World Bank Reprint Series: Number Seventy-two John A. Edelman and Hollis B. Chenery Aid and Income Distribution Re rinted from The NeC7' !iilertiatio?wl Econom?iic Order (MfT Press, 1977) Aid and Income Distribution John A. Edelman and Hollis B. Chenery Political and economic events of the past few years have greatly affected both the need for capital transfers to developing countries and the conditions under which they are supplied. The economic crisis of 1973/1974-the main features of which were the rise in oil prices, recession in the OECD countries, and worsened terms of trade for oil importers-has greatly increased the demand for capital flows to cushion the adjustmenit of the developing countries to new trading conditions. Although there has also been a substantial increase in the supply of official capital since 1972, that increase has been by no means adequate to offset the terms of trade losses suffered by the developing countries Irom 1972 to 1975. The impact of these events has fallen particllarly heavily on the poorer countries. Ter-!I. o, trade losses represent a larger proportion of imports for the poorer countries than for the middle-income group.' Moreover, the substantial increase in private capital flows has gone almost exclusively to the middle-income group. Private capital (and public capital on convelntional terms) will continue to be available mainly to this group because of creditworthiness considerations. As a result the middle-income countries are likely to be able to return to a sustainable growth rate of more than 6 percent for the rest of the decade. However, unless there is a substantial increase in the flow of aid to the poorer countries-as well as changes in policies that will permit more efficient use of this aid-their per capita incomes are likely to be little higher by 1985 than they were in 1970.2 Political prospects for concessional assistance are uncertain. The steady decline in aid from member countries of the OECD was reversed in 1974 and 1975, and there is reason to hope that political obstacles to increased aid will diminish as these countries recover from the recession. Ne iertheless, it hardly seems prudent to count on a rapid increase, especially since an important part of the recent increases was associated with the "emergency" situation created by the oil price increases. The most that can realistically be expected over the next several years is a modest expansion in real flows from recent levels. This outlook underscores the need to make more effective use of available cxternal resources. In this paper, we give an empirical analysis of the factors affecting the allocation of aid in the recent past. Our analysis provides a basis for judging the possibilities for improving aid all-cation in the future. Reprinted by permission from The New Intertiatioinal Economnic Order: The Nortli -Soitllh fliate, ed. Jagdish N. Bhagwati (© 1977 The Massachusettes Institute of Technology). J. A. Edelman, H. B. Chenery 28 Objectives and Criteria Concern with the effects of development on the internal distribution of incolne has become an accepted part of the philosophy of most aid donors and is increasingly reflected in the criteria used for selecting projects and sectors for assistance, In applying these criteria, greater weight is often given to increases in the income of lower-income groups.3 This can be done crudely by identifying a specific target group, such as poor farmers, or more systematically by using a welfare function that gives increasing weight to consumption or income in proportion to the poverty of the recipient. In applying this pr'nciple, Ahluwalia and Chenery (1974) proposed for the evaluation of domestic poverty programs a simple weltare function that can be extended to cover international allocations as well. It replaces the income weights implicit in the use of the increase in total GNP as a measure of welfare by a more egalitarian measure such as population. As adapted to the evaluation of international programs this welfare measure would be: W = Z( 6) (1) N where n is the population of country i, gi the rate of GNP growth in country i due to receipts of aid and N= En,. In contrast, the conventional measure of growth of total GNP is equal to: G - (yigi) A (2) y y where Yi is GNP in country i and Y = Eyi. In equation (2) an allocation of aid that produces a given increase in aggregate GNP will have the same impact on total GNP growth regardless of how it is distributed among countries, while in equation (1) a 1 percent increase in an income of $100 is weighed equally with a 1 percent increase in an incomne of $1,000. If there is no change of income distribution within countries, equation (1) can also be interpreted as the average growth of income for each person in the developing world.4 To judge the existing allocation of aid, it is also necessary to make allowance for variations in its marginal productivity. Although this cannot be determined with any accuracy, it will often be higher in richer countries. However, the effects of moderate productivity differences are not likely to offset the effects of large differences in income level.5 Allowvance should also be made in intercountiy comparisons for the effectiveness of different countries in reaching the poorer sections of their population. In principle, this can also be done by replacing the growth of GNP in country i in equation (1) by a poptilation--weighted average. The actual allocation of aid aimong countries is made by a number of individual agencies--bilateral and multilateral withi some form of consulta- tion concerning the larger aid recipients. Aklthough each agency has its own Aid and Income Distribution 29 mixture of motives and .;riteria, there are some common elements in the approaches taken to aid allocation, which combine notions of equity and efficiency with aspects of national interest. In recent years, most of these agencies have moved in the direction of criteria that are more consistent with those suggested above. There are, of course, many practical limitations to applying such criteria systematically. For one thing, it is not possible to obtain very satisfactory measures of the overall productivity of foreign aid, or of the effectiveness of domestic programs for improving income distribution. Considerations of national economic or political interest of the donors will inevitably modify considerations of global equity in the allocation of bilateral aid. For multilateral lenders, the bulk of whose funds must be raised from capital markets, important constraints are imposed on U.acreasing allocations to poor countries by their limited capacity to service debt. Nevertheless, we believe there is significant scope for further progress in the application of equity criteria in aid allocation, particularly if the total supply of concessional assistance can be expanded at least moderately in the future. Supply of External Capital In response to the large increase in demand for external capital in the past three years, there has been a more limited increase in supply and a greater need to ration concessional public funds. The major part of this increased demand was met by expanded private flows. Of a $19.4 billion increase in the total capital inflow to oil-importing developing countries between 1970 to 1q72 and 1974, some $12 billion was from private sources, while another $1.3 billion was provided by the IMF. Nevertheless, the increase in official long-term capital flows to this group of countries was impressive: in nominal terms, it rose from a net flow of $5.8 billion in 1970 to 1972 to $11.8 billion in 1974, and about $14.0 in 1975. The increase in net flows of official development assistance (ODA) from OECD countries in 1974 reversed-at least temporarily-the steady decline in its share of OECD GNP, from a low of .30 percent reached in 1973 to .33 percent in 1974, and .36 percent in 1975.6 Commitments of official assistance-which constitute the main focus of this paper-also doubled during the recent past from $11.5 billion in 1970 to 1972 to $24 billion in 1975 (see Table 1.1). At constant (1970-1972) prices, thle 1975 increase over 1970 to 1972 is nearly 50 percent. MTore than lhalf of this increase was provided by the OECD countries, primarily througlh the multilateral institutions. Nearly half was provided by OPEC member coun- tries. While the distributio-i of the increase in the OECD and multilateral aid was quite wide, the distribution of OPEC comnmitments was much more concentrated, with over half the total going to three countries: Egypt, Pakistan, and Syria. Despite these large increases in extemal flows (and future commitments), J. A. Edelman, H. B. Chenery 30 Table 1.1 Commitments of Official Concessional Assistance (OCA) to Identified Countries, 1969-75 (billion $) A. Current Prices 1969 1970 1971 1972 1973 1974 1975 est. Bilateral ODA 5.6 5.9 7.2 8.2 8.5 10.2 (10.8) Multilaterala 3.1 3.4 4.0 5.0 6.3 8.0 (9.4) Subtotal 8.7 9.3 11.2 13.2 14.8 18.2 (20.2) OPEC (.2) (0_.) (0.2) 0.3 0.7 3.8 (3.84 Total OCA 8.9 9.5 11.4 13.5 15.5 22.0 (24.0) B. 1970-72 Prices Bilateral ODA 7.4 6.9 7.1 7.3 6.9 7.7 (7.7) Multilaterala 4.1 4.0 3.9 4.4 5.1 6.0 (6.6) Subtotal 11.5 10.g9 11.0 11.7 12.0 13.7 (14.3) OPEC 0.3) () (0.2) 0.3 0.6 2.9 (2.7) Total OCA 11.8 11.1 11.2 12.0 12.6 16.6 (17.0) IBRD Commitment Deflator: (1970-72 = 100) 76.1 85.8 101.4 112.8 123.4 132.8 141.4 Source: OECD, supplemented by Bank staff estimates. The commitment de- flator takes account of projected inflation rates during the disbursement period, aIncludes IBRD commitments which strictly speaking do not meet the OECD definition of concessional assistance. The latter requires a grant element of 25 percent (calculated at a 10 percent opportunity cost) where IBRD commitments on this assumption had a grant element of only 19.5 percent during the 1969-73 period and a 12 percent grant element during the 1974- 75 period after the interest rate was raised. However, at a 12 percent official concessional assumption, the grant element of IBRD was 22.5 percent in this period. many developing countries have had to reduce their growth rates because of shortages of foreign exchange. For the low-income countries as a whole, the increase in total flows between 1972 and 1975 was only two thirds of the losses they suffered from the deterioration in their terms of trade in that period. For the middle-income group, the increase in total capital flows was moderately greater than the terms of trade loss, but they also suffered a substantial reduction in export volumes during 1975 as a consequence of the recession in the developed countries. Moreover, their medium-term debt outlook has become somewhat worse as a result of heavy private borrowing. The future growth of both groups of countries, but particularly of the low-income group, will depend in substantial measure on the volume of concessional assistance made available over the next several years, and on the way in which these funds are allocated. Trends in Aid Allocation Since even the most optimistic estimates of aid supplies fall substantially Aid and Income Distribution 31 short of the requirements of the poorer countries, we will concentrate our analysis on the possibilities for making more effective use of the amounts that may be available. Our starting point is an analysis of recent trends in aid allocation in which we compare the allocation patterns since the economic crisis of 1973/1974 to earlier periods. We then discuss the possibility of further reallocation to the poorer countries. Methodology In order to analyze the net relationship between aid allocation and poverty, it is necessary to allow for country size and other factors that have been shown to affect the allocation of aid. This is done by means of multiple regression analysis, taking advantage of the earlier findings of Strout (1966), Henderson (1971), and Isenrman (1975). In its 1974 Review (OECD, 1974), the Development Assistance Committee of the OECD also analyzed overall allocation patterns for the period from 1969 to 1974, but it did not separate the effects of individual factors. The principal innovations in the present study are: 1. the use of commitment data-which are indicative of donors' reactions to changing needs--instead of disbursements; 2. the extension of donors to include the OPEC countries, whiclh have ac- counted for nearly half of the recent increase in commitments;1 and 3. the use of grant equivalent, as well as nominal. values, which provides a more valid basis for comparing the value to the recipients of different forms of aid. Our sample consists of eighty-nine aid recipients that cover 85 percent of the commitments included in the more complete OECD tabulation for 140 countries.8 Separate regressions have been conmputedl for three timne periods: 1967-i969, 1970-1972, and 1973-1974. The low absorptive capacity and limited creditworthiness of mnany poor countries limits the aid they receive, so that per capita aid receipts first rise and then fall as per capita income rises. To capture this effect statistically, we have used a regression equation that is nonlinear in income.9 The form of the equation is adapted fromn the basic equation used by Chenery and Syrquin (1975) to analyze a variety of development patte.ns. The principal equation used is: logX = a +b logN+c logy +d(logy)2 + e(E/Y) where X = average per capita coimmrilitnients for the relevant time perioLd N = population, (3) Y = per capita income, E/Y = the ratio of exports to GNP for 1972.10 Six groups of per capita co,mmitments were tested for each time period: bilateral, multilateral, and total, both in nominal values and grant eqtuivalents. In addition, tests of the strength of the results wvere carried out by usinlg total J. A. L:delinan, Ff. B. Chenery 32 rather thani per capita coniiiiitnienlts and oni per capita commitmnents for a reduced sample, which eliminates seven couintries with extremely high per capita commitments. (Regression results are ..% .ilable from the authors.) Effects of Population Size Although our main interest is in the relation between aid allocation and income levels, it is necessary to allow first for the effects of size. A numln)ber of earlier studies have remarked on the apparent bias in per capita aid allocations in favor of small countries and against large countries.11 Table 1.2 shows these relationships for the three time periods we are considering. Between 1967 to 1960 and 1973 to 1974, average conmmlitnments to countries over 10 nmillion population increased by sipificantly more than those to smaller countries, thereby reducinlg the population bias somewhat,f2 Nevertheless, the differences remained striking; in 1973 to 1974, countries under 2 million population received more than twice the average for all other groups, while countries over 25 million received only one quarter to one tlhrd the average for smaller counitries. The regression results for total conllilmit- ments indicate that in 1973-74 a country with twice the population of another had on average a commitment level 35 percent lower in nominal terms and 44 percent lower on a grant c(quivaletlt basis."3 The separate regressions on bilateral and miulltilateral comimniitiienits in Table 1 . sliow a Table 1.2 Per Capita Commniitnments by Country Size (UMnveigh ted averages, based on 1970 population for all periods) Nunuber of 1970 Population Countries in Group Nomninal Values (US$) (millioni) 1967-69 1)70-74 1967-69 1970-72 1973-74 Less than 2.1 18 21 23.7 25.1 42.3 2.1-5.0 26 26 13.8 16.1 31.0 S.1-10.0 11 11 10.2 9.3 20.1 10.1-25.0 16 16 5.3 7.4 12.8 Over 25 14 15 3.1 5.4 9.7 Total/Average 85 89 12.0 14.0 25.5 Grant qLLlivalents (.L:';) 1967-69 1970-72 1973-74 Less than 2.1 16.7 18.2 31.6 2.1-5.0 10.4 11.7 23,8 5.1-10.0 7.3 6.8 16.3 10.1-25.0 3.5 4.2 8.2 Over 25 2.1 2.9 5.6 Average 8.7 9.8 18.8 Aid and Income Distribution 33 RELATIONSHIP BETWEEN PER CAPITA COMMITMENTS (GRANT EQUIVALENT) AND COUNTRY SIZE (COMMITMENTS DEFLATED TO 1970-72 PRICES) 2C 18 15 t: 14 0 z 1l2 C'U U* 10 0 "I 4 S t 20 3 40 50 100 700 .300 400 1000 POPULATION WMILL2ONS3 LOG SCALEi Notes: 1. These curves are based on the regression results given in Annex Table 1. Per Capita income and the ratio of exports to GNP have been fixed at their mean values, Vls. Mean of Income/Capita 1967- 69: 185.49 (US$) 1 970-72: 234.39 1 973-74: 262.96 Mean of Exp/GNP 1967-69: .271 1 97 0-72 & 73-74: .267 2. Commitment Deflators in both Fig. I & Fig. 2 are: 1 967^-69: .76 1973 74: 1.28 World Bank-16511 Figure 1.1 Relationseip bet.een per capita commitments (grant eqoivalent) and country size (commitments deflated to 1970-72 prices) J. A. Fdelinain, ff. 13. (henery 34 very similar pattern. Figure 1.1 shows these relationships in graphic form for total grant equivalents. There are a number of explanations for this strong negative relationisliip between per capita commitments and population size. The most important single deterniiiiinant is probably the fact that all the nation-states included have independent foreign policies that can sometimiies be illffluenced by forcign aid. Thus, these states tend to attract funds frorn a numiiber of donors who may wish to exert such influence. In addition, relatively more of the small countries than the largc tend to have close political, cultural, anld economic ties with one or mlore donor."4 These countries are also all members of one or more muLIltilateral lenidinig agencies. Niiltilateral leniders are virtually obliged to do some leniding to all members, and constraints of minimum project size often lead to relatively large per capita coiiiiiitniients in small countries. Another important factor is the relatively large import component of GNP (and investment) in these countries. The large country extreme in this respect is India, whose imports accoLunt for only 5 percent of GNP. Thus, although concessiotnal comnrilltiiuenlts are only sonme 1.7 percent of GNP, they corre- spond to 35 percenit of imtiports, a relatively large slhare. A substantial re(duictioni of the small country bias couild free a significanit amount of resources for reallocationi to larger poor coLunltries. The rorty-sevell countries with populations inider 5 million accoLun tedi in 1973/1974 for some $2.6 billion in grant equivalents or 20 percent of the total but only 7 percent of the population covered by this samlple. A reduction by one third ill these commitments would have pernnitted an inicrease in the averages for the fifteen largest countries by nearly $0.9 per capita which, in the case of India, would correspond to a 40 percent increase in the average 1973/1974 coninilitillent level. Clearly, it is unrealistic to expect that stulCh a large transfor could be achieved, althouglh some reduction in the hias in favor of very small countries did take place between 1967 to 1969 and 1973/1974. The allocation criteria outlined above would argue for efforts to achieve at least a miioderate further reduction in the future. Effects of Income In recent years, some 35 to 40 percent of total concessional comnmlitmlienits in nominial terms has gone to oil-importing cotutlries under $200 per capita. As shown. in Table 1.3, tlherc was little change in this slhare between 1967 to 1969 and 1973/i974.'5 Tlowever, the grant e(quivalents for this group increased sligltly from 44 percent in 1967 to 1969 to 46 percent in 1973/1974. This was the result of an incre:ise in the g¢ranit clemernt of commiiiiitmiients froni OECD) and mnultilatcral doonors from 77 percent to 80 percent. By conitrast, the grant celement of commlliitmiients to cotuntries over $300 per capita dropped fromii 56 to 47 percent over the period. The major teneficiaries of recent increases in coiminlitnl1enlts were in thi $200-300 group (mainly Fgypt and Syria). Betweeni 1967 to 1969 andl ,kid and Income Distribution 35 1973/1974 the share of this group rose from 16 percent to 24 percent of the grant equivalents. The share of all oil-importing countries under $300 in total grant equivalents had risen in 1973/1974 to 72 percent, compared with 61 percent in 1967 to 1969. Table 1.4 shows the grant equivalents of commitments on a per capita basis and as a percentage of GNP for the main income groups. During the period from 1967 to 1973 the average for the poorest countries (under $200) was consistently below that for all the other income groups except the hiighiest. The main reason for the low average of the poorest group is, of course, the great weight of India, which in 1973/1974 received only $2.2 per capita of grant equivalent commitments, compared to an average for all other coulntries of $9.5 per capita. Between 1967 to 1969 and 1974, conmnmitmnents to India rose only moderately at current prices, and declined in real terms. Commlnit- ments to all other countries in this group showed a rise in real terms of over 80 percent between 1970 to 1972 and 1974, partly in response to the oil crisis. The ratio of grant equivalent commitments to GNP rose from 2.6 to 4.0 percent for the poorest group of countries between 1970 to 1972 and 1973/1974. However, ratios for India were only 1.5 and 1.7 percent, respectively, while for the other poorest countries, the average ratio went up from 4.0 percent to 7.0 percent. In the regression analysis, the observations are unweighted, so that the Gambia has the same weight as India. The unweighted commiinitmiernt average for all countries in 1973/1974 was $19 per capita for grant equivalents, compared with the weighted average of $6.4 shown in Table 1.4. This difference, of course, reflects the fact that so many small countries receive relatively high per capita commitments. Even correcting for this small country effect, the allocation pattern shows a rise with income up to the $200400 range and then a decline, albeit a rather erratic one. The quadratic term for income used in the regression analysis is designed to capture this tendency for per capita allocations to rise and then fall. The curves reflecting the shape of the regression results for total grant equivalent commitments in the three time periods are given in Figure 1.2.16 The relationship between per capita allocation and income is significantly differ- ent as between bilateral and multilateral comnlitmenits. The regression curves for multilateral peak at a substantially higher income level than for bilateral, reflecting the fact that the largest component of mulltilateral lendin,g consists of relatively hard funds from the IBRD, the IDB, and the ADB, vliicili have to take account of creditworthiness considerations. The peak income levels foi per capita comminiitments implied by the sepai ate regressions on1 bilateral and multilateral commnitmnenits, togetlher with those for the totals, are shown on p. 38. The main reason that per capita lending in nominal terms tends to rise with income is that creditworthiness and absorptive capacity also rise with income. These considerations affect the pattern of grant equivalent commitments with J. A. Edelman, H. B. Chenery 36 Table 1.3 Distribution of Official Concessional Assistance by Income Group, 1967-74 No. of 1970 A. Concessional Cornmmitments Countries Population ($ billions) (70-74) (Million) 1967-69 1970-72 1973 1974 A. Non-Oil Exportersa Under $200 38 932 2.82 3.30 4.99 7.80 India 1 538 1.10 1.10 1.46 1.77 Other 37 394 1.72 2.20 3.53 6.03 $201-$300 14 148 1.13 1.61 2.72 4.78 Over $300 28 367 2.25 3.01 3.80 4.47 $300-$500 13 204 1.40 1.79 2.60 2.27 Over $500 15 163 0.85 1.22 1.20 2.20 Total of A 80 1448 6.20 7.92 1 1.51 17.05 B. Oil Exportersa 9 232 0.96 1.51 1.69 2.07 Indonesia 1 116 0.43 0.82 0.96 1.01 Other 8 116 0.53 0.69 0.73 1.06 Total of A + Ba 89 1630 7.16 9.42 13.20 19.12 Total of A + B in 1970-72 Prices 9.40 9.42 10.70 14.40 B. Percent of Total Commitmenlts 1967-69 1970-72 1973 1974 A. Non-Oil Exporters Under $200 39.4 35.0 37,8 40.8 India 15.4 11.7 11.1 9.3 Other 24.0 23.4 26.7 31.5 $201-$300 15.8 17.1 20.6 25.0 Over $300 31.4 32.0 28.8 23.4 $300-$500 19.6 19.0 19.7 11.9 Over $500 11.7 13.0 9.1 11.5 Total of A 86.6 84.1 87.2 89.2 B. Oil Exporters 13.4 16.0 12.8 10.8 Indonesia 6.0 8.7 7.3 5.3 Others 7.4 7.3 5.5 5.5 Total of A+B 100.0 100.0 100.0 100.0 aCurrent prices. Aid and Income Distribution 37 Table 1.3 (continued) C. Grant Equivalent ($ billions) 1967-69 1970-72 1973 1974 2.17 2.54 3.98 5.93 0.86 .84 1.12 1.27 1.31 1.70 2.86 4.66 0.85 1.08 1.90 3.34 1.27 1.27 1.72 2.19 0.87 0.83 1.23 1.06 0.40 0.44 0.49 1.13 4.29 4.89 7.60 11.46 0.62 0.95 1.14 1.39 0.27 0.58 0.73 O.'/ 7 0.35 0.37 0.41 0.62 4.91 5.84 8.74 12.85 6.46 5.83 7.08 9.68 D. Percent of Grant Equivalents 1967-69 1970-72 1973 1974 44.2 43.5 45.5 46.2 17.5 14.4 12.8 9.9 26.7 29.1 32.7 36.3 17.3 18.5 21.7 26.0 25.9 21.8 19.7 17.0 17.7 14.2 14.1 8.3 8.1 7.5 5.6 8.8 87.4 83.7 87.0 89.2 12.6 16.3 13.0 10.8 5.5 9.9 8.3 6.0 7.1 6.3 4.7 4,8 100.0 100.0 100.( 100.0 J. A. Edelman, H. B1 Chenery 38 Table 1.4 Commitments of Official Concessional Assistance: Grant Equivalent per Capita and as Ratio to GNP by Income Grourp, 1967-74 Per Capita Grant Equivalentsa (US$, current prices) Ratio to GNP 1967-69 1970-72 (percent) Ave.rage Average 1973 1974 1970-72 Av.b 1973/74 Av.c A, Oil Importers Under $200 2.3 2.7 4.3 6.4 2.6 4.0 India 1.6 1.6 2.1 2.4 1.5 1.7 Other 3.3 4.3 7.3 11.8 4.0 7.0 $201-$300 5.7 7.3 12.8 22.6 3.2 5.2 Over $300 3.5 3.5 4.7 6.0 0.6 0.5 $300-500 4.3 4.1 6.0 5.2 1.1 0.8 Over 500 2.5 2.7 3.0 6.9 0.3 0.4 Total of A 3.0 3.4 5.2 7.9 1.4 1.8 B. Oil Exporters 2.7 4.1 4.9 6.0 2.1 l.5 Indonesia 2.3 5.0 6.3 6.6 6.5 4.7 Other 3.0 3.2 3.5 5.3 1.0 0.7 Total of A +B 2.9 3.5 5.2 7.6 1.5 1.7 aAll based on 1970 population; this, of course, overstates the upward trend- by about 2.5 percent a year. bBased on 1970 GNP. cBased on 1973 GNP. Income Turning Points for Per Capita Commitments (per capita income, 1970 prices)a For Noiiinil Commitments For Grant Equivalents Bilateral Multilateral Total Bilateral M ultilateral Total 1967-69 259 675 396 225 439 275 1970-72 * * (392) * (193) 1973-74 153 406 278 * 174 154 arhe undterlying income data used in the regressions are at current prices, converted inito U.S. dollars at average exchiange rates. The estimates given here have hbe.n deflated to 1970 dollars using the GNP deflator for the United St;ates. *The income coefficients for these time periods are not statistically siLgnificant. 1970-72 data in b1rackets are derved from regressions for the reduced sample. For 1973/74, the reduced sample gives turning points of 262 (total nominal) and 169 (total grant equivalent). Aid and Income Distribution 39 RELATIONSHIP BETWEEN PER CAPITA COMMITMENTS (GRANT EQUIVALENTS) AND PER CAPITA INCOME (COMMITMENTS DEFLATED TO 1970-72 PRICES) 10 7- z . a: 0 U 2N 200 400 60 00 1000 1200 1400 1600 t800 2000 INCOME PER CAPITA (US$) Notes: 1, These results are derived from the regression results given in Annex Table 1. Population and the Ratio of exports to GNP are fixed at their mean values. vis. Pop. 1967-69: 5.18 m.*~ 1970-72: 5.47 ml. 1973-74: 5.69 m. Exports: GNP 1967-69: .271 t1970-72. .267 1 973-74: 2, The dotted line for 1970--72 indicates the co-efficients underlying this curve are n.ot statistically significant. World Bank-16512 F:igijre 1.2 R.elationship bvehveen per capita comm-itments (grant equivalents) and per capita income (commitments deflated to 1970 72 prices) J. A. Edelman, H. B. Clhencry 40 less force, since grants and grant-like commitments tend to dominate in that pattern. However, an awareness of the greater needs of middle- and lower- income countries has led both multilateral and bilateral donors to reduce lending to a number of higher-income countries in recent years, despite their high creditworthiness. The grant equivalent of multilateral commitments to countries over $300 per capita income declined from 40 percent to 26 percent of the multilateral total between 1967 to 1969 and 1973/1974. At the same time, the share going to countries under $200 (including Indonesia and Nigeria) rose from 44 to 56 percent. A third of the latter increase was accounted for by Indonesia, whose share rose from 2 to 6 percent of the multilateral total. Countries in the $200-300 income range had an increase in their share of the multilateral grant equivalent from 16 to 18 percent. The share of the bilateral ODA grant equivalent commitments allocated by OECD countries to countries over $300 per capita dropped from 24 percent to 17 percent between 1967 to 1969 and 1973/1974; while that going to the group under $200 per capita increased from 53 percent to 57 percent, despite a drop in bilateral allocations to India from 20 percent to 10 percent of the total. Bilateral ODA allocations to the $200-300 group rose from 23 percent to 26 percent over this period. OPEC assistance is also heavily concentrated in the latter category because Egypt, Syria, and Jordan are all in this group. Undoubtedly, somietlhing of a bandwagon effect has developed in lending to a number of countries in the $200-300 group. While some of these were hard hit by losses from the terms of trade in 1973-74, others are net oil-exporters and have benefited substantially from the recent increase in oil prices: e.g., Algeria, Congo, Syria, and Tunisia. In spite of a decline in the share of total concessional assistance going to countries over the $300 level, a number of these were also still receiving relatively large per capita commitments in 1973-74. However, most of these were countries under 5 million popu- lation.'7 Conclusions: The Potential for Reallocation The total welfare of developing countries can be increased by allocating a larger share of aid to poorer couintries so long as the productivity of aid at the mnargin is not so low as to offset differenices in income levels. Other things being equal, there is also a presumption that the marginal productivity of aid will be higher in countries receiving relatively low levels than in those receiving very highl levels. However, "other things" seldom are equal. Table 1.5 identifies fifteen oil-importing countries with per capita incomes under $201 in 1970 that received less than $10 per capita per annum in conces- sional aid (grant equivalents) during 1973 and 1974. Ten of these had receipts below $6.5 per capita--the %veiglifed average for all counitries covered by our sample in this period. On grounds of per capita income and aid levels alone, it would appear that Aid and Income Distribution 41 these ten would be the prime candidates for receiving higher allocations if aid were to be redistributed more equitably. However, one of the largest of these-Thailand-did not in fact have a pressing need for resource transfers in this period because of rising export earnings and comfortable foreign exchange reserves. Most of the others share two common characteristics-low absorptive capacity (a low marginal productivity of aid), and relatively weak political and economic ties with the major aid donors. The second factor is undoubtedly at least as important as the first in accounting for the large differences between the level of per capita aid for this group and that of the twenty-three small countries, grouped under "all other" in Table 1.5, that obtained average per capita commitments of $17.2 in this period."8 In the latter group, there are a number of countries where the productivity of aid appears to be quite high; however, there are some in which it seems to be as low as in those receiving low per capita aid allocations. The relatively high allocations made to these countries may be explained in part by an expectation that productivity will rise because of a willingness on the part of the recipients to make changes in domestic economic policies and adminis- Table 1.5 Grant Equivalent of Commitments of Concessional Assistance to Oil-Importing Countries Under $201 Per Capita Income, 1973/74 Average 1973/74 Grant Equivalent Commitments A. Commitments 1970 1970 Total Percent under $10 Per Capita Population Per Capita ($ of 1973 Per Capita (US$) (million) (US$) million) GNP India 110 538.1 2.2 1,193.4 1.7 Guinea 120 3.9 3.1 12.0 2.1 Burma 80 27.6 3.1 86.1 3.6 Sierra Leone 190 2.6 4.0 10.3 2.2 Nepal 80 11.1 4.3 47.9 4.4 Afghanistan 80 14.3 5.2 74.7 5.3 Thailand 200 36.2 5.2 186.9 1.8 Ethiopia 80 24.6 6.4 156.4 6.8 SriLanka 110 12.5 6.4 80.2 5.1 Haiti 110 4.5 6.4 29.0 5.1 Sudan 120 15.7 8.1 127.1 5.6 Bangladesh 70 67.8 8.5 574.0 9.7 Pakistan 100 62.4 8.9 556.8 7.2 Madagascar 130 7.3 9.1 66.6 5.3 Burundi 60 3.5 9.4 32.9 12.2 Subtotal 110 832.1 3.9 3,234.3 2.9 B. All Other Under $201 100 100.2 17.2 1,720.3 11.3 C. All CoLuntries Under $201 110 932.3 5.3 4,954.6 3.7 J. A. Edelman, II. B. Clienery 42 trative practices perceived by the donors to be important requirements for ensuring efficient use of aid. Even if it takes a long time for such changes to produce tangible results, donors are more likely to maintain relatively high commitments in countrles where they see such a willingness than in countries they believe to be unwilling or unable to make changes of this type. In these terms, a high degree of receptivity to aid may be said to "explain" relatively high commitments even if measured productivity remains low. In principle, of course, a reallocation in favor of the poor countries with low aid receipts could be made by the donors without evidence of greater receptivity or of increases in marginal productivity. In practice, however, it is unlikely that this will be done, if only because donor agencies need to justify their aid allocations to their legislatures or boards of directors in terms of some sort of positive response (actual or expected) to the allocations they provide. India is, of course, overwhelmingly the most inportant of the countries in this group. It is the lowest recipient of aid on a per capita basis, and the share of concessional aid going to India has declined steadily over the past decade. To some extent this is a reflection of competition from other countries for the available resources, resulting from thu crcationi of newly independent states in Africa and Asia and increased receptivity to aid in some colintries (for instance, Indonesia). However, the decline in India's share is also function of India's owIn policies directed at reducing dependenice on imports and on foreign assistance over the past decade. Althotugh plausible at the start-given the size of the Indian economy-these policies have proved to be a disappointment in that in recernt years they have produced very slow growth of real income (about 1 percent per capita since the early 1960s), food shortages, low productivity of capital, and severe foreign exchange con- straints. Moreover, receptivity to policy changes recommended by donors has generally been low. Recently, there have been signs of change. In addition to an excellent harvest in 1975, encouraging progress has been made in long-term agricultural development programs. Efforts to increase Indian exports have started to produce results, and some major import substitution investmlelnts in energy, steel, and fertilizers are -finally becoming productive. This has been accom- panied by significant increases in foreign aid commitments and flows in 1975 and 1976. If the recent improvements in India's economic performance can be sustained over the next five to ten years, it seems likely that aid agencies will be able to justify a further increase in India's share in the total of available aid resources. If there is to be a reallocation of aid in favor of the poorer cotiuntries with very low per capita aid receipts, the most obvious candidates for reduced shares would be higher-incoimie countries with relatively higlh per capita aid receipts. Table 1.6 lists by three incomle groups the couintries over $200 per capita with commitments that are substantially above the average for Aid and Income Distribution 43 countries under $200 per capita. It will be seen that a rough notion of progressivity is introduced by selecting lower commitment cutoff points to represent "substantial," as incomes rise, that is, for the $200-300 groups, fourteen dollars is used, while for the over $500 group, a level of seven dollars is used. The small country bias is clearly evident in this table. Of the thirty countries identified (in Group A) as being substantially above average, only six have populations over 5 million, and only two are over 20 million. Their total 1970 population was only 136 million, while commitments to them averaged $3.2 billion or twenty-three dollars per capita in the two years from 1973 to 1974. The remaining nineteen countries in our sample with incomes over $200 per capita had a total population of 441 million and received average per capita commitments of only $4.2 in this period. Special political and economic relationships with major donors undoubted- ly played an important role in determining the high levels of assistance to many of these countries. There is no reason to believe that these considera- tions will diminish greatly in importance for the main donors over the next few years. However, if the total supply of aid can be held at least constant in real tenrns, a gradual reallocation in favor of poorer countries might be achieved without unduly disturbing these special relationships. For example, if nominal commitments to these favored countries were, on average, maintained approximately constant, while inflation continued at about 6 percent a year, the result after six years would be to free about $1 billion in grant equivalents (at 1973/1974 prices) for transfer to the poorer countries. If this were devoted entirely to the fifteen poorer countries identified separately in Table 1.5, it would correspond to a 30 percent increase in their total commitment levels, raising per capita receipts from $3.9 to $5.0 in 1973/1974 prices. The corresponding reduction in real per capita allocations implied for the thirty countries identified in Table 1.6 (undey group A) would be from $23 to $16.3. One important qualification to the potential for reallocation needs to be noted here-namely, the limited scope for multilateral agencies that raise their funds on the capital markets to increase lending to poorer countries with very limited creditworthiness. About one quarter of the receipts of the countries shown separately in Table 1.6 are from imiultilateral agencies, and the great part of these derive from funds raised oni capital markets. As this implies, not all the comnmitments we have lumped together as "grant equivalents" are in fact fungible as among countries. At present, none of the poor countries listed separately in Table 1.5, except Thailand, can afford any substantial increase in their borrowings on conventional terms because of prospective debt servicinig problems. Of course, this outlook CoLuld change for some of them if recent imnprovemiients in ccoInomiiic policies can be reinforced and sustained over the next several years. But, in the meantime, this consideration means that any substantial reallocation to countries under $200 per capita will have to take place through shifts in the pattern of bilateral aid or in J. A, Edelman, H. B. Chenery 44 Table 1.6 Grant Equivalent of Concessional Commitments to Countries Over $200 Per Capita by Income Group and Level of Commitment, 1973/74 Grant Equivalent 1970 Per 1970 Capita Per Total Percent of Population GNP Capita ($ 1973 1970 Income Group (million) (US$) (US$) million) GNP $200-300 Per Capita Income A. Over $14 Per Capita Papua and New Guinea 2.4 300 113.7 272.9 26.0 Jordan 2.3 250 81.2 186.8 21.5 Syria 6.1 290 50,6 309.0 11.0 Mauritius .8 240 39.0 32.8 9.1 Congo .9 300 35.7 32.1 7.8 Senegal 3.9 230 32,5 126.7 10.9 Tunisia 5.1 250 32,5 165.7 6.5 Paraguay 2.4 260 24.6 59.0 5.9 Egypt 33.3 210 20.3 677.4 7.7 Honduras 2.5 280 18.8 47.0 5.3 Liberia 1.5 240 16.2 24.3 5.4 Algeria 14.3 300 14.1 201.9 2.4 Subtotal A 75.5 230 28.3 2,135.6 7.4 B. All Other in $200-300 Range 93.8 240 8.4 788.0 2,4 C. Group Total 169.3 240 17.3 2,923.6 4.7 $301-500 Per Capita A. Over $ 10 Per Capita Nicaragua 2.0 430 28.5 56.8 5.4 Fiji .5 430 2'/. 13.8 3.8 Ivory Coast 4.9 310 27.1 132.9 5.9 Zambia 4.1 400 26.3 107.9 5.3 Guyana .8 370 17.6 13.2 4.1 Dominican Republic 4.1 350 14.7 60.4 2.6 Subtotal A 16.4 360 23.5 385.0 4.6 B. All Other in $301-500 Range 216.6 380 3.8 818.2 0,5 C. Group Total 233.0 380 5.2 1,203.2 0.7 Over $500 A, Over $7 Per Capita Gabon .5 630 54.4 27.2 4,0 Barbados .3 610 41.5 10.8 4.5 Israel 2.9 1,960 39.6 114.8 1.2 Costa Rica 1.7 560 27.0 45.9 3.5 Aid and Income Distribution 45 Table 1.6 (continued) Grant Equivalent 1970 Per 1970 Capita Per Total Percent of Population GNP Capita ($ 1973 1970 Income Group (million) (US$) (US$) million) GNP Lebanon 2.7 590 26.8 72.3 2.6 Cyprus .6 950 23.0 13.8 1.5 Panama 1.5 730 21.9 32.9 2.3 Jamaica 1.9 670 14.4 27.3 1.4 Singapore 2.1 920 11.9 25.0 0.6 Chile 9.8 720 10.4 102.3 1.4 Trinidad and Tobago 1.0 860 9.5 9.5 0.7 Yugoslavia 20.5 650 7.5 153.4 0.7 Subtotal A 45.6 760 13.9 635.2 1.2 B. All Other over $500 129.0 900 1.8 237.5 0.1 C. Group Total 175.2 870 5.0 872.7 0.4 Total of Groups A 136.5 430 23.1 3,155.8 3,5 Total All Other 440.0 770 4.2 1,843.7 0.5 All countries over $200 Per Capita 576.5 690 8.7 4,999.5 1.1 decisions by those donors to channel more concessional funds through multilateral institutions. There may, however, be a case for some reallocation of conventional official capital to countries with relatively low receipts in the $200-500 income groups-especially to those where productivity of aid appears relative- ly high. Countries in this income group with per capita commitments of less than $7 on a grant equivalent basis in 1973/1974 include Brazil, Colombia, Peru, Philippines, and Turkey. For this group of countries, of course, another relevant consideration in any aid reallocition would be the extent of their access to private capital markets on reasonable temis. It is of interest to see what the impact on growth and welfare might be from the potential shift in aid to the poorest countries discussed. On the assumption that the illustrativ. $1 billion reallocation by 1980 were reached gradually, it would produce a cumulative shift of $4.4 billion during the six-year period, within an aggregate aid total of $38.4 billion for the fifteen oil importeis identified in Table 1.5 plus the thirty high aid recipients listed in Table 1.6. The average incremental capital ouLtpuit ratios (ICORs) for the past decade produce a handy (if limited) proxy of the productivity of aid for purposes of this illustration. For the poorest group, the average was about 5.5 for 1965 to 1973, while for the thirty high aid recipients, the ICOR averaged about 4.0 (For the three largest countries in this group-Algeria, J. A. Edelman, It. B. Chenery 46 Egypt, and Yugoslavia-the ICORS were 3 9, 4.6, and 4.2, respectively, while in most of the other count.ies, the ICOR was under 4.0.) Using these two sets of assumptions, the total growth in GNP attributable to aid in this period for these forty-five countries works out at 4.2 percent with constant shares and 3.8 percent after the assumed reallocation in favor of the poorest, However, as shown in Table 1.7, using population weights to aggregate country growth, the annual increase in W would be 3.5 percent on constant shares and 3.9 percent after the assumed reallocation. Even more substantial increases in total welfare than these could be achieved by an increase in the real level of aid, and/or increased productivity. For example, if total aid could be increased in real terms by 5 percent annually over the 1973/1974 level for the total of the forty-five countries considered in this illustration, their 1979/1980 level would rise to $9.1 billion, compared with $6.4 in 1973/1974, and the cumulative difference for the six years would be $7.3 billion. If this increment were allocated entirely to the poorest group, it would represent a 38 percent increase in the cumulative total for 1975 to 1980 projected on the assumption of a constant real level. The annual growth in welfare (based on population weights) would be 4.4 percent. The impact of a reduction in the ICOR for aid for the poorest Table 1.7 Illustrative Impact of Aid Reallocation on Growth and Welfare for 45 Countries 30 Upper and 15 Pooresta Middle Incomeb Totals Base Period GNP (billion $) 110 90 200.0 Base Period Population (nriflion) ?@9) 148 1047 Assumed ICOR (for aid) 5.5 4.0 Assumed Aid 197 5-80 (US$ million, 1973-74 prices) A (constant shares) 19.2 19.2 38.4 B (reallocated) 23.6 14.8 38.4 Percentage Growth in GNP, 1974-80 A (constant shares) 3.17 5.33 4.15 B (reallocuted) 3.90 3.73 3.82 Percentage Growth in W, 1974-80 (population weighted)c A (constant shares) 3.17 5.33 3.47 B (Teallocated) 3.90 3.73 3.87 aFrom Table 1.5. bFrom Table 1.6 (group A countries). cAs the identity between the GNP and population weighted growth rates for each group indicates, population weighting is applied here only to the totals for the two groups and not within each group. Aid and Income Distribution 47 group from 5.5 to 4.0 would be almost exactly the same. Productivity of aid measured in terms of its contribution to GNP growth is only one element in determining the welfare impact of aid. Another important element is the allocation of aid in relation to per capita income levels. Substantial progress has been made during recent years in reallocating aid in favor of countries with (1970) inicome levels below $300 per capita. However, the major beneficiaries have been in the $200 to $300 range. There remains a significant potential for further reallocation to the group under $200 per capita. The main "target" countries consist of a limited number (ten to fifteen) now receiving a comparatively low per capita level of assistance. The role of concessional aid is particularly important for these countries, which typically have very limited creditworthiness for private borrowing. The objective of increasing welfare in the developing countries as a whole will be best served by simultaneous efforts on the part of the donors to increase both the real level of total concessional assistance and the share of the total going to this limited group of poor countries. As a practical matter, however, achievement of the later objective will also require that the poorest countries improve their ability to make productive use of this aid. In determining the outcome for the poorest group as a whole, the performance of India, which comprises about two thirds the population of all countries in this category, will be crucial. Notes The views in this article represent those of the authors, and not necessarily those of the World Bank. We are indebted to Julian Bharier, Ian Little, Marcelo Selowsky, and Joris Voorhoeve for helpful comments. Pisei Phlong and Sayeed Sadeq assisted with the statistical analysis. 1 Measured on a 1967-1969 price base, the terms of trade loss between 1970 to 1972 and 1975 amounted to 14 percent of 1975 imports for oil importing countries with in- comes under $200 per capita. The loss for middle-income oil importers in the same peri- od correspond to 10 percent of 1975 imports. 2 Analyses of the factors leading to this conclusion are given in Chenery (1975), Tims (1975), McNamara (1975), and H19len and Waclbroeck (1976). 3 See, for example, Little and Mirrlees (1968) and Squire and van der Tak (1975). 4 This function has also been used by Kuznets (1972), wlho showed thait for the 1960s the value of W was about 1 percent lowver than the value of G because of the higher growth rates of the richer countries. 5 For example, assume that Mexico has a level of per capita income (in purchasing power terms) four times that of India and that a loan of $10 million will yield a net increase in Mexican CGNP twice as large as that in India, On these assuniptions, the allocation of a loan to India would produce twice as large an increase in W even though the alluca [ion to Mexico would yield twice as great an increase in Gr. It should also be noted that applica- tion of the same criteria would lead to a reallocation away from low-income countries where marginal productivity of aid is very low to couuntries with moderately Ihiglher in- come, but with substantially higher marginal productivity of capital. J. A. Edelnman, H. B. Chenery 48 6 OECD (1975) and OECD (1976). 7 Inclusion of concessional assistance from OPEC countries yields a commitment total that we have labeled official concessional assistance COCA), as shown in Table 1.1. 8 The latter tabulation is slhown in Table 1.1. Our smaller sample is limited to countries that are active borrowers from the World Bank. The bulk of the difference between the two is accounited for by the countries or Indocihin;i and the dependent overseas terri- tories of France, Netherlands, the United Kingdom and the United States. 9 Failure to use this quadratic form may account for the findings of Cline and Sargen (1975) that per capita income had no influence on the allocation of World Bank commit- ments in grant equivalent tetm for a sample of nineteen countries from 1969 to 1972. Separate regressions using the quadratic form prepared by the bank staff for a larger sample of World Bank commitments from 1970 to 1974 show per capita income was significantly correlated with per capita commitments nmcasured on botlh a nominal and grant equivalent basis. However, it is possible that even with the iiso of the quadratic form, per capita income may not hiave been significant for the time period chosen by Cline and Sargen. 10 Preliminary tests were also made on a number of other variables: losses or gains from terms of trade, savings rates, GNP growilll rates, and several subjective measures of per- formance in equity and economic management. Resul ts of these tests have so far proved inconclusive and are not discussed here. 11 Little and Clifford (1965), Strouit (1966), OECD (1969), Henderson (1971), Isenman (1975). 12 This decline io rtflected in a fall in the ntegative elasticities given by the population coefficients for these two time periods in the regressioni results. 13 The population oefficients have high T ratios (ranging from 6 to 8) indicating that these results are quite stable and would not be much affected by the elimination of a few extreme country cases. 14 Dudley and Montmarqtuette ( 1976) argue that the entire small country bias in bilater- al commitments can be explained by l actors such as these. 15 Comprehensive coninmitment t data on a comparable basis are not available prior to 1967. lowever, flow data for net official assistance compiled by OECD indicate that the share of countries with incomes under $200 per capita income in 1970 declined from 43 percent in 1960 to 1966, to 42 percent of the total in 1968 to 1970. However, there were major shifts within this group duirinzg the period, with the share of India and Pakis- tan dropping from 26 to 21 percent and that for Nigeria and Indonesia increasing from 2.5 to 7.5 percent. 16 As indica ted in Fligure 1.1, the coefficients for the income variables in the regressions on the fuil sample for 1970 to 1972 are not statistically significant at thie 10 percent level. However, they are marginally significant for the other two time periods, and for the reduced sample they are highll signifiLant in all three periods. 17 The supporting statistical evidence regarding the statements in the text is availahle from the authors. 18 The list of these countries is availahle from the authors. References Ahluwalia, M. S., and Hollis B. Chenery in Redistrilbitioti vith Growth, by ('henery, Ahluwalia, Bell, Duloy, Jolly. Oxford University Press, London, 1974, Aid and Income Distribution 49 Chenery, Hollis B., "Approaches to Development Finance" in Symposium on a New International Order, The Hague, Netherlands, May 1975. Chenery, Hollis, B., and M. Syrquin, Patterns of Development 1950-1970, Oxford Uni- versity Press, London, 1975. Clifford, J. M., and Ian M. D. Little, International Aid. George Allen and Unwin, Ltd., London, 1965. Cline, William R., and Nicholas P. sargen, "Performance Criteria and Multilateral Aid Allocation," World Development, Vol. 3, No. 6, June 1975. Dudley, L., and C. Montmarquette, "A Model of the Supply of Bilateral Foreign Aid," American Economic Review March 1976, 66, 132-142. Henderson, P. D. "The Distribution of Official Development Assistance Commitments by Recipient Countries and by Sources," Bulletin, Oxford University Institute of Eco- nomics and Statistics, Vol. 33, February 1971, pp. 1-20. Holsen, John A., and Jean L. Waelbroeck, "LDC Balance of Payments Policies and the International Monetary System," World Bank Staff Working Paper No. 226, February 1976. Isenman, P., "Biases in Aid Allocations Against Poorer and Larger Countries," IVorld Development, Vol. 4, No. 8, 1976. Kuznets, S., "The Gap: Concepts, Measurement and Trends" in The Gap Betveen Rich and Poor Nations, ed. G. Ranis, MacMillan, London, 1972. McNamara, Robert S., Address to Board of Governors, World Bank Group, September 1975. Organization for Economic Cooperation and Development (OECD), Development Assis- tance, 1969 Reviewv, Paris, 1969. OECD,Development Cooperation, 1974 Review, Paris, 1974. OECD, Development Cooperationi, 1975 Review, Paris, 1975. OECD, Press Release of the Development Assistance Committee, June 29, 1976. Squire, Lyn, and Herman G, van der Tak, "Economic Analysis of Projects," Johns Hop- kins University Press, Baltimore and London, 1975. Strout, A., and P. Clark, "Aid, Performance, Self-Help, and Need," AID Discussion Paper No. 20, Washington, D.C., 1969. Little, Ian M. D., and James A. Mirrlees, Manual of Industrial Project Analysis, OECD, Paris, 1968.