A study of the poor in Sri Lanka

lbPoo(o THE WORLD BANK F4 Internal Discussion Paper AsIA REGIONAL SERIES Report No. IDP 65 A Study of the Poor in Sri Lanka Cecilia Elena Rouse June 1990 The views presented here are those of the author, and they should not be interpreted as reflecting those of the Wrld Bank. ASIA REGION DISCUSSIO PAPER SERIES 1L itL Aqg Oainator IDP2 The Labor Force Participation of Women in the Republic of Korea: Evolution and Policy Issues C. Grootaert May 1987 F. Iqbal IDP15 The Role of Exchange Rate Policy in Sang-Woo Nam May 1988 D. Leipsiger Four East Asian Countries 78841 IDP28 The Small-Scale Enterprise Credit Program (S.S.E.P.) Under the Second and Third Calcutta Urban Development Projects F. Kahnert March 1988 F. Kahnert (CUDP II and CUDP III) - An Assessment 76376 IDP35 Improving Tax Policy Advicet Lessons and a. FleLaig June 1989 H. Fleisig Unresolved Issues from Asia Experience 76375 EDP36 Direct Taxes and Fiscal Policy Issues: A. Virmani June 1989 H. Fleisis An Illustration for East Asia 76375 IDP37 Commodity Taxation in Selected Countries Z. Shalist June 1989 H. Fleisig In South East and East Asia 76375 IDP38 Tax Analysis in Developing Country R. Muagrave June 1989 H. Fleisig Settings 76375 IDP39 Indonesia: External Shocks, Policy Sadiq Ahmed June 1989 Sadiq Ahmed Response and Adjustment Performance 73723 IDP42 An Anatysis of the Nature of Unemployment W. T. Dickens July 1989 R. Zagha in Sri Lanka and Kevin Lang 80433 IDP44 Assisting Poor Rural Areas Through Friedrich Kahnert August 1989 C. Chamberlin Groundvater Irrigation 81409 IDP51 Educational Development in Asia: A Comparative Study Focussing on Cost Jee-Peng Tan October 1989 Jee-Peng Tan and Financing Issues Alain Mingat 81408 IDPS2 Chinese Reforms, Inflation and the Allocation of Investment in a Socialist Oktay Yenal October 1989 Oktay Yenal Economy 81415/81416 IDP63 Public Policy to Promote Industrialization: The Experience of the East Asian NICs and David Dollar May 1990 David Dollar Lessons for Thailand 80518 IDP65 A Study of the Poor in Sri Lanka Cecilia Rouse June 1990 Yukon Huang 80419 11 Extra copLes may be obtained from the Asia Information Service Center. A STUDY OF TE POOR IN SRI LAU by Cecilia Elea Rouse prepared for The World Bank June 1990 A STUDY OF THE POOR IN SRI IANA Cecilia Elena Rouse Abstract The nature of poverty in any country is as unique as the society in which it is situated. While in many countries concern for the poor arises out of concern for their quality of life, the literacy, infant mortality, and life expectancy rates of Sri Lanka have historically rivaled those of cou.tries with much higher levels of GNP per capita, presumably because of the government's extensive social expenditures. On the other hand, the low level of GNP per capita suggests that there are probably Sri Lankans living in some state of deprivation. The focus of the current study is to characterize the poor in Sri Lanka as well as to analyze the roles played by education and employment in determining poverty. Calorie data from the Labor Force and Socio-Economic Survey of 1985186 conducted by the Department of Census and Statistics are utilized in order to define a calorie poverty line. The intent of the first portion of the paper is to update much of the work conducted by authors such as Sahn and Edirisinghe. Household expenditures, demographics, and other characteristics of the poor are analyzed with particular attention to the differences between the poor and non-poor; many of the previous results are corroborated. The poor have larger family sizes and larger ratios of non-working household members to working household members; there is also continued evidence of a rising food share in household expenditures for the very poor. Finally, the food stamp program still suffers from mis-targeting and leakage and estimates suggest that the government's proposed Jana Saviya Program exceeds the transfers necessary to bring families out of poverty. The paper also explores two main correlates of poverty: education and employment. The poor are generally less educated than the non-poor and we argue that the difference has two sources. Family perceptions of the value of education rather then a lack of resources are important determinants of whether young children attend school. On the other hand, older students may not receive advanced degrees due to the quality of their earlier schooling. There is evidence which suggests that students from poor families would like to pass their Ordinary Level exams but that they do not succeed due to a lack of adequate preparation from their elementary and secondary schooling. Finally, we find that policies aimed at reducing unemployment would also alleviate poverty but only marginally because only a small proportion of the poor are unemployed. The analysis suggests, however, that underemployment may be more of a function of one's sector than of one's poverty status. ACINOWLEDGMNM This paper could not have been produced without the kind assistance of the Department of Census and Statistics, which provided the tapes of the 1985/86 Labour Force and Socio Economic Survey. A STUDY OF THE POOR IN SRI LANKA Table of Contents Page No. INTRODUCTION............................................................. Background...................................................... 2 Review of the Literature........................................ 3 DATA AND METHODOLOGY..................................................... 5 Data............................................................ 5 Methodology..................................................... 5 THE ANALYSIS ............................................................. 9 Households...................................................... 9 Expenditures.................................................... 13 Assets.......................................................... 17 Food Stamps..................................................... 20 Individuals ..................................................... 21 Who are the Poor............................................... 21 The Role of Education........................................... 24 Labor Force Participation, Unemployment, and Underemployment.... 33 Regression Results.............................................. 43 CONCLUSION............................................................... 45 BIBLIOGRAPHY............................................................. 47 APPENDICES............................................................... 49 INTRODUCTION The nature of poverty in any country is as unijue as the society in which it is situated. 'While in many countries concern for the poor arises out of concern for their quality of life, the literacy, infant mortality, and life expectancy rates of Sri Lanka have historically rivaled those of countries with much higher levels of GNP per capita, presumably because of the government's extensive social expenditures. One consequence is that what it means to be poor in Sri Lanka is not easily comparable with other countries. On the other hand, the low level of GNP per cap.ta ($400 in 1987)1 suggests that there are probably Sri Lankans living in some state of deprivation. The focus of the current study is to characterize the poor in Sri Lanka as well as to analyze the roles played by education and employment in determining poverty. Calorie data -from the Labor Force and Socio-Economic Survey of 1985/86 conducted by the Department -.f Census and Statistics are utilized in order to define a calorie poverty line. Household expenditures, demographics, and other characteristics of the poor are analyzed with particular attention to the differences between the poor and non-poor. The paper also explores two main correlates with poverty: education and employment. In a country where 85Z of the population is literate and close to 882 of the population has received some schooling, educational attainment nonetheless remains highly correlated with poverty. Similarly, in a developing country with a large agricultural sector it is difficult to sort-out issues of unemployment and underemployment, and yet one's labor force status is an important determinant of one's poverty status. The paper highlights several aspects of Sri Lankan poverty. First, poverty, as measured by food consumption, appears to have decreased since 1981. This result is consistent with the observed increase in food consumption on the island. However, the majority of Sri Lankan children live in poverty. Similarly, many of the characteristics of the poor which previous authors, such as Sahn and Edirisinghe, have found regarding family size, expenditure patterns and food stamp recipients are supported by this latest data. The poor are generally less educated than the non-poor. We argue that the difference has two sources. For young children, perceptions about the value of education appear more important than a lack of family resources in determining whether a child is currently in school. On the other hand, for older students, many of the poor may not receive advanced degrees due to the quality of their earlier schooling. They would like to pass their Ordinary Level exam, but they do not succeed due to a lack of adequate preparation. Finally, analysis of the labor market characteristics of the poor suggests that reduction of unemployment would be an important contribution to the reduction of poverty both by employing those who are responsible for a family, the household head, and by reducing the dependency burden of families. We find, however, that underemployment, as measured by inadequate days of employment as opposed to insufficient wages, may be more a function of one's sector rather than of one's poverty status. 1 World Bank, "Table ls Basic Indicators," World Development Report, (New York: Oxford University Press, 1989), p. 164. -2- Background An island off the southern tip of India, Sri Lanka has a population of just over 16 million. The island was first colonized in 1505 by the Portuguese and was a British colony when it finally gained independence in 1948. One relatively unique aspect of Sri Lanka's economy is that there are three distinct sectors: urban, rural, and estate. The estates are large plantations where primarily tea, rubber, and coconuts are grown. There is typically a main employer which was previously British business interests, but since independence is the government of Sri Lanka or Sri Lankan nationals. Approximately 62 of the total population and 8% of the employed population live in the estate sector.2 Most of the workers are Indian Tamils (Indians who were brought over from India in the late 19th and early 20th centuries to work on the estates.) who live and work on the estate in houses provided by the estate. Entire families are often employed and because there is usually work available (although it is at a "minimum wage" in Sri Lanka), the workers do not generally go hungry. Indeed, out of 654 estate households surveyed, only 4 could be classified as oultr&-poor".3 Yet, other necessities and social services are scarce. There are fewer schools than in the rest of the country, and housing and health care are of relatively lower quality. In the area of health, in 1986, the infant mortality rate in the estate sector was 49.6, almost twice the rate in the rest of the country.4 On the other hand the rural sector, home of 73% of the country's population, is characterized by relatively good access to social services. Similarly, eighty-three percent of the households which own land reside in the rural sector as well as ninety-five percent of those who own paddy land. The majority of the residents are Sinhalese (the dominant ethnic group in Sri Lanka, most of whom are Buddhists) and over half of those employed work in agriculture, hunting, forestry, or fishing. Only about 16%, however, are 2 William T. Dickens and Kevin Lang, "An Analysis of Unemployment in Sri Lanka," Internal Discussion Paper 42, (Asia Regional Series, The World Bank, 1989), P. 3. 3 .Ultra-poor" is a classification proposed by Lipton (1983) which identifies households which consume less than 80% of their daily requirement of calories and for which food comprises more than 80% of their total expenditures. The poverty classifications are explained more fully in the section on methodology. 4 Population and Human Resources Division Country Department 1, Asia Region, *Sri Lanka Nutrition Review,* July 1989 (white cover), p. 1. 5 R.B.M. Korale and M.R.N.A. Fernando, "Employment, Unemployment, and Poverty," in Income Distribution and Poverty in Sri Lanka, ed. R.B.M. Korale, April 1987 (unpublished), p. 189. -3- regular paid employees, as compared with 25Z for the island as a whole, reflective of the seascnal nature of the work.6 In the early 1900s the country began to develop an extensive welfare state by providing free education and medical services, as well as income transfers. It is not clear, however, whether all of the programs precipitated the kinds of results envisioned by the government. Since the 1960s, unemployment rates have risen along with the level of educational attainment, reaching a high of 24% in the 1970s. Some argue that this rise resulted from inadequate growth unable to absorb the newly educated population (Korale and Fernando, p. 205). Similarly, the transfer programs have undergone significant changes since their conception. For example, in 1V79 the government replaced its Food Subsidy Scheme with a Food Stamp nogram. The Food Subsidy Scheme issued subsidized rice rations to half of the population and supported price subsidies on major foods such as wheat and sugar which benefitted the entire population.7 The food stamps are targeted at families with an income of less than Rs. 300 per month with adjustments made for families with six or more members. The households redeem their stamps at authorized shops and they can be used to buy a bundle of basic foods. The food stamp program is quite expensive, comprising 3Z of government current expenditure in 1984 (Edirisinghe, p. 9). Because the stamps are not indexed and hence lost their real value during a period of inflation in the early 1980s, in December 1988 the government doubled the value of food stamps ("Sri Lanka Nutrition Review," p. 41). Review of the Literature The literature on poverty in Sri Lanka is both rich and plentiful. Much of the economic literature has focused on the seeming anomaly of the economy: very high social indicators such as literacy rates and health statistics, and yet very low levels of per capita GNP. Hence, if one measures development in terms of "quality of life", Sri Lanka would appear to have fewer poor people than a country with a similar level of GNP. How Sri Lanka achieved such an outstanding quality of life is open to debate. Some believe that it was the result of government intervention (For example, Anand and Kanbur (1987)) while others contend that the state of the economy before independence was predisposed to such development (Isenman (1980), Sen (1986)). This literature is also part of the debate on the apparent conflict between growth and equity in Sri Lanka, and the effects of the 1977 liberalization. (See also Sahn (1984, 1987), Bhalla and Glewe (1985, 1986), and Glewve (1986, 1988).) In 1977 the new government moved to ease some of the government intervention in the economy. Many grew concerned, however, that the strides in equity that had taken place in the previous decades would be lost with the liberalization. 6 A.B.W. Nanayakkara and H.A.G. Premaratne, 'Food Consumption and Nutritional Levels,' in Korale, p. 146. 7 Neville Edirisinghe, The Food Stamp Scheme in Sri Lanka: Costs, Benefits, and Options for Modification, International Food Policy Research Institute No.58, March 1987, p. 9. -4- Analyzing the 1980/81 Socio-Economic Survey of the Department of Census and Statistics, Sahn (1987) finds that while unemployment declined and the growth rate of GDP increased, the poor were adversely affected through the high inflation following the liberalization which eroded their real wages as well as the real value of their food stamps. He further reports an increase in the level of malnutrition as measured both through the calorie consumption of low expenditure groups, as well as acute wasting. Sahn (1984) finds that while over one-third of rural household produce some paddy, only 16Z of these households are net producers. (A similar result holds for the producers of coconut.) Most of the households (especially among small farmers) are producing for home consumption. He estimates that an increase in these food prices would, in the short run, only benefit those farmers producing more than 90 bushels a year. In general, Sahn concludes that the liberalization policies, while potentially beneficial in the long run, were damaging to the poor in the short run, especially since they coincided with cutbacks in subsidies to the poor. Bhalla and Glewe (1985) conclude that the results on the effects of liberalization on the poor are more mixed. Comparing poverty rates from the 1969/70 and 1980/81 Socio-Economic surveys they find that the incidence of urban poverty and poverty among the better educated did rise, while that among female-headed households remained relatively stable. They caution, however, that using the 1969/70 survey as a baseline for comparison may be misleading as it was a particularly good year for rice production, and rice and wheat imports were high for political reasons as 1970 was an election year. Although they attempt to correct for such bias, the results may underestimate the level of poverty in 1970. Furthermore, Glewe and Bhalla (1985) find that unemployment rates fall after the liberalization such that the poorest Sri Lankans were made better-off as they were able to substitute wage income for government tiansfers. While much of the poverty literature has focused on equity and relative poverty measures and has attempted to assess the changes in poverty over time, there has been less emphasis on characterizing the poor and the non-poor at any point in time. Sahn (1984) explores some of the socio-economic characteristics of the ultra-poor and the nutritionally-at-risk. He, however, does not contrast them with the non-poor. Gunaratne (1987), utilizing the 1981/82 Socio-Economic Survey and the 1978/79 Consumer Finance Survey by the Central Bank of Sri Lanka also examines absolute poverty levels and the characteristics of the poor. He concludes that the poor are mainly landless agricultural workers and fishermen in the rural sector and hence the only way to eliminate absolute poverty is by increasing their productivity. The present study updates many of the earlier findings, and explores the issues of education and unemployment more closely. -5- DATA AND MTODOLOGY Data The data used in this paper are from the first six rounds of the Labor Force and Socio-Economic Survey of 1985/86 conducted by the Department of Census and Statistics in Sri Lanka.8 The survey is conducted over the eatire year and approximately 2,000 different households are interviewed each month. The current sample runs from May through October of 1985 and consists of approximately 11,800 household units. The survey contains data on household expenditure and consumption (of both food and non-food items) as well as income, demographic and socio-economic information about each household member. The data are well-suited to a study of the inter-relation of poverty, education, and unemployment. As with all surveys there are limitations in the data. Firstly, because there are only six rounds, extrapolation of some of the results to the entire year may lead to biases due to the seasonal nature of many of the economic activities. For example, rice is typically harvested in March and September leading to both increased employment and calorie intake. Secondly, the non- completion rate in this survey was unusually high in the areas plagued by political unrest. In the districts of Trincomalee, Vavuniya, and Kilinochchi the non-completion rates were 12?, 68%, and 83% respectively and as such they are undersampled (Preliminary Report, p. 29). These are districts made up of urban and rural sectors. Finally, in order to obtain estimates of the value of the housing in total expenditure for households who own their own home or who have their shelter provided for, the value of rent is imputed. As a result, the poverty classifications and expenditure shares are sensitive to errors in the amount imputed for each household.9 Methodology A fundamental problem in addressing issues of poverty is simply the definition of 'poverty*. One can focus on income or expenditure (which reflect the potential for fulfilling basic needs), or on the actual fulfillment of the basic needs, the quality of life, or the general well-being of the individual (which does not necessarily require a large income). Economists typically choose to examine either income or expenditure, although this study uses a combination of the two approaches. Relative and absolute measures of poverty are used to highlight different aspects of an economy. The study of relative poverty is intimately tied to that of income distribution, with the normative addition that everyone falling 8 Information about the survey can be found in Labour Force and Socio-Economic Survey-1985/86: Preliminary Report. Department of Census and Statistics, Ministry of Plan Implementation, 1987. 9 Angus Deaton, sAnalyzing the Food Shares in a Household Survey," in Three Essays on a Sri Lanka Household Survey, World Bank Living Standards Measurement Survey Working Paper No. 11 (1981). -6- below a certain portion of the distribution should be considered "poor". Relative poverty measures -mbody the notion that those who are the worst off suffer because they do not keep up with the "norm" of the society. By construction, for a given income distribution, relative poverty will always exist regardless of the growth of the economy. Measures of absolute poverty specify a particular cut-off in income or expenditure below which a person or household is considered to live in poverty. Hence, levels of absolute poverty are typically used to assess changes over time. Absolute poverty measures typically identify a minimum consumption bundle which is thought necessary in order to survive. Those who do not have sufficient income to purchase this bundle are probably living under extremely adverse conditions.?O This study analyzes the detailed food consumption data provided in the survey in order to define a "caloric" poverty line. The number of calories consumed by the entire household over the sample period (which in this case is one week) are totalled and compared to the total of the daily calorie requirements of the individual household members according to their age and sex (See Appendix A). If the household fails to meet its total caloric requirement, then all of the household members are considered to be living in poverty.11 One advantage of such a calorie line is that it strikes at the heart of well-being; one's nutritional status. If a person is hungry, he or she is less productive and hence less able to rectify his or her economic situation. This method of establishing a poverty threshold also obviates the need for intra-regional price indices. Nevertheless, as with any one definition of poverty there are also disadvantages to only looking at calorie consumption. First of all, as discussed above, it may be that if one examines expenditure data or a definition of poverty which includes housing, access to health care, or education, poverty would increase in the estate sector, but decrease in the urban sector. (Although, since prices are typically highest in the urban sector, high expenditure does not necessarily mean that a family is well-off.) Secondly, counting calories is a difficult and imprecise task. While the interviewers for the survey were instructed to record only quantities consumed during the survey period, one cannot know exactly how much of any food the family actually ate. Similarly, the calorie requirements are merely averages and they are based on healthy individuals of "normal" body weight. As discussed earlier, on the estates many people have parasites and all over the island stunting has been found to be a problem among children (See Sahn (1984) or "Sri Lanka Nutrition Review"). Ideally the calorie requirements would be based on age, sex, as well as weight and physical work demands. Finally, calorie counts often vary from one source to another (See Appendix B for the calories counts used in this study). These discrepancies 10 See Korale for a further discussion of poverty measures. 11 Some researchers either look at whether individuals are consuming below some per capita daily calorie requirement or a per adult equivalency requirement. The current method is equivalent to the latter. -7- are not a problem when families eat many foods, especially since the calories are divided among all of the family members. However, in Sri Lanka there are two major food staples in the economy: rice and coconuts. (Wheat flour is also consumed in large quantities in the estate sector.) Unfortunately different authors use different calorie counts for coconut. D6spite the fact that this study uses a count higher than the one used by Korale but lower than Sahn's, the estimated poverty levels are reasonably close to those reported by Korale who also used the 1985 data. Hence, while the poverty estimates seem relatively robust, it is just one more addition to the imprecision of counting calories. In order to mitigate some of the inaccuracies associated with any absolute povertl line, Lipton's three classifications of poor will be reported in this paper.1 The 'poor" will be defined as those households which do not achieve 10O% of their daily requirement of calories. The "nutritionally-at- risk" are those households which achieve less than a0? of their daily requirements but who allocate less than 80% of their total expenditures to food. The "ultra-poor" are those who achieve less than 80? of their daily requirements and who devote at least 80% of their total expenditures to food. Hence, the nutritionally-at-risk are those households which do not consume enough, possibly because of unwise nutritional choices as opposed to a lack of resources. On the other hand, the ultra-poor, after spending what they must on fixed costs such as rent and fuel, devote the rest of their income (expenditure) to feeding themselves. As illustrated in Table 1, the current results are very close to those found by Korale with the discrepancies due to the calorie counts. While one must compare studies employing different methodologies with caution, a comparison with Sabn's 1981 evidence suggests that there have been strides in lowering the ultra-poverty rate.13 On the other hand, the poverty rate in the urban sector appears to have remained relatively constant while that in the rural sector may have worsened and that in the estate sector improved, possibly due to the tea boom which peaked in 1984. 12 Michael Lipton, "Poverty, Undernutrition, and Hunger," World Bank Staff Working Papers No. 597, (1983). 13 It should be noted that these gains are an under-estimate due to the different calories that Sahn attributes to coconuts in his 1987 paper. -8- TABLE 1 POVERTY RATES (Individuals) Urban Rural Estate All-Island SAHN (1984) -- 1981 data Poor 49.9 42.9 32.6 Ultra 8.1 3.7 3.8 KORALE (1989) -- 1985 data Poor 53.8 50.1 23.3 49.2 At-rIsk 23.9 23.1 9.2 22.4 Ultra 3.2 2.9 1.8 2.9 AUTHOR'S CALCULATIONS - 1985 data Non-poor 51.4 53.8 76.1 54.4 Poor 48.6 46.2 23.9 45.6 At-risk 21.5 21.2 10.7 20.7 Ultra 2.8 3.0 1.0 2.8 CHILDREN (Age less than 14) All 29.8 33.8 37.2 33.1 Non-poor 36.8 38.4 68.1 39.9 Poor 63.2 61.5 31.9 60.1 At-risk 30.5 30.4 14.7 29.5 Ultra 4.5 6.0 1.3 4.7 Sources: Korale (1989), Sahn (1984), and Author's calculations. Note: All tables In this paper were calculated using the first six rounds of the Labor Force and Soclo-Economic survey of 1985/1986 unless otherwise noted. -9- One tool used to study poverty in Sri Lanka is the behavior of food shares. Food shares are the ratio of food expenditure to the total expenditure of the household. Since Engel first asserted that as a household's income (or total expenditure) rises, the proportion of income (or total expenditure) devoted to food falls (what is now known as "Engel's Law"), it has become accepted as an inverse indicator of welfare.14 Households with the same food share have the same level of welfare regardless of the household size. In Sri Lanka, however, the food share has been observed to rise over the first few expenditure deciles, before falling.15 Those families in the lowest expenditure deciles would like to consume more food, however because of other fixed costs of survival, such as shelter and clothing, they have only a little money left over for food. As their total expenditure rises they first devote more money to food before increasing their consumption of non-food items. Hence families rearrange their consumption bundles to insure that members have enough to eat and then they spend the additional income on non- food items. Furthermore, at the income rises so does the family switch from foods with low calorie costs to more expensive-per-calorie foods, a relationship referred to as "Bennet's Law.*16 As a result of the positive relationship between expenditure and food share over the lowest expenditure deciles, Sahn (1984) suggests that in order to help the ultra-poor, the government should turn to income and price interventions, while health education may be more appropriate for those in higher expenditure deciles who still fail to consume their daily requirement of calories, a group which is identified as the "nutritionally-at-risk" (Sahn (1984), p. 43). THE ANALYSIS HOUSEHOLDS Thirty percent of households in Sri Lanka live below the poverty line with 2.8% of them classified as ultra-poor (See Table 2). The difference in average calories consumed per person in each household between non-poor and poor households is quite remarkable. In general, the non-poor consume twice as many calories as do the poor and two and one-half times the calories consumed by the ultra-poor. The disparity between the non-poor and the ultra- poor is largest in the estate sector. 'While the estate sector households appear to consume their daily requirement of calories, the poor sanitation and harder work requirements (especially for the women) mean that they may be, on net, just as malnourished as those households in the rest of the country with lower caloric intake. This problem is evidenced by the poor child nutrition 14 See Deaton (1981) for a more in-depth discussion. 15 See Edirisinghe (1987) or Deaton (1981). 16 Thomas Poleman, 'Quantifying the Nutrition Situation in Developing Countries, I Food Research Institute Studies 18, No. 1 (1981). -10- TABLE 2 POVERTY RATES, 1985 Urban Rural Estate All-Island HOUSEHOLDS Non-poor 68.1 69.5 83.8 69.5 Poor 31.9 30.5 16.2 30.5 At-risk 11.3 11.7 6.3 11.4 Ultra 1.4 1.4 0.9 1.4 FEMALE-HEADED HOUSEHOLDS All 23.4 18.2 18.5 19.3 Non-poor 68.4 71.1 81.4 71.0 Poor 31.6 28.9 18.6 29.0 At-risk 13.1 9.9 6.3 10.5 Ultra 1.5 2.0 1.5 1.9 NUMBER OF HOUSEHOLDS IN SAMPLE All 4406 6769 654 11887 Non-poor 3018 4662 555 8270 Poor 1388 2107 99 3617 At-risk 496 802 40 1346 Ultra 67 99 4 175 Note: These rates are conditional on being in the column and one of of the row categories (Households, Female-headed Households, Children). Hence, under "Households", 868.1' is the percentage of urban households which are non-poor. TIE DISTRIBUTION OF HOUSEHOLD POVERTY GROUPS ACROSS SECTORS (Percentages) Urban Rural Estate All-Island ALL 20.9 73.3 5.8 100 Not-poor 20.3 72.7 6.9 100 Poor 22.3 74.6 3.1 100 At-risk 20.9 75.8 3.3 100 Ultra 20.7 75.7 3.7 100 -11- statistics such as stunting and wasting (See World Bank *Sri Lanka Nutrition Review'). While there is a distinct trend in the poverty status of households and their poverty status, there are a relatively large number of poor families in tlhe higher expenditure deciles (See Table 3). The apparent anomaly is mitigated by the fact that the expenditure deciles are calculated on an all- island basis, thus they are sensitive to the varying price levels in the sectors. For example, given that prices are higher in the urban sector, it isno surprise that there are more poor families in the higher expenditure deciles than in the other two sectors. Nevertheless, these numbers reflect that some families are efficient in their expenditure, able to reach the daily caloric requirements with little expenditure, while other families, despite relatively large outlays, still fail to consume enough calories. As is widely observed in the literature, the difference in individual calorie consumption between the poverty groups depends on whether the calories are measured per capita or per adult equivalences; the difference reflecting the larger number of children in poor households (See Tables 4 and 5).17 As the calories per adult equivalency attempt to allow for the varying calorie needs of the household members, it is the better over-all indicator of individual caloric consumption. Within Sri Lanka the average poor household falls 30% below the adult equivalence requirement of 2570 calories per day per adult.18 Although this number undoubtedly overestimates the amount by which the individuals fall short of their needs due to the fact that the stated 2570 daily requirement is probably an overestimate of that truly required (on average) in Sri Lanka (due to stunting when young which means that when adult 17 Each household member is converted into a fraction of an adult equivalent unit (which is based on males 25-45). See Appendix A for a table of the 1973 FA0/WHO recommended calorie requirements. 18 The estimated calorie consumption reported in Table 4 is quite high and in general is about 20Z more than that found by Sahn when he used the 1981/82 survey. To some extent this reflects differing methodologies, but further, as presented in Appendix C, there is evidence that food c3nsumption has increased in the time between the surveys. -12- TABLE 3 PERCENTAGE OF HOUSEBOLDS IN POVERTY GROUPS BY EXPENDITURE DECILES, 1985 Per Capita URBAN RURAL ESTATE Expenditure (Deciles) Poor At-risk Ultra Poor At-risk Ultra Poor At-risk Ultra 1 96.0 57.6 21.2 86.2 51.3 9.9 86.9 56.5 13.0 2 81.5 38.6 3.4 60.1 19.2 1.0 65.2 26.1 2.2 3 59.8 19.3 0.5 40.6 10.1 0.0 38.2 11.8 0.0 4 42.3 9.7 0.3 24.6 3.5 0.1 18.6 4.7 0.0 5 34.7 6.3 0.0 17.7 4.3 0.0 14.1 5.1 0.0 6 19.9 4.5 0.0 10.5 1.4 0.0 5.1 2.1 0.0 7 17.3 4.3 0.0 6.7 1.2 0.0 2.7 0.9 0.0 8 13.8 2.7 0.0 5.3 1.5 0.0 4.0 1.0 0.0 9 12.2 3.1 0.0 6.6 1.4 0.0 5.3 1.7 0.0 10 9.3 3.6 0.0 7.9 2.6 0.0 5.7 0.0 0.0 Total 32.2 11.4 1.4 31.1 11.9 1.5 16.7 6.6 0.9 NOTEt The expenditure deciles are based on expenditures for the whole island. -13- they are below "normal" body weight), it nevertheless reveals that malnutrition is a problem (Sabn (1984), p.25).19 Another familiar characteristic of the poor in Sri Lanka is their larger family sizes. Indeed, 60% of the children in Sri Lanka live in poverty (See Table 1). Families tend to be largest in the urban sector with an increase as the poverty level deepens. This general trend also holds for the ratio of "non-working" household members to working household members (the dependency burden) (See Table 4). As a result, the poor families must share their resources among more people. The larger household sizes may represent rational behavior, as poor families may invest in children in order to increase the number of people able to work. Expenditures Given the larger family sizes of poor households, one might surmise that households have the same total expenditure, but that poor families must divide these resources among more people. Table 6 reveals, however, that poor households have lower total expenditures than do the non-poor. Hence both the lower total expenditure as well as the larger household sizes, contribute to the low expenditures per household member. Similarly, on average, Sri Lankan households devote 62% of their total expenditure to food. Surprisingly, there is only a small difference between the food share of the non-poor and that of the poor. On the other hand, the ultra-poor devote on average 84% of their expenditures to food, suggesting that these families are struggling just to reach 80% of their calorie needs. Such a result arises partly by definition, but is also further evidence of the Engels relationship that has been documented in Sri Lanka.20 The very poor are unable to compress their expenditures on certain items such as housing and transportation and, as a result, they consume less food. As their income rises, instead of consuming more of all goods, they first increase their consumption of food. The result is a rising food share over the lowest expenditure deciles. Another implication of the high food share of the ultra-poor is that they are extremely vulnerable to price changes in some of the food staples. The ultra-poor allocate almost 30% of their total expenditure to rice alone and obtain 49? of their calories from rice, whereas the non-poor devote but 16% of their expenditure to rice in order to derive 44? of their calories from rice. Hence, while there are a few close substitutes to rice, such as coconuts and pulses, rice price changes are most likely to adversely affect the welfare of the poor, namely the ultra-poor. Indeed, Sahn has found that the poor have 19 It should be emphasized that these consumption figures are post- food stamp transfer. As a result, they do not reflect where these households would stand without the extensive food stamp program provided by the government. 20 See Edirisinghe (1984) and Sahn (1984) as discussed in the section on methodology. -14- TABLE 4 HOUSEHOLD CARACTRISTICS, 1985 Urban Rural Estate All-Island AVERAGE SIZE OF HOUSEHOLD All 5.3 5.1 4.6 5.1 Non-poor 4.7 4.6 4.4 4.6 Poor 6.4 6.1 5.7 6.1 At-risk 6.6 6.3 5.9 6.3 Ultra 7.6 6.9 5.2 7.0 AVERAGE DEPENDENCY BURDEN All 3.7 3.5 2.3 3.5 Non-poor 3.4 3.2 2.1 3.2 Poor 4.5 4.1 2.9 4.1 At-risk 4.8 4.2 2.9 4.3 Ultra 5.8 5.2 2.2 5.2 AVERAGE CALORIES PER CAPITA All 2338 2397 2717 2402 Non-poor 2680 2757 2948 2755 Poor 1617 1595 1573 1600 At-risk 1322 1324 1329 1326 Ultra 1269 1306 1304 1297 AVERAGE CALORIES PER ADULT EQUIVALENCY All 3178 3268 3654 3270 Non-poor 3741 3855 3993 3841 Poor 2003 1983 1981 1987 At-risk 1629 1637 1673 1638 Ultra 1560 1622 1555 1606 DEPENDENCY BURDEN is defined as the ratio of non-workers to workers per household. -15- TABLE 5 AVERAGE DAILY CALORIES CONSUMED PER ADULT EQUIVALENCY UNIT, 1985 Per Capita Expenditures (Deciles) Urban Rural Estate All-Island 1 1628 1888 1875 1832 2 2139 2420 2329 2338 3 2446 2748 2665 2651 4 2755 3094 2943 2978 5 2948 3349 3297 3212 6 3277 3743 3640 3564 7 3484 4021 4180 3813 8 3618 4261 4520 3993 9 3833 4333 4620 4084 10 4018 4441 4217 4171 Total 3178 3268 3654 3270 Note: The expenditure deciles are based on expenditures by the whole island. -16- TABLE 6 AVERAGE HOUSEOLD EPENDITURE, 1985 Urban Rural Estate All-Island TOTAL EXPENDITURE (Rupees per month) All 786.53 489.54 512.05 554.10 Non-poor 895.54 547.10 539.88 618.30 Poor 553.75 358.62 368.30 404.84 At-risk 468.27 301.66 317.87 337.02 Ultra 273.68 215.25 133.46 224.35 TOTAL EXPENDITURE PER CAPITA (Rupees per mouth) All 165.89 106.43 123.71 120.03 Non-poor 199.99 126.41 134.70 142.15 Poor 93.06 60.97 66.94 68.62 At-risk 75.07 50.09 56.16 55.68 Ultra 36.80 31.68 30.72 32.79 FOOD EXPENDITURE PER CAPITA (Rupees per month) All 72.22 56.08 64.87 60.00 Non-poor 84.30 65.49 70.47 69.73 Poor 46.23 34.64 35.92 37.36 At-risk 37.39 28.29 30.17 30.36 Ultra 30.79 26.74 25.63 27.62 -17- higher income and price elasticities of demand such that rice prices play an important role in determinants of poverty.21 One can gain insight into both the ultra-poor and the estate sector by looking at the share of expenditure devoted to other necessities, such as rent and health care, as well as education (See Table 7). In all sectors, the expenditure shares of these other goods for the ultra-poor is roughly half that of those for the nutritionally-at-risk group, providing further evidence that the ultra-poor spend the minimum possible on these non-food items. In the estate sector, the expenditure shares reflect that the estate sector is socio-economically different from the rest of the country. The rent share is low because it is mainly provided by the estate; the education share is low partly because of the high rate of school inattendance found in the estate sector (which is partly due to a lack of educational facilities on the estates). Finally, unless the health services are more highly subsidized in the estate sector than in other sectors, the low health share (expenditure on health care) probably reflects the lack of available health services. Assets As might be expected, an analysis of assets reveals that they distinguish the non-poor from the poor all over the island. In Table 8 it is clear that the ownership of land distinguishes the poor from the non-poor more in the urban sector than in either the rural sector (where most families own some land) or the estate sector (where almost no families own land). Similarly, the non-poor are much more likely to have some form of non-wage income (defined as pensions, remittances, or rent). The results suggest that the poor are more vulnerable to fluctuations in employment opportunities as is characteristic of largely agricultural economies. Again, the ultra-poor emerge as a group distinct from the rsst of the poor. They own less land and only 2-3Z of the households .ave any non-wage income. Clearly, the ultra-poor have little income security. Thus far, we have seen that the poor have larger families with more non- market-wage earning members than the non-poor families. While the overall average population growth in Sri Lanka over 1980-1987 was a low 1.5 percent per year (World Development Report, p. 214), it is undoubtedly much higher for the poor. Analysis of total expenditure per household and total expenditure per household member, however, suggest that poverty is not simply a matter of larger families. The poor seem to simply lack resources. They are much less likely to have either non-wage income or land such that they are more likely to rely on government transfers and their wage income as sources of income. 21 David Sahn, 'The Effect of Price and Income Changes on Food-Energy Intake in Sri Lanka,' Economic Development and Cultural Change, 36, No. 2, January 1988 (reprint). -18~- TABLE 7 AVERAGE EPEDITURE SHARES, 1985 Urban Rural Estate All-Island FOOD SHARE All 57.5 63.4 60.5 61.9 Non-poor 55.7 62.2 60.2 60.7 Poor 61.5 66.0 62.1 64.9 At-risk 60.5 64.2 58.9 63.2 Ultra 83.8 84.4 83.8 84.3 RENT SHARE All 9.5 4.8 3.3 5.7 Non-poor 10.1 4.9 3.3 5.9 Poor 8.5 4.7 3.4 5.5 At-risk 8.9 5.0 3.8 5.8 Ultra 4.1 2.8 4.6 3.1 EDUCATION SHARE All 4.6 3.5 1.3 3.6 Non-poor 4.6 3.3 1.3 3.5 Poor 4.5 4.1 1.3 4.1 At-risk 4.2 4.3 1.7 4.2 Ultra 2.5 1.9 0.0 1.9 HEALTH CARE SHARE All 3.5 3.8 2.9 3.6 Non-poor 3.5 3.9 3.0 3.7 Poor 3.4 3.6 2.7 3.5 At-risk 3.3 3.5 2.6 3.5 Ultra 1.3 1.2 0.6 1.2 CLOTHING SHARE All 4.1 4.8 5.0 4.7 Non-poor 4.3 4.8 5.0 4.7 Poor 3.7 4.8 4.9 4.5 At-risk 3.7 5.3 5.9 4.9 Ultra 1.4 2.1 2.9 2.0 NOTE: Shares are a percentage of total expenditure. -19- TABLE 8 HOUSEHOLD ASSET OWNERSHIP, 1985 Urban Rural Estate All-Island Land Ownership* All 63.3 87.9 6.1 78.0 Non-Poor 67.1 89.4 6.2 79.1 Poor 55.3 84.7 5.7 75.5 At-risk 57.5 83.6 2.1 75.3 Ultra 46.0 74.0 0.0 64.6 Paddy Ownership* All 5.0 31.0 0.5 23.7 Non-Poor 6.3 34.9 0.6 26.7 Poor 2.1 22.1 0.0 16.9 At-risk 2.1 19.6 0.0 15.2 Ultra 0.0 15.3 0.0 11.4 Non-Wage Income* All 28.5 15.7 18.1 18.5 Non-Poor 30.5 16.9 19.6 20.0 Poor 24.3 12.7 10.3 15.2 At-risk 22.3 11.0 8.8 13.2 Ultra 2.3 3.5 0.0 3.4 Land Ownership* Non-Poor 72.1 70.6 84.8 70.9 Poor 27.9 29.4 15.2 29.1 At-risk 10.3 11.2 2.2 11.0 Ultra 1.0 1.2 0.0 1.2 Paddy Land Ownership** Non-Poor 86.8 78.2 100.0 78.6 Poor 13.2 21.8 0.0 21.4 At-risk 4.9 7.4 0.0 7.3 Ultra 0.0 0.7 0.0 1.0 Non-Wage Income Distribution** Non-Poor 72.8 75.3 90.7 75.3 Poor 27.2 24.7 9.3 24.7 At-risk 8.8 8.3 3.1 8.1 Ultra 0.1 0.3 0.0 0.3 *Percentage of households conditional on being in both the sector and the poverty group, for example the first cell, 63.3, should be read as the percentage of all households in the urban sector that own land. **The distribution within the sector among landowners/non-wage income recipients; across poverty groups. (Within sectors, non-poor+poor=100%.) -20- Food Stams As one would expect and as seen in Table 9, the poor are more reliant on the government for a source of income for their food than are the non-poor, with the ultra-poor in the urban and rural sectors relying on them the most.22 Because the poverty line is derived after the government transfers have been taken into account, some of the non-poor families who receive food stamps are undoubtedly pushed over the poverty line by the government aid. Hence one cannot directly interpret the non-poor recipients as unworthy of the transfers. On the other hand, these results are evidence of mis-targeting and leakage within the program and of the inadequacy of food stamps for those who remain below the poverty line. Unfortunately, as also noted by Edirisinghe, the distribution of food stamps does not adequately reach either the ultra- poor or the poor in the estate sector. This result is consistent with other findings that social services in general do not reach either the ultra-poor or the estate sector. Being that the number of food stamps received by the family is supposed to reflect the number of family members, particularly children, with no limit, it does not seem that the large family sizes are contributing to the failure to reach the poorest families. Edirisinghe suggests that it may be due to errors in determining eligibility or that ultra-poor households may be just above the cutoff line (Edirisinghe, p.28). An estimate of the amount of money needed to bring all of the poor families up to their daily requirement of calories, not only represents the potential fiscal burden to the government, but it also lends insight into whether the proposed Jana Saviya Program (JSP) will prove adequate for the needy families, if indeed they are reached under this new program. The government would need to allocate approximately Rs. 72,000,000 per month or Rs. 330 per household, in addition to the current food stamp program.23 When the average value of food stamps per household is added to this number, the cost to the government is Rs. 516 per household. While this calculation, which is based on the calorie per rupee spent by each family, is not exact since it is not constant as a family's income increases, it does suggest that the Rs. 2500 per month household subsidy promised under the JSP is in excess of what is truly needed in order to alleviate inadequate caloric intake. 22 As noted previously, because this poverty line relies mainly on the calorie consumption of the households, there will be some households with a large income below the poverty line and there are households who because of the food stamp program are above the present poverty line, depending on how the household allocates its budget to food and depending on its food choices. Despite these anomalies one can see a trend in the food stamp share of total household expenditure. 23 In 1989 rupees. Source: Colombo Consumer Price Index, Central Bank of Sri Lanka. -21- INDIVIDUALS One gains insight into the demographic and household characteristics of the poor by analyzing the data at the household level, yet much more can be learned about their labor market experiences and perhaps some causes of poverty by looking at individuals, be they heads of households, new labor market entrants, or children. And yet, before embarking on an exploration of the data at the individual level, a cautionary note is in order. The poverty line is derived from statistics about the household and an assumption about the distribution of resources, food in particular, is needed to use the data to answer questions about individuals. The crucial assumption is that the available food is distributed equitably among the household members according to their needs and that no members receive preferential treatment. There is evidence that this assumption is more plausible in Sri Lanka in wealthier families than in the poorest. For example, by analyzing survey data of households in the Kandy district in 1984, Edirisinghe has found that the calorie consumption of children increased by only half that of older members of the household with the introduction of food stamps. On the other hand, this was mainly true for the ultra-poor such that once a family reached 80Z of its daily requirements, the children started to benefit more from the program. In fact, this may represent income-maximizing behavior by the household as it insures that the most productive members of the family are better fed (Edirisinghe, p. 63). Furthermore, while fewer assumptions is always better, in light of the lack of available data as well as the wealth of information to be explored in the labor force and socio-economic survey, it is reasonable to analyze the data at the individual level while never losing sight of this underlying assumption. Who Are the Poor? As reflected in Table 10, a person in poverty in Sri Lanka is likely to be a Sinhalese (although the Sri Lanka Moors are the most over-represented group in poverty) who resides in the rural sector and works as a farmer or agricultural worker, if employed, and while being literate, is unlikely to have advanced past the 0 Level exam in his or her studies. On the other hand, a Sri Lankan Moor who resides in the urban sector and in unemployed has the highest incidence of poverty. The first sentence reflects that most Sri Lankans are Sinhalese, live in the rural sector, and work in agriculture. Hence, these characteristics do not distinguish the poor from the non-poor, although they do reflect where policies might be targeted to reach the largest number of poor. The second sentence reflects which groups are over- represented among the poor and hence may provide clues to causes of poverty in Sri lanka. -22- TABL 9 THE FOOD STAMP PROGRAM, 1985* Urban Rural Estate All-Island PERCENTAGE OF HOUSEHOLDS RECEIVING FOOD STAMPS All 40.3 63.3 6.1 56.1 Non-poor 31.9 56.0 5.9 48.6 Poor 55.3 78.1 6.8 71.7 At-risk 63.1 81.7 4.5 76.1 Ultra 72.9 84.1 0.0 78.9 DISTRIBUTION OF RECIPIENTS WITHIN EACH SECTOR All 12.4 87.0 0.6 100.0 Non-poor 50.7 59.6 80.3 56.1 Poor 49.3 40.4 19.7 41.4 At-risk 21.0 16.5 4.9 17.0 Ultra 3.4 2.2 0.0 2.3 AVERAGE VALUE OF FOOD STAMPS PER CAPITA** (Rupees per month) All 27.81 21.11 19.13 21.92 Non-poor 33.31 20.94 20.32 22.26 Poor 22.14 21.36 14.28 21.44 At-risk 26.98 19.85 10.18 20.90 Ultra 18.60 16.19 0.00 16.58 FOOD STAMP SHARE OF HOUSEHOLD FOOD EXPENDITURE** All 9.7 11.1 11.0 10.9 Non-poor 7.8 9.3 10.8 9.2 Poor 11.6 13.8 12.1 13.5 At-risk 13.5 16.3 8.5 15.8 Ultra 15.3 16.2 0.0 16.0 * All numbers are conditional on being in both the sector &nd the poverty group. For example, the first cell, 40.3, should be understood as the percentage of all households in the urban sector who receive food stamps. Among food stamp recipients, 27.81 is the average monthly value of food stamps per household member received among households in the urban sector. ** Among food stamp recipients. -23- TABLE 10 WO ARE THE POOR In 1985? 69% of the poor are Sinhalese, although Sri Lankan Moors have the highest poverty rate. Over 70% of the poor live in the rural sector although the poverty rate is highest in the Urban sector. 40% of the poor work in agriculture. 15? of the poor are illiterate which is true for the country as a whole. Only 14? of the poor who are 15 years of age or older have passed their 0 Level as compared with 24? of the noa-poor. The labor force participation rate for the poor is 45Z compared with 54? for the non-poor. The unemployment rate of the poor is almost twice that of the non-poor. 15% of the employed poor are underemployed compared with 10% of the non-poor. -24- What is striking about the Sri Lankan situation and which makes analysis and description of the poor all-the-more challenging is the distribution of poverty across the sectors. Again, as noted previously, it is not surprising that the poverty rate in the estate sector is roughly half that of those found in the other two sectors because of the definition of poverty used in this investigation. On the other hand, while most of the poor are found in the rural sector, the poverty rate is highest in the urban sector. (This could be an artifact of the poverty line. It is plausible that urban sector work is, in general, less demanding than work elsewhere on the island such that the urban workers require fewer calories.) On the other hand, the incidence of ultra-poverty is highest in the rural sector perhaps reflecting a seasonality of employment opportunities. Other demographic trends are presented in Table 11. As is clear, the average age of the population varies both within and across the sectors. This result reflects both the larger family sizes of the poor as well as their higher mortality rates (especially in the estate sector) and basically provides insight into the age distribution of the population. The Role of Education Although its effectiveness may be challenged by some, education is one avenue which both developed and developing countries have stressed in trying to alleviate poverty. Indeed, Sri Lanka is a textbook example where the literacy rate rivals that of countries with much higher growth rates because of the government's dedication to education. The government emphasizes education both to raise the quality of life of Sri Lankans, as well as to provide an avenue by which people can rise out of poverty. Education is highly correlated with poverty. A chi-squared test of independence24 between the state of poverty (poor/non-poor) and educational attainment of the head of the household is significant at the 1% level. (See Table 12 for educational attainment and Appendix D for the chi-squared tests.) Similarly, whether one is literate or not is also highly correlated with one's state of poverty. (Because of the high literacy rate in the country as a whole and its nebulous definition, literacy is less of a distinguishing characteristic than is education.) As reflected in Table 13, the literacy rate is relatively high and improving in Sri Lanka; the literacy rate is 89% for children aged 5-14 as compared to 84Z for adults. 24 The null hypothesis of a chi-squared test is that the two characteristics are not correlated, and the alternative hypothesis is that they are correlated. A sufficiently high chi-squared statistic allows one to reject the null hypothesis of no correlation. -25- TABLE 11 INDIVIDUAL CHARACTERISTICS OF TEB POOR, 1985 Non-poor Poor At-risk Ultra All AVERAGE AGE All-Island 30.6 21.3 20.0 17.8 26.4 Urban 31.7 22.7 21.3 20.0 27.4 Rural 30.8 20.9 19.6 17.1 26.2 Estate 25.5 19.4 19.0 20.7 23.8 PERCENT FEMALE All-Island 50.3 48.7 47.8 47.7 49.6 Urban 52.4 49.5 49.2 49.3 51.0 Rural 49.8 48.3 47.4 47.2 49.1 Estate 50.1 52.9 49.7 50.0 50.7 PERCENT OF SECTOR POPULATION AGE LESS THAN 14 All-Island 24.3 43.7 47.1 54.8 33.1 Urban 21.3 38.7 42.2 47.0 29.8 Rural 24.1 45.1 48.5 57.1 33.8 Estate 33.3 49.5 50.8 50.3 37.2 -26- TABLB 12 8IGHEST EDUCATIONAL ATTAINMENT OF INDIVIDUALS 15 AND OLDERe 1985 LEVEL OF EDUCATION Urban Rural Estate All-Island NO SCHOOLING All 7.5 11.3 30.1 11.4 Non-poor 6.9 11.0 27.8 11.2 Poor 8.4 11.9 39.8 11.7 At-risk 9.9 13.1 36.3 12.9 Ultra 12.1 13.0 40.1 13.3 PASSED GRADES 0-4 All 16.9 25.2 37.4 23.9 Non-poor 15.2 22.9 39.3 22.3 Poor 19.2 28.9 29.2 26.4 At-risk 19.5 30.5 31.5 27.8 Ultra 27.9 27.5 59.9 28.3 PASSED GRADES 5-9 All 48.1 44.5 26.2 44.5 Non-poor 44.3 43.1 26.4 42.3 Poor 53.3 46.7 25.4 47.9 At-risk 53.9 46.5 23.7 47.8 Ultra 54.4 54.7 0.0 53.6 PASSED ORDINARY LEVEL EXAM All 20.0 15.0 5.5 15.7 Non-poor 23.7 18.0 5.5 18.4 Poor 15.0 10.3 5.4 11.4 At-risk 13.6 8.4 8.5 9.7 Ultra 5.1 4.5 0.0 4.6 PASSED ADVANCED LEVEL EXAM All 5.1 3.0 0.7 3.4 Non-poor 6.6 3.7 0.8 4.1 Poor 3.1 1.9 0.2 2.2 At-risk 2.4 1.5 0.0 1.6 Ultra 0.4 0.2 0.0 0.3 COLLEGE AND BEYOND All 2.4 0.9 0.1 1.2 Non-poor 3.4 1.4 0.1 1.7 Poor 1.0 0.2 0.0 0.4 At-risk 0.7 0.1 0.0 0.3 Ultra 0.0 0.0 0.0 0*0 NOTE The percentage of individuals conditional on being both in the row and the column. For example, 7.5% of all urban residents have no schooling, while 6.92 of the non-poor urban residents have no schooling. -27- Unfortunately, it is difficult to untangle the causality with respect to poverty and educational attainment. Is it that the poor are poor because of a lack of human capital, or is the lack of human capital a result of their poverty? As reflected in Table 14, on the island, 84% of the children between the ages of 4 and 15 attend school. The percentage is highest in the urban sector and lowest in the estate sector where the returns to education are probably lower and the opportunity cost of a child being in school higher than on the rest of the island. Furthermore, the educational system on the estates appears relatively less developed. More families cite that their children are not in school because there is no school in the neighborhood and most children attend schools which only teach through grade 5, as will be discussed later in the paper. A child's school attendance is highly correlated with whether the head of the household is literate or not. Hence, there is evidence of a perpetuation of poverty through the generations. If the child of an illiterate is less likely to attend school, and if school achievement is likely to determine future employment, then the child of a poor household is unlikely to escape from that poverty. Parents who did not attend school may not believe in the importance of education, or they may simply not know how to enroll their children in school. Similarly, in all sectors the non-attendance rate increases with the level of poverty suggesting that either the poor attach different value to education or that they find the education of their childrentoo costly either in foregone income or in direct outlays. One would imagine that for the poor, who almost by definition lack sufficient income, the issue of a high opportunity cost of investing in a child's education must be relevant. When asked why the children were not in school, 23% of the poor cited "financial difficulties" as the primary reason for non-attendance as compared to 19% of the "non-poor" (See Table 15). While this suggests a difference between the poverty groups, it is not as large as might be expected.25 One reason that the poverty groups do not widely diverge is that the educational system, as a whole, is so well-developed. The public schools are 'free* and the books and supplies are supplied by the state. Nevertheless, some families must pay for transportation to and from school and, more importantly, in a country where clothing and appearance are very important, they must provide acceptable school clothes for their children. Note that only 9? of the poor and 7? of the non-poor claimed that they needed the children to help with household work or other work. On the other hand, both groups cited "other' as the primary reason that their child was not in school. 25 See Appendix D for a chi-squared test of independence. While the Chi-squared statistic does not accept the null at the 1% level, 88% of the data are missing so that the test must be viewed with caution. -28- TABLE 13 ILLITERACY RATES, 1985 Urban Rural Estate All-Island HEADS OF HOUSEHOLDS All 11.1 18.1 32.3 17.4 Non-poor 9.2 16.8 32.0 16.3 Poor 15.0 20.8 33.7 19.9 At-risk 16.7 24.1 23.9 22.5 Ultra 20.6 29.9 35.4 28.1 ADULTS (age 15 and older) All 10.9 15.9 40.2 16.0 Non-poor 9.9 15.1 39.6 15.5 Poor 12.3 17.4 42.9 16.7 At-risk 13.4 18.9 39.8 18.1 Ultra 16.3 21.0 40.1 20.2 CHILDREN (ages 5-14) All 9.5 9.6 37.6 10.8 Non-poor 6.3 7.9 36.7 9.9 Poor 11.2 10.5 39.3 11.3 At-risk 11.5 11.7 43.5 12.5 Ultra 14.2 17.7 8.7 16.7 -29- TABLE 141 CHILDRE AND EDUCATION IN 1985* SCHOOL ATTENDANCE RATES Non-poor Poor At-risk Ultra All All-Island 85.7 83.6 82.9 77.0 84.4 Urban 89.0 85.9 85.1 82.2 86.9 Rural 87.3 84.1 83.5 76.9 85.3 Estate 65.0 51.1 53.5 18.5 60.2 MECNTAGE OF STUDENTS ATTENDING THE DIPFERENT TYPES OF SCHOOLS Ron-poor Poor At-risk Ultra All A LEVEL SCIENCE All-Island 20.8 14.3 12.8 8.2 16.8 Urban 40.4 27.4 25.8 15.0 32.0 Rural 17.0 10.7 9.3 6.4 13.1 Estate 5.6 7.1 3.8 0.0 6.0 A LEVEL COMMERCE &ARTS All-Island 19.4 19.3 19.7 13.4 19.3 Urban 14.1 16.1 15.9 14.3 15.4 Rural 22.1 20.5 21.1 13.2 21.1 Estate 3.5 7.9 5.7 0.0 4.8 TO GRADE 10 All-Island 39.6 49.1 51.3 60.9 45.4 Urban 29.7 41.3 41.8 53.4 37.2 Rural 43.5 52.1 54.6 62.7 48.8 Estate 24.7 21.1 24.6 100.0 23.7 TO GRADE 5 All-IslaAkd 20.3 17.3 16.2 17.6 18.4 Urban 15.8 15.2 16.5 17.3 15.4 Rural 17.5 16.8 15.0 17.7 17.1 Estate 66.2 63.9 65.8 0.0 65.5 *All numbers are percentages conditional on being in both the sector and the poverty group. For example, 85.7Z of the non-poor.children all over the island currently attend school. -30- *ABLE is REASONS FOR CHILD INAOFDANCB OF SCHOOL, 1985 Need for Need to Sickness, No School in Cannot Economic Help With Disability or Neighborhood Afford Activity Housework Handicap Other ALL-ISLAND All 6.1 22.0 2.1 6.2 5.6 58.0 Non-poor 6.6 19.7 2.4 5.0 4.8 61.5 Poor 5.8 23.3 1.9 6.8 6.1 56.1 At-risk 7.2 26.0 1.0 6.5 6.5 52.8 Ultra 2.7 35.7 3.9 6.5 2.5 48.7 URBAN All 1.8 28.0 2.9 5.5 5.3 56.5 Non-poor 1.1 27.6 4.0 2.7 3.9 60.7 Poor 2.1 28.2 2.4 6.7 5.9 54.7 At-risk 1.1 36.8 0.3 8.0 6.8 47.1 Ultra 1.3 40.4 10.3 0.0 4.4 43.6 RURAL All 3.6 20.5 1.9 6.3 6.1 61.6 Non-poor 1.5 20.0 2.6 4.6 5.3 66.1 Poor 4.7 20.8 1.5 7.1 6.5 59.5 At-risk 7.2 22.2 1.3 6.5 6.6 56.2 Ultra 2.0 35.8 0.0 8.2 2.3 51.7 ESTATE All 25.6 22.2 2.0 6.7 3.5 40.1 Non-poor 26.4 13.4 0.8 7.8 3.8 47.8 Poor 24.4 34.1 3.7 5.1 3.1 29.7 At-risk 21.9 36.0 0.0 3.6 4.9 33.7 Ultra 15.6 22.7 41.1 0.0 0.0 20.6 The distribution of reasons are percentages of the poverty group and sector, for example 6.6% of the non-poor all over the island claim their child does not attend school because there is no school in the neighborhood. -31- The estate sector distinguishes itself in this area. Fully one-quarter of the residents cited that their children did not attend school because there was none in the neighborhood as compared with approximately 1.5% in the urban and rural sectors. Similarly, over one-third of the estate poor claimed financial difficulty as the primary reason that their child was not in school and yet only 9Z claimed that they needed the child to work. Hence, either the children are working and the opportunity cost of school is too high, or because the families live so far from the school, the transportation costs are prohibitive. In either case, it is clear that sufficient educational services are lacking in the estate sector. In order to assess the importance of a lack of family income as a reason that a child is not in school, a binary probit was run with a dependent variable indicating whether the child is in school or not. The results are presented in Table 16. Expenditure per capita (i.e. per household member) was included to control for the financial resources of the family. There may be some simultaneity in this equation, because if the child is not in school and working, then she or he is contributing to the available household resources. However, given that these are children aged 5 to 14 is it most likely that any contribution is only a small share of total family resources. The expenditure variable does not appear to be a significant factor, except in the estate sector where the opportunity cost of attending is most likely the highest. On the other hand, the literacy (the variable "illiterate") and the educational attainment of the head of the household (These are dummy variables "Grade School'-grades 0-9; "0 Level and Beyond"-Ordinary Level and further) are generally statistically significant. These results suggest that the value of education of the head of the household is more important than finances, although it should be cautioned that only approximately 10Z of the children (higher in the estate sector) do not attend school. It is well-known that while the government has been -ather successful at providing education, there are varying qualities of schools. Hence it is possible that the difference between the sectors and between the poverty groups is not the quantity, but rather the quality of education. Of course, "quality" is a rather elusive attribute to quantify and define, but the survey does inquire into the type of school attended by the child which reflects school curriculum as well as the future prospects of the child's education. Forty-five percent of the children attend schools that continue until grade 10. It is after grade 10 that the student attempts to pass his or her Ordinary Level exam (0 Level). If the student is successful, then he or she will continue on to attempt the Advanced Level exam (A Level) and then possibly advance to university. As is clear from Table 14, however, the type of school that one is likely to attend depends on one's sector and poverty group. A child from a non-poor urban family is almost twice as likely to attend a school with A Level science as a child from a poor urban family or a child from a rural non-poor family. On the other hand, a child (either poor or non-poor) from the estate sector is over three times more likely to attend a school that is only through grade 5 than a child from either of the other two sectors. Hence, we see that there is a wide variety of school types and *32- TABLE 16 SCHOOL AhIN 1985 PROBIT IESULTS Dependent Variables 1CHILD IN SCHOOL/0-CHILD NOT IN SCHOOL URBAN RURAL ESTATE Coefficient Coefficient Coefficient Variable (Std. Error) (Std. Error) (Std. Error) INTERCEPT -0.1609 -0.0189 -1.0224 (0.1522) (0.1018) (0.3025) SRI LAMKAN TAMIL1 -0.1209* -0.1476** 0.1156 (0.0547) (0.0580) (0.1119) PEWALE -0.0974* -0.0205 -0.2872** (0.0473) (0.0347) (0.1021) POOR -0.1137* -0.1232** -0.1630 (0.0525) (0.0389) (0.1180) EXPENDITURE PER CAPITA 0.0001 0.0000 0.0038** (0.0001) (0.0002) (0.0011) AGE 0.0995** 0.1044** 0.0805** (0.0081) (0.0059) (0.0180) DEPENDENCY BURDEN 0.0867** 0.0235** -0.0241 (0.0114) (0.0083) (0.0423) GRADE SCHOOL1 0.1717 0.1783** 0.5227** (0.1151) (0.0721) (0.1998) 0 LEVEL AND BEYOND1 0.4549** 0.2811** 0.8880** (0.1296) (0.0936) (0.3438) ILLITERATE1 -0.2157* -0.2661** 0.1105 (0.1016) (0.0629) (0.1792) Number of Observations 4750 8178 682 Log-Likelihood -1764.0 -3316.2 -411.4 * Significant at the 52 level ** Significant at the 1% level 1 Refers tp the head of the household. Agriculture and Fishing are industry dummies. *33- therefore probably school quality. It should also be noted that one determinant in educational quality that exists between the poor and non-poor is attendance of private school. These schools are expensive and they tend to provide a much better education. Hence, as far as the educational attainment of a child is concerned, if a child stops attending school before grade 10, it is most likely due to a high opportunity cost of education in the estate sector, and due to more sociological factors in the other two sectors, such as the ability of the head of the household to enroll the child or possibly a difference in the value placed on education within the family. For students who pass grade 10, however, the determinants of educational attainment probably change. One explanation is that poor families deem that an education past grade 10 is too costly to be worth the possible future return (in the form of a go"d job). In this case, the family views the probability of the student actually receiving a well-paying job to be too low. (The poor may assess that their probabilities are lower since they may have fewer informal connections into good jobs.) Hence, they would rather that the student proceed directly into the labor force. Given that the poor were more likely than the non-poor to state that they were not in the labor force because they were studying, this explanation probably does not fully describe the current situation. It is more likely, however, that those students who achieve grade 10 would like to pass the 0 Level, and indeed they do try. Because many of the good government jobs require an 0 Level or A Level, the students have an incentive to achieve these degrees. However, because of the low quality of their grade school education, they lack the preparation necessary to pass the standardized tests in which they are competing with students from better equipped schools. The poor, by virtue of the fact that they cannot afford a private school education, or because they are not situated in an area with good public schools, are unable to advance in their education. Labor Force Participation, Unemployment, and Underemployment With regard to poverty and unemployment in Sri Lanka, people have argued two seemingly contradictory claims. On the one hand, unemployment is highly correlated with poverty. On the other hand, the poor cannot afford to be unemployed. The first statement reflects that given that one is unemployed, there is a high chance that one comes from a household that is below the poverty line. (Similarly, given that the head of the household is unemployed, there is a high probability that the family lives in poverty.) This finding is not surprising in a country with no unemployment transfers and where very few people have savings to help them smooth their consumption through hard times. On the other hand, this statement, as suggested by the second claim, does not suggest that poverty and unemployment are synonymous. By eradicating unemployment one would eage the plight of at most 9.5Z of the poor -- those poor who are unemployed. If one looks at unemployed heads of households, then one would only help 4% of the poor families. In other words, there would remain a core of those poor who are currently working but are apparently -34- underemployed. This next section tries to investigate these two aspects of the Sri Lankan labor market, as well as labor force participation. Examination of Table 17 reveals a surprising trend among labor force participants: the poor have a lower labor force participation rate than do the non-poor. (Labor force participation is defined as *being employed or available for work for a major part of the last 12 months.*) There are two possible explanations. It may be that the poor are more likely to be discouraged workers, those who have dropped out of the labor force because they believe that there is no available work and have hence lost hope. Or, since the labor force is defined for people aged greater than 10, it may be that the statistic is skewed by youth who are not actively looking for work due to their studies. Table 18 presents the labor force activity of heads of households. Here we see a reverse trend. The labor force participation rate is generally higher for the poor than the non-poor, except in the estate sector which reflects the unique nature of its labor market. Hence, labor force participation increases when a person assumes the responsibilities of a household. Furthermore, poor heads of households are less likely to cite old age or retirement as the reason for their non-availability for work. (Fifty percent of the non-poor heads of households who are not in the labor force claim that this as opposed to 38% for the poor.) The non-poor heads of households probably have retirement pensions or families that are better able to support them such that they can afford not to work. Among those poor who are labor force participants, unemployment is an important determinant of poverty.26 Unemployment rates are consistently higher for the poor, increasing with the level of poverty, than for the non- poor. The literature on the nature of unemployment in Sri Lanka has emphasized that it is mostly generated by highly educated youth who are waiting for good government jobs (Glewwe (1987) and Glewe and Bhalla (1985)). Dickens and Lang contend that because highly educated youth (i.e. those who had passed their A Level) comprise at most 10% of the unemployed (in 1985/86), that unemployment is better characterized as simply youth unemployment. In terms of poverty, if the bulk of the unemployed are young people who depend on their parents and relatives for assistance while unemployed (Over 90% of the unemployed claim that this is the case. See Appendix D.), then one reason why unemployment may lead to poverty is the larger number of people who must share each paycheck, i.e. the higher dependency burden. 26 See Appendix D for the distribution of employment across poverty groups and within each sector. TABLE 17 LABOR FORCE PARTICIPATION, UNEMPLOYMENT, AND UNDEREMPLOYMENT RATES IN 1985 The Distribution Across Poverty Poverty Group Rates Groups and Within the Sector Non-poor Poor At-risk Ultra All Non-poor Poor At-risk Ultra LABOR FORCE PARTICIPATION All-Island 53.7 44.6 43.4 39.2 49.9 62.4 37.6 16.1 1.8 Urban 48.7 43.1 42.9 38.2 46.1 57.2 42.8 17.9 2.2 Rural 53.1 44.6 43.0 39.0 49.5 62.0 38.0 16.5 1.8 Estate 77.0 60.5 59.1 60.7 73.4 82.2 17.8 7.9 0.8 UNEMPLOYMENT All-Island 12.8 21.4 23.1 28.1 14.0 49.9 50.1 23.3 3.2 Urban 15.5 28.8 34.1 34.9 21.1 41.8 58.2 28.9 3.6 Rural 12.7 19.3 19.9 27.1 15.2 51.7 48.3 21.5 3.3 Estate 8.1 13.0 16.7 0.0 8.9 74.2 25.8 14.8 0.0 UNDEREMPLOYMENT #1* All-Island 10.3 15.4 17.0 22.3 21.1 55.3 44.7 20.8 2.9 Urban 8.2 11.9 12.8 23.7 9.6 52.3 47.7 19.9 4.4 Rural 11.7 16.6 18.3 21.7 13.5 55.4 44.6 21.2 2.5 Estate 4.2 10.6 12.3 26.6 5.3 65.a 34.2 17.0 4.5 UNDEREMPLOYMENT #2** All-Island 5.9 10.2 11.6 16.1 7.4 51.8 48.2 23.1 3.4 Urban 4.1 7.2 8.1 11.2 5.3 47.8 52.2 22.7 3.8 Rural 6.9 11.1 12.6 16.9 8.4 52.3 47.7 23.3 3.2 Estate 2.3 9.3 9.7 26.6 3.5 54.5 45.5 20.3 6.9 * Those people who work fewer than 180 days per year as a percent of the usually employed population. ** Those people who work fewer than 180 days per year and are available for more work as a percent of the usually employed population. -36- TAN 18 LABOR ORCE PARTICIPATION, UNMULOIMMN, AND 0 1 r *0LOMB RATES 01 HEADS or I:SHO: S. 1985 Non-poor Poor At-risk Ultra All LABOR FORCE PARTICIPATION All-Island 76.7 81.7 82.0 81.2 78.2 Urban 69.4 72.9 69.5 71.9 70.5 Rural 77.3 84.1 85.0 84.7 79.5 Estate 91.9 85.4 93.1 66.1 90.8 UNEMPLOYMENT All-Island 3.1 4.4 5.5 6.6 3.5 Urban 5.1 7.1 9.7 13.0 5.8 Rural 2.7 3.9 4.7 5.2 3.1 Estate 2.4 0.7 1.5 0.0 2.1 UNDEREMPLOYMENT #1* All-Island 8.3 11.6 13.8 19.6 9.4 Urban 7.1 9.8 10.7 26.8 8.0 Rural 9.4 12.2 14.7 16.7 10.3 Estate 1.7 8.1 11.6 46.5 2.7 UNDERMPLOYMENT #2** All-Island 5.0 8.3 10.5 17.1 6.1 Urban 3.3 6.6 7.7 23.8 4.4 Rural 5.9 8.8 11.3 14.3 6.8 Estate 1.2 6.1 7.1 46.5 1.9 * Those people how work fewer than 180 days per year as a percent of the usually employed population. ** Those people who work fewer than 180 days per year and are available for more work as a percent of the usually employed population. *37- The unemployment rates for heads of households are much lower than that for all individuals. Hence, as observed in the unemployment literature, once individuals decide that it is time to start a family, they give up waiting for the good jobs and take other available work. Because heads of households are often among the main income earners, if they are unemployed, the family simply lacks resources. Vhile they may not account for the bulk of the unemployed, their employment status is largely indicative of the poverty status of the household. Given that most of the poor unemployed (i.e. the young) may not remain either poor or unemployed once they assume family responsibilities, the unemployment among the heads of households is a more serious concern. Essentially these are individuals whose unemployment may not reflect queuing, but rather a lack of necessary job skills, bad luck, or an unstable profession. A better understanding of this fraction of the unemployed would greatly benefit any attempts to alleviate poverty. Tables 19a-19c report the poverty rates of industries in each of the sectors.27 The poor are over-represented in industries such as manufacturing in the urban sector, agriculture in the rural sector, and construction in the estate sector. In all three sectors, the poor are over-represented in industries which are not defined, most likely because these are casual workers who take odd jobs. As evidence, the poor are generally over-represented in the industries in which the percentage of poor varies according to the month. (In other words, a chi-squared test of independence suggests that there is dependence between the percentage of the workers in the industry in poverty and the month of the survey. See Appendix D.) Specifically, the poor are over-represented in manufacturing, construction, wholesale and retail trade as well as "not- identified" industries, i.e., Industries with apparent seasonal variation. On the other hand, the poor are under-represented in the utilities, transportation, financing/insurance/real estate/business services, and community/social/personal services. The two industries which do not hold to this general rule are agriculture and forestry and mining which most likely reflects occupational differences within these two industries. 27 Note that these are industries and not 'occupations' (which are more conceptually relevant). The focus on industries is due to inconsistencies in the data on occupations. TABLE 19a POVERTY AND INDUSTRY IN THE URBAN SECTOR, 1985 Industry as a Percent of Poverty Group Poverty Group as a Percent of Occupation INDUSTRY Non-poor Poor At-risk Ultra All Non-poor Poor At-risk Ultra Agriculture 7.4 5.7 5.9 5.9 6.7 67.4 32.6 13.2 1.6 Fishing 4.0 4.5 4.2 9.5 4.2 58.3 41.7 15.1 4.0 Forestry/Mining 0.7 0.9 1.4 0.0 0.8 55.0 45.0 25.4 0.0 Manufacturing 18.0 21.7 20.3 14.8 19.5 56.8 43.2 15.6 1.3 Electricity, Gas and Steam 0.9 0.8 0.5 1.2 0.8 63.4 36.6 9.8 2.5 co Construction 5.9 8.7 8.3 17.4 6.9 51.6 48.4 17.7 4.4 Wholesale and Retail Trade /Restaurants and Hotels 19.3 21.8 21.7 19.8 20.3 58.3 41.7 16.0 1.7 Transport/Storage/ Communication 8.5 7.5 6.3 3.6 8.1 64.3 35.7 11.6 0.8 Financing/Insurance/ Real Estate/ Business Services 4.2 1.8 2.3 0.0 3.3 78.5 21.5 10.5 0.0 Comunity, Social, and Personal Services 29.4 22.2 23.6 10.9 26.6 67.7 32.3 13.2 0.7 Not Defined 1.9 4.4 5.6 17.0 2.8 40.6 59.4 29.3 10.7 TABLE 19b POVERTY AND INDUSTRY IN THE RURAL SECTOR, 1985 Industry as a Percent of Poverty Group Poverty Group as a Percent of Occupation INDUSTRY Non-poor Poor At-risk Ultra All Non-poor Poor At-risk Ultra Agriculture 50.5 52.2 52.0 57.5 51.1 63.0 37.0 15.8 1.8 Fishing 1.8 1.7 1.7 2.1 1.8 64.3 35.7 14.7 1.9 Forestry/Mining 1.7 1.7 1.6 2.8 1.7 64.7 35.3 14.7 2.6 Manufacturing 11.8 13.4 13.2 8.4 12.4 60.8 39.2 16.6 1.1 Electricity, Gas and Steam 0.3 0.3 0.2 0.0 0.3 60.4 39.6 13.0 0.0 Construction 4.2 5.3 5.0 4.6 4.6 58.2 41.8 17.1 1.6 Wholesale and Retail Trade/Restaurants and Hotels 8.5 7.7 6.2 4.3 8.2 66.3 33.7 11.8 0.9 Transport/Storage/ Communication 4.6 3.0 2.5 1.1 4.0 72.9 27.1 9.9 0.4 Financinglinsurance/ Real Estate/ Business Servicis 1.1 0.3 0.2 0.0 0.9 85.6 14.4 3.9 0.0 Community, Social, and Personal Services 12.3 5.9 6.0 2.9 10.0 78.5 21.5 9.4 0.5 Not Defined 3.2 8.5 11.2 16.4 5.1 40.1 59.9 34.1 5.1 TABLE 19c POVERTY AND INDUSTRY IN THE ESTATE SECTOR. 1985 Industry as a Percent of Poverty Group Poverty Group as a Percent of Occupation INDUSTRY Non-poor Poor At-risk Ultra All Non-poor Poor At-risk Ultra Agriculture 92.7 88.6 89.7 85.3 92.0 83.7 16.3 7.1 0.9 Fishing 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Forestry/Mining 0.5 1.9 2.5 14.7 0.7 56.2 43.8 25.6 18.2 Manufacturing 1.7 1.7 0.8 0.0 1.7 83.4 16.6 3.5 0.0 Electricity, Gas and Steam 0.0 0.4 0.9 0.0 0.1 0.0 100.0 100.0 0.0 Construction 0.6 1.1 0.8 0.0 0.7 71.0 29.0 9.1 0.0 Wholesale and Retail Trade/Restaurants and Hotels 0.7 1.9 2.5 0.0 0.9 64.2 35.8 20.2 0.0 Transport/Storagel Comuwunication 0.6 0.5 0.0 0.0 0.6 86.6 13.4 0.0 0.0 FinancingiInsurance/ Real Estatel Business Services 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Caoounity, Social, and Personal Services 3.1 2.1 0.9 0.0 2.9 87.7 12.3 2.3 31.2 Not Defined 0.1 1.8 1.7 0.0 0.4 23.3 76.7 0.0 0.0 -41- Hence the poor are employed in less stable industries. Combined with the earlier finding that they typically do not have non-vage income, aside from government transfers and remittances from abroad, and they have few assets, we see that their income fluctuates with their wage income and hence employment opportunities. Again, one of the general aspects of poverty in Sri Lanka is the vulnerability of tho households. Unemployment as a cause of poverty suggests that it is a transitory state alleviated with the finding of a job. However, the fact that most of those poor who are usually in the labor force (i.e. available for work) are employed for at least part of the year leaves one to suspect that underemployment is a major problem among the poor: they are working but not often enough, or at an insufficient wage. Yet, surprisingly, the evidence suggests that underemployment, as defined by the number of days worked rather than a sufficient wage, is more a function of sector than poverty status such that it is unlikely to be a major determinant of poverty on the island. Using a definition which should provide an overestimate (those who worked less than 180 days last year whether they declare themselves available to work more or not), the all-island underemployment rate of the poor is 15%. (The underemployment rate for those who worked less than 180 days last year and are available to work more, is 10%.) There are certainly problems with this definition of underemployment since it relies on people's recollection over the year and does not capture those who under-utilize their human capital, nevertheless it appears that underemployment more generally reflects one's sector rather than one's poverty status. Indeed, the underemployment rate in the rural sector is about 90% higher than that in both the urban and estate sectors. Most of the poor are found in the rural sector so a significant fraction of Sri Lanka's poor are underemployed. However, the underemployment rate of the non-poor in the rural sector is also relatively high. Hence, underemployment plagues the entire rural sector. This same finding is true for the urban, but not the estate sector. In the estate sector, underemployment strikes more heavily the poor and increases as the poverty level increases. Possible sources of the difference in income among the poor and the non- poor are reflected in Table 20. While these results should be viewed with caution since there is an average response rate of only 30% in the wage data, the poor appear to work slightly fewer hours per week than the non-poor and they earn a slightly lower (estimated) hourly wage. The small difference in hourly wages is rather surprising and raises the possibility that the poor suffer more adversely from fluctuating employment than from low wages. -42- TABLE 20 MONTELY WAGES, HOURS WORKED, AMD HOURLY WAGES ACROSS SECTORS* 1985 Urban Rural Estate All-Island AVERAGE MONTHLY WAGES All 682.35 486.96 464.69 522.53 Non-poor 733.49 539.12 473.01 568.66 Poor 633.51 431.93 447.30 489.33 At-risk 578.73 430.74 470.88 459.80 Ultra 622.72 382.64 338.03 437.94 AVERAGE HOURS WORKED IN THE LAST WEEK All 40.2 31.7 38.6 33.8 Non-poor 40.6 32.7 39.0 34.7 Poor 39.7 29.9 36.8 32.2 At-risk 39.1 29.0 34.9 31.2 Ultra 38.2 27.5 43.3 30.6 IMPLIED HOURLY WAGE RATES All 4.24 3.84 3.01 3.86 Non-poor 4.52 4.12 3.03 4.10 Poor 4.00 3.61 3.04 3.80 At-risk 3.70 3.71 3.37 3.68 Ultra 4.07 3.48 1.95 3.58 Note: These are nominal vages in 1985 rupees. *43- Regression Results Finally, in order to statistically control for the effects of education, unemployment, and underemployment as sources of poverty, an ordinary least squares regression with calories per adult equivalency as the dependent variable was estimated. Dummy variables for the education of the head of the household and whether the head of the household is underemployed or unemployed (with "fully-employed" as the base group), as well as socio-economic variables of the household head were included among the independent variables. The causality with respect to education is the most problematic.28 However, one can take adult education as pre-determined and then look at the causal structure of education on poverty conditional on the past educational outcomes for adults.29 Since the head of the household's job situation is most likely a large determinant of the poverty status of the family, and since his or her education was largely determined before assuming family responsibilities, one can take the education of the head of the household as pre-determined in the sense that it is irreversible. (On the other hand, the educational choices of the children of this family are more likely a function of the family's poverty.) The results are presented in Table 21. Because 'the coefficients are conditional on the pre-determined outcome of education, they are an over- estimate of the returns for the general population. Yet, in the rural sector there appear to be positive returns to education at all levels, while in the urban sector the returns accrue mainly to those who have at least passed their 0 Level.30 Finally, in the estate sector, where people have lower educational attainment than on the rest of the island, there are only returns to an 0 Level or beyond, most likely reflecting the value of formal education for the available jobs. It should be noted that very few people on the estates have an education beyond the 0 Level. 28 In light of efficiency wage theories of wage determination and work effort, it is also possible that the causality with respect to unemployment and underemployment in this regression runs both ways. It may well be, especially for the ultra-poor, that those who do not consume enough calories do not have the energy and strength necessary to maintain a job. This theory most likely applies to only a small percentage of the sample, notably the ultra-poor. 29 Paul Glewwe, "Investigating the Determinants of Household Welfare in Cote D'Ivoire," unpublished manuscript, July 1989. 30 Education appears to have more of an impact on household welfare in the rural sector than in the urban sector. One possible explanation is that given that those who live in the urban sector are generally more educated, education may be less of a 'signal" of being an able worker. If employers cannot perfectly observe, while making hiring decisions, whether someone would be a good worker or not, and they mainly judge from the level of one's education, then when everyone is well-educated, it is harder for employer's to feel confident about who they have hired, reducing the economic value of the education. This explanation, however, warrants further investigation. -44. TABLE 21 D:hRCNATS OF BOUSEBOLD WLARE IN 1985 Dependent Variables Calories per adult equivalency URBAN RURAL ESTATE Coefficient Coefficient Coefficient Variable (Std. Error) (Std. Error) (Std. Error) INTERCEPT 2687.64 2510.55 3895.72 (136.564) (102.409) (326.487) AGE 34.21** 45.03** 17.84 (1.808) (1.415) (5.152) FEMALE -49.39 -226.90** 298.11* (72.412) (61.103) (153.579) LAND 1187.42** -47.59 -- (407.804) (395.337) NONWAGE -775.82 849.33 -- (531.837) (927.726) HOUSEHOLD SIZE -210.99** -294.04** -174.56** (10.818) (9.261) (29.951) DEPENDENCY BURDEN -65.88** -45.11** -186.95** (11.759) (10.412) (51.109) GRADES 0-9 90.17 263.39** 179.81 (88.430) (57.157) (139.797) PASSED 0 LEVEL 289.60** 644.30** 639.50** (96.015) (75.254) (239.878) PASSED A LEVEL 516.14** 969.30** -- (128.544) (151.074) COLLEGE + 750.90** 1305.71** -- (132.425) (156.470) UNEMPLOYED -148.25 -354.59** 385.26 (91.271) (103.015) (345.049 UNDEREMPLOYED #1 -19.35 -12.10 -937.82 (73.979) (56.548) (285.697) R2 0.2692 0.3115 0.2067 Number of Observations 2996 5180 542 * Significant at the 5Z level. ** Significant at the 12 level. O LEVEL is for those who have passed the 0 Level exam and beyond in the estate sector. UNDEREMPLOYED 11 is those who worked less than 180 days last year and are available for more work. AGE, SEX, LAND (ownership of land), NQNWAGE (receipt of non-wage income), and education refer to the head of the household. Note: Other control variables not reported include round, race, and district dummies. -45- Part of the return to a higher education comes in the form of eligibility for high paying government jobs. In fact, given the high wage and the generous pension, employment in the government may be driving the returns to education. There is both a strong correlation between education and whether a person has a government job as well as a large and significant relationship between calories consumed (household welfare) and whether a household member has a government job. The dummy variable for whether the head of the household is employed by the government or not is relatively large and significant at the 5% level. This result holds for the urban and rural sectors, but not for the estate sector where over 80% of the labor force works for the 0government'. It appears that there is a statistical difference between unemployment and full-employment, but not between full-employment and underemployment. Of course, this test is not an ideal one since it excludes all of the poor heads of households who are out of the labor force, nevertheless, it does suggest that unemployment may be more strongly correlated with poverty than underemployment, which is more indicative of the individual's sector. Whether the household is headed by a woman or not is less important in either the urban or the estate sectors than in the rural sectors, possibly reflecting employment opportunities for women in each sector. The regression also includes controls for other family resources. The dummy variable for whether or not the family owned land is only important in the urban sector most likely because everyone owns land in the rural sector and no one owns land in the estate sector, as discussed earlier in the paper. The dummy variable for whether the family receives non-wage income is insignificant in both the urban and rural sectors. These results certainly warrant further investigation for the magnitude of the estimates suggest that it is important, but unfortunately the coefficients are imprecisely estimated as evidenced by the large standard errors. Finally, the fact that both the size of the household hold and the dependency burden are significant components in the equation suggests that supporting children as well as unemployed youth burdens the already small incomes of poor families. Hence, policies aimed at reducing youth unemployment will most likely improve, simultaneously, the poverty rates. CONCLUSION In characterizing the poor in Sri Lanka, this paper raises many areas for future research. While the study was not constructed as a investigation of whether poverty rates have declined since the early 1980s, the results suggest that they have, as reflected in the ultra-poverty rates across the country. On the other hand, the poverty rates suggest that there is a large fraction of the population which fail to fulfill one crucial basic need: sufficient caloric intake. These Sri Lankans tend to be landless, casual workers who are most likely agricultural workers and fishermen. Given that the nutritional adequacy does not appear to be uniform all over the island, it would be interesting to study how some of the other social indicators such as birth and mortality rates differed among the poor and the non-poor. Clearly the fact that 60% of the children in Sri Lanka live in poverty suggests that the government may want to direct programs to both protecting the welfare of these -46- young Sri Lankans as well as possibly enlarging existing family planning programs. Education remains highly correlated with poverty and yet, in light of the current labor market situation, it is not clear that higher education directly translates into an escape from poverty. Any investment in education must be viewed as risky given the limited number of good jobs. Hence, the lack of education among the poor may reflect that they are unable to pass the exam based on their lack of adequate preparation, or it may reflect a rational choice by the family which attempts to maximize the household income. In any event, the role of education as a road out of poverty is worthy of future consideration. The evidence suggests that unemployment is a stronger determinant of poverty than is underemployment, although further investigation is necessary. A definition of underemployment based on wages and hours worked and decomposed by occupation would shed more light on the relative importance of these two phenomena. Unemployment is a tricky area since the majority of the unemployed are young labor market entrants who will eventually find employment when the costs of waiting for a good job outweigh the potential benefits. Unemployment benefits for certain groups of people may help alleviate the poverty caused by heads of households being unemployed, but the government would do well to discourage the queuing unemployment of the young, as is encouraged in the unemployment literature. *47. BIBLIOGRAPHY Anand, Sudhir, and Ravi Kanbur. "Public Policy and Basic Needs Provisiont Intervention and Achievement in Sri Lanka." Unpublished manuscript (1987). Bhalla, Surjit, and Paul Glewwe. "Living Standards in Sri Lanka in the Seventies: Mirage and Reality." Mimeo, World Bank (1985). Country Monograph Series No. 4. Population of Sri Lanka. Bangkok, Thailands United Nations Economic and Social Commission for Asia and the Pacific, 1976. Deaton, Angus. Three Essays on a Sri Lanka Household Survey. Living Standards Measurement Study Working Paper No. 11, World Bank, 1981. Dickens, William T., and Kevin Lang. *An Analysis of the Nature of Unemployment in Sri Lanka." World Bank Internal Discussion Paper, Asia Regional Series No. 42 (1989). Edirisinghe, Neville. The Food Stamp Scheme in Sri Lanka: Costs, Benefits, and Options for Modification. International Food Policy Research Institute No. 58 (1987). Glewwe, Paul. "The Distribution of Income in Sri Lanka in 1969-1970 and 1980-81: A Decomposition Analysis.* Journal of Development Economics, 24 (1986), pp. 255-274. ------------. "Economic Liberalization and Income Inequality: Further Evidence on the Sri Lankan Experience." Journal of Development Economics, 28 (1988), pp. 233-246. ------------. "Investigating the Determinants of Household Welfare in Cote D'Ivoire." Mimeo, World Bank (1989). -------. 'Unemployment in Developing Countriess Economist's Models in Light of Evidence from Sri Lanka." International Economic Journal, 1, No. 4 (1987), pp. 1-17. Glewwe, Paul, and Surjit Bhalla. "Unemployment and Wages in Sri Lanka 1963-1982.* Mimeo, World Bank (1985). Guanaratne, Leslie. "The Poorest of the Poor in Sri Lanka,' in Alleviation of Poverty in Sri Lankas A Symposium. Colombo, Sri Lankat The Department of Information Central Bank of Sri Lanka, 1987. -48- Isenman, Paul. 'Basic Needes The Case of Sri Lanka.* World Development, 8, No ?7, (1980), pp. 237-258. Korale, R.B.M. (ed). Income Distribution and Poverty in Sri Lanka. Unpublished manuscript (1987). Labour Force and Socio-Economic Survey- 1985/86: Preliminary- Report. Colombo, Sri Lankas Department of Census and Statistics, 1987. Lipton, Michael. 'Poverty, Undernutrition, and Hunger." World Bank Staff Working Papers No. 597 (1983). Marga Institute. An Analytical Description of Poverty in Sri Lanka. Colombo, Sri Lankas Sri Lanka Centre for Development Studies, March 1981. Ravallion, Martin, and Sisira Jayasuriya. 'Liberalization and Inequality in Sri Lankas A comment." Journal of Development Economics, 28 (1988), pp. 247-255. Sahn, David E. "An Analysis of Food Consumption and Expenditure Patterns in Sri Lanka." Mimeo, U.S. Agency for International Development (1984). ------------. "Changes in the Living Standards of the Poor in Sri Lanka During a Period of Macroeconomic Restructuring.' World Development, 15, No. 6 (1987), pp. 809-830. ------------. 'The Effect of Price and Income Changes on Food-Energy Intake in Sri Lanka.* Economic Development and Cultural Change, 36, No. 2 (1988), pp. 315-340. Sen, Amartya K. "Sri Lanka's Achievementst How and When?." in Srinivasan and Bardhan (1986). Srinivasan, T.N., and Pranab K. Bardhan (eds). Rural Poverty in South Asia. New Yorks Colombia University Press, 1986. World Bank. 'Sri Lanka Nutrition Review.' White cover (1989). World Bank. World Development Report. New York: Oxford University Press, 1989. -49- APPENDICES -50- APPENDIX A DAILY RECOMMENDED CALORIE REQUIREMENTS AND ADULT EQUIVALENCY WEIGHTS WOMEN Weight Adult Equivalency Age (Kg.) Calories Weights 7-12 months 7.3 818 0.43 1-3 years 12.0 1212 0.54 4-6 years 18.2 1656 0.72 7-9 years 26.2 1841 0.87 10-12 years 36.0 2238 0.93 13-15 years 40.0 2300 0.80 16-19 years 43.8 2200 0.75 20-39 years 47.0 1900 0.71 40-49 years 47.0 1805 0.68 50-59 years 47.0 1710 0.64 60-69 years 47.0 1520 0.51 70 years 47.0 1330 0.50 MEN Weight Adult Equivalency Age (Kg.) Calories Weights 7-12 months 7.3 818 0.43 1-3 years 12.0 1212 0.54 4-6 years 18.2 1656 0.72 7-9 years 26.2 1841 0.87 10-12 years 36.0 2414 1.03 13-15 years 40.0 2337 0.97 16-19 years 43.8 2500 1.02 20-39 years 47.0 2530 1.00 40-49 years 47.0 2404 0.95 50-59 years 47.0 2277 0.90 60-69 years 47.0 2024 0.80 70 years 47.0 1771 0.70 Sources: 1/Energy and Protein Requirements - World Health Organization Report. Series No. 552, 1973. 2/National Sample Survey Organization of India -51- APPENDIX B CALORIE CONVERSION FACTORS Calorie per 100 g. of Units from of Edible Portion Food Item Survey (unless otherise marked) Rice g 346.0 Wheat Flour g 344.0 Kurakkan 5 328.0 Maize 5 86.0 Bread g 245.0 Buns number 115.8/bun Cakes g 100.0 Hoppers number 93.52/Hopper String Hoppers number 45.36/String Hopper Thosai number 46.48/Thosai Pittu number 206.0/Pittu Pappadam g 288.0 Noodles g 363.0 Oats, Rye, Barley g 338.0 Corn Flakes g 390.0 Infant's Cereal g 384.0 Dried Chillies g 291.0 Red Onions & 56.0 Bombay Onions 5 47.0 Garlic g 123.0 Cumin Seeds g 356.0 Fennel Seeds g 333.0 Mathe Seeds g 333.0 Corriander 5 288.0 Maldive Fish g 204.0 Ginger g 67.0 Turmeric g 349.0 Mustard g 541.0 Tamarind g 283.0 Green Chillies g 26.0 Pepper g 289.0 Limes g 45.0 Dhall (Mysoor. Thora, Kadala, etc.) g 343.0 Green gram g 348.0 Gram whole g 360.0 Cow pea gram g 323.0 Soya Beans g 432.0 Pumpkin g 8.0 Ash Plaintain g 44.0 Brinjal g 22.0 Bandakka g 31.0 Bitter Gourd g 20.0 Cucumber i 9.0 Drumstick i g 22.0 -52- Calorie per 100 g. of Units from of Edible Portion Food Item Survey (unless otherise marked) Kohila Yams g 24.0 Long Beans g 38.0 Snake Gourd 9 16.0 Ridge Gourd & 13.0 Sweet Pumpkin g 20.0 Beans g 43.0 Carrot g 43.0 Beet Root 9 36.0 Cabbage g 23.0 Khol Khol g 19.0 Tomatoes 5 19.0 Leeks 8 27.0 Capsicum Chillies g 24.0 Raddish g 12.0 Bread Fruit number 396.0/fruit Potatoes 8 92.0 Sweet Potatoes g 1.02 Manioc g 133.0 Kiriala g 96.0 Innala g 82.0 Coconut number 1513.1/coconut Beef a 210.0 Mutton g 87.0 Pork g 465.0 Chicken g 73.0 Liver g 136.0 Fresh Large Fish g 101.0 Fresh Small Fish g 64.0 Prawns 8 58.0 Crabs g 60.0 Dried Fish 8 179.0 Sprats 8 298.0 Jadi 8 149.0 Sardin g 263.0 Cow milk ml 700.0/1 Goat milk ml 740.0/1 Condensed milk g 325.0 Milk Powder g 496.0 Infant Milk Powder g 499.0 Cheese g 348.0 Coconut Oil ml 8120.0/1 Gingetty Oil ml 8280.0/1 Soya Oil m al 8120.0/1 Ghee ml 8280.0/1 Butter g 729.0 Margarine 8 765.0 Hen Eggs number 86.5 Bananas number 105.0/banana Pineapple number 376.2/pineapple Papaw number 117.0/papaw -53- Calorie per 100 g. of Units from of Edible Portion Food Item Survey (unless otherise marked) Mangoes number 135.0/mango Oranges number 72.0/orange Avocado number 302. 5/avocado Rambutan number 18.7/Rambutan Grapes g 67.0 Thambli/Kurumba number 500.0/coconut Dates g 276.0 Cashew nuts g 170.0 Groundnuts g 397.0 Plums g 47.0 Canned Pineapple g 223.0 Canned mango 5 237.0 Tea dust/leaves g 40.0 Coffee powder/seeds g 56.0 Sugar g 400.0 Juggery (Coconut, Kital, Sugarcane, Palmyrah) g 340.0 Honey g 319.0 Jam/Jelly g 260.0 Marmite/Vegemite g 6.0 Soya meat g 432.0 Milk tea number 50.0/cup Plain tea number 50.0/cup Milk coffee number 50.0/cup Plain coffee number 50.0/cup Soft drinks (bottled) number 327.-0 Sources: 1/ A.B.W. Nanayakkara and H.A.G. Premaratne, "Food Consumption and Nutritional Levels," in Income Distribution and Poverty in Sri Lanka, ed. R.B.M. Korale, April 1987 (unpublished). 2/ Composition of Foods: Raw. Processed, Prepared (Revised). U.S. Department of Agriculture. 1963-. -54- APPENDIX C PER CAPITA FOOD *NX TION YEAR RICE WHEAT FLOUR SUGAR COCONUTS FISH MEAT (Kglyear) (Kglyear) (Ig/year) (nuts/year) (Kg/year) (Kglyear) 1978/79 91.0 16.4 8.7 92.2 9.6 1.7 1980/81 110.0 8.7 8.9 87.6 9.8 2.2 1981/82 101.2 10.6 10.1 92.4 10.9 2.2 1985/86 106.4 9.4 14.1 90.3 13.4 3.2 Sourcess 1/Consumer Finance and Socio-Economic Survey 1978/79 Sri Lanka: Part II. Statistics Department Central Bank of Ceylon. September 1984. 2/Labour Force and Socio-Economic Survey 1980/81 Sri Lanka. Department of Census and Statistics, Ministry of Plan Implementation. 3/Report on Consumer Finances and Socio-Economic Survey 1981/82 Sri Lanka: Part II. Statistics Department. Central Bank of Ceylon. May 1985. 4/Labour Force and Socio-Economic Survey 1985/86 Sri Lanka. Department of Census and Statistics, Ministry of Plan Implementation. -55- APPENDIX C (cont.) CALORIE SHARES OF RICE. WHEAT FLOUR, AND COCONUT IN 1985 Rice Wheatflour Coconut All-Island 45.2 3.7 22.8 Non-poor 44.5 4.0 23.3 Poor 47.2 2.9 21.5 Urban Non-poor 38.0 2.9 21.2 Poor 41.1 2.3 19.1 Rural Non-poor 46.6 2.5 24.5 Poor 49.3 2.6 22.4 Estate Non-poor 40.9 22.4 16.4 Poor 46.0 15.2 17.6 Source: Author's calculations. .� � i � и о+ . С7 • � н DC v . , ' _ � -57- THE DISTRIBUTION OF EMPLOYMENT ACROSS POVERTY GROUPS AND WITHIN THE SECTOR 1985 Non-poor Poor At-risk Ultra All-Island 64.8 35.2 14.8 1.6 Urban 61.3 38.7 14.9 1.8 Rural 63.8 36.2 15.5 1.6 Estate 83.0 17.0 7.2 0.9 -58- TABLE OF POVERTY AND THE HOUSEHOLD RECEIPT OF REMITTANCES POVERTY RECEIPT OF REMITTANCES FREQUENCY DOES NOT PERCENT HRECIVE IRECVES ROw POT Iltt- MIT- COL POT TANCES TANCES TOTAL 7144 11261 8270 NOTPOOR 00.o10 9.47 69.57 88.38 13.82 - 69.00 73.45 3210 407 3617 POOR 27.00 3.42 30.43 88.75 11.25 31.00 26.55 TOTAL 10354 1533 11887 87.10 12.90 100.00 STATISTIC OF VALUE PROO *******.********* ********* .**** .** .** -*** -*-*;** .* . CHI*SoUARE 1 12.509 0.000 LIKELIHOOD RATIO CHI-SQUARE f 12-788 0.000 CONTINUITY ADJ. CHI-SQUARE I 12.300 0.000 MANTEL-HAENSZEL CHI-SQUARE I 12.508 0.000 FISHERS EXACT TEST 41-TAIL) 0.000 (2-TAIL) 0.000 PHI -0.032 CONTINGENCY COEFFICIENT 0.032 CRAMER'S V -0.032 SAMPLE SIZE a 11887 TABLE OF POVERTY BY HOUSEHOLD RECEIPT OF NONWAGE INCOM 4DIVIDENDS.RiMITTANCES.PROPERTY RENT. PENSION) POVERTY RECEIPT OF NONWAGE INCOME FREQUENCTOOS NOT PERCENT RECEIVE RECEIVES ROW POT NONWAGE NONWAGE COL POT I INCOME I INCOME I TOTAL 6473 1797 8270 NOT-POOR 54.45 15.12 69.57 70.2 21.73 G8.0. 75.80 3037 I160 I 3817 POOR 25.55 4.88 30.43 83.98 16.04 3t.93 24.40 TOTAL 9510 2377 t887 80.00 20.00 100.00 STATISTIC OF VALUE PROS CHI*SQUARE I 50.993 0.000 LIKELIHOOD RATIO CHI-SQUARE 1 52.643 0.000 CONTINUITY ADJ. CHI-SQUARE 1 50.638 0.000 MANTEL-HAEMSZEL CHI-SQUARE 1 50.989 0.000 FISHERS EXACT TEST 41-TAIL) 0.000 42-TAIL) 0.000 PHI -0. 085 CONTINGENCY COEFFICIENT 0.065 CRAMER'S V -0.085 SAMPLE SIZE 2 1887 -59- TABLE OF GOVEANMENT EMPLOYMENT BY LEVEL OF EDUCATION FOR HEADS OF HOUSEHOLDS GOVERNMENT EMPLOYEE LEVEL OF EDUCATION FREQUENOV PERCENT ROW PCT NO GRADES GRADES PASSED PASSED CL PCT SCHOOL I 0-4 1 9-9 1 0 LEVELI A LtVALICOLLEGEol TOTAL 123 334 681 601 7 1 128 1984 GOVERNMENT 1.43 3.88 7.901 6009 .38 1.49 23.03 EMPLOYEE 6.20 16.83 34.32 30.29 5.90 6.45 15.59 12.11 19.50 49.88 00.00 71.91 NOT A 668 2423 0 211 1 804 1 78 50 6832 GOVERNMENT 7.73 1 28.12 32.63 1 7.01 0.95 0.98 76.97 EMPLOYEE 10.04 I 38.53 | 42.39 i 9.11 f .86 0.75 84.41 87.89 80.50 50.12 40.00 ( 28.09 TOTAL 789 2757 3492 1205 i95 i78 86 9.1 32.00 40.53 13.99 2.26 2.07 100.00 FREQUENCY MISSING * 2886 STATISTIC DF VALUE PROS CHI-SQUARE S l14.766 0.000 LIKELIHOOD RATIO CHI-SUARE 5 989.310 0.000 MANTEL-AENSZEL CHI-SQUARE I 830.574 0.000 PHI 0.360 CONTINGENCY COEFFICIENT 0.338 CRAMER'S V 0.360 EFFECTIVE SAMPLE SIZE a 8656 FREQUENCY MISSING * 2886 WARNING: 25% OF THE DATA ARE MISSING. -60- TABLE OF POVERTY AND ROUND OF SURVEY INDUSTRY NOT DEFINED POVERTY ROUND NUMBER (MONTH OF INTERVIEW FREQUENCY1 PEACENT ROW POT COL PCT 11 21 31 41 51 81 TOTAL 30 33 S 37 45 60 258 NOT-POOR 4.75 8.23 8.08 .6 7.13 9.51 40. I18.72 I12.89 8 9.921 14.481 17.581 23.441 2.41 33.33 42.50 43.02 4.92 47.62 72 66 69 49 53 66 375 POOR 11.4 10.46 to.94 7.77 8.40 o.461 59.43 19.20 17.0 18.40 1 3.07 14.13 17.60 70.59 66.67 57.50 6.98 54.08 52.38 TOTAL i02 99 120 so 98 126 631 16.16 16.69 19.02 13.63 18.83 19.97 100.00 STATISTIC OF VALUE PROS CHI-SQUARE 5 ii.77 0.041 LINELIHOOD RATIO CHI-SQUARE 3 11.785 0.038 MANTEL-ASNSZEL CHI-SQUARE I 10.333 0.001 PHI 0.135 CONTINGENCY COEFFICIENT 0.134 CRAMER'S V 0.135 SAMPLE SIZE a 631 -61- TABLE OF POVERTY AND THE HOUSEHOLD OWNERSHIP OF PADDY LAND POVERTY OWNERSHIP OF PADDY FREQUENCY PGAOKUT DotsS Raw POT NOT OWN OWNS COL POT PADDY PADDY1 TOTAL 0240 I 2030 1 8270 NOT*POOR 82.40 1 7.08 6.57 75.45 24.55 07.23 77.93 3042 S75 3617 POOR 25.59 4.84 30.43 84.10 IS .90 32.77 22.07 TOTAL 9201 2605 11887 78.09 21.91 100.00 STATISTIC OF VALUE PROB CHI-SQUARE I 110,014 0.000 LIKELIHOOD RATIO CHI-SQUARE I II5.069 0.000 CONTINUITY ADJ. CHI-SQUARE I 109.509 0.000 MANTEL-*AINSZEL CHI-SQUARE I 110.005 0.000 FISHSR*S EXACT TEST 4l-TAIL) 0.000 (2-TAIL) 0.000 PHI -.0.090 CONTINGENCY COEFFICIENT 0.090 CRAMER'S V -0.090 SAMPLE SIZE * 11887 TABLE OF POVERTY AND HOUSEHOLD OWNERSHIP OF LAND POVERTY OWNERSHIP OF LAND FREQUENCY ow PCT iOWS OS NOT COL POT LAND |OWN LANDI TOTAL 1889 6381 8270 NOT-POOR 1S.89 53.68 69.57 22.84 77.16 07.30 70.28 918 2690 3687 POOR 7.72 22.71 30.43 25.38 74.82 32.70 29.72 TOTAL 2807 9080 11887 23.61 76.39 100.00 STATISTIC OF VALUE PROS ****.*.****************.********-******..-..*.*.**.... CHI-SQUARE I 8.990 0 003 LIKELINOO RATIO CHI-SQUARS I 0.908 0.003 CONTINUITY ADJ. CHI-SQUARE 1 8.850 0.003 MANTEL-AENSZEL CHI-SQUARE I 8.989 0.043 FISHER'S EXACT TEST f1-TAIL) 0.002 12-TAIL) 0.003 PHI -0.028 CONTINGENCY COAFFICIENT 0.027 CRANS*S V -0.028 SAMPLE SIZE * 11087 -62- Тд61.Е Oi POVEpTY 8У OlSTд1CT РОVЕдТУ O1STRICT PдHOUENQY РЕдСЕNТ дОМ РСТ мUМАдА MAM9AN- 001. бСТ COlOM80 �ОАМРАИА �KAIUTARA� KANDY � MATAL � Е�1УА � вдllЕ � МАТдда � ТОТА � TOTAI . . .. . .. .. . . . .....» . . ...�....-�' "' "�' "• '�- -- '�--- • '•�•---• ^�• 'в. ...-"�"- ---•�--"....• NOT-РООд � я84•!12 � 880�48 � 880,8? � 447'28 � 148'98 � 378'11 � 44я'00 � з$а183 � 44�8�� �84544�1 ....... " �' " •�' '--�'- '--'в..-'-'-'�'-'- " -в. ' ' •�• "' '-'� " " ' "�•' '--'-• . ." "•' -. . . .. . . . . POOR � 847?08 � 839'84 � 339038 � 887я88 � 1S?.3� � 14Iя8я � 48�,00 � з4iо88 � 787�28 �7048s89 - "•'-ь....-.-.в..-....-�..•--•--в..-•- " -`�--•-- " -�" --•-..�..'•• "•�-- " -'••ь---'--••• тота� 1в1471о 1ееlsап 87заsз 10?8вs8 з7оояз ssвllo евявlв 7зв�а7 47о1sэ 1а4ввзов 11.74 s.4в 8.в8 в.в4 ?.зя з.4о s.вз а.7в з.о4 1оо.оо lсоитlмиЕ01 TABtE OF POYERTY ВУ OlSiд1CT � POYERTY 0lSТд1СТ FдEQUENCY РЕдСЕNТ ROM РОТ MULI.AI- ВАТТ1- TRtмCOM- KUдU- СО� PCt JAFFNA = hANNAд �VAYANIYA� Т1УА = CAIOA = АМРАдА � дl�Е S ИОliALA jPUTTALAM= TOTAL -�--••---+-.-...-•�•-•-•---1----....�.-....-•�'••-----�•-------�--.-..--�•....•-•�•'••••••t NOT•РООд� 482833 �4•38?08 �3388$g� �S184oYa � 147 27 � 3S3'�8 �8885384 � 7888�8 � 388$83 �8484441 ••••..'-^+--------�-.-...'•�•------•�•---^ •-•----•---+-•^ -.••�_••.•-••�--•-•^ •� ^ •••--•• РООд , з4??88 �78848я2 � 34 43 �28З8?7я1 � 18?g73 � 148у38 �894�338 � 843584 � 743837 �70484Sa '•----•-'," ^•.'-'� " " " '-�-•" •'.-в.-.....•�.•.....-�'•'•'•--'--' ^ '--'-'._...•�--'-'-•"1i♦ . tOTAL 8311td 117382 49408.8 8о822.9 388187 428314 188288 1297085 831314 18488308 8.38 0.78 0.3? 0.82 3.38 2.77 t.0i 8.Зя 3.44 100.00 tCONT1NUE01 TABIE Oi POYERTY ВУ :�STRICT РОУЕАТУ D1STд1CT FдE4uENCY РЕдСЕNТ дОМ РСТ А8УАА0- POLON- МОМА- RATNA- K1LL1- COL РСТ МАРUдА � NAдUMA �BAOULIA � RAбALA � РUдА �KEбAILE =NOCHCNI � TOTAL ---.• .............+•-•••---•^-•-••-••--•----*-----••-+-•-•-._..-.-._-•_• i 348в10 181448 389712 188яоя 442193 3718я7 1182я.1 840я4о1 е 4.1Z 2.fв 4.28 1.87 8.2в а.42 о.1а NOT•РООд i 54.18 � 83.?7 � 84.88 � 84.10 �$1.88 � 83.я7 � t00.00 84.41 •---•-•--t•••----•-+•-•---._•_.•--•--•--------•.._...•-•------_••-.___._.. - РООд � З48 84 � 'Звз78 � �4S?4i � 147?я0 � 448'84 � 347?ОЗ � 0,00 �7048s8a ---•^ • ..................в- --- •в..•-••••�••- ••--�-•••-...в•- -----• TOTAL вз988в 2888я1 в8я008 301178 882308 704137 11829.1 18488308� 4.11 1.88 4.48 1.я8 5.81 4.84 0.08 100.00 iдEQUENCY MISSINC т я848 -63- TABLE OF POVERTT 8У РАСЕ POYERTY RACE FдEQUENC� РЕдСЕNТ SRI- Sд1- дОМ РСТ S1N- LANKA 1ND1AN 4АМКА MAIAY 8UдGМЕд ОТМЕд COL РСТ NA4ESE � ТАМ(L � TAMlI � МООд � ( � � TOTAI •••••••••�•'••••••�••.••••-�••••--••в.»•-•.••�-••-•---�•-••-•••+•-----°-+ NOT•РflОд �835$Sя� г10$�0�� � 4�5198 � B4Z8� �29Q 8$g �2088$34 �2988.80 �84g14A1 "'••"'•�.'••.•.-1ij4�••••ь••♦•-•�.�•-�••••.•-.�-'••--•'�•'•r...в�..•w.•..� POOR �52�$T;8 � 84S'ОЗ � '28$0 � '$739g �'Т38808 �1543 88 �1433.00 �9048 89 ----___•..---- ---••--- •--+-•_•---•..----- -•-- -- -�- - --•- -----ii• тотаi ��ввазуа �яо�22о 8яовев �22з�4а лв�4t�s зs��s�в еззs.ев iво8sзов �8.4о I2.зо з.в2 �.яl о.зо о.2з о.оз �о0.оо FдEaUENCT MISSINC + 9845 , , -64- TABLE OF LITERACY AND POV9RTV' POVERTY LITERACY FREQUNCY PERO ROW POT IL- COL POT LITERATEILITERAT11 TOTAL 24808 1 4252 1 2sase NOT-POOR 8.50 57.69 '82:2? 14-73 $7.80 1 58.75 .................. POOR 17022 1 324; 1 2;183 311,13 8.41 4 .31 S4.69 15.31 4 2.14 43.25 .................. ........ TOTAL 42528 7493 50021 85.02 14.98 100.04 FREQUENCY MISSING 13908 STATISTIC OF VALUE PROS ................................................ CHI-SQUARE 1 3.228 0,072 LIKELIHOOD RATIO CHI-SQUARE 1 3.223 0.073 CONTINUITY ADJ. CHI-SQUARE 1 3.162 0.074 14ANTEL-RAINSZEL CHI-SQUARI 1 3.22$ 0.072 FISHER'S EXACT TEST 41-TAIL) 0.037 (2-TAIL) 0.074 PHI 0.008 CONTINGENCY COEFFICIENT 0.008 ORAMER'S V 0.000 EFFECTIVE SAMPLE SIZE a 50021 FREQUENCY MISSING x 13908 WARNING: 22% OF THE DATA ARE MISSING. TABLE OF CHILD LITERACY AND THE LITERACY OF THE HEAD OF THE HOUSEHOLD (CHILDREN ACED 6 TO 14) CHILD LITERACY LITERACY Of HEAD OF HOUSEHOLD FREQUENCY1 PERCENT I ROW POT I IL- COL POT LITERATEILITERATEI TOTAI: ................. 75706 SW .1 8:54; LITERATE 74 .36 .1 87:23 :2.77 92.09 1 72.99 .................. + ......... 799 ILLITL fil 84:8 1 432091 10.89 81.33 38.87 7.91 27-01 1 ; 7340 TOTAL 6 96 1 4 84.41 ls.59 100.00 FREQUENCY MISSING 3 11751 STATISTIC OF VALUE PROS ...................................................... CHI-SQUARE 1 363.252 0.000 LIKELIHOOD RATIO C"I-SQUARI 1 290.193 0.000 CONTINUITY ADJ. CHI-SQUARE 1 361.285 0.000 MANTEL-NAENSZEL CHI-SQUARE 1 363.202 01000 FISHER'$ EXACT TEST 41-TAIL) 01000 42-TAIL) 0.000 PHI 0.222 CONTINGENCY CO FFICIENT 0.217 CRAMER' 0.222 EFFECTIVE SA14PLE SIZE a 7340 FREQUENCY MISSING x 11751 WARNING: 82% OF THE DATA ARE MISSING. �. -65- TABLE OF SEOTOR АИО CN1L0 AOTlVITY АМОИб EMPLOYEO CNILOREN 1N NOT-POOR MOUSEHOLOS SЕОТОд NАТидЕ OF CNlLO'S АСТIУIГУ fcOMOPCTY�C �TONON�OULAYION#fАдМ1Nб �8RЕЕОаИб�ицNТtмбУ#NTURINQ � Тдд0Е � NOUSE �SERV1CESt TOTAL �����'���}���� " ��'�w�..� " ��" "� " �'��" �" �'�.w��'�������������..��������'�'� " 'f��.�.��.♦ иаеАИ ! о:оо � о:оо � о:о0о � о:оо � о:оо � о:оо i о:оо � о:оо �з143g$$ �2483gaë ��""'�'}�"•'•"'�"'•'•"�""•"'�'•"•"'*""�"'�"•'•"'�"""'.в-.�.....в•"""•!• NUдAE � 100000 �1A00.00 � ggg,qg � 0.00 �22 0 70 �88100400 � 237:8i � g00.00 �з188g48 jsя83.4З •"'•'•'•�•"••"_+�.���'•'.""•"'."'^"''�'••"""'•'^'•"�" "'�"""^�""" '`+ о о вs1.87 �о о о а7в��в2 о� о s2е.вз2 �ЕSТАТЕ � �:00 � 0.00 � 44.84 � 0�00 � 0.00 � 0.00 � 82.Зя � О.ОО � 0.00 � 13.10 "'•""'�•"""'�""..�..в.���."•�"'•""�""""г""""�'•"""�"'•""�"""`•• TOTAL 1890.4 1819.45 1014.48 0 221.8t8 888.447 784.172 828.11 888.092 7090.я4 22.43 28.88 14.31 0.00 3.13 8.30 t0.78 7.42 7.я8 100.00 fдEOUENCY M1SS1Nб • 1388327 tABLE OF SЕСТОд ANO CNllO АСТIУIТУ АМОИб EMPIOYED CN1LO1iEN 1И РООЧ ИОU5ЕЧ0 �5 SECTON ИдТцдЕ Of CM110'S АСТ1УlТУ FOOOLCPCTY�CUVTTNON�Oц�AT10N�FARMIN4 �BREEO1Nб�MUNTlNбY�MTURING � ТдА0Е � NOUSE �SERV(CES= TOTAI "•"""в.."'•'•�'•'•'•'•�'• " "в..••"••�•"""•�••"""�""•'••�"'._..�в�����_ �♦ 0 0 88 128? 0 О 411.887 0 б 78.0133 872.9яТ идsдИ � о:оо � о:оо � 1i:as � о:оо � о:оо � iв:�s � о:оо � 0'о� � io:sв � 4 73 -•-------•--------•--------•--------•--------•-•-•----• .................•----•---•------ • �зя78.ов 7sе.77э 1я71.я 477.а22 э10.язs 17в7.7в эл8.27 о 28в.sя� яя1з.7 pURAL � 100.00 � 5�.38 � 68.85 � t00.00 � 1 0.00 � 88.28 � 100.00 � 0.00 � 79.0� � 81.8я �"•"'�'�""•"��•""...s�.."•"�•""•"�""""�""""�""""�""•"•�""""• ESTATE � О.ОО j7148862 � я30.47 � 0.00 � 0.00 � О.ОО � О.ОО � О.ОО ' 0.00'�1813.зе ��..�.����в�'�""'��"""'�""��"�"""�'�""'•�'�""""��""'..iг-�....�.i�.�......� тотА� зятв.ое 1аТв.вs 2яs8.в� 477.а22 з10.я2в 21яs.в2 за8.2т о зв2.воз 121ов.2 з2.8з 12.г0 2а.44 з.я4 2.87 18.17 2.в8 о.оо з.оо 1о0.о0 FaE4. ИСУ M1SS1Nб = 2190978 -66- TABLE Of POVERTY AND LABOR FORCE PARTICIPATION POVERTY LABOR FORCE PARTICIPATION 'MUC!"Sly Out OF 1:000 LABOR L fi FORCE FORCE TOTAL 12932 14632 27 64 4OT-POOR 7.42 30 47 57459 47 * 2 52:91 53.27 62-07 04 1:!67 20224 POOR 23 7: 9 42.41 58.10 43.90 46.13 37-93 TOTAL 24 ; 77 23411 47688 50.91 49.09 100.00 FREQUENCY MISSING 16241 STATISTIC OF VALUE PROS ...... ..... ARE 1 378.301 0.000 LML INOOD RATIO CHI-SQUARE 1 378.978 0.000 CONTINUITY ADJ. CHI-S4UARt 1 377.941 0.000 10ANTIL-NAINSUL CHI-SQUARE 1 378.294 0.000 FISHER'S EXACT TEST (1-TAIL) 0.000 I 12-TAIL) 0.000 PHI -0.089 CONTINGENCY COEFFICIENT .:-089 CRAMER'S V 089 EFFECTIVE SAMPLE SIZE s 47668 FREQUENCY MISSING a 18241 WARNING: 25% Of THE DATA ARE MISSING. TABLE OF POVERTY AND REASONS FOR NOT BEING IN THE LABOR FORCE POVERTY RIAS M PON NOT BEING IN THE LABOR FORCE Fagauguoll PERCENT "SUO99' OSTIRE-INFIRM, too NOW PCT COL POT I'STUDIES I WORK I MINT JOISABLE YOUNG I OTHER I TOTAL --------- * ........ * ........ * ...... I ....... ... ........ --------- mc 1 6 4 '* 132 0: 12575 11-PCICII 141: 1 245!9: 7 nG 1. 73 0 .56 11 53.12 32.73 47.04 13.49 3.28 t .09 2.43 41.35 61.93 72.82 1 52.07 32.04 46.79 ................. ........ ........ .................. 384 1280 1 .. 00 60 36 to 1!4 POOR 24. IS 8 2131 1.54 487 46 a 3 2.60 32.76 S.70 3.26 .52 3.14 58.85 38.07 .27.18 47.03 07.90 $3.21 i;-TA,z ---- .... 3-2,9-*,-*-,7-74"*'*"4,12,-, ----- 54-" 23675 42.05 40-34 9.84 3.27 1.74 2.76 100.00 FREQUENCY MISSING 40254 STATISTIC OF VALUE PROB -: .................................................. CHI S41UARE 5 129a.232 0.000 IKGLIHOOD RATIO CHI-SQUARE 5 1320-256 10.000 MANTEL-HAINSUL CHI-SQUARE 1 91.299 0.000 PHI 0.234 CONTINGENCY COEFFICIENT -0-228 AWSII'S V 0.234 EFFECTIVE SAMPLE SIZE 8 23675 FREQUENCY MISSING a M54 WARMINGs 63% Of THE DATA ARE MISSING. -67- TABLE OF POVERTY AND'REASONS FOR NOT SEEKING EMPLOYMENT POVERTY REASON FOR NOT SEEKING EMPLOYMENT FREQUENCY ENGAGED PERCENT S"8ELIEVES HOUSE- IN NON* RoW PCT DISCOUR- NO WORK HOLD ECONOMIC COL POT I AGED AVAIL. DUTIES STUDIES ACTIVITY OTHER I TOTAL ...... . . . . . . . . . . . . . .. . . . V T T E O A 346 S 54 38 I, 121 429 NOT*POOR 17.46 4B7.08 6.46 3.71 .5 4.47 51 .32 34.03 s3.75 12.99 7.23 4.20 28.21 41.01 39.07 63.53 59.62 S5.43 62.05 . -- --*.. .. ........ . : 472 92 St 2 7 74 407 POOR 20.57 1.00 3.7j 2.55 2.03 8.85 48.68 42.2: 22.60 7362 5.16 4118 18.18 54.09 o ? 516 4. 54--09 609 36.4 40.38 48.57 37.9 TOTAL 318 ISO 85 52 35 195 836 38.04 18.08 10.17 6.22 4.19 23.33 100.00 FREQUENCY MISSING . 63093 STATISTIC OF VALUE PROS CHI-SQUARE 5 28.282 0.000 LIKELIHOOD RATIO CHI-SQUARE 5 28.526 0.000 MANTEL-HAENSZEL CHI-SQUARE 16.048 0.000 PHI 0084 CONTINGENCY COEFFICIENT 0.184 CRAMER'S V 0.184 EFFECTIVE SAMPLE SIZE * 836 FREQUENCY MISSING 9 63093 WARNING: 99% OF THE DATA ARE MISSING. TABLE OF POVERTY BY SOURCE OF ASSISTANCE WHILE UNEMPLOYED POVERTY SOURCE OF ASSISTANCE FREOUENCY GOVT 0R PARENTS, INCOM9 NO PERCENT CHARITAS FROM IPRESENT ASSIS- COL PC BL HLRNPOERTY SALEOF OCuPA* TANC COL PCT INSTIT'N RELAT'NS (ASSETS) SAVINGS ASSETS TION REQ*D TOTAL ------ BLE*C.ILDREN PROERT .** **** * SALE*** OF* * TANCE 33 1920 2 SS to toot 39 3086 NOT-POOR 0.54 .531 0.46 0.90 0.16 16.44 0.64 50.68 ol.0 62.22 TE0.9 .76 0.32 32 44 1.26 * 61... 48.87 60.67 47.41 37.04 1 3.64 17.47 --* ** *** ** *** +* ------***-* *-*-- ***** **** *--** ***** ***** **** -1-- - -- . ... . 2t 3 2009 18 1 17 66 12 3003 POOR 0.34 32.99 0.30 1.00 10.28 4.21 0.20 49.32 S 0.70 66.90 0.0 2.03 0.57 26.80 0.401 . . 38.89 S.5.!3 ..39.13 82.59 62.98 46.36 23.53 TOTAL 54 3929 46 ... 27 1860 SI 6089 0.89 64,53 0.76 1.91 0.44 30.85 0.84 100.00 FREQUENCY MISSING s 439b2 STATISTIC OF VALUE PROS CHI*SQUARE 8 32.063 0.000 LIKELIHOOD RATIO CHI-SQUARE 6 32.883 0.000 MANTEL-HAENSZEL CHI-SQUARE I 13.379 0.000 PHI 0.073 CONTINGENCY COEFFICIENT 0.072 CRAMER'S V 0.073 EFFECTIVE SAMPLE SIZE * 69 FREQUENCY MISSING 43932 WARNING: 88% OF THE DATA ARE MISSING. SOURCE OF ASSISANCB WILE UNEMPIDO Parents, Government Spouse, Icm frm or Charitable Cildren Property/ Liquidation Present No Assistance Institution ReLakives Assets savings of Assets owcuatign Saie All-Islard All 0.7 91.8 0.7 2.2 0.4 3.8 0.5 Non-Poor 0.9 91.3 0.5 2.8 0.5 3.0 0.8 Poor 0.5 92.4 0.8 1.5 0.3 4.5 0.1 At-Risk 0.5 92.9 0.5 0.9 0.3 4.9 0.0 Ultra 0.0 90.3 0.0 0.1 0.8 2.8 0.0 Urban All 0.8 92.1 0.5 2.2 0.5 3.3 0.6 Non-Poor 1.5 89.3 0.5 3.6 0.3 3.5 1.2 Poor 0.3 94.1 0.4 1.3 0.7 3.1 0.2 At-Risk 0.5 93.8 0.6 1.6 0.7 2.9 0.0 Ultra 0.0 94.3 0.0 0.0 2.4 3.3 0.0 Rural All 0.7 91.3 0.8 2.3 0.4 4.2 0.4 Non-Poor 0.8 91.3 0.6 2.7 0.6 3.1 0.8 Poor 0.5 91.3 1.0 1.7 0.1 5.4 0.0 At-Risk 0.5 92.0 0.6 0.5 0.0 6.4 0.0 Ultra 0.0 88.3 0.0 9.1 0.0 2.6 0.0 Estate All 0.7 99.0 0.0 0.3 0.0 0.0 0.0 Non-Poor 0.0 99.6 0.0 0.4 0.0 0.C 0.0 Poor 2.7 97.3 0.0 0.0 0.0 0.0 0.0 At-Risk 0.0 100.0 0.0 0.0 0.0 0.0 0.0 Ultra 0.0 0.0 .0..0 0.0 0.0 0.0 0.0 TABLE OF POVERTY BY INDUSTRY POVERTY INDUSTRY FREQUENCY PERCENT ELECTRIC. CON- WHOLESALE/ TRANS- FINANCE/ ROW PCT AGRI- FISHING FORESTRY/ WANU- CAS.AND STRUC- RETAIL PORT/ INSUR- SOCIAL NOT COL PCT CULTUREJ I MINING IFACTURE ISTEAN I TION I TRADE ISTORAGE I ANCE ISERVICESI DEFINEDI TOTAL -**--**------*** ** ** *-- - - --*---* *--** ** * * ------- -------- ------ * I 255! 65 .8 1435 63 55 1399 * 592 222 2151 256 12712 NOT-POOR 2822 1 .85 P.40 7.30 0.32 2.62 7.1 3.01 1.13 t0.94 1.30 64.64 43.GO 2.88 8.31 st29 0:.50 4.05 to.** 4.66 1.75 88.92 2.0 88.28 60.60 61.94 59.0 64.29 55.92 62.46 67.50 79.00 72.98 40.5? 2824 236 tot 994 35 406 841 285 59 797 375 6954 1..20 0.52 5.05 0.18 2.0614.28 1.451 0.30 4.05 1.91 35.36 POOR 40.61 3.39 #.47 14.29 0.50 5.64 12.09 4.10 0.85 11.46 5.39 33.72 30.40 38.06 40.92 35.71 44.08 32.54 32.50 23.00 27.04 59.43 ------*- * - - * * * * * ** - * * - - - - * * **-* * - * **0- -- + - - - *%-- - -- - - - - - - - *C TOTAL 8374 99 268 2429 98 921 2240 877 281 2948 631 19686 %3 42.58 3.05 1.36 12.35 0.50 4.68 11.39 4.46 1.43 . 14.99 3.21 100.00 FREQUENCY MiSSING = 44263 STATISTIC OF VALUE PROB CHI-SQUARE 10 360.986 0.000 LIKELIHOOD RATIO CHI-SQUARE 10 356.457 0.000 MANTEL-MAENSZEL CHI-SQUARE 1 0.040 0.842 -PHI 0.135 CONTINGENCY COEFFICIENT 0.134 CRAMER'S V 0.135 EFFECTIVE SAMPLE SIZE a 19666 FREQUENCY MISSING a 44263 WARNING: 69% OF THE DATA ARE MISSING. -- -70- TABLE OF POVERTY AND ROUND OF SURVEY AGRICULTURAL INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY PEROtNT I R0W POT COL POT I 1 21 31 41 51 01 TOTAL 88 1 173 930 1952 973 11933 1 5550 NOT-POOR 10.52 9.23 11.20 12.56 II.02 t.14 I66.26 Is.87 13.93 16.90 1.95 I 7.53 16.81 65.07 59.03 70.16 6.40 07.48 0.12 POR 473 S%9 39 48 49 48 22 5.65 6.20 4.76 5400 5.60 5.71 33.72 POO I 1 9 0 0 470 2024 16.75 1.38 14.13 17.21 10.01 86.93 34.93 40.17 29.84 31.60 32.52 33.88 TOTAL 34 1292 0337 1538 1442 141l 8374 16.17 15.43 19.97 18.37 17.22 16.85 100.00 STATISTIC OF VALUE PROD ****************.... **.*... **... ******. ***************** CHI-SQUARE 5 37.967 0.000 LIKELIHOOD RATIO CHI-SQUARE 5 37.535 0.000 MANTEL-AENSIZEL CHI-SOUARE 1 5.798 0.018 PHI 0.067 CONTINGENCY COEFFICIENT 0.067 CRAMER'S V 0.067 SAMPLE SIZE a 8374 TABLE OF POVERTY AND ROUND OF SURVEY FISHING INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY PERCEMT ROW POT COL POT I 1 21 31 41 51 6t TlirAL 30 S 80 54 41 911 363 NOT-POOR 6.0: 10.113.36 9.02 0.41I.19100.60 9.92 1.0 22.04 14.8 1.2925.07 48.65 50.41 67.23 60.07 6.13 67.91 38 60 39 35 21 43 236 POOR 34 10.02 6.5115.84 3.51 7.18 39.40 1.0 1 25.42 16.53 14.3 8.90 18.22 S1.35 49.59 32.77 39.33 33.87 32.09 TOTAL 74 121 119 89 62 134 599 12.35 20.20 19.87 14.8 10.35 22.37 100.00 STATISTIC OF VALUE PROS ****.***************************.****..************** CHI-SQUARE 5 15.668 0.008' LIKELIHOOD RATIO CHI-SQUA S 5 15.588 0.008 MANTEL-HAINSZEL CHI-SQUARE I 10.449 0.001 PHI 0.162 CONTINGENCY COEFFICIENT 0.160 CRAMER'S V 0.1862 SAMPLE SIZE * 599 -71- TABLE OF POVERTY AND ROUND OF SURVEY FORESTRY AND MINING INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY1 PERCENT ROW POT COL PCT it 21 31 41 .51 61 TOTAL 29 25 29 32 St 20 86 NOT-POOR 10.62 9.33 10.82 11.94 1 .7 7.48 81.94 17.47 .08 17.47 19.28 1?8, st.0S 59.16 55.58 67.44 78.19 62.00 1.28 20 20 14 30 19 19 102 POOR 11:48 7.48 5. 22 3.73 7.0 .9 80 19.56 19.01 13.73 9.80 18.83 18.63 40.82 44.44 32.5 23.81 38.00 48.72 TOTAL 49 45 43 42 50 39 268 18.28 16.79 16.04 15.67 18.86 14.55 100.00 STATISTIC OF VALUE Po CHI*SQUARE 5 6.986 0.222 LIKELIHOOD RATIO CHI-SQUARE 5 7.184 0.20? MANTEL-HAENSZIL CHI-SQUARE I 0.004 0.950 PHI 0.161 CONTINGENCY COEFFICIENT .0.19 CRAMER'S V 0.161 SAMPLE SIZE * 268 TABLE OF POVERTY AND ROUND OF SURVEY MANUFACTURING INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY PERCENT ROW POT COL PCr 11 21 31 41 at 61 TOTA 240 27 218 20 241 239 1435 NOT-POOR 9.88 6.03 8.97 11.53 9.92 9.84 59.08 16.72 15.12 15.9 19.51 16.79 3 o.86 54.55 55.22 1 58.13 67.80 59.65 59.18 200 176 357 133 63 365 994 POOR .23 7.256 .46 S.48 6.79.7 40.92 20.12 17.71 '5.79 13.38 18.401. 16.60 4.45 44.78 4.87 32.20 40.35 40.54 TOTAL 440 393 375 . t3 404 404 2429 15.11 16.18 15.44 17.00 16.63 18.63 100.00 STATISTIC OF VALUE PRO --- - *-* *** *** *** *** --* *** *** *** *** ** -*** CHSSQUAR 5 19.344 0.002 LIKELIHOOD RATIO CHI-SQUARE 5 19.654 0.001 MANTEL-AENSZEL CHI-SQUARE I 5.322 0.021 PHI 0.059 CONTINGENCY COEFFICIENT 0.089 CRAMER'S V 0.089 SAMPLE SIZE 2 2429 -72- TABLE OF POVERTY AND ROUND Of SURVEY ELECTRICITY, GAS AND. STEAM INDUSTRY POVERTY ROUND NUMBER (MONTH OP INTERVIEW1 FREQUINOV PERCENT ROW PCT COL PCT 1 21 31 41 SI 6| TOTAL 9 8 14 9 16 9 1 63 NOT-POOR 9.8 .512 14.29 9.18I 6.33 9.18 4.29 14.29 1.52 22.22 14.29 25.40 14.29 75.00 54.55 5Q.00 0.00 88.89 64.29 3 5 14 8 2 35 POOR 3.08 1 .t 14.29 0.12 2.04 W.0 35.71 8,0.57 I14.209 40.00 17.14 5.7j 1 4.29 25.00 45.45 50.00 40.00 s1.1 35.71 TOTAL 12 i 20 IS to £4 90 12.24 11.22 28.5? 15.31 10.37 14.29 100.00 STATISTIC OF VALUE PROS CHI-SQUARE 5 8.409 0.135 LIKELIHOOD RATIO CHI-SQUARE S 0.270 0.099 MANTEL-HAENSZEL CHI-SQUARE 1 0.018 0.366 PHI 0.293 CONTINGENCY COEFFICIENT 0.281 CRAMER'S V 0.293 SAMPLE SIZE x 90 TABLE OF-POVERTY AND ROUND OF SURVEY CONSTRUCTION INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEWO PERCENT ROW POTI COL PCT I 21 31 41 51 61 TOTAL 4 07 05 lt1 53 1 51 NOT-POOR 6.03 10.31 £0.53 10.31 10.97 5.75 55.92 1 4.37 18.45 £0.03 18a.45 819.61 0.29 48.54 48.47 88.90 65.07 60 48 49.07 1 85 lot 40 S 686 55 400 POOR 9.23 10.9? .21 5.54 7.17 5.07 44.00 20.94 24.088 it.82 2.58 1 6.26 13.55 53.46 S 1.53 33.10 34.93 39.52 1 50.93 1 TOTAL 159 £00 145 146 167 108 921 17.26 21.28 15.74 15.85 18.13 1.71 100.00 STATISTIC DF VALUE PROS CHI*SQUARE S 25.594 0.000 LIKELIHOOD RATIO CHI-SQUARE 8 25.000 0 000 :ANTEL-HAtNSZEL CHI-SQUARE 1 4.145 0.042 PH 0.167 CONTINGENCY COEFFICIENT 0 184 CRAMER'S V 0.1I? SAMPLE SIZE a 921 -73- TABLE OF POVERTY AND ROUND OF SURVEY WHOLESALE AND RETAIL TRADE. AND RESTAURANTS AND HOTELS INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY1 PERCENT I ROW PCT COL PCT i1 21 31 41 51 61 TOTAL 203 264 ' 268 249 214 205 1399 NOT-POOR 9.06 11.61 1t 96 t 9.55 9.5 6 2.46 14 SI 1858 19.16 17.80 15.30 14.65 55.18 83:.1 65.53 86.94 | 59.81 63.68 15 50 149 5123 14 117 841 POOR 737 6.70 .29 549 8.47 5.22 37.54 19.62 17.84 10.77 14.63 37.24 13.91 44.84 36.59 ,3447 3306 40.39 a 36.34 TOTAL 368 410 409 372 359 322 2240 16.43 18.30 18 28 IO 6 16.03 14.38 100 00 STATISTIC DF VALUE PROS CHI'SOUARE 5 14.775 0.011 LIKLLIHOOO RATIO CHI-SQUARE 5 14.663 0 012 MANTEL-HAENSZEL CHI-SQUARE 1 2.279 0.131 PHI 0.081 CONTINGENCY COEFFICIENT 0 081 CRAMER'S V 0.081 SAMPLE SIZE a 2240 TABLE OF POVERTY AND ROUND OF SURVEY TRANSPORT. STORAGE. AND COMMUNICATION INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY1 PERCENT ROW PCT COL PCT 11 2, 31 4, 5 1 TOTAL 83 -99 I 5 100 * 96 1 99 592 -POOR 9 486 11.29 3 5 1.40 to.95 11 29 67 50 14.02 16.72 1943 16 89 18.22 16.72 63.85 I 68.75 169 28 74 07 64 43 I 64.71 *******-**------- -**--*--****----*---**-*------------------ a 47a 45 S1 35 6.53a 54 285 POOR 5.38 5 13 1 582 3.99 8 04 6 I6 32.50 16:49 15.79 17 89 12 28 a 5 86 I 18.95 36.15 * 31.25 ; 30.72 25.93 35 57 35:29 TOTAL 130 144 '66 135 149 153 877 14.82 t8.42 18 93 15.39 i6 99 17 45 100 00 STATISTIC OF VALUE PROS CHI-SQUARE 5 4 977 0 419 LIKELIHOOD RATIO CHI-SQUARE 5 5 057 0 409 MANTEL-HAENSZEL CHI-SQURE I 0 053 0.817 PHI 0 075 CONTINGENCY COEFFICIENT 0 075 CRAMER'S V 0 075 SAMPLE SIZE - 877 -74- TABLE OF POVERTY AND ROUND OF SURVEY FINANCING, INSURANCE, REAL ESTATE, AND BUSINESS SERVICES INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FAEQUSNCY PERCENT ROW POT COL POT I( 21 31 41 51 61 TOTAL 28 35 42 44 42 31 222 NOT-POOR 0.98 12.40 1 4.95 S.60 14.39 81.03 79.00 12.61 IS.77 18.92 1 9.82 18.92 13.96 84.85 76.09 79.25 80.00 84.00 70.45 S1 15I Sit 8 13 S9 POOR j .7 3.95 3.91 3.95 2.85 I 4.63 21.00 47 8.64 18.64 8.64 53.56 22.03 2.15 3.91 20.75 20.00 56.00 29.55 TOTAL 33 46 53 55 50 44 281 55.74 16.37 18.86 19.57 57.79 15.66 500.00 STATISTIC DF VALUE PROS CHI-SQUARE 5 3.641 0 802 LIKELIHOOD RATIO CHI-SQUARE 5 3.581 0.611 MANTEL-HAENSZEL CHI-SQUARE 1 0.5609 0.451 PHI 0.554 CONTINGENCY COEFFICIENT 0.153 CRAMER'S V 0.114 SAMPLE SIZE a 281 TABLE OF POVERTY AND ROUND OF SURVEY COMMUNITY. SOCIAL, AND PER%ONAL SERVICES INDUSTRY POVERTY ROUND NUMBER (MONTH OF INTERVIEW) FREQUENCY PERCENT ROW POT COL POT It 21 31 41 51 Of TOTAL .. ... .. . ..... .. .... .. . .. .. .... .. . 322 340 1373 406 384 356 2St NOT-POOR 50.92 t.S3 12.65 13.77 12. 0 52.0 72.96 14.97 15.81 t7.34 58.87 16.46 56.5 70.93 71.88 73.43 73.02 72.54 75.91 ............... .. ........ ....... ........ ........ ........- - !32 533 53 550 34 513 797 POOR 4.48 4.51 4.58 5.09 4.58 3.83 27.04 16.56 16.60 56.94 58.62 1.81 4.18 29.07 28.12 26.57 26.98 27.46 24.09 .... ..... ... . .. - - - TOTAL 4S4 473 SQ0 556 488 469 2948 15.40 56.04 17.23 18.86 18.85 15.91 100.00 STATISTIC DF VALUE PROS CHI-SQUARE 5 3.398 0.639 LIKELIHOOD RATIO CHI-SQUARE 5 3.429 0.634 MANTEL-HAEUSZEL CHI-SQUARE 1 2.317 0.28 PHI 0.034 CONTING$NCY COEFFICIENT 0.034 CRAMER'S V 0.034 SAMPLE SIZE * 2048