THE WORLD BANK RESEAR C m:1 OBSERVEI9' V 7 4 91 Lessons from the Southern Cone Policy Reforms Vittorio Corbo and Jaime de Melo Land Rights Systems and Agricultural Development in Sub-Saharan Africa Gershon Feder and Raymond Noronha The Relevance for Developing Countries of Recent Developments in Macroeconomic Theory W. Max Corden Foreign Trade Regimes and Economic Growth in Developing Countries Deepak Lal and Sarath Rajapatirana The Management of Common Property Resources: Finding a Cooperative Solutionu Robert Wade ____ Uaig COPSI THE WORLD BANK RESEARCH OBSERVER EDITORIAL BOARD EDITOR Armeane Choksi CO-EDITORS Vittorio Corbo, David Turnham MANAGING EDITOR Hernan Cortes MEMBERS Domingo Cavallo (Fundacion Mediterranea, Argentina) Juergen Donges (Kiel Institute of World Economics, Federal Republic of Germany), Mark Gersovitz (Princeton University, United States), Enzo Grilli, Peter Muncie, Demetrios Papageorgiou, George Psachar- opoulos, William Tyler DESIGNER Joyce C. Eisen The World Bank Research Observer is published twice a year by the Intemational Bank for Reconstruction and Development. The Observ- er is managed in the Office of the Vice President, Economics and Research, and vetted by the editor, co-editors, and members of an editorial board drawn from various departments throughout the World Bank and the international academic community. It seeks to keep nonspecialist readers informed about economic research currently being undertaken within the Bank and special areas of economics relevant for development policy. Each issue of the Observer will normally contain an article surveying an area of recent Bank research. In addition, it will contain reviews of special topics written by World Bank staff or outside experts in the field in a style readily comprehen- sible to the lay reader. The World Bank does not accept responsibility for the views expressed herein, which are those of the authors and should not be attributed to the World Bank or its affiliated organizations. Some of the findings, interpretarions, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitation of its boundaries or national affiliation. The World Bank Research Observer welcomes inquiries, comments, and editorial responses. Please direct correspondence to: The World Bank Research Observer Managing Editor, Room 5-9121 The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. informal Bank documents cited in the articles stemming from or referring to World Bank research projects are generally available through the Bank Research Documentation Center in the Office of the Vice President, Economics and Research Staff, Room S-9035. World Bank Staff Working Papers, Country Reports, and other official series that may be cited are published by the Bank. The most recent World Bank publications are described in the annual spring and fall catalogs, which are available free of charge from the Publications Sales Unit, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from the World Bank European Office, 66, avenue d'1ena, 75116 Paris, France. () 1987 by the International Bank for Reconstruction and Development/ The World Bank All rights reserved Manufactured in the United States of America ISSN 0257-3032 Material in this journal is copyrighted. Requests for permission to reproduce articles should be addressed to the Director, Publications Department, The World Bank, Washington, D.C. 20433, U.S.A. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the intended reproduction is for non- commerical purposes, without asking a fee. Permission is not required to make photocopies for classroom and similar immediate use. ImportantSubsiphon r\j0h1ce With the January 1988 issue ol 3 N a year's subscription (two issues) ofi Te WorWd &Y* Research Observer wXi cost- $20 for readers with sal i in couintris that are members Of dhe Organo Er perion and Devlopmet (OECD). Readrswith in non- QOEC) countries are elWble to r ee sub- scriptions on request. Please fill in the order form f y sh continue your subscription to The Word BnkARisa Obever. Those wodo not returni dt ore - ^ Swrieiv issUes. Detach card here and mail.- - - -- - Yes, I wish to receive The World Bank Research Observer. Name Title Address City State Postal Code Country I Subscribers in OECD countries: I El Enclosed is my check for US$20, payable to World Bank Publications. r U Charge my U VISA U MasterCard U American Express U] Choice I Credit Card Number Expiration Date I Signature U. Please bill me. (Institutional orders only.) I Purchase Order Number Subscribers in non-OECD countries: U I am eligible to receive a complimentary subscription to The World Bank Research Observer according to the terms listed above. Please continue to enroll me as a corn- plimentary subscriber. If paying by check, detach this card and return it with your payment in an envelope addressed to World Bank Publications, Dept. 0552, Washington, D.C. 20073-0552, U.S.A. Otherwise, detach this card, affix postage to the preaddressed side, and drop it in the IDmail. Affix l Postage l Here World Bank Publications 1818 H Street, N.W., Room J1-085 Washington, D.C. 20433, U.S.A. THE WORLD BANK RESEARCH OBSERVER VOLUME 2 NUMBER 2 JULY 1987 Lessons from the Southern Cone Policy Reforms Vittorio Corbo and Jaime de Melo 111 Land Rights Systems and Agricultural Development in Sub-Saharan Africa Gershon Feder and Raymond Noronha 143 The Relevance for Developing Countries of Recent Developments in Macroeconomic Theory W. Max Corden 171 Foreign Trade Regimes and Economic Growth in Developing Countries Deepak Lal and Sarath Rajapatirana 189 The Management of Common Property Resources: Finding a Cooperative Solution Robert Wade 219 LESSONS FROM THE SOUTHERN CONE POLICY REFORMS Vittorio Corbo Jaime de Melo ew reform packages have led to as much controversy as those in the Southern Cone countries of Argentina, Chile, and Uru- guay. Some observers, notably in the press, have concluded that the reform effort as a whole was a failure. Others, including the present authors (1985a), have suggested that many of the microecon- omic reforms were quite successful and that most of the problems were caused by macroeconomic management during the transition to a more open economy. Still others have blamed unfavorable exter- nal shocks (Sjaastad 1983). And some maintain that the sequence in which the reforms were implemented was a major cause of failure (Edwards 1985, Frenkel 1983). The lessons of the Southern Cone experience are of interest not only to the economic historian. They are also of pressing practical importance to policymakers in other developing countries, most of whom are either adopting or contemplating reforms very similar to many of the measures employed in the Southern Cone. This article reviews the lessons. It complements our earlier paper (Corbo and others 1986), in which we drew country-specific conclusions from what our research suggested were the major failures in each country. Here we ask whether the reforms were consistent with what seems to be the emerging consensus on how to implement stabilization and liberalization policies in developing countries. We review conditions before the reforms began and summarize and evaluate the reforms, paying particular attention to how far they deviated from today's consensus. We then interpret the outcome of the reforms and draw policy lessons. (C 1987 The International Bank for Reconstruction and Development/The World Bankn1 1 The Period By the early 1970s the three Southern Cone economies were among before the most distorted of the middle-income cleveloping countries. Expan- the Reform sionary demand policies combined with fixed or slowly adjusting exchange rates, pervasive price controls (over 90 percent of the basket of goods in the consumer price index was controlled in each country), and restrictive trade regimes; the result was accelerated inflation, bot- tlenecks in production, slow export growth, and chronic balance of payment difficulties.' For more than three decades, trade policies in all three countries were strongly biased in favor of import-substituting industrialization and against exports. All three countries had experimented with some mild trade liberalization: Argentina in the second half of the 1960s, Chile in 1956-62 and again in the late 1960s, and Uruguay in 1959. In each case, they had restored a very restrictive trade regime with widespread tariff and nontariff barriers. Fragmentary evidence shows high effective rates of protection to domestic sales in each country: 84 percent in Argentina (1969), 151 percent in Chile (1974), and 384 percent in Uruguay (1968) (see tables 1-3). In all three countries, protection also varied widely across sec- tors-an indicator of distorted incentives-the piecemeal result of decades of pressure from different interest groups. The three economies had become steadily less open since the late 1920s. As shown in tables 1-3, the ratio of total trade (exports plus imports) to gross domestic product (GDP) was 25 percent for Uruguay, the smallest of the three countries, in the early 1970s. Chile's ratio was 20 percent, and Argentina's 17 percent. These percentages were well below the norm for countries of similar size and development (see Chenery and Syrquin 1975). All three countries were also in severe macroeconomic disequili- brium (see table 4). Both Chile (1973) and Argentina (1976) had large fiscal deficits, rapid inflation, and acute shortages of foreign exchange. Public sector deficits averaged 10 percent of GDP in Argentina in 1973-75, 16.1 percent in Chile during 1971-73, and 3.2 percent in Uruguay during 1971-73.2 Only Uruguay was not in deep crisis by the early 1970s. It still had some foreign exchange reserves. But its per capita incomes had barely grown at all for twenty years. Stabilization What Was Done Reforms The first task- of the reform programs was to restore external bal- ance and to rein in galloping inflation.3 In each country, governments first adopted an orthodox program that included a large devaluation with a unification of the different exchange rates, and an attempt to restrain monetary growth and public expenditures. In a second phase, 112 Research Observer 2, no. 2 (July 1987) Table 1. Protection and Trade Openness in Argentina (percent) A. Protection Effective protection to Realized Effective domestic value Legal tariff, protection, protection, Sector added,1969 Industry December 1977 February 1977 February 1977 Primary activities -8.0 Textiles 57.4 41.1 85.3 Processed foods 44.0 Clothing 95.0 79.2 131.6 Beverages and tobacco 95.0 Paper and paper products 29.0 30.8 74.9 Construction materials 31.0 Industrial chemicals 35.2 36.6 60.0 Intermediate products 1 146.0 Other chemicals 17.0 0.0 - 14.8 Intermediate products 11 9.0 Rubber products 45.0 29.6 29.3 Nondurable consumer 50.0 Glass 41.8 12.3 14.6 goods Other nonmetallic mineral 11.0 0.0 -1.9 Consumer durables 145.0 products Machinery 120.0 Basic ferrous metal 48.2 60.7 84.8 Transport equipment 207.0 products Basic nonferrous metal 44.5 47.0 88.0 Equally weighted 83.9 products arithmetic mean Metal products 45.9 10.0 -11.5 Standard deviation 69.7 Nonelectrical machinery 65.5 19.7 -4.7 Variability coefficient 0.8 Electrical machinery 61.3 55.7 77.6 Range 215.0 Transport equipment 87.2 29.7 3.5 Scientific and other 50.0 73.3 92.6 equipment Weighted average 52.7 37.1 39.1 B. Openness Measure 1929 1951-55 1965-70 1971-73 1974-79 1980-81 Share of foreign trade in GDP 36.0 39.4 15.6 17.0 19.2 17.6 Note: Estimates are based on price comparisons; 1977 estimates are for ninety products and probably underestimate protection because of the prevailing high real exchange rate in 1977 (see figure 1). Source: For 1968: Berlinski and Schydlowsky 1982; for 1977: Nogues 1986. starting in 1978, they opted for exchange rate stabilization.4 Governments expected these "orthodox" measures to be contrac- tionary, but thought that their potential benefits would easily outweigh the temporary costs of recession. Restoring external balance was much more important in Argentina and Chile than in Uruguay. In Chile, the short-run recession was the deepest; the external shock in 1974-75 was the greatest (Corbo and de Melo 1987); and inflation was also reduced the most. In Argentina, where inflation was more rapid, concern about the political consequences of unemployment limited the stabilization effort (Fernandez 1985). Anti-inflationary measures were considered insufficient to cure bal- ance of payments difficulties, so each country also made a major effort Vittorio Corbo and Jaime de Melo 113 Table 2. Protection and Trade Openness in Chile (percent) A. Protection Effective protection Sector 1967 1974 1979 Consumer goods 138.8 189.7 13.2 Intermediate goods 172.9 139.6 14.0 Machinery and transport equipment 265.3 96.0 13.0 Equally weighted arithmetic mean 176.7 151.4 13.61 Standard deviation 279 60.4 1.7 Variability coefficient 1.57 0.399 0.124 Range 1,163 216 6 B. Openness Measure 1929 1951-55 1965-70 1971-73 1974-79 1980-82 Share of foreign trade in GDP 66.3 21.7 24.0 20.3 36.1 32.6 Note: Data on protection are simple averages. Source: For protection data: Corbo and Meller 1981, Aedo and Lagos 1984; for trade and GDP data: International Monetary Fund, International Financial Statistics, various issues. to switch spending.5 They also eliminated multiple exchange rates for commodity trade and, more important, followed their initial devalua- tions of the exchange rate with a passive crawling peg regime. It was intended to nriaintain purchasing power parity adjusted by changes in the terms of trade. What Should Have Been Done Countries with rapid inflation will face complications if they try to liberalize their economies at the same time as stabilizing them. On the one hand, the success of stabilization depends on squeezing the econo- my as a whole; on the other, trade liberalization calls for the contrac- tion of highly protected import-substituting firms and the expansion of export-oriented sectors. If both programrts are applied simultaneous- ly, the contractionary pressure on highly protected import-substituting activities might be too strong to withstand. A second complication is the downward inflexibility of prices. To overcome this phenomenon, trade liberalization has to be accompan- ied by an initial devaluation of the exchange rate to achieve the desired improvement in the relative prices of exportables. However, the devaluation will also push up import prices, temporarily accelerat- ing inflation. 114 Research Observer 2, no. 2 (July 1987) Table 3. Protection and Trade Openness in Uruguay (percent) A. Protection Type of protection 1968 1978 1980 1981 1982a Nominal protection Domestic sales Formal 263 86 40 46 60 Without reference prices - 72 - 39 - Implicit - 25 36 38 41 Adjustedb - - - 1 - Export sales 4 16 16 1 22 Adjustedb - - - -18 - Redundant protection on domestic salesc - 23 6 1 8 Effective protection Domestic sales Actual 384 - - 75 - Adjustedb - - - 27 - Unweighted - - - 118 - Potentiald - - - 89 - Adjustedb - - - 37 - Unweighted -- - - 107 - By sales category Durables - - - 317 - Nondurables - - - 37 - Intermediates - - - 101 - Machinerye - - - 286 - Coefficient of variation - - - 103 - Export sales Actual 37 - - 30 - Adjustedb - - - 5 - Unweighted - - - 39 - Potentiald - - - 20 - Adjusted - - - -13 - Unweighted - - - 33 - Coefficient of variation - - - 184 - B. Openness Measure 1951-55 1965-70 1971-73 1974-79 1980-82 $hareofforeigntradeinGDP 19.6 27.3 25.1 36.2 33.9 -Not available. Note: All rates, unless otherwise noted, are weighted averages; product weights are at world prices. a. Calculated with 1981 weights. b. Adjusted for exchange rate deviation from purchasing power parity. c. Computed as the difference between the formal nominal rate of protection and the landed price (inclusive of customs duties) of corresponding imported goods. d. Potential effective protection computed using formal nominal rates of protection. e. Includes transport equipment. Source: For 1968: Bension and Caumont 1981; for 1978-82: Centro de Investigaciones Econ6micas 1983 and Mezzera and de Melo 1985. Vittorio Corbo and Jaime de Melo 115 Table 4. Macroeconomic Indicators for Chile, Uruguay, and Argentina, by Period, 1941-83 Prereformn Reform Postreform Country . cnis and indicator 1941-SO 1951-60 1961-70 1965-70 1971-73 1974-76 1977-78 1979-81 1982-83 Chile Average annual growth (percent) Grossdomesticproduct 4.0 4.4 4.5 4.1 1.3 -1.8 7.8 6.9 -7.4 Expenditure 4.0 4.5 4.0 5.1 1.3 -8.2 11.9 10.2 -14.4 Exportsa 11.6 3.4 9.4 11.8 9.9 23.5 7.9 20.4 -0.1 Importsa 12.3 8.8 6.5 8.1 5.6 22.3 35.2 28.7 -30.2 Gross fixed investment - 41.8 1.7 5.3 -9.8 -7.8 16.7 17.8 -26.6 Consumer prices - 37.6 27.2 23.3 149.7 358.0 79.0 30.2 11.7 Average level Fiscal deficit/GDP - - 1.6 2.1 16.1 5.1 1.3 -2.1 3.1 Unemployment (percent) - - - 6.0 4.6 14.2 13.6 12.2 22.2 Real wage (1969 = 100) - - - 98.0 98.0 69.0 82.0 100.0 82.0 Gross investment/GDP - 10.0 15.4 14.4 12.1 16.0 14.0 16.7 13.3 Terms of trade (1968 = 100) 63.3 73.4 89.4 101.4 92.7 79.0 60.3 56.1 42.5 Current account/GDP - 1.1 2.4 1.4 2.9 2.6 5.6 9.1 7.4 Uruguay Average annual growth (percent) Gross domesticproduct 4 0.04 1.6 2.1 -0.4 4.3 3.2 4.7 -7.2 Expenditure 4 0.8 1.3 2.9 -0.2 1.9 3.6 5.6 -11.2 Exports, 16.5 -3.7 6.5 3.9 16.8 21.4 10.2 21.8 -2.9 Imports, 16.3 5.7 1.9 3.9 8.7 30.2 14.0 32.1 -30.7 Grossfixedinvestment - 5.8 -1.5 7.3 -10.8 25. 10.5 6.9 -24.8 Consumer prices - 23.2 47.9 49.8 62.7 69.2 51.3 54 33.3 Average level Fiscal deficit/GDP - - - 1.9 3.2 3.8 1.9 0 6.4b Unemployment (percent) - - - 8.2 8.1 9.7 12.4 8.4 13.7b For countries with annual inflation of 25 percent or more, today's emerging consensus would therefore suggest that stabilization should precede liberalization (Fischer 1986a, Sachs 1986). This is because inflation reduces substantially the information content of relative prices, and the main aim of liberalization is to adjust relative prices in accordance with economic costs. Inflation also has other side effects that damage economic performance. These are well documented (Fischer 1986b, Yeager 1981) and are of four main kinds. First, with rapid inflation, changes in the rate do nct affect all prices and costs uniformly and at the same time. This makes relative prices very volatile, reducing their information value. Second, in countries with high inflation, interest rates are usually controlled. The result is nega- tive real rates, which lead to credit rationing, distorted investment 116 Research Observer 2, no. 2 (July 1987) Table 4 (continued) Prereform Reform Postreform Country crises, and indicator 1941-SO 1951-60 1961-70 1965-70 1971-73 1974-76 1977-78 1979-81 1982-83 Uruguay (cont.) Real wage (1969 = 100) - - - 104 102 86 70 64 54b Grossinvestment/GDP - 13.0 12.2 11.3 10.1 13. 15.6 16.1 13.5b Terms of trade (1968 = 100) 114.5 115.6 109.5 107.2 134.9 76 90.7 89 80.3b Current account/GDP - - 3.5 0.05 -0.5 3.4 3.2 5.4 0.7b 1941-50 1951-60 1961-70 196S-73 1973-75 1976-78 1979-80 1982-83 Argentina Average annual growth (percent) Gross domestic product 2.5 3.5 4.4 4.2 2.9 0.8 0.9 -3 Expenditure - 6.5 3.1 4.2 3.3 -1.6 3.4 -6.1 Exportsa 13.0 2.2 5.7 11.5 21.3 29.9 2.7 -33.5 Imports, 20.1 5.6 4.3 9.0 29.6 2.1 55.8 -15.6 Gross fixed investment - 13.5 30.6 2.6 7.6 -1.1 -4.0 -3 Consumer prices 15.3 30.5 21.5 29.9 138.7 225.5 100.8 188.5 Average level Publicsectordeficit/GDP 5.9 6.0 4.4 3.7 10.1 11.6 10.1 17.8 Unemployment (percent) - - - 5.7 2.4 3.4 2.2 4.7 Realwage (1969 = 100) - - - 125 154 100 118 111 Gross investment/GDP - 22.5 18.8 19.1 21.4 26.4 22.8 17.2 Terms oftrade(1968 = 100) 124.6 100.8 103 120.1 133.4 89 86.5 84.0 Current account/GDP - 1.9 -0.5 0.2 1.5 -2.1 1.8 1.8 -Not available. a. Imports and exports are valued in dollars. b. Data are for 1982-84 Source: National accounts and Corbo and others 1986, table 1; for public sector deficit/GDP in Argentina: Cavallo and Pefia 1983. decisions, and a shrinking of the formal financial system. Third, be- cause of uncertainty about future inflation, financial transactions tend to be concentrated in short-term instruments, thus reducing the supply of long-term investment finance. Fourth, rapid inflation is also asso- ciated with external crisis, as periodic attempts to control inflation through the exchange rate mean that the currency appreciates in real terms. The resulting balance of payments difficulties and capital flight reduce both actual and potential output. The recommendation to start with a stabilization program also stems from the fact that successful liberalization depends on credibili- ty and on having a stable and competitive real exchange rate. Both these objectives are difficult to attain when inflation is rapid. Not surprisingly, there are few historical examples in which stabilization Vittorio Corbo and Jaime de Melo 117 and liberalization were achieved simultaneously. Indeed, one of the most extensive studies of trade liberalization has concluded that its failure stemmed mainly from the failure of the accompanying anti- inflationary programs (Krueger 1978, 1981).6 For countries with annual inflation rates of, say, 15-25 percent, stabilization still remains a high priority, but liberalization can be introduced at the same time. However, any stabilization program should avoid measures that could jeopardize successful liberalization: the most obvious example is real exchange rate appreciation.7 Nor should export taxes be used in countries where the antiexport bias of the trade regime needs to be corrected. In countries with inflation rates below 2'i percent a year, macropoli- cies should be designed to maintain an "appropriate" and stable real exchange rate. For economies operating fixed rate or crawling peg regimes, the initial reduction in tariffs should be accompanied by a devaluation that, although it would not restore the landed prices of imports, would ensure that the relative prices of exports improves (Mussa 1986). The same applies to those countries that have long discriminated against exports. Besides exchange rate policy, other macroeconomic reforms should also support and ensure confidence in the liberalization effort. Thus, monetary expansion should be compatible with the rules for pegging exchange rates, and fiscal policy should try to ensure that the budget deficit is compatible with the domestic credit expansion resulting from a stable pegging rule (Buiter 1986) and does not crowd out the financ- ing of sectors that are meant to expand. Credit policy should ensure access to credit at competitive rates for thie expanding sectors, while denying cheap credit to the previously heavily protected sectors (be- cause its availability could retard their adjustment). And the labor market should be flexible enough to allow for a fall in the consump- tion wage in previously heavily protected sectors or a reallocation of labor toward the sectors that were previcusly discriminated against. Otherwise, unemployment will result. Evaluation The initial thrust of reform in the three countries was certainly in line with the framework suggested above. Yet several years after the contractionary policies had been applied, inflation remained disturb- ingly rapid.8 In Argentina, much of the public sector deficit continued to be monetized until late 1978. In Chile, although the public sector deficit was transformed into a surplus by 1978, the reduction of inflation was bound to take time, because retroactive inclexation (Fischer 1984) was widespread (Corbo 1985a, 1985b). Uruguay's inflation, although ini- 118 Research Observer 2, no. 2 (July 1987) tially much lower than that of the other two countries, proved intrac- table because many nontradables were affected by spillover demand from Argentina (Hanson and de Melo 1985). Persistent inflation prompted a major shift in tactics in all three countries. In a second phase of stabilization policy, the intended path of the exchange rate was preannounced, so as to control the evolution in the price of tradables as well as overall inflationary expectations. This policy corresponded to an "active" crawling peg and was clearly a departure from orthodoxy.9 In practice, the schedule of devalua- tions, known as the tablita, proved to be less than the difference between domestic and world inflation. Proponents of the new approach thought that purchasing power parity (especially in Chile) and interest parity (in Argentina and Uru- guay) would come fairly rapidly as the result of competition in free commodity and capital markets. In Argentina and Uruguay, the anti- inflationary policy took precedence over other economic objectives with the adoption of the tablita in December 1978. In Chile, too, bringing down inflation became a main concern when the tablita was adopted in February 1978. All three countries, but especially Argen- tina and Uruguay, sometimes used tariff reductions to impose price discipline rather than to rationalize the trade regime (see tables 5-7). At the time, the exchange rate approach to stabilization was a seductive novelty. Several other countries-including Brazil, Israel, Peru, Portugal, Sri Lanka, and Turkey-flirted with it, hoping, like the Southern Cone countries, to avoid the recessionary costs known to accompany orthodox stabilization. But the others were more pragmat- ic than the Southern Cone countries, quickly abandoning the ex- change rate strategy when tradable sectors became seriously uncompe- titive. Each of the three countries departed from our suggested framework inasmuch as they began to liberalize before inflation had fallen below 50 percent a year. This departure could conceivably be defended on the grounds that, with distortions so widespread, there were substan- tial gains to be had from liberalization. However, the danger of an inappropriate mix of policies was quite high. In particular, the three countries ignored one of the key ingredients of successful liberaliza- tions-the maintenance of an appropriate and stable (but not neces- sarily constant) real exchange rate. In each country, the large real appreciation of the exchange rate after 1978 weakened the extent and credibility of the liberalization effort (see figure 1 on p. 126). What Was Done Liberalization With different timing and intensity, all three countries removed Reforms price controls, liberalized interest rates, reduced restrictions on trade Vittorio Corbo and Jaime de Melo 119 Table 5. Stabilization and Liberalization Measures in Argentina, 1976-82 Year Stabilization Liberalization 1976 Exchange rate policy: (April) stabilization of exchange Domestic product markets: (April) prices gradually liberalized. rates from multiple rate system to dual rate (commercial Labor markets: arrangements made for periodic adjustment of and financial) system with fixed but periodically nal wages. adjustable parities. (December) convergence of financial and commercial Domestic financial markets: (May) preliminary attempt to correct exchange rates. negative real interest rates for borrowing through new tax on loans. Externalfinancialflows: (July) liberalization of rules for negotiating foreign exchange loans. Commercial policy: (April) progressive removal of prior peso deposit requirements and of quotas on imports. 1977 Monetary policy: (March) imposition of 120-day price Domestic financial markets: (January) repeal earlier law that had control period to reduce inflation. nationalized bank deposits. Commercial policy- further relxair,nn of quotas on mpori-2 foreign exchange reserves increase. 0 1978 Exchange rate policy: (December) preannouncement of Domestic financial markets: (October) authorization of use of gold peso/U.S. dollar exchange rates through end of August coins in bureaus of exchange. t 1979. Externalfinancialflows: progressive removal of restrictions on foreign exchange transactions of less than US$5,000 (June) to less than US$20,000 (September). Commercial policy: (December) elimination of prior peso deposit requirement for financing foreign trade; program announced for reduction of tariffs to 16 percent average and elimination of export 4 taxes by 1986. 1979 Exchange rate policy: (January) preannouncement of Domestic product markets: prices progressively decontrolled until peso/U.S. dollar exchange rates through end of 1982. December 1979. (October) preannouncement of peso/ U.S. dollar Labor markets: wages progressively decontrolled until 1982. 0- exchange rate forJanuary 1980 and of formula for Domesticfinancial markets: introduction of fractional reserve determining future month-by-month rates. requirements for financial institutions; extension of central bank guarantee of deposits to all authorized institutions. Commercial policy: (June) 1978 tariff reduction program accelerated. 1980 Exchange rate policy: (September) announcement of Domestic financial markets: (April) guarantee on deposits raised. 1 percent devaluation for October 1980 and following months. (December) announcement of peso/U.S. dollar buy/sell rates for December 1980 to March 1981. 1982 Monetary policy: (July) imposition of 100 percent labor markets: (July-August) attempt to obtain voluntary reserve requirements for bank deposits and regulated wage-price agreement. credit allowances. Domestic financial markets: (May-June) guarantee on deposits Exchange rate policy: (July) reintroduction of dual lowered. exchange rate system. Externalfinancialflows: (September) authorization to sell foreign exchange obtained on commercial (85 percent) and financial (15 percent) markets. Table 6. Stabilization and Liberalization Measures in Chile, 1973-82 Year Stabilization Liberalization 1973-74 Fiscal policy: start of sale of publicly owned firms Domestic product markets: (October 1973) many product (500 sold by 1979). prices deregulated. Exchange rate policy: (October 1973) multiple Domestic financial markets: (May 1974) interest rates freed for exchange rate reduced to three-rate system; 300 capital market transactions by Financieras. percent devaluation and establishment of crawling (October 1974) interest rates freed for commercial banks; peg. 1974 maximum debt-capital ratio set at 20. Externalfinancialflows: regulations governing inflows of external funds into Chilean banks liberalized. Commercial policy: (late 1973 through 1974) new government removes quotas and reduces average tariff from 105 percent to 69 percent; maximum tariff rate cut from 750 percent to 140 percent. 1975 Monetary policy: restrictive monetary policy to cope Taxation: (early) extension of value-added tax coverage; with 1975 balance of payments crisis. one-year surcharge on personal income tax; and 10 percent Fiscal policy: (early) reduction of deficit from 30.5 consumption tax on luxury goods. percent of GDP to 2.6 percent in one year; program involved an across-the-board cut of at least 19 percent in government spending on goods and services, sale of government assets, and improved tax collection. a Exchange rate policy: introduction of unified exchange rate. 1976 Commercial policy: new tariff structure proposed with rates of 25, 30, and 35 percent for primary, semimanufactured, and manufactured goods. o 1977 Exchange rate policy: devaluation amounts linked to Externalfinancialflows: (September) commercial banks consumer price index; inflation reduced to 3-4 authorized to intermediate capital inflows up to monthly ceiling percent a month. of inflows of 5 percent of each bank's capital and reserves. Commercial policy: following Chile's withdrawal from the 00 Andean Pact, effective tariffs of 10-35 percent proposed for implementation by mid-1977. 1978 Exchange rate policy: tablita introduced with devaluation rate above monthly change in consumer Z, prices to compensate for tariff cuts and lower rates cm thereafter. ° 1979 Labor markets: labor union activity generally diminished since 1973; collective bargaining now authorized at plant level only; wage floor set at previous wage indexed for consumer price increases. a- Externalfinancialflows: global limits on external borrowing eliminated; controls now only overall 20:1 ratio of borrowing to capital and reserves and 5 percent limit on monthly inflows. a Commercial policy: uniform tariff of 10 percent set (except for cars with engines larger than 850 cubic centimeters). 1980 Exchange rate policy: nominal rate fixed at mid-1977 Externalfinancialflows: (April) limit on monthly inflows level. eliminated; only 20:1 overall borrowing ceiling and applicable reserve requirements retained. 1981 Labor markets: legislation enacted allowing workers to negotiate fringe benefits and employers to fire workers without giving cause; minimum coverage wage limited to workers age 21-65. 1982 Exchange rate policy: (June) fixed rate abandoned; Labor markets: (June) wage indexation suspended. 18 percent devaluation. (December) legislation enacted to correct distorting effect of (August) initially "clean" but later "dirty" float. wage floor. (September) new rate based on highest dirty float rate, with monthly devaluation in line with change in consumer prices. Table 7. Stabilization and Liberalization Measures in Uruguay, 1974-82 Year Stabilization Liberalization 1974 Exchange rate policy: exchange rates for capital Domestic product markets: (July) liberalization of domestic prices transactions freely determined; passive crawling peg for of nonessential goods begins (94 percent of consumer prices goods transactions. hitherto controlled). Taxation: (July) removal of personal income and inheritance taxes; corporate profits tax (25 percent rate) established with remission for exporters' reinvested profits. Domestic financial markets: (September) gradual lifting of interest ceiling on peso loans. Externalfinancialflows: (September) liberalization of capital market and regulations on foreign exchange holdings and transactions; de facto convertibility of the peso through unrestricted purchase or sale of assets denominated in foreign currency. Commercial policy: (July) removal of export taxes on beef and wool; removal of some administrative and financial restrictions on imports. 1975 Monetary policy: (May) establishment of system of Domestic product markets: (July-December) liberalization of alloting credit; central bank to pay interest on reserves 13 percent of consumer goods prices. required by law, Externaifinanciafiows: (May) authorization of repatriation of O earnings, profits, and capital by foreigners. Commercial policy: (January) removal of remaining quotas. J>, 1976 Domestic product markets: (February) liberalization of prices of z nonconsumer goods, except monopolies. ° (later) liberalization of another 25 percent of consumer prices. c ' :Domestic financial markets: (March-September) domestic interest rates effectively freed; interest ceiling of 62 percent a year. 05 Externalfinancialflows: (midyear) authorization of foreign currency trading through other than commercial banks. 1977 Domestic financial markets: (midyear) relaxation of banking law limiting number of financial intermediaries. c (November) commercial banks permitted to pay interest on cash amounts. 0 1978 Exchange rate policy: (October) effective 90-day Domestic product markets: (July) replacement of official predetermination of exchange rate by forward sale of price-fixing agency by new agency to promote competition and three-month treasury bills redeemable in U.S. dollars; price flexibility. official unification of formerly dual foreign exchange (August) liberalization of another 13 percent of consumer goods market. prices. Domestic financial markets: (October) introduction of nil marginal reserve requirement and 20 percent unified legal reserve requirement. Commercial policy: (December) initiation of tariff reduction program to achieve a uniform 30 percent tariff by 1985. 1979 Exchange rate policy: (October) capital and commercial Domestic product markets: (March) reduction of list of goods and market foreign exchange is unified. services with administratively fixed prices. Taxation: (November) social security taxes reduced; banking tax and tax concessions to exporters abolished; 18 percent value added tax imposed. Domestic financial markets: (May) elimination of 8.4 percent banking tax and legal reserve requirements. Commercial policy: (early) removal of interest subsidies for exports. (February) acceleration of tariff reductions. (September) further tariff cuts. (November) elimination of remission of profits tax for exports and of subsidized credit to exporters. 1980 Domestic product markets: (June) official fixing of car prices because of lack of foreign competition; only 29 percent of consumer goods prices and 14 percent of agricultural prices still fixed. 1982 Exchange rate policy: (June) imposition of 10 percent import surcharge and 10 percent increase in reintegros. and capital flows, and partly deregulated their labor markets (see tables 5-7). The sequence varied from cotntry to country, although financial market deregulation proceeded rapidly in all three. Uruguay removed all controls on capital flows and on many prices early on, but was slower to liberalize for- Figure 1 eign trade. It went furthest in re- Argentina Chile Uruguay forming its fiscal system, elimi- 130 180 nating income tax, and moving to a value added tax. Chile did most 100 t ; \ 140 to abolish price controls and re- w / ^ _1 ~~~~~~~~duce trade barriers, but it main- > ~~~ Br' 90 s><,^s_J 9 . >\tained controls on short-term 5.% - ~ '*., *s . J . -capital flows for a long period. It also retained some important la- \ VA, bor market regulations. Argen- A\ \ /61 8v. tina eliminated price controls M/ and removed most restrictions ,f \ / / \ / on medium-term (more than one 0% i j year) capital flows. It also got 20% I. l l l l l l l l l l rid of many import quotas (with 1977 1982 1977 1982 1977 1982 some important exceptions) be- Left scale: fore making some reductions in --Real interest rate (five-quarter moving average of annual lending rate tariffs. Uruguay virtually elimi- in percent) - Real growth of GDP (fourth quarter 1978=100) nated price controls by the end of Right scale: ----Real exchange rate (fourth quarter 1978=100) 1979, but did little to reduce pro- tection. Contrary to popular belief, only Chile liberalized its trade substan- tially. Effective rates of protection remained high in Argentina and Uru- guay; pressure from foreign competition was felt only at the height of the overvaluation of real exchange rates. For example, in Uruguay redundant protection was not eliminated until 1981 (see table 7 and figure 1). At that time the bias against export sales was still 35 percent.'0 As far as the reform of domestic financial markets was concerned, nonprice allocation of credit and strongly negative real interest rates had long been widespread in all three countries. Each began by pro- gressively eliminating ceilings on interest rates and then reduced re- strictions on financial intermediaries. Argentina went from 100 per- cent reserve requirements and directed credit programs to a decentral- ized fractional reserve system. The Chilean reforms began by allow- ing nonbank intermediaries to operate without interest rate controls. Then, over several years, interest rate ceilings for commercial banks were removed, and state-owned commercial banks were returned to the private sector. In Uruguay, starting in 1974, the government legal- ized dollar deposits and dismantled its directed credit programs; in 1977, it lifted controls on entry to the banking system. 126 Research Observer 2, no. 2 (July 1987) As for international capital flows, Uruguay legalized unrestricted movements of private capital as early as 1974 and reached full conver- tibility by early 1977. Argentina eliminated most controls on capital movements between 1977 and 1979. Chile progressively deregulated flows of medium-term capital, eliminating global limits on borrowing in 1979 and restrictions on monthly inflows in April 1980. But restric- tions on short-term capital inflows were not dismantled until late 1981. Labor markets were changed little in the three countries. They continued to be controlled through penalties or prohibitions on labor dismissals, together with legislated wages and wage indexation. How- ever, the weakening of trade union power in the early stages of the reforms amounted to some deregulation. What Should Have Been Done There is broad agreement on the general principles that define the contents of any liberalization package, but room for much disagree- ment on the sequencing, speed, and extent of liberalization. In highly regulated economies with widespread price controls, the right ap- proach would be to lift price controls so as to improve resource allocation, while simultaneously deregulating domestic labor and capi- tal markets. Financial markets should be deregulated (subject to ap- propriate rules for banking supervision) so as to improve the al- location of credit and thus of investment. Similarly, labor market restrictions should be lifted so as to facilitate the contraction of inefficient businesses and the expansion of new, efficient ones. As for trade policies, quotas should first be replaced with equiva- lent tariffs. Then the range of tariff rates and their average level should be reduced, with the aim of achieving a low, uniform tariff. As a rule, there should be no discrimination between exporting and import-competing activities (Little and others 1970, Corden 1974, Ba- lassa 1976)." The implementation of reforms is more complicated, because is raises questions of dynamics (about which little is known). The litera- ture covers two sets of issues: the speed of reforms and their sequenc- ing. Speed of reforms. In considering whether policies should be changed gradually or at a stroke, it is essential to remember that liberalization is not an end in itself, but a means to achieving a more efficient use of resources. Because resource allocation depends on expected prices, the credibility of the reforms is very important. They therefore need realistic timetables-and what is realistic may differ from one policy area to another and from country to country. Any reform package that ignores the pace at which individuals and organi- Vittorio Corbo and Jaime de Melo 127 zations can adjust-a variable that is partly determined by political circumstances-runs the risk of failure and undermines the credibility of future reforms. Indeed, the theory of rational expectations demon- strates that coherence and credibility are important in determining the likely effect of reforms on, for example, investment and labor hiring and thus on the program's success or failure (Calvo 1986a, 1986b). Credibility can be enhanced by including policies that are intended to speed up adjustment. In regard to foreign trade, for example, the pace of liberalization must depend on how quickly resources can be expected to be reallocated to the sectors that have hitherto been discriminated against; otherwise substantial unemployment would re- sult. The initial conditions in each country are an important factor: for example, the smaller the investment-GNP ratio, the slower should trade be liberalized. Similarly, the more competitive and flexible the labor market, the faster can trade liberalization proceed."2 Financial deregulation can also be problernatical if initial conditions are overlooked. For example, if a large proportion of the assets of financial institutions are held at rates below the market and lending rates are substantially negative in real terms, then deregulation of interest rates will create difficulties for existing institutions. In particu- lar, if deposit and lending rates are deregulated simultaneously and new banks are allowed into the financial system, existing banks will be forced to pay market rates. They may then suffer substantial capital losses, jeopardizing the banking system's solvency. This im- plies the need for a transition phase, in which lending rates are deregulated, with deposit rates following gradually. The liberalization of capital flows provides another example of the importance of taking initial conditions intc account. The vital ques- tion is whether domestic interest rates are hvigher or lower than those prevailing abroad (when expressed in the same currency), since the difference between the two will determine whether there are net capi- tal inflows or outflows. Large capital inflows will push up the real exchange rate, which in turn will jeopardize the success of trade liberalization (Bruno 1983, Corbo 1983, de Melo 1987). Whatever the initial conditions, a substantial reform program im- plemented briskly and to an agreed timetable offers major advantages. First, the required reallocation of resources will not occur unless the reforms provide strong and clear signals. Second, an unduly slow pace of reforms will delay the emergence of export activities and of interest groups whose support for the reforms could help counter the antagon- ism of those who have been benefiting from the protective regime (Papageorgiou and others 1986). Sequencing of Reforms. Here, too, economic theory offers little guidance about the optimal sequence for removing distortions when many markets are initially regulated. Given that the goal of liberaliza- 128 Research Observer 2, no. 2 (July 1987) tion is to improve resource allocation, however, some broad principles can be derived. The first stage is to deregulate domestic markets to ensure that resources are reallocated. The second stage involves liber- alizing foreign economic relations. It is usually argued that the current account of the balance of payments should be liberalized first, then the capital account. Much has been written on this issue (see McKin- non 1982, Frenkel 1982 and 1983, Krueger 1984, Edwards 1985). Two arguments have been advanced for liberalizing domestic mar- kets and the current account before the capital account. First, because asset prices are determined by the present value of income streams, income streams generated by distorted prices will result in assets being traded at distorted prices (Krueger 1984). Second, since asset markets typically adjust much faster than commodity markets, liberalization of the capital account could result in large capital flows-with unwanted consequences for the real exchange rate. By the same argument, the current and capital accounts must, by definition, be brought into line with each other, and thus the speeds at which they adjust must be harmonized. It is much easier to achieve this by slowing down capital flows than by accelerating the liberalization of the current account (Frenkel 1983, Edwards 1985). This point could be extended by argu- ing that, within the current account, imports respond faster than exports: thus, opening up the capital account first could jeopardize the trade liberalization by producing a sharp increase in imports much in advance of the export expansion. Evaluation The Southern Cone countries did not conform to the emerging consensus on the sequencing of liberalization. Argentina and especial- ly Uruguay deregulated capital flows early on. The Uruguayan experi- ence is interesting, since none of the perverse side-effects (such as real exchange rate appreciation) occurred so long as the exchange rate was not used to bring down inflation. Indeed, the contractionary effects associated with orthodox stabilization were avoided because of capi- tal repatriation and domestic financial deregulation (de Melo 1987). These two factors, combined with a stable real exchange rate, were the keys to faster growth in 1974-78-though other factors, including higher savings and investment rates, also helped. Chile, in contrast, followed the recommended sequence of liberaliz- ing the current account first and the capital account later. However, a combination of retroactive indexation of wages and interest rates and substantial capital inflows produced a large real appreciation of the exchange rate, which put the trade liberalization in jeopardy. In all three countries, the most far-reaching reform was the deregu- lation of financial markets. This is not surprising: one might reason- Vittorio Corbo and Jaime de Melo 129 ably expect much less resistance from threatened interest groups to financial liberalization than, say, a lowering of trade barriers or a freeing of labor markets. Eventually all three countries also decon- trolled short-term flows of external capital--something rarely done in developing countries-but only Uruguay went all the way in this. In foreign trade, only Chile virtually eliminated protection. As tables 5-7 show, deregulation was usually gradual. The excep- tions were the removal of controls on capital flows in Uruguay and to a smaller extent in Argentina, and the salle of public enterprises in Chile. The reform process cannot be criticized for abruptness; indeed, trade liberalization in Argentina and Uruguay was, if anything, too little, too late, and too slow. Outcomes The results of the reforms can be evaluated in three periods. The first corresponds to the management of the macroeconomic crisis and the reduction of some microeconomic distortions (1974-78 in Chile and Uruguay; 1976-78 in Argentina). The second starts with the use of the exchange rate to bring down inflaticn, often referred to as the tablita period, and ends with the sudden reduction in foreign finance in August 1982. The third is the post-August 1982 period. The First Period: Stabilization with Some Liberalization During the first period, all three countries were gradually adopting anti-inflation policies and liberalizing markets. In Chile, as progress was being made in reducing the fiscal deficit and in controlling infla- tion, the terms of trade fell sharply. This resulted in a loss of real income of close to 5 percent of GDP. The decline in the terms of trade plus the big reduction in the fiscal deficit produced a severe recession: GDP fell by 12.9 percent in 1975. It then grew at an average rate of 8.3 percent a year in 1975-78. The urban unemployment rate, which had reached 15.0 percent in 1975 and 16.3 percent in 1976, was reduced to 13.3 percent in 1978. The inflation rate, which had averaged 358 percent a year in 1974-76, fell to an annual average of 79 percent in 1977-78. In spite of the sharp deterioration in terms of trade, ex- ports grew in current dollars at an average of 23.5 percent a year in 1974-76 and 7.9 percent a year in 1977-78. The fiscal deficit was only 5.1 percent of GDP in 1974-76 and 1.3 percent of GDP in 1977-78. Real wages, which had dropped by 29.6 percent between 1971-73 and 1974-76, rose by 18.8 percent between 1974-76 and 1977-78 (table 4). The tariff rate inclusive of the real exchange rate peaked in 1975 and then declined, but was almost constant between 1977 and 1978. The most unexpected results were the slowness of disinflation (des- pite a big reduction in the public sector deficit) and the high real 130 Research Observer 2, no. 2 (July 1987) interest rates that followed financial deregulation. The real interest rate on peso loans was 127.2 percent in 1975, 65.2 percent in 1976, 58.0 percent in 1977, and 43.8 percent in 1978 (Corbo 1985a). Differ- ent explanations have been offered for these high rates: the fiscal and monetary squeeze; restrictions on capital inflows; the high cost of financial intermediation arising from reserve requirements; the reduc- tion in inflation which, though slow, may nevertheless have been unexpected; and distress borrowing by firms that had been heavily protected (Edwards 1986, Ramos 1984, and Zahler 1985). These ex- planations have not been corroborated, however, and they leave unan- swered the issue of how to evaluate credit risk when loans carry such high real interest rates (an issue discussed later). Argentina's terms of trade loss was smaller than Chile's and Uru- guay's at the beginning of the reforms. But its GDP growth was even slower than in the prereform period, although exports grew at an average of 29.9 percent a year in 1976-78, and the current account had a surplus equivalent to 2.1 percent of GDP. The average unem- ployment rate increased from 2.4 percent in 1973-75 to 3.4 percent in 1974-76 and would have been higher if the military government had not deliberately kept it low to prevent political unrest. Real wages fell by 33.3 percent between 1973-75 and 1976-78. The inflation rate, which had reached 443.2 percent in 1976, was reduced to 176.1 per- cent in 1977 and 175.5 percent in 1978 (Fernandez 1985). This slow pace of disinflation is not surprising, given that the public sector deficit increased from 10.1 percent of GDP in 1973-75 to 11.6 percent in 1976-78. The freeing of Argentina's domestic interest rates and most com- modity prices in the middle of 1977 also resulted in positive real interest rates between the last quarter of 1977 (when they reached an annual rate of 10 percent) and the last quarter of 1978 (when they again turned negative until the end of phase 1 [Fernandez 1985, table 2]). Not surprisingly, these rates were lower than Chile's, since capital inflows were not restricted. In Uruguay, output grew right from the start of the reforms. This achievement was all the more remarkable because of the huge fall in the terms of trade during the reform years (table 8) in comparison with 1965-70 and the especially favorable period of 1971-73 (see table 4). Several factors accounted for this turnaround. First, the real ex- change rate became less volatile after the passive crawling peg was introduced in 1972. Second, the antiexport bias of high tariffs was partially offset by the abolition of taxes on traditional exports and the introduction of incentives for nontraditional exports. On average, exports grew by 21 percent a year during 1974-76. Their growth was helped by the accompanying reduction in costs (for example, the fall in real wages) and by expenditure-switching policies. Vittorio Corbo and Jaime de Melo 131 The removal of quotas on imports of critical goods, together with increased public investment, dramatically boosted total investment (see table 4). Between 1974 and 1978, public investment rose by nearly 400 percent in real terms, while private investment doubled. As a result, GDP growth-which averaged barely 1 percent a year in 1955-73-rose to an annual average of nearly 4 percent between 1974 and 1978. One other factor contributed to this dramatic change: a rise in capital inflows. Capital repatriation by Uiruguayans and growing de- posits by Argentines averted a severe drop in real liquidity (de Melo 1987). The inflows allowed Uruguay to prepay debt incurred under a stabilization plan of the International Monetary Fund. The cumulative totals of "net errors and omissions" in the balance of payments-a crude proxy for capital flight-went from an outflow of us$250 mil- lion for 1970-74 to an inflow of us$51 million for 1975-81. The fiscal deficit, which averaged 3.8 percent of GDP during 1974-76, was brought down to 1.9 percent during 1977-78. Neverthe- less, inflation (which had peaked at 97 percent in 1973) proved stub- born and averaged 51 percent a year over 1977-78. As in Argentina, the gradual lifting of ceilings on domestic iinterest rates raised average real borrowing interest rates. They went from a negative 30 percent in 1973 to a positive 3.6 percent in 1978 (FHanson and de Melo 1985, table 3). The Second Period: Stabilization through Preannouncement of the Exchange Rate At the start of this second phase, Argentina's public sector deficit was still almost 10 percent of GDP; Chile still had a wage indexation mechanism that was bound to result in a real appreciation of the exchange rate; and Uruguay's fiscal deficit increased substantially in 1981. The anti-inflationary programs in all three countries were flawed. In this second period, domestic demand grew faster than output in the three countries (table 4). The gap was filled by foreign finance, and exchange rates rose in real terms (figure 1). The international depreciation of the dollar and the demand pressures on nontradables limited the effectiveness of anti-inflationary policies. In Chile, the annual inflation rate was reduced from 50 percent in 1978 to 20 percent in 1981 and to zero in early 1982, but the accumulated real appreciation of the peso was large. Argentina's inflation rate only fell from 175 percent in 1978 to 101 percent in 1980. Uruguay's rate actually rose, from 44.5 percent in 1978 to 66.8 percent in 1979; this was partly due to the deregulation of domestic beef prices in August 1978 and to demand pressures resulting from heavy spending by Ar- 132 Research Observer 2, no. 2 (July 1987) gentine tourists. Inflation subsequently fell to 63.5 percent in 1980 and 34.0 percent in 1981. In all three countries, increased imports and loss of export competi- tiveness combined to raise the current account deficit. In Chile, the deficit rose from 5.6 percent of GDP in 1977-78 to 9.1 percent in 1979-81; Argentina moved from a current account surplus equal to 2.1 percent of GDP in 1976-78 to a deficit of 1.8 percent in 1979-80; and in Uruguay the deficit increased from 3.2 percent of GDP in 1977-78 to 5.4 percent in 1979-81 (table 4). Because all three econo- mies were booming, the average unemployment rate was reduced from 14.2 percent to 13.6 percent in Chile, from 12.4 percent to 8.4 percent in Uruguay and from 3.4 percent to 2.2 percent in Argentina. At least two factors contributed significantly to the increase in domestic demand and the resulting current account deficits. First, particularly in Argentina and Uruguay, the real appreciation of ex- change rates led to extra spending on imported durables while their prices were low (Dornbusch 1985). Second, the rise in asset values in all three countries during the boom phase produced a wealth effect on spending (Barandiaran 1984, Corbo 1983, Dornbusch 1985, Fernandez 1985, Hanson and de Melo 1985, and Harberger 1983). As exchange rates continued to rise in real terms, so doubts grew about the sustainability of the tablita. These doubts were reflected in growing interest rate spreads despite the shrinking (Chile) or absence (Argentina and Uruguay) of impediments to short-term capital flows. Real interest rates rose sharply, adding to the difficulties of the trad- able goods sectors. Toward the end of phase 2, companies were doing more and more borrowing to stave off bankruptcy and awaiting a bailout after devaluation (Diaz-Alejandro 1985, Tybout 1987). The Third Period: Crisis The inconsistency in the three countries' economic policies became apparent in late 1980 in Argentina and in early 1982 in Chile and Uruguay. In Argentina, with an externally financed public sector defi- cit of over 10 percent of GDP and no prospect of fiscal reform, doubts about the sustainability of the exchange rate regime began as early as the first half of 1980. They were increased by the collapse in April 1980 of the BIR (Banco de Intercambio Regional), which prompted a 25 percent increase in the money supply in a single month. When President-elect Viola refused to make a commitment about future exchange rate policy, private capital outflows accelerated. The 10 percent devaluation in February 1981 was too little and too late and only exacerbated the crisis. In Chile, despite a fiscal surplus in 1979-81 equal to 2.1 percent of GDP, the current account deficit reached 14.6 percent of GDP in 1981. Vzttorio Corbo and Jaime de Melo 133 The exchange rate rose by 29.8 percent in real terms between the second quarter of 1979 (when the rate was fixed) to the last quarter of 1981 (Corbo 1985a). Doubts about the sust:ainability of the exchange rate started to set in, with inflows of private capital decreasing from us$1.6 billion in the second half of 1981 to only us$900 million in the first half of 1982. The monetary contraction that followed resulted in high interest rates and a sharp recession. In Uruguay, the fiscal deficit (which had fallened continuously through 1980) started to increase in 1981 with an underfunded social security reform. Meanwhile, the real exchange rate rose by 27.4 per- cent between 1978 and 1981-and even more relative to Argentina, after the latter's stabilization attempt collapsed (Hanson and de Melo 1985). As doubts grew about the sustairiability of the tablita, so outflows of private capital started to increase in 1981 (de Melo 1987). Thus capital flight started in Argentina and Uruguay (and to a lesser extent in Chile) before their economies were hit by the adverse exter- nal developments of the early 1980s. The debt crisis that began in August 1982 and the interruption of voluntary capital flows had severe consequences for all three coun- tries. Chile's private sector was too dependent on foreign financing; the same was true of Argentina's public sector; Uruguay was some- where in the middle. The debt crisis closed the option of using public borrowing to finance outflows of private capital, but the private sec- tor had in fact already started to adjust. The August 1982 crisis therefore implied a faster cut in domestic demand and a faster real depreciation of the exchange rate. The stickiness of nontradable prices and wages made the recession even worse. Chile abandoned its ex- change rate regime in June 1982; Uruguay did so in November 1982. How important were the external shocks? The interest rate shock, which began to be felt after U.S. interest rates starting rising at the end of 1979, affected not only the cost of new borrowing but also the interest on existing debt. This latter effect was particularly strong in the three Southern Cone countries during 1982-83, because much of their debt carried variable interest rates. As shown in table 8, the combination of declining terms of trade and increasing interest costs amounted to 12.2 percent of GDP in Chile and 6.7 percent of GDP in Argentina. However, external shocks were insignificant during 1979-81-which was when the unsustainability of the stabilization programs based on exchange rates became apparent. The results in table 8 were confirmed by simulation of an econo- metric model estimated with annual data for 1962-83 (Corbo and de Melo 1987). In Chile, where external shocks were heaviest before the crisis, the simulations suggest that, if the average interest rate of 1974-79 and the terms of trade of 1980 had prevailed during 1981-83, GDP growth would have been only 2.1 percent a year and external 134 Research Observer 2, no. 2 (July 1987) Table 8. External Shocks in Argentina, Chile, and Uruguay (percent of GDP) Country Terms of trade Interest rate Total and period (1) (2) (1) + (2) Argentina 1976-78 -2.9 - -2.9 1979-80 -0.3 2.8 2.5 1981-82 -0.3 -6.4 -6.7 Chile 1974-78 -5.7 - -5.7 1979-81 -1.9 0.2 -1.7 1982-83 -4.8 -7.4 -12.2 Uruguay 1974-78 -7.6 - -7.6 1979-81 -0.4 0.9 0.5 1982-83 1.9 -2.9 -1.0 -Not available. Note: The real income effect of changes in terms of trade is computed from import and export unit value indexes weighted by the import share of GDP. The interest rate effect is calculated as I Pp ; PT (PX + PM) 0.5 where Px and PM are percentage changes in the export and import price indexes, and i is computed from the World Bank, World Debt Tables 1986 as the ratio of interest payments to disbursed debt and expressed in terms of GDP by multiplying it by the debt-GNP ratio. PM is measured by the export unit value of industrial countries from the International Monetary Fund, International Financial Statistics. The same methodology is used in Sachs (1985). Changes are computed with respect to the previous period. Source: Corbo and de Melo 1987. debt at end-1982 would have still been 87 percent of its actual value. In a further simulation in which a more normal expenditure path was assumed, average GDP growth in 1981-83 would have been 2.7 per- cent, and end-1982 external debt would have been only 67 percent of its actual size. This more favorable outcome comes from smaller external borrowings and lower real wages which would have benefited employment, especially in nontradables. Simulation results for Argen- tina and Uruguay were even less significant than for Chile. The first lesson relates to the microeconomic efficiency effects of the Lessons reforms. These are difficult to detect, because the reforms did not last long. This was especially true of Argentina, where deregulation was mostly confined to the financial system. Vittorio Corbo and Jaime de Melo 135 Of the other two countries, Uruguay offers the clearest evidence. During 1955-73, GDP grew at 1.0 percent a year; during 1974-83, it grew at 2.4 percent. In the second period, too, private savings and investment rose relative to their earlier trend. Controlling for factors affecting private savings and investment, it is clear that the upward shift could be attributed not mainly to financial reforms, but to the reform package as a whole.'3 In Chile, GDP growth was sustained in spite of slow growth in investment and employment, suggesting that increases in productivity and in capacity utilization were major contributors. Estimates of the growth in total factor productivity (TFP) for manufacturing show a sharp increase in the reform period: during 1960-70, TFP had declined by an annual average of 0.6 percent; during 1977-81 it grew by 2.5 percent (Mierau 1986). Further calculations from a simulation model for 1977-81-which took account of increases in capacity utilization during 1977-79-suggest that productivity gains induced by reform raised GDP in 1981 by 19 percent over what would have been achieved had the TFP growth rates for 1965-74 still prevailed (Condon and others 1985). Evidence of productivity gains also comes from interviews with entrepreneurs in each country. Several noted that productivity had risen faster than they were expecting-the result of consolidation of product lines, manpower economies, and improvements in product quality to meet foreign competition (Corbo and de Melo 1985b). Given that the reforms ultimately faced severe problems, how much did external shocks contribute to their failure? Not much evidently- our second lesson. The previous section showed that even in Chile, which was most damaged by external shocks, their contribution can- not account for the slowdown in growth during 1981-83. The third lesson relates to the need for policies to be consistent. In particular, the macroeconomic framework tnust be supportive of liber- alization. Argentina never reduced its public sector deficit below 8 percent of GDP, and meanwhile pursued mutually inconsistent ex- change rate and fiscal policies. This was especially true of the tablita phase, when the deficit fed the growing expectations of devaluation during 1980-which were in turn reflected in rising interest rates. The collapse of Argentina's exchange rate regime has been throughly stud- ied (Cumby and van Wijnbergen 1984 and Connollay 1985). Inconsis- tency of macropolicies was at its peak when the three countries were pursuing their exchange rate-based stabilization programs. These pro- grams in due course reduced inflation temporarily, but meanwhile they set a financial trap for all three economies in the form of a powerful boom-squeeze-bust sequence that led to the abandonment of exchange rate policy. Furthermore, the anti-inflation measures were slow to work through commodity and financial markets, meanwhile 136 Researcb Observer 2, no. 2 (July 1987) creating unsustainably large capital movements and rising real ex- change rates. The fourth lesson is that each country would have benefited from closer scrutiny of its banks. There is a crucial distinction between wholesale liberalization of financial markets and properly monitored deregulation. A clear understanding of this distinction could at least have mitigated several unfortunate developments: * In Chile, banks allowed the debt of affiliated firms to rise even though these firms were doing badly and should have been forced to liquidate. Hence less credit was available for more profitable independent firms (Galvez and Tybout 1985). * Bankers suddenly placed in a free market environment failed to recognize that the increase in interest rates tended to redirect their loans away from low-risk, low-return activities, resulting in "ad- verse selection" (Stiglitz and Weiss 1981). Better bank monitoring might have resulted in less upward pressure on lending rates. * De facto deposit insurance provided incentives for undue risk- taking."' Banks with poor portfolios were able to attract new funds by raising deposit rates, thereby forcing less-risky banks to match these rates. The fifth lesson is that in Chile where a combination of exter- nal shocks and inappropriate macroeconomic management resulted in large debts and then a slump in 1982-83, the reforms were resilient enough to be maintained. Today, Chile is in a sustained recovery in spite of a large overhang of debt. This article assesses the design of stabilization and liberalization programs in the Abstract Southern Cone countries of Argentina, Chile, and Uruguay. With the exception of Chile, the reforms were not as widespread as some believed. Little trade liberalization took place in Argentina and Uruguay, although some of the antiexport bias was reduced by eliminating taxes on traditional exports. In all three countries, labor markets remained fairly highly regulated, though it was easier to dismiss labor. In general, liberalization was gradual: even Chile's trade liberalization spanned five years. The article also shows that the collapse of the three economies in the early 1980s cannot be ascribed mainly to terms of trade and interest rate shocks. The main causes of failure were poorly designed programs and poor implementation. These errors included restrictive wage legislation (Chile) or political instability combined with a preoccupation with keeping unemployment as low as possible (Argentina). Monetary policy to deal with growing fiscal deficits was inconsistent with the accompanying exchange rate policy (Argentina throughout its reform period and Uruguay toward the end of its reforms). Financial deregulation was not matched by appropriate supervision of the financial institutions. The article suggests several policy lessons for countries attempting to resume growth and restore external balance through a combination of liberalization and stabilization policies. First, it finds evidence that reductions in distortions produced efficiency gains in Chile and Uruguay even though Uruguay's reforms were short-lived. Second, the article shows that policy inconsistencies undermined the credibility of the later stages of reform in all three countries, eventually producing a crisis. Third, it presents data Vittorio Corbo and Jaime de Melo 137 that call into question the use of exchange rate-based stabilization, because of the slow convergence of domestic prices and interest rates to international levels, which in turn can produce unsustainably large capital movements. Fourth, the article stresses the need for caution in financial deregulation. Notes We thank Edgardo Barandiaran, Julio Nogues, and William Tyler for their comments. We also thank Ricardo Caballero and Gabriel Castillci for research assistance, Myriam Bailey and Jackson Magargee for word-processing, and Peter Bocock for editorial help. 1. The introduction of crawling peg exchange rate regimes in the mid 1960s in Chile and later in Uruguay (1972) and Argentina did reduce the more extreme fluctuations in the real exchange rate, but imbalances persisted. 2. Annual inflation rates approached 1,000 percent in Chile (September 1973), 2,300 percent in Argentina (March 1976), and 100 percent in Uruguay. The fiscal deficits were substantial well before the collapse of the civilian governments (table 4). In Argentina and Chile inefficient public enterprises contributed to high public sector deficits. 3. Much has been written in the last three years on this topic. Our purpose here is to summarize the main reforms. This section draws mostly on Corbo and others (1986) and Corbo and de Melo (1985a). Other references are Calvo (1986a), Edwards (1985), Harberger (1982), Rodriguez (1982), and Sjaastad (1983). 4. Chile's substantial and chronic fiscal deficit was eliminated by drastic across-the- board expenditure cuts (amounting to 15 percent in 1975 alone), followed by a tax reform. In Uruguay the fiscal deficit was reduced yearly until 1980. Much credit should be given to the rationalization of taxation, including the introduction of a value added tax, which improved fiscal performance compared with the poor record of the prereform period (see Harberger and Wisecarver 1977). In Argentina, on the other hand, the fiscal deficit was never controlled (Cavallo and Pefia 1983). 5. In Chile, the switching was achieved through a large real devaluation and reduction of barriers to imports. In Argentina, switching efforts included a combination of real devaluation, reduction of taxes on exports, and some reduction of import barriers. In Uruguay, expenditure switching was combined with a real devaluation, reduction of barriers to imports, and introduction of subsidies for nontraditional exports. 6. Other recent analyses of stabilization and liberalization policies (Killick 1984 and Lin 1985) have shown that simultaneous application of the two is unlikely to be sustainable and successful. 7. The view that real exchange rate appreciation to bring down inflation should be avoided owes much to Southern Cone experience with this policy, as will be discussed later. 8. Chile's rate of inflation was around 50 percent in late 1977, Argentina's was 166 percent in late 1978, and Uruguay's was roughly 50 percent in late 1978. 9. How stabilization based on exchange rate was supposed to work is described in Rodriguez (1983). Comparisons of the two approaches is provided in Dornbusch (1982). 10. For further discussion see Nogues (1986) and Petrei and de Melo (1985) on Argentina and Mezzera and de Melo (1985) on Uruguay. 11. For infant industries a timetable of reduction in protection over, say, a five-year period should be followed. See Balassa (1976) and Bell and others (1984). And, for countries with export earnings derived from natural resource-based products, it is appropriate and accepted to tax windfall gains during commodity booms and to offer rebates to producers during troughs. See Davis (1983). 12. Lessons from interviews with managers of manufacturing firm in the Southern Cone countries are summarized in Corbo and de Melo (1985b). They found that major efficiency gains were achieved in a short period for some firms but that others delayed 138 Research Observer 2, no. 2 (July 1987) adjustment because of skepticism about the reforms and, in some instances, because of high costs for severance payments. 13. Controlling for other factors, de Melo and Tybout (1986) showed that savings and investment rose during the reform period. However, they could not attribute this rise to reforms only in financial markets and suggested that fiscal reforms also played an important role. 14. For reference, the first major bank failures in each country occurred on the following dates: Argentina (BIR), March 1980; Chile (Banco Osorno), 1977; Uruguay (Banco Panamericano), 1979. In all cases depositors incurred no financial losses. The "moral hazard" effect is further elaborated in Diaz Alejandro (1985) and Tybout (1987). Aedo, C., and L. F. Lagos. 1984. "Protecci6n Efectiva en Chile 1974-1979." Documento References de Trabajo 94. Instituto de Economia, Pontificfia Universidad Catolica de Chile, Santiago. Balassa, B. 1976. "Reforming the System of Incentives in Developing Countries." World Development 3, no. 6: 365-82. Barandiaran, E. 1984. "Nuestra Crisis Financiera." Estudios Publicos (Fall): 89-122. Bell, M., and others. 1984. "Assessing the Performance of Infant Industries." Journal of Development Economics 16, no. 1-2: 101-28. Bension, A., and J. Caumont. 1981. "Uruguay: Alternative Trade Strategies and Employment Implications." In A. 0. Krueger and others, eds. Trade and Employment in Developing Countries. Chicago, Ill.: University of Chicago Press. Berlinski, J., and D. Schydlowsky. 1982. "Incentive Policy and Economic Development: Argentina." In B. Balassa and Associates. Development Strategies in Semi-Industrial Economies. Baltimore, Md.: Johns Hopkins University Press. Bruno, M. 1983. "Real versus Financial Openness under Alternative Exchange Rate Regimes." In P. Aspe and others, eds. Financial Policies and the World Capital Market: The Problem of Latin American Countries. Chicago, Ill.: University of Chicago Press for National Bureau of Economic Research. Buiter, V. W. 1986. Macroeconomic Responses by Developing Countries to Changes in External Conditions. National Bureau of Economic Research Working Paper 1836. Cambridge, Mass. Caballero, R. 1986. "Testing Liquidity Constraints: International Evidence." Massachu- setts Institute of Technology, Cambridge, Mass. and V. Corbo. 1985. Explaining the Trade Balances: A General Equilibrium Approach. Discussion Paper DRD141. World Bank, Development Research Depart- ment, Washington, D.C. Calvo, G. 1986a. "Fractured Liberalism: Argentina under Martinez de Hoz." Economic Development and Cultural Change 34, no. 3: 511-34. . 1986b. "Incredible Reforms." Columbia University, Department of Economics. Cavallo, D., and A. Pefia. 1983. "Deficit Fiscal, Endeudamieuto del Gobierno y tasa de Inflacion en Argentina: 1940-82." Estudios 6, Instituto de Estudios Economicos Sobre la Realidad Argentina y Latinoamericana. Centro de Investigaciones Economicas (CINVE). 1983. "Estudio Sobre el Impacto de la Reforma Arancelaria Sobre el Sector Industrial Uruguayo." Montevideo. Chenery, H., and M. Syrquin. 1975. Patterns of Development 1950-1970. London: Oxford University Press. Condon, T., and others. 1985. "Productivity Growth, External Shocks and Capital Vittorio Corbo and Jaime de Melo 139 Inflows in Chile during 1977-81: A General Equilibrium Analysis." Journal of Policy Modelling 7, no. 3: 379-406. Connolly, M. 1986. "Speculative Attack on the Pesa and the Real Exchange Rate: Argentina 1979-81." Journal of International Money and Finance 5, no. 1 (suppl.): 117-30. Corbo, V. 1983. "Desarrollos Macroecon6micos Recientes en la Economia Chilena." Cuadernos de Economia 20, no. 59: 5-23. -____ . 1985a. "Reforms and Macroeconomic Adjustments in Chile during 1974-84." World Development 13, no. 8: 893-916. -____ . 1985b. "International Prices, Wages and Inflation in an Open Economy: A Chilean Model." Review of Economics and Statistics 67, no. 4: 564-73. Corbo, V., and P. Meller. 1981. "Alternative Trade Strategies and Employment Implica- tions: Chile." In A. 0. Krueger and others, eds. Trad'e and Employment in Developing Countries. Chicago, Ill.: University of Chicago Press. Corbo, V., and J. de Melo. 1985a. "Liberalization with Stabilization in the Southern Cone of Latin America: Overview and Summary." World Development 13, no. 8: 5-15. ______ eds. 1985b. Scrambling for Survival: How Firmns Adjusted to Recent Reforms in Argentina, Chile, and Uruguay. World Bank Staff Working Paper 764. Washington, D.C. 1987. "External Shocks and Policy Reforms in the Southern Cone: A Reassess- ment." In G. Calvo and others, eds. Debt, Stabiliz2ation and Development. London: Basil Blackwell. Forthcoming. Corbo, V., and others. 1986. "What Went Wrong with the Recent Reforms in the Southern Cone." Economic Development and Cultural Change 34, no. 3: 607-40. Corden, W. M. 1974. Trade Policy and Economic Welfare. Cambridge, Mass.: Claren- don Press. Cumby, R., and S. van Wiinbergen. 1984. "Fiscal Policy and Speculative Runs on the Central Bank under a Crawling Peg Regime: Atgentina 1974-81." World Bank, Development Research Department, Washington, D.C. Davis, J. M. 1983. "The Economic Effects of Windfall Gains in Export Earnings, 1975-78." World Development 11, no. 2: 119-39. de Melo, J. 1987. "Financial Reforms, Stabilization and Growth Under High Capital Mobility: Uruguay 1973-83." In M. Connollay ancl C. Vega, eds. Economic Reform and Stabilization in Latin America. New York: Praeger. ______ and J. Tybout. 1986. "The Effects of Financial Liberalization on Savings and Investment in Uruguay." Economic Development and Cultural Change 34, no. 3: 561-88. Diaz-Alejandro, C. 1981. "Southern Cone Stabilization Plans." In W. R. Cline and S. Weintraub, eds. Economic Stabilization in Developing Countries. Washington, D.C.: Brookings Institution. . 1985. "Good-bye Financial Repression, Hello Financial Crash." Journal of Development Economics 19, no. 1-2: 1-24. Dornbusch, R. 1982. "Stabilization Policies in Developing Countries: What Have We Learned?" World Development 10, no. 9: 701-30. -_____ 1985. "External Debt, Budget Deficits, and Disequilibrium Exchange Rates." In G. Smith and J. T. Cuddington, eds. International Debt and the Developing Countries. Washington, D.C.: World Bank. Edwards, S. 1985. "The Order of Liberalization of the External Sector: An Analysis Based on the Southern Cone Experience." University of California, Los Angeles, Department of Economics. 140 Research Observer 2, no. 2 (July 1987) . 1986. "Monetarism in Chile 1973-1983: Analytical Issues and Economic Puzzles." Economic Development and Cultural Change 34, no. 3: 535-60. Fernandez, R. 1985. "The Expectations Management Approach to Stabilization in Argentina during 1976-82." World Developnment 13, no. 8: 871-92. Fischer, S. 1984. Real Balances, the Exchange Rate, and Indexation: Real Variables in Disinflation. National Bureau of Economic Research Working Paper 1497. Cam- bridge, Mass. . 1986a. "Issues in Medium-Term Macroeconomic Adjustment." World Bank Research Observer 1, no. 2: 163-82. . 1986b. Indexing, Inflation, and Economic Policy. Cambridge, Mass.: MIT Press. Frenkel, J. 1982. "The Order of Economic Liberalization: Discussion." In K. Brunner and A. H. Meltzer, eds. Economic Policy, in a World of Change. Amsterdam: North Holland. - 1983. "Panel Discussion on the Southern Cone." IMF Staff Papers 30, no. 1: 164-84. Galvez, J. and J. Tybout. 1985. "Microeconomics Adjustments in Chile During 1976-81: The Importance of Being a Group." World Development 13, no. 8: 969-94. Hanson, J., and J. de Melo. 1985. "External Shocks, Financial Reforms, and Stabiliza- tion Attempts in Uruguay during 1974-83." World Development 13, no. 8: 917-40. Harberger, A. 1982. "The Chilean Economy in the 1970s: Crisis, Stabilization, Liberal- ization, Reform." In K. Brunner and A. H. Meltzer, eds. Economic Policy in a World of Change. Amsterdam: North Holland. -____ 1983. "Welfare Consequences of Capital Inflows." Wforld Bank, Country Policy Department, Washington. D.C. and D. Wisecarver. 1977. "Private and Social Rates of Return to Capital in Uruguay." Economic Development and Culttural Change 25: 411-55. Hayashi, F. 1982. "The Permanent Income Hypothesis: Estimation and Testing by Instrumental Variables." Journal of Political Economy 90, no. 5: 895-916. Killick, T., ed. 1984. "The IMF and Stabilizationz: Developing Countrv Experiences. London: Heinemann. Kindleberger, C. 1978. Manias Panics and Crashes. New York: Basic Books. Krueger, A. 0. 1978. Foreign Trade Regimes and Economic Development: Liberalization Attempts and Consequences. Cambridge, Mass.: Ballinger. . 1981. "Interactions between Inflation and Trade Objectives in Stabilization Programs." In W. R. Cline and S. Weintraub, eds. Economic Stabilization in Developing Cou7ntries. Washington, D.C.: Brookings Institution. . 1984. "Problems of Liberalization." In A. Harberger, ed. World Economic Growth. San Francisco, Calif.: ics Press. . 1985. "How to Liberalize a Small Open Economy." In M. Connolly and J. McDermott, eds. The Econormics of the Caribbean Basin. New York: Praeger. Lin, Ching-Yuan. 1985. "Latin America and East Asia: A Comparative Development Perspective." International Monetary Fund, Washington, D.C. Little, I., and others. 1970. Industry and Trade in Some Developing Countries. London: Oxford University Press. McKinnon, R. 1982. "The Order of Economic Liberalization: Lessons from Chile and Argentina." In K. Brunner and A. H. Meltzer, eds. Economic Policy in a World of Change. Amsterdam: North Holland. Mezzera, J., and J. de Melo. 1985. "Adjustments by Industrial Firms in Uruguay during 1974-82." In V. Corbo and J. de Melo, eds. Scrambling for Survival: How Firms Vittorio Corbo and Jaime de Melo 141 Adjusted to Recent Reforms in Argentina, Chile, and Uruguay. World Bank Staff Working Paper 764. Washington, D.C. Michaely, M., and others. 1986. "Some Consequences of Trade Liberalization in Developing Countries." World Bank, Country Policy Department, Washington, D.C. Mierau, B. 1986. "The Impact of Trade Policies on Productivity Performance in the Chilean Manufacturing Sector." Ph.D. dissertation. Georgetown University, Wash- ington, D.C. Mussa, M. 1986. "Macroeconomic Policy and Trade Liberalization: Some Guidelines." World Bank Research Observer 2, no. 1: 61-78. Nogues, J. 1986. The Nature of Argentina's Policy Reforms during 1976-81. World Bank Staff Working Paper 765. Washington, D.C. Papageorgiou, D., and others. 1986. "The Phasing of a Trade Liberalization Policy: Preliminary Evidence." CPD Discussion Paper 1986-42. World Bank, Country Policy Department, Washington, D.C. Petrei, H., and J. de Melo. 1985. "Adjustments by Industrial Firms in Argentina During 1976-81." In V. Corbo and J. de Melo, eds. Scrambling for Survival: How Firms Adjusted to Recent Reforms in Argentina, Chile, and Uruguay. World Bank Staff Working Paper 764. Washington, D.C. Ramos, J. 1984. Estabilizaci6n y Liberalizaci6n Econ6mica en el Cono Sur. Estudios e Informes de la Comisi6n Econ6mica para America Latina (CEPAL) 38. Santiago, Chile. Rodriguez, C. A. 1982. "The Argentina Stabilization Plan of December 20th." World Development 10, no. 9: 226-38. -_____ 1983. "Politicas de Establizaci6n en la Economia Argentina." Cuadernos de Economia 20, no. 59: 21-42. Sachs, J. 1986. "Conditionality and the Debt Crisis: Some Thoughts for the World Bank." World Bank, Development Research Department, Washington, D.C. -_____ 1985. "External Debt and Macroeconomic P'erformance in Latin America and East Asia." Brookings Papers on Economic Activity 2, no. 2: 523-64. Sjaastad, L. 1983. "Failure of Economic Liberalism in the Cone of Latin America." World Economy 6, no. 1: 5-6. Stiglitz, J., and A. Weiss. 1981. "Credit Rationing with Imperfect Information." American Economic Review 71, no. 3: 393-410. Tybout, J. 1987. "A Firm-Level Chronicle of Financial Crises in the Southern Cone." Journal of Development Economics (forthcoming). Yeager, L. B. 1981. Experiences with Stopping Inflation. Washington, D.C.: American Enterprise Institute. Zahler, R. 1985. "Las Tasas de Interes en Chile: 1975-1982." Economic Commission for Latin America, Santiago, Chile. 142 Research Observer 2, no. 2 (July 1987) LAND RIGHTS SYSTEMS AND AGRICULTURAL DEVELOPMENT IN SUB-SAHARAN AFRICA Gershon Feder Raymond Noronha I n poor, but uncrowded rural societies, land rights are typically defined for groups rather than individuals. Within the groups, individual or family rights rest on elaborate traditions and cus- toms. Such customs enforce group control over the use and disposi- tion of land. Furthermore, to minimize social friction and ensure the group's survival, the entitlement of individuals to specific tracts of land is transitory. As a result, some efficiency is lost, since people lack incentives to improve the land. But these losses are small as long as land is abundant and farming methods primitive. The evolution of permanent and enforceable land rights is closely related to increases in population density, advances in farming tech- nology, and the emergence of agricultural markets. As land becomes scarce, societies can no longer rely on long fallow periods to maintain land fertility. They must adopt fertility-restoring technologies, which require investment of capital and effort-and thus also require incen- tives for farmers to change their practices. One such incentive is the right to cultivate land continuously and to bequeath or sell it. One nearly universal development is a unified system of land documenta- tion and registration, giving a land owner proof of ownership. If the registration system is effective and if the state can protect the owner from encroachment or unsubstantiated challenges to his land, then the system will indeed enhance security of ownership. Land rights link up with another feature of agricultural develop- ment, the emergence of rural credit markets. Credit transactions often D 1987 The International Bank for Reconstruction and Development/The World Bank 143 require some form of collateral. Land is an attractive collateral, pro- vided that the owner-borrower can assure the lender that the land can be transferred. Again, such an assurance is greatly enhanced by an effective system of land registration. The importance of land rights to agricultural development is the starting point of this article. It then describes the evolution of land right systems in Sub-Saharan Africa, reviews the evidence from Africa on the implications of existing systems, and discusses the policy op- tions. Ownership The main (and obvious) effect of a lack of secure ownership is the Security: Theory uncertainty in a farmer's mind about the value of improvements made and Evidence to the land. This uncertainty tends to increase as farming becomes more commercialized. There is ample evidence that the incidence of land disputes and land grabbing by larger or more powerful farmers increases as the potential return to land rises (Baron 1978, p. 27; Clark 1969; Feeny 1982, p. 95; Kemp 1981, p. 15; Tanabe 1978; Tomosugi 1980). Uncertainty regarding ownership will also tend to affect the sale and rent of land, which would otherwise allow land to be owned or used by those who are likely to put it to best use. Many studies have also highlighted the role of secure legal owner- ship in providing farmers with access to cheaper, longer-term, and more extensive credit. A land title is often a prerequisite for commer- cial or official bank loans (Dorner and Saliba 1981, p. 23; Sacay 1973; U Tun Wai 1957). As noted by Binswanger and Rosenzweig (1986), land has several attributes that are desirable as collateral. Farmers without secure title have to rely more on informal lenders, who usually charge much higher interest rates than those in the formal market. In some areas of India, for example, lenders charged 8-16 percent on secured loans, compared with 18-37.5 percent on unse- cured loans (Panandikar 1956, p. 75). (Hlowever, as explained by Stiglitz and Weiss [1981], interest rates cannot be allowed to rise to equate supply and demand, and credit rationing is optimal.) Titles may also increase the supply of all types of credit. In a study of land transactions in Thailand, Stifel (1976) observes that the title certificate is often deposited as security for loans from noninstitution- al lenders, a practice that is not registered or recognized by law, yet is common for small, short-term loans. The creditors have no legal rights to the land, but by holding the documents they prevent the owner from selling the land. They also restrict the owner's ability to borrow elsewhere and thereby incur excessive debt. In one village studied by Stifel, the number of these unregistered mortgages was three times that of registered mortgages. A study in Costa Rica by Seligson (1982) showed that, before the 144 Research Observer 2, no. 2 (July 1987) initiation of a titling program, 18 percent of the farms sampled ob- tained credit; after the program, 31.7 percent did so. Credit seems to have improved mainly for larger farms since the average farm size was 19 hectares for those title holders who received credit and 7.3 hectares for those who did not. (These results obviously reflected not only supply changes, but also shifts in demand.) Similarly, recent data from Jamaica (Inter-American Development Bank 1986) indicate that almost half of the recipients of titles under a new program went on to increase their borrowing. And a World Bank-sponsored study in Thailand (Feder and others 1986) found that more than 90 percent of the medium- and Figure 1 long-term loans were received by titled farmers-who formed only half of the sample. Legal Secure ownershi owners, who were able to pro- rl vide land collateral, obtained be- tween 52 and 520 percent more More security More security credit than farmers without col- t lateral. The interest rate on insti- * *- * . , . M~~~ore demand More cheap tutional credit was a third of the for investment long-term credit average rate charged on noninsti- tutional credit. The effects of constrained and investment dearer credit are significant. Fac- (input complementarity) tor ratios and input levels tend More demand for More cheap to be suboptimal and invest- variable input hort-te ment to be lower than it would otherwise have been (David and Meyer 1980; Rosegrant and input use Herdt 1981). Suboptimal inputs lead to lower output per unit of Higher output land, and suboptimal factor ra- peracr tios lead to lower output per unit of production cost. land valu These effects-less invest- ments and land improvements, lower use of various inputs, and lower productivity per unit of land for farmers without secure titles-are summarized in figure 1, which is adapted from Feder (1987). It shows that, even when the availabil- ity of credit is not a binding constraint, insecure ownership can reduce productivity because farmers have less incentive to invest. Even when eviction and land disputes are rare, the lack of secure formal ownership could reduce productivity significantly if credit is an effective constraint on farmers' activities. As land value is related Gershon Feder and Raymond Noronha 145 to its productivity, it follows that titled land is more valuable than untitled land. Apart from these implications for productivity, the ownership sys- tems also raise questions of equity. In some countries the procedures required to prove legal ownership are extremely complex and involve significant fees for lawyers, surveyors, and government departments. Since these costs vary little according to the size of the farm, larger landowners are better placed to afford them. By the same token, wealthier farmers usually have better access to information about land law, administrative procedures, and farm prices. They may therefore buy out poorer and less knowledgeable smallholders. One study of a land-titling system in Costa Rica found a high positive correlation (0.68) between farm size and the degree of ownership security in one of two provinces (Salas and others 1970). To safeguard the rights of the poorer farmers, governments could simplify procedures, dissemi- nate information on land registration, and take the initiative them- selves with a titling program. Land Tenure in The systems of land tenure in Sub-Saharan Africa today can be Sub-Saharan understood only in their historical perspective. Africa The Precolonial Era Much of what is known about tenure in this period is based on indirect evidence. Some general principles can be deduced from the literature (see Biebuyck 1963, Elias 1956, Glazier 1985, Hecht 1982, Lewis 1979, Maini 1967, Meek 1949, Synder 1981, and Thomson 1976). • The person who cleared land first was, in the absence of any more powerful claim, entitled to use it. The literature often refers to this person as the maftre du feu (the master of fire, a reference to a common method of land clearing). Anybody who later tried to establish rights of use within the area already cleared and controlled had to seek permission from the maitre (or his descendants, since rights were usually inherited by his progeny). Where there were no distinct lineages, permission had to be obtained from the chief (usually the village head). In such cases, the right to use land continued only so long as the farmer continued to live in the village head's jurisdiction and recognize his authority. * When land was abundant, access to it was not difficult. It was obtained either by residence or by acquiring "membership" in a group-which could be done by tracing real or fictional genealo- gies. The admission of outsiders, even slaves, was common (Barnes 1954, Gluckman 1941, Van Velsen 1964). Thus ethnic identities, so 146 Research Observer 2, no. 2 (July 1987) important during the colonial age, were less clearly defined and far more flexible. Once the right to use land was admitted, it could be passed on as a legacy. * The crucial element for the continued control and use of land was to have enough people, be they relatives or slaves, to work the land. * Given the primitive technology (mainly the hoe), the abundance of land, and the practice of shifting cultivation, land had little or no economic value. It thus made no sense to develop a system of rights to a particular parcel of land that could not be protected and had no utility until its fertility was restored. Accordingly, land was under group control, and individuals used particlar bits of it. These areas of control expanded and contracted with the rise and fall of leaders. * Under such systems of land use, one person could cultivate crops, while, on the same land, another could have rights to trees; or land could be used by cultivators during the cropping season and by herders in the off-season or during fallow periods. The system was somewhat different under Islamic law (which, through conquest and influence, spread over Mali, Mauritania, Niger, and northern Nigeria in West Africa and to Somalia, Sudan, and Zanzibar in the East). As with other indigenous systems, land be- longed to the "person who vivified it" (Anderson, 1954, Middle- ton 1961): the act of cultivation, or boring and enclosing underly- ing streams, gave the person doing so a right of ownership. But Islamic law differed from other indigenous rules in two respects: First, once land had been appropriated, nonuse did not mean a loss of ownership; that could happen only through conquest or sale. Second, Islamic law provided for defined rules of inheritance for both men and women, either as sharers or as residuaries (that is, after the sharers had received their specific shares of the property). In other indigenous systems, only patrilineal or matrilineal heirs could inherit rights. (In practice, though, under the Islamic system female entitlements to shares in land were usually bought by other heirs or residuaries.) The Colonial Era During the twenty-five years between 1885 and 1910, the African claims of nearly every major European nation-Britain, France, Ger- many, Portugal, and Belgium-were finally settled. The European powers were not troubled by considerations of ethnic homogeneity among the colonized; and as Hailey remarked, "the extent of the appropriation of indigenous lands has depended more on factors of climate or soil than on juridical arguments" (1957, p. 686). The colonial attitude toward Africans was influenced by theories of Gershon Feder and Raymond Noronha 147 evolution. "Just as races were supposed to have evolved from a primitive 'state of nature,' so too was land tenure expected to evolve from the simple form of tribal ownershilp to individual ownership" (Sorrenson 1967, p. 27). In British colonies, the colonialists regarded Africans as being on a lower evolutionary rung-and thus at a stage where land ownership would vest in chiefs, not individuals, and all rights to land would flow from membership of the chief's ethnic group. The French initially took the opposite view. They believed that only individual rights to land existed. The respective philosophies of the colonizers initially influenced their systems of government. The British chose indirect rule through traditional leaders, who were free to make rules within their "spheres of competence," including land rights. *The French policy was to assimilate: it drew no distinction between a dependency and metro- politan France. "Native chiefs" derived their powers from, and were subject to, the metropolitan government. Belgium merely stated that the Congo Free State was under the sovereignty of Belgium, but with a separate legal personality and laws. For Portugal, the practice of Christianity became the test of its policy of assimilation. These origi- nal distinctions gradually blurred as the practicalities of government affected the approach of all the colonial powers. In defining the concept of land ownership, the colonial powers drew a distinction between occupied lands (which were therefore owned) and unoccupied lands (vacant and "without a master"). The unoccupied lands were deemed open to settlement, as happened in Kenya, Zambia, and Zimbabwe; or to lease by foreign concessionar- ies, as, for instance, in the C6te d'Ivoire and Congo; or for use for other public purposes. These distinctions were much influenced by European conceptions of title and property. The colonialists were completely ignorant of the systems of shifting cultivation and of transhumance that were prevalent in Sub-Saharan Africa. As a result, the land available to Africans shrunk, despite a few protests (particularly by French scholars). The consequences were disastrous for the African population, particularly in the British settler colonies where logic was strained to provide land and benefits for the settlers. Even more influential was the colonial belief that all occupied land was held in "communal tenure." This meant that individuals had only user rights to land, without the power to sell or mortgage it. "Owner- ship" was vested in chiefs as trustees for existing and future genera- tions. Yet the British were not entirely consistent. In Uganda, for instance, Sir Harry Johnston (who was later to settle claims in Mala- wi, then Nyasaland) decided that the Ganda chiefs had absolute title to land. He thereby created a new system of tenure: mailo. The French, far more consistent, recognized individual title only if the 148 Research Observer 2, no. 2 (July 1987) applicant went through the complex procedures of titling and registra- tion; few did so. The Belgians tried for decades to discover what lands were used by, and needed for, subsistence cultivation by the Africans. The logical consequence of insisting on communal tenure and the trusteeship of chiefs was to raise the status of chiefs. Where there were none, chiefs were appointed. In some cases, as among the Ashanti, failing rulers were propped up, and the development of individual rights to property was stifled. And chiefs were quick to exploit their position to establish or strengthen their control over land. Indigenous farming was further affected by the creation of districts (British) or cercles (French and Belgian). Initially, the districts and cercles were based on ethnic identity, real or assumed. This method had two consequences. First, it necessitated and strengthened ethnic differences. Second, it restricted movement outside the district by the introduction of the "pass" system (a badge of both identification and confinement) and kept trade to within each district. In precolonial days, it had been relatively easy to become a member of a group, and the absence of well-defined boundaries permitted both shifting cultiva- tion and migration. All that was changed: land for cultivation now had to be found within the district or cercle. What the colonial powers did was to pacify a continent. No longer was war a means of acquiring land and labor. Colonialism also brought with it some improvements to sanitation and public health, which helped to increase the longevity of the African population. When, in the last decades of colonial rule, the conquest of malaria and control of the tsetse fly opened new areas for settlement and cultivation, faster population growth reduced the amount of cultiv- able land per capita. It became correspondingly more important to acquire land rather than labor. The development of a land market was also restricted by prohibit- ing Africans from selling land to non-Africans. This prohibition pre- vailed even in Uganda, where individual title to land was recognized. And land transactions had to be undertaken by chiefs, not common- ers. Only in the twilight of colonialism was a feverish attempt made to introduce individual land titling, and then only in some colonies. By then, it was too late. Interestingly, the introduction of individual titles in Kenya was justified by the need to promote economic development. The colonial- ists' earlier fears-for example, that in India individual titles had allegedly led owners to mortgage and sell their lands to moneylenders, resulting in widespread indebtedness and landlessness-no longer seemed to matter. The Swynnerton Plan for Kenya expressly recog- nized that some landlessness could be a consequence of introducing individual titles, but that this increase in landlessness was a necessary Gershon Feder and Raymond Noronha 149 price to pay for development and that the more progressive farmers would survive (Glazier 1985). This theme has been reaffirmed in sub- sequent five-year plans in Kenya. Independence and After After African nations attained independence, their approaches to land tenure varied. In Malawi, for example, the thrust toward individ- ual tenure has been tempered by registering only family tenures in the Lilongwe Agricultural Development Division. In Swaziland, the desire to return to customary tenure has not been applied to land that was privately owned before independence. In. Sierra Leone, the idea that individual tenure was conducive to economic development has been accepted, but not widely implemented. In Tanzania, the government then began experimenting with long-term leases to individual farmers (for thirty years, which could be extended to ninety-nine years). Only two countries have consistently maintained a single view: C6te d'Ivoire and Kenya. In both, the prevailing philosophy has been that individual tenure is essential to foster economic growth. Nonetheless, Kenya restricts the rights of an owner to sell, lease, or mortgage land by making the consent of a land board a precondition to such transac- tions. African approaches to land tenure can be divided into three main types: a. Countries that allow the acquisition of individual title: Cote d'Ivoire (without any restrictions oin the power of the title-hold- er); Kenya and Malawi (with restrictions on the title-holder). b. Countries that recognize different types of tenure: Senegal and Sudan (both individual title and nationalization of nontitled lands); Bostwana, Ghana, Lesotho, Liberia, Mali, Sierra Leone, Swaziland, Uganda, and Zimbabwe (individual title, indigenous systems and public lands); and Cameroon (individual, group, indigenous systems and public lands). c. Countries that vest title to land in the state, so that individuals have rights only of use and occupancy: Ethiopia, Mauritania, Nigeria, Tanzania, Zaire, and Zambia. This classification needs to be qualified. First, there is an overlap between the countries in (b) and (c): where indigenous systems are recognized in group (b), this usually means that the individuals or groups covered by those systems have rights only of occupancy and use. In that sense, they share the approach of countries in (c): exam- ples are Botswana and Zimbabwe. (A simnilar practice applies in Sene- gal and Sudan where, with the nationalization of untitled land, the government recognizes only user rights of occupants.) Second, in Ma- lawi, indigenous systems are recognized for untitled land. Third, in 150 Research Observer 2, no. 2 (July 1987) Ghana, rights of sale of land under indigenous systems of tenure are vested in chiefs-the approach first introduced by the British. Fourth, in all the examples it is assumed that the state holds paramount title to land. The classification above is not, of course, immutable. For example, both Mauritania and Sudan have passed legislation recog- nizing Islamic law as the governing framework of the nation. This is a contradiction in terms, since Islamic law recognizes individual title to land, the power to bequeath and sell land, and the power to make gifts of property, while Sudan still maintains the ideology of state ownership of land. In Mauritania, where the state grew out of con- quest, land title is arguably vested in the state. Land as a Commodity The most important factors contributing to the development of a land market are the growth of population, the use of new technology and inputs, the development of markets for products, the growth of communications, and the rise of alternative uses for land. These influ- ences are not easily separable. However, as Cohen (1980) points out, there has been a tendency to concentrate on the analytically clearest factor, population growth. In the African context, three studies are worth noting. Lunning (1965) studied land transactions in Nigeria: three villages in Katsina province and seven villages in Sokoto between 1960 and 1962. He classified the seven methods by which land could be acquired: inheri- tance, gitft, purchase, pledge, loan, lease, and share cropping. He defined purchase as "an outright transfer of land for cash"; pledge as "the right to use the land [that] is passed to another party in exchange for a money loan"; and loans as land entrusted to a friend for a year when the owner had more than he needed (to be distinguished from a lease, where there was a defined money payment). Lunning found that there was "to a certain extent, a relationship between density of population and the occurrence of sales but it does not explain all differences... In some ways, acreage of farm land available per head of population should be a more reliable 'yard-stick"' (1965, p. 79). But, although this high ratio of population to cultivable land could explain land sales and leases (amounting to 31 percent of all transac- tions) in one village, it was not compatible with the prevalence of sales and leases in other villages where twice as much land was available per person. In these villages a more plausible explanation was the development of the road system and the greater accessibility to markets, which appeared "to be related to a greater occurrence in sale transactions of land. Villages off the beaten track have a far greater incidence of such customary transactions as inheritance and loan... Nearness to markets and subsequently easier accessibility to Gershon Feder and Raymond Noronha 151 capital may be reasons for a greater incidence of pledging" (1965). Lunning concluded that increasing demand for land had a decisive impact on land tenure patterns. The four factors he identified as being the most important were population growth, the introduction of cash crops, infrastructural development, and distance to major markets. But, he added, "none of these factors can be singled out as having had a singular influence on the demand; a number of them are usually involved" (1965, p. 178). Roden (1971) studied farming among the Nuba of Sudan. Before 1940, they had cultivated both highlands and lowlands. In the high- lands, the farming had been intensive, with terracing, manuring, rota- tion of crops, and planting of legumes. In the lowlands, where few people lived for fear of raids, shifting cultivation was the norm. On the plateau, the Nuba drew a distinction between homestead plots, which were near the terraced plots, and the main arable lands, which were some distance away from the homes. Where population density was greatest, there was a market in homestead plots and strict rules of inheritance: only the eldest son could inherit the homestead plot. Although in theory the state owned the land, sales of homestead plots were common, particularly to avoid rules that would otherwise result in inheritance by the mother's brother's son. By 1940, however, the lowlands were no longer affected by raids. People started forming towns and growing cotton (which could not be grown on the plateau). In some highland villages, population pressure began to ease. Accordingly, the rules of inheritance were relaxed for the Nuba who had started farming on the plains, so that all sons could share equally in the property. The market in hill and homestead plots almost disappeared. A similar picture emerged from Netting's study (1965) of the Kofyar hill farmers on the Jos Plateau in Nigeria. The Kofyar farmed inten- sively on the plateau, with strict rules of inheritance and individual ownership of plots. On the hillsides, agriculture was shifting, and nobody had rights of access to a specific plot. When the Kofyar were studied again in the 1980s, they were faLrming the plains intensively and had developed a market in land: both changes had been assisted by the development of infrastructure and access to markets for their produce (Netting 1985). Other factors that encourage the development of a land market include proximity to means of communication, such as a new road (Cobb and others 1980, Haswell 1975, Lunning 1965) or the main railway line (Bruce and Dorner 1982). Proximity to urban centers is also important: the possibility of alternative uses increases the value of land and turns it into a tradable commodity. An estimated 80,000 to 100,000 hectares of land are annually converted from agricultural to urban use in Sub-Saharan Africa (Hamer 1986). Hill's (1963) study 152 Research Observer 2, no. 2 (July 1987) of cocoa farmers in Ghana seemed to indicate that the cultivation of a cash crop is related to the development of a land market. Finally, restrictions on mobility, particlarly during the colonial re- gimes in some cases, resulted in the development of a land market much earlier than if people had been allowed to migrate to other, less populated, areas. An outstanding example was the development of a land market among the Kikuyu in Kenya (who lost most of their lands to settlers) long before such a market had developed, say, among their neighbors, the Mbeere. Household Appropriation of Land With the development of a land market, is there also a correspond- ing increase in individual tenure? The evidence on this is mixed. Studies by Berry (1975) and Okali (1983) suggest that the kin group continued to play an important role in the determination of owner- ship. However, others (Adegboye 1974, Food and Agriculture Organi- zation 1984, Haswell 1975, and Levin 1976) suggest that, as the de- mand for cash crops grows, so more and more households appro- priate land in defiance of indigenous rules. Land rights may be viewed as a bundle of distinct privileges (right to use, right to plant trees, right to rent, right to sell, and so forth). Over time, more of these rights are transferred from large social groups to smaller groups and eventually to households. The pattern of this transfer appears to rest on a principle that is well known in Hindu law: where a farmer has acquired land with the help or common resources of his kin group, he has less individual control than if he has acquired the land from his own earnings and employs labor (paid in cash or kind). In all these examples, however, sales of land increase and so does the power of households (Haswell 1975, Hecht 1985). The first rule of indigenous tenure is that a person is entitled to undisturbed possession of some allotted land as long as it is being used. The period of use varies according to the type of crops grown, so farmers often lengthen the period of use by planting trees. Thus, in Cameroon, for instance, the main purpose of planting cocoa and coffee is to retain undisturbed possession of the land so long as the trees survive (Levin 1976). Tree planting in C6te d'Ivoire and Zanzibar serves a similar purpose (Koby 1979, Middleton 1961). This gradual process of appropriation begins with the best land-the valley bottoms (as in Tanzania and Zambia), the oualo, fadama, and swamplands (in the Gambia, Nigeria, and Senegal). Because they retain some moisture, these lands allow cash crops to be grown in the off-season. Over time, the rights of "outsiders"-the navettane in the Gambia and Senegal-are increasingly restricted. At first, they are allowed Gershon Feder and Raymond Noronha 153 access to marginal land. But as land values rise, other rights of out- siders are curtailed. They can no longer become members of the group, so cannot gain access to land (Haswell 1963, Hecht 1982). Restrictions are placed on the types of crops that tenants (or "stranger farmers") can cultivate. For instance, in Ghana and Cameroon, ten- ants are not allowed to plant cash crops (Adegboye 1974, Dravi 1984); in Zanzibar, they are not allowed to plant trees, particularly the valuable kola tree (Middleton 1961). The basis for the restrictions is to deny tenants the opportunity of claiming title to land by virtue of their length of possession. Implications As colonial views of land law have been carried through into today's independent countries, the land market in Africa remains distorted. In some areas only rights of use and occupancy of land are recognized by the state. In others, some Df the rights of ownership- to transfer, mortgage, or lease land-have been fettered. This section discusses the implications of such distortions. The Divorce of Law from Reality When the legal system decrees that land cannot be sold or can be transferred only with bureaucratic (and frequently arbitrary) approval, law gets divorced from reality. Land continues to be sold or pledged, but in an informal market. The only result is that these sales or pledges are unenforceable in a court, so prices contain risk premiums that cause a deviation between the social value of land and its market value. Land sales may be disguised as the sale of trees or houses, as in Malawi (Ibik 1971); or as a pledge, with the pledgee paying an amount equivalent to the purchase price of the land so as to avoid getting the permission of the village headman, as in Nigeria (Lunning 1965). During the colonial period the clearest examples of these market distortions came from West Africa, where the production of commer- cial crops (initially, oil palm) had led to the development of a land market even before colonialism began. In fact, in the Gold Coast (the coastal area of modern Ghana), land sales had started at the turn of the century and had been given judicial recognition (Grier 1981). In Nigeria, land sales had begun in the southern provinces even before the colony of Lagos was ceded in 1861 and were common in northern Nigeria in the first decade of the twentieth century (Rowling 1946, Watts 1983). However, the evidence of these transactions was dis- counted by the West African Land Committee in 1912 as being insuf- ficient and inconclusive. As a result, land sales took place only "in- formally" (illegally) except where chiefs claimed that sales were for the benefit of the clan. 154 Research Observer 2, no. 2 (July 1987) This inability to accept evidence of land sales, blurred by precon- ceptions of what "native" groups could actually do, afflicted other land commissions as well. In Kenya, the Carter Commission ignored evidence that land had been traded by Kikuyu and that, in the Kiam- bu region in particular, sales were common (Sorrenson 1967). In Tan- zania, the cultivation of marketable crops and land sales were com- mon among the Arusha, Sambaa, Hayya, and Chagga; the Sambaa did not even require the consent of kin to the sale (Hailey 1957, p. 782). Sales among the Sukuma in Tanzania began before the German occupation in 1891, but were stopped by the Germans (Malcolm 1953, p. 12). In Malawi, the Land Commission in 1921 did not accept evidence that in the Marimba and West Nyasa districts there were both land sales and individual titles to land. Independence has not put a stop to land transactions. In Tanzania, Pitblado (1981) reports that in one village in the North Mkata Plain, some 16 percent of land was acquired by purchase; in another, the figure was 36 percent. In Lesotho, where land cannot be legally sold (and where urban and rural lands have equal value in the eyes of the law), Mosaase notes that as a result of land scarcity "a clandestine land market had developed and the indiscriminate selling of arable land for residential and commercial sites has become uncontrollable" (1984, p. 90). In Mali, land is inalienable in theory. In practice, though, sales of less fertile lands to stranger farmers take place, even though it is difficult to obtain data on such sales. In Niger, sales of land are increasing, although indigenous rules say that land cannot be sold (University of Arizona 1979). Ega's (1979) survey of three villages in Zaria, Nigeria, showed that 18 percent of those surveyed had obtained their lands by purchase. He notes that "there is a significant prevalence of illegal commercial transactions in land and considerable mobility of land. In particular, purchase has become an important means of acquiring land" (1979, p. 291). Of the Volta region of Ghana, Nkunya says that "outright purchase... is becoming more and more common these days" (1974, p. 4). Even in areas where sales are recognized by law, cumbersome legal procedures mean that many transfers are not registered and the official record does not reflect reality. Access to Credit with Land as Collateral The widespread prohibition against mortgaging land does not stop land being used as collateral in informal transactions. The prohibition serves only to make the occupant more dependent on state largess or on the informal market where interest rates are much higher (Watts 1983). As Woodman (1967) demonstrates, lending practices increasing- ly take on the color of formal legal requirements (with witnesses to Gershon Feder and Raymond Noronha 155 transactions and documents); land mortgages are recognized in both Ghana and Nigeria. Nonetheless, the status of these transactions is blurred, which often results in litigation, especially when land prices rise. As Goody notes of southern Ghana "the different value that cocoa farming, large-scale mining and tirnber exporting had given to the land and its products created innumnerable disputes that found their way right up to the Privy Council" (1980, p. 144). Land pledging was aiso found by Pitblado (1981) in four of the villages he studied (in one of these the land pledged amounted to 40 percent of the farming acreage). The pledging of oil palms in N4igeria continues despite its illegality, increasingly maturing into the right of sale by the mortgagee (Abasiekong 1981). As in the case of land sales, data on pledges are difficult to obtain, particularly in areas where Islamic law (with its prohibition against usury) prevails (Scott 1984, Watts 1983). The Potential for Increased Inequality Increased inequality, the fear of which is often cited as a reason for prohibiting land sales and mortgaging, could in fact be an important consequence of the prohibition. Inequality also arises where govern- ments recognize sales only by people from a particlar group (chiefs, for instance) or where transactions in land involve complex proce- dures with uncertain results. During the colonial period, the insistence that chiefs were trustees of the land encouraged them to use their office to their own advantage (Fallers 1955, Glazier 1985, Goody 1980). With growing shortages of land, chiefs were not reclutant to revoke allocations of land, particularly when outsiders were involved (Hamnet 1975). Inequality is a consequence of prohibition for two other reasons. First, those who know the law (usually the wealthier and better-off) can use the system to their own advantage. Second, they are protected in land transactions because their status ensures that no action would be taken to dispossess them (see Bates 1981, pp. 53-61). And in Ghana, "many chiefs have benefited as a result of their control over land... to acquire bank notes, tractors... Where benefits did accrue to chiefs, they were not redistributed within the chiefs' communities, with the result that the chiefs have becomLe economically quite distinct from their subjects. At the same time, the institution of chiefs has been reinforced at a regional level by this new wealth of its office- holders" (Shepherd 1981, p. 177; see also Goody 1980). Similar conse- quences have occurred in Botswana (Lawry 1983, Peters 1983). In their studies of land adjudication in Kenya, in the Embu Dis- trict and among the Mbeere, Haugerud (1983), Brokensha and Riley (1980), Glazier (1985) and Njeru (1978) found that it was the influen- tial people (including the chiefl and the civil servants who used their 156 Research Observer 2, no. 2 (July 1987) knowledge of the law to acquire land, at the expense of the poorer and less knowledgeable. Glazier (1985) shows that the chiefs take advantage of laws requiring proofs of genealogy and residence to expand the numbers of their kin, so that more land would be granted to them. Burial sites are also scattered as evidence of prior occupation (Glazier 1984, West 1972). In Nigeria, occupancy certificates under the Land Use Decree of 1978 were granted according to occupation and income. Koehn concludes that "most applicants for statutory rights of occupancy are prominent businessmen and senior civil servants" (1983, p. 467). The advantages of knowledge (and literacy) tend to favor the urban dweller. In Senegal, for example, under the Law on National Domain of 1964 residents were allowed to establish title and request registra- tion within six months from the date of passage of the law. However, "rural people, including those of the river basin, were generally una- ware of this, and were not notified to present claims. Then all non- deeded lands became part of the National Domain" (USAID-RBDO 1982, p. 115). Women, too, have often been excluded from owning land: examples include the Tonga of Zambia (Colson 1963, Spring 1985), Nigeria (Spiro 1985), Kenya (Pala 1978), and Senegal (Gladwin and others 1987). And nomadic farmers have been affected-not only during the colonial era, when administrations regarded "unoccupied" lands as being lands without title (Baker 1975). After independence, the nationalization of unregistered lands has resulted in nomads' los- ing their traditional routes. In Mauritania, Niger, Senegal, and Sudan, they have only a licence (which can be withdrawn) to pass over transhumant routes, but no easement (which is both recognized in law and enforceable). Insecurity and Uncertainty Uncertainty is an obvious consequence of any transfer that is for- mally illegal, both because it is not clear that the seller has the right to transfer land, and because the buyer fears government action to cancel the transaction. In some areas, the insecurity arises "not so much in fear of interference (with possession) by members of (the cultivator's) own community but from the apprehension that the gov- ernment may, for its own purposes-such as the need of land for public use or for alientation to colonists-disturb him in the posses- sion of his holding" (Hailey 1957, p. 807). Thus, in Cote d'Ivoire, it is fear of expropriation by government that makes a farmer plant more coffee and cocoa, so widely spaced that the returns are "inefficient" and ten hectares produce what could have been obtained from three hectares of closely spaced trees. Another barrier to greater productivity is the application of indi- Gershon Feder and Raymond Noronha 157 genous rules on tenure. They provide for an individual household to be granted as much land as it needs for subsistence; this rule prevails, for instance, in Lesotho (Hamnet 1975). Families are naturally inhibit- ed in producing a surplus. If they do, they face additional pressure to redistribute the surplus-pressure that is "usually sufficiently great to inhibit both general economic development and rigid patterns of stra- tification" (Cheater 1984, p. xiv; see also 1Richards and others 1973). The rule that a person in possession of property has only a right of occupation, not genuine ownership, is not conducive to productivity. This is largely a matter of inference, for there are no detailed studies of the link. However, the final report of the Tanzania National Agri- cultural Policy Task Force notes that the present system of land ownership by village (with individual residents having a mere right of use) has the following constraints: a. reluctance to invest in the cultivated area for a long-term improvement of the land; b. unwillingness to expand crop acreage for fear of being identified as a person opposed to collective farming; c. the issuing of short-term leases discourages long-term investments leading to mining of the land; d. scaring off potential investors in agriculture due to uncertainty over ownership of land (1982, pp. 27-8). Another important consequence of the indigeous system of tenure is its impact on labor mobility and, implicitly, on productivity. Although there are studies of the impact of taxes imposed by colonial regimes on labor and migration (Hailey 1957, Palmer and Parsons 1977, Syn- der 1981, Ward 1976) and on the introduction of cash crops (Pelissier 1966), no study has examined the impact of indigenous systems of tenure on labor mobility (except, very indirectly, Collier 1983). When statutes affirm only rights of use and occupation, the indigenous rules would apply. Non-Islamic rules construe any absence from allocated land as abandonment. This implies, of course, that some time must elapse (for instance, five years in Lesotho) before rights to the land can be assumed to have been surrendered. The threat of such rules being applied is growing as land becomes scarcer. The result is that some members of each household must always stay on the land and cultivate it, however inefficiently. The others, if they can move at all, do so only to urban areas. Titles to Land and Security The term "security" is often misunderstood in the literature. When it refers to the ability to use land for a certain period and for a defined purpose without disturbance, security of possession is usually 158 Research Observer 2, no. 2 (July 1987) ensured under indigenous systems. It is clear that in most Sub-Saharan African societies, land under cultivation by an allottee cannot be taken away. Eckert (1980) notes that in Lesotho the average period of landholding is eighteen years which, adds Doggert, is "more than that prevailing in the United States" (1980, p. 20). The situation, however, is entirely different when security is defined as the ability of an occupant to undertake land transactions that would best suit his interests-for example, to offer land as collateral for a loan. Transactions in the informal market become problematic when there is a question of selling the property (Haugerud 1983) or borrowing from commercial lenders (Abasiekong 1981, Haswell 1975). As Goody says: In order to make loans for agricultural development, some "secu- rity" is required by both commercial and government banks. "Security" means a regular income (i.e., being an employee), a saleable (i.e., not a mud) house, or land (one's own). Everybody of course has land, although subject to the rights of others. But if one is going to purchase a tractor or even a plough, one needs access to an increased acreage and the problem is showing that one has rights to that land. These rights have to be of a certain kind if they are to act as security under the contemporary rules and identifiable by a court, bank or government department as being linked, in effect exclusively, to the particular individual. And they have, in principle, to be disposable by him. Such identification is achieved by the registration of land in the name of an individual, which proves definite proof of "ownership" and hence security for loan (1980, pp. 144-5). At times, even an officially granted occupation certificate can be valuable property. Seidman notes that possession of such a certificate in Nigeria confers private economic advantages: "major financial in- stitutions treat these certificates as a necessary collateral against var- ious types of loans, including bank mortgages, and commercial agri- cultural credit. Therefore, holders of statutory titles can gain access to domestic money markets and secure loans at favorable terms which can be utilized for private capital accumulation and investment" (1975, p. 642). Koehn adds that these statutory rights "provide a measure of legal, state-enforced security of tenure that is not afforded by customary rights, squatting or land purchases on the secondary market" (1983, p. 461). At times, however, title to agricultural land is regarded as inferior collateral compared with "urban properties, at- tachment of salaries and other forms of valuable property" (Okoth- Ogendo 1976, p. 175). This attitude could arise from the fact that in some areas it is difficult to enforce sales of mortgaged agricultural land when a mortgagor defaults. Gershon Feder and Raymond Noronha 159 Secure Ownership: Productivity and Efficiency While the evidence seems generally to favor the conclusion that titled land opens up the commercial money market, it is scanty on the question of whether titling in Africa leads to higher productivity. Studies from other parts of the world, however, suggest that that is indeed what happens. In Costa Rice Salas and others (1970) found positive correlations between the degree of ownership security and farm investment per unit of land. Data from three states in Brazil in 1978 show that capital per hectare is substantially higher on titled farms than on undocumented or encroached land (Villamizar 1984). The analysis was done for different farm sizes, and within most groups the link applied. Analysis of data from three provinces in Thailand (Feder and Onchan 1987) indicates that the probability of investing in land improvements is significantly affected by possession of secure ownership, holding constant other factors such as location, soil quality, and farmers' characteristics. Capital-land ratios were found to be markedly higher among titled farmers (again, controlling for various other attributes). Differences were more than 100 percent in some of the areas studied. Data from this study also indicated significantly greater inputs (labor, powet, cash) per unit of land by legal landowners, compared with squatters and untitled farmers. Several studies have focused on the impact of ownership security on output or income. The earliest study (Salas and others 1970) indicates a positive correlation of 0.53 between income per unit of land and ownership security on one province in Costa Rica. In another prov- ince, however, the correlation was negative, although quite weak (-0.07). A study of the state of Maranhao in Brazil (cited in Inter- American Development Bank 1986, pp. 186-89) concludes that the granting of full legal ownership to squatters and undocumented occu- piers would increase their income by 200 percent. The same report quotes recent data from Equador indicating that the incomes of titled farmers were double those of untitled farmers, holding the amount of land constant (p. 187). Data from Thailand show much lower, but nonetheless statistically significant, differences in output per unit of land, in the range of 12-26 percent (Feder 1987). The analysis for Thailand controls several factors that may vary between titled and untitled farmers (for example, location, soil quality, and initial wealth), which could explain the fact that the productivity differences are smaller than those in the studies from South America. Although it would have been more appropriate to consider the impact of secure ownership on output per unit of total factor use, the data deal with the more readily observable output per unit of land. Efficiency may also be damaged when farmers cannot use land as collateral for credit. For instance, in Costa Rica it has been reported 160 Research Observer 2, no. 2 (July 1987) that cattle are a surer collateral than land when the farmer does not possess a full formal title. Farmers without title therefore tended to shift from crops to cattle raising, even though their land may have been better suited to growing rice and beans (Dorner and Saliba 1981, p. 23). In general, credit constraints may cause farmers to shift to crops and activities that need less working capital and farm machi- nery. Uncertainty over continued access to land encourages squatters to prefer short-cycle crops. Recent survey results from Jamaica indicate that titled farmers had almost double the incidence of permanent and semipermanent crops than did untitled farmers. Indeed, a third of the recipients of titles under a government program reported that, follow- ing the change in their status, they planted more permanent and semipermanent crops (Inter-American Development Bank 1986, p. 189). The proposition that the prices of titled land are higher was con- firmed by Chalamwong and Feder (1986) in rural Thailand. Their results show that, in several areas, titled land of given quality and location was twice as expensive as squatters' land. However, the difference in land prices was smaller in an area where farmers had access to substantial amounts of credit from traders (who do not usually require a formal land collateral). There are few quantitative studies of the economic effect of secure ownership in Sub-Saharan Africa. Most are based on preconceived notions of the necessity for land titling. For example, Ike (1977) took data from western Nigeria and tested the hypothesis that a freehold system was inherently superior to communal land tenure. He conclud- ed that his hypothesis was proved by the higher average incomes of freeholders. But he provided no information on how he selected the sample, nor any data indicating that the study controlled for differen- tial access to inputs and extension, quality of land, and access to labor. Perhaps only two studies examine the consequences of granting individual title in a thorough and revealing way: Richards and others (1973) in Uganda and Cheater (1984) in Zimbabwe. Richards and others (1973) tried to establish which factors tended to induce a switch from subsistence cultivation to commercial agricul- ture. The area for study was Buganda in Uganda, where in 1900 the British had granted absolute title to land to the king, his kinsfold, and other senior officials (mailo tenure, as it came to be known). Richards and others point out that Buganda had several advantages not enjoyed by many other areas: a developed transport system, immigrant labor (especially in the first four decades of this century), and towns in the midst of the farmland providing a ready demand for food. Fifty-eight percent of the farmers in the sample had bought their land. The authors' conclude: Gershon Feder and Raymond Noronha 161 The mailo system itself... did not result in the commercial use of land by its owners for a period of some 40 to 50 years, though there were a few exceptions. Traditional attitudes to land ownership as a source of political power and prestige continued and during the first half of the century mailo land was important to its owners as providing rents from peasant tenants and also capital, through the sale and mortgage of portions of estates. The money so obtained was used to raise the standard of living of the owners and to provide for a European type of education for their children... Nevertheless the mailo system, by introducing the right of land purchase so early in the century as compared with the rest of East Africa, gave to the ambitious peasant the opportunities... and it certainly facilitated the purchase and development of the big coffee farms of the fifties which attracted Ganda traders, businessmen and others into com- mercial farming... A property market therefore grew up and a man with capital was able to buy land and start farming on his own account. This oepned the way for the emergence of a class of commercial farmers (pp. 295, 313). The authors emphasize the importance of capital. Those who did not hold mailo could raise capital through selling cotton, which was introduced in Buganda in the 1920s. It is interesting to note that the creation of a land market, permitting land mobility, allowed the commercially minded farmers (whom the authors characterize as "the men of profit) to buy land from the mailo holders who regarded land primarily as a source of status (they were 'men of affairs"). Cheater (1984) studied African Purchase farmers in Msengezi, west of the Zimbabwean capital of Harare. African Purchase farmers were those who, after the Carter Commission's recommendations in 1925, were allowed to buy land in areas designated as African Purchase areas. Although it was suggested that these farmers should ultimately obtain freehold title, there were initially many restrictions (for example, a prohi- bition against subletting and a restriction of rights of residence to family members and hired labor). Furthermore, all transfers "were to be subject to government approval, whether or not title had been granted" (p. 6). Some 69 percent of the sample farmers were monogamous. Although the farmers were not entirely free of social obligations to their kin-they let some kinsfolk live with them-it was clearly understood that this resi- dence was only temporary (except in the case of a mother or eldest son). In fact, the relatives were often evicted, unlike in the communal lands where such residents are permanent fixtures. Cheater says that "less than 10,000 Purchase Land farmers account for one-third of the total value of marketed produce from African growers, the remaining two-thirds coming from approximately 600,- 000... Communal Land households" (p. 11). These farmers, accord- 162 Research Observer 2, no. 2 (July 1987) ing to Cheater, now express themselves in a modern idiom of status and prestige: "durable goods such as farm machinery and bank ac- counts... In earlier African societies, there were few opportunities to retain wealth in such material forms and crop surpluses were, there- fore, converted into control over people, particularly over wives and children" (p. xiv). In 1974, 91.4 percent of the sample population possessed cattle; 86.3 percent, oxplows; 70.3 percent, cultivators; 61.7 percent, harrows; 41.4 percent, carts; and 12.1 percent, planters. It is difficult to accept Cheater's inference of higher productivity among Purchase farmers in the absence of comparative data for these farmers and those on Communal Lands of similar soil quality. But the study confirms the more consistent commercial approach of the Pur- chase farmers, which also characterized the farmers in the sample from Buganda. The evolution of land rights in Sub-Saharan Africa should not Conclusions necessarily be viewed as a natural process, because some of the changes were the results of government intervention (either colonial or postcolonial). Such intervention is not always conducive to efficien- cy or equity-whereas market forces have tended to circumvent any restrictions that cause inefficiency. The issue is whether those market forces achieve the same efficiency that could have been obtained under a different institutional set up. The evidence cited in this article dispels some of the popular mis- conceptions about land rights in Sub-Saharan Africa. In many areas there has always been individual possession; in others, it is growing. Even where communal ownership was imposed, cultivation and pos- session remained with individual households, and an increasing range of rights to land were appropriated by individual households. Land sales and mortgaging by individuals are common in many areas where such transactions are not legally recognized. The lesson from other parts of the world is that efficiency requires individual land rights to be recognized in a way that provides suf- ficient security (either in the form of long-term leases or land titles). That stage may not have been reached yet in parts of Sub-Saharan Africa. But in other parts (sometimes only a region within a country), the justification for a change in land arrangements already exists. In such cases, what is needed is a careful analysis of the benefits and costs of different systems (for example, title deeds, title registration, and long-term leases), including equity considerations. The gain in efficiency may or may not outweigh the costs of introducing a new system. Unfortunately, there is a shortage of rigorous quantitative research that would help to assess the costs and benefits of a policy change. Gershon Feder and Raymond Noronha 163 Such research is essential in Sub-Saharan Africa, where many policy- makers and donors have preconceived notions of the ideal system, based on ideological considerations or inadequate data. Obviously, quantitative research has its own limitations: it tends to highlight measurable variables, while disregarding administrative constraints. There is no point in introducing a system of title registration, for example, where the capacity continuously to update the registers does not exist. A proper analysis should include the costs of providing such capacity, rather than simply assuming its existence. Abstract Links between land rights and agricultural development provide a conceptual frame- work to analyze land rights systems in Sub-Saharan Africa. The discussion demonstrates that land rights in Sub-Saharan Africa evolved in response to changing political, social, and economic conditions, often the results of governmental interventions that may not have been conducive to efficiency or equity. The evidence dispels some popular misconceptions about land rights systems in Sub-Saharan Africa. There is increasing individualization of ownership, and in many areas possession has always been individual. Even in areas where communal ownership has been imposed, cultivation and possession remain with individual households and an increasing range of rights to land are appropriated by the individual household. Existing and indigenous systems are not inherently equitable. Land sales and mortgaging by individuals are observed frequently in many areas where such transactions are not recognized under the formal legal system. The lesson from others parts of the world is that efficiency ultimately requires formal recognition of individual land rights. That stage has not been reached yet in many parts of Sub-Saharan Africa, but in many other parts the justification for a change in land rights arrangements already exists. The practical problem in such cases is the careful analysis of benefits and costs, including equity considerations. References Abasiekong, Edet M. 1981. "Pledging Oil Palms: A Case Study on Obtaining Rural Credit in Nigeria." African Studies Review 24, no. 1: 73-82. Adegboye, R. 0. 1974. "Land Tenure Problems and Improved Practices." In R. A. Kotey and others, eds. The Economics of Cocoa Production and Marketing, 159-64. Legon: University of Ghana, Institute of Statistical, Social and Economic Research. Anderson, J. N. D. 1954. Islamic Law in Africa. London: Her Majesty's Stationery Office. Baker, Randall. 1975. "'Development' and the Pastoral Peoples of Karamoja, North- eastern Uganda: An Example of the Treatment of Symptoms." In T. Monod, ed. Pastoralism in Tropical Africa, 187-202. London: Oxford University Press for the International African Institute. Barnes, J. A. 1954. Politics in a Changing Society. A Political History of the Fort Jameson Ngoni. London: Oxford University Press for the Rhodes Livingston Institute. Baron, D. 1978. Land Reform in Sub-Saharan Africa: An Annotated Bibliography. Washington, D.C.: U.S. Agency for International Development, Office of Rural Development. Bates, Robert H. 1981. Markets and States in Tropical Africa. The Political Basis of Agricultural Policies. Berkeley: University of California Press. Berry, Sara S. 1975. Cocoa, Custom and Socioeconomic Change in Rural Western Nigeria. Oxford: Clarendon Press. 164 Research Observer 2, no. 2 (July 1987) Biebuyck, D., ed. 1963. African Agrarian Systems. London: Oxford University Press for the International African Institute. Binswanger, H. P., and M. R. Rosenzweig. 1986. "Behavioral and Material Determinants of Production Relations in Agriculture." journal of Development Studies 22, no. 3: 503-39. Brokensha, David, and Bernard W. Riley. 1980. "Introduction of Cash Crops in a Marginal Area of Kenya." In Robert H. Bates and Michael F. Lofchie, eds. Agricultur- al Development in Africa. Issues of Public Policy, 244-74. New York: Praeger. Bruce, John W., and Peter P. Dorner. 1982. Agricultural Land in Zambia: Perspectives, Problems, and Opportunities. Research Paper 76. Madison, Wisc.: University of Wisconsin, Madison, Land Tenure Center. Chalamwong, Y., and G. Feder. 1986. Land Ownership Security and Land Values in Rural Thailand. World Bank Staff Working Paper 790. Washington, D.C. Cheater, Angela P. 1984. Idioms of Accumulation. Harare, Zimbabwe: Mambo Press. Clark, R. J. 1969. "Problems and Conflicts over Land Ownership in Bolivia." Inter- American Affairs 22, no. 4 (Spring). Cobb, Richard, and others. 1980. Liberia: Ruiral Roads. Project Impact Evaluation 6. Washington, D.C.: U.S. Agency for International Development. Cohen, John M. 1980. "Land Tenure and Rural Development in Africa." In Robert H. Bates and Michael F. Lofchie, eds. Agricultural Development in Africa. Issues of Public Policy, 249-300. New York: Praeger. Collier, Paul. 1983. "Malfunctioning of African Rural Factor Markets: Theory and a Kenyan Example." Oxford Bulletin of Economics and Statistics 45, no. 2: 141-72. Colson, Elizabeth. 1963. "Land Rights and Land Use Among the Valley Tonga of the Rhodesian Federation: The Background to the Kariba Resettlement Program." In D. Biebuyck, ed. African Agrarian Systems, 137-56. London: Oxford University Press for the International African Institute. David, C., and R. Meyer. 1980. "Measuring the Farm Level Impact of Agricultural Loans." In J. Howell, ed. Borrowers and Lenders: Rural Financial Markets and Institutions in Developing Countries. London: Overseas Development Institute. Doggett, Clinton L., Jr. 1980. Land Tenure and Agricultural Development in Lesotho and Swaziland: A Comparative Analysis. Washington, D.C.: U.S. Agency for In- ternational Development, Bureau for Africa, Office of Development Resources, Agricultural/Rural Development Division. Dorner, P., and B. Saliba. 1981. "Interventions in Land Markets to Benefit the Rural Poor." Discussion Paper. University of Wisconsin, Land Tenure Center, September. Dravi, Mokpokpo Muki. 1984. The Alleviation of Rural Poverty in Cameroon. Rome: Food and Agriculture Organization. Eckert, J. 1980. Lesotho's Land Tenure: An Analysis and Annotated Bibliography. Lesotho Agricultural Sector Analysis Project Special Bibliography 2. Maseru, Lesotho and Ft. Collins, Colo.: Lesotho Ministry of Agriculture and Colorado State Univer- sity, Department of Economics. Ega, L. Alegwu. 1979. "Security of Tenure in a Transitory Farming System: The Case of Zaria Villages in Nigeria." Agricultural Administration 6, no. 4: 287-97. Eicher, Carl K., and Doyle C. Baker. 1982. Research on Agriculture Development in Sub-Saharan Africa: A Critical Survey. East Lansing: Michigan State University. Elias, T. 0. 1956. The Nature of African Customary Law. Manchester: Manchester University Press. Fallers, Lloyd. 1955. "The Modern African Chief." American Anthropologist 57, no. 2: 290-305. Gershon Feder and Raymond Noronha 165 Feder, G. 1987. "Land Ownership Security and Farm Productivity in Rural Thailand." Journal of Development Studies (forthcoming). Feder, G., and T. Onchan. 1987. "Land Ownership Security and Farm Investment." American Journal of Agricultural Economics (forthcoming). Feder, G., and others. 1986. "Land Ownership Security and Access to Credit in Rural Thailand." Discussion Paper ARU-53. World Bank, Agriculture and Rural Develop- ment Department, Agricultural Research Unit, Washington, D.C. Feeny, D. 1982. The Political Economy of Productivity: Thai Agricultural Development 1880-1975. Vancouver: University of British Columbia Press. Food and Agriculture Organization. 1984. Institutional Aspects of Shifting Cultivation. Prepared by Mary Triffin. Rome. Gladwin, C., and others. 1987. "Reaffirming the Agricultural Role of African Women: One Solution to the Food Crisis." Journal of African Studies (forthcoming). Glazier, Jack. 1984. "Mbeere Ancestors and the Domestication of Death." Man 19: 133-48. . 1985. Land and the Uses of Tradition among the Mbeere of Kenya. Boston: University Press of America. Gluckman, Max. 1941. Economy of the Central Barotse Plains. RLI Paper 7. Livingstone: Rhodes-Livingstone Institute. Goody, Jack. 1980. "Rice-burning and the Green Revolution in Northern Ghana." Journal of Development Studies 16, no. 2: 136-55. Grier, Beverly. 1981. "Underdevelopment, Modes of Production and the State in Colonial Ghana." African Studies Review 24, no. 1: 21-47. Hailey, Malcolm. 1957. An African Survey. Revised 1956. London: Oxford University Press. Hamer, Andrew Marshall. 1986. "Urban Sub-Saharan Africa in Macroeconomic Per- spective: Selected Issues and Options." World Bank, Water Supply and Urban Development Department, Operations Policy Staff, Washington, D.C. Hamnet, Ian. 1975. Chieftainship and Legitimacy: An Anthropological Study of Exec- utive Law in Lesotho. London: Routledge and Kegam Paul. Haswell, Margaret R. 1963. The Changing Pattern of Economic Activity in a Gambia Village. London: Her Majesty's Stationery Office. Haswell, Margaret R. 1975. The Nature of Poverty. A Case History of the First Quarter- Century After World War 11. London: Macmillan. Haugerud, Angelique. 1983. "The Consequences of Land Tenure Reform Among Smallholders in the Kenya Highlands." Rural Africana 15-16: 65-89. Hecht, Robert Michael. 1982. "Cocoa and the Dynamics of Socio-economic Change in Southern Ivory Coast." Ph.D. dissertation, Cambridge University, Cambridge. . 1985. "Immigration, Land Transfer and Tenure Changes in Divo, Ivory Coast, 1940-80." Africa 55, no. 3: 319-36. Hill, Polly. 1963. The Migrant Cocoa-Farmers of Southern Ghana. Cambridge: Cam- bridge University Press. Hopkins, Anthony G. 1973. An Economic History of West Africa. New York: Columbia University Press. - 1978. "Innovation in a Colonial Context: African Origins of the Nigerian Cocoa Farming Industry, 1880-1920." In Clive Dewey and Anthony G. Hopkins, eds. The Imperial Impact: Studies in the Economic History of Africa and India, 83-96. London: Athlone Press and University of London. Ibik, J. 0. 1971. "Restatement of African Law, 4." In Malawi II. London: Sweet and Maxwell. 166 Research Observer 2, no. 2 (July 1987) Ike, Don Nnaemeka. 1977. "A Comparison of Communal, Freehold and Leasehold Land Tenure: A Preliminary Study in Ibadan and Ife, Western Nigeria." American Journal of Economics and Sociology 36, no. 2: 187-95. Inter-American Development Bank. 1986. "Jamnaica Land Titling Project Feasibility Report." Washington, D.C. Kemp, J. 1981. "Legal and Informal Land Tenures in Thailand." Modern Asian Studies 15, no. 1: 1-23. Koby, A. T. 1979. "Projection des Formations Sociales sur l'Espace: Example du Pays Odzukru en Cote d'Ivoire." In Mabtrise de l'Espace Agraire et Developpement en Afrique Tropical. Paris: ORSTOM. Koehn, Peter. 1983. "State Land Allocation and Class Formation in Nigeria." Journal of Modern African Studies 21, no. 3: 461-81. Lawry, Steven W. 1983. "Land Tenure, Land Policy and Smallholder Livestock Devel- opment in Botswana." Research Paper 78. Madison: University of Wisconsin, Madi- son, Land Tenure Center. Levin, Michael David. 1976. "Family Structure in Bakosi: Social Change in an African Society." Ph.D. dissertation, Princeton University, Princeton, N.J. Lewis, John Van Dusen. 1979. "Descendants and Crops: Two Poles of Production in a Malian Peasant Village." Ph.D. dissertation, Yale University, New Haven, Conn. Lunning, H. A. 1965. "The Impact of Socio-Economic Factors on the Land Tenure Pattern in Northern Nigeria." Journal of Administration Overseas 4, no. 3: 173-82. Maini, K. M. 1967. Land Law in East Africa. London: Oxford University Press. Malcolm, D. M. 1953. Sukuimaland: An African People and Their Country. London: Oxford University Press. Meek, C. K. 1949. Land Law and Custom in the Colonies. 2d ed. London: Oxford University Press. (Reprinted London: Frank Cass, 1968). Middleton, John. 1961. Land Tenure in Zanzibar. London: Her Majesty's Stationery Office. Mosaase, A. 1984. "An Analysis of Existing Land Tenure in Lesotho and Experience in the Implementation of Current Land Policy." In Commonwealth Association of Surveying and Land Economy. The Development of African Land Resources (With Particular Reference to Central Southern Africa), 88-95. London. Netting, Robert McC. 1965. "Household Organization and Intensive Agriculture: The Kofyar Case." Africa 35, no. 4: 422-29. . 1985. Hill Farmers of Nigeria: Cultural Ecology of the Kofyar of the Jos Plateau. Seattle: University of Washington Press. Njeru, Enos Hudson Nthia. 1978. Land Adjudication and Its Implications for the Social Organization of the Mbere. Research Paper 73. Madison: University of Wisconsin, Madison, Land Tenure Center. Nkunva, G. K. 1974. "Land Tenure and Agricultural Development in the Angola Area of the Volta Region." Land Tenure Center Paper 120. Madison: University of Wisconsin, Madison, Land Tenure Center. Okali, C. 1983. COaoa and Kinship in Ghana. Ibadan, Nigeria: International Livestock Center for AfricA. Okoth-Ogendo, H. W. 0. 1976. "African Land Tenure Reform." In Judith Heyer and others, eds. Agricultural Development in Kenya, 152-86. Nairobi: Oxford University Press. Pala, A. 0. 1978., "Women's Access to Land and Their Role in Agriculture and Decision-Making on the Farm: Experience of the Joluo of Kenya." Discussion Paper 263. University of Nairobi, Institute of Development Studies, Nairobi, Kenya. Gershon Feder and Raymond Noronha 167 Palmer, Robin and Neil Parsons, eds. 1977. The Roots of Rural Poverty in Central and Southern Africa. Berkeley: University of California Press. Panandikar, A. 1956. Banking in India. Madras: Orient Longmans. Pelissier, Paul. 1966. Les paysannes de Senegal: Les civilizations agraires du cayor a la Casamance. Abregue: Saint Yrieux (Haute Vienne". Peters, Pauline Elaine. 1983. "Cattlemen, Borehold Syndicates and Privatization in the Kgatlen District of Botswana: An Anthropological History of the Transformation of a Commons." Ph.D. dissertation, Boston Universit:y, Boston. Pitblado, J. Roger. 1981. The North Mkata Plain, Tanzania: A Study of Land Capability and Land Tenure. Toronto: University of Toronto Press. Richards, Audrey I., and others, eds. 1973. Subsistence to Commercial Farming in Present-day Buganda: An Economic and Anthropological Survey. Cambridge: Cam- bridge University Press. Roden, David. 1971. "Changing Patterns of Land Tenure among the Nuba of Central Sudan." Journal of Administration Overseas 10, no. 4: 294-309. Rosegrant, M., and R. Herdt. 1981. "Simulating the Impacts of Credit Policy and Fertilizer Subsidy in Central Luzon Rice Farms, the Philippines." American Journal of Agricultural Economics 63, no. 4: 655-65. Rowling, C. W. 1946. Land Tenure in Kano. Kaduna: Government Printer. Sacay, 0. 1973. "Credit and Small Farmer Development in the Philippines." In Spring Review of Small Farmer Credit: Small Farmer Credit in the Philippines. Vol. 13. Washington, D.C.: U.S. Agency for International Development. Salas, O., and others. 1970. "Land Titling in Costa Rica: A Legal and Economic Survey." University of Costa Rica Law School, San Jose. Scott, Earl, ed. 1984. Life before the Drought. Boston: Allen and Unwin. Seidman, Robert B. 1975. "Law and Development: The Interface between Policy and Implementation." Journal of Modern African Studies 13, no. 4: 641-52. Seligson, M. A. 1982. "Agrarian Reform in Costa Rica: The Impact of the Titles Security Program." Inter-American Economic Affairs 35, no. 4 (Spring): 31-56. Shepherd, Andrew. 1981. "Agrarian Change in Northern Ghana: Public Investment, Capitalist Farming and Famine." In Judith Heyer and others, eds. Rural Development in Tropical Africa, 168-92. New York: St. Martin's. Snyder, Francis G. 1981. Capitalism and Legal Change. New York: Academic Press. Spiro, Heather. 1985. The Llora Farm Settlement in Nigeria. Hartford: Kumarian Press. Spring, Anita, and Art Hansen. 1985. "The Underside of Development: Agricultural Development and Women in Zambia." Agriculture and Human Values 2, no. 1: 62-67. Sorrenson, M. P. K. 1967. Land Reform in the Kikuyu Country. A Study in Government Policy. Nairobi: Oxford University Press. Stifel, D. L. 1976. "Patterns of Land Ownership in Central Thailand during the Twentieth Century." Journal of the Siam Society 64, no. 1 (January): 237-74. Stiglitz, J., and A. Weiss. 1981. "Credit Rationing and Markets with Imperfect Information." American Economic Review 71, no. 3 (June): 393-411. Tanabe, S. 1978. "Land Reclamation in the Chao Phraya Delta." In Y. Ishii, ed. Thailand: A Rice Growing Society, trans. by P. Hawkes and S. Hawkes. Honolulu: University Press of Hawaii. Tanzania Ministry of Agriculture, Task Force on National Africultural Policy. 1982. The Tanzania National Agricultural Policy: Final Report. Dar Es Salaam. Thomson, James Trevor. 1976. "Law, Legal Process and Development at the Local Level 168 Research Observer 2, no. 2 (July 1987) in Hausa-Speaking Niger: A Trouble Case Analysis of Rural Institutional Inertia." Ph.D. dissertation, University of Indiana, Bloomington. Tomosugi, T. 1980. "A Structural Analysis of Thai Economic History: A Case Study of a Northern Chao Phraya Delta Village." Institute for Developing Economies, Tokya. USAID (United States Agency for International Development)/RBoO (River Basin Devel- opment Office). 1982. Social Analysis. Vol. 3, sec. 2. Integrated Development Project 625-0621. Dakar, Senegal: USAIDORBDO. University of Arizona, Arid Lands Natural Resources Committee. 1979. Etude sur la commercialisation du betail et de la viande dans le departement de Zinder, Niger. Prepared by Amoul Kinni. Tucson. Van Velsen, J. 1964. The Politics of Kinship. A Study of Social Manipulation among the Lakeside Tonga. Manchester: Manchester University Press for the Rhodes-Living- stone Institute. Villamizar, F. 1984. "IBD Financing in Land Administration Programs." Paper presented at the International Workshop on Land Tenure Administration, Salvador, Brazil, August 20-24. Wai, U. Tun. 1957. "Interest Rates Outside the Organized Money Markets of Under- developed Countries." IMF Staff Papers 6 (1): 80-142. Ward, Jennifer Claudette. 1976. "The Bambara-French Relationship, 1880-1915." Ph.D. dissertation, University of California, Los Angeles. Watts, Michael. 1983. "Good Try, Mr. Paul: Populism and the Politics of African Land Use." African Studies Review 26, no. 2: 73-83. West, Henry W. 1972. Land Policy in Buganda. Cambridge: Cambridge University Press. Woodman, Gordon R. 1967. "Developments of Pledges of Land in Ghanian Customary Law." Journal of African Law 11, no. 1: 8-26. Gershon Feder and Raymond Noronha 169 THE RELEVANCE FOR DEVELOPING COUNTRIES OF RECENT DEVELOPMENTS IN MACROECONOMIC THEORY W. Max Corden Tn his article reviews the recent literature of macroeconomic theory in industrial countries and considers its relevance for developing countries. Much of the paper is concerned with the "rational expectations revolution" and its possible use for studying macroeconomic policies in developing countries. It would be helpful to specify some standard model for industrial countries, or perhaps some textbook written for them, and then relate it to a standard model for developing countries or one of them in particular. That would bring out differences between the models and show how the industrial-country model (or the textbook) needs ad- justing to suit a particular developing country. But that approach is impossible, because macroeconomic theory for the industrial countries has not been consolidated since the destruction of the Keynesian consensus, and the characteristics and experiences of the developing countries are so varied. Every idea in the recent literature on industri- al countries can be shown to have some relevance for some develop- ing country, and the literature on developing countries is so extensive that no important issue has been completely neglected. For this reason every generalization made here must be qualified. The central message of Keynesian economics is that demand man- Keynesianism agement through monetary and fiscal policies can successfully stabilize and Rational output and employment and possibly raise the average level of em- Expectations ployment over a longer period. The recent theoretical literature on © 1987 The International Bank for Reconstruction and Development/The World Bank 171 industrial countries seeks to explain how Keynesian policies can work when some allowance is made for rational behavior and expectations of private agents. This literature has three parts. The first part as- sumes that market-clearing wages and prices are continuously estab- lished, and the question is then whether Keynesian effects can still be generated through policy surprises. The second part seeks to explain the existence of nominal rigidities, especially wage rigidities. The third part-the most interesting for the present discussion-is concerned with the role of policy surprises and the scope for policy activism, given some rigidities. The Market-Clearing Approach Suppose that there were instantaneous market clearing; any increase in nominal demand in a particular market would thus lead to a rise in the price. Surprise shocks in nominal demand-for example, an in- crease in the money supply-would still have real effects if their implications were initially misunderstood. This is an instance of "in- complete information."' For example, if a growth in nominal demand leads to a rise in prices and nominal wages, without necessarily any change in real wages, but it is misunderstood as either a rise in real wages (which would lead to an increase in labor supply) or a rise in real profitability (which would lead to an expansion by firms, so boosting labor demand), then nominal demand policies would have real effects. This type of story (the Lucas supply function) has been given prominence in the literature on industrial countries. The rational expectations theorists argue that it is not possible to surprise private agents consistently. If the government is following some kind of policy rule, that rule will be discovered in due course. On average, changes in nominal demand will come to be expected, so will no longer be misunderstood when they happen. It does seem implausible that firms or workers misunderstand a change in demand. They may not understand why it happens and may have failed to anticipate it, but wit:h instantaneous price and wage flexibility it is not really necessary to anticipate events. If one believes that prices and wages are very flexible in response to market conditions, one should not expect nominaol demand policies to cause changes in real output and employment, even when the policies are expected. The assumption of instantaneous market clearing is patently unrealistic, so that the whole discussion becomes interesting only when it introduces some rigidities in prices and wages or at least significant lags in their adjustment to changing conditions. 172 Research Observer 2, no. 2 July 1987) Explaining Nominal Rigidities An extensive literature seeks to explain nominal rigidities, especially wage rigidity, as part of an efficient maximizing process. It is notable that nominal wages are largely rigid downward, and also somewhat sluggish upward, even where (as in the United States) trade unions are relatively weak. This is partly because explicit contracts with unions tend to be longer-term in the United States than in other industrial countries. The literature also tries to explain wage rigidities in terms of negotiation costs, including the costs and delays involved in the process of reaching agreed positions (the "logic of collective action"). There is also an implicit contract literature (see Azariadis and Stiglitz 1983). It seeks to explain wage rigidity by the willingness of employ- ers to accept the risks for their profits of fluctuations in demand, employees being risk-averse and unable to ensure adequately against income fluctuations. Most explanations of wage rigidities seem to be explanations of real rather than nominal rigidities, and only the difficulties of indexa- tion (which, surely, are not so great) can explain why rigidities that are intended to be real turn out to be nominal. If people learn from the experience of inflation, they would make increased use of indexed contracts-as has happened in several developing countries. Policy Surprises and Nominal Rigidities The most important issues in recent literature are (a) the interaction among nominal rigidities, especially wage rigidity; and (b) whether policy changes are expected or not. By introducing some nominal rigidities into a rational expectations framework, results that are es- sentially Keynesian can be obtained (Fischer 1977, 1980). The litera- ture assumes that nominal wages are fixed on some rational basis- taking into account expectations about market conditions, prices, and so on-but are then not changed for some time. Contracts may overlap, so that the general wage level gradually responds to changing expectations, even though particular wage rates stay fixed for a time. Given the assumption of short-term nominal rigidities, demand pol- icies would have no real effects if (a) demand policies are determined at the same time as prices and wages are fixed; (b) the principles of those policies are known to private agents; (c) the policies are based on information that is available to private agents as much as to the government; and (d) private agents interpret the information correctly. The absence of assumptions b and c means that there is "asymmetric information." Assumption d is usually implicit in the rational expectations litera- W. Max Corden 173 ture, though it is hardly discussed. Assumptions b and c are the critical rational expectations assumptions; for demand policies to be effective, say rational expectations proponents, at least one of these two assumptions must not apply. There must be some element of surprise. These rather extreme conclusions are changed radically when as- sumption a is removed. It is quite possible that wages are rigid longer than demand policy is. The government may have more flexibility than, say, the labor market. There is (my term) "asymmetric flexibility." In that case Keynesian demand policies may be effective even though there is no information asymmetry. The government can adjust its policies in the light of new information, but even thouagh the information is also available to and absorbed by private agents (so that expectations are rational) wages are not adjusted or are adjusted less. The policy regime, which includes the pattern of policy reactions to new information obtained by the government, is known by the private agents from the start and is adhered to by the authorities. The "surprise" is embodied in the new information, rather than in the policy. The general conclusion is that the more government demand poli- cies can surprise private agents (essentially by being unsystematic) or the longer the periods of wage rigidity relative to those of policy rigidity, the more likely is it that Keynesian demand policies will have some real effects. But if real wage outconies turn out to be consist- ently different from those intended by the private agents then presum- ably private agents will learn and adapt their behavior accordingly. If it is usual for significant new information about labor market condi- tions and demand policies to emerge after wage contracts have been made, then presumably the contract periods will be shortened. Are There How relevant is all of this for developing countries? Many macro- Keynesian economic models that have been constructed specifically for develop- Effects in ing countries have Keynesian elements: there is some degree of wage Developing and price rigidity, or at least some sluggishness of adjustment. Hence Countries? it is certainly a widespread view that Keynesian effects are relevant for developing countries. Notable examples are the paper by Krugman and Taylor (1978), which deals with the possible contractionary ef- fects of devaluation, and the paper by Khan and Knight (1981), which is concerned with stabilization policy when the exchange rate is fixed. Some people think that prices and nominal wages are more flexible in many developing countries than in industrial countries, mainly because trade unions are more powerful in the latter (though this does not mean that developing countries are free of structural unemploy- ment or of supply rigidities of various kinds). This view can be tied to the argument that significant Keynesian effects cannot be obtained 174 Research Observer 2, no. 2 (July 1987) from unexpected nominal demand policies purely because their effects are misunderstood by private agents. In such a framework nominal demand policies would have no real effects irrespective of whether they were expected or not. Only when there are significant short-term nominal rigidities are Keynesian policies and analysis relevant. Developing countries usually want to determine the effectiveness of devaluing the exchange rate. As is well known, the reaction of wages to a devaluation is crucial. Some rigidity in nominal wages or in the prices of nontradables is needed if a nominal devaluation is to lead to a real devaluation. If nominal wages are flexible downward, adjusting the exchange rate would be unnecessary, while if real wages are rigid, lowering the nominal exchange rate would have no real effects (aside from any real balance effect). This is the main form in which the issue of Keynesian effectiveness arises in small open economies. In such economies Keynesian policy essentially consists of using the exchange rate as a policy tool. If a developing country finds that it has unemployment in its tradables sector and if its government thinks that the Keynesian conditions for policy effectiveness apply, it would devalue the exchange rate and expand the money supply. Employment would then increase, even while the current account is kept at its original level. Keynesian assumptions may also apply to some nontradables, not- ably construction. It would then be possible to increase employment by expanding nominal demand, leading possibly to lower real wages through the exchange rate depreciation that demand expansion brings about. Monetary and fiscal policies could then be used to stabilize domestic demand in the face of fluctuating private investment. The crucial issues are empirical, and much research has been done. Some developing countries show evidence of having enough rigidity to allow devaluations or (unexpected) monetary acceleration to have short-term effects. But usually the effects are eroded after a few years.2 This suggests that there might be a short-term nonvertical Phillips curve, perhaps with adaptive expectations, which gradually shifts in an unfavorable direction as nominal demand expands. That in turn raises the question of whether devaluation, if it is to be effective, must be unexpected at the time that nominal wages and prices are initially fixed. If there is some nominal rigidity that lasts for a significant time (as in the United States), then all the issues discussed in the literature on industrial countries become potentially relevant. Yet that begs the question of why anything should ever be rigid in nominal, rather than real, terms. Long-term labor contracts like those in the United States are not common in developing countries. The strength of labor unions varies greatly from one developing country to another, but in general they are strong only in the urban sectors. W. Max Corden 175 The Rational In Keynes's General Theory expectations were exogenous (animal Expectations spirits) and influenced the inducement to invest. They played no part Revolution in determining nominal wages. The crucial role of expectations-and, above all, of expectations that proved to be consistently wrong-was seen by Friedman and Phelps when they sought the logic of the nonvertical Phillips curve. The expectations-augmented Phillips curve was derived from backward-looking expectations, with learning for- malized by the concept of adaptive expectations. This approach led to the conclusion that demand management could maintain unemploy- ment below the natural rate, but only aiL the cost of accelerating inflation. "Rational expectations" was the next logical step.3 Expectations would surely be forward looking. Mistakes would not be made con- sistently. Private agents would make use of all available information, not just past inflation, to develop their models of the economy. If they find these models to be consistently wrong, they will gradually discard them until they develop the right model. This approach raises several questions. Hlow does one know what is the right model? It has often been remarked that even respectable economists have many models, and the faith that "truth will prevail" may seem a little naive. Does history really teach that mistakes are not repeated? For many people (including politicians in power) the chance to learn from one's own mistakes is limited because decisions on particular matters are infrequent. In any case, the central theme of rational expectations theory is that individuals do not make system- atic mistakes in forecasting the future. This limits the ability of gov- ernments to steer the economy by tricking private agents. Clearly, the simplest versions of rational expectations theory re- quire considerable qualification. In particular, how far private agents are able to assimilate information is crucial. Presumably there is some kind of production function where information is the input and un- derstanding the output. The mental model of the economy into which the information is fed is also crucial. In all these respects, private agents can differ from one another. Information may be available to some and not to others, perhaps because the costs of obtaining the information vary. There may be economies of scale in obtaining infor- mation. Not all agents (like not all economists) will have the same model of the economy. If agents have different abilities to obtain or process information, some being more "rational" than others, the outcome may or may not be dominated by the more rational agents (Haltiwanger and Waldman 1985). One should not be surprised that these and other problems have not been ignored in the literature, so that the early, rather naive versions of rational expectations theory would no longer be widely 176 Research Observer 2, no. 2 (July 1987) accepted even by the enthusiasts. For example, Barro and Gordon (1983) explore in detail the relationship between the credibility of government monetary policy rules and the expectations of private agents. That relationship determines the real effects of monetary poli- cy shocks or changes. A lot depends on how private agents interpret government policies. The literature puts great emphasis on the effects of "systematic" The Nature (and hence, in principle, predictable) government policies. The two of Government policies usually referred to are the old-fashioned Keynesian ones of Policies fine-tuning demand to offset private sector instabilities and the Fried- man policy of steady monetary growth. Because of the difficulties of measuring money and the unpredictability of velocity, the Friedman prescription can be recast as a constant rate of nominal growth in gross national product. Are Policies Systematic in Developing Countries? In trying to apply these notions to developing countries, it is natur- al to ask how far their policies can be described as systematic. Some developing countries have stable governments and stable economic philosophies; in some cases, such as India, they are even more stable than those of some industrial countries. In other developing countries, it would be hard to describe policies as systematic in any sense. (Yet, if one stretches the term a bit, even apparently chaotic policies or nonpolicies could be described as systematic.) Taking all the inevitable variety into account, the weight given in the macroeconomic literature to private agents' reactions to systematic government policies seems out of proportion in the context of developing countries-and possi- bly of many industrial ones as well. Is Government Benign? Underlying Keynesian thinking was the idea of government as a benign, well-informed optimizer of the national interest, manipulating the various rigidities in the private sector and offsetting its destabiliz- ing activities. In an open economy the last point can be extended to allow for destabilizing effects coming from abroad, perhaps originated by the policies of foreign governments rather than by the private sector. In a broad sense rational expectations theory can be regarded as operating within the same framework as Keynesian theory, even though its main contribution has been to highlight the doubtful as- W. Max Corden 177 sumptions that crude Keynesianism requires for its validity. The impli- cit assumption is still an optimizing government that can pursue whatever policies are deemed desirable. Both approaches employ the concept of "the" government or "the" central bank. The government acts on the private sector, and the private sector reacts through the market place, that is, through price and wage reactions. The whole edifice of rational expectations theory is concerned with this two-way relationship, and above all with whether the government can "trick" the private sector into behavior that it would not engage in if it had full information and made full use of that information. Neither Keynesian nor rational expectations theory considers the motives of governments, or whether private agents can influence the behavior of governments directly (through lobbying, for example) rather than through the market. Political Economy The vital macroeconomic issues for the developing countries, and possibly for some industrial ones as well, seem to be outside the scope of rational expectations theory. What are the macroeconomic objec- tives of different governments at different times? How do pressure groups influence them? How do finance ministers and central banks balance one pressure against another? There is a "political market" in which pressure groups operate. Government policies certainly affect private behavior, but the private sector also affects the government. And policy is likely to be influ- enced by the interests of politicians as a class, and of various groups of government employees. Public choice theory might shed light on some of these issues, though its formal development appears to be relevant only for democracies. One theme in the political economy literature is that politicians are driven by short-term, electoral considerations (Nordhaus 1975). Given that a demand expansion may (in a Keynesian model) boost employ- ment before it raises inflation, the politicians' high discount of the future may lead to more inflation than the national welfare would favor. In the context of recent developments in developing countries, if Keynesian effects are not significant, the short-term benefit to politi- cians may be higher public spending financed by foreign borrowing, while the longer-term cost may be not just inflation but also an increased burden of foreign debt. One question is whether govern- ments' discount rates are consistently lower when governments are more stable, and thus feel themselves to be more securely in pow- er. To what extent do the more insecure governments buy support through current account deficits? 178 Research Observer 2, no. 2 (July 1987) Rational expectations theorists have assiduously explored many The Rational questions about the private sector. Surprisingly, they have not done Expectations the same with governments. Is all available information used by gov- of Government ernment decisionmakers? What are the costs of acquiring information, and when will it pay not to acquire all possible information? What is the model into which the public decisionmakers feed their informa- tion? If there are wrong and right models, do the right models even- tually prevail because politicians or decisionmakers with the right ones defeat those with wrong ones? (In other words, does truth pre- vail?) Could outcomes be improved if public decisionmakers obtained more information and more education in the right models? Perhaps this is where the policy advice of the International Monetary Fund and the World Bank fits in. This discussion suggests an empirical program that could usefully be applied to the public sector. The program would include consider- ing a government's ability to assimilate information even when the model is given. Do policy shifts reflect changing information, chang- ing ability to absorb information, changing models, or changing ob- jectives? If the objectives have changed, is it because pressure groups or individuals have changed their objectives, or because the weight- ing given to different pressure groups or individuals in power has changed, with each having a different but constant objective? These questions need to be considered in any assessment of macroeconomic policies for developing countries. An important feature of fiscal theory in the industrial countries is The Neo- the rediscovery of the neo-Ricardian equivalence theorem.4 In the case Ricardian of a bond-financed government deficit, it used to be argued that the Theorem and extra bonds represent net wealth in the hands of the public. The Developing public, it was said, ignores the future tax obligations incurred when Countries the government borrows. The new idea is that rational taxpayers will take this into account. The Ricardian theorem in its extreme form is that private savings will increase to provide for future tax obligations: a bond-financed deficit does not increase net wealth, because the increased holding of bonds in private hands is offset by the wealth- reducing effect of the higher expected taxes. Thus a budget deficit will not increase total expenditure; it will only shift the pattern of expen- diture. If the deficit resulted from a rise in government spending, the pattern of spending will shift from the private sector to the public. Extra public dissaving will be matched by extra private saving. Is this application of rational expectations theory to fiscal policy relevant for developing countries, where deficits tend to be money- W. Max Corden 179 financed? If the increase in the money supply is not expected to increase inflation or the current account deficit, but is just expected to boost output along Keynesian lines, there should not be any neo- Ricardian effect. If inflation is expected, the expectation that the real value of money holdings would be eroded (the inflation tax) would lead to higher savings. But the most relevant case for some developing countries may be essentially non-Keynesian, where a fixed exchange rate prevents or moderates inflation. Much or all of the budget deficit will then be reflected in a current account deficit. In that case the government does not pay any interest to the central bank or to local citizens; either foreign exchange reserves must be reduced (so forgoing interest earnings) or money must be borrowed abroad. One way or another, interest obligations and obligations eventually to repay foreign loans or rebuild the reserves are incurred. The perceived net wealth of taxpayers who foresee these conse- quences will decline; if they are rational, they will increase their savings-and hence produce an offsetting reduction in the current account deficit. If these reactions were instantaneous and completely offsetting-both very unlikely-a current account deficit would not result at all. Might there be any tendency to such offsetting savings behavior in developing countries? It is indeed possible that private households and firms may foresee that government borrowing abroad could create difficulties later. They may come to expect foreign pressures to raise taxes or cut public spending, leading to social disruption-and they might then wish to take precautions. But such precautions would not necessarily consist of building up financial assets denominated in domestic currency. They might wish to build up stocks of goods, or gold, or-most likely-assets denominated in foreign currency. The motive might be not to accumulate financial assets to finance future tax bills, but rather to prepare to evade future taxes by moving funds out of the domestic jurisdiction.5 Properly adjusted and broadened, the Ricardian theorem could have some relevance for developing countries. There is plenty of experience available for testing it, some of which suggests that fore- sight was not always shown by the private sector, possibly because the same analyses that led governments to think that certain policies would have favorable consequences led the private sector to the same conclusion. This section can be concluded with a general point: the fiscal policies of developing countries may create various expectations at home and abroad, which can have early effects on private sav- ings, private investment, and the desired portfolio balance between domestic- and foreign-currency-denominated financial assets, and 180 Research Observer 2, no. 2 (July 1987) hence on the exchange rate (in a floating rate system) or the overall balance of payments (in a fixed rate system). These expectational effects must be taken into account in any analysis of current fiscal policy. The tendency for real wages to be rigid downward, at least in the Real Wage short run, has been noted since 1973 in European and other industrial Rigidity countries outside the United States. A large literature has been built around the assumption of real wage rigidity in the industrial countries. Explaining the Rigidity It is not difficult to explain some involuntary unemployment and real wage rigidity by the "efficiency wage" hypothesis (Yellen 1984). In a static model it can also be explained by trade unions' exercising monopoly power at the expense of the unemployed. It is a little harder to explain rigidity in a dynamic model-why, for example, the real wage demanded by unions does not fall sufficiently in response to shocks, such as the oil price shock. There may be problems of infor- mation or of understarnding (using the wrong model or being unable to feed information efficiently into the right model). The logic of collective (in)action may also dictate slow adjustment. And if the negative shock is believed to be temporary, it might justify maintain- ing real wages when it is costly to negotiate and explain required real wage cuts to trade union members. Potentially, rational expectations theory is fully applicable to this issue, even though it is no longer concerned with explaining the determination of nominal wages and prices, and hence with analyzing the Keynesian question of the efficacy of managing nominal demand. The theory tries to explain the real wage expectations on which wage demands are based. What is the information base of these expecta- tions, what are the implicit models of the economy used, and so on? Raising Employment with Real Wage Rigidity Various policies have been proposed to get around the rigidity of real wages and thereby boost employment. All are applicable to devel- oping countries. One approach rests on the assumption that the rigidity is posttax, not pretax, and that it is not only downward but upward. In that model, the appropriate policy would be to reduce income tax or indirect taxes; this would lead to a decline in pretax real wages (to maintain posttax wages) and hence to higher employment (Corden W Max Corden 181 1981). The hope is that the wider tax base resulting from the extra employment will prevent a fall in government revenue. Another possible solution is short-term borrowing from abroad, the borrowed funds allowing taxes to be kept low and thus posttax wages high; the debts are repaid when productivity has increased and the rigid posttax real wage can be sustained without borrowing. Such borrowing may also allow public sector real wages to be higher than they would otherwise be. A third approach is to tax those sectors of the economy in which real incomes are not rigid, using the revenue to subsidize the sectors with real wage rigidity. An indirect and rather inefficient way of doing this would be through particular forms of trade protection, as suggest- ed in the implicit model of the Cambridge (England) protectionists. Real Wage Rigidity and Developing Couintries The assumption of real wage rigidity was implicit (and occasionally explicit) in the Latin American structuralist- models. Their themes may have been revived in the more recent literature dealing with European rigidities. Furthermore, real wage rigidity in one sector of a two-sector economy is central to the Harris-Todaro model, which seeks to ex- plain a form of structural unemployment in developing countries (though this model has not been used for analyzing macroeconomic policy). For developing countries the idea of rigid real wages may have to be extended to allow for at least two types of wage rigidity-in the public sector and the urban private sector. Other real rigidities- notably for particular categories of government expenditure and also for some rural incomes-might be added. The various real incomes in the country (presumably posttax and allowing for real incomes de- rived from certain forms of government expenditure) may have to be categorized according to how downwardly rigid they are. Presumably government would have two policy instruments: first, fiscal policy, which would include not only the size of the budget deficit (leading to a current account deficit and hence foreign borrowing) but also the mix of taxes and government spending and, second, incomes policy. Supply Side The literature also contains models that do not assume that the real Models and wage is rigid downward, but where all wage adjustments are made Sectoral explicitly in real terms. Such models allow for rational behavior and Income Effects seem relevant for many countries. They allow negative supply shocks to produce both an increase in unemployment and some fall in real wages as part of an optimizing response by trade unions. In such cases the real wage is unresponsive to nominal demand shocks, but it 182 Research Observer 2, no. 2 (July 1987) has some flexibility (though not enough to maintain employment) in response to real shocks. Models of this kind are expounded in Bruno and Sachs (1985), where further reference are also given.6 Some shocks, whether nominal or real in origin, may be "fa- vorable," causing real wages to rise; the original shock may be tempo- rary, yet real wages may not fall when it is reversed. This is surely a common feature of industrial and developing countries. Yet it is hard to derive such reactions from a model of rational behavior. If the favorable shock was known to be temporary to start with, so that its reversal is expected, rational behavior would not lead to a rise in real wages (or, at least, not one so great that it later needed reversing), unless the future were heavily discounted. If the favorable shock was initially expected to be permanent, rational behavior should later lead the trade unions to lower their real wage targets when new informa- tion about the reversal of the shock is revealed. One can imagine a body of real macroeconomic theory that is very different from the macroeconomics which is discussed now in text- books and which focuses on real rigidities and sectoral income effects in developing countries. A literature of this sort is forming, though it is not yet consolidated: see Prachowny (1984), Bruno and Sachs (1985), Corden (1985), and various papers by Scandinavian authors.' The "booming sector and Dutch Disease" literature comes into this category. It has been developed for both industrial and developing countries and is concerned chiefly with one issue: the sectoral real income effects, as well as overall macroeconomic consequences, of a sectoral real income boom (Corden 1984). In effect, the Dutch Disease models apply various aspects of macroeconomics to a particular prob- lem, but their special feature is the emphasis they place on real income and output effects on different sectors. Macroeconomic policy and events can be seen as the outcome of a Large Group game played between two or more large actors. Recent theorizing, Interaction mostly in Europe, might be relevant for some developing countries. and In the simplest model (originating in Scandinavia) the two actors Game Theory are the government and the trade unions. The first has fiscal policy as its instrument of policy; the second has the real pretax wage. The outcome yields the current account of the balance of payments and the level of employment (Calmfors 1982). Alternatively, the govern- ment's instrument is monetary policy (or the exchange rate), and the outcome yields the employment and price levels (Gylfason and Lind- beck 1986). There are several variants of such models, including those in which one party is committed in advance to a policy, or to a reaction function, while the other one then optimizes. Such models could be extended to more than two actors. The corporate sector W Max Corden 183 could be added, and the government could have both monetary and fiscal policy as instruments. Various game theory issues arise, includ- ing the crucial one of credibility of threats. For many industrial economies outside the United States, models of this kind seem more realistic than many of the contributions of ra- tional expectations theorists. The models can either assume that each party has full information, or that there is an asymmetry of informa- tion-the withholding of information and the surprising of the other parties being part of the game. The main weakness of the game theory approach is that govern- ments are not really independent of the private actors. Groups further their interests not just through their wage (or other income) demands in the market place, but also by directly influencing government poli- cies. Although in some developing countries the government controls the unions, in others (as in many industrial countries as well) the unions significantly influence the government. The applicability of this new body of theory to developing coun- tries depends on two considerations. First, if interest groups are not well organized, the model is inappropriate. Second, interest groups exercise their power directly on governments, through the political process, rather than by adjusting their nominal wages or product prices in response to varying government policies. Both types of inter- actions have to be taken into account. Conclusion The purpose of this article has not been to expound macroeco- nomic theory for developing countries. Instead it has been to assess the significance of various recent changes in theory (and also possible further changes) that are not yet fully ircorporated in the standard development theory and to determine whether these changes can be of use in studying the macroeconomic expeiriences of developing coun- tries.8 The main conclusions can be summarized as follows: 1. A crucial question concerns the extent to which there are nomin- al rigidities in developing countries, and especially-a familiar issue- how long it is that nominal devaluations have real effects. 2. The rational expectations approach raises interesting issues about the reaction of private agents to government behavior. The extent to which private behavior in developing countries suggests evidence of "rationality" in response to policy changes needs to be considered. 3. It is relevant to ask to what extent policies in various developing countries are systematic or can ever be so. 4. The political dimensions of macroeconomic policies need to be investigated to explain the effects that pressure groups have on gov- ernment policies. Orthodox theory studies the effects of the govern- 184 Research Observer 2, no. 2 (July 1987) ment on the private sector; it is also necessary to examine the effects, through the political market, of the private sector (as well as of government employees and politicians) on government policy. 5. The expectations of governments need to be analyzed. Are they rational? That is, what information and models do policymakers use, and what is their ability to process information? The article suggests the need for an empirical program, applying the insights of the ration- al expectations revolution to the public sector. 6. The neo-Ricardian theorem may have some relevance for devel- oping countries, particularly because it suggests that official borrow- ing abroad could lead to offsetting changes in private savings. More broadly, fiscal policies, notably the expectations of the difficulties that deficits might cause, are likely to provoke reactions by the private sector. Expectations must be taken into account in analyzing current fiscal policies. 7. Models involving the rigidity of posttax real wages, or at least some sluggishness in real wages, are relevant for many developing countries. The same is true of various solutions that have been ex- plored for maintaining or increasing employment when real wages are rigid. These models may need to be extended to allow for several sectors, and there is scope for a theory that focuses on real rigidities and sectoral income effects. 8. Game theory models may be useful for analyzing macroeco- nomic behavior in some developing countries, but if interest groups are not well organized such models are not really relevant. Interaction through the political process may be more important than through the market place. This article reviews recent developments in macroeconomic theory and considers their Abstract relevance for developing countries. Particular emphasis is given to the rational expecta- tions revolution. The article asks whether government policies are systematic and suggests the need to study the expectations on which governments base their decisions. Also discussed are the neo-Ricardian theorem, the implications of real wage rigidity and of supply side models with sectoral income effects, and recent models involving large group interaction. The views expressed in this article do not necessarily represent the views of the Notes International Monetary Fund. An earlier version of the article was prepared, at the suggestion of the late Carlos Diaz-Alejandro, as background for a World Bank project on macroeconomic policies, crises, and long-term growth in developing countries. A version of this paper will appear in a book of essays in his memory, to be published by Basil Blackwell, Oxford. The author is indebted to Carlos Diaz-Alejandro for stimulus and to valuable comments from John Cuddington and Mohsin Khan, as well as to Nuffield College, Oxford, Harvard University, and the Australian National University, where various versions of the article were written. W Max Corden 185 1. This market-clearing approach is set out systematically in Barro (1984, chap. 18), and originated with Lucas, Barro, and Sargent and Wallace (see note 3 below). 2. The literature on the effectiveness of devaluation is generally concerned with "competitiveness," rather than real wages. A well-known paper is Connolly and Taylor (1976). See also Warr (1984) and Khan and Knight (1985), which both contain further references. Recent Southern Cone experience suggests that nominal exchange rate policy has real effects, that is, that there are nominal wage and price rigidities. Some empirical work on the real effects of monetary policy, mostly in Latin America, is available. There is an excellent summary and review in Khan and Knight (1985). See especially Hanson (1980) and the references given there, as well as Edwards (1983) and Leiderman (1984). It appears, at least from Hanson (1980), that there is a significant relation between output and unexpected inflation-though there must always be doubt as to what the "expected" inflation really is. 3. Classic articles on the applicability of the rational expectations idea to macroeco- nomics are Sargent and Wallace (1976) and various papers by Lucas, reprinted in Lucas (1981). For a clear statement of implications, see Lucas and Sargent (1979). Important contributions have also come from Barro, and his current views can be found in Barro (1984). The textbook by Sheffrin (1983) is particularly useful. It contains an extensive bibliography. An excellent reference is Fischer (1980), which contains both sympathetic and critical analyses of the rational expectations approach. See also Buiter (1980); it is critical of "the new classical macroeconomics," at least as it was defined some years ago, and of claims about policy ineffectiveness that were made on the basis of this theory. 4. See Barro (1974), Barro (1984, p. 381), and references given in the latter. 5. As far as I know, no empirical work on the applicability of the neo-Ricardian hypothesis to developing countries has yet been published. Mohsin Khan has pointed out to me that the data suggests that the savings-CDP ratio is pretty stable even for countries with large shifts in public dissavings, and this could be a Ricardian equivalence effect. An interesting example is Chile in 1979-80, when a private financial deficit replaced the budget deficit, which had been practically eliminated. 6. See also Calmfors and Horn (1985) and Calmfors (1985) and references cited in these two papers, as well as various papers presented at a Stockholm conference on Trade Unions, Wage Formation, and Macroeconomic Stability published in the Scandi- navian Journal of Economics 87, no. 2, 1985. 7. The connection with elements of Latin American structuralism and the neostruc- turalist model-building of Taylor is obvious. I have not attempted to survey or analyze this literature. The best-known reference is Taylor (1983). 8. I have not dealt with the sophisticated literature of non-market-clearing macro- economic models. These assume price and wage rigidities (usually with no capital markets) and spell out the general equilibrium implications of various policies. The prices and wages are not endogenous, not even to the extent that they are in old- fashioned (non-expectations-augmented) Phillips curve models. Nothing is said about the explanations for the nominal rigidities. See Barro and Grossman (1976), Malinvaud (1977), and Muellbauer and Portes (1978); on these and the broader issues, including the causes of rigidities, see Cuddington, Johansson, and Lofgren (1984). References Azariadis, Costas, and Joseph E. Stiglitz. 1983. "Implicit Contracts and Fixed-Price Equilibria." Quarterly Journal of Economics 48, supplement: 1-22. Barro, Robert J. 1974. "Are Government Bonds Net Wealth?" Journal of Political Economy 82, no. 6 (December): 1095-117. . 1984. Macroeconomics. New York: Wiley. Barro, Robert J., and David B. Gordon. 1983. "Rules, Discretion and Reputation in a Model of Monetary Policy." Journal of Monetary Economics 12, no. 2 (July): 101-21. 186 Research Observer 2, no. 2 (July 1987) Barro, Robert J., and Herschel Grossman. 1976. Money, Employment and Inflation. Cambridge: Cambridge University Press. Bruno, Michael, and Jeffrey D. Sachs. 1985. Economics of Worldwide Stagflation. Cambridge, Mass.: Harvard University Press. Buiter, W. H. 1980. "The Macroeconomics of Dr. Pangloss: A Critical Survev of the New Classical Macroeconomics." Economic Journal 90 (March): 34-50. Calmfors, Lars. 1982. "Employment Policies, Wage Formation and Trade Union Behav- ior in a Small Open Economy." Scandinavian Journal of Economics 84, no. 2: 345-73. . 1985. "Trade Unions, Wage Formation and Macroeconomic Stability-An Introduction." Scandinavian Journal of Economics 87, no. 2: 143-59. Calmfors, Lars, and Henrik Horn. 1985. "Classical Unemplovment, Accommodation Policies and the Adjustment of Real Wages." Scandinavian Journal of Economics 87, no. 2: 234-61. Connolly, Michael, and Dean Taylor. 1976. "Testing the Monetary Approach to Devaluation in Developing Countries." Journal of Political Economy 84, no. 4 (August): 849-59. Corden, W. Max. 1981. "Taxation, Real Wage Rigidity and Employment." Economic Journal 91, no. 362 (June): 309-30. - 1984. "Booming Sector and Dutch Disease Economics: A Survey." Oxford Economic Papers 36, no. 3 (November): 359-80. . 1986. Inflation, Exchange Rates and the World Economy. Chicago: University of Chicago Press. Cuddington, John T., Per-Olov Johansson, and Karl-Gustaf Lofgren. 1984. Disequili- brium Macroeconomics in Open Economies. Oxford: Blackwell. Edwards, Sebastian. 1983. "The Short-Run Relation between Growth and Inflation in Latin America: Comment." American Economic Review 73, no. 3 (June): 477-88. Fischer, Stanley. 1977. "Long-term Contracts, Rational Expectations and the Optimal Money Supply Rule." Journal of Political Economy 85, no. 1 (February): 191-205. ed. 1980. Rational Expectations and Economic Policy. Chicago: University of Chicago Press for National Bureau of Economic Research. Gylfason, Thorvaldur, and Assar Lindbeck. 1986. "Wages, Money, and Exchange Rates: A Game-Theoretic Approach." Institute for International Economic Studies Seminar Paper 320. Stockholm. Haltiwanger, John, and Michael Waldman. 1985. "Rational Expectations and the Limits of Rationality: An Analysis of Heterogeneity." American Economic Review 75, no. 3 (June): 326-40. Hanson, James A. 1980. "The Short-Run Relation between Growth and Inflation in Latin America." American Economic Review 70, no. 5 (December): 972-89. Khan, Mohsin S., and Malcolm D. Knight. 1981. "Stabilization Programs in Developing Countries: A Formal Framework." International Monetary Fund Staff Papers 28, no. 1 (March): 1-53. -____ 1985. Fund-Supported Adjustment Programs and Economic Grouth. Interna- tional Monetary Fund Occasional Papers 41. Washington, D.C. Krugman, Paul, and Lance Taylor. 1978. "Contractionary Effects of Devaluation." Journal of International Economics 8, no. 3 (August): 445-56. Leiderman, Leonardo. 1984. "On the Monetary-Macro Dynamics of Colombia and Mexico." Journal of Development Economics 14, no. 1-2 (January-February): 183-201. Lucas, Robert E. 1981. Studies in Business-Cycle Theory. Cambridge, Mass.: MIT Press. W. Max Corden 187 Lucas, Robert E., and Thomas J. Sargent. 1979. "After Keynesian Macroeconomics." Federal Reserve Bank of Minneapolis Quarterly Review 3, no. 2 (Spring): 1-16. Malinvaud, Edmond. 1977. The Theory of Unemployment Reconsidered. Oxford: Blackwell. Muellbauer, John, and Richard Portes. 1978. "Macroeconomic Models with Quantity Rationing." Economic Journal 88, no. 352 (December): 788-821. Nordhaus, William. 1975. "The Political Business Cycle." Review of Economic Studies 42, no. 2 (April): 169-90. Prachowny, Martin F. J. 1984. Macroeconomic Analysis for Small Open Economies. New York: Oxford University Press. Sargent, Thomas, and Neil Wallace. 1976. "Rational Expectations and the Theory of Economic Policy." Journal of Monetary Economics 2, no. 2 (April): 169-83. Sheffrin, Steven M. 1983. Rational Expectations. Cambridge: Cambridge University Press. Taylor, Lance. 1983. Structuralist Macroeconomics: Applicable Models for the Third World. New York: Basic Books. Warr, Peter G. 1984. "Exchange Rate Protection in Indonesia." Bulletin of Indonesian Economic Studies 20 (August): 53-89. Yellen, Janet L. 1984. "Efficiency Wage Models of Unemployment." American Economic Review 74, no. 2, papers and proceedings (May): 200-05. 188 Research Observer 2, no. 2 (July 1987) FOREIGN TRADE REGIMES AND ECONOMIC GROWTH IN DEVELOPING COUNTRIES Deepak Lal Sarath Rajapatirana T he static case for free trade is as simple as it is powerful. The removal of barriers to foreign trade expands the feasible set of consumption possibilities. It does so by providing, in effect, an indirect technology for transforming domestic resources into the goods and services that yield current and future utility for consumers. This static case does not involve any commitment to laissez-faire;' the law of comparative advantage, as well as the gains from trade it underpins, applies to both socialist and capitalist economies. The dynamic version of the law incorporates investment in line with a country's changing comparative advantage, which minimizes the pre- sent value of the resource costs of its future demands. By widening the market, foreign trade also allows a country to exploit economies of scale. Furthermore, the competitive pressures exerted by imports pre- vent the emergence of welfare-reducing domestic monopolies and in- duce domestic producers to improve quality and reduce costs. To the extent the static gains are saved and invested efficiently, they will grow over time, while the introduction of new goods and (more important) new technology through foreign trade can affect an econo- my's rate of technical progress. Apart from this last factor, the result of moving toward free trade is a higher level of per capita income, not a permanently faster rate of growth. This is one argument currently being used to denigrate the case for free trade as a means of enhancing growth (see Lucas 1985); it will be discussed in the last section of this paper. Longer-standing skepticism includes, first, the claim that the static gains are fairly small for even large reductions in tariffs and, second, that dirigiste CC 1987 The International Bank for Reconstruction and Development/The World Bank 189 foreign trade regimes (which in 1945-65 encouraged import substitu- tion and then, after 1965, export promotion) are likely to do more to boost a country's growth rate. This article surveys empirical studies that seek to demonstrate the limited static gains from freer trade and then reviews studies of the dynamic effects of growth of exports on l-hat of per capita income. The following section summarizes the results of comparative studies of developing countries undertaken in the 1960s and early 1970s, which show fairly conclusively that "outward orientation" seems to be positively associated with faster growth and greater equity. The article then examines whether the conclusions of these studies hold for the more volatile conditions since then. It considers the various arguments that cast trade as an "engine of growth." The final section introduces certain insights of the classical writers-in particular Adam Smith-which have reemerged in the nec-Austrian, as well as the more recent neoclassical "new political economy" schools, which might explain the stylized facts about the links between trade and growth. These emphasize the importance of the nonquantifiable as- pects of a free trade regime in creating (in an irreducibly uncer- tain world) an economic framework that encourages entrepreneur- ship, productivity, and thrift. We argue that free trade is thus the handmaiden of growth, as it indirectly constrains the state from going beyond the bounds of providing those public goods essential for de- velopment. The Static Early studies of the costs of protection measured the static gains Gains from trade in terms of the familiar welfare triangles associated with from Trade complete or partial trade liberalization (as in the case of customs unions). Harberger (1959) estimated that the cost of protection in Chile amounted to "no more than 2'/2 percent of the national in- come" (p. 135). Scitovsky (1958) estimated the gains to the European Community from increased specialization at "less than one-twentieth of one percent of the gross social product of the countries involved" (p. 67). Johnson (1958) estimated the gain to Britain from the forma- tion of a free trade area as at most 1 percent of national income. Recently, several models have been developed to examine the gen- eral equilibrium effects of trade liberalization (AB in figure 1).2 These estimates are based on the standard Heckscher-Ohlin model with constant returns to scale. The gains are estimated as Hicksian-equiva- lent variations as percentages of gross dolmestic product (GDP) in a base period. The gains from trade liberalization appear to be small: Whalley (1984) estimates a global net gain of about 0.3 percent of world GDP in 1977, and the maximum for any region or country is 0.5 190 Research Observer 2, no. 2 (July 1987) percent of GDP (see Srinivasan 1986a, 1986b). Moreover, in Whalley's model, with a move to world free trade, the developing countries lose 4 percent of their GDP. As Srinivasan (1986a) has argued, however, the results of these models are not credible, partly because of how they manipulate data to make an internally consist- ent equilibrium set (when the data themselves come from Figure 1 nonequilibrium situations) and partly because of how they Importable specify some crucial elasticity d parameters. In particular, they all make use of estimated trade \ elasticities (see Stern and oth- \ ers 1976), which have a well- known bias to underestimation (see Orcutt 1950, Kemp 1962, Kakwani 1972).3 Most of these P. P models also do not take ac- count of scale economies and imperfect competition. An ex- ception is a model for Canada by Harris (1983), which shows that a multilateral reduction of D \ A \ all tariffs yields welfare gains of P, P, p more than 5 percent of GDP. Exportable Furthermore, most of these Note: PP is the production possibility frontier. F is the free trade output when models do not take account of world prices are given by tt. With a tariff and no lobbying or rent seeking, the production point is P, with the domestic price ratio being given by dd and the the "rent seeking" and "directly welfare cost by AB. With lobbying for the tariff, the production possibility curve shifts to P'P'. the production point to P~, and the welfare baas is CB. To avoid unproductive" activities trig- various immiserizing paradoxes" (which depend on the relative slopes of the gered by protectionism (see Rybczynski line between P and P*, and the world price ratio tt), which could imply a welfare gain from the lobbying equilibrium at P as compared with P, it is better Tullock 1967, Krueger 1974, to decompose the welfare loss CB into the loss due to the tariff CE (this is the Bhagwati 1980). In the case of usual triangle estimate) and that due to lobbying EB (see Bhagwati 1980). With rent seeking, the production possibility curve shifts inward to P"P", the production rent seeking, the deadweight point to P", and the welfare loss at world prices is DB. loss associated with, for in- stance, a tariff, is not merely the conventional net change in the consumer and producer surplus triangles, but also the spending by lobbies for and against the tariff and by those who aim to capture the rent for themselves. Thus in figure 1 the lobbying costs shift the production possibility (PP) curve inward, with an associated welfare cost of protection CB.4 If there is a struggle over the rents associated with the tariff, the welfare costs rise to DB, as the PP curve shifts further inward (see Srinivasan 1986b). One study by Grais and others (1984) has attempted to estimate the costs Deepak Lal and Sarath Rajapatirana 191 of rent seeking (but not lobbying) associated with quotas in Turkey. It found that if tariffs were removed but quotas maintained, there was little effect on real GDP; if quotas were also eliminated, real GDP rose between 5 and 10 percent. So far we have excluded the deadweight losses of domestic mono- poly and X-inefficiency associated with protection (see Leibenstein 1966, Corden 1974, Krueger 1984). Thus "a reduction in tariff levels might be expected to result in a downward shift in industry supply curves. The welfare costs of protection would then consist of the conventional production cost, plus an inefficiency cost and possibly a monopoly cost" (Krueger 1984, p. 544). The only attempt to measure all these costs is by Bergsman (1974). He found that the costs of protection as a proportion of GNP, consisting of the conventional allocative inefficiency costs (A), the X-inefficiency combined with monopoly costs (X), and the total costs (T=X+A) were as follows for four developing countries: A X T Brazil 0.3 6.8 7.1 Mexico 0.3 2.2 2.5 Pakistan 0.5 5.4 5.9 Philippines 1.0 2.6 3.6 The static welfare gains from trade liberalization could therefore be quite substantial. But these would still affect only the level of income, not its rate of growth. They do not explain why the growth rates of free-trading countries should be higher on a sustainable basis, as seems to have been the case. Statistical The links between trade and growth have many statistical studies, Tests which are summarized in tables 1 and 2. Michalopoulos and Jay of Exports (1973) estimated an aggregate neoclassical production function for and Growth thirty-nine countries. Exports were found to be highly significant, and GNP growth was significantly correlated to the growth rate of exports. By studying the change in the proportion of exports to GNP relative to the rate of GNP growth in forty-one countries during 1950-73, Michaely (1977) found a significant relationship at the 1 percent level for the Spearman rank correlation. The study attempted to avoid autocorrelation between exports and GNP by using the change in the share of exports in GNP to represent the growth of exports which was then regressed against the rate of change of per capita income (Mi- chaely 1977, p. 50). In her study for the National Bureau ol Economic Research (NBER) on foreign trade regimes and economic development, Krueger (1978) 192 Research Observer 2, no. 2 (July 1987) Table 1. Estimated Spearman Rank Correlation Coefficients between Export Growth and Output Growth in Developing Countries Average annual change in export Incremental export Increments in export GNP ratio vs. average Export growth GNP ratio GNP ratios annual per capita GNP vs. vs. vs. Author growth GNPgrowth GNPgrowth GNPgrowth Michaely For 41 observations, 1950-73 0.380` - - For 23 middle-income countries, 1950-73 0.523* - - For 18 low-income countries, 1950-73 -0.04 Balassa: for 11 semi-industrial countries, 1960-73 - 0.888*- 0.813'- 0.776** * *indicates 1 percent level of significance. Source: Michaely 1977, pp. 51-52, and Balassa 1978, p. 184. regressed GNP growth for each of ten countries against the rate of export growth. She found a positive and significant relationship be- tween the two. Similarly, Balassa (1978), by reestimating Michaely's equations and incorporating the Michalopoulos-Jay factors, noted a robust relationship between exports and GNP growth for eleven coun- tries. He recognized that it understated the effects of export growth. And Feder (1983) not only found a positive correlation between ex- ports and GNP growth, but also provided evidence to support the hypothesis that export-oriented policies both led the economy to an optimal allocation of resources and generally enhanced productivity.5 All these studies confirm a statistical relationship between export and income growth. But at best this provides a stylized fact, not a theory. As in most statistical matters however, even this association is disputed-a reflection of the emerging "law" that all econometric evidence is equivocal. Thus Helliner (1986) in a study of low-income countries heavily weighted toward Sub-Saharan Africa, concluded that the results for 1960-80 "show no statistically significant link between the change in export share of GDP and growth. Indeed, the sign on this relationship is consistently negative" (p. 146). Similar results were also reported by Michaely (1977): "the positive association of the econo- my's growth rate with the growth of the export share appears to be particularly strong among the more developed countries, and not to exist at all among the least developed.. . This seems to indicate that growth is affected by export performance only once countries achieve some minimum level of development" (p. 52). Deepak Lal and Sarath Rajapatirana 193 Table 2. Estimated Relationship between Export Grovwth and Output Growth in Developing Countries Dependent Independent va;viablesb variable, a 1AY\ AKd AKf AL AX AX AX Author RI1) Y1 Y1 L1 AY Xl Y1 Dilt Di2t R2 Michalopolous and - 0.25 0.20 0.66 - - - - - 0.53 Jay: for 39 (7.81) (3.35) (2.44) observations, - 0.24 0.12 0.60 0.04 - - - - 0.71 1960-66 (9.62) (2.33) (2.81) (4.82) Balassa: for 10 - 0.18 0.30 1.09 - - - - - 0.58 observations, (3.23) (2.42) (1.74) 1960-73 - 0.15 0.23 0.97 0.04 - - - - 0.77 (3.33) (2.40) (1.99) (3.57) Feder: for 31 - 0.284 0.739 - - - - - 0.37 observations, (4.311) (1-990) 1964-73' - 0.178 0.747 - - 0.422 - - (3.542) (2.862) (5.454) 0.689 - 0.124 0.696 - 0.131 0.305 - - 0.809 (3.009) (3.399) (4.239) (4.571) Krueger: for 10 - - - - 0.11 - 0.08 0.16 0.99 observations, (4.29) (0.85) (1.70) 1950-70d Note: Numbers in parentheses are t values. a. The dependent variable (AY/ Yj) is the GNP growth in Michalopoulos and Jay, Balassa, and Krueger. In the Feder study, it refers to GDP growth. b. The independent variables are AKdl Y1, domestically financed investment as a proportion of the GNP in the initial period; AKf/ Y1, foreign-financed investment as a proportion of GNP in the initial period; AL/L1, increase in population as a proportion of the population in the initial period; AX/AY, incremental export-GNP ratio; AX! XI, increase in exports as a proportion of the exports in the initial period; AX/ Y1, increase in exports as a proportion of GNP in the initial period; Di, a dummy variable which takes the value of 1 during Phases 1 and 2 of the trade regimes; Di2, a dummy variable which takes the value of 1 during Phases 3 and 4 of trade regimes; and t which is a time variable. c. The export variables in Feder are (AX/X) (XI Y) = (AX/ Y) and (AX/X). d. In Krueger's study log refers to a pooled sample of ten countries in which log GNP is regressed on a time trend and log X, a dummy variable for the country's Phases 1 and 2 trade regime (that is, Di,t), and another dummy variable for Phases 3 and 4 trade regime. Source: Michalopoulos and Jay 1973; Balassa 1978, p. 186; Feder 1983, pp. 65, 68; Krueger 1978, p. 273. As most of the low-income countries in the sample used by Helliner (using data from the World Development Reports6) can hardly be classified as having followed outward-oriented policies, the failure to find a link between exports and growth is not surprising. We would expect, however, that in the turbulent decade after the first oil shock of 1973, even among these dirigiste low-income countries, the rela- tively more outward-oriented would have had a better growth record. Using data from the World Development Report 1986, on GDP, labor force growth, and the growth of export and investment shares in GDP, 194 Research Observer 2, no. 2 (July 1987) we estimated the following regression for eighteen low-income coun- tries for which statistics were available: GDP = 2.251 + 0.225 Investment - 0.963 Growth + 0.1S2 Growth of growth share of of labor export share rate, GDP, force, of GDP, 1973-84 1984 1973-84 1965-73 (0.064) (0.786) (0.079) r' = 0.30; F = 3.2 The figures in brackets are standard errors. This equation shows that there was a positive and statistically significant association be- tween income growth rates in the turbulent decade after the oil shock and the growth of exports in GDP in the preceding period (1965-73). All the studies reviewed above use conventional statistical tests for establishing an association between exports and growth. Following classical statistical methodology, these correlations by themselves re- veal nothing about causation. To make causal inferences, an underly- ing theoretical model is required, whose validity is then tested by standard econometric techniques. Recently, however, a new school of econometrics has sought to make causal statements purely on the basis of a particular statistical technique called vector auto regression (VAR) and of a Granger-Sims causality test, which seeks to establish whether over time a particular variable regularly precedes another. Jung and Marshall (1985) have applied the Granger causality test to data for thirty-seven developing countries in 1950-81 to determine whether exports "Granger-cause" growth, or vice versa. They find that only Costa Rica, Ecuador, Egypt, and Indonesia provide evidence in favor of export promotion; "more interestingly, many of the coun- tries most famous for the miraculous growth rates that appeared to arise from export promotion policies (e.g. Korea, Taiwan, Brazil) provide no statistical support for the export promotion hypothesis" (p. 10). Darrat (1986) has also applied the Grangercausality test to the time series for exports and growth between 1960 and 1982 for Hong Kong, Korea, Singapore, and Taiwan and finds that for the first three "neither exports cause economic growth nor economic growth causes exports." For Taiwan he finds that "economic growth unidirectionally causes exports" (p. 697). Several points need to be made against this recent counterrevolu- tion. First, even within the atheoretical Grangercausality framework, the results showing that output growth causes export growth are not inconsistent with the export-growth link found by the more conven- tional studies. Consider this comment by Darrat: "The economic Deepak Lal and Sarath Rajapatirana 195 growth that Taiwan enjoyed during the estimated period (1960-82) appears to be an internal process perhaps diue to domestic technologi- cal advancement and enhanced accumulation of human capital (Jung and Marshall 1985). Given the country'; limited market capacity, Taiwan's producers were probably compelled to turn to foreign mar- kets for exports. It seems therefore, that economic growth (generated internally) has caused higher exports in Taiwan, contrary to the impli- cation of the export-led growth hypothesis" (p. 697-98, emphasis added). However, it is obvious that, if a small country is developing efficiently in line with its comparative advantage, it will specialize and hence be "compelled to turn to foreign markets for exports" of goods that use its most abundant factor of production most intensively. The statistical establishment of this fact hardly disproves the validity of the outward-oriented development strategy; in fact, it supports it. Thus in the Jung and Marshall (1985) study if the cases in which output growth causes export growth are also included as supporting the outward orientation theory (as they should be on the above argument), the list of countries rises to fourteen and also includes Bolivia, Greece, Iran, Israel, Kenya, Korea, Pakistan, Peru, South Afri- ca, and Thailand. Second, it is clear from the test of precedence (which is what the the Grangercausality test amounts to) that the statistical counterrevo- lutionaries are testing for the growth-enhancing effects of a develop- ment strategy that biases incentives toward exports. They are not concerned with the neutral free trade equilibrium point-F in figure 1-whereas (as we argue below) most proponents of outward orienta- tion do not favor this biased export-led growth. The same criticism applies to Fishlow's (1985) interpretation of the case for outward orientation as identical with that for export-led growth. To test this hypothesis, he rightly argues, "requires calculation of the relationship between aggregate performance and the extent to which the rate of growth of exports exceeds overall growth"; not surprisingly, he finds that "with such a specification, there is no statistically significant relationship" (p. 139). But again, it is clearly unwarranted to identify the case for outward orientation (or, more precisely, for neutral trade policies) with that for an export bias. Third, the statistical studies based on Granger causality, which test for the precedence of one variable over another, do not in fact reveal anything about causation as the term is normally understood. Leamer (1985), in a review of the new econometric fashion that uses vector autoregressions alone for causal inferences, rightly notes that: "this concept should be called 'precedence'. . . We can all think of contexts in which precedence is suggestive of causation and also contexts in which it is not ... It is altogether clear that precedence is not suf- ficient for causality. Weather forecasts regularly precede the weather, 196 Research Observer 2, no. 2 (July 1987) but few of us take this as evidence that the forecasts 'cause' the weather" (pp. 259, 283).7 There is one other, more compelling reason why judgments based solely on statistical tests (both conventional and novel) of dynamic effects of trade regimes must remain inconclusive. Economics, as Hicks (1979) has put it, "is on the edge of science and of history" (p. 38). The historical aspects are particularly important for what he terms "sequen- tial causality"-which is the relevant notion of causality for analyzing the dynamic effects of trade regimes. But in studying such dynamic historial processes, techniques of statistical inference may not be very useful, because "when we cannot accept that the observations along the time series available to us are independent or cannot by some device be divided into groups that can be treated as independent, we get into much deeper water. For we have then, in strict logic, no more than one observation, all of the separate items having to be taken together. We are left to use our judgment, making sense of what has happened as best we can, in the manner of the historian. Applied economics does then come back to history after all" (p. 126).8 This section looks at five comparative studies of particular develop- Evidence ing countries' trade regimes undertaken in the 1960s and 1970s (Little on Trade and others 1970, Balassa 1971, Donges 1976, Bhagwati 1978, and Krue- Regimes ger 1978). The studies provide fairly firm evidence that countries that adopted or moved toward an export-promoting (EP) strategy did much better in growth of per capita income and equity than those with an import-substituting (is) strategy. These terms, EP and is, have caused some confusion. The most common definition now is that a movement from the neutral free trade position is is, and a movement toward it (that is, from P to F in figure 1) is EP. Thus the EP strategy does not imply any subsidization of exports beyond the level that restores equali- ty between the effective exchange rates on imports and exports.9 For our purpose we need only note that the five comparative studies have established that is regimes produce a misallocation of resources. Although there are analytical doubts about the use of domestic re- source cost (DRC) measures as indicators of static efficiency in some of these studies, the general conclusion is reinforced by more appropriate indicators of allocative efficiency: the divergences between Little- Mirrlees (LM) shadow prices and market prices for several developing countries.'" More significantly, these studies (in particular the NBER study) showed that countries that reduced or removed the bias against exports had accelerated their growth rates of per capita incomes; those with an is strategy did not. In this context it is important to distinguish between the degree and pattern of protection. It has been argued that the existence of some Deepak Lal and Sarath Rajapatirana 197 highly protected industries in an economy whose trade regime on aver- age shows little is bias (for example, Korea) invalidates drawing any inferences from its experience in favor of neutral trade (see Wade 1985). Jagdish Bhagwati (1986) has given the correct response to this argument: Thus, within the broad aggregates of an EP country case, there may well be activities that are being import-substituted (i.e., their EERm exceeds the average EERX) [where EERlm is the effective exchange rate for imports, and EERX is the effective exchange rate for exports]. Indeed there often are. But one should not jump to the erroneous conclusion that there is therefore no way to think of EP versus IS and that the distinction is an artificial one-any more than one would refuse to acknowledge that the Sahara is a desert, whereas Sri Lanka is not, simply because there are oases (p. 93). Tables 3 and 4 summarize the divergence between market and LM shadow prices for traded goods in India and Korea.'" Though there are highly protected activities in both countries, even casual inspec- tion shows that Korea's trade regime is much more neutral than India's, and the dispersion of its protection is lower.'" Of the five studies, only the one from the NBER explicitly sought to quantify the possible effects of alternative trade regimes on savings rates, technical progress, and entrepreneurship-the dynamic factors that affect a country's growth rate. The evidence (surveyed in Bhagwati 1978) on entrepreneurship, innovation, and techniical change is inconclusive, though none of these factors is shown to benefit from is regimes. On savings, Bhagwati concludes that the evidence does not support the view that restrictionist exchange control regimes "will or are likely to contri- bute to increased domestic savings, and/or to augmented capital forma- tion. If anything, much of our evidence-at least on the domestic savings issue-suggests an opposite relationship" (p. 174). The NBER study also emphasized the importance of appropriate macroeconomic and exchange rate policies to maintain a realistic real exchange rate. As Krueger (1978) put it: It seems a fair conclusion that one of the policy mistakes of the two decades covered by the country studies was using devaluation to a new fixed exchange rate as an instrumtent designed to attain both domestic price stabilisation and a liberalised trade regime (p. 297). Exogenous The 1970s and 1980s produced two oil shocks, the worst recession Shocks since the Great Depression, and a huge switch in real interest rates in the 1970s (low or negative for most of the 1970s, unprecedentedly high in the and 1980s 1980s). These shocks were common to all developing countries, yet the relative performance of the EP countries was far superior (see Lal and Wolf 1986). 198 Research Observer 2, no. 2 (July 1987) Table 3. Accounting Ratios for Traded Commodities in India, 1973 Sectoral code Accounting number Commodity ratio T1 Electrical equipment 0.36 T2 Nonelectrical equipment 0.65 T3 Transport equipment 1.28 T4 Metal products 0.29 T5 Iron and steel 1.00 Pipes and tubes 1.00 Pig iron 1.00 T6 Cement 0.66 T7 Nonferrous metals 0.59 T8 Other minerals 0.61 T9 Rubber 0.60 T10 Leather 0.50 T11 Other leather products 0.50 T12 Leather footwear 0.81 T13 Animal husbandry 0.37 T14 Sugar 0.52 T15 Gur and khandsari 1.08 T16 Vegetable oils 1.14 T17 Vanaspati 0.65 118 Starch 0.87 T19 Milk products 0.29 120 Breweries and soft drinks 0.68 121 Confectionery 0.60 122 Cigarettes and cigars 0.39 T23 Other tobacco products 0.39 T24 Fruits and vegetables 0.32 T25 Cashew nut processing 0.27 T26 Cotton 0.51 T27 Cotton yarn 2.04 T28 Cotton textiles 0.46 T29 Jute 0.57 T30 Jute textiles 0.44 T31 Woolen yarn 0.60 T32 Woolen textiles 0.61 T33 Raw silk 0.71 T34 Silk textiles 0.50 T35 Man-made fiber (rayon) 0.13 T36 Artificial silk 0.43 T37 Other textiles 0.44 T38 Tobacco 0.43 T39 Fertilizers 1.00 T40 Ceramics and bricks 0.44 T41 Glass and glassware 0.72 T42 Wood products 0.97 T43 Timber 0.80 T44 Chinaware, pottery 0.50 T45 Wood, others 0.56 T46 Other forest products 0.27 T47 Petroleum products 0.65 T48 Rubber footwear 0.73 (Table continues on next page) Deepak Lal and Sarath Rajapatirana 199 Table 3 (continued) Sectoral code Accounting number Commodity ratio T49 Synthetic rubber 0.73 T50 Other rubber products 0.48 T51 Paper and paper products 0.44 T52 Plastics 0.47 T53 Dyestuff 0.39 T54 Paints and varnishes 1.35 T55 Insecticides and pesticides 0.91 T56 Drugs and pharmaceuticals 0.32 T57 Soaps and glycerine 0.57 T58 Perfumes and cosmetics 0.39 T59 Miscellaneous chemicals 0.53 T60 Coal and coke 0.72 T61 Matches 0.76 T62 Plantations 1.00 T63 Aluminum primary product 0.80 T64 Zinc 0.61 T65 Lead 0.58 T66 Tin 0.57 T67 Manganese 1.00 T68 Sulfur 0.65 T69 Sulfuric acid 0.65 T70 Rock phosphate 0.87 T71 Salt 1.00 T72 Wheat 0.87 T73 Soda ash 0.76 T74 Dry cells 0.52 T75 Ball bearings 0.40 T76 Electric fans 1.00 T77 Radio receivers 0.52 T78 Nonferrous metal alloys 0.69 179 By-products of foodgrains 0.87 T80 Gypsum 1.00 T81 Limestone 1.00 T82 Iron ore 1.00 T83 Bauxite 1.00 Note: The accounting ratio is the ratio of the social price to the market price; see note 11 to the text for a fuller definition. The unweighted means of these accounting ratios is 0.675, with a standard deviation of 0.294. Source: Lal 1980. The starting point for explaining this conclusion is a simple one: all countries need some foreign trade. In dealing with external shocks, the more inward-looking countries face greater costs of output for- gone through compressing imports (which are mainly capital goods), and they have more difficulty in expanding exports because of a smaller proportion of their output is tradable. For these reasons, is countries have not only had slower growth but also more serious debt 200 Research Observer 2, no. 2 (July 1987) Table 4. Accounting Ratios for Traded Commodities in Korea, 1973 Sector Accounting number Commodity ratio 1 Rice, barley, and wheat 0.84 2 Vegetables, fruits 0.79 3 Industrial crops 0.84 4 Livestock breeding 0.87 5 Forestry products 0.88 *6 Fishery products 1.27 7 Coal 0.85 8 Metallic ores 0.95 9 Nonmetallic minerals 0.87 10 Slaughtering, dairy products 0.85 11 Canning and processing 0.81 12 Grain polishing and milling 0.81 13 Other food preparations 0.80 14 Beverages 0.47 *15 Tobacco 0.95 ; 16 Fiber spinning 1.16 *17 Textile fabrics 1.08 * 18 Apparel and fabrications 1.08 *19 Leather and leather products 1.04 *20 Lumber and plywood 1.03 *21 Wood products and furniture 1.00 22 Paper and paper products 0.67 23 Printing and publishing 0.93 24 Inorganic chemicals 0.77 25 Organic chemicals 0.68 26 Chemical fertilizers 0.93 27 Drugs and cosmetics 0.80 28 Other chemical products 0.68 29 Petroleum refining 0.80 30 Coal products 0.91 31 Rubber products 0.69 32 Nonmetallic minerals 0.77 33 Iron and steel 0.86 34 Primary iron and steel manufactures 0.76 35 Nonferrous metal manufactures 0.79 *36 Fabricated metal products 1.05 37 Nonelectrical machinery 0.84 38 Electrical machinery 0.76 39 Transportation equipment 0.83 40 Measuring, medical 0.67 *41 Miscellaneous manufactures 1.33 *56 Unclassifiable 1.33 Note: These ratios have been derived as described in Lal (1978b). The asterisked items were taken to be export sectors. The unweighted mean of these ratios is 0.883 with a standard deviation of 0.178. problems than EP countries: witness the contrast between the newly industrializing countries (NICs) in Southeast Asia and the Latin Ameri- can Southern Cone countries. Deepak Lal and Sarath Rajapatirana 201 For comparative purposes, three groups of economies have been selected. Group A consists of Hong Kong, Korea, Singapore, and Taiwan, which have followed EP strategies as defined in the Bhagwati- Krueger studies.13 Group B includes the Southern Cone countries- Argentina, Chile, Uruguay-plus a South Asian country, Sri Lanka. They are referred to as moderately is countries, which made some effort to liberalize their trade regimes in the late 1970s.14 Group C consists just of India, an is country in which the effective exchange rate for imports markedly exceeds that for exports. Group A: Export-Promoting Countries The four economies in this group have been the most dynamic exporters in the world. During 1970-79. their exports grew at an annual average rate of 25 percent, with manufactured exports grow- ing at 30 percent. During 1978-81 their exports grew at 19 percent a year; then, with the world recession of the early 1980s, they actually declined. But they recovered more rapidly than the exports of any other group, despite experiencing greater external shocks than most other developing countries. The shocks were equal to 18 percent of GNP for Singapore, 10 percent for Taiwan, and 9 percent for Korea; for developing countries as a whole, the shock was 6 percent of GNP (Balassa 1984). By 1983, the GDP growth rates of all four economies were back to the 1970-82 average; so were their trade balances and export growth. The four EP economies adjusted in two ways. First, they expanded exports by raising their market shares even when world demand was depressed. Second, they raised domestic savings. Korea was an excep- tion to this rule, because its financial market was repressed. It fi- nanced part of the temporary loss in inconme by increasing its external borrowings-but, unlike most heavily indebted countries, it was able to service its debt without cuts in domestic output."5 Group B: Moderately Import-Substituting Economies The countries in this group tried to liberalize their trade regimes between the mid-1970s and the early 1980s. They are the Southern Cone countries of Argentina, Chile, and Uruguay and one South Asian country, Sri Lanka. All were initially successful in liberalizing their trade regimes. They speeded up their GDP and export growth and improved their external accounts. In the 1"80s, however, they ran into crises of macroeconomic stabilization. The Southern Cone countries shared a common path up to the mid- 1970s. By that stage, Argentina's GDP growth was slow (averaging 0.5 202 Reseazrch Observer 2, no. 2 (July 1987) percent a year during 1965-73), and its trade heavily protected. Infla- tion was high-around 180 percent in 1975-so was unemployment. Effective rates of protection ranged from 111 percent for manufactur- ing to -13 percent for agriculture. Since the 1930s it had followed an is strategy. For Chile, too, the initial conditions were very difficult. GDP shrunk by 5.6 percent in 1973 on the eve of the reforms. Inflation had reached 1,000 percent. In trade, rates of protection averaged 217 percent, varying from 1,140 percent for petroleum and coal products to -7 percent for agriculture. Uruguay had experienced prolonged stagnation during 1950-70. In 1965-73, GDP growth averaged 2.0 percent a year. Inflation had reached 97 percent, capital flight was substantial, and the currency was overvalued. By 1970 its is strategy had hit the limits of the small domestic market. All three countries undertook substantial economic reforms during 1975-80, of which a principal feature was trade liberalization. In Argentina, taxes on exports were reduced;16 so were import tariffs, thus lowering both the average level of effective protection and the variance. Following these reforms GDP growth averaged nearly 4 per- cent a year between 1978 and 1980. Manufactured exports increased by 216 percent in 1975-80. In 1978 the government started to prean- nounce exchange rates (the tablita). But it had failed to curb the fiscal deficit. The ensuing domestic inflation, coupled with a slowly adjust- ing nominal exchange rate, led to a rise in the real exchange rate, capital flight, and the collapse of the banking system (Calvo 1986). Because of this stabilization crisis, by 1982 the trade reforms had been reversed. The Chilean trade reforms were the most far-reaching. All quotas were eliminated except those on motor vehicles. All tariffs were re- duced to a uniform 10 percent by 1979. By 1977-78, the budget was balanced. The exchange rate was first put on a crawling peg, and then in 1979 fixed against the U.S. dollar. Also by 1979 the capital account was liberalized. The results of these reforms were dramatic. During 1976-81 GDP grew by 8 percent a year and manufactured exports by 30 percent a year. However, the opening of the capital account and the fixing of the nominal exchange rate led to a large appreciation in the real exchange rate. As in Argentina, this was followed by a balance of payments crisis, capital flight, and the collapse of the domestic banking system. In 1982 GDP declined by 14 percent, and unemployment reached 22 percent of the labor force. In Uruguay, taxes on traditional exports were lowered in the mid- 1970s. Nontraditional exports were given additional incentives. Price controls were reduced sharply, and restrictions on the inflow of pri- vate capital were eliminated. The results of these reforms were im- Deepak Lal and Sarath Rajapatirana 203 pressive. Having averaged 1.0 percent a year in 1955-73, GDP growth rose to an average of 4.5 percent during 1974-80. In 1982, however, the worldwide rise in interest rates, combined with an appreciation in the real exchange rate (because of its tablita) and an increase in the fiscal deficit, led to a balance of payments crisis. GDP declined by nearly 10 percent in 1982. The Southern Cone experience provicles several lessons. The trade reforms were successful in raising export and GDP growth, particularly in Chile and Uruguay. But the overall attempt at economic liberaliza- tion failed eventually, because of unsustainable macroeconomic and real exchange rate policies (Corbo and de Melo 1985, de Melo and Tybout 1986). Compared with the Southeast Asian NICS, the Southern Cone countries faced more difficult conditions initially and made errors in macropolicy when it came to dealing with external shocks. In Sri Lanka, the trade liberalization of 1977 reversed a long- standing is strategy. By that stage, the economy was highly distort- ed-but not subject to the inflationary instability of the Southern Cone. For many years the economy was sluggish: in 1965-77, for example, GDP growth averaged 2.9 percent a year. The country's share of world trade declined, as did its volurne of exports-by 1.5 percent between 1970 and 1977. The wide-ranging reforms of 1977 mainly involved the trade regime and led to a dramatic economic recoverv. Sri Lanka's GDP grew by 6 percent a year in 1978-85, and unemployment fell from 24 percent of the labor force in 1973 to 12 percent in 1981 (Bhalla and Glewwe 1986). By 1983, however, macroeconomic imbalances forced the gov- ernment to slow down its reform program. Heavy public expenditure and a reluctance to close loss-making public enterprises proved to be incompatible with the trade reforms (Lal and Rajapatirana 1987). As in the Southern Cone, macroeconomic imbalances-in this case brought about by the financing of public expenditure by capital in- flows-led to a real appreciation of the rupee and to a balance of payments and fiscal crisis. The experience of this group of countries points to three lessons for trade liberalization. First, appropriate macroeconomic policies are vi- tal to maintaining a more liberal trade regime. Second, an appropriate real exchange rate plays a bigger role in trade reform than was realized in the five comparative studies. Third, the order in which the various repressed markets are reformed seems to be important. In both Chile and Uruguay the liberalization of financial markets was destabilizing. It is still an open question whether this was due to the structure of the domestic financial market, or to poor macroeconomic and exchange rate management, or is an inherent property of financial liberalization itself. 204 Research Observer 2, no. 2 (July 1987) Group C: Import-Substituting Countries India's relative immunity to the external shocks of the 1980s is often used to suggest that, over the long haul and despite the ac- knowledged productive inefficiencies of the country's trade regime, India's "delinking" from the world economy has allowed it to main- tain much steadier growth in the past two turbulent decades."' This view is mistaken, because any stability of domestic incomes achieved by delinking can also mean that average income is lower than if the international roller coaster is ridden efficiently. India proves this point. Its trend rate of GDP growth has been 3-4 percent a year for three decades-much lower than the growth that could have been obtained by integration into a world economy, which boomed for two decades. To illustrate this loss, in 1960 the absolute size of Korea's manufacturing industry was a quarter of India's; in 1980 it was almost two-thirds of the size. Korean manufactured ex- ports rose from virtually nothing in 1960 to more than $15 billion in 1980. In the same period India's manufactured exports rose from $600 million to only $4.1 billion. "Even tiny Singapore has managed to export more manufactures in value terms ($11.7 billion in 1981) than India! India's share in world exports has declined steadily from 2.4 percent in 1968 to a minuscule 0.41 percent in 1981" (Srinivasan 1986). Since labor-intensive manufacturing is a major means of pro- viding employment and alleviating poverty in countries with a rapidly growing labor force and a scarcity of land, India's inward-looking policies have done long-term damage to both growth and equity. Though it is now fashionable to castigate Latin American countries for having followed "debt-led" growth in the late 1960s and 1970s, the subsequent income losses they may have suffered to service their debts have to be set off against the enormous previous gains in real incomes that debt-financed growth entailed (see Bhagwati 1986). Two of the most influential development economists, Nurkse (1961) Income Effects and Lewis (1980) have adduced a link between trade and growth in of Trade terms of the transmission of rapidly growing world demand to an and Growth open developing country: "trade as the engine of growth.""8 Nurkse (1961) argued that in the nineteenth century, trade had been such an engine for white settler communities, but predicted that it would no longer be so for developing countries in the second half of the twen- tieth century. Nurkse's historical analysis and his forecasts both proved false. The view that international trade assisted the growth of the countries of new settlement in the nineteenth century has been questioned by Kravis (1970). Essentially, he argues that economic growth is deter- Deepak Lal and Sarath Rajapatirana 205 mined by internal factors. Foreign trade provides an extension of the domestic opportunities available for converting resources into goods and services. Furthermore, by widening the market, it enables a coun- try to produce goods with decreasing costs of production. Probably most important of all, exposure to international competition is the best antimonopoly policy-and thereby prevents the development of high-cost industries. Most of these benefits concern the efficient use of available re- sources and hence the supply side of an eccnomy. The demand factors that preoccupied Nurkse and others cannot be as influential, because the countries that shared in the nineteenth century expansion of trade developed in such different ways. For instance, Australia seemed to develop whereas Argentina did not, despite similar natural resource, "white" populations (Argentina had none of the problems of assimila- tion posed for other countries in Latin America by an indigenous population), and a similar stimulus from the rise in foreign demand for their primary products. Thus, as Kravis emphasizes, though strong external demand for a country's exports may be helpful, it is neither a necessary nor sufficient condition for growth or even trade to play a helpful role in growth. ... The term "engine of growth" is not generally descriptive and involves expectations which cannot be fulfilled by trade alone; the term "handmaiden of growth" better conveys the role that trade can play (p. 869). More recently, Lewis (1980) has presented another model in which trade serves as an engine of growth. He bases his theory on the following empirical regularity: The growth rate of world trade in primary products over the period 1873 to 1913 was 0.87 times the growth rtate of industrial production in the developed countries; and just about the same relationship, about 0.87, also ruled in the two decades to 1973. World trade in primary products is a wider concept than exports from developing countries, but the two are sufficiently closely related for it to serve as a proxy. We need nro elaborate statistical proof that trade depends on prosperity in the industrial countries (p. 556, emphasis added). The italicized words contain by no means an innocuous assump- tion: whereas manufactures accounted for only 10 percent of develop- ing countries' nonfuel exports in 1955, their share had risen to over 40 percent by 1978. Primary product exports can no longer serve as a proxy for developing country exports, as Lewis asserts. Nor, except for Sub-Saharan Africa, does the picture change much when the fig- ures are disaggregated. There are big differences in the export struc- tures of developing countries: manufactures now account for 75 per- cent of the exports of the four East Asian superperformers, for exam- 206 Research Observer 2, no. 2 (July 1987) ple. But most countries in South Asia plus Brazil, Egypt, Mexico, Tunisia, and some smaller Latin American countries (together ac- counting for about two-thirds of the population of the developing world) have also raised the share of manufactures in their exports (on a trade weighted basis) from an average of 15 percent in 1950 to above 50 percent in 1978. Disaggregation also greatly weakens Lewis's link coefficient of 0.87 between the rate of growth of industrial production in industrial countries and developing country exports (see Riedel 1984). Broadly speaking, Lewis's hypothesized link is unstable over time, and the only commodities to which it seems to apply are tea and sugar. For manufactures, the dominant and growing element in developing coun- try exports, Riedel concludes that "the evidence ... suggests that supply rather than demand factors have principally determined LDC export performance in manufactures." This is also the conclusion of the numerous historical studies of the trade and industrialization poli- cies of developing countries cited earlier (Little and others 1970, Ba- lassa 1971 and 1982, Bhagwati 1978, and Krueger 1978). Despite creeping protectionism and the slowdown in industrial countries, Rei- del noted that whereas in the 1960s LDC exports of manufactures grew almost twice as fast as DC [industrial country] real GDP.. . in the 1970s, despite a general slowdown of growth after 1973, LDC exports maintained their rapid pace, growing four times as fast as DC real GDP. (p. 67). Thus the view of trade as the engine of growth cannot adequately explain the link between neutral or liberal trade regimes and growth. Other studies (Streeten 1982, Cline 1982) have argued that there is a fallacy of composition in generalizing the example of the East Asian NICs to the rest of the world. They claim that if all developing countries were to switch to export promotion, the industrial coun- tries would become protectionist in an effort to stave off a surge in developing country exports. Ranis (1985) provides a thorough cri- tique of these views (see also Havrylyshyn 1987). Ranis writes that, with the adoption of an EP strategy, "the much more substantial growth of per capita income resulting in the exporting countries would enable them to increase their imports from the North as well as each other" (p. 544). Unless developing countries were to run continual and massive trade surpluses, the industrial countries would boost their exports to them-a powerful counterweight to any pro- tectionist lobbies. Despite the fears expressed about the "new pro- tectionism" in industrial countries, the fact is that protectionism has by and large been kept at bay during the deepest recession since the 1930s. Deepak Lai and Sarath Rajapatirana 207 Level It seems to be as firm a stylized fact as any in the economics of and Growth developing countries: a sustained movement to an outward-oriented Effects trade regime leads to faster growth of both exports and income. How can this be explained? Krueger (1978) argues that mere neutrality of the trade regime is not enough: There are numerous countries where incentives for export and import substitution have been about equal, and the results have not been spectacular . .. Although econoniic theory suggests that incen- tives for exports and for import substitution should be equated at the margin, in fact neither Brazil nor South Korea did so; during the rapid growth years [Korea after 1964 and Brazil after 1968] the bias in their regimes was toward exports (p. 282-83). Krueger then compares two activist policies to encourage growth ("the alternative of a strictly laissez-faire regime is not explored") and argues that "a growth strategy oriented toward exports entails the development of policies that make markets and incentives function better, while an import-substitution strategy usually involves policies designed to frustrate individuals' maximizing behaviour under market incentives" (p. 284). These statements have misled some economists (see Streeten 1982 and the riposte by Henderson 1982) to suggest that an outward- oriented strategy necesssarily involves an export bias. Of course, ex- port promotion can be as inefficient and chaotic as protection, as India has shown (Lal 1979c). The liberal position on trade and growth (which we support) is different. As a first step it entails a neutral trade regime. The fact that such a regime does not necessarily lead to growth merely underlines the Kravis view of "trade as the handmai- den of growth," which sees internal factors as the biggest determi- nants of growth, with trade a helpful though not dominating influ- ence. However, a liberal trade regime (and an export-biased one) can help more directly than Kravis allowed to create a domestic economic system conducive to growth. This was one of Tumlir's insights devel- oped in a series of papers in the late 1970s and early 1980s but unfortunately not consolidated into the book he was planning to write when his life was so tragically cut short. Analytically, he accepted that the case for government intervention in foreign trade can be separated from that in the domestic economy-so that, whatever view one takes of the latter, the former is unjustifiable (except for the optimal tariff case). But he argued that the analytical separation in this "man- agement economics for governments" was misguided (Tumlir 1981). Though it mitigated some of the irrational dirigisme of governments, it nevertheless implied that market failure was ubiquitous. The canons 208 Research Observer 2, no. 2 (July 1987) of second-best welfare economics then allowed benevolent, omnis- cient, and omnipotent governments to intervene in the social interest. There is another, clearer view of government motives and foresight. It is associated with Adam Smith and the classicists, whose modern votaries are the so-called neo-Austrians, and it recognizes the ubiqui- tousness of government failure. The case for a liberal trade regime then becomes part of the general case for markets against mandarins. The ideal balance between the two is discussed in Lal (1986), not in this article. But, if one accepts the need for restraints on the natural and often irrational dirigisme of mandarins in most developing coun- tries, then the adoption of a liberal trade regime (irrespective of the ensuing gains from trade, static and dynamic) becomes an important means to this end. This line of thought can be developed by making use of some ideas attributable to Maurice Scott on investment and growth (see Scott 1976). They also provide an antidote to the criticism by Lucas that static gains from free trade merely affect the level of income, not its rate of growth. In the standard neoclassical growth model (Solow 1956, Swan 1956), the steady state growth of an economy-its "natural rate of growth"-is determined by the exogenously given growth rate of population (n), plus the rate of labor-augmenting (Harrod-neutral) technical progress (t) which falls like manna from heaven. Several authors, starting with Kaldor (1957), Kaldor and Mirrlees (1962), and including Arrow (1962),19 have argued that this exogeneity of produc- tivity growth does not explain one of the mainsprings of economic growth. In various forms they have sought to introduce the rate of investment (which in the neoclassical model only affects the level and not the rate of growth) as an endogenous determinant of technical progress. In the standard neoclassical framework, an improvement in allocative efficiency in economy A compared with (otherwise identi- cal) economy B leads to a higher level of income per person in A; it is as if A had saved more. But income does not grow faster. This result also holds in the Arrow (1962) and Kaldor-Mirrlees (1962) type of growth models, which seek to endogenize technical progress through a technical progress function that assumes a direct link between the growth of capital per person and the rate of labor-augmenting techni- cal progress. However, as Eltis (1973) has argued,20 it is more plausi- ble in endogenizing technical progress to include in the technical progress function the saving ratio instead of the rate of growth of capital. In that case an increase in allocative efficiency, which is equivalent to an increase in savings, will raise the growth rate.2' These ideas have been further developed by Scott (1976). He argues that "investment is ... by definition ... the cost of cbange, and so will cover all activities associated with growth" (p. 317) and that Deepak Lal and Sarath Rajapatirana 209 "growth due to capital and technical progress are both the result of investment" (p. 330) in the sense of "the cost, in terms of consump- tion forgone, of propelling the economy iForward instead of leaving it in a stationary state" (p. 318). "Incurring capital expenditure leads to a rearrangement of the things of this world. It does not lead to there being any more of some substance 'capital'. . . There is then simply change which is due to investment, and to population growth. We cannot separate change which is 'more capital' from change which is 'technical progress.' We must abandon the attempt to distinguish between movements along a production function whose arguments are labor, land and all capital, and a shift in that function due to technical progress" (p. 331). Within his proposed framework, "the rate of increase of static income is a function of only two variables: total savings and labor force growth. There is no independent techni- cal progress" (p. 331). The key aspect of Scott's analysis is its emphasis on "the impor- tance of allocation" for the growth rate. Unlike the conventional framework, which views allocative improvements as providing "a once-and-for-all increase in output and a temporary boost to the growth rate while it is occurring," Scott argues that "if, however, investment is essentially a matter of incurring costs to reallocate re- sources, then the efficiency with which this is done must affect the yield of investment, and so the proportionate rate of growth in the long run. So long as investment is occurrin-g, reallocation is occurring. It is not once-and-for-all, but a continuing process, and, indeed, the principal source of growth in many countries" (p. 332-33). Moreover, argues Scott, "investment at any given time is undertak- en in a state of ignorance about the future. We make changes whose consequences we cannot wholly foresee, and, simultaneously, others are making changes of which we can only become aware after they are made. In the light of these changes we are then in a better position to make the next round of changes." This implies "that there is an externality to investment" (p. 334).2" But "if the externality exists just because we are ignorant of the future effects of investment, it may be impossible to discover very much about the characteristics of investment that produce the externality' (p. 325). This argument needs to be extended. It suggests the importance of an economic environment that is conducive to this ignorance-based, externality-creating form of investment. This is the place for the neo- Austrian insights concerning the role of the entrepreneur in an eco- nomic environment characterized by ignorance (see Lal 1986 for refer- ences and a fuller discussion). The entrepreneur is redundant in neo- classical economics, which assumes an environment of purely actuari- al Knightian risk. But he is at the center of the neo-Austrian stage- creating and searching out investment opportunities and gambling on 210 Research Observer 2, no. 2 (July 1987) the future. Like the speculator and middle man, the entrepreneur is an economic agent who lives by making money out of irreducible Knigh- tian uncertainty. This entrepreneurial function must, for reasons to do with incentives and information, be decentralized. To the extent that an EP strategy has to rely on this entrepreneurial function (as export markets cannot be ensured by local mandarins), it will induce the creation of that economic framework in which Scott's externality- creating investments will lead to faster growth. The case for a free trade regime (or, as a second best, an export biased one) is thus close to that argued by the classical and neo- Austrian economists. As Keynes emphasized, the classical case against mercantilism was not based on laissez-faire, but rather on limiting state action to areas where such action was indispensable. These, broadly speaking, are to provide the public goods essential for the efficient functioning of market processes-law and order, stable money-and those infrastructural activities that affect public goods. The modern variant of the classical case, while accepting the need for an activist state, would seek to limit its activities. The state would be prevented from creating those policy-induced distortions that sup- posedly are there to cure endogenous distortions in the working of the price mechanism but which merely aggravate the level of distortions in the economy. Such distortions have led to large, though unquantifi- able, losses, through diverting energies and resources from productive activities into the wasteful lobbying and rent-seeking activities so common in most developing countries. In this task of confining public action to its proper place, a free trade regime could be an important component. It would help to create an economic framework that provides the necessary incentives for entrepreneurship, productivity, and thrift. In a formal sense, these qualities are only dimly understood by economists. But they are, at bottom, the mainsprings of sustained and sustainable economic growth. This article surveys empirical studies of the static gains from a movement toward free Abstract trade and studies of the dynamic effects of growth in exports on per capita income. It also summarizes comparative studies of the trade regimes of developing countries undertaken in the 1960s to 1970s, which show fairly conclusively that "outward- orientation" is associated with better economic performance. The conclusions of these studies are then tested for the more volatile global environment of the 1970s and 1980s. Various arguments are weighed about the dynamic income effects of the growth in world income and trade on a free-trading country's economic growth rate-the "trade as an engine of growth" view. The closing section introduces insights of the classical writers that have reemerged in the neo-Austrian and the more recent neoclassical "new political economv" schools, which might explain the links between trade and growth perfor- mance. These emphasize the importance of the nonquantifiable aspects of a free trade (as compared with a protectionist) regime in creating a general economic framework conducive to individual entrepreneurship, productivity, and thrift. In this context we argue that free trade is the "handmaiden of growth," as it indirectly constrains the state Deepak Lal and Sarath Rajapatirana 211 from going beyond the bounds of necessary public action for the provision of those domestic public goods that are essential for development. Notes The authors wish to thank Jagdish Bhagwati and Soogol Young for valuable comments on the draft of this article and Sophie Kim, Fayez Omar, and John Wayem for research assistance. This is a revised and considerably expanded version of a paper prepared for the conference on "Free Trade in the World Economy" organized by the Institute for World Economics, Kiel. This article is dedicated to the memory of Jan Tumlir. 1. Thus it is recognised that endogenous domestic distortions may require appro- priate domestic public interventions for their correction; where the country has mono- poly (monopsony) power in trade and can feasibly influence its terms of trade, taxes or subsidies on trade may be justified. 2. See Srinivasan and Whalley (1986) for a comprehensive discussion and evaluation of the principal models in this genre. Also see Srinivasan (1986a, 1986b, and 1987). 3. Also see Krueger (1984), Taylor and Black (1974), and de Melo (1978), for other model-based estimates of these static gains. 4. But see the note to the figure for the decomposition of this loss into that due to the tariff and lobbying. 5. See also Tyler (1981), Heller and Porter (1978), Michaely (1979). 6. The World Bank's World Development Report is published annually by Oxford University Press. 7. Leamer is also quite caustic about Granger's defense that: "provided I define what I personally mean by causation, I can use the term." Leamer rightly castigates Granger for misusing common language and for misleading persuasive definitions. He writes: "I detect a certain lack of concern for the human capital that is invested in our language. If I were to continue in that tradition I would propose that we henceforth refer to this notion of precedence by the word pair: fool's causation. This substitutes a loaded word 'fool' for the neutral 'Granger' just as 'causation' has replaced the neutral 'precedence.' Moreover, 'fool' is decidely simpler than 'Granger"-it contains only four letters, one of which is repeated-and, like 'cause,' it is rather difficult to define precisely. One man's fool is another man's genius. My definition of a 'fool' would be a friend of mine living in San Diego" (Leamer 1985, p. 284). But, as Leamer is at pains to emphasize, this does not mean to imply that the atheoretical statistical technique of vector autoregressions underlying the new econometrics cannot be useful as a descriptive and perhaps a forecasting device. 8. See also McKloskey (1983) who rightly argues that the evidence adduced in support of particular economic propositions must be ecclectic and cannot be confined to the positivist statistical testing that is currently so fashionable. As Hicks argues, "the usefulness of 'statistical' or 'stochastic' methods in economics is a good deal less than is now conventionally supposed... Thus it is not at all sensible to take a small number of observations (sometimes no more than a dozen observations) and to use the rules of probability theory to deduce from them a 'significant' general law. For we are assuming, if we do so, that the variations from one to another of the observations are random, so that if we had a larger sample (as we do not) they would by some averaging tend to disappear. But what nonsense this is when the observations are derived, as not infrequently happens, from different countries, or localities, or industries-entities about which we may well have relevant information, but which we have deliberately decided by our procedure to ignore. By all means let us plot the points on a chart and try to explain them; but it does not help in explaining them to suppress their names. The probability calculus is no excuse for forgetfulness" (p. 122). 9. See Lal (1981), and for an emphatic reassertion of this point see Bhagwati (1986). This point is also emphasized in the last chapter of Bhagwati (1978). 212 Research Observer 2, no. 2 (July 1987) In this context it may also be useful to distinguish a liberal trade regime from a neutral one. A failure to do so has caused some confusion about the nature of what different developing countries have done to liberalize their trade. A neutral trade regime is one where incentives for import substitution do not outweigh those for export promotion; but it does not rule out (essentially offsetting) trade interventions. A liberal trade regime ensures this neutrality of trade incentives, because of the absence of trade intervention. We owe this point to Soogil Young's comment on our paper at the conference. 10. For these LM shadow price estimates see the references in Little and Scott (1976); in addition see Little and others (1979) for Pakistan, Lal (1978a) for the Philippines, Lal (1978b) for Korea, Lal (1979a) for Jamaica, Lal (1980) for India, and Lal (1979c) for Sri Lanka. 11. These LM accounting ratios are the commodity-specific ratio of the domestic to border price (c.i.f./f.o.b.) of the traded good, with the dollar value of the border price converted into local currency at the official exchange rate. The effective exchange rate on exports (imports) is the units of domestic currency that can be obtained for a dollar's worth of exports (imports). Thus the rates are weighted averages of the relevant accounting ratios of traded commodities, where the weights are the actual shares in exports (imports) of the relevant commodities. 12. Wade in a private communication has stated that we have misrepresented his position, which he says is from Wade (1985): "In the comparison between Taiwan and Korea, on the one hand, and India and Latin America, on the other, the first important fact about trade regimes is that the East Asia type is more 'liberal' in the sense that the average level of protection is much lower. But the second important fact, which the neoclassical argument has tended to ignore, is that dispersion around the average is much higher in East Asia, because selective promotion of some industries requires high protection to a small number" (p. 27). However, as can be seen from tables 3 and 4, the dispersion around the average protection is also lower in Korea than India, as measured by the respective standard deviations of the accounting ratios in the two countries. 13. This is in relation to the effective exchange rates for exports and imports. 14. These countries attempted to move from Phase 11 to Phase Ill of the restrictive trade regimes in the Bhagwati-Krueger sense. See Krueger (1978). 15. Balassa's studies (1981, 1982, 1984) of forty-three countries subject to shocks in 1974-76 and 1979-81 confirm the superior response of the NICs to external shocks. 16. The export tax on wheat was reduced from 56 percent to 5 percent, on corn from 46 percent to 16 percent, and for wool from 33 percent to 16 percent from July 1976 to July 1977 (Nogues 1981). 17. The external shock to India was 2.1 percent of GNP in the 1974-78 period. See Balassa (1981). 18. Much of this is based on Lal (1983). 19. See Hahn and Matthews (1965) for a more complete survey. 20. See Dixit (1976, p. 81 and following) for a lucid discussion of the models that seek to endogenize technical progress. See also Hahn and Matthews (1965). 21. Formally in the standard Solow-Swan model, the determinants of the steady state growth rate are: gj= g,=s/v=n + t where g,= growth rate of output g,= growth rate of capital s= savings rate v= capital output ratio n= rate of population growth t= rate of Harrod neutral (labor-augmenting) technical progress. With n + t-the natural rate of growth-determined exogenously, changes in the savings ratio will not effect the steady state growth rate of output or capital, but through Deepak Lal and Sarath Rajapatirana 213 countervailing changes in the capital output ratio (v) merely lead to changes in the levels of capital and output per capita. In the Eltis framework, the exogenously given technical progress term t, is replaced by an endogenously determined term whose argument is savings (s), hence in this alternative framework, gy= g= s/v= n+ 4(s) where (l is an increasing concave function (see Dixit 1986, ch. 4). Actual and "as if" increases in savings because of improved efficiency will now lead to a rise in the growth rates of output and capital (y and k). 22. Lucas (1985) presents a neoclassical model of economic development in which a central element is an externality in human capital investment. Many of his insights would seem to complement those of Scott, except that he draws unwarranted dirigiste implications from them. References Arrow, K. J. 1962. "The Economic Implications of Learning by Doing." Review of Economic Studies 29, no. 3: 155-73. Balassa, B. 1971. The Structure of Protection in Developing Countries. Baltimore, Md.: Johns Hopkins University Press. . 1978. "Exports and Economic Growth: Furtlier Evidence." Journal of Develop- ment Economics 5, no. 2 (June): 181-89. . 1981. "The Newly Industrializing Development Countries after the Oil Crisis." Weltwirschaftliches Archiv 117, no. 1: 142-94. . 1982. Development Strategies in Semi-Industrial Economies. Baltimore, Md.: Johns Hopkins University Press. 1. 984. Adjustments to External Shocks in D)eveloping Countries. World Bank Staff Working Paper 472. Washington, D.C. Bergsman, J. 1974. "Commercial Policy, Allocative and 'X-Efficiency."' Quarterly Journal of Economics 58 (August): 409-33. Bhagwati, J. 1978. Foreign Trade Regimes and Economic Development Anatomy and Consequences of Exchange Control Regimes. Cambridge, Mass.: Ballinger. __ . 1980. "Lobbying and Welfare." Journal of Public Economics 14, no. 3 (Decem- ber): 355-64. _ . 1986. "Rethinking Trade Strategy." In J, P. Lewis and V. Kalleb, eds. Development Strategies Reconsidered. Washington, D.C.: Overseas Development Center. Bhalla, S., and P. Glewwe. 1986. "Growth and Equity in Developing Countries: A Reinterpretation of the Sri Lankan Experience." World Bank Economic Review 1, no. 2: 35-63. Calvo, G. A. 1986. "Fractured Liberalism: Argentina under Martinez de Hoz." Eco- nomic Development and Cultural Change 34, no. 3: 511-34. Cline, W. R. 1982. "Can the East Asian Model of Development Be Generalised?" World Development 10, no. 2: 81-90. Corbo, V., and J. de Melo. 1985. "Overview and Summary: Liberalization and Stabilization in the Southern Cone of Latin America." World Development 13, no. 8: 863-66. Corden, W. M. 1974. Trade Policy and Economic Welfare. Oxford: Oxford University Press. Darrat, A. F. 1986. "Trade and Development: The Asian Experience." Cato Journal 6, no. 2: 695-700. 214 Research Observer 2, no. 2 (July 1987) de Melo, J. 1978. "Protection and Resource Allocation in a Walrasian Trade Model." International Economic Review 19, no. 1: 25-44. de Melo, J., and J. Tybout. 1986. "The Effects of Financial Liberalization on Savings and Investment in Uruguay." Economic Development and Cultural Change 34, no. 3: 561-88. Dixit, A. K. 1976. The Theory of Equilibrium Growth. Oxford: Oxford University Press. Donges, J. 1976. "A Comparative Study of Industrialisation Policies in Fifteen Semi- Industrial Countries." Weltwirtschftliches Archiv 112, no. 4: 626-59. Eltis, W. A. 1973. Growth and Distribution. London: Macmillan. Feder, G. 1983. "On Exports and Economic Growth." Journal of Development Eco- nomics 12, no. 1/2 (February/April): 59-74. Fishlow, A. 1985. "The State of Latin American Economics." In Inter-American Development Bank. Economic and Social Progress in Latin America: Annual Report. Washington, D.C. Grais, W., and others. 1984. "A General Equilibrium Estimation of the Reduction of Tariffs and Quantitative Restrictions in Turkey in 1978." In T. N. Srinivasan and J. Whalley, eds. General Equilibrium Trade Policy Modelling. Cambridge, Mass.: Cambridge University Press. Hahn, F. H., and R. C. U. Matthews. 1965. "The Theory of Economic Growth: A Survey." In American Economic Association. Surveys of Economic Theory, vol. 2. London: Macmillan. Harberger, A. C. 1959. "The Fundamental of Economic Progress in Underdeveloped Countries: Using the Resources at Hand More Effectively." American Economic Review 49, no. 42: 134-46. Harris, R. 1983. Trade, Industrial Policy and Canadian Manufacturing. Toronto: Ontario Economic Council. Havrylyshyn, 0. 1986. "Penetrating the Fallacy of Export Composition." Background paper to the WVorld Development Report 1987. Washington, D.C.: World Bank, Economics and Research Staff. Heller, P. S., and R. C. Porter. 1978. "Exports and Growth: An Empirical Re- Investigation." Journal of Development Economics 5, no. 2: 191-94. Helliner, G. 1986. "Outward Orientation, Import Instability and African Economic Growth: an Empirical Investigation." In S. Lall and F. Stewart, eds. Theory and Reality in Development. London: Macmillan. Henderson, P. D. 1982. "Trade Policies and 'Strategies': Case for a Liberal Approach." World Economy 5, no. 3: 291-302. Hicks, J. R. 1979. Causality in Economics. Oxford: Blackwell. Johnson, H. G. 1958. "The Gain for Free Trade with Europe: An Estimate." Manchester School 26, no. 3: 241-55. Jung, W. S., and P. J. Marshall. 1985. "Exports, Growth and Causality in Developing Countries." Journal of Development Economics 18, no. 1: 1-12. Kakwani, N. C. 1972. "On the Bias in Estimates of Import Demand Parameters." International Economic Revietv 13, no. 2: 239-44. Kaldor, N. 1957. "A Model of Economic Growth." Economic Journal 67, no. 268: 591- 624. Kaldor, N., and J. A. Mirrlees. 1962. "A New Model of Economic Growth." Review of Economic Studies 29, no. 3: 174-92. Kemp, M. C. 1962. "Errors of Measurement and Bias in Estimates of Import Demand Parameters." Economic Record 38, no. 83: 369-72. Deepak Lal and Sarath Rajapatirana 215 Kravis, I. B. 1970. "Trade as a Handmaiden of Growth-Similarities between the 19th and 20th Centuries." Economic Journal 80, no. 320: 850-72. Krueger, A. 0. 1974. "The Political Economy of the Rent-Seeking Society." American Economic Review 64, no. 3: 291-303. _____. 1978. Foreign Trade Regimes and Economic Development: Liberalization Attempts and Consequences. Cambridge, Mass.: Ballinger. . 1984. "Trade Policies in Developing Countries." In R. W. Jones and P. B. Kenen, eds. Handbook of International Economics, vol. 1. New York: North Holland. Lal, D. 1978a. Men or Machines. Geneva: International Labour Office. . 1978b. Estimates of Shadow Prices for Korea. Discussion Papers in Public Economics 10. London: University College, Department of Political Economy. _____. 1979a. "Accounting Prices for Jamaica." Social and Economic Studies 28, no. 3: 534-82. _____. 1979b. "Estimates of Accounting Prices for Sri Lanka." London: University College, Department of Political Economy. . 1979c. "Indian Export Incentives." Journal of Development Economics 6, no. 1: 103-17. . 1980. Prices for Planning. London: Heinemann. . 1981. Review of Bhagwati (1978). Journal of l'olitical Economy 89, no. 4: 826. . 1983. The Poverty of Development Economics. London: Institute of Economic Affairs. . 1986. "Markets, Mandarins and Mathematicians." Paper prepared for Cato Institute Conference in Honor of Lord Bauer, May 1986. To be published in Cato Journal, forthcoming. Lal, D., and S. Rajapatirana. 1987. Impediments to Trade Liberalization in Sri Lanka. Thames Essay No. 51. London: Gower for the Trade Policy Research Centre. Lal, D., and M. Wolf, eds. 1986. Stagflation, Savings and the State: Perspectives on the Global Economy. Oxford: Oxford University Press. Leamer, E. E. 1985. Vector Autoregressions for Causal Inference? Carnegie Rochester Series on Public Policy 22. Amsterdam: North Holland. Leibenstein, H. 1966. "Allocative Efficiency vs. X-Efficiency." American Economic Review 56, no. 3: 392-415. Lewis, W. A. 1980. "The Slowing Down of the Engine cf Growth." American Economic Review 70, no. 4: 555-64. Little, I. M. D., and others. 1970. Industry and Trade in Some Developing Countries. Oxford: Oxford University Press. Little, 1. M. D., and M. F. Scott. 1976. Using Shadow Prices. London: Heinemann. Little, I. M. D., and others. 1979. "Shadow Pricing and Macro Economic Analysis: Some Illustrations from Pakistan." Pakistan Development Review 18, no. 2: 89-112. Lucas, R. E., Jr. 1985. "The Mechanics of Economic Development." University of Chicago, Department of Economics. Mckloskey, D. 1983. "The Rhetoric of Economics." Journal of Economic Literature 21, no. 2: 481-517. Michaely, M. 1977. "Exports and Growth an Empirical Investigation." Journal of Development Economics 4, no. 1: 149-53. . 1979. "Exports and Growth: A Reply." Journal of Development Economics 6, no. 1: 141-43. Michalopoulos, C., and K. Jay. 1973. "Growth of Exports and Income in the Developing 216 Research Observer 2, no. 2 (July 1987) World: A Neoclassical View." Discussion Paper 28. Washington, D.C.: U.S. Agency for International Development. Nogues, J. 1981. "Politica Comercial y Cambiaria: Una Evaluacion Cuantitativa de la Politica Argentina Durante 1961-1981." Technical Study 52. Buenos Aires: Banco Central de la Republica Argentina. Nurkse, R. 1961. Equilibrium and Growth in the World Economy. Cambridge, Mass.: Harvard University Press. Orcutt, G. H. 1950. "Measurement of Price Elasticities in International Trade." Review of Economics and Statistics 32, no. 2: 117-32. Ranis, G. 1985. "Can the East Asian Model of Development be Generalised?" World Development 13, no. 4: 543-45. Riedel, J. 1984. "Trade as the Engine of Growth in Developing Countries, Revisited." Economic Journal 94, no. 373: 56-73. Scitovsky, T. 1958. Economic Theory and Western European Integration. London: Allen and Unwin. Scott, M. F. 1976. "Investment and Growth," Oxford Economic Papers 28, no. 3: 317-63. Solow, R. M. 1956. "A Contribution to the Theory of Economic Growth." Quarterly Journal of Economics 70, no. 1: 65-91. Srinivasan, T. N. 1986a. "Development Strategy: Is the Success of Outward Orientation at an End?" In S. Guhan and M. Shroff, eds. Essays on Economic Progress and Welfare. New Delhi: Oxford University Press. . 1986b. "International Trade and Factor Movements in Development Theory, Policy and Experience." Twenty-fifth Anniversary Symposium on The State of Development Economics, April 11-13, 1986, Yale University, New Haven, Conn. . 1987. "Structural Change, Economic Interdependence and World Develop- ment." In J. Dunning and M. Usui, eds. Economic Independence. London: Macmillan (forthcoming). Srinivasan, T. N., and J. Whalley, eds. 1986. General Equilibrium Trade Policy Modeling. Cambridge, Mass.: IMIT Press. Stern, R. M., and others. 1976. Price Elasticities in International Trade: An Annotated Bibliography. London: Macmillan for the Trade Policy Research Centre. Streeten, P. 1982. "A Cool Look at 'Outward-looking' Strategies for Development." World Economy 5, no. 1: 159-70. Swan, T. W. 1956. "Economic Growth and Capital Accumulation." Economic Record 32, no. 2: 334-61. Taylor, L., and S. L. Black. 1974. "Practical General Equilibrium Estimation of Resource Pulls under Trade Liberalisation." Journal of International Economics 4, no. 1: 37-58. Tullock, G. 1967. "The Welfare Costs of Tariffs, Monopolies and Theft." Western Economic Journal 5, no. 3: 224-32. Tumlir, J. 1981. "The Contribution of Economics to International Disorder." Second Harry G. Johnson Memorial Lecture. Trade Policy Research Centre, London. Tyler, W. 1981. "Growth and Export Expansion in Developing Countries: Some Empirical Evidence." Journal of Development Economics 9, no. 1: 121-30. Wade, R. 1985. "The Role of Government in Overcoming Market Failure: Taiwan, South Korea and Japan." In Helen Hughes, ed. Explaining the Success of East Asian Industrialisation. Cambridge: Cambridge University Press. Deepak Lal and Sarath Rajapatirana 217 THE MANAGEMENT OF COMMON PROPERTY RESOURCES: Finding a Cooperative Solution Robert Wade W then will villagers come together to produce goods and ser- vices that they all need but cannot provide individually? In what circumstances will those who face a potential "tragedy of the commons" be able to formulate rules by which the tragedy is averted? Many writers on collective action are inclined to think that the circumstances are very limited. They argue that people in a situation in which all could benefit from cooperation will be unlikely to coop- erate without an external agent to enforce agreements. Likewise, many theorists on property rights argue that common property re- sources will be overexploited as demand rises, so only private enclo- sure or state regulation stands a chance of preventing such a result.' This article offers a critique of some of the analytical arguments used to reach these conclusions and argues that they have been inappro- priately applied to certain types of village resources. It then discusses how to judge whether villagers will be able to sustain local rules of restrained access to common property resources and interprets the evidence from a study of forty-one villages in South India. Clearly there can be no general presumption that collective action rather than privatization or state regulation will work: witness the frequency of degraded grazing commons, despoiled forests, overex- ploited groundwater, and depleted fisheries. But there are many exam- ples of villagers collectively managing property for long periods. Pri- vatization or state regulation is therefore not always essential. A third option-local collective action-needs to be taken seriously. Because CC 1987 The International Bank for Reconstruction and Development/The World Bank 219 less public money is likely to be needed for local collective action than for either privatization or state regulation, it makes financial sense to establish local rules where circumstances permit. Common On a continuum of property rights, exclusive possession (freehold) Property is at one end. At the other is no property, as in ocean fisheries or the and atmosphere. In between lies common property, where the rights to Common-Pool exploit a resource are held by people in ccnjunction with each other. Resources These rights may take several forms: they may allow unlimited exploi- tation for those within a specified group (as happened in commercial fisheries under national jurisdiction) or they may stipulate limits for each user (as in most commercial fisheries today or as in "stinting" of a grazing commons). Of course, the same type of resource may be exploited under a variety of property rights. This article deals with those resources that might be called "common-pool" resources--a subset of public goods, as that term is used in economics. All public goods have the property that many people can use them at once, because exclusion is difficult. But some public goods yield infinite benefits, in the sense that if A uses more there is not less available for others (lighthouses and weath- er forecasts, for example). Common-pool resources, by contrast, are public goods with finite, or subtractive, benefits: if A uses more, less remains for others. Common-pool resources are therefore potentially subject to congestion, depletion, or degradation (Blomquist and Os- trom 1985; Randall 1983). Groundwater is an obvious example of a common-pool resource. It can be used jointly, but use is subtractive. So when water is scarce, the groundwater table is likely to be depleted. Canal irrigation water, unfenced grazing land, and unfenced forest all meet the same criteria. These three resources-water, grazing, and trees-are vital to the livelihoods of millions of people in developing countries; the question of how to prevent their overexploitation as population grows is im- portant for development policy. The prevailing answer runs as follows: when people are in a situa- tion where they could mutually benefit if all of them restrained their use of a common-pool resource, they will not do so unless an external agency enforces a suitable rule. Each individual has an incentive to ignore the social costs of his behavior for fear that others will exploit the resource before he does. As a result, the rate of aggregate use exceeds the physical or biological rate at which the resource renews itself (Ostrom 1985b). This argument has been used to justify far-reaching proposals for changing the way that common-pool resources are managed (Ostrom 1985a; Runge 1986). According to one school, full private property 220 Research Observer 2, no. 2 (July 1987) rights over the commons are a necessary condition for avoiding over- exploitation (Demsetz 1967; North and Thomas 1977; Johnson 1972; Picardi and Siefert 1976). According to another, it is essential to give an external agency-usually the state-full authority to regulate the commons (Carruthers and Stoner 1981; Hardin 1968). For both schools, the policy issue is simply how to achieve the desired change with the least opposition from those involved. Defining the conditions under which users of common-pool re- Theories sources may voluntarily restrain their use can be considered as a of Collective subproblem of the theory of collective action, also known as the A theory of public goods. Collective action is action by more than one ction person intended to achieve a common goal or satisfy a common interest (that is, a goal or interest that cannot be obtained by an individual alone). Achievement means that a public or collective good has been provided. The collective action might be setting and observ- ing a rule of restrained access to a common-pool resource, and the public good might be the sustainable exploitation that results. One of the theories that has generated pessimism about the viability of collective action is Mancur Olson's "logic of collective action" (which might better be called the illogic of collective action, or the logic of collective inaction). His core proposition is this: "unless there is coercion or some other special device to make individuals act in their common interest, rational, self-interested individuals will not act to achieve their common or group interests" (Olson 1971, emphasis added). In other words, the theorem says that (a) voluntary collective action will not produce public goods, and (b) collective action based on selective (that is, excludable) penalties or rewards may produce public goods. Existing cases of common interest groups are thus to be explained by selective punishments or inducements. My findings question this argument. The conventional view of Indian villages is that they lack any real Indian public realm. A number of men are regarded as "big men," in some Villages sense first in the village. But there is no clearly defined social domain or institution separate from state authority where activities of a public nature are carried out, no center of community management other than the lowest levels of the state apparatus itself, and no machinery for raising resources for public (village) purposes other than through state-sanctioned taxation. I analyzed forty-one villages in South India (Kurnool district, An- dhra Pradesh), thirty-one of which are irrigated from one or another of two large canal systems, while the other ten are dry. Despite the Robert Wade 221 conventional view to the contrary, a significant number of these vil- lages do provide public goods and services through local arrange- ments that have nothing to do with outside bodies, whether govern- ment or voluntary agencies. The Public Kurnool district is semiarid; rainfall averages 620 millimeters a year Realm in a unimodal distribution. Population density averages 105 people a square kilometer (1971), up from 53 in 1870. Seventy percent of the cultivated area is under foodcrops; only 12 percent is irrigated. Thirty-four percent of villages are supplied with electricity (1971). There is one tractor for every one or two irrigated villages, and many fewer in rainfed villages (1980). Most variation in real wage rates is contained within the range of 1.5 to 4.5 kilograms of foodgrain a day. In those villages that have a public realm., it consists of four main institutions: a village council (distinct from the statutory Panchayat, which is moribund in all the forty-one villages); a village standing fund (distinct from local government moneys); a group of village guards, employed by the council to protect crops from livestock and thieves; and a group of "common irrigators," employed by the council to distribute water to the rice fields and to hielp get more water to the village through the government-run canal. The council, and through it the field guards and common irrigators, are loosely accountable to an annual meeting of all the village's farmers. The council also organizes the supply of many other public goods and services, such as repairing wells, ridding the village of monkeys, and making donations for a new primary school or a building where sick animals can be treated, and so on. All these services except water distribution are financed from the village standing fund, for which the council raises money in a variety of ways. Take K village as an example. Its population is just over 3,000. The council has about nine members (the number is fixed for any one year, but varies slightly from year to year). It has authority to make deci- sions affecting all the village. The village's standing fund spends about Rs 10,000 a year (in an economy where a male agricultural laborer gets Rs 4 a day outside of seasonal peaks). The standing fund pays the salaries of the field guards. Four are employed full-time for most of the year, and another two to four are added as the harvest approach- es. About twelve common irrigators are employed for up to two and a half months, for about 1,200 acres of first-season rice. At harvest the common irrigators supplement the field guards in protecting the crops. In the sample of thirty-one canal-irrigated villages, eight have all four of the main corporate institutions-council, fund, field guards, and common irrigators; eleven have some but not all; and twelve have 222 Research Observer 2, no. 2 (July 1987) none. These proportions are not necessarily typical of the whole area, since the sample was drawn not randomly but with an eye to ease of access and a representative range of water arrangements. Among the ten dry villages, eight have field guards; six have a village council; and six have a village fund. The impetus for central (village) control comes primarily from two The distinct sources of social conflict and production loss. Trespassing Management animals and thieves are one. Unrestrained use of irrigation water is of Grazing the other. These are discussed in turn, using K village as an example. K has a population density of 159 people a square kilometer. With this density goes a farming system of annual cropping (at least one crop on each plot each year) and multiple cropping where irrigation permits. Little waste or yearly fallow land is left; the village has no common, in the sense of a large area available for common grazing for a year or more. But oxen and buffalo are needed for traction, and they must be fed. During the growing season they graze close to the crops, on the verges or on small areas of fallow, which are treated as commons. Because fields are not fenced, the animals must be tethered or shepherded. But the incentives for careful shepherding or tethering are asymmetrical-I may not be unhappy to see my animals get fat on your grain. The rationale of the field guards is to make the incentives less asymmetrical. During the medieval and early modern period in Europe, the open-field system solved the problem primarily by regu- lating the cropping; these Indian villagers solve it mainly by regulating the livestock. If the field guards catch an animal grazing a standing crop, they take it to the village pound, where it remains until its owner pays a fine. If just a few animals are involved, the fine is a flat rate-Rs 2 per animal during the day, Rs 4 at night. The rate is set by the council; the field guards collect and keep the fine, dividing it equally among themselves. If many animals are involved, the council uses its discre- tion. The fine may run into hundreds of rupees. The field guards collect it, keep 25 percent, and hand the rest over to the standing fund. (In most villages the owner of the damaged crop is not compen- sated.) The field guards do not enforce a stinting. The decision about how many animals to own and graze is left to each individual. After the harvest of most of the rainfed crops in February, large areas of stubble become available for grazing. (Even in irrigated vil- lages, the area under rainfed crops is generally greater than the irri- gated area.) Each landowner could reserve his stubble for his own animals or those he chose to allow. He could do so by posting guards around each field or by fencing. But the cost of either method-the cost of privatizing the stubble-is very high; all the more so as each Robert Wade 223 landowner tends to have his holding divided into several scattered plots (McCloskey 1975). So, as in the open-field system of Europe, the stubble is put in common; private rights to the product of the land extend only to the crop, not to the crop residues. How, in the "corporate" villages, is this commons managed? Since the village's own stock of animals is adjusted to the availability of year-round grazing, after the harvest it has some surplus grazing, which it could rent out to herdsmen from drier parts of the district. The market for grazing and manure is organized in two ways. In one system, a small group of herders bargains with the council for exclusive access to the village's grazing. The bargain states how many sheep and goats they will bring, when they will come, how long they will stay, and how much they will pay for the franchise. Once the bargain is made, those herders have exclusive claim to the village's grazing, and others can enter only as some leave. Their flocks graze over the stubble by day. By night, when the animals drop most of their manure, they are put on the plots of particular landowners, who pay them an agreed nightly rate per animal. So the herders as a group pay the village a lump sum for access to the commons; and they individually get back part of what they pay through the sale of manure. The second method (used in K, among others) is more complex. Again, a group of herders obtains exclusive access. But instead of a group rent, an auction is held regularly (every four days in K) to decide who will have each flock on his land at night up to the next auction. The auction is arranged by the village council. Half the winning bid (for each flock) is then paid to the herder, and half goes to the village fund. In K, some 9,000 to 13,000 animals commonly enter the village for about six weeks, and the village fund gets about Rs 5,000 in return. Such a large number of animals entering the village when some crops (mainly the irrigated ones) are still standing poses a serious risk of loss for those crop owners. The response is to tighten the regula- tion of the livestock in two ways. One is to stipulate rules for both herder and landowner. These rules are read out at the first auction of the year and may be read out again if infringed. They are worth repeating here because they do not fit easily with the view that Indian villagers cannot, so to speak, get their act together (although they may seem unremarkable compared with the elaborate by-laws of open-field villages in medieval England cited in Ault 1973). For the herder: * He must take the flock to the designated field by 6:30 p.m. and keep it there until 8:00 a.m. * He must not allow the flock to graze standing crops. 224 Research Observer 2, no. 2 (July 1987) * He must deposit with the council half the money paid to him for the first "turn" (four nights); if he leaves before completing four turns (sixteen nights), he forfeits his deposit to the village fund. (This is to discourage herders from leaving before the farmers have had their fields manured and cleared of stubble.) * He must stay within the village boundary; if the farmer asks him to go to a field outside he must refuse. For the farmer: * He must keep the flock within the village boundary. * If he prefers to pay the fund or the herder in kind rather than cash, he must make the conversion at the rate (in early 1980) of Rs 1.25 per measure of hybrid sorghum or Rs 1.50 of "local" sorghum. * To help the herder guard the flock at night, he must provide two men for each 2,000 head. Hired guards must be paid Rs 3 a night, or the equivalent in grain. (This is to prevent the farmer from sending nonablebodied men, who could be paid less.) Such rules are not self-enforcing. Any one farmer would have an incentive to cheat, by not providing the stipulated number of guards or by bringing the flock to a field outside the village boundary. So joint regulation is carried further by means of village-appoint- ed guards. To pay the guards, it would be possible for the council to set a flat rate-so much per cultivated acre-which each landowner had to pay. But such an arrangement would be vulnerable to free riding: a farmer could delay payment indefinitely, expecting that others would not similarly delay. In most villages this free rider problem is avoided by raising the money for the guards' salaries from a collective source. The chief source is the money received from the sale of the grazing franchise, which is generally enough to pay for a semipermanent team of field guards. The "corporate" irrigated villages tend to have several other sources of revenue for the standing fund, based mostly on the sale of council-sanctioned franchises. In addition to the grazing franchise, a franchise to sell liquor is a valuable source of revenue. By law the franchise is sold in a government-run auction. However, the corporate villages usually send just one person from their village, thus acquiring the franchise at the lowest price. They then auction it within the village which, with competition, produces a higher price. The difference be- tween the two prices goes to the village fund. Or again, one village in the sample has an irrigation tank (a small reservoir). Each year the council stocks it with fingerlings, and later in the year auctions the franchise to catch the fish, the money going to the fund. Some villages auction the right to collect a commission on all their grain sales. Robert Wade 225 The Any irrigation system that experiences water shortages poses an Management inherent conflict between farmers upstream and those downstream. of Irrigation Those upstream have first access, and their supply is relatively abun- dant; how much they take determines how much the downstreamers wiiI get. Without some rules about access, continual conflicts are likely. The villages in the sample are irrigated from large, government-run canals and grow rice in the first (wet) season. The rice is transplanted in late July or early August and harvested, in December and January. By the end of September the heavy rains have normally stopped, so the crop depends on canal water. Common irrigators are then ap- pointed to allocate the scarce and fluctuating supply of canal water over the village's land and also to procure, by various means, more water from the government-run supply (vvhich may include surrepti- tiously blocking the outlets of upstream villages). This arrangement is notable in two respects. First, the common irrigators do not decide how much land will make a claim to the irrigation water; that is left to individual farmers (as are decisions about how many animals to graze). Second, however, once the com- mon irrigators are appointed, they take important decisions out of the hands of individual farmers, in the name of a villagewide authority. The criterion used by the common irrigators is "adequately wet- ted." Each field is entitled to be adequately wetted, but it cannot get more water until the other fields downstream from that outlet (most villages have several outlets) have also been adequately wetted. This criterion is quite different from the open access, first-come-first-served rule that prevails before the common irrigators are appointed. The criterion is also quite different from the approach in northwest India, where canal water is always scarce; there during a fixed period of the week each field is entitled to draw whatever water is flowing in the watercourse, but cannot draw more until the same time the following week. The difference in approach presumably derives from the difference between rice (in Andhra Pradesh) and other crops (in the Northwest). Rice is more adversely affected than other crops if the water available is less than potential evapotranspiration; however, rice does not suffer from excessive irrigation. As a result, one farmer's overgenerous irri- gation of a rice crop can cause drastic yield reductions for others. Therefore, whereas the Northwest's "fixed time per acre" method is self-policing-the next farmer in line knows exactly when his turn should start-the judgment of adequate wetting cannot be left to each irrigator and must be made by people answerable to a villagewide authority. Farmers who steal water-who try to influence how much water 226 Research Observer 2, no. 2 (July 1987) they get once the common irrigators have been appointed-are liable to be fined by the council. During a drought, the fines may be Rs 20 to S0 each time, but the main penalty is the stigma of being dressed- down in front of the council. The common irrigators are paid at harvest time, not from the village fund but by means of a levy-so much per irrigated acre-on each irrigator. The rate is set by the council. Is this not vulnerable to free riding? The short answer is no, because the levy is in kind rather than cash and is made at the harvest-the one time of the year when every farmer has no excuse to delay payment in kind. More impor- tant, common irrigators who are not paid one year can damage the nonpayer the next year. The withdrawal of common irrigator services from one individual's land has graver implications than does the withdrawal of field-guarding services. This article has so far discussed how things are done in the corpo- The Ecological rate villages, those with all or most of the four corporate institutions. Basis of Although most of the detailed information comes from K village, the Common organization of the four key institutions varies little from village to Property village, even though they were not imposed from above. Rules Many villages, however, have no corporate organization: no village council, no standing fund, no common irrigators, and no village- appointed guards (though private landowners may hire their own, sometimes forming small groups to do so). In these villages the rule of open access to canal water applies, although informal turn-sharing may develop along some watercourses. And uncoordinated groups of herders may enter a village's land at will, or with the permission of the headman, and will negotiate with individual farmers about where they fold their flocks at night. Why the difference between the corporate and uncorporate villages? The first point is that the corporate irrigated villages are located toward the tail-end of an irrigation distributary (roughly, the bottom one-third of its length; typical distributaries are ten miles long). The second point is that the corporate dry villages tend to be located in areas with black, rather than red, soil. The third point is that, in the semiarid tropics generally, black soil areas tend to be lower down a watershed than red soil areas. So irrigated villages toward the tail-end of a distributary also tend to have a higher proportion of black soils. Black soils are more water-retentive than red soils, allowing a wider range of rainfed crops and a higher yield-and hence a more abun- dant and varied supply of stubble. With unrestrained access, too many animals might come in, damaging the soil. Moreover, the risks of crop loss are higher: with the more varied cropping pattern of black soils, large areas of stubble from the early-harvested crops will become Robert Wade 227 available when other crops are still standing. This provides the impe- tus to arrange for the fields to be guarded, which can be financed by the herders' willingness to pay for the better grazing on black soil. In theory, the power structure of the village might be such that collective action was impossible because ix was not supported by a few key households with land close to the canal. In practice, however, holdings are typically scattered in small parcels, partly because land- owners want to diversify risk. A family with land close to one irriga- tion outlet may have another plot close to the tail-end of a block fed from another outlet. This greatly helps the consensus on the need for rules and joint regulation. Areas of rainfed cultivation higher up a distributory have more red soil, which dries out sooner after the rains stop. They therefore have less stubble, so herders are less interested in grazing there. The higher irrigated areas tend to be under rice in both seasons, but sheep and goat manure is wanted mainly for other crops. So both the demand and supply of animals and grazing is less in the higher villages-and canal water there is more plentiful and reliable. It is in villages lower down a watershed that the potential for conflict is greatest. The evidence of my sample suggests that these lower villages are very likely to have an organized public domain, with rules on the use of water and grazing and the provision of other public goods. How effective are the rules of restrained access in changing the way that resources are used? This question is difficult to answer, because it is hard to find a pair of similar villages, one with corporate institu- tions, one without. All one can say with confidence is that both production and equity are greater where rules and institutions apply than they would have been had the same villages been unorganized. Where individual benefits from joint action are high, joint action is likely to be forthcoming. This is not to say that the temptation for self-interested individuals to go for immediate gain is minor. The need to respond to the free rider problem affects a village's organization. The typical council has developed formidable arrangements for eniforcing the rules, precisely to convince each individual that his fellow villagers will probably abide by the rules, so that if he too does so he will not be the sucker (Runge 1984). These expectations are reinforced by the social compo- sition of the council, an elite body with no pretense at representation. Councils have increased their authority with the passage of time: in all these villages, they have been operating for several decades at least. Lessons What of Olson's argument? One of its main weaknesses is a lack of for Theory attention to the size and nature of the coilective benefit.2 It concen- trates instead on the size of the selective benefits and costs, those that 228 Research Observer 2, no. 2 (July 1987) can discriminate between people according to whether or not they help to provide the public good. It simply assumes the net collective benefit to be high, since free riders must by definition be among those who value the public good highly. So the argument is geared to interpret noncooperation as evidence for the free rider hypothesis, rather than for the hypothesis of low collective benefit. In these Indian villages, however, cooperation and noncooperation are explained in terms of high and low collective benefit, as indicated by downstream or upstream location on a water supply channel and by black or red soils. The presence or absence of selective punishments has little bearing on the variation among villages. Another qualification to Olson concerns the source of control and punishment. Olson's key proposition-that examples of collective ac- tion groups can be explained as the response to selective punishments or rewards-differs from the more conventional notion that joint be- havior is related to the presence of an external enforcer of agreements. Olson is not entirely clear on whether the source of selective punish- ment or reward is inside the group or outside. But he can be read to mean that the penalties must be organized from outside the group. This notion is contradicted by my findings and those of many others. There are many examples where villagers have established rules, monitored the conditions of the commons, checked for cheat- ing, and assigned punishment.3 There are also, of course, many more examples where attempts to do so have failed; in the absence of state regulation or private property, the commons has then degenerated. But the successful examples of local rules show that regulation of the commons does not have to be imposed from outside (McKean 1984; Ostrom 1986). Where Olson and other pessimists about collective action are surely right is in the need for coercion to back up agreements. Their empha- sis on the difficulties of voluntary collective action is a useful counter to the simple optimism of those who believe that community develop- ment projects, people's participation, water users' associations, and the like are mainly a matter of teaching people about their real common interests or promoting values that are less individualistic. On the contrary, rules that make people do what they may not imme- diately want to do are a necessary ingredient of managing common resources, so that while free riding tendencies may remain, they need not destroy the organization. The voluntariness of collective action, therefore, has to be consid- ered as a constitutional issue and as a matter of action. Constitution- ally, people can agree on a set of rules of restrained access or financial contributions, their incentive to do so being prospective net collective benefit. In action, compliance with the rules must also be mainly voluntary, not the result of a calculus of evasion and punishment. But Robert Wade 229 the rules must be backed by a system of punishment, which helps to assure each individual that if he follows the rules he will not be cheated, and which at times of crisis can directly deter violations.4 Lessons If an outside authority wanted to encourage the establishment of for some cooperative bodies, how would it choose to design them? One Organizational lesson is that, in the South Indian context at least, villagers are likely Design to follow joint rules and arrangements only to achieve intensely felt Design needs that could not be met by individual action (Johnston and Clark 1982). These are likely to involve primarily the defense of production (avoidance of crop or animal loss), seconclarily the enhancement of income, and last by a long way, education, nutrition, health, and civic consciousness. The opportunities for avoiding losses or boosting in- come by collective action will be taken only if the losses or gains are large. In my irrigated villages, corporate organization to manage com- mon property is found, with barely an exception, only toward the tail-ends of distributaries. The second lesson is that corporate organizations, to be effective, should be based on existing structures of authority. In practice, this means that the council will be dominated by the local elite, which is a disturbing conclusion for democrats and egalitarians. But rules made by the majority of villagers would carry little legitimacy in the eyes of the powerful. More important, the effectiveness of a council depends on its councillors all having a substantial private interest in seeing that it works, and that interest is greater the larger a person's land- holding (provided holdings are in scattered parcels). Moreover, the tendency of the nonelite to cheat can be checked by the sanctions of the village's power structure. Without these wider sanctions, the coun- cil's formal penalties would probably be ineffectual. This point tends to be overlooked in the public choice literature because it assumes a context of free and equal individuals. If the elite runs the council, will it not become another instrument of exploitation? That is not so in these Indian villages, which reflects the third design lesson: the council is concerned only with nonpriva- tizable benefits. It is not involved in supplying inputs other than water. It is not involved in settling disputes unrelated to husbandry or water. It does not try to compensate the owners of animal-damaged crops, for that would create conflict about privatizable value. In K village, the council once tried to intervene in the allocation of a privatizable good (rationed sugar): the conflicts over who got it be- came so strong that the council almost ceased to function. The fourth lesson is that a council should take on less vital func- tions (well repairing, monkey catching, and so on) only when it is very good at the core (income-defending or income-enhancing) activi- 230 Research Observer 2, no. 2 (July 1987) ties. Not all those councils that are organized to do the essentials also do much less-essential work. The fifth lesson is to keep the techniques of calculation and control simple. Some record keeping and accountability are needed, so as to "institutionalize suspicion," in Ronald Dore's phrase (1971). But the accountability procedures are straightforward in these Indian villages. There is an annual general meeting of all farmers to discuss the forth- coming season, ratify the new council, and receive nominations for guards. The records on standing fund income and expenditure are simple and are read out at the general meeting. Meetings of the council are held in the open, and anyone who passes can listen on the fringes. One specific lesson is that, where water users' associations are deliberately fostered, the right unit of organization is usually the whole village. Attempts to form a water users' association around each canal outlet are likely to be futile if such a group of farmers does not already do other things together. The group will simply not contain enough authority. Yet many programs for irrigation improve- ment in India assume that the natural unit of organization is the outlet or some other hydrologically defined unit. More generally, what can be said about the conditions on which successful collective action depends? In the extreme case-many users, unclear boundaries of the common resources, people scattered over a large area, rules easy to break-degradation of the commons can confidently be expected as demand increases, and privatization or state regulation may be the only options. The likelihood of successful collective action therefore depends on the following:5 *Resources. The smaller and more clearly defined the boundaries of the common resources, the greater the chances of success. * Technology. The higher the costs of exclusion technology (such as fencing), the better the chances of success. * Relation between resources and users Location: The greater the overlap between the location of the common-pool resources and the residence of the users, the greater the chances of success. Users' demands: The greater the demands (up to a limit) and the more vital the resource for survival, the greater the chances of success. Users' knowledge: The more users know about sustainable yields, the greater the chances of success. * Characteristics of users Size: The smaller the number of users, the better the chances of success. However, there is a minimum number below which the tasks able to be performed by such a small group cease to be meaningful. Robert Wade 231 Boundaries: The more clearly defined the boundaries of the group, the better the chances of success. Relative power of subgroups: The more powerful are those who benefit from retaining the commons and the weaker are those who favor enclosing private property, the better the chances of success. Existing arrangements for discussion of common problems: The more developed such arrangements are, the greater the chances of success. Extent to which users are bound by mutual obligation: The more concerned people are about their social reputations, the better the chances of success. * Noticeability. The more noticeable is cheating on agreements, the better the chances of success. Noticeability is a function partly of how clearly defined are the resource boundaries, how near they are to users' residences, and how large is the group of users. * Relation between users and the state, The less the state can or wishes to undermine locally based authorities and the less it can enforce private property rights effectively, the better the chances of success. As the list implies, there can be no presumption that collective action will generally work-any more than there can be a presump- tion that private property or state regulation will generally work. My argument is only that (a) the propensity to descend into anarchy or destruction is neither as strong nor as general as mainstream collective action theory implies and (b) that where circumstances look promis- ing for collective action government officials should treat this option as seriously as the other two. The government can help these local systems by providing a legal framework and perhaps technical assistance. The legal framework should enable village organizations to obtain legally enforceable re- cognition of their identity and rights and to call upon the state as an enforcer of last resort (Korten, forthcoming). Obvious though it may sound, few governments in Asia have given much attention to this task, at least with respect to rural (as distinct from modern urban) organizations. If governments did more, their efforts would widen the range of situations in which locally based common property regimes could be expected to work. Abstract When will villagers come together to supply themselves with goods and services that they all need but could not provide for themselves individually? Can locally based collective action be a viable way to manage common property resources? Many writers on collective action and common property are pessimistic about the ability of people who face problems with common property resources to organize sustainable patterns of use for themselves. Some writers favor privatization of the commons as the only viable solution; others, the imposition of state regulation. This article shows, with reference 232 Research Observer 2, no. 2 (July 1987) to Mancur Olson's "logic of collective action," that the analytical basis for this pessimism is weak for the village-based use of common property resources. There can thus be no general presumption that collective action will fail in the management of common property resources, any more than there can be a general presumption that it will work. The article suggests that the chances of success through collective action depend on the characteristics of the resources, the user group, and group-state relations. This article is based on a forthcoming book, Village Republics: Economic Conditions Notes for Collective Action in South India, Cambridge University Press. I am grateful to Haris Binswanger, Richard Kimber, Ford Runge, and especially Elinor Ostrom for discussions on various points of the argument. 1. For references, see citations later in text. 2. It is not that Olson says or implies that the size of the collective net benefit is irrelevant; he simply does not give it much attention. 3. For example, McKean (1984) on Japan; Gilles and Jamtgaard (1981) on Peru; Campbell and Godoy (1985) on the Andes; Hitchcock (1980), Peters (1983), and Thomsen (1980) on Africa; and Netting (1978) on Switzerland. See also Runge (1987) and Ostrom (1985b). 4. This argument is in line with some of the early writings in public choice theory, notably Buchanan and Tullock (1962) and Ostrom (1968). Later work in the public choice tradition has tended to focus too much on the issue of financial contributions. I have not discussed here the issue of group size. Olson's celebrated theorem, stated without qualification early in his book, is later restricted to large groups in a taxonomy of small, intermediate, and large. He says little about how to distinguish the three types of groups in practice, but he might argue that the groups under discussion here are intermediate groups and therefore outside the scope of his argument. 5. See also Ostrom (1985b), the starting point of my own formulation. Ault, W. 1973. Open-field Farming in Medieval England: A Study of Village By-laws. References London: Allen and Unwin. Blomquist, W., and E. Ostrom. 1985. "Institutional Capacity and the Resolution of a Commons Dilemma." Policy Studies Review 5, no. 2: 383-93. Buchanan, J., and G. Tullock. 1962. The Calcuilus of Consent. Ann Arbor: University of Michigan Press. Campbell, Bruce, and R. Godoy. 1985. "Common-field Agriculture: The Andes and Medieval England Compared." Queens University of Belfast, Geography Department. Carruthers, I., and R. Stoner. 1981. Economic Aspects and Policy Issues in Groundwater Development. World Bank Staff Working Paper 496. Washington, D.C. Demsetz, H. 1967. "Toward a Theory of Property Rights." American Economic Review 57: 347-59. Dore, R. 1971. "Modern Cooperatives in Traditional Communities." In P. Worsley, ed. Two Blades of Grass. Manchester: Manchester University Press. Gilles, J. L., and K. Jamtgaard. 1981. "Overgrazing in Pastoral Areas: The Commons Reconsidered." Sociologia Ruralis 21: 129-41. Hardin, G. 1968. "The Tragedy of the Commons." Science 162 (December): 1343-48. Hitchcock, R. K. 1980. "Tradition, Social Justice, and Land Reform in Central Botswana." Journal of African Law 24: 1-34. Johnson, 0. 1972. "Economic Analysis, the Legal Framework and Land Tenure Systems." Journal of Law and Economics 15: 259-76. Robert Wade 233 Johnston, B., and W. Clark. 1982. "Organization Programs: Institutional Structures and Managerial Procedures." In Redesigning Rural Development: A Strategic Perspective. Baltimore, Md.: Johns Hopkins University Press. Korten, F. 1987. "The Policy Framework for Community Management." In D. Korten, ed. Community Management: Asian Experiences and Perspectives. Hartford, Conn.: Kumarian Press. McCloskey, D. 1975. "The Persistence of English Common Fields." In W. Parker and E. Jones, eds. European Peasants and Their Markets: Essays in Agrarian Economic History. Princeton, N.J.: Princeton University Press. McKean, M. 1984. "Management of Traditional Common Lands in Japan." Duke University, Department of Political Science. Netting, R. 1978. "Of Men and Meadows: Strategies of Alpine Land Use." Anthropolog- ical Quarterly 45, no. 3: 132-44. North, D., and R. Thomas. 1977. "The First Economic Revolution." Economic History Review 30: 229-41. Olson, M. 1971. The Logic of Collective Action. Cambridge, Mass.: Harvard University Press. Ostrom, E. 1968. "Some Postulated Effects of Learning on Constitutional Behavior." Public Choice 5 (Fall): 87-104. __ . 1985a. "Are Successful Efforts to Manage Common-Pool Resources A Challenge to the Theories of Garrett Hardin and Mancur Olson?" Working Paper W85-31. Indiana University, Workshop in Political Theory and Policy Analysis. _ . 1985b. "The Rudiments of a Revised Theory of the Origins, Survival, and Performance of Institutions for Collective Action," Working Paper W85-32. Indiana University, Workshop in Political Theory and Policy Analysis. _____. 1986. "Institutional Arrangements for Resolving the Commons Dilemma: Some Contending Approaches." In B. McKay and J. Acheson, eds. Capturing the Com- mons. Tucson: University of Arizona Press. Peters, Pauline. 1983. "Cattlemen, Borehold Syndicates and Privatization in the Kgatleng District of Botswana." Ph.D. dissertation, Boston University. Picardi, A., and W. Siefert. 1976. "A Tragedy of the Commons in the Sahel." Technology Review 78: 42-51. Randall, A. 1983. "The Problem of Market Failure." Natural Resources Journal 23, no. 1: 131-48. Runge, C. Ford. 1984. "Institutions and the Free Rider: The Assurance Problem in Collective Action." Journal of Politics 46: 154-81. _ . 1987. "Common Property and Collective Action in Economic Development." World Development. Forthcoming. Thomsen, James T. 1980. "Peasant Perceptions of Problems and Possibilities for Local- Level Management of Trees in Niger and Upper Volta." Paper presented at the African Studies Association Meetings, October 15-18. Wade, Robert. 1985. "Common Property Resource Management in South Indian Villages." Agricultural Research Unit Discussion Paper 35. World Bank, Agriculture and Rural Development Department, Washington, D.C. _____. 1986. "The Management of Common Property Resources: Collective Action as an Alternative to Privatization or State Regulation." Agricultural Research Unit Discussion Paper 54. World Bank, Agriculture and Rural Development Department, Washington, D.C. ____. 1987. Village Republics: Economic Conditions for Collective Action in South India. Cambridge: Cambridge University Press. Forthcoming. 234 Research Observer 2, no. 2 (July 1987) World Development Report 1987 Featuring WORLD DEVELOPMENT INDICATORS-the most com- prehensive and current data available on social and economic develop- ment in 128 countries. Plus a special focus on industrialization and trade covering -industrialization trends and transformations -government and the industrialization process -trade strategies and industrialization -the threat of protectionism -prospects for a more open trading system "The best single volume on the state of economic development. -Eric Monke. University of Arizona, Tucson 'A untiiqute source of reliable statistical infornation. -David S. Richardson, Morgan Guaranty Trust Co., New York 298 pages/8 x 101/2 Order your copy of WORLD DEVELOPMENT REPORT 1987 with the coupon that follows. 4~~~~~~ " Only those who remain ideologically attached to outmoded ideas can ignore the powerful evidence of this book. oll -James Buchanan, 1986 Nobel Prize winner in economics Refuting conventional wisdom that only the public sector can supply certain services in developing countries, Gabriel Roth analyzes successful private delivery systerns in more than 50 countries. This fresh approach, Roth argues, could improve the quality of services, particularly for low-income people, while reducing government spending. 288 pages / 61/4 x 81/2 / ISBN 0-19-520544-8 Available in English-language hardback only. Other World Bank publications of interest M Agricultural Mechanization and the Evolution of Farming Systems in Sub-Saharan Africa Prabhu Pingali, Yves Bigot, and Hans Binswanger Looks at the conditions in Sub-Saharan Africa that have led to only sporadic use of tractors, oxen, or plows, and the failure of many agricultural projects that sought to move directly from hand hoes to tractors. Drawing from interviews with farmers at field sites in 10 countries, the authors discuss how various levels of mechanization affect crop yields, the cost-effectiveness of mechanization, and the negative consequences of govemment intervention to encourage use of tractors beyond what is economically justified. Published for the World Bank by The Johns Hopkins University Press 224 pages/6Gs x 9¼4/ISBN 0-8018-3502-X/$22.50 * Agricultural Mechanization: Issues and Options A World Bank Policy Study Analyzes patterns of mechanization and explores its effects on efficiency. Discusses the dis- tortions introduced when policies encourage high capital-labor ratios and accelerate mecha- nization too rapidly. Includes recommendations for developing countries to bring their mechanization policies in line with overall development objectives. 92 pages/8% x Il/ISBN 0-8213-0903-X/$7.50 * Financing Health Services in Developing Countries: An Agenda for Reform A World Bank Policy Study Proposes an altemative approach to financing health care: relieve govemments of the bur- den of spending public resources on health care for the rich in order to make more public resources available for the poor. Draws on examples from World Bank experience to illus- trate the successful application of policy alternatives in various countries. 104 pages/8%2 x I1/ISBN 0-8213-0900-5/$7.50 I Success in Small and Medium-Scale Enterprises: The Evidence from Colombia Mariluz Cortes, Albert Berry and Ashfaq M Ishaq Reviews the experience of small- and medium-scale manufacturing enterprises in Colombia in the 1970s when these enterprises grew rapidly and made important contributions to the country's growth. Analyzes the determinants of economic efficiency and includes both micro- economic data from specific enterprises and aggregate data from national sources. Published for the World Bank by Oxford University Press 296 pages/6Y8 x 94/ISBN 0-19-520593-6/$27.50 Order any of these fitles with the coupon that follows. World Bank Publications ORDER FORM Use this coupon to order any of the following titles: O AGRICULTURAL MECHANIZATION AND THE EVOLUTION OF FARMING SYSTEMS IN SUB-SAHARAN AFRICA (JH3502) copies, US$22.50 each O AGRICULTURAL MECHANIZATION: ISSUES AND OPTIONS (BK0903) copies, US$7.50 each Cl FINANCING HEALTH SERVICES IN DEVELOPING COUNTRIES (BK0900) ________ copies, US$7.50 each O SUCCESS IN SMALL AND MEDIUM-SCALE ENTERPRISES (OX520593) copies, US$27.50 each O THE PRIVATE PROVISION OF PUBLIC SERVICES IN DEVELOPING COUNTRIES (OX520544) copies, US$27.50 each O WORLD DEVELOPMENT REPORT 1987 (check edition desired) EU English, paperback (OX520563) copies, US$12.95 each O English, hardback (OX520562) copies, US$26.00 each C French, paperback (IB0930) copies, US$12.95 each El Spanish, paperback (BK0874) copies, US$12.95 each Air mail outside the U.SA. (US$4.50/copy) TOTAL Enclosed is my check for $ -_ Charge my U VISA El MasterCard E American Express Ul Choice (Credit cards accepted only for orders to Washington, D.C.) Credit Card Number Expiration Date Signature El Please bill me. (Institutional customers only) Purchase Order Number Name Title Address City State Postal Code Country Return to: World Bank Publications World Bank Publications P.O. Box 0552 or: 66, avenue d'1ena Washington, D.C. 20073-0552 75116 Paris, France Coupon orders accepted only from the U.SA and France. To order from other countries, request a list of local World Bank publications distributors. Contributors Vittorio Corbo is economic adviser and acting chief of the Macro- economics Division, Development Research Department, World Bank, and professor of economics, Catholic University of Santiago, Chile (on leave). W. Max Corden is a senior adviser in the Research Department of the International Monetary Fund and professor of economics, Austral- ian National University (on leave). Jaime de Melo is a senior economist, International Economics Divi- sion, Development Research Department, World Bank. Gershon Feder is a senior economist in the Research Unit, Agricul- ture and Rural Development Department, World Bank. Deepak Lal is research administrator, Economics and Research De- partment, World Bank, and professor of economics, University Col- lege, London (on leave). Raymond Noronha is a consultant for the World Bank on issues of land rights and rural development. Sarath Rajapatirana is staff director, World Development Report 1987, Economics and Research Department, World Bank. Robert Wade is an economist in the Agriculture and Rural Develop- ment Department, World Bank, and fellow of the Institute of Devel- opment Studies at the University of Sussex.