As LSM - 81 L.SJflS JULY 1991 Living Standards Measurement Studv Working Paper No. Si Patterns of Aging in Thailand and Cote d'Ivoire Angus Deaton and Christina H. Paxson LSMS Working Papers No. 10 Reflections on the LSMS Group Meeting No. 11 Three Essays on a Sri Lanka Household Survey No. 12 The ECIEL Study of Household Income and Consumption in Urban Latin America: An Analytical History No. 13 Nutrition and Health Status Indicators. Suggestions for Surveys of the Standard of Living in Developing Countries No. 14 Child Schooling and the Measurement of Living Standards No. 15 Measuring Health as a Component of Living Standards No. 16 Procedures for Collecting and Analyzing Mortality Data in LSMS No. 17 The Labor Market and Social Accounting: A Framework of Data Presentation No. 18 Time Use Data and the Living Standards Measurement Study No. 19 The Conceptual Basis of Measures of Household Welfare and Their Implied Survey Data Requirements No. 20 Statistical Experimentation for Household Surveys: Two Case Studies of Hong Kong No. 21 The Collection of Price Data for the Measurement of Living Standards No. 22 Household Expenditure Surveys: Some Methodological Issues No. 23 Collecting Panel Data in Developing Countries: Does It Make Sense? No. 24 Measuring and Analyzing Levels of Living in Developing Countries: An Annotated Questionnaire No. 25 The Demand for Urban Housing in the Ivory Coast No. 26 The C6te d'Ivoire Living Standards Survey: Design and Implementation No. 27 The Role of Employment and Earnings in Analyzing Levels of Living: A General Methodology with Applications to Malaysia and Thailand No. 28 Analysis of Household Expenditures No. 29 The Distribution of Welfare in Cote d'Ivoire in 1985 No. 30 Quality, Quantity, and Spatial Variation of Price: Estimating Price Elasticities from Cross-Sectional Data No. 31 Financing the Health Sector in Peru No. 32 Informal Sector, Labor Markets, and Returns to Education in Peru No. 33 Wage Determinants in C6te d'Ivoire No. 34 Guidelines for Adapting the LSMS Living Standards Questionnaires to Local Conditions No. 35 The Demandfor Medical Care in Developing Countries: Quantity Rationing in Rural Cote d'Ivoire No. 36 Labor Market Activity in C6te d'Ivoire and Peru No. 37 Health Care Financing and the Demand for Medical Care No. 38 Wage Determinants and School Attainment among Men in Peru No. 39 The Allocation of Goods within the Household: Adults, Children, and Gender No. 40 The Effects of Household and Community Characteristics on the Nutrition of Preschool Children: Evidence from Rural C6te d'Ivoire No. 41 Public-Private Sector Wage Differentials in Peru, 1985-86 No. 42 The Distribution of Welfare in Peru in 1985-86 No. 43 Profits from Self-Employment: A Case Study of C6te d'Ivoire No. 44 The Living Standards Survey and Price Policy Reform: A Study of Cocoa and Coffee Production in Cote d'Ivoire No. 45 Measuring the Willingness to Pay for Social Services in Developing Countries (List continues on the inside back cover) Patterns of Aging in Thailand and C8te d'Ivoire The Living Standards Measurement Study The Living Standards Measurement Study (LSMS) was established by the World Bank in 1980 to explore ways of improving the type and quality of house- hold data collected by statistical offices in developing countries. Its goal is to foster increased use of household data as a basis for policy decisionmaking. Specifically, the LSMS is working to develop new methods to monitor progress in raising levels of living, to identify the consequences for households of past and proposed gov- ernment policies, and to improve communications between survey statisticians, an- alysts, and policymakers. The LSMS Working Paper series was started to disseminate intermediate prod- ucts from the ISMS. Publications in the series indude critical surveys covering dif- ferent aspects of the Lsw data collection program and reports on improved methodologies for using Living Standards Survey (LSS) data. More recent publica- tions recommend specific survey, questionnaire, and data processing designs, and demonstrate the breadth of policy analysis that can be carried out using iss data. LSMS Working Paper Number 81 Patterns of Aging in Thailand and Cote d'Ivoire Angus Deaton and Christina H. Paxson The World Bank Washington, D.C. Copyright 0 1991 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing July 1991 To present the results of the Living Standards Measurement Study with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. Any maps that accompany the text have been prepared solely for the convenience of readers; the designations and presentation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries concerning the legal status of any country, territory, city, or area or of the authorities thereof or concerning the delimitation of its boundaries or its national affiliation. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to Director, Publications Department, at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to photocopy portions for classroom use is not required, though notification of such use having been made will be appreciated. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of subjecs,% authors, and countries and regions. The latest edition is available free of charge from the Publications Sales Unit, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'Iena, 75116 Paris, France. ISSN: 0253-4517 Angus Deaton is professor of economics and Christina H. Paxson is assistant professor of economics and public affairs, both at Princeton University. Library of Congress Cataloging-in-Publication Data Deaton, Angus. Patterns of aging in Thailand and C6te d'Ivoire / Angus Deaton and Christina H. Paxson. p. cm. - (LSMS working paper, ISSN 02534517; no. 81) Includes bibliographical references. ISBN 0-8213-1865-9 1. Aged-Thailand. 2. Aged-Ivory Coast. 3. Cost and standard of living-Thailand. 4. Cost and standard of living-Ivory Coast. 5. Household surveys-Thailand. 6. Household surveys-Ivory Coast. I. Paxson, Christina H. II. Title. m. Series. HQ1064.T5D43 1991 305.26'09593-.dc2O 91-24154 CIP ABSTRACr This paper is broadly concerned with the living standards of older people in two contrasting developing countries, C8te d'Ivoire and Thailand. We use a series of household surveys from these two countries to present evidence on factors affecting the living standards of the elderly: living arrangements, labor force participation, illness, urbanization, income and consumption. One of the issues we examine is whether life-cycle patterns of income and consumption can be detected in the data. The fact that few of the elderly live alone makes it difficult to accurately measure the welfare levels of the elderly, or to make statements about the life-cycle patterns of income and consumption of individuals. We find that labor force participation and individual income patterns follow the standard life-cycle hump shapes in both countries, but that average living standards within households are quite flat over the life-cycle. The data presented suggest that changes in family composition and living arrangements of the elderly are likely to be more important sources of old- age insurance than asset accumulation. v ACKNOWLEDGEMENTS We are grateful to the World Bank for providing data and other support; this is part of a broader research project on household saving behavior in Thailand and CMte d'Ivoire. We should also like to thank Noreen Goldman, Robert Hecht, Valerie Kozel, Anne Pebley, and Charles Westhoff for useful discussions. vi CONTENTS Introduction ........................................................ 1 1. Individual characteristics and age ................................. 4 1.1 Sample data and population characteristics ................................. 4 1.2 Living arrangements ................................. 13 1.3 Education, labor supply, and health status ................................. 14 1.4 Levels of living ................................. 19 2. Household life-cycles ................................. 31 3. Conclusions ................................. 40 4. References ................................. 45 LIST OF FIGURES AND TABLES Figure 1 (a) Age Distribution: Bangkok 1981 ..................................... 6 (b) Sex Ratio: Bangkok 1981 ......................................... 6 (c) Age Distribution: Thai Villages 1981 .................................. 6 (d) Sex Ratio: Thai Villages 1981 ...................................... 6 Figure 2 (a) Age Distribution: Bangkok 1986 ..................................... 7 (b) Sex Ratio: Bangkok 1986 ......................................... 7 (c) Age Distribution: Thai Villages 1986 .................................. 7 (d) Sex Ratio: Thai Villages 1986 ...................................... 7 Figure 3 (a) Age Distribution: C8te d'Ivoire 1985 .................................. 8 (b) Sex Ratio: C8te d'Ivoire 1985 ...................................... 8 (c) Age Distribution: C6te d'Ivoire 1986 .................................. 8 (d) Sex Ratio: C8te d'Ivoire 1986 ...................................... 8 Table 1.A Age Distribution by Sex: 55 and Over, Thailand 1981 and 1986 ................... 10 Table 1.B Age Distribution by Sex: 55 and Over, C8te d'Ivoire 1985 and 1986 ................. 10 Table 2 Urbanization and the Elderly: C6te d'Ivoire and Thailand ....................... 12 Table 3.A Marital Status of the Elderly: Thailand 1981 ................................ 12 Table 3.B Marital Status of the Elderly: C8te d'Ivoire 1986 ............................. 12 Table 4.A Living Arrangements of the Elderly: Thailand 1981 and 1986 ..................... 15 Table 4.B Living Arrangements of the Elderly: C8te d'Ivoire 1986 ........................ 16 Table 5.A Educational Attainment by Age, Sex and Location: Thailand 1981 .................. 16 Table 5.B Educational Attainment by Age, Sex and Location: C8te d'Ivoire 1986 ............... 16 Table 6.A Labor Force Participation and Work Hours, by Age Category: Thailand 1981 .... ....... 18 Table 6.B Labor Force Participation and Work Hours, by Age Category: C6te d'Ivoire 1986 ... ..... 18 Table 7 Health and Sickness by Sex and Age: C8te d'Ivoire 1986 ........................ 20 Table 8 Average Household Characteristics by Age of Household Members: C6te d'Ivoire 1986 ..... 23 Table 9 Average Household Characteristics by Age of Household Members: Rural C8te d'Ivoire 1986 . 24 Table 10 Average Household Characteristics by Age of Household Members: Bangkok 1981 ... ..... 25 Table 11 Average Household Characteristics by Age of Household Members: Rural Thailand 1981 ... . 26 Table 12 Distribution of Income, by Age, Sex and Location: Thailand 1981 .................. 28 vii Table 13 Income Composition and Living Arrangements: Thailand 1981 .................... 30 Table 14.A Age and Sex Composition of Household Heads: Thailand 1981 .................... 32 Table 14.B Age and Sex Composition of Household Heads: Cote d'Ivoire 1986 ................. 32 Table 15 Members, Income, Expenditure, and Assets by Head's Age: COte d'Ivoire 1985 and 1986 .... 33 Table 16 Regressions of Income, Consumption, and Assets on Household Composition: C6te d'Ivoire 1986 .............................................. 36 Table 17 Members, Income and Expendiure by Head's Age: Thailand 1981 .................. 37 viii Introduction This paper presents and discusses some facts about older people in two contrasting developing countries, CMte d'Ivoire and Thailand. We shall be concerned with standard questions in the aging literature, namely demographic structure, living arrangements, urbanization, illness, labor force behavior, and economic status. In this paper, we shall go little beyond the presentation of data from a series of household surveys from the two countries. Although recent years have seen an increased attention in the demographic and sociological literatures to questions of aging in LDC's, data are still relatively scarce, particularly for Africa, and we see our current task as providing stylized facts to help focus further discussion. There are two research issues that provide the structure for our discussion; household saving behavior, and, more broadly, the economics of aging in countries with low living standards but with rapidly expanding shares of old people in the population. Research on savings behavior in the United States, Japan, and Western Europe has been dominated by permanent income and life-cycle models since their introduction in the 1950's. There has been a good deal less work on household saving behavior in LDC's, and much of the work that has been done has simply transferred the analytical framework from the more to the less developed context. It is not clear that this is the best way of proceeding. While it makes sense to work with the same basic ideas, that saving can smooth consumption over time, and that assets provide a measure of insurance against an uncertain future, there are important differences in environment and in mechanisms, so that the same aims may be achieved in very different ways. A much larger share of the population in developing countries is engaged in agriculture, where incomes are very variable, and there are many poor people close to subsistence, so that consumption insurance may be of the greatest importance. Household size is typically larger in poorer countries. Extended families, or even simply large households, may play many of the roles that are performed by asset markets in more developed economies, so that, for example, wealthy older men may acquire additional young wives as an alternative to an annuity. At the same time, the internal organization of the family and its living arrangements are intimately tied to patterns of inheritance, so that the means of transferring assets from one generation to the next will themselves vary with household structure. Age composition within very large households may not vary very much over time, so that main motive for saving becomes the protection of living standards from short-term covariate risk, and has little to do with transferring resources between generations or between widely separated time periods. One of the issues we examine in this paper is the extent to which there are clearly defined economic and demographic characteristics of households that vary systematically with the ages of their members, particularly characteristics that are likely to provide motives for saving. A number of broader issues have been raised in the literature on aging in LDC's, and these also play a role in shaping our discussion. The dominant demographic fact for LDC's is the effect of the demographic transition on raising the fraction of old people in the population. In Thailand, where the demographic transition is largely complete, the share of over 60's in the population, which was 6.2% and 5.7% in 1960 and 1985 respectively, is expected to rise to 11.9% in 2020, United Nations (1986), figures which are repeated in much of South, South East, and East Asia, as well as in Latin America. United Nations (1987) lists 3.8, 3.5, 5.1 and 4.3 as the percentages aged 65 and over for these four regions in 1980, whereas the estimated figures for 2000 are 4.8, 4.6, 7.8 and 5.2, rising in 2025 to 8.2, 8.3, 13.3, and 8.3. In Africa, wvhere by contrast, there has been little decline in the rate of population growth, the percentages aged 65 and over are 3.1, 3.0, and 3.9 in 1980, 2000, and 2025. The two countries discussed in this paper are good examples of these two contrasting cases. It is also important to note that life-expectancy for older people in LDC's is high, and although not as high as in Japan or the United States, the difference is much smaller than the corresponding differences at birth. Life expectancy at birth in North America is 72.4 years for men, and 80.1 years for women, and at age 60, men can expect to live for 17.8 years and women for 21.8 years. In South Asia and Africa, respectively, life expectancy at birth is 59.4 and 54.1 for men, and 60.2 and 57.4 for women, while at age 60, the figures are 15.1 and 14.3 for men, and 16.3 and 15.9 for women, see Treas and Logue (1986) for these and other figures. Since women live longer than men, higher life-expectancy for all tends to exaggerate the predominance of women over men in the population, so that the ratio of males to females tends to decline with the level of development. In the more developed countries in 1980, there were 62 males per 100 females aged 65 and over, compared with 82 per 100 in Thailand and 80 per 100 in China, and in parts of South Asia where there is excess mortality among women, there are more men than women in the older age groups, see Martin (1988). Several West African countries also show a predominance of men over women, USAID (1982). The growing relative importance of the elderly, particularly in Asia, has led to an increased academic and policy debate mirroring much of the earlier debate in more developed economies. Two excellent reviews are provided by Treas and Logue (1986), and for Asia, Martin (1988). One of the dominant themes of this debate is the contrast between the status of the elderly in more and less developed countries. There are extreme idealized versions of both types of 2 societies. To some, the extended family provides insurance for old-age, unemployment, and sickness, as well as an environment in which the elderly are an integrated, useful and respected part of their families. This is seen as a stark contrast to the "Western" treatment of the old, whereby they are unproductive, isolated, and institutionalized, with social insurance providing only a poor substitute for family insurance. Cowgill (1974, 1986, Chapter 8) sees the victimization of the elderly as a natural concomitant of development, with education, urbanization, and technical change as "processes that strip the old of claims to respect, power, and independence," Treas and Logue (1986, p.666). To others, the security of the extended family is a romanticized myth that appeals mostly to those who have long escaped the grinding poverty, poor health conditions, and low life-expectancy with which it is typically associated. One person's isolation is someone else's individual freedom. It is perhaps not surprising that Asian policy makers, faced with the prospect of rapidly increasing absolute and relative numbers of old people, view Western systems pensions, social security, and public geriatric care with a mixture of envy and alarm. These "big" questions of the effect of development on the status of the elderly are not sufficiently well posed to be amenable to serious empirical evaluation. Nevertheless, good work has been done on more specific issues, particularly on the living arrangements of the elderly. Martin (1989) reviews a number of studies of Asian populations which suggest that the proportion of the elderly living with their children, although still high (typically between 70% and 80%) is declining over time, with a corresponding increase in the numbers living alone, a pattern that is consistent with a move towards living arrangements such as those in the U.S., where only 15 percent of the elderly live with their children. T-he rest of the paper is organized as follows. Section 1 is concerned with individuals, and reviews demographic characteristics and living arrangements for elderly people in C6te d'Ivoire and Thailand. It also presents data on urbanization, on health, on labor force parti- cipation and hours worked, and, to the extent possible, on levels of living. Section 2 is concerned with households, and looks for "life-cycle" type patterns in household size, income, and consumption patterns in relation to the ages of household members. Section 3 summarizes and concludes. 3 1. Individual characteristics and age 1.1 Sample data and pputon The data presented in this paper come from two series of household surveys from CMte d'Ivoire and Thailand. CBte d'Ivoire is listed by the World Bank (1989) in the lower middle-income division of its middle-income category, with per capita GNP in 1987 of $740, which grew at an annual average per capita growth rate of 1.0% from 1965-87. Its population in 1987 is estimated as 11.1 million, and grew at an annual rate of 4.2 % during both 1965-80) and 1980-87. The crude birth rate per thousand was 52 in 1965 and 51 in 1987, while life expectancy at birth in 1987 was 52 years. Thailand has a similar GNP of $850, but has had much faster growth, averaging 3.9% from 1965-87. If these figures can be taken seriously, the average Thai was 280% richer in 1987 than in 1960, as opposed to an increase of only 30% for Ivorians over the same period. Whatever the precise magnitude, young Thais are now very much better off than were their parents, either in terms of lifetime resources, or in terms of income at the same age, and this is much less true for young Ivorians. There were 53.6 million Thais in 1987, with a life expectancy at birth of 64 years. The population growth rate was 2.9% from 1965-80, 2.0% from 1980-87, and is projected to be 1.5% from 1987- 2000; the crude birth rate per thousand fell from 41 to 25 between 1965 and 1987. The Ivorian surveys are the Living Standards Surveys of 1985 and 1986, collected by the Department of Statistics of C8te d'Ivoire with the technical and analytical support of the World Bank. The survey design is described in Ainsworth and Mufnoz (1986), and is a non-traditional one, carried out on a simple random sample of 1600 households in each of the two years, with 800 households common to both surveys. Although the number of households is small compared with traditional designs, there are around 14,000 individuals in each of the two surveys. The emphasis is less on large sample size than on the collection of comprehensive data for each household, so that interlinkages between different economic and activities can be studied. The Thai surveys are the two Socioeconomic Surveys of the Whole Kingdom, collected by the National Statistical Office in the two years 1981 and 1986. These surveys are more like the traditional household income and expenditure surveys, they have sample sizes in excess of 12,000, they have no panel element, and there is less detailed information about many of the activities covered in the Living Standards Surveys. Even so, for the purposes of this paper, the two sets of surveys provide roughly equivalent information. There are earlier household surveys for Thailand which could be used to examine the same issues over a longer time period. However, after the 1975 survey, the definition of the 4 household was changed so as to exclude sub-unit households so that, for example, in 1975 a married son and his wife living with parents would have been included as part of the parents' household, but not in later surveys. As a result, it is not possible to make consistent comparisons about living arrangements over the two types of survey. This seemingly technical issue points to a deeper problem in the measurement of household structure in Thailand and, indeed, in developing countries in general. To quote Cowgill (1986, p.70), "In Thailand, however, the term household is somewhat elusive and ambiguous. The climate is semi- tropical, and a great portion of one's life is spent out-of-doors. To a very great extent, this includes cooking, eating, and visiting. Thus the physical structure of the home is little more than a bedroom situated within a compound, while the cooking, eating, bathing, visiting, and even much of the working takes place in the compound rather than in the physical structure of the home. Hence when we say that the young married couple lives with the parents of one of them, the young couple usually sleeps in a separate structure within the parental compound. This usually involves common cooking and eating facilities, but this too is flexible, especially since eating is more of an individual matter and less often a scheduled group activity. Westem definitions of household membership are not easily applied in this type of society." These issues must be constantly borne in mind when interpreting the figures given below. In particular, the "new" treatment of the household in the Thai surveys is likely to overstate the degree to which people live either alone or in small groups, and to understate household size. By contrast, the Ivorian survey used a more inclusive concept of the household, and tended to include subunits if they lived in the same compound. As a result, household size in the surveys is larger than household size in the 1975 Ivorian census, and the biases may be in the opposite direction from those in Thailand. Figures 1 and 2 show the age pyramids and sex ratios for Bangkok and for rural Thailand in 1981 and 1986, while Figure 3 provides the same information for CMte d'Ivoire. For most of the paper, we shall follow this practice of showing data for Bangkok and for rural Thailand, rather than for the more conventional urban-rural split. Bangkok contains nearly 70% of Thailand's urban population, and while the survey also collects data on other urban and semi-urban (sanitary district) data, these seem sufficiently different from Bangkok to merit separate treatment. In order to avoid a three, or possibly four way split for each table, we compromise with two. For the same reason, and when it is not misleading to do so, we shall normally present data from the 1986 Ivorian survey. On balance, the data from the second year are probably of somewhat higher quality. There are a number of problems with the Ivorian data that are apparent from the figures, although none are particularly serious for older people. There is very pronounced peaking at five year age intervals, particularly in 1985, and particularly among females. Such effects are not uncommon among uneducated populations, see for example Ewbank (1981, pp. 66-68) and are 5 Figure la Figure 1 b _ Age distribution: Bangkok 1981 Sex ratio: Bongkok 1951 9 0~~~~~~~~~~ o 1A CD 2~~~~~~~~~~~~~~ .n MaleJs@ ,igr Femles e . o 0 O AgMae dstrbto:Ta Femlles 18 e ai:Ta ilgs18 0 _ o~~~~~~~~~~~~~~~~~~~~~~~~~~o 4 E o0 10 20 30 0 50 60 70 8 o . -~~~~~~~~ I - 0 o /~~~~~~~~7 -1.6 -1.2 -0.8 -0.4 -0.0 0.4 0.8 1.2 1.6 2.0 age percent of population 6 Figure 2a Figure 2b 8 Age distribution: Bangkok 1986 Sex ratio: Bangkok 1986 8 ~~~~~~~~~~~~~~E43 O XMles Females 0 ~~~~~~~~~0 s SF- $o \ 1\~~~~~~- 00 0~~~~~~~~~~~~ ~0 10 20 ~0 40 50 610 70 30 oL 1.6 I.2 0.8 0.4 0.0 0.4 0.8 t.2 1.6 2.0 percent of population age Figure 2c Figure 2d Age distribution: Thai villages 1986 A 0. O' E, - 0 0 1 4)~~~~~~~~~~~~~~~~~~~~~~~~~~~4 0 0)~~~~~~~~~~~~~~~~~~~~~~~~~C 0 FemoI.g~~~~~~~~~~O 1 60 4 5 o 0 8 -1.6 1.2 O. 0.4 0.0 o.4 0.a i.2 1.6 2.0 age percent of population '0' 00~ ., 0" - ',70 o , _S,,0) 10 2 30 4 50 6 70 8 -,~~~~~~~~~~~~~~~ Figure 3a Figure 3b g Age distribution: Cote d'lvoire 1985 Sex ratio: Cote d'tvoire 1985 to. 0 0Q 0 E , 0~~~~~~~~~~~~~ N * Males 2 L Females -r /\ / \ 8 0 eo i0 a O \ 6 0° '0'~-' ~ '. E l _ -~~~~~0 __ __ __.__ __ _ __- Cv '00 1 0 20 230 40 50 60 70 SC 1.6 1.2 0.8 0.4 0.0 0.4 0.8 1.2 1.6 2.0 age percent of population 9 Figure 3c Fgsure 3d 8 Age distribution: Cote d'lvoire 1986Se rao:Ctdlore 98 0 o Mcies ,, Females Q-~ o 0--gN .0 .c ./ i. -|- c 1.6 1.2 0.8 0.4 .0 0.4 0.8 1.2 1.6 2.0 0 lO 20 30 40 50 60 70 80~~~~~~~~I. a ~ ecn of4pu,0o g O 0.0~~~~~~~~~ typically correlated with low education and low incomes; the 1975 Ivorian Census shows similar effects, see Ahonzo, Barrere, and Kopylov (1984, p.9). In 1986, interviewers placed less reliance on reported figures, and acquired more supporting information, and the problem is considerably reduced. Even so, it is wise not to make much of the precise age estimates, and to work instead with five or ten year age brackets. More serious is an apparent undercount of prime age males; in 1986 the sex ratios (males per 100 females) in the age groups 20-24, 25-29, 30-34, 35-39, 40-44, 45-49, and 50-54, are, respectively, 85.1, 74.2, 83.2, 65.3, 60.7, 78.9. and 77.6, with between 600 and 200 people in each sex-age cell, see Figures 3b and 3d. Neither we nor the World Bank currently have any explanation for these results. Again, there are similar, although not identical problems with the 1975 Census, and other demographic surveys, as well as with census data in other African countries. Ahonzo, Barrere and Kopylov (1984, Chapter 5), find similar patterns in surveys carried out in the early sixties, as well as in the 1975 Census once the predominantly young, male non-Ivorian immigrants (22% of the population) are removed. Although the Living Standards Surveys include non-nationals, there are only 14% in the two surveys in 1985 and 1986. Since it is the same age group missing in data ten years apart, and since there is no large-scale emigration from Cote d'Ivoire, the problems must come from - measurement errors, possibly in connection with the many prime-age males in the cities, where they are hard to count or survey. Respondents may also exaggerate their ages, and it is possible that men do so more often or by more than do women. The figures for the fractions of old people appear to be consistent with those from the Census, again see Ahonzo et al (1984), and USAID (1982); note that once again differential overreporting of age by men may account for at least part of the apparent excess of older men. Tables 1.A and 1.B show the fractions of people, by sex and urbanization, who are aged over 55 in both Thailand and CMte d'Ivoire. We have chosen the young cutoff age of 55 because, particularly in a young, rapidly growing population such as that in CMte d'Ivoire, there are relatively few old people. In the 1985 Ivorian sample as a whole, there are 994 individuals aged 55 or over out of 13,271 people in all, or 7.5% (in 1986, 1,046 out of 12,896, or 8.1%). For Thailand, the urban sector is relatively oversampled, so that when the appropriate weights are applied, the survey shows fractions 55 and over of 9.9 in 1981; weights for the 1986 survey are not currently available. National Economic and Social Development Board (1985) gives a lower figure of 7.83% over 55 in 1980. This publication notes a tendency for Thai survey data to underestimate the numbers of children under the age of 10 and this may explain some of the discrepancy. The two estimates for Cote d'Ivoire, which are only one year apart, provide some 9 TABLE 1.A AGE DISTRIBUTION BY SEX: 55 AND OVER THAILAND 1981 AND 1986 BANGKOK RURAL FEMALES MALES FEMALES MALES AGE: 1981 1986 1981 1986 1981 1986 1981 1986 55-59 3.2 3.5 3.5 3.0 3.3 3.8 3.0 3.4 60-64 2.3 2.3 2.1 2.2 2.5 2.6 2.1 2.5 65-69 1.6 2.1 2.0 1.3 1.8 1.9 1.5 1.8 70-74 1.1 1.0 1.0 1.3 1.4 1.7 1.3 1.5 75-79 0.5 0.7 0.7 0.5 0.8 1.3 0.6 0.6 80+ 0.8 0.8 0.2 0.8 1.0 1.0 0.4 0.8 >=55 9.5 10.5 9.5 9.1 10.7 12.5 9.0 10.6 OBS 3399 2209 3028 1879 11941 10260 11359 9862 TABLE 1.B AGE DISTRIBUTION BY SEX: 55 AND OVER COTE D'IVOIRE 1985 AND 1986 URBAN RURAL FEMALES MALES FEMALES MALES AGE: 1985 1986 1985 1986 1985 1986 1985 1986 55-59 1.5 1.8 1.9 2.0 2.7 3.0 3.6 2.9 60-64 1.1 1.3 1.5 1.5 2.1 2.7 2.3 3.1 65-69 0.7 0.8 0.5 0.6 1.7 1.7 2.2 2.5 70-74 0.3 0.3 0.6 0.5 0.9 1.1 1.3 1.4 75+ 0.4 0.5 0.3 0.4 1.4 1.3 1.3 1.5 >=55 3.9 4.7 4.8 5.0 8.9 9.7 10.6 11.4 OBS 2842 2805 2678 2662 4014 3846 3737 3583 Notes: Figures are percentages of the relevant group, so that, in Thailand in 1981, 3.2% of all women in Bangkok were aged 55-59. OBS is the total number of observations for all ages in the sample, so that, e.g., there are 3144 females in the urban Ivorian sample in 1985. Note that the Ivorian sample is a simple random sample, so that the sample numbers can be used to estimate the fraction urbanised. This is not true for the Thai survey, see Table 2 below. The Thai results exclude urban non-Bangkok and the suburban "sanitary districts" sector. 10 cross check on reliability, although remember that half of the households are common to both surveys. In rural areas of both countries, the age distributions of older men are very similar, with around 10% of men older than 54 in both cases. The Ivorian survey shows (absolutely) more men than women in all the age categories over 54, whereas Thailand shows the common pattern of more women than men, see Figures lb, ld, 2b, 2d, 3b and 3d. In Thailand, the proportions of elderly are increasing over time (except for males in Bangkok), as is to be expected given the continuing decline in fertility. The major difference between the two halves of the Table lies in the relative youth of the urban sector in C6te d'lvoire. Only 4% of the urban population is aged 55 or over, as opposed to 10% in Thailand, and the much higher level of urbanization in Cote d'Ivoire (42% opposed to 27%) is what reconciles the similarity between the rural sectors with the overall lower fraction of elderly in C6te d'Ivoire. Of course, both fractions are still much lower than those for the more developed countries of the world; in the U.S., 21.3% of the population is aged 55 or over in 1985, United Nations (1985), while for developed countries as a whole there were 15.8% of the population 60 years or over in 1985. The urbanization figures are given in Table 2 for the elderly and for the population as a whole. In Cote d'Ivoire, the urban population grew by 8.7% per annum from 1965-80, as opposed to 4.2% for the total population, World Bank (1988), and we see the picture that would be expected if it is largely the younger people who move to the cities; most (three-quarters of) old people live in the countryside, as opposed to only sixty percent of the population as a whole. The towns are predominantly young; there are relatively few old people in C6te d'Ivoire in any case, and a relatively small proportion are urbanized. For Thailand, the picture is different; the distribution of elderly across rural and urban regions is virtually identical to the distribution of all people across regions. For example, in 1981, 17.5% of people older than 54 were urbanized, as opposed to 17.8% for the population as a whole. However, these numbers mask the fact that that the fraction of older people urbanized exceeds the fraction of children urbanized (14.9%), and is less than the fraction of people aged 15-54 who are urbanized (20.9%). Thus, cities in Thailand have a slightly heavier concentration of younger adults than older adults. The difference between the fraction of older and younger adults who are urbanized is so small because there is relatively slow growth of urban areas in Thailand, and because there are fairly high rates of migration by the elderly to urban areas other than Bangkok. The growth of the urban population in Thailand averages 4.6% a year bewteen 1965 and 1980, as opposed to a 2.6% annual growth rate for the population as a whole, World Bank, (1988). Migrants to Bangkok tend to be young: only 2.6% 11 TABLJ 2 URBANIZATION AND THE ELDERLY COTE D'IVOIRE AND THAILAND Percentages of people living in: C8te d'Ivoire 1985 1986 Thailand 1981 Aged 55 or over: Urban 24.3 25.3 Urban 17.5 Semi-urban 10.3 Rural 75.7 74.7 Rural 72.2 All ages: Urban 41.6 42.4 Urban 17.8 Semi-urban 9.0 Rural 58.4 57.6 Rural 73.2 TABLE 3.A MARITAL STATUS OF THE ELDERLY THAILAND 1981 RURAL FEMALES BANGKOK FEMALES NEVER DIVOR- NEVER DIVOR- AGE: MARRIED MARRIED WIDOWED CED TOTAL MARRIED MARRIED WIDOWED CED TOTAL 55-59 1.8 63.9 29.6 4.8 396 2.8 73.2 15.7 8.3 108 60-64 2.0 57.6 37.3 3.1 295 3.9 39.7 48.7 7.7 78 65-69 1.8 43.6 51.8 2.7 220 1.8 49.1 43.6 5.5 55 70+ 2.2 21.4 74.5 1.9 364 3.7 9.8 79.3 7.3 82 RURAL MALES BANGKOK MALES NEVER DIVOR- NEVER DIVOR- AGE: MARRIED MARRIED WIDOWED CED TOTAL MARRIED MARRIED WIDOWED CED TOTAL 55-59 2.3 86.3 8.4 2.9 344 0.0 87.6 6.7 5.7 105 60-64 0.4 83.8 13.7 2.1 234 1.6 89.1 17.5 0.0 63 65-69 0.6 84.3 13.4 1.7 172 3.3 80.0 13.3 3.3 60 70+ 0.8 71.9 25.8 1.5 267 0.0 60.7 36.1 3.3 61 TABLE 3.B MARITAL STATUS OF THE ELDERLY COTZ D'IVOIRE 1986 FEMALES MALES NEVER DIVOR- NEVER DIVOR- AGE: MARRIED MARRIED WIDOWED CED TOTAL MARRIED MARRIED WIDOWED CED TOTAL 55-59 0.6 65.0 25.8 8.6 163 1.9 90.5 5.1 2.6 157 60-64 0.0 54.4 39.1 6.5 138 2.0 91.5 2.0 4.6 153 65-69 0.0 41.9 55.8 2.3 86 0.9 85.9 11.3 1.9 106 70+ 0.9 15.5 81.9 1.7 116 0.0 82.7 11.8 5.5 127 12 of migrants to Bangkok in 1982 were aged 65 or older, National Statistical Office, (1983). However, migration rates of older people to urban areas other than Bangkok have been quite high, with rates for those 65 and older exceeding rates for those aged 30-49, World Bank (1979). The Ivorian data also provide information on nationality of people sampled. Cate d'lvoire has been one of the more successful West African economies, and has attracted many migrants from its neighbors, particularly Burkina Faso, Mali, and Guinea. Of the two samples, 13.9% in 1985 and 13.1 % in 1986 are non-Ivorian, divided in the ratios 4:2:1:1 among the three countries listed and other Africans. As one might expect if many of these migrants are young, the proportions among those 55 and over are lower, 7.6% and 6.8%. 1.2 Livig aranenls Tables 3.A and 3.B tabulate marital status for those aged 55 and over. For women, the modal status at ages 55-59 is married, and at 70 and over, it is widowed, with the weight shifting from one category to the other as we move from the younger to the older women. These patterns are similar in the two countries. The modal status for men is married in all of these elderly age categories. In C8te d'Ivoire, where a quarter of men have more than one wife, 83% of men aged 70 and over have at least one spouse. Of the 543 men in Table 3.B, 492 are household heads, and for them we have data on numbers and ages of wives. Of these 449 have one or more wives in the household, 59% have one wife, 26% have two, 11% have three, and 4% have four or more. The average age of these 449 men is 64, that of the first wife 51, the second wife 44, and the third wife 40. It is difficult to become a widower in Cote d'Ivoire, and even among those aged 70 or more, there are only 12% in this category, compared with 26% in rural Thailand. For Ivorian men, there is a strong association between wealth, especially cash wealth, age, and the number of wives. Hecht (1982) describes how, in the 1920's, which were the early years of cocoa and coffee production in C8te d'Ivoire, the cash from the new crops, which were farmed by lineages, not families, was used to provide bridewealth for the acquisition for the lineage of new wives, and thus ultimately new labor. By the 1980's, the old lineage system had largely broken down and been replaced by one of small-scale peasant farming, with alienable land and wage labor, but the use of the surplus to acquire additional wives remains. Indeed, the acquisition of additional young wives for wealthy Ivorians is a standard way of purchasing old age security. The occurrence of polygyny rises with age until remarkably late in life, see Ahonzo et al (1984, Table 5.8). Only ten percent of men aged 25-29 have more than one wife, and the proportion 13 rises with age until it reaches nearly a third for 65-74 year olds. Indeed, 13% of men aged 70-74 have three or more wives. The effects of polygyny on living arrangements also appear in Tables 4.A and 4.B. Over 80% of Ivorian males in the table live in households with at least one spouse, as compared with only 60% of men aged 70 or more in Thailand. Elderly women, by contrast, are increasingly widowed, and live with their children or with others. About half of these "others" are brothers who take their sisters into the household, the rest are women living with a head of household who is more distantly related, perhaps a niece or nephew. Very few of the elderly, either men or women, live alone in C8te d'Ivoire; in the 1986 (1985) sample there are only 22 (17) people over 54 who live by themselves. Indeed, there are very few couples; less than 5% of the elderly live in households with only two members. Households are large in CMte d'Ivoire, averaging 8.1 persons in 1986, and neither the elderly (nor anyone else) are likely to live in small households; only 1 % of the people in the survey live in households with less than three members. The situation in Thailand is different, although the caveat about the definition of the Thai household must be kept in mind. Household size is smaller, with 4.2 and 4.6 persons per household in Bangkok and rural regions in 1981, and 3.6 and 4.5 per Bangkok and rural household in 1986. There are correspondingly more older people who live alone or with their spouses. Among elderly women in rural Thailand in 1981, 5.6% to 14% lived alone, and a substantial fraction among the younger elderly lived together with a spouse but with no, other family members. The numbers for 1986 do not reveal an increase in the tendency to live alone or with a spouse only. In fact, the fraction of rural females living alone decreased substantially between the two survey periods for all age groups. Older individuals who do not live alone or with a spouse only almost always live with adult children. The fraction of older people living with "others only" is small for all but Bangkok females. Of women who do live with "others only," the age of the household head is typically quite low, indicating that these women may live with adult grandchildren. The "Western" view of the elderly living either alone, alone with spouse, or with their children, is perhaps closer to the truth in Thailand than it is in CMte d'Ivoire. The larger, more complex families in West Africa allow a wide range of possible living arrangements, especially for the large fraction of widows. 1.3 Education, labor supply, and health status The data on education are not comparable between the two countries, but nevertheless Tables 5.A and 5.B show similar patterns across ages and sexes in both Thailand and C8te 14 TABLE 4A LIVING ARRANGEMENTS OF THE ELDERLY THAILAND, 1981 AND 1986 BANGKOK FEMALES RURAL FENALES 1981 1986 1981 1986 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ ALONE 3.8 7.9 7.3 9.8 9.0 8.0 10.9 1.8 5.6 13.2 12.3 13.7 3.4 5.6 9.0 7.6 SPOUSE 5.7 9.2 5.5 2.4 3.9 6.0 10.9 0.0 12.2 12.9 13.2 6.9 11.2 16.7 12.5 7.1 KIDS 18.9 31.6 32.7 24.4 37.2 36.0 34.8 31.6 23.9 22.7 28.8 22.5 23.6 24.1 28.5 25.1 OTHERS 8.5 11.8 9.1 22.0 6.4 16.0 13.0 14.0 5.3 3.4 9.6 14.8 3.1 3.0 4.5 11.1 SPOUSE+KIDS 49.1 22.4 25.5 4.9 33.3 26.0 21.7 5.3 42.6 38.0 23.7 7.7 49.9 41.5 29.0 10.9 SPOUSE+OTHERS 1.9 2.6 1.8 0.0 1.3 0.0 0.0 5.3 2.5 2.4 1.4 2.8 2.1 0.4 4.0 2.1 KIDS+OTHERS 3.8 13.2 9.1 36.6 5.1 2.0 6.5 42.1 3.1 4.8 ' 7.8 28.3 2.9 5.2 7.5 29.6 SPOUSE+KIDS+OTHERS 8.5 1.3 9.1 0.0 3.9 6.0 2.2 0.0 4.8 2.7 3.2 3.3 3.9 3.7 5.0 6.4 SUBTOTALS: WITH SPOUSE: 65.1 35.5 41.8 7.3 42.3 38.0 34.8 10.5 62.2 55.9 41.6 20.6 67.0 62.2 50.5 26.5 WITH KIDS: 80.2 68.4 76.4 65.9 79.5 70.0 65.2 79.0 74.4 68.1 63.5 61.8 80.3 74.4 70.0 72.0 NUMBERS 106 76 55 82 78 50 46 57 394 295 219 364 385 270 200 422 BANGKOK MALES RURAL MALES 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ ALONE 5.7 7.9 5.0 4.9 3.5 7.3 4.2 10.2 5.3 5.1 2.3 4.2 2.1 2.0 3.9. 3.9 SPOUSE 9.5 9.5 8.3 13.1 8.8 14.6 8.3 12.2 14.9 15.4 21.4 23.6 15.0 19.4 22.4 18.4 KIDS 6.7 9.5 11.7 19.7 3.5 14.6 0.0 14.3 7.3 8.1 11.6 12.6 6.3 6.9 11.2 14.5 OTHERS 1.9 1.6 5.0 3.3 . 1.8 0.0 4.3 6.1 2.0 3.0 1.7 2.7 1.2 2.0 1.7 3.2 SPOUSE+KIDS 61.0 55.6 55.0 36.1 66.7 48.8 75.0 30.6 61.5 58.1 52.6 35.7 64.9 59.5 48.0 39.4 SPOUSE+OTHERS 2.9 4.8 6.7 1.6 7.0 2.4 4.3 4.1 2.9 2.6 1.2 4.6 3.3 3.6 .3.9 4.3 KIDS+OTHERS 1.9 1.6 0.0 16.4 1.8 0.0 0.0 20.4 0.3 2.6 2.3 10.3 0.3 0.4 0.6 6.7 SPOUSE+KIDS+OTHERS 10.5 9.5 8.3 4.9 7.0 12.2 4.3 2.0 5.8 5.1 6.9 6.5 6.9 6.1 8.4 9.6 SUBTOTALS: WITH SPOUSE: 83.8 79.4 78.3 55.7 89.5 78.1 91.7 49.0 85.1 81.2 82.1 70.3 90.1 88.7 82.7 71.6 WITH, KIDS: 80.0 76.2 75.0 77.1 79.0 75.6 79.2 67.4 74.9 73.9 73.4 65.0 78.4 72.9 68.2 70.2 NUMBERS 105 63 60 61 57 41 24 49 343 234 173 263 333 247 179 282 Note: 'Alone' means ivsng alone. 'Spouse' means ivin with a spouse only. 'kids' means iving with children only, etc. 'Children' csn Include sons- or daughters-in-aw, and atep-children. Only peope aed 1S mo ovr w Incudod hi household member counts, and servants were excluded. For some observatIons, It is not possible to determine the rebltIonships between all people In the household. For example, an older person who Is an 'othtr relative' of the household head coul potentially be the parent of another person in the household (who would also be coded as n 'other relative' of the head.) In this case, the older person would be coded as lvIn with 'others onry.' Thus, the fractIon otpeople iving with others only is likely to be overstated and the fraction living with children understated, TABLE 4.B LIVING ARRRNONENTS OP THE ELDERLY COTF D'XVOIRB, 1986 FEMALES MALES 55-59 60-64 65-69 70+ 55-59 60-64 65-69 70+ ALONE 0.6 0.7 2.3 0.0 2.5 2.6 2.8 5.5 SPOUSE 4.2 0.7 0.0 0.9 1.9 4.6 1.9 3.1 KIDS 2.5 2.9 0.0 0.9 1.3 0.7 0.9 0.8 OTHERS 15.3 16.7 26.7 21.6 3.8 4.6 7.5 6.3 SPOUSE+KIDS 3.7 2.2 4.7 0.9 17.2 14.4 18.9 11.0 SPOUSE+OTHERS 20.2 23.9 16.3 5.2 3.8 11.1 6.6 10.2 KIDS+OTHERS 22.1 27.5 31.4 62.1 3.8 1.3 5.7 6.2 SPOUSE+KIDS+OTHERS 31.2 25.4 18.6 8.6 65.6 60.8 55.7 56.6 SUBTOTALS: WITH SPOUSE 59.5 52.1 39.5 15.5 88.5 90.8 83.0 81.1 WITH KIDS 59.5 58.0 54.7 72.4 87.8 77.1 81.1 82.6 NUMBERS 163 138 86 116 157 153 106 127 Note: Spouse means living alone with spouse and no others, kids with children and no others, and so on. Children are defined as biological children of the reference elderly person, living in the same household, so that a woman living with her spouse and the spouse's children who are not her own would be classed under 'spouse and others,' which is different from the treatment in Thailand. TABLE 5.A EDUCATIONAL ATTAINMENT BY AGE, SEX AND LOCATION THAILAND 1981 BANGKOK RURAL FEMALES MALES FEMALES MALES AGE SCH ELEM SEC SCH ELEM SEC SCH ELEM SEC SCH ELEM SEC 20-39 .96 .29 .17 .99 .38 .22 .91 .02 .02 .96 .04 .03 40-49 .77 .05 .04 .89 .14 .10 .84 .00 .00 .90 .01 .01 50-59 .58 .04 .02 .67 .09 .05 .65 .01 .00 .85 .01 .01 60-69 .27 .00 .00 .40 .03 .02 .32 .00 .00 .63 .01 .01 70+ .13 .01 .01 .36 .07 .03 .06 .00 .00 .47 .00 .00 Notes: SCH means that the respondent had completed at least one grade higher than kindergarten. ELEM means that the respondent had completed elementary school. SEC means that the respondent had completed high school or a technical/vocational school. TABLE 5.B EDUCATIONAL ATTAINMENT BY AGE AND SEX COTE D'IVOIRB 1986 FEMALES MALES AGE YEARS ARITH READ SCH? YEARS ARITH READ SCH? 20-39 2.60 0.35 0.32 0.37 5.94 0.70 0.66 0.68 40-49 0.28 0.04 0.03 0.04 2.36 0.37 0.32 0.32 50-59 0.05 0.01 0.01 0.01 0.96 0.19 0.17 0.18 60-69 0.05 0.00 0.00 0.01 0.49 0.11 0.09 0.09 70+ 0.00 0.00 0.00 0.00 0.29 0.06 0.05 0.04 Notes: YEARS is years of school completed, ARITH is fraction of people who can do written calculations, READ is the fraction who can read a newspaper, SCH? is the fraction who are attending or who have ever attended a school. 16 d'Ivoire. By any measure, educational standards are much higher in Thailand, and even among ural women, over 90% of the 20-39 age group have had a least one year of school, whereas only 37% of Ivorian women in the same age group have ever been to school. But even in Thailand, very few individuals have ever completed elementary school (seven years of education), and in the rural villages, less than one percent of men or women over 40 have done so. In C6te d'Ivoire, none of the sample women aged sixty or over can read a newspaper, or do a simple written calculation, and only a negligible fraction of women over 50 have ever been to school. But apart from the differences in levels, the patterns are the same; men have more education than women, and young people have much more education than their elders. Conventional concerns about education separating the generations are clearly relevant in these sorts of situations. Three quarters of Ivorian males and more than a half of Ivorian females between 15 and 19 can read a newspaper, something that be accomplished by about half of their fathers, and perhaps a quarter of their mothers, and almost none of their grandparents. One might legitimately wonder if the experience and wisdom of older farmers, real though it is, may not be offset by their inability to read the label on a bag of seeds or fertilizer. Experience may be more valuable than education in a stationary environment, but much growth in LDC's has come from exploiting new crops, and new techniques of growing them; indeed, there appear to be large gains to greater use of fertilizer and insecticide in coffee and cocoa production in CBte d'Ivoire, gains that have so far gone almost entirely unexploited, see Deaton and Benjamin (1988). Labor force participation and hours worked show the standard life- cycle patterns in both countries. In rural areas in Thailand, (Table 6.A) almost all prime-age males and females participate in (mostly agricultural) work, though substantial fractions of time are spent idle according to the dictates of the agricultural calendar. Participation rates are lower for women than for men in Bangkok, and fall off very rapidly among the elderly. Among those who continue to work, hours and weeks remain high. This contrasts with behavior in the rural sector, where hours and weeks decline along with participation among the elderly, perhaps because of the physical demands of agricultural work. Participation rates in CMte d'Ivoire are surprisingly low, especially among males in the 20-39 age group. (Table 6.B). Note that these figures, although covering a broad range of activities, relate to the last seven days, so that those farmers who did nothing in the past week would be counted as non-participants. Furthermore, the traditional allocation of tasks among many West African groups is for women to undertake food growing and trading activities, leaving men free for hunting, fishing, and fighting. Cocoa and coffee farming are, however, legitimate 17 TABLZ 6A LABOR FORCE PARTICIPATION AND WORK HOURS, BY AGE CATEGORY THAILAND, 1981 BANGKOK FEMALES MALES AGE OBS LFP WEEKS HOURS WEEKS OBS LFP WEEKS HOURS WEEKS IDLE IDLE 15-19 460 .37 43.3 51.6 2.1 365 .32 48.4 45.7 6.9 20-39 1338 .67 48.3 49.3 1.2 1131 .85 50.2 49.4 1.8 40-54 437 .57 49.7 52.7 0.0 375 .97 51.4 50.4 0.7 55-59 108 .45 50.2 52.2 0.0 105 .87 51.1 51.5 0.0 60-64 78 .32 51.5 53.0 0.0 63 .67 51.4 55.7 1.7 65-69 55 .16 51.3 62.2 0.0 60 .53 49.3 53.8 0.0 70-99 82 .06 52.0 58.8 0.0 61 .23 52.0 53.7 1.1 RURAL AGE OBS LFP WEEKS HOURS WEEKS OBS LFP WEEKS HOURS WEEKS IDLE IDLE 15-19 1304 .86 46.0 57.3 4.2 1270 .89 47.6 58.1 5.7 20-39 3295 .94 45.3 57.8 3.3 2927 .99 50.6 61.4 5.0 40-54 1569 .93 45.4 57.1 2.9 1493 .99 50.8 61.9 4.3 55-59 396 .80 44.7 54.2 3.8 344 .96 50.9 60.0 5.4 60-64 295 .62 44.2 52.0 2.0 234 .88 49.9 55.6 4.3 65-69 220 .47 42.3 48.6 3.2 173 .77 47.2 53.9 4.2 70-99 365 .24 41.3 41.0 3.0 267 .46 46.7 49.5 4.1 Notes: Labor force participation was defined as spending at least one week in the last year employed, self-employed (on or off farm), or working as free family labor. Average weeks in the labor force includes weeks unemployed. Weeks unemployed (called 'weeks idle" in the table) consists mainly of weeks spent waiting for the agricultural season or 'because no work was available." "Hours" is the individual's reported hours/week when working. TABLE 6.B LABOR FORCE PARTICIPATION AND WORK HOURS, BY AGE CATEGORY COTE D'IVOIRE 1986 FEMALES MALES AGE N LFP WEEKS DAYS HOURS N LFP WEEKS DAYS HOURS 15-19 690 .41 45.2 4.92 6.70 707 .43 44.0 5.12 7.60 20-39 1643 .59 45.5 5.09 6.84 1281 .66 44.7 5.43 8.02 40-54 792 .76 47.0 5.16 6.90 566 .86 47.0 5.37 8.10 55-59 163 .69 46.7 4.81 6.53 157 .73 49.4 4.90 7.56 60-64 138 .62 46.4 4.72 6.40 153 .71 45.9 5.06 7.55 65-69 86 .63 46.0 4.93 6.59 106 .63 47.2 4.72 6.61 70+ 116 .22 44.5 4.52 5.72 127 .40 45.5 4.53 6.25 Notes: These relate to household members. For a person to be a non-participant, he or she must answer "no" to the following three questions, "During the past 7 days have you worked for someone who is not a member of your household, e.g. an employer, a firm, the Government, or some other person outside your household?" "During the past 7 days, have you worked in a field or garden belonging to yourself or your household or have you raised livestock?" 'During the past 7 days, have you worked in a trade, industry, business, enterprise or profession belonging to yourself or your household? For example, as an independent merchant or fisherman, lawyer, doctor, or other self-employed activity?" WEEKS are number of weeks last 12 months in main job only, DAYS is days worked in the last 7 days, and hours hours per day in the last week, again in the main job. 18 activities for men, and are undertaken by a large fraction of Ivorian- households. Participation rates among older workers remain relatively high into the their late sixties, only falling off among the oldest group. Among older participants, weeks worked declines hardly at all, although both days and hours per day fall with age, which is exactly the pattern that might be expected in a predominantly agricultural economy. Note that the hours, days, and weeks figures for Cote d'Ivoire relate only to the activity defined as the main job over the last seven days. Many individuals have second jobs, and there are a large number of small family enterprises, many run by women. Table 7 presents information on the health of the respondents in the Ivorian survey. These are self-reported figures, and the investigators have no means of checking the reliability of these reports. Although all respondents were weighed and measured, such measurements are of relatively little value in determining health status, except for children. Except for those under 30, more than a quarter of all respondents in each age group report some sickness or injury in the last four weeks, with the fraction rising to well over a half among the older groups. For those aged 55 or over, 6 to 13 days a month are days of illness, and these illnesses are sufficiently severe to cause a suspension of normal activities in 3 to 10 days. Women show more illness than men until about 40 years of age, but subsequently show less, considerably less in some of the older age groups. Somewhat less than a half of all illnesses lead to a medical consultation, or the purchase of medicine, and the figures suggest that, among the elderly, a smaller fraction of illnesses in females are severe, or are treated as such. Comparable data from the Thai survey are not available. 1.4 Levels of living Although much of the concern about the elderly is a concern about living standards, it is remarkably difficult to measure their consumption or income levels, even in more developed countries, and the difficulties are much greater in poor countries. In the U.S., where many old people live alone or with their spouses, their household income and expenditure levels can give some idea of living standards in relation to the rest of the population. Indeed, work on the status of the low-income elderly in more developed countries, e.g. Coder, Smeeding, and Torrey (1990), effectively define the population of interest to be this group, typically female, one-person families and married couples, groups that together covered 91% of the U.S. population aged 65 and over in 1982, see Cowgill (1986, p.29). 19 TABLE 7 HEALTH AND SICRNESS BY SEX AND AGE COTE D'IVOIRE, 1986 FEMALES MALES AGES ILL DAYS1 DAYS2 CON MED ILL DAYS1 DAYS2 CON MED 15-19 .16 1.5 0.7 .51 .58 .12 1.0 0.4 .45 .56, 20-24 .21 2.0 1.0 .59 .61 .17 1.6 1.0 .51 .54 25-29 .27 2.4 1.6 .49 .60 .20 2.1 1.3 .57 .63 30-34 .27 3.3 1.6 .59 .61 .30 2.7 1.2 .52 .62 35-39 .34 4.8 2.8 .48 .55 .32 3.5 1.6 .58 .69 40-44 .36 4.7 2.1 .40 .53 .36 3.5 1.8 .50 .59 45-49 .40 5.4 2.7 .45 .50 .42 5.5 2.5 .41 .58 50-54 .36 5.2 2.2 .47 .54 .47 6.2 3.1 .43 .62 55-59 .40 6.2 2.9 .35 .45 .45 6.2 4.0 .47 .57' 60-64 .41 7.2 3.7 .30 .43 .52 7.6 4.4 .41 .47' 65-69 .37 5.2 3.4 .25 .38 .57 11.8 6.9 .45 .58 70-74 .50 9.8 6.0 .28 .60 .67 11.2 7.6 .26 .40 75+ .59 12.0 7.9 .10 .36 .66 13.0 9.6 .38 .48 Notes: These are self-reported figures for all household members. ILL is the fraction of respondents who, during the last 4 weeks, experienced an illness or injury, "for example, a cough, a cold, diarrhea, an injury due to an acci- dent, or any other illness." DAYS1 is the number of days in the last four weeks during which the respondent suffered from the illness or injury, counting in zero days for those not sick. DAYS2 is the number of days the illness prevented the respondent from carrying on his or her usual activities. CON is the fraction of persons reporting an illness, who consulted "a doctor, nurse, pharmacist, healer, midwife, or other health practitioner." MED is the corresponding fraction of cases where the respondent bought medicine. 20 In Thailand, and even more so in C6te d'Ivoire, the vast majority of the elderly live with other people, children, spouses, and other relations, and very few live alone. Household surveys collect data on household levels of living, not on those of the individuals within them. Disentangling who gets what within the household is difficult, even for "private' goods like food, and attempts to do so require costly and intrusive techniques of observation. For public goods, such as housing, entertainment, and many services, individual consumption levels are not even well defined. In contrast to consumption, many income flows can be assigned to individual members of the household, although only with great difficulty in farm households, although such assignment, even when possible, tells us only a limited amount about the distribution of welfare within the household, which is our main concern. There is a belief in much of the development literature that individuals who bring money into the household receive better treatment than those who do not, but there is little credible evidence to support the contention. This problems of isolating the living standards of the elderly is conceptually the same as that of isolating the living standards of children, a topic on which there exists a large and venerable literature. However, as argued in Pollak and Wales (1979) and elaborated in Deaton and Muellbauer (1986), much of this literature sets out by assuming what it wants to measure, and, even after more than a century of research, no generally acceptable methodology has been derived that would support the isolation of children's living standards from household level data. One possible avenue, suggested in Deaton, Thomas, and Ruiz-Castillo (1989), is to identify a set of goods that are not consumed by adults, for example children's goods, and, on the grounds that additional adults exert negative income effects, but no substitution effects on such goods, measure the 'cost" of old people versus that of younger adults by calculating their relative (negative) effects on the consumption of child goods. However, it is difficult to isolate commodities that are only consumed by children, especially in less developed countries where children consume little beyond food, shelter, and clothing. Moreover, (unreported) experiments with the Spanish data used in Deaton, Thomas, and Ruiz-Castillo, did not lead to sensible estimates. If these problems of measuring living standards are taken seriously, it is unclear that it possible for most LDC's to make statements about, for example, the fraction of old people living in poverty, let alone to address broad topics like the effect of development on the status of the old. Even so, something can be said, and we report some fragmentary but relevant evidence. The simplest procedure is to assume that everyone in each household is treated equally, and to impute to each person the per capita or per adult equivalent total expenditure or income for the household in which they reside. If the assumption is correct, the procedure yields the right 21 answer. If it is false, as it almost certainly is, then the calculations are still informative. If old people live predominantly in households with low average living standards, we are more likely to be concerned about their welfare than would otherwise be the case. Of course, it may be that it is the children or younger people in such households that we should worry about, not their likely, powerful elders. Table 8 shows the relevant calculations for C8te d'Ivoire in 1986. In computing adult equivalents, children under 5 have been assigned a weight of 0.25, and those from 5 to 14, a weight of 0.45. These numbers are essentially arbitrary, but they are relatively low in the light of the considerations discussed in Deaton and Muellbauer (1986), and it is better to make some such assumptions than to work with either total or per capita household expenditure. As the age of the individual increases, the average number of household members with which he or she resides decreases, from 12 at age 0 to 9 at age 70, but rises to around 10 for the oldest ages. The economic measures, income, consumption, and income and consumption per equivalent all have the same general shape, rising to their maxima for the 30-34 year age group and faling steadily thereafter. Among the oldest people, household total income and expenditures in the households in which they live are little more than a half the levels in the peak years, and the per equivalent measures are less than a half of the peaks. If consumption per equivalent is taken as a representative measure, the average for those 55 and over is 79% of the average for all individuals. Older Ivorians live in households that have less income and consumption than the national average. However, old people live mostly in rural areas, and the much better-off urban residents are typically young. Moreover, the rural-urban difference is likely overstated by the fact that no allowance is made for price differences between rural and urban areas, and because urban residents typically pay rent, or have rents imputed for them, something that cannot be done for rural residents. Table 9 repeats the information for rural areas only. Now the relationship between living standards and age has essentially disappeared; while total consumption and income fall with age, at least until the late sixties, adult equivalents fal at much the same rate, so that there is little or no relationship between age and the per equivalent measures. Tables 10 and 11 show income and expenditure by the age of the individual for Bangkok and rural areas in 1981. Unlike C8te d'Ivoire, the number of adult equivalents per household does not vary with age. However, family income and expenditure do not vary greatly with age either. There is a small peak in income in the 50-54 age range for males in both Bangkok and rural regions; in Bangkok, this peak in income is offset by a corresponding peak in the number of adult 22 TABDL 8 AV3ZAoU HOUSEHOLD CHAZACT3RSTZCS BY AGE OF HOUSEHLD MENDERS COTS D'IVOIRE 1986 nmrmn nae y cnd ype cndp- number 0-4 11.7 7.6 1760 1748 267 264 2176 5-9 12.0 7.9 1779 1878 248 262 2140 10-14 11.8 7.9 1995 2011 278 278 1841 15-19 12.0 8.7 2232 2264 294 295 1395 20-24 11.3 8.3 2127 2111 296 296 1021 25-29 10.7 7.6 2239 2080 366 341 763 30-34 10.1 7.0 2262 2143 385 387 608 35-39 10.4 7.1 1839 1890 313 325 528 40-44 10.4 7.2 1610 1661 258 252 489 45-49 10.5 7.4 1642 1780 239 270 423 50-54 10.2 7.4 1356 1407 211 220 443 55-59 9.3 6.9 1391 1534 224 243 320 60-64 9.4 6.8 1815 1468 340 220 288 65-69 8.9 6.6 1049 1135 169 185 190 70-74 9.8 7.2 1262 1378 191 246 113 75-79 10.5 7.7 1653 1540 227 224 64 80+ 9.8 7.5 1397 1348 175 178 66 All 11.2 7.7 1884 1895 278 279 12868 FEMALES AGED 55 AND OVER nmemu nae y cnd ype cndpe number 55-59 9.5 7.2 1382 1570 209 234 163 60-64 9.9 7.2 1991 1560 362 228 136 65-69 9.2 6.8 1047 1177 164 187 85 70-74 11.9 8.4 1379 1498 159 179 50 75-79 10.8 7.8 1744 1723 196 227 28 80+ 10.0 7.8 1234 1239 159 165 38 All 11.4 7.8 1877 1898 273 273 6636 MALES AGED 55 AND OVER nmems nae y cnd ype cndpe number 55-59 9.1 6.6 1400 1497 240 252 157 60-64 8.9 6.5 1657 1386 322 214 152 65-69 8.7 6.3 1050 1100 172 183 105 70-74 8.1 6.2 1170 1283 216 298 63 75-79 10.3 7.5 1583 1398 252 221 36 80+ 9.6 7.3 1617 1496 198 196 28 All 11.1 7.7 1890 1892 285 285 6232 Notes: These are calculated on an individual basis, i.e. each individual in the sample is assigned the number of household members, household income, or household income per equivalent, and then averages are calculated conditional on individual age. nmems is number of household members. nae is number of adult equivalents, where children aged 0-4 are counted as 0.25, aged 5-14 as 0.45, and 15 and over as 1. y is household income. cnd is household consumption excluding purchases of durables, and ype and cndpe are the corresponding figures per equivalent adult. number is the number of persons over which the means are calculated. Money amounts are in CFA '000 per annum (about $3.) 23 TABLE 9 AVERAGE HOUSEZOLD CHARACTERISTICS BY AGE OF HOUSEHOLD MEMBERS RURAL COTE D'IVOIRE 1986 rnems nae y cnd ype cndpe number 0-4 11.8 7.6 1307 1310 179 181 1327 5-9 11.8 7.7 1263 1384 167 185 1245 10-14 11.6 7.7 1334 1400 173 183 1039 15-19 12.3 8.7 1472 1471 170 168 690 20-24 11.2 7.9 1345 1300 180 169 471 25-29 11.5 7.9 1318 1349 194 200 354 30-34 10.6 7.3 1293 1301 187 190 300 35-39 10.7 7.2 1272 1319 184 195 310 40-44 10.2 7.0 1108 1175 165 174 305 45-49 10.3 7.2 1124 1243 165 185 272 50-54 9.7 7.0 1013 1088 156 169 320 55-59 8.9 6.6 1006 1130 166 185 218 60-64 9.2 6.6 1082 1158 163 172 211 65-69 8.2 6.1 842 945 148 170 151 70-74 9.0 6.6 1086 1190 186 245 92 75-79 9.0 6.7 1004 1132 150 182 49 80+ 9.5 7.3 1280 1138 165 153 55 All 11.2 7.6 1262 1313 173 181 7409 Notes: y is household income. cnd is household consumption excluding pur- chases of durables, and ype and cndpe are the corresponding figures per equivalent adult. number is the number of persons over which the means are calculated. Money amounts are in CFA '000 per annum (about $3.) For other notes, see Table 8. 24 TABLE 10 AVERAGE HOUSEHOLD CHARACTERISTICS BY AGE OF HOUSEHOLD MEMBERS BANGKOK, 1981 nmems nae y cnd ype cndpe number 0-4 5.0 3.4 7141 5693 2173 1729 565 5-9 6.0 4.3 8031 6600 1949 1615 515 10-14 6.5 4.8 7945 6760 1754 1500 629 15-19 6.2 5.5 9044 7540 1678 1423 825 20-24 5.4 4.9 9152 7185 1962 1574 816 25-29 4.7 4.1 8518 6260 2229 1680 783 30-34 4.6 3.7 8496 6121 2527 1800 546 35-39 5.1 3.9 7947 6240 2208 1752 324 40-44 5.4 4.3 8209 6671 2176 1759 317 45-49 5.7 4.9 9057 7159 2059 1574 252 50-54 5.5 5.0 9971 8051 2092 1697 243 55-59 5.1 4.8 9467 7135 2111 1591 213 60-64 4.8 4.4 9682 6781 2238 1662 141 65-69 5.3 4.6 7659 6354 1778 1518 115 70-74 5.4 4.6 8596 7733 1958 1842 69 75-79 4.6 4.3 7026 5872 1929 1596 39 80+ 5.3 4.7 8515 7151 2031 1719 35 ALL 5.5 4.5 8513 6739 2043 1624 6427 FEMALES nmems nae y cnd ype cndpe number 55-59 4.9 4.6 8334 6679 1905 1504 108 60-64 4.6 4.2 9196 6195 2220 1655 78 65-69 5.2 4.4 7856 6662 1908 1674 55 70-74 5.4 4.6 7757 7587 1795 1784 39 75-79 4.1 3.6 5307 4861 1632 1544 17 80+ 4.9 4.5 8517 7127 2157 1840 26 ALL 5.5 4.5 8486 6772 2010 1616 3399 MALES nmems nae y cnd ype cndpe number 55-59 5.3 4.9 10633 7604 2324 1680 105 60-64 5.0 4.6 10284 7506 2261 1671 63 65-69 5.3 4.8 7478 6072 1660 1376 60 70-74 5.3 4.6 9685 7924 2170 1916 30 75-79 5.1 4.8 8355 6653 2159 1637 22 80+ 6.4 5.2 8510 7220 1665 1369 9 ALL 5.4 4.5 8544 6702 2080 1634 3028 Notes: Money amounts are baht per month. Variables are defined in Tables 8 and 9 above. 25 TABLE 11 AVERAGE HOUSEHOLD CHARACTERISTICS BY AGE OF HOUSEHOLD MEMBERS RURAL THAILAND, 1981 nmems nae y cnd ype cndpe number 0-4 5.5 3.7 2316 2158 673 624 2536 5-9 6.0 4.1 2455 2270 636 582 3328 10-14 6.2 4.5 2789 2530 659 597 3272 15-19 6.1 5.0 3074 2716 643 569 2572 20-24 5.3 4.3 2892 2486 718 619 1880 25-29 4.8 3.6 2545 2234 766 685 1656 30-34 5.1 3.5 2622 2296 784 693 1458 35-39 5.5 3.9 2587 2374 705 640 1225 40-44 5.7 4.3 2828 2612 723 656 1123 45-49 5.6 4.4 2853 2596 696 635 1097 50-54 5.2 4.3 3288 2862 852 772 839 55-59 4.8 4.1 2877 2479 773 665 740 60-64 4.3 3.8 2858 2410 900 720 529 65-69 4.3 3.7 2525 2228 787 707 393 70-74 4.4 3.6 2336 2260 710 685 308 75-79 4.7 3.9 2650 2466 748 678 160 80+ 4.9 4.1 2711 2441 724 656 164 ALL 5.6 4.2 2706 2430 703 631 23280 FEMALES namems nae y cnd ype cndpe number 55-59 4.6 3.9 2583 2316 718 646 396 60-64 4.2 3.6 2684 2337 848 743 295 65-69 4.2 3.4 2482 2237 852 784 220 70-74 4.4 3.6 2309 2263 674 662 163 75-79 4.9 4.0 2504 2512 656 664 94 80+ 4.9 4.2 2834 2472 769 677 108 ALL 5.5 4.1 2699 2435 708 640 11935 MALES nmems nae y cnd ype cndpe number 55-59 5.1 4.3 3215 2666 837 688 344 60-64 4.5 4.0 3077 2501 965 690 234 65-69 4.6 3.9 2580 2216 705 610 173 70-74 4.3 3.6 2367 2256 750 712 145 75-79 4.4 3.8 2857 2399 879 697 66 80+ 4.9 4.1 2474 2382 639 616 56 ALL 5.6 4.2 2714 2425 697 621 11345 Notes: Money amounts are baht per month. See Tables 8 and 9 for definitions of variables. 26 equivalents. Overall, income and expenditure, as well as income and expenditure per adult equivalent, are very flat across age groups. The average consumption per adult equivalent of those 55 and over is 100.3% of the average for aU individuals in Bangkok, and 109% for rural regions. On average, older Thais in both Bangkok and rural regions do not live in poorer households. Unlike the Ivorian surveys, the Thai surveys provide a good deal of information on individual income levels and the sources of individual income. If the allocation of consumption to members within a household depends on the amount of income members bring to the household (and again, it is not known if this is true), then the patterns of individual income with age provide evidence on standards of living over the life-cycle. Information on the distribution of income between pensions, annuities, and property income, as opposed to remittances and gifts, provides evidence on the extent to which older individuals rely on asset markets for old-age support. It is possible to disaggregate individual income into that derived from wages, farming and self-employment (called business income), property, transfers (remittances, pensions and annuities), and other sources. The measures of profits from farming and self-employment are problematic in that they do not exclude the value of free family labor used, and are usually "assigned' to the household head "or to the operator of the enterprise if he could be identified." For most family businesses, it is not clear that the profits from the business should be assigned to any one person. In what follows, no adjustments were made for these problems. Table 12 provides information on individual income and the distribution of income for males and females in Bangkok and rural regions. Unlike the household income figures discussed above, there is a clear pattern of individual income over the life-cycle. For both rural and Bangkok males, income levels peak in the 50-59 age range and then decline rapidly. Female income levels in are flatter over the 30-60 age range but then also decline. These results are consistent with the declines in labor force participation for both males and females after the age of 60, and much of the declining income levels of older individuals can be accounted for by the increasing fraction of those who earn no income at all. As is to be expected, the share of income from wages and business declines for older people, although the share of income from farming and self-employment remains quite high for men (49% and 79% for Bangkok and rural men aged 60-69). This reflects the fact that the oldest man in the household is usually the head of the household and would typically have all family business income assigned to him. 27 TABLE 12 DISTRIBUTION OF INCOME, BY AGE, SEX AND LOCATION THAILAND, 1981 BANGKOK MEAN FEMALES INDIVIDUAL %INC SHARE OF INDIVIDUAL INCOME: INC/ FREE/ AGE: OBS INCOME >0 WAGE BUS. PROP TRANS OTHER FAMINC FAMINC 20-29 870 1177 .54 .75 .09 .01 .11 .04 .18 .05 30-39 468 1733 .62 .61 .25 .01 .06 .07 .24 .05 40-49 299 1648 .58 .39 .37 .04 .14 .07 .23 .05 50-59 245 1764 .58 .25 .35 .03 .32 .05 .23 .04 60-69 133 1523 .47 .05 .31 .06 .56 .03 .19 .07 70+ 82 456 .26 .00 .10 .12 .74 .05 .11 .05 MALES 20-29 729 2141 .71 .81 .09 .00 .08 .02 .34 .05 30-39 402 4517 .93 .73 .24 .01 .01 .02 .62 .06 40-49 268 4899 .96 .60 .37 .00 .01 .02 .65 .05 50-59 211 5418 .89 .48 .41 .01 .06 .04 .53 .03 60-69 123 2719 .67 .26 .49 .07 .15 .04 .34 .07 70+ 61 1168 .41 .07 .23 .11 .58 .01 .16 .08 RURAL MEAN FEMALES INDIVIDUAL %INC SHARE OF INDIVIDUAL INCOME: INC/ FREE/ OBS INCOME >0 WAGE BUS. PROP TRANS OTHER FAMINC FAMINIC 20-29 1893 241 .39 .67 .21 .00 .09 .03 .09 .06 30-39 1400 394 .42 .50 .36 .02 .07 .04 .12 .04 40-49 1134 473 .46 .35 .45 .03 .13 .04 .17 .04 50-59 830 369 .44 .23 .49 .03 .21 .05 .17 .05 60-69 515 384 .46 .10 .42 .07 .38 .03 .16 .08 70+ 365 129 .38 .08 .21 .12 .55 .04 .10 .10 MALES 20-29 1643 808 .71 .55 .42 .00 .01 .01 .37 .05 30-39 1283 1591 .94 .27 .69 .00 .01 .02 .66 .04 40-49 1086 2031 .98 .18 .77 .01 .02 .02 .68 .04 50-59 749 2278 .96 .16 .75 .01 .05 .02 .65 .04 60-69 407 1665 .87 .06 .75 .03 .13 .03 .55 .06 70+ 267 825 .63 .03 .54 .10 .29 .04 .29 .09 Notes: Individual income includes wages, business income (farm plus self-employment income), property income (interest, dividends, income from roomers and boarders), transfer income (pensions and annuities, remittances from friends and relatives), other income (lotteries, insurance, sales of durable goods.) Family income equals the sum of all member's individual incomes plus the rental value of owner occupied homes, home-produced goods not included in farm income, and goods received free. INCIFAMINC is individual income over family income. FREE/FAMINC is the value of goods received free as a fraction of family income. 28 The share of income derived from transfers (including pensions, annuities, and remittances) increases dramatically with age for both men and women in Bangkok and rural regions. Transfers account for a large share of individual income, particularly for women. These transfers consist mainly of remittances, presumably from family members or friends in other households. Although transfers cannot be divided up between remittances and pensions and annuities at the individual level, they can be disaggregated at the household level. Of aU households that receive transfers, 93% of those in Bangkok receive no pensions or annuities, and 97% of rural households receive no pensions or annuities. The share of income from property (including interest, dividends, and rents) increases with age but, like pensions and annuities, is quite small, reaching only 10-12% of income for both Bangkok and rural residents in the 70+ age group. Thus, sources of old-age income that are standard in developed countries have only a very small role in Thailand. Table 13 tabulates income by source for rural males and females who live alone or with a spouse only, and for those who live with at least one person who is not a spouse. Older people living with others are less likely to earn any income at all. However, the shares of income from different sources are not too dissimilar for those in different living arrangements. The fraction of total family income derived from goods received free does vary with living arrangements. For example, for rural females aged 60-69, goods received free accounts for 18% of family income for those alone or with a spouse only, and only 5% of family income for those living with others. For females living alone or with a spouse only, free goods and transfers make up a significant share of their income. 29 TABLE 13 INCOME COMPOSITION AND LIVING ARRANGEMENTS THAILAND, 1981 RURAL FEMALES, LIVING ALONE OR WITH SPOUSE ONLY MEAN INDIVIDUAL %INC SHARE OF INDIVIDUAL INCOME: INC/ FREE/ AGE: OBS INCOME >0 WAGE BUS. PROP TRANS OTHER FAMINC FAMINC 50-59 157 417 .64 .27 .37 .04 .27 .05 .25 .11 60-69 133 394 .69 .11 .35 .09 .43 .03 .28 .18 70+ 75 292 .84 .07 .25 .13 .52 .04 .32 .30 FEMALES, LIVING WITH AT LEAST ONE CHILD OR OTHER PERSON MEAN INDIVIDUAL %INC SHARE OF INDIVIDUAL INCOME: INC/ FREE/ AGE: OBS INCOME >0 WAGE BUS. PROP TRANS OTHER FAMINC FAMINC 50-59 669 359 .40 .21 .53 .03 .19 .05 .15 .04 60-69 381 382 .38 .10 .46 .06 .35 .03 .12 .05 70+ 289 87 .25 .09 .19 .11 .57 .05 .05 .05 MALES, LIVING ALONE OR WITH SPOUSE ONLY MEAN INDIVIDUAL %INC SHARE OF INDIVIDUAL INCOME: INC/ FREE/' AGE: OBS INCOME >0 WAGE BUS. PROP TRANS OTHER FAMINC FAMINC 50-59 144 1984 .99 .26 .63 .03 .08 .01 .68 .07 60-69 89 1479 .96 .09 .67 .06 .15 .04 .59 .11 70+ 73 1038 .96 .05 .47 .15 .31 .03 .50 .18 MALES, LIVING WITH AT LEAST ONE CHILD OR OTHER PERSON MEAN INDIVIDUAL %INC SHARE OF INDIVIDUAL INCOME: INC/ FREE/ AGE: OBS INCOME >0 WAGE BUS. PROP TRANS OTHER FAMINC FAMINC 50-59 602 2352 .95 .14 .79 .01 .04 .03 .64 .03 60-69 318 1717 .84 .05 .77 .03 .12 .03 .54 .04 70+ 190 761 .51 .01 .59 .07 .28 .05 .22 .06 Notes: See Table 12. 30 2 Household life-cycls In this section we move our focus away from individuals and towards households, and how they change with the ages of their members. Households in LDC's are typically larger than those in more developed countries, particularly so here for CMte d'Ivoire, so that, with several generations living together, the life-cycle patterns of the household as an aggregate may be much attenuated compared with the patterns observed in the West. Households with between ten and twenty members are not uncommon in C8te d'Ivoire, and in the limit, it is possible to imagine a state of affairs in which each household's demographic composition is a miniature version of that of the country as a whole, and the life-cycles of the individuals within the household offset are subsumed into a stationary structure for the household. In fact, such is far from being the case in Cote d'Ivoire. Table 14.B shows the breakdown of household heads by age and sex. If household composition were stationary, and the oldest male was always designated as the household head, there would be no heads outside this category. In reality, 42% of household heads are under the age of 55, and only 19% are men over the age of 70. Only five percent of households contain one or more married sons of the head, and less than one percent have two or more. Similarly, it is rare for married brothers to live together; only three percent of male headed households contain a married brother. These households seem to conform well to what Cowgill (1986, p.62) describes as the common pattern among polygynous households; "a man, his several wives, their (unmarried) children," and possibly "some additional consanguines, such as unmarried or widowed sisters of the husband, and perhaps his aged parents." New households are set up by married sons, so that, while there is a clear bias towards older heads and there are more heads in older groups than their share of the population would warrant, there are many households headed by younger men. The economic status of the household is also clearly related to the age of its head, as shown in Table 15. These data are presented for both years; they are probably a good deal less reliable, particularly for assets, than previous data, so that one year cannot be safely taken as representative for both. The figures show that older heads preside over bigger households, but that both household income and household total expenditure reach a peak among households headed by 30-34 year-olds, and then steadily decline. The pattern, if it is there, is a good deal less obvious in the rural areas. As was the case for patterns in the individual data, the hump in household incomes and expenditures is exaggerated by pooling older, poorer, rural individuals with younger, richer, urban ones. Since household size and the number of equivalents increase with the age of the head, deflation by either measure produces a pattern in which household living 31 TABL 1UA AGE AND SEX COMPOSITION OF HOUSEHOLD HEADS THAILAND, 1981 BANGKOK RURAL females males all females males all 15-19 1.05 1.31 2.36 0.20 0.42 0.62 20-24 2.16 4.52 6.69 0.52 3.87 4.39 30-34 2.62 11.67 14.30 0.86 9.29 10.15 35-39 2.36 12.33 14.69 0.86 11.08 11.94 40-44 2.30 8.46 10.75 1.08 10.51 11.58 45-49 2.30 8.79 11.08 1.68 10.21 11.88 50-54 2.23 7.34 9.57 2.14 10.29 12.42 55-59 2.82 6.23 9.05 2.00 7.87 9.87 60-64 2.30 5.70 8.00 2.34 6.55 8.89 65-69 2.16 3.54 5.70 1.94 4.51 6.45 70-74 1.25 2.95 4.20 1.78 3.24 5.01 75-79 0.66 1.25 1.90 1.24 2.52 3.75 80+ 0.66 1.05 1.70 1.30 1.74 3.04 ALL 24.85 75.15 100.00 17.91 82.09 100.00 TABLE 14B AGE AND SEX COMPOSITION OF HOUSEHOLD HEADS COTE D'IVOIRB 1986 males females all 20-24 1.3 0.1 1.4 25-29 1.9 0.1 2.1 30-34 3.5 0.2 3.7 35-39 6.1 0.8 6.9 40-44 8.8 0.9 9.7 45-49 9.2 1.1 10.3 50-54 11.3 0.9 12.1 55-59 10.6 0.8 11.3 60-64 9.4 0.9 10.4 65-69 10.9 1.0 11.9 70-74 10.4 0.5 10.9 75-79 6.6 0.6 7.3 80+ 1.8 0.1 1.9 All 91.9 8.1 100.0 32 TABLE 15 NEMBERS, INCOME, EXPENDITURE, AND ASSETS BY MEAD'S AGE COTE D'IVOIRN 1985 AND 1986 1985 AGE nmems nae y cnd a agass busass perass 20-24 3.5 2.7 926 933 -8 911 90 386 25-29 5.3 3.8 1491 1542 -51 560 6148* 804 30-34 7.0 4.8 1937 1916 21 2435 596 910 35-39 7.5 5.0 1885 1880 5 2687 275 656 40-44 8.8 5.9 1610 1832 -222 2632 893 716 45-49 9.7 6.6 1857 1749 107 3666 636 1108 50-54 9.2 6.5 1271 1408 -137 5167 230 796 55-59 9.2 6.7 1496 1377 120 3732 708 1008 60-64 9.7 7.0 1497 1537 -40 5199 737 1314 65-69 9.6 7.0 1470 1415 55 4655 752 1251 70+ 8.0 6.1 870 932 -62 6882 155 976 1986 AGE nmems nae y cnd s agass busass perass 20-24 4.1 3.0 843 912 -70 1194 27 198 25-29 5.5 4.0 1845 1659 186 613 211 726 30-34 6.7 4.5 2096 2050 46 1055 170 780 35-39 7.9 5.3 2132 2119 13 1173 398 979 40-44 8.4 5.6 1455 1677 -222 1575 296 969 45-49 9.4 6.4 1835 1831 4 1693 602 1326 50-54 9.0 6.3 1298 1340 -41 2187 274 658 55-59 8.7 6.4 1381 1481 -100 2710 523 1367 60-64 8.4 6.2 1700 1391 310 2199 684 1117 65-69 8.1 6.0 994 1041 -47 3275 287 628 70+ 8.5 6.4 1224 1277 -53 2568 492 786 RURAL 1986 AGE nmems nae y cnd s agass busass perass 20-24 5.5 3.8 1072 602 471 2982 9 101 25-29 6.5 4.4 1223 1088 135 1672 79 211 30-34 7.3 4.9 1185 1022 162 2394 89 187 35-39 8.6 5.6 1089 1100 -11 2334 111 263 40-44 8.6 5.5 831 1033 -201 2855 79 185 45-49 9.0 6.0 1025 1096 -71 2042 57 264 50-54 8.7 6.0 919 963 -44 2699 64 192 55-59 8.1 5.9 851 998 -146 3879 33 235 60-64 7.9 5.7 1116 1123 -6 2364 390 301 65-69 8.0 5.8 842 899 -57 3572 70 319 70+ 8.1 6.1 927 1052 -124 2938 41 176 Notes: nmems is number of members, nae, numbers of adult equivalents, y is household income, and cnd consumption excluding purchases of durable goods. s, for saving, is the difference bewteen y and cnd. agass is the value of agricultural assets, including a farmer estimated figure for the value of agricultural land. busass is the value of assets used in family business, and peress is the value of personal assets. * This figure is dominated by one outlier, a 28 year old head near Aboisso, in the extreme South-East of the country, who reported business assets worth more than CFA700,000 ($2.1 million). 33 standards decline with the age of the head, so that the hump is moved to the extreme left of the age distribution. The hump-shaped pattern, in which incomes and consumption shapes are closely matched, is one with a peak that occurs much earlier in the head's age distribution than is the case in many LDC's, particularly those in Asia, see Deaton (1990) for evidence on Thailand, Korea, Indonesia, and Hong Kong, the first of which we return to below. These cross-country patterns are important because, as pointed out by Carroll and Summers (1989), if tastes are common across countries, then the rapidly growing countries are those where young people are relatively much richer than their parents and grandparents, so that age consumption profiles should peak earliest in the most rapidly growing economies. But C6te d'Ivoire is a very slow-growing economy relative to Thailand and the other Asian countries listed above, and this slow growth is accompanied by the earliest peak in household consumption. As Carroll and Summers emphasize, these results make it difficult to believe that life-cycle saving is responsible for the cross-country correlation between growth and savings that exists in the data. Instead, the obvious alternative is that consumption tracks income over the life-cycle, a hypothesis that is fully consistent with the data in Table 15. Saving itself is as often negative as positive, and shows no clear pattern with age. The measurement of income for poor, largely illiterate, self-employed farmers in LDC's is an undertaking fraught with difficulty, and little weight should be attached to the magnitude of these figures. However, analysis of the micro data from CMte d'Ivoire provides evidence that farmers undertake short-run saving to smooth their consumption relative to their noisy incomes, and this evidence is also consistent with the earlier results on farmers' saving behavior in Thailand in Paxson (1989). The asset figures are likely to be almost as unreliable as the savings data, and there is a still unresolved question as to why the (largely self reported) figures for agricultural assets are so much lower in 1986 than in 1985. The data in the upper panels suggest that, over the country as a whole, asset levels continue to increase with the age of the household head, but some of this is an aggregation effect; in the rural panel, agricultural assets are more or less equally distributed across age groups, something that would follow from a process in which land is closely tied to household formation. Note that, at least until recent years, land has not been particularly scarce in C8te d'Ivoire (nor in Thailand), and, given permission from the lineage owning the land, new cocoa and coffee farms could be established by clearing virgin forest, with ownership gradually established by use. Even today, fathers would typically assume responsibility for providing their sons with land, and if uncultivated land is no longer available within the lineage boundaries, the 34 acquisition or use of land elsewhere will be arranged, preferably close by, but sometimes at some considerable distance. See again Hecht (1982) for a description of the evolution of land markets in response to increasing scarcities, first of labor, and later of land. Table 16 presents regressions of income, consumption, and the asset variables on household demographic structure, and on dummies for the five main urbanization and agroclimatic zones in the country. These results should not be interpreted as structural equations, but as an alternative and more informative description of the relationship between head's age and these economic variables. High income and consumption levels are associated with the presence of prime age males and females; in itself evidence of consumption tracking income. The presence or absence of individuals aged 55 and over seems to contribute little to household income or consumption levels. Asset levels, however, are positively associated with the presence of older men, (but not older women) particularly those aged from 55 to 64. This is cerainly consistent with a steady accumulation of assets by the male head, passing on to sons at or before death. Women aged 25-34 also attract a very large positive coefficient in the agricultural assets equation. Since daughters would not normally inherit land, there is no obvious explanation for this result, although it could conceivably reflect the propensity of older wealthy men to marry young second or thrid wives. There is no evidence of an association between business assets and women, although many small business in Cote d'Ivoire are owned by women. Thai households (as defined by the Socioeconomic Surveys) are much smaller than those in C8te d'Ivoire. Using 1981 data, approximately 50% of rural households have four members or less, and households of ten or more members are rare. Households in Thailand are also likely to have younger household heads. Sixty-three percent of rural household heads and 69% of urban household heads are younger than 55, see Table 14.A. The size of households also varies with the age of the household head. The first column of Table 17 shows that the number of household members first increases and then decreases as the household head ages. These numbers are consistent with the "Western" pattern of children marrying and setting up their own households, which grow as children are added, and then shrink as children move out. Cowgill (1986, pp.69- 70) describes the Thai system as a "residual stem family" system, in which young married couples often live with one set of parents but only until a younger sibling marries and takes their place. The last child married, often the last daughter married, stays with the parents until the parents die, and then inherits the land. This would explain why households with very old household heads have, on average, four members rather than one or two. 35 TABLE 16 REGRESSIONS OF INCONS, CONSUMPTION, AND ASSETS ON HOUSEHOLD COMPOSITION COTE D'IVOIRE 1986 income consumption agricultural business personal assets assets assets Variable Est It| Est it! Est it Est it! Est It| CONSTANT 124 (0.7) 259 (2.3) -603 (1.4) -378 (2.9) -791 (3.7) M0-4 -65 (0.9) -76 (1.7) 131 (0.8) 40 (0.8) 29 (0.3) N5-14 67 (1.4) 127 (4.3) -44 (0.4) -29 (0.8) -22 (0.4) N15-24 214 (3.7) 173 (4.7) 417 (3.0) 195 (4.5) 401 (5.7) N25-34 394 (4.0) 213 (3.5) 196 (0.8) 64 (0.9) 229 (2.0) M35-44 257 (1.7) 219 (2.4) 167 (0.5) 59 (0.5) 23 (0.1) M45-54 157 (1.0) 25 (0.3) 525 (1.4) 193 (1.6) 221 (:1.2) M55-64 302 (1.8) 29 (0.3) 947 (2.3) 375 (3.0) 635 (:3.1) M65-74 24 (0.1) -19 (0.1) 1475 (3.0) 248 (1.6) 373 (:1.5) M75+ 373 (1.2) 124 (0.7) 547 (0.8) 374 (1.7) 430 (1.2) F0-4 -47 (0.7) -107 (2.4) 180 (1.1) -30 (0.6) 15 (0.2) F5-14 64 (1.3) 90 (2.9) -40 (0.4) 10 (0.3) 29 (0.5) P15-24 199 (3.0) 279 (6.8) 164 (1.1) 20 (0.4) 150 (:1.9) P25-34 345 (3.5) 257 (4.1) 1033 (4.4) 75 (1.0) 291 (2.4) F35-44 43 (0.4) 89 (1.3) 713 (2.7) 26 (0.3) 222 (1.7) F45-54 -163 (1.4) 4 (0.1) 144 (0.5) 66 (0.8) -37 (0.3) F55-64 196 (1.4) 102 (1.2) 408 (1.3) 147 (1.5) 360 (2.2) F65-74 -144 (0.8) -62 (0.5) -257 (0.6) 38 (0.3) 142 (0.6) F75+ 17 (0.1) -5 (0.0) -262 (0.4) -113 (0.6) 101 (0.3) ABIDJAN 1742 (9.0) 1673(13.9) -937 (2.0) 1077 (7.6) 1907 (8B.3) OTHER URB 909 (4.8) 739 (6.2) -358 (0.8) 405 (2.9) 1266 (5.6) W.FOREST 34 (0.2) 36 (0.3) 1742 (3.7) 120 (0.8) 38 (0.2) Z.FOREST 186 (1.0) 96 (0.9) 1979 (4.7) 16 (0.1) -74 (0.3) Notes: The figures are total income, consumption, and assets, undeflated by any measure of household size. M is males, F is females, and the ind- ependent variables are the numbers of people in the relevant age category in the household. The omitted region is the northern savannah region. 36 TABLE 17 MEMBERS, INCOME AND EXPENDITURE BY HEAD S AGE THAILAND, 1981 BANGKOK AGE nmems nae y cnd s y/nae cnd/nae s/nae 15-19 1.6 1.5 2225 1961 264 1682 1467 215 20-24 2.3 2.0 3736 3285 451 1939 1779 160 25-29 3.1 2.5 6065 4570 1495 2564 1993 571 30-34 3.9 3.0 7239 5348 1891 2724 2002 722 35-39 4.5 3.3 7017 5540 1477 2294 1822 472 40-44 4.8 3.8 7598 6165 1433 2437 1940 497 45-49 5.3 4.4 6951 6245 706 1758 1543 214 50-54 5.3 4.9 9284 7570 1714 2073 1676 397 55-59 4.9 4.6 9831 7114 2716 2289 1670 619 60-64 4.3 4.0 10033 6806 3227 2449 1765 684 65-69 4.8 4.3 6933 5941 991 1832 1588 244 70+ 4.0 3.6 6293 6053 240 2128 1991 137 RURAL AGE nmems nae y cnd s y/nae cnd/nae s/nae 15-19 2.3 2.0 1459 1295 164 810 760 49 20-24 3.0 2.3 1845 1652 193 923 800 123 25-29 3.7 2.6 1969 1749 220 875 781 94 30-34 4.5 3.0 2381 2090 291 838 735 103 35-39 5.2 3.5 2302 2196 106 714 658 56 40-44 5.5 4.0 2852 2593 259 805 713 92 45-49 5.6 4.3 2762 2439 323 701 624 76 50-54 5.1 4.1 3137 2825 312 861 801 60 55-59 4.8 4.1 2883 2461 422 797 678 119 60-64 4.1 3.5 2826 2310 517 983 750 233 65-69 4.0 3.4 2483 2172 311 843 751 92 70+ 3.7 3.1 2095 2010 86 756 711 44 Note: See Table 15 for variable definitions for the first six columns. The last three columns are income, non-durable consumption expenditures, and saving divided by numbers of adult equivalents. 37 These patterns of household formation may make life-cycle models of household consumption more relevant for Thailand than CMte d'Ivoire. With smaller households it is less likely that household members span a broad range of ages, and the age of the household head should be a good indicator of where a household is in its life-cycle. Given the fairly rapid growth in Thailand, one might expect to see younger (richer) households both earning and consuming more than older households. The results in Table 17 indicate that income and consumption do follow a hump-shaped pattern similar to that seen in CMte d'Ivoire but with a much later peak in both income and consumption. Household income reaches its highest level in the 60-64 age category for Bangkok, and in the 50-54 age category for rural areas. Consumption tracks income closely, and saving also appears to follow a similar pattern, with those in the highest income groups saving most, although the pattern for saving is less pronounced. These patterns are consistent with the age patterns, of individual income shown in Table 12, and are also consistent with the patterns of household size shown in Table 17. Household heads in their 40's and 50's have the largest households, and it is likely that the children in these households are old enough to contribute substantially, to household income. Although total household income and consumption are both strongly related to the age of the head of the household, income and consumption after adjusting for the number of adult equivalents are not. Since family size and the number of adult equivalents follows the same hump- shaped pattern as income and consumption, adjustment for family size results in extremely flat income and consumption profiles that appear to have no relation to the age of the household head. The absence of any difference in income and consumption per equivalent adult between young and old households is puzzling, especially in a rapidly growing country such as Thailand. One possible explanation is that households in Thailand may be much more complex than the data suggest. As discussed earlier, a small "household" may actually be part of a larger group of several related households in a single compound, and there may be significant transfers between such households. The fact that older people "living alone" receive a large fraction of their incomes in the form of free goods (most of which are food) suggests this might be so. If each household, as measured by the survey data, is actually part of a network of closely linked households containing people in different generations, then it becomes quite unclear whether one ought to expect individual households to operate in ways predicted by life-cycle models. One can imagine a situation in which household formation is itself the mechanism that is used to smooth consumption (and income) across individuals in different generations: individuals may be 38 "allocated" across households so as to maintain roughly equal consumption levels across all family members within a group of households. Much more detailed data on links between households would have to be collected to determine whether or not this is so. 39 3. Couduskos We have presented a considerable mass of evidence, most of it not well structured by any theoretical concerns. This is perhaps inevitable given the current state of the subject; aging in developing countries is an issue that looks like it might be important, but concern is still unfocussed on any particular set of economic research questions, or even outstanding policy issues. There are many large and attractively wooly creatures at loose in the forest: the role of development and the status of the aged, the relationship between marriage patterns, polygamy, living arrangements, and the treatment of the elderly, and what policy steps, if any, should be considered by those Asian countries that are facing rapidly rising shares of elderly inhabitants. But we are very far, not only from answers, but even from a well-defined set of topics that economists could usefully think about. Even so, we feel that we have learned something by looking at these data and by writing this paper, and it is perhaps useful to conclude by summarizing some of what is known, and what might usefully be learned: 1. Questions of the economic status of the old in LDC's are not answerable and have to be rethought. In more developed countries, where perhaps nine tenths of the elderly live by themselves or with elderly spouses, household surveys can tell us a great deal about their living standards. In LDC's, to a greater or lesser degree, older people do not live by themselves, and until a method can be found for measuring intrahousehold allocations, we have no method of assigning welfare levels to them, or indeed to other members of the households in which they live. 2. More work needs to be done on the question of whether the source of income, i.e. who earns it, affects what individual members of the household receive. This cannot be done directly, but if the earnings of the elderly are spent differently from other household income, the fact should be detectable from consumption data. Data such as those from Thailand show considerable variation in source of income with age, although the patterns are quite different from those in the Unites States or Western Europe. 3. In the U.S. and other developed countries, where many elderly people live alone, there has been concern about the possible abandonment of the old. However, such cases seem to be rare, most old people live alone because they want to do so, and frequency of contact with children is generally high, see Mancini and Blieszner (1989) for a review. In C6te d'Ivoire, under current living conditions, abandonment would seem to be an unlikely event, because very few old people live alone. There are perhaps more grounds for concern in Thailand, but the population at risk is still small, and is probably overstated by the survey results quoted here. However, there 40 is evidence from elsewhere that suggests that these results should not be generalized to al poor countries. In many areas of India, living arrangements for newly-weds are strictly patrilocal, so that, after marriage, women are effectively cut off from their parents' family. In turn, they will be looked after in old age by their sons, their daughters having themselves moved to their husbands families. In consequence, women who fail to produce sons, or fail to produce surviving sons, are likely to fall into destitution as widows. Drbze (1988) provides evidence on this problem, and highlights it as an outstanding issue for social security and poverty policy in India. 4. The living arrangements of the elderly will vary from place to place according to marriage arrangements, agroclhmatic conditions, and the availability of labor and land. The position of Indian widows has already been cited. In C8te d'Ivoire, living patterns have been changing in response to the increasing scarcity of land, since sons, who previously were guaranteed land nearby, now are often required to set up households at considerable distances. The shortage of land itself reflects a great deal of immigration to the cocoa and coffee areas, an immigration that responded to original labor shortage, and that contributed to the destruction of the original lineage system of cocoa and coffee production. One may also wonder whether the pattern of inheritance in Northern Thailand, whereby as a result of the residual stem family system, the youngest daughter typically inherits the land, will continue unmodified into an era where land is increasingly scarce. 5. Individual participation and earnings patterns show the standard life-cycle hump shapes in CMte d'Ivoire and Thailand, and presumably do so more widely. However, households act so as to make average living standards within households much less variable over the life-cycle than are the individual patterns. The degree to which this happens in the data is different between the two countries, and depends on how household size is measured. Even so, sharing resources between household members is presumably one of the main economic functions of the household. What needs a great deal more research is the extent to which household size and composition itself adapts to facilitate sharing, and to guarantee the best possible living standards to its members. In both Thailand and C8te d'Ivoire, there is a great deal of migration, on both a seasonal and non- seasonal basis. In Thailand, the process of household formation is explicitly tied to the pressure on resources within the compound, so that the departure of a previously married child on the marriage of a younger sibling is as much a matter of economics as of immutable custom. In the panel households in C8te d'Ivoire, there are major differences in membership between 1985 and 1986, and while there is undoubtedly some measurement error, careful attempts were made to link household members from one year to the next, and there is certainly a great deal of movement. 41 Fosterage of children, often children not closely related, is a widespread phenomenon in West Africa, see Ainsworth (1989), and provides a mechanism, not only for education, training and apprenticeship, but also for sharing economic burdens between members of the same lineage. There has been a good deal of emphasis on the role of risk sharing in determining patterns of marriage and migration, see for example, Rosenzweig (1989). But there is scope for more modeling here, particularly for a simple unifying theory that explains how potential household members decide how to form household groups given the economic opportunities available to them. 6. There are a number of interactions between urbanization and age distributions. Migration tends to lead to young cities and an older countryside, as is the case in C8te d'Ivoire, but much urban growth in LDC's comes from reproductive behavior, as well as from migration. The fall in fertility in the demographic transition often begins first in the cities, so that cities are likely to age more rapidly than more rural areas. The balances between these forces will produce different age distributions in different countries, for example, younger cities in Africa, and older cities in Asia, and these have a number of repercussions for policy, for example in the provision of services, as well as in the likely effectiveness of older people as a political force. 7. Many LDC's are in a state of transition, not only demographic, but also educational. In both countries examined here, there are very large differences between the educational attainments of the different generations. The consequences of these differences are much less clear, and we do not wish to subscribe to the view that they always and everywhere undermine the status of the old. Nevertheless, models that provide a theoretical framework for the role of the elderly would do well to bear these facts in mind. S. The life-cycle model of saving and capital accumulation, which has brought so many insights in developed countries, cannot be applied without modification to economies where the functions of households are different. Asset accumulation for old age, with a large share of the capital stock being accounted for (or not accounted for) by life-cycle saving, is not likely to be a very useful model for savings in LDC's. Households can and do provide old-age insurance without an obvious need to accumulate and decumulate assets. Our data do not suggest any run down of assets with the age of the household head. Of course, as in more developed economies, heads have a range of other motives for keeping control of assets for as long as possible. 9. As in developed countries, there is a pronounced household life-cycle, with a hump shaped income, pealdng much earlier in Cote d'lvoire than in Thailand. However, we doubt that there is much long-term consumption smoothing associated with these humps, and tend to attach 42 more importance to saving as a means of smoothing consumption over short-term fluctuations in income that are typically associated with agricultural activities. Indeed, it is possible that variations in household structure contribute more to long-term smoothing than do varations in assets. 43 References Ahonzo, Etienne, Bernard Barrere, and Pierre Kopylov, (1984), Population de la Cote d 'lvoire: Analyse des donnes denogragraphiques disponibles, Abidjan. Direction de la Statistique, Ministere de l'Economie et des Finances, Republique de Cote d'Ivoire. Ainsworth, Martha, (1989), "Economic aspects of child fostering in C8te d'Ivoire," The World Bank, processed. (September) Ainsworth, Martha, and Juan Mufioz, (1986), The Cote dI'voire living standards survey, Washington, D.C. Living Standards Measurement Study Working Paper No. 26, The World Bank. Carroll, Christopher and Laurence H. Summers, (1989), "Consumption growth parallels income growth: some new evidence," Departments of Economics, Harvard University and Massachusetts Institute of Technology, processed. Codey, John, Timothy M. Smeeding, and Barbara Boyle Torrey, (1990), "The change in the economic status of the low-income elderly in three industrial countries: circa 1979-1986," Washington, D.C. Bureau of the Census, processed. Cowgill, Donald O., (1974), "Aging and modernization: a revision of the theory," in D.O. Cowgill and L.D. Holmes (eds.), Aging and Modernization, New York. Appleton-Century-Crofts. Cowgill, Donald O., (1986), Aging around the World, Belmont, Ca. Wadsworth. Deaton, Angus, (1990), "Saving in developing countries: theory and review," World Bank Economic Review (Supplement), Vol. , Deaton, Angus, and Dwayne Benjamin, (1988), "Coffee and cocoa in the C8te d'Ivoire," Washington, D.C. Living Standards Measurement Study Working Paper No. , The World Bank. Deaton, Angus, and John Muellbauer, (1986), "On measuring child costs in poor countries," Journal of Political Economy, 94, 720-44. Deaton, Angus, Javier Ruiz-Castillo, and Duncan Thomas, (1989), "The influence of household composition on household expenditure patterns: theory and Spanish evidence," Journal of Political Economy, 97, 179-200. Dreze, Jean P., (1988), "Social insecurity in India: a case study," London School of Economics, processed. Ewbank, Douglas C., (1981), Age misreporting and age-selective underenumeration: sources, patterns and consequences for demographic analysis. Washington, D.C. National Academy Press. Hecht, Robert, (1982), "The transformation of lineage production in Southern Ivory Coast, 1920-1980," Ethnology, W, 261-77. Mancini, Jay A., and Rosemary Blieszner, (1989), "Aging parents and adult children: research themes in intergenerational relations," Journal of Marriage and the Family, 51, 275-90. Martin, Linda G., (1988), "The aging of Asia," Journal of Gerontology: Social Sciences, 43, S99-S113. 45 Martin, Linda G., (1989), "Living arrangements of the elderly in Fiji, Korea, Malaysia, and the Philippines," Demography, 26, 62743. National Economic and Social Development Board, (1985), Population projections for Ihailand 1980-2015, Bangkok. National Statistical Office, (1983), The survey of migration in Bangkok metropolis, Nonthabui, Pathun Thani and Sanut Prakai, Bangkok. Paxson, Christina H., (1989), "Savings in Thailand: responses to income shocks," Princeton. Research Program in Development Studies, processed. Pollak, Robert A., and Terence J. Wales, (1979), "Welfare comparisons and equivalent scales," American Economic Review, 69, 216-21. Rosenzweig, Mark R., (1988), "Risk, implicit contracts and the family in rural areas of low income countries," Economic Journal, 98, 1148-70. Treas, Judith, and Barbara Logue, (1976), "Economic development and the older population," Population and Development Review, 12, 645-73. United Nations, (1986), World Population Prospects: Estimates and Projections as Assessed in 1984, New York. United Nations. United Nations, (1987), Demographic Yearbook 1985, New York. United Nations. United States Agency for International Development, (1982), Selected Statistical Data by Sex, Part 13, Ivory Coast, Washington, D.C. World Bank, (1979), "The population of Thailand: Its growth and welfare," World Bank Staff Working Paper #337, June. World Bank, (1988), World Development Report 1988, Washington, D.C. World Bank, (1989), World Development Report 1989, Washington, D.C. 46 Distributors of World Bank Publications ARGENTINA FINLAND MEXICO 5OUTH AFRIC, BOTSWANA Cain. Hhkad SEL AkasmdrAui Kijkamyp Baumr Fwmbqkldfr GaIsdaGuenwe P.O. Box 121 Apartsdo Po=1.l22.440 OidwdlUhivaaryPnwSwashow EIAld&163,41h'5lixAXL4S3/46S SP.0oi IM apne6 ASic. 1333_ARBuw 4Ai1m6 Hdinid 10 14( a0Medn.DF. PD. Du1141 MOROCCO CeeTeaan AUSTRALIA4 PAPUA NEW GUINEAi, FRANCE SociA drEbBa MmelugMarocMne FUL SOLOMONISLANDS, Wrd Bank Plubticas 12ruwMae ar Md. 'Ards Frubhi( , VANUATU, ANDWESTERN SAMOA 66, avened't6ra C_na,mcP Ar Intbfaaliind ShmoipMonS Ia DA.Books&Journais 75116 Pari PD.. 8=d4095 648 Whiftdoe Road NETHERLANDS Crgha fkhan3132 GERMANY, FEDERAL REPUBUC OF InO,.Publwtfwbv. Jdunombur21024 VidErla UNO-Veldg P.O. Dm 14 P.ppldadm*rAllsSOS 724D BA Lodtii SPAIN AUSTRIA D.531 Ben 1 M74di-Pm. EM UtE SA. Gtvdland Co. NEW ZEALAND ICatdIo37 Grabe,31 GREECE little lk,y ad bnnfogniaM. Sule 2M MdId A-lOtl Wbro KEME pNvaSeBa 24, Ippodamom SRnt Platla 1a4ew Market L11e(a Inadomd AEDOS BAHRAIN Abthiw11635 Auddatd C-11 de Cen. 391 Baw.i. Rearch and Cldtay sa- Anocdaies ld. GUATEMALA NIGERIA P.O. Bm 22103 LUfra Pledn Santa ,by Unhiqy t Aed SRI LANKA AND THE MALDIVES M.aia, Toan 317 S. CalUe 7-55 TeCownsudl,aicio LakeIHoauseliodinp Zn.. 1 FrIivteMi gRsmoS P.O. B=244 BANGLADESH Guelmela ty bAldata 10FB StrOhtlanpdazA.G fra Micro hidustir Dedopmet Mawasha A_id-an SaDdety AIDAS) HONG KONG MACAO NORWAY C wo8bo 2 House S, Road 16 A7CM lAd. Nav-e hdrationC Dhaznondi R/An 6 Fl. 146 Prina Edwrd BSokDeydment SWEDEN Dh.ke 1209 Rad, W. P.O. Em 6125 _ser.a sewdeg riwac KowlNo00 N-062 (Mo 6 FHitm Fackboloeet Bmuh vim Hong Kmg RegrhgatngIt 1,3 .16M56 156, Nwf Ahred Swsk OMAN S103275 oddim Chittagog4 tOg INDIA MEMSB hbaun S,i -1 Alied PublihePrsivate Ltd. P.0.1=1613.SeubAkprt Fcre 76, KDA. Avenue 751 Mount Road Muncat Wen.gna-WilharnAB Kuln Macre-EO00012 B20D4 PAKISTAN S-10425 J DeLAnnoy ISJ.N. HerediaMg 65, 5 h dd SWIERLAND Av. du Rad 202 Balard BaEale P.O. BNo. 72o Forsim5kliiw 100 Branda Bnbay-40D038 L*ore3 LibrqriePanet CANADA 13/14A.dAl Road PERU Coepostae3i LeLDtoeur NewDi. 110002 Edjkweal D.rllIo SA CHM1211 Gae.ll C.P 5.1 1531BreAnpine Apartd*3124 Bachravil Quhec 17Chittarnj;_Avenue Lizaa Fo. JIB SE6 Calcuta-700072 IPrSifrieePayot FHILIFFiNES Set,icedeAboawseenrts CHINA Jay eva Hood Building Interat iBookn B rrkec C_sepesde33t2 hil Russ"aA, Eanatic Pubhsr3 S lth Man Rod GndhnagrF PAw om Fbpina UAe Deiding CHII10G2 L _usam House Bargalore. 09 09 Ayda Avenue. Ma B. 1. P.StDatgjle MiebroMard1. TANZANIA Befg 35129KhdgrudaCe..Road Odai dUnbartyFin. Hyderabad-S0027 POLAND PD. Bm1S9 COLOMBIA ORPAN D rsale r hIo.00 Ltda. Prutlrnl RFloe 2trd Fboor PN KlbhoyI Narud Apar-do Arno 34270 NesrTh qkareB.Navrwxra O.91 waivwa THAILAND BegotaDE. Ahmedabad .350DO4 CntraDepterto e PORTUGAL 3D6 Sda.. Road COTE D'IVOIRE PaFla Houe Uvewla Pattral B-16kok Cd'Ednrd ar et deDlumien 16.AAAhokMag Rt. Do Ciano 70.74 ASHcEDEA Lacknow-226001 120 Ltdu TRINIDAD ATOBAGO, ANrIGUA 04B.P.541 IURSUDA. BARBADOS, AbidjnO4(Pl Peu INDONESIA SAUDIAAIUABQATAR DOMINICA,GRENADA.GUYANA, Pft bIdir UmLited PWO.Book31°6 JAMAAICAMONTFR SRAT. ST. CYPRUS JL. Sam Ratulangi 37 P.0.8=od 3196 KITrS & NEVIS. ST. ULUCIA, MEMRESf fauitionService P.O.Bm181 Riydlrl471 ST. VINCENT &GRENADINES P.O. Box 29 Jlkw Purt SpaatnatudieaUnit Niceda MEMRB Ioin_Wm Sevc #SWauSaeed ITALY Brandum Conepe DENMARK Ltcsa Cmatuda Sascrnt SPA AUAIl Steai Trhidad, West idi-e SaatfurdsLAra*tur Via Bmaedetiol FmtL 120/10 Al Dehn Cntcr RowneecasAD^IId CArda Ftlc,Sl2 Finst oor UNITED ARAB EMIRATES DrKl9W Frederflabrg C 5D175F ucr P.O.Bm718 MEMRBGulfCo. Riyadh ~~~~~~P.O.8=m0997 DOMINICAN REPUBLIC JAPAN Ryhaxja 9. EdiorTllUer,C.prA. Eatn BookSwv.im HoAb utah Alima tBudjng Reauard6.l.abdIsCgdEcaRE30 374-3 Hoogo 34hanut Bunkya.ku113 ling lald St UNltED KNGDOM Aprtado Potl 2190 Tokyo P.O. B0 3969 MkinfoLtd. SEito Datnirio Dsatutmn P.. Bm 3 KENYA &Jk HaatpailreGU134 2PG ELSALVADOR ASikaBookSkeiElA)lid. 33, Mdluatuned Haws, Awad StRet Erngland Fundes P.O. Bx 4245 P.O. DM 95 Avedda Manud BEnrique Arnvo #353) Nairbi jaddah VENEZUELA Edtldo SSA. la. FBa LUxeta del Ete S. Salvader KOREA, REUBUC OF SINGAPORE, TAIWAN, MYANMAR, Aptdo. 6037 Pan Koe Book Cpmaia BRUNEI Carae 1060-A EGYPT, ARAB REUBLIC OF P.O. B101, Kwagwhamun fdon Pub Al Airrau Puivao. [ld. YUGOSLAVIA Al Gala StrdK 02X06 It Fl. Pdi-Pu Induaial JUPOElooxm Kug cair, XKWAIT Bldg. P.O.8B=36 MIMRB lo nrmata Saen.. 24 New Indutrial Road Trg Reulke TheMiddleEnatObnerv P.O. B S465 Sengyrel1953 YU.LMIDBeigrade B CoawarM Sted Caro MALAYSLA Unrivnity oMayaCCooputive Bookdho, Liadted P.O. Box12J27,juan Prti Baru Kual Lbanp LSMS Working Papers (continued) No.46 Nonagricultural Family Enterprises in C6te d'Ivoire: A Descriptive Analysis No. 47 The Poor during Adjustment: A Case Study of C6te d'Ivoire No.48 Confronting Poverty in Developing Countries: Definitions, Infonnation, and Policies No.49 Sample Designs for the Living Standards Surveys in Ghana and Mauritania/Plans de sondage pour les enquetes sur le niveau de vie au Ghana et en Mauritanie No.50 Food Subsidies: A Case Study of Price Reform in Morocco (also in French, 50F) No. 51 Child Anthropometry in C6te d'Ivoire: Estimates from Two Surveys, 1985 and 1986 No.52 Public-Private Sector Wage Comparisons and Moonlighting in Developing Countries: Evidence from C6te d'Ivoire and Peru No. 53 Socioeconomic Determinants of Fertility in C6te d'Ivoire No.54 The Willingness to Pay for Education in Developing Countries: Evidence from Rural Peru No.55 Rigidite des salaires: Donne's microeconomiques et macro6conomiques sur l'ajustement du marche' du travail dans le secteur moderne (in French only) No.56 The Poor in Latin America during Adjustment: A Case Study of Peru No. 57 The Substitutability of Public and Private Health Care for the Treatment of Children in Pakistan No. 58 Identifying the Poor: Is "Headship" a Useful Concept? No. 59 Labor Market Performance as a Determinant of Migration No. 60 The Relative Effectiveness of Private and Public Schools: Evidencefrom Two Developing Countries No.61 Large Sample Distribution of Several Inequality Measures: With Application to Cote d'Ivoire No. 62 Testing for Significance of Poverty Differences: With Application to C6te d'Ivoire No.63 Poverty and Economic Growth: With Application to C6te d'Ivoire No. 64 Education and Earnings in Peru's Informal Nonfarm Family Enterprises No.65 Formal and Informal Sector Wage Determination in Urban Low-Income Neighborhoods in Pakistan No. 66 Testing for Labor Market Duality: The Private Wage Sector in C8te d'Ivoire No. 67 Does Education Pay in the Labor Market? The Labor Force Participation, Occupation, and Earnings of Peruvian Women No.68 The Composition and Distribution of Income in COte d'Ivoire No. 69 Price Elasticities from Survey Data: Extensions and Indonesian Results No.70 Efficient Allocation of Transfers to the Poor: The Problem of Unobserved Household Income No. 71 Investigating the Determinants of Household Welfare in Cote d'Ivoire No. 72 The Selectivity of Fertility and the Determinants of Human Capital Investments: Parametric and Semiparametric Estimates No.73 Shadow Wages and Peasant Family Labor Supply: An Econometric Application to the Peruvian Sierra No. 74 The Action of Human Resources and Poverty on One Another: What We Have Yet to Learn No. 75 The Distribution of Welfare in Ghana, 1987-88 No. 76 Schooling, Skills, and the Returns to Government Investment in Education: An Exploration Using Data from Ghana No.77 Workers' Benefitsfrom Bolivia's Emergency Social Fund No. 78 Dual Selection Criteria with Multiple Alternatives: Migration, Work Status, and Wages No. 79 Gender Differences in Household Resource Allocations No.80 The Household Survey as a Toolfor Policy Change: Lessonsfrom theJamaican Survey of Living Conditions The World Bank CI Headquarters European Office Tokyo Office E0 1818 H Street, N.W. 66, avenue d'Iena Kokusai Building Washington, D.C. 20433, U.S.A. 75116 Paris, France 1-1, Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan Telephone: (202) 477-1234 Telephone: (1) 40.69.30.00 o Facsimile: (202) 477-6391 Facsimile: (1) 47.20.19.66 Telephone: (3) 3214-5001 Telex: wUi 64145 WORLDBANK Telex: 842-40651 Facsimile: (3) 3214-3657 RCA 248423 WORLDBK Telex: 781-26838 Cable Address: INTBAFRAD WASFHINGTONDC