Report No. 16040 Fiscal Management in Adjustment Lending October 3, 1996 Operations Evaluation Department D)ocumnt of the World Bank Abbreviations and Acronyms AFR Africa Regional Office EAP East Asia and Pacific Regional Office ECA Europe and Central Asia Regional Office GDP Gross Domestic Product GNP Gross National Product ICR Implementation Completion Report IBRD International Bank for Reconstruction and Development IDA International Development Association IMF International Monetary Fund LAC Latin America and the Caribbean Regional Office MNA Middle East and North Africa Regional Office OED Operations Evaluation Department O&M Operations and Maintenance PARs Performance Audit Reports PCRs Project Completion Reports PRs President's Reports QRs Quantitative Restrictions SALs Structural Adjustment Loans SAS South Asia Regional Office SECALs Sectoral Adjustment Loans TAR Tax Administration Reform TFSAL Trade and Financial Sector Adjustment Loan TINs Tax Identification Numbers VAT Value Added Tax The World Bank Washington, D.C. 20433 U.S.A. Office of the Director-General Operations Evaluation October 3, 1996 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Fiscal Management in Adjustment Lending The attached report Fiscal Management in Adjustment Lending, prepared by the Operations Evaluation Department, analyzes the Bank's experience with fiscal management in 250 structural/sectoral adjustment operations approved during FY 1979-94 in 86 countries, for an amount of US$ 34 billion. The study examines the country record in fiscal adjustment and the Bank's role in diagnosis, design and implementation of conditionality, paying particular attention to the content of tax and expenditure reforms. The study finds that fiscal mismanagement rather than exogenous shocks was the principal cause of persisting fiscal deficits. Fiscal adjustment requires long, sustained effort and was achieved primarily through revenue measures and to a lesser extent through capital spending reductions. Reductions in current expenditures, while frequently an important objective of the reform programs, proved to be more elusive. The study confirms that deficit reduction leads to improved external balances and growth. . While deficits were reduced across all regions and country groups, there were significant reversals in recent years. Rising external debt burdens also suggest that fiscal adjustment has not been sufficient and that problems of sustainability and solvency remain in many countries. The record suggests that success was limited because fiscal issues were analyzed in isolation from other macroeconomic reforms, issues concerning the role of the state and deficit coverage and measurement were ignored, and conditionality was "soft" or vague. While tax reform was generally consistent with current professional norms, conditions on nontrade taxes and tax administration were imprecise and unmonitorable. The record was better for trade taxes. Expenditure reform, which focused on spending reductions and restructuring, was broadly consistent with the Bank's guidelines. However, countries with conditions did only marginally better than countries without them in improving sectoral and economic expenditure patterns. A major reason for the poor impact may be that the Bank emphasized the means rather than the ends of expenditure reform. The study recommends that the Bank take a more direct approach to fiscal analysis, while respecting the Fund's lead role in advising clients on fiscal matters. The Bank needs to help clients improve the coverage, measurement, and transparency of fiscal accounts and to better integrate fiscal considerations into the macroeconomic framework. This should allow analysis of sustainable deficits and the relationships between fiscal and other structural reforms. In tax reform, the Bank needs to stress simplification and base broadening, tax administration, sequencing, and concern for the environment, equity, and private sector development. In the expenditure area, the Bank should assist clients in defining the role of the state, clarifying the goals of reform, cutting spending where deficit reduction is a pressing concern, emphasizing poverty alleviation, and promoting equity. In general, the Bank needs to build better performance indicators that track outcome as well as the use of fiscal instruments. Contents Executive Summary 7 1. Setting the Stage 13 The Typical Crisis Profile 13 The Centrality of Fiscal Reform 15 Coverage of the Report 16 2. Fiscal Management in SALs/SECALs. A Close-Up View 19 What Was the Quality of Fiscal Diagnosis? 19 Measuring the Fiscal Deficit 20 Identifying Sustainable Deficits 23 The Role of the Government 24 Budgetary Processes 24 Total Expenditures 25 Public Investment 25 Social Expenditures 26 Current Expenditures 26 Public Enterprise Reform 27 Revenue 27 Fiscal Analysis and Projections 29 Was the Quality of Conditionality Adequate? 29 Was Conditionality Consistent? 30 Were Budget Deficits Specified? 30 Were Quantitative Fiscal Targets Specified? 31 Other Aspects of the Quality of Conditionality 32 Data Quality 33 Was Implementation Adequate? 33 Tranche Delays 33 Budget Deficit/Public Savings 33 Budgetary Process Reform 34 Public Enterprise Reform 34 Public Investment Programs 34 Current Expenditure 34 Tax Revenues 35 Other Issues 35 Conclusions 36 Lessons 38 The report was prepared by Jayati Datta-Mitra (Principal Evaluation Officer). Members of the team were: Amaresh Bagchi (Tax Administration Reform); Vandana Chandra and Salman Anees (Fiscal Management under SALs/SECALs: An Aggregate View and Conclusions); Poonam Gupta (Fiscal Management in SALs/SECALs: A Close-Up View); Anwar Shah (Principal Evaluation Officer), Jeff Huther, Sandra Roberts, Yibin Xiang (Public Expenditure Reform); Wayne Thirsk and Harry Watson (Tax Reform). John Diamond and Monica Rubio provided initial research support. Meta de Coquereaumont, Barbara Karni and Alison Strong were the editors. Jasmine Mason-Anderson, Norma Namisato and Geri Wise provided administrative assistance. The report was issued by the Country Policy, Industry and Finance Division (Manuel Pefialver, Division Chief) of the Operations Evaluation Department (Francisco Aguirre-Sacasa, Director). Director-General, Operations Evaluation: Robert Picciotto. 2 3. Tax Reform 41 What the Study Found 41 Nontrade Tax Conditionality 41 Trade Tax Conditionality 41 Principles of Tax Reform 42 Precision and Content of Tax Conditionality 44 Nontrade Tax Conditionality 44 Trade Tax Conditionality 48 How Successfully Were Tax Conditionalities Implemented? 50 How Effective Was Tax Reform? 51 Lessons and Directions for the Future 57 4. Tax Administration Reform 59 What the Study Found 59 The Ends and Means of Tax Administration Reform 59 Improving Taxpayer Compliance 60 Improving Taxpayer Enumeration and Registration 60 Estimating Taxable Income and Tax Liability 61 Improving Tax Collection 61 Improving Enforcement of Tax Regulations 61 Improving the Organization of the Tax Authority 62 Bank Involvement in Tax Administration Reform 63 Specificity and Precision of Conditionality 64 Content of Conditionality 66 Gaps in Tax Administration Reform Conditionality 67 Tracking Implementation of Reforms 68 Measuring Outcomes 69 Lessons 70 5. Public Expenditure Reform 73 What the Study Found 73 Initial Conditions 73 Quality of Conditionality and Compliance 73 Changes in Expenditure Patterns 74 Final Outcomes 74 Equity Considerations 74 Conditions Emphasized Means Rather Than Ends 75 The Role of Government 75 A Framework for Evaluating Public Sector Reforms 75 Emphasizing Primary Health and Education 76 Decentralizing Decisionmaking 76 Getting the Right Balance between Current and Capital Spending 76 Getting Control of the Wage Bill 76 Improving Conditions for Private Sector Involvement 76 Expenditure Patterns before Adjustment Lending 77 Bank Conditionalities 79 Regional Differences 79 Specificity 79 Content 80 Equity Considerations 82 3 Implementation of Conditionalities 84 How Effective Were Expenditure Reform Conditions? 85 Impact on Expenditure Levels 85 Comparisons between Countries with Conditionality and Those Without 89 Impact on Major Expenditure Components 90 Outcomes: The Ultimate Objective of Expenditure Reform 93 Lessons 95 6. Fiscal Adjustment under SALs and SECALs: An Aggregate View and Conclusions 99 Were Fiscal Deficits Reduced? 100 Not in the Medium Term 100 Deficits Were Reduced in the Long Term but Many Remained High 102 Though Most Countries Sustained Reductions in Recent Years, Significant Reversals Occurred 102 How Did External Factors Affect Fiscal Outcomes? 103 Terms of Trade Shocks 103 Foreign Interest Payments 103 Primary Deficits and Total Interest Burdens 104 How Was Fiscal Adjustment Achieved? 104 Fiscal Deficits and Other Macroeconomic Variables 105 Creditworthiness, Fiscal Solvency, and Sustainability 106 Recommendations 107 Bibliography 163 List of Boxes in the Text: 2.1. Basic Elements of Diagnosis 20 2.2. Examples of Quasi-Fiscal Activities of Financial Systems 22 2.3. Evaluating Conditionality 30 3.1. Fiscal Reform Needs to Precede Tariff Reform 57 6.1. Determining the Solvency of the Public Sector 107 List of Figures in the Text: 1.1. Macroeconomic Indicators during the Preloan Period Presaged the Onset of Crisis 14 1.2. Private Investment and Interest on External Debt were Low during the Preloan Period 14 1.3. FY 79-94: AFR and LAC Account for: Largest Number of Adjustment Loans and Greatest Share in Total Adjustment Lending 16 2.1. Coverage of Fiscal Accounts Was Specified in Only Half of the 129 SALs/SECALs 21 2.2. Proportion of SALs/SECALs with Budget Deficit Conditions 31 6.1. Change in Fiscal Deficits by Region and Country Group 101 List of Tables in the Text: 1.1. Representativeness of Samples Used in the Study 17 2.1. Quantitative Fiscal Targets in SALs/SECALs 32 3.1. Content of Tax Conditionality in Africa, Latin America and the Caribbean, and All Regions 47 4 3.2. Trade Tax Conditionality by Type and Region 49 3.3. Types of Trade Tax Conditionality by Region and Country 52 3.4. Tax Structure Before and After Adjustment 54 3.5. Diagnosis of Need for Revenue Reform and Performance 55 3.6. Successful and Less Successful Trade Adjusters 56 4.1. Sample SALs/SECALs Incorporating Tax Administration Reforms by Region and Period 64 4.2. Specificity of Tax Administration Reform Conditionalities by Region 65 4.3. Sample Conditionalities by Level of Precision 65 4.4. Specific Tax Administration Reform Conditionalities by Region 66 5.1. Composition of Central Government Expenditures, Regional Averages, 1970-79 77 5.2. Public Current Expenditures per Pupil by Level of Education 78 5.3. Loan Conditions by Expenditure Reform Objective, 1980-94 80 5.4. Countries Facing Civil Service Conditions 81 5.5. Loan Conditions by Equity Objective, 1980-94 82 5.6. Implementation of Bank Conditions on Public Expenditures 85 5.7. Diagnosis of Need for Total Expenditure Reform and Performance 87 5.8. Diagnosis of Need for Capital Expenditure Reform and Performance 88 5.9. Diagnosis of Need for Current Expenditure Reform and Performance 89 5.10. Changes in Wage Expenditure Shares, Before and After Adjustment 90 5.11. Changes in Central Government Expenditures on Education, Before and After Adjustment 91 5.12. Changes in Central Government Expenditures on Health 92 5.13. Expenditure Inflexibility: Fixed Categories as Percentage of Total Revenue 93 5.14. Primary Enrollment Ratios in Countries with Education Conditions 94 5.15. Enrollment Ratios and the Gender Gap in Primary Education 95 6.1. Average Fiscal Deficits by Country Group and Region 100 Annexes/Annex Tables: 2.1. Description of SAL/SECAL Sample 115 Table 2.2. SALs/SECALs with Budget Deficit/Savings Conditions 117 Table 2.3. Aspects of Conditionality 118 Table 3.1. Changes in Structure of Central Government Current Revenues, 1980-93 121 Table 3.2. Changes in Tax Structure by Income Category and Region, 1980-1993 122 3.3. The Instruments of Tax Reform 123 Table 3.4. Content of Nontrade Tax Reform Conditionality 125 Table 3.5. Content of Trade Tax Conditionality by Type and Region 126 Table 3.6. Countries by Degree of Implementation of Nontrade Tax Conditionalities 127 4.1. Checklist for Bank Staff in Formulating Strategies for Tax Administration Reform 129 Table 4.2. Distribution of All SALs/SECALs Incorporating Tax Administration Reforms by Region 135 Table 4.3. Level of Specificity of Tax Administration Conditionalities in Sample SALs/SECALs 136 Table 4.4. Distribution of Sample SALs/SECALs by Region and Coverage of TAR Areas 137 5 Table 5.1. Health Care before 1985 139 Table 5.2. Changes in Total Expenditure Shares, Before and After Adjustment 140 Table 5.3. Changes in Capital and Current Expenditure Shares, Before and After Adjustment 141 Table 5.4. Average Central Government Expenditure Shares 142 Table 5.5. Government Subsidies before and after Imposition on Subsidy Reduction 143 Table 5.6. Current Expenditure Adjustments 144 Table 5.7. Public Expenditures on Education in Countries with Conditions and Countries Without 145 Table 5.8. Infant Mortality Rates 147 Table 5.9. Trends in Public Employment in Selected Latin American Countries 148 Figure 6.1. Fiscal Deficits and Terms of Trade by Region and Exporter Group 149 Table 6.1. Adjustment Loans and Countries in Sample, by Region 150 Table 6.2A. Countries with Eight or more Years of Fiscal Deficit Data 151 Table 6.2B. Adjustment Operations in Chapter 6 Sample 152 Table 6.3. Sample Countries by Country Group 155 Table 6.4. Fiscal Deficit Reductions by Country Group and Region 156 Table 6.5. Fiscal Deficits, Interest Payments, and Primary Deficits 157 Table 6.6. Changes in Fiscal Deficits and Primary Tool Used, by Country Group and Region 158 Table 6.7. Changes in Total Expenditure, Current Expenditure, and Capital Expenditure by Country Group and Region Relative to Preloan Period 159 Table 6.8. Correlation Coefficients between Ratio of Fiscal Balances to GDP and GDP Growth and Other Macroeconomic Variables 160 Table 6.9. Required Adjustments for Fiscal Solvency and Sustainability, 1989-93 161 7 Executive Summary 1. The World Bank initiated structural adjustment lending in the early 1980s to help developing countries out of the economic crises into which external shocks and macroeconomic mismanagement had plunged them. By the early 1990s about two-thirds of the Bank's clients had initiated policy reforms supported by structural or sectoral adjustment loans (SALs/SECALs). A large number of these loans-about 250 SALs/SECALs to 86 countries between FY1979 and 1994-had fiscal reform components. This study assesses how well Bank- supportedfiscal adjustment worked and reflects on how loan performance can be improved in future operations. What the Study Found 2. The principal findings of the study are that: (i) successful fiscal management requires sustained, long-term effort, (ii) lower fiscal deficits are associated with faster economic growth, (iii) sustained deficit reductions were achieved primarily through revenue enhancement, while expenditure reduction targets were more elusive and (iv) the fiscal reform component of Bank- supported adjustment lending had only limited success because of the fragmented treatment of fiscal issues and the vague conditionality of loans. General Findings 3. Management offiscal deficits requires sustained, long-term adjustment effort. Deficits of SAL/SECAL countries rose in the medium term; only in the long term (years 5-8 following the first loan) were there substantive signs of budgetary improvements. Though deficits were lower in the most recent two years (relative to the 4-year pre-loan period) in SAL/SECAL countries as a whole, there were significant reversals: low-income adjusters, severely indebted adjusters, and primary exporters showed no improvement. 4. The principal source of large fiscal deficits was fiscal mismanagement not exogenous shocks. Though the foreign interest burden did not constrain adjustment in several SAL/SECAL countries, the growing ratio of external debt to GDP across most regions indicates that the improvements in fiscal management were not strong enough to sustain solvency in many of the Bank's clients. 5. Deficit reduction leads to improved external balances and stimulates growth. Because fiscal deficits can contribute to many economic ills (inflation, credit shortages, distorted interest and exchange rates, current account imbalances) they can crowd out private investment and hurt economic growth. Although growth performance was uneven across regions, the data confirms that while other conditions are also necessary for growth, good fiscal management helps. The data also indicate a strong association between improvements in fiscal deficits and in current account deficits. 6. Sustained deficit reductions were achieved primarily through revenue enhancements and to a lesser extent through reductions in capital spending. Targeted reductions in current expenditures proved to be more elusive. Most countries that made fiscal improvements did so by increasing revenue. Countries with worsening fiscal deficits generally increased current expenditures without increasing revenues. Current and capital expenditure strategies were 8 consistent across country groups and regions and over time: all regions and country groups except the Middle East and North Africa and the middle-income adjusters increased current expenditures, and all groups except East Asia and Pacific pruned capital expenditures relative to the preloan period. Reductions in current expenditures, while considered necessary in many cases, were only achieved in a small number of countries. 7. The record of borrower compliance with Bank conditions was excellent, yet Bank- supported adjustment lending had only limited fiscal success because fiscal issues were treated in a fragmented manner and conditionality was vague. Most operations paid little attention to coverage, measurement, or sustainability of the deficit or to the role of the public sector (a critical factor determining the level of the public finances). Though fiscal policy is only one aspect of a larger adjustment program of interdependent policy reforms, few operations discussed the relationship between fiscal deficits and instruments for reducing them, and other macroeconomic variables and structural reform measures. Absent too was an integrated approach to the complex issues connected with budgetary reform (budgetary discipline, involvement of local beneficiaries in expenditure choices, technical efficacy, accountability, information systems). Conditionality on the budget deficit or public savings was too general, inconsistent between loans for the same country, or subject to different interpretations. The vagueness of fiscal conditions made implementation and monitoring difficult or impossible (in 68 percent of operations covered by Project Completion Reports/Performance Audit Reports- PCRs/PARs). Conditions were "soft," calling for legislation to be drafted rather than implemented (47 percent) and stressing studies rather than implementation (16 percent). Conditionality was occasionally poorly timed (32 percent). Specific Findings on Aspects of Fiscal Reform Revenue Reform 8. The Bank's advice on tax reform has generally been consistent with best practice as set out in the Bank's Lessons of Tax Reform (1991). Tax administration reforms have generally been consistent with state-of-the-art recommendations for improving taxpayer compliance and operation of tax departments. But some problems remain. Revenues rose in only 17 of the 26 countries where tax measures (base broadening, simplification, or rationalization) had been identified as necessary and as likely to boost revenue. Weaknesses in tax administration appear to at least partly explain why revenue fell short of expectations in several countries (Bulgaria, C6te d'Ivoire, Hungary, Senegal, Venezuela) and why revenue increases, where they occurred, failed to have the desired fiscal impact (Bangladesh, Pakistan, Uganda). Tax administration was also neglected in some early SALs (Bangladesh, C6te d'Ivoire, Hungary, Pakistan, Uganda). 9. In general, conditions on nontrade tax reforms have constituted only a small component of SALs/SECALs and have been selective rather than comprehensive. In the sample of 92 SALs/SECALs covering 53 countries with nontrade tax conditions (out of 171 operations in 75 countries), conditionality sought largely to broaden direct and indirect tax bases and to lower marginal income tax rates in order to make tax systems simpler, more efficient, and easier to administer. Typically, conditionality supported government reform initiatives or piggybacked on other institutional initiatives. The Bank generally deferred to the International Monetary Fund's lead in nontrade tax reform in countries where the Fund was involved in the tax area. 9 10. About 40 percent of conditionality was quantitative and easily monitorable: meeting certain quantitative criteria or making specific changes in tax bases or rate structures. However, about 48 percent of conditions stipulated specific statutory changes but did not provide quantitative measures of performance; these conditions generally specified the direction of change in the tax code without mentioning the magnitude of change expected. About 12 percent of conditions lacked both quantitative and qualitative criteria. Thus about 60 percent of conditions left open the possibility of poor outcomes: some could be satisfied dejure by minor alterations in the tax code, others simply could not be adequately monitored or evaluated. 11. None of the SALs/SECALs specified pre-reform benchmarks such as marginal effective tax rates. Partly as a result, none of the PCRs/PARs could evaluate implementation and outcomes in terms of changes in these rates, and very few verified compliance by comparing the record before and after reform. Thus the effectiveness of tax conditionality was generally evaluated in terms of statutory compliance (fulfillment of conditions stipulating changes in tax statutes) or increases in tax ratios even though such ratios are affected by many other factors. Although few conditions called specifically for increases in tax rates, tax ratios did increase (through increases in tax bases) in a large number of countries. The shifts in tax structure that followed tax reform were generally consistent with Lessons of Tax Reform-in the direction of less reliance on trade taxes and greater reliance on domestic indirect taxes. 12. The record for trade taxes was much better. Conditionality was specific, and reform objectives were spelled out as clear-cut, quantified policy targets to be reached before a specified deadline. Such precision makes compliance relatively easy to monitor. Though compliance was good, the pace of reform varied, often depending on how well trade reform was synchronized with other reforms. Revenue pressures often held up trade reform. So did insufficient progress on stabilization. Little effort was made to quantify the revenue effects of trade reforms; inadequate attention went to harmonizing sales and excise taxes for imports and domestic goods. 13. Tax administration reform was generally imprecise and unmonitorable. Though the Bank's involvement in tax administration reform has increased significantly since 1987, the record suggests piecemeal reform rather than a comprehensive reform strategy. Only 40 percent of conditionality was couched in precise terms. PCRs/PARs that tracked implementation of conditionality pronounced compliance to be high, but without monitorable performance indicators such findings are not very persuasive. The most common yardstick was the ratio of taxes to GDP, even though tax collections are affected by many factors other than administration. As with nontrade tax reform, imprecision and a paucity of monitorable indicators impaired the quality of tax administration conditionality. Expenditure Reform 14. The Bank has frequently helped countries define expenditure reform; expenditure conditions were incorporated in 219 operations in 83 countries during FY1979-94. In general, expenditure reform efforts have been broadly consistent with the directions discussed in the Bank's report(s) on adjustment lending (1988, 1990, 1992). Conditions have focused on reducing and restructuring capital expenditures, reducing subsidies, reforming public employment and cutting the public wage bill, emphasizing operations and maintenance (O&M) of existing facilities, and increasing social sector spending (though attention to poverty reduction has been weak). 10 15. Countries with expenditure conditions did only marginally better than countries without them in improving expenditure patterns. Countries with conditions had greater success in increasing social spending and cutting subsidies and defense spending, and they ended the period with lower wage expenditures than countries without conditions. Although both groups started out with roughly equal ratios of total expenditure to GDP, countries with conditions had less success in cutting total expenditure. Success has been limited in restructuring current expenditures away from wages and subsidies and toward nonwage O&M, and attempts to cut public employment have generally been short-lived or modest. Countries with capital expenditure conditions ended the period with very low levels of capital expenditures, reflecting relatively large cuts in infrastructure spending. One reason for the relatively modest performance on the expenditure front may be the inflexibility that afflicts budgets in many countries; Bank adjustment lending made little headway in this area. 16. In countries with education and health conditions both sectors increased their shares in total spending. Yet education and health expenditure ratios were only marginally better in countries with such conditions than in countries without. For both groups of countries illiteracy rates declined, particularly for girls and women, as did infant mortality rates. Cost-recovery efforts in the social sectors were limited. One reason that Bank conditions may have had little additional impact on social sectors is that the Bank emphasized the means rather than the ends of expenditure reform. Conditions focused on changing expenditures in specific functional categories or sectors rather than on the relation between inputs (such as level of expenditures), outputs (achievement scores, graduation rates), and final outcomes (rates of literacy, enrollment, mortality; access to basic services, particularly by the poor and needy). What Can Be Done? Suggested Priorities for Bank Work 17. On Bank-Fund Coordination. The centrality of fiscal reform to macroeconomic stability and growth has been well recognized by the Bank and the Fund, and the two institutions have worked out a division of labor in recent years. The Fund takes a lead in issues relating to tax and aggregate level of spending, whereas the Bank leads on sectoral spending issues. Although this division of labor works well for the most part, it cannot be viewed inflexibly. Thus while respecting the Fund's lead role in giving fiscal advice, the Bank needs to take a more hands-on- approach to the design of fiscal reforms in adjustment lending, particularly where the Fund may not be directly involved. For example, the Bank needs to consider the relationships between tax policy reforms undertaken in adjustment programs and other structural reforms (as in private sector development, financial and capital market reforms); between the size and role of the government and the level of the public finances; between the financing of the budget and the level and composition of public expenditures; and between tax and expenditure decisions and poverty programs. The Bank needs to build up analytical capacity to advise staff on fiscal and adjustment issues, to identify best practice, and to analyze new fiscal policy issues that are likely to assume importance in the future (financial sector taxation, environmental taxes). In this respect, the Bank's Policy Research Department, which withdrew from tax analysis in October 1994 to concentrate on expenditure issues, needs to correct the balance. Such a capability will enhance Bank-Fund coordination and strengthen dialogue on fiscal issues. The Bank and Fund are exploring new and existing channels for improving coordination in the preparation of country macroeconomic frameworks (Policy Framework Papers, Public Expenditure Reviews, Bank 11 Country Assistance Strategy papers, Fund Country Assistance Briefs). The Bank and Fund also need to collaborate in improving their own fiscal databases and client fiscal data systems. 18. On loan conditions. Fiscal conditionality in adjustment operations needs to be more specific and precise, supported by specific and monitorable performance indicators, and focused on fiscal (and related macroeconomic) outcomes in addition to the means of reform. Conditionality should stress implementation of new laws as much as enactment and implementation of study recommendations more than the preparation of studies. 19. On fiscal deficits. The Bank needs to ensure greater precision on several fronts: * The coverage of fiscal accounts needs to be expanded and the transparency of accounts improved. This implies a need to incorporate state, local, and municipal accounts into the fiscal accounts, track quasi-fiscal deficits, and work out the implications of implicit taxes, subsidies, and government guarantees. * The appropriate measure of the deficit for tackling the problem at hand needs to be identified. In most countries, the overall deficit may be appropriate; in some, it may be the operational or the structural deficit; in others, public savings may need to be the focus; in yet others, primary deficits may need monitoring. * Fiscal accounts must be better integrated into the macroeconomic framework of analysis to enable more dynamic analysis of fiscal and other macroeconomic adjustments. This is essential for estimating the sustainable deficit; identifying tradeoffs among acceptable levels of borrowing, inflation, interest rates, and other macroeconomic variables; and setting quantitative targets for achieving a sustainable deficit and monitoring macroeconomic outcomes. 20. On tax reform. The main areas of Bank attention should be: * Simplification and base broadening. Bank work in tax reform should continue to stress broadening and simplifying tax bases for easier administration and better compliance. * Sequencing of tax reform. The Bank needs to emphasize the proper sequencing of tax reform. If trade taxes contribute significantly to revenue, nontrade tax reforms to enhance revenue may need to precede trade tax reform, which normally results in lower revenues. Reforms in tax administration should precede or accompany tax structure reform. * Environment, equity and private sector development concerns. Bank work on tax reform should involve considerations of environmental concerns; the tax system should also not adversely affect equity/poverty reduction or the international competitiveness of the economy. * Tax administration. Tax administration reform should focus on a conducive legal environment, taxpayer education, information systems (used by tax departments), audits, withholding and presumptive taxation, and improvements in VAT administration. 12 Monitoring progress. Building better performance indicators into tax and tax administration reform programs would benefit design, implementation, and evaluation and focus attention on the goals of reform. Performance indicators should cover effective tax rates and, for tax administration, cost and output measures. 21. On public expenditure reform. The Bank should focus on: * Establishing the basis for the recommendations. Recommendations on public expenditure reform should be derived from considerations of the appropriate role of government. Governments need to be persuaded to move away from a command and control culture toward a service delivery outlook. They also need to be encouraged to invite local governments, potential beneficiaries, and other stakeholders to participate in discussions on development and spending priorities, particularly in social sectors. This would entail reform of budgetary processes and institutions. * Clartfying the goals of reform. Focusing on the ultimate goals of expenditure reform right from the start is important both for the proper selection of expenditure (or other) instruments and for better implementation. The Bank needs to promote the development of appropriate performance monitoring indicators that include intermediate (short-run) outputs and final (long-run) outcomes as well as measures of the level and composition of expenditure. * Increasing attention to expenditure composition. When deficit reduction is a major objective, both current and capital expenditures need to be examined, with special attention to inflexible and unproductive expenditures (such as defense and subsidies not oriented to the poor). Although the Bank's emphasis on capital spending cuts was warranted in countries where large capital projects did not meet the basic needs of the population, in some countries capital spending shares (of total spending or GDP) have fallen to levels that may be retarding growth. In these countries and in others where much of the population lacks access to basic services, basic investments may need a boost. * Emphasizing poverty alleviation. Expenditure programs have greater potential than revenue reforms as instruments for reducing poverty. Thus the Bank needs to pursue expenditure solutions for better targeting of subsidy programs and better access by the poor to basic services. The Bank might also seek innovative solutions to the problems of poverty, such as using education programs to teach poor households about safe water and sanitation practices. The Bank also needs to encourage clients to direct resources to rural areas, where most of the poor live. * Promoting equity. Basic education and health services should be almost free to ensure universal coverage, while innovative cost-recovery solutions that also promote equity are needed for tertiary education and health services. 13 1. Setting the Stage 1.1 Developing countries as a group experienced a series of external shocks in the late 1970s and early 1980s in the form of rising international interest rates, falling terms of trade, and loss of access to commercial bank credit. Initially the crises were perceived as external account problems, but it soon became clear that there were internal imbalances to contend with as well. For countries that had had access to borrowing in private capital markets, the recycling of petrodollars had encouraged ambitious, but poorly conceived and low-return, public and private investment programs. When international interest rates rose, debtor countries found themselves unable to borrow their way out of their debt service obligations; for low-income countries the exacerbation of crises was attributed to deteriorating terms of trade. To help developing countries deal with these problems, the World Bank embarked on adjustment lending in 1978-79. The Typical Crisis Profile 1.2 Several macroeconomic developments presaged the crisis in developing countries They are summarized here and in Figures 1.1 and 1.2 for each of five Bank regions (data limitations precluded the inclusion of ECA).3 * Fiscal deficits were large and persistent in all regions except EAP and tended to rise over time. At their peak, deficits ranged from 5 percent in LAC to 9 and 10 percent in MNA and SAS to 13 percent in AFR. * GDP growth rates plummeted. Growth rates fell from more than 9 percent to 3 percent in EAP, from 6 percent to 3 percent in SAS, from more than 3 percent to less than 1 percent in AFR, and turned negative in LAC and MNA. * Inflation surged to double-digit levels in all regions except MNA, where it reached 9 percent. * Current account deficits were high and volatile. Peak levels of 4.5 percent were registered in SAS, 6-7 percent in LAC and EAP, 11 percent in MNA, and 15 percent in AFR. * Private investment was low. In AFR, LAC, MNA, and SAS private investment was 10 to 12 percent of GDP and stagnant or declining; in EAP it was higher, at 19 percent of GDP, but stagnant. These developments pertain to the four-year period prior to the Bank's first adjustment loans to individual countries in each region. 2 Referred to here by their Bank abbreviations: AFR for the Africa Regional Office, EAP for East Asia and Pacific Regional Office, ECA for Europe and Central Asia Regional Office, LAC for Latin America and the Caribbean Regional Office, and SAS for the South Asia Regional Office. This section draws on a background paper by Chandra and Anees (1996). 14 Figure 1.1. Macroeconomic Indicators during the Preloan Period Presaged the Onset of Crisis 25- 20 ~15--4 10_ 51 GDP growth Fiscal Current Inflation deficit/GDP account deficit/GDP isLAC m EAP 0 SAS * MNA *IAFR Rgure 1.2. Private Investment and Interest on External Debt Were Low during the Preloan Period 20 18 - _ _ _ ____ 14 0 (D 6 - 4-- 2- LAC EAP MNA SAS AFR * Interest payments on external debt/GDP Private investment/GDP Note: Preloan period refers to the four years preceding the first adjustment operation in a country. Fiscal deficits refer to overall deficits of consolidated central governments. Source: IMF, Government Finance Statistics and International Finance Statistics; World Bank, World Tables and World Debt Tables. 15 Contrary to common belief, on average neither the terms of trade4 nor interest payments on external debt were a predominant source of fiscal instability. Interest payments remained below 2 percent of GDP in EAP, SAS, and AFR and averaged 3 percent in LAC and MNA. 1.3 External pressures initially induced many governments to assume private debt and to finance their way out of fiscal pressures by creating money. Inflation rose, the tax base deteriorated, current account deficits widened, and governments resorted to devaluation to contain external imbalances. The public (and private) sector's external debt obligations grew in local currency terms, transforming the external debt problem into an even larger fiscal problem. The inability to raise taxes to cover expanded expenditure programs led to indiscriminate expenditure cutting and to the realization that tax systems were inherently flawed. 1.4 Policymakers in developing countries eventually came to recognize the need to distinguish between the short-run problems of stabilization and longer-run problems impeding the resumption of growth and to devise reform programs that addressed both types of problems. Stabilization measures addressed problems that needed urgent attention: inflation, loss of foreign exchange reserves, capital flight, and burgeoning current account deficits. Structural impediments to long-term growth could usually be traced to inadequate or malfunctioning institutions or to distortions in the policy framework: public enterprises that were mismanaged or required to pursue conflicting objectives, public sector banks that were induced to finance the losses of public enterprises, distorted production incentives, inefficient public expenditure programs, excessive taxes and subsidies, and controls on prices, interest rates, and imports, which sent the wrong signals to economic agents. When the IMF and the Bank stepped in to support developing countries with policy-based financial assistance and advice, the IMF generally took the lead in stabilization and the Bank in structural adjustment. As we shall see throughout this report, the Fund's lead role in fiscal reform has affected the Bank's approach to the design and scope of fiscal programs in structural adjustment loans (SALs) and sectoral adjustment loans (SECALs). The Centrality of Fiscal Reform 1.5 As awareness of the need for structural change grew, policymakers also realized that "prudent fiscal policy is the foundation of a stable macroeconomic environment" (World Bank 1991b)6 because fiscal policy affects private sector savings and investment and so the private sector's contribution to growth. Fiscal variables affect private economic decisions both directly by affecting taxation and public sector pricing policies and indirectly by affecting other macroeconomic variables. Depending on how deficits are financed, they can lead to inflation, the crowding out of private investment, current account deficits, or problems of external The regional terms of trade did not worsen sharply enough relative to their levels in 1980 to explain the onset of fiscal imbalances. The impact of rising international interest rates played itself out more forcefully in the 1980s. 6 The macroeconomic implications of fiscal deficits are well documented in several Bank studies including OED publications (Jayarajah and Branson 1995; Thomas, Chibber, Dailami, and de Melo 1991; and World Bank 1988a). See also Faini and de Melo 1990. 16 creditworthiness. Recent research in the Bank by Easterly, Rodriguez, and Schmidt-Hebbel (1991) demonstrated that public sector balances are correlated with a range of macroeconomic variables, including GDP growth, per capita growth, the real interest rate, money creation, investment and consumption rates, the real exchange rate, and the current account. Achieving fiscal balance is therefore central to the success of the adjustment process. While the level of the fiscal deficit is a good indicator of the need for fiscal restraint as a central part of the reform process, the size of the required fiscal adjustment is usually a matter of debate. The balanced budget rule has given way to a more flexible approach based on a sustainable budget deficit- one that can be financed without causing major macroeconomic imbalances.7 Looking at fiscal deficits in this context also places fiscal reform in perspective-as only one aspect of a larger adjustment program of interrelated policies. Coverage of the Report 1.6 This study analyzes the Bank's experience with fiscal management in about 250 completed and ongoing SALs and SECALs approved by the Board during FY 1979-94. The operations involved 86 countries in all six of the Bank's regions. Bank financial support amounted to $34.3 billion, with the largest shares and the greatest number of operations in AFR and LAC (Figure 1.3). Figure 1.3. FY 79-94: AFR and LAC Account for: Largest Number of Adjustment Loans Greatest Share in Total Adjustment Lending LAC MNA SAS 23% 5% AFR ECA SAS MNA 30% 6% 7 EAP EAP LAC 10% 6% FR 30% ECA 16% Source: ALCID. 1.7 The study looks at fiscal management from both a "micro" and an aggregate perspective. Chapters 2 through 5 take a micro stance, analyzing fiscal adjustment efforts loan by loan and at the country level. The objective is to identify common patterns in the diagnosis, conditionality, and implementation of fiscal programs (Chapter 2) and in the design (precision and content), implementation, and outcomes of tax policy (Chapter 3), tax administration (Chapter 4), and public expenditure (Chapter 5) reform. Because the areas of nontax revenue and public enterprise reform span a very large number of sectors and issues (public sector management, cost recovery, fee structure, privatization, restructuring, among others), this study focuses more narrowly on tax and expenditure concerns. The study builds on the common patterns identified in these areas and focuses on relatively simple and tractable lessons for improving the design, See Anand and Wijnbergen (1990), Dinh (1995), and Chapters 2 and 6 of this study. 17 implementation, outcomes, and monitoring and evaluation of fiscal programs and their components. 1.8 Chapter 6 pulls together a picture of the ultimate outcome of fiscal reforms after more than a decade of SAL/SECAL experience and draws lessons and recommendations. The picture is presented both in the aggregate and by groups of countries that share common characteristics. Was fiscal adjustment sustained? What instruments were used to reduce deficits? Did adjusting countries experience growth? What was the association between fiscal deficit reduction and other indicators of macroeconomic performance-inflation, current account deficits? What lessons emerge from linking outcomes to patterns of fiscal diagnosis across specific countries? 1.9 The study draws on the Bank's ALCID database to identify SALs/SECALs and the types of conditionality incorporated in adjustment operations. Because the universe of loans and countries is so large, each chapter draws on loan samples from the ALCID to analyze specific aspects of fiscal management; however, the size and composition of these samples vary depending on the subject and its importance in the SAL/SECAL population. The details on sample size and composition are set out in each chapter. Table 1.1 indicates the representativeness of each of the samples drawn from the varying populations of adjusting countries and adjustment operations relevant for evaluating the specific areas under consideration. Table 1.1. Representativeness of Samples Used in the Study Sample as Sample as Total number percentage of Total number percentage of of relevant total of relevant total Areas ofjstudy countries countries operations operations Diagnosis/conditionality implementation 86 31 252 51 (Chapter 2) Nontrade tax reform (Chapter 3) 53 100 92 100 Trade tax reform (Chapter 3) 75 31 171 51 Tax administration reform (Chapter 4) 47 74 67 61 Public expenditure reform (Chapter 5) 83 100 219 100 Macroeconomic overview (Chapter 6) 75 35 229 59 Source: ALCID. 1.10 The study also taps into Bank and Fund databases, President's Reports (PRs), Project and Implementation Completion Reports (PCRs/ICRs), Project Audit Reports (PARs), and other Bank documents and staff estimates. However, the data series used, particularly those relating to the fiscal accounts, are based on different definitions and conventions and draw on different institutional sources within the Bank's client countries. In view of this, the study has attempted to ensure consistency in data use within each chapter. However, there are instances, as in Chapter 5, when more than one data source has been enlisted for the same variable; such instances are pointed out in the text and tables. 19 2. Fiscal Management in SALs/SECALs: A Close-Up View 2.1 This chapter takes a close look at fiscal management by examining 129 completed and ongoing adjustment loans in 27 countries drawn from across the Bank's regions. The analysis tracks patterns underlying the diagnosis of fiscal ills, establishes whether diagnosis and conditionality were internally consistent, determines whether implementation was adequate, and identifies common problems. The emphasis is on identifying elements of commonality rather than on specific details (Chapters 3-5 analyze the details of tax and expenditure reform). Even so, not every issue raised is common across the board. This chapter advocates fuller treatment (diagnosis and conditionality) of specific issues without implying that such treatment is called for in every report; that is a matter of judgment specific to each case. 2.2 The 129 operations in the sample were chosen with a view to capturing the range of attributes of the countries and operations covered by the Bank's adjustment loans (see Annex 2.1). Random sampling could not be used because of the diversity of the criteria involved. The sample was selected to include loans that reflect: * Specific types of conditionality (including specific budget deficit targets, tax policy and administration reforms, expenditure reforms, and improvements in budget processes). * Regional shares in SALs/SECALs. * A large number of the Bank's top 15 borrowers. * Low-and middle income countries. * Operations covered by PCRs and PARs.2 * Successful and unsuccessful operations, as assessed by OED in PCRs and PARs. Although the number of sample countries represents only 31 percent of total SAL/SECAL borrowers during FY 1979-84, the number of operations represents over half of all SALs/SECALs and about 64 percent of the value of loans and credits approved during the period. What Was the Quality of Fiscal Diagnosis? 2.3 While this study recognizes that fiscal policy is only one aspect of a larger adjustment program of interrelated policy reforms (see. para. 1.5), it limits its purview to fiscal policy per se. The diagnosis of fiscal issues typically sets the stage for identifying fiscal conditionalities in the Bank's PRs for SALs/SECALs. To diagnose the factors behind fiscal imbalance, basic questions need to be asked about the nature and source of the deficit, the role of government, 1 This chapter is based on a background paper by Gupta (1996). 2 A few operations were covered by Implementation Completion Reports (ICRs), which replaced PCRs in July 1994. 20 budgetary processes, and the level and composition of public expenditures and revenues (Box 2.1). The study concludes that the quality of fiscal diagnosis in Bank SALs/SECALs was weak-PRs indicate that Bank programs did not systematically work through the issues that need to be addressed to ensure adequate diagnosis. On the average diagnosis was somewhat better in SALs than in SECALs, but this pattern does not necessarily hold in individual countries. Box 2.1. Basic Elements of Diagnosis 1. What was the major source of the deficit? Did the measure of the deficit reflect the intrinsic fiscal problem? Was the fiscal deficit sustainable? 2. Was the role of government at issue? In light of a more appropriate role for government, what was the source of the fiscal problem (budget systems, expenditures, revenues)? 3. Where budgetary processes were weak, how were they to be reformed? 4. Were total expenditures to be cut? Were public investments to be reduced or restructured? Or were current expenditures to be restrained or allocated differently across sectors? Which components? Were public enterprises to be reformed? 5. Would increasing public revenue raise public savings without substituting for private savings? How was revenue to be increased? By increasing tax rates or broadening tax bases or improving tax administrations? Or by increasing nontax revenues? 6. Were the fiscal accounts effectively integrated with other macroeconomic accounts? Were projections spelled out for total expenditures, revenues, and their components to facilitate ex post evaluation? Measuring the Fiscal Deficit 2.4 Good fiscal diagnosis depends on accurate measurement of the public deficit. Good measurement has two aspects: ensuring appropriate coverage of the various levels of government and selecting the measure of the fiscal deficit that best captures a country's specific fiscal ills. 2.5 Coverage of the Deficit. SALs/SECALs need to work through the fiscal accounts at the broadest possible level before zeroing in on the specific accounts that are largely responsible for the overall fiscal imbalance. The adequacy of coverage was difficult to determine because coverage was not specified in almost half (61 of 129) the operations. The source of the deficit was identified in only 68 loans or 53 percent of the sample reviewed (Figure 2.1). In these 68 operations, coverage was limited to the central government accounts in 30 operations, to general government (central government and state and local entities) in 6 operations, and to the nonfinancial public sector (the general government and state-owned enterprises) in 17 operations; in 15 operations coverage did not fit any of the above categories but varied from loan to loan (Figure 2.1). Even when the coverage was broad enough to represent the nonfinancial public sector, some levels of government were excluded, as in Argentina, which explicitly mentioned local government deficits as contributors to the overall deficit, but excluded 21 the provincial accounts. In general PRs did not separately quantify or present the various components of the fiscal accounts. Figure 2.1. Coverage of Fiscal Accounts Was Specified in Only Half of the 129 SALs/SECALs Nonfinancial public sector 13% , . Did not specify 47% General 5% Central Government Other 23% 12% Source: President's Reports for SAL/SECALs. 2.6 Though attention has recently focused on quasi-fiscal deficits, central bank losses were identified in only nine countries (Argentina, Chile, Costa Rica, Ghana, Honduras, Mozambique, the Philippines, Tanzania, and Uruguay) and even in those cases the losses were not uniformly reflected in fiscal deficit estimates.4 Evidence in PRs suggests that large quasi-fiscal deficits existed in other countries also (Box 2.2). Social security systems in Argentina, Bulgaria, Chile, Costa Rica, Hungary, Poland, Tunisia, and Uruguay were also found to have extended overly generous benefits covered by transfers from central governments. In Argentina (1988), the nonfinancial public sector deficit excluded the provincial accounts, but Argentina (1991) acknowledged that more than half of the nonfinancial public sector deficit was accounted for by the provinces and municipalities. In Venezuela, where the central government assumed the external debt of public enterprises and financial support was provided by the state-owned industrial bank, such support did not figure in the fiscal accounts (Venezuela 1989). Turkey PRs (1982-84) presented the consolidated budget but in some years the public sector borrowing requirement (including the consolidated budget, state enterprises, local governments and extra-budgetary funds) was twice the size of the borrowing requirement implied by the consolidated budget. In the Philippines (1983, 1987) the accounts referred only to the national government, but the national government deficit averaged only about 2 percent of GDP in 1983-85 when the total public sector deficit including central bank losses was about 8 percent. In Bangladesh (1992) where public enterprises accounted for 6 percent of GDP and required budgetary support, their losses were not reflected explicitly in the budget. Quasi-fiscal deficits reflect the activities of the central bank or other financial institutions that exert pressures on financial markets similar to those exerted by the financing of recognized government activities. Fiscal deficit estimates need to identify and incorporate arrears (of inter alia, public enterprises and banks). Arrears of insolvent enterprises and banks amounted to more than 4 percent of GDP in 1989 in Hungary. They were also sizable in Poland, Morocco, and Tunisia The Uruguayan social security system required transfers amounting to 3.8 percent of GDP and 25 percent of current revenues in 1985. In Poland budget subsidies to the social funds rose from 2 to 8 percent of GDP between 1990 and 1992. 22 Box 2.2. Examples of Quasi-Fiscal Activities of Financial Systems LAC. In Argentina the coverage of accounts did not include central bank losses till 1988, even though the central bank provided housing and other subsidies to favored sectors. In 1986 Argentina's fiscal deficit was relatively low, but the central bank was losing about 2 percent of GDP a year. Nonfinancial sector accounts improved with the Public Enterprise Reform Loan (FY91) and Public Sector Reform Loan (FY92, but the primary surplus was insufficient to cover the large nominal interest bill on domestic and external debt. In Chile the fiscal target in 1985 was public savings and the nonfinancial public sector deficit even though central bank losses were nearly 18 percent of GDP (the bank had assumed the exchange rate and portfolio losses of private banks). In Uruguay the central bank generated most of the deficit when it rescued four banks that failed in 1986. In Costa Rica the central bank accumulated heavy losses from paying most of the direct interest costs and exchange losses associated with the rescheduling agreement with foreign commercial banks. The conventional deficit was 1.8 percent of GDP in 1985; central bank losses were 5.3 percent. Losses fell to 2 percent in 1989/90 and were addressed in SAL III, which was canceled. In El Salvador the central bank extended many credit lines, and its losses were estimated at 1 percent in 1990/91. AFR. Evidence of central bank losses was found in five countries, yet the overall central government deficit was the measure used during the period of Bank operations. In the Gambia the central bank financed a large marketing board. The Bank of Ghana took on such quasi-fiscal activities as development finance, subsidized lending, direct participation in rural banks, and absorption of foreign exchange losses on behalf of the government. In Mozambique fiscal and credit activities were carried out by both Banco de Mozambique and the Ministry of Finance. The bank incurred quasi-fiscal losses when it extended interest-free short-term credit to the government or assumed foreign exchange losses. In Tanzania the central bank gave automatic access to credit to the agricultural marketing system. Quasi-fiscal activities are estimated to have added about 8 percent of GDP a year during FY 1984-88 to the more narrowly defined fiscal deficit. The Bank of Uganda took over the provision of crop finance from the commercial banks from 1989 to 1991 and provided credit to the Coffee Marketing Board, which the board was not able to repay. The unprogrammed expansion in credit caused inflationary pressures. EAP. In Indonesia the central bank extended subsidized credit to rice farmers and agricultural cooperatives. In the Philippines the central bank incurred heavy losses from assuming foreign currency liabilities of public and private enterprises and from advancing payment of foreign interest on publicly guaranteed debt. These losses accounted for 55 percent of the consolidated budget deficit in 1986-90. The deficits affected the capacity to carry out independent monetary policy and inhibited liberalization of the exchange rate. MNA. In Tunisia credit was extended to priority sectors at concessional rates. The resources came partly from banks, which became eligible for rediscounting at concessional rates, and partly from the treasury. In Morocco the treasury provided interest rebates on certain private investments and gave preferential credit to special institutions, in addition to covering their exchange risk on foreign borrowing. ECA. In Poland quasi-fiscal subsidies through the banking system amounted to 9.2 percent of GDP in 1989. SAS. The State Bank of Pakistan provided low-cost credit lines to enable financial institutions to earn an adequate spread. Concessional credit was also provided to locally manufactured machinery and to an export finance scheme. 2.7 Selecting the Appropriate Deficit Concept. The appropriate measure of the public deficit (whether overall or primary, public savings, structural or operational) depends on the purpose for which it is being used (Blejer and Cheasty 1991; World Bank 1988a). If the objective is to measure net claims on resources by the public sector (which affects inflation, domestic interest rates, and, to some degree, external balances), the appropriate concept is the public sector borrowing requirement, or the overall deficit of the nonfinancial public sector. The Fund, which puts considerable emphasis on measuring the fiscal stance, routinely relies on this concept as a first step. Other measures are seen as useful supplements to the overall balance in specific circumstances. Despite the diversity of country experience, 23 of the 33 operations that 23 specified a deficit measure relied on the overall deficit. Perhaps for consistency with IMF practice, eight SALs/SECALs (all in LAC) used public sector savings (the difference between total current revenue, usually excluding grants, and current expenditures) as a measure of the government's contribution to total savings.6 When the focus is on how to affect the public sector's net indebtedness, the primary (or noninterest) deficit is taken to be the appropriate measure: primary deficits were discussed in six operations but included in policy matrices in only two.7 In high-inflation economies the operational deficit-which incorporates an inflation correction to eliminate that part of interest payments that compensates creditors for the loss in the real value of the debt principal-may give a truer picture than a nominal measure. The operational deficit was mentioned in the text discussion in only one operation.8 And in economies subject to temporary disturbances from such exogenous factors as terms of trade shifts and interest rate shocks or from one-time sales of government assets, the structural deficit may be the appropriate one to use. Only I of the 129 loans made a reference in the text to the structural deficit (para. 2.32 and Figure 2.2).9 Identifying Sustainable Deficits 2.8 In addition to the appropriate coverage of the deficit, there is a need to determine what level of the deficit a country can live with: a sustainable deficit does not necessarily imply fiscal balance (see Chapter 6). Deficits are sustainable if they are consistent with feasible levels of other macroeconomic indicators: the balance of payments and foreign reserves, public debt (external and domestic), interest rates, and money creation and inflation (Anand and Van Wijnbergen 1989). Primary deficits are sustainable, according to Easterly, Rodriguez, and Schmidt-Hebbel (1994), if they do not push up the shares of debt and money to income. Dinh (1995) proposes a framework for calculating fiscal adjustments needed to achieve solvency and sustainability over time based on the path of the real interest rate, real GDP growth, and the interaction of fiscal, monetary, and exchange rate policies. 2.9 Estimates of sustainable deficits help determine what level of deficit is appropriate, what patterns of financing are feasible, and what path adjustment should take. Such estimates require the analysis of alternative scenarios spelling out the tradeoffs between fiscal and other macroeconomic variables. An important requirement is that the deficit be as broad as possible in its coverage; otherwise, the consistency of the macroeconomic framework is jeopardized. This study suggests that sustainable fiscal deficits be estimated for most countries, particularly for 6 However, its usefulness is marred by problems of classification (current or capital expenditures) and by the implicit assumption that all current expenditures represent consumption whereas all capital expenditures contribute to growth. Vito Tanzi (1993) points out that the concept has drawbacks in that it tells us "nothing about the impact of fiscal policy on the balance of payments, and perhaps not much about the impact of fiscal policy on growth." The primary deficit concept is central to the measurement of the sustainable deficit (see para. 2.8). It was discussed in Argentina (1991, two loans in 1992), C6te d'lvoire (1991), and Peru (1992), and Turkey (1988). 8 In Turkey (1988) the PR stated that the nominal public sector borrowing requirement overstated the government's claims on real resources compared with the operational deficit. The same loan also briefly discussed the primary balance. The discussions of the operational and structural deficits in Turkey (1988) and Argentina (1991) respectively were not followed up with conditionalities incorporated in their policy matrix. 24 those that have tended to overborrow or to rely excessively on printing money to finance their deficits. 2.10 This has not been the common practice. Bank SALs/SECALs generally embedded reforms in an "acceptable" macroeconomic framework usually estimated with the help of the RMSM or RMSM-X accounting framework. Although most SALs/SECALs arrived at some notion of acceptable fiscal deficit levels based on assumptions regarding target levels of growth, the current account balance, inflation, and creditworthiness, several SALs did not integrate public finances or money into the RMSM consistency framework. Even now, the RMSM-X does not generally link external debt and financing to the public finance accounts to examine the sustainability or solvency of public debt. Most SALs ignored internal debt. The distinction between domestic debt and external debt is becoming increasingly blurred, however. The linkages (through domestic debt) between fiscal policy and interest rates, exchange rates, and monetary policy are becoming stronger in the 1990s as world financial markets become more integrated. The Role of the Government 2.11 Only for a few countries (the transition economies of Bulgaria, Hungary, and Poland) did PRs explicitly discuss the role of government in relation to that of the private sector in the context of deficit reduction.10 Market failures and externalities and the need to alleviate poverty are issues to consider in assessing the government's rolell and in determining the appropriate mix of public interventions (through production or provision of goods and services, financing, or the regulation of private sector activities). Some PRs have moved in this direction, suggesting that governments concentrate on core activities (supply of public goods and services) and activities that confer large externalities or benefits to the poor. Only when the role of government is defined and generally accepted, can the level and composition of public expenditures and revenues be determined. Budgetary Processes 2.12 Budgetary processes were identified as the major institutional constraint to maintaining fiscal discipline in 16 countries (including Argentina, Bangladesh, Costa Rica, El Salvador, 12 Kenya, Morocco, and Uganda). Common problems were the absence of comprehensive budgeting, poor budget execution, vague accounts (particularly the links between the central '3 government and the provinces), inadequate control and audit of expenditures (Argentina), 10 The Bank's Lessons of Tax Reform notes that "it is important that tax measures not be chosen without having made a broader assessment of the role of government in the country concerned" (World Bank 1991 a, page 18). 11 A general appraisal of the role of the government is of course implicit in conditionalities calling for private enterprise reform, divestiture, elimination of marketing boards, and the like. 12 Budgetary process reform strategy was not discussed at all in the three transition economies in ECA, Indonesia and Philippines in EAP, Tunisia in MNA and Pakistan in SAS. 13 The links between central government and provincial fiscal accounts were cursorily discussed in the PRs for Pakistan and Tunisia. 25 earmarking and compulsory spending (Bangladesh, Costa Rica), and budget fragmentation (El Salvador). 2.13 In general the PR discussions suggested that budgetary process reform was an area in which the Bank needed to build up expertise. In LAC proposed strategies included modernization of auditing and control functions and reform of provincial processes in Argentina, amendment of laws to control earmarking in Costa Rica, and implementation of an integrated financial management system in El Salvador. In AFR the focus on strengthening ministries and developing capacity for economic management, particularly public investment programming. Only a few PRs discussed the need to extend the coverage of fiscal accounts and improve their transparency: Tanzania (1991) sought to make the quasi-fiscal activities of the central bank more transparent, Mozambique (1994) aimed to integrate foreign aid into the general budget, and Uganda (1991/92) planned to integrate its rehabilitation and development plan with the development budget. Although the depth of analysis improved in later adjustment loans in AFR (for example, in C6te d'lvoire [1991], Ghana [1987], Kenya [1986], Malawi [1992], Mozambique [1994], Senegal [1987], and Tanzania [1993]), not enough attention appears to have been paid to expenditure monitoring and control procedures, the databases required for such systems, forward budgeting, or management of extrabudgetary spending. Only a few PRs (Argentina [1991], Hungary [1991], Pakistan [1994], and Uganda [1994]) discussed decentralization and intergovernmental fiscal relations, which may have implications for overall expenditure efficiency and control. Total Expenditures 2.14 All sample countries focused on the need to adjust total expenditures. PRs identified a need to decrease total spending to curb the budget deficit in 22 countries. 14 Where total expenditure levels were low (El Salvador, Ghana, Peru, Philippines, and Uganda), PRs considered increases appropriate (Chapters 5 and 6). None of the PRs analyzed the consequences for the deficit of alternative levels of total spending.15 Public Investment 2.15 Operations in 24 countries focused on adjusting capital expenditures; in 10 the emphasis was on capital spending restraint or declines (Chapters 5 and 6). In Costa Rica (1985), C6te d'Ivoire (before 1991), the Gambia (both loans), Honduras (1988), Kenya (1980), Malawi (1981- 83), Morocco (1984), and Tunisia (1986-88) high capital expenditures reflected too many low- priority public investments. However, PRs considered that an increase in capital expenditures was warranted in as many as 14 countries where public investments had either been cut to undesirably low levels (below 10 percent in all countries, save Indonesia, and below 5 percent in as many as 7) or investments were needed to improve the core level of physical or social infrastructure to improve growth (see Table 5.8 in Chapter 5). Several countries (Bangladesh, Morocco, Pakistan, Philippines, Turkey, and several countries in AFR) sought to enhance efficiency by integrating the investment budget more fully into budgetary and expenditure 14 The transition economies identified expenditure cuts as necessary for a continued reduction in the role of the state. 15 This study did not draw on Public Expenditure Reviews (PERs) for its evaluation of public spending strategies. The assessment of PERs is the focus of a subsequent OED study of public expenditure which is currently in progress. 26 control systems and thereby highlighting the current expenditures needed to increase the productivity of investments. Social Expenditures 2.16 Reflecting the notion that the composition of investment is as important for growth as its level, the emphasis in almost all cases was on improving the structure of public investment. Social expenditures figured relatively little in strategy analysis in SALs/SECALs, particularly in the 1980s. The diagnosis of adjustment issues in El Salvador, Peru, Uruguay, and Venezuela focused on low social indicators, but there was no strategy follow-through in the SALs/SECALs; these issues were set aside for follow-on social sector projects or technical assistance projects. Discussion of social safety nets was also limited, perhaps because there was little analysis of the losers and gainers from expenditure reforms. In AFR only 15 of 54 loans (4 of them in Mozambique) mention social safety nets,. In EAP, MNA, and SAS the issue did not figure at all in strategy discussions, except in Pakistan (1989) and Morocco (1992). In the transition economies, which already had safety nets in place to compensate losers, the focus was on controlling expenditures and increasing efficiency. Current Expenditures 2.17 PRs discussed cutting or holding the line on current expenditures in 25 of the 27 sample countries; only in Ghana and the Philippines were increases considered. Reducing subsidies and transfers was discussed in 24 of the 27 sample countries (Chapters 5 and 6), though the impact of subsidies on the budget was not well analyzed, perhaps because financing was not always transparent. Explicit subsidies, or payments made by the government to consumers or producers, are reflected in the budget; implicit subsidies are not. But even explicit subsidies may be hard to detect in public finance data because they are buried in other public expenditure categories. The PRs indicate that implicit subsidies were sometimes financed by taxing producers (through controlled producer prices) or the banking system (through controlled interest rates to producers) or by amassing arrears on loans and taxes. Subsidy reform was particularly urgent in the three transition economies in the sample-Bulgaria, Hungary, and Poland-because it was linked to a wide range of structural reforms (the financial system through interest rate reform, the environment through reductions in energy subsidies, improved project selection through the phasing out of enterprise subsidies). Subsidies (consumer, production, enterprise investment, housing) accounted for 13 percent of GDP in Hungary in 1989 and 16 percent of GDP in Bulgaria in 1990. In other regions reform generally involved reductions in consumer subsidies and in transfers to loss-making enterprises. Transfers from the central government to state and local governments were not addressed. 2.18 Wage and salary reforms figured next in importance. Reductions in the wage bill were planned in 16 countries, mainly through job reductions. A few countries-Bangladesh (1986), Ghana (1987), Indonesia (1990), Tanzania (1986), and Uganda (1994)-had let wages and salaries fall so low, sometimes as a result of adjustment, that the strategy was to raise them. 2.19 The need to restructure current expenditures by squeezing subsidies and transfers so that operations and maintenance (O&M) were adequately funded was recognized in AFR, EAP, MNA, and SAS, but not in LAC or ECA. In general, later adjustment loans recognized the importance of adequate funding of O&M, perhaps in response to the recommendations of the 27 third report on adjustment lending (World Bank 1992b).16 However, explicit strategies, particularly for distinguishing between the wage and nonwage components of O&M, were not always spelled out. In the Gambia (1989), Ghana (1989), and Senegal (1990) public expenditure reform called only for monitoring O&M requirements. (For other examples of undifferentiated strategy, see footnote 30 in Gupta 1996.) Public Enterprise Reform 2.20 The need for public enterprise reform was identified in almost all countries. The weak financial position of public enterprises was usually at the heart of fiscal and banking sector problems in AFR, the transition economies in ECA, and in Argentina and Venezuela. State enterprise losses were often financed through the banking system, resulting in nonperforming loans and deteriorating bank portfolios in almost all AFR countries and transition economies. Losses by marketing boards were particularly large in Cbte d'Ivoire, Ghana, Kenya, Senegal, Tanzania, and Uganda. In C6te d'Ivoire price stabilization funds for basic consumption goods mushroomed into large, inefficient bureaucracies and became a constant drain on the budget. However, several PRs failed to take the next step and identify the fiscal or macroeconomic impacts of explicit and implicit support. In Senegal, where public enterprises accounted for 75 percent of the overall deficit in 1988-89, SALs in 1986, 1987, and 1990 continued to grapple with subsidies and cross debts between government and public enterprises.17 2.21 The strategies proposed varied. For transition economies defining and legislating the distribution of responsibilities between the private and public sectors were considered essential. Elsewhere, proposed reforms included reducing public employment, subsidies, and transfers; encouraging private sector activity; and increasing tax revenues and economic efficiency through liquidation, privatization, or rehabilitation to address production efficiency problems for enterprises that were to remain in public hands. Specifying clear objectives and providing incentives through performance contracts were discussed in the Gambia and Senegal. A variety of reforms were suggested for marketing boards: abolishing them (Senegal and Uganda), making them buyers and sellers of last resort (Kenya), and turning over processing and exporting activities to the private sector, adjusting producer prices in line with international prices, raising consumer prices, and reducing employment in state enterprises (Ghana and Tanzania). Revenue 2.22 Inadequate revenue was identified as a source of budgetary difficulties in all countries but Chile, yet only the Chile PR addressed the impact of revenue increases on public savings (and whether public savings add to or substitute for private savings). Nontax revenue was recognized as below potential wherever public enterprise charges did not cover marginal costs. In most countries tax revenue was diagnosed as insufficient because tax bases were too narrow and tax administration lax, not because rates were low (Chapter 3). 16 World Bank (I 992b) stressed the need to restructure current expenditures and to correct the balance between wage and nonwage O&M. 17 Explicit subsidies consisted of Treasury loans; implicit subsidies/taxes stemmed from tax exemptions, loan guarantees, preferential interest rates, and price controls. 28 2.23 Discussion of revenue strategies was extensive in LAC and ECA, including the transition economies. In retrospect, however, it appears that the loss of revenue from the demise of public enterprises in the transition economies was underestimated, since more aggressive tax strategies were not proposed. Despite an emphasis on tax instruments in LAC, the quality of the strategy discussions was thin in Costa Rica, Honduras, and Venezuela and improved in Argentina only in 1988. The limited focus on tax strategies in AFR and the Philippines may have been due to the fact that the IMF had taken the lead in this area of adjustment. Revenue strategies were not even mentioned till the 1990s in Bangladesh and Pakistan. 2.24 Broadening the tax base was the primary strategy identified in the tax area. Methods commonly included such tax-related changes as tightening exemptions, rationalizing incentives, expanding VAT coverage or the income tax base, and converting specific taxes to ad valorem taxes. Sometimes the strategy was to manipulate macro-policy instruments-indexing for inflation, freeing administered prices, and correcting for the overvaluation of the exchange rate (when exchange rate appreciation reduced the value of taxable traded goods in local currencies). Increases in tax revenues were also sought through changes in tax structure. Rarely were increases sought through tax rate increases alone. More commonly, rate increases for some taxes were offset by rate cuts for others. A reduction in the income tax rate was combined with an increase in the asset tax in Argentina, the VAT rate in Peru, and the sales tax and luxury goods tax in Ghana (Gupta 1996). (These strategies are spelled out in greater detail in Chapter 3.) 2.25 In line with the recommendations in the Bank's Lessons of Tax Reform, efficiency rather than equity was the focus of tax strategy discussions in PRs. Efficiency issues were discussed in 24 of the sample countries. Equity was mentioned in PRs in 16 countries, but the quality of the discussions was poor compared with the efficiency discussions. Attention to equity issues appears much later in the adjustment process: C6te d'Ivoire, Malawi, Senegal, Pakistan, and the Philippines received adjustment loans with fiscal conditions in 1980/81, but equity was not discussed until 1991 in C6te d'Ivoire, 1988 in Malawi, 1987 in Senegal, 1994 in Pakistan, and 1987 in the Philippines. (For a discussion of equity and tax reform, see Chapter 3.) 2.26 Tax administration strategies were identified in 21 countries, but in 5 (Costa Rica, Honduras, Hungary, Pakistan, Philippines, and Tunisia) the justification for the strategy was limited. In general, tax administration reforms were more often prompted by the failure of pure tax policy reforms to deliver increases in tax revenues, than justified as being integral to a revenue enhancement strategy. In LAC the rationale was reduction of collection lags; in AFR the reforms were wide-ranging and induced as much by revenue as by structural and institutional considerations. The quality of the strategy discussions tended to be more substantial in later loans (for example, the need for a taxpayer identification number system was identified in Malawi as early as 1981, but discussion of the measure was found in the 1988 and subsequent operations). (See Chapter 4.) 2.27 Nontax revenues were identified in all countries as a potential source of public funds. Such revenues were expected from adjustments in the prices of publicly produced goods, cost recovery in health and education services, and tariff increases for public utilities. Cost recovery for health and education services was stressed more in LAC than in AFR. (See Chapter 5.) 29 Fiscal Analysis and Projections 2.28 Fiscal reform has to be evaluated as part of a larger adjustment program of interrelated policies. PRs contained very limited discussion of the fiscal implications of reforms involving the exchange rate, the interest rate, and trade policy. Exceptions were some qualitative discussions of the fiscal effects of exchange rate changes in Ghana, Mozambique, Tanzania, and Uganda;18 of interest rate reform in El Salvador, Pakistan, and Tunisia; and of trade liberalization in Bangladesh, Morocco, and the Philippines. 2.29 Projections describe a medium-term perspective, alternative scenarios, and benchmarks for monitoring performance. Yet only 22 percent of the loan sample included projections of revenue and expenditure components, and only 59 percent provided projections of total revenues and expenditures; the rest had no revenue or expenditure projections. The extent of coverage of component categories of revenue and expenditure varied widely: none of the Bangladesh loans included projections, while about 53 percent of loans in LAC did. Though subsidies and transfers were an important preoccupation (see para. 2.17), none of the PRs tackled the issue of subsidies as a whole. Whether because of the lack of political will or because the effects of implicit subsidies are not always explicitly apparent as revenue or budget consequences, there was little attempt to work through the macroeconomic implications of making implicit subsidies (and taxes) transparent. Was the Quality of Conditionality Adequate? 2.30 While it is important to evaluate the success of fiscal management in the context of the broader adjustment program, not merely with respect to the performance of fiscal instruments but also with respect to its results (in reducing inflation, encouraging private sector activities, ensuring fiscal sustainability, etc.), this study limits its focus to fiscal conditionality per se. Chapters 3, 4, and 5 discuss whether the specific types of tax and expenditure conditions incorporated into SALs/ SECALs were consistent with Bank guidelines or best practice. Here, sample SALs/SECALs are examined to determine whether conditionality was consistent with diagnosis and to assess the patterns that emerge on the quality of the conditionality, targets, and projections (the issues investigated are set out in Box 2.3.) 19 The study concludes that while consistency with diagnosis was preserved, the quality of Bank conditionality was often poor. 18 In the CFA franc countries, C6te d'Ivoire and Senegal, PRs discussed the use of export subsidies to compensate for an overvalued exchange rate. 19 Judgments on discernible patterns in the quality of conditionality are drawn from PCRs and PARs. Conditionality refers to the policy actions set out in matrix form in every PR on adjustment operations. 30 Box 2.3. Evaluating Conditionality 1. Is conditionality consistent with diagnosis and strategy? 2. Is a budget deficit condition specified? Does conditionality address the deficit appropriate to the country's circumstances: primary, structural, or operational? 3. What patterns emerge in PCRs and PARs with respect to the quality of conditionality? * Are targets specified? * Did conditions stress or aid implementation? Did they stress actions, not studies? Did they take account of the electoral cycle? Did they focus on the enactment of laws, not merely their preparation? * Are budget deficit conditions too vague? * Are there too many conditions? 4. Do conditions address data problems? Was Conditionality Consistent? 2.31 The diagnosis and analysis of fiscal problems and the selection of conditions to deal with them were generally consistent across regions at the level of each operation. Once a problem and a fiscal instrument for dealing with it were identified at the diagnosis stage, conditionality was formulated for the instrument to deal with the problem. Tax administration was the one area where conditions were not always consistent with diagnosis. Not all loans that had extensive diagnostic and strategy discussions followed up with conditionality (as in the Gambia, Malawi, Mozambique, or the Philippines). Conditions were also often vague or replaced by technical assistance proposals (as in Costa Rica and Venezuela). (See Chapter 4.) Conditions on nontax revenues were generally more numerous than the discussions on diagnosis and strategy would have suggested. Were Budget Deficits Specified? 2.32 In 47 percent of the total sample, the coverage (level of government) of the deficit was not specified (see para. 2.5). In an even larger proportion (54 percent) of the sample there was no conditionality relating to the budget deficit (Figure 2.2). However, there was great variability across regions (see Annex Table 2.2). In another 20 percent of the loans the need to achieve a consistent macroeconomic framework was merely alluded to. These apparent omissions may have been due in part to the Fund's role in specifying the acceptable deficit level and the macroeconomic framework in general, though the PRs contained no clear record of the Bank explicitly assessing the Fund program before accepting its main outlines. In only 33 operations, or 26 percent of the sample, was a specific target proposed for the budget deficit (or the level of public savings). Moreover, the measure of the deficit varied between operations for the same country-often without adequate explanation (examples include Pakistan, the Philippines, Morocco, and Uruguay). 31 Figure 2.2. Proportion of SALsISECALs with Budget Deficit Conditions Public Overall deficit savings 18% 6% Primary deficit Consistent 2% macro- economic framework 20% No condition 54% Source: President's Reports for SALs/SECALs. 2.33 While SALs should emphasize the overall balance of the government entity under scrutiny, certain circumstances require narrower budget conditionality. Yet apart from eight loans that incorporated conditions involving public savings and two where conditionality involved the primary deficit, the loans dealt exclusively with the overall nominal deficit (see para. 2.7). Were Quantitative Fiscal Targets Specified? 2.34 Quantitative targets show the effort needed to keep programs on track, signal that programs are going off track, and indicate the magnitude and direction of any mid-course corrections required. On average, in all regions covered by the study sample, there was less than one quantitative target per operation on either the budget deficit (including public savings) or on 10 revenue and expenditure areas (Table 2.1). MNA was the only exception, with almost two revenue or expenditure targets per loan. Though not all reform conditions can be couched in quantitative terms (conditions relating to institutional reform, for example, may be precise but not quantifiable), the paucity of quantitative targets in sample SALs/SECALs suggests room for improvement. 32 Table 2.1. Quantitative Fiscal Targets in SALs/SECALs Number oftargets relating to 10 revenue/expenditure Budget deficit or Total operations per Region/country areas" public savings region LAC 27 12 28 AFR 29 8 54 ECA 6 8 16 (Bulgaria, Hungary, Poland, Turkey) EAP 4 0 10 (Indonesia, Philippines) MNA 16 2 9 (Morocco, Tunisia) SAS 11 3 12 (Bangladesh, Pakistan) Total sample 93 33 129 a Total expenditure, total revenue, tax revenue, nontax revenue, tax administration, current expenditures, employment/wages and salaries, subsidies/transfers, O&M expenditures, and capital expenditures. Source: President's Reports for SALs/SECALs. Other Aspects of the Quality of Conditionality 2.35 PCRs and PARs on 97 operations noted other problems that continue to mar the quality of fiscal conditionality, though they have been extensively discussed before in other Bank reviews: * Budget conditionality was too general (conditions commonly called for "a macroeconomic framework consistent with the objectives of the program") or inconsistent across loans for the same country, without adequate rationale. * Conditions were judged by PCRs/PARs to be too numerous (12 cases) or so vague as to be difficult to implement or monitor in as many as 25 operations. * Conditions were "soft" in many cases, calling for the drafting of legislation rather than implementation (in 14 cases) and stressing studies rather than actions (13 cases). * Conditions paid too little attention to timing (7 cases where, for example, the implications of the electoral cycle for the adoption of reform were often ignored). PCRs and PARs indicate that attention to design and timing issues would have helped implementation in about 56 loans (52 percent of the operations with completed evaluations). (See Annex Table 2.3 on these aspects of conditionality.) These flaws are not confined to the early adjustment loans. Of the 19 loans processed after 1990, PCRs/PARs pointed out that 16 percent focused on studies rather than actions; 47 percent called for the presentation of laws to 33 law-making bodies rather than their implementation; 37 percent had too many conditions; 32 percent had conditionality that were mistimed; and 68 percent had conditionality that was characterized as "vague" and difficult to monitor. Data Quality 2.36 Projections, scenarios, and targets provide transparency to the reform process and bolster borrower ownership. Yet conditionality did not address improvements in data needed for such analyses. The sample provided several examples of data gaps. Lack of data hampered budget formulation and mid-year budget reviews in Bangladesh and calculation of the consolidated budget deficit and the public sector borrowing requirement in Turkey. On the expenditure front, subsidies that are to be reduced often reappear in the guise of net lending, a category for which data are not often available. Expenditure data in education and health are unavailable by functional classification, and the data that are available are not disaggregated by primary, secondary, or tertiary services. Morocco (1992) aimed at redirecting spending to primary education, but these expenditures were not differentiated in the budget from expenditures on the first stage of secondary education. In AFR and SAS a key recommendation to increase expenditures for O&M could not be monitored because such data were generally not available as a line item classification in budgets (other examples, see Gupta 1996). Was Implementation Adequate? 2.37 The implementation record of detailed tax and expenditure conditions is examined in Chapters 3, 4, and 5. Here, patterns of implementation experience are examined across regions in the broad areas of fiscal reform touched on in previous sections. The implementation record was tracked by examining PCRs/ICRs/PARs for 97 operations; 31 operations had not been evaluated at the time of this study; 1 operation (in Costa Rica) was canceled. These documents reveal that borrower performance in implementing adjustment conditionality has been mixed: implementation has been weak in budgetary process and public enterprise reform and in the reduction or restructuring of current expenditures. Tardiness in implementation was also reflected in tranche delays. Tranche Delays 2.38 Tranche delays were reported in 45 loans. In 38 of them (about 85 percent), the delay was attributed wholly or in part to the nonfulfillment of fiscal conditionality. Budget Deficit/Public Savings 2.39 The budget deficit/public savings conditions were considered fulfilled in 23 of the 33 loans with such conditionality. Central bank losses had been explicitly identified as a factor in fiscal deficits in the diagnosis of fiscal imbalance in 9 countries.20 While they are no longer a problem in Argentina and Chile, they continue to be a concern in Costa Rica, Mozambique, the 20 In some countries, central bank losses were identified in later loans but ignored in earlier loans. Presumptive evidence was available on the existence of central bank losses in the PRs for another 8 countries. 34 Philippines, and Tanzania.21 Social security reform was implemented in Chile, but not in Bulgaria, Hungary, and Poland in ECA or in Costa Rica and Uruguay in LAC. Budgetary Process Reform 2.40 The record on budgetary process reform is poor. Implementation of loan conditions was still in progress in a number of countries that emphasized such reform, such as Argentina, El Salvador, Bangladesh, and Morocco. In Kenya several loans addressed budget rationalization, but implementation has been weak. In Costa Rica reforms have yet to be implemented. In Ghana, Tanzania, and Uganda, budget transparency has been achieved in some areas (for example, by reducing or eliminating the subsidies inherent in overvalued exchange rates), but weak budgetary processes remain as one of the more important challenges facing these countries. Public Enterprise Reform 2.41 Implementation has fallen short on public enterprise reform as well. PCRs reported satisfactory implementation in only 5 countries (Argentina, Chile, Peru, the Gambia, Philippines) and weak implementation in 20 countries (see Gupta 1996 for details). In Mozambique, Senegal, and Tanzania, indirect or implicit subsidies have continued. In Morocco transfers to enterprises were lower than in 1980 but remained significant. In Mozambique the government agreed to stop using bank loan guarantees to cover the operating losses of state enterprises, but the accounts of the Bank of Mozambique and Banco Commercial de Mozambique revealed substantial additional credit to state enterprises during 1987-92. Public Investment Programs 2.42 Though diagnosis and strategy discussions in PRs emphasized changes in the composition of public investments, most PCRs and PARs have not assessed progress in this area. (Chapter 5 attempts an assessment of changes in investment composition.) The Bank has been successful, however, in assisting countries to evaluate projects and formulate rolling investment plans. Few countries that had planned to refine estimates of O&M requirements by evaluating capital budgets have yet done so. Current Expenditure 2.43 Conditions calling for reductions of subsidies or transfers have been at least partially implemented in a range of countries: C6te d'Ivoire, Kenya, Malawi, and Mozambique in AFR; Bulgaria, Hungary, Poland, and Turkey in ECA; Costa Rica in LAC; and Bangladesh and Pakistan in SAS. Overall results, however, fell short of expectations (see Chapter 5), and success in implementing wage and salary conditionality has been mixed. The Gambia and Ghana reduced their wage bills as a ratio of GDP and improved the wage and salary structure. Some countries that had achieved significant employment reductions were rehiring (Kenya, Senegal, Uruguay). Where initial wage compression had been excessive, wage increases or decompression of salary scales were implemented (Argentina before 1991, Chile, El Salvador, 21 PCRs were silent about central bank losses in Honduras and Uruguay. The Ghana (1991) loan is still unrated. 35 Indonesia, Mozambique, Tanzania, and Uganda). Further curtailment of wages and salaries is needed in Bangladesh, Morocco, and Pakistan Tax Revenues 2.44 Implementation has generally been slower for direct tax reform than for indirect taxes and is continuing in a large number of countries (see Chapter 3, para. 3.45). Implementation of tax administration reforms was not addressed in adequate detail in PCRs/ICRs (see Chapter 4). Other Issues 2.45 Gainers and Losers. Identifying likely gainers and losers and strategies to deal with the losers might have speeded implementation. Such analysis could have highlighted problems with the Argentine land tax (1986) and the Uruguay social security reform (1992). In Malawi analysis of the impact of reductions in the fertilizer subsidy on smallholder productivity might have prevented delays in implementation. In Hungary the reluctance to accelerate bankruptcy proceedings for state enterprises stemmed from the failure to analyze the impact on losers or to build in measures to mitigate adverse impacts (see Chapter 3 on trade reform). 2.46 Bank-Fund Collaboration. Bank-Fund collaboration has been recognized in both institutions as essential for ensuring consistency in economic advice: guidelines for collaboration were formally endorsed by the heads of the two institutions in March 1989. However, the evaluation of Bank adjustment programs has been hampered by the lack of systematic reporting in Bank PRs on the principal Fund actions required. While coordination on stabilization has generally been good, the Bank has had some problems in synchronizing stabilization and structural adjustment. In Ghana, for example, the Bank failed to consider that the fiscal contraction effected under the IMF program would slow the investment program to be financed from the local currency proceeds of the Bank's Rehabilitation Import Credit (1983). In Malawi the Bank failed to realize that the country's agreements with the Fund to keep public sector wages frozen during 1982-86 would prevent the government from retaining key personnel by granting them compensatory wage increases when prices were deregulated under SAL III (1985). (For examples pertaining to other regions, see Gupta 1996.) 2.47 Coordination with Other Donors. Donor coordination has been particularly fruitful in some fiscal areas, notably in devising social safety nets. The Gambian civil service retrenchment program was particularly successful because donors helped set up counseling and retraining programs. Donors also financed the cost of staff retrenchment for the Ghana cocoa marketing agency. In Costa Rica USAID assistance was key to the successful privatization of public enterprises owned by the government agency CODESA. The unity of purpose among donors helped tighten reform objectives and promote the rapid adoption of financial sector reforms under Senegal's Financial Sector Adjustment Loan (1989). 2.48 Lack of coordination among donors can delay structural reform. Examples abound. In LAC, AFR, and Bangladesh the lack of coordination with donor groups worked against public expenditure control. A study on project aid utilization in Bangladesh found that donor collaboration problems contributed to cost overruns of 35-40 percent and prolonged implementation by 60 percent relative to estimates in project appraisals. Though the Ugandan government had agreed under SAC 11 (1994) to adopt the basic principles for distinguishing 36 between core and noncore projects, problems with getting other donors on board stood in the way of putting the agreement into practice. The PAR for the Kenya industrial SECAL (1988) stressed that donors working in isolation tended to send the wrong signals and to thwart government efforts at expenditure control. The PR for a subsequent agricultural SECAL recognized that the availability of numerous sources of external financing tended to soften the application of rigorous project selection criteria, while the excessive number of projects taxed the government's implementation capacity. In Bangladesh the large and unconditional budgetary contributions of donors may have hampered efforts to mobilize and rationally allocate public resources in the 1980s. 2.49 As donors became more aware of the need to support social expenditures and shifted from providing general balance of payments support to earmarking funding for social programs, budgetary control and expenditure management became more complicated. The use of special accounts by donors in El Salvador contributed to expenditure control problems. In Mozambique aid earmarked for poverty reduction could not be incorporated into the general budget, which jeopardized the efforts of the coordination groups set up to align government expenditure priorities and those of the donor community. Conclusions 2.50 Analysis of the 129 loan sample suggests that diagnosis, analysis and strategy, and conditionality were generally consistent across all regions: areas diagnosed in PRs as responsible for fiscal problems were generally also identified in the strategy discussions as eligible for follow-through with reform conditions. However, the quality of diagnosis and conditionality warrants improvement. Implementation was also weak in key areas. More specifically, * The coverage and specific measure of the fiscal deficit were not specified in about half the operations. Not surprisingly, therefore, conditionality on budget deficits was meager. Despite the diversity in fiscal experience across countries, almost 39 percent of operations that did specify a budget deficit condition confined their attention to the overall deficit. Moreover, the measure of the deficit varied between operations for the same country (often without adequate explanation), the same deficit terminology was applied to different measures of the deficit across countries, and none of the loans addressed the level of the sustainable deficit. * Not all loans worked systematically through the various elements of diagnosis that needed to be addressed. For example, loans that focused on revenues/tax reform did not always appraise the role of the public sector or explore possibilities for expenditure reduction before diagnosing the need for revenue reform. When strategies called for increasing revenues, PRs did not establish whether past performance provided assurance that public savings would rise when public revenue did. Nor did PRs discuss whether public savings add to or substitute for private savings. Even if such issues are hard to establish empirically for specific countries, discussions of this nature focus borrower attention on the need for public savings. Further, because diagnosis and strategy typically failed to identify those who stand to lose from adjustment operations, the loans included no mitigating measures that might have improved the chance of success. 37 * None of the PRs analyzed the relationships between fiscal and other macroeconomic variables with respect to a sustainable deficit or discussed the macroeconomic implications of making implicit taxes or subsidies transparent. Few considered the fiscal impacts of other macroeconomic policy reforms, and not all operations that needed to do so worked out the revenue effects of alternative fiscal reforms. One reason for these analytical gaps is the poor quality or outright lack of country fiscal data. Lack of data also hampered monitoring and evaluation. Yet conditionality did not stress the need to address data gaps. * Though budgetary processes were diagnosed, none of the PRs took an integrated view of the framework of governance in which the public expenditure system is embedded (Pradhan and Campos 1996). For example, none dealt with the incentives or organizational arrangements needed to ensure overall budgetary discipline, the involvement of local institutions and beneficiaries in formulating public programs, technical efficiency in drawing up alternative expenditure proposals and determining the choice of programs, incentive systems for fostering an efficient bureaucracy, or information systems for building in transparency and accountability. The failure to deal with budget transparency, monitoring and control, database systems, or the management of extrabudgetary spending was particularly evident in AFR, where the discussion was confined mainly to strengthening public ministries and investment programming. In general, the sample PRs suggest that the Bank needs to build up more expertise in budgetary systems. * Public enterprise reform was a focus of diagnosis, analysis, and conditionality in almost all countries, but implementation was incomplete or weak in about 25 (of 27) countries. * The level and allocation of expenditures within and between capital and current expenditures were identified as factors behind fiscal imbalances in almost all loans. While PRs focused on project evaluation and rolling investment plans, the restructuring of public investments received less attention. Particularly lacking were discussions of social sector investments. This gap carried over into analyses of current expenditure. The importance of holding down or reducing subsidies and the wage bill if expenditures on nonwage O&M were to be increased was discussed but not achieved. * PCRs/PARs on 97 of the 129 operations sampled noted a set of problems related to conditionality that had already been flagged in other Bank reviews. Conditionality on the deficit or public savings was often too general or inconsistent between loans for the same country. Other fiscal conditionality was judged to be so vague as to be difficult to implement. Conditions were also soft, calling for preparation of legislation rather than implementation or for studies rather than actions. Conditionality was occasionally poorly timed, calling for the most important actions to be taken at a time when election pressures were likely to affect the acceptability of reform. * While coordination with other donors was sought, the analysis suggests that the Bank needs to collaborate more effectively with the Fund in synchronizing stabilization and structural adjustment in the fiscal area. 38 * Hitches in implementation have been overwhelmingly responsible for tranche delays. PCRs/PARs have drawn attention to implementation problems in budgetary processes, public enterprise reform, the level and composition of public investments, employment and the public sector wage bill, and social security reform. Improvements in the design and timing of conditionality would have helped implementation in as many as half the operations with completed evaluations. These shortfalls also characterize operations approved as recently as 1990-94. Lessons 2.51 In addition to the specific sectoral and functional areas identified above, for which attention to design, conditionality, and implementation and the links among them could improve the quality of loans, there are some relatively simple lessons that may apply more generally across all fiscal programs in adjustment loans, as well as some broader issues that the institution needs to debate. 2.52 Specific issues include: * SALs and SECALs need to work through the fiscal accounts at the broadest institutional level before zeroing in on the specific account that is primarily responsible for the fiscal deficit. This implies more detailed attention to aspects of the deficit that are not immediately apparent: losses of the central bank and other state banks, arrears of public sector entities, implicit subsidies, the fiscal implications of other macroeconomic policy changes. While the overall deficit should be the starting point, the underlying nature of the deficit-whether primary, structural, or operational-merits attention. Moreover, all adjustment loans need to identify the level of the sustainable deficit that is consistent with the overall macroeconomic framework and to specify quantitative targets over time to achieve it. * The design and timing of conditionality and the success of implementation might be enhanced by making conditions more specific (using targeting where possible), limiting the number of fiscal conditions to help focus government attention and scarce technical skills, emphasizing actions rather than studies, identifying gainers and losers and compensating losers, and taking better account of the country's electoral timetable. * The need to work on improving donor coordination is suggested by concerns about synchronization of stabilization and structural adjustment reforms, the pace of adjustment, overall fiscal and expenditure discipline, a focus on strategic issues, performance monitoring indicators, and stricter implementation. 2.53 Four general issues need attention: * While the lead role of the Fund in offering fiscal policy advice to members should be respected and its long record of expertise and involvement in fiscal matters recognized, the Bank needs to take a more hands-on approach to the design of fiscal reforms in adjustment programs. The Bank needs to do more and better analysis to 39 supplement the Fund's perspectives with its own views on the fiscal measures necessary for long-term efficiency, growth, and poverty reduction. This does not mean that the Bank should offer different policy advice to members. Rather, what it suggests is the benefits of effective coordination on fiscal issues between the two institutions. As the 1989 Concordat puts it, both "the institutions and borrowing members [would] normally stand to benefit from analyses from different perspectives." The Bank could thus assist the Fund in presenting clients with a broader range of policy options by expanding the fiscal debate. The modalities for greater coordination between the two institutions on macroeconomic and fiscal matters (whether through the Policy Framework Paper, public expenditure reviews, the Bank's Country Strategy Paper, the Fund's Country Strategy Brief, or some other means) would need to be worked out. * The Bank needs to assist borrowers in determining the appropriate role of government relative to the private sector (a critical determinant of the level of the public finances), and the specific mix of public sector roles that suits a country's situation: Should the government continue to produce the same goods and services it currently does? Should it assist the private sector with financing? In which areas should it withdraw to a regulatory role? * The overall framework for public expenditures needs to be the focus of reform, not ad hoc measures. The Bank needs to look systematically at how to maintain overall budget envelopes, encourage local beneficiaries and governments to participate in formulating expenditure proposals, ensure technical efficiency in the choice of expenditure programs, build in incentives to ensure an efficient civil service, and ensure transparency and accountability to keep governance open and honest. The Bank's Development Economics Vice Presidency has recently started looking at these issues. More needs to be done. * Given the central importance of fiscal reforms in the adjustment process, assisting governments to improve their data systems (for identifying the components of the fiscal accounts and the fiscal implications of policy changes in other areas) may be the simplest and yet the single most helpful fiscal reform that the Bank could support. Focusing on the quality of fiscal data would enable countries and the Bank to detect crises early and to present alternative projections for guiding strategy, contingency planning, implementation, and monitoring. Better data and information dissemination are also critical for government transparency and accountability. This may be an area for stronger Bank-Fund collaboration. The Fund has recognized that despite its greater strength in this area, its published fiscal data have shortcomings; it has taken several initiatives over the past year to encourage its members to provide it with comprehensive, timely, and high-quality information.22 22 See IMF, IMF Survey, November 6, 1995. 41 3. Tax Reform 3.1 This chapter examines tax policy reforms supported by Bank adjustment operations. It briefly describes the problems that beset tax systems in developing countries in the early 1980s; summarizes the principles of tax reform promoted by the Bank and describes how the Bank sought to translate these principles into specific tax instruments; categorizes the conditionalities adopted in adjustment operations by content and regional distribution; and assesses the design, implementation, and efficacy of tax reform. What the Study Found Nontrade Tax Conditionality 3.2 The Bank's policy advice on nontrade tax reform has generally been in line with the recommendations in Lessons for Tax Reform (World Bank, 1991 a) (referred to as Lessons). Over half of the Bank's tax conditionalities aim to achieve broader direct and indirect tax bases and lower marginal income tax rates to make tax systems simpler, more efficient, and easier to administer. In general, however, tax conditionalities have represented a small component of SALs and have been selective rather than comprehensive in scope. Typically, the Bank's conditionality has supported the government's own reform initiatives or has piggybacked onto other institutional initiatives. 3.3 The Bank's design of tax conditionality has often been loosely structured: nearly half of all conditions specify only the direction of modifications to be made to tax statutes, not their extent. Only 40 percent of conditions had some quantitative element; 12 percent of all conditions had neither qualitative nor quantitative criteria attached. 3.4 More than half of a sample of SALs covering 31 countries had been successful in implementing all of the Bank's tax conditionality and about a third had been partially successful. Slightly less than 15 percent of the countries surveyed had failed to implement any of the conditionalities. However, given the imprecision of conditionality this relatively high level of compliance is difficult to assess. Furthermore, PCRs/PARs evaluated tax reforms only in terms of statutory compliance. None attempted to evaluate tax reforms in terms of marginal effective tax rates; very few checked on the compliance record. However, shifts in tax structure did take place in the direction of less reliance on trade taxes and greater reliance on indirect taxes. Although the Bank's conditions did not stipulate rate increases, shifts in tax structures were accompanied by increases in tax to GDP ratios in a large number of sample countries. Trade Tax Conditionality 3.5 Perhaps because the Bank's trade tax conditionality has been extremely specific and nonambiguous, compliance has been exceptionally good; for the sample of countries examined, the Bank has made substantial headway in reducing export barriers and in inducing countries to This chapter is based on a background paper by Watson and Thirsk (1995). Both take as their starting point the discussion of the microeconomic aspects of fiscal reforms in World Bank Structural and Sectoral Adjustment Operations: The Second OED Overview (Washington, D.C.: World Bank, 1992c). The ideas contained in this 1992 OED volume are amplified in this chapter and validated against a larger sample of Bank SALs/SECALs. 42 replace nontariff trade barriers with tariffs and to reduce both the level and the variability of effective protection. The pace of reform, which has varied from country to country, has been determined by the extent to which trade tax reforms have been synchronized with nontrade tax reforms. 3.6 Successful trade tax adjusters have for the most part also been successful fiscal adjusters; countries that had reformed their tax systems either before or during their period of trade reforms and had instituted a broadly based sales tax were able to implement large tariff reductions and compensate for any tariff revenue loss. Unsuccessful trade tax adjusters, by contrast, were unable to cope with the revenue losses accompanying lower trade taxes and so continued to rely on tariffs as a major source of revenue; they either did not undergo significant tax reform or introduced reforms too late to assist trade liberalization. Insufficient progress in stabilization also constrained trade tax reforms. 3.7 Very little effort was made to quantify the revenue implications of trade reforms, partly because the Bank deferred to the Fund on issues of revenue adequacy and tax reform. The response to revenue shortfalls has been higher trade taxes, something the Bank's conditionalities have sought to avoid. 3.8 The removal of quantitative restrictions (QRs) and their replacement by tariffs assumes that tariffs will be applied properly. Implementation of trade tax reforms therefore requires effective customs operations, something the Bank's trade tax conditionality may have neglected. Principles of Tax Reform 3.9 In the early 1980s when the Bank embarked on adjustment lending, tax systems in developing countries shared certain common characteristics (World Bank 1991 a): * Tax revenue to GNP ratios were low (18 percent, compared with 30 percent in industrial countries) (see Annex Table 3.1). * Reliance on commodity taxes, particularly trade taxes, was high: low-income countries collected almost three-fourths of their tax revenue from commodity taxes, more than half from trade taxes alone (see Annex Table 3.2). * Income tax systems were highly progressive and functioned poorly, with effective tax rates that might be characterized as proportional at best (Thirsk 1991). * Personal income taxes represented a small share of total income tax revenue, most of which came from business taxes. * Both direct and indirect tax systems had highly differentiated rate structures, narrow tax bases, and extensive exemptions and incentives. A narrow tax basis with highly differentiated tax rates tended to result in high and variable effective tax rates and low collection rates. * Compliance problems were serious. 43 3.10 It is not surprising that by 1980 many of the developing countries that faced fiscal crises were relying heavily on unsustainable levels of debt or inflation to finance their deficits, experiencing significant losses from tax-induced resource misallocation, and suffering an erosion of taxpayer confidence. It is against this backdrop that the Bank began to promote structural adjustment programs that incorporated tax conditionality. 3.11 The principles guiding the Bank's tax reform efforts are clearly set out in Lessons, and details on the characteristics of specific tax systems are described in World Development Report 1988. These two volumes Frovide the benchmarks against which the tax reforms of the Bank's SALs/SECALs are judged. 3.12 Lessons sets out the following tax principles: * The primary function of the tax system is to generate revenue. Where tax increases are not an explicit objective, reform efforts usually seek revenue neutrality, that is, to improve the structure of taxes (by shifting the tax system away from relatively inefficient revenue sources, such as trade taxes) without changing revenues. * A major goal of tax reform is to reduce the efficiency costs of the tax system. Compliance costs to the taxpayer and administrative costs to the public sector of enforcing tax laws are two of the most transparent of these costs. The cost associated with resource misallocation-excess burden-is the least transparent. To lessen excess burden, taxation of production should be avoided in favor of consumption-based taxation; if production is taxed, all activities of a similar nature should face the same effective tax rate. This principle reflects the neutrality objective of tax reform in the efficiency context. * Lessons and many recent tax reforms espouse more broadly based taxation at lower and more uniform rates with fewer incentives on the grounds that such rules simplify the determination of the tax base and tax liability, increase efficiency, offer fewer incentives to evasion, make evasion more difficult, strengthen the ability of the tax administration to collect taxes, and make the tax system politically more acceptable.4 * A fourth goal of tax reform is to lift the tax burden off the poorest households and to ensure that tax structures promote both vertical equity (distributing the tax burden according to ability to pay) and horizontal equity (treating taxpayers with the same amount of income equally, irrespective of the source of income). Broadly based tax systems promote vertical equity by carefully targeting exemptions to the poor, by 2 See Annex 3.3 on choosing instruments in line with these principles. Such misallocation occurs when taxation induces taxpayers to make substitutions along a number of margins, including consumption, savings, and investment decisions: saving-consumption tradeoffs, labor-leisure tradeoffs, and portfolio decisions. The simplicity principle seems to fly in the face of the optimal tax literature, which suggests that the tax structure that minimizes the excess burden of taxation (and also respects society's vertical equity norms) is likely to be based on a nonuniform pattern of tax rates. However, establishing such an optimal tax framework is difficult because of daunting informational requirements, the difficulties of identifying society's social welfare function, the likelihood that differentiated tax rates will encourage inefficient rent seeking activities and will be less politically acceptable than uniform rates, and the fact that tax administrations may not be perfect. 44 relying on tax bases that apply largely to the rich (higher excise taxes on luxury items), and by generating enough revenue to prevent expenditure cuts or increases in inflation, which harm the poor. They promote horizontal equity by extending the tax net to cover the informal sector and hard-to-tax formal activities. Lessons and recent tax reforms recognize, however, that tax systems can avoid taxing the poor but cannot increase their incomes. Public expenditure policies are a better instrument for income redistribution. Precision and Content of Tax Conditionality 3.13 What was the quality of conditionality in the Bank's SALs/SECALs? For the sake of analytical convenience, conditionalities involving nontrade taxes are discussed separately from trade taxes. Nontrade tax conditionalities are analyzed by looking at 92 SALs/SECALs approved by the Bank's Board for 53 countries between 1980 and 1993; for trade tax conditionalities, the study looks at a sample of 87 SALs/SECALs involving 23 countries.5 Nontrade Tax Conditionality 3.14 Although the geographical spread of nontrade tax conditionality was wide, the range of such conditionality in most countries was limited. Of the 53 countries sampled, 49 percent had a single loan involving tax conditionality and 36 percent had two loans. For most of these countries conditionality was limited in both the number and the scope of reforms. Typically, countries faced two or three conditions on changing a few tax code parameters to improve compliance. In about 30 percent of countries more significant reform of a particular part of the tax system was stipulated (usually the creation of a value added tax-VAT); in only a few countries, including Burundi, Costa Rica, Hungary, Morocco, and Thailand, did conditionality involve major changes in different parts of the tax system. Eight countries had three or more SALs/SECALs incorporating tax conditionality (Bangladesh, Ghana, the Philippines, Tunisia, and Turkey had three loans each; Jamaica and Malawi four loans each; and C6te d'Ivoire five loans). For C6te d'Ivoire, Ghana, and Malawi the scope of tax conditionality was broad, requiring reforms across several parts of the tax system as well as the rehabilitation of investment codes. Elsewhere, as in the Philippines and Jamaica, tax conditionality was more limited. 3.15 Specificity. How explicit were the Bank's conditionalities on nontrade taxes? About 40 percent of conditionality included quantitative criteria or stipulated adoption of particular tax code provisions. This type of conditionality typically involves alterations in tax base or rate structures, which are easily monitored. For instance, a country may be required to switch from specific to ad valorem excise taxes, to lower its sales tax rate to a certain level, or to remove certain exemptions from an income tax base. 3.16 The most common form of conditionality, accounting for about 48 percent of conditions, stipulated specific statutory changes but did not provide any quantitative measure of performance. These conditions usually specified the direction of change in the tax code- Of the total of about 250 adjustment operations (in 86 countries) that incorporated fiscal reforms during FY1979-94, trade tax conditionality was included in 171 operations in 75 countries. The sample therefore represents 50 percent of all operations and 31 percent of all countries involved in trade tax reform between 1980 and 1993. 45 reducing exemptions and marginal income tax rates and expanding indirect tax bases-without mentioning the magnitude of change. Because these conditions were not quantified, statutory compliance could be met through minor alterations in the tax code. 3.17 Some 12 percent of conditions examined lacked any explicit performance criteria, qualitative or quantitative, making it difficult to evaluate tax reform efforts. An example of this form of conditionality is "adoption by government of final draft of the new ... investment code" (Gabon). 3.18 Content. Policy advice on nontrade taxation has been consistent over time and across countries and has reflected "best practice" (Table 3.1). (See Annex Table 3.4 for background data and definitions of conditionality covering all regions. See Watson and Thirsk 1995 for country-specific conditionalities in individual nontrade tax areas.) About 40 percent of conditionality dealt with rationalizing tax bases (base broadening) or rates, indicating the importance the Bank places on simplifying tax codes. About one-third of conditionality involved reform of the indirect tax system, with several reforms focusing on reorienting indirect taxation toward the VAT, a tax that reduces the scope for inefficiency and evasion. These conditions were appropriate, given the serious compliance problems facing most developing countries in the early 1980s and their heavy reliance on inefficient indirect taxation, especially trade taxation. Rationalizing business incentives and harmonizing rates on individual and business income taxes and financial transactions has also received a great deal of attention. Hoping to promote domestic production or foreign investment, many developing countries in the 1980s cobbled together a welter of poorly designed or misguided investment incentives that generated few benefits and lost revenue. These incentives were a natural target for reform. As capital markets developed in these countries during the 1980s, investors became more sensitive to differences in income tax rates-individual and corporate, within and across countries. Thus a relatively large number of conditions involved rate harmonization. 3.19 Since tax reform is futile if evasion is high, improved compliance was also a significant focus of reform in indirect taxes and individual and business income taxes. The focus on corporate compliance probably reflects the increasing importance of business taxes in the direct tax systems of most developing countries and the fact that reducing evasion of business income taxes is easier than reducing evasion of individual income taxes. (Conditionality involving compliance is discussed in detail in Chapter 4.) 3.20 About 5 percent of conditionality was aimed at eliminating nuisance taxes (stamp taxes, taxes on financial intermediaries) that fall outside the income tax system. These taxes are easy to enforce, but they distort capital market activities. 3.21 Another 5 percent of conditionality was devoted to reforming property and wealth taxes, which take a lot of preparatory work to develop. Property and wealth are often undertaxed or not taxed at all in developing countries, and neglect of this area has caused tax bases to evaporate in some countries. This area deserves more attention because some property taxes (such as taxes on land) can be very efficient and permit significant redistribution of income. 6" Collectibility," or the ease of revenue collection has also been a factor in the choice of tax instruments. Here such cultural factors as literacy rates and public attitudes toward taxes also play a role. See Annex 4.1. 46 3.22 Examination of a sample of detailed conditions relating to specific nontrade tax instruments confirms that these conditions were consistent with the reform guidelines set forth in Lessons (see Watson and Thirsk 1995 for details on the content of conditionality described in paras. 3.18-3.21). 3.23 Because of the desirability of shifting taxation away from production toward consumption and of limiting tax evasion, VAT conditions have been relatively popular. These conditions cover broadening the tax base, lowering rates, or moving toward full crediting of inputs. Most countries adopt VATs to replace sales or excise taxes, but administrative difficulties usually require the introduction of a modified VAT that does not cover all stages of production. Conditionality ranges from specific, easily monitorable conditions, such as the elimination of exemptions, to broader requirements, such as the adoption of a VAT at the manufacturing level. 3.24 Most individual income tax conditions aim to reduce the efficiency losses from high rates of taxation and the unequal tax treatment of similar forms of income. In most cases the magnitude of statutory reform is left up to the country, presumably in consultation with the Bank (or the IMF). 3.25 Business tax conditionality emphasizes reductions in tax rates, clarification of deductions, and broadening of tax bases. In some inflationary economies conditionality includes revaluation of balance sheet items. In many cases conditionality aims at harmonizing rates with those of capital-exporting countries or harmonizing business and individual income taxes to eliminate tax arbitrage opportunities or double taxation of dividends. Conditions on investment incentives usually involve adopting a new investment code or revising an old one that distorts investment incentives. In most cases, however, the conditionality is vague. In no case does a PR or Letter of Development Policy refer to an underlying analysis of the costs and benefits of the incentives, or the cost-effectiveness of different forms of incentives. 3.26 Conditionality involving taxes on financial transactions aims to reduce taxation of capital market activities, especially of financial intermediaries. Such conditionality is usually reasonably well defined, in part because the taxes involved are simple. 3.27 Gaps in Nontrade Tax Conditionality. Some areas of nontrade tax conditionality that have received inadequate attention in SALs/SECALs need to be revisited because they are likely to become important in the future (see Shome 1992). These include: * Investment incentives. The Bank has focused on the most egregious investment incentives, while neglecting investment tax holidays and tax credits (sometimes combined with accelerated depreciation allowances), which offer little marginal incentive to invest but many opportunities for abuse and reduce tax revenues while increasing their variability (see Watson and Thirsk 1995). * Taxation ofglobal income. The prescription in Lessons advocating a move toward a global income tax (one that taxes all forms and sources of income at the same rate) needs to be modified to take account of a country's administrative capacity and the fact that some sources of income (for example, farm incomes) may be hard to identify. A modified schedular approach, such as that adopted in the Philippines, 47 may permit the evolution of the system toward a global income tax as administration improves.7 Table 3.1. Content of Tax Conditionality in Africa, Latin America and the Caribbean, and All Regions (percent) Conditionality All regions AFR LAC By type of tax instrument Indirect taxation 34 35 34 Value added tax 7 4 5 Individual income taxation 23 18 30 Business taxation 33 39 25 Taxation of wealth and property 5 3 9 Taxation of financial transactions 5 5 2 Total conditionality (by type of tax 100 100 100 instrument) By type of conditionality Rationalize/broaden tax base 27 26 22 Rationalize tax rates 22 27 22 Harmonize tax structure with other systems 10 10 9 Rationalize incentives/ deductions 12 13 2 Adopt value added tax 7 4 5 Adjust for inflation/ revaluation 5 5 7 Institute compliance measures 14 11 30 Others 3 3 2 Total conditionality (by type of 100 100 100 conditionality) Source: ALCID. * Taxation of capital income. The taxation of interest is an unresolved issue. Some countries that fear capital flight do not tax interest income and also exempt dividends and capital gains at the personal income level out of concern for equity and efficiency in the treatment of debt- and equity-based income. A related issue concerns the deductibility of interest expense in a system that does not tax interest Several developing countries, such as the Philippines, have settled on a compromise that partitions the income tax system into four components, or schedules: a progressive tax on labor incomes withheld at source, a final flat withholding tax on interest and dividend payments, a flat corporate income tax, and a progressive tax on net business incomes of unincorporated enterprises. Withholding labor and capital income generates timely tax payments, eliminates the need for taxpayers to file tax returns, and allows the tax administration to focus on the business sector. This four-pronged approach works better when tax rates are aligned for similar levels of income earned as either labor or business income because it removes the incentive to represent income in its most tax-favored form. As administrative capacity improves, the approach permits the evolution of the system toward a global tax: the final tax on dividends and interest can eventually be transformed into a creditable tax that effectively integrates the treatment of capital incomes at the business and personal levels. However, this four-pronged approach has pitfalls-inefficiencies may result if similar activities do not face the same tax rate. Countries that have implemented the system with significantly misaligned tax rates, such as the Philippines, may have created efficiency losses. 48 income. Disallowance puts purely domestic companies at a disadvantage relative to companies that are able to deduct interest costs abroad on funds obtained from foreign capital markets. Other unresolved issues concern the appropriate tax treatment of all forms of capital income. To promote efficiency and avoid capital flight requires low tax rates on capital income, but this limits the tax system's ability to redistribute income (since in most countries capital income usually accrues to the highest-income groups). Further, fear of capital flight and the absence of any international mechanism for coordinating capital income taxation constrain countries from unilaterally moving toward more uniform tax treatment of different forms of capital income and saving, resulting in efficiency and revenue costs. Moreover, the lack of an international mechanism for exchanging information about incomes of residents and foreign investors makes evasion relatively easy. * Taxation of the financial sector. SALs/SECALs have aimed to reduce or eliminate taxes on financial intermediaries because such taxes distort capital market activities. In many countries, however, the financial sector represents a growing and easily accessible tax base that cries out to be tapped. Income taxes and the VAT are hard to apply because it is difficult to measure net annual taxable income or sectoral value added; moreover, a tax on gross interest income receipts risks encouraging inefficient intermediation. The challenge is to find a tax that neither subsidizes nor penalizes financial sector growth. * Taxation of property. Although some SALs/ SECALs do incorporate conditions on property taxation, this remains a neglected area that is likely to become increasingly important if countries move toward greater fiscal decentralization and grapple with tax assignment issues. * Environmental taxes. The challenge for developing countries, particularly middle- income countries that are likely to consider such taxes in the 1990s, is to identify environmental tax bases, devise proxy measures for environmental damage, and design taxes that are politically acceptable and induce changes in behavior. Trade Tax Conditionality 3.28 Specificity. In contrast with the often imprecise nontrade tax conditionality, trade tax reform objectives have almost invariably been spelled out in clear-cut, quantified policy targets to be reached within a precise time frame. For example, a country might be required to reduce its maximum tariff from 100 percent to 50 percent within the following year or to reduce the proportion of its imports subject to quantitative restrictions from 40 percent to 10 percent within the next six months. When conditionality is laid out in such concrete policy steps, compliance is easy to monitor. 3.29 Content. Import conditions (57 percent of sample conditionality) clearly dominated trade conditionality (see Table 3.2 and Annex Table 3.5). They involved: * Simplifying import licensing and removing QRs on imports and replacing them with tariffs ("stage 1" of the Bank's model reform process), * Reducing the level and average dispersion of tariff rates ("stage 2" of trade reform), 49 * Reducing the number of tariff bands, applying the same rates to similar commodities, and eliminating official reference prices. 3.30 One-fourth of trade conditionality covered exports. The Bank stipulated the removal of QRs on exports (bans and quotas) and the elimination of explicit and implicit export taxes (such as multiple exchange rates, export retention schemes, marketing board price controls). The removal of QRs and export taxes reduces the anti-export bias of trade regimes and may have revenue effects. Bank trade conditions also stipulated the introduction of drawback schemes to remove customs duties on intermediate and capital goods used in export production by all exporters (including indirect and partial). 3.31 Some 19 percent of trade conditionality related to revenue raising reforms, equalization of the domestic treatment of imports and import-competing production, customs reform to ensure equal treatment of sales and excise taxes at the border, and improvements in customs procedures. Compensatory revenue measures were explicitly considered in the design of trade tax reform in only 33 percent of sample countries. Table 3.2. Trade Tax Conditionality by Type and Region (percent of total conditionality) Type of conditionality AFR LAC SAS EAP MNA All regions Export 24 29 9 38 0 24 Import 61 58 64 46 57 57 Revenue, harmonization, and 15 13 27 15 43 19 customs reform a Includes export subsidy conditions for Cte dIvoire, which attempted to implement a uniform export subsidy and a uniform tariff to simulate a devaluation in the absence of the exchange rate instrument, and Costa Rica, which eliminated export subsidies as devaluation occurred. Source: Sample of 87 loans from ALCID. 3.32 The Bank has generally adopted a sequence of trade reform, with the removal of QRs (stage 1) usually preceding trade tax reform (stage 2). Making the level of import protection and export taxation visible and measurable by replacing QRs with tariffs is a logical first step toward reducing the level and variability of taxation of both exports and imports. This sequenced approach to trade tax reform implies that the revenue effects may be quite different depending on the stage of reform. On the export side, removing QRs in countries that levy export taxes may have the effect of raising revenues. On the import side, trade reform and revenue enhancement go hand in hand in stage 1, as newly imposed tariffs absorb quota rents and transfer funds from quota recipients to the government treasury. During this first stage of reform the absence of conflict between trade tax reform and revenue requirements implies that there is no compelling need to combine trade tax reform with nontrade tax reform. 3.33 Matters become more complicated during stage 2. Lower export taxes and tariff levels will usually result in lower revenues. Governments need to anticipate these effects by adopting compensatory revenue measures (see the section "How Effective Was Tax Reform?"). 50 3.34 How well designed were the Bank's trade tax conditionalities? On the basis of a study by Rajaram (1992) of the Bank's performance in trade reform in 12 countries before 1987, this study concludes that the Bank's record in designing trade reform has improved, though problems remain. The Rajaram study concluded that the Bank's performance in ensuring that the revenue effects of trade tax measures were reflected in the design of trade taxes was extremely uneven (see also Mitra in Khalilzadeh-Shirazi and Shah 1991). Only in a few cases did the Bank made specific recommendations on adjusting the country's indirect tax system to compensate for the revenue loss expected to accompany trade reform. Rajaram also found that the Bank's trade reform proposals paid insufficient attention to the coordination of trade reform and complementary tax reforms. As a result, several Bank borrowers failed to reduce their reliance 8 on trade taxes for revenue purposes. 3.35 This study finds that in 8 of 18 countries (in the study's trade reform sample of 23 countries) in which conditionalities affected revenues, trade reforms were accompanied by (sometimes partially) compensatory tax reform measures. In as many as 10 countries, however, revenue issues were completely neglected.9 This neglect was evident in the case of both import duties and export taxes. Moreover, the removal of unequal tax burdens imposed by sales and excise taxes on imports and domestic production was addressed in only 12 of 23 countries (see Table 3.3). 3.36 Revenue factors and harmonization of the impact of indirect and trade taxes on domestic production and imports may not be the only impediments to trade tax reform. Few PRs examine ways of dealing with the losers in trade reform. Resistance to tariff cuts can be softened by providing adjustment assistance for injured parties and allowing producers adequate time to adjust to the new policy environment. The need for adjustment assistance was ignored in every Bank SAL/SECAL except those in Ghana, Mauritius, Mexico, and Nigeria. Moreover, the one- to two-year period of adjustment appears to have been insufficient. The Philippines experience suggests that in some countries a longer period may have been necessary. An early trade reform effort in the Philippines failed because of political opposition; a subsequent and so far more successful trade reform program envisages less rapid adjustment stretching over five years. How Successfully Were Tax Conditionalities Implemented? 3.37 The Bank assesses the implementation of tax conditionality in its PCRs/PARs by examining statutory compliance with conditionality. Compliance with trade tax conditionalities appears to have been better than compliance with nontrade tax conditionalities. 3.38 Review of a sample of PCRs/PARs on 35 of the 92 SALs/SECALs that incorporated nontrade tax conditions reveals that over half (52 percent) of conditions were fully implemented, about a third (34 percent) were partially implemented, and 14 percent were not implemented (see 8 Predicting revenue effects is not easy. Much depends on initial conditions (whether tariffs were actually paid or avoided) and the impacts of changes in exchange rates and tariff rates on imports of different goods. In a few of these cases, neglect reflected the inconsequential revenue effects of trade reforms or the presence of well- functioning VATs capable of replenishing revenue losses from trade reform. 10 A counterargument is that if the adjustment period is too long, it gives producers sufficient time to mobilize effective political opposition. For this reason policymakers in Mexico argued for a shock approach to tariff reform. 51 Annex Table 3.6). However, even though 85 percent of conditions were either fully or partially implemented, it is difficult to assess what was actually achieved because 60 percent lacked both quantitative and qualitative criteria (see the section "Precision of Nontrade Tax Conditionality"). 3.39 Lack of success was not confined to any particular region or income group. In most countries in which conditionalities were not fully implemented tax reform was an insignificant element in the loans. While usually a small element in SALs and SECALs, tax policy appears to be better integrated in those with a greater macroeconomic focus (for example, financial sector and public sector management) than in SECALs dealing with narrower sectoral issues such as agriculture or industry. Such narrow SECALs were overrepresented in the partial success/failure group. 3.40 Analysis of PRs and PCRs/PARs for a sample of SALs/SECALs for 23 countries with trade reform conditionalities suggests that the Bank's record in obtaining compliance with commitments to dismantle trade barriers has been impressive. All 16 countries in which the Bank sought either to reduce or to eliminate export taxes did so. With the possible exception of Bangladesh, every country that committed itself to replacing QRs with tariffs did so. In 22 out of 23 countries progress was made toward reducing both the average tariff level and the dispersion in tariff rates. Conditionalities to rationalize tariff structures, reduce import exemptions, introduce new or improved duty drawback procedures for exports, apply export subsidies, or reform customs procedures have been consistently met, although the quality of these adjustments varied. Finally, customs reform complemented trade reform in all seven countries in which customs reform conditionality was stipulated. Such reform has included improving valuation procedures, cutting red tape and speeding up clearing procedures, and enhancing revenue collection. How Effective Was Tax Reform? 3.41 The Bank's PCRs/PARs have tended to assess efficacy in terms of statutory compliance. Assessing statutory compliance is at best a first cut, however. Translating a country's tax statutes into economically effective provisions may founder at various stages: most tax code changes need to be adopted as administrative laws or regulations, tax legislation has to survive the test of the law courts, administrative structures need to be created to implement the laws, and the tax administration has to enforce the laws and regulations. Moreover, when conditionality is vague, compliance may have little real effect. 3.42 Shifts in tax structure are another broad indicator of tax reform. Table 3.4 indicates the shifts in tax structure that took place by the postadjustment period in 26 countries for which detailed revenue data were available. Overall reliance on trade taxes diminished (from 27 to 25 percent) while reliance on indirect taxes increased (from 41 to 43 percent) in all regions except EAP. The number of countries that experienced shifts is significant: 15 out of 21 countries lowered their reliance on trade taxes, and 15 out of 24 countries increased their reliance on indirect taxes. The share of direct taxes dropped in AFR and ECA, perhaps because direct taxes are easier to evade. These shifts in broad tax categories are confirmed by the shifts in tax structure that occurred between 1980 and 1993 across regions and income groups in a larger sample of countries drawn from the World Development Report 1995 database (see Annex Table 3.2). Although these shifts are consistent with reform, the effects of reform cannot be isolated: other factors, including recession and inflation, may have played a role. Table 3.3. Types of Trade Tax Conditionality by Region and Country Export policies Import policies Reductions in Tariffication average tariffs; Rationalization Harmonization of of increases in and tariffs and domestic Lower Du.Ty quantitative uniformity of simplfication of sales and excise Revenue raising Region export taxes drawbacks restrictions tariffs tariff rates taxes reforms Customs reforms Burundi Burundi Burundi Burundi Burundi C6te d'lvoire Mauritius Nigeria C6te d'Ivoire Ghana C6te d'Ivoire C6te d'Ivoire Kenya Ghana Ghana Kenya Kenya Ghana Malawi Malawi AFR Mozambique Mauritius Mauritius Kenya Mauritius Nigeria Nigeria Nigeria Mozambique Malawi Mozambique Zambia Nigeria Mauritius Nigeria Mozambique Zambia Nigeria Zambia Indonesia Indonesia Indonesia Indonesia Indonesia Indonesia Philippines Indonesia EAP Philippines Philippines Philippines Philippines Philippines Philippines Thailand Thailand Thailand Pakistan Bangladesh Bangladesh Bangladesh Bangladesh Bangladesh Bangladesh Bangladesh SAS Pakistan Pakistan Pakistan Pakistan Pakistan MNA Morocco - Morocco Morocco Morocco Morocco Morocco Morocco Argentina Costa Rica Argentina Argentina Costa Rica Jamaica Argentina Argentina Costa Rica Jamaica Costa Rica Costa Rica Jamaica Costa Rica Mexico Honduras Peru Honduras Honduras Mexico Honduras Peru LACa Mexico Jamaica Jamaica Panama Uruguay Panama Mexico Mexico Peru Panama Panama Uruguay Peru Peru Uruguay Uruguay In LAC conditionality prohibiting tariff surcharges was imposed in Argentina, Honduras, Mexico, Peru, and Uruguay. Source: President's Reports for SALs/SECALs. 53 3.43 Although few SALs/SECALs called for tax rate changes, changes in tax structure have led to increases in tax revenues as a share of GDP across regions: conditionality on revenue increases was incorporated in the adjustment programs of 26 countries. Differences in ratios of average revenue to GDP before and after adjustment (typically the year prior to approval of the first SAL/SECAL and the two most recent years for which data are available) were used to assess whether these countries took action to increase revenue performance. 1 Revenues rose in 17 ( 65 percent) of the countries and declined in 9 (Table 3.5). Success may have come easily in 5 of the 17 countries because of very low pre-adjustment ratios. Moreover, in Bangladesh, Hungary, Pakistan, the Philippines, and Uganda the revenue increase was below expectations. In some countries tax efficiency problems are still acute (Costa Rica, Honduras, the Philippines). In the 9 countries where revenues declined, weaknesses in tax administration and failure to foresee the full extent of the revenue losses from trade and to devise compensatory nontax revenue measures during adjustment were largely responsible for the declines. Another factor was the Bank's decision to leave tax policy and tax administration reform issues to the Fund, particularly in the early years, in Bulgaria, C6te d'Ivoire, and Uganda. 3.44 Did tax reform improve efficiency by reducing excess burden and administrative and compliance costs (see para. 3.12)? There is no quick or easy way to quantify the reduction in excess burden or to compare differences across countries. But it is possible to determine whether reform has moved the tax system in the right direction by estimating marginal effective tax rates (as a proxy for the change in excess burden) and by assessing their variance across different types of investments, savings instruments, and intermediaries. Although there are differences of opinion on the appropriate methodology for these measures, using the same methodology over time should provide a rough measure of the tax system's impact on resource allocation. Administrative and compliance costs can also be tracked using the type of indicators in the checklist in Chapter 4 (see Annex 4.1). While many of these more informative measures are used in the Bank's reports on economic and sector work (ESW) and in conducting its dialogue with governments, they have not been used in evaluating conditionality. 3.45 Assessing the efficacy of tax reform in countries that have received multiple SALs raises other issues. Tax reform is usually an ongoing process, with reforms in one period leading to reforms in subsequent periods. In the absence of in-depth studies of these multiple SALs, this study judges efficacy by the record on statutory compliance and revenue performance. Two broad generalizations may be made. First, success has been most evident where the Bank has played a continuing role in reform, where better tax administration has been part of the reform package, and where governments have been committed to the goals of reform (as in Malawi). This definition of the pre-adjustment period differs from the definition adopted in Chapters 1 and 6, which look at the four-year period prior to the initial year of the first loan. The difference in approach is due to differences in data availability and to differences in data sources: the data in this chapter are drawn largely from Bank and Fund documents and staff estimates; the data in Chapters 1 and 6 are drawn from IMF and Bank databases. 12 Measuring marginal effective tax rates is a natural extension of the Bank's efforts to measure changes in the level and dispersion of effective rates of import tariff protection. Although both approaches have their limitations- marginal effective tax rates typically assume perfect tax administration, perfect competition, and perfect certainty, while effective protection rates are grounded in market prices rather than shadow prices-both offer reasonable first approximations as to whether policy reforms are moving in the right direction. Table 3.4. Tax Structure Before and After Adjusiment Ratio ol lax revenue to GDP Ratios to total tax revenue Direcl taxes Taxes on goods and services Trade taxes Beore Ale' Cr.ange Beor Ale, Cnange Belare Atler Change Belore Abel Change LAC Argentina 10.0 11.4 + - - - Chile 22.8 19.6 29.9 30.6 + 50.6 54.3 + 123 11.7 - Costa Rica" 21.0 22.6 + 42.4 43.4 + 58.1 56.6 - El Salvador 7.6 10.0 + 32.2 25.6- 49.8 47.4 - 18.4 16.3 Honduras 15.1-_ 23.8+ ______ ______ Peru 68 107 + 22.2 28.0 + 57.0 58.5 + 19.2 12.0 - uu- - 11 + 8.6 10.0 + 73.5 80.0 + Average 14.1 16.4 + 27.1 27.5 + 57.8 59.4 + 16.6 13.3 cote di voire 22.1 17 .7 29.0 18.4 - 26.2 41.2 + 43.9 41.2 - Gambia, The 17.5 21.3 + 17.7 15.8 - 10.3 37.1 + 71.4 47.0 - Ghana 4.8 11.2 + 35.4 20.5 - 45.8 _ 42.9 __-22.9 ___ 37.5 Kenya 21.2 18.7 32.5 35.3 + 44.3 50.0 + 22.2 11.8 - Ma-awi 16.6 15.3 38.6 35.0 -34.9 38.2 7 21.5 Mozarribique 9,3 21.9 + 28.0 18.3 - 58.1 45.7 - 9.7 27.9 + Senegal 21.1 14.7 - 23.2 239 + 31.8 27.3 - 40.8 43.7 + Tanzania 17.3 17.9 + 35.3 297 - 43.9 56.6 + 19.1 12.9 Uganda_ 2.3 6.8 + 13.0 15.2 41.5 - 39.1 31.1 Average 14.7 16.1 + 28,1 236 38.1 42.3 32.6 30.5 - ECA Bugaria 40.5 _ 33.1 - 51.6 32.9 - 22.2 30.2 + 2.5 8.5 + Hungary 53.3 40.5 - 25.3 24.0 - 39.8 35.3 11.1 8.6 - Poland 216 25.3 + 66.7 49. 33.3 40.0 + 9.3 10.1 + Turkey' 14.7 _ 18.4 + 57.4 41.3__ _ 41.9 57.4 + ______ __ Aveg - 7.6 9.1 + Indonesia 13.9 16.5 + 57.4 62.1 + 33.1 34.8 69 5.7 - Philippines 11.9 15.5 + 24.5 33.2 + 40.1 30.4 - 25.6 33.3 + Average 12.9 16.0 + 41.0 47.7 36.6 32.6 - 16.3 19.5 Pakiatan~~~~3. 33 13. 12929.1721. 4049 ___+4. 86- Bangladesh 7.2 -+ -13.9 13.- - Average 10.3 11-1 + 15.5 16.7 + 42.1 37.5 - 41.0 34.2 Moroc 19.2 22.0 + 24.0 26.3 + 28.4 34.2 + 38.4 ___35.7 - ý I ___ - --._ --_ - -_ _ý +_ 2682.2+. Tunisia 4.8 23.734.9 26.8 28.2 + 38.3 34.6 - Average 22.0 22.9 + 28.2 30.6 +27.6 31.2 + 38.4 35.2 - TOTAL AVERAGE 17.3 1.4 3.6 28.6- 40.9 4 + 26.6 24.7 a. Taxes on goods and services include trade taxes. b. Revenue data refer only to state budget revenues, not general government. Note Values before adjustment refer to the average for the two years before the initial year of the first adjustment operation. Values after refer to the most recent two-year period for which data were available. See Gupta 1996 for country details. Source: President's Reports, Project Completion Reportslimplementation Conpletion Reports, and Bank and IMF documents and staff estimates. 55 Where the Bank has been only peripherally involved and where tax administration reforms have not been central to reform, performance has been much worse (as in the Philippines). Second, success has been more easily attained in reform of indirect taxation, where the Bank's efforts have been concentrated, than in reform of individual income taxation. This result is due to the importance of indirect taxes and to the fact that direct tax reforms encounter greater political resistance and require greater tax administration effort. Several countries in LAC (Bolivia, Mexico, Peru) abandoned traditional corporate and personal income tax systems in favor of presumptive methods of taxation and withholding. Table 3.5. Diagnosis of Need for Revenue Reform and Performance Diagnosis Revenue performance Revenue increased Revenue decreased Increase revenues 17 countries 9 countries (26 countries) AFR LAC AFR LAC The Gambia, Ghana, Argentina,a Costa C6te d'Ivoire, Kenya, Venezuela Mozambique, Rica, El Salvador, Malawi,b Senegal Tanzania, Uganda Honduras, Peru, Uruguay EAP MNA EAP MNA Philippines Morocco Indonesia Tunisia ECA SAS ECA Poland, Turkey Bangladesh, Pakistan Bulgaria, Hungary Decrease revenues a Adjustment period performance was compared with the post-adjustment performance because consistent data were unavailable for the pre-adjustment period. b In Malawi revenues in the post-adjustment period were affected by the 1992 drought. Source: President's Reports, Project Completion Reports/Implementation Completion Reports, and Bank and IMF documents and staff estimates. See also Gupta 1996. 3.46 Two measures have been used to measure the efficacy of trade tax reform: a country's ability to wean itself away from tariffs as a source of revenue, and a country's ability to implement both stage 1 and stage 2 reforms (see section "Content of Trade Tax Conditionality," para. 3.29). With these criteria in mind, the countries in the sample can be divided into four groups (see Table 3.6): * Successful adjusters, which do not now use tariffs for revenue reasons and which have navigated their way through both stages of trade tax reform. * Erratic trade adjusters, which have completed the first stage of reform and have gone part way through the second stage but have had significant policy reversals as a result of revenue declines. * Limited trade tax adjusters, which are stuck at the first stage of reform. * Nonadjusters. 56 Table 3.6. Successful and Less Successful Trade Adjusters Successful trade adjusters Erratic trade adjusters Limited trade adjusters Nonadjusters Burundi Argentina' Bangladesh C6te d'lvoire Costa Rica Kenya Pakistan Panama Ghana Malawi Philippines Honduras Mauritius Indonesia Morocco Jamaica Mozambique Mexico Nigeria Peru Thailand Uruguay Zambia a Since 1991, Argentina has achieved a stable and liberal trade regime. Source: Project Completion Reports/Performance Audit Reports for SALs/SECALs. 3.47 What distinguishes the successful adjusters from the other countries? Successful adjusters tended not to rely heavily on trade taxes as a source of revenue, were able to substitute alternative revenue sources for trade taxes, and were able to rely on the adequacy of their stabilization efforts in facilitating (tax) reform. Initial reliance on trade taxes as a revenue source was low in these countries, and almost all of them had achieved significant tax reforms that increased the relative importance of revenues from indirect taxes (the VAT) and helped them offset the revenue losses from lowering trade taxes. All of the successful adjusters were also able to effect sizable real devaluations and avoid balance of payments crises that could trigger the reimposition of trade or payments controls and the reversal of trade tax reforms. The successful adjusters also enjoyed some luck: none faced an unfavorable external environment (sharp terms of trade shifts or abrupt losses of export markets). Uncertainty can also be policy induced: the record suggests that where governments have committed themselves to a preannounced, phased schedule of tariff and other trade and tax reforms, the more certain policy environment has facilitated adjustment. Where dithering or lack of credibility has engendered uncertainty, the private sector response has been slower.13 3.48 Erratic and limited adjusters also shared common characteristics. Their reliance on trade taxes was relatively high, and they either had not carried out significant tax reform or had been unsuccessful in doing so. When faced with revenue shortfalls, these countries have typically resorted to import surcharges that have set back their trade tax reforms (see Box 3.1). Moreover, in several countries, particularly Nigeria and Mozambique, poor customs administration undermined trade liberalization. 3.49 The nonadjusting countries (Panama and C6te d'Ivoire) were handicapped by the limited number of policy instruments at their disposal; both were deprived of exchange rate flexibility in pursuing trade tax reforms. The experience of these countries illustrates the need for fiscal reform to precede tariff reform if the latter is to be successful. The experience of specific 13 In C6te d'Ivoire, for example, the export subsidy policy failed to induce investment in export activities because the private sector correctly perceived that the subsidy would not materialize. In Pakistan, where tariff reform conflicted with the goal of fiscal deficit cuts, the business community, aware that tariff and tax reforms go hand in hand, fell prey to investment uncertainty. 57 countries in all groups-Kenya, Pakistan, Panama, and, for a while, Jamaica-highlights the importance of adequate stabilization efforts in facilitating trade reform. Box 3.1. Fiscal Reform Needs to Precede Tariff Reform The experience of Kenya and Jamaica illustrates the need for fiscal reforms to precede tariff reforms. In Kenya most of the trade liberalization achieved under the SALs of the 1980s was reversed by the end of the decade, largely because of failure to achieve macro stabilization: Kenya removed QRs before it devalued the shilling, but balance of payments problems forced it to reimpose trade controls. Between 1979 and 1989 Kenya's export to GDP ratio declined from 20 percent to 12 percent. However, under the 1991 Export Development Loan, Kenya successfully devalued in 1993 and began to liberalize its trade regime. It replaced almost all its QRs with equivalent or lower tariffs, lowering the average tariff level from 50 percent to 33 percent and reducing tariff dispersion. Revenue losses were so large, however, that the government felt compelled to raise all rates by 25 percent, and the average tariff rate now stands at 40 percent. The government plans to compensate for the revenue loss by eventually increasing VAT rates and coverage. For revenue reasons Jamaica failed to meet most of the trade tax conditionality in its SALs between 1981 and 1984. Only under the 1987 Trade and Financial Sector Adjustment Loan (TFSAL) did Jamaica begin to relax some of its QRs and to lower tariff levels. A major tax reform also began to bear fruit at that time. The pace of trade reform began to pick up in 1991 under a second TFSAL. Jamaica took the lead in negotiating a new CARICOM agreement, which calls for three-stage reductions in tariffs to a 5 to 20 percent range by 1997. Jamaica has also opted for the fast track in reaching this tariff target before 1997. There is no doubt that the success of Jamaica's new VAT has given it the revenue space it needed to reduce its reliance on trade taxes. Between 1986 and 1992, when the contribution of import duties to total revenue fell from 36 percent to 10 percent, taxes as a share of GDP rose from 21 percent to 29 percent, largely as a result of the new VAT. Lessons and Directions for the Future 3.50 What can be learned from the Bank's experience with tax conditionality? Nontrade tax conditionalities have been a relatively small component in SALs, and their incidence in SALs/SECALs appears to have been on the wane since 1992. Reform also has tended to be piecemeal. The literature on tax reform in developing countries confirms that "comprehensive tax reforms, in which all or many components of a country's tax system are adjusted simultaneously, are a relatively rare occurrence" (Thirsk 1996). Comprehensive reform is rare because the political process in most countries is better able to deal with marginal reform than with wholesale reform. The lack of emphasis on comprehensive reform in the Bank's SALs/SECALs reflects both recognition of the difficulty of comprehensive tax reform and deference to the IMF in the area of tax reform. The division of roles between the IMF and the Bank has gradually crystallized since a formal understanding (the Concordat) was reached in March 1989 on Bank-IMF collaboration. The Bank's Policy Research Department has withdrawn from tax work since October 1994 and now focuses almost entirely on public expenditure issues (see para. 2.53). 3.51 As a result, the Bank devotes limited attention to research on evolving tax issues for developing countries, although cross-country studies by the Bank in the past generated much of the available documentation on "best practice." Such research, enormously useful for practitioners in developing countries, does not normally attract academic scholars, and the Bank may have a role to play here in the future. The Bank should complement the Fund's shorter-term 58 research efforts with longer-term evaluation of cross-country experiences and with empirical research on special topics that are likely to be important in the future (environmental taxes, decentralization of revenue authority, taxation of the financial sector, consumption-based business taxes, presumptive taxation methods, rate of return taxes for natural resources). A cooperative approach to such research by the two institutions would help avoid overlapping work (para. 3.54). 3.52 PCRs/PARs have not proved to be appropriate for evaluating tax reforms. Assessing the efficacy of Bank tax conditionality, particularly in countries with multiple SALs, with a view to drawing lessons for the future requires in-depth studies of specific country tax experiences, rather than general macroeconomic evaluations of individual SALs/SECALs. 3.53 The different components of tax reform need to be properly sequenced . If export taxes or import tariffs contribute significantly to total revenues, tax reform should precede tariff reform. Reforms in tax administration should either precede or accompany the structural reform of tax policy: there is no point in reforming tax structure if tax administration is poor. 3.54 In the area of nontrade taxes some reforms (involving investment incentives, for example) need to be buttressed by better analysis and pushed further. Other reforms-including taxation of the financial sector, particularly in high-inflation economies; environmental taxes; and property taxes, particularly as fiscal decentralization advances-need to be emphasized because they are likely to assume greater importance in the future. Possible reforms need to be examined in other areas-such as the use of schedular taxes with an emphasis on withholding because of the difficulties of implementing the global income tax, and the treatment of capital income in countries that are being increasingly drawn into global capital markets. 3.55 Conditionality on nontrade tax reform needs to be more precisely structured. Both the design and the evaluation of nontrade tax reform would benefit if better analysis and performance indicators were built into reform programs from the beginning. Such indicators include marginal effective tax rates and such compliance indicators as arrears collected, penalties, and revenues collected relative to forecasts (see also Chapter 4 and Annex 4. 1). 3.56 In trade tax reform the design of Bank loans needs to take better account of revenue effects. It also needs to ensure that sales and excise taxes apply with equal force to imports and to domestic sales. 3.57 Two neglected aspects of trade tax conditionality are the need to compensate losers and the appropriate adjustment period. The experience of the erratic and limited adjusters suggests that policy reversals may be avoided if adjustment assistance in the form of retraining or retooling is provided and the adjustment period is extended to five years (rather than the traditional one to two years).14 14 More recent SALs appear to have recognized the need for longer adjustment periods: the last round of trade reform in Jamaica, Pakistan, and the Philippines provided for adjustment periods of four years or more. 59 4. Tax Administration Reform 4.1 The Bank has recognized the need to buttress tax reforms with reforms in tax administration since the mid-1980s. But unlike in the area of tax policy, where the Bank has spelled out generally accepted principles in an accessible handbook-Lessons-reviewed by the Bank's Board, in tax administration reform no guidelines have been formally articulated. This chapter briefly summarizes the reform measures that have been endorsed in the literature and that should serve as benchmarks for the evaluation of the tax administration reform (TAR) incorporated in SALs/SECALs.I (See Annex 4.1 for a checklist for assessing tax administration reforms.) What the Study Found 4.2 The Bank's involvement in tax administration has increased significantly since 1987, but the overall record still suggests piecemeal reform-isolated measures that do not add up to a comprehensive effort grounded in a definite reform strategy. Some SALs/SECALs recognized the need for improvements in tax administration. But even among these, analysis and strategy were either perfunctory or totally lacking in the early loans, and conditionality was therefore weak or nonexistent. Only in the later loans were tax administration issues addressed, usually in response to revenue pressures. About a fifth of conditionality was general or nonspecific about the measures to be applied, and only about 40 percent was couched in precise terms. Areas that warrant attention include the creation of a conducive environment (particularly a legal environment); education of taxpayers; estimation methods; audits; VAT administration; withholding and presumptive taxation; the planning, organization, and staffing of information systems; and pay structures. 4.3 PCRs/PARs that tracked the implementation of conditionality pronounced compliance to be high, but given the paucity of monitorable indicators, such findings are difficult to assess. Moreover, most evaluations did not track the outcomes of tax administration reforms. The most common yardstick used to evaluate tax administration reform was the tax to GDP ratio, even though many factors other than tax administration affect tax collection. A few evaluations focused on increases in tariff collection, attributing them to improvements in customs procedures and administration; others focused on a few intermediate output indicators (for example, increases in the number of taxpayers); and a handful referred to actions involving inputs (for example, drafting of tax legislation). As with nontrade tax reform, imprecision and a paucity of monitorable indicators impaired the quality of tax administration conditionality. The Ends and Means of Tax Administration Reform 4.4 The literature on tax administration sets forth several goals for reform: * Reduce the costs of taxpayer compliance. * Raise the costs of taxpayer noncompliance. This chapter draws on background work by Amaresh Bagchi. 60 * Increase the efficiency with which tax laws are enforced. 4.5 To achieve these goals, TAR must articulate a strategy that includes a "comprehensive plan that assigns clear priorities to the tasks that must be performed, tailored to the available resources" (Bird and Casanegra de Jantscher 1992), and focus on some key functional areas and supporting measures. The six key areas are taxpayer compliance, enumeration of taxpayers, estimation of taxable income, collection of taxes, enforcement of tax regulations, and the organization. of the tax authority. Improving Taxpayer Compliance 4.6 TAR usually aims to simplify the tax system so that it can be applied effectively. Moderating tax rates reduces compliance costs, while reducing the number of exemptions and discretionary elements in the tax code eliminates opportunities for tax evasion and thus increases compliance. Eliminating opportunities for collusion between taxpayers and tax collectors by switching from an official assessment system to a self-assessment one may also increase compliance. 4.7 Compliance is less burdensome when tax laws are simple and taxpayers do not need the help of tax specialists to file returns. Simplifying tax laws, forms, and procedures makes tax administration simpler and more efficient. Compliance may also be improved by the availability of administrative and judicial grievance procedures. 4.8 Compliance can also be increased by improving the taxpayer environment: treating the taxpayer as a client, becoming more service-oriented, providing amenities in tax offices, accelerating refunds, and providing easy access to the tax authorities for clarification of the tax laws and advance rulings on ambiguous regulations. A related area is taxpayer education: taxpayers need to be convinced that "taxes are the price paid for civilized society" (Bird 1989) and that their taxes are being used effectively. Improving Taxpayer Enumeration and Registration 4.9 Entities and individuals about whom the tax authorities have no information are unlikely to pay taxes. The first requirement of good tax administration is thus to identify all those required to pay taxes and issue unique tax identification numbers (TINs). TINs serve as the building blocks for the taxpayer master file that is updated every year. The master file also stores other information that constitutes the taxpayer's current account records. These records help identify evaders or potential taxpayers who have not been brought into the tax fold, aid cross-checks of information, and reduce the likelihood of error. 4.10 Data storage, updating, and retrieval are best done with computers run by trained staff who ensure that agency employees have timely access to taxpayer current account records. The 2 Computerization has proved to be a powerful tool in supporting tax administration objectives in several key functional areas. It is invaluable for setting up and maintaining taxpayer master files, for estimation exercises, and for strengthening enforcement and audits. Computerization also helps ensure timeliness, accountability, and managerial efficiency (see paras. 4.17, 4.20, and 4.21). For these reasons, TAR programs must have well-designed plans for setting up computer hardware and software and for hiring trained personnel. 61 taxpayer current account system may be complemented by a third-party information system consisting of other government agencies, employers, banks, and other corporate entities legally required to provide information. Estimating Taxable Income and Tax Liability 4.11 Because of the high cost of administrative assessments, most industrial countries have switched to a system of "self-assessment" under which taxpayers assess their own liability, file a return reporting that assessment, and pay the taxes due. Self-assessment is likely to result in an increase in the number of taxpayers, but an underassessment of taxes due. It therefore needs to be bolstered by selective checks, audits, and penalties and by simplified procedures (assessment committees and court appeals) for conflict resolution. 4.12 To ensure compliance, many countries have also resorted to withholding and presumptive taxation. Withholding works best when the number of income payers is relatively small (and smaller than the number of income receivers). Presumptive taxation helps tax authorities to collect taxes (personal, corporate, property, and land taxes) from groups of taxpayers that find it difficult to maintain accounts. Presumptive tax assessments are usually based on surveys (cadastral surveys, sector studies, surveys of specific types of establishments) that identify a few visible indicators of standard of living or scale of business. To work well, presumptive taxation requires monitoring, vigilance, and regular updating of the assessment base. Improving Tax Collection 4.13 Improving tax collection requires providing numerous and easily accessible facilities at which taxes can be voluntarily paid as well as mechanisms for coercing payment by noncomplying taxpayers. It also requires adequate withholding systems, an appropriate penalty structure that is rigorously enforced, and the use of audits and the removal of administrative bottlenecks (such as delays in government verification of taxes paid or refunds). Critical, too, are the recordkeeping capabilities of the tax authority. When the number of taxpayers is large, computerization is cost-effective. 4.14 To improve collection, some countries have turned over parts of the collection process to private agents, with mixed results (see para. 4.21). Others have offered officials incentives linked to tax recovery, and taxpayers incentives for prompt payment. Improving Enforcement of Tax Regulations 4.15 The first step in enforcement is identifying gaps between potential and declared taxes, declared and paid taxes, and taxes paid and payments received by the treasury. These gaps result from lapses in identification, filing, reporting of current income, payment, and the transfer of tax payments to the treasury. 4.16 To distinguish between genuine errors in tax returns and deliberate misreporting, tax departments rely on audits. The authorities need to check all returns for genuine arithmetic errors, and audit a sample of returns thoroughly to uncover deliberate evasion and to impress upon all taxpayers that detection and punishment are possible. For the threat to be credible, 62 sample selection needs to be based on appropriate statistical models (such as the discriminant index function model used in the United States). However, such models can only be used when auditors have access to sufficient information on taxpayers and the tax authorities have adequate information processing capacity. Where these conditions are not met, a better use of scarce auditing resources may be to identify norms for tax returns for specific activities or lines of business and to systematically audit the tax returns that deviate from these norms. To do this, the tax administration needs to develop an audit plan and audit techniques and provide audit training. Some countries have addressed evasion by setting up special tax units that concentrate on auditing large taxpayers. 4.17 Adequate and frequently updated information systems on registered taxpayers are key to the success of enforcement audits. Computerization of taxpayer records has proved to be critically important in many countries. In some countries tax agencies have access to third-party sources of information, such as withholdings, bank transactions, foreign exchange transactions, transactions in publicly traded instruments above a specific threshold, and large transactions (involving real estate, cars, tax-deductible transactions, social security payments, customs payments). 4.18 Audits can identify nonfilers, allow better assessment of declared incomes, and reveal discrepancies between potential and declared taxes. For audits to have real clout, however, the public must be convinced that tax evasion will be punished and that punishment (fines, penalties, publicizing of evasion, seizures of property) will be significant. An important issue in dealing with delinquent taxpayers is whether to confiscate assets to recover the tax liability. Where such practices are permitted, clearly established procedures need to govern tax recovery (such as requiring judicial permission for extreme steps) to prevent abuse of authority. 4.19 Tax amnesties have been used effectively in some industrial countries, but have not proved effective in developing countries, where measures to tighten administration have seldom followed the amnesties. Improving the Organization of the Tax Authority 4.20 In most developing countries tax agencies are characterized by poor training, poor equipment, low salaries, and low status (Acharya 1986). Because tax authorities usually are poorly funded, a continuing challenge is to identify and abandon unproductive tasks (such as extensive rather than selective audits).3 The development of better management information systems through computerization has helped tax authorities identify the best uses of administrative resources, the output generated by various resource uses (collections, number of cases audited), and potential and actual tax bases. 4.21 One of the more important tasks of the tax administration authorities is to set up internal control and accountability systems to detect errors in tax assessments. To ensure that taxpayers are not taxed incorrectly, the tax authorities must also remedy deficiencies in laws, procedures, or practices that lead to errors. To deter corruption by tax officials, many developing countries There is a growing body of literature on the economics of resource allocation by tax administration authorities. Vazquez-Caro, Reid, and Bird (1992) attempt to equalize marginal product from different activities as an indicator of resource allocation. Bagchi, Bird, and Dasgupta (1995) use average cost-effectiveness indicators. 63 establish revenue targets for each tax collector, district tax office, and auditor; maintain detailed records of the eventual outcomes of assessments by tax officers; entrust sensitive negotiations to special teams; minimize contacts between taxpayers and tax collectors and reduce the discretionary powers of tax officials; set up supervisory systems with at least three hierarchical levels to reduce opportunities for collusion; and devise incentive systems that better match public and private interests. Some countries have sought to rely on private agents for tax collection. In Bolivia, Chile, Colombia, and Uruguay banks have assumed an important role in tax collection; in Indonesia foreign contractors have been entrusted with customs collection (see para. 4.14). The experience has been mixed. TAR programs should therefore evaluate such proposals with caution and in the context of an overall reform strategy, not as an isolated option. 4.22 The growing literature on TAR reveals that no single approach can be applied everywhere successfully: each country requires a unique strategy that fits its circumstances and makes the best use of its available administrative resources to minimize noncompliance. Nevertheless, the checklist in Annex 4.1 provides a useful basis for assessing the effectiveness of tax administration. Bank Involvement in Tax Administration Reform 4.23 How extensive has the Bank's role in tax administration reform been? In general, such reform has not been a large part of the Bank's reform agenda: of the approximately 250 SALs/SECALs approved by the Bank during FY1979-94, only 67 operations, about 26 percent, incorporated any TAR conditionality. Conditions on TAR were only rarely imposed before 1987 (13 loans included TAR conditions between 1980 and 1986), perhaps because of the IMF's early and substantial involvement in the area. After 1987 they became more common (54 loans made between 1987 and 1994 included such conditions) (see Annex Table 4.2). 4.24 To examine TAR conditionality, a sample of 41 completed and ongoing loans was drawn from the universe of 67 loans with TAR conditionality. To reflect the change in the frequency of conditions on TAR, 6 of the loans were drawn from those approved before 1987, 35 from those approved after 1987. Conditions on TAR occurred more often in SALs than in SECALs. The distribution of the sample operations by region, type of loan, and time period are shown in Table 4.1. 4 The checklist in Annex 4.1 draws on Vazquez-Caro, Reid, and Bird (1992). 64 Table 4.1. Sample SALs/SECALs Incorporating Tax Administration Reforms by Region and Period Regional shares in Overall overall SALs SECALs Total total total Region 1980-86 1987-94 1980-86 1987-94 1980-86 1987-94 1980-94 (percent) AFR 0 16 0 3 0 19 19 46 EAP 2 1 0 2 2 3 5 12 ECA 2 2 1 0 3 2 5 12 LAC 1 6 0 1 1 7 8 20 INA 0 1 0 1 0 2 2 5 SAS 0 0 0 2 0 2 2 5 Total 5 26 1 9 6 35 41 100 Source: ALCID. Specificity and Precision of Conditionality 4.25 Not all the conditionality in the 41 loans spelled out a specific or precise agenda. About 13 percent of the operations included only general stipulations that referred to the need to reform tax administration without suggesting particular strategies ("prepare and implement package of improvements in tax administration ...." Mozambique [1989]) or that suggested the need for TAR only as part of such tax structure reforms as base broadening, rate reductions, or reforms that require organizational change ("simplify tax budget administration, broaden revenue base, and reduce distortions"-Guinea Bissau [1989]). (See Annex Table 4.3.) 4.26 The remaining loans (87 percent) identified one or more of the six key areas of reform cited earlier in the section "Ends and Means of Tax Administration." Even here, however, some of the conditionality was nonspecific. In Ghana (1991), for example, the Bank stipulated that the borrower "continue reforms to encourage voluntary compliance"; in its 1990 loan to Senegal the Bank mandated "implementation of tax administration measures aiming at improving tax collection." Only 80 percent of all conditionalities across the six regions referred to specific reform measures in the six TAR categories (Table 4.2). General or nonspecific stipulations were most prevalent in MNA and ECA, specific conditions in LAC and SAS. 65 Table 4.2. Specificity of Tax Administration Reform Conditionalities by Region (percent) Type of reform measures AFR EAP ECA LAC MNA SAS Total General and nonspecific 25 20 54 10 100 16 20 Specific 75 80 46 90 0 84 80 (76) (28) (6) (95) (0) (16) (221) Note: Figures in parentheses refer to the number of specific conditionalities in the six identified TAR categories. Source: ALCID. Table 4.3. Sample Conditionalities by Level of Precision (percent) Reasonably Region precise Qualitative Imprecisea Total AFR 21 35 45 101 EAP 40 31 29 100 ECA 62 8 31 101 LAC 52 16 32 100 MNA - - 100 100 SAS 63 11 26 100 Total 40 24 36 100 Note: Rows may not add to 100 percent because of rounding. a Aggregate of general, nonspecific, and vague conditionalities. Source: ALCID; based on Annex Table 4.2. 4.27 How a specific conditionality is couched can affect the success with which it is implemented, tracked, and monitored. Because many of the conditionalities deal with institutional reform, which cannot be captured by a few technical parameters, assessing their precision is somewhat subjective. The general picture that emerges is that many of the conditions were stated imprecisely (Table 4.3). Only 40 percent of all conditionalities were couched in reasonably precise terms; 36 percent were couched in imprecise language, and another 24 percent in qualitative language. Taken together, the high proportion of imprecise and qualitative conditions (60 percent) suggests considerable room for devising conditionality in terms of more precise, monitorable criteria. Such criteria might include tax gaps to be narrowed, cost-effectiveness criteria, arrears ratios, proportions of cases subject to appeals, or measures that focus on the outputs rather than the inputs of tax administration (see Annex 4.1). 66 Content of Conditionality 4.28 TAR conditionalities generally conformed to the best practice guidelines suggested by the literature and summarized earlier in this chapter. They also covered all six TAR areas. The distribution of specific conditionalities is set out in Table 4.4. About 35 percent stipulated organizational reform; of these, more than two-thirds focused on the internal reorganization of tax departments, and the rest involved improvements in training, supervision, and inspection. Measures to induce taxpayer compliance were stipulated in 21 percent of conditionalities, with almost as much emphasis on simplifying forms and procedures as on simplifying laws. Improving enumeration was cited in 18 percent of conditionalities, with most focusing on computerization and adoption of taxpayer identification numbers and master rolls. Collection measures were stipulated in 10 percent of conditionalities, and enforcement measures in 13 percent. Collection reform measures focused on both tax arrears and collection processes, with relatively little emphasis on withholding or incentives for quick payment. The wider coverage of enforcement included audit reform; reform of judicial procedures; establishment of special measures for large taxpayers; reliance on third-party informants, searches, and seizures; and introduction of penalties. Measures to improve estimation were rarely stipulated, appearing in fewer than 4 percent of all conditionalities. The coverage of issues in specific conditionalities mirrored the coverage of total conditionalities (see Annex Table 4.3). Table 4.4. Specific Tax Administration Reform Conditionalities by Region (percent) Type of TAR AFR EAP ECA LAC SAS Total Induce taxpayer compliance 18 29 33 18 38 21 Improve enumeration 16 7 0 22 25 18 Improve estimation 5 4 33 1 0 4 Improve collection 14 4 0 9 0 10 Improve enforcement 13 7 0 18 0 13 Improve organization of tax authority 33 50 33 32 38 35 Total 100 100 100 100 100 100 Note: Specific conditionalities represent total conditionalities less general and nonspecific conditionalities. No identified (and therefore no specific) conditionalities were stipulated in MNA. Source: ALCID. 4.29 Regional differences are also apparent when loans are grouped by the number of TAR areas they cover (see Annex Table 4.4). More than half of all loans in AFR, LAC, and EAP covered three or more TAR areas. In contrast, only one loan in ECA and in SAS had relatively comprehensive coverage, and in MNA only general conditionalities were stipulated. 67 Gaps in Tax Administration Reform Conditionality 4.30 Although some operations have included comprehensive recommendations for tax administration reform5 and multiple rounds of tax administration reform conditionalities have been imposed in several countries,6 outside of LAC the Bank has not yet established a solid record in TAR. The picture that the record suggests-one of piecemeal reform-is confirmed by the general lack of discussion of the overall TAR strategy in PRs and the rarity of certain types of TAR conditionalities. 4.31 Among the early SALs, some recognized the need for improvements in tax administration (Bangladesh, Bulgaria, C6te d'Ivoire, Ghana, Honduras, Hungary, Malawi, Kenya, Pakistan, Senegal, Uganda, and Venezuela). But analysis and strategy were perfunctory or totally lacking, and as a result conditionality was either nonexistent or weak. For example, conditionality in Honduras and Hungary only called for tax laws to be presented to the legislature, and in Kenya it left it to each forward budget proposal to meet the mandated 15 percent increase in revenues. The quality of conditionality improved only in later loans in some countries (as in Bangladesh, C6te d'Ivoire, Ghana, Malawi, Pakistan, and Uganda), usually in response to revenue pressures (see Chapter 2). 4.32 Casanegra and Bird (in Bird and Casanegra de Jantscher, 1992) emphasize that TARs need to be impelled by a "definite strategy." The strategy needs to specify with precision the program's goals, include a comprehensive plan that assigns clear priorities to specific tasks, and be tailored to a country's circumstances. Few Bank SALs cite the underlying strategy of reform; most appear to consist of ad hoc steps. 4.33 Bank conditionalities have tended to ignore estimation and the use of such methods as censuses of tax groups or systematic surveys. Although SALs/SECALs have stipulated actions on enforcement, conditionality appears to have been adopted piecemeal. Moreover, the Bank has ignored such important issues as enforcement codes and incentives for informants. In organizational reform it has not raised issues of better pay for tax administrators-a surprising gap given the corruption, poor morale, and low productivity that are alleged to prevail in many tax departments. It has also largely ignored taxpayer education and assistance. Several areas warrant attention: * Creating a conducive (legal) environment. Where judicial processes are lengthy or make excessive demands (maintaining excessively high standards of evidence or placing the burden of proof on the administration), noncompliance cannot be prosecuted; when tax systems try to short-circuit the judicial process, the potential gains may be wiped out through a loss of public confidence in the fairness of the system. Bank SALs need to reflect awareness of these realities, and the Bank needs to examine specific environmental factors, particularly the legal tax system- something it did for the first time in the economic and sector work for Argentina in 1990. Six operations (Argentina [1991], Bangladesh [1992], El Salvador [1993], Ghana [1991], Guinea [1988], Peru [19921) are examples of fairly substantial TAR efforts. 6 About seven countries have also gone in for multiple rounds of TAR conditionalities in consecutive SALs/SECALs. 68 * Taxpayer education. Taxpayer education campaigns have been undertaken in some developing countries, with positive results. Public campaigns are sometimes credited with Mexico's success in implementing reforms. In Indonesia and Argentina publicized actions against tax evaders and defaulters improved tax collection. Korea introduced the VAT with a massive public education effort. Bank SALs need to reflect the value of involving the public in majortax reform. * The information system. The Bank has promoted computerized information systems in several countries. But computerization has not always been preceded by adequate planning, organizational restructuring, and staff training. Bank SALs need to emphasize that tax information systems must have access to data from other government agencies and such third-party sources as banks. Moreover, with increasing globalization, tax departments need to pay greater attention to the opportunities for tax evasion or avoidance through treaty shopping: double tax treaties should help promote the international flow of information. Member countries are already seeking Bank advice on the implications of globalization. * Audits. Bank SALs have focused on the need to devise sound audit strategies. Greater attention needs to be devoted to audit selection criteria, however, and to the scope and efficiency of audits. Economic and sector work reports on Argentina and Indonesia mark a beginning in this direction. * Withholding and presumptive taxation. Bank experience in these areas points to the following problems: (a) Withholding, if too stringent, can lead to "overpayment" (as in Indonesia). Also, if not final, withholding can generate excessive paperwork and add to the compliance burden of taxpayers. (b) While presumptive taxation is a convenient device for taxing "hard-to- tax" sectors, the approach requires background work on setting standards and norms for each sector that are neither so lenient as to be unfair, nor so onerous as to cause business to undertake inefficient marginal adjustments in activity. * VAT administration. The Bank has not pursued ways in which the VAT can extend its reach to small traders and farmers, or sought to implement a practical way of cross-checking credits claimed by taxpayers for taxes paid on purchases. Tracing and checking all VAT invoices is too cumbersome and costly (as experience in Korea and Indonesia demonstrates). Instead, taxpayers should be required to prepare annual lists of sales data by customer for use in developing lists of purchases that can be cross-checked at random against taxpayers' books at audit. Tracking Implementation of Reforms 4.34 Of the 41 sample loans, only 26 operations involving 25 countries have been evaluated through PCRs/PARs, all but one of which examined TAR performance. The PCRs/PARs follow up on the implementation of 64 percent of the conditionalities; no data are available on the other 69 36 percent, suggesting that some PCRs/PARs accord little importance to TARs. This study reviewed the TAR conditionalities examined in the 25 PCRs/PARs to assess how successfully they were implemented. 4.35 Of these conditionalities, 97 percent were reported as implemented, though 7 percent as only partially implemented. Because few PCRs/PARs describe the extent or quality of implementation, however, it is not possible to distinguish between fully and partially implemented tax administration reforms. Given that only a very small share of conditionalities (35 percent) are couched in precise terms, the information on implementation (whether full or partial) may not add up to much. 4.36 Because to be successful, TAR must be tailored to the specific needs of each country, proper assessment of TAR requires a detailed country study, not a quick evaluation. The weakness of TAR assessments partly reflects the Bank's deference to the Fund's superiority in resources, experience, and expertise in the area. It may also reflect the fact that tax administration is a relatively esoteric field for development economists. Finally, tax administration is an area in which best practice has not been codified and the Bank is still learning. If the Bank is to continue to design, monitor, and assess TAR, it must develop relevant performance indicators. Measuring Outcomes 4.37 The efficacy of TAR can be determined by measuring final outputs (revenues as well as fairness and equity in taxation), intermediate outputs (such as the number of audits), and the completion of actions involving inputs (such as setting up a tax analysis unit). TAR generally aims to achieve two distinct final outputs: tax revenues and equity. Because the revenue effects of tax policy reforms and TAR are difficult to disentangle and equity outcomes are difficult to measure, assessments of TAR often focus on intermediate outputs. Two approaches that measure changes through "marginal" conditions (essentially marginal productivity indicators of resource use) and average cost-effectiveness are discussed in the literature. Where lack of data precludes assessment of either final or intermediate outputs, results have been tracked in terms of inputs. 4.38 Most of the 26 PCRs/PARs reviewed did not explicitly track the outcomes of TAR. Where outcomes were examined, the most common yardstick used was the tax to GDP ratio (Peru PCR, 1994; Nicaragua PCR, 1995). Other measures used included the increase in the rate of tariff collections attributed to improvements in customs procedures and administration (Mozambique PCR, 1995); other PCRs (Congo PAR, 1991; Mexico PCR, 1993) focused on such intermediate output indicators as improvements in the ratio of collections to assessments and increases in the number of taxpayers. The Zaire PAR (1993) focused on actions involving inputs for improving the efficiency of tax administration: the preparation of draft legislation to harmonize the tax and tariff codes, the setting up of a fiscal analysis unit, and the setting aside of 5 percent of tax receipts to improve the organization and salary premia of the direct and indirect tax units. 4.39 In most situations the tax to GDP ratio cannot serve as the only measure of TAR. Since tax policy reforms generally accompany TAR in SALs/SECALs, it is difficult to disentangle their effects. Moreover, tax revenue is affected by several factors beyond the control of the tax authorities, and administrative measures often take a long time to implement. Finally, because 70 the Bank is not always the only player involved in TAR, it is difficult to isolate its influence. Two of the more informative of the completion documents-the PCR and PAR on the Mexico Financial Sector Adjustment Loan of 1989-therefore adopted the twofold approach of describing the measures actually implemented and supplementing the description with a quantified assessment of final and intermediate outputs (increase in number of taxpayers, improvements in cost-effectiveness of audits, the increased yield of direct taxes attributed qualitatively to improvements in the efficiency of the tax system, the increase in the overall tax to GDP ratio). 4.40 The Mexico PAR points to the need for more in-depth country studies if TAR is to be better assessed. Some of the in-depth assessment is available in Bank economic and sector work, and studies of tax reform in Mexico, Morocco, and Turkey (see Thirsk, 1996, forthcoming). In the former category are the reports on Indonesia (Report No. 13166-IND, 1994), Argentina (Morisset and Izquierdo 1993), and Guinea-Bissau (Report No. 9443, 1992). These are isolated examples that focus on issues other than tax administration. More such in-depth cross-country studies could flush out common problems, point to issues for the future, and provide the basis for best practice guidelines. Lessons 4.41 TAR in SALs/SECALs needs to be part of a clear reform strategy that spells out the problems of the tax department, the administration's priorities, and the plans for achieving those priorities with the resources at hand. Experience differs. In some countries (Bolivia) shock treatment paid dividends; in others (Colombia, Uruguay) a more gradual approach was warranted (Bird and Casanegra de Jantscher 1992). Introducing new taxes such as the VAT may call for meticulous planning, preparation, and implementation (Trinidad and Tobago). 4.42 Several areas of Bank conditionality need strengthening: * Estimation techniques have generally been neglected everywhere. There is considerable room for using more surveys, honing methods of presumptive assessment, shoring up the capabilities of central valuation authorities, and relying more on their findings. * Judicial procedures are sometimes so onerous that they encourage tax authorities to bypass the system. That undermines the credibility of the tax authorities and reduces their effectiveness. * Information and management systems need to be strengthened. Several tax departments have computerized their activities, but many systems are not yet performing at their peak. Reasons range from inadequate planning and training (data storage may be stressed but not fast retrieval), to inattention to client needs (such as the extent of decentralization required), to neglect of means for coordinating conditionalities with other government agencies and third parties. 4.43 The Bank has much to learn. TAR is a relatively new field for the Bank, and given the Fund's greater expertise in TAR, it stands to gain from closer coordination with the Fund in Morisset, Jacques, and Alejandro Izquierdo, "Effects of Tax Reform on Argentina's Revenue," Policy Research Working Paper 1192 (World Bank, Washington, D.C., 1993). 71 devising reform programs. The practice in the Bank has been to concentrate on in-depth studies of the functions, methods, and organization of tax administration departments and to compare country practices with the best cross-country examples. Such an approach depends greatly on the quality and experience of the expert employed to assess a country tax department. It also focuses more on tax administration activities than on final indicators of outputs and efficiency. More work therefore needs to be devoted to measuring outputs and costs and developing monitorable indicators for cost-effectiveness, resource allocation, and productivity of tax administration agencies. Assessments of tax administration reforms may have to rely on a mixed approach, one that combines the use of such indicators with cross-country comparisons (see Vazquez-Caro, Reid, and Bird 1992). If resources for such tax administration research pose a constraint, this may be another area for joint Bank-Fund work. 73 5. Public Expenditure Reform 5.1 Chapter 2 described the broad patterns of expenditure reform design and conditionality in the Bank's adjustment loans. The final chapter of this report (Chapter 6) will discuss the importance of expenditure cuts relative to revenue increases in reducing fiscal deficits across regions and country groups. This chapter focuses on the problems with the level and composition of expenditures before Bank adjustment lending gained momentum, assesses Bank conditions for changing the structure of expenditures, and evaluates the effect of Bank-supported reforms not only in terms of actual shifts in expenditure composition but also in terms of final outcomes.I The chapter looks at 219 operations in 83 countries that adopted expenditure reform under the Bank's adjustment lending during FY 1979-94.2 What the Study Found Initial Conditions 5.2 Before the initiation of adjustment lending, public spending patterns in developing countries favored capital spending over operation and maintenance of existing facilities, public sector wages over nonwage expenditures, university education over primary education, and tertiary health services over basic health services. This spending pattern led to an expanded public sector but limited access to basic services. Quality of Conditionality and Compliance 5.3 The Bank sought to introduce public expenditure reform primarily through conditions on the size and composition of expenditures. The general goal (developed in World Bank 1988b, 1990b, and 1992b) was an initial move away from low-return projects followed by a restructuring of priorities and a shift of resources to high-payoff projects; curbs on subsidies, public sector employment, and the wage bill; shifts in favor of nonwage operations and maintenance (O&M) and social expenditures; and within the social sectors a move away from tertiary services toward primary education and health services. The Bank's conditionality was generally in line with these broad expenditure guidelines; equity considerations assumed importance only in the late 1980s. 5.4 The adjustment of capital expenditures received the most attention; in recent years a number of countries that had either overly reduced capital spending or were unable to provide basic services were called upon to increase capital expenditures. Next in importance were reform of public sector employment and reduction of subsidies. In more recent years Bank lending has increasingly emphasized the importance of spending on social sectors. It sought to address equity concerns primarily through shifts in spending toward health and education and, within these sectors, toward basic services. The work on this chapter was envisaged as an intermediate step to a broader OED study of the substance and process of Bank public expenditure reviews and their efficacy and equity impacts on public spending. 2 This chapter is based on the background paper by Huther, Roberts, and Shah (1996). 74 5.5 The degree of specificity in Bank conditions varied widely, with a recent trend toward detailed conditions. Reflecting tighter resource situations, low-income countries tended to face more binding conditions than higher-income countries, and conditions tended to be stronger in AFR and ECA than in LAC. Compliance with Bank conditions has been good. However, a third of the conditionalities studied have yet to be evaluated, with half the conditions on O&M and social sector expenditures imposed in recent operations. Changes in Expenditure Patterns 5.6 In general, the study found that countries with conditions did marginally better than countries without in improving expenditure patterns between the early 1980s and the early 1990s (though problems of data quality and comparability need to be kept in mind in interpreting the results). Countries with conditions had greater success in increasing social spending and cutting defense expenditures. They also had more success in cutting subsidies, and they ended the period with lower wage expenditures than countries without conditions. 5.7 Even though both groups started out with roughly equal ratios of total expenditure to GDP, adjusters with conditions had less success in cutting expenditures, however. In allocating expenditures between current and capital spending, countries with conditions devoted higher shares to current spending than did countries without, in both the initial and the final years of the study. Despite smaller cuts in capital spending, countries with conditions ended the period with alarmingly low levels of capital expenditure, which was already low in the early 1980s. These low spending levels may have partly reflected relatively large cuts in infrastructure spending during the period under review. 5.8 Data from Bank reports or staff estimates for the 27 countries studied in Chapter 2 indicate that SAL/SECAL recipients have not had much success in restructuring expenditures away from wages and subsidies toward nonwage O&M after a period of adjustment. Attempts to cut public employment have generally been short-lived or modest. Cost recovery efforts in the social sectors also have been limited. One reason for the relatively modest performance on the expenditure front may be the inflexibility that afflicts budgets in many countries; Bank adjustment lending made little headway in this area. Final Outcomes 5.9 As with expenditure trends, final outcomes do not differ substantially between countries with conditions on health and education and those without such conditions. In primary enrollment, adjusters with conditions appear to have done marginally better than those without. In both groups illiteracy rates declined, particularly for girls and women, as did infant mortality rates. Equity Considerations 5.10 The decline in illiteracy rates and the increase in primary enrollment rates in countries facing conditions on education suggest that the access of the poor to educational services increased. Similarly, the evidence that infant mortality rates declined in almost all countries indicates that the poor's access to health services also probably increased as a result of adjustment. However, conditions on nutrition, measures to increase the earning potential of the poor, and income support programs were few. Receiving the least attention was the provision of 75 safe water and sanitation, particularly in rural areas. Both nutrition and safe water and sanitation services have received much attention in the Bank's investment lending, however. Conditions Emphasized Means Rather Than Ends 5.11 One reason that Bank conditions appear to have had little additional impact in these areas is that the Bank emphasized the means of expenditure reform rather than its ends. Expenditure conditions focused on changing expenditures in specific economic categories or sectors rather than on the relation between inputs (such as the level of expenditures), outputs (such as graduation rates), and final outcomes (such as literacy, enrollment, or mortality rates or access to basic services, particularly by the poor and needy). The Role of Government 5.12 Moreover, loan conditions appear to have been formulated without any integrated overview of expenditure issues and objectives. Conditions were not always grounded in an appraisal of the role of government in the Bank's client countries; possibilities for expanding the role of the private sector, particularly in the provision of infrastructure and social services; and the scope for shifting the government's role away from public provision toward more extensive regulatory activities. There appear to have been no systemic efforts to ensure overall budgetary discipline, to identify what expenditure programs were needed through discussions with program beneficiaries, to encourage technical soundness at the sectoral or project level, or to improve and maintain the efficiency of the civil service. These issues were not wholly neglected, but they were not placed in the context of a public expenditure framework viewed as an integrated system.3 A Framework for Evaluating Public Sector Reforms 5.13 Assessing the effects of structural adjustment programs on public sector spending requires some notion of the proper role of government and government spending in a market economy. Efficiency and equity arguments are usually marshaled to explain government interventions. Public sector protagonists argue that extensive intervention is needed to protect infant industries and vulnerable groups, jump-start underdeveloped credit markets, relieve critical shortages of basic necessities, and capture the positive externalities associated with education, health care, and infrastructure development-in short, to remedy market failures and redress inequity. Yet market failures may be compounded and equity worsened if government delivers services inefficiently. State firms often require subsidies or protected markets to survive, while inefficiency in the public provision of social services can restrict, rather than improve, access for whole segments of the population, particularly the poor. In addition, there is often disagreement on which redistributive goals are important and on how much the government should do to achieve them. The role of government is perhaps discussed more fully in country ESW reports. However, PRs do not spell out how such analysis has shaped expenditure conditionality. 76 Emphasizing Primary Health and Education 5.14 To establish the need for government spending and to assess the adequacy of reforms, both output mix and input use need to be considered. From the output perspective, economic considerations suggest that governments should invest in public goods with the highest social returns. Studies have generally found the highest returns to public investment in education and health (Baffes and Shah 1993). They have also found strong growth effects from cutbacks in military spending (Arora and Bayoumi 1994). In education returns are highest in primary education (Psacharopolous 1994; Hossain 1995), while improving girls' access to schooling also has well-documented benefits (Appleton and Collier 1995). In health a dollar spent on basic health care or on preventive care brings higher returns than the same amount spent on hospital or curative care. Equity considerations reinforce these priorities, since these programs tend to reach the poor more than others do (Jimenez 1986; Selden and Wasylenko 1995; van de Walle and Nead 1995). Decentralizing Decisionmaking 5.15 Changing the government decisionmaking process for determining the mix of spending programs may also be important. Countries with participatory democracy at the local level can often ensure more responsive and cost-effective service delivery by devolving responsibility for basic education and health services to the local level. A system of national norms and standards, accompanied by conditional grants from national to local levels, can help ensure quality (Shah 1994). Getting the Right Balance between Current and Capital Spending 5.16 Efficiency is also important on the input side, particularly in balancing current and capital spending. Countries with seriously deficient infrastructure will need to spend more on the capital side to increase the number of schools, health facilities, roads, water and sewerage systems, and so on (Alderman and others 1995; Lockheed, Verspoor, and others 1991). Countries that already have a good infrastructure base will need to spend more on operation and maintenance, to keep it functioning properly (Lockheed and Verspoor 1991). Getting Control of the Wage Bill 5.17 The public sector wage bill is often one of the largest single items in current budgets. Here, there are two key issues. One is the size of the overall wage bill. A large wage bill may reflect overstaffing (too many administrators and managers, for example, at the expense of too few teachers and maintenance personnel). The other problem is too much compression in the wage structure, so that the salaries of higher-level staff do not adequately reflect their greater contributions (Nunberg and Nellis 1995). This can make the public sector unattractive to people whose skills are better rewarded in the private sector. Improving Conditions for Private Sector Involvement 5.18 Making sure that government provision is efficient and that spending is properly distributed between capital and current budgets is one side of the public expenditure reform equation. The other side is the promotion of the private sector. In some areas the private sector already competes with the government (education, health). In others the government monopoly 77 may be maintained artificially, through barriers to entry or an inhospitable environment for the private sector. In some of those cases efforts to encourage and promote private sector development can reduce "market failures" by stimulating private activity and by reducing government regulation and interference. However, the government would need to retain its role in setting standards for private providers and ensuring compliance. Expenditure Patterns before Adjustment Lending 5.19 The expenditure patterns of developing country recipients of the Bank's adjustment lending reveal that in the 1970s a disproportionate share of public spending went to capital projects, wages, and general administration-generally at the expense of basic services. Spending on capital and infrastructure projects as a share of total expenditures was higher in five regions (AFR, EAP, LAC, MNA, and SAS; lack of data precluded ECA's inclusion in this comparison) than the average in industrial countries. Capital expenditures generally ranged between 20 and 30 percent of total public spending, compared with 12 percent in industrial countries. The need to develop infrastructure facilities, which were particularly deficient, explains some of the higher capital spending. 5.20 Governments in developing countries in AFR and MNA also devoted large shares of their spending to the civil service (both for wages and for general administration; Table 5. 1). Governments in AFR devoted 32 percent of central government expenditures to wages and 25 percent to general administration, and those in MNA 28 percent and 15 percent; the averages for industrial countries were 20 percent for wages and 10 percent for general administration. As might be expected, these higher expenditures were also reflected in higher levels of public sector employment. The civil service accounted for more than 30 percent of total employment in Algeria, Egypt, Jordan, Mali, Nigeria, Senegal, and Zambia (Shihata 1 994). 4 Table 5.1. Composition of Central Government Expenditures, Regional Averages, 1970-79 (percent) General Economic Region adm in. affairs Defense Education Health Total Wages AFR 25.4 23.6 8.8 15.1 5.8 100 31.8 EAP 13.4 26.3 22.5 15.1 2.5 100 18.8 LAC 11.0 17.1 5.2 12.9 6.2 100 30.3 MNA 15.2 24.5 9.5 16.2 4.6 100 28.1 SAS 10.7 31.5 10.8 7.9 4.5 100 18.9 Industrial countries 10.4 17.8 11.0 11.3 8.3 100 19.9 Note: Expenditure categories not listed include debt charges and subsidies. The country composition of regions is listed in Huther, Roberts, and Shah (1996). Source: IMF, Government Finance Statistics. 5.21 This emphasis on infrastructure expenditure and wages left inadequate resources for delivering basic services, especially in health and education, two of the more important social sectors. Spending on health ranged from 2.5 percent to 6.2 percent of central government total expenditures in the five regions, compared with 8.3 percent in industrial countries. These low 4For the AFR countries the civil service data relate to 1981; for the MINA countries 1984-90. 78 funding levels contributed to poor health care in many developing countries and to serious deficiencies in access. Pre-1985 data for 14 recipients of Bank adjustment lending indicate high infant mortality rates (in 9 out of the 14 countries, more than 125 deaths per thousand live births) and wide disparities in access to health care services (in 6 countries, more than half the population lacked access) (see Annex Table 5.1). 5.22 Spending on education fluctuated during the 1980s in some regions; in AFR and SAS education spending fell between 1980 and 1985 (from 5.2 percent of GNP to 4.5 percent for AFR, and from 4.1 percent to 3.3 percent for SAS). Spending began to recover for both regions in the late 1980s. 5.23 A breakdown of spending on education during the 1980s shows that public spending on primary education declined as a share of total education spending, while spending on secondary education increased in all regions except SAS. Spending on higher education also increased in EAP, LAC, and MNA (see World Bank 1995b). 5.24 On a per-pupil basis education expenditures reveal a more pronounced bias toward tertiary education in developing than in industrial countries (see Table 5.2). Ratios of per-pupil expenditures on university education to per-pupil expenditures on primary and secondary education were higher in all developing regions than in industrial countries, with spending in AFR particularly skewed toward higher education. Table 5.2. Public Current Expenditures per Pupil by Level of Education (U.S. dollars) University! University/ primary + Region Year Primary Intermediate University primary intermediate AFR 1980 47 163 2,808 59.7 13.4 1990 42 150 1,405 33.5 7.3 EAP 1980 22 57 752 34.2 9.5 1990 56 128 668 11.9 3.6 LAC 1980 147 252 992 6.7 2.5 1990 256 319 1,911 7.5 3.3 MNA 1980 - 179a 980 - 5.5 1990 - 263a 1,394 - 5.3 SAS 1980 50 91 284 5.7 2.0 1990 78 156 593 7.6 2.5 Industrial countries 1980 - 1,327a 2,580 - 1.9 'Figure is for preprimary, primary, and intermediate education. Source: UNESCO 1993a, 1993b. 79 Bank Conditionalities 5.25 Through its adjustment lending program, the Bank has sought to introduce reform by imposing expenditure-related conditions on 83 countries since 1980. Lending was directed mainly to lower-income countries in AFR and SAS, upper-middle-income countries in LAC, and lower-middle-income countries in ECA. The goal of expenditure conditions is "a desirable pattern of expenditure shifts [that] includes, in general, an initial move away from capital expenditure, followed by strong efforts to contain the government wage bill, mainly by reducing the numbers employed, and to restructure social expenditures away from tertiary services and toward primary education and health services" (World Bank 1992b). Conditions imposed to achieve all elements of this goal are shown in Table 5.3. 5.26 Capital expenditures were addressed most frequently (52 countries). Civil service reform, requiring reductions in public sector wages or employment, and streamlined budget processes and improved financial accounts were each required in 42 countries. Other conditions that were widely adopted involved limiting subsidies (33 countries) and reducing the role of public enterprises (21 countries). Conditions on education expenditures were imposed on 17 countries, and conditions on health spending on 15 countries. Though the Bank drew attention to military spending in El Salvador, Pakistan, and Uganda, military expenditures were subject to conditions only in Bank lending to Nicaragua and Zimbabwe. Regional Differences 5.27 Bank adjustment lending in the three regions that received the most attention-AFR, LAC, and ECA-had conditions on cutting civil service wages, staff, and subsidies, but the emphasis varied by region. AFR faced the broadest range of conditions, covering virtually all expenditure categories. Far fewer conditions were imposed on countries in LAC, where few or no conditions were imposed on capital, O&M, education, or health expenditures; social sector conditions focused instead on reforms of pension systems. Only limited conditions were imposed in ECA as well. In addition to cuts in civil service spending and subsidies, countries were advised to increase priority capital investments. No conditions were imposed on O&M, education, or health expenditures. Specificity 5.28 The degree of specificity in Bank conditions varied widely. Although some adjustment operations contained only general statements, the broad trend has been toward increasingly detailed conditions (the imposition of at least one quantitative constraint on a project). Low- income countries tend to face more binding conditions than higher-income countries, and conditions tend to be stronger in AFR and ECA than in LAC. For example, Malawi in 1984 undertook to "ensure adequate budget allocations to key development departments," but Gabon in 1988 undertook to "reduce public expenditure on salaries by 10.4 percent." 80 Table 5.3. Loan Conditions by Expenditure Reform Objective, 1980-94 Countries on which condition Objective Condition imposed Create governments that serve Rationalize central and local planning processes 8 Reform civil service (primarily public employment and wages) 42 Streamline budgetary processes and improve financial accounts 42 Limit public sector interventions in marketplace (public enterprises) 21 Enhance relevance and accountability through participation 1 Set public spending priorities across sectors General 16 Reduce military spending 2 Set public spending priorities within sectors General 16 Adjust capital expenditures 52 Adjust operating expenditures 14 Adjust current expenditures-wages 14 Adjust current expenditures-nonwages 12 Introduce cost recovery 4 Improve educational access Adjust allocations to primary education 17 Adjust allocations to secondary and tertiary education 16 Improve access to health Adjust allocations to basic health 15 Adjust allocations to other health 12 Improve infrastructure Adjust allocations to water and sanitation 4 Adjust allocations to roads 8 Adjust allocations to irrigation 3 Adjust allocations to other infrastructure 8 Adjust operations and maintenance expenditures (see intrasectoral allocations above) Improve agriculture's contributions to development Adjust allocations to agriculture 18 Limit market interventions by reducing or eliminating subsidies 33 Reach out to the poor and the disadvantaged Alleviate poverty 5 Promote regional development 3 Promote rural development 2 Enhance service delivery through increased private sector participation Privatization 6 Source: ALCID. Content 5.29 Capital and Current Expenditures. The Bank tailored its advice on capital and current expenditures to the infrastructure and capital investment needs of the borrowing country. In countries where large sections of the population lacked access to basic services (Gabon, Niger, Papua New Guinea, Senegal), the Bank recognized and supported an increase in capital investments in basic facilities as perhaps the most efficient use of public resources. In countries with many large capital projects that might not meet the basic needs of the population, however, the Bank generally called for a reduction in capital expenditures or for joint review of large new capital projects with the Bank before undertaking them. More emphasis was placed on maintenance expenditures, to prolong the useful life of past capital investments. Increased O&M 81 expenditures were required in many countries (Chad, the Central African Republic, the Lao People's Democratic Republic, Papua New Guinea, the Philippines, Somalia, Tanzania, Togo, Uganda, and Zambia). In some cases, Bank conditions called for appropriate funding of O&M before a country funded additional capital projects; in many others, countries undertook to establish spending priorities or to streamline the decisionmaking process. 5.30 Civil Service Reform. Civil service reform was addressed through a variety of conditions (Table 5.4). Most common were mandated staffing reductions and freezes or limits on increases in wages or staff. Conditions requiring reductions in the public sector wage bill to deal with the excessive allocation of government funds to payrolls have featured in SALs/SECALs since 1980. In several countries Bank conditions supported the retraining or redeployment of displaced public sector workers (Costa Rica, Hungary) or the provision of severance pay or other assistance to redundant employees (Mauritania, Zambia). In Benin and the Central African Republic voluntary departures were encouraged by abolishing family, housing, and travel benefits. Since 1985 Bank conditionalities have supported government-funded retraining when public sector employment has been cut. Table 5.4. Countries Facing Civil Service Conditions Condition Countries Reduce public employment Honduras, Nicaragua, Niger, Turkey, Zaire Change compensation terms Bangladesh, Hungary, India, Mauritania, Tanzania, Tunisia, Zambia Reduce public employment and change Argentina, Benin, the Central African Republic, compensation terms Comoros, Costa Rica, C6te d'Ivoire, Ethiopia, Gabon, the Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, the Lao People's Democratic Republic, Mali, Papua New Guinea, Peru, Sdo Tom6, Senegal, Sierra Leone, Uganda, Uruguay Source: ALCID. 5.31 The Bank has also addressed simplification of the salary structure, wage decompression, equalization of pay discrepancies, and the monetization of benefits. Incentives for public servants were addressed through conditions aimed at improving salaries in the public sector. In several countries-including Bangladesh (1984), Ghana (1987), India (1990), Tanzania (1986), and Uganda (1994), where wages had fallen to very low levels-the Bank encouraged increases in public sector wages. Some countries were to increase salaries by lowering employment (Argentina, Mali, Uganda); others agreed to adjust public sector salaries to compensate for real wage declines (Ethiopia, Uruguay). In Ghana, where civil service reforms were addressed in a series of adjustment operations, some of the reforms were designed to increase salaries for senior staff. 5.32 Subsidies. Bank conditions have consistently required reductions in subsidies (both explicit and implicit). Reductions in fertilizer subsidies were promoted in Bangladesh, Indonesia, Mexico, and Senegal. Conditions focused on reductions in subsidies to public enterprises in Burkina Faso, Burundi, the Central African Republic, the Gambia, Guyana, Jamaica, the Lao People's Democratic Republic, Mexico, Mozambique, Nicaragua, Senegal, Sierra Leone, Turkey, and Venezuela. Other Bank conditions called for the elimination of 82 guaranteed prices for commodities or agricultural products in Honduras and Mexico and for the elimination of implicit subsidies on interest rates (through preferential credit) in Poland, Tunisia, Turkey, and Zaire. Energy subsidies received scant attention in Bank conditions on adjustment lending, even though subsidies on transport fuels are generally regressive as well as environmentally damaging (see Larsen and Shah 1995). Energy subsidies receive considerable attention in investment loans, however. Equity Considerations 5.33 The Bank has focused greater attention on equity concerns since the late 1980s, using its adjustment lending to encourage shifts in expenditures toward human resource development, particularly basic activities that yield the highest returns in both equity and efficiency. Better access to health and education was addressed most often, with conditions focusing on three issues: the overall budget share for health and education, the share of sectoral budgets allocated to personnel, and the allocation of resources between tertiary and primary activities. Other equity concerns were reflected in conditions for ensuring adequate nutrition or increasing the income and earning potential of the poor, which were each addressed in adjustment lending to 11 countries. Income support programs were subject to conditions in 9 countries. The distribution of conditions by equity objective is summarized in Table 5.5. Table 5.5. Loan Conditions by Equity Objective, 1980-94 Countries on which condition Objective Condition imposed, Improve access to basic health and education 30 Improve intersectoral allocation 20 Increase nonwage expenditures 12 Reduce tertiary and increase primary expenditures 9 Introduce fee system 4 Increase rural investment 2 Increase role of private sector 1 Ensure adequate nutrition for the poor 11 Target subsidies 10 Introduce food stamps 1 Sponsor direct government provision 1 Increase income and earning potential 11 Sponsor government programs (employment, retraining) 7 Improve road access to markets 5 Reach poor through income support programs 9 Increase support for affected workers 7 Reform social security 2 Improve access to safe water and sanitation 3 Increase share of government budget 2 Introduce fee system I Increase rural investment 1 Source: ALCID. 83 5.34 Better Access to Education and Health. Bank conditionalities emphasized a reallocation of funds toward the social sectors, particularly education and health. Some conditions allowed borrowers considerable discretion in how to achieve such shifts: in these countries Bank conditionality called only for public expenditure reviews. In two countries the Bank encouraged attention to the budget process itself through development of "core" budgets to protect essential social services. More common were Bank conditions on the share of the social sectors in the budget. In 24 operations in 17 countries the Bank stipulated that the budget share be agreed on, maintained, or increased. Half the PRs included only general statements of intent (the Gambia, for example, was encouraged to increase the share of health and education in the recurrent budget). The remaining Bank operations included detailed conditions that left little discretion to the recipient government (Mali was to allocate at least 25 percent of the recurrent budget to education in FY 1989; Malawi was to maintain specific expenditure floors). 5.35 The size of the wage bill in health and education was the second most frequently addressed issue in the social sectors, with reducing the share of personnel expenditures a typical general condition. More specific conditions in 14 operations in 12 countries required higher expenditures on such nonwage inputs as teaching materials and health supplies, which have been shown to be important for effective service delivery (see Lockheed, Verspoor, and others 1991). There was a clear trend toward increasing specificity: 70 percent of operations approved in the 1990s included specific conditions, some quite detailed. An FY1990 credit to Guinea, for example, specified minimum expenditure levels on materials per pupil; a 1992 credit to Kenya included ceilings on the number of primary and secondary teachers. 5.36 Conditions requiring a shift in health or education resources from tertiary to primary activities were included in 12 operations in 9 countries, consistent with empirical evidence that expenditures at the primary level are pro-poor, while tertiary benefits flow to the more affluent (Jimenez 1986; Selden and Wasylenko 1995). In this area the Bank relies heavily on detailed conditions: all but one of the nine projects funded in the 1990s included specific conditions. 5.37 User fees, properly designed, can also promote the shift from tertiary to primary activities while improving service delivery. Charging income-adjusted fees to recover costs in higher education and tertiary health care, while imposing low or no fees for basic services, can generate revenue and allow governments to reallocate funds toward primary education and basic health, thereby increasing access to basic services (Barnum and Kutzin 1993; Lockheed, Verspoor, and others 1991). The Bank has sought to promote cost recovery in only four countries, however. Bank conditions in these countries supported the introduction, testing, or review of fee systems for education or health. In Malawi the fee system was introduced as part of a broader expenditure strategy. The government was to reduce contributions to tertiary education by 20 percent over a five-year period as student contributions were increased and recurrent outlays for education were maintained at an adequate level. A narrower focus was taken in Kenya, where user charges were to be introduced for health, education, and other public services to fund an increasing share of current expenditures. These examples suggest that a consistent approach was not adopted in designing fee systems. 5.38 In two countries (C6te d'Ivoire and Guatemala) the Bank imposed conditions requiring the development of investment plans for rural service delivery to address the unequal access to education and health in urban and rural regions. However, a more consistent approach may be necessary: in Cte d'Ivoire, for example, the operation requiring a plan for rural social sector investment was followed two years later by another that froze construction of new health and 84 education facilities (presumably also in rural areas) so that targets for the overall investment budget could be met. 5.39 Bank conditionality for directly promoting nutrition has been relatively sparse. Direct government provision of food was addressed in only one case, Madagascar, where it was to be eliminated. Introduction of food stamps was sought only in a 1988 loan to Mexico (see Table 5.5). More common were conditions requiring reductions in subsidies on food and consumer staples. Here too the Bank approach was not monolithic: a 1994 credit to Niger included the condition that food staples be subsidized. Cutting broad subsidies can improve efficiency and equity by eliminating significant benefit leakages to higher-income groups. However, Bank conditions rarely took the next step in improving targeting to reach those in need. An exception is Macedonia, which replaced a general price subsidy with a targeted bread subsidy. Program design was not addressed in most conditions. 5.40 Increased Income and Earning Potential of the Poor. Bank conditions on income or earning potential emphasize direct government involvement in the labor market. In seven countries governments undertook to provide employment or to fund retraining programs; in five countries governments were required to establish funds or severance policies to provide support to workers affected by the adjustment. Expanding coverage of income support programs typically is not addressed through Bank conditions. Only in an adjustment loan to Chile did a condition require the expansion of the coverage, as well as benefits, of an unemployment program. India, Poland, and Uruguay were required to increase the funding (or benefits) for unemployment programs or old age and family assistance, but expanding the coverage of these programs was not a condition of the loans. 5.41 Safe Water and Sanitation. Access to safe water and sanitation services, particularly sanitation, remains poor in all regions, especially in rural areas, but was addressed in only three countries receiving adjustment loans: C6te d'Ivoire, Malawi, and Zambia. Investment lending has been the primary tool used to address these deficiencies. Implementation of Conditionalities 5.42 A review of 124 SALs/SECALs to 69 countries to assess compliance with specific expenditure conditionalities (173 in all) suggests that Bank conditions have generally been implemented: 85 percent of all conditions for which implementation information was available in PCRs/PARs were met fully or in part (Table 5.6). Except for conditions relating to defense expenditures, this pattern of compliance (more than 84 percent of conditions met in all sectors) holds for each area of conditionality: subsidies, civil service reform, O&M expenditures, and spending on education, health, and poverty. In no case has the share of unfulfilled ("not met") conditionalities exceeded 16 percent. However, no PCRs/PARs were available for tracking the implementation of 32 percent of the conditions. Nearly half the conditions on O&M and social sector expenditures were imposed in recent operations that have not been completed or have not yet been reviewed following completion. 85 Table 5.6. Implementation of Bank Conditions on Public Expenditures Education, Total Civil health, and conditions Subsidy service O&M poverty Defense No. % No. % No. % No. % No. % No. % Total conditions 173 100 60 100 66 100 15 100 29 100 3 100 Nodata 55 32 9 15 23 35 7 47 14 48 2 67 Data 118 68 51 85 43 65 8 53 15 52 1 33 Conditions on loans with data 118 100 51 100 43 100 8 100 15 100 1 100 Met 75 64 31 61 27 63 6 76 11 74 0 0 Partially met 25 21 13 25 9 21 1 12 2 13 0 0 Not met 18 15 7 14 7 16 1 12 2 13 1 100 Note: The loans/credits and countries included in this table are listed in Annex 3 of Huther, Roberts, and Shah (1996). Source: PCRs/ICRs. How Effective Were Expenditure Reform Conditions? 5.43 This section assesses the efficacy of public expenditure reforms by examining the impact of Bank conditionalities on overall and sectoral and functional expenditure categories and by identifying final outcomes in each expenditure area and sector. Two caveats need to be borne in mind. First, data problems preclude more formal analysis; moreover, even regional comparisons of expenditure performance in particular sectors for adjusters with conditions and those without are not possible because country composition varies for each type of expenditure. Second, many of the SALs/SECALs, particularly those with social sector expenditure conditions, have been approved too recently for their full effects to be felt. As Summers and Pritchett (1993) point out, before-and-after comparisons for short periods are likely to underestimate the effects of Bank conditions. Impact on Expenditure Levels 5.44 The impact of Bank diagnoses and conditionalities on expenditures is assessed by looking at changes in expenditure performance in 27 countries where a need for reform of total, capital, and current expenditures was diagnosed in PRs, and by comparing the performance during the 1980s of adjusters that faced conditions in specific sectoral or functional categories of expenditure with the performance of countries that faced no conditions.7 5.45 Total Expenditures. A need for actions affecting total expenditure was diagnosed in the PRs of all 27 sample countries (Table 5.7 and Annex Table 5.2). Performance matched the diagnosis in 70 percent of the sample. Total expenditures increased in all 5 of the countries 6 This sample of 27 countries is the same as that studied in Chapter 2. For consistency reasons this comparison is based on a different data source, the IMF's Government Finance Statistics. The countries in the sample also differ from those analyzed in Chapter 2. 86 where initial levels of expenditure were low and increases were diagnosed as necessary for fiscal health; in 2 countries (the Philippines, Uganda), however, the increases were larger than required. Among the 22 countries whose PRs diagnosed a need for a spending cut or restraint, total spending declined in only 14, and in 3 of these the cuts were smaller than diagnosed. Moreover, in several countries-C6te d'Ivoire, Kenya, and Tunisia--the decreases were achieved largely through reductions in capital spending, not through the current expenditure cuts called for in their SALs/SECALs. 5.46 Capital Expenditures. Capital spending performance was even more consistent with the Bank's diagnosis of the need for reform, with success in 79 percent of the sample (Table 5.8 and Annex Table 5.3). Of the 15 country cases in which an increase in capital spending was called for, II actually stepped up expenditures. However, even among the countries encouraged to increase capital expenditures were some whose subsequent capital spending levels remained modest (Bangladesh, El Salvador, Ghana, Pakistan, Peru, the Philippines, Uruguay). This may partially explain the failure of growth rates to recover in Pakistan, Peru, the Philippines, and Uruguay. In others increases exceeded targets, as in Uganda. Of the 10 that undertook to decrease or restrain capital expenditures, only one failed to comply with the Bank's diagnosis. The performance of a few (C6te d'Ivoire, the Gambia, Kenya, Malawi) may have overshot expectations. 5.47 Current Expenditures. All 27 countries undertook to modify current spending (Table 5.9 and Annex Table 5.3). However, current spending behavior proved to be the most recalcitrant; the performance of only 44 percent of the sample matched the diagnosis in Bank PRs of the need for current expenditure reform. Of the 25 countries where the Bank called for expenditure decreases or restraint, 10 were successful in curbing current spending, though in Honduras and Hungary the actual reductions were smaller than targeted. Two countries (Ghana, the Philippines) undertook to increase current expenditures; both were successful, but in the Philippines actual spending outstripped targets. 5.48 What factors account for the poor current expenditure performance? Almost all countries that failed to cut current expenditures had weak budgetary systems and processes. Another common problem was the inability to reduce and sustain cuts in employment and therefore in wages and salaries. Finally, Bank conditionality on expenditures was often weak. 87 Table 5.7. Diagnosis of Need for Total Expenditure Reform and Performance Diagnosis Total expenditure performance Expenditure increased Expenditure decreased Increase expenditure 5 countries (5 countries) AFR LAC Ghana, Uganda' El Salvador, Peru EAP Philippinesa Decrease/restrain expenditure 8 countries 14 countries (22 countries) AFR LAC AFR LAC Mozambiqueb Argentina, Costa C6te d'lvoire, the Chile, Venezuela Rica, Honduras, Gambia, Kenya,c Uruguay Malawi, Senegal, Tanzania ECA SAS EAP MNA Poland, Turkey Pakistan Indonesia Morocco, Tunisia ECA SAS Bulgaria, Hungary Bangladesh a Actual increases in expenditure were higher than diagnosed. b Part of the increase in expenditure identified in Mozambique was due to better accounting, for example, expanded coverage of expenditures included in the budget. c Despite earlier adjustment efforts, the decline in expenditures in Kenya was small. Source: PRs, PCRs/ICRs, and Bank and IMF documents and staff estimates. See also Gupta 1996. 88 Table 5.8. Diagnosis of Need for Capital Expenditure Reform and Performance Diagnosis Capital expenditure performance Capital expenditure increased Capital expenditure decreased Increase capital expenditure II countries 4 countries (15 country cases) AFR LAC AFR ECA Ghana, Mozambique, El Salvador, Peru, C6te d'Ivoire,a Polanda Uganda Uruguay" Kenya,a Senegal EAP MNA Indonesia,a Moroccoa Philippines SAS Bangladesh,' Pakistan Decrease/restrain capital expenditure I country 9 countries (10 country cases) LAC AFR MNA Honduras C6te d'Ivoire,b the Morocco,b Gambia, Kenya,b Tunisia Malawib EAP LAC Indonesiab Costa Rica ECA Turkey Note: Where different operations for the same country diagnosed a need for both increases and decreases in capital spending, the episodes are reflected as two separate "country cases" in the table. a Because capital expenditure increase was introduced not in the initial adjustment operation but in later ones, the post-adjustment performance is compared with the capital expenditure to GDP ratio during the adjustment period. b Because capital expenditure decrease was called for in the early loans for countries for which later loans called for capital expenditure increases, pre-adjustment performance is compared with adjustment period performance. Source: PRs, PCRs/ICRs, and Bank and IMF documents and staff estimates. See also Gupta 1996. 89 Table 5.9. Diagnosis of Need for Current Expenditure Reform and Performance Diagnosis Current expenditure performance Current expenditure increased Current expenditure decreased Increase current expenditure 2 countries (2 countries) AFR EAP Ghana Philippinesa Decrease/restrain current expenditure 15 countries 10 countries (25 countries) AFR LAC AFR LAC C8te d'Ivoire, Argentina, Costa The Gambia, Chile, Honduras, Kenya, Malawi, Rica, El Salvador, Senegal, Tanzania Venezuela Mozambique, Peru, Uruguay Uganda ECA MNA EAP MNA Poland, Turkey Tunisia Indonesia Morocco SAS ECA Bangladesh, Bulgaria, Pakistan Hungaryb a Actual current expenditure increases were significantly higher than targeted. b Actual declines in current expenditure were smaller than targeted. Source: PRs, PCRs/ICRs, and Bank and IMF documents and staff estimates. See also Gupta 1996. Comparisons between Countries with Conditionality and Those Without 5.49 Though the pattern varied across regions, overall total public expenditure ratios to GDP in the early 1980s were roughly similar in countries with conditions and in those without. During the 1980s total expenditure declined as a share of GDP in both groups of countries, but declined more in countries without conditionality. EAP and SAS, where conditionality apparently paid off, were the only exceptions to this pattern. The overall decline in the total expenditure ratio was due primarily to a reduction in capital expenditures, with the decline more marked in countries without conditions than in countries with them. However, in countries with conditions the capital expenditure ratio, already low in the early 1980s, appears to have dropped to alarmingly low levels by the early 1990s (4 percent of GDP). Current expenditures rose as a share of both GDP and total expenditures in both groups during the last decade (see Annex Table 5.4). 5.50 Countries with conditions made a greater effort to increase spending on education and health by reducing expenditures on defense and infrastructure; however, the decline in the group's infrastructure spending as a share of total expenditure (about 9 percentage points) may have been too sharp. Countries without expenditure conditions actually increased the share of defense in total expenditure. 5.51 The share of the wage bill in total expenditure increased in both groups during the 1980s. Although data on subsidies (particularly for the early 1980s for countries without conditions) are hard to come by, they suggest that countries with conditions were more mindful of the need to reduce subsidies. 5.52 Regional patterns varied widely across expenditure categories. AFR stands out in two ways: initial and final levels of expenditure on health and education by countries with conditions 90 were higher than for countries without, and defense expenditures by countries with conditions were lower at both times than for countries without. Shifts in expenditure patterns were in the wrong direction in education and defense for both groups of countries in AFR. AFR countries with conditions increased the share of wages in total expenditures, but their wage expenditure remained lower than that in countries without conditionality. In SAS, despite expenditure shifts in favor of the social sectors, countries with conditions continued to devote smaller shares of total expenditure to these sectors. Impact on Major Expenditure Components 5.53 Subsidies. The limited data on subsidy reductions indicate that although expenditures on subsidies were often reduced in response to Bank conditionality, the reductions often are not sustained (see Annex Table 5.5). 5.54 Civil Service Reform: Wages. Data are even more limited on public sector wage and salary expenditures, covering only 6 of the 14 countries with explicit conditions on wages. Comparing these countries with 9 others that had no wage conditions suggests that, overall, wage conditionality did get results (Table 5.10). Wages fell as a share of total expenditure in 4 of the 6 countries with conditions, though the decline was minimal in Hungary. The wage share increased in 2 of the 6 countries (Costa Rica, Guinea). Wage shares fell in only 4 of the 11 comparator countries with no wage conditions. The data-incomplete as they are-suggest the stability of wages as an expenditure category, though Bank conditionality appears to have been a relatively successful tool. Table 5.10. Changes in Wage Expenditure Shares, Before and After Adjustment (percentage of total expenditure) Country Adjustment period Share before Share after Change Wage conditions Costa Rica 1985-89 37.9 41.0 + Guinea 1990 16.9 22.1 + Hungary 1986-88 7.0 6.9 - Togo 1983-85 29.5 24.3 - Uruguay 1984 28.8 23.5 - 1989 21.7 18.0 - Zimbabwe 1983 28.8 24.6 - No wage conditions Bolivia 1987 35.7 40.9 + 1992 39.2 29.4 - Brazil 1984 7.5 5.5 - Chad 1989 23.4 25.7 + Chile 1986-88 20.2 18.2 - Colombia 1985 18.7 18.8 + Mauritius 1981-87 30.3 35.3 + Mexico 1983 23.1 17.5 - 1989 13.5 26.9 + Morocco 1984-89 32.7 36.5 + Panama 1984-87 28.2 47.2 + Source: IMF, Government Finance Statistics. 91 5.55 Operations and Maintenance. Data on current expenditures generally are not broken out by detailed expenditure categories; data on subsidies in developing countries are very limited or incomplete, and data on nonwage O&M are particularly difficult to come by. However, data on the ratios to GDP of current expenditures, wages and salaries, and subsidies for 17 of the 27 sample countries studied in Chapter 2 (not all of which had wage and subsidy conditions) suggest that even in countries where current expenditures increased as a share of GDP, these increases were often swamped by increases in wages, subsidies, and transfers. This suggests that shifts in current expenditures toward nonwage O&M in this sample of adjusting countries were correspondingly very limited (Annex Table 5.6). 5.56 Education Expenditures. Data are especially weak on education expenditures. Almost 60 percent of SALs/SECALs with conditions on education expenditures were approved after 1990, so it is too early for the expenditure effects to show up fully. And data availability at the country level is also a problem. Although during the early 1990s SAL/SECAL countries with conditions devoted a higher share of public expenditures to education than did countries without conditions, the regional and country experience underlying the overall average is varied (Annex Table 5.4 and Table 5.11). The share of central government expenditures allocated to education increased in Bulgaria, Indonesia, and Zambia, and decreased in Burkina Faso, Cameroon, and India (in India, where states are responsible for funding education, the decline in central government expenditures probably reflects cutbacks in tertiary funding only). Table 5.11. Changes in Central Government Expenditures on Education, Before and After Adjustment (percentage of total expenditure) Fiscal year of Share Share Country condition before after Bulgaria 1992 3.1 3.4 Burkina Faso 1991 17.9 17.3 Cameroona 1989 2.9 2.7 India 1993 2.2 1.9 Indonesia 1989,1991 9.0 9.8 Zambia 1991 10.2 13.4 Note: The adjustment period is the fiscal year listed; for Indonesia the adjustment loan approval period is 1989-91. Pre-adjustment periods are three years prior to the approval year. The most recent periods are as follows: Bulgaria, 1993-94; Burkina Faso, 1992; Cameroon, 1993; India, 1994; Indonesia, 1992-93; Zambia, 1992-94. a. For Cameroon, education expenditures are shown as a ratio to GDP. Source: IMF 1994b. 5.57 Eight of the nine adjustment loans that called for shifts of funding from tertiary to primary education were approved after 1990, so detailed analysis of the outcomes of these conditionalities is not possible. The only available evidence on expenditures is for Ghana. Between 1984 and 1990 the share of current expenditures at the tertiary level declined by slightly less than 2 percentage points and the share of primary expenditures increased by slightly less than 5 percentage points. 5.58 The pattern of education spending in adjusting countries during the 1980s shows that social sector spending does not necessarily suffer when an adjustment program is adopted (see Annex Table 5.7, which separates countries with education conditions from those without). Of 92 the 55 countries in Annex Table 5.7, 65 percent showed increases in education spending as a share of both total public spending and GDP; the same percentage (but not the same group) of countries also showed a positive average annual growth rate in education spending between 1980 and 1990. Some countries with Bank conditions on education (Cameroon, Ghana, Kenya, Malawi, Uganda) experienced positive growth rates in education spending. In others (the Gambia, Mali, Mozambique, Sierra Leone, Trinidad and Tobago, Zambia) the Bank called for increases in the allocation of funds to education to halt a falling trend in education expenditures. In LAC, a region where SALs/SECALs did not include conditions on education spending, 10 of 16 countries listed in Annex Table 5.7 had negative growth rates in education spending during the 1980s. 5.59 Health Expenditures. Annex Table 5.4 suggests that Bank conditionality encouraged higher levels of health spending, as well as increases in health spending as a share of total spending, in countries with conditionality relative to countries without. The limited country- specific empirical evidence also suggests that Bank conditionalities exerted a positive influence on central government health expenditures. The share of health expenditures in total spending increased after adjustment in 5 of the 6 countries in Table 5.12. Table 5.12. Changes in Central Government Expenditures on Health (percentage of total expenditure) Year of loan Share Share Country approval before after Bulgaria 1992 2.1 3.1 Burkina Faso 1991 6.4 6.9 Cameroona 1989 0.8 0.7 Indonesia 1989, 1991 1.7 2.6 Uruguay 1989 4.5 5.0 Zambia 1991 7.3 9.8 Note: The adjustment period is the year of loan approval; for Indonesia this period is 1989-91. Pre-adjustment periods are three years. The most recent period for Bulgaria is 1993-94, Burkina Faso 1992, Cameroon 1993, Indonesia 1992- 93, and Zambia 1992-94. a. For Cameroon health expenditures are shown as ratios to GDP. Source: IMF 1994b. 5.60 Budgetary Inflexibility. The mixed performance of adjustment lending recipients can be traced in part to budgetary inflexibility. Several large expenditure categories, including statutory transfers to lower levels of government, defense funding, and debt payments, cannot be adjusted quickly. In Brazil, India, and Zaire, for example, over 90 percent of expenditures were not easily adjustable during the past decade (Table 5.13). Adjustment lending programs appear to have had little effect on such budgetary inflexibility, and so to have had limited opportunities to reshape spending priorities. Among regions, SAS had the least budgetary flexibility (69 percent of expenditures were fixed on average during 1986-94), followed by AFR (41 percent fixed). 93 Table 5.13. Expenditure Inflexibility: Fixed Categories as Percentage of Total Revenue Region/country 1977-85 1986-94 Industrial countries 39 39 AFR. 27 41 EAP 37 34 ECA 35 20 LAC 29 37 SAS 54 69 Countries with greatest inflexibility Zaire 39 122 Brazil 37 99 India 81 91 Bahrain 78 81 Note: Figures reflect simple period averages of the results for each year calculated using the following formula: (defense expenditure + interest payments + transfers to other levels of national government) / (total revenue and grants - grants - entrepreneurial and property income). Source: IMF, Government Finance Statistics. Outcomes: The Ultimate Objective of Expenditure Reform 5.61 Education. The ultimate goal of Bank conditions on education expenditures is improving educational outcomes. The overall effectiveness of Bank conditions must be evaluated by looking at the basic indicators of education: illiteracy rates and primary enrollment ratios. Though illiteracy rates adjust very slowly, the data confirm that illiteracy rates are declining, particularly for girls and women. 5.62 Relevant data on primary enrollment rates are available for only nine countries that faced conditions on education spending. Data for the years before and after the imposition of the conditions show that primary enrollment rates increased in six countries immediately after conditions were imposed (Table 5.14). Enrollment declined in two countries (Burkina Faso, Trinidad and Tobago), and in one (Senegal) showed no immediate change but declined within two years. In Ghana enrollment figures rose initially but fell two years after the credit was extended. Among five comparator countries with no education conditions, enrollment increased steadily in two countries during the 1980s (Burundi, the Central African Republic), rose after a brief downturn in the late 1980s in one (Mauritania), and fell or remained stagnant in the late 1980s in two (Tanzania, Zaire). 94 Table 5.14. Primary Enrollment Ratios in Countries with Education Conditions (percentage of age group enrolled) Fiscal year of Country conditions 1985 1986 1987 1988 1989 1990 1991 1992 Impact Countries with education conditions Benin 1989 - - 66 67 58 61 66 - + Burkina Faso 1991 - - 32 34 35 37 37 31 - Gambia, The 1989 - - 65 62 64 65 66 69 + Ghana 1987, 1989 76 72 70 73 74 77 74 - +, - Guinea 1990 - - 34 34 34 37 37 42 + Mali 1989 - - 23 23 23 24 25 - + Nigeria 1990 - - 68 73 75 73 76 - + Senegal 1990 - - 58 58 58 59 59 58 0 Trinidad and Tobago 1990 - - 99 97 96 96 95 - - Average - - 57 58 57 59 59 - Countries with no education conditions Burundi - 60 69 70 71 72 70 - + Central - 63 65 65 68 - - - + African Republic Mauritania - 51 52 51 49 50 55 - + Tanzania - 69 66 69 69 69 69 - Zaire - 77 76 - 55 70 72 - - Average - 64 66 64 63 65 66 - Source: World Bank, 1995g. 5.63 Improving basic education for girls has received little direct attention in Bank conditions, and although some narrowing has taken place, the gender gap persists. Enrollment figures for girls show a mixed pattern (Table 5.15). Enrollment of girls increased in over half the sample of countries with general conditions on education spending, suggesting that girls may gain greater access to basic education without explicit Bank conditions. However, the persistence of the gender gap and the precipitous decline in female enrollment in countries such as Mozambique indicate a need for greater Bank attention to eliminating the gender gap quickly and preventing further erosion in the education of girls. 95 Table 5.15. Enrollment Ratios and the Gender Gap in Primary Education (percentage of school age population) 1985 1990 Country Female Male Gender gap Female Male Gender gap Bangladesh 50 70 20 73 84 11 Burkina Faso 21 36 15 28 45 17 Chad 24 63 39 37 82 45 C6te d'Ivoire 62 88 26 59 83 24 Gambia, The 52 84 32 52 77 25 Guinea 23 47 24 24 50 26 India 80 110 30 84 111 27 Kenya 96 102 6 93 97 4 Malawi 52 68 16 62 74 12 Mali 17 30 13 17 30 13 Mozambique 75 97 22 52 75 23 Niger 18 33 15 21 37 16 Nigeria 84 106 22 76 97 21 Pakistan 30 56 26 30 57 27 Senegal 46 67 21 50 68 18 Trinidad and Tobago 97 95 -2 96 96 0 Note: Table includes countries with conditions on education spending, but not specifically on the education of girls. Gender gap refers to the difference between male and female enrollment ratios. Source: World Bank, 1995g. 5.64 Health. A useful indicator of changes in health care is the infant mortality rate, which reacts quickly to such changes. Data for 18 countries before and after a Bank loan or credit that imposed conditions on health care expenditures show that infant mortality declined in all but one (Annex Table 5.8). The same general trend is apparent for a set of countries with no health expenditure conditions, although in these countries mortality rates dropped by slightly less. 5.65 Civil Service Reform: Employment. Recent data on public employment in developing countries are available for only a handful of countries (Annex Table 5.9). The data reveal an almost flat trend in public employment as a percentage of the population or the labor force during the 1980s. Most countries participating in Bank-assisted structural adjustment programs met Bank conditions on retrenchment of public service and containment of public wages, but in some cases the retrenchment was short-lived. In Guinea and Senegal, for example, reductions carried out during the first two years of the program were later reversed. Lessons 5.66 The following lessons emerge from this review of public expenditure reform: * All public expenditure reform programs need to start with an appraisal of the role of government and of the possibilities for shifts away from the public provisioning of goods and services toward more extensive regulatory activities or toward financing that encourages the expansion of private sector activities in infrastructure and social 96 services. The specific mix of government roles will of course differ according to country circumstances. * Public expenditure reforms also need to take account of the sources of financing, whether tax revenues or other forms of domestic or foreign financing. An evaluation of the costs of and constraints on financing and the efficiency costs of different types of revenue is likely to cultivate a culture of accountability-by ensuring that spending programs will be examined in terms of both their social benefits and their social costs. The Bank needs to engage the Fund in discussions on issues concerning the financing of public expenditures. The formal arrangements instituted in June 1995 for collaboration between the two institutions on public expenditure issues should provide a means for such discussions. * Governments need to be persuaded to move away from the culture of "command and control" and to orient themselves to a "service delivery" outlook. They also need to be encouraged to invite local governments, potential beneficiaries, and other stakeholders to participate in discussions on development and spending priorities, particularly in the social sectors. The institutional aspects of the budgetary process therefore require attention (see Chapter 2). * While the Bank's emphasis on capital spending reductions was warranted in countries where large capital projects did not meet the basic needs of the population, capital spending shares (in total spending or GDP) have fallen to low levels in some countries and may be retarding growth. In these countries and in others where large sections of the population lack access to basic services, an increase in investments in basic services may be warranted. * Where budget deficit reductions continue to be a goal, the Bank may need to emphasize cuts in current spending, particularly given its large share in total expenditure. Expenditure efficiency can also be improved through reorientation of current expenditure toward O&M or the social sectors. Where caps on current expenditure need to be maintained, restructuring in favor of nonwage O&M can only be brought about through sharper reductions in subsidies and the wage bill. * The dialogue on public expenditure reform will need to focus on inflexible and unproductive expenditures (such as defense and general subsidies not targeted to the poor). * Because expenditure programs have greater potential than revenue reforms as instruments for reducing poverty, the Bank needs to pursue expenditure solutions for providing the poor with better access to basic services and for improving targeting of subsidies (by restricting them to the type and quality of goods consumed by the poor). (See Chapter 3.) The Bank also needs to seek more innovative solutions to the problems of poverty, such as using education programs to teach poor households about safe water and sanitation practices. And it needs to encourage clients to direct resources to rural areas, where most of the poor live. 97 * To promote equity, basic education and health services should be almost free. For tertiary education and health services, more innovative cost-recovery solutions that also promote equity need to be explored. * The implied inconsistency between the relatively adequate compliance with Bank conditionality and the relatively modest impact of the conditionality on expenditure levels and outcomes can probably be attributed to the Bank's focus on means rather than ends. This focus may invite purely "accounting" responses (budgetary accounting changes and shifts in resources between departments) rather than improvements in efficiency or in service delivery. The Bank therefore needs to establish a more detailed and intensive dialogue on public sector organization and delivery systems and capabilities if shifts in public spending are to be translated into improvements in intermediate and final outcomes. * Focusing on the goals of public expenditure right from the start is likely not only to lead to a better choice of expenditure (or other) instruments, but also to aid implementation. The Bank therefore needs to focus on formulating better performance indicators that are tied to sectoral objectives and to the removal of institutional impediments to public sector performance. Thus in education, in addition to monitoring expenditures at different education levels, the Bank needs to monitor enrollments, student-teacher ratios, graduation rates, and drop-out rates in the short run, and achievement scores, literacy rates, and supplies of skilled professionals in the long run. 99 6. Fiscal Adjustment under SALs and SECALs: An Aggregate View and Conclusions 6.1 Before drawing lessons about fiscal management in SALs/SECALs, it is important first to examine the record on fiscal adjustment-was it achieved, how, and did it last?-and the relationship of fiscal adjustment to other macroeconomic variables.I The uniformly high levels of initial deficits motivated this focus on deficit reduction. The study did not anchor its assessments of fiscal adjustment in independent evaluations of macroeconomic programs (other OED and Bank research reports have carried out such exercises to establish the centrality of fiscal reform). Whether the magnitude of deficit reductions is appropriate is discussed in the context of a solvency and sustainability exercise carried out by Dinh (1995). 6.2 To examine the fiscal record, the study looked at a sample of 26 countries (134 operations in 5 regions) with a long-term history of fiscal adjustment. The sample included 35 percent of countries with adjustment lending and 59 percent of all adjustment loans in the 5 regions. (Annex Table 6.1 reflects the weight of the sample in the universe of SALs/SECALs approved between FY1979 and FY1994; Annex Table 6.2A lists the sample countries and Annex Table 6.2B the sample operations.) Each country's fiscal record is divided into four phases: the preloan period (the four years preceding the first loan); the medium term (years 1-4, beginning with the initial year of the first loan); the long term (years 5-8 following the first loan); and the two most recent years. In addition to being grouped by region, countries are grouped by two lending characteristics-OED ratings of macroeconomic performance (successful or unsuccessful)3 and frequency of adjustment operations (more than three SALs/SECALs qualify a country as loan-intensive)--and three structural characteristics-indebtedness, income level (low or middle), and type of exporter (primary or diversified).4 (Annex Table 6.3 lists the countries in each group.) The aggregative findings reported here draw on Chandra and Anees (1996). 2 Though doubts have been expressed about the usefulness of regional groupings (because country composition is not homogeneous), they do serve a purpose. The tables indicate that the fiscal problems of specific groups defined by "lending" or "economic" characteristics, were in reality confined to a few regions. This is why the regional groupings have been retained. Successful adjusters (as assessed by the Bank and OED) were considered so on the basis of their overall success in macroeconomic reform. Since fiscal stability and sustainability are integral to macroeconomic stability, the successful group would be expected to score higher on fiscal adjustment as well. A finding that that was the case would reinforce the belief that good macroeconomic management overall is necessary for fiscal stability, while the opposite finding would indicate that governments need to do more than achieve external equilibrium to attain fiscal equilibrium and that the Bank needs to take a strong position on the centrality of fiscal stabilization in SALs/SECALs. The choice of these country groups may appear arbitrary. However, Chandra (1996a and 1996b) found that in analyzing a handful of successful and unsuccessful adjusters, the structural characteristics-indebtedness, income level, and type of exporters-were important in explaining fiscal performance. 100 Were Fiscal Deficits Reduced? Not in the Medium Term 6.3 In the medium term the average fiscal deficit for the sample increased from 6.7 percent of GDP to 6.8 percent (Table 6.1). This overall pattern dominated the fiscal performance of countries in a number of the country groups: unsuccessful, loan-intensive, and low-income adjusters and primary and diversified exporters. The pattern of rising deficits also characterized two of the regions, AFR and EAP. In the medium term deficits worsened for almost all the AFR subgroups: successful and unsuccessful adjusters, loan-intensive adjusters, low-income adjusters, and primary and diversified exporters; the only exceptions were the AFR middle- income adjusters and, surprisingly, the severely indebted adjusters (see Figure 6.1 and Annex Table 6.4). Deficits were reduced in the medium term only in the successful and middle-income adjusters and, somewhat unexpectedly, in the severely indebted adjusters; they were also reduced in LAC and MNA and marginally so in SAS. Table 6.1. Average Fiscal Deficits by Country Group and Region (percentage of GDP) Preloan Medium term Long term Most recent two- year period Sample average (26) -6.7 -6.8 -5.2 -4.7 COUNTRY GROUP Loan categories Successful adjusters (15) -6.1 -5.8 -3.8 -3.7 Unsuccessful adjusters (8) -6.4 -7.7 -6.7 -6.1 Loan-intensive adjusters (23) -7.2 -7.3 -5.8 -5.5 Economic categories Low-income (14) -8.2 -8.7 -7.8 -8.4 Middle-income (12) -5.0 -4.6 -2.0 -0.4 Severely indebted (11) -8.0 -7.2 -5.9 -6.9 Primary exporters (10) -7.6 -7.8 -7.2 -8.4 Diversified exporters (14) -6.8 -6.9 -4.6 -2.8 REGION AFR (13) -8.5 -9.0 -7.4 -8.1 EAP (3) -2.4 -3.2 -1.1 0.8 LAC (5) -4.0 -3.0 -1.6 0.6 MNA (2) -8.5 -5.4 -3.0 -2.6 SAS (3) -8.0 -7.9 -7.1 -5.9 Note: Sample includes long-term adjusters (at least eight years of adjustment). Fiscal deficits refer to overall deficits of consolidated central government. Preloan period refers to the four years preceding the first adjustment operation in a country; the medium term refers to years 1-4 beginning with the initial year of the first operation; the long term refers to years 5-8 beginning with the initial year of the first operation; and the most recent period refers to the most recent 2 years for which data are available for the country. Numbers in parentheses refer to the number of countries in each group. Source: IMF, Government Finance Statistics; Bank and IMF staff estimates; World Bank, World Tables, several years. All changes in deficits are in reference to the four-year preloan average unless otherwise indicated. 101 Figure 6.1. Change In Fiscal Deficits by Region and Country Group Regional Averages and Income Categories 9.0 9.0 -------- ---------_______ -______ ___--_ 7.0 - --- 1e.0 3.0 - 2.0 10 Regional Average t.. . eddle Income Loan Categories -1.0 S3.0 2.0 .51.0 0.0 - I -1.0 U -2.0. -3.0 Successful Unsuccessful Loan Intensive Economic Categories 8.0 7.0 so - ----- - - -- - 50 o S4 0 ---- I C Uo ... ---- -2.0 . - ......... .. ... .....- - - --- ...- --- - --.-...- --. Iimary Exporters Diversified Exporters Medium term in Long term Most recent two years Note: Fiscal deficits refer to overall deficits of consolidated central government. Preloan period refers to the four years preceding the first adjustment operation in a country; the medium term refers to years 1-4 beginning with the initial year of the first operation; the long term refers to years 5-8 beginning with the initial year of the first operation; and the most recent period refers to the most recent 2 years for which data are available for the country. Changes are relative to the preloan average. Numbers in parentheses refer to the number of countries in each group. Source: IMF, Government Finance Statistics; Wodd Bank and IMF staff estimates; World Bank, World Tables, several years. 102 Deficits Were Reduced in the Long Term but Many Remained High 6.4 What happened to countries over the long term? Did countries that increased their deficits in the medium term reduce them in the longer term? Were countries that reduced their fiscal deficits in the long term able to sustain the reductions? 6.5 Deficits were reduced in the long term across all regions and country groups (except the unsuccessful adjusters). However, deficits remained high-more than 3 percent of GDP- except in the middle-income adjusters and in EAP and LAC. In AFR the unsuccessful and the primary exporting adjusters had deficits higher than those in the preloan period. (See Table 6.1, Annex Table 6.4, and Figure 6.1). 6.6 Among the country groups, the performance of the middle-income and successful adjusters, diversified exporters, and, surprisingly again, the severely indebted adjusters stands out. The contrasting performance of the successful and unsuccessful adjusters suggests that good macroeconomic performance overall is associated with strong fiscal adjustment. That some of the unsuccessful adjusters (in LAC and SAS) also did well suggests that strong fiscal performance is necessary but not sufficient for good overall macroeconomic performance. 6 Though Most Countries Sustained Reductions in Recent Years, Significant Reversals Occurred 6.7 Though deficit reductions were sustained in the sample as a whole, in EAP, LAC, and MNA, and in several country groups, there were significant reversals as well. The groups that saw reversals were the low-income adjusters, the severely indebted, and the primary exporters. For the low-income adjusters and primary exporters, deficits actually rose above levels reached in the preloan period. What is more, deficits remained well above 3 percent in the sample as a whole, in all country groups (except the middle-income adjusters and diversified exporters), and in AFR and SAS. 6.8 Several themes emerge from this history of 8 or more years of fiscal adjustment. First, fiscal adjustment turned out to be a long-term process with a greater likelihood of success in regions and groups with a 7- to 8-year trend of fiscal deficit reduction. Second, the successful adjusters sustained their fiscal advantage over the unsuccessful group. Third, the middle-income adjusters and diversified exporters were able to pull off a substantial and sustained adjustment. Their performance is in sharp contrast to that of the low-income group and primary exporters. Fourth, the performance of the poorly performing country groups (unsuccessful, low-income, severely indebted, and primary) is wholly attributable to the large and overlapping AFR representation in these groups (see Figure 6.1 and Annex Table 6.4). Fifth, AFR's performance is not uniformly poor; witness the deficit reductions in the region's middle-income and diversified exporter subgroups. The factors underlying poor performance in AFR therefore have to do with low income, severe indebtedness, and primary exports. Finally, while countries in the loan-intensive group have sustained fiscal adjustment, the slow pace of deficit reduction and the 6 These long-term patterns are corroborated by the larger sample of 198 operations in 47 countries. Though the number of countries varies between the preloan period and the long term, changes in deficit levels are all in the same direction for all country groups and regions. However, in AFR significant deficit reductions were achieved not merely by the middle-income adjusters and diversified exporters, but also by the successful adjusters. 103 persistence of high deficits (5.5 percent of GDP on average) in this group is of concern, since these are the countries that have received more than three adjustment loans from the Bank. How Did External Factors Affect Fiscal Outcomes? 6.9 Can fiscal deficits be attributed to terms of trade shocks or movements in world interest rates? How did budgetary interest payments on domestic and foreign debt affect fiscal deficits? Terms of Trade Shocks 6.10 Terms of trade shocks did not hold back fiscal adjustment in EAP, LAC, MNA, and SAS. Fiscal deficits fell as a percentage of GDP in LAC and SAS despite a deterioration in terms of trade over the adjustment period. In MNA fiscal adjustment was strong, possibly indicating some contribution from the improved terms of trade for the region. In EAP, though fiscal adjustment coincided with a gradual strengthening in the terms of trade, the relationship between the two events remains unclear. In AFR, where both terms of trade and fiscal balances worsened, the deficit for the region as a whole did not follow the terms of trade. Only for primary exporters in AFR can terms of trade be more certainly blamed for holding back fiscal adjustment (see Annex Figure 6.1). Foreign Interest Payments 6.11 There is no denying that long-term foreign interest payments have been a heavy burden.7 They rose particularly sharply in the medium term and ranged between 1.5 percent and 3 percent of GDP for the sample as a whole. However, the issue here is not the burdensomeness of these external payments, but whether they have constrained fiscal adjustment in the sample as a whole, particularly among the severely indebted, the low-income adjusters, and primary exporters. (As the study revealed, it is the AFR subgroups within these categories that experienced the greatest fiscal difficulties.) An affirmative answer is not obvious. The foreign interest burden of the severely indebted, the low-income adjusters, and the primary exporters generally remained smaller than that for the sample as a whole. Moreover, other groups with higher foreign interest payments were able to overcome the constraint and reduce their fiscal deficits.9 6.12 While the foreign interest rate is often considered an exogenous factor affecting the deficit, it is important to remember that current stocks of external debt are the result of past policy decisions on foreign borrowing. Moreover, the need for continuous adjustment led to 7 This study focused on long-term foreign interest payments because the data are more consistent and more readily available than short-term debt data. 8 The data here refer to interest actually paid. The interest burden may actually have been higher in countries that ran up arrears on interest payments. These conclusions should not be taken as running counter to the concerns underlying the program of action envisaged by the Bank and the Fund for the heavily indebted countries. That program is concerned with external debt stocks and not foreign interest payments. Moreover, this study focuses only on actual interest payments on long-term external debt whereas the multilateral debt initiative paper is concerned with projections of total debt service due in the next 5 to 10 years. The background papers presented to the Board do not deal in detail with the fiscal performance issues raised in this study. 104 repeated SALs that also contributed to external debt buildup in the poorly adjusting countries. Finally, the public investment expenditures financed by external borrowing did not lead to adequate growth in these fiscally constrained countries (see discussion of expenditure management in Chapter 5). Thus the foreign interest burden represents very real endogenous- not exogenous-solvency and sustainability issues that need to be tackled. Primary Deficits and Total Interest Burdens 6.13 That external factors such as the foreign interest rate need not be an insuperable hurdle in fiscal adjustment is demonstrated by the behavior of primary deficits in EAP and LAC and in several country groups: the successful and unsuccessful adjusters, loan-intensive adjusters, the middle-income group, and the diversified exporters. Almost all of these groups grappled with high levels of total budgetary interest payments (foreign and domestic) that rose steeply over the medium term, the long term, and the most recent period to between 3 and 4 percent of GDP. Nevertheless, all of these groups managed to reduce their fiscal deficits by sharply reducing their primary deficits (see Annex Table 6.5). The middle-income adjusters and the diversified exporters actually achieved primary surpluses; so too did EAP and LAC . The groups that stand out because of their high primary deficits (more than 3 percent of GDP) are again the low- income adjusters, severely indebted adjusters, and primary exporters. The countries in these groups are drawn largely from AFR. The primary deficit picture therefore suggests that the fiscal crisis was primarily a product of fiscal mismanagement rather than of terms of trade shocks or foreign interest payments. 6.14 The AFR story of fiscal problems (among the primary exporters, the severely indebted, and the low-income adjusters) warrants a closer look. The study indicates (see para. 6.10) that terms of trade were a factor in constraining fiscal adjustment among the AFR primary exporters. But terms of trade appear not to have constrained the LAC primary exporters from adjusting fiscally. Is indebtedness then to blame for the lack of fiscal adjustment in AFR? The data show that the severely indebted subgroups in both LAC and MNA carried out significant fiscal adjustment throughout the period. Is the AFR story of fiscal difficulties attributable then to low income? Perhaps, because there is a great deal of overlap between the low-income countries and the primary exporters and severely indebted adjusters in AFR. Low income has been a severe handicap because it constrains revenue, but in addition economic management has not succeeded in adequately increasing growth in low-income AFR. Here the comparison with SAS (the other region dominated by low-income countries) is instructive: the SAS low-income countries were able to diversify their export base, keep debt levels down, and also grow during the period under study. Thus, while low income was a constraint in AFR, so too was the quality of economic management. How Was Fiscal Adjustment Achieved? 6.15 How countries addressed their fiscal deficit problems-with tax increases, expenditure cuts, or a combination-matters because the choice affects incentives to save, invest, and produce. Over the long term and during the most recent period fiscal improvements were associated with revenue increases (see Annex Table 6.6). More than half the instances of fiscal 10 The broader sample of 47 countries studied by Chandra and Anees also confirms that LAC primary exporters were able to adjust fiscally despite terms of trade deterioration. 105 improvement for the country groups were related to increases in revenue effort. Both instruments, revenue increases and expenditure cuts, were used by the middle-income and diversified exporter groups in the long term and most recent period and by the severely indebted countries in the long term. Expenditure increases were responsible for the worsening of the fiscal deficit in the medium term for the sample as a whole. Worsened fiscal deficits were generally associated with expenditure increases alone (with the only exception the unsuccessful adjusters in the long term, where revenues collapsed relative to preloan levels as expenditures increased). Only in two cases (in the medium term for the middle-income countries and during the most recent two years for the unsuccessful adjusters) was expenditure reduction associated with fiscal improvement. Among the regions, MNA was successful in cutting expenditure, making cuts as large as 4 percent of GDP and maintaining them throughout the adjustment period. 6.16 The data indicate much uniformity across country groups and regions and over time in current and capital expenditure strategies. All country groups save one increased current expenditures, and all groups save one pruned capital expenditures, relative to the preloan period. For the sample as a whole and for all country groupsI current expenditures were higher throughout the period relative to preloan levels (Annex Table 6.7). And for all groups (except the unsuccessful adjusters in the medium term) capital expenditures were lower relative to preloan levels throughout the period (see Chapter 5). Fiscal Deficits and Other Macroeconomic Variables 6.17 Fiscal deficits, if they can be financed, do not represent an immediate problem. Problems arise because of the economic disequilibria that fiscal deficits cause: inflation, credit shortages, distorted interest and exchange rates, current account imbalances, and a large public debt. All are common outcomes of fiscal deficits, and all can crowd out private investment and hurt economic growth. But because many other factors can also cause these macroeconomic problems, only associations between fiscal deficits and growth and macroeconomic indicators are explored here, not causal relationships.12 6.18 For the sample as a whole the correlation between central government fiscal balances and the GDP growth rate is fairly strong (Annex Table 6.8). The association was also strong for AFR, EAP, and LAC, for the better fiscal performers (the middle-income adjusters and diversified exporters), and for the weak fiscal performers (the low-income group and the primary exporters). Although growth performance was uneven across regions, in general, growth accelerated as fiscal deficits improved in the long run. Thus the data support the view that while other conditions are also necessary for growth, good fiscal management helps. For inflation the association with deficit reduction was ambiguous. It was fairly pronounced in EAP, MNA, and SAS but was either weak or ran counter to the expected result (that large fiscal deficits are associated with high inflation) in the sample as a whole, in AFR and LAC, and in the country groups. For external balances the sample as a whole indicates a strong association with fiscal deficits. So do the data for all regions (except LAC) and for the four country groups. In the The sole exception was the middle-income adjusters in the most recent period. 12 The causal relationships have been explored in depth in other OED publications (Jayarajah and Branson 1995), in World Development Report 1988, and in Easterly, Rodriguez, and Schmidt-Hebbel (1994). 106 medium and long term improvements in fiscal deficits were associated with improvements in current account deficits in all five regions. Creditworthiness, Fiscal Solvency, and Sustainability 6.19 Fiscal deficits were reduced to some degree across all country groups (except the OED- rated unsuccessful adjusters) and all regions in the long term, but were they reduced enough? Fiscal deficits can be a problem if they raise doubts about creditworthiness. In the 1980s the assessment of a country's creditworthiness rested mainly on its debt and debt servicing capacity at the time of the assessment. Today lenders weigh fiscal solvency and sustainability more heavily because they are good predictors of future indebtedness and growth (see Box 6.1).13 Since the private sector's decision to invest is based on the marginal return on alternative financial assets, the sustainability of fiscal deficits plays an important role in investment decisions. Unsustainable deficits are perceived as signaling future tax increases and consequently capital flight, or deficit financing through money creation and consequently higher inflation and devaluations. In either case, unsustainable deficits may lead to a financial or balance of payments crisis. These arguments point to the importance of achieving sustainable fiscal deficits. 6.20 Dinh (1995) has estimated the fiscal adjustments needed during 1989-93 to restore fiscal solvency and sustainability in specific countries. His estimates, set out in Annex Table 6.9, suggest that fiscal adjustments, some larger than 3 percent of GDP, continue to be needed in all regions save EAP (the Philippines is the only country in EAP requiring solvency adjustments). 6.21 Traditional creditworthiness analysis looks at the external debt to GDP ratio. For the sample in this study the ratio rose from 37 percent in the preloan period to more than 50 percent in the medium and long terms, to about 64 percent in the most recent period. Among country groups, external debt ratios have risen to between 48 percent and 102 percent of GDP in recent years. The primary exporters, the severely indebted, and the low-income adjusters had the highest debt ratios in the most recent period, ranging from 93 percent to 101 percent. Among regions, recent debt ratios have ranged between 23 percent in EAP and 96 percent in AFR. The largest growth of debt since the preloan period took place in AFR. 6.22 The rapid buildup in debt in most regions and Dinh's 1995 estimates of the fiscal adjustments required to restore fiscal solvency and sustainability in a range of Bank client countries suggest that the fiscal deficit reductions of the 1980s and early 1990s have not been sufficient for the majority of countries studied. 13 Solvency is defined as the point at which the intertemporal budget constraint is satisfied (that is, the present discounted value of the primary balance is equal to the initial level of public debt). It is measured at a point in time. Sustainability refers to the discounted value of a country's public sector debt approaching zero as time progresses. It is measured over a period of time. 107 Box 6.1. Determining the Solvency of the Public Sector Three approaches are used to evaluate the solvency of the public sector. The accounting approach (Buiter 1983, 1985, 1990; van Wijnbergen 1989; Dinh 1995) defines a sustainable deficit level as one that can be financed by keeping the debt to GDP ratio constant under reasonable assumptions about growth, real interest rates, and inflation. The difference between actual and sustainable deficit to GDP ratios is then calculated. Dinh calculates the fiscal adjustments required over time to achieve solvency and sustainability. He argues that with the introduction of dynamic components, such as expected real interest rates and real GDP growth rates, the conventional debt to GDP ratios, which are static components, are not sufficient to assess fiscal sustainability. Dinh argues that it is important to make a distinction between solvency and sustainability and to calculate the degree of fiscal adjustment needed to achieve both. The solvency adjustment is used to assess the potential of the public sector to repay debt in the future, while sustainability adjustment looks at the current debt payment capacity, taking into account current and potential government measures in the areas of the exchange rate and the financial sector. The neoclassical approach (Hamilton and Flavin 1988; Grilli 1989; Wilcox 1989; Buiter and Patel 1990) compares the growth rate of the public debt to GDP ratio with that of the real interest rate. A country is deemed solvent if the public debt to GDP ratio grows no faster than the real interest rate. The price-oriented approach focuses on prices rather than quantities of debt. Under this approach the discounts on public debt paper are taken as indicators of the public sector's solvency. Recommendations The study suggests that greater fiscal discipline remains a major objective for adjusting countries. The first step is to redefine the role of the public sector-an important determinant of the level of the public finances. How much fiscal discipline is required needs to be systematically identified for each country on the basis of estimates of the sustainable deficit. Looking at fiscal deficits in this context also places fiscal reform in perspective-as only one aspect of a larger adjustment program of interrelated policies. Estimates of sustainable deficits ought to be routinely incorporated into Bank economic and sector work and should underlie adjustment lending proposals for most countries. Following are the study's more specific recommendations. Regional and Country Performance 1. Though fiscal deficits have declined on average, the Bank needs to work toward greater fiscal discipline in AFR and SAS and among low-income, primary exporting, and severely indebted adjusters (particularly in AFR). Contrary to popular belief, fiscal mismanagement, not external shocks, has been largely responsible for the poor fiscal performance in these groups; the only exception is the primary exporter group in AFR. Fiscal Deficits 2. The study recommends that: * Fiscal accounts be expanded to include the broadest possible coverage of the deficit and their transparency improved. This implies a need to incorporate state, local, and 108 municipal accounts into the fiscal accounts and to track quasi-fiscal deficits and work out the implications of implicit taxes and subsidies. * The appropriate measure of the deficit for tackling the problem at hand be identified. In some countries, this measure may be the operational or the structural deficit; in others, public savings may need to be the focus; in yet others, primary deficits may need monitoring. * Fiscal accounts be better integrated into the macroeconomic framework of analysis. This will make it easier to estimate the sustainable deficit, identify tradeoffs among acceptable levels of borrowing, inflation, interest rates, and other macroeconomic variables, and set quantitative targets for achieving a sustainable deficit and related macroeconomic outcomes. Analysis 3. The study calls for increasing transparency in the analysis and design of fiscal programs, through: * More explicit analysis (in alternative scenarios) of relationships between fiscal and other macroeconomic variables and of alternative revenue and expenditure packages. * Identification of the revenue effects of alternative fiscal policies. * Analysis of implicit taxes and subsidies. * The generation of better, more timely, and more complete fiscal data and the development of more widely accessible data systems in the Bank's client countries. PRs need not document the detailed results of all such exercises, but they do need to indicate that fiscal programs have taken such alternatives into consideration. Conditionality 4. Renewed attention to some well-known conclusions about what is required for the design of effective conditionality is likely to improve implementation and outcomes: * Couching conditionality in more precise terms. * Following through on the reforms in one SAL with related reforms in another. * Making implementation of new laws-rather than their presentation to the legislature-a condition of effectiveness or tranche release. * Moving beyond studies (which should certainly precede reform) into actions as the object of conditionality. * Improving the timing of conditionality. 109 Instrument Choice: Revenue Increases versus Expenditure Cuts 5. The study suggests that, in addressing the choice of revenue and expenditure instruments for reducing fiscal deficits, the Bank needs to focus on the current components of the budget. It needs to emphasize: * Revenue expansion through broadening tax bases and simplifying and improving taxpayer compliance and tax administration. * Cuts or restraint in current expenditure combined with restructuring. A fresh assessment of the role of government is likely to suggest areas for reducing or redeploying expenditures. Most countries have already made large cuts in capital expenditure, however, so the Bank needs to ensure that governments keep their sights focused on maintaining or enhancing infrastructure for the delivery of basic services. Tax Reform * The Bank needs to emphasize the proper sequencing of tax reform. If trade taxes contribute significantly to revenue, nontrade tax reform needs to precede trade tax reform. Reforms in tax administration should precede or accompany structural reform of tax policy. * Conditionality needs to be more precisely structured, especially for nontrade tax and tax administration reform. * Building better performance indicators into reform programs from the beginning would benefit design, implementation, and evaluation and focus attention on the goals of reform. For nontrade tax and tax administration reform, SALs/SECALs need to build in performance indicators on marginal effective tax rates and compliance indicators based on costs and outputs. * Nontrade tax reforms need to be anchored in better analysis (for example, estimates of marginal effective tax rates). Some reforms, such as the reduction of investment incentives, need to be pushed further. Other areas of reform likely to assume greater importance in the future need to be emphasized, such as property, environmental, and financial sector taxes. In yet other areas-the use of schedular taxes and withholdings and the treatment of capital income-possible reforms need to be examined. Bank-Fund collaboration is important in evaluating cross-country experience, "best practice" in reform, and special topics of future importance in tax policy. * Certain issues need to be emphasized in tax administration reform: a conducive legal environment, taxpayer education, information systems (used by tax departments), audits, withholding and presumptive taxation, and improvements in VAT administration. * In trade tax reform the design of Bank loans needs to take better account of revenue effects. Bank loans also need to ensure that sales ands excise taxes apply with equal 110 force to imports and competing domestic production. Attention to allowing appropriate adjustment periods and compensating losers is also important. Public Expenditure Reform * SALs need to keep the ultimate goals of expenditure reform in mind. Focusing on the goals right from the start is likely not only to lead to a better choice of expenditure (or other) instruments, but also to aid implementation. The Bank needs to promote the development of appropriate performance monitoring indicators that focus not only on the level and composition of expenditure but also on intermediate (short-run) outputs and final (long-run) outcomes. * Where cutting budget deficits continues to be a goal, the Bank may need to emphasize reducing current expenditures, particularly given their large share in total expenditures. Restructuring current expenditures in favor of nonwage O&M and social sectors is only possible through sharper reductions in subsidies, public employment, and the wage bill. * While the Bank's emphasis on capital spending cuts was warranted in countries where large capital projects did not meet the basic needs of the population, capital spending shares (in total spending or GDP) have fallen to low levels in some countries and may be retarding growth. In these countries and in others where much of the population lacks access to basic services, an increase in basic investments may be warranted. * The dialogue on public expenditure reform will need to focus on inflexible and unproductive expenditures (such as defense and subsidies) and reform of budgetary processes. Governments need to be persuaded to move away from a "command and control" culture and toward a "service delivery" outlook. They also need to be encouraged to involve local governments, potential beneficiaries, and other stakeholders in discussions on development and spending priorities, particularly in the social sectors. * Expenditure programs have greater potential than revenue reforms as instruments for reducing poverty. Thus the Bank needs to pursue expenditure solutions for providing the poor with better access to basic services and for improving the targeting of subsidies (by restricting subsidies to the type and quality of goods consumed by the poor). The Bank might also seek more innovative solutions to the problems of poverty, such as using education programs to teach poor households about safe water and sanitation practices. The Bank also needs to encourage clients to direct resources to rural areas, where most of the poor live. * To promote equity, basic education and health services should be almost free. For tertiary education and health services, more innovative cost-recovery solutions that also promote equity need to be explored. * The neglect of the ends of reform can be traced in part to the lack of fiscal data in most of the Bank's client countries. But at the same time that neglect bears some Ill responsibility for the absence of data. 14 If there were more concern about access to services by specific population groups, there would be more focus on generating and disseminating the public expenditure data needed to make the proper assessments. Very few SALs/SECALs have tried to promote the collection and dissemination of information that is indispensable for improving the design, implementation, and evaluation of adjustment policies and for enhancing transparency and accountability in government. Some Larger Issues 6. The study also points to some larger issues that the Bank needs to consider and address: * The Bank has made a beginning in formulating sectoral performance monitoring indicators. This study has drawn attention to the need to extend that exercise to all aspects of fiscal reform. Performance indicators are best framed if they start from the perspective of the specific goal that individual reforms address. Keeping the program's sights on the final outcome is likely not only to suggest better performance indicators for the fiscal instrument identified for reform, but also to prevent the final outcome from being forgotten during implementation. * A difficult issue is the role of government relative to the private sector. Should the government continue to provide basic services? Should it provide financial assistance for private provision? In which areas should it withdraw to a regulatory role? The answers to these questions will differ from country to country. * The overall framework for public expenditures needs to be clearly drawn in every case. Years of effort by the Bank in Bank-borrower expenditure dialogues, public sector expenditure reviews, and the design of expenditure conditions have failed to yield commensurate results. It may be time for the Bank to look at expenditure issues from a different perspective. It needs to look systematically at how to maintain overall budget discipline across core and line ministries and at different levels of government, encourage local beneficiaries and governments to participate in formulating expenditure proposals, ensure technical efficiency in the choice of expenditure programs, build in incentives to ensure an efficient civil service, and ensure transparency and accountability to keep governance open and honest. The literature on such institutional issues offers no immediate solutions, but the Bank has made a beginning in discussing these questions. The Bank's research efforts on policy analysis in these areas need to be supported and expanded. * The Bank needs to evaluate its role in the design of fiscal programs. The Fund's lead role in advising clients on fiscal matters needs to be respected; in addition, the Fund has superior resources and expertise and a long tradition of involvement in the area. However, recognizing the strong convergence of client, Fund, and Bank interests in the fiscal area, the Bank needs to do more searching analyses of fiscal issues particularly in adjustment areas where the Fund may not be directly involved. For example, the Bank needs to consider the relationships between tax policy reforms 14 The data issue was also flagged in World Bank 1992b. 112 undertaken in adjustment programs and other structural reforms (as in private sector development, financial and capital market reforms); between the size and role of the government and the level of the public finances; between the financing of the budget and the level and composition of public expenditures; and between tax and expenditure decisions and poverty programs. The Bank needs to retain the analytical capacity to advise staff on fiscal and adjustment issues, to identify best practice, and to analyze new fiscal policy issues that are likely to assume importance in the future (financial sector taxation, environmental taxes). In this respect, the Bank's Policy Research Department, which withdrew from tax analysis in October 1994 to concentrate on expenditure issues, needs to correct the balance. Such a capability will enhance Bank-Fund coordination and strengthen dialogue on fiscal issues. As a first step, the President's Reports need to report on the Fund's principal actions so that the Bank's contribution to adjustment is more transparent. The Bank and Fund are exploring new and existing channels for improving coordination in the preparation of country macroeconomic frameworks (Policy Framework Papers, Public Expenditure Reviews, Bank Country Assistance Strategy papers, Fund Country Assistance Briefs). The Bank and Fund also need to collaborate in improving their own fiscal databases and client fiscal data systems. 113 Annexes 115 Annex 2.1. Description of SAL/SECAL Sample 1. Country composition by type of conditionalities (ALCID database) 23 out of 40 countries with specific budget deficit conditions 11 out of 25 countries with budgetary process reform conditions 10 out of 23 countries with tax conditions only 11 out of 48 countries with expenditure conditions only 6 out of 22 countries with tax and expenditure conditions 11 out of 13 countries with conditions related to IMF 20 out of 48 countries with other conditions 2. Regional composition (includes only countries with adjustment lending) Number of countries Number ofsample Percentage of Region in region countries countries in region LAC 20 8 40 AFR 37 9 24 ECA 12 4 33 EAP 7 2 29 SAS 5 2 40 MNA 5 2 40 Total 86 27 31 3. Share of total SAL/SECAL lending by region, FY 1979-94 Total lending Lending to countries Share in Bank lending Region (US$m) selected (US$m) to region (%) LAC 10,126 5,816 57 AFR 10,214 5,673 56 ECA 5,501 4,331 79 EAP 3,308 2,652 80 SAS 2,600 1,650 63 MNA 2.,565 L,84 72 Total 34,314 21,971 64 116 4. Share of total SAL/SECAL operations by region, FY 1979-94 Total SAL/SECAL Share of total Region operations Sample selected SALs/SECALs (%) LAC 58 28 48 AFR 128 54 42 ECA 23 16 70 EAP 15 10 67 SAS 16 12 75 MNA 12 9 75 Total 252 129 51 5. Top SAL/SECAL borrowers Eleven out of 15 top SAL/SECAL borrowers during FY 1979-94 were selected (Mexico, India, Colombia, and Algeria were not included). 6. Income classifications Twelve low-income and 15 middle-income countries were selected. Low-income and middle- income country classifications are based on 1993 GNP per capita income data as set out in OP 3.10. 7. Coverage in PCRs/ICRs/PARs Ninety-seven operations were covered by PCRs/ICRs/PARs. 8. OED ratings Sample includes the following: 59 satisfactory operations in 22 countries 25 unsatisfactory operations in 16 countries 45 unrated operations in 21 countries 117 Annex Table 2.2. SALs/SECALs with Budget Deficit/Savings Conditions (percent) Focus of budget deficit condition General conditions: Budget deficit consistent macro- No conditions condition Overall Public Primary economic on budget specfled deficit savings deficit framework deficit/savings Region (a) (b) (C) (d) (e) (f) LAC 43 25 67 8 43 14 AFR 15 88 0 12 19 66 ECA 50 100 0 0 19 31 EAP 0 0 0 0 0 100 MNA 22 100 0 0 11 77 SAS 25 100 0 0 8 67 Total sample 26 70 24 6 20 54 Note: For each region shares in columns a, e, and f add up to 100 percent, as do shares in columns b, c, and d. Source: President's Reports on SALs/SECALs. 118 Annex Table 23. Aspects of Conditionality Presentation Vague, rather than dfficult to Studies rather implementation Too many Election monitor/ Country Loan than actions oflaw conditions cycle implement LAC Costa Rica SAL I (1985) / SAL II (1989) / V- SAL IIIa(1993) - Argentina ASAL (1986) / / BSL (1988) Honduras SALI(1989) / / - SAL 11(1991) 9/ V v v- Venezuela SAL (1989) / / Uruguay SALI(1987) / - SAL 11(1989) AFR Ghana RIG I, 11(1983, - 1985) Export Rehab (1984) Private Inv (1991) Gambia, The SAL 11 (1989) Kenya SAL11(1983) 9' ASAO (1986) EDAC (1992) Malawi ASAL (1990) SALI(1991) Benin SALII(1991) Cameroon SAC (1994) / - Central SAL (1990) / V- Africa Republic C6te dIvoire SAL 1(1982) / - SAL 11(1984) d- SAL 111(1986) d- HRD (1992) Guineau Private Inv Prom - (1990) Senegal SAL 11(1986) SAL 11l (1987) SAL IV (1990) / V- Tanzania FSAC (1992) - Togo SAL IV(1991) - Uganda 3rd Recons (1984) / ERP (1988) - Zimbabwe SAL (1992) 119 Annex Table 2.3. Aspects of Conditionality (continued) Presentation Vague, rather than difficult to Studies rather implementation Too many Election monitor/ Country Loan than actions of law conditions cycle implement ECA Hungary ISAL (1988) SAL(1990) Bulgaria SAL (1992) Turkey SAL (1980) SAL 111 (1982) SAL IV(1983) SAL V (1984) Poland SAL (1991) EAP Philippines ERP (1987) FSAP(1989) MNA Tunisia ASAL (1987) / ITPAL (1987) / SALI(1988) Morocco ITPAL (1986) SAL(1992) SAS Pakistan Import Credit / (1981) 1 SAPI (1982) / / EDL (1986) FSAL (1989) Bangladesh IPC (Twelfth) I (1984) (Thirteen) (1986) a Costa Rica's SAL III was canceled and has no PCR. However, SAL III incorporated conditionality on presentation of a law to Parliament. Source: Project Completion Reports and Performance Audit Reports. u 121 Annex Table 3.1. Changes in Structure of Central Government Current Revenues, 1980-93 (percent) Total current Total tax revenue to GNP revenue to GNP Total nontax ratio ratio revenue to GNP ratio 1980 1993 1980 1993 1980 1993 Income categorya Total low and middle income 21.8 23.2 17.2 17.5 4.5 5.7 Low income 18.4 20.0 15.1 16.1 3.1 3.9 Middle income 23.5 25.1 18.3 18.3 5.2 6.7 Lower middle 20.7 23.4 16.8 17.1 3.9 6.2 Upper middle 27.5 27.6 20.4 20.0 7.1 7.5 High income 34.2 35.0 27.4 30.2 6.8 4.9 Region AFR 21.0 23.5 17.3 18.7 3.4 4.8 EAP 19.4 19.5 16.8 16.2 2.5 3.3 SAS 14.1 15.5 12.2 12.3 1.8 3.2 ECAb 22.3 18.7 17.7 14.0 4.6 4.7 MNA 35.9 33.3 18.7 18.8 17.2 14.6 LAC 21.3 22.3 18.3 18.2 3.0 4.0 a Classification of countries is the same as in World Development Report 1995, and is based on GNP per capita as of April 30, 1995. b Data for Europe and Central Asia refer only to Turkey; data for other countries in the region were unavailable. Source: World Bank, World Development Report 1995 (New York: Oxford University Press, 1995). 122 Annex Table 3.2. Changes in Tax Structure by Income Category and Region, 1980-93 (percent) Taxes on goods and Income to Social security services to Trade taxes Other taxes to total tax to total tax total tax to total tax total tax revenue ratio revenue ratio revenue ratio revenue ratio revenue ratio 1980 1993 1980 1993 1980 1993 1980 1993 1980 1993 Income category' Total low and middle- 30.9 29.9 6.2 5.7 27.4 33.8 31.0 25.8 4.5 4.8 income Low-income 21.5 23.7 2.6 1.8 32.6 37.8 40.0 33.6 3.4 3.2 Middle income 37.3 34.4 8.2 8.0 24.4 31.3 25.0 20.5 5.1 5.9 Lower-middle 33.7 31.7 7.9 6.2 23.1 32.1 29.1 24.2 6.2 5.8 Upper-middle 42.8 38.4 8.7 10.8 26.3 29.9 18.8 14.9 3.4 6.0 High income 41.4 38.2 23.5 23.4 27.0 27.8 4.3 5.5 3.9 5.1 Region AFR 34.3 30.2 2.1 0.9 22.1 31.3 39.1 34.8 2.4 2.8 EAP 40.9 42.8 0.3 1.8 35.2 32.3 22.0 19.4 1.7 3.8 SAS 15.4 18.0 0.0 0.0 41.0 46.2 40.3 32.8 3.2 3.0 ECAb 61.8 47.6 0.0 0.0 24.8 43.3 7.6 5.6 5.8 3.3 MNA 28.9 33.6 8.2 7.3 14.3 19.1 39.5 30.9 9.1 9.1 LAC 24.8 22.5 16.8 16.2 32.4 41.0 19.5 13.9 6.6 6.5 a Classification of countries is the same as in World Development Report 1995, and is based on GNP per capita as of April 30, 1995. Data for Europe and Central Asia refer only to Turkey; data for other countries in the region were unavailable. Source: World Bank, World Development Report 1995 (New York: Oxford University Press, 1995). 123 Annex 3-3. The Instruments of Tax Reform 1. Few tax reforms begin with a tabula rasa-most are anchored in existing tax structures that are riddled with distortions and opportunities for evasion. The diversity of baseline conditions makes it difficult to devise a set of generally acceptable rules for ensuring that the Bank's tax reform principles are translated into practice. The need to grapple with existing tax structures means that reform is generally implemented incrementally rather than comprehensively. Notwithstanding these difficulties, it is possible to provide a general set of guidelines that, modified to take account of country circumstances, could help implement the Bank's principles of tax reform. Indirect Taxes 2. Incremental reform of indirect taxation usually requires that countries broaden their indirect tax bases, eliminate cascading by providing credits for taxes paid on inputs, and reduce the range of tax rates. To ensure an adequate tax base, especially in an inflationary environment, countries need to move from specific to ad valorem taxation. When possible, indirect tax reform should aim at shifting taxation away from production toward consumption. If a more substantial overhaul of the indirect tax system is possible, countries might move to a consumption-based VAT that includes few zero-rated commodities apart from exports. Designed correctly, a VAT avoids taxation of production, limits the potential for evasion, and provides a relatively large and predictable source of revenue. 3. Policy reform of trade taxation requires actions in four areas. First, all explicit export taxes and such implicit taxes as export bans, multiple exchange rates, and export retention schemes need to be eliminated, and duty drawbacks on imported inputs used in export production need to be introduced. Second, QRs need to be converted into transparent tariffs, and the level and dispersion of tariffs must be reduced. Third, trade and domestic taxes need to be harmonized to ensure that domestic taxes do not undo (or do not exaggerate) the degree of protection intended by the trade tax system. To do so, domestic excise and sales taxes must be designed to apply to both imports and competing domestic production. Fourth, domestic indirect taxes may need to be adjusted to compensate for any revenue losses associated with trade tax reforms. To achieve these harmonization and revenue goals, tariffs should be designed with only protective purposes in mind, and the sales tax/VAT should be used only for revenue purposes. Because a tariff raises the producer price over the world price, it serves as a subsidy to domestic producers, and since the sales tax/VAT, levied on the tariff-inclusive value of imports, raises the import price to users above the world price, it constitutes a tax on domestic users. As long as the sales tax/VAT applies at the same rate to domestic production and imports, it will not discriminate among alternative sources of supply and therefore will not encourage an inefficient production structure. Income Taxes 4. Guidelines for income taxes are problematic because the economic effects of statutory changes are difficult to predict. On efficiency grounds, Lessons presses for base broadening and uniform treatment along the following lines: * Ensure that all forms of business proprietorships, partnerships, and corporations (including state enterprises) face the same business tax statutes and that all individuals face the same income tax statutes. These guidelines draw heavily on Lessons and on the tax reform experience of 10 developing countries described in Thirsk (1991). 124 * Ensure uniformity in the statutory tax treatment of financial instruments whenever possible. * Broaden the income tax base, eliminate schedular taxation (but see below for another point of view), and eliminate deductions for individuals unless the deductions represent a cost in earning income. * Eliminate most investment and savings incentives. * Index nominal magnitudes in the tax code to neutralize the impact of inflation. * Equalize marginal effective tax rates across all forms of real investment, savings, and intermediation. * Harmonize tax codes across countries by ensuring that the tax rates faced by residents are close to those faced by foreign investors. To preclude fiscal competition among home and host countries, avoid subsidizing the output of foreign-owned subsidiaries and investments, both directly and indirectly (through tariff protection or lower sales taxes).2 Wealth and Property Taxes 5. Wealth and property tax bases should be defined broadly and adjusted for inflation. Complicated tax structures should be avoided to lessen the likelihood that effective tax burdens will vary across types of property or wealth. Equity Improvements 6. Increased taxation of luxuries, increased reliance on property (especially land) taxation, and a rise in income tax thresholds provide some scope for redistribution of income. Revenue Constraints 7. The effects of changes in specific tax instruments on revenue levels and variability need to be taken into account in designing reforms. Some of the tradeoffs and interactions among tax instruments that have revenue effects have already been discussed. Tax revenue is also affected by two other common problems--evasion and capital outflows. Increased compliance efforts or a shift to a VAT can help redress tax evasion, which is facilitated when capital is free to flow across borders into tax havens, thereby eroding the income tax base. For example, the U.S. practice of not taxing nonresident interest income and of not providing information on interest income earned by residents of other countries may impose significant income tax revenue losses on some developing countries and force them to abandon efforts to tax the worldwide income of their residents. Capital-originating countries that rely on their income tax bases but have little information about their residents' income from foreign sources can try to shift away from income taxation toward a tax with a less mobile base, such as a VAT. 2 For a fuller discussion of harmonization of tax codes across countries, see Watson and Thirsk (1995). 125 Annex Table 3.4. Content of Nontrade Tax Reform Conditionality (percent) Feature of tax Frequency of tax system Frequency of all system Conditionality conditionality conditionality Indirect taxation Adopt VAT 20 Rationalize tax base 35 Rationalize tax rates 28 Impose compliance measures 17 All conditions 34 Individual taxation Broaden tax base 42 Rationalize tax rates 23 Harmonize rate structure with other tax systems 15 Adjust for inflation 5 Impose compliance measures 15 All conditions 23 Business taxation Broaden tax base 11 Rationalize deductions 12 Rationalize tax rates 22 Harmonize rate structure with other tax systems 14 Rationalize incentives 24 Adjust for inflation 4 Impose compliance measures 13 All conditions 33 Wealth/property Broaden tax base 43 taxation Revalue tax base (including 57 inflation adjustments) All conditions 5 Taxes on financial Reduce or eliminate taxes on 64 transactions (stamp financial transactions taxes, GRT) Harmonize tax rates with other 36 tax systems All conditions 5 Source: ALCID. 126 Annex Table 3.5. Content of Trade Tax Conditionality by Type and Region (percent) Conditionality AFR LAC SAS EAP MNA All regions Export conditionsa 24 29 9 38 0 24 Reduce export taxes 55 78 100 60 - 65 Introduce duty drawbacks/reduce subsidies 45 22 - 40 - 35 All export conditions 100 100 100 100 100 100 Import conditions 61 58 64 46 57 57 Reduce quantitative restrictions 32 33 28 33 25 32 Reduce level and dispersion of tariffs 32 39 28 50 25 35 Reduce exemptions 18 11 15 - - 13 Reduce number of categories 11 11 - - 25 13 Rationalize rates 3 6 29 17 25 7 All import conditions 100 100 100 100 100 100 Harmonization conditions 15 13 27 15 43 19 Equalize domestic tax on imports and domestic output 44 71 50 60 60 57 Introduce revenue raising reforms 30 13 35 20 20 22 Introduce customs reform 26 16 25 20 20 21 All harmonization conditions 100 100 100 100 100 100 Note: Based on sample of 87 loans covering 23 countries. a Includes export subsidy conditions for CMte d'Ivoire and Costa Rica. In the absence of the exchange rate instrument, CMte d'Ivoire adopted a uniform export subsidy and a uniform tariff to simulate a devaluation. Costa Rica eliminated export subsidies as devaluation occurred. Source: President's Reports. 127 Annex Table 3.6. Countries by Degree of Implementation of Nontrade Tax Conditionalities Fully implemented Partially implemented Failed (52%) (34%) (14%) Peru Jamaica Jamaica Dominica C6te d'Ivoire Philippines Korea, Rep. of Hungary Bangladesh Mauritius Senegal Argentina Tunisia (2) Pakistan Guinea-Bissau Malawi (3) Guyana Philippines Uruguay Jamaica Central Africa Republic (2) C8te d'Ivoire Gambia, The Benin Honduras Mexico Gabon Morocco Algeria Lao PDR Sierra Leone Source: Project Completion Reports/Performance Audit Reports. 129 Annex 4.1. Checklist for Bank Staff in Formulating Strategies for Tax Administration Reform General issues 1. Determine structure of existing tax system with brief description of each tax exemption and other information on working of system and its economic impact. 2. Assess the level of taxation and tax trends. Measure the tax to GDP ratio and elasticities of total revenue from the principal taxes with respect to GDP. To identify points of weakness in the tax system and tax administration, decompose the tax to GDP elasticities into elasticities of tax collection to tax due (leakage results from evasion and failure to monitor), tax due to assessed tax base (leakage results from exemptions), and tax base to the proxy tax base (leakage results from identification problems). 3. Determine tax obligation calendar (deadlines for filing and payment). 4. Determine the costs of tax administration by measuring the proportion of each type of tax that is spent on administering the tax (generally 3-4 percent of the income tax, 1 percent of customs duties); determine the costs of enforcing taxpayer compliance (where studies are available for other countries, compare results to determine efficiency of the system relative to other systems). 5. Assess the cultural, economic, and legal environments. * Cultural environment: What is the level of literacy? What social ethics govern the society? How much corruption is there in public life? How prevalent is economic crime? How are taxes perceived by the public-as oppressive or fair? How are tax evaders perceived? * Economic environment: What is the level and structure of income and employment? What is the size of the organized sector? What are the main forms of business organization? How high is inflation? How open is the economy? * Legal environment: How simple, efficient, and prompt are the judicial system and judicial procedures? What laws and customs govern privacy? What standards of evidence govern tax matters? 6. Identify statistical data. * Identify regularly produced statistical data and dates of availability. * Construct table of economic activities used and the methods for classifying them. This checklist is designed to provide broad guidelines, not to serve as a succinct questionnaire. It draws on Vazquez-Caro, Reid, and Bird (1992) and the IMF's questionnaire "Information Requirements for the Analysis of Tax Administration." 130 Specific issues 1. Compliance Are tax laws, forms, and procedures for registration, filing, and tax payment clear and simple? Are indicators of compliance costs for taxpayers covering the quantity and quality of services in use? (Some quantity indicators: the number of forms, the number of items per form, length of time required to fill a form. Possible quality indicators: the number of tax forms automatically distributed to taxpayers, the number of taxpayer inquiries received by the agency, the average time that elapses before inquiries are answered.) * Are tax office facilities user-friendly (telephones, seating)? * Is there adequate taxpayer education? * Do taxpayers have easy access to the tax department for answers to queries and for advance rulings? * Are rates moderate and generally well accepted (number of complaints received from chambers of commerce about rates, exemptions, disallowances of exemptions)? * Do trends in number of complaints received suggest a good or bad relationship between taxpayers and tax administration? * Are assessments generally accepted or contested (ratio of appeals to assessments, proportion of assessments contested)? 2. Enumeration * How are potential taxpayers identified (censuses of specific taxpayer groups, systematic surveys)? * What are the sources of information (internal documents of tax agencies, other government agencies, third parties such as banks, municipal authorities)? Does the system track payment flows between taxpayers? * Are master rolls of taxpayer identification numbers broken down by sector, source of tax, and so on? * How is taxpayer information stored and retrieved (by taxpayer identification number, payments and receipts, other activities of taxpayer, past history)? 3. Estimation * What is the usual method of determining tax liability? If administrative assessments, are they rational and fair and based on evidence (are assessed rates of profit comparable across similar businesses and similar localities)? Do taxpayers get a hearing? * Are accounting standards and practices prevalent in the country? 131 * Is reliance on self-assessment high? * Are presumptive norms based on consultation with taxpayer constituencies? Are cadastral surveys used? Are the same valuations used for other related taxes to ensure consistency? 4. Collection * Is there a single current account for each taxpayer that tracks payments, refunds, and tax liabilities over time and that integrates information for single taxpayers across all taxes paid by the taxpayer? Does the system calculate totals and averages for tax payments, liabilities, and balances across taxpayers and by tax categories (type of tax, specific periods)? Does it calculate the number of taxpayers covered and the average lag between tax payment and records updating? * Does the tax system rely on withholding by third parties or advance payment by taxpayers? * How are tax payments monitored? Are large taxpayers monitored monthly? How are withholding and advance payment arrangements monitored? * Is the collection process public or private? Are there collection lags? Are arrears records checked? * Are there discounts for timely payment? Do they induce prompt payment? Do penalties escalate with delay? * What are the ratios of arrears to taxes due? * Are penalty rates of interest close to market rates? Are there arrangements for property seizure and garnishing (withholding of worker salaries by the state to pay off taxes due)? * Are payment arrears within norms allowed by the tax collection agency or the appeals authority? 5. Collection data * Last five years revenue: nominal and real amounts by tax; refunds and net collections by tax. * Collection patterns (as percentage of total collections for each tax) by major, medium-size, and small taxpayers; by economic sector. * Number of returns received per tax. Number of registered taxpayers by economic activity. Number and percentage of nonfilers (as share of total registered taxpayers) by tax period and by tax. 132 * Number of delinquent accounts. Amount collected and amount pending collection by tax, maturity, and amount due and ratio to total collection. 6. Enforcement * Are tax gaps measured? Ratio of identified taxpayers to relevant population? Ratio of taxpayers who file to identified taxpayers? Ratio of income returned to assessed income? Ratio of taxes paid to taxes due? * Are there any estimates of tax evasion? Are quick estimates possible? * How are audit samples chosen? Are systematic sample selection criteria used? Are rational and systematically chosen indicators used? How is the effectiveness of audits assessed (share of audit-assessed tax liabilities uncontested by taxpayers, share of appealed audits that are upheld, share of audit-assessed tax liabilities collected, share of detected arithmetical errors or cross-checked discrepancies collected, share of audits yielding assessed tax liabilities above some threshold, average audit-assessed tax liability per audited return)? * Is there a special cell for large taxpayers? * Is the information system centralized or decentralized? (Large countries need a more decentralized system but a single master roll.) Computerized? How are records maintained and updated? * Is there a system for collecting and matching information across tax agencies? Does it have access to bank accounts? What are the rules of privacy? How often is the third-party database used by agency staff to identify unregistered taxpayers or audit taxpayer returns? What are the costs of maintaining the third-party system? * What are the problems with search and seizure procedures? Are there checks against harassment and invasion of privacy? * How many penalty cases? Total penalty levied? What proportion of penalty cases are sustained by courts? Number of cases prosecuted for fraud? Number of convictions? * What are the annual number and value of appeals (by type of appeal, type of tax or taxpayer, value of appeal, or basis of appeal)? What are the number and value of appeals resolved, rejected, or partially rejected? What is the average length of the appeal process? What portion of the appeal board's decisions are appealed or overturned? What is the cost per appeal resolved? * How many amnesties are offered or given within an amnesty period? Do they increase tax collections? Does compliance improve perceptibly during the amnesty period? 133 7. Organization * How is tax administration organized? Organizational chart and description of organizational structure. Are taxes administered by central or local government? Are separate functions entrusted to different branches? Description of operational units (regions, districts). Description of units responsible for conceptualizing procedures. * Are the functions of different sections clearly demarcated? What is the degree of decentralization? * What principles govern resource allocation? What is the marginal productivity of each activity? Do costs include all types of resources used (labor, materials and supplies, capital assets) and cover multiple time periods to ensure reliability? * What is the extent of automation? Is there scope for expansion? * Description of tax administration facilities (buildings, office equipment). Is there adequate space for records storage and retrieval? Is the capital budget adequate for instruments, hardware, transport? Are arrangements for equipment maintenance adequate? * How many staff are assigned to the main functions in the organizational structure, and what is the level of formal training? Is the promotion system automatic or performance-based? Are staff temporary or permanent? How is merit recognized? Are pay scales in line with labor market rates? * Do internal controls and accountability ensure supervision at all layers? 135 Annex Table 4.2. Distribution of All SALs/SECALs Incorporating Tax Administration Reforms by Region Regional share in Region 1980-86 1987-94 Total total (percent) AFR 3 28 31 46 EAP 3 2 5 7 ECA 4 2 6 9 LAC 1 14 15 22 MNA 1 4 5 7 SAS 1 4 5 7 Total 13 54 67 100 Source: ALCID. 136 Annex Table 4.3. Level of Specificity of Tax Administration Conditionalities in Sample SALs/ SECALs % of Type of conditionality AFR EAP ECA LAC MNA SAS Total total Identified conditionalities Induce taxpayer compliance 17 8 2 17 0 6 50 18 Nonspecific measures 3 0 0 0 0 0 3 1 Improve enumeration 12 2 1 22 0 4 41 15 Nonspecific measures 0 0 1 1 0 0 2 1 Improve estimation 4 1 2 1 0 1 9 3 Nonspecific measures 0 0 0 0 0 1 1 0 Improve collection 14 2 2 9 0 0 27 10 Nonspecific measures 3 1 2 0 0 0 6 2 Improve enforcement 14 2 0 19 0 0 35 13 Nonspecific measures 3 0 0 2 0 0 6 2 Improve organization 25 16 2 30 0 6 79 29 Nonspecific measures 0 2 0 0 0 0 2 1 Subtotal for identified 86 31 9 98 0 17 241 87 conditionalities Nonspecific conditionalities 10 3 3 3 0 1 20 7 General conditionalities 15 4 4 8 3 2 36 13 Total conditionalities 101 35 13 106 3 19 277 100 Regional shares in total 36 13 5 38 1 7 100 conditionalities Note: Table refers to sample of 42 loans. Source: ALCID. 137 Annex Table 4.4. Distribution of Sample SALs/SECALs by Region and Coverage of TAR Areas Conditionalities AFR EAP ECA LAC MNA SAS Total General 0 0 3 1 2 1 7 Specific All six areas 0 0 0 0 0 0 0 Five areas 0 0 0 3 0 0 3 Four areas 2 1 0 2 0 1 6 Three areas 9 3 1 1 0 0 14 Two areas 4 0 1 1 0 0 6 One area 4 1 0 0 0 0 5 Total loans 19 5 5 8 2 2 41 Source: ALCID. 139 Annex Table 5.1. Health Care before 1985 Country Access to health carea Infant mortality rate Bangladesh 45 133 Burkina Faso 70 157 Gambia, The 90 162 Ghana 64 100 Guatemala 59 77 Malawi 54 171 Mali 20 187 Mozambique 40 158 Niger 48 153 Pakistan 64 126 Somalia 20 147 Uganda 42 116 Zambia 70 - Zimbabwe 71 84 aPercentage of population with access to health care; 1985 (except Mali, 1980). bPer thousand live births, 1979. Source: World Bank, Social Indicators of Development. 140 Annex Table 5.2. Changes in Total Expenditure Shares, Before and After Adjustment (percentage of GDP) Adjustment Country period Share before Share after Change in share AFR C6te d'Ivoire 1981-91 40.1 35.8 - Gambia, The 1986-91 33.8 26.8 - Ghana 1983-90 11.1 16.0 + Kenya 1981-91 31.9 30.5 - Malawi 1981-91 35.0 27.9 - Mozambique 1987-91 31.1 53.4 + Senegal 1981-91 28.2 18.8 - Tanzania 1981-91 36.7 28.6 - Uganda 1982-91 11.4 20.5 + EAP Indonesia 1987-90 23.0 18.3 - Philippines 1980-91 13.6 19.1 + ECA Bulgaria 1991-92 57.8 44.9 - Hungary 1989-92 62.2 60.5 - Poland 1990-91 39.9 49.0 + Turkey 1980-92 16.4 17.1 + LAC Argentinaa 1990-92 16.9 17.0 + Chile 1985-91 32.6 21.4 - Costa Rica 1985-91 27.0 28.1 + El Salvador 1990-92 15.8 16.4 + Honduras 1988-91 26.7 32.8 + Peru 1990-92 14.2 15.3 + Uruguay 1987-91 33.8 37.0 + Venezuela 1989-91 33.2 29.8 - MNA Morocco 1984-92 33.4 28.6 - Tunisia 1986-90 38.9 32.7 - SAS Bangladesh 1985-92 18.4 17.8 - Pakistan 1981-92 17.5 23.4 + a. Argentina's "share before" data correspond to the adjustment period, not the pre-adjustment period. Note: + denotes increase in share, - denotes decrease in share. Source: Bank and IMF documents and staff estimates. 141 Annex Table 5.3. Changes in Capital and Current Expenditure Shares, Before and After Adjustment (percentage of GDP) Share before Share after Change in current Change in capital Adjustment Current Capital Current Capital expenditure expenditure Country period expenditure expenditure expenditure expenditure share share AFR Cote d'lvoirea 1981-91 22.0 18.1 32.7 3.1 + Gambia, The 1986-91 19.8 14.0 17.3 9.5 - Ghana 1983-90 10.0 1.1 11.4 4.6 + + Kenya' 1981-91 23.3 8.6 24.9 5.6 + Malawib 1981-91 17.5 17.5 21.5 6.4 + Mozambique 1987-91 25.3 5.8 27.1 26.3 + + Senegal 1981-91 23.7 4.5 14.6 4.2 - Tanzania 1981-91 24.8 11.9 23.3 5.3 - Uganda 1982-91 6.7 4.7 10.3 10.2 + + EAP Indonesia' 1987-90 12.5 10.4 9.1 9.1 - Philippines 1980-91 9.6 2.1 15.7 3.4 + + ECA Bulgaria 1991-92 54.7 3.1 43.3 1.7 - Hungary 1989-92 54.6 7.6 53.8 6.8 - Polanda 1990-91 36.4 3.4 45.9 3.1 + Turkey 1980-92 10.0 6.4 11.6 5.5 +- LAC Argentinab 1990-92 15.8 1.1 15.8 1.2 + + Chile 1985-91 29.7 2.9 18.3 3.1 + Costa Rica 1985-91 21.0 6.0 22.6 5.5 + El Salvador 1990-92 11.8 4.0 12.2 4.2 + + Honduras 1988-91 21.6 5.1 20.4 12.4 - + Peru 1990-92 10.3 3.9 10.6 4.7 + + Uruguaya 1987-91 29.6 4.0 32.5 4.5 + + Venezuela 1989-91 20.2 13.0 18.7 11.1 - MNA Moroccoa 1984-92 23.7 9.7 21.3 7.3 - + Tunisia 1986-90 25.8 11.3 25.9 6.8 + SAS Bangladesha 1985-92 6.4 7.0 8.9 7.8 + + Pakistan 1981-92 14.5 3.1 20.0 7.0 + + Note: + denotes increase in share, - denotes decrease in share. a. Because capital expenditure increase was introduced not in the initial adjustment operation but in subsequent ones, performance in the post-adjustment period is compared with that during the adjustment period. b. Argentina's "share before" data correspond to the adjustment period, not the pre-adjustment period. Source: Bank and IMF documents and staff estimates; data relate to Chapter 2 sample. 142 Annex Table 5.4. Average Central Government Expenditure Shares (percent) Country group EAP ECA LAC MNA SAS AFR Overall TotExplGDP With conditions 1 20.0 54.9 25.9 34.1 20.5 27.5 30.5 II 16.7 52.4 25.5 30.8 21.3 26.7 28.9 Without conditions I 27.5 42.4 23.6 38.3 34.0 18.5 30.7 II 30.5 34.6 17.4 26.7 34.9 22.1 27.7 Cur ExplGDP With conditions I 13.9 47.9 21.9 23.3 16.0 18.8 23.6 I 11.6 50.4 24.7 24.1 18.7 20.9 25.1 Without conditions I 17.2 27.3 17.2 27.4 12.7 16.0 19.6 II 18.8 29.7 13.1 20.5 18.7 19.6 20.1 Cap ExplGDP With conditions 1 6.1 7.0 4.2 10.8 6.5 6.6 6.9 II 5.0 2.0 1.9 6.7 4.2 5.7 4.2 Without conditions I 10.6 15.1 6.3 10.9 21.3 3.5 11.3 II 11.8 4.9 4.2 6.2 16.1 3.4 7.8 Cur ExplExp With conditions I 71.3 78.7 84.7 68.3 76.6 72.3 75.3 II 70.5 90.7 95.2 78.3 83.6 75.8 82.4 Without conditions I 65.1 82.1 72.7 70.9 37.8 84.4 68.8 II 65.6 92.7 75.9 76.8 49.6 88.4 74.8 Cap ExplExp With conditions I 28.7 21.3 16.1 31.7 23.4 27.7 24.8 II 29.2 9.3 8.1 21.7 16.4 23.4 18.0 Without conditions 1 35.6 17.9 26.8 --29.1 62.2 20.0 31.9 II 34.6 7.3 23.9 23.1 50.4 15.4 25.8 Education/Exp With conditions I 15.0 9.3 13.4 16.6 6.4 15.1 12.6 II 15.4 10.8 12.8 17.7 7.6 13.3 12.9 Without conditions I 12.5 1.5 10.0 10.9 13.6 15.1 10.6 II 16.0 3.2 11.8 14.4 10.3 7.6 10.6 Health/Exp With conditions I 2.6 2.4 10.7 5.3 3.4 7.3 5.3 11 4.0 5.6 12.8 4.7 3.8 7.8 6.5 Without conditions I 7.4 0.3 6.1 4.2 5.3 3.2 4.4 11 6.3 4.5 8.0 5.6 6.6 1.9 5.5 DefenselExp With conditions 1 22.9 9.8 6.7 13.7 9.5 4.3 11.2 II 15.2 7.3 7.0 9.9 10.1 5.4 9.2 Without conditions I 12.3 27.0 7.1 21.0 0.0 20.7 14.7 II - 29.9 5.4 16.9 0.0 25.5 15.5 Infrastructure/Exp With conditions 1 25.8 34.4 19.7 29.4 36.1 28.1 28.9 II 23.9 21.8 9.7 17.8 27.6 20.0 20.1 Without conditions 1 39.5 35.6 24.4 23.5 57.6 19.0 33.3 II 29.8 30.8 20.8 22.0 49.6 19.9 28.8 Wages/Exp With conditions I 18.1 17.1 22.5 29.8 13.4 29.3 21.7 II 20.8 22.1 26.5 33.2 14.5 32.7 25.0 Without conditions 1 25.2 2.6 36.3 32.2 14.1 36.7 24.5 II 31.3 16.2 30.3 33.7 19.0 36.3 27.8 Subsidies/Exp With conditions I 11.4 - - - 2.9 - 7.1 11 1.9 - - - 2.7 - 2.3 Without conditions I - - 0.0 - 0.0 - 0.0 11 - - 0.2 - 2.5 - 1.3 Note: Country composition in each region varies by type of expenditure. Period I refers to the early 1980s and period II to the early 1990s. For each country included in the table, the early 1980s refers to any two contiguous years during 1980-84 and the early 1990s refers to any two contiguous years during 1990-94 for which expenditure data are available. Source: IMF, Government Finance Statistics. 143 Annex Table 5.5. Government Subsidies before and after Imposition of Condition on Subsidy Reduction (1990 prices) Year of Country condition Units 1986 1987 1988 1989 1990 1991 1992 1993 Brazil 1984 Reais (thousands) 208 122 120 212 76 229 151 - Bulgariaa 1991 Leva (billions) - 6.3 6.4 6.1 6.8 3.4 8.8 - Hungary 1988,90,91 Forints (billions) - 347 - 322 149 - - - Romania 1992 Lei (billions) - - - - 56 41 72 - Sierra Leone 1992 Leones (billions) - - - - - - 2.4 2.8 Sri Lankab 1990 Rupees (billions) 3.91 2.66 3.18 4.84 3.68 3.67 2.46 2.43 Tunisia 1988, 90 Dinars (millions) 54 53 152 111 79 66 71 Note: Figures are deflated by GDP deflator. a Figures are nominal. b Figures are deflated by CPI, all others. Source: International Monetary Fund 1994 and World Bank 1995. Annex Table 5.6. Current Expenditure Adjustments trercentage c' G.DP'____________________ Current expenditure Wages Subsidies and transfers Adjustment Country period Share before Share after Change iShare before Share after Change Share before Share after Change AFR Ghana 1983-90 10.0 11.4 1.4 29 2.8 -0.1 2.3 2.9 0.6 Kenya 1981-91 23.3 24.9 1.6 5.7 8.6 2.9 3.6 2.8 -0.8 AFR AVERAGE 16.7 18.1 1.5 4.3 5.7 1.4 2.9 2.9 -0.1 EAP Indonesia 1987-90 12.5 9.1 -34 3.3 2.9 -0.4 5.0 2.6 -2.3 Philippines 1980-91 9.6 15.7 6.1 3.5 5.2 1.7 0.8 2.3 1.5 EAP AVERAGE 11.0 12.4 i 1.4 3.4 4.1 0.6 2.9 2.5 -0.4 ECA Bulgaria 1991-92 54.7 43.3 -11.5 5.5 5.8 0.3 14.9 1.7 -13.3 Hungary 1989-92 54.6 53.8 -0.9 8.6 84 -0.3 129 4.4 -8.5 Turkey 1980-92 10.0 11.6 1.6 3.9 7.8 3.9 5.2 11.0 5.8 ECA AVERAGE 39.8 . -3.6 6.0 7.3 1.3 .11.0 5.73 LAC Chile 1985-91 29.7 18.3 -11.4 6.4 3.9 -2.5 18.3 10.5 -7.8 Costa Rica 1985-91 21.0 22.6 1.6 9.3 8.5 -0.8 6.3 7.7 1.4 EI Salvador 1990-92 11.8 12.2 0.4 7.7 6.6 -1.1 1.5 2. 8 1.3 Peru 1990-92 10.3 10.6 0.3 3.1 2.5 -0.6 2.6 5.3 2.8 Uruguay 1987-91 29.6 32.5 2.8 5.7 4.8 -0.9 11.7 18.2 6.5 Venezuela 1989-91 20.2 18.7 -1.5 5.2 4.9 -0.3 6.7 6.7 0.0 LAC AVERAGE 20.4 19.2 -1.3 6.2 5.2 -1.0 7.8 8.5 0.7 MNA Morocco 1984-92 23.7 21.3 -2.4 11.9 10.5 -1.4 4.6 2.6 -2.0 Tunisia 1986-90 25.8 25.9 0.1 9.8 9.6 j -0.2 10.8 10.8 0.0 MNA AVERAGE 24.8 23.6 -1.2 10.8 10.1 -0.8 7.7 6.7 -1.0 SAS Bangladesh 1985-92 6.4 8.9 2.5 2.8 2.7 -0.1 1.5_ 2.2 0.7 SAS AVERAGE 6.4 8.9 25 2.8 2.7 -0.1 1.5 2.2 0.7 SAMPLE AVERAGE 22.1 21.3 -0.8 6.0 6.0 0 6.8 5.9 0.9 Note: Regional averages and the sample average do not include countries for which data are not available. Source: Bank and IMF documents and staff estimates. 145 Annex Table 5.7. Public Expenditures on Education in Countries with Conditions and Countries Without (percent) Education expenditures As share of Region and Education condition, As share of total Average country loan approval year GNP expenditure growth ratea 1980 1990 1980 1990 1980-90 AFR Burkina Faso Yes(1991, 1994) 2.2 2.3 19.8 17.5 3.9 Cameroon Yes (1989) 3.2 3.4 20.3 19.6 2.6 Gambia, The Yes(1989) 3.3 3.8 8.7 11.0 -0.4 Ghana Yes(1987, 1989, 3.1 3.3 17.1 24.3 7.6 1990) Kenya Yes(1989, 1991) 6.8 6.8 18.1 16.7 5.0 Malawi Yes (1981, 1984, 3.4 3.4 8.4 10.3 3.9 1992) Mali Yes (1989, 1991) 3.8 3.2 30.8 17.3 -0.4 Mozambique Yes (1994) 3.8 6.3 12.1 12.0 -0.1 Senegal Yes (1990) 4.5 3.7 23.5 24.1 0.6 Sierra Leone Yes (1992, 1994) 3.8 1.4 11.8 - -12.1 Togo Yes (1988, 1991) 5.6 5.7 19.4 24.7 0.7 Uganda Yes (1990) 1.2 2.9 11.3 22.5 10.9 Zambia Yes (1991, 1994) 4.5 2.9 7.6 8.7 -6.8 Zimbabwe Yes (1993) 6.6 10.6 13.7 - 8.1 Burundi No 3.0 3.5 17.5 16.7 4.9 CAR No 3.8 2.8 20.9 -. -2.2 Congo No 7.0 5.6 23.6 14.4 1.4 Ethiopia No 3.3 4.8 10.4 9.4 4.9 Gabon No 2.7 5.7 - - 8.3 Mauritania No 5.0 4.7 - 22.0 -1.9 Mauritius No 5.3 3.7 11.6 11.8 2.8 Rwanda No 2.7 4.2 21.6 25.4 2.7 Zaire No 2.6 0.9 24.2 6.4 -10.6 EAP China No 2.5 2.3 9.3 12.4 8.4 Korea, Rep. of No 3.7 3.6 23.7 22.4 9.5 Philippines No 1.7 2.9 9.1 10.1 5.8 Thailand No 3.4 3.8 20.6 20.0 6.7 ECA Bulgaria Yes (1992) 4.5 5.4 - - 4.0 Annex Table 5.7 (continued) 146 Education expenditures As share of Region and Education condition, As share of total Average country loan approval year GNP expenditure growth ratea 1980 1990 1980 1990 1980-90 Former Czech. No 4.0 4.6 - 8.2 2.6 Hungary No 4.7 6.1 5.2 7.8 3.0 Romania No 3.3 3.1 6.7 7.3 -0.7 LAC Guatemala Yes (1993) 1.9 1.4 11.9 11.8 5.6 Trinidad and Yes(1990) 4.0 4.1 11.5 11.6 -1.4 Tobago Bolivia No 4.4 3.0 25.3 20.1 -3.3 Brazil No 3.6 4.6 - - 3.9 Chile No 4.6 3.7 11.9 - -2.6 Colombia No 1.9 2.9 14.3 21.4 5.0 Costa Rica No 7.8 4.6 22.2 20.8 -2.4 Ecuador No 5.6 2.8 33.3 19.1 -5.3 El Salvador No 3.9 1.8 17.1 - -7.1 Guyana No 9.7 4.7 14.0 - -7.6 Honduras No 3.2 4.6 14.2 15.9 5.6 Jamaica No 7.0 6.1 13.1 12.9 -1.0 Mexico No 4.7 4.1 - - -2.5 Panama No 4.9 5.5 19.0 - 1.7 Uruguay No 2.3 3.1 10.0 15.9 4.3 Venezuela No 4.4 4.1 14.7 18.8 -1.2 MNA Algeria No 7.8 9.1 24.3 27.0 5.1 Egypt No 5.7 6.7 9.4 - 6.3 Morocco No 6.1 5.5 18.5 26.1 1.8 Tunisia No 5.4 6.1 16.4 14.3 5.0 SAS Bangladesh Yes (1992) 1.5 2.0 7.8 10.3 8.6 India Yes(1993) 2.8 3.5 10.0 11.2 6.9 Pakistan Yes (1994) 2.0 3.4 5.0 - 12.4 Sri Lanka No 2.7 2.7 7.7 8.1 5.3 a. Average annual growth rates are based on estimated total public expenditure in constant prices (deflated using the implicit GDP deflator) and computed by fitting trend lines to the logarithmic values of the expenditure data for each year of the period. Source: World Education Report, 1993. 147 Annex Table 5.8. Infant Mortality Rates (per thousand live births) Fiscal year Country of condition 1987 1993 Change Health conditions imposed Bangladesh 1992 119 106 -13 Benin 1989, 1991 90 85 -5 Bulgaria 1992 14 14 0 Burkina 1991 138 129 -9 Cameroon 1989 74 61 -13 C6te d'Ivoire 1992 97 91 -6 Gambia, The 1989 143 130 -7 Ghana 1987, 1989 90 79 -11 Indonesia 1989, 1991 75 56 -19 Kenya 1991 63 61 ---2 Malawi 1992 151 142 -9 Mali 1991 169 157 -12 Morocco 1992 82 66 -16 Somalia 1989 132 120 -8 Togo 1988, 1991 94 83 -9 Trinidad and Tobago 1990 24 18 -6 Uganda 1990 120 114 -6 Zambia 1991 109 103 -6 Average 99 90 -9 No health conditions imposed Burundi 111 101 -10 Central African Republic 107 101 -6 Mauritania 110 99 -11 Tanzania 89 84 -5 Zaire 99 92 -7 Average 103 95 -8 Source: World Bank, Social Indicators ofDevelopment. 148 Annex Table 5.9. Trends in Public Employment in Selected Latin American Countries Public employment as Public employment as percentage percentage ofpopulation of labor force Country 1985 1990 1985 1990 Argentina 6.2 6.4 17.2 17.9 Chile 2.5 - 8.0 - Uruguay 8.3 8.7 21.7 22.1 Venezuela 6.5 6.5 21.3 18.6 Trinidad and Tobagoa 11.7 10.6 29.5 27.0 Ecuadorb 3.1 3.3 - - Note: May not include military personnel. a. Data are for 1987 and 1991. b. Data are for 1985 and 1991. Source: Population and labor force data: World Tables and Social Indicators ofDevelopment; Trinidad and Tobago 1992; Chile: El Empleo Pdblico Frente a la Crisis, Adriana Marshall; Argentina 1993; Uruguay 1991; Venezuela: CORDIPLAN D.P.E.P.P. en Base a Datos de OCEPRE y Encuesta de Hogares de OCEI. 149 Annex Figure 6.1. Fiscal Deficits and Terms of Trade by Region and Exporter Group AFR Diversified Exporters AFR Primary Exporters 2 160 2 ---- ------- - . -----...--.-...---------- 160 120 a-2 so oa 1 2 3 4 5 6 7 8 9101112 1 2 3 4 5 6 7 8 9101112 Years Years --- Deficits -0TOT -- Deficis --TOT EAP Diversified Exporters LAC Diversified Exporters 2 160 2160 120 120 -2a-2 80 80 z 40 40 10 0 -14 -4 1: 0 -14 4 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Years Years -4-Deficts -0O--TOT --eiit 0 O MNA Diversified Exporters SAS Diversified Exporters 2 160 1 2- - --- --- - ----- - --------- --- ----- 160 120 120 0 40 -66 -a 0-10 -0. 0 - 0 1 4 0 114 -------4 12 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Years Years -4.- Diefit -0Tots SS--Deficits -0-TOT Note: Tunisia was excluded from the sample because its terms of trade data ended in the seventh year after the first loan. Year 5 marks the year of approval of the first Bank SAL/SECAL. Source: IMF, Government Finance Statistics; Bank and IMF staff estimates; World Bank, Wodld Tables, severai years. 150 Annex Table 6.1. Adjustment Loans and Countries in Sample, by Region Countries with Adjustment loans adjustment lending Percentage Percentage Region Total Sample of total Total Sample of total AFR 128 73 57 37 13 33 EAP 15 10 67 7 3 43 LAC 58 25 43 20 5 28 MNA 12 10 83 5 2 50 SAS 16 20 75 6 3 50 Regional total in study 229 134 59 75 26 35 Note: Due to data limitations, ECA was excluded from the sample. Source: ALCID. Annex Table 6.2A. Countries with Eight or More Years of Fiscal Deficit Data AFR EAP LAC MNA SAS Cote d'Ivoire Korea, Rep. of Argentina Morocco Bangladesh Ghana Philippines Chile Tunisia India Kenya Thailand Costa Rica Pakistan Madagascar Mexico Malawi Uruguay Mauritius Senegal Sierra Leone Tanzania Togo Uganda Zambia Zimbabwe 152 Annex Table 6.2B. Adjustment Operations in Chapter 6 Sample Country Region Pcy Loan # FY COTE.IV AFR ly L2058 82 COTE.IV AFR ly L2332 84 COTE.IV AFR ly i L2711 86 COTE.IV AFR ly i L3127 90 COTE.IV AFR ly L3240 90 COTE.IV AFR ly L3408 92 COTE.IV AFR ly L3428 92 COTE.IV AFR ly L3429 92 GHANA AFR ly C1777 87 GHANA AFR ly C2005 89 GHANA AFR ly C1393 83 GHANA AFR ly C1435 84 GHANA AFR ly C1573 85 GHANA AFR ly C1672 86 GHANA AFR ly C1744 87 GHANA AFR ly C2140 90 GHANA AFR ly C2236 91 GHANA AFR ly C2318 92 GHANA AFR ly C2345 92 KENYA AFR ly C0999 80 KENYA AFR ly L2190 83 KENYA AFR ly C1717 86 KENYA AFR ly C1927 88 KENYA AFR ly C2049 89 KENYA AFR ly C2197 91 KENYA AFR ly C2204 91 KENYA AFR ly C2295 92 MADAGAS AFR ly C1541 85 MADAGAS AFR ly C1691 86 MADAGAS AFR ly C1834 87 MADAGAS AFR ly C1941 88 MALAWI AFR ly C1427 84 MALAWI AFR ly C1644 86 MALAWI AFR ly L2026 81 MALAWI AFR ly C2396 92 MALAWI AFR ly C1920 88 MALAWI AFR ly C2121 90 MAURITIUS AFR my L2010 81 MAURITIUS AFR my L2361 84 MAURITIUS AFR my L2791 87 S.LEONE AFR ly C2352 92 S.LEONE AFR ly C2546 94 S.LEONE AFR ly C1501 84 SENEGAL AFR ly C1656 86 SENEGAL AFR ly C1802 87 SENEGAL AFR ly C2090 90 SENEGAL AFR ly L1931 81 SENEGAL AFR ly C2077 90 153 TANZANIA AFR ly C1133 81 TANZANIA AFR ly C1741 87 TANZANIA AFR ly C1969 89 TANZANIA AFR ly C2308 92 TOGO AFR ly C1365 83 TOGO AFR ly C1599 85 TOGO AFR ly C1892 88 TOGO AFRF lyy C2194 91 TOGO AFR ly C2211 91 UGANDA AFR ly 1252 82 UGANDA AFR ly C1474 84 UGANDA AFR ly 01844 88 UGANDA AFR ly 2087 90 UGANDA AFR ly C2314 92 UGANDA AFR ly C2608 94 UGANDA AFR ly 02190 91 ZAMBIA AFR ly C2214 91 ZAMBIA AFR ly C2577 94 ZAMBIA AFR ly C1630 86 ZAMBIA AFR ly C1720 86 ZAMBIA AFR ly L2391 84 ZAMBIA AFR ly C2405 92 ZIMBABWE AFR ly C2527 93 ZIMBABWE AFR ly L3434 92 ZIMBABWE AFR ly L2239 83 KOREA EAP my L2071 82 KOREA EAP my L2354 84 PHILIPPIN EAP my L1903 81 PHILIPPIN EAP my L2266 83 PHILIPPIN EAP my L3539 93 PHILIPPIN EAP my L2469 85 PHILIPPIN EAP my L2787 87 PHILIPPIN EAP my L3049 89 THAILAND EAP my L2097 82 THAILAND EAP my L2256 83 ARGENTINA LAC my L2675 86 ARGENTINA LAC my L2815 87 ARGENTINA LAC my L2923 88 ARGENTINA LAC my L2996 89 ARGENTINA LAC my L3291 91 ARGENTINA LAC my L3394 92 CHILE LAC my L2625 86 CHILE LAC my L2767 87 CHILE LAC my L2892 88 COSTARICA LAC my L2518 85 COSTARICA LAC my L3005 89 COSTARICA LAC my L35 93 MEXICO LAC my L2331 83 MEXICO LAC my L2745 87 MEXICO LAC my L2882 88 154 MEXICO LAC my L2918 88 MEXICO LAC my L3085 89 MEXICO LAC my L3086 89 MEXICO LAC i my L3087 89 MEXICO LAC my L2919 88 MEXICO LAC my L3309 T 91 MEXICO LAC my L3357 91 URUGUAY LAC my L2836 87 URUGUAY LAC my L3081 89 URUGUAY LAC my L2468 84 MOROCCO MNA my L3001 89 MOROCCO MNA my L3463 92 MOROCCO MNA my L2377 84 MOROCCO MNA my L2604 86 MOROCCO MNA my L2820 87 TUNISIA MNA my L2962 88 TUNISIA MNA my L3424 92 TUNISIA MNA my L2754 87 TUNISIA MNA my L2781 87 TUNISIA MNA my L3078 89 BANGLAD SAC ly C1194 82 BANGLAD SAC ly C1471 84 BANGLAD SAC ly C1655 86 BANGLAD SAC ly C2361 92 BANGLAD SAC ly C1816 87 BANGLAD SAC ly C2427 93 INDIA SAC ly L3421 92 INDIA SAC ly L2629 86 INDIA SAC ly C2448 93 INDIA SAC ly L3267? 93 PAKISTAN SAC ly L2166 82 PAKISTAN SAC ly C1066 81 PAKISTAN SAC ly L2701 86 PAKISTAN SAC ly L3029 89 PAKISTAN SAC ly L3107 89 PAKISTAN SAC ly L3645 94 Note: The sample includes 134 adjustment operations. Pcy is per capita income level. Ly and my denote low-income and middle-income country, respectively, as classified in World Development Report 1994. Source: ALCID. Annex Table 6.3. Sample Countries by Country Group Loan- Severely Middle Primary Diversified Country Successful Unsuccessful intensive indebted income Low income exporters exporters COTE.IV AFR GHANA COTE.IV COTE.IV COTE.IV MAURITIUS COTE.IV COTE.IV KENYA GHANA AFR MADAGASCAR KENYA GHANA GHANA SENEGAL GHANA GHANA MAURITIUS KENYA AFR MALAWI SENEGAL KENYA KENYA KENYA MADAGASCAR S.LEONE MADAGASCAR AFR MAURITIUS TOGO MADAGASCAR MADAGASCAR MADAGASCAR MALAWI SENEGAL MALAWI AFR TANZANIA ZAMBIA MALAWI S.LEONE MALAWI TANZANIA MAURITIUS AFR UGANDA MAURITIUS TANZANIA S.LEONE TOGO S.LEONE AFR ZIMBABWE S.LEONE UGANDA TANZANIA UGANDA SENEGAL AFR SENEGAL ZAMBIA TOGO ZAMBIA TANZANIA AFR TANZANIA UGANDA ZIMBABWE TOGO AFR TOGO ZAMBIA UGANDA AFR UGANDA ZIMBABWE ZAMBIA AFR ZAMBIA ZIMBABWE AFR ZIMBABWE KOREA EAP KOREA PHILIPPINES PHILIPPINES KOREA PHILIPPINES PHILIPPINES EAP THAILAND PHILIPPINES THAILAND THAILAND EAP THAILAND ARGENTINA LAC ARGENTINA URUGUAY ARGENTINA ARGENTINA ARGENTINA ARGENTINA COSTA RICA CHILE LAC CHILE CHILE URUGUAY CHILE MEXICO COSTA RICA LAC COSTA RICA COSTA RICA COSTA RICA URUGUAY MEXICO LAC MEXICO MEXICO MEXICO URUGUAY LAC URUGUAY MOROCCO MNA MOROCCO MOROCCO MOROCCO MOROCCO MOROCCO TUNISIA MNA TUNISIA TUNISIA TUNISIA TUNISIA BANGLADESH SAS BANGLADESH BANGLADESH BANGLADESH BANGLADESH INDIA SAS INDIA INDIA INDIA PAKISTAN SAS PAKISTAN PAKISTAN PAKISTAN Total 15 8 23 11 12 14 10 14 156 Annex Table *. Fiscal Deficit Reductions by Country Group and Region Average fiscal deficit Change' Most In the In the most Medium recent two- medium In the long recent two- Preloan term Long term year period term term year period Region AFR -8.5 -9.0 -7.4 -8.1 -0.4 1.2 0.4 EAP -2.4 -3.2 -1.1 0.8 -0.8 1.3 3.2 LAC -4.0 -3.0 -1.6 0.6 0.9 2.4 4.6 MNA -8.5 -5.4 -3.0 -2.6 3.0 5.4 5.9 SAS -8.0 -7.9 -7.1 -5.9 0.1 0.9 2.1 COUNTRY GROUPS Loan categories Successful adjusters AFR -7.8 -8.1 -6.3 -7.8 -0.3 1.5 0.0 EAP -3.0 -3.2 -0.1 1.3 -0.2 2.9 4.3 LAC -4.0 -3.2 -1.8 0.6 0.9 2.3 4.6 MNA -8.5 -5.4 -3.0 -2.6 3.0 5.4 5.9 Unsuccessful adjusters AFR -8.1 -9.3 -8.3 -8.6 -1.2 --0.2 -0.5 EAP -1.1 -3.2 -3.1 -0.2 -2.0 -1.9 0.9 LAC -3.7 -2.4 -0.9 0.8 1.3 2..8 4.4 SAS -9.0 -9.4 -7.6 -5.9 -0.4 1.5 3.1 Loan-intensive adjusters AFR -8.5 -9.0 -7.4 -8.1 -0.4 1.2 0.4 EAP -1.1 -3.2 -3.1 -0.2 -2.0 -1.9 0.9 LAC -4.0 -3.2 -1.8 0.6 0.9 2.3 4.6 MNA -8.5 -5.4 -3.0 -2.6 3.0 5.4 5.9 SAS -8.0 -7.9 -7.1 -5.9 0.1 0.9 2.1 Economic categories LOW-INCOME AFR -8.3 -8.9 -8.0 -9.1 -0.5 0.3 -0.8 SAS -8.0 -7.9 -7.1 -5.9 0.1 0.9 2.1 MIDDLE-INCOME AFR -10.5 -9.6 -3.7 -2.5 1.0 6.9 8.1 EAP -2.4 -3.2 -1.1 0.8 -0.8 1.3 3.2 LAC -4.0 -3.0 -1.6 0.6 0.9 2.4 4.6 MNA -8.5 -5.4 -3.0 -2.6 3.0 5.4 5.9 Severely indebted AFR -8.6 -8.5 -7.6 -9.3 0.1 1.0 -0.7 LAC -4.6 -2.1 -0.5 0.8 2.5 4.0 5.4 MNA -10.6 -6.4 -3.2 -2.8 4.2 7.4 7.8 PRIMARY EXPORTERS AFR -7.9 -8.5 -8.0 -9.5 -0.5 -0.1 -1.5 LAC -5.5 -1.8 -0.2 0.9 3.7 5.3 6.3 DIVERSIFIED EXPORTERS AFR -10.0 -10.1 -6.0 -5.1 -0.2 4.0 4.9 EAP -2.5 -4.0 -1.9 1.1 -1.5 0.6 3.6 LAC -4.1 -4.5 -3.1 0.4 -0.5 0.9 4.5 MNA -8.5 -5.4 -3.0 -2.6 3.0 5.4 5.9 SAS -8.0 -7.9 -7.1 -5.9 0.1 0.9 2.1 Note: Fiscal deficits refer to overall deficits of consolidated central government. Preloan period refers to the four years preceding the first adjustment operation in a country; the medium term refers to years 1-4 beginning with the initial year of the first operation; the long term refers to years 5-8 beginning with the initial year of the first operation; and the most recent period refers to the most recent 2 years for which data are available for the country. Changes are relative to the preloan average. Souce: IMF, Govemment Finance Staffsfics; Bank and IMF staff estimates; World Bank, World Tables, several years. Annex Table 6.5. Fiscal Deficits, Interest Payments, and Primary Deficits (percentage of GDP) Preloan average Medium-term average Long-term average Most recent two-year period Fiscal Interest Primary Fiscal Interest Primary Fiscal Interest Primary Fiscal Interest Primary Country group or region deficits payments deficits deficits payments deficits deficits payments deficits deficits payments deficits Sub-sample (22) -6.5 1.7 -4.8 -7.0 3.3 -3.7 -5.5 3.4 -2.1 -5.4 3.9 -1.6 COUNTRY GROUPS Loan categories Successful adjusters (11) -5.7 1.4 -4.2 -5.9 3.4 -2.9 -4.0 3.6 -1.3 -4.6 3.6 -2.1 Unsuccessful adjusters (8) -6.4 1.8 -4.5 -7.7 4.0 -3.7 -6.7 3.9 -2.8 -6.1 5.5 -0.9 Loan-intensive adjusters (19) -7.1 1.8 -5.3 -7.7 - 3.5 -4.1 -6.3 3.7 -2.6 -6.4 4.3 -2.2 Loa-iteniv adjusters (19)__ -7.1 1.8- Economic categories - -_ Low-income adjusters (14) -8.2 1.9 -63 -87 3.9 -4.8 -7.8 3.9 -3.9 -8.4 4.7 -3.7 Middle-income adjusters (8) -3.6 1.3 -2.3 -4.1 2.2 _ -1.9 -1.4 2.6 1.2 -0.1 2.4 2.0 Severely indebted adjusters (9) -8.0 1.7 -6.3 -7.9 3.8 -4.1 -6.8 3.5 -3.4 -8.2 4.8 -3.4 Primary exporters (9) - - 2.0 -5.9 -8.5 46 -3.9 -8.0 4.2 -3.8 -9.5 4.9 -4.6 Diversified exporters (12) -6.4 1.6 -4.8 -7.0 2.5 -4.4 -4.6 3.2 -1.5 -3.1 3.6 0.3 REGION AFR (13) -85 2.0 -65 -90 4.2 -4.8 -7.4 4.0 -3.4 -8.1 4,7 -3.4 EAP (3) -24 1.1 -1.3 -3.2 1.7 -1.5 -1.1 2.6 1.5 0.8 2.2 2.1 LACa (3) -2.5 1.3 -1.2 -1.4 2.1 0.7 -0.2 2.3 2.1 0.7 2.1 2.8 SAS (3) -8.0 1.6 -6.4 -7.9 2.3 -5.6 -7.1 3.0 -4.1 -5.9 3.9 -1.9 a. Mexico and Argentina were excluded from LAC average due to lack of data on interest payments. Consistent data were not available for MNA countries. Note: Fiscal deficits refer to overall deficits of consolidated central government. Interest payments pertain to both domestic and external debt. Numbers in parentheses refer to the number of countries in each group. Source: IMF, Government Finance Statistics; World Bank and IMF staff estimates 158 Annex Table 6.6. Changes in Fiscal Deficits and Primary Tool Used, by Country Group and Region Medium term Long term Most recent period Country group or Change Primary tool Change Primary tool Change Primary tool region in fiscal used in fiscal used in fiscal used deficit deficit deficit Sample average X Expenditure V Revenue / Revenue Country groups Loan categories Successful / Revenue / Revenue / Revenue Unsuccessful X Expenditure X Both / Expenditure Loan-intensive X Expenditure / Revenue / Revenue Economic categories Low-income X Expenditure / Revenue X Expenditure Middle-income / Expenditure / Both / Both Severely indebted V Revenue / Both V Revenue Primary exporters X Expenditure / Revenue X Expenditure Diversified exporters X Expenditure / Both V Both Regions AFR X Expenditure / Revenue / Revenue EAP X Both V Both V/ Revenue LAC V Revenue / Revenue V Both MNA V Expenditure / Expenditure V Expenditure SAS / Revenue / Revenue V Revenue Note: Fiscal deficit and time period definitions are the same as in Table 1. In the case of worsened fiscal deficits, the "primary tool used" refers to the instrument that worsened the deficit. X denotes worsening of fiscal deficit. V denotes improvement in fiscal deficit. Source: Annex Table 6.5. Annex Table 6.7. Changes in Total Expenditure, Current Expenditure, and Capital Expenditure by Country Group and Region Relative to Preloan Period (in percentage points of GDP) Medium term Long term Most recent 2 years Total Current Capital Total Current Capital Total Current Capital Country group and region expenditure expenditure expenditure expenditure expenditure expenditure expenditure expenditure expenditure SAMPLE AVERAGE 1.7 2.1 -0.3 0.3 1.5 -1.1 0.8 1.7 -0.7 COUNTRY GROUPS Loan categories Successful adjusters (15) 1.2 2.4 -0.9 0.0 1.5 -1.1 0.7 1.7 -0.6 Unsuccessful adjusters (8) 3.3 2.0 __ 0.6 0.5 1.4 -1.4 0:5 1.5 -1.4 Loan-intensive adjusters (23) 1.9 ____ 2.3 -0.4 0.3 1.7 -1.2 0.8 . 1.8 -0.9 Economic categories Low-income adjusters (14) 3.8 3.6 -0.2 2.1 2.9 1.1 3 1 35 -0.7 Middle-income adjusters (12) -0.6 0.4 -0.6 -1.8 0.0 -1.2 -1.8 -0.4 -0.9 Severely indebted (11) 1.3 1.5 -0.4 -0.1 0.6 -0.8 1.7 1.9 -0.1 Primary exporters (10) 4.5 4.3 -0.3 3.0 3.6 -0.9 38 4.0 -0.4 Diversified exporters (14) 0.5 1.4 -0.6 -0.7 1.3 -1.5 -0.5 1.2 -1.2 REGION AFR (13) 3.1 3.7 -0.3 0.5 2.2 -1.2 1.0 2.2 -0.7 EAP (3) 0.4 0.3 0.1 -0.5 0.3 -0.8 1.3 1.1 0.1 LAC (5) _____ 0.1 0.0 -0.2 0.2 0.6 -0. 7 -0.4 . 0.1 - MNA (2) -4.2 -0.7 -3.4 -4.6 -1.6 -3.1 -3.9 -0.8 -3.1 SAS (3) 1.3 1.6 -0.3 1.2 2.7 -1.4 1.9 3.4 -1.5 Note: Preloan period refers to the four years preceding the first adjustment operation in a country; the medium term refers to years 1-4 beginning with the initial year of the first operation; the long term refers to years 5-8 beginning with the initial year of the first operation; and the most recent period refers to the most recent 2 years for which data are available for the country. Numbers in parentheses indicate number of countries in each group. Source: MF, Govemment Finance Statistics; World Bank and IMF staff estimates. 160 Annex Table 6.8. Correlation Coefficients between Ratio of Fiscal Balances to GDP and GDP Growth and Other Macroeconomic Variables Growth rate Inflation Current account/GDP Sample average 0.77 0.18 0.63 Country group ' Middle-income 0.68 -0.09 0.74 Diversified exporters 0.73 0.08 0.69 Low-income 0.64 0.40 0.46 Primary exporters 0.59 0.59 0.59 Region AFR 0.7 0.5 0.6 EAP 0.5 -0.3 0.5 LAC 0.5 -0.2 0.2 MNA 0.1 -0.7 0.8 SAS 0.2 -0.4 0.5 Source: Fiscal deficit data from IMF, Government Finance Statistics; growth and current account data from World Bank, World Tables; and inflation data from IMF, International Finance Statistics. 161 Annex Table 6.9. Required Adjustments for Fiscal Solvency and Sustainability, 1989-93 Low adjustment Severe adjustment Region (0-3%) (more than 3%) A. Fiscal Solvency LAC Argentina El Salvador (3-10%) Colombia Peru (3-10%) Costa Rica Venezuela AFR Gabon Gambia (3-10%) Mauritius Ghana (3-10% Nigeria Kenya (3-10%) Zimbabwe Senegal (3-10%) Uganda (3-10%) EAP Philippines SAS None Pakistan (3-6%) India (6-10%) Bangladesh (6-10%) Sri Lanka (6-10%) MNA None Tunisia (3-6%) Morocco (6-10%) 162 Annex Table 6.9 (continued) Low adjustment Severe adjustment Region (0-3%) (more than 3%) B. Fiscal Sustainability LAC Argentina El Salvador (3-6%) Peru (3-6%) Honduras (6-10%) AFR Gabon Benin (6-10%) Gambia, The Burundi (3-6%) Kenya Cameroon (3-6%) Senegal Central African Republic (10-20%) Congo (6-10%) C6te d'Ivoire (6-10%) Ghana (6-10%) Madagascar (6-10%) Malawi (6-10%) Mali (6-10%) Mauritania (10-20%) Mozambique (10-20%) Niger (6-10%) Sierra Leone (3-6%) Uganda (6-10%) Zambia (1-20%) EAP None None SAS Pakistan Bangladesh (6-10%) India (6-10%) Sri Lanka (3-6%) MNA Morocco None Tunisia Note: Estimates refer to the overall public sector not the consolidated central government. Source: Dinh 1995. 163 Bibliography Acharya, Shankar N. et al. (1986). Aspects of the Black Economy in India: Report Submitted to the Ministry of Finance, Government of India. New Delhi: National Institute of Public Finance and Policy. Akin, J. et al (1987). Financing Health Services in Developing Countries: An Agenda for Reform. A World Bank Policy Study. Washington, D.C.: World Bank. Alderman, Harold, et al. (1995). "Public Schooling Expenditures in Rural Pakistan: Efficiently Targeting Girls and a Lagging Region ", in Dominique van de Walle and Kimberly Nead (eds), Public Spending and the Poor: Theory and Evidence. Baltimore: The Johns Hopkins University Press. Alesina, Alberto (1995). The Political Economy ofMacroeconomic Stabilizations and Income Inequality: Myths and Reality. Paper prepared for IMF Conference on Income Distribution and Sustainable Growth, June 1995. Anand, Ritu, Sweder van Wijnbergen (1989). "Inflation and the Financing of Government Expenditure: an Introductory Analysis with an Application to Turkey." World Bank Economic Review, Vol. 3, No. 1. Washington, D.C.: World Bank. Appleton, Simon and Paul Collier (1995). "On Gender Targeting of Public Transfers", in Dominique van de Walle and Kimberly Nead (eds), Public Spending and the Poor: Theory and Evidence. Baltimore: The Johns Hopkins University Press. Baffes, John and Anwar Shah (1993). "Productivity of Public Spending, Sectoral Allocation Choices, and Economic Growth." Policy Research Working Paper Series, WPS#1 178. Washington, D.C.: World Bank. Barnum, Howard and Joseph Kutzin (1993). Public Hospitals in Developing Countries: Resource Use, Cost, Financing. Baltimore: The Johns Hopkins University Press. Bagchi, Amaresh, Richard M. Bird and Arindam Das-Gupta (1995). Developing a Strategy for Tax Administration Reform. Washington, D.C.: World Bank. Bagchi, Amaresh, Richard M. Bird and Arindam Das-Gupta (1994). An Economic Approach to Tax Administration Reform. Washington, D.C.: World Bank. Bagchi, Amaresh (1991). "Tax Reform in Developing Countries: Agenda for the 1990s." Asian Development Review, Vol. 9, No. 2. Washington, D.C.: World Bank. Binswanger, Hans and Klaus Deininger (1995). Towards A Political Economy of Agriculture and Agrarian Relations. Processed. Bird, Richard M. (1983). "Income Tax Reform in Developing Countries: The Administrative Dimension." International Bureau of Fiscal Documentation Bulletin. 164 Bird, Richard M. (1989). "The Administrative Dimension of Tax Reform in Developing Countries," in M. Gillis (ed.), Tax Reform in Developing Countries. Durham: Duke University Press. Bird, Richard M. and Milka Casanegra de Jantscher (eds.) (1992). Improving Tax Administration in Developing Countries. Washington, D.C.: International Monetary Fund. Blejer, Mario I., and Adrienne Cheasty (1993). The Measurement ofFiscal Deficits. Washington, D.C.: International Monetary Fund. Boadway, Robin, Sandra Roberts, and Anwar Shah (1994). "The Reform of Fiscal Systems in Developing and Emerging Market Economies: A Federalism Perspective". Policy Research Working Paper 1259. Washington, D.C.: World Bank. Boadway, Robin and Anwar Shah (1994). "Fiscal Federalism in Developing/Transition Economies: Some Lessons From Industrialized Countries." National Tax Journal, Proceedings Issue, April 1994, pp.64-71. Buiter, Willem H. (1982). The Proper Measurement of the Public Sector Deficit and its Implications for Policy Evaluation and Design, DM/82/84. Washington, D.C.: International Monetary Fund. Campos, Ed and Sanjay Pradhan (1996). The Impact ofBudgetary Institutions on Expenditure Outcomes: Binding Governments to Fiscal Performance. Washington, D.C.: World Bank. Chandra, Vandana (1996a). "A Macroeconomic Evaluation of Fiscal Adjustment in Africa." (1996b). "A Macroeconomic Evaluation of Fiscal Adjustment During SALs/SECALs." Chandra, Vandana and Salman Anees (1996). "Fiscal Adjustment under SALs and SECALs: A Macroeconomic Overview." OED Background Paper. de Melo, Martha (1988). Public Investment/Expenditure Reviews: The Bank's Experience. Public Economics Division, Country Economics Department. Washington, D.C.: World Bank. Dinh, Hinh T. (1995). Fiscal Solvency and Sustainability in Creditworthiness Analysis. Washington, D.C.: World Bank. Easterly, William, C.A. Rodriguez, and K. Schmidt-Hebbel (eds.)(1994). Public Sector Deficits and Macroeconomic Performance. New York: Oxford University Press for the World Bank. Faini, Riccardo and Jaime de Melo (1990). "Fiscal Issues in Adjustment: An Introduction". Ricerche Economiche, No. 2-3, April-September 1990, Department of Political Science, Universita degli Studi Ca'Foscari Venezia, Venice. 165 Fleisig, Heywood (1989). Improving Tax Policy Advice: Lessons and Unresolved Issues from Asia Experience." Asia Regional Series, Report No. IDP 35. Washington, D.C.: World Bank. Gupta, Poonam (1996). "An Analysis of Revenue versus Expenditure Focus in Bank-Supported Lending." OED Background Paper. Hasan, Aynul (1995). Optimal Enrollment and Cost Effective Expenditures for Public School System. Paper presented at the Conference on Resource Mobilisation and Expenditure Planning, Lahore. Hasan, Aynul (1995). Optimal Mix ofHealth Sector Expenditure. Paper presented at the Conference on Resource Mobilization and Expenditure Planning, Lahore. Home, Jocelyn (1991). Indicators ofFiscal Sustainability. WP/91/5. Washington, D.C.: International Monetary Fund. Husain, Shaikh (1994). Education Finance: Public and Private Interface. Processed. Huther, Jeff , Sandra Roberts and Anwar Shah (1996). Adjustment Lending and Public Spending. Operations Evaluation Department, World Bank. International Bank for Reconstruction and Development (1994). Minutes of the Joint Audit Committee, August 3, 1994. Washington, D.C.: World Bank. (1994). Approach Paper, Selected Issues in Public Sector Management: Fiscal Management Study. Joint Audit Committee. Washington, D.C.: World Bank. International Monetary Fund (1994a). Overview ofDevelopments in Countries with Stand-By and Extended Arrangements Approved During 1988-91. Washington, D.C. (1994b). Government Finance Statistics Yearbook, 1994. Washington, D.C.: International Monetary Fund. . Several Years. International Finance Statistics. Washington, D.C. (1989). Tax Administration in Developing Countries: Strategies and Tools of Implementation. Washington, D.C.: International Monetary Fund. International Monetary Fund, Fiscal Affairs Department (1995). "Unproductive Public Expenditures. A Pragmatic Approach to Policy Analysis." Pamphlet Series No. 48, Washington, D.C.: IMF. Jayarajah, Carl and William Branson (1995). Structural and Sectoral Adjustment: World Bank Experience, 1980-1992. A World Bank Operations Evaluation Study. Washington, D.C.: World Bank. Jimenez, E. (1986). "Public Subsidization of Education and Health in Developing Countries: A Review of Equity and Efficiency". World Bank Research Observer (1): 111-129. January 1986. 166 Khalilzadeh-Shirazi, Javad and Anwar Shah (eds.) (1991). Tax Policy in Developing Countries. The World Bank Symposium. Washington, D.C.: World Bank. Larsen, Bjorn and Anwar Shah (1995). Global Climate Change, Energy Subsidies and National Carbon Taxes. In Public Economics and the Environment in an Imperfect World., ed. by Lans Bovenberg and Sijbren Cnossen. Boston: Kluwer Academic Publishers. pp. 113-32. Lockheed, Marlaine E., Adriaan M. Verspoor, et al. (1991). Improving Primary Education in Developing Countries. World Bank: Oxford University Press. Mitra, Pradeep (1992). "The Coordinated Reform of Tariffs and Indirect Taxes." The World Bank Research Observer, Vol. 7, No. 2. Washington, D.C.: World Bank. Morisset, Jacques and Alejandro Izquierdo (1993). Effects of Tax Reform on Argentina's Revenues. Working Papers WPS 1192. Washington, D.C.: World Bank. Nashashibi, Karim, Sanjeev Gupta, Claire Liuksila, Henri Lorie, and Walter Mahler (1992). The Fiscal Dimensions ofAdjustment in Low-Income Countries. Occasional Paper No. 95. Washington, D.C.: International Monetary Fund. Nunberg, Barbara and John Nellis (1995). Civil Service Reform and the World Bank. Discussion Paper No. 161, May. Washington, D.C.: World Bank. Pradhan, Sanjay (1996). Evaluating Public Spending: A Framework for Public Expenditure Reviews. World Bank Discussion Papers No. 323. Washington, D.C.: World Bank. Rajaram, Anand (1992). Tariff and Tax Reform: Do World Bank Recommendations Integrate Revenue and Protection Objectives? Policy Working Papers, WPS 1018. Washington, D.C.: World Bank. Ridker, Ronald G. (1994). The World Bank's Role in Human Resource Development in Sub Saharan Africa. Washington, D.C.: World Bank, Operations Evaluation Department. Ron, A. et al (1990). Health Insurance in Developing Countries: The Social Security Approach. Geneva: International Labour Office. Schmidt-Hebbel (1995). Fiscal Adjustment and Growth: In and Out ofAfrica. Washington, D.C.: The World Bank. Selden, Thomas M. and Michael J. Wasylenko (1995). "Measuring the Distributional Effects of Public Education in Peru", in Dominique van de Walle and Kimberly Nead (eds), Public Spending and the Poor: Theory and Evidence. Baltimore: The Johns Hopkins University Press. Sen, Amartya (1995). "The Political Economy of Targeting", in Dominique van de Walle and Kimberly Nead (eds), Public Spending and the Poor: Theory and Evidence. Baltimore: The Johns Hopkins University Press. 167 Shah, Anwar (1992). "Dynamics of Public Infrastructure, Industrial Productivity and Profitability." The Review of Economics and Statistics, Vol.LXXIV, No.1, February, pp.28-36. Shah, Anwar (1991). "The New Fiscal Federalism in Brazil." World Bank Discussion Paper No. 124. Washington, D.C.:World Bank. - (1988). "Public Infrastructure and Private Sector Profitability and Productivity in Mexico. " The World Bank Working Paper Series No. 100, September. Shihata, Ibrahim (1994). Civil Service Reform in Developing Countries. In Shahid Amjad Chaudhry, Gary Reid and Waleed Malik, editors, Civil Service Reform in Latin America and the Caribbean. World Bank Technical Paper No.259, Washington, D.C.: World Bank, pp.1 1-13. Shome, Parthasarathi (1992). Trends and Future Directions in Tax Policy Reform: A Latin America Perspective. WP/92/43. Washington, D.C.: World Bank. Skinner, Jonathan (1989). "Do Taxes Matter? A Review of the Effect of Taxation on Economic Behavior and Output." Policy, Planning, and Research Working Paper No. 48. Washington, D.C.: World Bank. Stiglitz, Joseph E. (1996). The Role of Government in Economic Development. Annual Bank Conference on Development Economics. Washington, D.C.: World Bank. Subbarao, Kalanidhi, Jeanine Braithwaite, and Jyotsna Jalan (1995). "Protecting the Poor During Adjustment and Transitions". HCO Working Paper 58. Washington, D.C.: The World Bank. Summers, Lawrence and Lant Pritchett (1993). "The Structural-Adjustment Debate." American Economic Review, May, p. 383-389. Talvi, Ernesto (1994). "Fiscal Policy and the Public Sector in Exchange Rate Based Stabilizations: Evidence from Uruguay's 1978 and 1991 Programs." The University of Chicago and Central Bank of Uruguay. Tan, Jee-Peng, Vito Tanzi, Jr., Abdo Yazabeck (1995). Public Spending on Health in the 1980s: The Impact ofAdjustment Lending Programs. Background Paper Number 16. Johns Hopkins University. Tanzi, Vito (1995). "Corruption, Government Activities and Markets." Finance and Development, Vol. 32, No.4, pp.24-27. (1994). "Corruption, Government Activities, and Markets," IMF Working Paper, Washington, DC: International Monetary Fund, Fiscal Affairs Department. Thirsk, Wayne (1991). Lessons from Tax Reform: An Overview. PPR Working Paper. Washington, D.C.: World Bank. 168 Thirsk, Wayne (1996, forthcoming). The Substance and Process of Tax Reform in Ten Developing Countries. Thirsk, Wayne and Harry Watson (1995). "Overview of Tax Conditionality." OED Background Paper. Thomas, Vinod, A. Chibber, M. Dailami, and J. de Melo (eds.) (1991). Restructuring Economies in Distress: Policy Reform and the World Bank. New York: Oxford University Press for the World Bank. UNDP (1995). The 1995 Human Development Report. New York: Oxford University Press. UNESCO (1993a). World Education Report 1993. France: Oxford University Press. (1993b). Statistical Yearbook 1993. Paris: UNESCO. Van de Walle, Dominique (1995). Public Spending and the Poor: What We Know, What We Need to Know. Policy Research Working Paper 1476. Washington, D.C.: The World Bank. Van de Walle, Dominique and Kimberly Nead (eds) (1995). Public Spending and the Poor. Theory and Evidence. Baltimore: The Johns Hopkins University Press. Van der Gaag, Jacques (1995). Private and Public Initiatives. Washington, DC: The World Bank. van Wijnbergen, Sweder (1990). "External Debt, Inflation, and the Public Sector: Toward Fiscal Policy for Sustainable Growth." Economic Review, Vol. 3, No. 3: 297-320. Washington, D.C.: World Bank. Vazquez-Caro and Gary Reid with Richard Bird (1992). Tax Administration Assessment in Latin America. Latin America and the Caribbean Technical Department. Washington, D.C.: World Bank. World Bank (1 995a). Review ofPublic Expenditure Work. Office of the Vice President, Development Economics. Washington, D.C. (1995b). Bureaucrats in Business. Washington, D.C.: World Bank. (1995c). Evaluation Results, 1993. Washington, DC: Operations Evaluation Department, World Bank. (1 995d). The Social Impact ofAdjustment Operations: An Overview. Operations Evaluation Department, Report No 14776. Washington, D.C.: The World Bank. (1995e). World Development Report. New York: Oxford University Press. (1995f). "ALCID. Computerized Data Base." Washington, D.C. (1995g). Social Indicators ofDevelopment. Washington, D.C. 169 World Bank (1995h). World Tables. Washington, D.C. (1995i). World Debt Tables. Washington, D.C. _ _ (1992a). Trade Policy Reforms under Adjustment Programs. Operations Evaluation Department. Washington, D.C. (1992b). Adjustment Lending and Mobilization ofPrivate and Public Resources for Growth, Policy and Research Series (22). Washington, DC: Country Economics Department, World Bank. (1 992c). World Bank Structural and Sectoral Adjustment Operations: The Second OED Overview. Washington, DC: Operations Evaluation Department, World Bank. _ (1991a). Lessons of Tax Reform. Washington, D.C. - (1991b). World Development Report 1991. Washington, D.C. (1990a). World Development Report 1990: Poverty. Washington, D.C.: The World Bank. (1 990b). Adjustment Lending Policies for Sustainable Growth. Policy and Research Series No. 14. Washington, D.C.: The World Bank. (1989). "Bank-Fund Collaboration in Assisting Member Countries." Memorandum to the Executive Board of the International Monetary Fund and the Board Executive Directors of the World Bank. Washington, D.C.: World Bank. (1988a). World Development Report 1988. Washington, D.C. (1988b). Adjustment Lending. An Evaluation of Ten Years of Experience. Country Economics Department. Policy and Research Series No.1. Washington, DC: World Bank (1988c). Public Investment/Expenditure Reviews: The Bank's Experience. Country Economics Department. Washington, D.C.: World Bank.