1'7Y88 THE WORLD MEIi RESEARCH OBSERVER The Three Faces of the Intetnational Comparison Project Irvitng B. K rat is Agricultural Mechanization: A Comparative Historical Perspective Hans Binswanger Aid in the Development Process Anne 0. Krueger Stabilization Policies in Developing Countries Liaquat Ahamed The Public Subsidization of Education and Health in Developing Countries: A Review of Equity and Efficiency Emmanuel Jimenez II THE WORLD BANK RESEARCH OBSERVER EDITORIAL BOARD EDITOR George Psacharopoulos MANAGING EDITOR Ocy Astra Meesook ASSISTANT EDITOR Francesca Jessup MEMBERs Vittorio Corbo, Enzo Grilli, Peter Muncie, Demetrios Papageorgiou, David Turmham, William Tyler CONTRIBUTING ED1TOR Rupert Pennant-Rea The World Bank Research Observer is published twice a year by the International Bank for Reconstruction and Development. The Observ- er, which is managed in the Office of the Vice President, Economics and Research, and vetted by a board drawn from various departments throughout the World Bank, seeks to keep nonspecialist readers informed about economic research currently being undertaken within the Bank and about developments in special areas of economics relevant for development policy. Each issue of the Observer will 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, interpretations, 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 all correspondence to: The World Bank Research Observer Editorial Office, Room S-9123 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 inrthe 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'Ilna, 75116 Paris, France. Subscription price to The World Bank Research Observer is $10 a year. The single copy price is $6. New subscriptions are entered to begin with the first issue of the current volume and should be sent to: World Bank Publications P.O. Box 37525 Washington, D.C. 20013, U.S.A. t 1986 by the International Bank for Reconstruction and Development/ The World Bank All rights reserved Manufactured in the United States of America ISSN 0257-3032 A Note from the Vice President, Economics and Research The last three decades have seen an explosion of research on develop- ing countries, to which the World Bank has contributed through a large and wide-ranging research program. The resulting stock of knowledge is important both in improving the advice that the Bank can offer to developing countries and in illuminating contemporary debates-some of which may be long standing-about the processes of, and policies to promote, economic development. Nonetheless, the increasing technical sophistication and the quantity of this research are threatening to overwhelm the ability of policymakers and nonspecialists concerned with the problems of developing countries to keep abreast in the field. As a contribution toward the dissemination of the growing stock of findings from research relevant to the develop- ment community, the World Bank with this issue begins biannual publi- cation of The World Bank Research Observer. It is our hope that Ob- server surveys of research on policy issues, written for the reader with some knowledge of economics, will help communicate research results to policymakers and inform the general public about the ongoing debates on development. Anne 0. Krueger The World Bank Notice to Readers This first issue of The World Bank Research Observer is being distri- buted without charge to readers of Research News, which, in its original guise, discontinued publication with the Fall/Winter 1984 issue (vol- ume 5, number 3). The name, Research News, lives on in the form of a newly designed quarterly publication, whiose inaugural issue (volume 6, number 1) was published in September 1985, and the concept of having a substantive lead article on a topic related to a major activity in the World Bank's research program continues in the Observer. Because the Observer is much more a new-in the sense of different- publication than a successor to an old one, readers are urged to fill out and mail to the World Bank the order form at the back of this volume if they wish to receive future issues. A token charge has been applied to the Observer to help defray the costs of manufacture and distribution. Please note that prepayment of the yearly subscription price of $10 must accompany all orders. THE WORLD BANK RESEARCH OBSERVER VOLUME I NUMBER 1 JANUARY 1986 The Three Faces of the International Comparison Project Irving B. Kravis 3 Agricultural Mechanization: A Comparative Historical Perspective Hans Binswanger 27 Aid in the Development Process Anne 0. Krueger 57 Stabilization Policies in Developing Countries Liaquat Ahamed 79 The Public Subsidization of Education and Health in Developing Countries: A Review of Equity and Efficiency Emmanuel Jimenez 111 Recent Publications 131 THE THREE FACES OF THE INTERNATIONAL COMPARISON PROJECT Irving B. Kravis he system of international income and purchasing power com- parisons produced by the United Nations International Compari- son Project (icP) is improving our perceptions of the world econ- omy in three major ways.' The most direct and widely known revela- tions, of course, are the true size of countries' average incomes and the differences between them. The non-icP method commonly used-the conversion of income in own-currency to a numeraire currency by the use of the exchange rate-exaggerates the dispersion of per capita in- comes by systematically understating those of poorer countries. What has attracted less attention is the insight the icP study brings to comparative national price levels. Until now the vast literature on prices and exchange rates has been confined to comparative movements of the price levels of different countries over time. This approach requires adopting some base period when price levels are assumed to be in equilibrium. There has been no way of measuring the differences in the levels of prices for GDP at a given period. The icP fills this gap. It enables us to observe, for example, that Germany's price level was 130 percent of the U.S. level in 1980 and 76 percent of it in 1984, not just that German prices declined by more than 40 percent relative to U.S. prices. The icP's third and still less noticed contribution is the opportunity to compare the relative quantities and prices of the goods that make up the The author is grateful to Laszlo Drechsler, Alan Heston, and Robert E. Lipsey for their helpful comments on an earlier draft of this paper, but none of them is responsible for the views expressed here. David Robinson did the statistical work and Nancy Bonsall prepared the manuscript. 3 GDPS of different countries. Take, for example, the share of spending on producers' durable goods. For most countries, the share expressed in own-currency is available from the national accounts. Without the icp comparisons, however, one cannot tell to what extent a country's spend- ing share is higher than another's because it absorbs larger, quantities of producers' durable goods and to what ex:ent it is merely because their prices are higher. All three of these new insights arise because purchasing power parities (PPPS) replace the exchange rate as the means of converting GDP and its components to a common currency. The exchange rate has great import- ance and many uses, but it is not a reliable indicator of the purchasing power of a currency. For example, the average German mark (DM)/U.S. dollar exchange rate in 1984 was DM 2.85 but it took only DM 2.17 to buy the bundle of GDP that $1 would buy. The German GDP per capita was DM 24,186; the U.S. was $11,363. Converted at the exchange rate of DM 2.85, the German figure was only 75 percent of the U.S. GDP per capita. The correct ratio was in fact 98 percent, based on the relative purchasing power of the two currencies. One purpose of this paper is to elaborate on these three faces of the IcP. Another is to place these results in the context of the history of international income comparisons, the methods used to obtain them, and the problems of maintaining and developing the system of comparisons. A Little The story of real income comparisons goes back to Gregory King History (1936). He found that in 1688 per capita incomes in England and Hol- land were about the same, but almost a fourth lower in France. Nearly three centuries later, the United Kingdom and the Netherlands are still very close; France is roughly 10 percent higher. Since King's work, some other estimates have been made. But until the last few decades, major, well-organized efforts were rare. The mos't ambitious was an investiga- tion of the real wages and welfare of industrial workers in about 200 towns and cities of Belgium, France, Germany, the United Kingdom, and the United States carried out by the British Board of Trade and pub- lished in five reports between 1908 and 191.1. Data were collected in the field on money wages in certain ubiquitous trades; price comparisons were made for foods, fuel, and rents; and descriptions of workers' wel- fare (such as the extent of overcrowding) were provided. At the risk of pressing the data beyond what they can support, it may be said that weekly real wages in October 1905 were about 60 percent higher in the United States than in England, while in the other three countries they were some 30 percent below the English level (Kravis 1984). Another landmark study of a very different character was Colin Clark's compari- sons of real income in his Conditions of Economic Progress (1940). Using a remarkable range of secondary sources, Clark compared the 4 Research Observer 1, no. 1 (January 1986) purchasing power of thirty currencies for consumption goods with a 1929 reference date and more than a dozen with a 1946 reference date. Two advances in statistics and technology paved the way for the next stages of international comparisons. The first was the development of national accounts; the other was the computer. The first widened the focus from wages and wage earners to all types of income and income recipients and provided a statistical framework of concepts (for example, GDP) and accounting rules. The computer made it possible to compare many more countries in a completely even-handed and consistent way (KHS 1982, p. 71). In particular, researchers could satisfy the requirement that the index numbers for income should be "base-country invariant"- that is, the quantitative relationships among countries should be inde- pendent of the choice of the numeraire country. Another important requirement is that the index numbers should be transitive: that is, ji.b 'k/b should equal Ilk1 when each index is computed directly and where I is an index of quantities or prices, and j, k, and b, are countries. The first study based on national accounts was the Gilbert and Kravis (1954) comparison for four western European countries and the Unit- ed States, sponsored by the Organisation for European Economic Co- operation (OEEC). In several important respects, this study established the broad approach adopted in later work, including not only the Gilbert and Associates (1958) extension to four more western European coun- tries but also the Icp. All worked with a breakdown of GDP in terms of final expenditures for functional categories (such as food and clothing) instead of trying to compare the output of the GDP originating in each industry. All also used the relationship that prevails for a given good for any pair of countries: the price ratio between the countries times the quantity ratio equals the expenditure ratio. These basic principles were applied to work on Latin America (Braith- waite 1968; Salazar-Carrillo 1973) and eastern Europe. However, many other studies were made during the 1950s and 1960s, mainly involving pairs of countries. Almost all were based on secondary data sources and differed in scope, method, timing, and quality. The result was a patch- work that fell far short of a cohesive picture of world incomes (KKHS 1975, pp. 2-3). In 1968 the United Nations Statistical Office (UNSO) felt the time was ripe to meet head on the major source of incomparability that prompted these scattered efforts-the deviation of exchange rates from purchasing power parities. The UNSO had already promoted international compar- ability of national accounts by sponsoring the development of a standar- ized system of national accounts (United Nations 1968). With help from the Ford Foundation, the icP was organized. Half of its small central staff came from UNSO, the rest from the University of Pennsylvania. The icP's first effort, phase I, covered seven countries with a 1967 reference date and ten countries for 1970. Phase II extended the comparisons to Irving B. Kravis 5 sixteen countries with 1970 and 1973 data. Phase III included thirty-four countries with a 1975 reference date. The basic methodology was esta- blished in phase I, but was improved and extended in the later phases. In phase IV, the icP was taken over by the UNSO. This phase included about sixty countries with a 1980 reference date. Much of the work was done by regional groups of countries, with much less coordination in the selection of specifications and in methods of quality matching. How It The heart of international income comparisons is making price, rather Is Done than quantity, comparisons. For most kinds of goods, it is easier to obtain a representative sample of country-to-country price ratios than quantity ratios, and the sampling variability of price ratios tends to be smaller. If aggregation problems are ignored, the approach may be sim- ply described as deriving a quantity (real income) comparison by divid- ing a price ratio into an expenditure ratio: Qb Eb Pb where j and b are countries; Q's are physical quantities, E's are expendi- tures (for GDP or its appropriate components), and the P's are prices, the E's and P's being in own-currencies. (1P,/Pb is the purchasing power parity.) The tasks involved are: * Dividing GDP into categories for which expenditure data and price comparisons can be obtained * Selecting and pricing a sample of specifications for each expenditure category * Aggregating the price relatives at the category level * Aggregating the categories to form price and quantity indexes for GDP and its subaggregates. The methods used by the European Comrmunity in its parallel series of comparisons through which its member countries participated in the icP were broadly similar, but involved greater detail in classifying expendi- tures and different ways of matching qualities in price comparisons (SOEC 1983, pp. 22f). The way these tasks have been perforrmed in the iCP may be summa- rized briefly, with particular attention to those elements that affect the quality of the results. Final expenditures on GDP are subdivided into about 150 detailed categories, following closely the United Nations (1968) recommendations. The degree of disaggregation depended pri- marily on the availability of expenditure data in various countries. The classification employed is mainly a functional one: household consump- tion, capital formation, and government consumption are differentiated and then subdivided-for example, food and then beef within consump- 6 Research Observer 1, no. 1 (January 1986) tion. Some effort is made to select classifications that have homogeneous international price relationships for their products. The next step is to make price comparisons for each detailed category, with a few exceptions (such as number of teachers and of govern- ment employees) for which quantity comparisons are more feasible. This phase of the work has the greatest influence on the quality of the income comparison. The sample of specifications for each category must be representative of price formation influences in each country, and the items actually priced in the different countries must be equivalent in quality. To insure such equivalence, the icP arranged international ex- changes of samples, on-the-spot inspection of goods in shops by visiting price experts, plus the advice of merchants, manufacturers, and engi- neers. Once specifications were established, the country's statistical au- thorities provided the national average price for each. In phase III the thirty-four participating countries each priced an average of nearly 400 consumer goods, 38 construction specifications, and nearly 100 producer durable goods. This approach ran into a major aggregation problem because the countries varied widely in their consumption habits. This variety made it impossible to price a standard list of specifications in all countries and still retain the principle that only representative items be compared. Thus, for most detailed categories, few if any countries reported prices for the full list of items. Methods were devised for deriving transitive, base-country invariant index numbers from incomplete sets of prices for the detailed categories. The missing prices in country A were inferred from the price relationships found in other countries between them and other items for which prices were available in country A (Summers 1973). The formula for aggregating across the categories valued the quanti- ties of goods in each country's GDP at world average prices. Added together, these values yield the desired real income comparisons for GDP or its subaggregates. The formula, which was suggested by Robert Geary and amplified by S. Khamis, involves deriving the price compari- sons and the average world prices simultaneously in two subsets of equations (KHS 1982, p. 90). Although there has been a lively academic discussion of aggregation methods, it seems unlikely that equally plausi- ble methods would produce significantly different results (KHS 1982, pp. 95f). In the first three phases of the icP, the methods described above were applied uniformly to all countries, regardless of their location or politi- cal affiliation. In phase III, for example, the Geary-Khamis formula was used to calculate "world" average prices, taking into account the price structures of all thirty-four countries. Malawi and the United States were treated alike-although each was assigned a "supercountry" weight, based on the importance of that type of country in the world, not just Irving B. Kravis 7 the country per se (KHS 1982, p. 79). A single set of prices-a common measuring rod-was therefore used to value the product of every coun- try or supercountry. The same is not true of the phase IV results, because the exercise has been regionalized. For example, comparisons among fifteen participating African countries were carried out using average African prices to value each country's components of GDP. At a second stage this method was applied to several "core" countries from each region to produce interre- gional comparisons. The regional GDP totals obtained from the second stage were applied pro rata to the countries within the region in the proportions of real GDP produced by the first stage. The regional group- ings included two sets of countries in Europe (the European Communi- ties, or EC, and one led by Austria), Africa, Asia, and Latin America. In addition, the EC insisted that the relative per capita GDPS of its member countries obtained in exclusively EC calculations be kept fixed in the worldwide comparisons of the UNSO (and also in the OECD compari- sons for its member countries). As a result, a common measuring rod was no longer used for comparisons between countries in different re- gions. The EC's average prices were used to value and compare the GDPS of its member countries, and world prices were used to determine the total GDP of the EC. This meant that a different set of prices was used to value the quantities in, say, a Germany/Norway comparison than in a Germany/United States comparison, and both are different from the EC average prices that produced the internal EC comparison of, say, Ger- many and France. A valid argument for having regional estimates is that the sample of specifications compared in price and the weights used in aggregation will be closer to those of the countries being compared when the comparison is limited to a region. For internal policy purposes within the EC, com- parisons involving only EC members are preferable. But the EC'S fixity rule requires that the worldwide comparisons may not present alterna- tive results for EC member countries based on the application of world average prices. The aim is to avoid undermining the confidence that policymakers would have in the estimates were there alternative sets of indexes available. However, the fixity rule imposes a high price on worldwide uses and users. Fixed regional results cannot be embodied in the worldwide index- es so as both to preserve the GDP relationships among countries in a region and to have the sum of the region's quantities for each category equal to the sum established by the interregional comparison. It is impossible to produce a table characterized by "matrix consistency" like the one below, that is, with all the rows adding up correctly for each column, each row showing the proper quantitative relationship between the country columns, and the regional and world versions differing by a fixed multiple. 8 Research Observer 1, no. I (January 1986) International units Country A Country B Region Consumption 150 200 350 Investment 30 15 45 Government 20 35 55 GDP 200 250 450 All in all, the case for taking a world view in phase V, now being planned with a 1985 reference date, is compelling. The UNSO would produce "universal" comparisons, leaving it to regional bodies to prod- uce purely regional estimates where there is a practical need for them. The results of phase IV are available for some of the regional group- Real Income ings. But the worldwide comparisons being prepared by the UNSO are not Comparisons completed at the time of writing. In this article, we have linked the European (Ward 1985) and African (soEc 1985a) comparisons for 1980 and extrapolated Asian and Latin American benchmark comparisons from 1975 phase III results to 1980. These procedures give only rough interim estimates, but they serve to illustrate the icP findings. To prepare table 1, the forty-nine countries for which such estimates could be made were listed in ascending order of real per capita income and divided into seven income classes, each with seven countries. The table shows the simple averages of the real (PPP-converted) per capita Table 1. Per Capita GDP Converted to Dollars by PPPs and by Exchange Rates, Averages for Countries Grouped by Income Class, 1980 Average GDP per capita converted by Exchange rate Income class PPP Exchange rate deviation index (U.S. = 100) (1) (2) (3) 1 7.0 or less 4.8 2.3 2.1 2 7.0-9.9 8.7 5.4 2.0 3 10.0-21.9 15.6 9.1 1.8 4 22.0-34.9 26.1 16.2 1.7 5 35.0-59.9 46.5 37.3 1.3 6 60.0-84.9 76.5 86.2 0.9 7 85.0ormore 93.6 109.1 0.9 Note: Forty-nine countries were arrayed according to their Ppp-converted per capita GDP in 1980 (from low to high) and divided evenly into the seven income classes. Only OECD countries (including Canada, Japan, and the United States) are found in classes 6 and 7, but a few OECD countries are found in classes 4 and 5. Classes 1 to 5 comprise mainly developing countries: fifteen African, ten Asian, and five Latin American and Caribbean countries. Sources: See tables 2-4. Irving B. Kravis 9 Table 2. Real and Nominal GDP Per Capita, European Countries, 1980 and 1983 1980 GDP per capita 1983 GDP per capita Exchange rate (U.S. = 100) (U.S. = 100) deviation index converted by converted by 1980 1983 PPPs Exchange rates PPs Exchange rates (1) (2) (3) + (4) Country (1) (2) (3) (4) (5) (6) Portugal 34.4 21.8 35.1 14.7 1.58 2.39 Yugoslavia 35.6 27.7 n.a. n.a. 1.29 n.a. Greece 43.8 36.4 40.9 24.2 1.20 1.69 Ireland 51.1 48.4 52.2 36.7 1.06 1.42 Spain 55.7 49.5 56.5 29.7 1.13 1.90 Italy 69.1 61.3 67.4 44.4 1.13 1.52 United Kingdom 73.1 82.8 75.8 57.9 0.88 1.31 Austria 75.2 89.7 76.8 63.5 0.84 1.21 Finland 77.9 81.5 82.0 68.4 0.96 1.20 Netherlands 82.2 104.5 79.2 65.9 0.79 1.20 Belgium 84.7 106.2 84.7 59.6 0.80 1.42 France 86.0 106.5 87.3 67.9 0.81 1.29 Denmark 85.8 113.0 89.1 78.8 0.76 1.13 Germany 88.7 115.5 88.7 76.2 0.77 1.16 Luxembourg 93.8 111.1 88.2 62.5 0.84 1.41 Norway 98.9 123.3 102.5 95.3 0.80 1.08 n.a. Not available. Sources: Ward (1985), except Yugoslavia (UN 1984). GDP (column 1) and the nominal (exchange-rate-converted) per capita GDP (column 2). The ratio of the real to the nominal per capita GDPS set out in column 3 has been called the exchange rate deviation index (ERDI); it provides a gauge of the extent to which the exchange rate, valuable though it is for many purposes, fails as a measure of purchasing power. It can be seen that the lower a country's real GDP per capita, the higher its ERDI tends to be. This generalization also ocurred in earlier phases of the icP. In phase III, for example, the ERDI averaged 2.6 for eight countries with 1975 real per capita GEIPS of 15 percent or less of the U.S. level, but it was unity for nine countries with real incomes from 60 to 90 percent of the U.S. level (KHS 1982, p. 22). Thus, the dispersion of real per capita incomes is smaller than that of nominal per capita incomes; the coefficients of variation for the underlying 1980 benchmark observations in columns 1 and 2 are 84 percent and 108 percent respec- tively. The larger number of countries included in phase IV makes possible a closer examination of the relationships between real and nominal per capita GDP. Column 5 in table 2 shows the 1980 ERDI for sixteen Euro- pean economies. As in table 1, the lowest-income countries have the highest ERDIs. In 1980, the average ERDI for the five countries with the 10 Research Observer 1, no. 1 (January 1986) lowest real incomes was 1.25; for the next five it was 0.91; for the remaining six it was 0.80. Among the more industrialized countries of western Europe, however, the relationship between the ERDI and real GDP per capita is much more tenuous. For the countries in the lower half of the table (rows 7 to 16), the negative association between per capita income and the ERDIS is very weak despite a wide spread of per capita incomes (Norway's is 35 percent greater than the United Kingdom's). The underlying reason for this weak relationship may be seen by consi- dering the ERDI in terms of its reciprocal-that is, the price level: ERDI = Own-currency GDP . PPP = Exchange rate Own-currency GDP -Exchange rate ppp In terms of the earlier illustration, a Ppp of DM 2.17 to the dollar and an exchange rate of DM 2.85 to the dollar means that German prices are on average 76 percent of the U.S. level. Since the price levels of western Table 3. Real and Nominal GDP Per Capita, African Countries, 1980 1980 GDP per capita (average for 15 countries = 100) Exchange rate converted by deviation index PPPs Exchange rates (1)+(2) Country (1 ) (2) (3) Ethiopia 37 20 1.85 Mali 43 29 1.48 Tanzania 47 37 1.27 Malawi 53 30 1.77 Madagascar 73 55 1.33 Kenya 82 62 1.32 Senegal 88 76 1.16 Zambia 94 94 1.00 Nigeria 115 145 0.79 Zimbabwe 115 105 1.10 Cameroon 117 121 0.97 Morocco 154 128 1.20 Ivory Coast 176 186 0.95 Botswana 205 167 1.23 Tunisia 256 199 1.29 Africa (15 countries) 100 100 1.00 Note: To prepare table 1, the figures in column 1 were converted to a basis of U.S. = 100 by comparing the phase III and SOEc results for the three countries included in both studies (Malawi, Kenya, and Zambia). The average phase IlI/soEc 1980 ratio for these three countries, 8 percent, was applied to the column 1 figures in this table to shift them to a U.S. base. Source: SOEC (1985a). Irving B. Kravis 1l European countries are closely linked by extensive trading and institu- tional ties, differences among them in real per capita income are more weakly correlated with the ERDI. A similar analysis of the phase IV results for fifteen African countries may be carried out on the basis of table 3. Here, ERDIs are highest for the six lowest-income countries. An exception is found in that some of the richer countries in the table have higher ERDIS than those in the middle. For the world as a whole, however,the basic relationship still applies: the poorer a country, the higher its ERDI. This receives further support from the data for Asian countries in table 4, the remaining region for which we can assemble more than a handful of benchmark countries. Comparisons of There are two possible explanations for the fact that the purchasing Price Levels power of the currency of low-income countries tends to be greater than is suggested by their exchange rates. In one, the productivity differential model, the productivity gap between rich and poor countries is held to be smaller in nontradable goods (especially services) than in tradables. This is consistent with the widespread observation, confirmed in the IcP, that nontradables tend to be cheap in low-iincome countries. In 1975, for example, Average price levels (U.S. = 100) Tradables Nontradables GDP Eight countries with per capita incomes less than 15 percent of U.S. per capita income 60 25 41 Nine countries with per capita incomes from 60 to 90 percent of U.S. per capita income 118 97 108 All goods were cheaper in the low-income countries, but the nontrad- ables were cheaper by a larger margin (KHS 1982, p. 196). The reason is that wages in a low-income country are set to a national standard, and the relatively higher productivity in nontracdables ensures low prices (KHS 1982, p. 333). With the prices of tradables more nearly aligned with world levels, low prices for nontradables mean a low price level for GDP as a whole. However, even tradable prices are pulled down because they are always sold with a mixture of services (retailing, for example). The alternative explanation, cast in terms of factor proportions, as- sumes that low-income countries are labor-abundant and that services (an important part of nontradables) are labor-intensive. Since the price equalizing tendencies of international trade operate weakly on nontrad- 12 Research Observer 1, no. I (January 1986) Table 4. Real and Nominal GDP Per Capita, Asian Countries, 1980 and 1983 1980 GDP per capita 1983 GDP per capita Exchange rate (U.S. = 100) (U.S. = 100) deviation index converted by converted by 1980 1983 PPPs Exchange rates PPPs Exchange rates (1) (2) (3) (4) Country (1) (2) (3) (4) (5) (6) India 6.4 2.2 6.7 1.9 2.91 3.59 Pakistan 7.2 2.6 7.8 2.2 2.77 3.55 Sri Lanka 10.0 2.4 11.1 2.4 4.17 4.63 Thailand 15.0 6.3 16.5 5.9 2.38 2.80 Philippines 11.8 6.5 13.8 4.8 1.82 2.88 Korea 25.3 14.4 29.4 13.8 1.76 2.13 Malaysia 24.9 15.1 27.5 14.2 1.65 1.94 Syria 25.2 12.9 27.2 14.8 1.95 1.84 Israel 59.1 47.8 59.0 44.2 1.24 1.33 Japan 73.5 77.1 80.3 69.3 0.95 1.16 Sources: Ppp conversions: extrapolations from phase 111 (1975) results (KHS 1982), except for Israel (SOEC 1985b) andJapan (Ward 1985) for which 1980 benchmarks were available; 1983 Israel extrapolated from 1980; Japan from Ward (1985). Exchange rate conversions: based on [MF (1985) data on current GDP, population, and exchange rates, except for Israel and Japan. ables, they will be cheap in low-income countries and again produce a low price level (Kravis and Lipsey 1983; Bhagwati 1984). These and other structural factors affecting prices in different coun- tries are also joined by short-run influences. Obvious examples include changes in the supply of and demand for money in a given country relative to changes in the numeraire country, and changes in expecta- tions that produce a volatility of exchange rates relative to goods prices (Kravis and Lipsey 1983, p. 18). The literature on the determination of exchange rates has paid most attention to these and other short-run influences. However, it has been argued (Kravis and Lipsey forthcoming) that the price levels of different countries tend to conform to a long-run structural relationship. Such a relationship should prove more useful than the assumption that equilibrium exchange rates conform to pur- chasing power parities, but the issue needs further research. The overall variation in price levels may be inferred from the relation- ships between nominal and real GDP per capita shown in tables 1 to 4. The ERDI for countries in income class 1 of table 1 indicates, for exam- ple, that the 1980 average price level for the seven lowest-income coun- tries was 48 percent of the U.S. level-that is, their price level is the reciprocal of their ERDI. Table 5 examines more closely price level data for four key countries. The first group of figures shows the variability of exchange rates in recent years-a variability muted here by the use of annual averages, but still large. It is far greater, for example, than the tariff changes achieved by arduous trade negotiations. The next two Irving B. Kravis 13 Table 5. Comparative GDP Price Levels, Selected Countries and Dates France Germany Japan United Kingdom (F/US$) (DM/US$) (-Y/US$) (£/US$) Exchange rates 1975 4.29 2.46 297 0.452 1980 4.23 1.82 227 0.430 1983 7.62 2.55 238 0.659 1984 8.74 2.85 238 0.752 PPPs 1975 4.62 2.79 277 0.349 1980 5.24 2.37 240 0.487 1983 5.93 2.19 207 0.503 1984 6.18 2.17 202 0.511 Price levels 1975 108 113 93 77 1980 124 130 106 113 1983 78 86 87 76 1984 71 76 85 68 Source: Ward (1985). banks of figures, the PPPs and the price levels, are unique to icp studies. In the absence of icp data, an analyst might choose a base year and then measure relative changes in price levels. If 1975 were picked as the base and the German PPPs correctly reflected the relative movement of Ger- man and U.S. prices, what is often called the "real exchange rate" might be calculated as shown in table 6. The irndex of the real exchange rate (column 4) is the nominal rate adjusted for the relative price changes in Germany and the United States. The same information may be derived from the price-level data by converting it to an index series with 1975 = 100 and taking the reciprocals. The icp price levels for Germany-113, 130, 86, 76-give information that the real exchange rates do not. To be sure, the latter imply big fluctuations in relative price levels, but they do not show that the Ger- man level shifted from 30 percent above that of the United States in 1980 to 24 percent below it in 1984. Comparisons This article has concentrated so far on an exchange rate's deviation of Economic from its purchasing power over the bundle of goods that comprise GDP. Structure However, the deviation of the exchange rate from iPp is not uniform. It varies from one good to another. This is, of course, a way-of saying that the relative price structures of countries differ. The varying deviations of exchange rates from PPPs also mean that intercountry quantity relation- 14 Research Observer 1, no. 1 (January 1986) Table 6. Calculation of the Real Exchange Rate Real exchange rate Nominal Nominal rate Index of DM/US$ Index of adjustedfor price-adjusted ICP exchange German prices price change rate (3) with price rate to U.S. prices, (1) (2) 1975=100 levels (1) (2) (3) (4) (5) 1975 2.46 100.0 2.46 100 113 1980 1.82 86.0 2.12 86 130 1983 2.55 80.0 3.19 130 86 1984 2.85 79.0 3.61 147 76 a. Ratio of German GDP deflator to U.S. GDP deflator. Sources: Column 1: table5; column 2: IMF (1984, 1985); and column5: tab]e 5. ships for components of GDP are not what exchange rate conversions make them appear to be. Thus, the ICP results open the path to the study of both price and quantity structure. Here we illustrate the possibilities by showing the contrast for two sets of goods-producer durables and services-seen first in terms of real quantity comparisons and then in exchange rate conversions. For convenience, table 7 groups the thirty-four countries from IcP phase III into six classes according to the size of their 1975 real GDP per capita. The figures in column 1 suggest that the four lowest-income groups, consisting mainly of developing countries, devoted a larger pro- portion of their GDP to producer durable goods than did the industrial countries. However, when all components of GDP (including the various kinds of producer durables) are valued in every country at a common set of world average prices, a different picture emerges. The real share of GDP spent on producer durables (column 2) tends to rise with income. This difference stems from the fact that producer durables are relatively expensive in the price structure of a typical low-income country and inexpensive in the price structure of a high-income country. The twist in price relationships which produces this result may be seen in columns 3 and 4. Column 3 shows no marked association between the level of producer goods prices per se and real GDP per capita. But most of the other components of GDP do become more expensive as countries grow richer, and the price-level indexes for GDP in column 4 reflect this. When the prices of producer durables are expressed in relative terms-as col- umn 5 does, taking them in relation to the country's GDP price level- they are indeed seen to be expensive relative to other GDP components in poor countries, but comparatively cheap in rich countries. Developing countries spend a relatively large part of their incomes on producer durables, but do not get as much as appears for what they spend. Irving B. Kravis 15 Table 7. Expenditures and Price Levels for ProducerDurable Goods, by Income Class, 1975 Share of expenditures Price levels Relative pro- in GDpb (u. S. 1 100) ducer goods Income In nominal In real Producer price levels class GDP GDP durables GDP (3) . (4) (U.S. =100). (1) (2) (3) (4) (5) 1 0-14.9 10.7 5.1 130.1 40.7 3.20 2 15.0-29.9 11.6 8.8 105.6 51.7 2.04 3 30.0-44.9 10.5 7.5 13'.8 64.5 2.11 4 45.0-59.9 11.9 10.5 116.4 73.6 1.58 5.60.0-89.9 8.9 11.2 123.8 107.4 1.17 6 90.0-100.0 6.9 9.9 101).0 100.0 1.00 a. U.S. per capita GDP of $7,176 is equal to 100. The ccuntries in each real income class are: (1) Malawi, Kenya, India, Pakistan, Sri Lanka, Zambia, Thailand, and Philippines; (2) Korea, Malaysia, Colombia, Jamaica, Syria, and Brazil; (3) Romania, Mexico, Yugoslavia, Iran, Uru- guay, and Ireland; (4) Hungary, Poland, Italy, and Spain; (5) United Kingdom,Japan, Australia, Netherlands, Belgium, France, Luxembourg, Denmark, and Germany; and (6) United States. b. Averages of country shares. Source: KHS (1982), tables 6-6, 6-5, and 6-8. This general argument can be strikingly illustrated by examples of individual countries. In the phase III report, the Philippines was found to have spent 13.9 percent of its 1975 GDP on producer durables while Korea spent only 11.8 percent. Translated into dollars via exchange rates, their respective expenditure per capita was $52.40 and $68.97. Such figures are grossly misleading, however. Producer durables were 36 percent more costly in the Philippines than in the United States, while in Korea they were 24 percent cheaper. When world prices are used as a common measuring rod, Koreans bought nearly 2.5 times as many prod- ucer durables per capita as did Filippinos ($131.00 compared with $55.60). The Korean share of real Ppp-converted GDP going to producer durables was 8.8 percent; the Philippine share was 5.9 percent. Another illustration of the misleading comparisons that can be prod- uced by exchange rate conversion is the widely held belief that richer countries produce and consume more and more of their output in the form of services. This view seems to be justified by the figures in column 1 of table 8, showing the shares of GDP spent on services by countries in the six income classes. However, when world average prices are used to value the goods and services in each country's GDP, the share of services remains virtually constant (column 2). The contrast can be explained by the behavior of prices (columns 3 and 4). For both goods and services, prices levels rise with income; but service prices are higher in the rich countries by much larger margins than goods prices. For example, the ratio of the price index for class 5 to class L countries is 4.6 for services, 16 Research Observer 1, no. 1 (January 1986) Table 8. Share of Services in GDP and Services Price Levels for Countries Grouped by Income Class, 1975 Share of services in GDP World National average Pnce levels Income prices prices Goods Services GDP class, (1) (2) (3) (4) (5) 1 22.2 33.8 57.2 20.7 40.6 2 28.4 31.7 65.9 34.1 51.7 3 27.3 31.8 83.1 41.2 64.7 4 25.6 30.3 94.0 46.3 73.5 5 36.8 31.2 119.0 94.6 107.5 6 43.9 32.3 100.0 100.0 100.0 a. See table 7. Source: KHS (1982), tables 6-10 and 6-12. only 2.1 for goods. Richer countries do not devote more of their real GDP to services than poorer ones. The cross-country income elasticities of demand for services and for goods are close to 1.0 (KHS 1983). However, if the calculations are confined to consumption rather than all of GDP, the income elasticities of demand are 1.13 for services and 0.93 for goods. The system of international comparisons developed through phase IV Filling the will include benchmark estimates for about sixty countries, most of them Gaps in the for only one or two reference dates. Extensions will clearly involve System of extending the system to all countries and providing annual estimates of Comparisons at least real GDP per capita and the GDP price level. However, these extensions will be a slow process. The UNSO is finding it difficult to finance an adequate central staff, and some countries will lack the data to participate in future rounds of the icP. For nonbenchmark countries, estimates of real per capita GDP or of the price level can be made by one of several methods. The "reduced information" approach, experimented with at the World Bank, aims to produce close-to-benchmark results with data that meet two criteria: (1) they do not require special collection, but are at hand or easily made available; and (2) the PPPs they yield are close to the average PPP for GDP in the cases of the benchmark countries. Another approach is based on relationships found between real GDP per capita in benchmark countries and certain physical indicators such as per capita milk consumption (Beckerman 1966; Economic Commission for Europe 1980). These physi- cal indicators are then used to estimate the real per capita GDP of Irving B. Kravis 17 nonbenchmark countries. A third approach, often referred to as the "shortcut" method, is based on estimating for the set of benchmark countries an equation relating real GDP per capita (or the price level) to certain variables that are widely available for all countries. The equation is then used to estimate real per capitav GDP (or the price level) for nonbenchmark countries. An example of such an equation based on the Phase III results is as follows (with t-ratios in parentheses): In r= 1.57 + 0.89 In n - 0.038 (In n)2 - 0.103 In OP R2 = 0.97 (7.0) (7.7) (1.9) (2.1) SEE = 0.157 where r and n are indexes of real (Ppp-converted) and nominal (exchange-rate-converted) GDP per capita, with the United States = 100; OP (openness) is the sum of exports and imports divided by GDP. The rationale for including n among the independent variables rests on the earlier finding that price levels (PL) are positively correlated with r; this implies that n, the product of PL and r, is positively associated with r. The quadratic term picks up curvature in the relationship between n and r. A high degree of openness tends to raise the prices of abundant factors. In a low-income, labor-abundant country the result is likely to raise the price of services, which tend to ibe labor-intensive, and thus to raise the price level for GDP as a whole. WVith n held constant, a higher price level means a lower real income. In addition to estimates for nonbenchinark countries for the bench- mark years, estimates are needed for nonbenchmark years. Although annual repetition of benchmark surveys would be burdensome, different components of the comparisons might be carried out at different times, completing the cycle over, say, five years. However, benchmark studies once every five years seem the most probable pattern. (Phase III was for 1975, phase IV for 1980, and phase V is being targeted for 1985.) At first sight, the problem of filling in the between-benchmark years seems simple. One has first to alter the benchmark year PPP estimate by the percentage change in a country's implicit deflator relative to that of the numeraire country. If, for example, country A's PPP was 10.0 pesos per dollar in the benchmark year of 1980, and its implicit deflator rose by 10 percent in 1981 while U.S. prices rose by 5 percent, the 1981 ppp would be 10.0 x (1.10/1.05) = 10.48. The country's 1981 real per capita income would then be obtained by dividing the new PPp into the coun- try's own-currency per capita GDP. However, closer inspection reveals some drawbacks in this approach. Country A's increase of 10 percent in prices is based on its own quantity weights, and the U.S. increase of 5 percent on U.S. weights. The relative change is applied to a 1980 Ppp for country A based on yet another set of quantity weights. This disadvan- tage could be moderated by extending into the nonbenchmark years PPPs for individual components of GDP, not just GDP as a whole. Each Ppp would be used to convert a particular component to the numeraire 18 Research Observer 1, no. 1 (January 1986) currency and the results summed to get real per capita GDP. This has the benefit of restricting domestic weights to within-category effects; the aggregation of categories can be based on a common set of international weights applied to all countries. Several studies have applied these methods both to nonbenchmark countries and nonbenchmark years. One study set out a basic approach to a shortcut estimation method (KHS 1978b); another offered a refine- ment of the technique and provided annual estimates for 1950-80 (Sum- mers and Heston 1984). Such efforts will have a firmer basis when the phase IV comparisons for sixty countries become available. Despite the availability of these extrapolations from benchmark coun- tries and years to nonbenchmark countries and years, widespread use has continued to be made of exchange rate conversions. The use of the exchange rate as a converter can be justified only on the ground that it is the best approximation to the Ppp that is available. The question is then whether exchange rate conversions do in fact come closer to benchmark estimates than the shortcut estimates based on Prp concepts. Some newly available benchmark data for European and African countries from phase IV (Ward 1985, SOEC 1985a) provide an opportuni- ty to make some preliminary tests of this issue. Specifically, we can match these new 1980 benchmark figures with the previous estimates for 1980 made on an exchange rate basis (see World Bank 1983) and with the estimates of Summers and Heston (1984) derived from shortcut methods. The new OECD phase IV estimates of 1980 real GDP per capita included five countries for which no previous benchmark studies had been done.2 The OECD benchmark, World Bank, and shortcut estimates Table 9. Estimates of Real GDP Per Capita, 1980 Absolute deviation Shortcut World Bank OECD minus minus benchmark Shortcut, World Bankb benchmark benchmark Country (1) (2) (3) (4) (5) Portugal 34.4 38 20 3.6 14.4 Greece 43.8 49 36 5.2 7.8 Finland 77.9 71 84 6.9 6.1 Norway 98.9 90 111 8.9 12.1 Canada 102.1 93 88 9.1 14.1 a. The shortcut method (Summers and Heston 1984) was applied to obtain 1975 estimates for these and other nonbenchmark countries. The estimates were then extrapolated to other years, including 1980 as shown here. b. World Bank (1983) estimates are for GNP. Sources: Column 1: Ward (1985); column 2: Summers and Heston (1984); and column 3: World Bank (1983). Irving B. Kravis 19 are listed in table 9. In four of the five new benchmark countries, the shortcut method came closer to the benchmark estimate than the World Bank's estimates based on exchange rate conversions. Another comparison between World Bank and shortcut estimates is possible for fifteen African countries for which 1980 benchmark esti- mates were recently published by the SOEC (1985a). These will eventually be linked to the United States and other countries by the UNSO; at the moment, the only official version is a within-Africa comparison. In column 1 of table 10, the benchmark estimate of each country's real GDP per capita is expressed as an index number based on the unweighted average of the fifteen original indexes as 100. (The original indexes were based on the weighted average of the real GDP of the fifteen countries; the shift to unweighted averages is made here to facilitate comparisons with World Bank and shortcut estimates.) Comparing columns 4 and 5, showing the absolute deviations, it can be seen that in ten of the fifteen cases the shortcut methods come closer to the benchmark results; their Table 10. Comparison of Shortcut and World BankPer Capita GDP Estimates with Benchmark Estimates, Fifteen Countries, 1980 Absolute deviations Shortcut World Bank SOEC minus minus benchmark Shortcut World Bank benchmark benchmark Country (1) (2) (3) (4) (5) Ethiopia 33.5 38.1 21.7 4.6 11.8 Mali 39.0 27.6 30.1 11.4 8.9 Tanzania 42.6 54.3 45.2 11.7 2.6 Malawi, 48.0 37.5 31.8 10.5 16.2 Madagascar 66.2 60.2 58.5 6.0 7.7 Kenyaa 74.3 49.8 65.2 24.5 9.1 Senegal 79.8 75.7 70.2 4.1 9.6 Zambia, 85.2 77.1 97.0 8.1 11.8 Nigeria 104.2 171.6 145.5 67.4 41.3 Zimbabwe 104.2 103.6 130.4 0.6 26.2 Cameroon 106.0 103.3 122.1 2.7 16.1 Morocco 139.6 137.4 138.8 2.2 0.8 Ivory Coast 159.5 163.5 185.6 4.0 26.1 Bostwana 185.8 186.2 157.2 0.4 38.6 Tunisia 232.0 214.1 21.0.7 17.9 21.3 Unweighted meanb 100.0 100.0 100.0 - - a. 1975 benchmark estimates, which were available for these countries, extrapolated to 1980 by Summers and Heston (1984). b. The unweighted mean of the original estimates has been taken as 100 as a normalizing device for columns 1 to 3. Sources: Column 1: SOEC (1985a); column 2: Summers and Heston (1984); and column 3: World Bank (1983). 20 Research Observer 1, no. I (January 1986) average absolute deviation is nearly 30 percent less than that of the World Bank estimates. The superiority of shortcut methods over ex- change conversions is the more notable because all the fifteen countries had low incomes; Tunisia, the richest country in the set, has a real GDP per capita roughly one-fifth that of the United States. Even if shortcut methods were used to complete the IcP's coverage, an important gap in our ability to compare economic structures would remain. The system of comparisons that has been developed is based on a final expenditure breakdown of GDP, because that was the easier ap- proach. However, much valuable information about the structure of the world economy-for example, the comparative productivity of given industries in different countries-requires comparisons of GDP by indus- try of origin. This work, begun by Paige and Bombach (1959), needs to be picked up again. It also requires careful price comparisons-which are more demanding, for an equal degree of disaggregation, than those of the final expenditure approach. There is little justification for the bland assumption that goods with the same name are of equal quality in different countries. Rice, for example, seems like a simple product whose price can easily be compared. But rice comes in numerous varieties which dominate different national markets. And the average bag of rice in a low-income country is likely to contain more impurities and broken kernels than in a rich country. International comparisons of quantities that ignore quality differences can be far off the mark. Although the iCP reports have been favorably reviewed in the profes- What sional journals (Hill 1976; Isard 1983), they have also provoked doubts Has Been that the real-income gap between rich and poor countries is really re- Questioned duced by so much. One objection has been to the use of world average prices to value the components of the countries' GDPS (Isenman 1980); it is claimed that world average prices are dominated by the large weights of the rich countries. The Gerschenkron effect thus comes into opera- tion; price weights that are very different from those of the low-income countries push up their relative quantity indexes. While this argument is valid in principle, empirical investigation shows that its quantitative importance is small (Kravis 1984). For example, dividing the weights equally between the set of poor countries and the set of rich ones lowers the estimate of real per capita GDP for the eight lowest-income countries (class 1) in phase III by between 9 and 13 percent. This difference is much smaller than the one between the exchange-rate-converted GDP and almost any reasonably devised PPP-type estimate. (The phase III report put the average ERDI for the eight lowest-income countries at 2.6.) As this suggests, the narrowing of the income gap between poor and rich countries depends more on using a common set of average prices to value the components of the different countries' GDPS than on which Irving B. Kravis 21 particular set is chosen. For example, applying Indian prices in a 1975 binary comparison with the United States yields an India/United States index of 4.1, twice the exchange-rate-converted index of 2.03. Another area of icp methods that has been questioned is the treatment of services. Most services, such as haircuts and telephone calls, can be compared in price. But some, like the services of teachers and govern- ment employees, are "comparison resistant"; they are in a sense "un- priceable," having no readily identifiable unit of output which can be priced. In the national accounts of individual countries, price and quant- ity changes of such services are tracked over time in terms of their inputs. The icp comparisons used the same approach over space. Phase III tried to correct for international differences in productivity of the providers of the services-that is, for medical personnel, and teachers (KHS 1982, chap. 5). Some allowances were also made for the use of capital inputs. These corrections to the raw quantity data were based on a careful sifting of the available evidence, but they cannot claim to be solid. Perhaps the most explicit challenge to the icP treatment of services is that of Maddison (1983). His estimates of 1965 real GDP per capita for ten developing countries are on average almost one-third less than the 1965 estimates extrapolated from icp results for 1975. Maddison says that the "most important" source of the difference is the treatment of services, but he does not offer any other explanation. Maddison's statistical procedures raise serious doubts. His estimates, based on a breakdown of GDP by industry of origin, utilized only secon- dary sources and were made without the benefit of careful price compar- isons. Maddison focuses his criticisms on icp treatment of comparison resistant services and on what he calls "disguised services." For the former, he simply assumes that their prodactivity in different countries varies from one-quarter to three-quarters of the U.S. level, depending on productivity differentials in industry (Maddison 1970, p. 292). He offers no justification for this assumption. Even if Maddison's productivity ratios were correct, however, the change in the phase III results would be small. As Maddison recognizes, comparison resistant services accounted for only 10 percent of the GDP of most low-income countries; so if their true productivity were one-third or one-half of the icp estimate, the estimate of real per capita GDP would be 5 to 7 percent smaller-tiny compared with the difference between exchange rate and PPP conver- sions. Maddison ascribes greater importance to "disguised services," which he defines as the difference between the service industry share in GDP and the share of final product services in GDP. Substantively, Maddison seems to have in mind distributive services. He criticises the ICP for failing to adjust for a larger amount of distributive service per unit of consump- tion in the rich countries, which would therefore lead to an overstate- 22 Research Observer 1, no. 1 (January 1986) ment of the relative real income of developing countries. Again, the figures suggest that the Maddison criticism is exaggerated. The share of distributive services in total services is probably in the 15 to 18 percent range for developing countries and below 15 percent for the developed countries.3 The true value of distributive services would have to be massively understated in order to swell the 15 percent by enough to raise real per capita incomes in rich countries and thereby significantly widen the spread between poor and rich countries.4 It is in any case doubtful that services alone can account for the bulk of the differences between the Maddison and phase III estimates. Ser- vices constitute a little under one-third of the real income of developing countries. For all of the difference between Maddison and icp estimates to be due to an icp exaggeration of service consumption in poor coun- tries, the absorption of services in low-income countries would need to be zero. The icP results show that poor and rich countries divide their income between goods and services in roughly the same proportions. Critics who claim the ICP estimates for service consumption in poor countries are too high are suggesting some perverse economic behavior. They imply that people in poor countries where goods are dear consume lots of them, while they buy only small amounts of relatively cheap services. People in rich countries also behave perversely, buying expen- sive services and economizing on cheap goods. The constant real share of the icP (column 2 in table 8) seems much more plausible. It may be added that the set of ICP outputs-prices, quantities, and real incomes- performed well econometrically, when fitted both to simple demand functions and to a complete demand system (KHS 1982, chap. 9). There is little doubt that purchasing power parities are the right way Concluding to convert GDP and its components from domestic currencies to dollars Comments (the usual numeraire). Using standard exchange rates as converters prod- uces biased results, because exchange rates systematically understate the purchasing power of the currencies of low-income countries. These points are now well understood and accepted. Nonetheless, exchange rates continue to be widely used as converters. One reason given for not using Ppps is that the icP has unresolved methodological problems. In fact, methodological improvements can and doubtless will be made, but the indexes of real per capita GDP are little changed by the use of alternative methods. In the European Community rPp conversions, prod- uced by the methods described above, are being used for policy purpos- es. Furthermore, it is not enough to claim that there are errors and uncertainties in the Ppp conversions to reject them in favor of exchange rate conversions; one has to ensure that exchange rates are better ap- proximations to Ppps than are produced by the icP. Irving B. Kravis 23 A second objection, that Ppps were available for only a limited number of countries, will lose some of its force when the UNSO publishes bench- mark comparisons for sixty countries later this year. More important, shortcut methods could be used to estimate PPP figures for nonbench- mark countries. They are a better guide to real GDP per capita than figures obtained from exchange rate conversions. In addition, if bench- mark studies can be repeated at five-year intervals, estimates for years between the benchmarks can be obtained by extrapolation. The world statistical system could therefore produce an annual set of estimates of real GDP per capita for all the countries of the world. Abstract Purchasing power parities (PPPs), this article confirms, are the correct converters for translating GDP and its components from own-currencies to dollars (the usual numeraire); the alternative measure, exchange rates, obscures the relationship between the quantity aggregates of different countries. Drawing on the reports of the United Nations Interna- tional Comparison Project (Icp), the article contends that exchange rates systematically understate the purchasing power of the currencies of low-income countries and thus exaggerate the dispersion of national per capita incornes. Where full-scale (benchmark) Ppp estimates are not available, estimates based on shortcut methods better approximate what the benchmark estimates would be than do the exchange rate conversions. The icP results also illuminate price and exchange rate relationships among countries by providing a measure of the difference in the levels of prices in different countries. ICP price comparisons for components of GDP make possible the analysis of comparative price and quantity structures of different countries and provide the raw rnaterials for many types of analytical studies. Notes 1. The first three phases of the icP are reportecl in Kravis, Kenessey, Heston, and Summers 1975; Kravis, Heston, and Summers 1978a; and Kravis, Heston, and Summers 1982 (hereafter, KKHS and KHS). A report on phase IV is expected from the United Nations Statistical Office late in 1985. 2. They included thirteen other countries for which 1975 benchmarks were available from phase 111. For these countries Summers and Heston had only to update from 1975 to 1980; they did not have to use shortcut methods. 3. Based on data for Brazil, Colombia, India, Korea, Mexico, Pakistan, Philippines, Spain, and Thailand; and for France, Germany, Italy, Japan, and the United Kingdom (United Nations 1984). 4. It is not clear whether Maddison considers that nondistributive services are also "disguised." In fact, they enter into the final product comparisons in appropriate ways. Some, such as government employee compensation, are also final products and enter directly. Others, such as financial services, enter into the prices of goods and services and are captured in the price comparisons. References Beckerman, Wilfred. 1966. International Comparisons of Real Income. Paris: Organisation for Economic Co-operation and Development (OEci)) Development Center. Bhagwati, Jagdish N. 1984. "Why Are Services Cheaper in the Poor Countries?" Economic Journal 94 (June): 279-86. Board of Trade. 1908a. Report on an Enquiry by the Board of Trade into Working Class Rents, Housing and Retail Prices Together with the Rates of Wages in Certain Occupations 24 Research Observer 1, no. I (January 1986) in the Principal Industrial Towns of the United Kingdom. Cd. 3864. London: Printed for H.M. Stationery Office by Darling & Son. . 1908b. German Towns. Cd. 4032. . 1909. French Towns. Cd. 4512. . 1910. Towns of Belgium. Cd. 5065. . 1911. American Towns. Cd. 5609. Braithwaite, Stanley N. 1968. "Real Income Levels in Latin America." Review of Income and Wealth 14, no. 2 (June): 113-82. Clark, Colin. 1940. The Conditions of Economic Progress. London: Macmillan. Economic Commission for Europe. 1980. Economic Bulletin for Europe 31 (2). Gilbert, Milton, and Irving B. Kravis. 1954. An International Comparison of National Products and the Purchasing Power of Currencies: A Study of the United States, the United Kingdom, France, Germany, and Italy. Paris: Organisation for European Economic Cooperation (OEEC). Gilbert, Milton, and associates. 1958. Comparative National Products and Price Levels. Paris: OEEC. Hill, T. P. 1976. "Review of A System of International Comparisons." Economic Journal (March): 161-64. IMF (International Monetary Fund). 1984a. International Financial Statistics 37, no. 7(July). . 1984b. Yearbook of International Financial Statistics. Washington, D.C. . 1985. International Financial Statistics 38, no. 7(July). Isard, Peter. 1983. "Review of A System of International Comparisons." Journal of Interna- tional Economics 15 (August): 177-81. Isenman, Paul. 1980. "Inter-Country Comparison of 'Real' (Ppp) Incomes: Revised Estimates and Unresolved Questions." World Development 8, no. 1 (January): 61-72. King, Gregory. 1936. Two Tracts. George C. Barnett, ed. Baltimore, Md.: Johns Hopkins University Press. Kravis, Irving B. 1984. "Comparative Studies of National Incomes and Prices." Journal of Economic Literature 22 (March): 1-39. Kravis, Irving B., Alan W. Heston, and Robert Summers. 1978a. International Comparisons of Real Product and Purchasing Power. Baltimore, Md.: Johns Hopkins University Press. . 1978b. "Real GDP Per Capita for More than One Hundred Countries." Economic Journal 88, no. 350 (June): 215-42. . 1982. World Product and Income: International Comparisons of Real Gross Product. Baltimore, Md.: Johns Hopkins University Press. Kravis, Irving B., Zoltan Kenessey, Alan W. Heston, and Robert Summers. 1975. A System of International Comparisons of Gross Product and Purchasing Power. Baltimore, Md.: Johns Hopkins University Press. Kravis, Irving B., and Robert E. Lipsey. 1983. Towards an Explanation of National Price Levels. Special Studies in International Finance no. 52. Princeton, N.J.: Princeton Univers- ity Press. . Forthcoming. "The Assessment of National Price Levels." In S. Arndt and D. Richardson, eds. Real-Financial Linkages in Open Economies. Washington, D.C.: Ameri- can Enterprise Institute. Maddison, Angus. 1970. Economic Progress and Policy in Developing Countries. New York: Norton. . 1983. "A Comparison of the Levels of GDP Per Capita in Developed and Developing Countries, 1790-1980." Journal of Economic History (March): 27-41. Irving B. Kravis 25 Paige, Deborah, and Gottfried Bombach. 1959. A Comparison of National Output and Productivity in the United Kingdom and the United States. Paris: OEEC. Salazar-Carrillo, Jorge. 1973. "Price, Purchasing Power and Real Product Comparisons in Latin America." Review of Income and Wealth 19, no. 1 (March): 117-32. SOEC (Statistical Office of the European Community). 1983. Comparisons in Real Values of the Aggregates of ESA, 1980. Luxembourg: European Economic Communities (EEC). . 1985a. Comparison of Price Levels and Economic Aggregates: The Results for African Countries. Luxembourg: EEC. . 1985b. Comparison of National Accounts Aggregates between Israel and the European Community. Luxembourg: EEC. Summers, Robert. 1973. "International Comparisons Based upon Incomplete Data." Review of Income and Wealth 19, no. 1 (March): 1-16. Summers, Robert, and Alan W. Heston. 1984. "Improved International Comparisons of Real Product and Its Composition: 1950-80." Review Df Income and Wealth 30, no. 2 (June): 207-62. United Nations. 1968. A System of National Accounts. Studies in Methods, ser. F, no. 2, rev. 3. New York: United Natons Statistical Office. United Nations, Economic and Social Council, Statistical Commission and Economic Commission for Europe, Conference of European Statisticians. Summary of the Results of the European Comparison Programme CES/514 (May). New York. Ward, Michael. 1985. Purchasing Power Parities and Real Expenditures in the OECD. Paris: OECD. World Bank. 1983. 1983 World Bank Atlas. Washington, D.C. 26 Research Observer 1, no. 1 Uanuary 1986) AGRICULTURAL MECHANIZATION A Comparative Historical Perspective Hans Binswanger he mechanization of farming in developing countries has been very uneven. In certain parts of Africa, in Java, and in many hilly regions, farmers still till their fields with hand tools even though animal tillage has been common in other parts of the world for thousands of years. While draft animals have completely disappeared in North America, Europe, and Japan, they have been widely accepted in Senegal only in the past few decades. Even in countries where farming is beginning to be mechanized, power tillers and tractors are still restricted to tillage and a few other operations. This paper discusses the history of mechanization, the major reasons for the wide diversity observed, the options for developing countries in extending mechanization, and the role of government policy in influenc- ing the choice of technology. The emphasis is on the adoption of me- chanized techniques in farming systems which are already using animal draft. The issues surrounding the introduction of animal draft where only hand cultivation is practiced are discussed in Pingali, Bigot, and Binswanger (1985). Instead of a summary or conclusions, a set of gener- alizations is presented in the text. The pattern and speed of mechanization is heavily influenced by rela- Economic tive scarcities of capital and labor, and other macroeconomic variables. Influences The responsiveness of invention and innovation to economywide factors has become known as the process of induced innovation (Hayami and Ruttan 1971; Binswanger and Ruttan 1978). Generalization 1. The rate and pattern of mechanization are governed substantially by an economy's land and labor endowments, by the nona- gricultural demand for labor, and by demand for agricultural products. 27 The history of agricultural growth and mechanization in some of today's industrial countries illustrates this generalization. In 1880, factor endowments differed widely among these countries, with Japan having only 0.65 hectares of land per male worker and the United States about forty times as much (see Binswanger and Ruttan 1978, tables 3-1 and 3-2). European countries fell in between, with land in the United King- dom about twice as abundant as on the continent. These differences in endowments were reflected in massive clifferences in factor prices. In Japan, a worker had to work nearly 2,000 days to buy a hectare of land, while his counterpart in the United States needed to work only one-tenth of that time. During the ninety years until 1970, land-labor ratios increased in all countries, especially after 1950. These increases reflect the rapid declines in agricultural labor forces as people moved to industry and off the land. The United States, however, had an increase in agricultural land as well as a reduction in the agricultural labor force, so that differences in land-labor ratios between it and other countries increased. Despite their differences in natural resources, Japan, European countries, and the United States managed to expand their agricultural output by up to 1.7 percent a year. Japan and the continental European countries achieved their rapid growth because yields (outpuit per hectare of arable land) grew at about 1.5 percent a year, or roughly twice as fast as in the United States. Japan and the United States relied on different technological paths to expand their agricultural output. Research summarized by Hayami and others (1975) and Binswanger and Ruttan (1978) has established that Japan has long emphasized biological, yield-raising technology, much of it supported by heavy investment in irrigation. This emphasis continued with systematic investment in agricultural research initiated after 1868. Until the 1950s mechanization played only a minor role (see table 1). The emphasis on biological technology was supported by conscious government choice: in the late nineteenth century Japan imported machi- nery from the United States, but did not find it useful. It then hired biologists from Germany to assist in developing its biological research program, which was successful. The United States, however, emphasized mechanical technology even before 1880 (see tables 2 and 3). Although publicly funded biological research was initiated in the 1870s, it did not produce big increases in yields until about 1930, well after the major land frontiers had been closed and mechanization was far advanced. Successful agricultural growth in various developed countries has therefore capitalized on abundant factors of production: land and me- chanization in the United States; labor, land improvements, and biologi- cal technology in Japan. Continental Europe also emphasized biological technology before shifting the emphasis to mechanical technology.' 28 Researcb Observer 1, no. 1 (January 1986) Table 1. PatternofAgriculturalMechanizationinJapan (thousands) Draft Number and Power of beef Rice sprayers, Power Riding Rice Year farms cattle Horses Motors Pumps Threshers bullers dusters Cultivators tillers tractors Binders Combines transplanters 1880 5,500 1,152 1,626 n.a. n.a. n.a. n.a. n.a. n.a. n.a n.a. n.a. n.a. n.a. 1900 5,502 1,204 1,542 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1910 5,518 1,259 1,564 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1920 5,564 1,256 1,468 2 2 0.5 0.6 n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1931 5,632 1,361 1,477 92 28 56 77 0.2 n.a. n.a. n.a. n.a. n.a. n.a. 1939 5,492 1,767 1,168 293 83 211 132 5 3 3 n.a. n.a. n.a. n.a. 1945 5,670 1,827 1,049 424 87 364 177 7 8 7 n.a. n.a. n.a. n.a. 1951 6,145 -_ 1,112 1,295 92 1,080 460 20 29 16 n.a. n.a. n.a. n.a. 1955 6,027 n.a. 888 2,140 122 2,060 700 87 82 82 n.a. n.a. n.a. n.a. 1960 5,966 n.a. 618 2,799 288 2,651 878 305 791 514 n.a. n.a. n.a. n.a. 1966 5,665b n.a. 396' 3,108b n.a. 3,172 1,008b 1,126 n.a. 2,725 39 146b n.a. n.a. 1971 5,342b n.a. n.a. n.a. n.a. n.a. n.a. 2,400 n.a. 3,201 267 582 84 46 1976 4,835b n.a. n.a. n.a. n.a. n.a. n.a. 2,898 n.a. 3,183 721 1,498 428 1,046 1979 4,742 n.a. n.a. n.a. n.a. n.a. n.a. 2,618 n.a. 3,168 1,096 1,704 747 1,601 n.a. Not available. a. Continued as beef cattle. b. Figure corresponds to nearest adjacent year. c. Figure corresponds to 1963. Sources: Kazushi Okawa, M. Shinohara, and M. Umemura, Estimates of Long-Term Economic Statistics of Japan since 1868: Agriculture and Forestry, no. 9 (Tokyo, 1966); and Farm Machinery Statistics (1981). Table 2. Sources of Farm Power in the United States (thousands) Tractors (exclusive of steam and garden) Number Workstock above Horse- of two years Steam Gas power Year farms Oxen Mules Horses Windmills engines engines Number (millions) Trucks 1870 2,660 1,319 1,125 7,145 n.a. n.a. n.a. n.a. n.a. n.a. 1880 4,009 994 1,813 10,357 200 24 n.a. n.a. n.a. n.a. 1890 4,565 1,117 2,252 15,266 400 40 n.a. n.a. n.a. n.a. 1900 5,737 960 2,753 15,506 600 70 203 n.a. n.a. n.a. 1910 6,406 640 3,787 17,430 900 72 60D 10 0.5 0 1920 6,518 370 4,652 17,221 1,000 70 1,009 246 5 139 1930 6,546 n.a. 17,612a n.a. 1,000 25 1,131 920 22 900 1940 6,350 n.a. 13,029 n.a. n.a. n.a. n.a. 1,567 62b 1,047 1945 5,967 n.a. 11,116 n.a. n.a. n.a. n.a. 2,354 88' 1,490 1950 5,648 n.a. 7,415 n.a. n.a. n.a. n.a. 3,394 93 2,207 1955 4,654 n.a. 4,101 n.a. n.a. n.a. n.a. 4,345 126 2,675 1959 4,105 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1960 3,963d n.a. 2,883 n.a. n.a. n.a. n.a. 4,685 153 2,826 1965 3,356 n.a. _ n.a. n.a. n.a. n.a. 4,787 176 3,030 1970 2,949 n.a. n.a. n.a. n.a. n.a. n.a. 4,619 203 2,984 1975 2,767 n.a. n.a. n.a. n.a. n.a. n.a. 4,469 222 3,031 1979 2,672f n.a. n.a. n.a. n.a. n.a. n.a.. 4,3509 243 3,045 n.a. Not available. a. From 1930 onward refers to tota] workstock on farm. b. Average horsepower for 1930-34 multiplied by number of tractors in 1930. c. Average horsepower for 1940-44 multiplied by number of tractors in 1940. d. After 1960 corresponds to 1969 definition. e. Discontinued. f. Figure corresponds to 1978. g. Tractors over 40 horsepower only. Sources: Number of farms: up to 1959, U.S. Department of Agriculture, Century sf Agriculture in Charts and Tables; 1960-79: U.S. Department of Commerce, Statistical Abstract of the United States (1980). Oxen, mules, horses, windmills, gas engines, and steam engines, 1850-1930: W. M. Hurst and L. M. Church, Power and Machinery in Agriculture (1933), table 8, p. 12; 1930-79: U.S. Department of Commerce, Historical Statistics of the United States: Colonial Times to 1970 (1975). Tractors, horsepower, and trucks, 1870-30: W. M. Hurst and L. M. Church, Power and Machinery in Agriculture (1933), table 8, p. 12; 1940-59: U.S. Department of Agriculture, Changes in Farm Production and Efficiency, 1964 and 1973; 1960-79: U.S. Department of Commerce, Statistical Abstract of the United States (1980). Generalization 2. Mechanization leads directly to increased yields only in exceptional circumstances, such as when high-yielding seeds, pesti- cides, and fertilizers are also used.2 Thus, extra machinery usually sub- stitutes for labor or-where they are already in use-for animals. This generalization corresponds to the substitution view of agricultur- al mechanization (Binswanger 1978). It differs from the net contributor view, which assumes that more machinery-in particular, tractors- 30 Research Observer 1, no. 1 (January 1986) Table 3. Production or Sales of Horse-Drawn and Tractor-Drawn Machines in the United States (thousands) Corn surface planters only Number Harrows (hand Self Threshers Hay making of Plows (all Seed drills and rake Grain Horse Steam Horse Year' farms Horse Tractor types) Cultivators Horse Tractor horse) reapers binders (small) (large) Combines Mowers rakes Loaders Stackers 1870 2,660 865 n.a. 9 89 n.a. n.a. 22 60 n.a. 23 n.a. n.a. 81 n.a. n.a. 1880 4,009 1,326 n.a. 128 318 n.a. n.a. 69 35 n.a. 10 n.a. n.a. 96 9 n.a. 1890 4,565 1,249 n.a. 269 445 n.a. n.a. 132 9 n.a. 11 n.a. n.a. 115 3 n.a. 1899" 5,737' 1,075 n.a. 478 505 n.a. n.a. 208 36 n.a. 1.3 3.6 n.a. n.a. 216 7 n.a. 1899d n.a. 973 n.a. 478 296 92 n.a. 208 36 233 1.3 3.7 n.a. 399 216 7 12 1909 6,406, 1,358 n.a. 701 435 68 n.a. 219 58 129 2.2 8.0 0.5 359 266 35 17 1920 6,518 714 145 604 579 107 3 132 2 100 16.5 4.2 2.7 173 118 32 10 1929 6,512 324 117 540 398 36 16 93 n.a. 65 9.6 1.3 19.6 115 91 26 6 1938 6,527 137 124 351 214 28 57 n.a. 31 2.7 3.6 41.5 76 54 19 1 n.a. Not available. a. Figures for years before 1920 represent numbers manufactured. The earliest sales figures available are for 1920. b. Data from U.S. Department of Commerce. c. Figure corresponds to 1900. d. Data from McKibben, Hopkins, and Austin. e. Figure corresponds to 1910. Sources: 1870-99: U.S. Department of Commerce, Historical Statistics of the United States: Colonial Times to 1970. 1899-1938: E. G. McKibben,J. A. Hopkins, and Griffin R. Austin, Changes in Farm Power and Equipment Field Implements (Philadelphia: Work Projects Administration, National Research Project, August 1939). produces higher yields or other gains in output, regardless of the econ- omic environment in which it is introduced. Such a view usually confus- es the direct effects of mechanization with the indirect productivity effects of factor savings. For example, in an extensively farmed area of Africa where hoe cultivation is used yields may be low, while in an intensively farmed tractorized region of India yields may be much higher. The yield differences may be caused in part by differences in other inputs, such as fertilizers or seeds. They could also be caused by better tillage in India-but this does not mean that good tillage is achieved only by tractors and cannot be achieved by hand. Examples from Java show that cultivation by hand can be as thorough as by ox or tractor. The lower tillage intensity in Africa may simply reflect the abundance of land: in order to maximize labor productivity, people work thinly over a large area. Under the substitution view, the profitability of mechanization and its contribution to economic growth depend on the opportunities available to workers (and sometimes draft animals) released from their tasks. Hence the third generalization: Generalization 3. Mechanization is most profitable and contributes most to growth where land is abundant, where labor is scarce relative to land, and where labor is moving rapidly off the land. Several cases, listed in table 4, illustrate the effects of mechanization on employment. In case 1 unused land is available and mechanization leads to output growth-the more so, the higher the elasticity of final demand.3 The best example is the United States in the second half of the century: an impressive horse-based mechanization led to massive agricul- tural growth because land was rapidly opened up and export markets in Europe provided a highly elastic demand for agricultural products. Total farmland more than doubled between 1870 and 1920. Since average farm size stayed roughly constant, total farm employment must have nearly doubled as well. The agricultural work force, far from being displaced, was redeployed within agriculture, along with large numbers of immi- grants. Mechanization did not produce higher yields, however; they came later (see table 5) and were linked to fertilizers and biological innovations. In all these changes, the elastic demand provided by export markets played a crucial role. Without such export possibilities, areas planted, employment, and agricultural output would have expanded less and mechanization would probably have happened more slowly. (If final demand is very inelastic, mechanization could lead to a reduction in agricultural employment even if extra land is available.) Mechanization can also be induced by labor scarcity arising out of nonagricultural demand for labor (case 2). Production costs rise because 32 Research Observer 1, no. 1 (January 1986) Table 4. Direct andIndirectEffects of AgriculturalMechanization Indirect Indirect Forces Immediate effect effect leading consequence on on to of agricultural agricultural mechanization mechanization output employment Examples 1. Land available Labor used on larger Expands, and more Expands if demand is Nineteenth-century areas, production quickly the more elastic; stagnates or United States costs drop elastic is final falls if demand is demand inelastic 2. Wages rising in Production costs rise Falls (or grows more Falls United States after response to less than in absence slowly), but by less 1940; Japan, Europe nonagricultural of mechanization than in absence of after 1955 labor demand mechanization 3. Unmechanized A new method of Expands, and more Expands, and more Pumping in technique production becomes quickly the more quickly the more contemporary Asia unprofitable profitable elastic is final elastic is final demand demand 4. Subsidies on capital Production costs may Small expansion at Falls, sometimes Contemporary energy drop modestly or best sharply Brazil, Pakistan, stay constant China wages rise rapidly. Other things being equal, farming output will there- fore fall (or grow more slowly), depending on the elasticity of final demand. Farmers mechanize, although they can seldom prevent some increase in their production cost. The best example of these trends comes from the United States after 1940. The use of tractors, combines, and other machines expanded at unprecedented rates (see table 2). Al- though labor input per acre or per animal had declined a little between 1915 and 1939, it fell sharply after 1940 (see table 5). Agricultural employment also fell substantially, both in absolute and relative terms, and labor was redeployed outside agriculture. The number of workers per farm was stable, while farm sizes grew rapidly from an average of 167 acres in 1950 to 401 acres in 1978. Europe went through equally dramatic changes after 1955. Cases 1 and 2 show that the labor effects of mechanization depend on the alternatives available to the economy. The Indian Punjab provides an opposite example. The green revolution initiated in the mid-1960s led to sharply increased demand for labor, which caused a big rise in real wages around 1968 (Gupta and Shangari 1979). This in turn led to increased seasonal and permanent migration, primarily from Eastern India.4 But it also led to the use of more tractors and threshers by Punjab farmers. The combined effect of these developments was a dec- line in real wages after 1972-73, which brought them closer to the stagnant real wages in the rest of India's agriculture. Because India's Hans Binswanger 33 Table 5. ProductivityIndicators: LaborHoursper UnitofProduction and Related Factors, Selected Crops and Livestock in the iUnited States, 1915-78 Crop and livestock 1915-19 1925-29 1935-39 1945-49 1955-59 1965-69 1974-78a Crop Corn for grain Hours per acre 34.2 30.3 28.1 19.2 9.9 5.8 3.7 Yield (bushels) 25.9 26.3 26.1 36.1 48.7 78.5 87.8 Sorghum grain Hours per acre n.a. 17.5 13.1 8.8 5.9 4.2 3.9 Yield (bushels) n.a. 16.8 12.8 17.8 29.2 52.9 50.8 Wheat Hours per acre 13.6 10.5 8.8 5.7 3.8 2.9 2.9 Yield (bushels) 13.9 14.1 13.2 16.9 22.3 27.5 30.0 Hay Hours per acre 13.0 12.0 11.3 8.4 6.0 3.8 3.5 Yield (tons) 1.25 1.22 1.24 1.35 1.61 1.97 2.15 Potatoes Hours per acre 73.8 73.1 69.7 68.5 53.1 45.1 38.3 Yield (cwt.) 56.9 68.4 70.3 117.3 178.1 212.8 257.0 Sugarbeets Hours per acre 125 109 98 8.5 51 33 26 Yield(tons) 9.6 10.9 11.6 13.6 17.4 17.5 19.7 Cotton Hours per acre 105 96 99 83 66 30 10 Yield (pounds) 168 171 226 273 428 484 462 Tobacco Hours per acreb 353 370 415 460 475 427 259 Yield (pounds) 803 772 886 1,176 1,541 1,960 2,049 Soybeans Hours per acre 19.9 15.9 11.8 8.( 5.2 4.8 3.7 Yield (bushels) 13.9 12.6 18.5 19.6 22.7 25.8 27.8 (continued) economy was growing slowly, a slower rate of mechanization and a larger volume of migration could have solved the labor shortages in Punjab at a lower capital cost. And the extra employment would have meant that the benefits of the green revolution were shared more widely with workers in poorer regions. Mechanization can be a powerful stimulus to growth when it makes a new method or crop profitable (case 3). The best example is pump irrigation. Although it is always possible to lift water with animal or human power, it may often not be profitable to do so, even at extremely low wages. The pump therefore enables output to rise-but the size of the increase is determined by the elasticity of final demand. Since the extra production requires extra labor, agricultural employment expands more or less in step with output. 34 Research Observer 1, no. 1 (January 1986) Table 5 (continued) Crop and livestock 1915-19 1925-29 1935-39 1945-49 1955-59 1965-69 1974-78a Livestock Milk cows Hours per cow 141 145 148 129 109 78 48 Milk per cow (pounds) 3,790 4,437 4,401 4,992 6,307 8,820 10,783 Cattle other than milk cows Hours per cwt. of beef producedcl 4.5 4.3 4.2 4.0 3.2 2.1 1.4 Hogs Hours per cwt. produced' 3.6 3.3 3.2 3.0 2.4 1.4 0.6 Chickens (laying flocks and eggs) Hours per 100 layers n.a. 218 221 240 175 97 61 Rate of lay n.a. 117 129 161 200 219 234 Chickens (farm raised) Hours per 100 birds 33 32 30 29 23 14 12 Hours per cwt. producedd 9.4 9.4 9.0 7.7 6.7 3.7 3.0 Chickens (broilers) Hours per 100 birds n.a. n.a. 25 16 4 2 0.6 Hours per cwt. producedd n.a. n.a. 8.5 5.1 1.3 0.5 0.2 Turkeys Hours per cwt. producedd 31.1 28.5 23.7 13.1 4.4 1.3 0.6 n.a. Not available. cwt. Hundred weight. Note: Labor hours per acre harvested include preharvest work on areas abandoned, grazed, and turned under. a. Preliminary. b. Per acre planted and harvested. c. Production includes beef produced as a by-product of the milk cow enterprise. d. Liveweight production. Source: Economics, Statistics and Cooperative Service-Economics. The pace of mechanization is influenced by three other economic factors: capital scarcity and energy costs, farm size, and subsidies. Capital scarcity and energy costs. Poor societies have smaller capital stocks than rich ones, and the cost of capital (in terms of labor) is higher. High capital costs retard mechanization in several ways. First, they reduce the profitability of all forms of agricultural investment, including land improvements, irrigation, animals, and buildings. Second, they may cause farmers to allocate whatever investment funds are avail- able away from mechanical inputs. This trend will be stronger the more expensive and long-lived the mechanical inputs are and the easier it is to produce other forms of capital (such as land improvements) by hand. A third effect, discussed in detail in the next section, is that higher capital costs produce a bias in mechanization toward power-intensive opera- Hans Binswanger 35 tions. Finally, higher capital costs influence the design of machines; if repair costs are relatively low, designs emphasize repair over durability. Generalization 4. High capital costs (relative to labor) retard mechani- zation and lead to selective emphasis on power-intensive operations. Machinery design adjusts to high capital costs by lack of convenience features, simplicity, and reduced durability. Energy is only one of the costs of using rnachines. Capital and mainte- nance costs are often larger. Since the profitability of machines-their comparative advantage-is tied closely to labor costs, expensive energy is likely to retard mechanization much nnore in countries with cheap labor. Farm size. The size of the average farm is largely a reflection of the scarcity of land relative to labor and thus need not be an independent influence on the pace of mechanization. Mechanization can certainly facilitate the growth of large farms, however, as it did in the United States after 1940 and later in Western Europe. Generalization 5. Mechanization is the main facilitator of the trend toward bigger farms. This goes along with another lesson from historical data and contem- porary experience: Generalization 6. Large farms adopt new forms of machinery consi- derably faster than small farms. Because larger farms offer more collateral, they make it easier to bor- row to invest in new machinery.5 In addition, some (but not all) mechani- zation is subject to genuine economies of scale: it is technically more efficient to design a large rather than a small machine. Even machines invented in countries with abundant labor (and therefore smaller farms) were first developed for the largest farms, because they had the lowest costs of capital relative to labor.6 The market for machines expanded to smaller farms only when labor costs rose or capital became more abun- dant. In the history of engineering, technical developments have often been embodied in smaller and smaller machines.7 Japan, in particular, has developed many machines for small farms and plots. For certain operations, mechanization spreads to small farms when machinery can be rented rather than bought. For a rental market to be established, the optimal farm size for owning a machine must be bigger than that of numerous small farms. In addition, it is easier to establish rental markets for operations that do not need to be done on all farms at the same time: threshing and milling are examples. It is thus no accident 36 Research Observer 1, no. 1 (January 1986) Table 6. Ownership and Use of Farm Equipment in the Philippines, 1971 (thousands) Number Number of machines Number of farms using machines Ratio Type of of farms owned by Owned fully Rented or provided of renters equipment reporting farm operator or party by landlord to owners Total number of farms 2,355 - - - - Plows 1,170 1,511 1,366 129 0.09 Harrows 887 1,069 1,031 94 0.09 Tractors 11 16 12 78 6.50 Stripping machines, crushers, shellers 16 19 18 85 4.72 Harvesters and threshers 14 26 16 132 8.25 Power-producing machines 5 7 6 3 2.00 Carts and wheelbarrows 262 292 306 46 0.15 Motor vehicles 14 19 15 69 4.60 Sprayers 79 90 89 61 0.69 Source: National Census and Statistics Office, Philippine Census of Agriculture 1971 (Manila: National Economic and Development Authority, 1971). that rental markets for threshing machines were well established in the nineteenth century in the United States and are now common all over Asia (Gardezi and others 1979; Walker and Kshirsagar 1981). The contract-hire system for combines in the United States illustrates the problem of synchronized timing. The contractors achieve higher rates of machinery utilization by migrating to follow the harvest from the Texas- Oklahoma area to the northern states, where harvesting takes place months later. Milling rice for home consumption can also be done over a long period, and in Asia it is common for a mill owner to "rent" his machine to customers. Rental markets for land preparation, with the use of animals or tractors, were once common in the United States, when plows were scarce, and in Europe. Such rental markets are now common in Asia wherever tractors or power tillers have penetrated. These characteristics of rental markets are confirmed by some data for tractor rentals in South Asia and machinery rentals in the Philippines (see table 6). The Philippine data show that most farmers own their animals, carts, plows, and harrows. Harvesting and threshing equip- ment, tractors, and motor vehicles, however, are used on about five to seven times more farms than own them, which indicates that rental markets are extremely well developed. Generalization 7. Where rental markets are fairly easy to establish, farm size has much less influence on the size of machines. Subsidies. Subsidies may speed up mechanization (case 4). Because the direct effects of mechanization on yields are small, however, any effect of Hans Binswanger 37 subsidies on agricultural output must be an indirect one that arises from the cost reduction made possible by machines. But when mechanization is not spontaneously driven by some form of labor scarcity, the impact on production costs is not large; the output effects of subsidies therefore cannot be large either. When mechanization is caused by subsidies, reductions in the agricul- tural work force can be substantial. Unlike in cases 1 and 2, workers who lose their jobs will find only infericir alternatives, and some may remain unemployed; this redeployment of labor is not a productive benefit, but a loss. Nor is there any potential relief from drudgery for the redeployed workers, since their inferior work options may in fact entail more drudgery. Patterns of The most dramatic aspect of mechanization is the shift from one Mechanization source of power to another. In ancient China cattle began to replace human labor more than 3,200 years ago. Between the second and fourth century A.D. fairly widespread use of water power is reported from China for rice pounding, grinding, and water lifting (Liu 1962). Water power was widely used for milling purposes in Europe during the Mid- dle Ages; at about the same time wind power is reported to have been used in China and Europe as well. In nineteenth-century Europe and North America, oxen were displaced by horses, which provided power for many mechanical devices from about 1850 to as late as 1965. Steam engines were widely used for only about fifty years between 1870 and 1920. After 1900 they were rapicdly displaced by internal com- bustion engines and electric motors. Tractors came into widespread use in North America after about 1920, but coexisted with horses for rough- ly twenty-five to thirty years. Except for Great Britain, where tractors began to be adopted in the 1930s, the tractorization of European and Japanese agriculture was delayed until about 1955, after which it hap- pened very quickly (table 7). The emphasis on shifts in power sources, especially the shift to trac- tors, can cause misunderstandings about which operations are the most likely candidates for mechanization in developing countries. This section therefore discusses mechanization in terms of operations and pays only occasional attention to power sources. Most of the evidence comes from machinery stock data; though lacking detail, no other data can give so comprehensive a picture over long periods of time.8 Operations can be grouped in terms of the intensity with which they require power (or energy) relative to the control functions of the human mind (or judgment). Regardless of the stage of mechanization, new power sources are always used first for power-intensive operations. Furthermore, it appears that the price of labor matters less for the mechanization of power-intensive operations than for control-intensive 38 Resea,ch Observer 1, no. 1 (January 1986) ~5 Table 7. Growth in Number of Tractors in Selected Countries (thousands) Japan United United Korea Year Two wheel Four wheel Germany Denmark France Kingdom, States Spain Yugoslavia Two wheel Four wheel India Mexico Philippines 1920 n.a. n.a. n.a. n.a. n.a. 10 246 n.a. n.a. n.a. n.a. n.a. n.a. n.a. 1930 n.a. n.a. n.a. n.a. 27 30 920 n.a. n.a. n.a. n.a. n.a. 4 n.a. 1938-39 3 n.a. 30 4 36 55 1,545 3 n.a. n.a. n.a. n.a. 5 0.2 1945-47 8 n.a. 69 4 77 244 2,613 5 n.a n.a. n.a. 5 n.a. 1 1950 16 n.a. 140 17 137 325 3,394 10 6 n.a. n.a. 9 23 n.a. 1955 82 n.a. 462 58 305 436 4,345 25 10 n.a. n.a. 21 na. n.a. 1960 514 n.a. 857 111 680 456 4,688 39 36 1 n.a. 31 55 8 1965-66 2,725 39 1,164 161 996 482 4,787 148 45 1 1 n.a. 54 na. n.a. 1970 3,201 267 1,371 175 1,230 514 4,619 260 80 44 0 148 91 11 1975-76 3,183 721 1,425 185 1,363 541 4,469 379 226 60 1 228 102 n.a. 1979 3,168 1,096 1,456 190 1,430 508 4,350 492 385 n.a. n.a. 310 114 n.a. n.a. Not available. a. Great Britain and Northern Ireland. Sources: Binswanger (1984), tables 3, 9, 12, 13, 14, 16, 18; Organisation for Economic Co-operation and Development, Development of Farm Motorization and Consumption and Prices of Motor Fuels in Member Countries (Paris, 1962); and Food and Agriculture Organization, Production Yearbook, various issues. Table 8. Machinery Patterns in France (thousands) Hay and Sprayers Cream Root straw Pick-up Sowing Fertilizer of Year Plows separators Threshers cutters presses trucks machines distributors traction 1852 2,578 n.a. 60 n.a. n.a. n.a. n.a. n.a. n.a. 1862 3,206, n.a. 101 28 n.a. n.a. 11 n.a. n.a. 1882 3,267 n.a. 211 n.a. n.a. n.a. 29 n.a. n.a. 1892 3,669 n.a. 234 n.a. n.a. n.a. 52 n.a. n.a. 1929 1,190b 666 204 n.a. 10 74 322 119 142 1937 n.a. n.a. 152 n,a. 9 n.a. n.a. n.a. n.a. 1941 n.a. n.a. 141 n.a. 9 n.a. n.a. n.a. n.a. 1946 1,325 626 206 1,007 12 n.a. 385 151 85 1950 1,385 686 218 1,099 17 n.a. 410 165 104 1955 1,427 696 215 1,152 26 n.a. 447 221 122 1960 n.a. 672 191 1,152 33 n.a. 514 321 153 1965 n.a. n.a. 122 n.a. n.a. n.a. n.a. n.a. 223 1970 n.a. n.a. n.a. n.a. n.a. n,a. n.a. n.a. 304 1977 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 406 n.a. Not available. a. Includes 794 improved plows. b. Double-sided plows only after 1929. c. Includes motor-driven mowers. d. Only self-propelled combines. ones-that is, it often pays to move to a higher stage of mechanization in power-intensive operations, even at low wages, when mechanization of control-intensive operations is not profitable. The rest of this section provides support for the following: Generalization 8. When new power sources become available, they are initially used only for selected operations for which their comparative advantage is greatest. Power-intensive operations are shifted most rapid- ly to new power sources. Control-intensive operations are shifted to the more mechanized techniques when wages are high or rapidly rising. Power-intensive processing and pumping. Milling, threshing, chop- ping, sugarcane crushing, pumping of water, and the like are extremely power-intensive but need little control. Mioreover, both stationary and mobile sources of power can be used for them. Among the stationary sources, water was first used for milling, pounding, and grinding in the first century B.C. in China. Water-powerecL milling was also invented in France in the fourth century A.D., though not until the twelfth century was it adopted throughout Europe. Wincl power has historically been used almost exclusively for milling and for lifting small amounts of 40 Research Observer 1, no. 1 (January 1986) Sugar Reaper/ Motor Hay Potato beet Milking binders Mowers mowers Combines balers diggers diggers Rakes Tedders machines n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 9 9 n.a. n.a. n.a. n.a. n.a. n.a. 6 n.a. n.a. 19 n.a. n.a. n.a. n.a. n.a. n.a. 27 n.a. 23 39 n.a. n.a. n.a. n,a. n,a. n.a. 51 n.a. 420 1,389 n.a. n.a. n.a. 60 13 739 354 4 341 n.a. n.a. 0.3 n.a. n.a. n.a. n.a. n.a. n.a. 481 1,279 n.a. 0.3 n.a. n.a. n.a. 733 448 n.a. 501 1,373' n.a. n.a. n.a. 67 10 740 n.a. n.a. 529 1,470' n.a. 5 n.a. 77 11 785 n.a. 46f 560 1,547' n.a. 18 17 90 12 839 n.a. 80 534 n.a. 97 50 51 91 15' n.a. n.a. 124 361 n.a. 104 102 169 100 20 n.a. n.a. 186 133 n.a. 105 133 292 92 25 n.a. n.a. 283 n.a. n.a. 84 148d 445 n.a. n,a. n.a. n.a. 392 e. Includes only diggers up to 1960 and only complete harvesters from 1965 on. f. Figure corresponds to 1951. g. Reaper binders only after 1937. Source: Ministere de l'Agriculture, Statistique Agricole (Retrospectifs 1930-1957) (Paris, 1959); and Statistique Agricole de La France and Statistique Agricole Annuelle (Paris: Imprimerie Nationale), various issues of each. water. Mills and threshers were the most common users of steam power in the late nineteenth and early twentieth century in both Europe and the United States. Mechanical threshing based on human power, but especially on hors- es, became widespread in the United States and Britain as early as 1830. By 1850 virtually all grain in the United States was threshed by large mechanical threshers, which went from farm to farm during the winter months. Rental markets were extensive. By 1852 the number of threshing machines in France had already reached nearly one-third of its peak 1929 level (see table 8), though they spread more slowly in Germany. Except for some animal-drawn primary tillage, stationary machines for power- intensive operations preceded all other forms of mechanization in Japan. In South Asia animals have long driven Persian wheels, sugarcane crushers, and oil crushers. Animals used in these operations are increas- ingly being replaced by diesel and electric engines. In India in 1972 the number of stationary engines for power-intensive operations was about twenty times that of tractors (see table 9). And in China (table 10) the number of threshers alone exceeded the combined total of tractors and power tillers, even in 1980. In all of Asia mechanical rice milling for large trade quantities had already been introduced in the late nineteenth Hans Binswanger 41 Table 9. Pattern of Farm Mechanization in India (thousands) Year l945 1951 1556 1961 1966 1972 Draft animals 59,333 67,383 70,690 77,986 78,517 80,137 Persian wheel n.a. n.a. n.a. 600 680 638 OilPumps 12 83 123 230 471 1,558 Electric pumps 9 26 47 160 415 1,618 Tractors 5 9 21 31 54 148 Plows Bullock Wooden 27,306 31,796 36,142 38,372 39,880 39,294 Iron 487 931 1,376 2,298 3,521 5,359 Tractor n.a. n.a. n.a. n.a. n.a. 57 Other tillage implements Bullock n.a. n.a. n.a. n.a. 2,724 17,119 Tractor n.a. n.a. L.a. n.a. n.a. 111 Sugarcane crushers Power 9 21 23 33 45 87 Bullock 481 505 545 590 650 678 Oil extractors Above 5 seers, n.a. 243 66 78 74 40 Less than 5 seersa n.a. 20 2.12 172 159 76 Shellers Indigenous n.a. n.a. n.a. n.a. n.a. 175 Power n.a. n.a. n.a. n.a. n.a. 16 Threshers Indigenous n.a. n.a. n.a. n.a. 249 n.a. Power n.a. n.a. n.a. n.a. n.a. 207 Chaff cutters Rotary n.a. n.a. n.a. n.a. 4,729 n.a. Power n.a. n.a. n.a. n.a. n.a. 161 Transport Bullock 8,483 9,862 10,968 12,072 12,695 12,960 Tractor n.a. n.a. n.a. n.a. n.a. SS Seed drill / planter Bullock n.a. n.a. i.a. n.a. 1,135 4,047 Tractor n.a. n.a. n.a. n.a. n.a. 34 Sprayer/duster n.a. n.a. n.a. n.a. 211 413 n.a. Not available. a. A seer equals about 2.05 pounds. Source: Directorate of Economics and Statistics, Agricultural Situation in India (1976), p. 141; andCentral Statistical Organization, StatisticalAbstractoflndia 1975, pp. 57-61. century, usually based on steam and later on internal combustion en- gines. Smaller rice mills have swept across Asia since the 1950s; it is hard to find villages where rice is still pounded by hand. Thus mechanical milling is even more widespread than mechanical threshing. But where the green revolution raised wages and increased harvests (as in Indian Punjab, Philippines, and central Thailand), the small threshers were rapidly adopted once efficient designs were available. The new threshers are now also penetrating into other South Asian regions (Walker and 42 Research Observer 1, no. 1 (January 1986) Table 10. Patterns of Mechanization in China (thousands) Four-wheel Garden Farm tractors tractors Threshers Combines trucks 1957 n.a. n.a. n.a. 2 4 1962 55 n.a. n.a. 6 8 1965 73 4 110 7 11 1970 125 78 455 8 16 1975 345 599 1,553 13 40 1979 667 1,671 2,328 23 97 1981 790 2,030 n.a. n.a. n.a. Note: In 1980 there were 560 tractor trailers and 36,000 wheelbarrows with rubber tires. In 1979 draft animals, including young stock, totaled 94,591, broken down as follows: oxen, 52,411; cows used for draft, 558; water buffaloes, 18,377; horses, 11,145; donkeys, 7,473; mules, 4,023; and camels, 604. Sources: Ministry of Agriculture, Animal Husbandry, and Fisheries, Agricultural Yearbook of China, 1980 and China Academy of Agricultural Engineering. Kshirsagar 1981). As with earlier American and European experience, mills and threshers are usually rented. Generalization 9. The mechanization of power-intensive processing and pumping operations always precedes the mechanization of harvest- ing and crop husbandry operations and can be profitable at low wages. Land preparation. Unlike the power-intensive operations, land prepar- ation requires mobile sources of power, such as animals, tractors, or power tillers (hand tractors). Of all land preparation operations, primary tillage (breaking soil, often combined with turning its top layer) is the most power-intensive. It is also usually the first use of a new source of power. Investment in animal-drawn harrows occurs later and is usually much less than investment in plows. The widespread use of modern steel harrows in the United States was delayed until the 1880s, roughly fifty years after the massive shift to cast iron and steel plows. When tractors were introduced, they began to be used universally for primary tillage, while animals continued to be used for other types of soil preparation. Generalization 10. Primary tillage is one of the first operations to be mechanized when a new source of mobile power becomes available. Secondary tillage operations often continue to be performed by the old power source for a long period. Transport. Carrying loads is the earliest use of domesticated work animals, even preceding tillage. Shifts to animal-drawn sleds or carts follow, especially when marketed quantities increase. The cart and plow Hans Binswanger 43 are the basic farmer-owned implements of early animal-drawn mechani- zation (see table 11). When mechanical power- becomes available, it is soon used for farm- to-market transport. Early tractors had no tires and in the 1920s were rarely used for farm-to-market transport in the United States or Great Britain. Instead, mechanizing farmers bought both tractors and trucks at about the same time. That also happened in Mexico after 1960 (see table 12). For on-farm transport, American farmers continued to use horses well into the 1940s. In Asia, where farms are rarely big enough to support the purchase of a truck, farm-to-mrarket transport is increasingly done by hired trucks or tractors. Rubber tires have given tractors a strong comparative advantage in most forms of transport. Generalization 11. Transport, along with primary tillage, is one of the first uses of new sources of mobile power. Where distances are long, trucks rather than tractors are used for farn-to-market transport. Harvesting. Without machinery, harvesting is very labor-intensive. Different crops vary widely, however, in the type of labor required-that is, in their power- and control-intensity. Harvesting of root crops is probably the most power-intensive, althotugh it still requires significant Table 11. Pattern of Farm Mechanibation in Senegal (thousands) Year 1950 1955 19S9 1965 1970 1975 Animals Horses n.a. n.a. 98 160 200 210 Asses n.a. n.a. 78 147 185 196 Work oxen n.a. n.a. 1 1 2 8 Tractors n.a. n.a. 0.2 n.a. 0.5 0.4 Plows 0.1 0.6 2 7 8 39 Hoes 0.8 2 4 36 102 219 Harvestors/threshers n.a. n.a. 0.1 0.1 0.3 0.1 Carts Horse n.a. n.a. 1 1t8 23 38 Ox 0.3 3 6 5 5 14 Ass n.a. n.a. 0 0.3 6 14 Sowing machines 11 31 46 94 120 189 Groundnut lifters n.a. n.a. 0 6 18 42 n.a. Not available. a. Figure corresponds to 1960. Sources: Tractors and harvester/threshers: Food and Agriculture Organization, Production Yearbook, various issues, Work oxen, 1959-65: World Bank, Senegal: Tradition, Diversification, and Economic Developmfent (Washington, D.C., 1974). Others: up to 1955, Y. Marie-Saite, La Culture attele au Senegal (Dakar: Direction de l'amenagernent, 1963); 1959 onward: Ministere du Plan et de l'Industrie, Situation Economique de Senegal, various issues. 44 Research Observer 1, no. 1 (January 1986) control. At the other extreme, cotton, fruit, and vegetables require inten- sive control: in the case of apples, the threat of damage is so large that their harvesting has still not been successfully mechanized. During the nineteenth century many attempts were made to develop harvesting machinery in Europe and the United States (van Bath 1960; USDA 1940). Reapers for small grains became widely adopted in North America after 1850, with grass mowers for the dairy regions following shortly afterward. But in France and Germany these machines did not make a substantial impact until 1890 or 1900. This time lag cannot be explained by lack of engineering knowledge in Europe: the same coun- tries were using mechanical threshers for virtually all their crops and seed drills had already been widely adopted. The difference was that labor was more abundant in Europe, farms were smaller, and the har- vesting machines were therefore not profitable. The United States started moving from reapers to wheat binders in the 1870s and to corn binders in the 1880s. These changes coincided with, or even preceded, the development of modern harrowing technology: spring tooth harrows and disk harrows. European farmers did not adopt reaper-binders until the first decade of the twentieth century (Bogart Table 12. Pattern of Farm Mechanization in Mexico (thousands) Year 1930 1940 1950 1960 1970 Numberof holdings 858 1,234 1,383 1,365 1,020 Work animals n.a. n.a. 3,920 3,476 4,150 Engines, n.a. 9 14 18 47 Electric motors n.a. n.a. n.a. n.a. 28 Tractors 4 5 23 55 91 Plows Indigenous 904 925 1,135 1,100 916 Iron n.a. 720 1,128 1,286 1,301 Harrows and cultivators n.a. 102 240 308 387 Threshers (fixed) 4b 2b 3b 5 3 Shellers Engine n.a. 2 3 5 13 Hand n.a. 4 5 9 18 Forage choppers n.a. 2 3 6 6 Carts 106 131 175 211 161 Trucks 4 6 18 40 104 Seed drills 26 27 60 93 122 Mowers/reapers 8 5 8 10 12 Hay balers n.a. 2 3 5 12 Combines n.a. n.a. n.a. 4 7 n.a. Not available. a. Fixed and movable engines. b. May include some combines. Source: Direcci6n General de Estadfstica, Censos Agricola: Ganadero y Ejidal, decennial. Hans Binswanger 45 1942). In Japan reaper-binders had a perceptible impact only after 1967, almost a hundred years after the United States and a good thirty years after Japan started mechanizing pumping, threshing, and winnowing in earnest. Again, technological ineptitude in Japan cannot have been the cause for such long delays. Practical development of horse-drawn harvesting combines started in the 1860s in California, where labor was extremely scarce. By the 1880s combines drawn by between twenty-four and forty horses reaped ten to fifteen hectares a day in California. In the 1890s combines drawn by steam tractors had a capacity of up to twenty hectares of wheat a day (van Bath 1960; USDA 1960), but combines did not spread beyond Califor- nia until 1914. They did not appear in Great Britain until 1928, nor in most of continental Europe until 1935, and not in Japan until about 1970. At each level of mechanization, machines for harvesting maize have tended to lag a few years behind those for small grains. Hay harvesting equipment, horse rakes, and tedders became important during the Amer- ican Civil War of the 1860s and remained so until World War II. In France in 1892, hay-raking machines had not reached 10 percent of their 1955 peak number. Hay loaders became widely used in the United States after 1880 but did not spread in continental Europe until after World War II, only to be quickly replaced by hay balers and other more sophisticated machines. Most of the animal-drawn harvesting machines derived their power from horses. Oxen could not be used successfully because sufficient power could be generated only at the higher speed of horses. The demise of the oxen in American and European agriculture was largely the result of their inability to work harvesting machines. When tractors became available, they could pull harvesting machines with only minor modifica- tions. Nevertheless, horses did not lose ELheir comparative advantage, even in harvesting, for some years. Generalization 12. Because mechanization of harvesting is directly dependent on labor costs, it is rarely profitable in low-wage countries. The higher the control intensity of the operation, the higher must labor costs be to warrant using a machine. Crop husbandry. Weeding and cleaning of crops, fields, and orchards are control-intensive operations. In animal systems, people go on weed- ing by hand long after the introduction of the plow and cart-until rising wages make herbicides profitable. Wages are so low in South Asia that, except for tea plantations, it is still cheaper to weed by hand than to use herbicides (Binswanger and Shetty 1977). Mechanical weeding between the rows with animals becomes feasible only when line seeding is practiced. Inter-row cultivation also tends to be performed by animals long after tractors are used for tillage. 46 Research Observer 1, no. 1 (January 1986) Generalization 13. Crop husbandry shifts to new sources of power only after tillage, transport, threshing, and seeding have done so. Seeding and planting. Animal and tractor-drawn machines are capable of greater precision than hand methods for only a few agricultural operations, especially seeding and planting. Mechanical means of seed- ing may lead to modest direct improvements in yields and may be attractive in land-scarce countries with intensive farming. The first seed drills were developed in China and Mesopotamia in the third millenium B.C. (van Bath 1960). The Mesopotamian drill required three workers- one to drive the oxen, one to put grain in the hopper, the third to hold the drill steady. It was apparently possible to use this instrument profit- ably only in the fertile soil of Mesopotamia, where high yields could be achieved and labor was abundant. The drill soon fell into oblivion. Between the sixteenth and nineteenth centuries, farmers in Europe tried to design better seed drills for small grains. Seed drills with me- chanical dribbling devices were commonly used in the United States in the 1860s and 1870s. In continental Europe their use started slightly later, became widespread in the late nineteenth century, and was fol- lowed a decade later by maize drills and cotton seeders. The use of seed drills similar to the Mesopotamian drill has been growing rapidly in India since 1966 (see table 9). In Senegal, where animal traction is primarily a post-1945 development, seed drills have become one of the most popular implements (see table 11). Improved seed drills, with mechanical dribbling of seeds, are becoming popular in South Asia and are one of the more successful machines in Mexico (see table 12). In all these cases it is not labor saving which leads to their success, but the improvement in yields, the saving of seed, and the ease of interculture. For most developed and developing countries for which data are available, the spread of seed drills is paralleled by the spread of inter-row cultivators or, at an earlier stage, simple animal-drawn hoes or blade harrows for interculture. Generalization 14. In labor-abundant countries, seeding of grains tends to be mechanized before grain harvesting, but the order is usually reversed when labor is scarce. Fertilizer and pesticide placement. Although fertilizer can be spread by hand, it produces higher yields if it is dispensed precisely. Thus, animal- drawn machines for spreading fertilizer were developed as fertilizer use increased. Since fertilizer was more intensively used in Europe in the interwar period, fertilizer distributors were common there. Large cart- mounted barrels for spreading liquid cow manure were also widely used, as were elaborate pumping systems. By contrast, in land-abundant North America the use of liquid manure was virtually unknown. Hans Binswanger 47 Liquid pesticides cannot be applied without at least a hand pump. Even for pesticides in dust form, mechanized dusters achieve higher precision and reduce waste. Sprayers were developed at the same time as pesticides. In France, for example, spraying carts were widely used in vineyards in 1929. But in Japan hand-carried power sprayers for rice and other crops became popular only with the development of a much wider range of pesticides after World War II. Such power sprayers are now used throughout Asia, often hired on a contract basis. Generalization 15. The growth of hand and power sprayers is driven by the availability and use of pesticides aned is widespread even at very low wages. Higher wages lead to the use of larger sprayers which may be animal- or tractor-drawn. Assessment. The selective use of new power sources (particularly trac- tors) for power-intensive operations has often been viewed as a sign of inefficiency. Since a farmer makes a huge investment in a tractor, why not use it for all operations? U.S. studies carried out in the 1920s and 1930s show clearly that there is nothing inefficient: in the selective use of tractors for power-intensive operations. As long as agricultural wages were rela- tively low, large farmers found it more efficient to maintain a tractor and truck along with some horses. Horses did virtually all the jobs for which power was not the overriding input. Each power source specialized in the tasks for which it had the greatest comparative advantage. Tractors were mainly used for tillage and as power sources for stationary machines such as threshers, saws, silo fillers, and choppers. The same pattern of tractor use was common in Europe until about 1960 and is now common in South Asia, Southeast Asia, and China. 'rhe only differences are that direct power takeoff has replaced the belt and pulley and that tractors are now more frequently used for transport. Although modern tractors are more efficient than prewar ones, wages in Asia are much lower than in the prewar United States. Asian countries are therefore likely to make continued use of animals along with tractors, until rising wages make the animals' drivers, and thus the animals, too expensive. The Speed of During the twentieth century and especially since World War II, the Mechanization speed with which farmers adopt new machines has quickened. In Japan, for example, the number of motors, threshers, and hullers increased five- to tenfold between 1939 and 1955. Power tillers grew from less than 100,000 to more than 3 million between 1955 and 1975. Binders, com- bines, and rice transplanters spread even more rapidly in the 1970s. Such spurts are not unique to Japan. Continental Europe experienced many similar surges in 1955-70. In Taiwan, after 1968, it took only about a decade to shift primary tillage completely to power tillers. 48 Research Observer 1, no. 1 (January 1986) Central Thailand, in about fifteen years from the late 1960s, went over entirely to tractor tillage with locally designed power tillers and small four-wheel tractors. The adoption of small paddy mills in Southeast Asia was also very quick. Rapid change has not been confined to the twentieth century. In the United States threshers spread enormously from 1830 to 1850 once satis- factory designs were available (USDA 1940). The same seems to have happened in Europe. However, the speed of change cannot be captured by historical statistics that focus on power sources at a national level; it is observed in operations at a regional level. For example, the growth of tractors in the United States was spread over a fifty-year period with occasional spurts, but once tractors became available, they took over primary tillage within a much shorter time. Their further growth in- volved shifting extra operations from horses to tractors. Today few farms in the Indian Punjab plow land with animals, thresh wheat by hand, or use Persian wheels. This is only fifteen years after tractors, threshers, and pumps became an important factor in Punjab farming. The aggregate Indian data in table 9 hide these facts because animals continue to be used for other work, even in Punjab, and because many other regions have not yet shifted massively to tractor plowing or me- chanical threshing. In the case of threshers, adoption cycles have always been fairly short. Once locally adapted designs are available, the cost advantage seems to be overwhelming. For other machines, the explosive growth of the post- war period must be understood as a response to unprecedented rates of growth in agricultural wages. This section therefore concludes with two generalizations. Generalization 16. Where cost advantages are large or change rapidly, individual operations are mechanized very quickly. Within smaller re- gions, adoption periods are often as short as ten to fifteen years. This speed of adoption implies directly: Generalization 17. In market economies, supply bottlenecks in the production, distribution, and servicing of machines are rarely a major cause of their slow adoption. The previous sections imply: The Process of Mechanical Generalization 18. Neither power sources nor the basic engineering Invention solutions for particular farming tasks are very sensitive to variations in soil and weather. However, the power sources must be embodied in specific machines and the basic engineering solutions adapted to different Hans Binswanger 49 environments. Both agroclimatic factors (;oil, terrain, rainfall) and econ- omic factors (land, labor, capital, farm size, and materials available) require adaptive innovation on a scale that has been vastly underesti- mated. The extent of adaptive innovation required is best illustrated by pa- tent statistics from the United States, which Evenson (1982) has put together on a regional basis. For other developed countries and the developing world, the investigation of innovation in agricultural machi- nery has been much less systematic. Nevertheless, case studies, actual observation, and discussions with engineers and machinery manufactur- ers reveal very similar trends. For example, the emergence of a diversi- fied machinery industry from small shops is well known for the Indian Punjab. The Thai power tiller industry has been well documented by Wattanutchariya (1981). Innovations in ithe Philippines have been de- scribed by Mikkelsen and Langam (1981). Generalization 19. In the early phases of machinery invention, subin- vention and adaptation are done almost exclusively by small manufac- turers or workshops, working closely with farmers. Public sector re- search has contributed little to machinery development, but more to education. The contribution of large corporations increases over time, but continues to be largest in the area of engineering operations. The reasons for these patterns are threefold: • In sharp contrast to biological innovation, for which public funding is crucial, private makers of machinery can capture the gains from their innovation by selling machines. The innovator's rights are more protected, the more developed the patent system is and the better it is enforced. (For a full discussion of alternative patent systems, see Evenson 1982.) * Because many adaptive discoveries are specific to particular regions, farmers, blacksmiths, and small firms have an important advantage over public research institutes or large corporations. * Unlike biological or chemical inventions, mechanical innovation does not usually depend on the university-acquired skills of chemistry, genetics, or statistics. Mechanically minded individuals with little formal education are thus not at a disadvantage. Metallurgical and mathematical knowledge become more important when the design of complex or self-propelled machines needs to be refined; at this stage, engineering staffs of corporations are more effective. These characteristics produce the industrial structure that Evenson described for the United States, but which has occurred wherever me- chanization proceeded rapidly. 50 Research Observer 1, no. 1 (January 1986) Generalization 20. At the beginning of a mechanical spurt, many small firms enter with alternative designs. The most successful either grow or are bought up by larger firms; small producers disappear or revert to servicing machinery.9 Evenson also shows evidence for the following: Generalization 21. Inventive work on a particular operation often precedes by decades the widespread use of machinery. It reaches a peak during the initial adoption cycle, when derivative invention, refinements, and adaptation to different environments are required. The longest lags between inventive activity and adoption of machines occur when inventors are trying to mechanize operations for which there is as yet little demand. Inventive work on seed drills and harvesting in early nineteenth-century Europe provides one of the clearest examples of this lag, but the same phenomenon occurs in many developing countries, especially in the machinery parks of publicly funded programs in agricul- tural engineering. Developing countries have a wide range of machine processes and Policy basic engineering solutions from which to choose. They thus seldom Implications need to solve basic engineering problems for operations they want to for Developing mechanize. Their bigger task is to foster a healthy climate for the Countries reinvention, adaptation, and straightforward copying of existing designs. By the very nature of agriculture, this process must be decentralized and carried out separately for different countries or agroclimatic regions. In choosing how to mechanize, different countries should expect dif- ferent results from mechanization. In general, mechanization will contri- bute little to growth in countries without a land frontier and with densely populated farmland-such as Bangladesh, most of India, and China. Given the fact that a high proportion of the work forces in these countries is still engaged in farming, even very rapid growth in the rest of the economy will not lead to rapid wage increases. Labor scarcity cannot be expected to arise from nonagricultural growth in the near future as a driving force for mechanization. By contrast, labor market pressures are quite different in the middle-income countries of South America, such as Brazil. Developing countries are also faced with the question of which opera- tions to mechanize next. Local farmers tend to be the best judges. Outsiders often know too little about local farming-or worse, they may try to solve perceived problems with solutions imported from their home environment. The patterns of mechanization discussed in this paper may be helpful in anticipating future developments somewhat better. Hans Binswanger 51 As for government intervention, the historical record is quilte clear. Mechanization in today's industrial countries did not depend on direct government involvement in machinery development, production, tech- nology choice, or finance.'0 The same is true of the most successful experiences in the developing- world, such as the mechanization of mill- ing, pumping, or harvest processing. Once economic conditions have produced effective demand for machinery, private firms have responded rapidly in the developed world. On a smaller scale, responses have been equally rapid in economies as diverse as Thailand and Mexico. In the developed world, government policy toward mechanization has been confined to patent laws for enforcing innovator's rights and en- couraging disclosure; testing of machinery, support of standardization measures, and dissemination of information; and support of agricultural engineering education and some university-based research. These are clearly appropriate interventions. Unlike the case of agricultural re- search, it is difficult to make a case for any further intervention on the grounds of economic welfare. Where governments have intervened more, they have either had little success, as in numerous publicly funded research efforts, or they have made wrong or controversial choices." Pakistan not only subsidized big tractors, but also prohibited imports of all but a few brands,'2 Its trade policies restricted imports of many smaller machines and implements and made it almost impossible for sma ll innogvating firms to import foreign designs for local adaptation. The contrast between Pakistan and Thailand could hardly be sharper. Thailand's laissez-faire policy has resulted in the development of indigenous power tillers and small trac- tors, a wider choice of machinery, and few adverse social consequences. Brazil's approach has been to encourage mechanization by subsidizing loans for buying machinery. Because interest rates were often lower than inflation, real borrowing costs were negative. There is ample evidence that credit subsidies, especially for tractors and other big machines, were obtained mainly by large farms and lat:ifundia. They gained a cost advantage over small farmers and expanded at their expense. This proc- ess has been documented for Pakistan in t-wo studies over a fifteen-year period (McInerney and Donaldson 1975; Lockwood 1981)" and for Bra- zil (Sanders and Ruttan 1978). Furthermore, the subsidies often favored the better-off regions: Sanders and Ruttan show that they have increased imbalances between Sao Paolo province and the poor Northeast. In China, subsidized credit and energy have undoubtedly benefited the richer regions over the poorer ones, where investment in machines is still very limited. Generalization 22. Subsidies to mechanization tend to have little effect on output, while damaging employment. They also tend to favor larger farms and richer regions. 52 Research Observer 1, no. 1 (January 1986) By contrast, where mechanization has occurred spontaneously in res- ponse to rising demand for labor in the rest of the economy, the social consequences have rarely been harmful. Released workers have been redeployed in jobs where they were more productive and received higher wages, and others have ended up farming larger areas.14 Even when mechanization is spontaneous, however, governments may sometimes be faced with severe distributional dilemmas. For example, it appears that combine harvesters would be modestly profitable in the Indian Punjab, but their introduction would displace many migrant wQrkers frQm 'the poorest regions (Laxminarayan and others 1981). In the absence of rapid employment growth elsewhere, the Indian government may be justified in banning combines. Mechanical rice milling has been controversial in Indonesia (Timmer 1974, Collier 1974) and is now penetrating Bangla- desh. It reduces the demand for women workers, who-because of social customs-have few employment options. On efficiency grounds, milling machines are clearly warranted; on equity grounds, their effects may he harmful. Any analysis of the potential contribution of mechanization to growth Implications applies as much to a socialist as to a capitalist economy. In both, payoffs for China depend on the opportunity costs of labor, land, and capital. Despite numerous policies and programs to influence mechanization in China over the past thirty years, the results are surprisingly similar to those of other labor-abundant developing countries: a limited spread of tractors, which are used for tillage and transport, plus substantial mechanization of power-intensive operations such as harvest processing and pumping. Since China has abundant labor but a shortage of land, the policy implications for mechanization do not differ much from those for simi- larly endowed economies. Equity issues, however, may differ a little in China. In principle, when a brigade or production team invests in a machine, all its members save on the work and share in the returns from agricultural production. They should therefore be able to mitigate the potential for distributional ine- quities within the commune and judge mechanization primarily in terms of profit. This might help to explain the complementarity between ani- mal and mechanical power in Chinese agriculture found by Ramaswamy (1981). Interregional equity is more complicated. As long as migration is restricted in China, labor cannot be redeployed from slow-growing to fast-growing regions. The fast-growing regions may therefore experience increased demand for labor and find it necessary and profitable to me- chanize. The alternative-allowing migration to solve regional labor scarcities-may be better than mechanization. Migrants from poorer regions could share the faster growth of the richer ones, and scarce Hans Binswanger 53 capital could be used for investments other than machines. Current Chinese policy emphasizes "sideline activities," the redistribution of in- dustrial and service activity to rural areas to overcome interregional problems of income distribution. Although such decentralization is cer- tainly desirable and necessary, locational and natural disadvantages in many regions limit how much can be achieved. Abstract The article provides a detailed comparative historical review of the patterns of agricultural mechanization by operation. The first section reemphasizes the major conclusions of the induced innovation literature, and the second section shows the remarkable similarity in the early mechanization experiences of the developed and developing countries. New mechanical power sources were first used on power-intensive operations such as processing, pumping, transport, and tillage, while mechanization of contro]i-intensive operations came much later and usually in association with high wages. Finally, an investigation of the process of agricultural machinery innovation shows that the public sector and corporate research has contributed little to machinery invention. Machinery innovation has generally been the domain of small manufacturers, with corporations becoming significant in later stages of product development and engineering optimization. Notes 1. Induced innovation processes can also be documented in the developing world. Mechanization in central Thailand, one of the most successful cases (discussed in detail in World Bank 1983), appears to have been clearly induced by increasing scarcity of labor. 2. For evidence that tractors have no direct effect on yield in South Asia, see Binswanger (1978). 3. Final demand is said to be elastic or inelastic according to whether an increase in quantity supplied leads to a small drop (elastic) or a large drop (inelastic) in the price received. 4. For a discussion and estimates of temporary migration for harvest work, see Laxmi- narayan and others (1981). s. For a thorough discussion of this issue, see Binswanger and Rosenzweig (1982). That discussion distinguishes carefully between the effects of operational holding size and ownership holding size on costs of capital and labor. Here we assume that the operational and ownership holdings are closely related. 6. For careful investigation of the impact of scale on machinery adoption in nineteenth- century United States and Britain, see David (1975). 7. This is so despite the well-publicized growth of tractor sizes and harvesting machines in the United States. That growth occurred in response to the growth in farm size and in the number of tasks performed by tractors; it was an indirect response to the unprecedented wage increases of the past forty years. 8. For the data on which this section is based, see Binswanger (1984). 9. Switzerland, for example, had at least five producers of tractors in 1950, none of which survived. 10. Not included here are general policies, such as agricultural price policy, which have side effects on mechanization but affect all agricultural investments. 11. A good case in the developed world is the invention of the tomato harvester in California. For a recent summary of the controversy, see de Janvry, LeVeen, and Runstein (1980). 12. The brand choices had usually been made under donor pressure rather than as conscious economic choices. Brands from different ccuntries were added whenever govern- 54 Researcb Observer 1, no. 1 (January 1986) ments donated or helped finance tractors. Several were later dropped when aid stopped (Lockwood 1981). 13. In Pakistan many farms grew very rapidly as a result of tenant eviction, purchases, additional renting of land, and a modest amount of reclamation. 14. It must be recognized, however, that wages might have risen even faster in the absence of mechanization. Binswanger, Hans P. 1978. The Economics of Tractors in South Asia: An Analytical Review. References New York: Agricultural Development Council; and Hyderabad, India: International Crops Research Institute for the Semi-Arid Tropics. . 1984. Agricultural Mechanization: A Comparative Historical Perspective. World Bank Staff Working Paper no. 673. Washington, D.C. Binswanger, Hans P., and M. R. Rosenzweig. 1982. "Productive Relations in Agriculture." Research Program in Development Studies, Discussion Paper no. 105. Princeton, N.J.: Princeton University, Woodrow Wilson School. Binswanger, Hans P., and Vernon W. Ruttan. 1978. Induced Innovation: Technology, Institutions, and Development. Baltimore, Md.: Johns Hopkins University Press. Bogart, Ernest. 1942. Economic History of Europe. London: Longmans, Green. Collier, William L., Jusuf Coulter, Sinarhadi, and Robert D'A. Shaw. 1974. "A Comment." Bulletin of Indonesian Economic Studies. 10, no. 1 (March):106-20. David, Paul A. 1975. Technical Choice, Innovation and Economic Growth. Cambridge: Cambridge University Press. de Janvry, Alain, Philip LeVeen, and David Runstein. 1980. "Mechanization in California Agriculture: The Case of Canning Tomatoes." Berkeley: University of California, Department of Agricultural Economics. Processed. Evenson, Robert E. 1982. "Government Policy and Technological Progress in U.S. Agricul- ture." In Nelson Richard, ed. Government Support of Technological Progress: A Cross Industry Analysis. New York: Pergamon Press. Gardezi, J., A. Rauf, M. Munir, K. Altaf, Q. Mohy-ud-sin, and B. Lockwood. 1979. "A Study of Mechanical and Traditional Wheat Threshing in Multan District, Punjab, Pakistan: Some Preliminary Results." Paper presented at the Workshop on Consequences of Small Farm Mechanization on Rural Employment, Incomes and Production in Selected Countries of Asia, International Rice Research Institute, Los Banos, Philippines, October 1-4. Gupta, G. P., and K. K. Shangari. 1979. Agricultural Development in Punjab, 1952-53 to 1976-77. Delhi: University of Delhi, Agricultural Economics Research Center. Hayami, Yujiro, and Vernon W. Ruttan. 1971. Agricultural Development: An International Perspective. Baltimore, Md.: Johns Hopkins University Press. Hayami, Yujiro, Masakatsu Akino, Jasahiko Shintani, and Yamada Saburo. 1975. A Century of Agricultural Growth in Japan. Minneapolis: University of Minnesota Press; and Tokyo: University of Tokyo Press. Liu, Xianzhou. 1962. "The Invention of Agricultural Machinery in Ancient China." Acta Agromechanica Sinica 5, no. 1:1-36; no. 2:1-48. Laxminarayan, H., H. P. Gupta, P. Rangaswamy, and R. P. S. Mali. 1981. "Impact of Harvest Combines on Labour-Use, Crop Pattern and Productivity." New Delhi: Agricole Publishing Academy. Lockwood, B. 1981. "Farm Mechanization in Pakistan: Policy and Practice." Paper presented at the Workshop on Consequences of Small Rice Farm Mechanization Project, International Rice Research Institute, Los Banios, Philippines, September. Mclnerney, John P., and Graham E. Donaldson. 1975. The Consequences of Farm Tractors in Pakistan. World Bank Staff Working Paper no. 210. Washington, D.C. Hans Binswanger 55 Mikkels8h, K, W., and N. N. Laagam. 1981. "Technology Change in the Philippine Agricultural Machinery Industry." Paper presented at the Workshop on Consequences of Small Rice Farmi MechahiizAtion Project, International Rice Research Institute, Los Bafios; Philippihes; Septembet Pingali, Prabhti, Yves Bigot, and Hans P. Binswanger. 1985. "Agricultural Mechanization and the Evolution of Farming Systems in Sub-Saharan Africa." World Bank, Agricultural Research Uinit Discussion Paper no. 40. Washington, D.C. Ramaswamy, N.S. i9g1. Report on Draught Animal Power as a Source of Renewable Eknergy. Rome: P1ood and Agricilture Organization. Sanders; John 1.; and Verion W. Ruttan. 1978. "Biased Choice of Technology in Brazilian Agriculture." In Hans P. Binswanger and Vernon W. Ruttan, Induced Innovation: Technology, Institutions and Development. Baltirnore, Md.: Johns Hopkins University Press. Timmer, W. Peter. 1974. "Choice of Technique in Rice Milling in Java: A Reply." Bulletin of Indonesian Economic Studies 10, no. 1 (March :121-26. U.S. Department of Agriculture (USDA). 1940, 1960. "Farmers in a Changing World." Yearbook of Agriculture. Washington, D.C.: Government Printing Office. van Bath, B. H. Slicker. 1960. "The Influence of Economic Conditions on the Development of Agricultural Tools and Machines in History." In J. L. Meij, ed. Mechanization in Agriculture. Amsterdam: North Holland. Walker, Thomas S., and K. G. Kshirsagar. 1981. "The Village Level Impact of Machine Threshing and Implications for Technology Development in Semi-Arid Tropical India." Economics Program Progress Report no. 27. Hydtrabad: International Crops Research Institute for the Semi-Arid Tropics, November. Wattanutchariya, S. 1981. "Economic Analaysis of Farm Machinery Industry and Tractor Contractor Business in Thailand." Paper presented at the Workshop on Consequences of Small Rice Farm Mechanization Project, International Rice Research Institute, Los Baiios, Philippines, September. World Bank. 1983. Thailand: Rural Growth and Employment. Washington, D.C. 56 Research Observer 1, no.1 (January 1986) AID IN THE DEVELOPMENT PROCESS Anne 0. Krueger he history of economic thought on foreign aid is somewhat peculiar. The phenomenal success of the Marshall Plan in the late 1940s and 1950s led many to believe that similar transfers to developing countries would permit their comparably spectacular trans- formation. That belief had two intellectual underpinnings. The first was the Harrod-Domar model, which extended the Keynesian emphasis on investment to include its capacity-increasing effects. The second was economists' emphasis on physical capital and the view that shortage of capital largely accounted for the poverty of developing countries. While it was recognized that many other factors would be needed to achieve satisfactory growth, the critical bottleneck was believed to be the shor- tage of investment because of low savings rates. The role for foreign aid followed logically from this analysis. If invest- ment was the bottleneck, the return on additional investment in develop- ing countries would be higher than in developed countries. In the late 1940s and early 1950s, it seemed inappropriate to assume that the inter- national capital market functioned smoothly. Hence few could doubt that, if capital were to flow to poor countries, it would of necessity be official capital-which was equated with foreign aid. Since the 1950s, understanding of development has deepened enor- mously. The optimism inherent in the view that capital was the main This essay examines the role of aid in development. It focuses on particular issues and is by no means a complete survey of the literature. In the course of writing, I have benefited from valuable comments and suggestions from many people, including Jagdish Bhagwati, Hollis Chenery, Gregory Ingram, Uma Lele, Ronald McKinnon, Allan Meltzer, Constantine Michalopoulos, Sherman Robinson, Vernon Ruttan, T. W. Schultz, T. N. Srinivasan, and Ernest Stern. Needless to say, on a subject as broad as this, none of them would associate himself with everything stated herein. I am also grateful to Suan Ying for valuable research assistance. 57 thing lacking has been replaced by an appreciation of the complexity of development. There has been increased recognition of the need for hu- man capital formation, the importance of developing well-functioning markets, the challenges of agricultural development, the role of trade, the interactions between economic policy and politics, and so on. Cur- rent thinking would place equal stress on resource accumulation in both a quantitative and qualitative sense, and on increased efficiency of re- source use in the economic, managerial, and engineering senses of the term. Achieving these goals would depend on several factors, including the incentives facing individuals for the accumulation and efficient use of resources, the development of well-functioning markets, efficient govern- mental provision of infrastructural services, and institutional develop- ment in both the private and public sectors. Despite this advance in understanding, tihere has not been a systematic reexamination of aid and its role in development. This essay will exam- ine that question. It concentrates on the tole of aid in facilitating devel- opment and growth, so its emphasis is on the macroeconomic aspects of aid. Much that has been learned about specific sectors -ducation, pow- er and irrigation, agricultural research and extension, and so on-is not covered here.' Throughout, the focus is on the econornic effects of aid and the ways in which its utilization affects the growth rates of developing countries. Despite that focus, it is important to recognize that much foreign assist- ance has motives that may have little tc do directly with accelerating economic growth. Donors may wish to enhance the military prowess of a recipient country, to promote their commercial interests, to support a friendly government in power, and to acquire goodwill now in the expectation that it will be politically valuable later. In some of these cases, diplomatic realities may preclude using the donor's resources for developmental purposes. In others, use of aid in support of development could be consistent with the donor's objectives. Hence, although the analysis will be conducted throughout in terms of the impact of aid on the recipient's growth prospects, it should be borne in mind that the objectives of aid have often been military or political, and those objec- tives may or may not have been consistent with using the resources to enhance growth. Ergo, the developmental impact of aid can be substan- tially reduced. As background for the analysis of the potential and actual role of aid in development, the first section below presents the various concepts and definitions of aid that are used; the second concerns the development process as it is at present understood. The following three sections examine in turn the rationale for aid, "aid effectiveness"-that is, the degree to which different types of aid are conducive to accelerating development-and some criticisms of foreign aid. A final section sum- marizes the lessons that have been learned about what aid works and 58 Research Observer 1, no. 1 (January 1986) what does not, and outlines some issues on which research could shed further light. An official flow from one country to another takes place whenever the Capital and government of the originating country provides command over resources Official Flows, to the other without a current commercial quid pro quo. It might obtain ODA, and goodwill, rights to military bases, or political support, but it does not Concessional receive direct payment simultaneously. Aid An official flow could consist, at one extreme, of a loan at near- market rates of interest. At the other, it could be an outright grant of convertible currency. Between these extremes, it could be a loan at a below-market interest rate or with a grace or maturity period longer than that commercially obtainable. In these cases, the value of the flow can be calculated as the difference between the amount received and the present value of the repayment stream (see Pincus 1963 or Schmidt 1964). Thus, the value of an untied grant would be its face value;2 the value of a loan at commercial interest rates would be zero. The grant component of an official flow is the percentage by which the present value of the repayment stream falls short of the current value of the flow. Conceptually, this measure is well defined when interest rates are equalized throughout the world. In practice, the Development Assistance Committee of the OECD defines the grant element as the excess of the loan's (or grant's) value over the present values of repayments where, by convention, present values are calculated using a 10 percent rate of interest, and the costs of restricted procurement or other side conditions are not taken into account. Over the past five years, slightly more than 60 percent of net official flows have been official development assistance (ODA). The rest were official and officially supported export credits (20 percent) and other flows at near-commercial rates, including nonconcessional lending of the multilateral organizations (10 percent). ODA is defined as a flow with a grant element greater than 25 percent, the purpose of which is at least loosely related to economic development. In this essay, ODA will be used synonymously with foreign aid and foreign assistance. Official development assistance can take several forms. It might cons- ist of loans at terms more favorable than available commercially.3 It might be food aid, or a grant or loan that has to be spent in the donor's home market (tied aid). In each case, the grant component can be calculated-although in practice the fair market value of tied aid, food shipments, Russian-built steel plants, and the like is usually not estimat- ed; instead, the donor's accounting of the costs of aid is used as a basis for valuation. In the case of the United States, the value of food assist- ance was substantially distorted when its shipments of commodities under PL 480 were valued at the farm support prices, which have at Anne 0. Krueger 59 times been well above world prices.' Beyond these forms of ODA, some aid is given by nongovernmental organizations (NGOs).s Defining aid as capital flows may result in a serious omission. Many observers believe that a very important component of aid is knowledge: education and training, the transfer of technical and institutional know- how (including interactions of donor and recipient that may improve policies and the functioning of domestic rnarkets), and so on. Sometimes this knowledge component may be embodied in aid; it should not be overlooked just because it cannot be quantified. The Most developing countries have factor payments that are a small Development fraction of their counterparts in developed countries. The average daily Process wage of an Indian factory worker in 1979, for example, was about $3.35 while that of an American factory worker was $53.20. The traditional trade-theoretic presumption, that factor proportions and a relatively abundant Indian supply of labor might account for this difference, is not-at least superficially-borne out by the facts. In India neither the absolute reward to skills (human capital) nor that to physical capital is evidently higher than in developed countries. While explanations such as the differing quality of labor undoubtedly account for part of the income differential., they cannot explain it all (see Krueger 1968 for a fuller discussion of this issue). In the early postwar years, the magnitude of the income differences between developed and developing countries led most development economists to the conclusion that conventional economics somehow failed to apply to developing countries.6 It seemed self-evident that markets had not functioned ef- ficiently, given the extreme poverty and slow growth in most poor countries. From this observation, and the view that capital shortage was the chief bottleneck to development, two propositions followed: (1) gov- ernments had a responsibility to intervene to ensure that the capital stock would grow and be appropriately allocated; and (2) although additional investment would have a high rate of return, it would be constrained by low domestic savings rates (as a result of very low incomes)-without foreign assistance, growth would be held back. The belief that markets failed to work in developing countries might have led to a search for prescriptions to improve them. Instead, consist- ent with the Harrod-Domar model and shortage-of-capital explanation of underdevelopment, this perception led ao an almost exclusive emphas- is on increasing capital as the way to raise incomes. To be sure, five-year plans addressed such issues as taxation, education, land tenure, and family planning, but they focused on planned increases in output and investment by economic activity. Reflecting this thrust, much develop- ment research centered on techniques for estimating output levels, input- output relations, and investment by sector. 60 Research Observer 1, no. 1 (January 1986) By the end of the 1950s, however, perceptions were changing. Three major, apparently independent, new lines emerged. First, evidence began mounting that markets were functioning substantially better than had earlier been appreciated. Second, Schultz's (1961) pioneering work on the importance of human capital was followed by growing evidence on the significance of the quality of human resources, including education, health, and nutrition. Third, experience with development led many to point to foreign exchange shortages as a critical bottleneck to develop- ment. The major breakthrough with respect to the functioning of markets was also forcefully put forward by Schultz (1964). Schultz's hypothesis, that small farmers were typically maximizers who responded to incen- tives, was greeted with great skepticism at the time (see, for example, the review by Balogh 1964). Nonetheless, it offered a testable hypothesis that was examined by others whose findings reinforced the original results. Although the implications for development-that behavior will not alter until incentives are changed-were recognized only slowly, the reassertion of the vital role of individual behavior and its determinants was in the longer run crucial to reassessing the role of government in development. Recognition that human capital and foreign exchange might be scarce helped to move development economists toward a general equilibrium view of development and away from unicausal theories. Although the addition of human capital and foreign exchange to the list of problems for developing countries was a big advance in appreciating the complexi- ty of growth, the phenomenon of poor countries was still attributed to a shortage of resources, which in turn implied that the appropriate remedy remained resource accumulation. Until Schultz's insistence on the ration- ality of individual actors was more fully appreciated, underlying think- ing on the role of government was not challenged.7 The foreign-exchange-shortage view is sufficiently important in rela- tion to analyzing the role of aid to receive more consideration. As a starting point, obtaining foreign exchange through concessional aid can be viewed as superior to earning it through exports, since it requires scarce resources to produce the goods that are exported, while the receipt of aid does not (see Johnson 1967 for a full exposition). From this basic line of analysis, aid came to be regarded as a transfer of resources. In the 1950s, however, many development economists thought the developing countries had little hope of expanding their export earnings (see the classic statement of this view by Prebisch 1950). This "elasticity pessimism," combined with the infant-industry argument, led to the prescription that developing countries would have to foster import sub- stitution if they wanted to develop. Simultaneously, the commodity price boom of 1950-51 broke sharply in 1953-54, when almost all the devel- Anne 0. Krueger 61 oping countries' export earnings were concentrated in a few primary commodities.8 The policy response was to intensify protection for balance of pay- ments reasons, often with little regard to the original objectives of industrialization. The alternative-adjusting incentives for exporting- was not adopted, partly because of elasticity pessimism. As a consequ- ence of (1) domestic inflation at fixed norninal exchange rates, (2) rising costs of inputs as imports were prohibited once domestic production started, and (3) sharp increases in demand for imports (because of the import intensity of import substitution; see Diaz Alejandro 1965), the growth of export earnings was typically slow, making elasticity pessi- mism a self-fulfilling prophecy. Many development plans proved overambitious precisely because the foreign exchange bottleneck forced more import restraint than had been anticipated, and thus checked economic growth. In that context, the two-gap model was developed by Bruno and Chenery (1962) and elabor- ated in Chenery and Strout (1966). Thi:s model had three potentially binding constraints on growth: a savings constraint (which might limit investment), a foreign exchange constraint (which might limit investment because of the high import content of investment), and an absorption constraint (which set an upper limit for the rate of growth). As an empirical assertion, the foreign exchange constraint appeared binding at low levels of income. In these circumstances, foreign aid was doubly powerful. It not only permitted higher investment via the transfer of resources, but by relaxing the foreign exchange constraint it also allowed the utilization of domes- tic savings. The model had its critics, notably Findlay (1971) and Mc- Kinnon (1964), who highlighted the neglect of price (or incentive) res- ponsiveness inherent in the fixed-coefficients formulation.9 Nonetheless, the model's elegance and simplicity provided a powerful argument for foreign assistance, as it demonstrated the high marginal productivity of aid. Later conceptual developments-the responsiveness of small producers to incentives, the importance of human capital-might by themselves have challenged the dominance of the two-gap model. In fact, the chal- lenge came from evidence about the effectiveness of incentives. Both cross-country analysis and individual countries that altered incentives (of which Korea was perhaps the most notable; see Mason and others 1980) showed that exports were indeed responsive to incentives, and that their failure to grow was due more to normal supply responses to, for exam- ple, overvalued exchange rates, than to any failure of world demand or structural rigidity. Once it is recognized that individuals respond to incentives, and that "market failure" is the result of inappropriate incentives rather than of nonresponsiveness, the separateness of development economics as a field 62 Research Observer 1, no. 1 (January 1986) largely disappears (see Lal 1983). Instead, it becomes an applied field, in which the tools and insights of labor economics, agricultural economics, international economics, public finance, and other fields are addressed to the special questions and policy issues that arise in the context of devel- opment."0 In the conventional framework of economics, differentials in income per head must originate either in differences in the quality and quantity of productive factors available per worker or differences in the efficiency with which factors are employed. To be sure, technology can change over time. Once known, however, its application is seen as an economic phenomenon that entails human capital and appropriate incentives as well as blueprints (see Teece 1977). The modern view of development would therefore focus on both resource accumulation and improving the efficiency of resource use. This latter concern concentrates on improving markets-by institutional methods such as developing the capital mar- ket, by governmental provision of public goods such as agricultural research and extension services, and by removing government-imposed impediments to economic efficiency such as import licensing, currency overvaluation, and other regulations that drive a wedge between private and economic profitability. This is not to downgrade the importance of resource accumulation. All observers recognize that countries with low per capita incomes have little physical and human capital per head, and that increasing that capital is essential to growth. However, given the earlier emphasis on resource accumulation and the fact that many developing countries have dramatically increased savings rates without raising their growth rates, analysis of the efficiency issues has understandably come to the fore in recent years. The next question is whether there is an economic (as opposed to Why humanitarian or political) rationale for official development assistance, Official concessional or otherwise. Critics of aid have alleged that, if profitable Assistance? investments are available, private international capital markets will fi- nance them. Insofar as the motive for aid is humanitarian, that is not a criticism (as long as aid does not impair development). Nonetheless, the bigger questions are whether world economic efficiency can be enhanced by official flows at market terms, and whether it can be increased by an official flow that could not be financed at market terms. Clearly, a recipient's potential welfare could always be increased by a grant, whereas the donor's potential welfare might be reduced. The interesting questions are: (1) given optimal policies in recipient countries (those that maximize the economic welfare of their citizens), can foreign aid enhance worldwide efficiency and be in the economic self-interest, narrowly defined, of both donor and recipient? and (2) are there cir- Anne 0. Krueger 63 cumstances in which the answer to the first question is yes, but it would not pay the recipient to accept capital on commercial terms? These two questions are the subject of this section. It is useful to start by assuming, first, that economic efficiency exists in a cleveloping country, in the sense that, given resource constraints, domestic policies are consistent with efficient allocation of domestic resources; and, second, that aid is no- thing more than a capital flow. The end of this section will consider how to modify the analysis if aid is viewed as part of a bundled transfer of resources, including institution building, technical know-how, policy leverage, and capital. Throughout this section and the next, the question is the impact of foreign aid on the economic well-being of donor and recipient. Later, when attention turns to particular criticisms of aid, it must be borne in mind that the objectives of donors in giving aid are not always as high-minded or as economic as this discussion suggests, and that aid may fail to achieve greater economic efficiency either because recipients use it inappropriately or because donor objectives are incom- patible with the use of aid for economically efficient purposes. Question 1. Could official flows improve the welfare of both donor and recipient? The answer to this question hinges on whether imperfections in the private international capital market preclude it from equalizing (risk- adjusted) rates of return between donor and recipient. If returns were higher in developing countries, the welfare of both donor and recipient could be improved through official flows. The recipient could service its debt and nonetheless have a higher future income stream than would otherwise be possible. Simultaneously, the donor could obtain a rate of return equal to or greater than that obtainable on other assets. Consider, first, maximization of world welfare in the context of a simple two-factor neoclassical model. If the usual conditions for econ- omic efficiency are met within individual countries, in the absence of capital flows free trade might fail to equalize factor returns."1 Gains in world efficiency could then be achieved by developing a means for capital to flow from low-return to high-return countries. A first conclusion, therefore, is that official flows on commercial terms could not reduce world welfare. Indeed, they would normally be expect- ed to increase world welfare if private financing were not available. This conclusion is based on the assumption that the development assistance permits incremental investments with real returns at least as great as the return to the donor, but for present purposes that is subsumed by the assumption of policy optimality in borrowing countries, even after they receive aid. Another underlying assumption is that behavior in the bor- rowing country is not influenced-or at least not negatively influenced- by the receipt of official development assistance. Since there is some 64 Research Observer 1, no. 1 (January 1986) basis for thinking that official development assistance is more likely to influence behavior positively (see below), this assumption does not alter the analysis.12 Subject to these various qualifications, official flows on commercial terms would not reduce welfare but would increase it if they encouraged improved policies or if private markets failed to supply the capital instead, despite the higher real rates of return. The remaining question is thus why the private international capital market might fail in this way. In the 1940s and 1950s, thinking on aid hardly addressed this question because the capital markets had broken down in the 1930s and during World War II. By the 1970s, however, they were functioning-if anything by providing too much capital.13 After the debt crisis of the early 1980s, however, many analysts doubt whether private flows will resume, at least on a scale that could match the supply of profitable opportunities in developing countries. Some believe that there is a herd instinct among commercial bankers, who overlent in the 1970s and now will be irrationally unwilling to resume lending-even to countries that appear able to borrow and achieve the returns to service their debt (see Guttentag and Herring 1984, for an elaboration of that view). Although other types of private capital may in the future partly compensate for reduced commercial bank lending, they are unlikely to do so to any great extent. This view, if correct, would certainly suggest that official flows at commercial terms could play a larger role than in the 1970s and increase world economic efficiency. A second source of concern about developing countries' ability to use the private international capital market centers on the "debt overhang." For some developing countries, it is argued, current debts are so great relative to their existing income that increases in future earnings must be tapped to finance their existing obligations. Because foreigners correctly perceive this claim on future income, they will not lend even for new projects that would yield acceptable returns. This inability to insulate new claims from existing debt leaves countries in a vicious circle: they cannot restore creditworthiness without growth, and they cannot grow until creditworthiness is restored. The private capital market may thus fail despite the rational behavior of all participants (see Krueger forth- coming for a fuller discussion), and there is a strong analytical case for official assistance on commercial terms. Question 2. Why concessional aid? Even though international commercial capital is available and there are projects that would yield the necessary returns, are there circumst- ances in which it would not pay anyone in the recipient country (includ- ing the government) to undertake the project at commercial terms? Most analysts of aid have focused on two reasons why this might be the case: Anne 0. Krueger 65 the gestation or payout period of projects is too long, and the investor cannot fully capture the stream of returns. Many investments by developing countries do have long gestation or payout periods. One obvious example is in the area of education proj- ects, which occur over a period of a decade or more, followed by an even longer payout. In addition, some irnvestments (in roads, ports, and power stations, for example) have such large indivisibilities that returns are low in their early years. At a microeconomic level, if a project's repayments stream is not matched with the earnings stream it generates, an investor would not undertake it unless he had other earnings streams (or borrowing possi- bilities) to service his obligations in the project's early years. One could ask, of course, why the country could not refinance (or borrow more) to cover debt-servicing obligations in the years prior to high returns. In reality, the capital flows to developing countries have had a maturity structure of ten years or less (implicitly even less in the late 1970s, when the inflation premium in the nominal interest rate rose). At a macroecon- omic level, therefore, poor countries may be unable to borrow at or near commercial terms to finance much of their infrastructural investment, despite adequate real rates of return in the long run. A second, related difficulty is that many investments in the early stages of development entail public financing of activities that have significant externalities and for which user charges may not be appropriate. Roads are an example: with initially low utilization and negligible congestion costs, charging users is neither feasible (because collection costs exceed potential revenue) nor desirable (because the marginal cost of use is very low). And the government may not be able to finance the project on anything like commercial terms, even though incomes may be rising and increasing the tax base, because in a poor country only a fraction of incremental income is taxed. These considerations would appear to have most relevance for the very poorest countries: the ones with high levels of illiteracy, rudimen- tary transport and communications systems, and low savings rates. It must be emphasized that this case for concessionality is based on the productivity of investment in these countries and is additional to the case based on need or humanitarian motives. Question 3. What of aid as a "bundle"? This paper has so far treated aid purely as a capital flow. In practice, aid (and other capital, such as direct private investment) can be much more: donors may provide technical assistance with project design, know-how on organization and management, and so on. This assistance has undoubtedly been of great importance in many areas: the green revolution is perhaps the most visible and dramatic example. 66 Research Observer 1, no. 1 (January 1986) For present purposes, however, one aspect of the aid bundle, the policy dialogue, deserves special attention. Recipients may be influenced in their choice of macroeconomic policies in the course of the dialogue, which can take many forms: discussion and persuasion, information on policy effectiveness and techniques for reform, support for reform ef- forts, and "conditionality"-that is, aid is given only if certain policies are changed. Regardless of the influence used, foreign assistance could certainly become a means of speeding policy reform in developing coun- tries. And, as discussed in the previous section, the main lesson from the past two decades has been the need for an appropriate framework of policies and incentives. To the extent that donors have influence over recipients' policies, and are willing to use it, they may perform a function for which private capital markets seem ill-suited. Indeed, one might even imagine a world in which donor influence produced such improvement in world econom- ic efficiency that they made large-scale private investment an attractive proposition. In this light, aid and private capital might well be more complementary than substitutes. The previous section examined the economic rationale for aid, on the Aid assumption that it would be effectively utilized by the recipient. This Effectiveness section turns the issue round. If a developing country receives foreign aid, how can it best use it to increase its people's welfare? In assessing empirically the effectiveness of aid, two major difficulties arise. The first, already mentioned, is that the objectives of donors may not be purely economic. However, this section assumes that the donor's aim is to maximize the recipient's welfare. (See Little and Clifford 1965, chap. 3, for an excellent discussion of why such objectives might be consistent with political motives for aid.) The second difficulty is that aid is typically only a small part of a country's investment. Although the contribution of individual aid- financed projects can be evaluated, it is much harder to estimate aid's contribution to overall growth. Suppose that the cumulative lending of the World Bank has yielded a real rate of return of 10 percent over and above interest charges. According to one estimate, all that lending might have increased developing countries' incomes by less than 2 percent in 1980 (see Krueger 1983). If during 1950-80 aid from every donor had had the same productivity (a highly optimistic assumption), per capita incomes in developing countries might have been 20 percent above their actual levels. While 20 percent is significant, it does not make a poor country into a rich one. And, since the allocation of aid was skewed among countries (with some early recipients such as Brazil and Korea now regarded as middle-income countries), the impact for many of the others would have been smaller. This point is relevant to the extent that Anne 0. Krueger 67 some criticisms of aid seem to imply thac, if aid had been truly effective, it would have solved the development problem. Any fair assessment of aid must recognize that it has been small relative to the task of develop- ment. The question, then, is what uses of aid are likely to enhance economic efficiency and growth? Several related issues must be addressed. The first is that the government-to-government nature of aid suggests constraints on and guidelines for its effectiveness. T'he second is the fungibility of aid, which again highlights the importance of the policy environment. A third is whether and how aid can be effective if policies are not condu- cive to economic efficiency. On the first: two issues, the discussion as- sumes that the recipient government's aim is to use aid to maximize the economic welfare of its citizens, and that this goal could be achieved by allocating resources to activities with the highest rate of return.14 Government-to-Government Aid Foreign aid is usually extended by one government to another. Since private economic activities might normally be expected to qualify for commercial financing, there is some presumption that foreign aid might have a comparative advantage in financing government expenditures, and especially investment. However, the fact that governments receive foreign aid does not mean that they must decide its allocation. Aid can finance private ventures through institutional means such as develop- ment finance banks and agricultural credit institutions. And, where poli- cies have been interventionist, aid can dilute the effect of these policies on the private sector-notably by providing foreign exchange that eases the impact of import licenses. Nonetheless, some aid can and should be directed to government activities, and especially investments. Some in- vestments increase capacity for individual economic activities (such as machinery, equipment, and on-the-job training); others are for infras- tructure or social overhead capital (such as roads, communications, ports, primary education, and so on). The emerging view of develop- ment questions the desirability of governmental involvement in the first category. But the second group is generally accepted to be the responsi- bility of government largely because it has some features of a public good. Either there are large indivisibilities (so that additional utilization can initially take place without sharply increasing the costs to existing users) or user charges would have to be so high as to be infeasible. In addition, an efficient infrastructure created by the government brings direct benefits to the private sector. Although traditional thinking emphasized the physical aspect of devel- oping an infrastructure, evidence increasingly shows the importance of institutional development (such as agricultural research and extension activities,1" delivery systems for public services such as education, health, 68 Research Observer 1, no. 1 (January 1986) and communications). Here, too, there are aspects of public goods, especially in the early stages of development. Moreover, the efficiency of some of these investments may be significantly increased by the organi- zational or technological know-how of foreign donors.'6 Fungibility of Aid The aid literature has long discussed whether aid should finance indi- vidual projects or support a country's overall development program. The argument for project finance has largely been presented in the preceding paragraphs: along with capital, it permits the transfer of skills, organiza- tional procedures, and technology. The counterargument has two related parts: (1) because money is fungible, aid has the effect of freeing re- sources for other projects (see Singer 1965 for the classic statement of the dilemma); and (2) the macroeconomic environment affects the re- turns on individual projects to such an extent that aid used to improve that environment might have a much higher productivity than if it simply increased the infrastructure within a framework of inadequate policies, and program aid may be more effective than project aid for this purpose. These issues cannot be easily resolved. On the one hand, even with project finance, donors may be able to influence the policies of recipi- ents. On the other hand, no government will abandon all its domestic policies to the dictates of foreign donors, and the extent to which program aid permits policy influence will vary. In practice, the mix of project and program aid that most improves the welfare of recipient's nationals probably depends, for example, on the degree to which donor and recipient objectives coincide, the recep- tiveness of the government to policy advice, the appropriateness of exist- ing policies, and a recipient's stage of development. As a country's savings rate rises and its creditworthiness improves, the productivity of foreign aid will depend increasingly on the extent to which it supports policy reform. Inappropriate Policies In some countries, prospects for growth will be unsatisfactory until policies are significantly reformed. Sometimes the government is unwill- ing or unable to make such reforms. If, nonetheless, a donor provides aid for humanitarian or political reasons, are there types of aid that can improve living standards? Clearly, in these circumstances, program aid cannot be defended. But two possibilities deserve consideration: (1) projects that will yield posi- tive real returns under existing policies and would continue to do so if policies were to be reformed later, and (2) projects that might not yield Anne 0. Krueger 69 a positive return under the existing regime but would raise the payoff to policy reform at a later date. There have been positive real rates of return on aid-financed projects, even in countries with manifestly inap- propriate policies and low growth rates. These include investments that raise peasant and smallholder productivity, that increase transport and power capacity, and that do not depend on existing inappropriate incentives for their relative profitability. The experience of countries that have reformed their policies shows that certain projects can lay a basis for faster economic progress by absorbing educated labor rapidly into productive employment and increasing the rates of utilization of their infrastructure. The spectacular spurt of growth in Korea after the policy reforms of 1960-64 was undoubtedly far more rapid and sus- tained than would have been possible had the educational standards of the labor force in 1960 been no greater than what would have yielded a satisfactory rate of return in the prerefcrm years. Although empirical evidence is lacking, there are a priori grounds for thinking that invest- ments in increasing capacity with long gestation periods may make sense even if the payoff under the existing policy regime is small.'7 Criticisms Many of the critics of foreign aid reject it on ideological grounds. of Aid Either aid perpetuates dependency and perverts domestic development (a view put forward, for example, by Hayter 1971), or it permits govern- ments to "escape the burdens of their foolish economic policies" (Krauss 1983, p. 158). The first argument is based on a political model of economic behavior that rejects most economic analysis and thus chal- lenges the tenets of economics. The second criticism accepts the basic neoclassical view and argues that aid is unproductive in conventional terms because it is aid. Despite these extremes, most professional economists tend to think that whether aid is effective-that is, whether it has a sufficiently high marginal product-is an empirical question that can be addressed only with empirical evidence. Moreover, even when aid is less productive than might have been hoped, a secondary question is whether that low productivity is inherent in the aid process or whether instead lessons can be learned and productivity raised. In fact, there has been surprisingly little empirical work on many facets o f aid-another feature of its peculiar history. This section discusses three issues with potentially serious implications for the productivity of aid. The first is the degree to which aid may simply substitute for domestic savings rather than raise investment. The second is the role of aid in affecting the allocation of resources between the private and public sectors. The third is the concern that aid diverts scarce administrative resources from other, higher-productivity work. 70 Research Observer 1, no. I (January 1986) Aid and Savings On the issue of aid and savings, two related criticisms have been made. One is that countries receiving aid do less to provide incentives for domestic savings; the other is that they are likely to have more overvalued currencies than would otherwise be the case."8 The first proposition has two versions: (1) at a given level of income, the domestic savings rate in a recipient is less than it would be in the absence of aid; and (2) at a given level of income, investment is lower than it would be without aid. The first proposition asserts that the recipient will allocate its aid partly for present and partly for future income. The second proposition is more extreme, implying that aid is more than offset by increased domestic consumption. The first proposition, that the marginal propensity to save is less than one, accords with economic theory and, therefore, is empirically testable. The second proposition, which essentially posits a negative propensity to save, is more extreme and is a priori impossible. Since government macroeconomic policy is a prime determinant of how an economy reacts to the receipt of aid, any outcome, in principle, is possible. Governments committed to the goal of economic growth would adjust macroeconomic policies to foster higher investment as a consequence of receiving aid. Alternatively, they might fail to adjust policies in ways conducive to higher investment, and domestic savings would decline or domestic consumption rise in response to aid. This latter path would be more likely the more the exchange rate appreciat- ed when aid was received. Economists have therefore attempted to estimate, usually by cross- section analysis, the effects of aid on domestic savings rates. These estimates suffer from various defects. To the extent that aid has success- fully stimulated domestic policies and resulted in faster growth, coun- tries leave the ranks of aid recipients. Since the motive for aid is often that domestic prospects are otherwise unsatisfactory, aid may be biased to countries with below-average performance. Several studies have nonetheless attempted to estimate the relation between aid and invest- ment. A study by Weisskopf (1972) reached the most negative conclu- sion. He posited an ex ante relationship between savings, income, all inflows of foreign capital (including aid), and exports of the form S = f(Y, F, E), where S is ex ante savings, Y income (regarded as exogenous), F the foreign capital inflow (also regarded as exogenous and defined as the trade account deficit in the balance of payments), and E total ex- ports. Weisskopf pooled time-series and cross-section data on seventeen countries, found a highly significant impact of capital inflow on savings, and estimated that about 23 percent of foreign capital inflows were offset by declines in domestic savings. Other authors, including Bhagwati and Srinivasan (1976), Gupta (1970), and Papanek (1972), have shown Anne 0. Krueger 71 different results, generally suggesting a positive marginal propensity to save, so that a part of aid might be offset by extra consumption.19 Given the variety of macroeconomic policies and objectives of devel- oping countries (not to mention levels of income and other differences), it would be surprising if the impact of aid on domestic savings were the same in all cases. Additional research could usefully be focused on analyzing the types of policies that recipients can adopt to affect the domestic savings response to aid. The Public and Private Sectors The second concern is that, since most aid goes to governments, it necessarily strengthens the role of government in the economy and weakens the private sector. While this may have happened in some instances, the proposition, as a logical necessity, does not withstand close inspection on three counts. 1. Some government-to-government assistance in fact leads to a relax- ation of controls and improved functioning of markets and the private sector. Consider a developing country in which past economic misman- agement has resulted in a highly overvalued exchange rate and stringent import licensing. The latter implies that all private entities requiring imports must conform to governmental criteria-a major instrument of control for the government. If a donor were to propose lending to finance imports in return for an exchange rate adjustment and abandon- ment of import licensing, that would represent a major reduction in the degree of government control over the private sector."0 There seems to be an implicit assumption that it is only command over resources that gives the government power to influence the private sector; in fact, bureaucratic and administered controls can be far more detrimental. 2. The assumption that any governmental activity is necessarily at the expense of the private sector ignores the essential infrastructure invest- ments-in roads, ports, flood control, and the like-that are not only inherently the business of government but- are essential to satisfactory economic growth. Although it is widely recognized that many devel- oping countries' governments have allocated resources to activities in which they have little or no comparative advantage, it is less frequently recognized that those same governments have also hampered growth by failing to allocate enough to those necessary functions in which they do have a comparative advantage. In most developing countries, resources are so scarce that it is implausible for the government to be "too big," in the sense of having adequately provided infrastructural support for growth, while simultaneously having enough resources to take over pri- vate sector activity. 3. The fact that resources are given to governments does not necessar- ily mean that they have command over them. For example, aid chan- 72 Research Observer 1, no. 1 (January 1986) neled to development finance corporations or other financial intermedi- aries can give those bodies the real control over how the money is spent, even though governments were the national recipients. Thus, the conclusion must be that proper allocation of scarce govern- mental as well as private resources is an important condition for satis- factory growth. That allocation depends on several factors: appropriate macroeconomic signals (including the exchange rate, agricultural and energy pricing, the absence of high levels of protection to domestic industry, and a well-functioning capital and labor market) and the provi- sion of infrastructure to support private sector development. Aid can assist or hinder that allocation, depending on the conditions under which it is extended and the uses to which it is put. Judgment as to what aid has actually accomplished would have to be made individually for each country and each period. Administering Aid The third concern, that foreign aid may have diverted scarce adminis- trative talent in recipient countries from other essential tasks, is the least documented of the three criticisms of aid considered here. Stories abound of multiple aid missions to countries that tie up top officials for long periods. Again anecdotally, some aid-financed projects have been successful but have diverted skilled manpower from other, possibly more important tasks. Although such instances have no doubt occurred and suggest the desirability of coordination among donors, the argument has another side: foreign aid has often been a major source of technical assistance and finance for improving administrative skills. Aid has fi- nanced countless programs to train officials overseas, to establish and expand educational institutions in developing countries, and to finance technical assistance programs which employ expatriates as managers and administrators. In the absence of careful empirical research, there is no basis on which to evaluate the impact of aid, both positive and negative. On a priori grounds, however, it is difficult to argue that aid has had a negative impact on administrative resources. Since the 1950s, much has been learned about development and also Lessons about the practice of aid. That learning has been uneven: much new Learned knowledge about development has been useful for formulating and run- ning aid programs, but research on aid and its effectiveness has lagged far behind. Knowledge of development policy and how it affects aid ranges from details about the best ways of providing technical assistance to a broad understanding of the development process. To report on all the lessons is far beyond the scope of this survey.2. Anne 0. Krueger 73 There are, however, some general precepts on which most observers of aid and development would agree. These include the importance of channeling development programs in ways that are compatible with individual incentives and prevailing markerl conditions, and the vital role of macroeconomic policies in determining the return on individual pro- jects in developing countries. With regard to the first, many well-intentioned programs that pin- pointed an area for improving development prospects have foundered because they failed to recognize individual incentives and their import- ance in determining individual behavior. These failures have ranged from family planning programs that did not understand peoples' motives for wanting large families to the provision of subsidized rural credit directly to the rural poor. In general, programs that attempt to alter individual behavior will seldom work unless the incentives confronting people are understood and appropriately modified to make it in their self-interest to change. Similarly, when programs are designed to "fight the market," they are likely to fail or, at best, to be extremely cost- ly. Examples of this latter group include the attempts in some labor- abundant countries to provide jobs for rural workers at wages above going rates. In most such cases, the jobs ccntractor was found to pocket a significant fraction of the excess wage or to charge workers a front-end fee for their jobs. The second lesson, about the role of macroeconomic policies, is no less important. Although aid has undoubtedly financed some costly pro- jects, the biggest difficulties have arisen in the context of inappropriate macroeconomic policies. Indeed, regardless of the soundness of an indi- vidual aid-financed project, growth is likely to remain slow when domes- tic policies discourage it. In that environment, one major contribution of aid will be in the policy dialogue and donor influence, if any, that can bring about effective and timely reforms. Because individual projects earn their returns in a milieu influenced by overall economic policy, project financing is not likely to maximize< the donor's impact on the recipient's growth unless macroeconomic policies are either already ap- propriate to growth or beyond influence. In the latter case, some types of projects may nonetheless yield positive returns and thus speed up overall growth (even within the unsatisfactory framework), or else lay the foun- dation for higher growth should policy reform later become feasible. As the influence of the policy framework has become increasing- ly appreciated, policymakers in developing countries (as well as aid officials) have altered their emphasis. To a large extent, increased under- standing has informed the aid process-which, one hopes, will make it more effective. This still leaves a large agenda for research. If understanding of development-and even of what aid can do as a theoretical proposi- tion-has increased, empirical research has lagged far behind. Many 74 Research Observer 1, no. 1 (January 1986) developing countries, some now in the middle-income group, received aid thirty or more years ago. There have been natural experiments, such as community development schemes, overseas training programs, public health programs, and so on. Although much has been learned, econo- mists could do far more to examine empirically the impact of aid and its effects. This article examines the economic effects of aid and the relationships among private Abstract capital flows, official flows, and concessional assistance. It considers whether official and concessional flows improve the welfare of both recipient and grantor, and it outlines the economic rationale for concessional assistance. The impact of aid on a recipient's growth performance is analyzed, and some criticisms of aid are evaluated. 1. For a survey of current understanding of some of these project and sectoral issues, see Notes Krueger, Michalopoulos, and Ruttan (forthcoming). There is also a significant literature on individual country experiences with aid, and some analysts have addressed the question of the impact of foreign assistance on donor countries and the global economy, but these issues are not dealt with here. 2. If procurement is tied, the value of the grant is reduced by the excess price over the international price. This, however, is hardly ever computed and is not taken into account in the official data. 3. An important question is the classification of loans which carry market rates of interest but maturities longer than can be obtained commercially. The market rate for such loans is not observable, so the grant component cannot be calculated. However, it would clearly be wrong to treat such a loan as entirely nonconcessional. This consideration matters in evaluating the role of official and concessional development assistance. 4. See Bhagwati (1970) for a good discussion of these issues. Schultz (1960) provided the classic estimate of the effect of Public Law 480 on the recipient. 5. Usually motivated by humanitarian concerns, private donors contribute to these organizations, which in turn undertake assistance programs. Some NGOs also receive cash, services, and commodities from official sources. In recent years NGOs have provided about $2.3 billion a year in aid, compared with about $32 billion of official concessional flows. Economic analysis of NGO activities could follow lines similar to those spelled out here. In practice, NGOs have been little involved in discussions with recipient governments of macroeconomic and incentive policies and have concentrated their assistance on particular target groups or projects. 6. See Hirschman (1982) for an exposition of this view. 7. With maximizing agents, removal of distorted incentives can bring about a once-and- for-all gain. Thereafter, resource accumulation and changing technology provide the basis for further growth, although the returns to resource accumulation and new technology are higher under appropriate incentives. 8. See Michaely (1962) for an indication of the extent of concentration at that time. The situation has changed considerably since: in 1983, it was estimated, about 55 percent of exports from developing countries consisted of primary commodity exports. 9. Chenery and many others recognized the importance of reducing the bias of incentives toward the home market and against exports. As Chenery pointed out in private corres- pondence, "In retrospect, it was the need to analyse the demand for external capital from many aid recipients together that led to the use of the simplifying assumption of exogenous exports (varied in alternative solutions) as in Chenery-Strout [19661. The persistence of this formulation is probably due more to the simplicity of its algebra than to a belief that Anne 0. Krueger 75 elasticities are zero." It might also be argued that the two-gap model led many adherents to ignore issues of substitution and incentives, even though the model's developers were aware of these shortcomings and their implications. 10. There are still "structuralists," who believe that response to changes in incentives is so slow that reliance on incentives and the usual tools of analysis is inappropriate. Taylor (1983) is a prominent exponent of that view. The diflerence is, of course, one of emphasis. The structuralists would surely not deny some response to altered incentives, and no one who has worked in a developing country would claim that all markets work efficiently. The question is whether governments should substitute for perceived market imperfections or devise policies to reduce them, while simultaneously providing necessary governmental services (see the next section). 11. This would happen in the simple 2 x 2 Hecksher-Ohlin-Samuelson model of trade if factor endowments differed so much between countries that trade could not fully substitute for factor mobility. Complete specialization would result in a model with constant returns to scale and no transport costs, and free trade (but no factor mobility) would be consistent with a higher rate of return on capital in the relatively labor-abundant countries. 12. This discussion ignores the risks associated with investing and how they may be affected by implicit or explicit guarantees from recipient governments. For the bulk of infrastructure investments, however, this consideration does not apply. 13. The capital flows of the 1970s were heavily oriented toward the so-called middle- income countries. Although some low-income countries increased their use of these markets a little, by the early 1980s it was clear that they had been ill-advised to do so. Their development has not yet proceeded enough for them to rely on commercial borrowing for as large a fraction of capital as they attempted in the 1970s. See below. 14. In effect, this is equivalent to assuming that the best way of helping poor people is to increase their future income and earnings, and that investments for this purpose have a sufficiently high return to be included in the investment program. Especially in poor countries, the scope for improving the welfare of the poor through redistribution is extremely limited; moreover, empirical evidence strongly suggests that poor people's incomes rise faster in countries achieving faster growth. Evidence is also mounting that, where income distribution has worsened with growth, much of the fault has been due to inappropriate incentive policies (see Myint 1985). 15. See Ruttan (1982, pp. 17-44 and 237-61) for an analysis of these issues. 16. Although I have nowhere seen it documented, observers in both Korea and Turkey believe that the entrepreneurs and skilled workers wh1o achieved such success in winning foreign construction contracts in the 1970s and 1980s gained their skills on American aid-financed projects in the 1950s. 17. The investments in human capital that have high potential returns probably center largely on primary education. Many developing countries with inappropriate policy frameworks have expanded their university enrollments faster than warranted and neglect- ed primary education. 18. As will be discussed below, it is also possible that the "knowledge component" of aid might render investment more productive than would otherwise be the case; there would thus be gains over and above those resulting from the incremental investment and consumption financed by aid. 19. See Krueger, Michalopoulos, and Ruttan (forthcoming, chap. 4 by Vasant Sukhatme) for a fuller survey of this literature. 20. There are several instances in which this type of reform has produced major benefits for economic growth. For a summary of some episodes, see Krueger (1978, chap. 10). For an analysis of the reasons why the recipient government could not successfully relax controls without assistance, see Krueger (1978, chap. 7). 21. See Krueger, Michalopoulos, and Ruttan (forthcoming), on which this section draws, for more detail on the aid experience. 76 Research Observer 1, no. 1 (January 1986) Balogh, Thomas. 1964. "Review of T. W. Schultz's 'Transforming Traditional Agricul- References ture."' Economic Journal 74 (December): 996-99. Bhagwati, Jagdish. 1970. Amount and Sharing of Aid. Overseas Development Council Monograph Series. Washington, D.C. Bhagwati, Jagdish, and T. N. Srinivasan. 1976. Foreign Trade Regimes and Economic Development: India. New York: Columbia University Press for the National Bureau of Economic Research. Bruno, Michael, and Hollis B. Chenery. 1962. "Development Alternatives in an Open Economy: The Case of Israel." Economic Journal 72(March): 79-103. Chenery, Hollis B., and Alan M. Strout. 1966. "Foreign Assistance and Economic Devel- opment." American Economic Review 56(September): 679-733. Diaz Alejandro, Carlos. 1965. "On the Import Intensity of Import Substitution." Kyklos 18 (fasc. 3):495-509. Findlay, Ronald. 1971. "The Foreign Exchange Gap and Growth in Developing Countries." In Jagdish Bhagwati and others, eds. Trade, Balance of Payments, and Growth. Amsterdam: North Holland. Reprinted as chapter 10 in Ronald Findlay. International Trade and Development. New York: Columbia University Press, 1973. Gupta, K. L. 1970. "Foreign Capital and Domestic Savings: A Test of Haavelmo's Hypothesis with Cross-Country Data." Review of Economics and Statistics 52 (May): 214-16. Guttentag, Jack M., and Richard J. Herring. 1984. "Commercial Bank Lending to Less Developed Countries: From Overlending to Underlending to Structural Reform." Brook- ings Papers in International Economics 16(June):1-52. Hayter, Teresa. 1971. Aid as Imperialism. New York: Penguin Books. Hirschman, A. 0. 1982. "Rise and Decline in Development Economics." In Mark Gersovitz and others, eds. The Theory and Experience of Economic Development. London: Allen and Unwin. Johnson, Harry G. 1967. Economic Policies toward Less Developed Countries. Washington D.C.: Brookings Institution. Krauss, Melvyn B. 1983. Development without Aid. New York: New Press. Krueger, Anne 0. 1968. "Factor Endowments and Per Capita Income Differences among Countries." Economic Journal 78 (September): 641-79. _. 1978. Foreign Trade Regimes and Economic Development: Liberalization Attempts and Consequences. Cambridge, Mass.: Ballinger for the National Bureau of Economic Research. . 1983. "The Role of the World Bank as an International Institution." Carnegie- Rochester Series on Public Policy no. 18, pp. 281-312. Amsterdam: North Holland. . Forthcoming. "Some Aspects of Capital Flows to Developing Countries." In Assaf Razin and Efraim Sadka, eds. Economic Policy in Theory and Practice. London: Macmillan. Krueger, Anne O., Constantine Michalopoulos, and Vernon Ruttan. Forthcoming. The Impact of Development Assistance to LDCs. Baltimore, Md.: Johns Hopkins University Press. Lal, Deepak. 1983. The Poverty of 'Development Economics.' London: Institute of Economic Affairs. Little, I. M., and J. M. Clifford. 1965. International Aid. London: Allen and Unwin. McKinnon, Ronald. 1964. "Foreign Exchange Constraints in Economic Development and Efficient Aid Allocation." Economic Journal 74 (June): 388-409. Anne 0. Krueger 77 Mason, Edward S., Mahn Je Kim, Dwight H. Perkins, Kwang Suk Kim, and David C. Cole. 1980. The Economic and Social Modernization of the Republic of Korea. Cambridge, Mass.: Harvard University Press. Michaely, Michael. 1962. Concentration in International Trade. Amsterdam: North Hol- land. Myint, Hla. 1985. "Growth Policies and Income Distribution." Economics and Research Staff Policy Discussion Paper no. 1. Washington, D.C.: World Bank. Papanek, Gustav. 1972, "The Effect of Aid and Other Resource Transfers on Savings and Growth in Less Developed Countries." Economic Journal 82 (September): 934-50. Pincus, John A. 1963. "The Cost of Foreign Aid." Review of Economics and Statistics 45 (November): 360-67. Prebisch, Raul. 1950. The Economic Development of Latin America and Its Principal Problems. Lake Success, N.Y.: United Nations, Department of Economic Affairs. Ruttan, Vernon. 1982. Agricultural Research Policy. Minneapolis: University of Minnesota Press. Schmidt, Wilson. 1964. "The Economics of Charity: Loans versus Grants." Journal of Political Economy 72 (August): 387-95. Schultz, T. W. 1960. "The Value of U.S. Farm Surplus to Underdeveloped Countries." Journal of Farm Economics 42 (December): 1019--30. _.____ 1961. "Investment in Human Capital." American Economic Review 51 (March): 1-17. . 1964. Transforming Traditional Agriculture. New Haven, Conn.: Yale University Press. Singer, Hans W. 1965. "External Aid: In Plans or Projects?" Economic Journal 75 (September): 539-45. Taylor, Lance. 1983. Structuralist Macroeconomics: Applicable Models for the Third World. New York: Basic Books. Teece, David. 1977. "Technology Transfer by Multinational Firms: The Resource Cost of Transferring Technological Know-how." Economic Journal 87 (June): 242-61. Weisskopf, T. E. 1972. "The Impacts of Foreign Capital Inflow on Domestic Savings in Underdeveloped Countries." Journal of International Economics 2, no. 1 (February): 25-38. 78 Resecrch Observer 1, no. 1 (January 1986) STABILIZATION POLICIES IN DEVELOPING COUNTRIES Liaquat Ahamed A broad definition of stabilization measures might be those poli- cies that attempt to reduce excessive or unsustainable balance of payments deficits or the rate of inflation. Both are symptoms of too much demand relative to output, so stabilization could also be defined as a better balance between aggregate supply and aggregate demand. Either definition accommodates a wide range of policies, from fine-tuning aggregate demand to making radical changes in the economic role of the state. But in every case, two factors will decide whether a particular policy is appropriate: first, the nature of the disturbance that produced the imbalance between demand and supply; second, the trade and exchange rate regime in force. Imbalances between aggregate demand and aggregate supply can emerge for a variety of reasons. Until the 1970s, the most common source of balance of payments deficits was overexpansionary demand- side policy-large budget deficits, financed either through the banking system or by unsustainable levels of foreign borrowing. But other factors can lead to balance of payments difficulties. This paper discusses two of them: domestic supply shocks and external shocks. The important fea- ture of a domestic supply shock is that returns received by some domes- tic factor come to exceed its supply price-as would happen, for exam- ple, after a rise in real wages in excess of productivity growth. A typical external shock is a deterioration in the terms of trade. The appropriate policy response in each case is analyzed below. Then, focusing more narrowly on one instrument of policy, the exchange rate, the paper looks at the effects of a nominal devaluation on the real exchange rate, on the trade balance, and on domestic output. The final section turns to the special difficulties of reducing inflation in both closed and open economies. 79 Balance of To see why the trade and exchange rate regime matters, consider a Payments country with excess aggregate demand. It could choose to maintain its Adjustment nominal exchange rate, allowing the real exchange rate to appreciate and an open balance of payments deficit to emerge. Or it could prevent an open balance of payments deficit by imposing import quotas. Finally, it could let the nominal exchange rate fall to offset the rise in domestic prices, while maintaining the real exchange rate and protecting the balance of trade-at the cost of higher inflation. Each policy implies a different outcome for relative prices and hence for the structure of the stabilized economy. Economies with fixed exchange rates cannot sustain domestic inflation higher than international inflation without running into balance of payments difficulties: the focus of policy in such cases is balance of payments adjustment. By conl:rast, economies which adjust their exchange rates can at least partly insulate their trade accounts from domestic demand: macroeconomic imbalance emerges for them as an acceleration in the rate of inflation. Countries using import quotas to bring their international payments into balance are a hybrid case. Deficits caused by excess demand In an economy initially disturbed by higher government spending unmatched by increased revenues, the government's budget deficit will rise. This affects demand in two ways: directly, by increasing the public sector component of demand and, indirectly, where the deficit is fi- nanced through domestic credit creation, by raising the rate of growth of domestic liquidity and thus of nominal private spending. Figure 1 shows how this increase in demand might affect the economy with a fixed nominal exchange rate. First, the story told by monetarist models. Starting from full employment, at a point such as A, Figure I higher spending leads to higher inflation and hence to an appre- Nontradables ciation of the real exchange rate, a loss of competitiveness in the B tradables sector, and a shift in the mix of output from A to B. C *c Spending rises from A to C, creating a balance of payments . \\ A deficit equal to BC, financed by a loss of reserves. D \ Keynesian models, by contrast, assume that the economy starts with unemployment, at a point such as D. With sticky nominal F \ wages, the demand-induced rise in inflation leads to a lower real product wage in the nontradables sector. So the supply of non- Tradables tradable goods expands, and output moves from D to B. The increased demand for tradables, however, causes a balance of payments deficit. Keynesian and monetarist models both suggest that higher government spending will mean faster inflation, a higher real exchange rate, and a deterioration in the balance of payments. But Keynesian models also predict a rise in output and lower real wages, 80 Research Observer 1, no. 1 (January 1986) while full employment models emphasize the changes in the composition of output.' A fixed exchange rate regime puts a limit on the size of the deficit that the government can finance without running out of reserves. Harberger and Edwards (1980) found evidence that this does impose discipline. They looked at monetary policy in developing countries that did not change their exchange rates to restore external balance. They found that countries which avoided devaluation in general followed a "reservers target" policy; they let bank credit expand when reserves were increasing and reduced it when reserves were falling.2 At the same time, a fixed exchange rate can automatically offset excess demand through the money supply process. Financing the balance of payments deficit with reserves implies a loss of foreign assets of the central bank and, in turn, a lower domestic money supply. If private spending bears a stable rela- tionship to the money supply, and the monetary authorities take no steps to sterilize the loss of reserves, private spending will fall. This implies that part of the fillip provided to real private spending by an expansion- ary budget deficit will be temporary. In terms of figure 1 the economy may move initially to E but eventually reverts to B. In full-employment models, the balance of payments deficit will tend toward equality with the budget deficit, with no long-run effect on private spending.3 Restoring balance of payments equilibrium with full employment re- quires a smaller budget deficit and a lower real exchange rate. If prices on nontradables are sufficiently flexible, the government should focus directly on reducing aggregate spending. Lower spending will automati- cally reduce the nominal price of nontradables and hence the real ex- change rate. But if prices of nontradables or nominal wages are rigid, external and internal balance can be reconciled only by raising the price of tradables through a devaluation. Devaluation reduces private spending by cutting the purchasing power of financial assets denominated in domestic currency. Devaluation also affects the value of external debt in terms of domestic purchasing power: if domestic prices do not rise to match the price of foreign exchange, the real value of external liabilities will rise. This, too, may affect the level of aggregate demand. The more flexible the structure of the economy- that is, the easier the substitution between tradables and nontradables in both production and consumption-the greater the impact of the deval- uation on domestic prices. And the more likely it is that the "real balance" effect of lower asset values will overshadow the deflationary influence of higher external liabilities. This is an overview of the orthodox theory of stabilization policy. Aside from the political difficulties, implementing such a program meets two economic problems: the cost of the adjustment process and the technical issue of synchronizing monetary and fiscal policy with ex- change rate adjustments. Liaquat Ahamed 81 The Costs of Adjustment The issue of cost is more complicated than is often supposed. A starting point is the welfare loss of moving from C to A in figure 1. This depends on the ability of the economy to move resources from produc- ing nontradables to tradables. One measure for this concept of cost is the change in the level of real expenditures associated with a given reduction in the balance of payments deficit-the ratio of AC (the reduction in expenditures) to BC (the payments deficit). Structuralist models of adjustment (see, for example, Taylor 1981) implicitly assume that tradables are totally inelastic in supply, nontraded goods are perfectly elastic, and the pattern of spending is highly unres- ponsive to relative prices. Reductions in the current account deficit therefore have a large impact on both output and spending. By contrast, in relatively open economies where tradables account for a large propor- tion of output and where spending is responsive to prices, reductions in the current account deficit can be achieved with much less of a fall in real expenditures. But this concept of cost is not very helpful to policy- makers. The important policy question is what is the least-cost way to get from B to A. There are two sides to this. The first concerns the costs of reallocating resources between sectors. Existing macroeconomic models give little insight here. The costs will depend on a variety of factors, including the mobility and specificity ol both capital and labor, the underlying rate of growth of the economy, and the rate of investment. The second issue is how to manage aggregate demand to move the economy from B to A. Theoretically, it should be possible to cut spend- ing, shift the pattern of output, and maintain demand so as to ensure full employment. In practice this is extremely difficult. Inevitably, unem- ployment causes transitional welfare losses. The transition will also affect the distribution of income. Because of the shift in the pattern of production, the demand for labor will change. If nontradables are labor-intensive, then a shift in relative prices against nontradables results in a fall in the demand for labor and a fortiori a fall in real wages. Real wages will, however, also be affected on the con- sumption side. If wage earners are heavy consumers of tradables, real wages will fall even further. By contrast, if nontradables account for a large proportion of the consumption basket, this will offset the fall in real wages. The policies used to cut aggregate demand also can affect the real wage.4 Reductions in government benefits (such as food subsidies) or increases in taxation (particularly indirect taxation) affect wage ear- ners' real incomes. Where labor markets are perfectly flexible, cutting the real wage does not pose problems of adjustment. But if real wages are sticky, the devaluation will have no real impact; exchange rate policy by itself is 82 Research Observer 1, no. I (January 1986) not enough to bring about a real devaluation and its associated switch- es in spending and resource reallocation. The only way to cut deficits in the balance of payments is to reduce real spending-at the cost of higher unemployment. Real wage rigidity apparently makes it impossi- ble to reconcile domestic employment and external balance objectives. In theory, this dilemma may be defeated by taxing capital and subsidiz- ing wage earners (Corden 1981). But capital is internationally mobile: a tax on profits would result in a decline in the returns to investment and thus a fall in inflows of long-term capital. The best policy, therefore, would be to tax the quasi rents on the existing capital stock while maintaining returns to new investment (possibly by some form of in- vestment incentive). Synchronizing Policies Coordinating exchange rate adjustments with monetary and fiscal policy is difficult because a nominal devaluation itself has aggregate demand effects: a devaluation does part of the job of reducing real spending. To see this, suppose that a government eliminates its budget deficit and devalues at the same time. The devaluation, by raising the overall price level, increases the private sector demand for money bal- ances and thus temporarily reduces private expenditures. If domestic credit creation stops as a result of a balanced government budget, pri- vate spending will fall further as the private sector rebuilds its money balances. For a time, the country could actually have a balance of payments surplus. (In figure 1, the economy might move from B to F before reverting back to A). A better policy would be to reduce the rate of credit creation gradually. Policy design must therefore take into ac- count the effects of devaluation on aggregate demand.5 Short-term capi- tal flows may aggravate these problems of coordination. A delay be- tween the first sign that a devaluation is necessary and the actual deval- uation is common under fixed exchange rates. Even if international capital mobility is highly restricted, expectations of a devaluation can lead to import hoarding and delays in the repatriation of export earn- ings. The result is a sharper fall in reserves. After the devaluation, the process is reversed and reserves rise again. Such instability leads to two problems. First, the initial decline in reserves means a bigger fall in the money supply before the devaluation-individuals, speculating on a de- valuation, shift from holding domestic financial assets in order to hoard foreign goods. Large shifts in the demand for money make it hard for the government to control monetary conditions. The authorities may even try to offset the monetary contraction and thus add to the original pressure of excess demand. Second, speculative capital flows make it difficult to judge the underlying balance of payments position and the size of the devaluation that is needed to correct it. Liaquat Ahamed 83 In general, devaluation and reductions in fiscal deficits tend to be mutually reinforcing. For most developing countries, indirect taxes on tradables are an important source of revenue, while most government spending is on nontradables. Devaluations therefore tend to raise public revenues and automatically cut fiscal deficits and the growth of domestic credit. But not necessarily. If domestic financial assets are indexed and, as is often the case in highly indexed economies, some proportion of the loans from the domestic banking system is provided at fixed nominal interest rates, then inflation no longer acts as a tax on the private sector. In fact, it has the opposite effect. Inflation, and hence devaluation, increases the net real wealth of the private sector and boosts aggregate demand. This is likely to happen, for example, in Brazil and Israel. The expenditure-reducing effect of the devaluation may also be muted be- cause of its fiscal impact. If the public sector has a large external debt, a devaluation raises the domestic cost of debt service and thus, at any given level of taxation, increases the budget deficit. Alternatively, if government revenues respond to higher inflation only after a delay (be- cause of collection difficulties, for example) while spending rises at once, higher inflation will mean a bigger budget deficit.6 And if the budget deficit is financed by expanding the money supply, a devaluation could mean faster monetary growth. The key to a successful stabilization program is a lower fiscal deficit. If political constraints rule that out, it becomes impossible to maintain a fixed exchange rate. The government has to choose between a policy of progressively higher import restrictions combined with periodic large devaluations and a policy that allows the exchange rate to crawl with the rise in domestic prices. Most authors have recommended some var- iant of the crawling peg as a second-best way to insulate both the balance of payments and the structure of relative prices from fiscal incompetence (see Krueger 1981; McKinnon 1981). A crawling-peg regime works, in effect., by allowing the rate of infla- tion to rise to the point at which the inflation tax on financial balances is enough to finance the budget deficit. The higher the budget deficit, the higher the rate of inflation and the faster t:he rate of depreciation needed to achieve external balance. But, in practice, crawling-peg systems face three problems.7 First, how to choose the rate of crawl? One option is to tie the exchange rate to the differential between the domestic and the interna- tional rate of inflation, thus stabilizing thie real exchange rate. But this provides no means of changing the real exchange rate, as might be warranted, for example, by changes in the terms of trade or in the rate of growth. Another possibility is to tie the exchange rate to the level of international reserves-slowing the rate of devaluation when reserves build up and raising it when reserves decline. But this fails to distinguish between temporary changes in the demand for or supply of foreign 84 Research Observer 1, no. 1 (January 1986) exchange, which should presumably be accommodated by changes in reserves, and permanent changes, which will require an adjustment in the real exchange rate. Other rules have been proposed, but no single rule will suffice: crawling pegs need active management based on several criteria. The second problem concerns the capital account. When the rate of inflation is high, the opportunity cost of holding non-interest-bearing financial assets is high. This prompts a shift into financial assets denomi- nated in foreign currency. Economies with high fiscal deficits also tend to have high reserve requirements on bank deposits and therefore a sub- stantial gap between interest rates on domestic deposits and loans. This form of taxation encourages depositors to buy financial assets denomi- nated in dollars (or other hard currencies) and borrowers to borrow abroad. The domestic financial system-the base underlying the inflation tax-shrinks. Thus for given levels of the budget deficit, inflation will speed up. Exchange controls may be necessary to retain even imperfect monetary control. The third problem with crawling-peg regimes is that they make infla- tion all the more difficult to control. The system promotes the view that the authorities are willing to tolerate high rates of inflation; this rein- forces inflationary expectations. Wage demands are made on the as- sumption that the authorities will prevent unemployment by depreciating the exchange rate and accommodating the increase in domestic prices. This builds inflation into the wage-setting process.8 Two Case Studies of Stabilization Programs Two case studies of stabilization programs designed to reduce current account deficits are presented below. The first, Mexico in 1976, is an example of a stabilization program with little cost in output. The sec- ond, Peru from 1976 to 1978, is a contrasting case in which the output cost was large. A combination of cuts in government expenditure and a depreciation in real exchange rate did reduce Peru's current account deficit, but only at the expense of high inflation. Mexico. The balance of payments crisis of 1976 had its roots in a fiscal problem. The government had increased its spending to meet social goals, but without corresponding tax increases. As a result, its fiscal deficit, which had averaged 2.5 percent of GDP between 1969 and 1970, rose to a peak of 6.5 percent of GDP in 1976. That meant faster inflation, growing external debt, and massive capital flight in anticipa- tion of a devaluation. For the same reason, people shifted from peso deposits to dollar deposits in Mexican banks (see table 1 on page 86). The stabilization program had two major elements: a cut in the public sector deficit and a substantial devaluation of the exchange rate. These succeeded in reducing the current account deficit from a peak of 4.2 Liaquat Ahamed 85 Table 1. Mexico: Macroeconomic Data 1974 1975 1976 1977 1978 GDP growth 6.1 5.6 4.2 3.4 8.3 Real expenditures 8.0 6.2 3.1 1.2 9.1 As percentage of GDP Investment 23.3 23.7 22.3 22.8 23.6 Savings 19.6 19.5 19.0 21.8 21.0 Current account deficit 3.7 4.2 3.3 2.0 2.6 Overall budget deficit 3.9 6.5 4.7 3.3 2.7 Financing from the banking system 4.3 2.7 2.6 3.1 2.7 Annual growth rate Money supply 20.9 18.0 48.2 23.8 33.3 Inflation 23.8 15.7 15.8 29.1 17.3 1974 = 100 National exchange rate 100 100 160.0 181.6 181.6 Real exchange rate 100 95.3 140.2 131.9 120.6 Real wages 100 10.5.4 117.4 119.1 116.6 International reserves (US$ million) 1,395 1,53.3 1,253 1,723 1,928 Sources: World Bank, Economic Analysis and Projections Department data bank; International Financial Statistics. percent of GDP in 1975 to 2.0 percent of GXDP in 1977. There was some slowing in output growth which was quickly reversed by 1978. The program's effect on real wages was also small, although the measures did curb the high rates of growth that had been apparent before 1976.9 Monetary policy was a problem. To offset the contractionary effect of falling reserves in 1976, the authorities sharply expanded the rate of domestic credit creation. This policy of sterilization blocked one of the financial system's automatic stabilizers and added to excess demand and the fall in reserves.10 The program was only partly successful in reducing "dollarization": the devaluation created expectations of further devaluations to come. When the government liberalized the Mex-dollar deposit rate in 1977, in effect making Mex-dollars closer substitutes for dollars held abroad, it partially solved the problem. The new policy stemmed the capital out- flow and diverted from the U.S. government to the Mexican government the seigniorage extracted from Mexican residents holding dollar assets.11 Peru. Table 2 provides a summary of the major macroeconomic devel- opments in Peru between 1973 and 1979. From 1975 onward the public sector embarked on a large spending program unmatched by increased revenues and financed by external borrowing. When private external financing dried up in 1976, the government resorted to deficit financing from the banking system-and hence inflation-to finance the budget."2 86 Research Observer 1, no. 1 (January 1986) Table 2. Peru: Macroeconomic Data 1973 1974 1975 1976 1977 1978 1979 GDP growth 4.3 7.5 4.6 2.0 0.0 -0.5 3.7 Real expenditures 13.4 11.1 3.1 0.5 -1.4 -6.7 3.0 As percentage of GDP Investment 16.7 21.6 19.6 18.1 15.5 14.3 13.3 Savings 14.6 15.4 9.6 9.8 8.2 12.4 17.4 Current account deficit 2.1 6.2 10.0 8.3 7.3 1.9 -4.1 Overall budget deficit 3.6 2.8 4.9 5.8 6.9 4.5 0.5 Financing from the banking system 1.6 -0.5 2.1 5.2 3.4 3.6 -0.4 1972 = 100 Export value 67.9 69.9 71.6 71.6 80.0 91.2 102.5 Import value 111.7 144.2 135.4 122.3 122.7 85.7 94.5 Termsoftrade 150.6 145.6 99.8 108.6 114.8 103.6 143.2 Nominalexchangerate 100 100 116.3 179.2 336.9 506.9 646.3 Realexchangeratea 106.6 108.2 100.6 110.2 125.4 - - Realwages, 101.8 94.1 86.5 81.2 67.1 57.8 - Percentage change Monetary growth 22.6 35.1 15.8 24.2 24.9 46.4 92.2 Inflation 9.5 16.9 23.6 33.5 38.1 57.8 66.7 a. From Cline (1981). The figures for the real exchange rate are based on a trade-weighted measure of the exchange rate. The figures for the nominal exchange rate provided here are, how- ever, based on the exchange rate relative to the U.S. dollar. Sources: World Bank, Economic Analysis and Projections Department data bank; International Financial Statistics. Initial attempts at stabilization in 1975 and early 1976 included a deval- uation of 30 percent, but they were flawed by a lack of complementary deficit-cutting measures. Inflation accelerated, eroding much of the real impact of the nominal devaluation, and the current account deficit re- mained at an unsustainable level. The problem was the familiar one that devaluation has only temporary effects on real spending and on the real exchange rate if it is not accompanied by measures to reduce the budget deficit. In late 1976 the Peruvian authorities shifted to a crawling-peg exchange rate. Action on the budget deficit was still limited, however, and a serious effort to bring the budget into some sort of balance was not made until 1979. As a result, the government had to keep a high rate of devaluation and a corresponding rate of inflation to ensure that the inflation tax was high enough to finance the budget deficit. Deficits Caused by a Domestic Supply Shock Balance of payments difficulties can also arise from a domestic supply shock, as when real wages rise more than productivity and lead to lower Liaquat Ahamed 87 "competitiveness." The essential characteristic is that the returns re- ceived by some domestic factor come to exceed its supply price. This has happened in Pakistan between 1979 and 1972 and in Portugal after 1974. Figure 2 shows how a supply shock affects an open economy.13 Sup- pose that output is initially at A and spending at B, so that the economy is running a current account deficit of AB. In the short run a rise Figure 2 in real wages will affect both the supply side and the demand Nontradables side, as increases in real wages leacl firms to reduce their demand for labor. Output shifts from A to a point such as C, and - A unemployment emerges. On the demand side, higher real wages -- \ E B /and lower profitability reduce investment. Spending shifts from B C\\