Levels of Povert Policy d Change SWP401 World Bank Staff Workirng Paper No. 401 July 1980 I.41 ERNATiONAIV ONtTA2Y FUND JOINT LERARLY SEP I. 0 1931 IDTERjNATUNAL SANKFOR RECONSTRUMON AND DFVELOPhiENT WAP&IIINGON, D.C. 20161 Prepared by: Amartya Sen' Developmeri j Copyright e 1980 The World Bank 1818 H Street, N.W. Washington, D.C. 20433, t The views and interpretations in this document are those of the author and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. LEVELS OF POVERTY: POLICY AND CHANGE Contents Page No 1. Concepts of Poverty 1 1.1. Absolute and Relative Standards 1 1.2. Variability of Nutritional Norms 3 1.3. Results, Not Inputs? 5 1.4. Distribution and Vulnerability 7 2. Poverty, Longevity and Income Levels 9 3. Changes in Longevity and Literacy 17 4. Growth, Private Enterprise and Public Policy: Case Studies" 31 5. Distribution, Social Services and Public Policy: Case Studies 40 6. Entitlements and Levels of Explanation 53 7. Starvation and Entitlement Failures: Case Studies 65 8. Advise, Dissent and Critical Assessment 75 LEVELS OF POVERTY: POLICY AND CHANGE 1. Concepts of Poverty 1.1. Absolute and Relative Standards It is natural to view poverty as failure to meet the basic requirements of a "decent" life. The concept of a "decent" life does, of course, vary from society to society (as Adam Smith (1776) had noted). While biological requirements and nutritional norms provide the most elementary concept of "basic requirements", the notion of deprivation has to be considered in broader terms to come to grips with the modern understanding of poverty. Over the last century, not only has the context of "basic requirements" been progressively revised, so has been the very method underlying the specificat- ion of such requirements. The contrast between the nutritional norms used by, say, Booth (1889) and Rowntree (1901), and the rather more demanding requirements used for analysing "relative deprivation" in works by such authors as Runciman (1966) and Townsend (1974), reflect one such shift. While this shift has been much discussed, it is possible to argue that both absolute and relative deprivation are essential ingredients of the common understandinq of poverty. If people are dying of hunger in a famine situation, it is legitimate to see it as a case of acute poverty even without supplementing the analysis of the obvious absolute deprivation 1 For helpful comments and suggestions, I am most grateful to Akbar Noman and Adrian Wood. 2 by some detailed probe into the relative picture. On the other hand, even if no one goes hungry, but some are terribly deprived compared with others and see their relative deprivat- ion as acute, then it is legitimate to diagnose poverty, even though the criteria here are entirely relative rather than absolute. It is fair to say that there have been many influences working over the years towards making the relativist perspect- ive more important. Not the least of these influences Xs the very fact of economic progress, which has made poverty and deprivation in terms in which Booth or Rowntree saw them increasingly rare in many countries of the world. The attent- ion has naturally shifted to more demanding requirements reflecting the community's new-found standards. There is, however, a danger in being too "modern" about the notion of poverty, since economic progress has been rather uneven internationally, and for many countries in the developing world, the s-tandards used by Rowntree and Booth still remain depressingly relevant. In studying poverty in the so-called Third World, tlhe older notions of absoluLe deprivation still have much to contribute. But,at the same time, the observation of the progress of advanced capitalist and socialist countries has aroused expectations and led to the realization of possibilities which make it natural to go beyond the calculus of just hunger and nutrition in estimating and analysing poverty. The standards of absolute deprivation themselves have had relativist influnc(..:es arising from international contrasts. There is, of coursc 3 nothing astonishing in this development of the concept of poverty. While amalgamating absolute and relative considerations is a natural way of taking a broad international view of poverty, a number of more specific issues have been raised recently which require attention. These are taken up in the next three subsections. 1.2. Variability of Nutritional Norms First, the cogency of nutritional norms has been questioned because of the interpersonal variability of' nutritional requirements and the existence of "adaptive mechanisms" operating over time (see particularly Sukhatme (1977, 1978), and Srinivasan (1979)). The critique acquires particular force in view of the sensitivity of the poverty measures to the precise nutritional norm chosen, e.g., varying the nutritional norm from 90% of average requirement to 110% of average requirement apparently increases the proportion of the malnourished in Pakistan from 0% to 90% (Scandizzo and Knudsen (1979), cquoted by Srinivasan (1979, p.14)). Variation of nutritional needs can be related to group parameters such as race, climatic conditions, etc., but it can also occur as interpersonal variation even for the same group-characteristics. The former can be explicitly taken-> into account in choosing nutritional norms for different countries and even different groups within a country. The latter is more problematic, but if the interpersonal variations follow some simple pattern, then the average deprivation can 4 be used to get an idea of the pattern of deprivation clustered around that average. Given the adaptive mechanisms at work, it is important that nutritional statistics be taken over longer periods, and not be hooked to day to day variations (see Srinivasan (1979)). The problem of variability of nutritional norms is a deep one, and the uncritical use of the "nutritional approach" does deserve criticism. -However, in making this criticism it is possible to overlook a simple point, to wit, malnutrition can provide a basis for a standard of poverty without poverty being identified as the extent of malnourishment. The level of income at which an average person will be able to meet his nutritional requirements has a claim to being considered as an appropriate poverty line even when it is explicitly recognized that nutritional requirements vary interpersonally around that average. To state that, say, "20 per cent of the population failed to have incomes adequate for buying enough food to meet the average nutritional requirements for that community" is a statement about poverty of some interest of its own, even though it is not at all equivalent to saying that "20 per cent of the population failed to meet their nutritional requirements". The two statements are of interest for rather different reasons: the first enlightens us on income deprivation related to some average standard (paying no attention to the fact that some are luckier than others in terms of need for income in meeting nutritional requirements), while the second throws light on the prevalence-of actual malnourishment. Indeed, Rawlsian analysis of justice in terms of holdings of 5 "primary goods" amounts to deliberately choosing the. former focus (see Rawls (1971, 1975), and while I have elsewhere tried to argue that it is an inadequate focus (Sen (1979c)), it can hardly be denied that it is of some substantial interest of its own. Thus, considerations of average nutritional requirements can be used for one perspective on poverty even when nutritional requirements vary from person to person. 1.3. Results, Not Inputs? Second, there have been strong arguments in favour of making poverty indicators geared towards measuring "results, not inputs" (Morris (1979, p.32)), and there is by now a long established tradition of preferring social indicators such as life expectancy, literacy, etc., over malnutrition or shortfall of incomes. Various composite indicators of results have been proposed, and recently such indicators have been extensively used instead of nutritional statistics and income data. There is indeed much to be said for focussing on achievements rather than on contributory causes. Longevity, in particular, is a widely approved index, and it is natural that it should play an important role in the construction of modern social indicators, such as Morris's (1979) "physical quality of life index". On the other hand, there might be a danger in taking too narrow - and too "final" - a view of results. "Death" is, of course, an important possible result of nutritional and other deprivations, but it is not the only possible result. The suffering from hunger and other 6 deprivations is also a "result". Lack of food may lead to death (a very bad thing1), but it can lead also to hunger without leading to death (also a bad thing, even if not quite so bad as death). The typical "quality of life index" takes no note of the latter when it occurs without the former. But it would be difficult to claim that suffering from hunger does not affect one's quality of life unless one happens actually to die frcm it! One can, in fact, argue that the expectation of Wife as an indicator should be used not in place of income statistics, but married to income statistics. A poor person with low income and low expectation life suffers on both counts, and neither can represent the other in any fundamental sense. In an earlier paper (Sen (1973b)) the case was made for judging individual advantage in terms of expected life-time income, which multiplies the expectation of life with the expected value of mean annual income, and thus the GNP statistics and longevity statistics have to be put together to judge over-all advantage or disadvantage of a person. One result of such a procedure is to increase - rather than decrease - the interpersonal and international gaps, since lower income is typically associated positively with lower longevity, and the product of the two will slhow a bigger relative gap than either. This result is in marked contrast 1 Wilfred Beckerman (1979b) argues powerfully that it isn't a bad thing in itself, and while I don't agree, I also believe that anyone believing life to be valuable in itself must sort out for himself precisely where one might disagree with Beckerman's reasoning. 7 with replacing income statistics by longevity statistics (or using a "physical quality of life index" based on non- income variables such as life expectation), which lowers the gaps identified by income-based exercises (see Morris' (1979) findings with his index based on life expectation at age 1, infant mortality up to age 1, and the literacy rate). The important issue isn't the choice of a particular grand measure of poverty or opulence, but to take note of the various parameters that would enter an acceptably broad picture of poverty or opulence. The non-income factors captured by physical quality of life indices (particularly longevity and literacy) are important, but so are income and consumption statistics which have relevance that go well beyond longevity or education. 1.4. Distribution and Vulnerability Third, no matter what parameter is chosen to reflect individual poverty, there is the further problem of aggregation to arrive at an over-all poverty irndex. The most commonly used measures tend to neglect the distributional questions implicit in aggregation. The "head-count" measure of poverty, for example, simply adds up the number of people whose income fell short-of: the "poverty line", irrespective of how far . short they respectively happened to be. Various distribution-sensitive methods of aggregation have been recently proposed (see, for example, Sen (1976a)).2 2 See also Sen (1973a, 1977a, 1979a), Anand (1979), Kakwani (1978), Osmani (1978), Hamada and Takayama (1978), Takayama (1979), Thon (1979), and Blackorby and Donaldson (1980), among others. . 8 These take note not merely of the number of people falling below the poverty line and their average "poverty gap",3 but also the distribution of the actual poverty gaps. However, there is an aspect of the "distribution problem" that is overlooked in these measures of poverty based on realized deprivation. People's suffering from poverty in situations of variability depends not merely on actual shortfalls, but also on the possibility of shortfalls and the sense of vulnerability. For a peasant to have to look nervously at the sky searching for monsoon clouds is als.o poverty. In fact, sources of vulnerability for many classes of people in the developing countries extend well beyond uncertainties of nature, and the economic systems create fragility of various kinds.4 A person acquires income, or food, or other necessities of life,by using the economic system in operation with its legal and social structure. A labourer acquires his means of survival by selling his labour power and buying goods with it. Unemployment without social security,or other social support, can make him go under. A craftsman may have to starve 'if the demand for his craft products falls suddenly. This question of vulnerability is difficult to capture in quantitative terms, but it is an important aspect of poverty, and does deserve attention. It will be taken up later in terms 3 See Beckerman (1979a) fo-r illuminating use and defence of the alternative poverty-gap approach. 4 See Sen (1977b, 1980a, 1980b). 9 of particular experiences of economic insecurity. 2. Poverty. Longevity and Income Levels Reliable poverty estimates are difficult to find for many countries and they are very often non-comparable. Also, the typical measure chosen is the simple one of head-count ratio, i.e., the proportion of population having less t an 5 the poverty line income. Even when good estimates exist, it is usually quite difficult to use them for intertemporal or international comparison. In checking comparative performance of different countries in reducing poverty, the comparability needed is both over time and across countries, since it requires a ranking of the intertemporal performance of different countries. There have, however, been some attempts at providing comparative poverty data, and the results of one such attempt will be used here. Ahluwalia, Carter, Chenery and Development Policy Staff (1979) provide comparative estimates of head- count poverty measure for 36 countries. ti pTvhrty i1Lne chosen is based on calorie requirements and constmlption behaviour observed in the Indian economy (a country which is distinguished both in terms of a large number of detailed poverty studies6 as well as in terms of having almost certainly 5 There have, however, been some interesting empirical attempts recently to apply the distribution-sensitive measures of poverty to the data of particular countries or regions; see, for example, Ahluwalia (1976), Alamgir (1978), Anand (1977), Bhatty (1974), Dutta (1978), Fields:(1979), Fish]ow (1977), Kakwani (1978), Osmani (1978), Sastry (1978), Seastrand and Diwan (1975), among others. These have enriched our understanding of poverty problems in particular countries, but cannot be easily used to construct an internationally comparable picture. 6 See Srinivasan and Bardhan (1974) and the references to the Indian literature there. See also Lipton (1977). 10 the largest.absolute number of the poor in the world according to virtually all standards of comparison). The specified calorie requirement of 2250 per day seems to be reached in India by around the 45th percentile income group, and .Ahluwalia et.al. took that income level as the poverty line for other countries as well. In Table 1 the poverty estimates for 1975 are provided for 34 countries, being all countries in the Ahluwalia-list which happened to have an income level below $1,000 at the official exchange rate (and, equivalently' in this particular case, below $2,000 with the Kravis-adjustment- factor). The GNP per head of each country for 1975 with Kravis- adjustment is also provided. Table 1 presents in addition the expectation of life at birth at each country fDr 1977 as given by the World Development Report 1979. The first remark that I would like to make on the data presented in Table 1 is the obvious one that GNP per head and poverty relates quite closely. (The value of Spearman's rank correlation coefficient is, in fact, .89, with z having the convincing value of 5.10,) However, there are some differences between the two rankings and they are, of course, very worth pursuing to get insight into policy issues. In terms of lowness of poverty, the top three countries in this list are Yugoslavia, Taiwan and South Korea. This is of interest, but also of interest is the tendency of some countries to have poverty levels a good deal lower than their income rank would suggest. There are many ways of tracing that difference, and the approach that will be used here is the purely "ordinal" one of comparing reverse poverty 11 ranks with GNP ranks and looking at the excess of the former over the latter as an indicator of success. This devised index of "net excess of poverty score" might appear to be somewhat obscure. Certainly the more commonly used procedure would be some type of "regression analysis" and looking at how a country is placed vis-a-vis the regression line. There-is, however, some spurious precision in using numerical values of income and poverty to fit regression lines. A more "ordinal" approach would be provided by using rank correlation, and then checking the deviation of each country's actual rank of poverty with its predicted rank of poverty. The method of "net excess of poverty score" provides essentially a similar measure. If the lowness of poverty coorelates positively with income, then the income rank is the predicted poverty rank, and the ''net excess of poverty score" measures the difference between these two ranks. The "net excess of poverty score" figures are presented in col'xmn (5) of Table 1. Here the highest scorers are South Korea and Sri Lanka, with Zambia coming third some distance away. It should be noted that such "net excess scores" become more and more difficult to achieve as we move to countries with higher and higher GNP per head, and indeed for the richest country in this list (viz., Yugoslavia) a positive excess score will be logically impossible. These two ways of looking at poverty removal display, in fact, two opposite biases. The net excess score, as I have just remarked, is 12 biased against the richer countries in the list. On the other hand, to look just at lowness of the level of poverty would tend to bias things in favour of the relatively richer countries since income correlates positively and strongly with lowness of poverty. It may be argued that the two measures provide two different perspectives on poverty removaLThe poverty measure itself reflects over-all achievement, and this is naturally rather better for richer countries. The net excess score measures achievement over the handicap. of low income, and this gives the scope to poorer countries to do comparatively well having some income handicap to get over. At the other end of the scale, the worst excess scores (deficit scores really) in this list of countries come from Peru, Mexico and Brazil, all from Latin America. But this is only in terms of comparison with income position. If we look instead at worst over-all positions as such, then the countries that get isolated as being left with the worst poverty problems are Ethiopia, Burma and Bangladesh, respectively. Turning now to expectation of life at birth, the best over-all achievements are those of Taiwan, Yugoslavia and Sri Lanka, while the highest net excess scores over GNP ranking can be found in Sri Lanka, Burma and Uganda respectively. The statement about contrasting biases in the two perspectives made in the context of poverty measures applies correspondingly in this case as well, and both lists are worth considering. The worst net excess scores on longevity are those of Iran, Ivory Coast and Zambia, while the worst over-all 13 Table 1 Income, Poverty and Life Expectation Country GNP per Percentage Reverse Net excess Expectat- Life ex Net excess head 1975 of populat poverty of poverty ion of expectat- of life ($ with ion in rank score life at ion rank expectat- Kravis poverty in birth 1977 ion score adjustment) 1975 (years) (1) (2) (3) (4) (5) (6) (7) (8) 1. Bangladesh 200 64 3 2 47 6 5 2. Ethiopia 213 68 1 -1 39 1 -1 3. Bunna 237 65 2 -1 52 *14% 11% 4. Indonesia 280 59 4 -½ 48 8½ 4 5. Uganda 280 55 5½ 1 53 16% 12 6. Zaire 281 53 8 1k 46 4 -2½ 7. Sudan 281 54 7 % 46 -2% 8. Tanzania 297 51 9 1 51 12 4 9. Pakistan 299 43 11 2 51 12 3 .10. India 300 46 10 0 51 12 2 11. Kenya 413 55 5½ -5½ 53 16½ 5½ 12. Nigeria 433 35 12½ % 48 8½ -3½ 13. Philippines 469 33 14 1 60 23 10 14. Sri Lanka 471 14 24 10 69 32% 18½ 15. Senegal 550 35 12½2 -2½ 42 2 -13 16. Egypt 561 20 19 3 54 18 2 17. Thailand 584 32 15 -2 61 24½ 7½ 18. Ghana 628 25 17½ -k 48 8½ -9½ 19. Morocco 643 26 16 -3 55 19 0 ;20. Ivory Coast 695 25 17½ -2k 46 4 -16 21. S- Korea 797 8 32 11 63 28 7 22. Chile 798 11 28 5½ 67 30% 8 23. Zambia 798 10 30 6½ 48 8½ -14 24. Coloitbia 851 19 20 -4 62 26½ 2½ 25. Turkey 914 14 24 -1 61 24k -½ 26. TurLisia 992 10 30 +4 57 21½ -4½2 27. Mblaysia 1006 12 27 0 67 30½ 3½ 28. Taiwan 1075 5 33½ 5½ 72 34 6 29. Guatemala 1128 10 30 1 57 -21½ -7% 30. Brazil 1136 15 22 -8 62 26½ -33 31. Peru 1183 18 21 -10 56 20 -11 32. Iran 1257 13 26 -6 52 -14% -17½ 33. Mexico 1429 14 24 -9 65 29 -4 34. Yugoslavia 1701 5 33k -% 69 32½ -1½ Sources: Ahluwalia, Carter, Chenery and Develolnmefnt Policy Staff (1979) and World Developmfent Report 1979. 14 performance can be seen in Ethiopia and Senegal, followed by the cluster of Ivory Coast, Zaire and Sudan. The good and bad performances as isolated by these procedures are summarised in Table 2. Three different types of countries seem to stand out in terms of achievements. First, there is the group of export-oriented, early capitalist countries in the shape of Taiwan and South Korea. Second, there is the socialist country Yugoslavia. Third, there is the mixed economy in the form of Sri Lanka, well-known for its policies of social security. All these countries have appeared in the top list at least twice, and indeed they are the only ones to do so. It is worth noting that the other countries that have appeared in the top list, viz., Zambia, Burma and Uganda, not only appear there only once, but are also compromised in terms of achievement assessment in other ways. Zambia and Burma also ma!ke the bottom list once, and there is some evidence that Uganda's high net excess score of longevity over GNP rank has been much helped by an unusually large fall in GNP per capita during 1970-75 (see Morawetz (1977, p.78)). Turning back now to the cases of clear success, it is interesting that the list includes various political systems. In pursuing the clue revealed by this preliminary analysis, it would be obviously useful to see how the performance of these countries stand up when other criteria are chosen, and also to see whether other countries of similar political systems show any such noticeable distinction. In particular, Table 1 is limited by having only one socialist economy, viz., 15 Table 2 Best and Worst Scorers in Poverty League Best Worst Poverty measure Taiwan } Ethiopia (headcount) Yugoslavia } Burma Korea Bangladesh Expectation of life Taiwan Ethiopia at birth Yugoslavia } Senegal Sri Lanka } Ivory Coast } Sudan Zaire Net excess score of Korea Peru reverse poverty rank Sri Lanka Mexico over GNP rank Zambia Brazil Net excess score of Sri Lanka Iran life expectation Burma Ivory Coast rank over GNP rank Uganda Zambia Source: see text 16 Yugoslavia, and it will be useful to look at other socialist countries in terms of those indicators for which data exist for these countries. Also, since the similarity in the economic performance of Taiwan and South Korea with that of Hong Kong and Singapore have been frequently noted, we shall have to watch out for these other Asian early capitalist economies relying on export-expansion.7 Finally, as far as Sri Lanka is concerned, it is rather unique because of its ambitious social security system (despite its low GNP level), but some times comparison has been made with Tanzania, another mixed economy committed to government-led social policy. Tanzania's performance in terms of the indicators in Table 1 seems better than average, but not remarkably so. When other indicators are considered, Tanzania's record would have to be carefully reviewed. Turning away now from good to bad performance, it would appear that the distinction of having low net excess scores goes to some Latin American economies, viz., Peru, Mexico and Brazil. This may or may not be very significant, since these are very rich countries comparatively speaking, and have a tendency towards being pulled down in the calculation of net excess scores. What is of some significance is the fact that they do badly in terms of net excess score not in the context of longevity, but in that of poverty. Since poverty is measured here - as outlined-earlier - in terms of income deprivation (vis-a-vis the Indian poverty line), -it is of some considerable interest that Peru, Mexico and Brazil 7 See Chenery and Keesing (1978). and Little (1979b). 17 do so Lac2iv daspite having a high average level of income. The finger points clearly at the distribution of income prevailing in these countries. The worst poverty levels are seen in Ethiopia, Burma and Bangladesh. Of these, Burma has the redeeming feature that it also makes one top list, viz., that of net excess score of longevity. In addition, it is worth recalling that Ethiopia and Bangladesh - but not Burma - are countries in which severe famines have taken place quite recentlyt viz., in 1973 in Ethiopia, and in 1974 both in Ethiopia and in Bangladesh. In the context of a malnutrition-oriented view of poverty, famines clearly reflect cases of acute poverty, and their continued prevalence in these countries within the last decade call for special attantion. They also reflect the vulnerability aspect of poverty rather starkly. 3. Changes in Longevity and Literacy The analysis of the preceding section was based on current performance of different countries. It is interesting to look also at changes in levels of performance over the last couple of decades. Unfortunately, no poverty estimates exist that can be used for international comparison of intertemporal changes of poverty. We have, therefore, no option but to look,_-, at other indicators. In particular, the indicators that figure most often in "physical quality of life" indices may be chosen for this purpose, and they are longevity and literacy.8 8 Morris (1979) chooses infant mortality up to age 1 as an additional parameter, and correspondingly measures longevity in terms of expectation of life at age 1. Expectation of life at birth combines the two. 18 Table 3 presents data for 100 countries with GNP per capita less than $3,000 in 1977 as given by the World Development Report 1979. Indeed these are all the countries for which data are given and which have income below the specified limit.9 Estimates of expectation of life at birth and of adult literacy rates are given for 1960 and 1977, taken from the World Development Report 1979.10 There is an interesting issue as to how best to calculate the magnitude of change. Should it be the size of absolute change in either of these magnitudes, or relati've change in the form of percentage increase. The answer, on reflection, is: neither! It can be argued that as, say, longevity becomes high, it becomes more of an achievement to raise it further. Raising the expectation of life from 25 to 35 may be easily achieved by cutting out epidemics, etc., but to raise it from 65 to 75 is quite another matter. And to raise it from 70 to 80 has not yet been done by any country whatsoever. Thus taking absolute magnitudes of change under- estimates the performance of countries with already high longevity. And to lookc at percentage chanjges of the expectat- ion of life introduces a change precisely in the wrong directiox e.g., 65 to 75 is a smaller percentage change than 25 to 35. So neither of these methods will do. 9 The World Development Report classifies countries not merely on the basis of income, but also on other character- istics. What has been done in Table 3 is to choose all countries with GNP per capita less than $3,000 whether they are placed by the World Development Report under "low income countries", "middle income countries", "industrializec countries", or "centrally planned economies". 19 A better procedure is to look at the absolute short- fall of actual longevity from some chosen upper bound, e.g., . 80, and then to examine the percentage decline of this short- fall. This is the method that has been used in Table 3 to calculate change in both longevity as well as literacy. For longevity the upper bound is taken to be, the unachieved (and arguably unachievable) 80 years, while for literacy, a figure of 100 per cent provides a natural upper bound.11 In the last column of Table 3 distinguished perfSrmances have been noted. For each of the two success variables (viz., percentage decline in longevity shortfall, and percentage decline in literacy shortfall), the countries with four best performances have been put in category A, and countries in the top quarter (25 per cent) of each list have been awarded B (if not already a recepient of A). The others all get C, but this has been written up only in those cases in which a country has scored A or B on one variable, but C in the other. (That is, all blanks in the last column either reflect C or unknown data.) While longevity data have been 10 In some cases the figures refer not exactly to 1960 or 1977, but to a neighbouring year. This applies to the set of literacy figures only, which also has some gaps, viz., countries for which data could not be found for 1966 (or its neighbouring years) or for 1977 (or its neighbouring years). 11 For longevity there is no natural maximum. The current best performance is 76 years. By choosing 80 as the upper bound, we have added four years to that, which is certainly most unlikely to be achieved in the absence of some remarkable medical breakthrough. One result of choosing 80 years as the upper bound is to make the maximum percentage rise from an increase in one year in longevity to be no more than 25 per cent. 20 found for all the countries, the literacy data refer to 61 countries (out of the 100) for which the World Development Report 1979 provides information for both 1960 and 1977. Thus while 25 countries have got A or B on the first score (longevity), only 16 countries have been awarded these marks in the second (literacy), with the last two tying for the fifteenth position. For those countries which score A or B under longevity but for which literacy data have not Meen presented, a question mark ? has been put as the second score. What kind of an over-all picture emerge from this? The best performances are the following: Longevity Literacy 1. Vietnam 1. South Korea 2. Cuba 2. Hong Kong 3. Taiwan 3. Tanzania 4. Hong Kong 4. Taiwan. 38 countries have shown distinction (A or B) in one or both of the fields. It should be noted that since literacy data were not available for 39 countries, this is biased against them, since they had the chance of showing distinction only in one field of the two. 21 Table 3 Changes in Longevity and Literacy Country GNP per head Expectation of life Adult literacy rate 1977 at birth (years) (percentage) (US $) 1960 1977 Percentage 1960 1977 Percentage decline in decline in longevity literacy shortfall shortfall (1) (2) (3) (4) (5) (6) (7) (8) 1. Bhutan 80 36 41 11 2. Cambodia 42 48 16 36 3. Bangladesh .90 42 47 13 22 22 0 4. Laos 90 40 42 5 28 5. Ethiopia 110 36 39 7 10 6. Mali 110 37 42 12 3 10 8 7. Nepal 110 37 45 19 9 19 11 8. Somalia 110 36 43 16 2 50 49 CB 9. Burundi 130 37 45 19 14 10 -5 10. Clad 130 35 43 18 15 11. Rwanda 130 37 46 21 16 23 8 12. Upper Volta 130 37 42 12 2 5 3 13. Zaire 130 40 46 15 31 14. Burma 140 44 52 22 6Q 67 18 15. Malawi 140 37 46 21 25 16. India 150 43 51 22 28 36 11 17. Mozambique 150 37 46 21 8 continued ... 22 Table 3 (continued) Country GNP per head Expectation of life Adult literacy rate 1977 at birth (years) (percentage) (US $) 1960 1977 Percentage 1960 1977 Percentage decline in decline in longevity literacy shortfall shortfall (1) (2) (3) (4) (5) (6) (7) (87 18. Niger 160 37 42 12 1 8 7 19. Vietnam 160 41 62 54 .. 87 .. A? 20. Afghanistan 190 34 42 17 8 12 . 4 21. Pakistan 190 44 51 19 15 21 7 22. Sierra Leone 190 37 46 21 15 23. Tanzania 190 42 51 24 10 66 62 CA 24. Benin 200 37 46 21 8 11 3 25. Sri Lanka 200 62 69 39 75 B? 26. Guinea 220 35 44 20 7 27. Haiti 230 42 51 24 15 23 9 28. Lesotho 240 42 50 21 .. 40 29. vadagascar 240 37 46 21 .. 50 30. 250 37 46 21 7 Enpire 31. Kenya 270 47 53 18 20 40 25 32. Mauritania 270 37 42 12 5 17 13 33. Uganda 270 44 53 25 35 34. Sudan 290 39 4b 17 13 20 8 35. Angola 300 33 41 17 5 continued... 23 Table 3 (continued) Country GNP per head Expectation of life Adult literacy rate 1977 at birth (years) (percentage) (US $) 1960 1977 Percentage 1960 1977 Percentage decline in decline in longevity literacy shortfall shortfall (1) (2) (3) (4) (5) (6) (7) (8) 36. Indonesia 300 41 48 18 39 62 38 CB 37. Togo 300 37 46 21 10 16 7 38. Egypt 320 46 54 24 26 44 24 39. Cameroon 340 37 46 21 19 40. Yemen (PDR) 340 36 47 25 .. 27 41. Chana 380 40 48 20 27 30 4 42. China 390 53 64 41 .. .. .. B 43. Hlonduras 410 46 57 32 45 57 22 44. Liberia 420 40 48 20 9 45. Nigeria 420 39 48 22 15 46. Thailand 420 51 61 34 68 82 44 CB 47. Senegal 430 37 42 12 6 10 4 48. Yanen Arab 430 36 47 25 3 13 10 49. Philippines 450 51 60 31 72 87 54 CB 50. Zambia 450 40 48 20 .. 39 51. Congo 490 37 46 21 16 50 40 CE 52. Papua 490 41 48 18 29 32 4 New Guinea 4 continued ... 24 Table 3 (continued) Country GNP per head EXpectation of life Adult literacy rate 1977 at birth (years) (percentage) (US $) 1960 1977 Percentage 1960 1977 Percentage decline in decline in longevity literacy shortfall shortfall (1) (2) (3) (4) (5) (6) (7) (8) 53. Rhodesia 500 45 52 20 39 54. El Salvador 550 50 63 43 49 62 25 BC 55. 1%brocco 550 47 55 24 14. 28 t 16 56. Albania 630 62 70 44 .. .. .. B? 57. Bolivia 630 43 52 24 39 63 39 CB 58. N. Korea 670 54 63 35 .. .. B? 59. Ivory Coast 690 37 46 21 5 20 16 60. Jordan 710 47 56 27 32 59 40 CB 61. Colombia 720 53 62 33 63 81 49 CB 62. Paraguay 730 56 63 29 75 80 20 63. Ecuador 790 51 60 31 68 74 19 64. Guetemala 790 47 57 30 32 46 21 65. S. Korea 820 54 63 35 71 91 69 BA 66. Mbngolia 830 52 63 39 .. .. .. B? 67. Nicaragua 830 47 55 24 .. 57 68. 'Dcuinican 840 51 60 31 65 67 6 Rep. 69. Peru 840 48 56 25 61 72 28 70. Tunisia 860 48 57 28 16 55 46 CB cont.inuied ... 25 Table 3 (continued) Country-, SW per head Expectation of life Adult literacy rate 1977 at birth (years) (percentage) (US $) 1960 1977 Percentage 1960 1977 ercentage decline in decline in longevity literacy shortfall hortfall (1) (2) (3) (4) (5) (6) (7) (8) 71. Cuba 910 64 72 50 .. 96 .. A? 72. Syria 910 48 57 28 30 53 33 73. Malaysia 930 57 67 43 53 60 15 BC 74. Algeria 1, u0 47 56 27 10 35 28 75. Turkey 1,110 51 61 34 38 60 35 76. Mexico 1,120 58 65 32 65 76 26 77. Jamaica 1,150 64 70 38 82 86 22 BC 78. Lebanon . 58 65 32 79. Chile 1,160 57 67 43 84 88 25 BC 80. Taiwan 1,170 64 72 50 54 82 61 AA 81. Panama 1,220 62 70 44 73 78 14 BC 82. Costa Rica 1,240 62 70 44 .. 88 .. B? 83. South Africa 1,340 53 60 26 57 84. Brazil 1,360 57 62 22 61 76 38 03 85. Uruguay 1,430 , 68 71 25 ,. .. 94 86. Iraq 1,550 46 55 26 18 87. Rcmania 1,580 64 70 38 .. 98 .. B? continued .... 26 Table 3 (continued) Country GNP per head Expectation of life Adult literacy rate 1977 at birth (years) (percentage) (US $) 1960 1977 percentage 1960 1977 Percentage ecline in decline in longevity literacy shortfall shortfall (1) (2) (3) (4) (5) (6) (7) (8) 88. Argentina 1,730 65 71 40 91 93 22 DC 89. Portugal 1,890 62 69 39 62 70 21 BC 90. Yugoslavia 1,960 62 69 39 77 85 35 DC 91. Iran 2,160 46 52 18 16 50 40 CB , &92. Trinidad 2,380 63 70 41 93 95 2 BC 93. Bulgaria 2,580 67 72 38 .. .. B? 94. Hungary 2,580 67 70 23 97 98 33 95. Hong Kong 2,590 65 72 47 70 90 67 AA 96. Venezuela 2,660 59 66 33 63 82 51 CB 97. Greece 2,810 68 73 42 81 .. .. B? 98. Israel 2,850 69 72 27 84 88 25 99. Ireland 2,880 69 73 36 .. 98 .. B? 100. Singapore 2,880 64 70 38 _ 75 .. B? Source: See text 27 Table 4 Performance Summary in Life Expectation and Literacy 2 A 1 A, 1 ? 1 A, 1 B 1 A, 1 C 2 B 1 B, 1 ? 1 B, 1 C Hong Kong Cuba S. Korea Tanzania Albania Argentina Taiwan Vietnam Bulgaria Bolivia China Brazil Costa Rica Chile Greece Colcnibia Ireland Congo Mongolia El Salvador N. Korea Indonesia Romania Iran Singapore Jamaica Sri Lanka Jordan Malaysia Panama Philippines Portugal Somalia Thailand Trinidad Tunisia Yugoslavia Venezuela What can we say about this list? One thought that is bound to occur is that communism is good for poverty removal. There are 10 communist countries in the total list of 100,. and of them 9 show some dis,tinction (i.e.,.'. .: qualify in the list of 38 in Table 4). In fact, 8 of the 9 (all except Yugoslavia) do this despite not having literacy figures reported in Table 3, i.e., they do it on the basis of one entry alone. That entry, of course, is 28 longevity, which is arguably a more basic indicator of success than literacy.12 As against this, it should also be noted that many of the communist countries are, in fact, richer than the mean or median developing country, and while the indices are relative ones, it may be argued that richer countries demon- strably do better, on the whole, in the comparisons made in Table 3. On the other hand, it is easily checked that even among comparable countries in terms of GNP per head, the longevity performance of the communist countries is typically superior. This applies to the poorer group also. For example, if we look at countries with GNP per head less than $1,000, the following is the breakdown of communist and non-communist countries in terms of longevity performance. Table 5 Comparative Performance in Longevity for Poor Countries All oountries Commuist countries Non-coIxunist Total number 73 6 67 Number of countries 2 2 0 with A Number of countries 10 6 4 with A or B Source: Table 3. 12 The educational performance of communist countries, especially in generating literacy, is also typically good, as is clear from other sources of evidence. 29 Clearly the relative performance of communist countries is superior in terms of this particular indicator. The other indicator, viz., literacy performance, cannot be used as a basis for comparison, since the data for this variable are not presented for most of the communist countries in the World Development Report 1979, which has been used as our basis for comparison. A second point to note is that some of the high-growth early capitalist countries (e.g., Taiwan, S. Korea, Hong Kong and Singapore) also have very good performance in terms of the chosen indicators. Indeed, Taiwan and Hong Kong have the best over-all performance record in terms of the two criteria for those 61 countries for which both sets of data are available. The countries that seem to do relatively worse are those in the middle, neither communist, nor successfully capitalist. Even here there are clear exceptions, and we have already pointed towards Sri Lanka and Tanzania as interesting cases of relative success in terms of poverty removaLl and longevity (Sri Lanka) and expansion of literacy (Tanzania). One does not., however, learn very much about causation at this remote level of inter-country comparisons, and it will be useful to go into the details for a number of chosen cases. 13 Vietnam has been put in the communist list, but even if it is transferred to the non-communist group on grounds of the nature of the south Vietnamese polity preceding unification, the over-all picture will not change much. Note also that the performance of the communist countries is superior not only in terms of the percentage decline in the longevity shortfall used as the basis for comparison in this study, but also in terms of the absolute level of longevity achieved (see Table 3). 30 One group that requires attention is clearly the successful early capitalist countries, in particular the Asian foursome: S. Korea, Taiwan, Hong Kong and Singapore. Among them the former two are more interesting to study, both because they are poorer (Singpore, it better be remembered, is richer than Ireland), and also because they are larger and not city states. South Korea in particular requires special attention since its growth rate has gone up recently quite sharply along with good performance in non-growth indicators, and it has been seen as something of a model for other develop- ing countries. The second group that deserves attention are Sri Lanka and Tanzania in terms of their superior performance, especially remarkable since they are among the poorer countries even in the list of developing economies. Insofar as removal of poverty and expansion of longevity are taken to be more important indicators of success than expansion of literacy, Sri Lanka's case deserves special attention. In the next two sections (4 and 5) these two groups are considered in turn. After that we return to more general analysis again on a wider canvas, and also go into the tricky methodological issue as to what "level" of explanation to use in drawing lessons from cases of success in the comparative 14 picture. Then we study some cases of "failure" 14 The communist countries deserve special study too in view of their superior performance, but there is already quite a literature on this. 31 (in particular, Bangladesh and Ethiopia) focusing particularly on the extreme form-.of poverty that takes the shape of starv- ation and famines and draw some lessons from that. The main conclusions are gathered together in the final section. 4. Growth, Private Enterprise and Public Policy: Case Studies The excellent performance of Taiwan, Hong Kong, South Korea and Singapore was noted earlier, and as Ian Little (1979b) puts it, "there is a good reason to ask what can be learned from ... the four" (p.1). The possibility of emulation has been explicitly raised,15 and Little himself has stated the central issue clearly: "Except for Hong Kong, very rapid growth began only in the sixties, in each case after a marked change of policy from import substitution to export promotion. It will be the thesis of the paper that the success is almost entirely due to good policies and the ability of the people - scarcely at all to favourable circum- stances or a good start." 16 It is worth clarifying that while Hong Kong is treated as something of an exception in this statement, this is only because it had a policy of export promotion anyway even before the sixties, and it is also to be noted that Hong Kong's over-all growth performance,has been very gooed for,,-a long time. The restrictive trade pattern of preceding years appears on 15 See particularly Chenery and Keesing (1978), Rao (1978), Westphal (1978), Little (1979b), and Krueger (1979). 16 Little (1979b, p.4). 32 this view to have been abandoned with great advantage in those countries which had such restrictions. What makes the experience of high growth in these countries particularly worthy of attention is the so-called "quality" of that growth, viz., that it seems to have taken place without accentuation of inequality (as is often, sadly, the case), and there is evidence of fast removal of poverty in these countries judged from each of the perspectives that have been used in the analysis above. Questions have been raised about the nature of the data used, especially for South Korea, and it has been particularly disputed that in the seventies - the period of rapid growth in Korea - there 17 has been no accentuation of inequality. What is not, however, disputed is that these countries have acheived very high growth rates with at least no sharp deterioration of the extent of inequality, so that the poorer sections of the population have shared substantially in the benefits of economic growth. What is also not in dispute is the fact that the growth in these countries has been accompanied by a remarkable rate of expansion of exports in general, and of manufacturing exports in particular. For example, during 1965-75, the Korean exports increased at an annual rate of 31 per cent per year at constant 1970 prices. "The major lesson," Little 17 See Choo (1975, 1978), Choo and Kim (1978), and Bhalla (1979). Bhalla's estimation suggests a decline in the share of bottom 20 per cent in national income from 6.5 per cent in 1970 to 5.2 per cent in 1976, along with a decline of the share of the bottom 40 per cent from 18.0 per cent to 15.4 per cent. 33 (1979b) concludes, "is the labour-intensive export-oriented policies, which amounted to almost free trade conditions for exporters, were the prime cause of an extremely rapid and labour-intensive industrialization which revolutionized in a decade the lives of more than fifty million people, including the poorest among them" (p.34). There is indeed much in the experience of "the four" to cheer Adam Smith, and the invisible hand would seem to have done a good deal of visible good. But is this really the "major lesson" to draw from the experiences of the four? I would now like to argue that perhaps it is not. First a preliminary point about the four. Two of them are very tiny: the combined population of Hong Kong and Singapore is exceeded by the population of many cities, e.g., Shanghai, or Tokyo, or even greater Calcutta, and they also occupy rather little land mass. Hong Kong and Singapore are also very rich, compared with the typical developing countries. On both these grounds it would make sense to pay more attention to tha experiences of South Korea and Taiwan. Especially South Korea, since it is a fairly large country (in Lerms of population size it is comparable to Egypt, more than twice the size of Taiwan), and also the poorest of the four (and had been,in fact, poorer than Egypt in the 1950s). How invisible were the hands that the reared South Korean economic expansion? Not much, it would appear. First, while the massive export promotion was going on, the Korean government was carrying out quite a strong programme of import substitution in selected areas. As Mrinal Datta-Chaudhuri 34 (1979) notes: "In 1960, chemical fertilizers, synthetic fibre yarns and thread, petroleum products, paper, cement, iron and steel products and plastic materials were some of the principal items in Korea's import trade", while "by 1971, import substitution eliminated all these items except synthetic fibre yarns and plastic materials from the list of major -importables" (p.23). In 1970, out of the 1312 basic items, the import of 73 items was banned, and that of 524 restricted.18 Even in 1976, 60 items were on the prohibited list and another 616 items on the restricted list. While the numbers of prohibited and restricted items provide very crude quantificat- ion of import restriction, the over-all picture to emerge is certainly very far from one of free trade. There seems to be also some evidence for the view that fostered import substitution in Korea prepared the ground for eventual export 19 expansion. Indeed, even in export expansion itself, the invisible hand was buttressed by various incentives, including "merit medals". Within the country the Korean government had overwhelming control of the organised banking sector, owning not only the central bank, but also the five nation-wide commercial banks, in addition to the specialized banks and financial institutions (e.g., Korea Development Bank, the Medium Industries Bank, the Korea Exchange Bank, etc.). It carried out a policy of 18 See Frank, Kim and Westphal (1975). 19 See Datta-Chaudhuri (1979, pp.24-5). See also Erank, Kim and Westphal (1975), Hong and Krueger (1975), Kim (1977a, 1977b), and Krueger (1979). 35 spectacular encouragement to private savings through high return on bank.deposits. This was combined with an already high tax base being expanded, and public enterprises generat- ing further resources at the disposal of the government. The speeding of the Korean economic development during 1964 to 1970 was accompanied by an increase in government saving of 20.6 per cent per year, and by 1970 government savings accounted for 43 per cent of total domestic savings. Adding government savings to deposits in nationalized banks, it appears that the government controlled two-thirds of invest- ment resources in the economy. As Datta-Chaudhuri (1979) puts it, "no state, outside the socialist bloc, ever came anywhere near this measure of control over the economy's investable resources" (p.16). The government has not been a passive channel for these investment funds. On the contrary, it used its power to control the direction of investment through various means, including varying the interest rate charged with the field of invrestment (e.g., 8 to 33 per cent depending on the field). The public sector itself absorbed nearly a third of the total investment during 1963-72, growing at 14.5 per cent per annum, compared with the economy's over-all growth rate of 9.5 per cent. In addition to including the traditional fields of railways, communication and electricity, the public sector in South Korea covers many industrial units, producing 15 per cent of the manufacturing output and owning a larger share of industrial capital (since the public sector industries are typically more capital-intensive). 36 It is not clear how cheered Adam Smith would have been with the South Korean experience. It is, in fact, a classic case of the government playing an active part in industrialization and growth. The fact that export expansion provided a very good vehicle for such expansion must not be denied, but it has to be viewed in the perspective of the total economic picture in South Korea. Indeed, in many ways the Korean experience has been fully in line with the traditional wisdom of development economics, exemplified by the writings of Nurkse (1953) and Lewis (1955), and even Fei and Ranis(1964). A labour-abundant economy embarked on a fast economic growth with a corresponding increase in employment and a remarkable acceleration of capital accumulat- ion. ° The ratio of gross domestic investment to GDP rose from 11 per cent in 1961 to 26 per cent in 1976, and during the sixties the South Korean rate of growth of domestic investment (23 per cent per year) was higher than that of any other country for which the World Development Report 1979 provides data (see Table 4, which gives these data for 95 countries). The governrment played a major part in the generation and use of these investment funds. What made the fruits of Korean growth spread so widely was clearly the fast expansion of employment and eventually also of real wages, which went with this growth.21 There can 20 The capital accumulation was certainly helped initially by external capital inflow, and continued to be important (see (Krueger (1979)). Also the low value of capital output ratio,related to the investment pattern and use of labour and skill,made the accumulation of capital unusually effective. 21 See Hong and Krueger (197T), especially Cole and Wiestphal (1975), Adelman and Robinson (1978), Westphal (1978), Little (1979b), and Krueger (1979). 37 be little doubt that the labour-intensive nature of the growth was central to Korean success in poverty reduction. In understanding the Korean experience, the role of employment expansion and growing demand for labour belongs to the most immediate level of explanation. Behind that immediate explanation stand several other factors causally linked to it in the Korean case, and these include rapid capital accumulation, reliance on export expansion, availability of a labour force suited by education to modern manufactures, etc. The merits of export-led growth proved to be substantial for the Korean economy in the particular circumstances in which it found itself, and as Adelman and Robinson (1978) demonstrate, "the strategy of labour-intensive export led growth has produced significantly higher incomes for the poorer half of the population and a far more equal relative distribution compared with a strategy of import substitution" (p.127). But this belongs to a remoter level of explanation. Even if a country is shown to be not in a position to emulate the Korea:i export-expansion success, the imTiediate lesson of the importance of employment expansion would still stand, and any set of policies - whether export-expansive or import- substituting - which can be shown to have these employmaent characteristics would be supported by the Korean experience. This methodological question will-be further taken up in section 6, but a few more points should be noted on the specifics of the Korean case. The growth of employment related to exports has been extraordinarily rapid in South Korea. During 1960-70 the 38 volume of direct employment in export production increased at the astonishing rate of 34 percent per year, and total export-induced employment - both direct and indirect taken together - also at the very high rate of 30 percent per year.22 The value of capital-output ratio fell sharply, and even the capital-labour ratio showed substantial decline, in the manufacturing sector of the economy, and these declines were helped by the export-orientation of production in this sector. 3 But despite that, the size of export-related employment has remained a small fraction of the total employ- ment in the Korean economy, and a good deal less than half even of manufacturing employment. The point of mentioning this is not to belittle the remarkable achievements of the Korean export-expansion programme, but to note that there is a good deal more in the Korean experience than this one feature which seems to have received more attention than any other. The fuller picture is that the South Korean economy was experiencing a remarkably fast economic growth with a massive capital accumulation, which was being invested in very labour- intensive ways, and the most dynamic part of the growing economy was undoubtedly the export sector. Turning now to Taiwan, its experience also demonstrates 2Q See Cole and Westphal (1975). 23 See, among others, Cole and Westphal (1978), Adelman and Robinson (1978), and Westphal (1978). 24 In 1970 direct employment in production for exports seems to have been about 18.9 percent of manufacturing employment and 5.1 percent of total employment. Adding indirect employment due to exports would increase these figures by a little more than a half. See Little (1979b, p.31) and Hsia (1979, pp.30-1). The data used come from Cole and Westphal (1975) and Westphal (1978). 39 the great importance of employment expansion in poverty removal.. The proportion of visible unemployment fell from 6.3 per cent of the labour force in 1953 to 3.7 per cent in 1975.25 But perhaps more importantly, hidden unemployment and underemployment showed sharp decline in terms of the change of their usual habitat (see particularly Galenson 1979, pp.404-7)). For example, unpaid family workers fell from being 28.5 per cent of total employment in 1964 to 15.9 per cent in 1975. 6 It is the expansion of employment at a reasonable wage that has permitted the poorer Taiwanese "to feed, clothe, and house his family and send his children to school" and to enjoy a steady increase in "the availability of consumer goods". 27 At a remoter level of explanation, the successful expansion of exports has certainly been a major factor behind this employment-led prosperity. There is evidence that the export industries provided greater scope for labour utilisation than the manufacturing sector in general, and the existing import-replacing industries in particular. 8 There has also been massive capital accumulation, with a slightly higher percentage invested than even in South Korea (though the proportionate rate of growth has not been so fast, since it was substantially higher than in Korea in 1960). In both these 25 See Kuo (1976). See also Hsia and Chao (1978). 26 See Galenson (1979, p.406). 27 Galenson (1979, p.447). Also Hsia and Chao (1978). 28 See Little (1979a) 40 respects the Taiwanese experience is rather similar to the Korean. And as far as the role of government is concerned, there is clear evidence of the importance of government's policy measures in determining the character of economic development in Taiwan, especially the promotion of exports.29 Even though the government did not have the kind of overwhelming command that was enjoyed in South Korea, it did have a good deal of power and used that to promote actively its chosen pattern of economic expansion. 5. Distribution, Social Services and Public Policy: Case Studies As was noted earlier, in poverty removal and in expand- ing longevity Sri Lanka's record has been distinguished. While the absence of later literacy data prevented comparison with other countries on that score, its level of literacy is indeed high. By 1973 the proportion of adult literacy was already 78 per cent in Sr. Lanka, more than twice the figure for neighbouring India, and more than three and a half times the figures for Pakistan and Bangladesh. It can be checked that in terms of relation with per capita GNP, Sri Lanka's actual values of (i) life expectancy, (ii) adult literacy, (iii) infant mortality, and (iv) fertility, all lie comfortably away from their expected values (in the right direction, i.e., higher for the first two, lower for the last two).30 For a poor country, with incomes comparable 29 See Scott (1979, pp.336-345). 30 See Isenman (1980). 41 (only a little higher) than India or Pakistan, Sri Lanka's record in removing poverty and providing a higher quality of life is quite remarkable. Regarding the causation of this superior performance, attention has tended to concentrate on Sri Lanka's programme of food subsidy and other social services. There is clear evidence that this programme reduces the concentration of income distribution quite significantly.31 Food subsidies also have the effect of improving nutrition of the poorer section of the population. This is partly because it is an income supplement, but also because offering income supplement in the form of food does typically have the effect of increasing food consumption rather more than the same amount offered 32 through direct income grant. There is also some evidence that the cut-back in the food subsidy programme in 1973 was followed by an increase in the death rate in Sri Lanka in 1974.33 While the important role of the food distribution programme in Sri Lanka is undeniable, it seems appropriate to view it as one part of the actively redistributive orientation of government policy in Sri Lanka. The health programme is also efficient and ehergetic. While there remain inequalities 31 See Jayawardena (1974), and Alailima (1978). 32 There is much evidence for the greater impact of income ,subsidy when given in the form of food, on the actual consumption of food, and this effect has been particularly studied in the case of Kerala state in India; see United Nations (1975), Kumar (1979) and George (1979). One would expect such an effect on theoretical grounds as well, since the subsidy on food changes relative prices adding a sub- stitution effect to the income effect, and also because transaction costs would push the consumer further in that direction. 33 See Isenman (1980). 42 in the coverage in terms of region and class,34 there is little doubt that Sri Lanka achieves a widespread health coverage of a kind that is quite rare among developing countries.35 It is also interesting to observe that the proportion of GDP devoted to health expenditure in Sri Lanka is not exceptionally high, and indeed many developing countries with much worse performance in health spend a higher proportion of GDP on this item.36 Given the relatively low per capita income level of Sri Lanka, the contrast is all the more impressive. The important difference seems to be brought about by the orientation of government policy in Sri Lanka, with its emphasis on securing a comparatively wide coverage, and being more active in trying to reach the poorer sections of the community. The food distribution programme is best seen as a part of this general approach. The same applies to Sri Lanka's education programmes. The educational facilities of Sri Lanka have been widely distributed, and are responsible for its rather superior performance. While Sri Lanka's poverty level measured in terms of income is also rather lower than that of countries with comparable income per head, its performance is very substantially more distinguished when judged in terms of results such as longevity, literacy, etc. (see Tables 1 and 3 above). Behind these achievements lie active government policies providing 34 See Viasaria (1979) and Richards and Gooneratne (1980). 35 See Alailima (1978), Isenman (1980), Visaria (1979), and Richards and Gooneratne (1980). 36 See Richards and Gooneratne (1980). 43 concrete services with an orientation towards wide distribution. Inequalities in income distribution are, thus, significantly 37 reduced by direct public provisions. The question has been frequently asked whether a poor country like Sri Lanka can afford to have such a high volume of social welfare expenditure, and it has also been argued that its growth rate may have been negatively influenced by the allocation of resources to these programmes. These "counter-factual" hypotheses are rather difficult to assess. There is, of course, the simple fact that the annual rate of growth of GNP per head in Sri Lanka for the period 1960-77 has been only 2 per cent, and in the seventies in particular it has been much less than that. But the question is whether the growth rate would have been much higher in the absence of these programmes. It can, of course, be argued that the rate of investment in the economy very likely would have been higher if these programmes were abandoned, since they absorb more than 10 per cont of GNP.38 It is also true that Sri Lanka's rate of domestic investment is lower than that of neighbouring Pakistan and India, and also -lower than the average for developing countries and even for "low-income" developing countries.39 37 See Jayawardena (1974), Alailima (1978), and Richards and. Gooneratne (1980). 38 In 1973 these programmes absorbed about 11% of GNP (see Alailima (1978, p.46)). This figure does, however, include the entire educational budget, some of which must be seen to be something other than social welfare expenditure. 39 See Table 5 of the World Development Report 1979. 44 It must, however, be noted that Sri Lanka's ratio of gross domestic savings (as opposed to investment) is comparable to that of the average of low income developing countries, and a bit lower than in India and rather substantially higher than in Pakistan. The comparatively low value of Sri Lanka's investment ratio seems to be due to less foreign aid and investment. Furthermore, nutrition does contribute to productivity, as several studies have shown, and the impact of literacy on growth - especially agricultural growth - has been noted.40 In fact, Sri Lanka's rate of expansion of agricultural production has been good, and remarkably enough, so far as food is concerned, the rise of 13 per cent in availability per capita over the period 1969-71 to 1975-77 experienced by Sri Lanka in unequalled by any country with a GNP per head below $600 (in 1977 U.S. dollars). There are 54 countries in that income bracket, with Sri Lanka coming more than half way down, and its comparative performance in the expansion of food output must be regarded as exceptionally good. There is little doubt that the social welfare programmes of Sri Lanka place it at an advantaged position in terms of poverty removal and longevity increase given its income level. In the context of assessing growth-based vis-a-vis welfare- based strategy, it might be useful to measure that advantage in terms of years of lead for appropriate rates of growth. The methodology used is, first, to convert the life expect- ation and poverty-head-count advantages into income advantages in terms of the general cross country relation holding around 40 Cf. Sen (1970) and Chaudhri (1979). 45 the world, and then to convert the income advantages into the corresponding measures of years of lead at the appropriate growth rates. Isenman (1980) has obtained the following power curve fit for life expectancy on per capita income:41 (1) Ln(life expectancy) = 3.263 + 0.123 Ln(per capita income), using 1975 data for 59 countries. From this, it is easily calculated that the income per capita corresponding to Sri Lanka's life expectancy of 69 years is $2,684 as opposed to its actual income of $130 in 1975 U.S. dollars. Thus the per capita income of Sri Lanka would have to be raised by a factor of 20.65 for it to have its actual life expectancy as its expected life expectancy based on income. There is no corresponding calculation for percentage of population below the poverty line. But this is easily done by using Table 1. Two fits were tried out, respectively linear y = a + bx, and power curve y = p xr, and the results are the following:42 (2) head-count-poverty ratio = 60.92 - 0.04409 Kravis-adjusted-GNP; (3) In(head-count-poverty-ratio) = 10.396 - 1.137356 Ln(kravis-adjusted-GN The Kravis-adjusted-income levels at which Sri Lanka's head- count-poverty ratio of 14 per cent is reached are respectively $1064 and $916ivis-a-vis Sri Lanka's ,actual Kravis-adjusted- income level of $471. Thus, the income level of Sri Lanka would 41 See Isenman (1980) for the details of the fit. The value of r2, incidentally, is .65. 2 42 The values of r for the two fits are respectively .75 and .83, and the number of observations is 33, since Sri Lanka was not included in doing the analysis. 46 have to be raised, on this approach, by factors of 2.26 and 1.94 respectively for achieving poverty removal through income growth. I turn now to the growth assumptions. The Sri Lankan growth rate of GNP per capita over 1960-77 has been 2.0 per cent per year. In the absence of the social welfare programme what kind of a growth rate would Sri Lanka have had? This is, of course, a terribly speculative counter-factual question, but there would be very little causal economic analysis if such speculations were to be entirely shunned. Perhaps the most favourable assumption is one in which all of the 11 per cent of expenditure on social services are put instead into gross domestic investment, increasing the latter by the same amount, and enhancing the growth rate of per capita income by a rate determined by the marginal capital-output ratio. The gross domestic product in Sri Lanka increased at 4.6% and 3.1% respectively over 1961-70 and 1970-76, with 15 and 17 per cent 43 of GDP invested in 1961 and 1976 respectively. With a slight liberty, these figures yield values of marginal capital output ratio of 3.26 and 5.47 respectively, and with an 11% of GDP added to investment, the growth rates would be raised by 3.37% and 2.01% respectively. Added to the prevailing growth rate of 2 per cent per annum (overlooking the niceties of the distinction between GNP growth rate and GDP growth rate), 43 See World Development Review 1979, Tables 2 ard 5. Regarding investment ratios, one should have used not the end-year ratios but the ratios for the whole period, which were not, however, readily available. 47 we obtain enhanced values of growth rates of GNP per capita of 5.37% and 4.01% respectively. The least favourable assumption would appear to be an unchanged growth rate as social welfare programmes are eschewed.44 In Table 6 the years taken to match non-income factors behind poverty removal and longevity enhancement by income growth are presented with alternative assumptions about growth performance. Table 6 Years Needed to Match Social Welfare Achievements of Sri Lanka through the Growth of Income Growth rate Inccra-based Inocme-based Incomx-based Growth of per capita poverty removal poverty renoval longevity increa. assumption GNP per year (power fit) (linear fit) (power fit) No change 2.C0% 33 years 41 years 152 years Full transfer at 1970-76 ~ 4.01% 17 years 21 years 77 years capital-output ratio Full transfer at 1961-70 5.37 13 years 16 years 58 years capital-output ratio Source: see text. The over-all impression is one of a long haul in match- ing social welfare achievements of Sri Lanka with income growth. 44 Note, in this context, that Sri Lanka's ratio of gross domestic savings is rather similar to the average of the "low income" developing countries, viz., 11% in 1961 (compared with a "low income" average of 11%), and 20% in 1976 (compared with a "low income" average of 18%). It is, thus, by no means obvious that imitation of countries with less social welfare programmes would raise Sri Lanka''s savings rate. 48 In fact, these hauls must be rather longer if it is recognized that the assumptions - even the "unfavourable" one of no change - may prove to be too optimistic for the growth strategy. There are several distinct reasons for this. (1) The existing growth rate of 2% per year in per capita GNP is based on the actual growth rate of population in Sri Lanka, but this is lower than in other countries in the region, to a great extent due to the fertility decline in Sri Lanka, which has been, with reason, related to the 45 social welfare programme. Thus "no change" in total growth rate might mean a lower growth rate of per capita income, had the social welfare programme been eschewed. (2) Cutting out social welfare programmes would not really add as much as 11% of GNP to investment, since this 11% reflects the government's total expenditure on these services, including the entire education and health budgets, which could hardly be reduced to zero! (3) While it is appropriate to consider the favourable effects on growth rate of an increase in investment in physical capital, it would be absurd to claim that a reduction of expenditure on health and education would have no unfavourable effects on growth, but that is the assumption on which the growth rates in Table 6 have been calculated. (4) The capital-output ratio of Sri Lanka would tend to rise at the margin if the rate of capital accumulation is substantially raised (as envisaged in the last two rows of Table 6). This would pull down the growth, and increase the 45 In 1970-77, Sri Lanka's population growth rate was 1.7 per cent per year, compared with India's 2.1%, Bangladesh's 9 R% Ana PAki .anl's5 3.1%. 49 measure of advantage in terms of years. (5) Several of the countries in the cross-country fits have some social welfare programme, and the fits there- forereflect the relation between income and poverty, or income and longevity, not in the absence of social welfare programmes, but in the presence of social welfare programmes as they actually exist in the countries covered. By removing all social welfare programmes through reduction of the 11% allocation to 0%, Sri Lanka would not be joining the cross- country lines of relationship, but would tend to go under them. On all these counts the estimated measures in terms of years presented in Table 6 might understate the values of Sri Lanka's social welfare achievements. While the impressive - almost magical - powers of compound growth rates have been frequently noted, it appears that such compounding may still take quite a long time to catch up with the advantages of social welfare achievements in Sri Lanka. It is worth noting, in the context of assessing the longevity and poverty-removal achievements of Sri Lanka, that the per capita availability of food is not significantly different in Sri Lanka compared with other low-income countries.46 Indeed, according to some estimates, the calorie per head and the proportion of the average daily calorie requirement met by average availability of food are virtually the same for Sri Lanka as for the average of low income countries taken 46 See World Development Report 1979, page 168. 50 together. 47 The relative absence of undernourishment and the high longevity in Sri Lanka compared with conditions in other poor countries would seem to be largely the result of government's active policies, including food programmes (substantially altering the way food gets allocated under market forces)48 as well as other public facilities especially health services. This is particularly important because of 49 the high level of unemployment ruling in the Sri Lankan economy, and the low purchasing power of the poorer section of the non- food growers, in particular estate labourers. The question of food entitlements will be pursued further in section 7 in the context of analysing starvation and famines in Ethiopia and Bangladesh. I turn now briefly to Tanzania's truly remarkable achievement in the field of literacy. The raising of the adult literacy rate from around 10 per cent in early sixties to about 66 per cent by 1977 is a rather unique record. The big push came around 1970 when the government announced a national literacy programme to eradicate illiteracy by the end of 1975. This did not quite happen, but by the summer of 47 World Development Report 1979, Table 22. For a different assessment, see Richards and Gooneratne (1980, pp.10-11). 48 The food distribution arrangements have varied over the years in Sri Lanka. In the early sixties 4 lbs were offered per person per week at subsidized prices. In 1966 this was replaced by offering 2 lbs per person per week completely free, which was supplemented from 1970 by another 2 lbs at subsidized prices. In 1973 both the free part and the sub- sidized part were reduced, and the eligibility to get free rice was restricted to non-income-tax payers, ie., the poorer section. Subsidized flour, dal, etc., are also available, supplementing the rice offer. 49 See Morawetz (1977, Table 11) for comparative data. However, a substantial part of the Sri Lankan unemployed population seems to be educated - very often coming from other than the poorest section of the community. 51 1975, 90 per cent of the estimated number of illiterates were in fact enrolled. The "effective" literacy rate'jumped from 33 per cent in 1967 to 59 per cent in 1975, and since the 1967 criteria were known to have been less demanding than in 1975, the increase was close enough to an actual doubling of the literacy rate in this short period. The Musoma Resolut- ion in 1974 set the target of universal primary education by November 1977, and the initiative of the preceding years was consolidated into a decisive policy for basic primary education for all Tanzanians. A programme of this kind does, of course, require an enormous allocation of resources. The main resource, in fact, is that of teachers. here remarkably enough, the majority of instructors for the adult literacy campaign were volunteers, responding to the call for the removal of illiteracy. The teachers' salaries did form a significant part-of the national income, but given the magnitude of the task, 1.4 per cent of GDP being absorbed this way, as was the case in mid-seventies, can hardly be regarded as prohibitively expensive. Like Sri Lanka's programme of social welfare, Tanzania's literacy programme illustrates how much can be achieved by a determined effort, sensibly directed towards specific goals. If the Tanzanian educational pattern were to move along the lines of cross-country fits, it would have to wait for a long 52 time before rising income would make it "natural" for it to have a high level of literacy and primary education. The experiences of Sri Lanka and Tanzania illustrate the positive role of state action, as did the Korean case, though in a very different way. Strategies adopted by these economies are quite different. What is common is the use of deliberate and decisive public action. It is perhaps also worth remarking in this context that the experiences of Sri Lanka and Tanzania show that it is possible for a dedicated government to have a substantial impact on particular aspects of poverty and deprivation by manipulating certain crucial levers on which it has control, even when it does not have anything like the power that, say, a fully socialist government would wish to have. Issues of feasibility come up often enough in the context of the assessment of what a government can or cannot do when it does not have the kind of control with which their own philosophies have been directly concerned. An example perhaps is the dilemma faced by the communist government in West Bengal in India, dedicated to a radical social transformation of a kind that it is hardly in a position to carry out given the federal political structure of India and its over-all fragile political position.50 The experiences of Tanzania and Sri Lanka have some 50 While the party in power, viz., the Communist Party of India (Marxist), has the distinction of not being a follower of either of the two leading communist countries, viz., Soviet Union and the People's Republic of China, and is commited to the Indian Parliamentary system, its ultimate aim does, of course, remain the achievement of a classless communist society. 53 clear relevance to the wisdom of the fundamentalist position that "all or none" are the only viable alternatives. While questions of long-run objectives and short-run political survival may, with good reason, overshadow this rather more mundane issue, the issue nevertheless is clearly of some importance to the strategy of policy making. 6. Entitlements and Levels of Explanation Each economic and political system produces a set of entitlement relations governing who can have what in that system (see Sen (1976b, 1977, 1980a, 1980b)). Ultimately poverty removal is a matter of entitlement raising. There are both aggregative and distributive influences on the determination of entitlements, and while balanced growth might work through the former, institutional arrangements affecting the division of the cake work through the latter. The precise characterization of entitlement relations will, of course, vary with the nature of the economy and society (see Sen (1980b, Chapters 1, 5-10, Appendices A and B)), but some distinguished cases are easy to identify. For a market economy, the determining variables of entitlements can be broadly split into (i) ownership vector (e.g., the land, capital or labour power which a person owns), and (ii) exchange entitlement mapping (e.g., for each ownership 51 Cf. the experience of Allende's government in Chile. See, however, Foxley, Aninat and Arellano (1979) for an assessment of the important redistributive achievements of government programmes in Chile during the period of 1964 to 1973, ending up with the Allende years. 54 bundle the set of alternative bundles of commodities any one of which the person can acquire through production or trade). The ownership vector for a particular group may be enhanced either through over-all increase (e.g., capital accumulation), or through asset redistribution (e.g., land reform). These influences on entitlements through ownership changes are rather more palpable and tend to be more often discussed. Entitlement raising through improving the exchange entitlement mapping is a bit more complex to analyse, but may be no less important. It includes such diverse factors as, say, guaranteeing better terms of trade for poor peasants, or ensuring employment at a living wage, or providing social security protections. The method of removing poverty used in South Korea and Taiwan, as was discussed, is one of guaranteeing employment at a tolerable wage, and this has been possible by a very fast expansion of these economies using labour-absorbing production processes. In contrast, the level of unemployment in Sri Lanka is high, and the wages, especially of estate workers, quite low. There the guarantee of basic entitlements has not come through the market, but outside it, in the form of a direct right against the state, as was discussed in the last section. The contrast is not, of course, a pure one. Some social security provisionstdo exist in Taiwan, and to a lesser extent in South Korea.52 Similarly, in Sri Lanka even the pre-tax "gross" distribution of income is not as unequal as in some developing countries. But the crucial and distinctive features 52 See Galenson (1979) and Adelman and Robinson (1978). 53 See Jayawaraena (1974, p.273). 55 of entitlement enhancement in the two types of economies are different. In Taiwan and South Korea the social services only rather marginally improve what is provided by a buoyant labour market. Sri Lanka's rather "middling" performance in market-based entitlements is vastly improved by public 55 policy of distribution and social service. More importantly, there is an important similarity between the two strategies, which superficially might look poles apart. Ultimately, poverty removal must come to grips with the issue of entitlement guarantees. The two strategies differ in the means of achieving this guarantee. While one relies on the successfully fostered growth and the dynamism of the encouraged labour market, the other gives the government a more direct role as a provider of provisions. The similarities can be seen more clearly if we look at the distributional pattern through which these means work to guarantee entitlements. It is not easy to obtain internationally comparable income distribution data, but to get some broad idea Table 7 presents income distributional data for 16 countries taken from the World Development Report 1979. (These are all the countries with income per head less than $3,000 for which such data are given.) The incomes in question are of "total disposable household income", net of taxes. The World Develop- ment Report 1979 gives the data in the,.form of shares of each successive quintile. For ready comparison of parameters of 54 See Little (1979b), Adelman and Robinson (1978) and Galenson (1979). 55 See Jayawardena (1974) and Alailima (1978). 56 Table 7 Comparisons of Relative Equalities of Income Distribution Country Relative Relative Relative Relative Borda share of share of share of share of score bottom 20% bottan 40% bottan 60% bottan 80% Taiwan 8.7 21.9 38.5 60.8 4.0 Sri Lanka 7.5 19.2 34.7 56.6 10.0 19 69-70 Yugoslavia 6.5 18.4 36.0 60.0 11.0 1973 1964-65 6.7 17.2 31.5 51.1 17.0 South Korea 5.7 16.9 32.3 54.7 18.0 1976 Argentina 4.4 14.1 28.2 49.7 25.5 1970 Chl6e8 4.4 13.4 27.2 48.6 27.5 Philippines 3.7 11.9 25.1 46.1 34.0 1970-71 Cost Ric Costa Rica 3.3 12.0 25.3 45.2 37.5 Turkey 3.4 11.4 23.9 43.4 42.0 1973 Mexico 2.9 10.3 23.5 45.5 45.5 19 77 Venezuela 3.0 10.3 23.2 46.0 45.5 1970 Malaysia 3.3 10.6 22.8 43.5 46.5 1970 Peru 1.9 7.0 18.0 39.0 59.5 Honduras 2.3 7.3 15.3 32.2 60.0 1967_ Brazil 2.0 7.0 16.4 33.4 60.5 1972 C f I I Source: Calculated from Table 24 of the World Development Report 1979. 57 inequality, these have been presented here in the form of cumulative aggregates, i.e., the shares of the bottom 20%, bottom 40%, bottom 60% and bottom 80%, respectively. The most satisfactory method of comparison with these data is the so-called "Lorenz dominance". Country A Lorenz- dominates country B in terms of equality of income distribution, if compared with country B, country A has at least as high an income share for each x% poorest population, and a strictly higher share for some x% poorest population. When Lorenz dominance holds between two distributions of incomd, many interpretable conclusions can be drawn, and in general this may be regarded as the most satisfactory method of comparing the income distribution of two countries. On the other hand, this is a demanding criterion, and very often neither country Lorenz-dominates the other in a pairwise comparison. Lorenz domination yields a partial ordering of "clear" rankings, but given its incompleteness, some less demanding niethod of ranking has to be chosen to supplement Lorenz comparison if completeness is sought. A simple method of doing this - surprisingly little used in emnpirical comparisons is to follow Borda's (1781) method of rank-order scoring, giving points equal to the rank value of each country in each criterion of comparative ranking. This produces a coimplete ordering based on all the criteria taken together in terms of lowness.. of the sum of ranks (Borda scores). Both the Lorenz and Borda methods of ranking the 16 56 See Kolm (1969), Atkinson (1970, 1976), and Sen (1973c). 58 countries in Table 7 have been' used, and the results are represented in the form of two Hasse diagrams (with descending lines representing superiority of the higher placed country vis-a-vis the lower). The partial ordering of Lorenz dominance given by Hasse diagram 1 is subsumed by the complete ordering of Borda ranking in Hasse diagram 2. Interestingly enough, the countries which were commended for their performance in poverty removal and longevity extension come very much at the top of these rankings. Countries which were separated out in section 2 for disappointing performance in terms of poverty removal commensurate with their income, viz., Brazil, Peru and Mexico, also figure here towards the bottom of the .58' two rankings. Poverty removal and related features, including longevity enhancement, is ultimately dependent on a wide distribution of effective entitlements, and this - for any given level of per capita income - would tend to be reflected in the low level of inequality in the distribution of income. At this immediate level of explanation, in terms of entitlements and income distributions, Yugoslavia, Sri Lanka, Taiwan and South Korea have much in common despite their widely different political systems and economic strategies. At the remoter level the explanations differ, and - as discussed earlier - Taiwan 57 India is the joker in the pack showing up in a high position (fourth among the 16 according to the Borda ranking, below Taiwan, Sri Lanka and Yugoslavia, but above South Korea). 58 In fact, there is evidence that the income distributional situation in Peru and Brazil are even worse than that represented in Table 7, and alternative estimates suggest higher inequality. See World Development Report 1979, pp.185-6. 59 Taiwan 1971 Yugoslavia Sri Lanka 1973 1969-70 South Korea India 1976 1964-65 gen 1970 Ch le 1968 Costa Rica Philippines 1971 1970-71 Malaysia Mexico Turkey Venezuela 1970 1977 1973 1970 Per Honduras Bai 1972 1967 1972 Partial Order of Lorenz Dominance Hasse Diagram 1 60 Taiwan 1971 SrilLanka 1969-70 Yugoslavia 1973 India 1964-65 South Korea 1976 Argentina 1970 Chile 1968 Philippines 1970-71 Costa Rica 1971 Turkey 1973 Mexico & Venezuela 1977 1970 Ma aysia 1970 Peru 1972 Hlonduras 1967 Brazil 1972 Complete Order of Borda Scores of Relative Equality Hasse Diagram 2 61 and South Korea rely on high employment at a reasonable wage as an instrument, while Sri Lanka leans on security outside the market mechanism. At a still remoter level, the former pair of countries have profited from the strategy of export expansion, while Sri Lanka has not used such a strategy, nor is Yugoslav planning especially export oriented. As explained earlier, learning from the experience of Taiwan or Korea need not take the form of seeing virtues in export expansion irrespective of the nature of the country, and indeed the remoter we move in the chain of explanation, the more conditional, in general, the explanation becomes. The specific nature of the economic problems faced by these countries to which their economic strategies responded has often tended to be ignored. But on these specificities, the remoter explanations may substantially depend. Consider, for example, the strategy of relatively free trade with absent or refunded import duties, used by Taiwan. The necessity of such a policy for export expansion in the Taiwanese context has been expla.-ned admirably clearly by Maurice Scott (1979). "Taiwan has very few mineral resources and strictly limited agricultural resources. It could only prosper, and has in fact prospered, by exporting what effectively are labor and capital services. These services, however, have mainly taken the form of adding value to imported materials rather than being services pure and simple (as are tourism, banking, transportation, and the like). Thus the most important single condition for an export 'take-off' was access by manufacturers to imported materials, components, and capital equipment at world prices. For a typical manufacturer in one of Taiwan's export industries ... materials account for around 65 to 70 per cent of the selling value of output, capital 62 altogether receives a further 5 to 12 per cent of selling value, and, of this, over half can be attributed to costs of plant machinery, vehicles, and stocks of materials. Hence, the total percentage of selling value represented by the cost of materials or equipment might be around 70 to 75 per cent or even more for a typical exporter. If his exports are sold at world prices, without benefit of any subsidy, he must be able to buy these materials and equipment at or close to world prices if the export is to be profitable. A 40 per cent tax on materials and equipment would wipe out, not merely all potential profit, but also all potential wages, fuel costs, or payment for other bought-in services (advertising and other selling costs, rent of buildings and land, and so forth), and a much smaller tax, say 10 per cent, would make many exports totally unprofitable." 5 Under the circumstances the compelling advantages of Taiwan's policy regarding imports are clear enough, but these may or may not be shared by other countries with different economic characteristics (e.g., larger economies, greater mineral resources, less skill of the kind needed for export expansion in the present world markets, etc.). Similar discriminating analysis can be done with other instruments, e.g., high savings rate, social welfare delivery systems, and there is need for caution in reading lessons from success stories. The ways of achieving widespread entitlements differ, and while different economic strategies have things in common both in terms of ultimate consequences as well as in underlying causation, the precise combination of instruments and policies 59 Scott (1979, p.321). low proportion total income industrial growth below poverty achievements C m uwith exports growing line 0 employment- I' much faster oriented fast widespread b C more equal growth industrial growth cal6u2IAUd O low entitlements distribution without exports b and deter- poverty /of income growing much < mined / t \ c. - - faster / public C~~~~~~~~C public distri- policy I \ bution system high longevity non-income - - - -rNsocial welfare advantages programmes 0C - South Korea, Taiwan l K Yugoslavia C - Sri Lanka and Yugoslavia Conseauential Links and Levels of Explanation Diagram 3 64 varies tremendously with the conditions in different economies. In a somewhat stylized form Diagram 3 brings out such contrasts. Roughly path a tries to capture the causal mechanism in South Korea and Taiwan; path c that in Sri Lanka; and paths b and c that in Yugoslavia. It is common to all these strategies that low poverty in the form of high longevity and low proportion of population below the poverty line is brought about by a widespread distribution of entitlements, but beyond this immediate link there are sharp differences in the way such entitlement delivery is made. There is similarity also in the role ultimately played by calculated and determined public policy (see sections 3, 4 and 5). But while the Korean- Taiwanese causal mechanism goes through industrial expansion with exports expanding much faster, employment-oriented fast growth, achievements in total income enhancement and in income distribution, the Sri Lankan path goes through social welfare programmes, public distribution systems, achievements in income distribution and in non-income advantages (e.g., in people having better health when the income subsidy comes in the form of food, medical provisions, etc.). The yugoslav experience fits neither since it is somewhat similar to the Korean-Taiwanese strategy of high growth with good distribution, but differs from it in not having export expansion noticeably faster than the growth of national product 60 and in having a more developed social security system. 60 Yugoslavia's growth rate per year of GNP per capita during 1960-7f has been 5.6 per cent, which is a good deal higher than that of Sri Lanka's 2%, and closer to Taiwan's 6.2% and South Korea's 7.4%. On the other hand, while the annual percentage growth rates of exports of Yugoslavia, being 7.8% and 5.4% respectively over 1960-70 and 1970-77, were higher than those of Sri Lanka (4.6% and -5.3%), they fell far short of Taiwan's 23.7% and 16.7% and South Korea's 35.2% and 30.7%. 65 It is only natural: that "lessons' will be sought from the experience of those countries which have been unusually successful in poverty elimination. But the lessons to be drawn have to concentrate on understanding the functional roles of various instruments rather than on blind imitation of the instruments themselves. Other countries can learn much from the experieAce of success stories if the causal mechanisms are fully studied and remoter causes distinguished from the less remote ones.61 To what extent, say, fast growth is possible with export industries taking the lead, is a question that can be answered for any particular country only by looking .62 at the details of its economic circumstances. The same applies to the possibility of delivering provisions through social welfare programmes. Neither of these circumstantial condition- alities renders the experience of the successful countries unproductive for policy formulation, but blind imitation may well have counter-productive consequences. 7. Starvation and Entitlement Failues: Case Studies It was noted earlier that Ethiopia and Bangladesh are 61 This applies also to learning from the experience of - communist countries which have had an especially good record in the expansion of expectation of life (see section 3). 62 Taiwan and South Korea are quite different from the general run of developing countries in many respects, e.g., educational background, urbanization, port facilities, successful land reform carried out earlier, history of power:using industrial production, and the nature of the government. 66 among the poorest countries in the world in terms of GNP per capita as well as the usual parameters of poverty, e.g., proportion of people below the poverty line, and expectation of life. They are also among the few countries which have had actual famines in recent years, viz., in 1973 and 1974 in Ethiopia and in 1974 in Bangladesh. Insofar as there is hardly any manifestation of poverty that begins to compare with famines, the experience of these two countries in this respect deserve particular attention in an international view of poverty. In the context of success stories discussed in the preceding sections, much attention was paid to the widely shared pattern of entitlements guaranteeing the means of subsistence to the population. In turning to the grim tales of starvation and famines, the concept of entitlement remains especially relevant, and it may be useful at this stage to discuss the conceptual apparatus a bit more. Entitlement relations are primarily legal relations - who is given the right to have what according to (i) the prevailing legal system, and (ii) the economic, political, social and cultural characteristics affecting the position of different people within that system. A person starves when he cannot establish entitlement to an adequate quantity of food. Such failure of entitlement may or may not have very much to do with the total supply of food in that country. Further, to say that starvation depends not merely on the total supply of food, but also on its distribution, asserts hardly anything at all, since this just redescribes the fact of the starvation. What has to be examined is the 67 economic reality that leads to such unequal distribution of food and to starvation. It was mentioned earlier that in a market economy a person's entitlements depend on his ownership as well as on the terms at which he can exchange - through trade or product- ion - what he owns, for other commodities. A person's ability to acquire any bundle -with enough food can collapse either because of a sharp fall of his ownership, or because of a worsening of his "exchange entitlement", i.e., the terms at which he can exchange his ownership bundle into other bundles. 63 The importance of loooking into entitlement relations as opposed to over-all availability data is brought out by examining the inability of the availability data to give us clues as to the genesis of famines in Ethiopia and Bangladesh. The Ethiopian famine killed, according to various estimates, 64 between 50,000 to 200,000 people, most of them dying during 1973; while the famine continued into 1974 with a somewhat different geographical focus, the mortality was a good deal lower then. Was 1973 a year of exceptionally bad food avail- ability in Ethiopia? It would appear that this was far from the case. Table 8 provides estimates of calorie avail- ability per head in Ethiopia in 1973 compared with previous years, and it would seem that the availability in 1973 was more or less the same as that in previous years - higher than some and lower than others. 62 Formal presentation of entitlement relations can be found in Sen (1980b, Appendices A and B). 64 See Rivers, holt, Seaman and Bowden (1976) and Gebre-Medhin and Vahlquist (1977). 68 Table 8 Food Availability in Ethiopia: 1973 Compared with Previous Years Standard calories per capita Reference Vis-vis the year reference year the 1973 food availability was: 1972 3% less 1970 5% more 1967-69 7% more 1964-66 9% more 1961-65 1% less. Source: FAQ Production Yearbooks for 1971 and 1976. For methodological issues and some minor problems with these data, see Sen (1980b, Chapter, 7). The 1971 figures could not be obtained. The genesis of the famine can, in fact, be traced to varying patterns of entitlements. The famine concentrated in the province of Wollo, and it is certainly true that Wollo did have a. substantial decline in the output of foodgrains during that year. But the question that arises then is why didn't food move into Wollo to meet the unsatisfied demand since there was no real shortage of food for the country as a whole? This is where the nature of market entitlements comes in. The Wollo farmer who lost his crop not merely lost the food he could eat, but also his main ownership bundle with which he could enter the market. He did not have anything to buy food with in the market, and the good availability of food in the country as a whole will tell us nothing about his 69 ability to command food through the market mechanism. This entitlement explanation of starvation in Wollo can be contrasted with another explanation that has been offered, viz., the lack of transport facilities - especially the absence of a road network in Wollo. This explanation comes into difficulty for at least three different reasons. First, while roads within Wollo are underdeveloped, the two main Ethiopian highways pass through Wollo, and indeed the north-south highway passes right through the most affected region of Wollo. Food shortage in this region cannot, therefore, be explained by the lack of road development. Second, there are reports throughout the famine period of food moving out of Wollo into other regions of Ethiopia, especially Addis Ababa and Asmera.66 Third, despite widespread starvation and famine, food prices rose remarkably little in Wollo, and indeed for some crops even came down. Clearly, something more than food output collapsed in Wollo, viz., the power of the Wollo population to buy food and indeed even to retain the much reduced food supply of the province. Aside from the farmer, the pastoralist was also affected both in Wollo in 1973 (mainly nomads of the Afar community) and in Harerghe in 1974 (mostly nomads of the Ogaden Somali and Issa Somali communities). In analysing their deprivation, 65 For a more detailed discussion, with data, see Sen (1980b, Chapter 7). 66 Such food "counter-movement" is, in fact, a common phenomenon of a particular type of famines ( see Sen (1980a, 1980b)), and was a major political issue in the Irish famine of the 1840s, Chinese rice riots of906 and 1910, and on other occasions. 70 it can be seen that they suffered not merely from a fall in the animal stock due to drought, but even more from a precipitate fall in the exchange rate of animals for food- grains; these exchanges normally meet about half the calorie 67 requirements of the pastoralist. The entitlements were hit at both ends, and this pattern seems to be quite common 68 in the region (and in Sahel countries), whereby a fall in both animal stock and foodgrains stock goes with a sharp decline in the exchange rate against animals. The phenom- enon is connected with animals being a store of value for the pastoralist which he has to flog in the market when short of cash. It also relates to the important fact that foodgrains being a cheaper food, the pastoralist moves to consuning more foodgrain offering more animals in the market when the death 69 of animals makes him poorer. Hit by drought, the pastoralist was smashed by the market mechanism. In understanding the nature and the magnitude of .starvation in Ethiopia during the famine years, the entitlement relations have to be closely examined. While it is tempting to think of famines in Ethiopia being caused by general poverty, since it is one of the poorest countries in the world, the 67 See Sen (1980b, Chapter 7). 68 See Sen (1980b, Chapter 8). 69 A similar phenomenon can be seen with fishermen in Bengal, though somewhat less starkly. 70 The Afar pastoralist also suffered from the growth of commercial crops in the valley of the Awash river, depriving him from the best grazing ground for the dry season. 71 actual famine cannot be explained in terms of low average income or low food availability per head.71 The same is true of the Bangladesh famine of 1974. The foodgrains availability in Bangladesh in 1974 compared with that in the previous years is reported in Table 9. Table 9 Food Availability in Bangladesh: 1974 Compared. with Previous Years Vis-a-vis the reference year, the 1974 food availability was: Reference Alamgir's estimates FAO estimates year of foodgrains of calories availability per per capita person 1973 4% more 6% more 1972 4% more 6% more 1971 7% more n.a. 1970 7% less n.a. 1969 4% less n.a. 1968 1% more n.a. 1967 6% more n.a. 1966 same i n.a. 1965 8% less n.a. 1964 1% less j n.a. Sources: Table 6.23 in Alamgir (1979), and FAO Production Yearbook 1976. 71 The 1974 availability in Ethiopia was worse but not disastrously so, and the bulk of the famine deaths had taken place in 1973 itself. 72 While there were some years in the preceding ten in which food availability was higher than in 1974, In most years that was not the case. Indeed, if one were to look for a plausible candidate for being a "famine year" in the first half of the seventies, one would have been least likely to pick 1974, which had higher food availability than any other year during 1971-75. (The comparison with 1975 is not given in Table 9, but in fact the availability per head in 1974 compared with 1975 was 7% higher.72 In fact, unlike in Ethiopia in which starvation could be related to regional food availability (even though that itself requires an entitlement explanation, as we saw), in Bangladesh districtwise food availability data give no clue to the geographical distribution of hunger. The highest rates of destitution were observed in the districts of Rangpur, Mymensingh, Dinajpur and Sylhet. In terms of food availability per head, of the 19 districts these four were among the top five (indeed ranked 1, 2, 3 and 5)' 73 The explanation has to be sought in terms of variations in entitlements in the market economy of Bangladesh. The famine was precipita~ted by flooding of the Brahmaputra river during June to September of 1974. It affected the standing rice crop (aus crop, harvested in July-August), and more importantly, it affected the planting and transplanting of the winter-harvest of rice (aman crop), the most important crop in Bangladesh. The availability figures for the famine year 72 Alamgir (1979, Table 6.23). 73 See Alamgir (1979). 73 1974 - the famine raged from July to November - were not much reduced by these losses, since the aus crop was otherwise good, and since the aman crop, which did decline, is harvested in December - after the famine was over. On the other hand, the flooding had an immediate impact on the employment of rural labour, since the work requirement for harvesting aus, and more importantly, for planting and transplanting aman, was immediately and sharply reduced. An occupational breakdown of the destitutes shows that the most affected group - both in terms of absolute number of destitutes as well as in terms of intensity of destitution (percentage of population destituted) - was that of rural wage labourers. The entitlements of the rural labourers were slashed both by the decline in employment opportunity as well as by a downward movement of the exchange-entitlement of employed labour through a fall in the amount of rice commanded by the wage rate. The rice-exchange rate of rural labour fell by 42% for Bangladesh as a whole during June to October of 1974, and t.he fall was highest in Mymensingh, Rangpur and Sylhet - three of the four districts of top destitution.75 The market adjustments that have to be considered concern not merely the labour market, but also the commodity markets, in particular the market for rice (the staple food of the Bengali). While the decline of the aman harvest of,rice 74 See Sen (1980b, Chapter 9). 75 The fourth - Dinajpur - also had a decline but not so high. It appears, however, that a large proportion of destitutes in Dinajpur "came from the adjoining district of Rangpur" (Alamgir (1979)). 74 affected the total supply only after the famine was over, the anticipated decline had an earlier impact on the market supply of rice.76 This was particularly sharp since the anticipated decline came after a good deal of inflation, especially affecting rice prices,which had been going on for some years previously, and since the usual buffer of American grains was denied precisely at this moment because of a dispute involving Bangladesh's trade with Cuba, with the United States insisting on the suspension of jute exports to Cuba. 77 This is not the occasion to go into the details of the process that led to the 1974 famine of Bangladesh.78 However, it is important to recognize that despite the general. low level of mean income in Bangladesh, the famine struck not in a year of especially low total real income or low food availability, but, in fact, in a comparatively favourable year in terms of these aggregate magnitudes. It would, therefore, be a mistake to think of Bangladesh's - or Ethiopia's - poverty in primarily Malthusian terms focusing on relative movements of food output and population size. The ability of particular groups - the Wollo farmer or the Rangpur rural labourer - to avoid starvation is a question of their entitlements, and not of aggregate availability. The variability of market entitle- ments in the absence of social security provides the real clue to the famines. 76 See Alamgir (1979) and Sen (1980b, Chapter 9). 77 See McHenry and Bird (1977) and Sobhan (1979). 78 On this see Alamgir (1979) and Sen (1980b, Chapter 9). 75 The elimination of famines from China - the country that W.H. Mallory had called "the land of famine" 9 - relates to the policy of public guarantee of elementary entitlements. The vulnerability to shifts in exchange entitlement has been effectively controlled by the government providing guarantees 80 *of employment and basic means of subsistence. While poverty in the sense of relative deprivation and even of some.types of absolute deprivation clearly continueSto exist in China, the importance of eliminating the most extreme type of vulnerability can hardly be over-estimated. 8. Advise, Dissent and Critical Assessment A wide range of issues have been covered in this paper, taking an overview of levels of poverty, their changes over time, and some of the underlying causal mechanisms governing these changes. While there will be no attempt to summarize the discussion, it may be useful to put together some of the more Veneral findings to place the discussion in perspective. In discussing the issues with which this paper is concerned, it is sometimes asked: "What are the practical lessons? What would you advise the Finance Minister to do?" The format provided by answering these questions is rather limited from several points of view. First, the line between theory and practice is a difficult one to draw. Second, not all practical lessons happen to interest Finance Ministers. 79 Mallory (1926). 80 See, among others, Riskin (1975), Perkins (1976), Eckstein (1977), Ishikawa (1977), Lardy (1978), and Rawski (1979). 76 Third, many policy issues require more organization than any Finance Minister can provide. It is also the case that one might not wish to advise all Finance Ministers, and indeed one would very likely not wish to speak to some at all. Nevertheless, it is useful to think of what one might tell a sympathetic Finance Minister treating this as a format for gathering together what one may have learned. I have a go at this in what follows, facing a grim Finance Minister. (1) A poverty removal programme must, naturally, depend on the concept of poverty used. The focus here has recently shifted from viewing poverty as inadequacy of means e.g., insufficient income, to poverty as unsatisfactory results, e.g., low expectation of life. (Indeed, Minister, I could have lent you Morris's book on "Physical Quality of Life Index" but for the fact that your stupidly prurient Customs Officers have confiscated my copy.) There is much to be said for this change of focus, and also for the typical result- variables that have been chosen, e.g., longevity and literacy. On the other hand, there has been a tendency to take an unduly narrow view of results. Inability to buy enough food and prevalence of hunger are not merely "causes" of possibly disastrous consequences including death, but are also "results" in their own right, and should figure in an estimation of poverty even when they don't cause death (or illiteracy). Indeed, suffering from inadequate income, hunger and malnutrit- ion are features of poverty additional to the shortness of life. Even when the former contributes to the latter, both the shortness of life and suffering while alive have to be considered as separate (though not independent) parameters 77 of poverty. Thus, the exercises on longevity and literacy, important as they are, do not eliminate the need for looking at income deprivation, malnutrition and other variables representing current deprivation and suffering (section 1.3). (2) There are great methodological problems in basing poverty standards on average requirements of nutrition. Some authors have stressed the need to view poverty as relative deprivation. This provides, once again, an additional perspective on poverty, supplementing but not supplanting the approach of absolute deprivation (section 1.1). Another criticism of basing poverty on malnutritional information has come from authors such as Sukhatme and Srinivasan who have pointed out the variability of nutritional requirements from person to person. (This, Minister, is rapidly becoming the new wisdom on the subject, and your bright, young research officer is bound to tell you: "Malnutrition? Whatever precisely can you mean?") The issue of variability is indeed a serious one, and how one can tackle it will, of course, depend on the extent. to which regularities of distribution over groups can be observed. But aside from that there is also the question as to whether some status of its own should not be given to deprivation in the form of not having enough income to buy the average requirement of nutrition for that community, ,even when it is not clearly known what this person's own nutritional necessities are (section 1.3). This criterion, which leads to very old-fashioned measures, gets some support from Rawlsian arguments in favour of the Difference Principle. (3) Distributional questions have to be considered in any measure of poverty, and such crude indicators as the 78 head-count measure are inherently defective because they aggregate without taking note of the distributional issue. Distribution-sensitive aggregation procedures can be and have been devised (section 1.4). But all these measures dealing with realized deprivation leave out an aspect of poverty that relates to the sense of insecurity arising from the vulnerability of one's economic position. This insecurity is itself an aspect of poverty, even though the data on which all the existing poverty measures are based are inadequate in reflecting this aspect. It is important to look into the sources of vulnerability, if any, of different sections of the population (section 1.4). (4) Some countries were separated out for good performance in an international contrast, using data on longevity, literacy, income and the proportion of population below poverty line. (You asked, Minister, which political system wins these contests, but the answer is less than straightforward, as you will see.) Three groups stood out in these comparisons. First, the communist countries did relatively well whenever comparable data were available. Second, the four rapidly industrializing early capitalist countries in east Asia, viz., South Korea, Taiwan, Hong Kong and Singapore, have very good records, and the experience of the first two, which are sizeable countries, are particularly important to study. Third, Sri Lanka and Tanzania can be seen as having unusually good achievement in some specific features of poverty removal, despite their persisently low general income level (sections 2 and 3). 79 (5) While concrete policies going with poverty removal in these different types of countries have varied a great deal, there have been fundamental similarities in the variables through which these policies have worked. There is much evidence that the success of poverty removal depends ultimately on the ability to spread widely guarantees of economic entitlements, and there are striking similariti'e's in the experiences of these different types of countries in (i) the coverage of entitlements, (ii) lowness of relative income inequalities, and (iii) use of state power and reliance on public policy (sections 4, 5 and 6). The differences have rested in the way public policy has been used to provide entitlement enhancement. For example, in the Korean-Taiwanese experience this has happened, to a great extent, through employ- ment oriented rapid growth reaching the bulk of the population, while the Sri Lankan strategy has been to rely on social services providing a similarly widespread coverage. (6) There is much to be learned from the success of Sout1 Korea and Taiwan in removing poverty. Their policies in recent years have depended heavily on export promotion, which has been recommended to other developing countries. (You asked me, Minister, about the growing demand for concentrating all efforts on export promotion. In warning you against this, I would echo Orlando's words from As You Like It: "Thou art not for the fashion of these times./ Where none will sweat but for promotion" (II, iii, 56).) The relevance of export promotion to the essence of South Korean and Taiwanese strategy may, in fact, be a good deal less central than is sometimes claimed. 80 First, export promotion is only one part of a whole package of policies used by these countries, including massive capital accumulation and building up an industrial base through import substitution. Second, while the particular advantages and difficulties of these countries made export promotion both more easily achievable and more necessary, and other developing countries may or may not find it possible to - or may or may not need to - rely so much on export promotion, there are still other things to be learned from the experiences of South Korea and Taiwan (sections 4 and 6). The more immediate antecedent of poverty removal in these countries has been rapid expansion of employment. There are many aspects of this experience aside from export promotion, that require attention, including the combination of labour-intensive technology and rapid accumulation of capital. If the circum- stances of a particular country make the rapid expansion of employment more easily achievable through import substitution, then the Korean and Taiwanese experiences can be interpreted as supporting such import-substituting growth, at a deeper level (section 6). (7) It is important to view Sri Lanka's achievements not in terms of just one or two particular policies (such as food distribution), but as a general approach of spreading entitlements widely through direct public policy. (You asked, Minister, whether these social welfare programmes do not adversely affect economic growth. They might or might not, but even when they do, the trade-off between direct welfare advantages and growth advantages 81 seems to be rather favorable to the former.) Despite the well- known powers of compound growth, the time taken to catch up with the favourable results of direct welfare programmes by enhancing growth seems to be very long indeed (section 5). (8) The role of the government in poverty removal seems to be a common feature of the experience of such politically diverse countries as South Korea, Taiwan, Sri Lanka, Tanzania, and Yugoslavia, not to mention other communist countries. (You refer, Minister, to the role of the "invisible hand" in the South Korean breakthrough, but it was fortified by a visible arm that had few rivals in terms of muscle.) (9) Poverty and over-all inequality need not go together even when the average income level is given, since the former concentrates on the distribution towards the bottom while the latter takes a broader picture. As it happens, however, the countries which have had identifiably bad performance in terms of poverty in comparison with their income levels, e.g., Brazil, Mexico and Peru, also have had relatively more unequal over-all distributions of income. On the other side, countries that have performed relatively well in poverty removal also have done comparatively better in reducing the over-all inequality of incomes (section 6). (Minister, you can do worse than aim at a general reduction of inequality in seeking poverty removal.) (10) The worst cases of poverty seem to go not merely with low average prosperity, but also with especially vulnerable entitlement systems. This applies not just to the "security" view of poverty, which relates directly to such vulnerability (see section 1.4), but also to poverty judged in more usual 82 terms. Both Bangladesh and Ethiopia, which have bad records in poverty, have also actually experienced severe starvation and famines in recent years. (Food shortage, you say, Minister? Not quite.) The famines have been connected with sudden failures of entitlement systems. In fact in both countries the famines took place without any substantial decline in over-all food avail- ability, but the market mechanism played havoc with the entitlements of particular economic groups (section 7). The entitlement failures can be related to ownership declines or to the collapse of exchange entitlement, reflecting the terms at which a person can exchange his ownership into commo.dity bundles which he wishes to acquire. The Malthusian focus on food supply per head provides little clue to the genesis of the famines. (11) The point is sometimes made, correctly, that studies that concentrate on the relative performance of different developing countries in poverty removal (like this one) tend to miss an important general point, to wit, almost all developing countries have made a good deal of progress in longevity, literacy, etc., in the last few decades (see Table 3). Very often this progress has been substantially more than in the preceding thousands of years - a rather astonishing fact. Progress in technology had clearly been a factor in this, but it is also worth noting that there has been much more of deliberate public policy aimed at these goals in the last few decades than ever before. Achievements may have frequently fallen short of attempts, but certainly even limited attempts have contributed a great deal to achievements (section 5). (Even when, Minister, public policy makers have had control of only a few levers which they could manipulate, considerable achievements have 83 sometimes been possible when these levers have been used systematically.) To believe that deep down it is "all or nothing" has the aura of wisdom, but it can also be crucially misleading. Things are sometimes what they seem. (12) The issue of entitlement guarantee is the central question in poverty policy. While various alternative methods of providing such guarantees contrast with each other, it is on the existence of such guarantees that the success of poverty removal has ultimately rested. The destituted Bangladeshi rural labourer in the summer of 1974 would not have been thus fired in China, would have found employment elsewhere in South Korea, would have survived through social support (especially free rice) in Sri Lanka, and would have been saved by both employment opportunities and social security in Yugoslavia. The alternative strategies have to be seen, ultimately, in this light. Amartya Sen 84 References Adelman, I., and Robinson, S. (1978). Income Distribution Policy in Developing Countries: A Case Study of Korea, Oxford: OUP. Ahluwalia, M. (1978). "Rural Poverty and Agricultural Performance in India", Journal of Development Studies, vol. 14. Ahluwalia, M.S., Carters, N.G., Chenery, H.B., and Development Staff (1979). "Growth and Poverty in Developing Countries", World Bank Staff Working Paper No. 309 (Revised), Washington, D.C. Alailima, P.J. (1978). "Fiscal Incidence in Sri Lanka", ILO Working Paper, WEP.2-23/WP69, Geneva. Alamgir, M. (1978).. Bangladesh: A Case of Below Poverty Level Equilibrium Trap, Dacca: Bangladesh Institute of Development Studies. Alamgir, M. (1979). Famine 1974 - Political Economy of Mass Starvation in Bangladesh, manuscript, Bangladesh Institute of Development Studies; to be published. Anand, Sudhir (1977). "Aspects of Poverty in Malaysia", Review of Income and Wealth, vol. 23. Atkinson, A.B. (1970). "On the Measurement of Inequality", Journal of Economic Theory, vol. 2. Atkinson, A.B. (1975). The Economics of Inequality. Oxford: Clarendon Press. Aziz, Sartaj,ed. (1975). hunger, Politics and Markets: The Issues in the Food Crisis, New York: N.Y. University Press. Beckerman, Wilfred (1979a). Poverty and the Impact of Income Maintenance Programmes, Geneva: ILO. Beckerman, Wilfred (1979b). "human Resources: Are They Worth Preserving?", Paper to be presented at the 1980 Congress of the International Economic Association, Mexico. Bhagwati, J.N., and Krueger, A.O., eds. (1975). Trade Strategies for Economic Development, National Bureau of Economic Research. Bhalla, Surjit (1979). "The Distribution of Income in Korea: A Critique and a Reassessment", mimeographed, World Bank, Washington, D.C. 85 Bhatty, I.Z. (1974). "Inequality and Poverty in Rural India", in Srinivasan and Bardhan (1974). Blackorby, C., and Donaldson, D. (1980). "Ethidal Indices for the Measurement of Poverty", Econometrica, vol. 48. Booth, C. (1889). Life and Labour of the People in London, London. Chaudhri, D.P. (1979). Education, Innovations, and Agric- ultural Development, London: Croom Helm. Chenery, H., Ahluwalia, M.S., Bell, C.L.G., Duloy, J.H., and Jolly, R. (1974). Redistribution with Growth, London: Oxford. U.C. Chenery, H., and Keesing, D.B. (1978). "The Changing Role and Composition of LDC Exports", paper presented at the Saltsj6baden Symposium on the Past and Prospects of the Economic World Order, Institute for International Economic Studies, Stockholm. Choo, Hakchung (1975). "Some Sources of Relative Equity in Korean Income Distribution: A Historical Perspective", in Japan Economic Research Center and Council for Asian Manpower Studies, Income Distribution, Employment and Economic Development in Southeast and East Asia, Tokyo: JERC. Choo, Hakchung (1978). "Economic Growth and Income Distribution in Korea", Working Paper 7810, Korea Development Institute. Choo, hakchung, and Kim, Daemo (1978). "Probable Size Distribution of Income in Korea: Over Time and by Sector", Report Submitted to Council for Asian Manpower Studies, Korea Development Institute. Cole, D.C., and Lyman, P.N. (1971). Korean Development: The Interplay of Politics and Economics, Cambridge, Mass.: Harvard University Press. Cole, D.C., and Westphal, L.E. (1975). "The Contribution of Exports to Employment in Korea", in Hong and Krueger (1975). Datta-Chaudhuri, M.K. (1979). "Industrialization and Foreign Trade: An Analysis Based on the Development Experience of the Republic of Korea and the Philippines", ILO Working Paper, Asian Employment Programme, WP II-4, ARTEP, ILO, Bangkok. Dutta, Bhaskar (1978). "On the Measurement of Poverty in Rural India", Indian Economic Review, vol. 13. 86 Eckstein, A. (1977). China's Economic Revolution. Ann Arobor: University of Michigan Press. Fei, J.H.C., and Ranis, G. (1964). The Development of the Labour Surplus Economy: Theory and Policy, Homewood, Ill.: Irwin. Fields, G.S. (1979). Poverty, Inequality and Development, manuscript, Cornell University; to be published. Frank, C.R., Kim, K.S., and Westphal, L. (1975). Foreign Trade Regimes and Economic Development, New York: National Bureau of Economic Research. Foxley, A., Aninat,'E., and Arellano, J.P. (1979). Redistrib- utive Effects of Government Programming, Oxford: Pergamon Press. Galenson, Walter, ed. (1979). Economic Growth and Structural Change in Taiwan: The Post-War Experience of the Republic of China, Ithaca: Cornell University Press. Gebre-Medhin, M., and Vahlquist, B. (1977). "The Famine in Ethiopia - The Period 1973-75". Nutrition Reviews, vol. 35. George, P.S. (1979). Public Distribution of Foodgrains in Kerala - Income Distribution Implications and Effectiveness, Washington, D.C.: IFPRI. Hamada, K., and Takayama, N. (1978). "Censored Income Distributions and the Measurement of Poverty", Bulletin of the International Statistical Institute, vol. 67. Hsia, B., and Chao, L. (1978). Industrialization, Employment and Income Distribution: A Case Study of Hong Kong, London: Croom Holm. Hong, W., and Krueger, A.O. (1975). Trade and Development in Korea, Seoul: Korea Development Institute. Isenman, Paul (1980). "Basic Needs: The Case of Sri Lanka", World Development, vol. 8, March 1980. Ishikawa, S. (1977). "China's Food and Agriculture", Food Policy, May. Jayawardena, Lal (1974). "Some Country Experience: Sri Lanka", in Chenery, Ahluwalia, Bell, Duloy and Jolly (1974). 87 Kakwani, Nanak (1978). Income inequality and poverty; methods of estimation and policy applications, World Bank, Washington, D.C. Kakwani, Nank (1980). "On a Class of Poverty Measures", Econometrica, vol. 48. Kim, C.K., ed. (1977a). Planning Model and Macroeconomic Policy Issues, Seoul: Korean Development Institute. Kim, C.K., ed. (1977b). Industrial and Social Development Issues, Seoul: Korea Development Institute.. Kolm, S. Ch. (1969). "The Optimal Production of Social Justice", in J. Margolis and H. Guitton, eds. Public Economics, London: Macmillan, 1969. Krueger, K.O. (1979). The Developmental Role of Foreign Sector and Aid, Cambridge, Mass.: Harvard U.P. Kumar, Dharma (1974). "Changes in Income Distribution and Poverty in India: A Review of the Literature", World Development, vol. 2. Kumar, S.K. (1979). Impact of Subsidized Rice on Food Consumption in Kerala, Washington, D.C.: IFPRI. Kuo, Wan-yung (1975). "Income Distribution by Size in Taiwan Area - Changes and Causes", in Japan Economic Research Center and Council for Asian Manpower Studies, Income Distribution, Employment, and Economic Development in Southeast Tokyo: JERC. Kuo, Wan-yung (1977). "Labour Absorption in Taiwan 1954-71", Economic Essays, Taiwan University, vol. 6. Lardy, N.R. (1978). Economic Growth and Distribution in China, Cambridge: C.U. Press. Leibenstein, H. (1980). Inflation, Income Distribution and X-Efficiency Theory, London: Croom Helm. Lewis, W.A. (1955). The Theory of Economic Growth, London: Allen and Unwin. Lipton, Michael (1977). Why Poor People Stay Poor: Urban Bias in World Development, London: Temple Smith. Little, I.M.D. (1979a). "An Economic Reconnaissance", in Galenson (1979). Little, I.M.D. (1979b). "The Experience and Causes of Rapid Labour-Intensive Development in Korea, Taiwan, Hong Kong, and Singapore: And the Possibilities of Emulation", ILO Working Paper, Asian Employment Programme, WP II-1, ARTEP, ILO, Bangkok. 88 Mallory, W.H. (1926). China: Land of Famine, New York: American Geographical Society. Marris, Rohin (1979). "A Survey and Critique of World Bank Supported Research on International Comparisons of Real Product", World Bank Staff Working Paper No. 365, Washington, D.C. McHenry, D.F., and Bird, K. (1977). "Food Bungle in Bangladesh", Foreign Policy, No. 27, Summer. Morawetz, David (1977). Twenty-five Years of Economic Development 1950 to 1975, Baltimore: Johns Hopkins University Press. Morris, M.D. (1979). Measuring the Condition of the World's Poor: The Physical Quality of Life Index, Oxford: Pergamon Press. Nurkse, R. (1953). Problems of Capital Formation in Under- developed Countries, Oxford: Blackwell. Osmani, S.R. (1978). Income Distribution and Social Welfare, mimeographed, London School of Economics; to be published by Oxford University Press. Perkins, D. (1976). The Central Features of China's Economic Development, manuscript, harvard University; to be published. Rao, D.C. (1978). "Economic Growth and Equity in the Republic of Korea", World Development, vol. 6. Rawls, John (1971). A Theory of Justice, Cambridge, Mass.: Harvard University Press. Rawls, John (1975). "A Kantian Concept of Equality", Cambridge Review, February. Rawski, T.G. (1979). Economic Growth and EmLployment in China, Oxford: OUP. Reutlinger, S.,,and Selowsky, M. (1976). Malnutrition and Poverty: Magnitude and Policy Options, Baltimore: John Hopkins University Press. Richards, P., and Gooneratne, W. (1980). Basic Needs, Poverty and Government Policies in Sri Lanka, Geneva; ILO. Riskin, Carl (1975). "Incentives for Industrial Workers", in Joint Economic Committee, People's Republic of China: A Reassessment of the Economy, Washington, D.C.: U.S. Government Printing Office. Rivers, J.P.W., Holt, J.F.J., Seaman, J.A., and Bowden, M.R. (1976). "Lessons for Epidemiology from the Ethiopian Famine", Ann. Soc. belge Med. trop., vol. 56. 89 Rowntree, S. (1901). Poverty: A Study of Town Life, London: Macmillan. Runciman, W.G. (1966). Relative Deprivation and Social Justice, London: Routledge. Sastry, S.A.R. (1978). "Poverty, Inequality and Development: A Study of Andhra Pradesh", mimeographed, Sardar Patel Institute of Economic and Social Research, Ahmedabad. Scandizzo, P., and Knudsen, 0. (1979). "Nutrition and Food Needs of Developing Countries", World Bank Staff Paper No. 328, Washington, D.C. Scott, M. (1979). "Foreign Trade", in Galenson (1979). Seastrand, F., and Diwan, R. (1975). "Measurement and Comparison of Poverty and Inequality in the United States", mimeographed, presented at the Third World Econometric Congress, Toronto. Sen, A.K. (1970). "Aspects of Indian Education", Lal Bahadur Shastri Memorial Lecture, New Delhi; reprinted in S.C. Malik, ed., Management and Organisation of Indian Universities, Simla: Indian Institute of Advanced Study; also in P. Chaudhri, ed., Aspects of Indian Economic Development, London: Allen and Unwin. Sen, A.K. (1973a). "Poverty, Inequality and Unemployment: Some Conceptual Issues in Measurement", Economic and Political Weekly, vol. 8; reprinted in Srinivasan and Bardhan (1974). Sen, A.K. (1973b). "On the Development of Basic Economic Indicators to Supplement GNP Measures", United Nations Economic Bulletin for Asia and the Far East, vol. 24. Sen, A.K. (1973c). On Economic Inequality, Oxford: Clarendon Press. Sen, A.K. (1976a). "Poverty: An Ordinal Approach to Measurement", Econometrica, vol. 44. Sen, A.K. (1976b). "Real National Income", Review of Economic Studies, vol. 43. Sen, A.K. (1977a). "Social Choice Theory: A Re-examination", Econometrica, vol. 45. Sen, A.K. (1977b). "Starvation and Exchange Entitlement: A General Approach and Its Application to the Great Bengal Famine", Cambridge Journal of Economics, vol. 1. 90 Sen, A.K. (1979a). "Issues in the Measurement of Poverty", Scandinavian Journal of Economics, vol. 81. Sen, A.K. (1979b). "The Welfare Basis of Real Income Comparisons: A Survey", Journal of Economic Literature, vol. 17. Sen, A.K. (1979c). "Equality of What?" Tanner Lecture, Stanford University, May 197' to be published in the first volume of Tanner Lectures. Sen, A.K. (1980a). "Famines", World Development, vol. 8. Sen, A.K. (1980b). Poverty and Famines: An Essay on Entitle- ment and Deprivation, ILO World Employment Programme; to be published by Clarendon Press, Oxford. Smith, Adam (1776). Wealth of Nations; reprinted London: Dent, Everyman's Library. Sobhan, Rehman (1979). "Politics of Food and Famine in Bangladesh", Economic and Political Weekly, vol. 14. Srinivasan, T.N. (1979). "Malnutrition: Some Measurement and Policy Issues", mimeographed, World Bank and Yale University. Srinivasan, T.N., and Bardhan, P.K. (1974). Poverty and Income Distribution in India, Calcutta: Statistical Publishing Society. Sukhatme, P.V. (1977). "Malnutrition and Poverty", Lal Bahadur Shastri Memorial Lecture, New Delhi. Sukhatme, P.V. (1978). "Assessment of Adequacy of Diets at Different Income Levels", Economic and Political Weekly, vol. 13. Takayama, N. (1979). "Poverty, Income Inequality and Their Measures: Professor Sen's Approach Reconsidered", Econometrica, vol. 47. Thon, D. (1979). "On Measuring Poverty", Review of Income and Wealth, vol. 25. Townsend, P. (1974). "Poverty as Relative Deprivation", in Wedderburn (1974). United Nations (1975). Poverty, Unemployment and Development Policy: A Case Study of Selected Issues with Reference to Kerala. Report prepared by the Centre for Development Studies, Trivandrum. New York: United Nations. Visaria, P. (1979). "The Incidence of "'Absolute' Povertv in SriLanka 1969-70", Working Paper No. 6, Joint ESCAP-IBRD Project on the Evaluation of Asian Data on Income Distribution, U-_ -1, '91 Wedderburn, Dorothy, ed. (1974). Poverty, Inequality and Class Structure, Cambridge: C.U. Press. Westphal, L.E. (1978). "The Republic of Korea's Experience with Export-Led Development", World Development, vol. 6. RECENT PAPERS IN THIS SERIES No. TITLE OF PAPER AUTHOR 400 Rural Poverty Unperceived: Problems and Remedies R. Chambers (consultant) 401 Levels of Poverty: Policy and Change A. Sen (consultant) 402 Education and Income T. King, D. Jamison A. Berry (consultant) M.J. Bowman (consultant) G. Fields (consultant) L. Lau (consultant) M. Lockheed (consultant) G. Psacharopoulos (consultant) 403. Implementing Programs of Human Development P. Knight N. Coletta, J. Meerman M. Esman (consultant) J. Mbindyo (consultant) J. Montgomery (consultant E. Rogers (consultant) C. Safilios-Rothschild (consultant) N. Uphoff (consultant) 404 Population and Poverty in the Developing N. Birdsall World 405 The Effects of Education on Health S. Cochrane D. O'Hara (consultant) J. Leslie (consultant) 406 Poverty and the Development of Human Resources: W. Bussink Regional Perspectives D. Davies, R. Grawe B. Kavalsky, G. Pfeffermann 407 Human Resource Development and Economic Growth D. Wheeler (consultant) in Developing Countries : A Simultaneous Model 408 Economic Growth and Human Resources N. Hicks J. Boroumand 409 The Changing Nature of Export Credit Finance A. Cizauskas and its Implications for Developing Countries 410 Trade in Non-Factor Services: Past Trends and A. Sapir, E. Lutz and Current Issues 411 Population Policy and Family Planning Programs: K. Kanagaratnam Trends in Policy and Administration C.S. Pierce 412 Health Problems and Policies in the Developing F. Golladay Countries B. Liese