World Bank Reprint Series: Number 197 Paul Isenman Basic Needs: The Case of Sri Lanka pi Reprinted with permission from World Development, vol. 8 (1980), pp. 237-58. WVorld Developmenlt, Vol. 8, pp. 237-258 0305-750X/80/0301-0237/$02.00/0 © Pergamon Press Ltd. 1980. Printed in Great Britain Basic Needs: The Case of Sri Lanka PAUL ISENMAN* The World Bank, Washington, D. C. Summary. - This article lhas two related objectives: to judge Sri Lanka's success in meeting its 'basic needs' and growth objectives, and to use Sri Lanka's experience to cast light on general hypotheses regarding basic needs. The analysis suggests that Sri Lanka's social expenditures had a substantial cost in growth and unemployment. However, largely because of these social pro- grammes, it has the best social indicators, compared to its income, of any country for which data are available. Its growth has, surprisingly, been above average for low-income countries. Implications for basic-needs programmes include the need to 'target' social programmes, the high priority of primary education and the potential high impact, but also high costs, of assur- ing minimum caloric intakes. Recent interest in 'basic needs' has called following what we now call a basic-needs favourable attention to Sri Lanka's extensive approach, but rather was responding to political social programmes and unusually favourable pressures for a set of social programmes. In social indicators. It has also been argued, how- addition, since these programmes were available ever, that there has been a serious sacrifice of to the whole population, rather than just to growth in output and employment for improve- those whose basic needs had not been met, it ments in income distribution and welfare. This could also be said that they were not basic- paper seeks to address these questions. It also needs programmes per se. But the only way seeks to draw on the Sri Lankan experience to a basic-needs programme can succeed is if it address some broader questions which have has strong domestic political support. Simi- been raised regarding basic needs. For example, larly, if political realities mean that health and to what extent are there trade-offs and to what education programmes in most developing extent complementarities between growth and countries cover the well-off groups much better basic needs? Similarly, what are the trade-offs than they do the poor, then it seems politically and complementarities among health, education unrealistic to insist that approaches which meet and other elements of a basic-needs approach? the needs of both the poor and others should And to what extent is redistribution of income * be excluded from the definition of basic needs. a precondition in poor countries for meeting Rather, the Sri Lankan case is an instance of basic needs? how, in a private-enterprise-oriented and demo- cratic society, it was politically feasible to meet basic needs. 1. BASIC NEEDS Sri Lanka's accomplishments in meeting healtlh and education basic needs are indicated For purposes of this paper, a basic-needs approach to development can be defined as one which includes meeting basic health and edu- , cation needs as one among the major develop- I would like to thank P. Alailima, M. ul Haq, K. S. ment objectives and which emphasizes increasing Lateef, P. P. Streeten and E. B. Waide for helpful com- e h n c ments and encouragement in the preparation of this the human capital, productivity and incomes of paper. However, neither they nor the World Bank bear the poor as means of meeting basic needs and any responsibility for the views expressed. This paper increasing growth.' will also be published, in slightly altered form, by the It could be said that Sri Lanka was not World Bank. 237 238 WORLD DEVELOPMENT Table 1. Selected social indicators 1946 1953 1963 1973 Adult literacy (%) 58 65 72 78* School enrolment (% ages 5-14) 41 58 65 86* Life expectancy (yr) 43 56 63 66* Infant mortality (per 1000) 141 71 56 46 Death-rate (per 1000) 19.8 10.7 8.6 7.7 Birth-rate (per 1000) 37.4 38.7 34.3 27.9 Natural population growth rate (% 1.8 2.8 2.6 2.0 Population growth rate (%) 2.3 3.3 2.5 1.6 (including migration) * 1971. in Table 1.2 Sri Lanka has an exceptionally good over the past 1 5 yr Sri Lanka has spent a record, in relation to its per capita income, remarkably high percentage - about half - of on life expectancy, infant mortality, fertility its current expenditures for its food subsidy, and literacy. To compare its social indicators health and education programmes. However, to those of other countries at different income as Table 1 shows, there has also been a good levels, regression analysis was carried out on a deal of progress in social indicators since 59-country sample of countries for which independence. The links between social expen- complete data sets were re-dily available. The ditures and social indicators are discussed in the results of that analysis are summarized in following sections. Figure 1. In each case, Sri Lanka's social indi- Moreover, the 'initial conditions' of com- caters, relative to its income, were the best paratively high literacy and life expectancy at among the 59 countries.3 independence, are themselves the result of a previous political commitment, by the Ceylonese people and to some extent by the British (a) Pre-independence 'initial conditions'and colonial govemment, to the meeting of basic post-independence social programmes education and health needs. Increasing life expectancy to over 60 yr or attaining near- The roots of Sri Lanka's social progress go universal primary education is a time-consuming back at least before independence (1948) and process. Success in implementing extensive to some extent to its Buddhist heritage. Food health, education or small farmer programmes rationing was begun in 1942; primary education may itself depend partly on investments in was made nominally (although not in practice) basic education over the previous generation. compulsory in 1901. In 1946, life expectancy And while Sri Lanka's commitment to basic in Ceylon was 43, while in India, for example, education and health services may have been it was only 32 (for 1941-1950); in 1946, adult unusual for the 1940s, many poor countries literacy (over age 10) in Ceylon was 5 8%, while today have internal social pressures for the in India total literacy (all ages) was only 17%. spread of these basic services at least as strong Also, Sri Lanka has benefited from a benevol- as those felt by Sri Lanka in the 1940s. ent ecological situation. It has had, until re- cently, ample land resources and a relatively abundant supply of nutritious foods, particu- (b) Education larly coconut, which provides about one-fifth of total calorie consumption. In addition, large Despite attempts at educational reform export earnings and tax revenues from tea and dating back to pre-independence, Sri Lanka rubber have until recently made it \feasible for has essentially followed the education system it the Government to afford large expenditures inherited from the British, although the for subsidies and services. medium of instruction has been changed to One recent article goes so far as to suggest Sinhala gind Tamil. The curriculum has been that Sri Lanka's accomplishments are essentially focussed primarily on the academic needs of due to these pre-independence initial conditions, the small minority going beyond the 'O' (sec- but gives Sri Lanka full 'credit' for slowing ondary) level, rather than on the development- down growth by overspending on social pro- oriented learning needs of the majority who grammes during the past 15 yr.4 It is true that drop out before then. The education system BASIC NEEDS. THE CASE OF SRI LANKA 239 Life expectancy and income Fertility and income 750 F 220.0 - 700 2000 \Sri Lankan expected value >. 65 0 1800 - ,s 60.0 - +Sri Lankan actual value x 1600 c .v 1400 - 550 -. 1200 0. 0. 50.0 + 1000 + Sri Lankan actual value 45.0 - /Sri Lankan expected value 45-0 80.0 - 40.0 600 35 0 l 400_ 0 100 200 300 400 500 600 700 .00 100 200 300 400 500 600 700 Per copita income, U.S. $ Per capita income, U.S. $ Infant mortality and income Adult literacy and income 1800-o 1000 900 1600 - + 800 1400 - ri Lankon expected value 70 +Sri Lankan actual value 700- 1200 600 E 100.0 500 E 500 - C 00 - \ 400 Sri Lankan expected value 600 -300 z + Sri Lankan actual value 200 400 - 200 200 ZioZ 0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 Per capicta income, U. S. $ Per copita income, U. S. S Figure 1. Sri Lanka's social indicators in relation to expected values for its per capita income ($130) in 1975. (The curves show expected values at any given income level.) Source: Appendix Table Al. Sri Lankan social indicators are for 1971 rather than 1975, thus slightly under- stating its record (except for infant mortality, which had not declined) is highly centralized on a nationwide basis, the 20-24 age bracket in 1971, 71% of men with little involvement of local communities and 64% of women had at least some education and little attention to non-formal education.5 beyond the initial 4 yr of primary education, The serious problem of educated unemployment and 26% each of men and women had at least a (Section 4) indicates that marginal social secondary school ('O' level) certificate.7 The retums to the expansion of secondary edu- open access to education also contributed to cation were low. In short, education in Sri social and economic mobility and, thus, to a Lanka has deviated sharply from what might weakening of the political and economic power be called a basic-needs education strategy. of the traditional landed elite. This is not to say On the other hand, Sri Lanka's educational that these accomplishments might not have accomplishments have been remarkable for been surpassed with a more 'relevant' system, a country of under $200 per capita. As a result but that the Sri Lankan case seems to contra- of high expenditures and high enrolment rates dict the view that basic needs in education at all levels of education, adult literacy increased cannot be met through relatively conventional from 58% in 1946 to 78% in 1971.6 Of those in approaches. More importantly, Sri Lanka's 240 WORLD DEVELOPMENT seemingly 'irrelevant' education system appears Asian countries, Sri Lanka's ration and subsidy to have contributed substantially to its ac- programmes have had effective coverage of the complishments in health, fertility reduction and poor and of rural areas. These programmes have agriculture. been roundly criticized for the high percentage of the government budget it preempted (aver- (c) Health aging about 20% of current expenditures in the past 15 yr if rough adjustment is made for The coverage of the Sri Lankan health overvaluation of the rupee) and for its disin- system and its impact on the reduction of centive effect on agricultural production. The mortality shown in Table 1 was dramatically former criticism appears well justified, since the illustrated when the number of reported deaths ration coverage was nearly universal, while attributed to malaria averaged only about 5/yr severe nutritional problems affected only low- from 1970-1974, although the resurgence of income groups. The latter criticism is more corn- malaria caused the number of clinically con- plex."2 Rice production grew by an extremely firmed cases to average about 215,000. (The high 5.8% annually during the 1960s. Compara- control of malaria alone in 1946 through DDT tively high paddy prices and high subsidies on spraying had led to a drop in the crude death- inputs provided an adequate incentive to farm- rate from 20 to 14.)8 Similarly, maternal ers, by international standards during the 1950s mortality declined from 16/1000 in 1946 to and 1960s.13 The reasons that there was not 1.2/1000 in 1970, partly because of the spread the expected disincentive effect on prices in the of maternity care; at present, more than two- 1960s from the large-scale food imports were thirds of births take place in hospitals or l that the subsidy on rice plus the growth of the maternity centres. population and per capita income generated a Two aspects of Sri Lanka's health system substantial amount of additional demand for seem particularly relevant for other countries foodgrains; and that Sri Lanka's worsening now trying to address basic health needs. First, balance of payments kept pressure on the Sri Lanka has had both extensive coverage of Government to increase domestic foodgrain the population through primary health care production in order to save foreign exchange. facilities staffed by paramedical workers and However, there has probably been a moderate a strong back-up 'referral' system of clinics and disincentive effect on rice in some years of the hospitals manned by both physicians and para- 1970s. Also, the flour subsidy of the past sev- medical workers. The need for the former has eral years has undoubtedly had a disincentive been too often neglected, but is now becoming effect on minor cereal crops, particularly sor- a part of the accepted wisdom on health ghum, for which there have been no price sup- planning; with this change in thinking, there is ports a danger that the pendulum might swing too Less attention has been paid to positive far, and that in the planning of extensive health effects of the food ration and subsidy pro- systems insufficient attention will be paid to grammes.14 In general, where incomes are close the need for an adequate back-up supervisory to bare subsistence levels and are variable (due and referral system.9 Second, the 'Western' in the case of Sri Lanka to the vagaries of rain- health care system coexists with an important, fall and the varying demand on tea estates for and Government-regulated, system of traditional labour), it seems quite plausible that a food 'ayurvedic' medicine. All economic groups ration programme that reaches the poor effec- make use of the ayurvedic system for certain tively would substantially reduce malnutrition. kinds of illness or injury.10 The traditional This would occur partly through the direct health system appears in general to complement effects of the ration in increasing consumption; and to help take the strain off government also, to the extent that the ration substitutes 'Western' health facilities."l for food that would otherwise have been bought and, thus, represents an income supplement, (d) Food subsidies nutrition will still improve, since the poorer the family the larger the percentage of income spent Sri Lanka has had food rationing for 35 yr on food. (In addition, among very poor families, and has also subsidized wheat and sugar for much whatever of the income supplement is not spent of the 1970s. The rice ration (or in the 1970s on food will still be spent in large part in meet- the wheat and rice ration) has varied frequently. ing basic needs.) Since a large percentage of It has generally stayed between 2 and 4 lb/week, deaths among children in poor developing coun- 1-2 lb of which was free, with the remainder tries are directly or indirectly related to malnu- sold at a subsidized price. Unlike other South trition, an effective food ration programme BASIC NEEDS: THE CASE OF SRI LANKA 241 would be expected to reduce mortality, particu- cept to bakeries, from October 1973 to August larly in years of high food prices. Also, food is, 1975.19 In many cases, supply problems meant politically, a special commodity. The severe i that people could not obtain their ration en- political constraints to substantial redistribution titlements. Because staple foods are a basic of incomes or assets by taxation or other non- necessity, the reduction of government grain revolutionary means have often been noted." sales caused a sharp increase in grain prices. In However, the constraints appear to operate less an attempt to acquire more food for distri- stringently whein food, rather than incomes, is bution, the government then restricted inter- to be distributed, as indicated by experience district rice trade, but this exacerbated the with feeding programmes in a number of coun- price increase in deficit districts. The net result tries, including the 'food stamp' programme in was that many poor people could not afford the US.16 enough food for minimum nutritional stan- These points are readily demonstrated in dards. This suggests a link between the cut in Sri Lanka. Data from the 1969/1970 Socio- foodgrain distribution and the increase in the Economic Survey show that in 1970, the ration death-rate. provided about 20% of total caloric intake for The hypothesized link receives statistical families with household incomes under Rs. 400/ support from the multiple regression analysis month (and 15% for those from Rs. 600-1000/ reported in Appendix Table A2, which indicates month); even those in the under-Rs. 400 per that the price of rice is statistically related to capita category had average intakes of about changes in the death-rate. If this hypothesis 2050cal/day. Only about 25% of the popu- was valid, one would expect the death-rate to lation in that year had consumption under the be highest among those most dependent on generally accepted (but rather high) 2200 cal/ government distribution programmes. In fact, a day 'requirement' and only 5% under 1900 cal/ disproportionately large part of the increase in day. In Bangladesh, for example, a country deaths occurred among the Indian Tamil estate with much worse nutrition problems and a workers, who purchased rather than grew most much higher death-rate, 25% of the population of their food and had adopted wheat as the consumes less than 1700cal/day.'7 As an indi- main staple of their diet. cation of the importance of food subsidies in The regression analysis indicates that the increasing incomes of the poor, the 1973 Sur- sharp increase in the death-rate cannot be dis- vey of Consumer Finances shows these sub- missed as a random annual fluctuation.20 Simi- sidies to be equal to about 14% of the income larly, such a spurt could not be caused by the of those with 'spending-unit' (a slight variant change in the age structure of the population or on 'household') incomes under Rs.400/month. the deterioration of the health care system, It is difficult, and not required for purposes of which occur much too gradually, or by the rela- this paper, to separate out the improvements in tively insignificant number of malaria or cholera welfare resulting from provision of subsidized deaths. While shortage of food distribution food per se from improvements resulting from under government programmes was no doubt the income supplement implicit in the sub- not the sole cause of the increase in deaths, the sidies.18 In either case, nutrition will improve results of the statistical analysis and the judg- and some additional funds will be freed up for ments of a number of Sri Lankan doctors, tea non-food expenditures, although to varying estate personnel and others with whoam I spoke extents. leave little doubt but that it was a major There is also evidence of increases in mor- cause.2' tality from cuts in the ration and subsidy pro- In sum, both the low death-rate in general gramme during years of high food prices. In and its increase in 1974, indicate that the rela- 1974, the death-rate increased from 7.7 to 8.9/ tively equally distributed supply of foodgrains 1000. Rice production in that year was much in most years, to which the ration and subsidy above average and overall calorie availability programmes made a substantial contribution, only slightly below average. But due to the has helped Sri Lanka to avoid the high degree rapid surge in international food prices, imports of malnutrition-related death common in low- of rice and wheat flour declined substantially. income countries. At the end of 1977 the new This shortage of imports, combined with cut- government took the politically bold and highly backs introduced in 1973 in the per capita desirable step of reducing the coverage of the ration allocation, caused distribution through ration programme to those in the bottom half the ration prograinme to decline by 1 5%. Off- of the income distribution, partly to focus sub- ration sales of wheat flour (a government im- sidies on those who need them, and partly be- port monopoly) were terminated entirely, ex- cause the November 1977 devaluation and float 242 WORLD DEVELOPMENT of the rupee almost doubled the direct budget- between education and health programmes. ary cost of subsidizing imported foodgrains. In This kind of linkage, while undoubtedly present September 1979 a food stamp programme was to some extent in Sri Lanka, seems rather substituted for the ration programme, and the deficient there and certainly does not stand government announced that the general subsidy out as a major cause of its social accomplish- on wheat flour would also be phased out. Tar- ments. Another important aspect of basic- geting on those with low incomes will not only needs linkages is at the output level. Multiple save money but is the best way to avoid rep- regression analysis of the 59-country sample etition of the experience of 1974. In shortage referr3d to previously indicates a series of such years, food available will be able to be concen- linkages. While the application to particular trated on the poor and malnourished rather countries of results of cross-section regression than spread more thinly. analysis must be approached with caution, the Sri Lanka's experience, both positive and results (reported in Appendix Table Al) were negative, with food ration and subsidy program- statistically highly significant and fit with mes suggests that while these programmes are widely accepted behavioural hypotheses. very expensive and must be approached with The regression results suggest that literacy caution, they can have substantial benefits. has an important impact on life expectancy and Their feasibility in any given country would be irfant mortality (after correcting for the effect determined in part by the extent to which pol- of income). This is not surprising, since a high itical factors allow a relative focus on the poor percentage of deaths in developing countries is (to hold down costs), and avoidance of disin- related to home hygiene and other health and centive effects on agricultural production. Agri- nutrition practices, and since literacy increases cultural product-on, budget, balance-of-pay- knowledgeability and ability to screen and ments and food aid prospects would obviously evaluate new information about changing these also be major determinants. Suggesting con- practices.23 The regression equation linking sideration of ration programmes is unlikely to literacy to life expectancy suggests that Sri be a popular rucommendation; tough-minded Lanka's very high literacy rate (78% rather than economists are supposed to oppose such 'hand- the expected value of 36%) would bring the outs'. However, most of the few non-Communist expected value of life expectancy at Sri Lanka's poor countries which have very low death-rates income level from 44 to 58. Thus, education, - Sri Lanka, Burma, Mauritius (which had a and factors associated with it, appear to explain low per capita income up to about 1970), Egypt a large share - the equation suggests about two- and Kerala (although it is a state rather than a thirds - of the 22-yr difference between its country) all have heavy subsidies on food- expected and actual (66) values.24 grains. (This is, of course, not to argue 'hat a No regression analysis has been carried out ration or subsidy programme would be a suf- on the linkage of primary education (literacy) ficient condition for replicating the social to growth in agricultural production. However, accomplishments of these countries.) Unfortu- there is a growing body of empirical studies nately, political, human resource, technological indicating that education has a positive effect and other constraints mean that in most coun- on agricultural output when technology is tries neither redistribution nor likely growth in changing.25 The reasons why literacy increases output or agricultural production will solve the the utilization of improved agricultural prac- problem of inadequate food consumption by tices are essentially the same as those why it poor families within any reasonable time increases the utilization of improved health frame.22 If targeted ration or subsidy program- practices. Given the risk aversion of small mes have fewer political constraints than other farmers and the relatively high financial risks redistribution programmes and if they appear of input-intensive new agricultural technologies, feasible by the criteria discussed above, they education would appear helpful in absorbing, deserve careful consideration. accurately recalling and evaluating proposed new technologies. It may also have an effect on (e) Linkages among steps to meet basic needs attitudes toward trying things that are new. Thus, high literacy could well be one of the A major tenet of basic-needs approaches to reasons that rice yields in Sri Lanka were the development is that there are important re- highest in South Asia in 1960 and that rice inforcing linkages among the steps required to production increased by 5.8% annually during meet basic needs. Emphasis is often put on the 1960s. linkages at the input level, as in integrated The regression results show that longer life rural-development projects, or in co-ordination expectancy contributes to lower fertility rates BASIC NEEDS: THE CASE OF SRI LANKA 243 (again after allowing for the effect of in- perienced rapid growth in agricultural out- come).26 The impact of longer life expectancy put without substantially higher literacy than (or lower infant mortality) on fertility relates Bangladesh's level of under 30%. And there are primarily to the expectation that children will equally few which (in the mid-20th century) survive into adulthood.27 These results indicate have experienced a rapid decline in population that Sri Lanka's high life expectancy appears growth without a substantially lower infant to explain about two-thirds of the difference mortality rate than Bangladesh's 140+/1000. between its expected and (much lower) ob- served fertility levels. While literacy had no additional explanatory power when added to (f) Substitution among steps to meet the regression analysis, it has a strong indirect basic needs effect through its impact on life expectancy. Recent analyses of the relationship of devel- Since basic needs has become a focus for lopment to fertility suggest some hypotheses analysis so recently, little analysis has been which appear applicable to Sri Lanka and which done on substitutability among what have come would amply explain the remaining diffc.ence to be considered as the minimum package of between Sri Lanka's expected and observed steps to meet health and education basic needs. fertility rates.25 One is the effect of the level of Amor- the elements that have been identified female education. (This effect is additional to as es,~ential to meeting basic needs are potable that of literacy per se, which is attained with water (as a means of achieving the broader only primary-school education or less.) The basic need of good health) and 'participation', 1971 census shows that, for example among which some consider to be a means to achieving women aged 40-44, those who had reached 'O' basic needs at low income levels and others to level qualifications had an average of 3.6 chil- be a basic need per se.31 dren, while those with only primary education In general, the quality and quantity of drink- had an average of 5. 1, and those with no school- ing water available to Sri Lankans is considered ing an average of 6.0.29 In addition, female inadequate by the WHO for good health.32 education has contributed to the relatively However, Sri Lankans tend to be well awarc of high female labour force participation rate (a the dangers of polluted water and frequently crude rate of 22% in 1975, up from 16% in treat or boil it, particularly for use by young 1963), to the increase in average female age at children. Also, there is general awareness of marriage (23.5 in 1971, compared to 22.1 in when medical attention is needed to treat gas- ] 963) and to the related decline in the percent- trointestinal and other water-borne diseases.33 age of married women.30 Another apparent These health practices seem most plausibly reason for low fertility has been the ration (and related to the high degree of literacy (basic food subsidy) and health programmes, which education).34 For example, infant mortality, have lowered the old age and disability in- which is closely linked to water-borne diseases surance motivation for having large families. A is highest in those areas (i.e. the 'estates') where further reason is Sri Lanka's family planning education is lowest. Thus literacy and the pri- programme, which works largely through the mary health care system appear to substitute in health system. Overall, the rapid decline in part for potable wmater supply in meeting basic fertility shown in Table 1 may be attributed health needs. largely, although not completely, to the direct In part, Sri Lanka has also 'substituted' for and indirect effects of Sri Lanka's social pro- local participation.35 In some programmes, it grammes. has used a highly centralized ('top-down') In sum, the strong linkages among basic approach in situations which some argue require needs suggested by the inter-country regression highly participatory ('bottom-up') local insti- analysis and by the case of Sri Lanka indicate tutions. In other ways, though, participation substantial indirect benefits from health and clearly has been important in Sri Lanka's social education programmes. These linkages raise accomplishments. some interesting questions of emphasis and As noted previously, Sri Lanka's education sequencing for development strategies in other and health programmes are highly centralized. countries. For example, it is often said that des- This centralization has had obvious disadvan- perately poor countries like Bangladesh should tages in terms of foregone financial and other concentrate their limited development efforts support from local communities. On the other on agricultural production and family planning. hand, the centralization helped avoid or miti- These undoubtedly are key priorities. But there gate situations where the best organized - often are extremely few countries which have ex- most prosperous -- localities would end up with 244 WORLD DEVELOPMENT the best services and with a widening gap be- their parties. This assistance includes, in part, tween them and mnore disadvantaged localities. lobbying for schools, clinics and agriculture Obviously, the question here is one of the right programmes. balance between 'top-down' centralization and 'bottom-up' participation. The many advantages of the latter have been increasingly recognized (g) Uiinmet basic needs of estate labourers in recent years, but it is easy to go overboard in this, as in other areas. It has even been argued Sri Lanka's progress in human resources by some who view participation as a necessary should not obscuru the fact that it remains a aspect of a basic-needs approach that because very poor country where malnutrition is such Sri Lanka has not had a good record on partici- that there is death from nutrition-related causes pation in local institutions, it has not had a in food-short years, where the infant mortality good record in meeting basic needs. This view, rate has stagnated at about three times that in however, seems rather doctrinaire, especially in Northern Europe, and where dropout rates light of Sri Lanka's extraordinary record on from primary school remain high. The most social indicators.36 severe human resource problems tend to be On the other hand, Sri Lanka's experience in found on the tea 'estates'. The vast majority of agriculture generally supports the importance estate workers are 'Indian Tamils', whose famil- of participation. In the 1 960s, the local 'Rural ies were brought by the British to Sri Lanka Development Committees', 'Cultivation Com- within the past century specifically to work on mittees', and co-operatives were, taken as a the estates. Whether the estate labourers fall group, considered generally successful in help- into the bottom third of the income distri- ing to improve agricultural productivity. As is bution in any given year depends largely on the discussed in the following section, the 1960s number of days of work available for them on were a period of rapid agricultural growth in the estates. When there is a strong demand for non-plantation crops. In the 1970s, experience labour, as has recently been the case, their with local rural institutions was much less household income tends to fall roughly in the favourable. There were widespread charges that middle of the income distribution. However, these institutions had become highly politicized their health, education and nutrition38 status is and inefficient. This period of institutional muclh worse than that of the population as a decline (from 1970 to 1977) coincided with a whole. For example, their death-rate was 55% period of very poor agricultural growth. This is, above that of the country as a whole for 1973 of course, not to deny the primary importance and 130% higher during the food-short year of of other factors - including input supplies, 1974. In 1969/1970, 52% of estate women had pricing policies and weather - in explaining the no schooling at all, compared to 23% in the difference in agricultural growth between the rural sector (excluding estates) and 15% in the 1960s and 1970s. urban sector. Only 1.5% of the estate popu- 'Participation' usually refers to participation lation had 'O' level qualifications, compared to in local institutions, regardless of the extent of 7.5'/.(" for the population as a whole. Also, low- participation in choosing local or national pol- income estate dwellers get much lower benefits itical leaders. For example, both the People's from government subsidy and service program- Republic of China and the Republic of China mes than does th(- test of the population (even (Taiwan) are often cited as examples of highly before taking account of the indirect incidence participatory systems in rural areas. However, of the heavy taxes on tea).39 as the Sri Lankan example shows, the extent of The reasons are easy to see: the cultural and local participatior. in choice of political leader- physical isolation of the estates; insufficient ship is also highly relevant to the meeting of political power to command a fair share of basic needs. Indeed, if we redefine the term par- services and subsidies; the conflict for mothers ticipation in this broader way, then Sri Lanka is between working and looking after their young a highly participatory society, and much of its children; low economic returns t't education, social accomplishments can be traced to local partly because children do not have to be liter- participation.37 Sri Lanka's democratic political ate to have first priority for work on the estate system is highly competitive at the local as well on which they live; and the fact that funds for as national level. Interest among the generally improving health, education and shelter directly literate populace in local and national elec- reduced the profits of the tea estates (except tions is high. MPs and Village Council chair- for those health measures which increased rev- men seek to barter assistance to their constitu- enues by more than costs). Nevertheless, Sri encies for political support for thenmselves and Lanka's general commitment to and success in BASIC NEEDS: THE CASE OF SRI LANKA 245 meeting basic needs calls added attention to the per capita. Sri Lanka's per capita income disparity between the conditions of estate growth in the I 960s was ninth of 30 countries labourers and those of the rest of the pOpU- under $250 per capita (in 1 975) for which data lation. Fortunately, there is some hope for were available.40 This growth was eroded, how- improvement. A large percentage of the Indian ever, by declining terms of trade (with stagnant Tamils are due to return to India under a long- tea prices and increasing import prices) so that standing agreement between the two coun- real national income corrected for terms of tries. With the need to recruit Sinhalese workers trade grew by 3.5% rather than by 4.4%. to the estates, it is becoming increasingly necess- The growth record from 1970-1977 was ary for social services and housing to be raised unambiguously poor. GDP grew at only 3%, closer to average Sri Lankan rural levels. Also, with agriculture and industry, the key pro- the nationalization of estates o.ver 50 acres ductive sectors, growing at only 1.9 and 2.3%, (carried out from 1973-1975) is forcing the respectively. Much of the growth in services, government to take a more active role in social recorded at 4%, was simply an unaffordable services on the estates. increase in public-sector employment. A lot of the problem was due to bad luck. The monsoon rains were poor for three consecutive years 2. TRADE-OFFS AND COMPLEMEN- from 1971 to 1973; this was then followed by TARITIES 'WITH GROWTH the surge in international prices of grains and petroleum. There was also the balance-of- Sri Lanka has spent about 10% of GDP on payments problem inherited from the 1960s. social programmes, including the food ration But the bad luck was exacerbated by poor programme, during the past 15 yr. This high growth management. Confrontation with the expenditure level for social programmes has private sector discouraged private-sector invest- been compared to other countries' high expen- ment, but the Government did not have the ditures for defence (about one-half of 1% of financial resources, manpower or policies GDP in Sri Lanka); both have a high oppor- required to keep growth up through public- tunity cost but the social expenditures seem sector investment. Much of what was invested less wasteful. In 1964 and 1965, total social by the public sector was in highly inefficient programmes (including the food subsidy pro- industrial projects. Nationalization of large tea gramme) averaged a very high 54% of current estates, an understandable step, was undertaken expenditures and 11.3% of GDP. Since that in a way that caused the estates to be run down time, education and health expenditures have before nationalization and to be inadequately been declining fairly steadily as a percentage of managed thereafter. The padding of the public total current expenditures or GDP. The decline payroll, combined with the sharp rise in the was from 32% of current expenditures in 1965 price of food imports required for the ration to 23%S in 1976. The case of the food subsidy is and subsidy programmes, made both the fiscal more complex. It too has been declining relative and balance-of-paymnents situations increasingly to total current expenditures and GDP, and in untenable. Unfortunately, the political situ- physical quantities per capita as well. But this ation, with the 1 971 insurrection and its after- decline has been largely offset by the continu- math, and the policy differences within the ing implicit depreciation of the rupee, which Government coalition, were not conducive to means that rupee costs of food imports in firm and consistent economic magement. As a recent years are substantially understated. result of the fiscal deterioration, real per capita The growth record of the 1960s was mixed. expenditures on health and education declined, Major policy weaknesses included insufficient as did food distribution per capita and nutrition- incentive for replanting of tea to keep up al standards. growth in production; an industry policy that The relatively better record on both growth emphasized relatively inefficient import substi- and social programmes in the 1960s than in the tution rather than exports, and inadequate 1970s suggests important complementarities, adjustments in exchange rates to reflect the real as well as trade-offs, between the two. In the depreciation of the rupee. On the other hand, 1960s, fairly good growth allowed improve- growth in industry, at 8%, was good, even if ments in social programmes although their primarily in import substitution, and growth in budget share was declining. In the 1970s, it rice production, at almost 6%, was outstanding. would be more correct to say that poor growth Inflation, growth in wages and growth in was the prime cause of the deterioration in government employment were held to modest social programmes than that social programmes levels. Overall, GDP growth was 4.4%, or 2.1% were the prime cause of the deterioration in 246 WORLD DEVELOPMENT growth. Thus, while the experience of many though, between sufficiently narrow cover- countries has shown that growth per se is not a age to be affordable and - sufficiently broad sufficient condition for rapid progress in meet- coverage to get necessary political support, can ing basic needs, the Sri Lankan experience make it hard to exercise needed economies; sharply illustrates the commonsense point that without political constituencies, there is little without adequate growth countries will not hope for basic-needs or other equity-oriented have enough financial resources to maintain, let programmes. alone expand, basic-needs programmes. Third, if the social programmes cause cur- There are also complementarities running in rent expenditures, or, in the case of food, im- the other direction - from basic needs to ports to get out of hand, there are likely to be growth. The previous section referred to the additional hidden costs - in the form of an effect of education (literacy) on rice pro- adverse effect on policies. In Sri Lanka, the duction. With less progress in health and edu- heavy social expenditures have had a net nega- cation, Sri Lanka would have worse current tive impact on pricing and other policies. To prospects for growth in per capita income: some extent, they have encouraged excessive population growth would be higher and the emphasis on short-term budget and foreign labour force less educated and less healthy. exchange problems, at the expense of longer- The Sri Lankan experience also indicates a run problems, and have fostered a 'welfare number of other commonsense points relevant mentality' which has sapped individual or group to the costs of social programmes. First, there is self-help efforts (e.g. see the subsequent dis- a need to target basic-needs programmes on cussion of the role of government programmes those who need them in order to make the in contributing to unemployment). programmes more affordable and to protect the On the other hand, it is easy to slip into the basic needs of the poor during bad years. If the trap of blaming all of Sri Lanka's policy prob- ration programme had been restricted to the lems on the social programmes. For example, it poorer half of the population in the mid-1960s is not clear that Sri Lanka has had worse-than- rather than in 1978, there would have been an average pricing policies for a low-income annual average savings of about 2% of GDP. country. As has been noted, its farm price poli- Similarly, if education expenditures had put cies have been much better than average, and less emphasis on the secondary and higher levels because of the 'foreign exchange entitlement and if government had not hired such an excess- certificates', its foreign exchange pricing policy ive number of teachers in the 1970s that the has been substantially better than the official student-teacher ratio declined from 28 in 1970 exchange rate would suggest. Similarly, the to 22 in 1 977, then there would have been welfare-mentality argument, which has been ample funding to have prevented the deterio- hotly debated at least since the time of the ration of the health system that occurred in the introduction of the 'poor-law' reforms in 1970s. With these changes in the ration pro- England in the early 19th century, can also be gramme, education and health, there would carried too far. It hardly seems consistent, for have been major savings with no sacrifice - example, with the nearly 6% growth rate of perhaps an improvement - in indicators of rice production in the 1960s. basic needs. An extreme case of blaming all Sri Lanka's Second, -the longer a programme has been in policy problems on the social programmes is operation, the more it builds up a political the argument that, with lower social expendi- constituency and the more difficult it is to cut tures, Sri Lanka could have grown as fast as back its benefits or its coverage. The record of East Asian countries like Malaysia or Korea. It Sri Lanka's ration programme shows literally seems inconsistent, though, to argue that ex- dozens of changes in the ration entitlement as penditures relevant to human-capital develop- governments sought to save funds by reducing ment should have been lower, because they the ration; then, particularly about the time of were a drain on growth, and at the same time to elections, the ration would again be increased. argue that Sri Lanka's growth should be judged The Sri Lankan programme had been in exist- by the standards of countries whose rapid ence for about 35 yr before it was targeted on growth is widely attributed to their human lower-income groups. The point here for other capital. In addition, to imply that all that stood countries is that since it is politically easier to between Sri Lanka's and Korea's growth records expand coverage or benefits than to contract was Sri Lanka's social expenditures would be to them, emphasis on low costs and targeting attribute to them all the economic, cultural and should be built in at the initial stage of pro- political differences between South Asian coun- gramme design. Finding the right balance tries (among which Sri Lanka's long-term per BASIC NEEDS: THE CASE OF SRI LANKA 247 Table 2. 1953 1963 1969/1970 1973 Gini coefficient 0.46 0.45 0.41 0.35 Income share of the poorest 20% (%) 5.2 4.4 5.5 7.2 capita growth rate looks fairly good) and the complishments could have been achieved at less fast growing export-oriented East Asian coun- cost, and growth could have been much better. tries. With lower social expenditures, would Sri One could, thus, hardly recommend Sri Lanka Lanka have avoided labour strife? Would it have as a model for countries interested in basic followed an export-oriented rather than import- needs to emulate. At the same time, the facts substitution industrial strategy? Would the do not seem to support either those who would SLFP have had a strongly positive attitude condermn Sri Lanka's development record as a toward private-sector industry? At some point, dismal failure or those who see the social pro- the argument ceases to have much to do with grammes as all that stood in the way of excep- social expenditures per se and becomes, in tionally highlt- effect, 'if Sri Lanka were not Sri Lanka'. If social expenditures were substantially lower and if, as is likely, political constraints 3. INCOME DISTRIBUTION had prevented a sharp redistribution of social programmes from the population as a whole to The household surveys of 1953, 1963, 1969/ the poor, caloric intake, and health and edu- 1970 and 1973 show a continuing improvement cation services received by the poor would have in the distribution of incomes, as indicated by been less; and there would have been corre- Table 2.41 sponding deterioration in social indicators. It is interesting to note that in 1963, when There are too many unquantified relationships the Gini coefficient was at the not very low and too many assumptions required to enable a level of 0.45, Sri Lanka's social indicators useful quantitative comparison of the growth (Table 1) were already exceptionally favour- benefits and social costs from substantial cut- able for its income level. This illustrates that backs in social programmes. There remains equality of income distribution and satisfac- ample room for disagreement in judgements on tion of basic needs are quite different things, Sri Lanka's development record as a whole - although improving income distribution can be with its increasingly unaffordable social expen- an important means of meeting basic needs.42 ditures, its variations in economic management The record of improvement in Sri Lanka's and growth rates, and its unusual social ac- income distribution seems extraordinary. A complishments. There is also room for disagree- study carried out at the World Bank indicates ment on the cross-country standards against that Sri Lanka was the only country of the 35 which Sri Lanka should be judged. Should one countries for which data were available where say, so to speak, that the bottle is half empty or the marginal share accruing to the bottom 60% that it is half full? But clearly, the social ac- of the population (between 1963 and 1973 in Table 3. Average consumption expenditure by quintiles, 1963 and 1973 (rupees per 2 months per spending unit) 1973 Percentage of 1963 (at 1963 prices) Ratio of cols. spending units (1) (2) (2): (1) 0-20 191.3 163.8 0.856 20-40 248.7 276.4 1.075 40-60 334.7 337.8 1.009 60-80 430.4 440.2 1.023 80-100 717.3 829.2 1.156 Average 382.6 409.5 1.070 Source: E. L. H. Lee, op. cit., derived from the 1963 and 1973 Surveys of Consumer Finances. 248 WORLD DEVELOPMENT the case of Sri Lanka) was higher than 60%.43 4 However, some recent analysis has questioned 400 these findings.44 The major argument is that - Incidence of services the distribution of real consumption (in con- 3QA and subsidies stant prices) worsened between 1963 and 1973, -- Incidence of services and the real consumption of the bottom quin- E and subsidies, 0 tile declined (see Table 3). _ 20.0 net of taxes While one might have thought that a higher - share accruing to the poor of a higher GDP (in 4, constant prices) must mean higher real con- a sumption for the poor, this is not necessarily , the case. What happened essentially is that in OD 0o - (Neutral impact) 1973 the cost of living of lower-income deciles CL rose by more than that of the population as a whole, primarily because the relative price of j 10.0 foodgrains had increased. Thus incomes of the m poor (except paddy farmers who produced -200 - more than their own requirements) rose more slowly than the prices of wheat and rice. The contradiction between trends in the dis- -30.0 - tribution of income (at current prices) and in _____________ IIIIIIIIIIIIIIIIII the distribution of consumption (at constant 0 400 a00 1200 1600 2000 prices) is due not only to this shift in the Monthly income of 'spending units, Rs relative price of foodgrains. It also reflects a likely understatement of the degree of income Figure 2. Incidence of services and subsidies as a per- inequalities. Comparison of expenditure (con- centage ofincome, 1973. sumption) and income data for upper-income Source: P. J. Alailima, 'Fiscal incidence in Sri Lanka', groups suggests an increase in 1973 (over 1 963) op. cit. in the degree to which their incomes were understated; this is not surprising, given the ible: relative stability in formal-sector wages much greater emphasis in 1973 on redistri- (i.e. in the middle to upper-middle part of the bution as a government policy. 5 income distribution) at a time when rural in- In a narrow sense-i e. the comparison of comes of small farmers were increasing (due to 1963 and 1973- the critcs have a strong case, fairly rapid growth in production of paddy and But although 1973 was chosen as the termial other non-estate crops without a decline in crop year because of the availability of data from the prices). Food balance sheets also indicate an Survey of Consumer Finances, it was not typi- improvement in 3-yr averages of per capita food cal of the late 1960s or early 1970s.46 The sur- consumption. Per capita food consumption was vey was conducted in January/February 1973. 1924cal for 1951-1953, 2040cal for 1961-- Foodgrain availability (based on production, 1963 and 2201 cal for 1971-1973.47 imports and .stock charges for 1972) was at its In addition, Sri Lanka's extensive service and lowest annual level since 1966. In addition, the subsidy programmes had a significant equalizing 1973 Survey was conducted during a seasonal influence on' incomes. A comprehensive recent low point of supply, while that of 1963 was study48 based on critical analysis of relevant roughly at the time of a major festival. survey data concludes that the distribution of The 1969/i970 Socio-Economic Survey, on government subsidies and services was substan- the other hand: came in an unusually good year. tially more equal than that of income, even Nevertheless, the distribution of income in before taking account of the effect of taxes 1970 was probably about typical for the good (see Figure 2).49 Taking the distribution (inci- growth years of the end of the 1960s. Some dence) of all taxes, subsidies and services as a relatively minor questions havi,- also been raised whole, there was a reduction ir income in- about the data in the 1969/1970 Socio-Econ- equality of about one-fifth.50 In effect, social omic Survey, particularly its exclusion of programmes appear to have substituted to some single-person 'households' (only about 5% of extent for the more radical redistribution of the total); but these questions are not such as assets and incomes which some have hypoth- to discredit the improvement in distribution of esized to be essential to meet basic needs in income that it reports. The major economic low-income countries. cause of the improvement appears quite plaus- There is no sample survey data available for BASIC NEEDS: THE CASE OF SRI LANKA 249 the distribution of incomes since 1973. The about 1 million people, in 1977.52 The unem- increase in the death-rate in 1974 and 1975 ployed represent not just underutilized re- indicates, to say the least, that the poor did not sources, but a threat to Sri Lanka's political fare well particularly on the estates, where stability. They do not, however, represent a the increases in mortality was highest. However, basic-needs problem, per se. They are mostly in response to the inflation-induced deteric- young (90% under 30 in 1975),53 generally ration in the living standards of the poor, and (75%) with at least some post-primary edu- resultant political and union pressures (includ- cation. Eighty-five per cent are supported by ing international attention to the plight of the their parents and three-fourths are looking for tea workers), there were general increases in their first jobs.54 In contrast, the problem of wages in 1974 and 1975. The decline in the poverty is severest among those who are em- relative price foodgrains since 1975, combined ployed but have low incomes. They tend to be with the increased demand for labour on the landless labourers, marginal farmers, estate tea estates, and the (very expensive and overly labourers, or unskilled workers in the informal broad) subsidy on wheat flour, have improved sector. Their households have a higher-than- the position of the poor. At the same time, the average proportion of dependants, few assets combination of slow growth and nationalization and less-than-average access to education or probably worsened the relative position of the even the rice ration (because they cannot afford rich, including particularly earnings from divi- to buy their full entitlement).55 They need not dends and rents.51 The net effect on income more jobs, but more productive, better paid, distribution was probably a worsening from jobs. Thus, the Sri Lankan situation indicates 1973 to 1975 and then an improvement. the danger of the recently popular approach to In sum, the distribution of income and con- development of treating 'employment' - sumption appears to have improved during the usually with primary focus on minimizing un- fairly rapid growth period of the 1960s, even employment - as the central development before taking account of the incidence of taxes, objective which unifies growth and equity ob- subsidies and services, or more importantly, of jectives. the real improvements in the distribution of Although unemployment in Sri Lanka is welfare indicated by the data on social indi- primarily not a basic-needs problem, there are cators. The improvement (or, at worst, stab- some connections between the factors that led ility) in income distribution during this period to the meeting of basic needs and the factors is iinportant, since it is often thought that Sri that have led to high unemployment,56 which Lanka achieved its improvement in income may have implications for other countries. distribution by sacrificing growth for equity. While the drive for education leads to a series Rather, the period of the severest redistribution of development benefits (discussed previously), measures, the first half of the 1970s, seems to it also tends to lead to a mismatch between the have been one of a decline in real incomes of aspirations of new entrants to the labour force the poor, aind of deterioration in 'basic-needs' and the requirement of the labour market programmes. Thus, although there have been for people to fill relatively unskilled, lQw- trade-offs between social programmes and productivity jobs.57 Hence the well-chosen growth, there have also been complementarities title of the innovative and comprehensive 1971 of growth in ouput with the absolute, and to ILO Employment Report on Ceylon - 'Match- some extent relative, incomes of the poor and ing employment opportunities and expec- with the distribution of welfare. This occurred, tations'. It is easy to blame this mismatch on though, not because growth always has this 'inappropriate' curricula and teaching methods, effect, but because of the structure of growth and they have undoubtedly contributed to the in Sri Lanka, which involved extensive social problem. But Sri Lanka's curricula and teaching programmes and substantial increases in the methods have been no more 'inappropriate' incomes of small farmers. than those of many other countries with much lower unemployment. More important than curricula, though, in Sri Lanka is that a very 4. BASIC NEEDS AND UNEMPLOYMENT poor economy simply does not require 75% of new entrants to the labour force to have more Unemployment is a very serious problem in than a basic primary education. Sri Lanka. Sample surveys have shown a growth This mismatch need not automatically lead in the unemployment rate for at least 15 yr, to unusually high open unemployment, Rather, reaching about 20%G of the labour force, or one might think that, over time, the excess 250 WORLD DEVELOPMENT supply of relatively educated people looking for can now live with their families and take ad- white-collar or skilled jobs in government and vantage of lower rural food prices. The ability in the formal private sector would drive down to live in a rural area while remaining un- the wage differential attributable to education employe d may thus cause an increase in unem- and drive up the minimum education stan- ploymznt.62 dards for any given job; after a while, unsuc- Furthermore, the spread of female edu- cessful job-seekers would give up and take cation is likely to mean that women will make whatever jobs were available. Such adjustment up an increasing proportion of the unemployed, mechanisms have operated to some extent in partly because there may be more social or Sri Lanka.58 Nevertheless, for a variety of other restrictions on jobs open to them and economic and non-economic reasons, the wage partly because the opportunity cost of waiting differential in Sri Lanka, as in many other is often lower than for men. This would be countries, has not been driven down to 'market- particularly the case for women whose parents clearing' levels.59 In addition, public- or formal couldI not otherwise force them to work, or private-sector jobs tend to be more secure and would set limits as to what kinds of jobs, geo- to offer much better career prospects; such jobs graphic areas or residential arrangements were are worth waiting for. In other words, relatively acceptable. For someone who has the option of educated young people in Sri Lanka have a high withdrawing from the labour force entirely, the 'reservation wage'. They tend to choose to costs of remaining unemployed while waiting remain ('voluntarily') unemployed in large for a good job are obviously reduced. In 1975, part because, from their point of view, it is the unemployment rate for women was 38% financially profitable in the long run to do so; while for men it was 15%. unfortunately, from the point of view of the In addition, the Sri Lankan political situ- society at large, the high unemployment that atiori increases the returns to waiting for a good results is the country's major economic and job. The insurrection of 1971 and the fact that political problem. At any given time, it may the 20% of the labour force who are unem- (rightly) appear rational to an individual to ployed are relations of a very large percentage wait, even if the chances of getting a good job of the total population indicate that the unem- are low and the waiting time is long (3 yr or ployed and their families can put considerable more for half of those unemployed in Sri Lanka pressure on the government to 'solve' the unem- in 1973). If, for example, a government white- ployment problem by creating permanent jobs collar job offers lifetime income, security, for the unemployed. Successive governments status and career prospects worth together five have tried to do so. Since such programmes times those in the agriculture or informal sec- tend to be restricted to those who are 'unem- tor, then it may be considered worthwhile to ployed', rather than those who are overedu- wait for that job, even if one's chances are only cated for their current jobs, there is a substan- one in three or four.60 In such, rather typical, tial incentive to remain unemployed, rather circumstances for a developing country, it is than to accept an inferior job while looking for likely that, the greater the mismatch between a better one. There are also other reasons to job opportunities and the qualifications and remain unemployed: more time to look for a expectations of new entrants to the labour job; loss of self-esteem from accepting a low- force, the higher will be the open unemploy- status job; and an alleged preference of em- ment.61 ployers for those who are unemployed over In addition, for any given mismatch, the those who have accepted a low-status job. lower the costs of waiting and the higher the In spite of the economic rationality of wait- benefits from waiting, the higher the unem- ing for several years for a good job, it would be ployment. In Sri Lanka, the welfare pro- wrong to imply that all voluntary employment grammes such as the food ration or free health among educated youth is 'rational', or that care reduce the costs of waiting and, thus, caste and class behaviour norms do not also play increase the time that parents or other rela- a part. While those from families with higher tives are willing to support a job-seeker. Also, socio-economic status undoubtedly are over- the spread of welfare programmes into rural represented at the university and 'A' levels, areas makes it more attractive, depending on Sri Lankan parents, like those elsewhere, have the nature of the job search progress, for the seen education as the key to social and econ- unemployed to live in rural areas. If this is omic mobility. Parents work hard and forego feasible (and 70% of the unemployed in Sri labour inputs from their children so that the Lanka live in rural areas), then the costs of children can have access to the better life- waiting are reduced further, since those waiting chances that education provides; many might Ur BASIC NEEDS: THE CASE OF SRI LANKA 251 well be willing to go on supporting an educated the other hand, governments should take steps unemployed child long after it becomes appar- in both employment and basic-needs policies ent that the child will eventually have to settle to minimize the extent to which welfare for whatever low-status, low-pay work is avail- programmes subsidize or otherwise encourage able. It would also be wrong to imply that the unemployment. unemployed do no productive work at all (although this remains an area where insuf- ficient empirical evidence is available). Unem- 5. CONCLUSION ployed youth being supported by their parents do contribute to their families' livelihoods and often, particularly in peak agricultural seasons, Sri Lanka has had serious development prob- will do work for others. The number of hours lems, including high unemployment, preemp- spent with no economically productive output tion of a high percentage of the budget and of would depend, among other things, on the imports for social expenditures, and an ex- parents' tolerance and their feelings about the tremely adverse turn in external terms of trade. social status they hope that their child's edu- These problems were exacerbated by some cation will bring. policy weaknesses in the 1960s and more wide- Finally, it 'ould be wrong to imply that spread policy weaknesses in the 1970s. But con- anything that governments do to reduce the trary to what is commonly assumed, it was not unemployment problem will only increase the primarily excessive expenditure on social pro- incentive to remain unemployed and, thus, grammes that caused growth to deteriorate in make things worse. There are constructive steps the first half of the 1970s, but the poor growth that can be taken, in Sri Lanka as elsewhere, policies and terms of trade. The growing un- that have fewer 'side effects' than the steps affordability of the food ration programme discussed above. These would include, for indicates that there are indeed difficult trade- example, elimination of pricing distortions that offs between basic needs and growth and that subsidize capital in relation to labour; changes it is important that programmes be lean and in curricula and counselling aimed at reducing targeted to enable them to remain viable. On student and parental expectations; fixed-term the other hand, Sri Lanka's relatively good employment on productive public-sector proj- record on growth and social programmes in the ects, but with an explicit condition that no 1960s (and the apparent deterioration in the permanent government employment will fol- distribution of real consumption and in social low; and continuing reductions in the overall programmes in the early 1970s) confirms the income and security advantage of a white- importance of complementarities as well as collar, public-sector job. trade-offs. In sum, while the elimination of unemploy- Sri Lanka's record on social indicators ment is primarily not a problem of basic needs, suggests that expenditure over a number of some basic-needs programmes - food subsidies years at a high percentage of the total govern- or the spread of social programmes to rural ment budget can lead to striking progress in areas - have the inadvertent effect of lower- the underlying social objectives of the econ- ing the cost of remaining unemployed while omic development process. Even the highly searching for a good job. Other programmes not controversial food ration and subsidy pro- directly a part of a basic-needs approach but grammes seem to have had nutrition, mortality derived from common political and economic and income distribution benefits. The widely sources - rapid extension of post-primary edu- accepted view that the social benefits were cation for both boys and girls, or well-intended greatly outweighed by the economic costs now government steps to ease the plight of the un- appear open to question, particularly since Sri employed - Ic-wer costs of waiting further and Lanka has initiated steps to restore growth, and increase expectations of benefits. The greater to cut back and target subsidies. Had the econ- affordability of unemployment in countries omic deterioration of the 1970s continued for where basic needs have been largely met means a good deal longer, however, there might be no that its level is likely to remain higher than in suich room for question. And there is no room countries where this is not the case.63 One for question but that Sri Lanka could have wouid hardly wish to argue that this is not a achieved even more in meeting basic needs if it price worth paying - that, in effect, the risk of had had better growth management. death from diseases related to malnutrition or To put the costs of Sri Lanka's social expen- health practices should be higher so the cost of ditures in perspective, in the first half ol the remaining unemployed will also be higher. On 1970s they were about $1 5 per capita/yr, split 252 WORLD DEVELOPMENT roughly equally between the food subsidy and many ways a special case in large part depen- social services. By the standards of domestic dent on its own historical circumstances. Be- poverty programmes in aid-donor countries, or cause of this, and because of the serious de- even by the standards of donor-funded develop- ficiencies in its past growth efforts, and its ment projects in poor countries, $1 5 per capita essentially untargeted social programmes, Sri hardly sounds high in relation to Sri Lanka's Lanka cannot be held out as a basic-needs social accomplishments. Even that $15 over- 'model' for others. Yet, it is one of the very states costs of direct basic-needs programmes, few non-socialist countries to have been em- since it includes both benefits to those whose barked for many years on a basic-needs ap- basic needs have already been substantially met proach. Its experience appears to shed light on and social programmes with objectives other a number of issues and hypotheses relevant to than basic needs (e.g. higher education). basic needs, including: trade-offs and comp- For the future, acceleration of growth seems lementarities between basic needs and growth; much more feasible in Sri Lanka than in most the tension between the lower economic costs other slow growing countries. Sri Lanka's po- of targeted programmes and the greater politi- tential for per capita income growth has been cal feasibility of less targeted program1mes; rela- enhanced by its relatively highly developed tive priorities among basic-needs programmes; human resources, its low rate of population and the complex relationship of basic needs to growth and its comparatively low wage rates. income distribution and unemployment. Over- Whether it will achieve accelerated and sus- all, since Sri Lanka has accomplished more in tained growth64 depends heavily on the quality meeting basic needs than most countries with and stability of growth policies and, in indus- three or four times its per capita income and try, on labour discipline. Both are primarily appears to have average or better growth pros- political, rather than economic, questions. pects, its development experience can no longer Like most other countries, Sri Lanka is in be written off as a failure. NOTES 1. Thus, there is no single basic-needs strategy; rather are computed from 2-yr average base and terminal basic-needs approaches to development are compatible periods. with a broad range of market-oriented or socialist economic and political systems. The basic-needs 3. Sri Lanka's indicators were all significantly better approach by this definition comprehends the recent than expected (at p < 0.05). See Appendix Table Al trend in development economics toward more empha- for the relevant regression equations. A good deal of sis on issues of poverty and income distribution. It controversy has surrounded the 'physical quality of differs primarily in the explicit emphasis it puts on life index' (PQLI), on which Sri Lanka scored ex- attainment of basic needs of the poor, with improve- tremely high, because of the arbitrariness of its choice ments in their absolute and relative incomes being of indicators and weighting among indicators. [The viewed as important nmeans of doing so. It does not 'PQLI' is explained in J. W. Sewell et al., The United matter, for purposes of this paper, whether one con- States and World Development: Agenda 1977 (New siders the means required to achieve good health - York: Praeger, 1977).] However, if one accepts the such as adequate nutrition, potable water, housing four indicators above (either as direct measures or as or clothing - as basic needs themselves. See Employ- proxies for other measures), then Sri Lanka would be ment Growth and Basic Needs: A One World Problem the highest scoring country in relation to its income (Geneva: ILO, 1976). Also see M. ul Haq, 'Progress no matter what weighting was used. report on basic needs' (World Bank, August 1977), for further development of some of the concepts and 4. D. Morawetz, 'Economic lessons from some small unresolved issues of the basic-needs approach. socialist developing countries', Mimeo (World Bank, 1978). His argument of lower than expected improve- ment in social indicators rests primarily on a regression 2. Note on data in this paper: Years for which data equation with an extraordinarily low R2 of 0.01. This are presented are largely determined by their avail- equation does not include a term which would allow ability from population censuses or surveys. Data on for the increasing difficulty of further progress as a demographic factors and employment for 1 946, 1953, country comes closer to maximum attainable levels. 1963 and 1971 are from the censuses. Data for 1970 Morawetz dismisses this problem by noting that (1969/1970) are from the Socio-Economic Survey. Korea, which also had good social indicators in the Employment data for 1973 and 1975 are from the base year, made faster progress than Sri Lanka. Unfor- 1973 Labour Force Participation Survey and the 1975 tunately, this use of the best performing counrtries (the Land and Labour Utilization Survey. Income distri- fast growing East Asian countries), rather than a more bution data for 1963 and 1973 are from the Surveys typical group of developing countries, as the standard of Consumer Finances. Data on growth in output are for comparison, detracts from Morawetz's otherwise from the Central Bank of Ceylon; output growth rates quite interesting analysis. BASIC NEEDS: THE CASE OF SRI LANKA 253 5. See E. J. Wijemanne, 'Educational reforms in Sri Economic Development and Cultural Change, Vol. 25, Lanka' (Marga Institute, 1977). No. 2 (January 1977), for a discussion of the con- ditions under which large-scale food imports will 6. For comparative data on expenditures and enrol- lead to a disincentive effect on domestic agricultural ment rates, see M. Zymelman, 'Patterns of education production. expenditures', World Bank Staff Working Paper No. 246 (November 1976). 14. In addition to the benefits discussed above, the ration also helped to keep wages in Sri Lanka rela- 7. A major reform intended largely to address these tively low and stable (until the early 1970s). The problems was undertaken in 1972, but it ran into im- ration prograsnme originated on the estates and was plementation problems and present resistance. As a initially conceived as a way to ensure wage stability result of this, and of the politicization by the previous rather than as an income supplement or nutrition government of the social-science curriculum, the new programme. government abandoned many of the changes made. However, it is premature to determine how education 15. See, for example, C. L. G. Bell, 'The political policy will evolve over the next several years. framework', in H. Chenery, M. S. Ahluwalia, 3. H. Duloy and R. Jolly, Redistribution with Growth (Lon- don: Oxford University Press, 1974). 8. On the importance of malaria control in relation to other causes of the decline in Sri Lanka's mortality 16. Also, because of farm price support programmes rate, see R. H. Gray, 'The decline of mortality in Cey- in food-surplus donor countries, food aid from these lon and the demographic effects of malaria control', countries seems to be to a substantial extent ad- Population Studies, Vol. 28, No. 2 (July 1974). ditional - i.e. to result in a higher net total than if food aid were not a part of the aid package. 9. Nevertheless, the cancellation from 1975 to 1977 17. See D. R. Gwatkin, 'Nutritional planning and of training for 'assistant medical practitioners' is one physical well-being in Kerala and Sri Lanka', Paper among several indications of a remaining imbalance presented to an AAAS symposium on 'Nutrition and favouring intensive, physician-based, curative health Agriculture: Strategies for Latin America' (February care in Sri Lanka. This imbalance is not surprising, 1978). since the extensive system developed from a conven- tional intensive system. 18. To the extent that recipients of rationed rice buy rice on the open market to supplement their ration, 10. The criteria determining how choices are made the marginal price they pay for wheat is the market between Western and ayurvedic (and other indigenous) price. In this situation, economic theory suggests that treatment in rural Sri Lanka are discussed in C. F. the ration would increase calorie and protein con- Arndt, 'Health seeking behavior in Wegeriya', Mimeo sumption by only the same amount as would an in- (George Washington University, 1977). come supplement (of an amount equal to the rice subsidy). On the other hand, since the wheat subsidy 11. This is not meant to imply that ayurvedic medi- has no quantity limit, the marginal price is still the cine is in most circumstances as effective as Western rationed price, so there would be a larger increase in medicine, or that traditional medicine in other coun- calorie and protein consumption than with an equiv- tries should be presumed as effective as that of Sri alent income transfer. (This result would most likely Lanka. However, other countries have also made ex- hold in spite of partially compensating cuts in con- tensive use of traditional health systems, most notably sumption of other foods, since wheat is a relatively China, as well as India (where Sri Lanka's ayurvedic low-cost source of calories and protein.) system originated). 19. Also, the ration for sugar was cut from 2 to 1 lb/ 12. ' . . . the (government procurement) paddy price month in 1973, with very little available off-ration, has been above the c.i.f. (paddy) equivalent price and that at 3-5 times the 1973 price. The reduction in since the early 1950s'. P. Richards and E. Stoutjesdijk, availability was due to the rapid rise in the import Agriculture in Ceylon until 1975 (Paris: OECD Devel- price. With these demand pressures, the price of dom- opment Centre, 1969), p. 54. On comparative rice/ estic 'jaggery' sugar undoubtedly multiplied as well. fertilizer ratios in 1970, see P. Timmer and W. P. Fal- Sugar had been a major source of calories. con, 'The political economy of rice production and trade in Asia', in L. Reynolds (ed.), Agriculture in 20. The difference between the observed and expec- Development Theory (Yale University Press, 1975), ted death-rates for 1974 was significant at 0.01. See Table 14.2. The heavy input subsidies - primarily on Appendix Table A2. While there was a sharp increase irrigation, fertilizer and credit - undoubtedly had a in reported cholera deaths in 1974 (333 vs 13 in 1973), positive impact on output; but they also had a high this would have caused an increase in the death-rate of fiscal cost and encouraged inefficiencies in the allo- only 0.025/1000. The regression analysis is for the cation of scarce inputs. years 1963-1976. Preliminary data from 1977 con- firm the hypothesis, since the per capita food supply 13. See P. Isenman and H. W. Singer, 'Food aid: was the highest in recent years and the death-rate fell disincentive effects and their policy implications', to 7.5/1000. 254 WORLD DEVELOPMENT 21. The same general points also apply to a lesser 30. The high level of unemployment among young extent to 1975, when food supplies and ration distri- people may also contribute to the increase in the aver- bution were below average and the death-rate, at 8.5/ age female age of marriage, due to the delay in a 1000, was still above the trend line. couple's ability to support themselves; however, there was no comparable increase in the male age at mar- 22. See M. Selowsky and S. Reutlinger, 'Malnutrition riage. Another suggested cause for the increase in and poverty', World Bank Occasional Paper, No. 23 female age at marriage and decline in the percentage (1976). While there is some dispute about aspects of of women married is a demographic shift in which their methodology, no serious challenges have been there are relatively fewer men of appropriate age for raised to their conclusions. marriage (which is generally about 5 yr older than for women). See M. B. Duza and C. S. Baldwin, Nuptiality 23. The regression results and the Sri Lankan experi- in Population Policy (New York: Population Council, ence suggest that education may not have to be 'rel- 1977), Chapter 4. evant' to development in order to have a positive impact on it. Primary and secondary education which, 31. See [LO (1976), op. cit., and Haq, op. cit. as in Sri Lanka, is not applied to develcipment prob- lems or to the needs of the majority terminating at 32. 'Sri Lanka Water Supply and Sanitation Sector these levels is undoubtedly less desirable than that Study' (World Health Organization, 1977). which is. But, it appears that in the developing coun- tries, as previously (and to some extent currently) in 33. Arndt, op. cit., indicates that a quite rational the developed countries, students manage to acquire approach to when and where to seek treatment often useful learning skills in spite of irrelevant curricula; coexists with traditional beliefs about the role of the same might even be said for rote learning which, spiritualism as the cause and/or cure for some diseases. for all its inadequacies, does teach some skills in re- (The same point could be made about a large percen- membering and following complicated instructions. tage of the population in developed countries as well.) 24. The equation ascribes to education the effects of 34. This is not to say that potable water is not im- other variables, such as health practices, affected by portant, but it does have some potential implications or otherwise correlated with education. Separating for priorities among basic needs. Similarly, it is not to out the ways in which education interacts with these say that literacy is a necessary condition for acquiring variables is well beyond the scope of this paper. knowledge about water pollution, or that literacy is a sufficient condition for good hygiene. 25. See M. E. Lockheed, D. T. Jamison and L. S. Lau, 'Farmer education and farm literacy: a review of the 35. I would like to thank, without implicating, Nor- literature', Mimeo (Educational Testing Service, April man Uphoff for a number of useful comments on 1978). The reasons suggested for the positive effect of earlier drafts of this discussion of participation. education on adoption of improved technology are not meant to imply that the farmers who are illiterate 36. This view is reported, but not necessarily en- are not 'rational' in their decision-making or not re- dorsed, in Sri Lanka: An Experience in a Need- sponsive to opportunities to increase their income. Oriented Development - Achievements, Conditions, Limits, Altematives (International Foundation for 26. The regression analysis of fertility rates reported Development Alternatives, 1978). in Appendix Table Al is a slight modification of that of D. Morawetz, 'Basic needs policies and population 37. Sri Lanka also has a number of private voluntary growth', World Development, Vol. 6, Nos. 11/12 organizations which address poverty and social ills, (November/December 1978), pp. 1251-1259. one of which, the Sarvodaya movement, has become known internationally as a symbol of participatory 27. See R. Cassen, 'Population and development: a non-governmental efforts to alleviate poverty. The survey', World Development, Vol. 4, Nos. 10/11 work of the voluntary organizations is not discussed (November 1976), for a survey of the literature on here because of their relatively limited impact on the the relationship of development to fertility. overall extent to which basic needs have been met in Sri Lanka. For further information on the Sarvodaya 28. Cassen, op. cit. movement, see, e.g., N. Colletta, 'Building the socio- psychological infrastructure for popular participation 29. Unfortunately, the analysis in the 1971 census and village development: the Sarvodaya experience', was not multivariate, and thus did not allow the effect International Development Review (3rd Qu rter, of female education to be separated from other vari- 1979). ables such as income. Data from the 1969/1970 Socio- Economic Survey indicate that female education is 38. The 1969/1970 Expenditure Survey data suggest not acting simply as a proxy for female employment, that calorie consumption on the estates is above aver- since middle-education levels show both low employ- age. However, not much faith can be put in this data. ment and low fertility. The age group of 40-44 was The number of wage earners per household on the used to show that higher levels of female education estates was about double that of the estates, so their reduce completed family size, rather than merely calorie requirements were substantially higher. A delaying births. higher incidence of nutrition-sapping diseases and less BASIC NEEDS: THE CASE OF SRI LANKA 255 access to non-market sources of food not counted in and the problem of understatement of incomes by the surveys also account for some of the apparent upper-iincome groups. higher calorie intake on the estates in 1969/1970 and 1973. The 'Sri Lanka Nutrition Survey' conducted 47. 'Participatory development and dependence' from September 1975 to March 1976 found nutrition (Marga Institute, 1977). status to be substantially lower on the estates than in other rural (or urban) areas (US Centre for Disease 48. P. J. Alailima, op. cit. Control, 1976). This strengthens the argument above on the effects on mortality of the cutbacks in food 49. However, as might be expected , the distribution distribution. It is likely that a nutrition survey in of subsidies and services in total, excluding social- 1969/1970 would also have shown above average service charity programmes, was still somewhat un- malnutrition in spite of the above average calorie equal; some items, such as higher education or the consumption. transport subsidy, were distributed highly unequally. 39. P. J. Alailima, 'Fiscal incidence in Sri Lanka', 50 P. J. Alailima, op. cit, World Employment Program Working Paper No. 69 (Geneva: ILO, 1978). 51. However, some individuals were able to make substantial amounts from windfall gains resulting 40. D. Morawetz, 25 Years of Economic Develop- from government controls and allocations Or from ment, 1950-1975 (World Bank, 1977), p. 81. production of consumer products under complete protection from foreign competition. Others became 41. 1953, 1963 and 1973 data are from the Surveys wealthy fronm the boom in gem exports. of Consumer Finances of those years. Data from them are by 'spending unit', which differs slightly frrm 52. Extrapolated from the Land and Lalt ,ur Utiliz- 'household'. (Data by 'income receiver' are con- ation Survey, 1975. sidered less reliable.) The data for 1969/1970 are from the Socio-Economic Survey of 1969/1970, and 53. Eighty-three per cent under 34 in 1971. The 1971 are by 'household'. data are in general more reliable. Either estimate indi- cates a high concentration of unemployment among 42. Regression analysis of a 47-country sample for the young. which data were available on both income distribution and social indicators shows a simple correlation of the 54. Labour Force Participation Survey, 1973. income share of the poorest 40% with life expectancy to be 0.27 and with adult literacy to be 0.26 (both 55. P. J. Alailima, 'A profile of poverty in Sri Lanka' significant at p = 0.10). (unpublished, 1976). This study was based on data from several surveys. 43. M. S. Ahluwalia, N. G. Carter and H. B. Chenery, 56. Among the reasons advanced for high open un- 'Growth and poverty in developing countries', Journal employment in Sri Lanka are slow growth, rapid of Development Economics (1979), pp. 299-341. growth in the labour force due to a demographic bulge and subsidies to capital. All three are valid con- 44. See E. L. H. Lee, 'Rural poverty in Sri Lanka tributors to unemployment, but the situation in India 1963-1973' in ILO, Poverty and Landlessness in appears to have been as bad or worse on all three, Rural Asia (Geneva: ILO, 1977). D. Morawetz, 'Econ- while open unemployment remained insignificant. omic lessons from some small socialist developing Since Sri Lanka's per capita growth rate has been countries'. above average for countries below $200 per capita, slow growth cannot explain the unusually high un- 45. Both the 1963 and 1973 Surveys report under- employment rate. statement of incomes of the upper-income groups to be a problem; this is an endemic problem in sample 57. There are generally unskilled jobs available in surveys of incomes. Sri Lanka, particularly during peak agricultural periods, in the less populous 'dry zone' and, recently, 46. The problems with the 1973 data illustrate, inter on the estates. alia, the dangers both of excessive reliance on distri- bution of incomes at current prices and of excessive 58. The income differential in Sri Lanka between focus on years for which good data are available, but those with 'O' level qualifications and those with only which, as is often the case, are untypical. This is a primary education has declined from 3 :1 to 2 :1 particular problem for economies dependent on the between the Surveys of Consumer Finances of 1953 monsoon, since over a decade yearly fluctuations in and 1973. This effect of the increased supply of food availabilities can be as large as trend changes, educated people on the wage rate indicates in part and since in these economies the real consumption of that the 'market' is working. In part, it also reflects the poor tends to depend heavily on food avail- the shift in the terms of trade in favour of agriculture, ability. Expenditure data are preferable to income since many farmers have a primary education but few data in handling both this problem of relative prices have reached the 'O' level, 256 WORLD DEVELOPMENT Also, an increasing number of relatively educated 62. This would be quite different from, but not in- young people in Sri Lanka are apparently willing to consistent with the spirit of the Harris-Todaro model, accept relatively-low-status jobs, in which secondary which related the modern/traditional wage differential (or in some cases primary) education does not add to rural-urban migration on grounds that the job significantly to productivity. There are also many search required living in the town. More work needs who are not willing to do so. See 'Pattern of job ex- to be done on the nature of the job search. For what pectations and unemployment among rural youth' kind of jobs do applicants have to queue up almost (Marga Institute, 1977). daily? What kind of jobs tend to be filled by formal On the 'screening process' by which employers use applications or by recommendations from current academic qualifications not directly relevant to the employees, so that prospective candidates could job vacancies to choose among applicants, see C. V. P. remain in the rural areas and make occasional trips Deraniyagala, R. P. Dore and A. W. Little, 'Certifi- into the city? Sri Lanka's relatively small size and its cation and employment in Sri Lanka' (Sussex: Insti- good and highly subsidized transport system may tute of Development Studies, 1977). make it unusually easy for those seeking modern- sector employment to do so while living in a rural 59. Economic reasons would include a premium for area. discipline, reliability and lower turnover; non-econ- omic reasons would include the political power of public-sector and other educated workers. 63. In other words, the basic-needs and related pro- grammes, as well as government help to the unem- 60. Another way of putting this argument is that the ployed, contribute to a divergence between private Sri Lankan case can be explained by a so-called 'seg- and social costs or unemployment that in turn con- mentation' model of the labour market, of which the tributes to an undesirably high unemployment rate. Harris-Todaro model of unemployment in East Africa is the best known example. See A. Berry and R. Sabot, 'Labour market performance in developing countries 64. If its growth is fairly rapid, the Sri Lankan experi- - a survey', World Development, Vol. 6, Nos. 11/12 ence would support the argument made in favour of (1978), pp. 1199-1242, for a broad review of relevant an initial heavy investment in human capital as an issues. important (but neither necessary nor sufficient) cause of long-run rapid growth. It would help to generalize 61. See M. Blaug, Education and the Employment this argument beyond the East Asian and European Problem in Developing Countries (Geneva: ILO, countries (e.g. Japan, Korea and Israel) from which 1973). it derives. BASIC NEEDS: THE CASE OF SRI LANKA 257 APPENDIX Table Al. Regression equations* A. Equations used in Figure 1: (1) LIT = 41.87 + 0.0323Y -3.94 X 10-6 Y2 R2 = 0.54 ('t' values) (9.28) (5.79) (-4.25) S.E. = 17.92 tFE = 18.36 (2) LnLIFEX = 3.197 + 0.132lnY q2 = 0.70 (41.26) (11.71) S.E. = 0.10 FE = 0.10 (3) LnINF = 8.121 - 0.621lnY R2 = 0.72 (22.90) (-12.01) S.E. = 0.43 FE = 0.45 (4) LnFERT = 7.320 - 0.389lnY R = 0.66 (28.96) (-10.56) S.E. = 0.31 FE = 0.32 B. Equations used in Section 1 (e) (5) LnFERT = 7.495 - 0.188lnY -0.025LIFEX R2 = 0.70 (30.94) (-3.18) -3.76 S.E. = 0.29 (6) LnLIFEX = 2.83 + 0.065lnY +0.199lnLIT R2 = 0.88 (45.30) (7.27) (10.69) S.E. = 0.06 (7) LnINFMOR = 8.900 - 0.4541nY -0.4641nLIT R2 = 0.74 (20.70) (-7.41) (-3.63) S.E. = 0.41 Where LIT = adult literacy, LIFEX = life expectancy, INFMOR = infant mortality, FERT = fertility index = (number of births/' of population aged 15-64), Y = per capita income. Number of observations = 59 countries. Source of data: World Tables, World Bank, 1977. Data are for 1975 or the closest available year. Notes: *The specifications used were those that gave the best fits in the range of the sample values, rather than, for example, specifications which provided asymptotes (e.g. 100% literacy) but weaker fits. There is likely to be some simultaneous equations bias in Equations 1, 4 and 5, since literacy and fertility affect per capita income, but the conclusions drawn from these equations are not sensitive to small changes in the coefficients. Sri Ianka is excluded from Equations 1-4, since indicators are to be tested against expected values derived from indi- cators for other countries. t FE = forecast error. I am grateful to T. N. Srinivasan for pointing out to me that to test the significance of the residual of a given observation, the forecast error should be used rather than the standard error to determine 't' values. Using this test, residuals for Sri Lanka in Equations 1-4 were all significant at p < 0.05. A one-tail 't' was used since, in addressing the question of whether Sri Lanka's residuals are better than expected, there is no ambiguity about the pre- dicted signs. 258 WORLD DEVELOPMENT Table A2. Sri Lankan deathl-rates 1. Significance of the increase in the death-rate in 1974 Lndeath-rate 2.143 - 0.012T 2 = 0.51 ('t' values) (106.57) (-3.39) S.E. 0.04 FE = 0.04 where T= time, from 1963 to 1973, with 1963 as 0. The coefficient of 0.012 indicates that the death-rate was declining (with statistically significant consistency) at a rate of 1.2%/yr from 1963 to 1973. The projected value for 1974 would have been 7.5, while the actual value was 8.9. The difference (residual) is significant at p = 0.01. 2. Cause of the increase in the death-rate in 1974 The following equation shows that the open-market price of paddy was, along with the time trend, a significant determinant of the death-rate. Lndeath-rateT = 1.754 - 0.011T + 0.159Ln(PP), R2 = 0.45 (16.47) (-3.09) (3.43) S.E. - 0.04 Where T= time, from 1963 to 1976, with 1963 as 0. PP - open-market wholesale paddy price, deflated by the Central Bank wage rate index for government employees. (A historical series on the price of wheat was not available, but the markets for wheat and rice are sufficiently closely related that the shortage of wheat in 1974 was promptly reflectcd in the price of rice.) THE WORLD BANK Headquarters: 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. European Office: 66, avenue d'Iena 75116 Paris, France Tokyo Office: Kokusai Building, 1-1 Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan The full range of World Bank publications, both free and for sale, is described in the World Bank Catalog of Publications, and of the continuing research program of the World Bank, in World Bank Research Program: Ab- stracts of Current Studies. The most recent edition of each is available with- out charge from: PUBLICATIONS UNIT THE WORLD BANK 1818 H STREET, N.W. WASHINGTON, D.C. 20433 U.S.A.