online
credit reporting system (Argentina). Beyond information, creditors need
incentives to monitor in the form of the assurance that they will be
allowed to suffer losses. Although small depositors are unlikely to be
good monitors of banks, large debt holders have a much greater po-
tential to fulfill this role. At the very least, large debt holders need to
be reminded that they are not covered by any explicit or implicit de-
posit insurance scheme. Mandating that banks periodically issue large
blocks of uninsured, subordinated debt, as recently instituted in Ar-
gentina, could in some circumstances further enhance market moni-
toring and also could create a class of future bank owners; if the current
owners fail to ensure a safe and sound bank, the subordinated debt
holders could take over the bank. The incentives of subordinated debt
holders may thus be appropriately balanced.
Incentive Structure for Supervisors
Although owners and markets can be motivated to provide oversight,
banks, given their special nature, also are subject to government su-
pervision. Historically, bank supervision in developing and transitional
economies was oriented toward ensuring compliance with government
directives on credit allocation. Though lagging relative to other parts
of financial reform programs, authorities in most developing countries
have moved to engage in prudential supervision. Less attention has
been devoted to providing supervisors with the incentives both to
monitor better and to take actions based on this effort. If there are no
incentives to monitor, and thus no consequences for banks for violat-
ing a regulatory framework, the trend toward greater supervision will
be completely ineffective. It follows, then, that one way to promote
better supervision is to give authorities better incentives. In many coun-
tries, supervisors are paid poorly relative to their counterparts in banks.
At the very least, low pay makes it difficult to attract qualified person-
nel, and may negate the effects of even the best training programs as
skilled supervisors move to the banking sector. Moreover, the lure of
eventual high-paying jobs leaves open a form of corruption: less rigor-
ous supervision now in exchange for a lucrative salary later. This dis-
incentive for effective supervision can be reduced only by raising
supervisory pay reasonably close to private sector levels.
Furthermore, it may be useful to tie the hands of supervisors and
lay down the course of action to be followed. In the context of dealing
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 331
with weak banks, it has become increasingly common to recommend
that countries adopt the "prompt, corrective action and structured,
early intervention" approach analogous to that embodied in U.S. leg-
islation. Structured, early intervention calls for (a) higher capital; (b)
st.ructured, prespecified, publicly announced responses by regulators,
triggered by decreases in a bank's performance (such as capital ratios)
below established levels; (c) mandatory resolution of a capital-depleted
bank at a prespecified point when capital is still positive; and (d) mar-
ket value accounting and reporting of capital. Although this approach
appears to have yielded promising results in the United States, it is by
no means certain that this model vworks at all times or can be exported
to other countries.
Conc:lusion
Recent experience suggests that the fiscal costs of a banking crisis can
be substantial and can, in extreme cases, threaten fiscal stability. On
average, a banking crisis in an emerging market has added 14 percent-
age points to government spending relative to GDP, and in some coun-
tries as much as 50 percentage points of GDP. Governments have taken
on liabilities in banking sector restructuring because of explicit guar-
antees-either extensive public deposit insurance or direct government
guarantees on bank liabilities-or implicit government guarantees.
'While many of the causes of banking crises have been identified, it has
proven difficult to predict the incidence and exact timing of a financial
crisis. It also has been hard to predict with any precision the magni-
tude of contingent liabilities arising from banking system distress. Ex-
isting approaches to measuring contingent liabilities in the banking
sector, while they represent a good start, often fall short. Most are
based on static models, which focus on the typical causes of a banking
c risis, rather than on policy responses. Most important, these models
do not analyze what governments can do to control fiscal liabilities
arising from the banking sector, because they fail to recognize the dy-
namic dimensions of managing contingent liabilities.
Recent research has identified more precisely the main policy mea-
sures contributing to banking crisis costs. Investigators found that li-
quidity support, unlimited guarantees, and forbearance contribute the
most to costs. These findings suggest some crisis management strate-
gies that can limit the fiscal costs. In addition, cross-country evidence
indicates that the comprehensiveness of restructuring efforts can deter-
mine the robustness of the banking system in the medium term. The
probability of banking crises can. be reduced through a variety of in-
struments that improve the incentive framework.
332 CLAESSENS AND KLINGEBIEL
Notes
1. Honohan and Klingebiel (2000) do not include these medium-term strat-
egies in their empirical analysis because this level of detail would be difficult
to cover in a regression framework, but many case studies highlight the value
of these principles (see Sheng 1996).
2. It also should be noted that guarantees might not be credible if the scale
of the losses is estimated to be so large that the government is perceived to
lack the resources and capacity to fulfill the guarantee. This happened in
some countries, where a depositor run turned into a currency panic.
3. In Argentina, for example, the minimum capital adequacy requirement
is 11.5 percent, with higher requirements for banks engaging in riskier ac-
tivities and having a weaker risk management capacity. The average actual
capital adequacy ratio in Argentina, for example, was close to 16 percent in
1997. Furthermore, before the financial 2001 crisis, banks in Argentina were
subject to high liquidity requirements. Singapore also has higher capital ad-
equacy requirements (12 percent). Moreover, most banks in countries with 8
percent capital adequacy requirements have capital adequacy ratios that greatly
exceed those: the average capital adequacy ratio in the United States, for
example, is about 12 percent.
4. In a world of derivatives, balance sheets can be altered in minutes.
5. In evaluating market risk, supervisors around the world have recently
moved toward assessing the quality of the risk management tools banks use
rather than the actual positions. Banks are then fined if they violate risk
management arrangements ex post.
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Alba, Pedro, Leonarda Hernandez, and Daniela Klingebiel. 1999. "Financial
Liberalization and the Capital Account: Thailand, 1988-97." Policy Re-
search Working Paper 2188. World Bank, Washington, D.C.
Baer, Herbert, and Daniela Klingebiel. 1995. "Systematic Risk When De-
positors Bear Losses: Five Case Studies." In George G. Kaufman, ed.,
Research in Financial Services: Private and Public Policy. Vol. 7. Green-
wich, Conn.: JAI Press.
Barth, James R., Gerard Caprio, Jr., and Ross Levine. 2000. "Banking Sys-
tems around the Globe: Do Regulation and Ownership Affect Performance
and Stability?" Policy Research Working Paper 2325. World Bank, Wash-
ington, D.C..(Forthcoming in Frederick Mishkin, ed. Prudential Regula-
tion and Supervision: What Works and What Doesn't. Chicago: National
Bureau of Economic Research/Chicago University Press.)
GOVERNMENT CONTINGENT LIABILITIES IN BANKING 333
- -. 2001. "Bank Regulation and Supervision: What Works Best." World
Bank, Washington, D.C. Processedl.
Blejer, Mario, and Liliana Schumacher. 1998. "Central Bank Vulnerability
and the Credibility of Commitments: A Value-at-Risk Approach to Cur-
rency Crises." IMF Working Paper WP/98165. International Monetary
Fund, Washington, D.C.
Bongini, Paolo, Stiin Claessens, and Giovanni Ferri. 2001. "The Political
Economy of Distress in East Asian Financial Institutions." Journal of Fi-
nancial Services Research 19 (1): 1-25.
Caprio, Gerard, Jr., and Asli Demirgbi-Kunt. 1997. "The Role of Long Term
Finance: Theory and Evidence." Policy Research Working Paper 1746.
World Bank, Washington, D.C.
Caprio, Gerard, Jr., and Daniela Klingebiel. 1997. "Bank Insolvency: Bad
Luck, Bad Policy, or Bad Banking?"' In Michael Bruno and Boris Pleskovic,
eds., Annual World Bank Conference on Development Economics. Wash-
ington, D.C.: World Bank.
--. 1999. "Scope and Fiscal Cost of Banking Crisis: Compilation of
Information on Systemic and Non-systemic Crises from 1970s Onward."
Financial Policy website . World Bank, Washington, D.C.
Claessens, Stijn. 1999. "Experiences of Resolution of Banking Crises." In
Strengthening the Banking System in China: A Joint BIS/PBC Conference
Held in Beijing, China, 1-2 March, 1999. BIS Policy Paper 7. October
(Bank for International Settlements).
Demirguc-Kunt, Asli, and Enrica Detragiache. 1998. "Financial Liberaliza-
tion and Financial Fragility." In loseph Stiglitz and Boris Pleskovic, eds.
Annual World Bank Conference on Development Economics. Washing-
ton, D.C.: World Bank.
-- . 2000. "Monitoring Banking Sector Fragility: A Multivariate Logit
Approach." World Bank Economic Review 14 (2): 287-307.
Goldstein, Morris, Graciela Kaminsky, and Carmen Reinhart. 2000. Assess-
ing Financial Vulnerability: An Early Warning System for Emerging Mar-
kets. Washington, D.C.: Institute of International Economics.
Gonzales-Hermosillo, B. 1999. "Determinants of Ex-ante Banking System
Distress: A Macro-Micro Empirical Exploration." IMF Working Paper
WP/99/33. International Monetary Fund, Washington, D.C.
Hardy, Daniel, and Ceyla Pazarbasic,glu. 1998. "Leading Indicators of Bank-
ing Crises: Was Asia Different?" IMF Working Paper WP/98/91. Interna-
tional Monetary Fund, Washington, D.C., June.
Honohan, Patrick. 1999. "Fiscal Contingency Planning for Banking Crises."
Policy Research Working Paper 2228. World Bank, Washington, D.C.
Honohan, Patrick, and Daniela Klingebiel. 2000. "Controlling Fiscal Costs
of Banking Crises." Policy Research Working Paper 2421. World Bank,
Washington, D.C.
334 CLAESSENS AND KLINGEBIEL
Kaminsky, Graciela, and Carmen Reinhart. 1999. "The Twin Crises: The
Causes of Banking and Balance-of-Payments Problems." American Eco-
noinic Reviewv 89 (3): 473-500.
Keefer, Philip. 1999. "Political Institutions and Crisis: The Effects of Politi-
cal Checks and Balances on the Dynamics of Financial Sector Distress."
World Bank, Washington, D.C., June. Processed.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, E H. Rogers, and L. Squire, eds.,
World Bank Economists' Forum. Washington, D.C.: World Bank.
Klingebiel, Daniela. 2000. "The Use of Asset Management Companies in the
Resolution of Banking Crises." Policy Research Working Paper 2284.
Washington, D.C.: World Bank.
Laeven, Luc. 1999. "Risk and Efficiency in East Asian Banks." Policy Re-
search Working Paper 2255. World Bank, Washington, D.C.
La Porta, Rafael, Florencio Lopez de Silanes, and Andrei Shleifer. 2000. "Gov-
ernment Ownership of Banks." Harvard University, Cambridge, Mass.,
August. Processed.
Sheng, Andrew, ed. 1996. "Bank Restructuring: Lessons from the 1980s."
World Bank, Washington, D.C. Processed.
Standard and Poor's Sovereign Ratings Service. 1999. "Global Financial Sys-
tem Stress: 24 Show Adverse Trends in Credit Quality." New York, May.
World Bank. 1993. East Asia Miracle. Policy Research Report. New York:
Oxford University Press.
. 2001a. Finance for Growth: Policy Choices in a Volatile World. Policy
Research Report. New York: Oxford University Press.
- 2001b. World Development Report: Institutions for Markets. New
York: Oxford University Press.
CHAPTER I 5
Gover:nment
Insurance Programs:
Risks and RiskI Management
Ron Feldman
Federal Reserve Bank of Minneapolis
INSURANCE CAN IMPROVE THE allocation of resources in at least two
ways. First, risk-pooling can reduce the likelihood that the insured
will suffer a catastrophic loss. Second, the central administration of
risk-pooling allows for cost minimization by reducing administrative
expenses and transactions costs and exploiting economies of scale. As
a result, insurance provides an efficient method for firms and house-
holds to manage some of the risks they face.
These benefits explain the long history of insurance provision, the
large size of private insurance markets in many countries, and the in-
creasing array of risks for which insurance has become available. But
offering insurance is a risky business, and this truism is particularly
applicable when the government d[ecides to provide insurance. In par-
ticular, governments tend to offer insurance-with a face value in the
trillions of dollars in developed countries-for risks that are difficult
to insure because of problems of correlated losses, moral hazard, ad-
verse selection, poor data, and limited demand. And partly as a result,
governments have suffered very high losses across a range of programs
from flood to pension insurance. Indeed, a poorly designed insurance
program-deposit insurance, for example-can actually retard a
country's growth.'
Policymakers thus face difficult tradeoffs. Government insurance
offers the potential to improve welfare by providing unique risk man-
agement for catastrophes such as natural disasters and banking panics.
At the same time, government must avoid the high costs that can arise
335
336 RON FELDMAN
when it absorbs risks without the policies and institutional capabilities
to manage them. To that end, tools are available to government to
help manage the exposure created by insurance programs. Many of
these tools are commonplace in the private sector's provision of insur-
ance, including risk-based pricing, the use of loss-sharing techniques
such as coinsurance, deductibles and caps on losses, discretion in the
coverage of risks, and effective verification and adjustment of loss.
Government also has unique powers that can assist it in the manage-
ment of insurance exposure. For example, the government can increase
opportunities for risk-pooling by compelling firms and households to
participate in insurance programs. The government also can use regu-
latory powers to try to manage the risk-taking of the insured. At times,
however, these techniques have proven ineffective or at odds with the
political objectives of policymakers.
The rest of this chapter provides analysis and background specifi-
cally on government insurance programs in the hopes of assisting
policymakers who face tradeoffs in managing them. The next section
discusses the inherent and political challenges that governments face
in offering insurance. It is followed by a section that examines some of
the methods available to the government insurer to mitigate the risks
it assumes. The concluding section emphasizes the need to reassess
continually the direct government provision of insurance given ad-
vances in private risk-bearing.
Challenges to Policymakers Offering
Government Insurance
The challenges to governments providing insurance are twofold. First,
insurance programs are more likely to remain solvent if certain con-
ditions hold, but governments typically provide insurance programs
when these favorable conditions do not exist. Second, the use of tech-
niques to manage the risks that government insurers assume can re-
duce the political benefits that insurance programs provide policy-
makers. As a result, policymakers may not implement these tools
(discussed in detail in the next section), thereby increasing the chance
the government insurer will become insolvent or distort resource al-
location or both.
Inherent Challenges
Several conditions facilitate effective and sustainable provision of in-
surance. First, the probability that one person or firm in the risk pool
will suffer losses should be independent of or uncorrelated with the
probability that others in the pool will suffer losses. However, virtu-
RISKS AND RISK MANAGEMENT 337
ally all of the insurance programs offered by governments lack inde-
pendence of loss. Banks whose deposits are insured, for example, tend
to fail in bunches. A similar situation may exist for firms that sponsor
pensions insured by the government (for example, many large steel
firms with underfunded pensions can face difficulty at the same time).
Likewise, large areas of agricultural production can face drought or
excessive rain. Conversely, hail is a localized phenomenon, and pri-
vate insurers offer policies against hail damage. A lack of indepen-
dence of loss means that government could suffer substantial losses
over a short period.
Second, losses suffered by the insured should result from chance
and not from the actions of the insured. However, acquiring insurance
introduces the moral hazard problem that the insured will more likely
engage in behavior that raises their chances of making a claim. Con-
sider the case of government pension insurance. Workers may have
previously reviewed the funding of their pension or had an agent, such
as a union, do it for them. If future pension benefits were uncertain,
workers would typically demand more cash income and fewer pension
benefits. The same workers, however, have an incentive to reduce re-
views after their benefits become insured. The workers may now agree
to lower current compensation irt return for guaranteed future ben-
efits. Firm cash flows that used to support current compensation may
increasingly benefit equity holders via dividends or increased invest-
rnent in high-return/high-risk projects. Moral hazard becomes particu-
larly problematic when insurance covers 100 percent of losses, the
insured entity is insolvent, and the premiums charged do not vary with
the risk-taking of the insured (some government programs have all
three characteristics).
Third, the insurer must be able to estimate accurately the expected
claims of the insured. An insurer cannot make adequate forecasts if it
does not have sufficient data. There may be pressure for the govern-
inent to step in if private insurance providers do not offer insurance in
markets with limited data. The U.S. government took such a role with
mortgage insurance (Weicher 19.94).
Fourth, and closely related to point three, the insurer must have the
ability to distinguish among the insured based on the risk of loss they
pose. If not, the insurer will price risk inaccurately and, because of
adverse selection, will face higher-than-expected losses.
Finally, limited demand from consumers prevents successful provi-
sion of insurance, particularly insurance against natural disasters. A
lack of demand has several sources, including the high cost of insur-
ance due to a lack of independen,ce of loss and adverse selection and
the low incomes of potential beneficiaries.2 A second cause of low
demand is ex post assistance from government and charities to those
who have suffered catastrophic losses from natural disaster. In a recent
338 RON FELDMAN
example, the U.S. government provided ex post assistance to agricul-
tural producers in the summer of 1999 who had explicitly waived the
right to such payments when they refused to purchase crop insurance.3
The Political Challenge
Policymakers tend to provide insurance to achieve social and distribu-
tional objectives. Deposit insurance, for example, supports smaller fi-
nancial institutions that might otherwise have to pay more for deposits
than larger banks with more diversified asset portfolios. Similarly, pen-
sion insurance encourages establishment of pensions that provide a
fixed or defined benefit to workers.
More generally, governments can use insurance to subsidize certain
constituents by charging them premiums below the fair amount needed
to cover expected losses. In some cases, the very act of providing insur-
ance generates the subsidy, as no other insurer would offer the con-
stituent an insurance policy at any price. For example, government
insurers have been known to offer insurance against a natural disaster
when the probability of imminent loss is near 100 percent. Insurance
is an especially attractive method for providing subsidies because of its
low visibility. Taxpayers have limited information on insurance pro-
grams and muted incentives to carry out the challenging analysis needed
to determine the long-term costs of such programs.
The bottom line is that methods of creating the subsidy require that
the government insurer avoid practices that give it greater control in
managing its exposure, such as charging risk-based premiums. The
difficulty in monitoring the fiscal position of the insurance program
removes a potential check on policymakers' discretion to increase ex-
pected losses. As a result, governments may not enact the policies dis-
cussed in the next section that guard against fiscal loss and resource
misallocation.
Methods for Addressing the Risks Posed
by Government Insurance Programs
Government insurers have two sets of tools available to them to man-
age the risks they assume. One group of methodologies reflects the
standard insurance management techniques used by private firms, in-
cluding risk-based pricing, loss-sharing, and coverage rules. The sec-
ond set of tools is uniquely available to the public insurance entity and
includes compulsion, regulation, and budgeting. Policymakers have
pointed to the availability of these instruments of government policy
to justify the provision of government insurance. However, these unique
RISKS AND RISK MANAGEMENT 339
attributes of government are not excuses for ignoring generic risk man-
agement techniques.
Generic Risk Management Tools for Insurers
Private insurers have had to develop a variety of methods to address
effectively the inherent challenges of providing insurance. Insurers must
address moral hazard, in particular. Observers have attributed some of
the significant losses that government insurers have suffered to a re-
fusal to apply these techniques.
Pricing. An insurance program will ultimately have a cash flow
deficit if the assessment charged to a household or firm in conjunction
with other sources of income is less than the expected costs of insuring
them. But perhaps the more profound importance of the actuarially
fair or risk-based premium is its influence on the behavior of the in-
sured. The insured will continue to take actions that increase her ex-
pected losses as long as the premium and other marginal costs that she
faces are less than the marginal benefit the actions produce. As a re-
sult, a fair premium encourages the insured to avoid actions that could
increase the chance of claim and loss (Blair and Fissel 1991).
Despite the importance of addressing moral hazard, government
insurance programs have at times deliberately charged less than the
fair premium. Some programs have charged no premium at all and
instead have relied on ex post levies. Governments justify this practice
as preventing misuse of funds helcl in reserve. An alternative common
practice has the insurer charging the same premium to all.
Although the setting of risk-based premiums has not been the norm
for many government insurance programs, the ability to set them is
not beyond the ken of government.4 Standard actuarial approaches to
premium-setting could be appropriate for programs such as crop in-
surance where data on crop yields and insurance performance go back
rnany years. The crop insurer should establish rates for broad catego-
ries of agricultural producers (for example, all wheat growers in a
given location). The insurer then adjusts the rates to account for indi-
vidual differences that influence losses such as the use of irrigation and
the variant of crop grown. The insurer can calculate initial rate classi-
fication and adjustments based on descriptive statistics-the average
loss for wheat farmers in a given location using irrigation, for ex-
ample-or through statistical techniques that isolate the effect of a
given attribute on the probability of loss. Putting the insured into groups
based on their general characteristics and charging an appropriate pre-
mium to each group are an effective response to the potential for ad-
verse selection.
340 RON FELDMAN
Premium-setting for insurance against the insolvency of firms (for
example, pension and deposit insurance) could benefit from advances
in both computing and practices in statistics and finance. Monte Carlo
simulations and the use of option-pricing theory provide new avenues
for estimating expected claims and fair premium-setting. (For details
on modeling the exposure of pension guarantee programs, see Chapter
13 by Pennacchi in this volume.) The utility of these models goes be-
yond premium-setting in one time period. Analysts could use these
models to judge exposure and simulate how changes in insurance poli-
cies might alter expected losses. In addition, some government insur-
ers can look to private markets for information that would help in
setting premiums. For example, risk assessments of banks captured in
fixed-income instrument or equity prices could serve as input into the
calculation of insurance assessments. More directly, a government in-
surer can contract with market participants to bear some of the risk of
an insured event occurring. The pricing of such contracts can provide
information for the government insurer in setting premiums for other
insured entities.
Loss-sharing. Seminal economic analysis of insurance by Arrow and
others recognized that insurers address moral hazard by having the
insured bear some risk of loss (Arrow 1977). Insurers share losses with
the insured in several ways. Capping the losses the insurer will bear
fosters risk-sharing. The International Monetary Fund (IMF) identi-
fied the provision of a "low" amount of coverage as a best practice in
government deposit insurance (Garcia 1999: 9, 12). The insurer can
also require the insured to pay a set amount, the deductible, before the
insured receives loss protection. Finally, the insurer can split losses through
a coinsurance arrangement. For example, the insurer may decide to cover
75 percent of losses. In implementing loss-sharing, governments should
consider, among other factors, the ability of private markets to provide
supplementary coverage, the benefits of targeting insurance coverage to
those least able to bear the risk themselves, the effect of loss-sharing on
the risk-taking of the insured, and the amount of exposure the govern-
ment can manage. Government insurers use loss-sharing to varying de-
grees, although it appears less often in public programs than in private
policies. This reluctance likely reflects social objectives and the poten-
tial for loss-sharing to produce an outcome, such as increased instability
in the banking system, that is at odds with the goals underlying the
establishment of the insurance program.
Loss-sharing policies also require policymakers to create time-con-
sistent commitments for imposing losses, especially if the insurer shifts
from a regime where it absorbed all losses, even when not legally re-
quired. Potential methods for establishing a credible loss-sharing re-
gime include passing laws that mandate loss-sharing, implementing
RISKS AND RISK MANAGEMENT 341
schemes that punish regulators for not imposing losses, and enacting
reforms that limit the damage loss-sharing could cause. Another method
for establishing credibility requires policymakers to appoint insurance
program managers who are more likely to impose losses than the gen-
eral populace. Other creative proposals address the time-consistency
problem. For example, the governmnent can agree to provide full pro-
tection for the first uninsured entity, such as a bank, that suffers loss
but not for other uninsured entities. This proposal could assuage
policymakers' fears of spillover failures. At the same time, the plan
increases creditors' incentives to price risk correctly so that they are
compensated in the event that their bank is the first to fail (Stern 2000).
Coverage Limits and Decisions, The expected claims for the gov-
ernment insurer depend significantly on the type of coverage it pro-
vides. Each coverage decision depends, in turn, on the particular
circumstances of the insurance program. In an obvious example, the
insurer would have a better chance at solvency if it insures risks for
which it has good data. Without sufficient data, the insurer should
take on only a limited amount of exposure and establish tight controls
on growth. The government insurer also reduces its chances for sol-
vency if it provides insurance against very high-probability losses or
routinely agrees to insure those whose record and behavior indicate a
very high probability of making a claim. In fact, the government would
discourage future risky behavior by those who are currently insured if
it excluded those posing a very high risk of claim from the pool.
Government disaster insurance programs provide numerous ex-
amples of the types of coverage decisions the insurer must make.
Consider insurance against flood damage. The government must de-
termine at what point prior to the flood it will offer insurance. If the
government provides coverage when flooding is imminent, it will be
nearly impossible to balance premiums with losses. The government
flood insurer in the United States had to increase the time between
the receipt of an application and the provision of insurance from 5 to
30 days to limit the number of applicants with very high probabili-
ties of making a claim. The flood insurer also had to restrict cover-
age on structures that suffered repeated damage over a short period
of time. Whatever and whomever it decides to cover, the government
insurer must produce a written contract or policy that clearly enu-
rnerates its decisions. Exposure can become virtually unlimited with-
out such documentation.
Verification and Adjustment of Loss. The insurer must have a pro-
cess in place to verify that the loss claimed by the insured actually
occurred and to determine the amount rightfully paid under the con-
tract. The absence of a proficient claims adjustment process invites
342 RON FELDMAN
fraud and unnecessary expenses. The adjustment process for govern-
ment insurance programs involving property damage could be similar
to practices by private insurers. In fact, the government insurer could
contract with private insurers to benefit from their potentially lower
costs and greater expertise. (Such contracting, of course, has monitor-
ing costs and requires incentives for the private provider to meet the
government's objectives.)
Adjustment in firm insolvency insurance programs may require dif-
ferent skills. For example, the government insurer needs to acquire
expertise in the disposition of assets held by the firm at insolvency.
The amount the government may owe the insured can depend on the
money the government raises through the sale of assets from the failed
firm. The ability of the government to sell assets for higher returns
after failure means lower losses to the insurer. Disposition of assets has
proven to be particularly challenging in a number of countries, par-
ticularly in the aftermath of mass bank failures. Creating the infra-
structure to sell the assets is costly and difficult, particularly when the
assets to sell are illiquid. Nonetheless, experience suggests that use of
private sector firms, financial market innovations such as securitization,
and disposition strategies that allow for the quick sale of assets can
support cost-effective collection efforts (FDIC 1998). The insurer can
also lower losses by recovering funds through lawsuits against the in-
sured firm that has unlawfully dissipated the value of its assets.
Unique Risk Management Tools for Government Insurers
In theory, governments might have a greater ability than private firms
to address problems of moral hazard, lack of independence of loss,
and adverse selection. These powers arise from the sovereign status of
the government and include means to compel participation in insur-
ance pools, regulate the behavior of insured entities, and raise funds
through taxes. In theory, these powers would allow the government to
provide risk-pooling where private firms could not, although their ef-
fectiveness has been less impressive in practice.
Compulsionz. Governments can attempt to compel firms and house-
holds to participate in risk pools. Governments often try to accom-
plish this objective by tightly linking receipt of a benefit that the
government provides to insured status. In many countries, only the
government has the right to grant a banking charter. The bank must
then accept deposit insurance in order to obtain the charter. This form
of compulsion could help address adverse selection, because the risk
pool will contain a larger representation of eligible candidates, not
just those posing a higher risk of loss. Uniform coverage could also
potentially provide more independence of loss. For example, govern-
RISKS AND RISK MANAGEMENT 343
ment could require all those who live near rivers to participate in the
government flood insurance program. Losses could be less correlated
across a risk pool of all those living near all rivers in a country com-
pared with a risk pool of those living near one river.
This form of compulsion has its limits, especially if the insurer tries
to force lower-loss entities to subsidize the higher-loss entities. The
lower-loss entities in such a scenario will take steps to reduce their
exposure to insurance assessments. Overcharged banks will shift away
from the assets or deposits on which premiums are levied toward un-
taxed items. In the extreme, overcharged insureds will give up the
government benefit tied to insurance and shift to untaxed options.
This substitution can leave the government in the position it sought to
avoid-namely, an insurance pool vvith an overrepresentation of high-
loss entities. Governments can try to limit shifts out of the risk pool by
inefficiently banning substitutes or taxing exits, although the latter
step will limit pool participation in the first place. More commonly,
governments will accompany attempts at compulsion with government
regulation of risk-taking.
Regulation. Governments can regulate the behavior of insured enti-
ties to address the potential for moral hazard and adverse selection
arid, more generally, to try to reduce the probability that an insured
entity will make a claim. Before granting insurance, the government
can set regulatory standards that artempt to limit the expected loss of
the insured. Entities that want to receive government protection from
natural disaster may have to first mitigate against loss by, for example,
ensuring that buildings meet quality codes. Similarly insured financial
institutions must meet certain levels of financial strength, often mea-
sured by equity-to-asset ratios, before government protection becomes
available. The government can also try to manage the amount of risk-
taking of the insured through laws and regulations that govern their
ongoing operations. The insurer rmay require that communities pre-
vent new building in disaster-prone areas in order to continue to re-
ceive disaster insurance. Likewise, the insurer could limit the asset
holdings of insured financial institutions.
In addition to promulgating regulations, the government insurer
and its agents could engage in ongoing supervision and monitoring. In
one form, the regulator audits the insured to ensure compliance with
regulations. But even this basic level of monitoring can be tricky given
the inevitable subjectiveness of compliance. A bank's compliance with
capital rules, for example, depends on the assessment of hard-to-value
assets. Regulators also go beyond checking for compliance and often
use quantitative and qualitative inputs to produce assessments of the
insured's level of risk. The government insurer may need supervisors
to generate such private information on the insured, because, as noted,
344 RON FELDMAN
it may not be able to rely on observable and easy-to-verify characteris-
tics to determine expected losses.
Academics and some policymakers have expressed significant skep-
ticism about the ability of regulation to limit effectively and effi-
ciently the risk-taking of the insured.' Moral hazard results when
economic agents do not bear the marginal costs of their actions. Regu-
latory regimes can alter marginal costs, but they accomplish this task
through very crude and often exploitable tactics. In particular, reli-
ance on lagging regulatory measures, restrictive regulatory and legal
norms, and the ability of insureds to quickly alter their risk profile
have often resulted in large claims despite regulation. Moreover, while
regulators have access to inside information, there still appear to be
areas of profound informational asymmetry between regulators and
insured entities.
A second concern focuses on forbearance-that is, the regulator
and the insurer may not take action against the insured in the hope
that the riskiness of the insured declines in the future. Forbearance can
result from political pressure or attempts by the regulator and insurer
to avoid damage to their reputations.
Finally, even if regulators had the ability to assess risk-taking accu-
rately, they do not have a sound basis for determining how much risk
is too much or too little. Supervision and regulation of insured entities,
in other words, should not be expected to result in the economically
efficient amount of risk-taking.
Some have used such criticism to call for less, or even an abolish-
ment of, regulation. But, for the most part, analysts have called on the
insurer and regulator to augment regulatory assessments of the insured's
risk-taking with market assessments. To accomplish this goal, as noted,
the government insurer can shift some of the risk of loss to private
agents. The price charged for such reinsurance would provide the pri-
mary government insurer and regulator with an assessment of risk. A
second source for market risk assessments are market participants also
at risk of loss from the insured's actions. The prices that these market
agents set on debt issued by the insured firm, for example, could pro-
vide useful information. A variant of this form would rely on the credit
ratings of the insured. The insurer could use these market signals to set
premiums or determine eligibility for insurance. The signals could also
trigger regulatory restrictions on the insured's actions. The market risk
assessments also have the advantage of being visible to the public.
They could therefore make it difficult for the government to ignore or
cover up potential exposure to the insurance program.6
Loss Absorption and Smoothing. Some governments can withstand
larger losses than private firms. In private markets, the more uncertain
the costs of providing the insurance and the larger the expected claims,
RISKS AND RISK MANAGEMENT 345
the more capital the insurer will have to hold to ensure it can survive
large claims. But most government programs do not raise funds from
private investors to withstand losses. Most government insurers raise
cash to pay for future losses through the government's taxing author-
ity and benefit from de facto taxpayer support. As a result, the govern-
ment insurer may not have to hold any funds to ensure that it can pay
off losses. Instead, the government could borrow funds and repay these
funds with future premiums and taxes if current premium income falls
below current claims. This means that governments could provide in-
surance even if there were not independence of loss in the risk pool.
U.S. Treasury Secretary Lawrence Summers summed up this position,
noting that "the Federal government is uniquely capable of spreading
risk over time. A private insurer .. . can only lose so much money in
any given period without being declared bankrupt. By contrast, the
capacity of the Federal government to borrow for the purpose of meet-
ing short-term contingencies dwarfs that of any private sector entity."7
Two very important caveats accompany this observation. In par-
ticular, the government must have a near uninterrupted ability to ac-
cess funds, through taxes or borrowving, at a "manageable" cost. If this
is not the case, the government insurer may be no better off than a
private firm in absorbing large losses. Moreover, even if the insurer
does not hold capital, it must still incorporate the lack of indepen-
dence of loss in its pricing if it wants to manage future losses and have
premium income pay for claims.
Public Budgeting and Management Strategies. A well-designed public
budgeting regime can give policyrriakers incentives to manage the ex-
posure of the government to expected insurance losses. Budgetary rules
and accounting that require policymakers to face the opportunity costs
today of potential costs in the future could achieve that objective.
Accounting for the finances of 'government insurance programs on
an accrual rather than a cash basis could make opportunity costs more
explicit.8 As noted, many public budgeting systems do not provide
useful data on future exposures. In Chapter 3 of this volume, Allen
Schick describes how short-term, cash-driven recognition of govern-
ment budgets can make insurance programs appear to be cash genera-
tors until the moment they suffer massive financial losses. Conventional
budgeting systems do not give policymakers reason to act to limit fis-
cal, and related deadweight, losses.
In a similar vein, policymakers can structure the incentives of those
running the government insurance program to facilitate management
of claims exposure. Managerial strategies must address the fact that
insurance programs require frequent modifications to keep losses within
clesired ranges. Elected bodies with rules designed to foster debate and
inclusion are often incapable of making the timely decisions required
346 RON FELDMAN
to keep insurance programs in fiscal shape. Environments in which
managers have the contractual incentives and the power to make nec-
essary adjustments could prove more effective. Establishing clear ob-
jectives, agreeing how to measure achievement of the objectives, and
linking rewards and performance are important steps in shifting insur-
ance management responsibilities from policymakers to managers.
These steps could also help address the principal-agent concerns that
such a delegation will produce.
Public Information Campaign. Public information campaigns de-
scribing the risk of loss from rare events could increase the demand for
government insurance programs, particularly in cases where underes-
timation of such losses helps explain limited demand. Some govern-
ment disaster insurance programs attribute an increase in participation
rates to more extensive and more effective advertising. In particular,
government insurers have shifted from more traditional and staid warn-
ings about risk to advertising campaigns that are closer in spirit to
those used by firms with well-known consumer brands. And there are
several reasons why the government may need to provide such infor-
mation. Information on the probability of a natural disaster could serve
as a public good. The public cannot easily verify the quality of forecasts
and could discount those provided by private insurers as biased toward
insurance purchase. The government forecasts usually come from an
agency with no clear ties to or financial gain from insurance purchase.
Finally, marketing efforts could convey the steps the government is
taking to reduce the chance residents will receive ex post assistance.
Importance of Regularly Reassessing
Direct Insurance Provision
The previous discussion focused on designing and managing govern-
ment insurance programs to increase their probability of remaining
solvent. That discussion implicitly presumed underlying social or eco-
nomic justifications for government provision of insurance. However,
these justifications may never have been strong and may have become
weaker with advancements in private insurance markets. Policymakers
thus have a strong reason to review insurance programs regularly and
to determine whether alternative, more attractive methods to achieve
their goals exist. In terms of social objectives, a direct subsidy pro-
gram may prove more effective and efficient than government insur-
ance. Additionally, improvements in insurance and risk-shifting
technology may allow private firms to offer insurance where govern-
ment had previously. In another option, the government could regu-
late private insurance markets in order to increase coverage. However,
such regulation could have significant drawbacks.
RISKS AND RISK MANAGEMENT 347
Alternative Methods of Subsidization
Several attributes of insurance that have been discussed suggest it is
not a good distribution method for subsidies. Policymakers cannot
target the subsidy very well through insurance. The benefits of govern-
ment insurance often go to unintended parties (for example, the man-
agers of insured firms) and those who could otherwise protect
themselves from loss. In addition, the policies that create the subsidy,
such as mispricing, lead to poor resource allocation and increased
chances of program insolvency. Finally, the overhead to deliver a sub-
sidy through an insurance program makes it an expensive option rela-
tive to direct payments, vouchers, or tax credits. For the recipient,
cash payments have a higher value than in-kind transfers.
This discussion juxtaposes direct subsidies with direct provision of
insurance. But, for example, how does insurance as a subsidization
tool compare with ex post disaster assistance? This comparison de-
pends on the actual management and structure of the insurance pro-
gram. Poorly designed government insurance programs can produce
the same moral hazard problems associated with disaster assistance.
Moreover, the provision of insurance may not make delivery of ex
post disaster assistance less likely. As such, it is not clear that replacing
ex post assistance with government insurance will reduce the
government's fiscal exposure.9
7echnological Advancements in Private Risk-Shifting Markets
The private markets that can absorb risk of loss from uncertain out-
comes have become much larger and more robust over the last decade.
It is now more feasible for these private markets to bear risks that
government may have assumed previously. Some of the private market
expansion has occurred in traditional risk-bearing organizations. In
particular, consolidation in the insurance industry has led to larger
firms, some of which have increased cross-national exposure.'" The
resulting firms should have more capital and diversified portfolios.
The growing liquidity of capital markets and increased ability to evalu-
ate risk have also led to increased capital for the primary and reinsur-
ance markets. This increased capacity has been accompanied by lower
prices, making reinsurance more widely available for primary insurers
(Standard and Poor's 1999b). In total, U.S. property and casualty in-
surers, for example, significantly increased their capacity to finance
major catastrophic property loss during the 1990s (Cummins, Doherty,
and Lo 1999).
There has also been significant growth in traditional financial in-
struments that allow firms and households to better manage their ex-
posure to loss. For example, the volume of agricultural options traded
348 RON FELDMAN
on the Chicago Board of Trade has more than doubled over the last
decade, and the volume of agricultural futures has gone up by about
30 percent." Despite this growth, there remains ample opportunity
for expansion as only a small minority of agricultural producers, in
the United States at least, make use of financial contracts to manage
their risk. And financial contracts are not the only traditional yet
underused method that producers can utilize to manage exposure to
loss. For example, farmers can hedge risk of loss through planting
diversification, vertical integration, and marketing contracts.'2 In ad-
dition, advances in financial modeling have made it more feasible for
producers, in fields such as agriculture, to simulate future revenue and
optimize risk management strategies (Falloon 1999).
In addition to growth in traditional markets, advancements in the
technology of insurance provision have increased the private market's
capacity to absorb loss, especially with regard to natural disasters.13
Huge strides in computing and analytical techniques have made it more
feasible to model and forecast the likelihood and cost of natural disas-
ters (although limited historical data remain a concern). Advances in
financial technologies in areas such as securitization and derivatives
have led to the development of financial instruments that shift the loss
caused by catastrophic events to capital market investors. Property
catastrophe put and call options are exchange-traded derivatives that
would allow an insurer to transfer risks above some limit, based on
the total claim payments made by the insurance industry for property
damage due to catastrophe, for example, to investors. Another choice
with increasing appeal is a catastrophe or act-of-God bond. Issuers
structure the bonds so that the bonds reduce principal and interest
payments when a catastrophe occurs. Thus the holder of the bond
takes on some of the risk of loss. In addition, there are markets for
catastrophe swaps where those bearing risk can exchange exposures
to loss.
All of these financial instruments, and there are other examples
not mentioned here, tap into a much deeper pool of capital with a
greater ability to diversify than is available to a single firm. More-
over, capital market investors can diversify their portfolios more com-
pletely than an insurer. In combination, these developments increase
the ability of the private insurance markets to bear risk and reduce
the likelihood that capacity and pricing will fluctuate as much as
they have in the past.
Many of the new financial instruments discussed are geared to in-
surers themselves, but innovation has also allowed producers to trans-
fer risk directly to capital markets. For example, producers have a
greater ability to hedge against weather-related loss directly in finan-
cial markets through so-called weather options. In the past, these pro-
ducers may have been able only to hedge or insure against more general
RISKS AND RISK MANAGEMENT 349
changes in the price of their output.'4 The combination of these devel-
opments could dramatically change how governments provide insur-
ance. Rather than bearing all the risk from weather loss through crop
and flood insurance programs, the government could make use of the
increased capacity of the insurance industry and the new technology
of risk-shifting to purchase reinsurance from a private firm or share
the risk of loss through a financial instrument. The government could
eventually encourage or require the insured themselves to go directly
to private markets.
It is not only recent developments in insurance markets that make
private insurance substitution for government programs possible. Some
scholars believe that market-basecl solutions to the problem of bank-
ing panics have developed in the past and could develop again in the
future if government stopped providing deposit insurance (see Bentson
and Kaufman 1995; Calomiris and Mason 1997). Provision of pension
insurance by private firms could also be more feasible as a result of
developments in insurance markets and the experience gained through
the government insurance prograrn (Weaver 1997: 157).
While private markets will not offer every insurance product a
policymaker may want for society, the trend is clearly toward a more
complete array of private risk-bearing options. Policymakers should
therefore regularly assess the viability of private market alternatives to
government insurance. Indeed, the greatest impediment to private risk-
bearing solutions may come from government. In particular, the pub-
lic may not believe that the government would actually allow firms or
households to bear large losses. As noted, previous government sup-
port and the likelihood that governments will provide support in the
future could discourage firms and households from purchasing private
insurance, encourage riskier behavior, and even lead insurers and their
creditors to believe that the government will bail them out if they do
suffer large losses. The steps for establishing credibility discussed in
the previous section would be particularly important for a transition
to more private risk-bearing.
Regulation in Place of Direct Provision
Governments can also expand insurance coverage for its residents by
regulating the private market. Regulation can take the form of price-
setting/rate suppression or requirements that insurers provide certain
types of coverage or that private firms accept all applicants for cover-
age. These steps have significant drawbacks as replacements for a di-
rect insurance provision that has run its course. For one, regulation
may simply shift the government's exposure to loss rather than reduce
it. Some governments provide guarantees to the policyholders of pri-
vate insurance firms and may offer implicit coverage for creditors of
350 RON FELDMAN
large insurance firms. The chance of a claim to the government may
increase if rate suppression or mandatory coverage rules make private
insurance providers more likely to fail. In addition, there are virtually
no sound justifications for price regulation, and the actual manner in
which governments force mandatory coverage reduces the benefits that
such programs might theoretically produce.'5 For some of these very
reasons, there has been a growing call for movement away from inef-
ficient economic regulation of insurance firms."6 Finally, government
could more effectively and transparently achieve social goals through
direct payments or vouchers than through regulation.
These points argue against rate regulation and forced coverage, but
do not preclude other government actions that could indirectly reduce
the need for government insurance provision. On the most general
level, governments can create and enforce legal and financial infra-
structures that have been shown to lead to a growing financial sector.i7
These steps seem to have significant rewards, especially for developing
countries. Part of this basic legal structure could include measures that
address fraud in the provision of financial services, such as insurance,
and perhaps provide for some limited regulation of insurers' safety
and soundness.'8 Governments could also ensure that other policy tools,
such as taxation, do not unnecessarily thwart the development of pri-
vate sector firms. At a later point, the government could determine
whether additional legal rules governing insurance products and firms
such as supplementary disclosure would prove beneficial. In total, the
steps the government can take to increase the depth and scope of pri-
vate insurance markets should prove most beneficial in the shift away
from government provision of insurance.
Notes
1. The World Bank notes, "There is a strong positive correlation between
economic development and the growth of [privatel insurance utilization" (World
Bank 2001).
2. For evidence of the effect of financial resources on the demand for disaster
insurance, see Browne and Hoyt (1998).
3. Schnepf and Heifner (1999) discuss the limited demand for crop insurance
in the United States despite subsidized premiums.
4. Details on estimation methods for federal insurance programs are found
in U.S. GAO (1997: 152-217).
5. For important discussions of regulation, moral hazard, and insurance, see
Ippolito (1989) and Kane (1989).
6. The use of market signals with regard to deposit insurance reform is dis-
cussed in Stern (1999)
Rl:SKS AND RISK MANAGEMENT 351
7. Lawrence H. Summers, testimony before the Committee on Banking
and Financial Services, U.S. House of Representatives, April 23, 1998.
8. See Redburn (1993) for a discussion of budgeting and accrual accounting.
9. For a similar view, see Harrington (2000b). For more discussion on di-
saster risk, see Chapter 20 by Alcira Kreimer in this volume.
10. Standard and Poor's (1999a) characterized recent mergers and acquisi-
tions in the property and casualty insurance market as significant and driven by
factors including the desire for increased scale and globalization.
11. Data from Chicago Board of Trade .
12. A discussion of the many methods producers can use to manage uncer-
tainty can be found in Harwood and others (1999).
13. This discussion is largely based on Froot (1999) and Borden and Sarkar
(1996).
14. A discussion of financial instruments to protect against weather-related
loss is found in WeatherRisk (1999).
15. A detailed analysis of regulation to increase coverage can be found in
Harrington (2000a).
16. For a discussion of the current state and trends in regulation of insurance
firms, see OECD (1998).
17. Levine (2000) discusses the importance of legal structures for economic
and financial growth.
18. See Savage (1999) for a discussion of implementing insurance regula-
tion. Savage argues that financial regulation could help establish public confi-
dence in insurance systems in developing countries.
References
Arrow, Kenneth. 1977. "Insurance, Risk and Resource Allocation." In Kenneth
Arrow, Essays in the Theory of Ri.,k-Bearing. New York: Elsevier.
Bentson, George, and George Kaufman. 1995. "Is the Banking and Pay-
ments System Fragile?" Journal of Financial Services Research 9 (Decem-
ber): 209-40.
Blair, Christine E., and Gary Fissel. 199:1. "A Framework for Analyzing Deposit
Insurance Pricing." FDIC Banking Review (fall): 26.
Borden, Sara, and Asani Sarkar. 1996. "Securitizing Property Catastrophe Risk."
Current Issues in Economics and Finance (August), Federal Reserve Bank of
New York.
Browne, Mark, and Robert Hoyt. 1998. "The Demand for Flood Insurance:
Empirical Evidence." Paper presented at the 25th Seminar of the Geneva
Association, Geneva, September.
Calomiris, Charles, and Joseph Mason. 1997. "Contagion and Bank Failures
during the Great Depression: The Chicago Banking Panic." American Eco-
nomic Review 87: 863-83.
352 RON FELDMAN
Cummins, J. David, Neil A. Doherty, and Anita Lo. 1999. "Can Insurers Pay
for the 'Big One'? Measuring the Capacity of an Insurance Market to Re-
spond to Catastrophic Losses." Wharton Financial Institutions Center Work-
ing Paper 98-11-B. Philadelphia.
Falloon, William. 1999. "AgRisk: Farm Quants Plant a New Model." Risk
(July): 31-33.
FDIC (Federal Deposit Insurance Corporation). 1998. Managing the Crisis: The
FDIC and RTC Experience. Washington D.C.
Froot, Kenneth. 1999. "The Evolving Market for Catastrophic Event Risk."
NBER Working Paper 7287. National Bureau of Economic Research, Cam-
bridge, Mass.
Garcia, Gillian. 1999. "Deposit Insurance: A Survey of Actual and Best Prac-
tices." IMF Working Paper 99-54. International Monetary Fund, Washing-
ton, D.C.
Harrington, Scott E. 2000a. "Insurance Deregulation and the Public Interest."
AEI-Brookings Joint Center for Regulatory Studies, Washington, D.C.
- 2000b. "Rethinking Disaster Policy." Regulation 23: 40-46.
Harwood, Joy, Richard Heifner, Keith Coble, Janet Perry, and Agapi Samwaru.
1999. "Managing Risk in Farming: Concepts, Research, and Analysis." USDA
Agricultural Economic Report 774. Economic Research Service, U.S. De-
partment of Agriculture, Washington, D.C.
Ippolito, Richard. 1989. The Economics of Pension Insurance. Homewood,
Ill.: Irwin.
Kane, Edward. 1989. The S&L Insurance Mess: How Did it Happen? Wash-
ington D.C.: Urban Institute.
Levine, Ross. 2000. "Bank-Based or Market-Based Financial Systems: Which is
Better?" Paper presented at the World Bank Conference on Financial Struc-
ture and Economic Development, Washington, D.C., February 10-11.
OECD (Organisation for Economic Co-operation and Development). 1998.
"Competition and Related Regulation Issues in the Insurance Industry."
Roundtables on Competition Policy 19.
Redburn, E Stevens. 1993. "Measuring Revenue Capacity, Effort and Spend-
ing: How Should the Government Measure Spending? The Uses of Accrual
Accounting. "Public Administration Review 53 (May/June): 228-36.
Savage, Lawrie. 1999. "Re-engineering Insurance Supervision." Policy Research
Working Paper 2024. World Bank, Washington, D.C.
Schnepf, Randy, and Richard Heifner. 1999. "Crop and Revenue Insurance:
Bargain Rates but Still a Hard Sell." Agricultural Outlook (August): 15-19.
Standard and Poor's. 1999a. "Standard and Poor's Maintains a Negative Out-
look for the U.S. Property/Casualty Insurance Industry through 2000." Stan-
dard and Poor's Insurance Rating Service Commentary. April.
. 1999b. "U.S. Reinsurance Pricing Is Expected to Firm in 2000." Stan-
dard and Poor's Insurance Rating Service Commentary.
Stern, Gary. 1999. "Managing Moral Hazard with Market Signals: How Regu-
lation Should Change with Banking." Proceedings of the 35th Annual Con-
RISKS AND RISK MANAGEMENT 353
ference on Bank Structure and Competition. Chicago: Federal Reserve
Bank of Chicago.
- -. 2000. "Designing Credible Policies to Address TBTF and Moral Haz-
ard after Financial Modernization." Speech presented at the George J. Stigler
Center for the Study of the Economy and the State Conference on the Future
of Finance: Globalization, the Internet, and Regulatory Reform, April 2000.
U.S. GAO (General Accounting Office;. 1997. Budgeting for Federal Insurance
Programs. Washington D.C.
WeatherRisk. 1999. A Risk Special Report.
NVeaver, Carolyn. 1997. "Government Guarantees of Private Pension Ben-
efits: Current Problems and Market-Based Solutions." In Sylvester Schieber
and John Shoven, eds., Public Policy Toward Pensions. Cambridge, Mass.:
MIT Press.
Weicher, John. 1994. "The New Structure of the Housing Finance System."
Economic Review (July/August): 47-65.
World Bank. 2001. "Financial Sector: Insurance and Contractual Savings"
.
CHAPTFER i6
Contingenit Liabilities
of the Central Bank:
Analyzing a Possible Fiscal
Risk for Grovernment
Mario I. Blejer
Central Bank of Argentina
Liliana Schumacher
International Monetary Fund
CENTRAL BANKS, NO MATTER HOW independent from the fiscal au-
thorities, may become a source of fiscal risk for the government. Should
the central bank's net worth drop below zero, fiscal authorities will be
called on to act. As it turns out, ithe major risk to the central bank's
financial position arises from its contingent commitments. Although
central banks have frequently undertaken contingent commitments as
part of their regular operation in the economy, it is only recently, through
central bank involvement in the derivatives markets, that these trans-
actions have drawn the attention of policymakers, market participants,
and international agencies. This renewed interest and the fact that
these contingent operations came to light in the aftermath of the 1997
East Asian financial crisis increased the interest in the analytics and
the measurement of these commitments.
The goal of this chapter is twofold. First, it seeks to explore the
rationale for central banks to undertake this type of commitment-in
particular, to intervene in derivatives markets-and to assess the argu-
tnents that are commonly made for and against this sort of involve-
ment. Second, it seeks to suggest a methodology for analyzing the whole
spectrum of central bank contingent liabilities. Specifically, we aggre-
gate all on- and off-balance sheet transactions in a single framework
355
356 BLEJER AND SCHUMACHER
in order to draw meaningful conclusions about the consequences of
these central bank operations for important policy issues such as the
volume of the central bank's available reserves, the potential burden
arising from instability in the banking sector, and the overall solvency
of the central bank.
The chapter is organized as follows. The next section briefly classi-
fies the different types of contingent liabilities that central banks typi-
cally undertake. It is followed by a discussion of the rationale for central
bank operations involving contingent liabilities and of some of the ad-
vantages and disadvantages of these operations. The next two sections
apply general finance principles to the problems of valuation, aggrega-
tion, and measurement of the risk arising from contingent liabilities.
The chapter concludes with a description of some policy implications.
Central Bank Contingent Liabilities: A Classification
The general definition of central bank contingent liabilities coincides
with the standard concept used to classify government contingent li-
abilities in the context of the fiscal accounts. However, some distinc-
tive central bank commitments require a specific conceptual treatment.
Moreover, the concrete quiantitative valuation of certain types of cen-
tral bank obligations requires a special analytical approach.
One point is important to stress at the outset. An analytical distinc-
tion exists between the lack of proper economic accounting of some
assets and liabilities that would result in off-balance sheet items (such as
implicit credit subsidies) and the conceptual and practical consequences
of contingent assets and liabilities-that is, the management of items
that have an uncertain value because their financial implications hinge
on the realization of conditions that depend on future uncertain events
(such as the provision of credit guarantees).
Like government contingent liabilities, central bank contingent li-
abilities can be divided into implicit and explicit ones, depending on
whether they arise from a legal or contractual source. Explicit central
bank contingent liabilities originate in formal statements in regula-
tions or in contracts entered into by the central bank with specific
counterparties. These liabilities can be divided into three types.
The first type is liabilities that arise from formal central bank com-
mitments to supporting the soundness of the banking sector. These
commitments include providing liquidity to individual institutions (the
central bank's role as lender of last resort), as well as providing spe-
cific deposit and other guarantees.
The second type is liabilities created by central bank operations in
non-spot foreign exchange and other financial markets. Specifically,
when the central bank intervenes in derivatives markets, these opera-
tions give rise to potential gains and losses that are contingent both on
the state of the world and on other central bank actions. These types
CONTINGENT LIABILITIES OF THE CENTRAL BANK 357
of operations could be a very important source of contingent liabilities
because they include not only straight intervention in the markets for
forwards, futures, options, and currency and interest rate swaps, but
also monetary operations involving foreign exchange swaps and re-
purchase agreements (repos). The focus of contingent liability analy-
sis, however, is on the use of derivatives as policy tools-that is,
operations designed to influence variables such as the exchange rate or
the interest rate. In particular, currency forwards and options and for-
eign exchange swaps can be used as instruments in the foreign ex-
change market to affect the exchange rate, and repurchase agreements
can be used in the money market: to influence the interest rate. It is
important to distinguish these operations from the use of derivatives
undertaken as part of the central bank's routine management of its
own foreign exchange reserves.
The third type of contingent liabilities is other potential guarantees
to private sector activities, such as guaranteed repayment of directed
credit to selected sectors and export and investment guarantees.
Among the implicit central bank contingent liabilities, the most sa-
lient is the commitment of central banks to ensuring the systemic sol-
vency of the banking (and financial) sector, over and above the explicit
commitment to providing liquidity to individual institutions and to
guaranteeing certain types of deposits or other private sector bank
assets. These commitments would include providing financial cover-
age over and above the legal guarantee scheme and the bailout and
recapitalization of banks and nonbank financial institutions.
It could be argued that certain macroeconomic central bank com-
mitments such as the preservation of a stable exchange rate regime or,
more generally, the attainment and maintenance of price stability also
should be considered, in themselves, implicit contingent liabilities and,
consequently, also should be subject to quantification. While in prin-
ciple this argument would seem consistent with the general frame-
work suggested here, we submit that they should not be part of the
same analytical framework. Our reason: the financial consequences of
the event (that is, deviations from implicit central bank policy targets)
are hard, if impossible, to quantify because they can give rise to nu-
merous types of responses. We postulate, therefore, that only when
specific policy actions can be taken to protect the implicit commit-
ment and they are embedded in legal norms (such as a stock of repo
transactions or forward exchange market operations), the implicit com-
mitment should be considered part of the framework suggested here.
The Rationale for Central Bank
Operations Involving Contingent Liabilities
The rationale for the involvement of the central bank in activities that
result in contingent liabilities varies according to the type of operation.
358 BLEJER AND SCHUMACHER
Much has been written about what motivates central banks to assume
specific commitments to strengthening the soundness of the financial
markets-for example, about the role of the central bank as lender of
last resort or the rationale for deposit insurance. The main arguments
in favor of these institutional devices include the illiquid nature of
banks, together with the potential for systemic risk (Diamond and
Dybvig 1983) and the existence of asymmetries of information and the
protection of the small depositor (Tirole and Dewatripont 1994). Simi-
larly, the issuance of guarantees for private sector activities has been
the subject of abundant inquiries. In general, the view is that such activ-
ity is in fact clearly quasi-fiscal in nature, and therefore its rationale
also is fiscal in nature. In particular, central banks may be induced to
undertake these types of operations to hide undesirable budget out-
comes from public scrutiny.
What has been much less well researched is the rationale for central
bank intervention in derivatives markets. Therefore, in what follows
we develop in more detail the main arguments that could be used to
justify the implementation of these operations. In general terms, it is
possible to assert that central banks tend to engage in derivatives op-
erations for the following reasons: (a) to provide additionality to in-
complete or illiquid markets; (b) to defend a fixed exchange rate regime
or an exchange rate band; (c) to alleviate the conflict between the
defense of an exchange rate regime and the stability of the financial
system; (d) to act as an automatic stabilizer of the foreign exchange
market; and (e) to act as an alternative instrument for monetary man-
agement under some specific circumstances.
Providing Additionality to Incomplete or Illiquid Markets
In many countries, the derivatives market is not deep enough and there-
fore does not provide the range of instruments needed for appropriate
hedging and risk management. In these circumstances, the rate of growth
of the underlying market would tend to be lower than desired, and the
central bank's provision of additional innovative instruments and li-
quidity could be seen as a means of developing both the spot and the
derivatives markets and of eliminating, or at least smoothing, volatil-
ity in the spot market.
Defending a Fixed Exchange Rate Regime
or an Exchange Rate Band
A central bank's engagement in derivatives operations, including for-
ward and swap operations, has also been used repeatedly to reduce
exchange rate fluctuations and, more specifically, to protect a fixed
exchange rate regime or an exchange rate band. Central banks have
CONTINGENT LIABILITIES OF THE CENTRAL BANK 359
two important reasons to prefer this form of intervention over inter-
vention in the spot market. First, derivatives allow defense of the ex-
change rate without an immediate use of foreign exchange reserves
and without an impact on the money supply. Such a defense is there-
fore similar to sterilized intervention, but it has ex ante an opposite-
that is, a positive-fiscal outcome.
Second, intervention in the derivatives market is an efficient way of
releasing some of the pressure that dealers and banks may exercise on
the foreign exchange spot market at times of particularly heavy specu-
lative stress. During normal times, banks and dealers can easily find
counterparts for hedging their foreign exchange operations. But at times
of uniform expectations, when there is a widespread market belief
that the exchange rate would likely change in one particular direction,
they may find it difficult to hedge in the derivatives market. Clearly,
when the generalized expectation is that the domestic currency is bound
to depreciate (or will be devalued), market participants will seek to
shorten the domestic currency, buying put options or taking short for-
ward positions. However, banks and other foreign exchange dealers
will take the long side of the market only if they can hedge their expo-
sure. But hedging in these circumstances could mean facing some im-
pediments. In the absence of agents that need to hold a natural long
position in the domestic currency, banks and dealers may be able to
hedge only synthetically. In a syntlhetic hedge, dealers aim to replicate,
with an opposite sign, the cash flows that emerge from the derivatives
transactions to which they have committed. There are two cash flows
to hedge: (a) a long position in the weak currency equal to the total
amount of their forward commitments plus their put options commit-
ments times the probability that the put options will be exercised (the
hedge ratio); and (b) a short position in the strong currency for an
amount equal to the long position times the forward rate. As can be
seen easily, these two cash flows, with an opposite sign, can be easily
replicated in the spot market by, for example, taking a loan in the
weak currency and opening a deposit in the strong currency.
While hedging synthetically coLild be, from the point of view of risk
management, satisfactory for the dealers, it may create a problematic
situation from the central bank's perspective that provides the motiva-
tions for stepping into the derivatives markets. Several of the major
concerns that central banks have with synthetic hedging are of par-
ticular interest. The first one relates to the impact of synthetic hedging
on the foreign exchange spot marlcet. As for the second, synthetic hedg-
ing can distort the response of agents to increases in the domestic in-
terest rates.
Impact of Synthetic Hedging on the Spot Market. Clearly, during
times of turbulence in the foreign exchange market, it is reasonable to
360 BLEJER AND SCHUMACHER
expect that a central bank committed to defending a peg would try to
avoid additional selling pressures on the domestic currency. However,
a dealer hedging synthetically will tend to do precisely that by short-
selling the domestic currency and using the proceeds to buy foreign
currency. This would indeed put additional pressure on the spot mar-
ket that can only be released by increasing the liquidity of the deriva-
tives market. The central bank's willingness to sell forward contracts
or to write put options is therefore intended to provide dealers with
appropriate hedges, removing in this manner the additional pressure
that synthetic hedging exerts on the spot market. In other words, cen-
tral banks may intervene in the derivatives markets to prevent specula-
tion from spilling over immediately into the cash/spot markets.'
Synthetic Hedging and Interest Rates. Central banks also are inter-
ested in containing synthetic hedging, because it is well recognized
that these types of operations tend to disrupt the typical central bank
defense of a pegged foreign exchange system. Garber and Spencer (1995)
show that an increase in the domestic interest rate results in an in-
crease in the hedge ratio (that is, in the inverse of the ratio between the
number of puts and the units of foreign currency necessary to hedge
those puts).2 This situation means that an increase in interest rates
raises the demand for foreign exchange in the spot market on the part
of the synthetically hedged agents. Therefore, whether a higher do-
mestic interest rate will succeed in reducing speculation, by inducing
market participants to continue to hold the domestic currency, de-
pends on the relative importance of market agents that are syntheti-
cally hedging versus the rest of the market participants that are caught
in the interest rate squeeze.
Alleviating the Conflict between the Defense of an Exchange
Rate Regime and the Stability of the Financial System
The conflict may arise when expectations of devaluation accelerate,
provoking a surge in capital outflows. Given the importance of banks
in the intermediation of capital flows, the intensifying pressures in the
foreign exchange market could result in serious liquidity problems for
the banking system. These problems might be further complicated by
the fact that the increase in the expected rate of devaluation will lead
to higher domestic interest rates. The central bank, in its role as lender
of last resort, would tend to provide liquidity loans to banks that have
experienced losses because of the higher interest rates' in the interbank
market or because of the fire sale of bank assets when the interbank
market dries up. However, because the lender of last resort cannot
discriminate among banks with a legitimate liquidity problem of this
sort and other banks that may attempt to borrow from the central
CONTINGENT LIABILITIES OF THE CENTRAL BANK 361
bank in order to hedge or to speculate in the foreign exchange market,
the provision of liquidity by the central bank may end up feeding the
short-selling of domestic currency, increasing in this way the pressure
on the foreign exchange market. In other words, central banks may
prefer to step into illiquid derivatives markets in order to provide banks
and dealers with an alternative way to speculate, through forwards or
options, without exerting further pressures on the foreign exchange
spot market.
Acting as an Automatic Stabilizer of the
Foreign Exchange Market
The money (American or Europeatn) put option written by the central
bank on the reserve currency serves as an automatic stabilizer of the
foreign exchange market.4 When there is an inflow of foreign currency
and the exchange rate appreciates, the put buyers exercise the option
and deliver the reserve foreign currency to the central bank. In addi-
tion, to stabilize the market, this mechanism allows the central bank
to accumulate reserves precisely when the foreign currency weakens
and avoids the negative signaling effect of open central bank interven-
tion in the spot market. The foreign currency reserves accumulated
during such episodes of appreciation can be used to reduce outstand-
ing foreign currency liabilities or, when there are pressures on the ex-
change rate to depreciate, to provide the additional supply required by
the market.
Acting as an Alternative Instrument for Monetary
Management under Some Specific Circumstances
Some arguments have also been voiced justifying the use of foreign
exchange swaps as an instrument for domestic liquidity management.
In particular, in countries running fiscal surpluses or where the out-
standing stock of public-including central bank-debt is low, central
banks may find it expensive (or disruptive) to inject domestic liquidity
using repos based on domestic bonds. For that reason, some countries
have resorted to the use of foreign exchange swaps, which are basi-
cally repos in foreign exchange currency, as a temporary mechanism
for managing domestic liquidity.'; These operations do not change the
level of net international reserves, but they do temporarily increase
domestic liquidity.
Conclusion
There is therefore a positive policy rationale for central banks to accu-
mulate contingent liabilities through intervention in the derivatives
362 BLEJER AND SCHUMACHER
market. However, these operations carry significant risks. A prolifera-
tion of contingent liabilities distorts the financial statements of central
banks. The solvency of the central bank also can be compromised by
potential losses. Moreover, intervention in the derivatives market may
have serious drawbacks. They could be difficult to support when these
markets are very thin, and they are bound to result in a loss of the
informational content provided by these markets. In addition, the ability
to intervene in the derivatives markets at a low cost, and the lack of a
material constraint to the intervention levels, could lead to a potential
postponement of important policy decisions.
Valuation and Aggregation
One of the main problems posed by contingent liabilities is the issue of
how to record them and, in particular, how to aggregate these contin-
gent liabilities (which are by definition off-balance sheet) with the on-
balance sheet central bank transactions for valuation purposes.
Valuation
We propose here using a portfolio approach to all central bank trans-
actions, as the only way in which both on- and off-balance sheet trans-
actions can be aggregated and as a way of providing some meaningful
information on variables such as the central bank's available reserves,
the potential burden caused by preserving banking sector stability, and
the overall solvency of the central bank. In a portfolio approach, trans-
actions are aggregated according to their sign (short or long) and their
value. The theory of financial instruments provides the tools needed
for pricing these transactions, and therefore the procedures are not
reviewed here in detail.6 However, for illustration, and because some
operations are particularly relevant for central banks, we discuss here
two specific cases: the value of a currency forward contract and the
value of a deposit insurance commitment. Then, we discuss some ex-
amples of proper aggregation in the central bank portfolio.
The Value of a Currency Forward Contract. The economic value of
a currency forward contract can be derived from covered interest rate
parity
(16.1) e-'F,, r = S,,e-rl
where F., T is the forward rate for the foreign currency, for maturity T,
as of the day the contract is signed; S. is the spot rate for the foreign
currency as of the day of the contract; r,,, is the domestic interest rate
CONTINGENT LIABILITIES OF THEI CENTRAL BANK 363
as of the day of the contract; and r, is the foreign interest rate as of the
day of the contract.
The meaning of (16.1 ) is that according to the covered interest rate
parity, a forward contract can be viewed as two zero coupon bonds.
The left-hand sign of the equality represents a zero coupon bond de-
nominated in domestic currency, with a face value equal to the for-
ward rate of the foreign currency for maturity T, as of the day of the
contract, and with maturity T. The value of this zero is found by dis-
counting the forward rate by the domestic interest rate. The right-
hand side of the equality represents a zero coupon bond denominated
in foreign currency, with a face value equal to one unit of the foreign
currency (converted into domestic currency using the spot price of the
day of the contract). The value of t:his zero is found by discounting the
unit of the foreign currency by the foreign interest rate.
We can now rearrange (16.1) to find the value of the forward con-
tract, any day after the contract was signed, as the difference between
the value of the two zeros-that is,
(16.2) e-rd'F. r- e-'.'S, O.
Equation 16.1 revealed that the value of a forward, as of the day of
the contract, is zero. But for any other day after the original date, the
value of the forward contract in the book of the central bank can be
different from zero and can be cletermined by calculating equation
16.2 with information that is generally readily available.
Deposit Insurance. Following Merton (1977), deposit insurance can
be seen as the equivalent of a put option held by the banks and written
by the central bank on each unit of bank assets, with a strike price
equal to the value of bank-insured, debts. The equivalence goes as fol-
lows: if banks become insolvent, the value of bank assets by definition
is lower than the value of bank clebts. Given limited liability for the
shareholders, bank debts will suffer the full loss. But in the presence of
deposit insurance, banks have the ability to "exercise the put option"-
that is, they "sell" their assets (the underlying asset of the put) to the
central bank and they get in exchange an amount equal to the face
value of the insured liabilities (the strike price), which is used to pay
for bank-insured debts.
Following this equivalence, the value of deposit insurance is
(16.3) G(T) = TBe-r O(X2)- V0(xJ)
where X, = (log(B/V) - [r + (2/2 )T] /1.T; X, = X2 + 84T; B = face
value of bank liabilities (exercise price); V = value of the banks' assets;
o = volatility of the banks' assets; T = maturity of bank liabilities; and
364 BLEJER AND SCHUMACHER
p (.) = cumulative probability distribution function for a standardized
normal variable-that is, the probability that such a variable will be
less than (.).
As Merton (1977) indicates, (16.3) also can be applied to valuing a
government guarantee of loans made to private (financial or nonfinan-
cial) corporations.
Aggregation
Based on an economic valuation of the off-balance sheet contingent
positions of the central bank, as described above, all central bank trans-
actions can be aggregated. We now illustrate this procedure using eight
hypothetical central bank portfolios, which are described in Tables
16.1 and 16.2. Table 16.1 contains the basic information used to con-
struct the on-balance and the off-balance sheet accounts of the assumed
central bank, together with the prices and interest rates used to value
the positions. We chose the British pound (GBP) as the domestic cur-
rency and the German mark (DEM) as the foreign currency. Table 16.2
is an estimate of the portfolio values of the central bank positions, and
Table 16.1 Hypothetical Central Bank Portfolio: Data
On-balance sheet items:
Reserves are invested in a one-year zero coupon bond denominated in DEM.
Face value of reserves: DEM] 18
Domestic debt: GBP30
The central bank holds a loan against the Treasury for GBP30.
Monetary base: GBPIO
Off-balance sheet items:
The central bank is short DEM30 in the forward market. The maturity of
the forward is one year.
For the calculation of the deposit insurance guarantee, the following data
were used:
o Bank leverage (ratio of bank liabilities to bank assets) = 0.8.
o Volatility of bank assets (measured by the standard deviation of
annual changes of the value of bank assets) = 0.5.
o Xl = 0.5964.
o Value of one put = 0.0515.
Prices
Spot exchange rate I DEM = GBPO.338.
Interest rate (GBP) = 0.1000.
Interest rate (DEM) = 0.0839.
Forward rate I DEM = GBPO.333.
Source: Compiled by the authors.
CONTINGENT LIABILITIES OF THE. CENTRAL BANK 365
Table 16.2 Hypothetical Central Bank Portfolio: Values
Positions in the central bank portfolio Value
1. Base case
Foreign exchange reserves GBP36.9
Domestic debt GBP27.27
Long leg of forward
Loan to Treasury 0
Monetary base GBPI00
Short leg of forward
Financial sector guarantee
Value of portfolio -35.84
2. Base case + forward
Foreign exchange reserves GBP36.9
Domestic debt GBP27.27
Long leg of forward GBP9.09
Loan to Treasury 0
Monetary base GBP100
Short leg of forward GBP9.36
Financial sector guarantee
Value of portfolio -36.21
3. Base case + guarantee
Foreign exchange reserves GBP36.9
Domestic debt GBP27.27
Long leg of forward
Loan to Treasury 0
Monetary base GBP100
Short leg of forward
Financial sector guarantee (bank assets: GBP400) -20.62
Value of portfolio -56.39
4. Base case + 2* guarantee
Foreign exchange reserves GBP36.9
Domestic debt GBP27.27
Long leg of forward
Loan to Treasury 0
Monetary base GBP100
Short leg of forward
Financial sector guarantee (bank assets: GBP800) -41.23
Value of portfolio -76.95
S. Base case + 3* guarantee
Foreign exchange reserves GBP36.9
Domestic debt GBP27.27
Long leg of forward
Loan to Treasury 0
Monetary base GBP100
Short leg of forward
Financial sector guarantee (bank assets: GBP1,200) -61.85
Value of portfolio -97.5
(Table continues on the folloiving page.)
366 BLEJER AND SCHUMACHER
Table 16.2 (continued)
Positions in the central bank portfolio Value
6. Base case + guarantee + forward
Foreign exchange reserves GBP36.9
Domestic debt GBP27.27
Long leg of forward 9.09
Loan to Treasury 0
Monetary base GBP100
Short leg of forward 9.36
Financial sector guarantee (bank assets: GBP400) 20.62
Value of portfolio -56.76
Source: Compiled by the authors.
from these economic-rather than accounting-values the true valua-
tion of the central bank equity is calculated. The economic values of
the balance sheet items were calculated by converting all notional
amounts into British pounds and discounting these amounts by the
relevant interest rate. For the forward positions and the deposit insur-
ance, we used the formulas just described.
The base case or Case I is the simplest one, where the central bank
issues monetary base in exchange for foreign reserves or domestic debt.
In addition, there is a loan to the Treasury whose economic value is
assumed to be zero.7 In Case 2, the central bank, in addition to its
activities in Case 1, is active in the forward market for foreign cur-
rency. The central bank intervenes in the forward market in only one
direction: it buys domestic currency forward. In Case 3, the central
bank also provides deposit insurance to the banking system. Cases 4
and 5 are identical to Case 3, except for the size of the banking system:
in Case 4, the assets of the banking sector are twice as large as in Case
3, and in Case 5, the bank assets are three times those in Case 3.
Finally, Case 6 combines intervention in the forward market for for-
eign currency with deposit insurance.
The value of the portfolios, when properly accounted for the eco-
nomic value of assets and liabilities, is negative in all cases and be-
comes more negative as contingent liabilities are added to the portfolio.
The fact that the central bank's equity is negative is not an anomaly. In
fact, it is interesting to note that a central bank with negative eco-
nomic equity is a likely outcome, because the usual accounting ap-
proach to the value of central bank activities does not consider the
economic value of explicit or implicit commitments such as a deposit
insurance guarantee and a forward contract. In addition, most central
bank assets are registered at nominal values and are not economically
valued (for example, the probability of repayment of certain loans, the
CONTINGENT LIABILITIES OF THE, CENTRAL BANK 367
time value of domestic and foreign assets, and the credit risk of re-
serves invested abroad are not considered). The standard approach
therefore usually results in an overestimation of central bank equity.
Using a methodology based on economic valuation rather than on
nominal accounts, we also can est.imate the value of specific compo-
nents of the portfolio. For example, for Case 2, we could compute the
value of reserves net of forward contracts (in domestic currency) as
36.90 + 9.09 - 9.36 = 36.36. It is easy to see in this example that as the
domestic currency (the pound) depreciates in the spot market, the eco-
nomic value of the reserves net of forward contracts decreases. The
same happens if the interest rate in1 pounds goes up. This is so because
the value of the long leg of the forward that is worth 9.09 in this
example decreases as the pound depreciates and the interest rate in
pounds increases.
In another example, the unit value (per unit of bank assets) of the
deposit insurance is GBPO.0515 (see Table 16.1). Note, however, that the
value of the insurance depends on the size of bank assets. When bank
assets are GBP400 (Case 3 in Table 16.2), the insurance is worth GBP20.62;
when bank assets are GBP800, the insurance is worth GBP41.23; and
when bank assets are GBP1,200, the insurance is worth GBP61.85.
The risk of the bank system also has an influence on the value of the
contingent liability. Consider the case of a rapid deterioration in the
cluality of bank loans. The value c,f bank assets, adjusted by risk, goes
clown and the leverage ratio goes up, making the value of the central
bank contingent liability higher.
Concluding Remarks
Central banks perform a large variety of operations that give rise to
contingent liabilities, defined as financial commitments that are trig-
gered by the occurrence of an event whose realization is uncertain.
Because these operations cover a wide array of areas, the motivation
for central banks to engage in this type of activity stems from a myriad
of reasons. In this chapter, we provide a taxonomy for classifying
these operations and elaborate on- their analytical aspects, as well as
on the operational motivations that induce central banks to utilize
these instruments.
We conclude that although some of a central bank's contingent li-
abilities arise from anomalous circumstances,8 some positive reasons
explain their apparent popularity. Some of these positive implications
are well recognized-particularly those that arise from the central bank's
role in guaranteeing the stability of the banking sector-but the con-
structive aspects of central banks' involvement in derivatives markets
are less understood. We attempt here to provide a broader and more
368 BLEJER AND SCHUMACHER
positive perspective, but, at the same time, we must stress that, be-
cause most of the operations that give rise to contingent liabilities also
tend to be off-balance sheet, they reduce the transparency of central
bank accounts. This, in turn, may result in serious problems regarding
the proper assessment of the financial position of the monetary au-
thority and, by implication, of the overall macroeconomic conditions
of the country. We suggest, therefore, that a comprehensive portfolio
approach that values, in an economic rather than purely accounting
sense, all off- and on-balance sheet assets and liabilities of the central
bank be adopted. We provide some examples of how this could be
done, particularly for some contingent liabilities that are characteris-
tic of central banks.
Note, however, that even though proper valuation and aggregation
of central bank financial positions would solve some of the transpar-
ency problems posed by contingent liabilities, their presence in the
central bank portfolio also would tend to increase financial risks. In
addition to reducing, ceteris paribus, the net equity of the central bank,
as shown in our illustrative simulations, formal risk indicators would
tend to rise in tandem with the volume of this type of liabilities. It
would indeed be a useful research endeavor to attempt a full quantifi-
cation of these effects, using available central bank information.9
Notes
1. Examples of such interventions are the Bank of Spain intervention in
the options (put) market during the 1992-93 European exchange rate mecha-
nism crisis and the Bank of Thailand's sale of forward contracts in 1997.
2. This also can be interpreted as the probability that the put will be
exercised.
3. When banks have positive duration gaps, an increase in interest rates
will lead to bank losses.
4. This was the case in the stabilization scheme adopted by Mexico in July
1996.
S. The Reserve Bank of Australia has resorted to this mechanism with
relative frequency. For example, the need to increase liquidity arising from
the Y2K problem led to the doubling of the stock of outstanding forward
obligations, matched by a similar increase in the total holding of official
reserve assets.
6. One of the many textbooks that have addressed this topic is Hull (1999).
7. Under the assumption that the Treasury will not repay the loan, the
discount factor is infinite.
8. Such as those that simply reflect quasi-fiscal operations transferred
from the government budget to the central bank for purely political or "cos-
metic" reasons.
CONTINGENT LIABILITIES OF THI, CENTRAL BANK 369
9. One possibility is to utilize risk measurement methodologies such as
value at risk. For a framework of this type of application in the context of
central bank portfolios, see Blejer and Schumacher (1999).
References
Blejer, Mario I., and Liliana Schumacher. 1999. "Central Bank Vulnerability
and the Value of Its Commitments: A VaR Approach." Journal of Risk
(fall).
IDiamond, Douglas W., and Philip H. Dybvig. 1983. "Bank Runs, Deposit
Insurance and Liquidity." Journal of Political Economy 91: 401-19.
Garber, Peter M., and Michael G. Spencer. 1995. "Foreign Exchange Hedg-
ing and the Interest Rate Defense." IMF Staff Papers 42: 490-516.
HIull, John. 1999. Options, Futures and Other Derivative Securities.
Englewood Cliffs, N.J.: Prentice H-all.
Merton, Robert. 1977. "An Analytic: Derivation of the Cost of Deposit In-
surance and Loan Guarantees. An Application of Modern Option Pricing
Theory." Journal of Banking and Finance 1: 3-11.
Tirole, Jean, and Mathias Dewatripont. 1994. The Prudential Regulation of
Banks. Cambridge, Mass.: MIT Press.
Practice
CHAPTER 17
Contingent Liabilities
in Infrastructure:
Lessons from the
East Asian Financial Crisis
Ashoka Mody*
International Monetary Fund
FOR EAST ASIAN GOVERNMENTS in the midst of financial crisis in
1997 and 1998, the transformation of their contingent liabilities into
immediate obligations proved to be an additional blow in an already
challenging situation.' In all crisis countries (Indonesia, Republic of
Korea, Malaysia, and Thailand), the banking sector was the major
source of such liabilities. However, except in Korea, infrastructure
projects also added to the fiscal stress. Specifically, in so-called public-
private partnerships, governments had contingent contractual obliga-
tions-and these became due as the crisis worsened. With ongoing
economic recovery, the fiscal pressure from these obligations will likely
decline. The pressures will remain, however, where the problems
stemmed from inadequate project: design and ineffective sector strat-
egy and regulation.
This chapter draws on the experience in Indonesia and Malaysia,
both of which adopted a distinctively Asian style of infrastructure
privatization.2 Indeed, throughout: East Asia, governments sought pri-
vate capital and private management skills, yet they also provided
contractual commitments to enhance the financial attractiveness of
the projects. Some of the governmnent commitments were in the form
"The author was with the World Bank when he wrote this article; he is currently with
the IMF.
373
374 ASHOKA MODY
of cash subsidies, but most were contingent in nature, promising, for
example, to top up revenues if they fell below a threshold or promising
to honor a pricing formula.
The principal guarantees provided by the Philippines government,
summarized in Table 17.1, represent the types of commitments made
by governments throughout the region. The commitments were based
on two premises. First, the transition from government infrastructure
monopoly to multiple private infrastructure providers requires signifi-
cant investment in regulatory capacity. Second, because such capacity
cannot be built overnight, contractually specified public-private part-
nerships are necessary intermediate steps in a rapid infrastructure de-
velopment strategy. The World Bank's 1994 World Development Report
endorsed the Asian approach as an appropriate transitional strategy
(World Bank 1994). Transport and power projects were identified as
especially suitable for applying the approach.
Based on the Indonesian and Malaysian experiences, this chapter
argues that contingent liabilities are not intrinsically associated with
privatized infrastructure. Their manifestation in East Asia reflects a
specific privatization strategy undertaken in a period of rapid growth
Table 17.1 Main Sources of Infrastructure Contingent
Liabilities, the Philippines
Item gtiaranteed Cost Sector
Buyout clause or Buyout or termination Power and transport
termination price
Force majeure Buyout or termination Power, transport,
price and water
Minimum revenue Payment obligation to Power and transport
("Take-or-pay" meet the minimum
contract for the revenue threshold
power sector)
Toll changes; aut- Costs of inability to Transport
matic toll adjust- implement toll ad-
ment formula justments
Assumption of "old" Cost of principal and Water
(preprivatization) intcrest
loans being paid by
concessionaire
Loser of appeal to pay Cost of appeals process Water
total cost of appeals
process for both parties
Souirce: Reside (2000).
CONTINGENT LIABILITIES IN INFRASTRUCTURE 375
with the objective of rapid new investment in infrastructure. However,
even where justified, transitional sr.ructures created with government's
direct support and contingent commitments should give way to com-
petitive infrastructure provision with greater risk shifted to the private
sector (World Bank 1994). This has not yet happened in East Asia.
Latin American economies-especially Argentina and Chile-demon-
strate the practical implementation of privatization without the gov-
ernment acquiring contingent liabilities.
The rest of this chapter is organized as follows. The sources of con-
tingent liabilities are discussed first for the transportation sector, then
for the power sector, and briefly for the water and sanitation sector.
Each section critiques the sector strategy pursued and outlines alterna-
tives based on international experience. The chapter concludes by em-
phasizing that it would be a mistake to view the postcrisis realization of
contingent liabilities as mainly due to the crisis and urges a systemic
reform that can both improve efficiency and lower government risks.
Transportation
In the mid-1990s, transportation projects dominated private infrastruc-
ture projects in East Asia, both in numbers and especially in dollar
outlays (World Bank 1994). These projects were typically built and
operated by the private sector, with the assets to be transferred to the
government after a contractually agreed on period. Several legal varia-
tions of the build-operate-transfer model exist, but these are not eco-
rnomically significant. Private debt and equity, supported by specific
government financial support, financed the projects. For transporta-
tion projects, governments provided some direct subsidies, often in the
form of rights to land. More controversial was support that guaran-
teed minimum revenues to the project sponsors.
The East Asian crisis notwithstanding, few countries worldwide have
succeeded in using a stand-alone build-operate-transfer model to de-
liver a significant transportation network. Especially in the early stages
of implementation, demand projections have proven to be overly opti-
tnistic and cost overruns have proven to be endemic. A major toll road
program in Mexico suffered from both problems. In Bangkok, Thai-
land, private expressway and urban rail financing has been scaled back,
because the government and private concessionaires failed to agree on
and adhere to contractual terms. Even on heavily traveled roads, such
as the one from Guangzhou to Shenzhen in southern China, cost over-
runs have led to unanticipated financing requirements. Overall, the
number of private transportation projects has fallen sharply. In 1998,
in the aftermath of the East Asian crisis, only one new toll road-in
Croatia-was brought to successful financial closure in non-OECD
(Organisation for Economic Co-operation and Development) countries.
376 ASHOKA MODY
Malaysia
In Malaysia, contingent liabilities have arisen principally from two
Kuala Lumpur light rail projects and an extensive toll road program.
The light rail projects have suffered much lower-than-expected de-
mand. By October 1998, concessions were signed for 26 expressway
and toll bridge projects. Of these, 12 projects were open to traffic, 6
were under construction, and the remaining 8 were under negotiation.
The full program was expected to create about 1,700 kilometers of
toll roads. Although the roads under the toll road program represent
only about 2 percent of the total length of the country's road network,
they are important economic arteries. Some roads constructed earlier,
including the major North-South Expressway, enjoyed initial financial
success. Problems have arisen in part because of the crisis-related eco-
nomic downturn that reduced demand, especially for new highways,
and also affected users' willingness to pay tolls. However, the long-
term economics of the new generation of toll roads is suspect.
Because a comprehensive account of the government's contingent
liabilities is not possible, estimates were obtained for three transporta-
tion projects essentially underwritten by the government. Failure by
the project to make its commercial debt payment required a declara-
tion of a default and the takeover by the government of the commer-
cial debt and therefore of the project. The government consequently
had significant financial exposure to these projects (Figure 17.1).
A Monte Carlo simulation approach was used to estimate the
government's exposure. Under this approach, the stochastic (uncer-
tain) cash flows are projected into the future, and 2,500 scenarios are
run. Under some scenarios, the project is economically viable and no
default occurs. Under other scenarios, the project defaults on its com-
mercial debt and is taken over by the government. The estimated ex-
pected cost to the government is the average cost over all the
scenarios-or the cost the government can expect to pay out on aver-
age. However, if some unfavorable factors coincide, such as low rider-
ship and higher-than-anticipated costs, the payout can be larger. The
unexpected cost is defined as the amount paid out in the 95th percen-
tile case (that is, there is only a 5 percent chance that such an eventu-
ality would occur). In addition to these costs arising out of project
default, the government is committed to subsidies, such as low interest
rate loans, that would have to be paid.
As Figure 17.1 shows, the government's financial obligations in trans-
portation projects are highly sensitive to the assumption on ridership/
traffic flows. For the light rail projects, traffic projections had been
steadily revised downward after initiation of the projects. Each revi-
sion raised the costs to the government substantially. Since the 1997
crisis in East Asia, ridership has been even lower than the latest precrisis
CONTINGENT LIABILITIES IN INFRASTRUCTURE 377
Figure 17.1. Monte Carlo Estimates of Contingent Exposure
to Transport Projects in Malaysia
_ Cost at 95th percentile
[w Cost at 50th percentile
O Mean cost to government
Light Rail #1
100% of projected revenue realized
70% of projected revenue realized
50% of projected revenue realized
with gradual recovery
Light Rail #2
100% of projected revenue realized
70% of projected revenue realized
50% of projected revenue realized
with gradual recovery
Dedicated Highway Project
100% of projected revenue realized E
70_ of projected revenue realized
50% of projected revenue realized
with gradual recovery
-20 0 20 40 60 80 100
Government exposure as a
percent of project cost
projections. If ridership is about half of the most recent projections,
then the expected costs to the government for one of the projects could
be as high as about two-thirds of project costs. The dedicated highway
project is performing better than the light rail projects. Assuming 50
percent less traffic than projected by the project and a gradual recov-
ery after the crisis, the average cost to the government will be around
10 percent of the total project.3
The estimates suggest that about half of the expected postcrisis costs
arise from the optimistic projections made when the projects were
initiated and the other half are the result of the downturn. These esti-
mates do not take into account obligations from the freezing of tolls
378 ASHOKA MODY
on certain highways, which have created significant payments by the
government to the project sponsors (Box 17.1).
Indonesia
In Indonesia, the government's obligations for toll road projects are
more modest than those in Malaysia and also are more modest than
those in Indonesia's own power sector. This is the case for two rea-
sons. First, many of the roads carry intraurban traffic, which is rela-
tively well established and steady. These roads, therefore, have not
been seriously affected by the economic crisis. Second, the govern-
ment has few explicit guarantees, unlike in Malaysia where typically
the government must pay the outstanding debt when a project de-
faults on its loan and take over the project. However, the full extent
of the obligations may be larger than is apparent. Much of the fi-
nancing of the toll roads came from domestic banks. The nonperfor-
mance of the toll roads would show in the nonperforming portfolio
of the banks, which in turn are backed by the government. Thus
there may actually be more exposure to the government from the toll
road projects than is evident.4
Lessons for Private Transport Concessions
Because individual toll road and rail projects rarely generate positive
cash flows in their early years, a possible approach lies in creating
Box 17.1 From Contingent to Real Liabilities:
Malaysia's North-South Highway (PLUS)
Under the 1986 concession agreement for the North-South Highway,
automatic toll increases equal the greater of 6 percent or the annual
increase in the consumer price index. If the eligible toll increase is
denied, the concessionaire must be compensated an equivalent amount.
Thus permitting a formula-driven toll rate increase creates a contin-
gent liability for the government that grows with each additional year
the eligible increase is denied. Since 1996, the government has denied
the full increase permitted under the concession agreement. The gov-
ernment budgets for this expense, which for the PLUS concession was
a total of MYR161 million for 1996 and 1997 and which increased
sharply to MYR145 million for 1998 (because of the cumulative de-
nial of rate increases for successive years).
CONTINGENT LIABILITIES IN INFRASTRUCTURE 379
regional transport utilities that are able to cross-subsidize the different
transportation segments within their jurisdiction. Such cross-subsidi-
zation could be justified on the basis of network externalities. In China,
this model has been piloted by the provincial communications minis-
tries in toll highway development. In several provinces, provincial toll
road companies securitize existing toll road assets on the domestic and
Flong Kong stock exchanges, using the proceeds to finance new toll
highway development.
Lessons from France and Spain, which have among the more exten-
sive toll road programs, also may be relevant. In particular, they have
experimented with private toll road companies that operate multiple
roads (Gomez-Ibanez and Meyer 1992). In France, three of the four
toll road companies went bankrupt in the 1970s, after the energy cri-
sis, and were taken over by the government. The one remaining pri-
vate company, which was the largest, has maintained efficient
operations. Since then, France has used a system of cross-subsidization
across companies to assist the investment programs in the underdevel-
oped regions. Spain has dispersed its toll road holdings among about a
clozen or so companies. The impact of the 1970s energy crisis also was
felt in Spain but with a less devastating effect. The government took
over some of the smaller companies; however, the bulk of the toll road
network continues to be in private hands. Spain has avoided cross-
subsidization of companies on the grounds that it reduces the incen-
tives for improved efficiency. In both France and Spain, initial tolls
wvere set at different levels, but the subsequent increases were based on
a common formula. For example, in France the tariff increase allowed
is 95 percent of the increase in the consumer price index.
Lessons from the Metropolitan Transportation Authority (MTA) in
New York, an example of a relatively successful transport utility, may
be relevant for urban public transport. The MTA operates the subway,
bus, and commuter railroad systems, which run chronic operating defi-
cits, and also owns the Triborough Bridge and Tunnel Authority (TBTA).
TBTA generates operating profits well in excess of half a billion U.S.
dollars a year, which subsidize transit operations or enable infrastruc-
ture investments. The experience highlights the importance of coordi-
nating the different transport modles.
Lessons from the international experience therefore caution that
the objective of speeding up the development of urban and intercity
city transportation networks through private capital may be untenable
except in select circumstances. As such, while the economic crisis, and
the consequent fall in ridership and traffic, undoubtedly created addi-
tional fiscal stress in Malaysia (and in Thailand), the very basis of a
strategy of accelerated development that does not take into account
the integrated nature of transport: networks must be reexamined.
380 ASHOKA MODY
Power Sector
Government contingent obligations in the power sector arise mainly
through long-term "take-or-pay" contracts with independent power
producers (IPPs). Under these contracts, the government-owned power
utility makes a commitment to buy ("take") a minimum amount of
power at a prespecified price or pay for that contracted power in any
case. Typically, the power utility is financially strained, and so the
utility's obligation is backed by the sovereign government, creating a
government contingent liability. In developing countries, the IPP strat-
egy was pioneered in the early 1990s by the Philippines (World Bank
1994). The Philippine government also waived sovereign immunity
against claims made by international creditors.
The IPP strategy is attractive because it facilitates the rapid installa-
tion of power generation capacity at no immediate cost to the govern-
ment. The government takes on a contingent obligation, which may
never be called, especially if the high economic growth projections
materialize. The cost of not adding to capacity can be large, because
the lack of power can constrain growth. Following the Philippine ex-
ample, other countries contracted with IPPs, especially Indonesia, Pa-
kistan, and Thailand. But, except in the Philippines, disagreements
and contract renegotiations have been a serious problem.
Malaysia
In Malaysia, the obligations for IPPs are not borne directly by the
government but by Tenaga Nasional Berhad (TNB), the partially priva-
tized power utility.5 These obligations arise in the context of long-
term take-or-pay contracts with IPPs. If the exchange rate is allowed
to float, the depreciation also can increase the costs of imported in-
puts and debt denominated in foreign currencies. These costs (unlike
those for hikes in interest rates) are contractually passed on to Tenaga.
How well Tenaga is able to bear these costs depends on its financial
condition.
In the run-up to the banking crisis, Tenaga's financial situation was
conditioned by its large U.S. dollar-denominated borrowings, which
resulted in significant foreign exchange losses as the Malaysian ringgit
weakened (Box 17.2). As a consequence, while demand growth re-
mained surprisingly strong after the crisis, Tenaga's internal funding
resources shrank, requiring fresh additional foreign borrowings. Al-
though Tenaga sought to renegotiate its commitments to five IPPs,
they did not accept the proposal. In a small gesture, however, they
agreed to receive payments monthly instead of weekly.
CONTINGENT LIABILITIES IN INFRASTRUCTURE 381
Box 17.2 Tenaga's Approach to the Financial Crisis
The government of Malaysia continues to be the principal shareholder
of the electricity utility Tenaga Nasional Bhd, which suffered cash losses
for 1998 that were significantly higher than in the previous year (see
table). The big loss stemmed from the currency depreciation. Demand
growth slowed, but it continued to be surprisingly strong. Electricity
demand is expected to grow between 6 and 8 percent annually.
For financial year ending
August 31 ( billions of ringgit)
1997 1998
Turnover 10.0 11.4
Operating profit 1.4 0.6
Foreign exchange loss (1.3) (3.5)
Loss after tax (0.1) (3.1)
As a utility, Tenaga normally has access to significant internal fund-
ing sources. Recently, however, these sources have been limited, re-
quiring fresh foreign borrowings. To reduce its debt service burden,
Tenaga swapped part of its dollar-denominated debt for yen-denomi-
nated debt, thereby reducing its dollar exposure from 40 to 30 percent.
The yen debt carries a significantly lower interest rate. However, Tenaga
now takes on the risk of possible yen appreciation. (Source: Star [Kuala
Lumpur], November 11, 1998.)
Indonesia
In Indonesia, concerns existed prior to the crisis about the country's
capacity for excess power generation and the unsolicited, long-term
take-or-pay contracts that the government-owned utility, Perusahaan
Listrik Negara or the PLN, signed with some of the 26 independent
power producers. The cost of this power is high by international stan-
dards and, more to the point, in relation to the tariffs that can realisti-
cally be charged the final consumers. The banking crisis aggravated
the PLN's debt burden in two ways. First, the rupiah value of its U.S.
dollar-denominated obligations (including debt service to the govern-
ment of Indonesia on on-lent loans, gas purchases for its own plants,
and contracted purchases of power from 1PPs) skyrocketed as the ru-
piah depreciated. Second, demand growth fell. For example, energy
sales in Java-Bali (80 percent of the PLN's market) are now projected
382 ASHOKA MODY
to remain flat. The PLN's current weak financial situation implies that
it will be unable to meet all its contractual obligations.
The PLN's obligations may ultimately be the government of
Indonesia's obligations. It remains true that the government's "letter
of comfort" to the sponsors of private projects does not oblige the
government to pay creditors and IPPs on behalf of the PLN under the
take-or-pay contracts. However, as more IPPs come onstream, the PLN's
fiscal situation is likely to come under increasing pressure. As a state-
owned entity, the PLN would have recourse to the government on at
least some of its other obligations.
The implications of the crisis are serious. The IPP capacity con-
tracted for by the PLN is just coming onstream. During fiscal 1999/
2000, additional monthly payments of about US$50-60 million (about
0.4 percent of GDP) came due. These payments will continue to in-
crease for the next four to five years as successive IPPs get ready to
deliver power. Because virtually none of the new capacity that comes
onstream will be used in the next four to five years, given the high
level of contracted capacity and the sharp decline in demand, these
new obligations will present a huge drain on the financial resources of
the PLN.
The PLN and the Indonesian government also are exposed to risks
associated with legal action. In view of the PLN's growing obligations,
the government has put on hold some negotiations for additional ca-
pacity (Keppres 39/97, September 1997). In principle, that is the right
decision. However, some of the negotiations were at a relatively ad-
vanced stage. One promoter of such a project has taken the view that
a commitment already existed on the part of the government and has
sought claims under the United Nations Commission on International
Trade Law.
Beyond IPPs: A More Competiti've Model
As with the transport infrastructure, some of the problems faced by
Indonesia (and also other countries such as Pakistan and Thailand)
stem from the underlying strategy. While negotiations of project terms
to lower the prices paid to the IPPs can help, ultimately far-reaching
sectoral reform will be required. Such a strategy will entail more ac-
tive steps toward containing the generating capacity and creating a
sector structure that permits the transfer of more risks to the private
providers.
The recent international experience offers several pointers. First, to
create more competition in power generation, countries that have con-
tracted with IPPs will have to deal with the stranded assets from past
contracts-assets whose economic value will potentially decline after
deregulation, either because the market test shows IPPs are no longer
CONTINGENT LIABILITIES IN INFRASTRUCTURE 383
needed or because they face greater market risks. Considerable con-
troversy surrounds how much compensation is due to the owners of
the stranded assets and how the compensation will be funded-the
two choices are the taxpayers or the consumers. In general, the con-
sumer pays through a levy on thc electricity bill. This need not raise
the cost of electricity if competition lowers the basic charges, but it
will delay the benefits of lower overall bills.
Second, creating and operating ain independent system operator (ISO)
are crucial. The ISO ensures equal access to transmission lines and the
economic utilization of generation assets through overseeing the func-
tioning of a spot market for electricity. Although the details of the spot
market vary somewhat across countries, the principle is the same ev-
erywhere. Generators bid for defined time slots by declaring their
marginal cost of supply. The ISO (lispatches requirements for that slot,
choosing the generators with the lowest marginal costs; all generators
receive a price equal to the marginal cost of the lowest-cost supplier,
thus creating an incentive for efficiency. The price paid by the electric-
ity distributor for electricity received from the ISO is regulated and
equals some variable cost elements (including the spot price of the
electricity in the relevant time periods) and transmission charges, which
are limited through a price cap formula.
Third, competition also occurs through direct contracting between
generators and the larger users of electricity as well as with distribu-
tors who then sell power retail to the final consumers. Typically, these
contracts are not regulated, and thus the price charged and the length
of the contracts are determined through direct negotiation between
the buyer and the seller. The spot: market for electricity and the direct
contracts provide benchmarks for each other, and, where competition
is effective, the average prices in the two markets should converge. A
key regulatory issue, however, does arise. The generator will not typi-
cally transmit electricity to the customer buying the power. Rather, the
customer "will take physical service from the local utility which will
deliver power commingled from an undifferentiated array of genera-
tors who are making common, simultaneous use of the transmission
grid and distribution facilities" (California Public Utilities Commis-
sion 1996). Thus the generator requires the agreement of the local
utility to conclude its obligation to its customer, and the pricing of
such access requires regulatory oversight.
Finally, distribution rates are typically regulated through a price
cap formula. Once again, certain variable costs of the energy are passed
through, while the fixed charges for distribution (for connection and
making capacity available) are capped. Competition among distribu-
tors is increasingly permitted, especially for larger consumers, although,
as the Argentine example shows, small consumers are increasingly
benefiting from a choice of energy sources (see Box 17.3).
384 ASHOKA MODY
Box 17.3 Competitive Power in Argentina
Just prior to its recent crisis, Argentina may well have had the most
competitive power industry in the world. At the end of 1997, genera-
tion capacity of 19,000 megawatts was shared by over 40 generation
plants. Another 14 plants with total capacity of over 5,000 megawatts
planned to be operational by the end of 2001. Most of the new plants
were expected to employ the combined cycle technology, which allows
greater operational efficiency and is economical at smaller scales than
traditional thermal and hydro-powered generation.
Distribution concessions were awarded to 17 principal distributors
plus several smaller ones. Distributors are organized largely in line
with state boundaries. Large consuLmers are able to bypass the dis-
tributors. In fact, because Argentina has one of the most liberal defini-
tions of large users, many relatively small users are able to contract
directly with generators. Two categories of large users are recognized.
In the first are those with a requirement of more than 1.0 megawatt
and a minimum annual consumption of 4,380 megawatt-hours; they
must contract directly for at least 50 percent of their needs. In the
second are those with peak requirements of between 0.1 and 2.0 mega-
watts; they must contract directly with generators for all of their en-
ergy needs. Most contracts, in practice, are of relatively short duration
because prices have been falling.
The competitive system is held together by a transmission and dis-
patch system and an overall regulatory framework for ensuring coor-
dination and efficient pricing. The National Interconnected System
covers about 90 percent of the Argentine population, and the rest is
served by a smaller grid. The system is managed by Compania
Administradora del Mercado Mayorista Electrico S.A. (CAMMESA),
which is owned half by the government and half by various suppliers
and users. Although CAMMESA has been successful in operating the
spot market for electricity, through which the various generators sup-
ply power to the transmission grid, expansion of the network has been
more difficult to coordinate, because it necds joint approval by several
parties.
The regulatory task includes setting price caps for the transmis-
sion and distribution segments of the network. The regulator also is
supposed to supervise the periodic reauctioning of the distribution
networks. To augment the independence of the regulator, the govern-
ment has set high standards for qualifying as a regulator, the regula-
tors have fixed terms and cannot be easily removed, the terms are
staggered in the style of U.S. public utility commissions to provide
continuity to the regulator's activities, and an independent source of
funding has been created through a levy on the various industry par-
ticipants. (Sources: Duff and Phelps 1998 and Estache and Rodriguez-
Pardina 1998.)
CONTINGENT LIABILITIES IN INFRASTRUCTURE 385
Malaysia's National Sanitation Project
Malaysia has made important strides in using the private sector for de-
livery of water and sanitation services (Haarmeyer and Mody 1998).
Several water supply and treatment projects have been constructed on a
build-operate-transfer basis. In addition, the Indah Water Konsortium
(IWK) national sewerage project demonstrates both the benefits and
limits of the Malaysian approach. Prior to the IWK concession, local
governments bore responsibility for providing sewerage services. How-
ever, they often lacked the necessary financial or technical resources,
and, as a result, the existing physical plant was very neglected and needed
investments were not undertaken. The unsolicited project proposal of-
fered an innovative approach to dealing with an increasingly urgent
sewerage problem. IWK's proposal was processed and approved very
rapidly by the government, and the award in 1994 yielded a dramatic
improvement in the level of investment and in the quality of service.
However, IWK quickly began to encounter problems that became
more serious over time and threatened the financial viability of the
concession. Consumers protested against the rates charged soon after
the concession was awarded, well before the economic crisis. As a
result, IWK's contractually agreed to tariff structure was disallowed
without clear compensatory arrangements and was not reestablished
until 1997 at much lower levels. Indeed, it appears that several impor-
tant details relating to the tariff and compensation structure, interim
performance targets, and contract management were not addressed
with sufficient detail in the concession agreement, perhaps in part be-
cause of the speed with which it was processed.
Notwithstanding the tariff structure change, poor collection rates
are a critical ongoing problem, complicated by the economic slow-
down and the lack of an effective enforcement mechanism. The eco-
nomic crisis also prompted a .30 percent reduction in charges to
commercial sector customers in July 1998. Compounding the revenue
woes, the full magnitude of the physical rehabilitation needed was not
anticipated, and IWK's management was further disrupted by three
ownership changes during its first four years. As a consequence, the
government has had to make more than MYR450 million in long-
term soft loans to IWK, in addition to other support.
Though many of IWK's difficulties are not uncommon, the approach
taken has aggravated them. At least in four respects, the Malaysian
approach runs counter to international trends:
* By eschewing a competitive bidding process in favor of a speedier
negotiated transaction, the government compromised the legitimacy
of the prices charged. Consumers are more likely to agree to pay for
services where the prices charged are perceived as fair.
386 ASHOKA MODY
o By awarding a nationwide contract, the government further lim-
ited its capacity to regulate the private supplier. In the United Kingdom
and in some developing country cities such as Manila and Mexico City,
benchmarked competition has been employed. By splitting the conces-
sion area into distinct jurisdictions, the regulatory authority is able to
compare and assess the performances of the different operators, even
though the operators do not directly compete with each other.
o By guaranteeing a minimum rate of return, the government went
against the trend of using "incentive" regulation. As is well known,
under a guaranteed return the operator has no incentive to control
costs. In contrast, under incentive regulation such as a price cap, there
is a stronger incentive to limit costs.
o A final unusual feature of the contract was the separation of sew-
erage from water. In major concession contracts (such as those in Buenos
Aires and Manila), water and sewerage have been jointly awarded to
the private concessionaire. This allows charging for sewerage services
as an add-on to the water service. In Malaysia, water services are un-
der the provincial authorities and sewerage is under the national au-
thority. The inability to reconcile these two authorities led to the
separate award of sewerage services to private operators. This repre-
sents a missed opportunity from a financial standpoint, insofar as water
utilities generally have much better success collecting revenues, thereby
possibly resolving a very serious problem for IWK.
From the Crisis to the Long View
The financial crisis in East Asia presents governments with a chal-
lenge-and an opportunity-to rethink their strategies in key infra-
structure sectors. It is possible, as in Malaysia, to view the postcrisis
problems as temporary and seek financial engineering solutions to tide
over. This approach requires the assumption that the projects and the
sector structures are basically sound and that with the recovery of
growth the projects will be financially viable. Under this interpreta-
tion, a thorough overhaul of sector strategies is not required as a re-
sponse to the crisis; a more modest effort will provide short-term relief.
Alternatively, the governments could undertake a more fundamental
review of project and sector economics to create a sounder basis for
future operation and investment decisions. Even after recovering from
the crisis, East Asian growth rates are unlikely to achieve the high
levels they reached just before the crisis. As such, the policy emphasis
needs to shift from the creation of new assets to the more efficient use
of current assets. As Argentina demonstrated in the early 1990s, a
crisis allows a fundamental questioning of old infrastructure strategies
and presents an opportunity to mobilize the political will to take diffi-
cult decisions.
CONTINGENT LIABILITIES IN INFRASTRUCTURE 387
A longer view also is desirable from a risk management perspective.
Risk management entails three complementary tasks: mitigating the
risk at source, transferring the risk to parties better able to bear the
risk, and monitoring and managing any residual risk that cannot be
mitigated or transferred.
Ultimately, risk mitigation is the most desirable long-run strategy,
because it reduces the vulnerability of the economy to shocks and thus
reduces the government's direct and indirect contingent liabilities.
Mitigation is also beneficial because it can be typically associated with
strategies that enhance efficiency of resource use, thereby enabling
faster growth and lower risk. For example, a power sector that is orga-
nized to permit competitive generation and distribution will foster ef-
ficient use of resources while lowering the risks arising from excessive
installation of capacity. Risk also would be more effectively trans-
ferred to private providers than under the current system.
Here a comparison of sector strategies in Malaysia and Chile might
be useful. Malaysia is unusual in targeting its privatization initiatives
to the "hard" sectors: transportation and water/sanitation. Although
the shares of privatized infrastructure investment in Malaysia and Chile
are at comparable levels, significant sectoral differences exist between
the two countries (Figure 17.2). I:n interpreting Figure 17.2, however,
note that the true extent of private participation in Malaysia is exag-
gerated, because the figure more closely reflects the share of project
value under private operation than the financial risk borne by private
operators (see Box 17.4).
Malaysia and Chile differ in their philosophical motivations for
privatization. Whereas Chilean privatization is integrally linked to a
competition policy, competition within the infrastructure sector is vir-
tually absent in Malaysia. Instead, privatization in Malaysia has been,
and continues to be, driven by the aim to fill a perceived financing
gap, although other considerations including socioeconomic factors
are at play as well. In Chile, the power sector is almost entirely private
but transportation and water/sanitation investments continue to be
largely in public hands; by contrast, Malaysia has pursued nearly the
opposite strategy. Chilean privatization of transport and water has
been limited, because competition in these sectors is difficult to achieve.
Also contrast the differences in the privatization approaches of the
electric power sector. In Chile, power generators compete to supply to
power grid. In Malaysia, the five private power producers have long-
term take-or-pay contracts with Tenaga, giving them a virtually as-
sured market for the power they produce irrespective of efficiency. So
while Chilean authorities have been concerned with maintaining ad-
equate separation between the generation, transmission, and distribu-
tion of electricity, until recently unbundling power sector services has
not been a high priority policy matter in Malaysia.
388 ASHOKA MODY
Figure 17.2. Public and Private Investment in Infrastructure
Across the World, 1995
(percent of total investment)
Telecommunications Power
100 100
80 80
60 60
40 40
20 20
0 0
Transportation Water and Sanitation
100 100
60 60
40 40
20 20
0 0
E 3~~~~~~ ~E U
Public
Private
Son(rce: World Banik staff estimates.
CONTINGENT LIABILITIES IN INFRASTRUCTURE 389
Box 17.4 How Much Risk (Jo Malaysian Project
Sponsors Really Take?
As in many countries, the true extent of private participation in Ma-
laysia is less than the value of the "private" projects. The private capi-
tal at risk is substantially lower than the project cost, because the
government has provided support through several mechanisms: "soft"
loans, equity investments, directed lending through banks and provi-
dent funds, and various explicit and implicit guarantees.
* Government soft loans carry low interest rates (between 0 and 8
percent a year, long grace periods of 10-15 years, and maturities as
long as 25 years). These loans often finance the acquisition of land and
are justified on the grounds that the projects would not otherwise be
economically viable and, by contract, are transferred to the govern-
ment at the end of the concession period.
* Equity participation is carried out through government-owned
holding companies and corporations, such as Khazanah, Employees
Provident Fund (EPF), Petronas, and Tenaga.
* Government guarantees take different forms. In some transpor-
tation projects, the minimum revenue or traffic volume is guaranteed,
while minimum revenue levels are assured to IPP project sponsors in
the take-or-pay capacity payment arrangements. The government also
stands ready to compensate lende:rs and project equity providers in the
event of project termination. In the Indah Water Konsortium (IWK)
concession, the government's guarantee is technically broader and as-
sures a minimum rate of return to the equity holders.
* Lending by domestic financia I institutions (including EPE and other
government-directed entities) has been a strength of Malaysian infra-
structure development, because it has reduced reliance on foreign cur-
rency borrowings, a significant source of financial dislocation in other
developing Asian nations. However, although such borrowing was in-
tended to take place on "commercial terms," lending institutions lacked
incentives and, in some cases, perhaps the sophistication to exert ex-
tensive due diligence given the government's support and sponsorship
of projects.
Conclusion
Much of the discussion of guarantees for infrastructure in East Asia
was initially conducted in terms of "optimizing" the guarantees
(Johnston, Mody, and Shanks 1996). The assumption was that guar-
antees, in one form or another, were likely to be a permanent (or, at
least, a long-term) element of infrastructure financing. As such, the
390 ASHOKA MODY
focus was on paring down the guarantees from the relatively extensive
obligations taken on by governments in the early stages to more lim-
ited obligations that constituted a "core" minimum. The core obliga-
tions, in turn, were defined as those that reinforced government
regulatory commitments for rights of operation and pricing formulas.
Financial guarantees were a device to hold governments to their word.
Proposals in vogue included one that recommended a timetable for
reducing the lists of risks for which governments provided guarantees
and also the benchmarks (such as credit rating upgrades) that would
trigger the reductions.
The East Asian financial crisis can be treated as an unusually ex-
treme event unlikely to repeat itself and thus having no bearing on the
basic approach to infrastructure privatization and so no additional
bearing on the forms of government commitment. However, in this
chapter I argue that that would be a mistake. It is likely that the con-
tingent liabilities that governments will eventually have to pay for will
be less than thought at the height of the crisis. But that should not lull
governments into assuming that the infrastructure strategy being pur-
sued is without significant risks. Throughout the region, as in other
parts of the developing world, projects based on government contin-
gent support (either implicit or explicit) have encountered problems
even outside of crisis conditions, with significant fiscal costs to gov-
ernments. Long-term sustainability, for both risk management and ef-
ficiency reasons, requires a shift to a new model rather than tinkering
with the old one.
Notes
1. This chapter builds on inputs provided to World Bank (1999) and World
Bank (2000). For comments and suggestions, I am grateful to Hana Polackova
Brixi, Sudarshan Gooptu, and Mona Haddad. The chapter also reflects ideas
developed in long-running collaboration with Christopher Lewis and Robert
Shanks.
2. Other Asian countries that adopted similar strategies include China,
India, and Thailand. Thai contingent liabilities are discussed in Arthur
Andersen (1999). In contrast, Korea undertook limited infrastructure
privatization. In Latin America, Colombia adopted the "Asian" strategy with
several stand-alone public-private infrastructure projects (Lewis and Mody
1997), as did Mexico for a major toll road program and a more modest
independent power program.
3. For an earlier application of a similar Monte Carlo simulation to esti-
mate contingent liabilities in Colombia, see Lewis and Mody (1997).
4. To the extent that this is already part of the overall bad debt portfolio
of the troubled banks in Indonesia, it is not an additional liability for the
government.
CONTINGENT LIABILITIES IN INFRASTRUCTURE 391
5. Electricity is supplied in Malaysia by three entities, each operating in
its own area. In Peninsular Malaysia, the partially privatized Tenaga Nasional
Berhad (TNB), formerly the National Electricity Board, is responsible. The
Sabah Electricity Sdn. Bhd, under the majority ownership of TNB with the
state government retaining a minority share, supplies power in Sabah. Fi-
nally, the Sarawak Electricity Supply Company, wholly owned by the state
government, supplies power to Sarawak.
References
Arthur Andersen. 1999. The Fiscal Costs of State Enterprises and Private
Participation in Infrastructure. Draft Rcport. Bangkok.
California Public Utilities Commission. 1996. "Executive Summary and In-
troduction" .
Duff and Phelps Credit Rating Co. 1998. Latin Power. New York.
Estache, Antonio, and Martin Rodriguez-Pardina. 1998. "Light and Light-
ning at the End of the Public Tunnel: Reform of the Electricity Sector in
the Southern Cone." EDI Regulatory Reform Discussion Paper. World
Bank, Washington, D.C. .
Gomez-Ibanez, Jose A., and John R. Meyer. 1992. "Toll Roads and Private
Concessions in France and Spain." John F. Kennedy School of Govern-
ment, Harvard University, Cambridge, Mass.
Haarmeyer, David, and Ashoka Mody. 1998. "Tapping the Private Sector:
Approaches to Managing Risks in WVater and Sanitation." Journal of Project
Finance 4 (2): 1-28.
Johnston, Felton Mac, Ashoka Mody, and Robert Shanks. 1996. "A House
of Cards: Government Guarantees." Project and Trade Finance 154
(February).
Lewis, Christopher, and Ashoka Mody. 1997. "The Management of Con-
tingent Liabilities: A Risk Management Framework for National Gov-
ernments." In Timothy Irwin, Michael Klein, Guillermo Perry, and
Mateen Thobani, eds., Dealing with Public Risk in Private Infrastruc-
ture. World Bank Latin American and Caribbean Studies. Washington,
D.C.: World Bank.
Reside, Renato E., Jr. 2000. "Estimating the Philippine Government's Expo-
sure to and Risk from Contingent Liabilities in Infrastructure Projects."
School of Economics, University of the Philippines, Manila.
World Bank. 1994. World Development Report: Infrastructure for Develop-
ment. Washington, D.C.
1999. Malaysia: Public Expenditure Review. Washington, D.C.
2000. Indonesia: Public Spending in a Time of Change. Washington,
D.C.
CHAPTrER I 8
Monitoring Fiscal Risks of
Subnational 'Governments:
Selected Country Experiences
Jun Ma*
Deutsche Bank in Hong Kong
IN MANY COUNTRIES, SUBNATIONAL (hereafter local) governments are
charged with responsibility for delivering most public services.' The
fiscal health of local governments is thus essential to the stability and
efficiency of the country's entire public finance system. The central
governments, however, often have limited information on local public
finance for the purpose of assessing local fiscal risks and planning for
fiscal emergencies. In a typical cleveloping country, the ministry of
finance receives monthly reports on budgetary revenues and expendi-
tures from local governments on a cash basis, but does not have suffi-
cient information on many extrabiudgetary activities, especially local
government guarantees, extrabudgetary and off-budget capital expen-
ditures and associated borrowings, operations of local financial insti-
tutions, pension funds, unemployment insurance funds, and other
transactions that could generate government liabilities.
Some countries have recognized the need to establish a monitoring
system to record and assess local government obligations (both direct
and contingent liabilities) and local fiscal risks in general. This system
can potentially be used to generate a ranking of local governments by
fiscal health and serve as a basis for central government intervention
and emergency financial assistance. Over the longer term, this moni-
toring system-the database and analytical indicators-can provide
'"The author was with the World Bank when he wrote this article; he is currently with the
Deutsche Bank in Hong Kong.
393
394 JUN MA
a useful information base for a local government credit rating system,
a prerequisite for opening up subnational borrowing and developing a
municipal bond market.
This chapter provides some relevant international experiences and
suggests an illustrative set of risk indicators for a hypothetical develop-
ing country. It also includes a brief discussion of what is needed institu-
tionally for a system of monitoring local fiscal risk to work effectively.
Specifically, it points out the need to widen the coverage of fiscal report-
ing, strengthen auditing, and, over the medium term, establish a legal
framework for dealing with fiscal emergencies and a clearer division of
expenditure responsibilities between levels of government.
Rationales for Monitoring Local Fiscal Risks
Unlike private companies, subnational governments provide basic public
services-such as education, health, police, infrastructure-that are
essential to the general welfare of the population in their jurisdictions
and to the functioning of the regional economy. In most cases of local
government default or bankruptcy, the resolution is not to cease the
operation of the government and sell its assets, but to restructure its
debt and other obligations. In many unitary states, if private creditors
are not willing to restructure or reschedule subnational debt, the cen-
tral government is pressured to provide a financial rescue package. In
some countries, local governments are not permitted to borrow from
banks or issue debt, but when local financial institutions fail, local
governments and major local government corporations default on guar-
antees as well as on wage, pension, and unemployment benefit pay-
ments, or local governments become unable to provide basic services
at minimum standards, the resulting social pressures may force the
central government to step in. This mechanism serves as an implicit
guarantee by the central government for local government liabilities.
This type of guarantee leads to a serious moral hazard problem-that
is, local governments (or corporations owned by local governments)
have strong incentives to borrow, issue guarantees, and even incur ar-
rears, because the costs may be borne by the central government. The
central government therefore needs to be aware of the size of the
subnational obligations, its capacity to meet payments obligations, and
its future expenditure profile in order to prevent local government
defaults and protect its own fiscal position. Otherwise, the central
government's own fiscal stability could be jeopardized.
In many developing countries where the municipal bond market
and private credit rating systems are not developed, local governments
are not subject to market discipline. In addition, the lack of transpar-
ency of local fiscal operations in many countries exacerbates the ten-
dency for local governments to engage in imprudent fiscal activities.
MONITORING SUBNATIONAL FISCAL RISKS 395
These problems further strengthen the need for the central govern-
ment to monitor and exert control over local liabilities.
Local fiscal emergencies do not arise without causes. Therefore, it is
possible to design an early warning system that monitors the fiscal health
of local governments and predicts potential fiscal crises. Through expe-
rience, many countries' central or provincial governments recognize that
local fiscal crises are often caused by, among other things, a rapid in-
crease in expenditure relative to revenue, persistent deficits, high gov-
ernment debt, and liquidity problems (short-term liabilities exceed liquid
assets). By closely watching indicators of this sort, the higher-level gov-
ernment may detect which local government is on the brink of financial
trouble and take measures (such as providing financial planning advice,
tightening expenditure control, improving revenue collection, and limit-
ing future borrowing) to prevent it from slipping deeper into fiscal trouble.
Local Fiscal Monitoring Programs:
The United States, Brazil, and Colombia
This section briefly describes the local fiscal monitoring systems in the
United States, Brazil, and Colombia. Among these, the U.S. model (more
specifically, the Ohio model) appears to be the most sophisticated one;
it was explicitly designed to provide an early warning system. The Bra-
zilian and Colombian systems aim to provide a basis for central govern-
rnent control over local indebtedness. Similar systems exist in many other
countries where subnational borrowing is permitted up to certain limits
and requires central government approval (see Table 18.1).
These monitoring systems do have important limitations, however.
The risk indicators used in the fiscal monitoring systems described in
this section are basically static and backward looking. They do not
reflect the dynamics of government revenue, expenditure, and debt
and, as a result, do not provide a direct assessment of fiscal sustain-
ability. These problems can be addressed by requiring local govern-
ments to produce medium-term fiscal projections. A more fundamental
flaw of this type of monitoring system is that most indicators are based
on traditional fiscal accounting--government cash flows and direct
debt obligations. They do not, or only to a very limited extent, reflect
government contingent liabilities. In this chapter, the examples of New
Zealand and Australia (and the example of the Czech Republic dis-
cussed in Chapter 9 by Brixi, Schick, and Zlaoui in this volume) illus-
trate how fiscal analysis can inccorporate contingent liabilities.
The United States: Ohio's Fiscal Watch Program
and Fiscal Emergency Law
The United States experienced some significant local government de-
faults in the 1970s and 1980s, including the 1975 New York City note
396 JUN MA
Table 18.1 Limits on Local Borrowing, Selected Countries
Debt-to- Some kinds of
Counrltry Debt service ratio revenue' ratio restrictions
Japan Three-year average None Mainly for local infra-
20 percent of gen- structure projects; no
eral revenue; more foreign borrowing
restrictions if ex-
ceeding 20 percent
India' None None No foreign borrowing;
long-term credits for
investment only; case-
by-case approval by
state government
Italy S 25 percent of own Only for capital
revenue nct of cer- expenditure; no
tain earmarked funds foreign borrowing
Russia 15 percent of gen- 30 percent
eral revenue for provinces
15 percent for
municipalities
Lithuania < 15 percent of gen- < 30 percent of No state guarantee;
(proposed) eral reventc total revenue ministry of finance
able to approve
lower ceiling for
municipalities;
long-term credits
for investment only
Spain < 25 percent of Long-term credit for
total revenuc investment only; ap-
proval required for
foreign borrowing
a. There is a proposal to impose quantitative limits on subnational borrowing.
Souirces: Counlcil of Europe ( 1993); Ma (I1994); Peterson (1997); Ter-Minassian (1997);
Budget Code of the Russiain Federation, Federal Law No. 145-FZ, July 31, 1998.
default, the 1978 Cleveland default, and the Washington power sup-
ply system default in 1983. As a response, the U.S. Advisory Commis-
sion on Intergovernmiental Relations (ACIR) conducted a series of
studies on local fiscal emergencies and recommended to the states that
they improve the monitoring of local government fiscal health and
prevent or correct local fiscal emergencies (see ACIR 1985 and Box
18.1). Ohio is one of the states that has adopted much of the
commission's advice and developed a local government monitoring
system, the Fiscal Watch Program. Ohio's Code on Local Fiscal Emer-
gencies, originally passed in 1979 and amended in 1985, stipulated the
MONITORING SUBNATIONAL FISCAL RISKS 397
Box 18.1 ACIR Indicators of Local Fiscal Health
Between 1945 and 1969, 431 state and local debt defaults occurred in
the United States. In the early 1970s, the financial stability of America's
cities became a matter of public concern. In response, in 1973 the U.S.
Advisory Commission on Intergovernmental Relations (ACIR) con-
ducted a study of city financial emergencies and proposed some mea-
sures as warning signals. These measures were modified and applied in
its 1985 follow-up report "Bankruptcies, Defaults, and Other Local
Government Financial Emergencies"(ACIR 1985). The eight measures
developed in the 1985 report are:
* deficits: general fund expenditures exceeded revenues by more
than 5 percent;
* persistence of deficits: general fund expenditures exceeded rev-
enues for two consecutive years with the second year larger;
* trend of deficit growth: expenditure growth rates exceed revenue
growth rates;
* balance sheet gap: general fund-accumulated deficits (net liabili-
ties) as a percentage of gencral fund revenues;
* liquidity: net liquid assets (cash and liquid assets minus short-
term debt outstanding) as a percentage of general fund expenditures;
* debt maturity: existence of short-term debt as of the end of the
fiscal year;
* tax compliance: property tax collection rate; and
* unfunded pension liabilities: ,a) net amount of payments for ben-
efits and withdrawals, shown as a percentage of receipts; (b) benefit
and withdrawal payments from the local fund as a percentage of the
total assets of the fund.
procedure for implementing this monitoring system.2 Similar to a
"Watch" issued by the National Weather Service, the Ohio Fiscal Watch
l'rogram acts as an early warning system to prevent local governments-
including counties, municipalities, school districts, state universities
and colleges-from slipping further into fiscal distress.
The Fiscal Watch Program, implemented by the Office of Auditor
of State, conducts fiscal watch reviews to determine whether a local
government is approaching a state of fiscal emergency. Any of the
following conditions constitutes grounds for a fiscal watch:
* The existence of either of the following situations: (a) all accounts
that were due and payable from the general fund and had been due
and payable for at least 30 days at the end of the fiscal year, less that
398 JUN MA
year-end balance in the general fund, exceeded one-twelfth of the gen-
eral fund budget for that year; (b) all accounts that were due and pay-
able for at least 30 days from the general and special accounts at the
end of the fiscal year, less that year-end balance in the general fund
and the respective special funds, exceeded one-twelfth of the available
revenues-excluding nonrecurring receipts-during the preceding fis-
cal year.
o The aggregate of deficit amounts of all deficit funds at the end of
the preceding fiscal year, less the total of any year-end balance in the
general fund and in any special fund that may be transferred to meet
such a deficit, exceeded one-twelfth of the total of the general fund
budget for that year and the receipts to those deficit funds during that
year other than from transfers from the general fund.
o At the end of the preceding fiscal year, moneys and marketable
investments in or held for the unsegregated Treasury of the local gov-
ernment, minus outstanding checks and warrants, were less in amount
than the aggregate of the positive balances of the general fund and
those special funds, and such deficiency exceeded one-twelfth of the
total amount received into the unsegregated Treasury during the pre-
ceding fiscal year.
Upon determining that one or more of the conditions just described
are present, the auditor of state issues a written declaration of the
existence of a fiscal watch to the local government. The fiscal watch
remains in effect until the auditor determines that the conditions are
no longer present and cancels the watch, or until the auditor deter-
mines that a state of fiscal emergency exists.
This Fiscal Watch Program-as an early warning system-provides
a major impetus for local governments to improve their fiscal manage-
ment. It also triggers the state government's advisory service. In many
cases, after receiving a warning signal, local authorities and institu-
tions immediately began to increase their cash reserves instead of spend-
ing down their funds at the fiscal year-end. At the same time, the Office
of Auditor of State provides immediate advisory help at no cost to
local governments under watch. For example, in 1997 it provided guid-
ance on improving the city of Silverton's accounting system, and within
three months the office issued a performance audit with several op-
tions for budget reduction and operational improvement. These advi-
sory services were greatly welcomed by local governments in fiscal
distress, because they could not afford to obtain consulting and other
support services from the private sector (Petro 1997).
The Ohio Code on Local Fiscal Emergencies sets out in detail the
conditions constituting a fiscal emergency. Some of these conditions
were tailored to fit the unique budgeting and accounting characteris-
tics of Ohio local governments. They can be summarized as follows:
MONITORING SUBNATIONAL FISCAL RISKS 399
* test 1 (default on debt): the existence of a default on a debt obli-
gation for more than 30 days;
* test 2 (wage arrears): failure to pay employees within 30 days of
when such payment is due unless two-thirds of the employees have
agreed to a delay of up to 90 days;
* test 3 (request for transfers). the need to reallocate tax levies,
within the constitutional tax limitation, from other local governments
to the municipality;
* test 4 (payments arrears to vendors): accounts payable that are
delinquent by more than 30 days in either the general fund or all funds
that, after deducting cash available to pay them, exceed one-sixth of
the prior year's general fund or all funds revenues;
* test 5 (deficits): total deficit for a combination of funds, less bal-
ances in any other funds that can be transferred to reduce such deficits,
exceeds one-sixth of the prior year's revenues of those funds that are in
deficit; and
* test 6 (cash shortage): uncommitted cash and investments in the
general cash accounts of the government are less than the book bal-
ances of the funds by an amount greater than one-sixth of the total
cash received in those funds in the prior year.
The results of applying the fiscal emergency tests to the ten local
governments declared to have financial emergencies during 1979-85
(Table 18.2) reveal that test 5, the measure of fund deficits relative to
receipts, was failed by eight of the ten governments. By contrast, no
Table 18.2 Local Fiscal Emergency Tests Failed, Ohio,
1979-85
Local government Test I Test 2 Test 3 Test 4 Test 5 Test 6
Niles X X X
Cleveland X X x x
Norwood X X
Plymouth x x x
Ashtabula X
Freeport X
Ironton X X X X
Lincoln Heights X X
East Liverpool x X
Manchester X X
Total number of
failed tests 3 1 0 6 8 6
Source: ACIR (1985).
400 JUN MA
governments failed test 3. Only one government failed to meet its pay-
roll for 30 days (test 2) and that failure occurred after the government
had already been declared to be in a state of fiscal emergency.
According to the Fiscal Emergency Law, as soon as a local govern-
ment is declared to be in a state of emergency, the state of Ohio shall
establish a "financial planning and supervisory commission" to as-
sume the supervisory power of the locality's fiscal management. Within
120 days of the commission's first meeting, the chief executive of the
local government (also a member of the commission) shall submit a
financial plan that contains actions to:
* eliminate all fiscal emergency conditions;
* eliminate the deficits in all deficit funds;
* restore to construction funds and other special funds moneys that
were used for purposes not within the purposes of such funds;
* balance the budgets;
* avoid any fiscal emergencies in the future; and
* restore the ability of the local government to market long-term
general obligation bonds.
The main powers and functions of the commission include:
* reviewing all tax, expenditure, and borrowing policies to require
that they are consistent with the financial plan;
* bringing civil actions to enforce the fiscal emergency law;
* ensuring that books of account, accounting systems, and financial
procedures and reports are in compliance with the auditor of state; and
* assisting the municipal executives in the structuring of the terms
of, and the placement or sale of, debt obligations.
Brazil: Limits on Subnational Borrowing
Over the past decades, Brazil has experienced three subnational debt
crises, resulting in considerable cost to the national government (this
section is based on Dillinger 1999). In most of these cases, the national
government had to take over a large portion of subnational debt. In a
response to the most recent debt crisis, in 1998 the Senate issued Reso-
lution 78, which significantly tightened the central government's moni-
toring and control over subnational borrowing. This resolution consists
of a set of extremely restrictive controls on both the demand for and
supply of subnational debt. The main provisions of this resolution in-
clude the following:
* Subnational governments are not permitted to borrow from their
own enterprises or suppliers.
* Borrowing must be equal to or less than the capital budget.
MONITORING SUBNATIONAL FISCAL RISKS 401
* New borrowing cannot exceed 18 percent of net current revenue;
debt service cannot exceed 13 percent of net current revenue; and the
debt stock must be less than 200 percent of net current revenue.
* Any borrowing government must have a primary surplus; default-
ers are not permitted to borrow.
* The total outstanding guarantees issued by the government must
be less than 25 percent of net current revenue.
* Short-term revenue anticipat:ion borrowing may not exceed 8
percent of net current revenue.
* New bond issues other than :rollover are prohibited.
* At least 5 percent of any bond issue must be retired at maturity,
and any borrowing government whose debt service obligations are less
than 13 percent of net current revenue must retire up to 10 percent of
bonds at maturity or spend 13 percent of net current revenue, which-
ever is less.
Passage of Resolution 78 was followed in September 1999 by Reso-
lution 2653 of the national monetary council, which imposed a comple-
mentary set of restrictions on the supply of credit to subnational
governments. This resolution has two major components. First, it au-
thorizes the central bank, in its capacity as supervisor of the domestic
banking system, to control the supply of credit to subnational govern-
rnents by the domestic banks.3 Second, it authorizes the central bank
to enforce Senate Resolution 78's controls on subnational borrowing.
Specifically, all borrowing petitions for Senate approval must be sub-
rnitted first to the central bank, which has 30 days to analyze the
petitions and forward its recommendations to the Senate. If the cen-
tral bank's analysis shows that the borrowing government is violating
the criteria listed in Resolution 78, it has the right to refuse to forward
its loan request to the Senate.
The new system appears to have significantly reduced new borrow-
ing, because only subnational governments that have little existing debt
and little capital investment are qualified to borrow. There are, how-
ever, two concerns. First, the current system relies heavily on restrictive
government regulations rather than on market dicipline. This may rein-
force the perception that the federal government stands behind all
subnational credit operations. Second, even under the new system, the
Senate is still free to ignore its own resolutions, making the control
mechanism highly vulnerable to political pressures from the states.
Colombia: The "Traffic Light System"
The government of Colombia is divided into 32 provinces (departments)
and 1,064 municipalities. It is one of the most decentralized govern-
ments in Latin America, with about 40 percent of total government
402 JUN MA
spending at the subnational level (this section is based on Rentaria, Steiner,
and Echavarria 1999).
Until 1993 Colombia's subnational government borrowing was under
no control by the Ministry of Finance, other than a registration re-
quirement. In 1993 Law 80 forced financial intermediaries to monitor
the destination of loans and the indebtedness capacity of subnational
governments. Banks became responsible for grading loans to govern-
ment and for monitoring the nature and quantity of acceptable guar-
antees. In 1997 Law 358 limited subnational debt to payment capacity,
by means of associating payment capacity with the generation of op-
erational savings. One ratio (interest payment/operational savings)
proxies the liquidity of the subnational government. The other ratio
(debt/current revenue) evaluates debt sustainability in the medium and
long run. For each new loan, the above indicators act as traffic lights
and must be calculated (Table 18.3).4 When the test result is yellow or
red, the loan requires permission from the Ministry of Finance, and
the borrowing government must sign a performance agreement with
the financial institutions (lenders).
Table 18.4 presents an estimation of the legal capacity of indebted-
ness of the provinces and municipalities of Colombia for 1998. Of the
27 provinces for which there are estimates, 17 are in red, 1 in yellow,
and 9 in green. Most of the liquidity problems can be tracked down to
negative savings. Of 26 provincial capitals, 13 have a red light, 4 have
yellow, and 9 have green.
A typical performance agreement consists of a series of targets with
which the local government must comply, within a predetermined time
frame. These include increases in own resources, expenditure cuts,
generation of current surpluses, and an improved debt profile. When
the local government does not comply, its access to future credit will
be limited. For example, the performance agreement signed by Valle,
the most indebted province, with 23 financial institutions includes the
following conditions:
o The province must contract an irrevocable trust deposit with a fidu-
ciary society, which will administer all the provincial government's funds.
o The province must contract an irrevocable trust deposit with a
fiduciary society, which will administer and sell shares of the two main
corporations owned by the provincial government.
o The province must have authorization from the central govern-
ment to roll over or refinance its short-term loans, to increase its level
of indebtedness.
o The province must, in 1999 and 2000, reduce personnel and other
current expenditures by at least 5 percent a year. Any increase in cur-
rent revenue must first be used to service debt. Only if the current
revenue increase in real terms exceeds 2.5 percent a year can the prov-
ince allocate 50 percent of the additional revenue.
MONITORING SUBNATIONAL FISCAL RISKS 403
Table 18.3 Indebtedness Alert Signals (Traffic Light System),
Colombia
Autonomous Intermeditite Critical
indebtedness indebtedness indebtedness
Indicator (green light) (yellotv light) (red light)
Debt interest/ < 40 percent Between 40 > 60 percent
operational pe:rcent and
savings (liquid- 60 percent
ity indicator),
Debt stock/ < 80 percent < 80 percent > 80 percent
current reve-
nue (solvency
indicator)
Effect Local gov- (a) If the new loan Authorization
ernment is cloes not increase is required to
allowed to clcbt stock by start credit op-
contract new rnore than the erations, and
credit auto- inflation target a performance
nomously. set by the central agreement with
bank, local gov- the lending
ernment can coIn- financial in-
tract new credit stitutions must
autonomously. be signed.
(b) Otherwise, the
indebtedness auth-
orization of the
Miliistry of Finance
is required, with the
I:ollowing condition:
t:he signing of a per-
formance plan with
the lending financial
institutions.
a. Operational savings is defined as current income minus operational expenses and
transfers paid by the local government. Current income mainly includes tax revenues,
nontax revenues, royalties and fees, transfers from the central government, national rev-
enue sharing, and interest income. Operational expenses include wages and salaries, hono-
raria, social welfare benefits, and social security expenditures.
Source: Law 359, 1997.
Accounting for Government Contingent Liabilities:
Australia and New Zealand
In the cases just discussed, governments build their local monitoring
systems only on information about government cash flows (such as rev-
enues and expenditures) and explicit and direct liabilities.' Contingent
404 JUN MA
Table 18.4 Legal Capacity of Indebtedness of Subnational
Governments, Colombia, 1998
(millions of Brazilian reals)
Province Liquidity Solvency Liquidity Solvency Situation
Valle -23.7 22.3 Red Red Red
Antioquia -163.4 46.9 Red Green Red
Atlintico 339.0 65.2 Red Green Red
Tolima -13.1 86.6 Red Red Red
Narifio -64.7 186.5 Red Red Red
Magdalena -18.2 57.0 Red Green Red
Huila -30.5 11.5 Red Green Red
Cesar 70.6 38.0 Red Green Red
Santander 106.9 11.1 Red Green Red
Caldas -7.6 16.0 Red Green Red
Bolivar 273.5 5.2 Red Green Red
Sucre 233.8 22.7 Red Green Red
Quindio -4.8 12.2 Red Green Red
Aniazonas -1.2 101.3 Red Red Red
Vaupes -0.5 81.2 Red Red Red
Vichada -4.2 28.0 Red Green Red
C6rdoba -326.3 - Red Green Red
N. de Santander 47.4 10.9 Yellow Green Yellow
Cundiniamarca 9.6 30.8 Grecn Green Green
Risaralda 19.8 39.6 Green Green Green
Guajira 12.7 76.9 Green Green Green
Arauca 16.8 14.8 Green Green Green
Cauca 30.1 18.3 Green Green Green
Boyac6 17.4 11.6 Green Green Green
Caqueti 7.2 9.2 Green Green Green
Putumavo 27.5 13.6 Green Green Green
Meta 30.7 - Green Green Green
Municipality
Medellin -87.5 26.4 Red Green Red
Barranquilla 94.2 29.2 Red Green Red
Santafe Dc
Bogota DC 105.9 7.0 Red Green Red
Cartagena -72.6 44.7 Red Green Red
Tunja 78.3 27.5 Red Green Red
Florencia 209.5 36.5 Rcd Green Red
lopayin 67.4 42.7 Rcd Green Red
Monteria 103.9 333.9 Red Red Red
Neiva -12.3 32.6 Red Green Red
Santa Marta 102.5 44.6 Red Green Red
Bucaramnanga 128.7 94.3 Red Red Red
Pasto -18.5 48.5 Red Green Red
Inirida -7.3 99.5 Red Red Red
MONITORING SUBNATIONAL FISCAL RISKS 405
Province Liquidity Solventcy Liquiidity Solvency Situation
Pereira 52.9 26.1 Yellow Green Green
Manizales 45.9 74.0 Yellow Green Yellow
Sincelejo 44.0 35.6 Yellow Green Yellow
Ibague 43.1 5.1 Yellow Green Yellow
Cali 49.7 53.6 Yellow G3reen Yellow
Villavicencio 22.7 19.'; Green Green Green
Armenia 16.8 19.2 Green (Green Green
Arauca 27.1 63.7 Green Green Green
Yopal 30.0 67.2 Green Green Green
NMocoa 16.4 31.6 Green Green Green
Leticia 21.4 15.6 Grcen Green Green
Mit6 - 26.8 Green Green Green
Puerto Carrefno 5.2 6.A Green Green Green
- Not available.
Source: Renteria, Steiner, and Echnavarria (1999).
liabilities are not taken into account in assessing local fiscal risks be-
cause, in most cases, data are difficult to collect or even do not exist.
Australia and New Zealand are two of the very few countries that have
developed a system for accounting and reporting governmiient contin-
gent liabilities. This section briefly describes the systems used in the
state of Victoria in Australia and by the New Zealand central govern-
rnent. I have not found good case, in developing and transition coun-
tries, but a series of recent World Bank studies on Eastern European
countries provides some relevant experience for other transition coun-
tries about the types of governmert actions that have led to contingent
liabilities. The Czech Republic case presented in Chapter 9 in this vol-
time is a good example. Although the New Zealand and Czech cases are
at the central government level, the concept and methodology of these
exercises also could provide useful references for local governments.
Victoria, Australia: A Statement of Fiscal Risks
rhe state of Victoria requires that the state treasurer include a state-
ment of risks in his or her annual and semiannual budget reviews pre-
sented to Parliament and the public. This statement describes the factors
that could have a significant effect on the fiscal outcome of the state,
including:
406 JUN MA
o changes in economic parameters such as the generalized system
of preferences (GSP), employment, wages, prices, and interest rates;
o fiscal risks associated with the occurrence of identifiable events
that affect specific revenues or expenditures but that are of uncertain
likelihood or timing; and
o the realization of contingent liabilities arising from nonquantifiable
commitments made by the government.
The risk statement of the 1999-2000 midyear budget review dis-
cusses some quantifiable and nonquantifiable events that form contin-
gent liabilities of the government. These events involve, among others,
lawsuits against the government, government guarantees, environmental
damage, and possible changes in demand for public services. What
follows are a few examples:
o The State Revenue Office has a contingent liability of about
AUD93 million related to the possible outcome of the appeal against
the Supreme Court's decision of 1998 in the case of Drake Personnel v.
Commissioner of State Revenue (subject: payroll tax).
o In May 1994 the Public Transport Corporation (PTC) entered
into contracts with the OneLink Consortium, which will provide au-
tomated ticketing and fare collection services to the PTC over a period
of 10 years. The treasurer has guaranteed the payment obligations of
the PTC under the service contract. The service provider has a con-
tractual right to claim compensation for any losses stemming from
public transport reform and privatization of the PTC.
o Some properties have been identified as potentially contaminated
sites. Although the state does not formally admit any liability with
respect to these sites, remedial expenditures may be incurred to restore
the sites to an acceptable environmental standard if they are devel-
oped in the future.
o Key services provided by the Department of Human Services such
as acute care in public hospitals are experiencing strong demand growth.
There is a risk that this growth will require unplanned additional fund-
ing of these services by the department.
New Zealand: A Statement of Contingent Liabilities
New Zealand's Fiscal Responsibility Act of 1994 requires the central
government to include, on an annual and semiannual basis, a state-
ment of contingent liabilities in its financial statements. Reporting
government contingent liabilities is thought to be essential to ensuring
the robustness of the reported fiscal position and outlook. All finan-
cial statements, including those on contingent liabilities, are submit-
ted to Parliament and also published on the government's website.
MONITORING SUBNATIONAL FISCAL RISKS 407
A summary form of New Zealand's contingent liability table is pre-
sented in Table 18.5. The coverage of the contingent liabilities includes
those of the Reserve Bank of New Zealand (the central bank), state-
owned enterprises, and Crown entities (central government budgetary
institutions). The important examples of contingent liabilities included
in these statements are as follows:
* Guarantees and indemnities: government guarantees for local
government or enterprise borrowing from foreign or domestic sources;
claims for indemnification from private corporations/individuals for
property damage or loss of value:, and a government guarantee for
deposits.
* Uncalled capital: the government's uncalled capital subscriptions
to international financial institutions such as the Asian Development
Bank (ADB), European Bank for Reconstruction and Development
(EBRD), and International Bank for Reconstruction and Develop-
ment (IBRD).
* Legal proceedings and disputes: interest and principal costs that
may be claimed if legal cases were decided against government agen-
cies (for example, ministries, police, defense force, tax authorities, social
welfare agency) and state-owned enterprises.
* Other quantifiable contingent liabilities: contingent liabilities re-
lating to fulfillment of conditions for payment by government agen-
cies of grants and compensation; claims against the government for
people's personal injuries; promissory notes issued by the government
to international financial institutions; other claims against state-owned
enterprises.
In addition to the above, the Crown government also is obliged to
provide details of those contingent liabilities that cannot be quantified.
Table 18.5 Statement of Government Contingent Liabilities,
New Zealand, 1997-99
(millions of New Zealand dollars)
June 30, June 3, June 30,
1997 1998 1999
Guarantees and indemnities 496 373 541
Uncalled capital 2,922 2,250 2,820
Legal proceedings and disputes 362 669 464
Other contingent liabilities 1,286 1,203 1,373
Total quantifiable contingent liabilities 5,066 4,495 4,902
Source: Financial statements of the government of New Zealand, 1997, 1998, 1999,
New Zealand Treasury.
408 JUN MA
Lessons from Experience
This section draws several general lessons from the above case studies.
Definition of Fiscal Emergencies/Crises
Two types of definitions are used in the countries studied here. A broad
definition is used in ACIR (1985), which suggests that fiscal emergen-
cies generally fall into one of the following categories:
o Bankruptcy: applies only to instances in which there has been a
formal filing of a bankruptcy petition under Chapter 9 of the federal
1978 Uniform Bankruptcy Reform Act. To file for bankruptcy, the
government must declare itself insolvent.
o Default on government bonds/notes/bills: failure to pay interest
or principal when due.
o Failure to meet other obligations (such as payrolls, payments to
vendors, pension obligations).
A narrower definition of fiscal crisis refers to the second bullet only-
that is, the local government defaults on explicit and direct debt. This
is a typical interpretation in countries where the bankruptcy law does
not apply to governments and fiscal accounting does not provide ad-
equate information on payments arrears.
For countries where local governments have not been permitted to
borrow (except for central government on-lending), there have been
no defaults on local government debt. A local fiscal emergency would
therefore be better described as a local government running large pay-
ments arrears (including wage, pension, and unemployment benefits
arrears) and failing to deliver basic public services at minimum stan-
dards. These events often generate sufficient social pressures on the
central government that it will provide a rescue package.
Selection of Early Warning Indicators
Debt service burden, deficits, debt levels, and liquidity positions (for
example, cash balances as a percentage of expenditure), and their
medium-term projections, are indicators commonly used for measur-
ing solvency and liquidity risks. However, these indicators suffer from
two major problems. First, for the selection of specific indicators, most
countries seem to base their design on the principles of data availabil-
ity, relevance (based on experience), and simplicity. Most of these sys-
tems lack a solid theoretical foundation and often are not supported
by rigorous statistical tests.
Second, this system works relatively well in countries where the
budgets of the local governments cover most of their fiscal activities
MONITORING SUBNATIONAL FISCAL RISKS 409
and where contingent liabilities are relatively unimportant in relation
to the size of the budgetary operations. This is the case in the United
States. In developing countries, where off-budget and contingent li-
abilities are presumably significanit, the traditional indicators do not
necessarily reflect the possibility of a sudden increase in debt service
obligations due to contingent liabilities. Therefore, identifying and
accounting for the major contingernt liabilities would be essential if the
monitoring system is to become a rneaningful early warning system. In
most Western countries, government guarantees, unfunded public pen-
sion and insurance liabilities, and legal claims against the government
are the most important items. For developing and transition countries,
insolvent financial institutions and state enterprise debt are likely to
be significant government liabilities as well.
Accounting, Auditing, and Reporting
of Local Fiscal Activities
Unsound fiscal management is one of the most significant causes of
the fiscal crises of local governments. In many countries, inadequate
accounting and reporting have caused some local governments to drift
into fiscal crisis without realizing how serious their problems had be-
come. By the time they became aware a crisis was imminent, the size
of the problem had already made fiscal adjustment (that is, a sharp
increase in taxes and a reduction in services) politically impossible.
Without a sound fiscal reporting system at the local level, an early
warning system, no matter how well designed, does not work effec-
tively. Any useful risk measurement must be supported by reliable fi-
nancial data and have sufficiently broad coverage. Significant efforts
by the central government, along with its political will, are required to
provide technical assistance to local governments, to widen the cover-
age of fiscal reporting, and to strengthen the auditing of local govern-
ment reports (including independent auditing). Budget reforms,
including establishment of a single Treasury account, introduction of
departmental budgeting, and reform of budget classifications, should
all help improve the quality and coverage of local fiscal reporting. In
addition, certain requirements for publicity, such as publication and
local congressional review of detailed local budgets, execution reports,
and off-balance sheet government transactions (such as guarantees)
will further improve local accountability.
Central Government Interven.tion
in Local Fiscal Emergencies
Establishing a local monitoring system is not a self-serving task. It
should act as a trigger for central government intervention and should
410 JUN MA
provide incentives for local governments classified as "highly risky" to
take quick and decisive actions to adjust their fiscal positions and re-
verse the tendency toward default.
In terms of administrative intervention by a higher-level govern-
ment in local default or near default, country experiences differ widely.
Most countries do not have a formal legal framework that governs the
procedure for dealing with local defaults, and, as a result, each situa-
tion is handled on a case-by-case basis. This legal ambiguity often goes
hand in hand with uncertainties in the areas of revenue and expendi-
ture assignments as well as negotiation-based intergovernmental trans-
fers. As a result, crisis resolutions often call for the central government
to extend additional transfers and loans, reschedule or write off local
government debt, or assume local government debt and other pay-
ment obligations, and for state banks to provide credits to local gov-
ernments. Such a system is often understood as providing implicit
guarantees by the central government on local government obligations,
and it generates incentives for local governments to pursue excessive
spending through borrowing or provision of guarantees.
Good examples exist of legally formulated procedures for higher-
level government intervention in local fiscal emergencies. In the United
States, although the federal government does not have the legal re-
sponsibility to intervene in municipal defaults (with a few exceptions
such as Washington, D.C.), some states have developed formal legal
frameworks for state government intervention. Most noteworthy is
the two-phase approach of Ohio, based on the state's Code of Local
Fiscal Emergencies. In Ohio, if a local government's fiscal position is
close to but does not yet qualify as an "emergency," the state auditor
would announce that the local government is included in the Fiscal
Watch Program, and the auditor's office would provide it with advi-
sory services on improving local fiscal accounting, auditing, and ex-
penditure control. In many cases, after receiving such a warning signal,
local authorities immediately begin to increase their cash reserves
instead of spending down their funds at fiscal year-end. If a local
government's fiscal health further deteriorates (that is, it suffers sig-
nificant defaults and arrears), it would be declared an "emergency."
In this case, the state would establish a fiscal planning and supervi-
sory commission, which would temporarily take over the fiscal au-
thority of the distressed government. This commission would be
charged with responsibility for restoring the distressed government's
debt service capacity by cutting expenditures (including freezing sal-
ary and staff levels), raising additional revenues, and building up
reserves. In recent years, legal procedures involving a "control board"
or "performance agreement" have been put in place in some devel-
oping and transition countries, including Argentina, Colombia, Hun-
gary, and South Africa.
MONITORING SUBNATIONAL FISCAL RISKS 411
An important advantage of having a transparent procedure for iden-
tifying and handling local fiscal emergencies is that local governments
know in advance the conditions thai: will trigger an emergency declara-
tion and the higher-level government actions that will ensue. Otherwise,
they cannot be sure whether, when, and how the higher-level govern-
ment will act. International experiences tend to indicate that the exist-
ence of a set of criteria for determining fiscal emergencies helps to reduce
the severity of the problems, because it permits the resolution of these
problems at a stage at which they are still manageable.
Toward a More Comprehensive Reform
ol Intergovernmental Relations
In many countries, a fundamental difficulty in implementing a cred-
ible local fiscal warning system and the ensuing central government
intervention-both aimed at reducing the local government's incen-
tive to engage in short-sighted and imprudent fiscal activities-is the
lack of a clear division of labor between levels of government in rev-
enue sources and expenditure responsibilities. Without a clear and trans-
parent legal framework for these relationships, the central government
has to fill the gap, which encourages local spending increases, reduces
local revenue efforts, and results in excessive bargaining.
Another important aspect of transparency in intergovernmental
relations is that the central government ensure that the rules for deter-
mining grants and loans to local governments are known and under-
stood by all local governments. Thi.s can serve as a useful device by the
central government for fending off local government requests for supple-
mentary grants and loans that are not justifiecd by the announced crite-
ria (for example, central government policy changes, natural disasters).
The rationale for the effectiveness of this device is simple: should the
center grant additional resources to a local government with poor fis-
cal management, it needs to defend this decision in front of all other
localities.
Establishing an Implemerntable Monitoring System
This section illustrates how a developing country can draw from rel-
evant international experiences and design a monitoring system that
meets the following criteria:
* the indicators reflect the fiscal risks associated with major quan-
tifiable contingent liabilities;
* the measurement of fiscal risks reflects the potential pressures on
the central government for rescue packages; and
* the system's predictive power is testable with empirical data.
412 JUN MA
Meastiring Contingent Liabilities
Most countries' ministries of finance receive information on local bud-
getary revenues and expenditures, extrabudgetary revenues and ex-
penditures, nonlending from the central government (including from
the proceeds of central government bonds and foreign loans), and debt
service payments. These data provide a basis for calculating local gov-
ernments' current fiscal positions, but do not necessarily reflect future
expenditure obligations and borrowing requirements. To build a use-
ful early warning system, the central government also must compile
information on contingent liabilities of local governments.
Because different countries may have very different sets of local
government responsibilities-explicit and implicit-it is impossible to
design a system that can be applied universally. This section considers
a sample developing country with significant state involvement in the
banking and productive sectors, a vague definition of legal responsi-
bilities for the central and local governments, substantial de facto fis-
cal autonomy at the local level but no legal right to borrow, and
accounting and reporting mechanisms. Under this system, at least four
major sources of contingent liabilities should and can be captured by
some simple measures:'
o Solvency gaps of local government-owned financial institutions.
These financial institutions mainly include local government-owned
commercial banks, credit cooperatives, and nonfinancial institutions
such as investment and trust corporations. While there are several dif-
ferent resolution strategies for problem local financial institutions,7
the worst scenario, from a fiscal cost perspective, is that the solvency
gaps (between the current levels of net worth and zero solvency) are
filled with government funds (or debt).' In the case of a large state
bank absorbing failed local financial institutions, the central govern-
ment in fact directly assumes the fiscal cost of recapitalizing (or pur-
chasing nonperforming loans from) the state bank. The central bank
that regularly receives financial statements from all local financial in-
stitutions (albeit of dubious quality) should be able to provide the sol-
vency information to the ministry of finance.
o Guarantees provided by local governments. Many local govern-
ments view issuing guarantees as a zero- or low-cost approach to mo-
bilizing resources for their jurisdictions. In countries in which local
government borrowing is prohibited or restricted, the incentive to is-
sue guarantees tends to be stronger. These guarantees have increas-
ingly become a main source of local government contingent liabilities.
While difficult, it is possible for the central government to require that
all guarantees be registered with local fiscal bureaus and their aggre-
gates be reported to the ministry of filance. The ministry of finance
MONITORING SUBNATIONAL FISCAL RISKS 413
can then calculate the total amount of outstanding local government
guarantees that are to mature within a certain period weighted by
some measures of default risk (see the case of the Czech Republic in
Chapter 9 in this volume).9
* Unfunded pension liabilities.. In many countries, local govern-
ments own and operate public pension funds. When these funds begin
to run deficits (under cash-basis accounting) and arrears, they are of-
ten met by local or central government budgetary transfers. These defi-
cits or arrears are likely to widen further as a result of population
aging and persistent unemployment problems. However, because of
the complexity of actuarial projections for many pension funds with a
large number of varying assumptions, it is not feasible to estimate
accurately the unfunded pension liabilities for all pension funds. There-
fore, the ministry of finance can consider a simplifying assumption
that the current levels of pension fuind deficits plus will persist over the
next three years and that they need to be financed/cleared with bud-
getary resources.10
* Unfunded unemployment benefits. In some countries, local gov-
ernments are responsible for providing unemployment insurance. When
the benefits levels mandated by government regulations cannot be fully
met with contributions, the financing gap becomes a local government
liability. Again, any accurate measure of the financing gap/deficit re-
quires detailed forecasts of the unemnployment rates, contribution bases,
collection compliance, and benefit: rates in all cities, and will be diffi-
cult to implement. For a rough estimate, one can assume that the fi-
riancing gaps of the unemployment insurance will continue for the
next three years at the current levels.
Other sources of contingent liabilities also are worth considering.
For example, environmental degradation in many regions poses a
major risk to local government budgets. It is clear that, to reach the
national environmental standard, a region with poorer environmen-
tal indicators will need to spend more on cleaning up than other
regions. In theory, some formulas for calculating the cleanup cost
can be developed. This factor is not listed as a quantifiable govern-
rnent liability for the current exercise, however, because its estima-
tion is complicated by the fact t:hat different regions have different
compositions of financing sources for environmental projects. This
can be a task for future research.'
Selection of Risk Indicators
This section first discusses the basic data requirements and then de-
scribes a set of indicators to measure local fiscal risks. The basic data
set will look like that in Table 18.6.
414 JUN MA
Table 18.6 Summary Data Sheet: [Name of Local Government]
1998 1999 2000 Data source
1. Revenue and transfer (R)
Total revenue MOF
Current revenue MOF
Higher-level government transfers MOF
2. Expenditure (E)
Total expenditure MOF
Current expenditure MOF
Wage and pension payments MOF
Interest payments ([) MOF
Expenditure on contingencies (EC), Local FB
3. Deficit
Overall deficit (OD) MOF
Current deficit (CD) MOF
Primary deficit (PD) MOF
4. Financing
Central government on-lending MOF
Other (specify) MOF
5. Cash balance in reserve funds Local FB
6. Outstanding debt, end of period (D)
6.1 Loans on-lent by the CB MOF
o/w: Local currency-denominated MOF
Foreign currency-denominated MOF
o/w: Short-term MOF
Long-term MOF
6.2 Loans contracted by the LG Local FB
6.3 Bonds issued by LG corporations" SSC
6.4 Payments arrears Local FB
7. Outstanding contingent liabilities,
end of period (CL)
7.1 Solvency gap of local financial
institutionsc CB
7.2 Risk-weighted guarantees provided Local FB
by LG, maturing by 2003w Local FB
7.3 Three times 2000 pension fund MOL, local
deficits, BOL
7.4 Three times 2000 unfunded MOL, local
unemployment benefitsf BOL
Note: MOF = ministry of finance; FB = fiscal bureau; CB = central bank: LG = local
government; SSC = state securities commission; MOL = ministry of labor/social security;
BOL = bureau of labor/social security; o/w = out of which.
a. Spending on financial sector restructuring, payments to creditors on behalf of de-
faulted borrowers, and transfers to cover pension and unemployment benefits.
b. For public infrastructure projects.
c. This number should be backed up by a separate table showing the solvency gaps of
all locally owned financial institutions and the percentage of shares owned by the local
government.
d. To be supported by a separate table showing all outstanding guarantees and their
maturities. The MOF should define the criteria for LG corporations in this category.
e. To be supported by a separate table showing deficits of all public pension funds.
f. To be supported by a separate table showing deficits of all reemployment centers and
unemployment insurance funds.
Source: Compiled by the author.
MONITORING SUBNATIONAL FISCAL RISKS 415
Most of the data in Table 18.6 are available in the local financial
reports submitted to the ministry of finance, or they can be obtained
from the local fiscal bureaus. The difficult part of data collection is
that of contingent liabilities, listed in section 7 of Table 18.6. These
data are to be collected from the central bank, the ministry of labor or
social security, local fiscal bureaus, and local departments of labor or
social security. Detailed data sheei-s will have to be designed to permit
the calculation of the summary numbers in Table 18.6.
Using the above data, one can calculate indicators to assess govern-
ment fiscal risks (see Table 18.7). Some of these are familiar ratios,
such as the debt-to-revenue ratio, debt service-to-revenue ratio, and
current surplus-to-revenue ratio, which are used by many countries'
central governments and by credit rating agencies. However, none of
these ratios will fully reflect the future obligations of the government
(especially those arising from conringent liabilities) and provide a pro-
jection of the pressure on the central government's budget created by a
local government (see the Appendix for a fuller discussion on the ad-
vantages and disadvantages of these ratios).
To address this problem, I propose two new measures as the main
criteria for ranking local governments in terms of fiscal risk: (a) ex-
pected minimum borrowing requirement (MINBRt, l) as a percentage
of current period revenue; and (2) expected maximum borrowing re-
quirement (MAXBR, ,) as a percentage of current period revenue. The
Appendix presents a detailed derivation of these two measurements.
In this exercise, I assume that the central government considers two
planning periods-the current period (t) and the next period (t + 1)-
and wishes to use information available at the end of period t to assess
the fiscal risks facing a local government during period t + 1. As an
example, period t can be interprei&ed as 1998-2000 and period t + 1 as
2001-2003.
Intuitively, MINBR, estimates the expected level of pressures from
a local government on the central government for a rescue package
(loans or grants) under an optimistic scenario-that is, the local gov-
ernment can roll over all realized contingent obligations with pay-
ments only covering the interest in period t + 1. MAXBR, l, on the
other hand, estimates the expected level of pressures for such a rescue
package under a pessimistic scenario-that is, all realized contingent
liabilities need to be paid off with cash within period t + 1.
Local governments with higher revenue capacities tend to face less
difficulty in achieving the fiscal adjustment needed to eliminate BR.
This suggests that MINBR and MAXBR as a percentage of local gov-
ernment revenue (denoted by minbr and maxbr) are better indicators
for local fiscal risk and its resullting pressure on the central govern-
ment budget.
The ministry of finance can determine the cutoff thresholds for minbr
and maxbr based on experience (for example, the observed relationship
Table 18.7 Selected Indicators for Local Fiscal Risks
Indicator Definition Remarks
Debt-to-revenue ratio (t) Stock of direct debt at year-end 2000 divided by Total revenue can be replaced with current revenue
total revenue in 2000. (which applies to all items below). The ratio of net debt
defined as debt minus liquid assets)-to-revenue ratio
also can be calculated.
Debt service-to-revenue Three-year average (1998-2000) of ratios of
ratio (t) debt service (interest and principal payments)
to total revenue.
Overall deficit-to-revenue Three-year average (1998-2000) of overall Includes arrears in overall deficits. Also applies to the
ratio (t) deficits-to-revenue ratios. following two items.
Current deficit-to-revenue Three-year average (1998-2000) of current
ratio (t) deficit (current expenditure minus current
expenditure)-to-revenue ratios.
Primary deficit-to-revenue Three-year average (1998-2000) of primary
ratio (t) deficit (noninterest expenditure minus total
revenue)-to-revenue ratios.
Payments arrears-to-revenue Outstanding payments arrears at year-end
ratio (t) 2000 divided by total revenue in 2000.
Balance in reserve funds-to- Three-year average (1998-2000) of ratios of Can also calculate the ratio of reserves to expenditure.
revenue ratio (t) balance of budgetary reserves to revenue.
Risk-adjusted contingent Risk (r)-weighted contingent liabilities that
tiabilities (t + 1) are to mature within the next three years
(2001-03). For solvency gap of locally
owned financial institution, r can be inter-
preted as the percentage of the financial
institution's shares owned by the local
government; for a guarantee, r represents
the default probability of the guaranteed
borrower; for pension deficits and unfunded
unemployment benefits, r should be set at 1.
Expected minimum bor- The expected level of local government borrow- Note that the minimum borrowing requirement can only
rowing requirement ing from the central government under an be measured in a probability sense-that is, using the
(t -- 1) optimistic scenario-that is. the local govern- vahle-at-risk approach. This indicator-expected max-
ment can roll over all the contingent obliga- imum borrowing requirement-produces a value based
tions with payments only covering the interest. on the assumption of a fixed set of risk weights (r,).
See Appendix for formula.
Expected maximum bor- The expected level of local government borrow- Note that the maximum borrowing requirement can only
rowing requirement ing from the central government under a be measured in a probability sense-that is, using the
(t + 1) pessimistic scenario-that is, all realized con- value-at-risk approach. This indicator-expected min-
tingent liabilities need to be paid off with cash imum borrowing requirement-produces a value based
within period t + 1. See Annex 18.1 for formula. on the assumption of a fixed set of risk weights (r ).
Source: Compiled by the author.
418 JUN MA
between these two indicators and the frequency of local requests for
additional central government lending and transfers beyond the origi-
nal budget). If a local government exceeds either of these two thresh-
olds, it should be declared on the watch list and qualify for certain
intervention (such as fiscal policy advisory service by the central gov-
ernment and tightening of central government monitoring over debt-
financed capital expenditure; see the earlier discussion of these
interventions in this chapter).
Empirical Tests and Future Work
In the future, as data on local defaults/arrears become more available,
the ministry of finance can construct a model to predict local govern-
ment defaults, with minbr, maxbr, and other financial ratios as inde-
pendent variables. This model will then permit the ministry to
determine, more scientifically, the predictive power of the proposed
indicators and their cutoff thresholds for risk classification purposes.
For example, these cutoff thresholds can be derived by minimizing the
combination of type I and type 11 errors of a logistic default predictive
model. Many readily available techniques, such as those used in credit
risk models in commercial banks and early warning models for pre-
dicting currency and banking crisis, can be easily adapted to model
local defaults.
Consideration also should be given to the following extensions of
the above-described monitoring system: (a) in compiling data on liq-
uid government assets (such as marketable securities, deposits, and
cash) and using this information to calculate net government debt (gross
debt minus liquid assets); (b) in developing a set of indicators to assess
the interest rate, currency, and rollover risks of government debt; (c) in
estimating future borrowing requirements, taking into account extra-
ordinary expenditure needs; and (d) when data on local government
defaults/arrears become available, in estimating the probability distri-
bution of defaults on contingent liabilities, and applying the value-at-
risk (VAR) approach to generate a more precise description of fiscal
risks associated with contingent liabilities.
Annex 18.1: Derivation of Expected Minimum
and Maximum Borrowing Requirements
in the Presence of Contingent Liabilities
This derivation considers two periods of a local government's opera-
tion: the current period (t), and the next period (t + 1). The objective
of designing the indicators is to predict the borrowing requirement-
that is, the demand of the local government for central government
MONITORING SUBNATIONAL FISCAL RISKS 419
lending"2 in period t + 1, using information available at the end of
period t. Under the current Chinese system, local governments are not
permitted to borrow from banks or issue public debt. Therefore, any
local government borrowing requirement can be met only by central
government lending or transfers if the central government does not
want to see a local government default. Compared with any other
indicators discussed in Table 18.8, the local borrowing requirement
best reflects the magnitude of pressures placed on the central govern-
rnent by local fiscal operations.
In this exercise, let us assume that each period lasts for three years;
period t being 1998-2000 and period t + 1 being 2001-03. The bor-
rowing requirement in period t + 1 can be written as
(18.1) BR, 1 = (PD,+,I - EC,, 1) + I,, + EC, t
where PD,+ , is the primary deficit in period t + 1, I,+, is the interest
payment on formal debt (excluding promissory notes, see below) in
period t + 1, and EC,. 1 is the expected expenditure on contingencies in
period t + 1. Let us assume that P]D,1 - EC,,,, or the noncontingency
primary deficit (deficit excluding expenditure on contingencies) in pe-
riod t + 1, grows at the same rate as revenue growth a-that is, PDl,+
- EC,, = (1 + a) (PD, - EC,). This is a simplifying but plausible as-
sumption for the purpose of cross-locality comparison given that no
information on future revenue and noncontingency expenditure is avail-
able. This assumption gives rise to
(18.2) BR,, =(1 + a) (PD, - EC,+I, +EC,,,.
To calculate EC+, ,p one first needs to know the magnitude of the
contingent liabilities that are likely to become direct government obli-
gations during period t + 1. Denote this amount by RACL,,, (risk-
adjusted contingent liabilities). It can be written as
(18.3) RACL,+, :: r,* CL, , ,1
where ri is the estimated risk weight attached to the ith contingent
liability CL,, ,. For the solvency gap of a locally owned financial
institution, r can be interpreted as the percentage of the financial
institution's shares owned by the local government; for a guarantee, r
represents the default probability of the guaranteed borrower; for pen-
sion deficits and unfunded unemployment benefits, r should be set at
1, because experience shows that budgetary transfers are the only major
source of financing.
EC,+, can range between two extremes. The expected lower limit
can be derived by assuming that all realized contingent liabilities are
420 JUN MA
Table 18.8 Comparison of Various Indicators
for Fiscal Health
Indicator Advantages Disadvantages
Debt-to- Indirectly reflects future A high but stable debt level
revenue debt service burden. may still be sustainable.
ratio This measure does not
fully reflect the impact of
overall/primary fiscal
balance on sustainability.
Debt service- Reflects part of govern- Current debt service may
to-revenue ment obligations. not reflect future obliga-
ratio tions. This measure does
not fully reflect the
impact of fiscal balance
on sustainability.
Deficit-to- Reflects current borrow- Does not fully reflect future
revenue ratio ing requirement. borrowing requirements.
Total contingent Partially indicates future Fails to discount for default
liabilities obligations. risk and adjust for revenue
capacity.
Liquid assets/ Captures liquidity risk. Does not reflect revenue
expenditure capacity, overall balance,
needs and debt sustainability.
MINBR and Directly measures the The selection of coefficients
MAXBR as fiscal pressure on the (the revenue growth rate
percentage central government and interest rate) is
of revenue budget. Reflects risks somewhat arbitrary.
arising from the debt
stock, future service
burden, and the current
fiscal balance. Unlike
the above indicators,
br is sensitive to all
relevant fiscal health
measures: debt stock,
contingent liabilities,
default risks, interest
rate, current fiscal
balance, and revenue
capacity.
Source: Compiled by the author.
AMONITORING SUBNATIONAL FISCAL RISKS 421
met by government promissory notes to the creditors (including bank
depositors, pensioners, and the unemployed) with a maturity longer
than the length of period t + 1. Thte only government payment obliga-
tion stemming from the realization of contingent liabilities during pe-
riod t + 1 is the interest payment on the promissory notes. The lower
limit of EC, 1 can thus be written as EC,, I = i*RA(-L, + 1/2, where i is
the average interest rate for period t + 1.13 With the above assump-
tions, the expected minimum borrowing requirement (to be met by the
central government) that corresponds to the expected lower limit of
E.C, can be derived from
(18.4) MINBR, = (1 + a) (PD, - EC,) +
i*[D, + (D, + MINBR.,+ ,)]/2 + i*RACL, + /2.
Note that the interest payment on formal debt (excluding promis-
sory notes) is derived by multiplying i by the average debt stock during
period t + 1. Solving equation 18A4 yields
(18.5) MINBR, , = [(1 + Ct) (PD, - EC,) + i*D, +
i*RACL,+ 1/2]/(1 - i/2).
The expected upper limit of EC,, is derived on the assumption that
all realized contingent liabilities during period t + 1 need to be fi-
nanced with cash outlays from the local government's budget, because
creditors do not accept any promissory notes offered by the local gov-
ernment-that is, EC,., = RACL, . This assumption gives rise to the
equation for the expected maxirnum borrowing requirement (to be
met by the central government):
(18.6) MAXBR,+, = (1 + a) (PD, - EC,) +
i* [D, + (D, + MAXBR, + 1)]/2 + RACL, l
Solving equation 18.6 yields
(18.7) MAXBR,+, = [(L + a) (PD, - EC,) +
i*D, + RAC;L, +,]/(1 - i/2).
Intuitively, MINBR, , estimates the expected minimum level of pres-
sure exerted on the central government by the local government for a
rescue package under an optimistic scenario-that is, the local govern-
ment can roll over all the contingent obligations with payments only
covering the interest. MAXBR, ,, on the other hand, estimates the
expected maximum level of pressure for such a rescue package under a
pessimistic scenario-that is, all realized contingent liabilities need to
be paid off with cash immediately.
Note that local governments vwith higher revenue capacities tend to
face less difficulty achieving the liscal adjustment needed to eliminate
422 JUN MA
BR. This suggests that MINBR and MAXBR as a percentage of local
government revenue (denoted by minbr and maxbr) are better indica-
tors of local fiscal risk and its resulting pressure on the central govern-
ment budget.
In measuring local fiscal risks, BR as a percentage of revenue (br)
has the following desirable features:
o a higher debt stock implies a higher br;
° a higher current fiscal deficit implies a higher br;
o a higher interest rate implies a higher br;
° higher contingent liabilities imply a higher br;
o higher default risks associated with the contingent liabilities
imply a higher br; and
o a higher revenue capacity implies a lower br.
Table 18.8 compares br with other indicators in other countries'
local monitoring systems. It concludes that all existing indicators only
partially reflect local government fiscal sustainability or default risk,
while br captures the impact of most measures of fiscal symptoms on
local government risk.
Notes
1. The author would like to thank Phillips Dearborn, Dana Weist,
David Shand, Robert Ebel, Bill Dillinger, Eleoterio Codato, Fernando Rojas,
Sergei Shatalov, Ed Gomez, Nobuki Mochida, Tom Richardson, David Shand,
Hana P. Brixi, E.C. Hwa, Homi Kharas, Christine Wong, Zhang Chunling,
Wang Weixing, Zhang Tong, Guo Chuiping, and Xiang Zongzuo for helpful
discussions and comments. The views expressed in this paper are those of the
author, and do not necessarily represent those of the World Bank.
2. The code may be found online <204.89.181.223/cgi-bin/>.
3. For example, it stipulates that a domestic bank's outstanding loans to
the public sector may not exceed 45 percent of its equity. It also sets a ceiling
of BRL600 million on the total exposure of the banking system to the public
sector, with any amount below this ceiling allocated among banks on a first-
come, first-served basis.
4. Note that the presentation of the Colombian example does not imply
that the author recommends these specific thresholds to other countries. See
other country cases for a comparison (Table 18.1).
5. According to Polackova (1998), government liabilities can be classified
into four categories: explicit and direct (such as formal government debt,
budgeted expenditure), explicit and contingent (such as government guaran-
tees and deposit insurance schemes), implicit and current (such as social in-
surance expenditures), and implicit and contingent (such as defaults by
state-owned enterprises, banks, social insurance funds).
MONITORING SUBNATIONAL FISCAL RISKS 423
6. One may argue that some of the following items-for example, the
solvency gaps of local government-owned financial institutions, would be
better labeled implicit liabilities than contingent liabilities. In reality, the di-
viding line between implicit and contingent liabilities often is not clear. In the
case of solvency gaps, they are government implicit liabilities in the sense
that the insolvency has already occurred, but they also can be contingent
because the actual amount of government support needed at a particular
future date depends on the macroeconomic conditions, the loss-absorptive
capacity of the private sector, and the bank restructuring strategy adopted,
none of which are deterministic.
7. Some typical forms of local government support to problem local
financial institutions are: (a) for liquidated institutions, the local budget pays
off individual deposits; (b) the local government requests other locally owned
state enterprises to inject capital intro problem local financial institutions,
imposing an indirect cost to the loczal budget through a reduction in tax
revenues from these enterprises; (c) the local government offers tax conces-
sions to the problem local financial institutions; (d) the local government
accepts deposit certificates in failed local financial institutions as payment to
the government; (e) because of the liquidation of local financial institutions,
local budgetary institutions lose their deposits, resulting in a demand for
additional budgetary allocation or arrears; or (f) the local government pays
off foreign debt owed by local financial institutions (for example, some trust
and investment companies).
8. If data are available, off-balance sheet guarantees provided by finan-
c ial institutions also should be taken into account in the projection of sol-
vency gaps.
9. To the extent possible, the ministry of finance alsco should consider the
inclusion of other implicit forms ol guarantees extended by local govern-
ments. Examples include: (a) local government agreement with foreign in-
vestors that guarantees the purchase of services over an extended period; and
(b) local government agreement with foreign direct investors that guarantees
a minimum rate of return and the repayment of the initial investment within
a certain period of time.
10. A slightly more complicated but accuracy-enhancing assumption is
that the deficit level in each of the next three years is the 1999 deficit multi-
plied by a coefficient (uniform across regions), determined by projected
changes in the demographic structure, wage level, unemployment rate, ben-
efit rates, and so forth. This alternative also applies to the estimation of
unfunded unemployment benefits (see the next bulleted section).
11. Here the Bulgarian experience provides some useful lessons. Bulgaria's
Ministry of Environment and Wate.r estimates that the 1999-2015 cost of
containing past environmental damage and financing the required treatment
of water supplies, management and disposal of solid and hazardous wastes,
and air cleanup will come up to US$8.5 billion. Assuming equal annual
amounts, the government will need nearly $570 million (about BGN1 billion)
424 JUN MA
annually. The 2000-06 Public Investment Program envisages annual amounts
of $400 million (for discussion, see Chapter 9 by Brixi, Schick, and Zlaoui in
this volume).
12. This also can be interpreted as a demand for a budgetary transfer from
the central government. In terms of the impact on the central government's
overall fiscal balance, lending and offering grants to local governments do not
make a difference. Therefore, I concentrate on the case of local borrowing
from the central government to cover its deficit.
13. Let us assume the interest rate applies to the average stock of realized
contingent liabilities during period t + 1.
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
ACIR (U.S. Advisory Commission on Intergovernmental Relations). 1985.
"Bankruptcies, Defaults, and Other Local Government Financial Emer-
gencies." Washington, D.C.
Council of Europe. 1993. Borrowing by Local and Regional Authorities.
Strasbourg: Council of Europe Press.
Dillinger, Bill. 1999. "Brazil: Regulations on Subnational Borrowing." World
Bank, Washington, D.C. Processed.
Ma, Jun. 1994. "Intergovernmental Fiscal Relations: The Cases of Japan and
Korea." Working Paper 41-1994. Economic Development Institute, World
Bank, Washington, D.C.
Peterson, George. 1997. "Measuring Local Government Credit Risk and
Improving Creditworthiness." World Bank, Washington, D.C. Processed.
Petro, Jim, Auditor of State (Ohio). 1997. "The Ohio Fiscal Watch Pro-
gram." Columbus. Processed.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Renteria, Carolina, Roberto Steiner, and Juan Jose Echavarria. 1999. "Bail-
out of Territorial Entities by the Central Government in Colombia." World
Bank, Washington, D.C. Processed.
Ter-Minassian, Teresa, ed. 1997. Fiscal Federalism in Theory and Practice.
Washington, D.C.: International Monetary Fund.
CHAPTER I9
Guarantees as Options:
An Evaluation of Foreign Debt
Restructuring Agreements
Sweder van Wijnbergen
University of Amsterdam and
Centre for Economic Policy Research
Nina Budina
World Bank
GUARANTEES ARE CONTINGENT LIABILITIES that come into play once
a specific guarantee clause is triggered.' As such, the associated pay-
ment streams can often be duplicated using a properly structured se-
ries of option contracts. In that way, option-pricing techniques can be
used to price guarantees.
In fact, public debt itself has option characteristics, because of the
option the government has not to service the debt in the years that the
debt servicing requirements exceed the country's ability to pay. Thus
option pricing also can be used to model the seccindary market dis-
count of sovereign debt. This chapter adopts and applies an approach
to debt pricing, based on option-pricing theory, to estimate the market
values of different menu options available under Bulgaria's debt and
debt service reduction (DDSR) agreement concluded in 1994 with the
London Club.
In Bulgaria, a part of the principal and interest payments of the
discount bonds and the front-loaded interest arrears bonds are secured
by 30-year, zero coupon U.S. Treasury bonds. Claessens and van
Wijnbergen (1993) show a way to account for the principal and inter-
est collateralization. The principal collateralization is equivalent to an
42s
426 VAN WIJNBERGEN AND BUDINA
unconditional guarantee of the full repayment at maturity and can
be presented as a put option written by the guarantor to the creditor
on the repayment capacity in every period, with a strike price equal
to the contractual obligation. Thus full collateralization of the prin-
cipal will fully offset the put option representing the sovereign risk
on this payment.
Pricing the interest guarantees is more complicated, because the guar-
antees are not fully backed to their maximal exposure; rather, they are
backed by a fund with much less coverage. Thus the interest guarantee
is conditional on the existence of at least one payment in the fund. Pric-
ing therefore requires a derivation of that probability. This can be done
using the same model used to price the nonguaranteed debt.
This chapter is organized as follows. The next section discusses the
issues surrounding foreign debt management and the terms of the 1994
Bulgarian DDSR agreement with the London Club. It is followed by
sections that present the results of our model and our conclusions.
An Option-Pricing Approach
to Sovereign Debt Valuation
How are the benefits of the injection of official funds as part of debt
restructuring exercises divided between the debtor country and its credi-
tors? The answer to this question depends on estimates of the market
value of commercial claims both before and after the agreement; the
difference in market value then represents the value accruing to the
banks. It is tempting to use the secondary market prices quoted by
traders and brokers before and after the agreement as a starting point
for such an analysis (see Bulow and Rogoff 1989 for such an attempt).
One would then equate the gains of the banks associated with the
injection of official funds to the difference in the market value of the
commercial claims between, say, the announcement of debt restructur-
ing and the actual time that the agreement was concluded and the
claims were reduced.
The result of such an exercise, however, could be far off, because
for this kind of analysis secondary market prices suffer from several
shortcomings. First, until recently, volumes of trade in the secondary
market were often low, and the market has been illiquid for claims on
smaller countries and is easily influenced by a single, relatively small
trade. Second, the regulations under which commercial banks operate
heavily influence prices and often change over the period considered.
Third, the prices over the period of time in which an agreement is
negotiated are often influenced by exogenous factors such as the price
of oil. This influence clearly contaminates the comparison between ex
ante and ex post prices as a measure of gains accruing to banks. Fourth,
GUARANTEES AS OPTIONS 427
ex ante prices incorporate the market's anticipation of the upcoming
debt reduction agreement, further contaminating them. Moreover, sec-
ondary market prices for sovereign debt are often very volatile in the
period leading up to debt negotiations as rumors about defaults do the
rounds. Finally, agreements often involve the creation of claims with
clifferent seniority status (the new money claims are junior to the debt
relief bonds). This factor too makes an evaluation using secondary
rnarket prices alone difficult, because not all claims are quoted.
The use of secondary market prices is further complicated by the
fact that ex post no prices of pure commercial claims are quoted. The
only market prices available are t:hose of the debt relief bonds, which
have third party guarantees in the form of principal collateralization
and rolling interest guarantees. Consequently, the prices of these bonds
will reflect the expected repayment by the debtor as well as expecta-
tions about the repayment from the collaterals. Simply comparing the
ex ante market value of the commercial claims with their ex post mar-
ket value would involve comparing claims that differ greatly in the
nature of the risk.
These considerations point to the need for an analytical pricing model
of sovereign claims with and without third party enhancements. Such
a model would allow setting benchmarks by pricing the claims before
the deal and the new instruments afterward using the same model.
The choice of a particular pricing model should be led by the nature of
the repayment behavior on a sovereign claim, the different seniority
structures of the old and new claims, the particular enhancement
schemes used, and the contingent nature of some of the repayment
obligations (the recapture clauses).
Option pricing is a logical choice. Methodologies developed in finance
for pricing contingent claims can Ibe adapted to the problem of pricing
sovereign debt (see Claessens and van Wijnbergen 1989, 1991, 1993).
The main issue with option pricing is to specify the underlying sto-
chastic process driving the debtor country's repayments to the com-
mercial banks. In general, repayments will be difficult to predict; they
will be the result of a complex bargaining process between the country
and its banks. Repayments will be influenced by factors such as export
earnings, terms of trade shocks, the benefit of future access to interna-
tional capital markets, the position of other lenders, and domestic
political and economic factors (Cohen 1989).
Claessens and van Wijnbergen (1991, 1993) used the following pro-
cedure to derive the stochastic process for the net amount of financing
available each period to service loreign commercial bank debt. First,
they model nonoil exports, import requirements, and net scheduled
capital in- or outflows consistent with the constraints imposed by do-
mestic economic and political factors to obtain the expected nonoil,
noninterest current account (see Claessens and van Wijnbergen 1989).
428 VAN WIJNBERGEN AND BUDINA
Second, they adjust the nonoil, noninterest current account for debt
service to more senior claim holders, for foreign direct investment flows,
and for capital account transactions such as reserve accumulation (see
van Wijnbergen and Pena 1989). Third, oil earnings are added to the
flows. In accordance with the large share of oil in total exports and the
very high variance in oil prices, the behavior of oil exports introduces
the stochastic element in the net amount of financing available to ser-
vice the commercial bank debt.
The net amount of financing available for debt service may itself
depend on the debt reduction agreement and the provision of third
party guarantees. However, theory gives no clear indication which way
this effect goes. On the one hand, because of moral hazard strategic
interactions could reduce the financing available from the country once
a guarantee is provided. On the other hand, such interactions may
increase available financing because of the stronger bargaining power
of the third party providing the guarantee (see Bulow and Rogoff 1989
for a multilateral bargaining framework). Or financing might increase
because of the country's incentive to avoid a possible adverse signal
associated with defaulting. In the absence of strong analytical support
either way, we assume that the derived process is not altered by the
conversion of commercial bank debt into claims with guarantees.
Pricing Sovereign Debt
Within the above framework, we now introduce the option-pricing
model. Each period, the debtor country will pay to its creditors as
much as available financing allows, but never more than its contrac-
tual obligations in the period-that is,
(19.1) R^(t) = min[R,, FX,]
where R-(t) is the actual repayment, R, the contractual obligation, and
FX, the resources available to service commercial debt, all in period t.
Equation 19.1 can be rearranged to yield
(19.2) R-(t) = R,- max[O, R, - FXJ.
But max [0, R, - FX,] equals the value of a put option written on FX,
with a strike price of R,. Thus the uncertain repayment can be repre-
sented by a certain repayment, R,, minus a put. Having thus replicated
the payoff stream at maturity, we can obtain its current value, V, as the
current value of the certain repayment Rt minus the current value of
the put-that is,
(19.3) V(R,) = exp(-rt)*R, - P(FX,, R,, r, t, a)
where r is the (continuously compounding) interest rate and P(FX,, Rt,
r, t, ay) is the current value of a put written on FX, with exercise price
GUARANTEES AS OPTIONS 429
R,, interest rate r, maturity t, andl standard deviation (y.2 The current
value of a loan with a series R, falling due over time (where R, can be
different for each period, depending on the terms of the loan) is then
simply the sum of the current values of a series of these claims over the
maturity of the contract.
Pricing Guarantees and Recapture Clauses
The option-pricing framework also allows for straightforward pricing
of the various guarantees and recapture clauses if the latter exist. Con-
sider the principal and interest guarantees first.
The principal collateralizatiorn is equivalent to an unconditional
guarantee of the full payment, RT' at maturity. The implicit guarantee
is equivalent to a put option with an exercise price RrT maturity date T,
and written on FX, provided by the guarantor to the creditor. This put
exactly offsets the put option representing sovereign risk on this pay-
ment (see the discussion in the previous section).
Pricing the interest guarantee is more complicated. Recall that this
is done through an escrow account with a fixed number of interest
payments, say N, in it and no retention of interest: earned in the ac-
count. This is clearly a conditional guarantee, of value only if there is
in fact money left in the escrow account at the time it is called. This
would not be the case if in the intervening years there were at least N
calls on the fund. If there is at least one payment left in the account,
the pricing conditional on there being at least one payment in the fund
is the same as the unconditional guarantee pricing.
Thus if FX, is serially independent over time, as assumed here, the
pricing of the conditional put, PC, for the rolling guarantee in year t is
(19.4) PC(FX,, R,, r, t, a) = ON(t, Rit,, FX,,,, c)*P(FX,, R,, r, t, a)
where (t') is the set of t' preceding t. ON(t. . I, ) denotes the probabil-
ity that the guarantee has been called at most N - 1 times during ft'},
leaving at least one payment in the account. The value of ON(t, RWI,,
FX lp, a) can be derived using a simple recursive formula on the level
of the funds in the escrow account in conjunction with some obvious
initial conditions., The total value of the rolling interest guarantee
provided through the escrow account is then equal to the summation
of equation 19.4 over all years that interest payments are scheduled.
Pricing the Various Options
To obtain the values of the various options, we adopt the Black option-
pricing formula, which uses as its inputs forward prices for the under-
lying state variables (Black 1976). This is in contrast to the Black-Scholes
model, which uses the spot price as input. We therefore need the forward
430 VAN WIJNBERGEN AND BUDINA
prices for the foreign exchange available, the volatility of the forward
prices, and the interest structure. As noted, the variability of FX stems
predominantly from the uncertainty of the price of oil. We therefore
model the maturity structure of forward prices to be consistent with
market expectations for forward oil prices at the time of agreement.
For the forward prices, lognormality could not be rejected at rea-
sonable confidence levels, so the options can be priced using the Black
option-pricing formula (see Black 1976). We obtain an estimate of the
standard deviation of the forward prices by calculating the volatility
of futures prices for contracts of all maturities that were consistently
quoted on the New York Mercantile Exchange between July 1986 and
July 1990. This excludes the effects of the Persian Gulf crisis. We use
daily (settle) prices. The resulting term structure of volatilities for dif-
ferent maturities (number of months to expiration) indicates that the
volatility of the longer maturity contracts is much below that of the
nearby contracts and that volatility converges to a level of about 22
percent. In accordance with this decline in volatility, commercial insti-
tutions priced long-dated options using volatilities of about 20 percent
at the time the agreement was concluded. A standard deviation of
about 20 percent is also confirmed by the standard deviations implied
by actual market prices of oil options at the time the agreement was
reached. We therefore use a volatility of 20 percent as input in our
option-pricing model.
The Brady Debt Restructuring Package
The Brady plan for heavily indebted developing countries, announced
for the first time in 1989 and named after Secretary of the Treasury,
Nicholas Brady (1988-1993), sought to remove inefficiencies caused
by a debt crisis. The plan was based on a combination of commercial
bank debt reduction and adjustment lending programs designed by the
international financial institutions (IFls). Commercial bank debt re-
duction was to be achieved through market-based debt instruments
(see Krugman 1989) using the menu approach to debt rescheduling
and combining the concerted and voluntary characteristics of com-
mercial lending, thus keeping the advantages of pure market and pure
concerted mechanisms. The main role of the IFIs in these debt and
debt service reduction programs is to provide adjustment lending, con-
ditional on specific adjustments and policy reforms in country debt-
ors, which would allow them to utilize fully the benefits of such a debt
reduction (see Diwan and Rodrick 1992).
At the outset of its transition to a market economy, Bulgaria faced
difficult conditions produced mostly by the inherited heavy debt bur-
den, the collapse in exports stemming from the breakup of the former
Council for Mutual Economic Assistance (CMEA), and the overvalued
GUARANTEES AS OPTIONS 431
domestic currency. These difficult initial conditions resulted in a deep
economic recession and the virtual collapse of the country's repayment
capacity, which prompted Bulgaria to announce a debt moratorium in
1990. During 1990-1996, economic reforms were implemented in a
stop-and-go fashion, and inconsistent macroeconomic policies triggered
domestic currency and banking crises (see Dobrinsky 1996, 1997b;
Budina 1997). These factors contributed to lengthy negotiations on
Bulgaria's foreign debt reduction, which began in 1992, when Bulgaria
partially resumed servicing its co:mmercial debt, and were concluded
only in March 1994. Since then, both the principal and the interest on
the foreign debt have been reduced through exchanging the liabilities
for new instruments using U.S. Treasury bonds as collateral. The subject
of the agreement was all the Bulgarian Foreign Trade Bank (BFTB) debts
in foreign currency and the debts to foreign creditors guaranteed by
BFTB as of March 1990 when the moratorium on Bulgarian foreign
debt was announced (Alexandrov 1995).
Table 19.1 shows the structure of the foreign obligations and the
amount of different options chosen. The 1994 DDSR agreement con-
verted US$8.3 billion, which included the original debt and the past
due interest (all dollar amounts are in current U.S. dollars). The market-
Table 19.1 Structure of Bulgaria's Foreign Debt Subject
to the DDSR Agreement
Face value Percent of
(millions of debt eligible
U.S. dollars) for DDSR
Debt eligible for DDSR agreement 8,088.0 100.00
Original debt' 6,186.2 76.49
Buybacks 798.2 9.87
Discount bonds (DISCs) 3,730.4 46.12
Front-loaded interest reduction bonds (FLIRBs) 1,657.5 20.49
Past due interest (PDI) 1,901.8 23.51
Interest down payment 63.8 0.79
Buybacks 223.3 2.76
PDI bonds 1,614.7 19.96
Up-front debt reductionb 179.3
Debt prior to DDSR' 8,267.2
a. Original debt represents the face value debt to the London Club before the debt
moratorium period.
b. Up-front debt reduction is a World Bank estimate of the face value of the interest
reduction from the favorable treatment o0: interest since September 1992, the debt mora-
torium period (see World Bank 1994: 11-12).
c. Debt prior to DDSR refers to what would have been the face value debt to the London
Club if interest arrears had been accumulated based on market interest rates, or it would be
just the debt eligible for the DDSR plus the face value of up-front debt reduction.
Sources: Ministry of Finance and the Wlorld Bank.
432 VAN WIJNBERGEN AND BUDINA
based debt and debt service reduction was estimated at about 42 per-
cent in present value terms or at about 38 percent when excluding the
up-front costs of the deal.
Here "original debt" refers to the face value of the debt to the Lon-
don Club before the announcement of the debt moratorium, or about
$6.2 billion. Past due interest includes all unpaid interest on this original
debt since declaration of the debt moratorium in 1990. Since September
1992, Bulgaria has resumed partial interest payments. Up-front debt
reduction is a World Bank estimate of the face value of the interest
reduction from favorable treatment of interest since September 1992,
the debt moratorium period.4 Debt prior to the DDSR ($8.3 billion)
refers to what would have been the face value debt to the London Club
if interest arrears were accumulated based on market interest rates, or it
would be just the debt eligible for the DDSR agreement ($8.1 billion)
plus the face value of up-front debt reduction ($0.179 billion). Thus the
DDSR agreement contained some reduction of accrued interest for the
period in which partial interest payments were made.
The debt allocated to each of the available debt reduction options
as a share of debt eligible for the DDSR agreement ($8.1 billion) was:
debt buybacks, 9.87 percent buyback of the original debt and about 3
percent mandatory buyback of the adjusted interest arrears; discount
bonds (DISCs), 46.12 percent; front-loaded interest reduction bonds
(FLIRBs), 20.49 percent; and past due interest, out of which past due
interest bonds (PDls) were 19.96 percent and about 1 percent of past
due interest was down paid (see World Bank 1994: 12, table 3).
A graph of the pre-debt deal secondary market price appears in
Houben (1995). This price, which was extremely low in the beginning
of the 1990s, was estimated by the World Bank to be about 20 cents
per U.S. dollar. After Bulgaria's agreement in principle with the deal,
the secondary market price went up to about 30 cents per dollar.
Menu of Options
The menu of options available was standard for such an agreement:
cash buybacks (total of about $1 billion); FLIRBs (approximately $1.7
billion) with lower interest payments in the first seven years; DISCs
($3.7 billion) with a 50 percent discount, but carrying market interest
rates; and PDIs covering interest arrears ($1.6 billion). Table 19.2 pro-
vides a detailed description of the menu of options under the DDSR
agreement (see Alexandrov 1995 and Houben 1995).
Under the cash buyback option, agreement was reached for the re-
purchase of $1.02 billion, or 12.63 percent of the qualified foreign
debt, at a buyback price of 25.8 cents per dollar face value debt. The
collateralized discount bonds (DISCs) were issued at $250,000 par
value with a 50 percent discount from the face value of the principal.
GUARANTEES AS OPTIONS 433
Table 19.2 Bulgaria's Brady Dlebt Agreement with the London
Club: Menu of Options
Interest rate
Instrument Maturity (percent) Collateral Comments
Cash buy- 25.8 cents per
backs U.S. dollar of
original debt plus
related unpaid
adjusted interest
FLIRBs 18 years Year 1: 2.00 One-year interest Bearer bonds
(par (8 grace) Year 2: 2.00 initially at 2.6 Limited to no more
exchange) Year 3: 2.25 percent rising to than 30 percent of
Year 4: 2.25 3.8 percent. In- original debt
Year 5: 2.50 terest collateral 30 percent of
Year 6: 2.75 is required only bonds issued for
Year 7: 3.00 for the first seven short-term debt
Year 8: LIBOR years. will carry a 0.5
+ 13/16 No principal percent higher
collateral. interest rate.
DISCs 30 years LIBOR + Principal secured Registered bonds.
(50 per- bullet 13/16 by zero coupon 30 percent of
cent dis- U.S. Treasury bonds issued for
count bonds. short-term debt
on face Twelve months will carry a 0.5
value) interest guaran- percent higher
tee calculated interest rate.
at 7 percent
interest rate.
Past due
interest
Cash n.a. n.a. n.a. 3 percent of interest
payment arrears, paid on the
(3 percent) closing date
PDI bonds 17 years LIBOR + None Bearer bonds
(par (7 grace) 13/16 Amortization is
exchange) payable semi-
annually as the
following scheduled
installments:
1-6, 1 percent
7-1l, 3 percent
12-16, 6 percent
17-31, 9.8 percent
n.a. Not applicable.
Note: FLIRBs = front-loaded interest re(luction bonds; DISCs = discount bonds; LIBOR
= London interbank offered rate; PDI = past due interest.
Source: World Bank (1994).
434 VAN WIJNBERGEN AND BUDINA
They have a maturity of 30 years and a bullet amortization. The prin-
cipal and one year of the interest payment of the DISCs are collateral-
ized with U.S. Treasury zero coupon bonds. The DISCs also are subject
to the value recovery clause. Finally, the DISCs yield a six-month LIBOR
(London interbank offered rate) plus 13/16 percent and their interest
payments are made semiannually. Under the third option, the FLIRBs,
the overall principal bid was exchangeable one to one with the overall
loan principal. The FLIRBs were issued in two tranches at $250,000
par value with maturity of 18 years and were subject to a mandatory
buyback in 21 semiannual installments. The first buyback is due in
July 2002 and the last one in July 2012. Interest payments are due
each January and July. Seventy percent of the bonds issued under the
first tranche were in exchange for short-term debts with maturity of
up to one year. The FLIRBs yield lower interest payments in the first
seven years: years 1-2, 2 percent; years 3-4, 2.25 percent; year 5, 2.5
percent; year 6, 2.75 percent; year 7, 3 percent. After that, they yield
the market interest rate of LIBOR plus 13/16 percent a year. One year
of their interest payments is collateralized.
Bulgaria accumulated a large amount of interest arrears between
the announcement of the debt moratorium in mid-1990 and the initia-
tion of debt negotiations with the London Club. Several schemes were
used to deal with the unpaid interest. First, part of the interest arrears
were written off (the interest payments due were calculated by using a
lower fixed dollar interest rate of 3.5 percent). The estimated face
value of the written-off interest was $0.1793 billion. Second, $0.223
billion of the interest due was subject to the buyback. In addition,
Bulgaria had to make a down payment of $0.0638 billion of the inter-
est due. The rest of the interest arrears, $1.615 billion, was converted
to PDls. PDls, the last option, were issued on the closing day of the
agreement at $250,000 par value and were used to purchase all ad-
justed interest arrears on the qualified debts or related to them, which
were exchanged for FLIRBs or DISCs. The PDls mature in 17 years
and are subject to buyback requirements. They yield LIBOR plus 13/
16 percent a year. Their principal and interest are not secured. Al-
though it was proposed that the FLIRBs and DISCs be accepted in
debt-equity swaps (one to one for DISCs and for FLIRBs at a 50 per-
cent discount), this proposal was not embodied in the regulations on
privatization through foreign debt conversion.
Up-front Costs of the DDSR Agreement
At the time of the debt deal negotiations, the Bulgarian government
was faced with conflicting objectives: achieving greater debt reduc-
tion-but at the lowest cost. Table 19.3 lists the up-front costs of the
Bulgarian debt and debt service reduction agreement.
GUARANTEES AS OPTIONS 435
Table 19.3 Up-front Costs of DDSR Agreement
(millions of U.S. dollars)
Closing
Buybacks PDIs DISCs FLIRBs date
Down payment 257 64 n.a. n.a. 321
Principal collateral n.a. n.a. 220 n.a. 220
Interest collateral n.a. n.a. 131 43 174
T otal 257 64 351 43 715
Foreign exchange reserves,
excluding gold n.a. n.a. n.a. n.a. 1,300
n.a. Not applicable.
Note: PDls = past due interest bonds; DISCs = discount bonds; FLIRBs = front-loaded
interest reduction bonds.
Source: World Bank (1994).
The most important question is whether Bulgaria succeeded in
achieving an efficient agreement. Houben (1995) expresses the view
that the Bulgarian foreign debt reduction was comparable to the aver-
age reduction with the other debt deals. Because this reduction has
been achieved at a remarkably lovv cost, it has, on average, received a
positive evaluation. Yet Houben points out some important lessons for
the other countries that plan to initiate the Brady-type debt restructur-
ing agreements. First, the debtor country must orient its macroeco-
riomic and financial policies toward preserving financial stability
throughout the debt restructuring process. In addition, to conclude a
debt deal that ensures the desired outcome, the debtor country must
look for the necessary external financing before reaching the final debt
restructuring agreement. Last but not least, the debtor country must
specify explicitly in the debt agreement the amount of debt reduction
it would like to obtain.
Estimation Procedure
Now it is necessary to estimate the underlying stochastic process driv-
ing Bulgaria's repayments to the commercial banks. Figure 19.1 pre-
sents the daily data on the secondary market prices of various Brady
bonds from their issue on October 14, 1994, to July 10, 1997. Second-
ary market prices of various Brady bonds are expressed as ratio to 1
IJ.S. dollar face value debt. For example, low secondary market prices
would represent deep discounts of face value debt and often reflect
low demand to hold debt instruments of the particular country debtor.
Frequently, they reflect the low credibility of domestic macroeconomic
436 VAN WIJNBERGEN AND BUDINA
Figure 19.1. Secondary Market Prices of Bulgarian Brady Bonds
100
80
60
DICs' :
PDls
FLIRBs
20
10/14/94 2/26/96 7/10/97
Note: DISCs = discount bonds; FLIRBs = front-loaded interest reduction bonds;
PDIs = past due interest bonds. Secondary market prices of various Brady bonds are
expressed as ratio to 1 U.S. dollar face value debt.
Source: Authors' calculations.
policies, but at times they also may be subject to external shocks, such
as contagion effects from financial crises in other countries.5
Secondary market prices also reflect the presence of third party guar-
antees. Brady bonds often have (partial) guarantees on the principal or
interest. Because various claims differ, we cannot compare directly the
debt of different countries because they have different contractual terms.
And these contractual terms can differ over time-for example, bonds
that yield variable versus fixed interest rates. For these reasons, raw
data on the secondary market prices cannot be used as an approxima-
tion of the underlying stochastic process that drives the debtor country's
repayment capacity.
Our approach to this problem follows that of Claessens and van
Wijnbergen (1993). They used the prices of forward oil contracts as
underlying the stochastic process of Mexico's noninterest current ac-
count, because Mexico is an oil exporter. First, we estimate the volatil-
ity of prices of the dominant import commodity, oil; Bulgaria is an oil
importer. Then we estimate the initial value and growth rate of the
underlying process, FX.
Determination of the Repayment Capacity,
FXo and Its Growth Rate
The next parameter needed to apply the option-pricing model is the
starting value of the noninterest current account balance. We use the
(GUARANTEES AS OPTIONS 437
actual and estimated data on external capital and debt and the DDSR
agreement from the World Bank (see World Bank 1994: 55, annex 3).
Working with the actual and projected current account in millions of
IJ.S. dollars for the period 1991-2002, we net it out with the interest
payments to obtain the noninterest current account for the same pe-
riod. In addition, we subtract the net transfers to the official credi-
tors.6 Then we average the noninterest current account over the period
1994-2002 to obtain the initial value for Bulgaria's repayment capac-
ity, $0.26 billion. However, when we include the debt moratorium
period, 1991-1993, in our average value for the repayment capacity,
we obtain $0.142 billion. We use this amount to value the pre-debt
deal original debt, including the period of the debt moratorium. When
we take the average of the noninterest current account without the
official creditors' disbursements for the period 1994-2002, we obtain
$0.26 billion, which we use as an estimate of Bulgaria's repayment
capacity to value the debt restructiured through DDSR. We also use the
two different estimates for Bulgaria's repayment capacity to assess the
changes in the secondary market price of the pre-debt deal debt stem-
ming from resolution of the interest arrears problem.
The next problem is the growth rate of the noninterest current ac-
count. For our analysis so far, we have assumed that the drift in the
repayment capacity is equal to the risk-free interest rate, which is the
assumption of the Black-Scholes formula.
Risk-free Interest Rates
We approximate the risk-free interest rates by using the U.S. Treasury
bond yields and calculate the term structure of future interest rates on
the U.S. Treasury bonds. Data on the interest rates on U.S. Treasury
bonds with maturities of 1, 3, 5, 10, and 30 years were used to derive
the term structure of future interest rates.
If we assume that the interest rate on a one-year bond is the same
for years t + 2 and t + 3, and because we know the one-year bond rate
in the current year and the three-year bond rate in the current year, we
can express the future interest rate on the two-year bond. In this way,
we are able to provide the annualized interest rates on bonds with
maturities of from zero to 30 years.
According to the term structure hypothesis of the interest rates, the
interest rate on the three-year U.S. Treasury bond is represented as
(19.5) (1 + i3)3 = (I + il) (1 + i2t+1)2
where it is the yield on a three-year U.S. Treasury bond that is known
at time t; it is the yield of a one-year U.S. Treasury bond at time t; and
i2t+' is the future yield of a two-year U.S. Treasury bond. Therefore, we
can determine the future interest rate of two- and three-year bonds
438 VAN WIJNBERGEN AND BUDINA
from (19.1).7 The future yield of a four-year U.S. Treasury bond is
determined by
(19.6) (1 + it)5 = (1 + i;) (1 + it'+)2(1 +i4+1)2
where we assume that the interest rates of a four- and a five-year U.S.
Treasury bond are the same.
Next, we can represent the current yield on a 10-year U.S. Treasury
bond as
(19.7) (1 + it )'O = (1 + i,) (1 + it+')2(1 + i4t+)2(1 + it+')5
where we assume that the future yields of 6- to 10 -year U.S. Treasury
bonds are the same.
Finally, we represent the current yield on a 30-year U.S. Treasury
bond as
(19.8) (I + it )30- = (1 + i') (1 + i2')2(1 + i41)2(1 + i6 1)5(1 + ill)20
where we once again assume that the future yields of 11- to 30-year
U.S. Treasury bonds are the same. In this way, we are able to deter-
mine all future yields on U.S. Treasury bonds with different maturities
(see Figure 19.2).
Figure 19.2. Future Annual Interest Rates on U.S. Treasury
Bonds with Different Maturities
Percent
6.3
6.2
6.1
6.0
5.9
5.8
5.7
5.6
5.5
5.4
5 .3 . . . . . . . . . . . . . . . . .
USlY USSY USlOY US1SY US20Y US25Y US30Y
Maturity
Source: Authors' calculations.
GUARANTEES AS OPTIONS 439
Market Valuation of 'Bulgaria's Foreign Debt
before and after the DDSR Agreement
We begin with the market valuation of the external debt before the
clebt deal. Based on our model, we are able to present the effect of the
changes in the starting value of the repayment capacity and changes in
its volatility on the secondary market price of the debt. Next, we ana-
lyze the relationship between the face value and the market value of
the debt. In doing so, we present: the impact of different face value
cliscounts on the secondary market price and the market value of the
debt. We then price separately a 50 percent discount bond with a face
value equal to the original debt. We also evaluate the total Brady pack-
age and compare it with its pre-debt deal value and price. The debt
cleal will be beneficial for Bulgaria if the market value of the claims
without enhancements falls because of the agreement., which also means
a reduced foreign debt burden. If the enhancements are included, the
market value may still go up, but by less than the full value of the
enhancements. The debt deal will not be beneficial for Bulgaria if the
market value goes up with the full value of the enhancements. This
implies that there was no debt relief and that all the benefits accrue to
the private creditors at the expense of the third party's enhancement
effort. We also calculate the gross and the net debt relief and its equiva-
lent buyback price.
Valuation of Bulgaria's External Debt
before the DDSR Agreement
Here we apply the option-pricing approach outlined earlier in this
chapter to the value of the original amount of Bulgaria's external debt
before the DDSR agreement. However, one should be careful in deter-
mining the original value of the debt; there are two different measures
for it. This difference can be explained with the interest arrears that
accumulated between the announcement of the debt moratorium and
the beginning of the negotiations between the Bulgarian government
and the London Club. According to the DDSR agreement, the country
was obliged to recognize partially the interest arrears incurred since
the announcement of the debt moratorium. These interest arrears were
treated favorably, because the interest rate applied in this period was
fixed at 3.5 percent in dollar terms. This would therefore imply an up-
front debt reduction of about $0.179 billion. We use these two differ-
ent values for the pre-debt agreement debt, as well as our calculations
of Bulgaria's repayment capacity as explained in the previous section,
in order to find out the pre-debt deal secondary market price. To value
the pre-debt deal original debt, including the period of the debt mora-
torium, we estimate Bulgaria's repayment capacity as the average of
440 VAN WIJNBERGEN AND BUDINA
the noninterest current account without the net disbursements to the
official creditors, using both actual values and projections for the pe-
riod 1991-2002, or $0.142 billion. To value the DDSR restructured
debt, we estimate the repayment capacity as the average of the
noninterest current account without the official creditors' disburse-
ments for the period 1994-2002, $0.26 billion.8 We start by using the
option price model discussed earlier in this chapter to value the debt
prior to DDSR agreement ($8.27 billion). In doing so, we use the higher
estimate of the present value of the repayment capacity ($0.26 bil-
lion), which is obtained after excluding the debt moratorium period.
This simulation yields a market value of $2.4 billion, which corre-
sponds to an implied secondary market price of 29 cents per dollar of
face value debt (after settlement of the interest arrears).
We then use the same model to calculate the market value and the
implied secondary market price of the debt prior to the DDSR agree-
ment ($8.27 billion) before the official settlement of the accumulated
interest arrears. In this case, we also use the lower value for Bulgaria's
repayment capacity, estimated at $0.142 billion. Thus we obtain 20
cents per dollar face value debt as an estimate for the secondary mar-
ket price of the original debt, $8.27 billion, during the period of the
debt moratorium. This means that the market value of the original
debt before the official settlement of the interest arrears was equal to
$1.65 billion.9 Houben (1995) presents a graph with the secondary
market prices of Bulgaria's external debt. The secondary market price
of the foreign debt during the period 1990-1992 was about 20 cents
per dollar of face value. However, once the agreement was reached in
principle in 1993, but before the agreement was signed, the secondary
market price went up to the high twenties.
Table 19.4 presents the secondary market price and market values
of the pre-debt deal debt during the debt moratorium period and after
resolution of the interest arrears problem. Thus, to analyze the ben-
efits of the DDSR agreement, we compare the results for the new sec-
ondary market price and market value with the two different market
prices of the old debt-first, before the settlement of the interest ar-
rears and, second, after the agreement on the interest arrears was
reached.
Market Value of the Foreign Debt after the DDSR Agreement
In this section we analyze the total impact of Bulgaria's DDSR agree-
ment with the London Club. We obtain the market value of the post-
DDSR agreement debt by summing up the market values of the
individual instruments. When we price each instrument, we take into
account its share of the total original debt, any principal or interest
guarantees, and the present value of Bulgaria's repayment capacity'°
GUARANTEES AS OPTIONS 441
Table 19.4 Secondary Market Price and Market Value
of Original Debt before and ai.ter Debt Management
Including the Without the debt
debt moratorium moratorium
Face value $8.27 billion $8.27 billion
Secondary market price $0.20 $0.29
Market value $1.65 billion $2.40 billion
Repayment capacity $0.142 billion $0.26 billion
Source: World Bank and authors' calculations.
and its volatility. We also take into account the interest arrears down
payment and the amount of buyback and its price.
Figure 19.3 presents the market value of the debt derived from our
valuation model. The first column represents the market value of the
pre-DDRS agreement debt ($8.2 7 billion), prior to settlement of the
interest arrears problem, which is estimated at $1.65 billion. The im-
plied secondary market price, therefore, is 20 cents per dollar face
value debt. The second column represents the market value of the pre-
DDSR agreement debt after the resolution of the interest arrears prob-
lem, estimated at $2.40 billion. This yields an impliecl secondary market
price of 29 cents per dollar face value debt. This estimate is consistent
with the actual secondary market: price of the Bulgarian foreign debt
Figure 19.3. Pre- and Post-Debt Deal Market Value
of the DDSR Debt
Market value (billions of U.S. dollars)
3.0
Pre- and Post-Debt Deal Market Value
2.5 -.
2.0-
1.01-
0.51
0 .. - ......-...- . . - ^ ....
ValueOLD ValueOLD1 ValueNG ValueG
Source: Authors' calculations.
442 VAN WIJNBERGEN AND BUDINA
when the DDSR agreement was concluded in principle. The third col-
umn represents the market value of the reduced debt after the DDSR
agreement, but without the value of the principal and interest guaran-
tees. The face value of the new debt is now equal to $5.13 billion. The
secondary market price of the reduced debt after the DDSR agree-
ment, but not taking into account the collateral, amounts to 38 cents
per U.S. dollar face value debt, which means that the market value of
the debt after the DDSR agreement without enhancements is now $1.97
billion. However, the price of the debt after the DDSR agreement with
guarantees increases up to 50 cents per U.S. dollar, and the market
value of the reduced debt with enhancements rises to $2.56 billion.
Thus, although the face value of the debt has been reduced with the
DDSR agreement, its market value with enhancements has increased
from $2.40 billion to $2.56 billion. Without guarantees, the market
value of the debt is lower ($1.97 billion) as compared with $2.40 bil-
lion before the DDSR.
Amount of Debt Relief Implied. Table 19.5 lists the absolute amount
of the debt reduction provided under the different options, the debt
reduction for the various instruments expressed as a percentage of the
new face value ($5.13 billion), and the percentage of debt relief rela-
tive to the old face value ($8.27 billion). Note that in the debt reduc-
tion provided by the DDSR agreement we include the up-front debt
relief and a favorable treatment of the interest arrears accumulated
Table 19.5 Debt Relief Implied by All Options
Debt relief as Debt relief as
Debt relief percent of percent of
(billions of new face value old face value
Debt instrument U.S. dollars) ($5.13 billion) ($8.27 billion)
DISCs 1.86 36.2 22.5
FLIRBs 0.32 6.2 3.9
Buybacks 1.02 19.9 12.3
PDIs 0.00 0.0 0.0
DIAP 0.06 1.2 0.8
Subtotal 3.27 63.6 39.5
Up-front debt reduction 0.18 3.5 2.2
Total 3.44 67.1 41.7
Note: DISCs = discount bonds, FLIRBs = front-loaded interest reduction bonds; PDls =
past due interest bonds; DIAP = interest arrears down payment.
a. The up-front debt reduction is made by capitalizing the nonpaid interest using a
lower dollar interest rate of 3.5 percent. This estimate is provided by the World Bank.
Source: Authors' calculations.
GUARANTEES AS OPTIONS 443
between the moratorium announcement in 1990 andI settlement of the
arrears before the debt deal was concluded with the commercial credi-
tors. This relief amounts to $0.175c billion. Therefore, we calculate the
percentage of debt relief, taking into account what the original debt
would have been without such a favorable treatment, $8.27 billion.
T hus the total value of the gross debt relief amounts to $3.4 billion, or
about 42 percent of the pre-DDSR agreement face value debt. This
suggests that Bulgaria's debt deal was on average successful.
The largest percentage of debt relief was provided by the discount
bonds, 22 percent of the original face value; the next type of instru-
ment, buybacks, accounted for about 12 percent of the original face
value. The FLIRBs accounted for about 6 percent of debt relief as a
percentage of the new face value, or as 3.9 percent of the old face
value. Finally, the favorable treatment of the interest arrears provided
in debt relief 3.5 percent of the new face value and about 2 percent of
the old face value.
Secondary Market Valuation ol the New Instruments. This section
describes the results of applying the option valuation model for the
market values and the secondary market prices of various instruments
available under the DDSR agreement. The face value of DISCs that
yield 50 percent debt reduction accounts for the largest share of the
face value of the pre-DDSR debt. Both the FLIRBs (interest reduction
bonds) and the PDIs (new money claims) account for about 20 percent
in the face value pre-DDSR debt.
Table 19.6 presents the percentage share of market value (with and
without guarantees) of different instruments in the total pre- and post-
Table 19.6 Projected Secondary Market Valuation of New
Instruments
(share of the new instruments)
Without guarantees W'ith guarantees
Percent of Percent of Percent of Percent of
old face new face new face old face
value' valueh value' value'
DISCs 13.19 21.24 30.59 18.99
FLIRBs 7.74 12.47 14.61 9.07
PDIs 2.90 4.68 4.68 2.90
Total 23.83 38.72 49.88 30.97
Note: DISCs = discount bonds; FLIRBs = front-loaded interest reduction bonds; PDIs =
past due interest bonds.
a. The pre-DDSR face value of $8.27 billion.
b. The new face value of restructured deht, $5.13 billion.
Source: Authors' calculations.
444 VAN WIJNBERGEN AND BUDINA
debt deal debt. A comparison of the second and the third columns
reveals the impact of the guarantees on the market value of the debt.
The difference between the two columns stems from the principal and
interest collateral of the discount bonds-their market value increases
from 21.2 to 30.6 percent of the new face value debt. The difference of
about 10 percent in the two market values is due to the effect of the
principal and interest guarantees. The market value of the FLIRBs
increases from 12.5 to 14.6 percent of the new face value debt. The
increase in the market value in this case is smaller because only one
year of payments is guaranteed.
Assessment of Bulgaria's DDSR Agreement
with the London Club
The debt deal will be beneficial for Bulgaria if the market value of the
claims without enhancements goes down because of the agreement,
which also means a reduction in the burden of the foreign debt. If the
enhancements are included, it may still go up but by less than the full
value of the enhancements. The debt deal will not be beneficial for
Bulgaria if the market value goes up with the full value of the enhance-
ments. This means that there was no debt relief and that all the ben-
efits accrue to the commercial banks at the expense of the third party's
enhancement effort.
Figure 19.4 presents the pre- and post-debt deal market value of
the Bulgarian foreign debt subject to the DDSR agreement. The first
two columns of the graph represent the market value of the debt deal
before the DDSR agreement. The first column is calculated under the
assumption that during the debt moratorium in the early 1990s, the
secondary market price of the debt was 20 cents per dollar. Given the
original face value, this yields a market value of $1.65 billion during
the time of the debt moratorium. However, during the DDSR negotia-
tions the secondary market price of the old debt went up-our model
estimates it at 29 cents per dollar face value debt. This was the situa-
tion in 1993, when Bulgaria had recognized the interest arrears in-
curred in 1991 and 1992 and there was an agreement in principle on
the debt reduction. Given the original face value of the debt, this price
implies a market value of the debt of $2.40 billion. We have used the
second value, $2.40 billion, as a basis for our comparison, because the
debt moratorium is not sustainable in the long run.
The third and the fourth columns of Figure 19.4 represent the mar-
ket value of the debt after the DDSR agreement, without and with the
enhancements. The difference between the market value of the post-
deal debt with and without enhancements represents the market value
of the enhancements, $0.59 billion-that is, $2.56 billion minus $1.97
billion. Assuming that Bulgaria has borrowed all the funds to provide
GUARANTEES AS OPTIONS 445
Figure 19.4. Evaluation of the Benefits of the DDSR
Agreement, Including Guarantee Costs
iMarket value (billions of U.S. dollars)
3.0
Pre- and Post-Debt Deal Market Value
2.5
2.0 -
1.5
1.0
0.5
ValueOLD ValueOLD1 ValueNG ValueG
Source: Authors' calculations.
the guarantees, we now look more closely at how the benefits from
these third party enhancements are distributed between Bulgaria and
its commercial creditors.
First, we compare the market value of the debt before the deal,
$2.40 billion, with the market value of the debt after the debt deal but
without enhancements, $1.97 billion. The difference of $0.43 billion
represents the benefits accrued to Bulgaria. Second, we compare again
the pre-debt deal market value of the debt, $2.40 billion, with the
market value of the debt after the deal, but including the enhance-
ments, $2.56 billion. The difference of $0.16 billion represents the
benefits accrued to the commercial banks. In terms relative to the total
market value of the guarantee funds, Bulgaria has gained approxi-
mately 73 percent of the total benefits, and the comnmercial creditors
have gained about 27 percent. This result suggests that the DDSR agree-
ment was beneficial for Bulgaria, and it is comparable with the Mexi-
can debt deal (see Claessens and van Wijnbergen 1989, 1993).
According to Claessens and van Wijnbergen (1993), Mexico struck a
very good bargain because its benefits from the agreement varied be-
tween 76 and 97 percent of the provision of official funds.
The situation looks very much different once we realize that not all
of the guarantee funds were provided by the official creditors. If we
assume in the extreme case that Bulgaria provided all the guarantee
funds from its own resources, then the result is that all the benefits
from the increase in the market value due to the guarantees went to
446 VAN WIJNBERGEN AND BUDINA
the commercial banks, not Bulgaria. However, this is not a plausible
conclusion, because, for one thing, Bulgaria was partially reimbursed
for its up-front costs. According to the World Bank figures (1994), the
total up-front costs of the DDSR agreement amounted to $0.715 bil-
lion. Of that amount, $0.22 billion was used as principal collateral on
the discount bonds; $0.174 billion was used as interest collateral on
both the discount and FLIRB bonds; and the rest, $0.321 billion, was
used for buybacks and the interest arrears down payment. Therefore,
the total up-front cash used for guarantees is $0.394 billion, whereas
we estimate their market value at $0.59 billion." The total official
support for the DDSR agreement was $0.241 billion,'2 or approxi-
mately 61 percent of the total up-front costs of the guarantees. If we
assume that the market value of the official support is again 61 per-
cent of the market value of the total guarantee costs, $0.59 billion,
this implies that the market value of the third party guarantees fund is
equal to $0.35 billion. This means that Bulgaria has to add the part of
the market value of the guarantees that was not covered by the inter-
national institutions to the market value of the nonguaranteed debt,
implying that the debt reduction in this case is much lower. This is also
shown in Figure 19.4.
In this case, the difference is that the direct benefit to Bulgaria is now
much lower: Bulgaria now gains only $0.19 billion, as compared with
$0.43 billion, which represents the potential gains for Bulgaria if the
country had been reimbursed in full for these up-front costs. The gain
accrued to the commercial banks is the same as before, $0.16 billion.
The total amount of the benefits is $0.35 billion. Thus the distribution
of the benefits is as follows: 54.3 percent of the benefits accrue to Bul-
garia and 45.7 percent to the commercial banks. This result shows that
if we account for the fact that part of the guarantee funds were paid by
Bulgaria, the debtor country succeeded in capturing a much lower share
of the total benefits from the DDSR agreement.
However, such a debt reduction agreement also has indirect ben-
efits, including future access to the official credits after the debt deal.
Currently, Bulgaria has gained access to World Bank and IMF loans
intended to provide the foreign exchange needed to introduce the cur-
rency board system and to speed up the structural reforms.
Conclusion
We have applied the option valuation approach outlined to assess the
benefits of Bulgaria's DDSR agreement with its commercial creditors.
Under the assumption that Bulgaria has borrowed all the up-front cash
needed for the guarantee funds, our analysis shows that Bulgaria suc-
ceeded in retaining most of the benefits (about 73 percent of the mar-
GUARANTEES AS OPTIONS 447
ket value of the guarantee funds). Commercial banks also gained from
the DDSR agreement (approximately 27 percent of the funds). In this
respect, the Bulgarian DDSR agreement is comparable with the Mexi-
can agreement, which was rated favorably among other debt restruc-
turing agreements. However, if we take into account that Bulgaria
provided about 40 percent of the total costs of the guarantee funds
from its own sources, lower debt relief is revealed, as well as a differ-
ent distribution of the benefits between commercial banks and the
debtor country: Bulgaria now captiares slightly above half of these ben-
efits, as compared with 73 percenit in the previous case.
Bulgaria's DDSR agreement provided a gross debt reduction of 42
percent of pre-debt deal face value debt, and about 38 percent net
debt reduction, accounting for up-front costs of the deal.
The agreement can lead to less uncertainty about future fiscal and
foreign exchange developments associated with the debt overhang. It
can therefore increase the probability of success of the reform pro-
gram and can lead to restoration of future growth through regaining
access to the international capital markets. Bulgaria benefited in three
ways: the debt overhang was reduced significantly; the maturity struc-
ture of its debt obligations was improved; and the future uncertainties
associated with the debt overhang were reduced, making it possible to
regain access to the international capital markets and renew growth
recovery prospects.
However, slow structural reforns and frequent political and finan-
cial instability prior to 1997 resulted in negligible external commer-
cial inflows and a relatively low level of foreign direct investment
compared with that in relatively more advanced transition economies.
Furthermore, although the debt overhang was decreased significantly,
the debt burden remains high, which means that during the next de-
cade Bulgaria must undertake significant fiscal tightening to be able to
service its debt without difficulties.
The introduction of the currency board system in 1997 (see
Dobrinsky 1997a), together with the strong political commitment to-
ward radical restructuring, will give Bulgaria an opportunity to grow
out of its debt.
Notes
1. The authors thank the European Commission for the financial
support provided through the ACE l'hare Research Project 1996 Programme
(ACE-PHARE Research Project P96? 6028-R).
2. We use a constant interest rate r for notational convenience only. In the
application, we allow for the maturity structure of interest rates. Even though
the state variable FX is a nontraded asset, and as such is not priced directly in
448 VAN WIJNBERGEN AND BUDINA
the market, it is likely spanned by instruments that are traded and whose
current values are known. For example, forward and futures contracts on oil
can span Mexico's oil earnings. As a result, the option-pricing model can he
used identically as in the case of traded assets. In particular, we use the Black
(1976) formula for commodity-based options, which uses forward prices for
FX rather than the spot price as in the more standard Black-Scholes pricing
formula.
3. Note that 0N(t, Rlt.). FX1f, C) = 1 for t < N. Also note that it is logically
impossible to get more than N - 1 calls in less than N payment periods.
4. For more details, see the description of past due interest bonds at the
end of this section and World Bank (1994: 1112).
5. For a more detailed discussion and empirical analysis of the determi-
nants of the secondary Brady bond prices of Bulgaria's debt, see Budina and
Mantchev (2000).
6. We calculate the net transfers to official creditors as repayments due
to official creditors minus the disbursements from the official creditors plus
the interest owed the official creditors.
7. We assume that the future yields of the two- and three-year U.S. Trea-
sury bonds are the same.
8. The use of the two different estimates for Bulgaria's repayment capac-
ity to assess the changes in the secondary market price of the pre-DDSR
agreement debt is needed to account for the impact of resolution of the inter-
est arrears problem on the secondary market price of the debt.
9. Our results are consistent with the World Bank estimates and with
data shown in Houben (1995).
10. The present value of the repayment capacity is calculated as the aver-
age noninterest current account, without the net disbursements to the offi-
cial creditors for the period 1994-2002.
11. The difference accounts for the fact that the up-front cash payment
for interest collateral is smaller. According to the World Bank (1994), at the
beginning the interest payments are collateralized at a level of 2.6 percent,
and the earnings on the interest collateral are to be retained until the interest
collateral reaches 3 percent of the outstanding amount of FLIRBs. The earn-
ings on the interest collateral in excess of 3 percent are to accrue to the
Bulgarian government.
12. The main official creditors were the IMF, the World Bank, and the
government of the Netherlands.
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Alexandrov, S. 1995. "Bulgarian Brady Bonds." Bank Review, Quarterly Jour-
nal of National Bank of Bulgaria (1).
GUARANTEES AS OPTIONS 449
Black, Fischer. 1976. "The Pricing of Commodity Contracts." Journal of
Financial Economics. January-March.
Budina, Nina. 1997. "Essays on Consistency of Fiscal and Monetary Policy
in Eastern Europe." Ph.D. diss. Tinbergen Institute Research Series No.145.
Budina, Nina, and T. Mantchev. 2000. "Determinants of Bulgarian Brady
Bond Prices: An Empirical Assessment," World Bank Working Paper 2277.
Washington, D.C.
Bulow, J., and K. Rogoff. 1989. "Sovereign Debt Repurchases: No Cure for
Overhang." NBER Working Paper No. 2850. National Bureau of Eco-
nomic Research, Cambridge, Mass.
Claessens, Stijn, and Sweder van Wijnbergen. 1989. "An Option-Pricing
Approach to Secondary Market Debt (Applied to Mexico)." International
Economics Department and Latin America and the Caribbean Country
Department II, Working Paper WPS 333. World Bank, Washington, D.C.
. 1991. "The 1990 Mexico and Venezuela Recapture Clauses: An Ap-
plication of Average Price Options." Journal of Banking and Finance 17:
733-45.
. 1993. "Secondary Market Prices and Mexico's Brady Deal." Quar-
terly Journal of Economics 108 (November): 965-82.
Cohen, D. 1989. "A Valuation Forniula for LDC Debt with Some Applica-
tions to Debt Relief." World Bank, Washington, D.C. Processed.
Diwan, I., and D. Rodrik. 1992. "Debt Reduction, Adjustment Lending, and
Burden Sharing." NBER Working Paper No. 4007. National Bureau of
Economic Research, Cambridge, Mass.
Dobrinsky, R. 1996. "Monetary Policy, Macroeconomic Adjustment and
Currency Speculation under Floating Exchange Rates: The Case of Bul-
garia." Economics of Transition 4 (1): 185-210.
. 1997a. 'The Currency Board in Bulgaria: First: Experience." Sofia.
Processed.
. 1997b. "Transition Failures: Anatomy of the Bulgarian Crisis." Pa-
per prepared for the ACE Workshop on Exchange Rate and Capital Flows
in Transition Economies. Sofia.
Houben, A. 1995. "Commercial Bank Debt Restructuring: The Experience
of Bulgaria." IMF Paper on Policy Analysis and Assessment PPAA/9S/6.
International Monetary Fund, Washington, D.C.
Krugman, P. 1989. "Market-Based Debt-Reduction Schemes." In J. A. Frenkel,
M. P. Dooley, and P. Wickham, eds. Analytical Issues in Debt. Interna-
tional Monetary Fund, Washington, D.C.
van Winjbergen, Sweder, and S. Pena. 1989. "Mexico's External Debt: Struc-
ture and Refinancing Options." World Bank, Washington, D.C. Processed.
World Bank. 1994. "Proposed Debt and Debt Service Reduction Loan to the
Republic of Bulgaria to Support I)DSR Program of Bulgaria." Report No.
P-6403-BUL, September 7. Washington, D.C.
CHAPT'ER 2ZO
The Fiscal Risk of Floods:
Lessons of Argentirna
Alcira ]Kreimer
Worla! Bank
TYPICALLY, AFTER A DISASTER OCCURS in a developing country its
government is forced to reallocate funding from development activi-
ties to relief and reconstruction efforts. In many countries, devastating
natural disasters are a frequent occurrence. A natural disaster occurs
when an extreme event such as a flood affects vulnerable investments
or structures and the consequences surpass the society's ability to re-
spond. In Argentina, like in other countries, accelerated changes in
demographic and economic trends have disturbed the balance existing
in ecosystems, increasing the risk of human, social, economic, and
financial losses. Rapid population growth in vulnerable areas increases
pressures on natural resources and the environment and raises the con-
sequent risks associated with hurnan activities. Major disasters not
only damage capital assets but have long-term effects on the economy
as well. Disasters can seriously dislocate a country's economy and trig-
ger shortfalls in tax revenues, create fiscal deficits, increase levels of
public debt, and affect the overall economic performance measured by
negative changes in the gross domestic product (GDP), balance of pay-
ments, and gross capital investments
Despite the repeated experience of disasters, government officials
often react with surprise at the results of an "act of God." Awareness
of risk to extreme natural events is not well established in the process
of public and private investment decisionmaking. According to Munich
Re (1999), in the decade of 1988-97 floods accounted for over half of
the 390,000 recorded fatalities and a third of the damage from all
natural catastrophes worldwide. Kunreuther and Linnerooth-Bayer
451
452 ALCIRA KREIMER
(1999) found that, despite the far greater capital stock in the devel-
oped world, most of the US$233 billion in global flood losses over the
past decade occurred in developing countries (all dollar amounts in
chapter are current U.S. dollars).
While the precise scale, time, or location of disaster events cannot
be predicted, areas of hazard and vulnerability can be identified. Ag-
gregate losses can be anticipated at the national level, and contingent
provisions can be made for extreme events. Governments can then
incorporate disaster losses into the economic and fiscal planning pro-
cess, as indicated in the sections that follow.
Assessing the Risk: Exposure of the
Argentine Government to Flood Risk
Flooding is the major natural hazard in Argentina, where the phenom-
enon poses a major challenge to development. Floods have destroyed
human, social, and physical capital, and they have derailed economic
and fiscal planning, because funds are reallocated from ongoing pro-
grams to finance relief and reconstruction assistance.
The most distinctive geographic feature of Argentina is the delta
formed by the conjunction of three great rivers: the Paraguay, Parana,
and Uruguay. The waters of the combined Parana and Uruguay Riv-
ers, which drain one-fourth of the waters of South America for a wa-
tershed that covers more than 4.1 million square kilometers, flow into
the Rio de la Plata and then out to sea. Heavy precipitation through-
out the year and the country's flat topography often combine to inun-
date the floodplains of all four rivers. The coastal areas of the
northeastern and central parts of the country also are at risk, and storm
surges pose a threat to regions along the coast.
Dangerous conditions occur alarmingly often. For example, hydro-
logical data for the city of Corrientes reveal that flooding takes place
in two of every three years. Since 1957, Argentina has had 11 major
floods. The floodplains in the country cover more than a third of Ar-
gentina. Within that area are the nation's most developed agricultural
and industrial zones, an extensive transportation network, and two major
hydroelectric dams (Kreimer, Kullock, and Valdes 2001). Of the 11 major
floods in Argentina, three of them have caused direct damage in excess
of $1 billion: the 1983 flood with damages of $1.5 billion; the 1985
flood with damages of $2 billion; and the 1998 flood with damages of
$2.5 billion. According to statistics developed by Swiss Re (1998),
Argentina is among the top 18 countries worldwide with potential
flood losses in excess of $3 billion. As for measuring flood losses as a
percentage of GDP, Argentina is one of 14 countries whose potential
losses to floods are greater than 1 percent of GDP (Freeman 2001). In
THE FISCAL RISK OF FLOODS 453
Latin America, only Ecuador has a higher GDP exposure from flood
risk. In pure economic loss terms, Argentina has the highest risk in
Latin America; however, not all members of society perceive the risk
as high and respect safety in planning and development.
Incorporating Catastrophic Exposure
into Government Fiscal Planning
A model incorporating the impaci: of floods into Argentina's national
economic and fiscal planning estimates that the country faces an an-
nual contingent exposure of $650--950 million in lost capital stock due
to flooding (Freeman and others forthcoming). The model traces the
potential sources of replacement capital finance and shows how incor-
porating contingent catastrophe losses can lead to better estimation of
macroeconomic performance.
Using estimates of $2 billion in losses from one-year flooding events,
$9 billion in losses from 100-year events, and $12 billion in losses
from 1,000-year events, Freeman and others (forthcoming) trace the
macroeconomic costs of those losses assuming that Argentina rebuilds
immediately after a catastrophe. Incorporating catastrophes into mac-
roeconomic planning indicates that accounting for flood exposure con-
tributes to the following effects:
* Increases estimates of requiired public investment by an average
of $801 million a year.
* Reduces projected levels of government consumption by $200
million a year.
* Decreases private consumption projections by as much as $1,647
million a year, or increases private consumption projections by as much
as $1,671 million a year. In the year of a catastrophe, private con-
sumption drops dramatically. If, however, the catastrophe is followed
by inflation and an appreciation of the real exchange rate, then in the
years following the catastrophe imports will increase, exports will de-
crease, and private consumption will increase, to the detriment of an
already deteriorating resource balance and growing foreign debt.
* Increases import projections by, on average, $429 million a year
and decreases export projections by, on average, $211 million a year.
* Increases foreign debt projections by $2,689 million a year.
Risk R.(eduction
Because government exposure to flood risk has such an important
impact on fiscal planning, it is important to understand what can be
454 ALCIRA KREIMER
done to reduce such exposure. Efforts to prevent flood losses consist of
(a) protecting existing development through flood control works (for
example, reservoirs, channel improvements, diversions, flood warning
and evacuation, and flood proofing); (b) removing existing develop-
ment through public acquisition, urban redevelopment, public nuisance
abatement, nonconforming uses, conversion of use of occupancy, and
reconstruction; (c) discouraging development through public informa-
tion, warning signs publicizing hazards, tax assessment practices, pub-
lic facility expansion, and flood insurance; and (d) regulating floodplain
uses through zoning ordinance districts, special floodplain regulations,
subdivision ordinances, and building codes (United Nations 1976). Early
on, the predominant efforts in Argentina focused on structural mea-
sures, such as the construction of dams, embankments, or levees.
But structural mechanisms to control floods are not enough; they
must be coupled with nonstructural measures. Furthermore, some flood
control projects may be counterproductive, because they may foster
unrealistic expectations that all flooding can be prevented and stimu-
late people to move onto floodplains, thereby increasing total cata-
strophic exposure. For example, investigators found that in the part of
Argentina affected by the floods of 1991, works implemented in the
past, such as makeshift protective earthworks, had been built without
a well-thought-out plan or a basic understanding of the local topogra-
phy. Thus settlements commonly protected themselves by building a
levee along the river, but then they did nothing to protect themselves
from the nearby creek that fed into it. Rising water in the river raised
the level of the creek, and floodwaters came into the settled area from
an unanticipated direction.
Recent efforts such as the Argentina Flood Rehabilitation Project,
prepared by the government with financing from the World Bank in
1992, have moved from purely structural measures to support for the
development of a national strategy to help the country cope more ef-
fectively with the recurrent floods. Thus, in addition to financing flood
recovery needs, the project financed the development of an institu-
tional framework for the coordination and implementation of govern-
ment rehabilitation efforts (World Bank 2000).
Risk Management: Legal and Institutional Framework
River basin and floodplain management in Argentina is a fragmented,
disorganized sector that lacks adequate infrastructure, appropriate
environmental standards and regulations, and institutional capacity.
The two most generic and notable features of the sector can be sum-
marized as follows. First, the provinces have primary jurisdiction over
water resources (except for navigation) even though many river basins
TIHE FISCAL RISK OF FLOODS 455
are interprovincial, and the national legal and institutional framework
does not adequately address this imbalance. Second, the characteris-
tics of water resources vary greatly among the country's regions and
provinces, and there is no consolidated policy and institutional frame-
work, which has resulted in significantly uneven and poorly coordi-
nated infrastructure development.
Although Argentina has significant national water legislation, it lacks
the basic framework that would allow provinces to establish consis-
tent regulations for users' rights and obligations, the economic value
of water, the river basin as the primary management unit, and
multisectoral user's organizations, among other things. At the provin-
cial level, legal frameworks are unevenly developed and inconsistent.
Some provinces have well-developed regulations, while others have no
specific laws on some of the issues that affect theni. In other cases,
existing legislation is very old and does not incorporate new concepts
such as multisectoral management, economic criteria, institutional
development, and so forth.
Gaps in the management of the flood problem are twofold. On the
one hand, there is a lack of norms for environmental management for
flood risk. For example, the resources devoted to meteorological and
warning systems are not well used, and the programs of public invest-
ment in water resource management are not part of an overall inte-
grated planning system. The fiscal policy system exerts only a minimal
degree of efficiency in the ex post -assistance efforts to support recov-
ery after floods. Finally, the handling of flood management issues by
government agencies is characterized by weak coordination, poor fi-
nancial control, and poor programming of public investments and fis-
cal policy. On the other hand, there are important gaps in the
coordination among the different jurisdictions that administer the flood
problem. In the metropolitan area of Buenos Aires, which is at high
risk from floods, 14 municipalities, in addition to the provincial gov-
ernment and the federal governmenrt, have jurisdiction over the flood
problem. There is a duplication of functions in some of these institu-
tions, and resources are devoted to similar functions., but they are ap-
plied in the context of government policies that are often opposed and
that are defined independently of each other. There is no articulation
of policies and programs, and resources and knowledge are not shared
across institutions. Furthermore, a unified and integrated fiscal policy
on1 government exposure to flood r isk is lacking.
Risk Transfer
Kunreuther and Linnerooth-Bayer (1999) identify two mechanisms avail-
able to governments to fund the costs of recovery: hedging instruments
456 ALCIRA KREIMER
and financing instruments. Hedging instruments are predisaster arrange-
ments in which the government incurs a relatively small cost in return
for the right to receive a much larger amount of money after a disaster
occurs. Insurance and capital market-based securities are examples of
hedging instruments. Financing instruments are arrangements whereby
the government either sets aside funds prior to a disaster or taps its own
funding sources after the event occurs. An example of a predisaster mea-
sure is a public catastrophe fund in which the government implicitly
self-insures by setting aside money to finance some of the recovery needs
after a disaster. An example of this type of mechanism in an emerging
economy is a calamity fund, such as FONDEN in Mexico (Kreimer and
others 1999). Funds earmarked for relief and reconstruction by the fed-
eral government are maintained in FONDEN, which provides funds for
the repair of uninsured infrastructure, immediate assistance to restore
the productivity of subsistence farmers, and relief to the low-income
victims of disasters. There is no such fund in Argentina.
Flood insurance has been introduced successfully in many parts of
the world, and it is particularly effective in areas where floods are
frequent. However, in most developing countries the full cost of insur-
ance in high-risk areas is beyond the means of many property owners.
Insurance premiums must be based on a full assessment of risks by
using flood hazard maps for the flood-prone areas. In some developed
countries such as the United States and Australia, the national govern-
ments have introduced subsidized flood insurance programs to pro-
vide some relief for disaster-prone areas. In the United States, flood
insurance is tied to measures to reduce losses. Flood-prone communi-
ties may buy subsidized flood insurance only if both the homeowners
and the community make efforts to reduce the risk of the insured struc-
tures to floods.
Historically, and like other emerging markets, the government of
Argentina has absorbed the losses from floods. In principle, flood cov-
erage is available as a supplement to fire insurance, because noncom-
mercial and commercial risks are covered under fire policies. However,
even though private insurance that can absorb a portion of the costs of
flooding is available in the country, the penetration of the private mar-
ket is minimal. Farmers absorb most losses from floods. However, as
Freeman (2001) notes, multiperil crop insurance has been available in
the country since 1990. This coverage, which includes losses from floods
occurring from excess of rain, builds on the 100-year history of hail
insurance in Argentina. As with fire insurance, the penetration of the
market has been very small. The total premium for multiperil crop
insurance is less than $1.5 million a year. Stand-alone flood coverage
is no more than $180,000 a year.
Currently in Argentina, the costs of catastrophes are absorbed by
the victims of the disaster and by the federal, provincial, or municipal
THE FISCAL RISK OF FLOODS 457
governments. In turn, the federal government, which is the primary
source of recovery funding, has relied on borrowing from international
lending institutions (mainly the World Bank and the Inter-American
Development Bank) to fund flood losses. Under the procedures now in
place, the federal government pays 60 percent of the costs of the re-
construction of flood-damaged infrastructure, and the provincial gov-
ernments pay the remaining 40 percent. Essentially, those funds cover
the immediate needs of the population and infrastructure projects, and
no funds are allocated to reimburse agricultural losses. Thus the flood
recovery plans proposed by the provinces have an emphasis on infra-
structure: roads, bridges, sanitation, schools, hospitals, and low-in-
come housing projects for people living on floodplains.
Additional funding for disaster relief is provided in a fragmented
fashion. However, a procedure is in place to declare an emergency
area and to make budget allocations to fund disaster relief. Often,
these allocations are accompanied by additional taxes to fund the re-
lief process. Yet this procedure certainly causes budget and fiscal prob-
lems, is subject to political manipulation, and brings about reallocations
from developmental initiatives to relief.
Alternatives to Flood Insurance
Alternatives for risk transfer exist. One alternative is the creation of
an insurance system for the government along the lines of the one
proposed for FONDEN in Mexico. Another alternative is to explore
the feasibility of using risk transfer based on weather indices.
A program similar to the one proposed to modify FONDEN in
Mexico would be based on providing transfer and financing options
for the government for losses from catastrophic flooding. Instead of
the government being the primaryr insurer for flood risk, which it is
now, there would be ex ante financial planning for flood losses. The
program would entail a budget allocation for flood losses, as opposed
to the post-event financing options currently employed. This early al-
location would provide more stability for the budgeting process, en-
abling funds for other programs to be more effectively managed.
Moreover, it would provide a consistent set of resources for dealing
with major catastrophic flood losses. And it would demand a system-
atic risk assessment by the govern,ment of the essential infrastructure
subject to flood losses. The program would include cooperation be-
tween the federal government and reinsurance companies to absorb a
portion of the federal government's flood risk. The program also would
define the financing required for the portion of the risk not subject to
transfer. In this alternative, the government would retain a portion of
the risk (about 50 percent). The reirisurers would provide a credit facility
458 ALCIRA KREIMER
to offer liquidity in the event of a flood with damages in excess of the
amount covered by the government. For this layer of risk, the govern-
ment would pay an annual premium to the reinsurers. The program
would cover losses from property damage for which the Argentine
government is now responsible-that is, the government is responsible
for reconstruction, mainly of infrastructure: roads, bridges, schools,
hospitals, dams, and dikes. The program also would require that the
Argentine government undertake steps to reduce its exposure to flood
losses. A financial guaranty by a third party or international organiza-
tion or a similar instrument may be required to initiate the program.
As noted by Freeman (2001), this type of program could have sig-
nificant benefits for both the Argentine government and the interna-
tional lending community. The process of lending for natural disasters
after a disaster occurs is both inefficient and disruptive for the federal
government. The process eliminates the benefits of risk-spreading as
an effective policy tool. As mentioned earlier, in the developed world
risk transfer is regarded increasingly as a relatively effective and effi-
cient tool for dealing with the costs of catastrophes. In the developing
world, the ready availability of post-event lending may be a disincen-
tive for exploring the benefits of pre-event risk transfer. Furthermore,
the repetitive financing of infrastructure projects destroyed by flood
reduces lenders' abilities to fund new projects with their poverty-re-
ducing benefits. In Argentina, the issuance of loans to restore the same
infrastructure from flood loss twice within a decade is an inefficient
use of scarce lending resources. This is particularly the case when a
private market option exists to deal with the problem. The program
would create an incentive for the government to encourage the use of
private insurance by other governmental agencies, industrial firms, and
commercial enterprises. By quantifying the cost of flood risk, and pay-
ing current budget funds to absorb the risk, the government would
have a powerful incentive to force those who incur the risk to bear
their fair share of the cost of the risk.
Another potential strategy for transferring disaster risk that could
be explored in the future would build on pilot work currently being
conducted in Morocco, Tunisia, Ethiopia, and Nicaragua, and a pro-
posed project in Nicaragua. These initiatives are targeted to crop pro-
tection and develop the idea of insurance based on a weather event
rather than on actual losses such as crop failure. The underlying as-
sumption is that certain weather events (such as rainfall above or be-
low a specified amount) are highly correlated with crop failures and
therefore the income risks of rural people. Drought insurance is one
example. Insurance contracts are written against severe rainfall short-
falls (say 30 percent or more below the norm) that can be measured by
a regional weather station. The key advantages of this kind of insur-
ance are that the trigger event (in this case rainfall shortages) can be
THIE FISCAL RISK OF FLOODS 459
independently verified, and therefore are not subject: to the possibili-
ties of manipulation that are present: when insurance payouts are linked
to actual farm losses.
Conclusion
Argentina is among the top seven countries in losses from floods as a
percentage of the gross domestic product. Managing the risk requires
an integrated approach that entails risk identification, risk reduction,
and risk transfer. A major aspect of risk management is a coordinated
and well-established institutional and legal system and a program of
risk reduction that incoporates structural and nonstructural mecha-
nisms. In addition, in Argentina feasible private sector options for trans-
ferring risk are available. All of this, however, requires moving from a
fatalistic approach to dealing with floods to a culture of prevention;
from a system of bearing the losses to a more efficienit system of man-
aging the risk.
References
Freeman, Paul K. 2001. "Recomendaciones para la transferencia de los riesgos
de inundactions." In Alcira Kreimer, David Kullock, and Juan B. Valdes,
"Inundaciones en el Area Metropolitana de Buenos Aires." Disaster Risk
Management Working Paper Series No. 3. World Bank, Washington D.C.
Freeman, Paul K., Leslie A. Martin, Re:nhard Mechler, Koko Werner, with Peter
Hausman. Forthcoming. Catastrophes and Development. Disaster Risk
Management Series No. 3. Washington, D.C.: World Bank.
Kreimer, Alcira, Margaret Arnold, Christopher Barnham, Paul Freeman, Roy
Gilbert, Frederick Krimgold, Rodney Lester, John D. Pollner, and Tom Vogt.
1999. Managing Disaster Risk in Mexico. Disaster Risk Management Series
No. 1. Washington, D.C.: World Bank.
Kreimer, Alcira, David Kullock, and Juan B. Valdes. 2001. "Inundaciones en el
Area Metropolitana de Buenos Aires. " Disaster Risk Management Working
Paper Series No. 3. World Bank, VWashington, D.C.
Kunreuther, Howard, and Joanne Linnerooth-Bayer. 1999. "The Financial
Management of Catastrophic Flood Risks in Emerging Economy Countries."
Paper presented at the Conference on Global Change and Catastrophic Risk
Management. International Institute for Applied Systems Analysis,
Laxenburg, Austria.
Munich Re. 1999. "Natural Disasters. Annual Review of Natural Catastro-
phes, 1998." Munich Reinsurance Company, Munich.
Swiss Re. 1998. "Floods-An Insurable Risk?" Swiss Reinsurance Company,
Zurich.
460 ALCIRA KREIMER
United Nations. 1976. "Guidelines for Flood Loss Prevention and Manage-
ment in Developing Countries." Department of Economic and Social Af-
fairs, United Nations, New York.
World Bank, Operations Evaluation Department. 2000. "Performance Audit
Report: Argentina Flood Rehabilitation Project." Washington, D.C.
CONCI.USION
Toward a Code 'of Good Practice
on Managing Fiscal Risk
Allen Schick
University of Maryland
iHE STUDIES PRESENTED IN THIS book indicate that countries differ
greatly in their treatment of contingent liabilities and other fiscal risks.
There are no generally accepted risk management principles, no model
practices to guide governments that want to go beyond conventional
budget and debt data to analyze their fiscal performance and regulate
their financial exposure to contingencies. Many, perhaps most, coun-
tries have taken no significant steps toward analyzing and managing
such risks, and those that have had to invent their own means of deal-
ing with the problem. There is as yet no agreement on what should be
reported on, or annexed to, financial statements, nor on the extent to
which contingent liabilities should be recorded in budget documents
prior to the point where paymeni: occurs. The difficulty of codifying
practices is greatest when the liability is implicit, but problems also
emerge when a government's obligation is explicit.
The conceptual studies in this book mirror the divergence in prac-
tice. Although the conceptual boundaries have been stretched and some
of the ideas proposed by various contributors may find future applica-
tion, at present there is no agreement on what is gocid or feasible prac-
tice. In contrast to risk managernent in the market sector, which is
highly developed, the stockpile of ideas for government is disparate.
Much work lies ahead before the concepts will solidify into practical
guideposts.
Given the disarray in theory and practice, one might conclude that
it is premature to propose normative rules for managing contingent
liabilities. This concluding chapter, however, takes the opposite posi-
tion. Precisely because practices are so disparate and the risks of doing
461
462 ALLEN SCHICK
nothing so high, we believe it prudent to move forward with advice on
regulating government's exposure. As the title of this book indicates,
contemporary governments are at risk. The absence of concrete stan-
dards means that many governments may not take even the most el-
ementary steps toward cushioning their fiscal position against the types
of shocks and disturbances that have destabilized the finances of emerg-
ing market, developing, and transitional countries.
Fiscal risks do not evaporate just because government fails to rec-
ognize them. In good times, the risks hibernate "below the line," off
the books and out of sight; when they come due in bad times it is too
late to do anything about them. It is likely that fiscal risks will escalate
in the years ahead, especially in developing and transitional countries
where national governments will become more active in stabilizing
economic conditions and household income. Moreover, as Maastricht-
type fiscal norms spread, governments may be impelled to substitute
contingent liabilities for cash payments-for example, by providing
guarantees rather than payouts. If this occurs, the short-term budget
situation may become more favorable, but the longer-term outlook
will deteriorate. The temptation to behave in this manner is great,
because contingent liabilities rarely come due in the same fiscal period
in which they are entered.
Taking the initial steps to codify good practice would spur many
governments to improve their management of fiscal risks. It also would
provide usable benchmarks for assessing existing benchmarks and for
identifying matters in need of correction. Furthermore, over time it
may lead to greater convergence in country practices and to more re-
fined standards. The guidelines set out in this chapter are provisional;
with additional experience, it should be possible in due course to specify
widely accepted principles for managing fiscal risk.
Two issues have to be faced in this first effort to devise standards.
One is whether the recommended norms should be limited to explicit
liabilities or also should cover implicit ones. The other is whether the
norms should reflect the most commonly used practices or should go
beyond these to novel approaches that may offer the most promise in
controlling fiscal risk. With regard to the first issue, it bears noting
that standards are better suited for commitments formalized in law or
in contracts than for those arising out of expectations or informal
understandings. For one thing, it may be hard to identify implicit li-
abilities or to assess the extent to which they may be called; for an-
other, recognizing implicit liabilities may spur the affected parties to
behave in morally hazardous ways and thereby increase the govern-
ment's exposure. Yet commitments embedded in expectations should
not be ignored. After all, doing so will not diminish expectations or
eliminate the prospect of future payment. The position taken here is
that explicit and implicit liabilities should be identified separately, and
CONCLUSION 463
that implicit liabilities should be recognized only to the extent that it is
strongly probable that government will be called on to make payments.
In no case should they be recognized when doing so would materially
increase moral hazard or the risk to government. On this basis, some
implicit liabilities would be covered; others would not. For all implicit
contingent liabilities, however, government should examine its expo-
sure, take steps to reduce the associated risks (such as improving bank-
ing regulation and supervision systems), and prepare confidential
contingency plans that would offer guidance to policymakers in case
of failure.
Proposed Standards
The standards proposed here build on the discussion of actions that a
government may take to control its r isks, which was presented in Chap-
ter 1 as well as in the chapters that describe specific country experi-
ences. The standards are based on excisting methods; they do not require
technical breakthroughs. It does little good to set the standards so high
that few countries will strive to reach them. The aim should be to
promote practices that can be adopted without undue difficulty by
most countries. For this reason, the standards do riot include some
promising ideas, such as the credit budgeting system introduced in the
United States (and explained by Schick in Chapter 3), or the reserve
fund conceived by Daniel Cohen (see Chapter 6).
Standard 1. Before it accepts a new contingent liability, a government
should assess the risk to its fiscal condition, including the probability
of future payouts. The assessment should be conducted by an indepen-
dent entity.
The best, and often the only, time to regulate fiscal risk effectively
is before it is taken. Some countries-including Canada, Colombia
(see Chapter 12 by Echeverry and others), the Netherlands, South
Africa, Sweden, the United States (,ee Chapter 3)-have implemented
this requirement. This is not easy to do, because the parties to the
risky transaction have an incentive to underestimate risk, sometimes
insisting that guarantees are costless because no money is leaving
government hands, or that the cost will be amply covered by the
future dividends of a burgeoning economy. In goocd times, it is easy
to forget that contingent liabilities and other risks typically come
due when conditions turn adverse; in bad times, it is tempting to
argue that things will get better if the government provides guaran-
tees or other assistance that enables enterprises and households to
stay afloat. In other words, the whole process is skewed in favor of
underestimating the government's exposure. This is especially so when
464 ALLEN SCHICK
the same government agency is responsible for tendering the guaran-
tees and assessing the risk.
To counter these tendencies, it is essential that governments sepa-
rate risk-taking and risk assessment by placing them in different orga-
nizations. The entity that processes guarantees or other contingent risks
should have no role in assessing the probability of default or of other
adverse events. This task should be entrusted to an independent unit
that has no decisionmaking authority. Separation of the two functions
is standard procedure in prudently managed financial institutions, but
it is still rare in government. In fact, few governments have formal risk
assessment rules or procedures; many treat this as a policy decision
that is taken regardless of the risks involved. Moreover, many govern-
ments see themselves as a guarantor of last resort, assuming risks that
would otherwise be held by weak enterprises or poor households. Be-
cause they knowingly take on bad risks, governments are indifferent
to the losses they face, especially when these do not appear in current
budgets or financial statements.
The standard proposed here is based on the principle that govern-
ment should assess risk even when it has already decided to provide
guarantees, and even when it does so because the beneficiary is a bad
risk. It should do so in order to have an accurate picture of its financial
condition and of potential downstream losses and payments. Ideally, it
should use this information to estimate the money needed to cover
losses in the current budget or future ones.
These assessments should be conducted independently; if they are
not, they will be tainted by conflicts of interest, misassessments of
risk, and increases in avoidable losses. Several alternative organiza-
tional arrangements may be appropriate for assuring independence.
One would be to assign responsibility to a specialized, independent
unit whose only task would be to carry out objective risk assess-
ments. Alternatively, responsibility might be assigned to an existing
entity that enjoys independence from political considerations. For
example, in countries that have an autonomous debt management
office, such as Sweden and Ireland, this office would be a good can-
didate. In other countries, it may be the supreme audit office, as it is
in the United States.
A third option would be to assign overall risk assessment to a central
unit and responsibility for particular actions to the agency running the
program. For example, a central agency might analyze programs to pro-
mote home construction or ownership, while the housing agency as-
sesses applications for guarantees from builders or home purchasers.
Standard 2. The government should periodically compile an inventory
of outstanding contingent liabilities and report on the volume of these
liabilities, their legal basis, and the probability of losses.
CONCLUSION 465
This standard recognizes that even if government is prudent is tak-
ing new risks, it already has a portfolio of contingent liabilities and
other risks that may come due in the future. In this regard, practice in
Australia and New Zealand (described by Petrie in Chapter 2) and in
the Czech Republic (see Chapter 9 by Brixi, Schick, and Zlaoui) is ex-
emplary. In managing its finances, a government must be cognizant of
these commitments and take them into account in budget plans and
other policy statements. It would be sensible for government to report
annually on outstanding liabilities and the risks associated with them.
The report may be appended to financial statements (in those countries
that regularly publish such statements) or be published as a free-standing
document. It would be preferable for the report to be reviewed by audi-
tors who would comment on the adequacy of the methods for docu-
menting contingent liabilities and estimating their costs.
Many governments may have difficulty at the outset compiling such
a report, but the requirement to do so should spur them to improve
their database. A government should describe risks even when it lacks
sufficient data to measure their volume or probable cost reliably. More-
over, the report should distinguish between quantifiable and non-
qLiantifiable risks. Quantifiable risks would be those whose volume or
probability of loss can be estimated with reasonable confidence;
nonquantifiable risks would be those that lack a basis for measuring
these amounts. But even risks that cannot be measured should be de-
scribed in the report.
In many cases, it would be prudent for the report to specify a range
within which the risk is expected to occur rather than to provide point
estimates. Point estimates often turn out to be wrong, and they can
mislead government so that it takes inappropriate action. It is prefer-
able to estimate a range of possible outcomes and to discuss the fac-
tors that may determine the actual liability or cost. Finally, the report
should project the fiscal period(s) (luring which the government may
be called on to indemnify losses.
Standard 3. Government fiscal analysis published in the annual bud-
get or other documents should discuss the major risk factors affecting
revenues and expenditures for the next fiscal year or beyond.
Every budget is subject to an array of risks that may result in actual
revenues and expenditures that deviate significantly from the planned
levels. This fact has been elaborated in the fiscal vulnerability frame-
work proposed by Richard Hemming and Murray Petrie in Chapter 7.
The risks generally do much more damage to fiscal stability in develop-
ing and transitional countries than in developed ones. In developed coun-
tries, unbudgeted losses may compel. the government to make relatively
minor adjustments in revenue or spending plans; in developing and
466 ALLEN SCHICK
transitional countries, variances may be so large that they force the
government to discard the approved budget and prepare an entirely
new one. In some poor countries, the government does not bother to
prepare a new budget, but pretends to be implementing the approved
one, while it covertly spends different amounts.
Many of the risks to the estimates are known when the budget is
tabled, but the near-universal practice is to ignore them, either in the
hope that things will turn out better or in the expectation that the
budget will have to be remade during the year. Commenting on major
risks would prepare the government for events-such as a shortfall in
revenue, deterioration in exchange rates, or natural disaster-that will
upset its budget plans, and also spur it to prepare more realistic rev-
enue and spending estimates. Doing so should not be difficult, because
governments typically are required to discuss risk factors in offering
statements or similar documents when they seek private financing.
These documents have narrow circulation and receive none of the at-
tention that is given the annual budget.
Fiscal analysis is incomplete if it omits government contingent li-
abilities and other fiscal risks. Speaking to this point, Chapters 1, 6, 7,
8, 9, 10, and 1 1 illustrate the weaknesses of conventional fiscal analy-
sis, outline possible frameworks for a more comprehensive analysis,
and apply such frameworks to the analysis of the fiscal position of
several countries. Although it may be difficult for governments to ana-
lyze the risks associated with their whole portfolio of assets and their
contingent as well as direct liabilities (as advocated by Brixi and Mody
in Chapter 1 and by Easterly and Yuravlivker in Chapter 8), or to
assess all sources of their fiscal vulnerability (as encouraged by Hem-
ming and Petrie in Chapter 7), it may be relatively simple to calculate
the government's hidden deficit and hidden debt (as was done for the
Czech Republic in Chapter 9). Consistent with government account-
ing practice, expenditures and revenues that occur outside the budget
can be taken into account. In addition, for noncash programs such as
guarantees, the present value of the expected fiscal cost, and the likely
cash consequences for the budget in the years ahead, can be estimated
in a simple manner. This is illustrated for the Czech Republic, Bul-
garia, and Hungary in Chapter 9 and for Thailand in Chapter 11 (and,
later, with the help of options-pricing methodology, for pension guar-
antees in Chapter 13 by Pennacchi, for central banks in Chapter 16 by
Blejer and Schumacher, and for credit guarantees in Chapter 19 by van
Wijnbergen and Budina). Other fiscal risks, such as those arising from
the country's banking sector (discussed by Claessens and Klingebiel in
Chapter 14), from government insurance schemes (surveyed by Feldman
in Chapter 15), and from private participation in infrastructure (out-
lined by Mody in Chapter 17), also may be analyzed in simple terms,
reflecting their face value and possible scenarios (as was done for In-
donesia in Chapter 11). As data and analysis improve, it will be fea-
CONCLUSION 467
sible to give fiscal analysis a longer-term perspective and to expand the
coverage fiscal sustainability analysis (as presented by Cohen in Chap-
ter 6 and applied to China by Krurnm and Wong in Chapter 10).
The budget discussion should cover at least the next fiscal year and
should reflect not only the risks arising from the levels and structure of
government budget and debt, but also those outside the budget and
debt portfolio. Governments that have a medium-term framework or
multiyear projections should extend the coverage to the next several
years and present different scenarios showing the possible impacts of
the various risks on the budget according to the macroeconomic, policy,
and other underlying variables. The discussion need not be compre-
hensive; it should concentrate on those risks that have the greatest
capacity to force the budget to veer off course.
Standard 4. The government should establish a risk management strat-
egy to guide public organizations when they take actions that expose
them to financial liability.
In contrast to most direct spending decisions that are made through
regular budget procedures, governments often take on contingent li-
abilities in an ad hoc manner, without regard for the impact on their
financial condition. Each guarantee program, and sometimes each guar-
antee, is treated as a special case, with its own rules and procedures. The
result is a hodgepodge of contingent liabilities that lack uniformity but
add up to a significant risk to government. Some governments impose
initiation (or origination) fees, but most do not. Sorne have rules and
procedures for recovering assets in case of default, but most do not.
Some assess the government's exposure before a commitment is made,
but most do not. The standard recommended here (and elaborated by
Brixi and Mody in Chapter 1) would have the government develop a
policy that would have to be followed before new liabilities are under-
taken. In some countries, it may be appropriate to codify the policy in
legislation, but in most an administrative decree could suffice. Ideally,
the policy should be comprehensive, covering all contingent liabilities,
but, as a practical matter, there are likely to be exceptional cases that are
guided by expediency rather than by rules.
The recommended standard is neutral on the question of whether
government, or an entity related to government such as a municipality,
state-owned enterprise, or agency benefiting from any kind of govern-
rnent guarantee (such as export guarantee funds, agriculture guarantee
funds, state insurance entities), should take on risk. It focuses instead on
the procedures to be followed and the data to be compiled and reviewed
before government, or such entities, accept these liabilities. As envisioned
here, the guidelines should cover issues that arise wihen contingent Ii-
abilities are proposed, including:
468 ALLEN SCHICK
* procedures to assess the risk of loss;
* initiation or origination fees;
* form of guarantee contracts and the role of the finance ministry
or others in reviewing contracts before they are executed;
* risk- or cost-sharing arrangements;
* procedures for monitoring performance under guarantees or other
contingent liabilities; and
* what protection the government or entity should have in case of
default or other events that cause losses.
These and other issues should be addressed in an authoritative manual
that covers all the main steps in the contingent liability process and
pertains both to programs that expose the government to risk and to
specific transactions. Few countries currently have comprehensive guide-
lines. Canada and the Netherlands (see Chapter 3) appear to be among
the most advanced in this area. In these ccuntries, central agencies oper-
ate a process for reviewing and implementing guarantees that parallels
the direct expenditure process. Their guidclines cover most of the steps,
from proposal through financing, and emphasize the importance of miti-
gating the government's risk. Although the guidelines are designed to
discourage the issuance of guarantees, they can be easily modified to
suit countries that prefer to use this policy instrument in a prudent man-
ner. It is perhaps more difficult, however, to regulate and ensure proper
risk-taking incentives in entities outside the framework of the central
government. Krishna Ramaswamy in Chapter 5 and Jun Ma in Chapter
18 point to some possible approaches for state-owned enterprises and
subnational governments.
Because policy without enforcement would be a hollow gesture, it
is essential that implementation of the guidelines be entrusted to a
central agency, such as the finance ministry, which, in most countries,
already has responsibility for direct expenditures. Another possibility
would be to assign responsibility to the debt management (or similar)
office (as in Colombia, Sweden, and Thailand), which is experienced
in assessing the risk associated with government borrowing. In rare
cases, the government may wish to establish a special agency for han-
dling contingent liabilities. The important thing is not that an agency
be established but that it have sufficient authority and resources to
safeguard the government's interests.
Standard 5. The government should promote cost- and risk-sharing to
discourage moral hazard, ensure the economic viability of guarantees,
and reduce the probability and amount of loss.
Even if it takes the position that contingent liabilities are appropri-
ate policy instruments, a government should be cognizant both of the
impacts of its actions on the behavior of others and of the risk it is
CONCLUSION 469
assuming. All guarantees change behavior, and all impose costs. If they
did not, the market for guarantees would dry up quickly. Even though
a government cannot fully escape moral hazard ancl the prospect of
loss when it assumes contingent liabilities, it can take steps to greatly
diminish its exposure. The steps are technically easy (and have been
discussed in several chapters, including Chapter 4 by Sundaresan),
though they may be politically dilficult. They involve shifting some
portion of the cost and risk to those who benefit from guarantees and
other contingent liabilities.
As a general rule, government should charge risk-based premiums
or establish deductibles for compensating losses. These premiums rec-
ognize that not all contingent liabilities are equally risky. In some cases,
a guarantee may merely lower the cost to the beneficiary without add-
ing much to the government's risk. For example, a borrower who can
obtain funds privately may nevertheless seek a government guarantee
in order to obtain lower interest charges or other more favorable terms.
More commonly, credit-unworthy borrowers seek government guar-
antees because they cannot obtain private loans. In these cases, it would
be appropriate for the government to charge premiums that reflect its
true cost (that is, the present value of the expected fiscal cost). If the
government decides to charge premiums lower than the true cost, it
should openly acknowledge the size of the subsidy. There may be in-
stances in which government subsidies, such as a recluced premium or
government support through guarantees, are appropriate. Yet such sub-
sidies and support should be subject to the same scrutiny as any spend-
ing items in the budget.
Government also can reduce its exposure by paying for the last,
rather than the first, loss-that is, by setting high deductibles that must
be satisfied before it makes payment. The problem, however, is that
because risky borrowers cannot otherwise obtain capital, they have a
strong incentive to lobby governraent for special treatment. Govern-
ment also can seek to mitigate moral hazard and loss by regulating the
actions of those benefiting from guarantees and other forms of contin-
gent assistance. It is common business practice for lenders to impose
covenants that restrict risky behavior by firms. A government should
act in the same way to protect its interest, by restricting those who
rnight otherwise behave in a morally hazardous manner. Finally, a gov-
ernment can safeguard its interests by monitoring the actions of bor-
rowers and other insured parties and by giving its claims seniority over
those of other parties. These practices are common in business, but
still rare in the public sector.
Standard 6. The budget should limit the amount of guarantees and
other contingent liabilities to be tendered during the year, as well as
the total amount that each institution authorized to issue guarantees
may have outstanding.
470 ALLEN SCHICK
As explained by several contributors to this book, the budget is
inherently a weak tool for regulating contingent liabilities. In cash-basis
budgeting, losses are recognized when the payment is made, not when
the liability is incurred. Accrual-basis budgeting provides for recogni-
tion when the liability is incurred, but rules and practices for treating
contingent liabilities have not yet been standardized. Few countries
have accrual-basis budgets, and fewer systematically provision for
losses. For this reason, efforts to strengthen budgetary control of con-
tingent liabilities will likely develop through cash-basis budgeting.
The cash basis enables government to control the volume of guar-
antees issued during the year or the total outstanding. These amounts
may be specified for the budget as a whole as well as for each institu-
tion authorized to issue guarantees or to enter into other contingent
commitments. These controls are easy to apply because they do not
depend on estimates of losses or payments; rather, the control is exer-
cised over volume, without regard for the riskiness of the transaction.
This method has been applied in Hungary (see Chapter 9) and in some
other countries where the annual budget specifies the maximum amount
of guarantees that may be tendered during the year. These aggregates
are then parceled out among institutions authorized to make guaran-
tees, much in the same manner that direct spending is allocated among
government departments and ministries.
Standard 7. The budget should set aside funds, within an overall fiscal
constraint, for expected losses during the year.
Although limits on the volume of guarantees can be separate from
the expenditure budget, the government should provide in the budget
for payments expected to be made during the year. These reserves should
be included in calculations of total expenditure and of the surplus or
deficit. Of course, the provisions are not effective constraints, for the
government would have to pay for losses that exceed the budgeted
levels. Nevertheless, setting aside funds in the budget would serve two
laudable purposes: it would sensitize the government to the fact that
contingent liabilities entail costs, and it would induce the government
to devise more reliable procedures for estimating risk.
Conclusion
Implementing these standards may appear to be a daunting task, yet
they are within reach of most governments. The critical step involves
collecting the information on outstanding contingent liabilities, esti-
mating the government's exposure, and incorporating the information
into fiscal analysis and government budgets and financial statements.
CONCLUSION 471
The task is ongoing, for it is a rare government that knows all that it
should about guarantees and other contingent schemes. In the same
way that governments monitor and regulate direct expenditures, up-
dating their information on a regular basis, they should invest in the
capacity to identify, measure, and update information on contingent
liabilities. If they do so, the quality and relevance of their information
will improve rapidly.
Where some standards are beyond immediate attainment, govern-
ment should seek corrective action-first, by imposing the standards
on new contingent commitments, and, second, by making strong ef-
forts to identify the most critical risks, those that pose the greatest
threat to fiscal stability. Ideally, governments should strive for a com-
prehensive accounting of risks. But they should move ahead even when
there are gaps in their database.
The standards themselves should be moving targets. Efforts should
be made both through government action and through policy research
to elaborate fuller standards and practices. In terms of action, govern-
ments at the cutting edge of practice, with strong fiscal institutions,
should devise fuller means of dealing with risks. Some may wish to
apply the business-type or experimental practices discussed in this book.
It is especially important that accounting standards be devised to deal
with fiscal risks. With the rapid spread of financial reporting, one can
expect that financial statements will be a powerful conduit for dis-
seminating standards in the future.
Finally, there is much need for research both on existing and on
experimental practices. For example, it would be fruitful to study sys-
tematically how New Zealand uses statements of contingent liabilities
to manage risk, the manner in which Hungary budgets for expected
losses, and the credit budget moclel introduced in the United States.
The study of fiscal risk is still in its infancy; much research remains to
be done.
Conventional fiscal analysis fails to address contingent fiscal
risks. Similarly, the government budget process and documentation generally fail
to scrutinize the substantial claims on public resources that are associated with
government contingent liabilities, realized and potential.
Government at Risk fills gaps in our understanding of fiscal risks and develops
suitable frameworks for managing them. It offers new analytical concepts, presents
country case studies, and based on the country case studies, provides a menu of
practical ideas for policymakers and scholars to bring fiscal risk within the ambit
of public finance.
The ideas and examples presented in this book should prompt governments to
bring their contingent liabilities and other fiscal risks under control and to carefully
manage their exposure to fiscal risk. Recent history in many countries has proven
that the benefits of such an effort are enormous, not only with respect to the
government's future fiscal stability, but also with respect to its capacity to achieve
broader policy objectives. Government at Risk points the way ahead by presenting
general principles of sound fiscal management and by providing specific examples
of innovative country practices.
48135214 835