L.S.ITl5S LSM - 68 Living Standards MARCH 1990 Measurement Study Working Paper No. 68 The Composition and Distribution of Income in Cote d'Ivoire LSMS Working Papers No. 1 Living Standards Surveys in Developing Countries No. 2 Poverty and Living Standards in Asia: An Overview of the Main Results and Lessons of Selected Household Surveys No. 3 Measuring Levels of Living in Latin America: An Overview of Main Problems No. 4 Towards More Effective Measurement of Levels of Living, and Review of Work of the United Nations Statistical Office (UNSO) Related to Statistics of Levels of Living No. 5 Conducting Surveys in Developing Countries: Practical Problems and Experience in Brazil, Malaysia, and the Philippines No. 6 Household Survey Experience in Africa No. 7 Measurement of Welfare: Theory and Practical Guidelines No. 8 Employment Data for the Measurement of Living Standards No. 9 Income and Expenditure Surveys in Developing Countries: Sample Design and Execution 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 Measuremient 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 Cote d'Ivoire Living Standards Survey: Design and Implementation No. 27 The Role of Employment and Earnings in Analyzing Levels of Living: A General Methodology wuith 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 Cote d'Ivoire No. 34 Guidelines for Adapting the LSMS Living Standards Questionnaires to Local Conditions No. 35 The Demand for Medical Care in Developing Countries: Quantity Rationing in Rural Cdte d'Ivoire (List continues on the inside back cover) The Composition and Distribution of Income in Cote 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 household 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 government policies, and to imnprove communications between survey statisticians, analysts, and policy- makers. The LSMS Working Paper series was started to disseminate intermediate products from the LSMS. Publications in the series include critical surveys covering different aspects of the LSMS data collection program and reports on improved methodologies for using Living Standards Survey (LSS) data. More recent publications recommend specific survey, questionnaire, and data processing designs, and demonstrate the breadth of policy analysis that can be carried out using LSS data. LSMS Working Paper Number 68 The Composition and Distribution of Income in Cote d'Ivoire Valerie Kozel The World Bank Washington, D.C. Copyright i 1990 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 March 1990 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. 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 subjects, authors, and countries and regions. The latest edition is available free oi 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. Valerie Kozel is an economist in the Welfare and Human Resources Division of the World Bank's Population and Human Resources Department. Library of Congress Cataloging-in-Publication Data Kozel, Valerie, 1952- The composition and distribution of income in C6te d'Ivoire / Valerie Kozel. p. cm. - (LSMS working paper, ISSN 0253-4517; no. 68) Includes bibliographical references. ISBN 0-8213-1446-7 1. Income-Ivory Coast. 2. Income distribution-Ivory coast. I. Title. II. Series. HC1025.Z915144 1990 339.2'096668-dc2O 90-11932 CIP ABSTRACT Empirical work on the distribution of welfare typically uses total household consumption expenditures in lieu of total income in measuring welfare levels. However, households must produce income to obtain consumption goods, and thereby welfare levels. Thus the study of income determinants and composition is a necessary adjunct to the study of welfare distribution, and ultimately to the study of poverty. The paper uses data collected under the auspices of the World Bank Living Standards Unit (now the Welfare and Human Resources Division, in the Population and Human Resources Department) in the Republic of C8te d'Ivoire in 1985. The major findings of the paper are (i) roughly one-third of private income is obtained in the wage sector, slightly less than one-third from agriculture, one-fifth from non-farm self-employment, and the remainder from a variety of other sources (imputed rents, rents and dividends, social security and pensions, private transfer payments); (ii) households obtaining income from wage activities are among the wealthiest (measured in terms of income and consumption) in the C8te d'Ivoire, while farm households are among the poorest; and (iii) physical assets, with the exception of land, tend to be highly concentrated in urban households at the upper end of the income distribution; in contrast, land is more equitably distributed in rural areas, and human capital, measured in terms of education, more equitably distributed in urban areas. - vii - TABLE OF CONTENTS I. Introduction .... 0... 0.... ................ *............... ......... 4-*0*01 II. Composition of Income .... ..................... 4 III Income Distribution.**.......................................... 20 Cumulative Shares of Income and Consumption Expenditures.....30 IV. Composition and Distribution of Assets .......................... 36 Composition of Household Assets ........................ ..... .36 Distribution of Assets.*..********..** ... ........ **#e60646 V. Summary and Conclusions ............. . ..... .. ............ ......50 ANNEX I Imputed Rents for Urban Households..* ........................52 ANNEX II Annualized Value of Durables ............66 References ............................................. LIST OF TABLES Table 1 Components of Household Income ....5 Table 2 Average Annual Wages and Hours Worked per Earner in C8te d'Ivoire By Employment Sector and Region. ....... ..........6 Table 3 Composition of Farm Income in C6te d'Ivoire .........8...........8 Table 4 Composition of Net Farm Income in C6te d'Ivoire by Agricultural Region ............................9..00*0*000*9 Table 5 Non-Farm Self-employment Income and Expenditures per Household by Region in C8te d'Ivoire .12 Table 6 Composition of Household Income in Cote d'Ivoire ....9..........15 Table 7 Total and Adjusted Per Capita Income by Region and Activity Classification in Cote d'Ivoire ..... .*. ... .18 Table 8 Composition of Household Income in C8te d'Ivoire, by Region and Adjusted per Capita Income Quintile within each Regioneo...21 Table 9 Composition of Household Income in Cote d'Ivoire, by Region and Adjusted per Capita Expenditure Quintile within each Region ...... ....... ***e*24 - viii - Table 10 Composition of Household Income in C8te d'Ivoire by Per Capita Expenditure Decile.... ............ 27 Table 11 Percentage Distribution of Households by Adjusted Per Capita Expenditure Quintile and Activity Classification in C6te d'Ivoire ... s - e .s e..... .*... e * .o .*e*......................o.......29 Table 12 Distribution of Adjusted Household Income and Consumption Expenditures in C6te d'Ivoire by Adjusted Per Capita Income and Expenditure Deciles ........................... .31 Table 13 Composition of Household Income and Consumption Expenditures in C8te d'Ivoire by Region and Adjusted Per Capita Income and Expenditure Quintiles within each Region........ 33 Table 14 Gini Coefficients for Household Income and Consumption Expenditures by Region and Total Country: C6te dvoire ...... ... **go . .... * * * .35 Table 15 Composition of Household Assets by Region and Total Country: C6te d'Ivoire. . ............ ........ ... .... o.40 Table 16 Composition of Household Assets by Income Source Category in Cote d'Ivoire .............. .............. 44 Table 17 Gini Coefficients for Selected Household Assets by Region in C8te d'Ivoire.........................** **.**... *. *..... 48 ANNEX TABLES Table I-1A Means and Standard Deviations of Independent Variables: Indicator Function for Tenure Choice .................... 59 Table I-1B Means and Standard Deviations of Independent Variables: ..... 61 Hedonic Rent Equation Table I-2 Indicator Functions for Choice of Housing Tenure, Renters Versus Owners ..................................... ......62 Table I-3 Hedonic Rent Equations Corrected for Sample Selectivity ..... 64 Table II-1 Percentage of Households Owning Durables, By Durable Category and Region..... *..........*...............67 Table II-2 Estimated Depreciation Rates by Type of Durable.............69 - 1 - I. INTRODUCTION Interventions designed primarily to maintain a household's living standard -- for example, food subsidies, transfers, public health care, and other social welfare programs -- are effective only in the short-term in alleviating the conditions of poverty. Longer-term solutions necessarily entail increased productivity and broader income earning opportunities for those most in need and therefore most vulnerable to income shortfalls. In developing strategies to improve the income earning potential of the poor, one must first identify what are the sources of income and what determines its absolute level (for example, fixed assets, land, labor, and human capital endowments). Low income is typically the result of low private asset endowments, augmented by problems of access to credit and critical public services, such as schools, health facilities, production subsidies in the form of fertilizer or seeds, extension services, and so on. The paper takes a very careful look at the structure and distribution of income and private assets in the Republic of C6te d'Ivoire. The purpose is to construct a base of information which would allow one to identify some of the causes of income inequality and poverty levels, and to assess the impact of macroeconomic growth policies on future income levels and living conditions of the poor. The data used are from a 1985 survey of 1600 households conducted nationwide in the C6te d'Ivoire under the auspices of the World Bank, Living Standards Measurement Study. The C6te d'Ivoire Living Standards Survey (CILSS), described in detail elsewhere (Grootaert, 1986, Ainsworth and Mufnoz, 1986), includes extensive information on earned income derived from wage activities, farming, and non-farm family enterprises, as well as unearned - 2 - sources of income. Information on productive and so-called non-productive assets is also available. Similar data were obtained from surveys in 1986 and 1987; an analysis of changes in income composition and distribution will be completed shortly. The results presented here are disaggregated along two dimensions -- urban (Abidjan, other urban areas) versus rural location, and by income source categories. In the last case, five categories are used: (i) households wholly dependent on wage income, (ii) households wholly dependent on income from farming and livestock, (iii) households dependent on income from self- employment, but not solely farm earnings, (iv) households receiving in,come from both self-employment and wage activities, and (v) households who receive no earned income in the reference year. There is a high correspondence between the two classification schemes; for example, nearly all wage employment is located in large cities, while farming (and thereby farm income) is a predominantly rural phenomenon. This correspondence has important potential policy implications. If, for example, we were to find (as we have) that urban households are substantially better off than rural househoLds, then growth policies designed to increase incomes and thereby improve the welfare of rural households to be effective must be oriented towards the agricultural sector or towards explicitly expanding the role of rural off-farm activities. Further, if urban wage earning households are found to be better off than their self-employed urban counterparts (as once again they are), then growth strategies directed at wage sectors (for example, modern large-scale industry) will not only exacerbate urban/rural welfare differentials, but will also worsen the distribution of income within urban areas. The paper has three main sections and a short summary section. The first section describes how income has been measured, and analyzes component shares by region and productive activity category. The second section introduces a poverty dimension to the analysis, and discusses the distribution of income (by source) based on per capita income and adjusted per capita expenditure (or welfare) quintiles. The composition and distribution of household assets is the subject of the final main section. The paper concludes by a brief summary of research findings, policy implications, and directions for future work. - 4 - II. COMPOSITION OF INCOME Income is defined as the returns to household labor inputs and capital stocks, plus the annualized flow of services from durable goods and the housing stock. It has nine major components which are defined in Table 1: (i) wages; (ii) farm income; (iii) non-farm self-employment income; (iv) capital and interest income; (v) income from forced savings; (vi) other unearned income; (vii) private transfer income; (viii) imputed rents; aLnd (ix) income from durable stocks. For purposes of this paper, no effort was made to impute a flow of income to time spent in home production activities (housework, child care, and the like), nor to leisure time -- in short, no effort was made to measure full income (see Becker, 1965, Gronau, 1976, Evenson and Quinzon, 1977). Wage income is derived from Section 5 of the CILSS questionnaiLre, which describes time use and earnings for all individuals aged 7 years or older. Respondents described earnings and time inputs for up to four unique wage-paying jobs over the 12 month period preceding the interview. Information was solicited on in-kind as well as monetary payments; of the 785 individuals who described themselves as employees (out of a universe of 10,067 potential employees), some 43 percent received some form of in-kind payments, which constitute an average 12 percent of total wage remuneration. Note that 142 persons, or some 18 percent of employees, did not receive cash payments for their work. Upon inspection, the bulk of these appear to be young (15 to 25 years old) persons serving unpaid apprenticeships. Table 1: Components of Household Income Income Component Description 1. Wage Cash and in-kind income from employment 2. Farm Net revenues from crop and livestock activities Net revenues from agriculture product sales 3. Non-farm Family Net revenues from non-farm self-employment Enterprise 4. Capital and Interest Land and buildings rental income, dividends and interest payments 5. Social Security, Income from social security and pensions Pensions, etc. 6. Other Unearned Income Income from grants, scholarships, gifts, dowery and inheritance, etc. 7. Private Transfer Income Gross income from private transfer payments 8. Imputed Rents Net imputed value of housing services from residing in one's own dwelling unit (rural households excluded) 9. Service Flows from Annual imputed value of services from Durables durables Table 2 shows average annual wages and hours worked per earner for total workers, public sector workers, and private sector workers (roughly forty percent and sixty percent of the total salaried labor force, respectively), by region. In all regions (although most markedly in Abidjan and in rural areas), employees in the public sector earn more than their private sector counterparts. However, as shown by van der Gaag and Vijverberg (1987), wage differentials primarily reflect differences in human capital - 6 - endowments (for example, work experience, educational attainment) rather than differential returns to endowments. For example, individuals employed in the public sector completed an average of 9.2 years of schooling, while private sector employees completed only 5.3 years of schooling. Not surprisingly, the within-region variance in private sector wages is higher than for public sector wages, reflecting greater heterogeneity of private sector activities -- low skill laborers are intermixed with senior-level managers. Table 2: Average Annual Wages and Hours Worked per Earner in CMte d'lvoire by Employment Sector and Region Abidjan Other Urban Rural Total Country Total Wages per Year (CFA) - overall 1,420,185 1,362,437 543,573 1,259,753 - public sector 2,164,657 1,518,032 1,358,278 1,784,851 - private sector 1,009,972 1,159,211 300,972 900,385 Percent of Wages Paid in Cash 89.4% 88.1% 88.1% 88.8% Total Hours Worked Per Year - overall 2195 2169 1312 2041 - public sector 1926 2014 1413 1922 - private sector 2130 2005 1144 1897 Source: CILSS 1985 Survey estimates. Farm income is derived from Sections 9 and 12 of the CILSS questionnaire. Section 9 is the source of all agriculture related measures except the value of home-produced agricultural goods consumed at home, which is drawn from Section 12. Following Singh and Asokan (1981), farm income is defined as the returns to family labor and productive assets. It is calculated as the difference between gross farm income and expenditures on variable inputs, and includes (i) net income from crop sales, (ii) net income from livestock sales and changes in livestock holdings due to non-market transactions (gifts, births, deaths), (iii) net income from the sales of products made from agricultural outputs, (iv) the estimated value of agricultural outputs consumed at home, and (v) land and equipment rents. Measurement problems were encountered in apportioning sharecropping income between the landlords and the sharecroppers; survey results indicate that sharecropping arrangements are more akin to a labor-buying than a land- leasing system in the C8te d'Ivoire. For the most part, the farmer retains control of the land and provides necessary inputs. The sharecropper essentially provides muscle, for which he or she receives one-half to one- third of total output. Most of the sharecropping arrangements in the sample involved cash crops, for example, coffee and cocoa. Sharecroppers in Cate d'Ivoire tend to be immigrants (or, autochtones) from other parts of the country or other West African countries, attempting to work their way into the rural economy in areas with a land abundance and rich growing conditions, (Ruf, 1984). It is important to note that only 36 percent of farm households surveyed claimed to be able to sell their land should they choose. For the remainder, hired labor or sharecropping arrangements are the only way to adjust factor proportions between land and labor inputs. Civen this situation, it is not surprising that nearly 27 percent of all farm households participate in sharecropping arrangements. - 8 - Table 3: Composition of Farm Income in G6te d'lvoire (CFA) Value of Hore Percentage Shares Gross Income Expenditures Net Income Consumption of Net Income Crop Production 836,726 187,264 649,462 285,328 88,2 Livestock 41,317 4,315 37,002 13,104 5,0 Agriculture Product Sales 37,765 1,483 36,282 - 4.9 Agriculture Rents 16,536 3,168 13,368 - 1.9 Total Agriculture Activities 932,344 196,230 736,114 298,432 100,,0 Source: 1985 CILSS Survey estimates. Table 3 shows the composition of agricultural income in the whole country. In our estimates, the average annual value of farm output (including outputs consumed at home) is CFA 932,344 per household, which is produced at a cost of 21 percent of gross output, or CFA 196,230, and thus yields a net farm income of CFA 736,114 per household. Farm output is mainly derived from crop cultivation. The value of home-consumed goods constitutes some 40 percent of the average value of net farm output (CFA 285,328 for crop consumption, and CFA 13,104 for consumption of livestock-related outputs). Table 4 shows the composition of net farm income stratified by the three major agricultural regions in Cote d'Ivoire -- West Forest, East Forest, and the Savannah. Farmers in the West and East Forests earn significantly more on average than do farmers in the northern Savannah region. In all cases, however, income -9- from crop cultivation constitutes the main portion of total income from farming activities. Table 4: Composition of Net Fare Income in C6te d'lIvoire by Agricultural Region West Forest (N=239) East Forest (N=478) Savannah (N=329) Avg. CFA Percent Avg. CFA Percent Avg. CFA Percent Crop Production 829,675 88.1 721,582 88.8 413,765 87.6 Livestock 37,330 4.0 39,775 4.9 32,737 6.9 Agriculture Product Sales 69,967 7.4 24,425 3.0 26,133 5.5 Agriculture Rents 4,790 0.5 27,168 3.3 -448 0.0 Total Net Income 941,852 100.0 812,950 100.0 472,187 100.0 Source: 1985 CILSS Survey estimates. Income from non-farm self-employment is derived from Sections 5 and 10 of the CILSS questionnaire. Severe problems were encountered in attempting to compute the measure based on Section 10 alone, although this was the intended purpose of the section. Section 10 has three parts; (i) information on revenues and structure of the family enterprise, (ii) information on business expenditures, and (iii) an accounting section on business assets. Information is collected for a maximum of three enterprises per household. On average, a third of these were managed by the household head, a third by his/her spouse, and the remainder by some other household member. Over 50 percent of enterprises sold foodstuffs, and 25 percent were involved in other forms of commerce. - 10 - Based on the data reported for Section 10, 65 percent of enterprises reported negative profits. Clearly these figures are not realistic. Evidence suggests that spurious net revenue estimates are caused by an underestimate of gross business revenues. It is particularly difficult to obtain estimates of gross revenues for small enterprises (involving one or two family members), which constitute over 80 percent of all enterprises sampled in the CILSS, probably because the budget for the business and the budget for the household are seldom maintained separately. Consider the following example: A woman makes meat pies at home to sell at noon: she buys meat, vegetables, and flour in bulk. She uses some of the flour to make bread for the family (which the questionnaire attempted to account for), some of the vegetables for dinner, and her children take some of the meat pies to school for lunch (also supposedly accounted for in the questionnaire). She makes a tray of pies and goes to the market to se]Ll them. Later in the day, she gives some of the money received from sales to one of her children for school supplies. She stops on her way home and buys rice and milk for the next day's meals. After she gets home, a man from a CILSS survey team comes to ask some questions about her "business" (selling meat pies). In particular, he asks her to estimate her total revenues since the last visited (roughly two weeks ago). What he would like her to do is estimate the number of meat pies sold and multiply by some price per pie. What she does may be very different, however. For example, she might think of the money in her pocket at the end of the day, which means that her relply would reflect net business revenues minus payments for household consumption expenditures. Alternatively, if she buys supplies in bulk and infrequently, the money in her pocket might represent gross revenues minus some payments for - 11 - household consumption expenditures. Recent field experiments suggest that both kinds of responses are not infrequent, and are far more typical of an individual's responses than attempts to estimate actual measures of gross business revenues. In either case, our prototypical respondent does not think of her business budget as logically separate from the household budget. Because of this problem, net income from family businesses is estimated using a mix of information from Section 10 and Section 5 (which describes individual time use and earnings). This requires that individuals be merged across households in Section 5 and enterprises be merged across households in Section 10 (it is not possible to obtain an accurate match of individuals with enterprises). Finally, information is matched between the two sources at a household level. Thus, the basic unit of observation is the household rather than the individual enterprise for income estimates, and enterprise level information cannot be fully recovered. Table 5 shows average business revenues and expenditures per household involved in non-farm self-employment for Abidjan, other urban areas, and rural areas, further stratified by a proxy measure for large (net revenues greater than CFA 3,000,000 per annum) and small enterprise households. Thirty seven percent of all households in C8te d'Ivoire report some non-farm business activity, which encompasses 46 percent of households in Abidjan, 54 percent of households in other urban areas, and 27 percent of rural households.!1 Net / These percentages are slightly different than those obtained by dividing counts in Table 3 due to a slight underreporting of business activities in the section of the questionnaire dealing specifically with family enterprise activities (Section 10), rectified through use of information from the general employment section (Section 5). - 12 - Table 5: Non-Farm Self-employment Income and Expenditures per Household by Reglion in C8te d'lvoire (CFA) Number of Gross Total Net Households Revenues Expenditures Revenues Abidjan Large Enterprise Households (>3,000,000 CFA/Year) 17 45,452,463 37,854,213 7,598,250 Small Enterprise Households (<3,000,000 CFA/Year) 137 1,780,137 1,122,959 657,178 Total Households 154 6,601,108 5,177,708 1,423,400 Other Urban Areas Large Enterprise Households (>3,000,000 CFA/Year) 7 7,107,402 3,213,464 3,893,938 Small Enterprise Households (<3,000,000 CFA/Year) 179 1,942,746 1,399,013 679,915 Total Households 186 2,138,165 1,467,299 801,527 Rural Areas Large Enterprise Households (>3,000,000 CFA/Year) 7 9,371,968 4,518,736 4,853,231 Small Enterprise Households (<3,000,000 CPA/Year) 247 1,676,867 1,341,638 374,960 Total Households 254 1,892,330 1,429,196 500,351 Total Country Large Enterprise Households (>3,000,000 CFA/Year) 31 28,646,692 22,504,743 6,141,949 Small Enterprise Households (<3,000,000 CFA/Year) 563 1,787,036 1,306,667 541,530 Total Households 594 3,200,702 2,412,963 836,289 Source: 1985 CILSS Survey estimates. - 13 - and gross business incomes are on average highest in Abidjan and lowest in rural areas, and a greater proportion of households have "large" businesses (in a revenue sense) in Abidjan than in other regions of the country. Capital and interest income is derived from Section 14 of the CILSS questionnaire, and includes rents on buildings and land (with the exception of land used for agriculture), dividends, and interest payments. Income from "forced" savings includes payments from pension plans and social security, which are also reported in Section 14. Information on private transfer payments is also provided in Section 14. Some 23 percent of households in the Cote d'Ivoire receive income through private support systems, with the vast majority of payments received from individuals related by blood or marriage. Two imputations are made for total household income estimates, one for the annual value of housing services (net of maintenance costs) for home owners, and one for an imputed annual stream of income from durables owned by the household. These imputations are described in detail in Annexes I and II for housing and durables, respectively. Imputed rents were estimated using selectivity-corrected hedonic rent equations (see Heckman, 1979, Lee and Trost, 1978, Malpezzi, et al., 1985). An estimate of the annual flow of services from durables is simply computed as the product of the annual depreciation rate times the present value of the durable, summed over the total stock of durable goods owned by the household (van der Gaag, mimeo, 1984). Table 6 shows the percentage of households receiving income in each of the nine income categories, average income per household, and category shares by region and for the country as a whole. From this table, it is clear - 14 - that income sources vary significantly between regions. In Abidjan, for example, 71.6 percent of households have at least one member working for wages, and 45.8 percent receive income from non-farm self-employment. In contrast, only 10.5 percent of rural households receive wage income and 27.2 percent receive non-farm family enterprise income. However, nearly all rural households (93.5 percent) receive income from farm activities, while few urban households do (4.5 percent). The category labeled "other urban areas" typically evidences a pattern of income composition that lies between Abidjan's and that of rural areas -- urban households are more likely to receive wage income than rural households, less likely to receive farm income, but more likely to receive non-farm self-employment income than either households in Abidjan or rural areas. It is interesting, although not surprising, to find that some 24.6 percent of households receive income from pensions or social security payments in Abidjan, while only 16.9 percent do in other urban areas, and 1.6 percent in rural areas. The reasons for this are simple; only individuals who hold government or private sector jobs are eligible for pensions (government) or social security (private sector) coverage upon retirement, and most employees work in urban areas. The poverty implications are important, however. Rural households almost entirely lack access to official safety nets, and instead must depend on community or kinship ties in times of need. This may increase their vulnerability to exogenous income shocks (such as those caused by droughts). Rural households also receive little income from rents or dividends in contrast to urban households, further exacerbating potential problems. Table 6: Composition of Household Income In CMte d'Ilvoire Abidjan (N=334) Other Urban (N-322) Rural (N=898) Total Country (N=1564) Percent Average Percent Percent Average Percent Percent Average Percent Percent Average Percent Receiving Income Share Receiving Income Share Receiving Income Share Receiving Income Share CFA CFA CFA CFA (1) Annual Wage Income 71.6 1,454,201 51.5 52.1 907,291 46.9 10.5 65,628 6.4 32.4 540,830 33.8 (2) Net Agriculture Income 4.5 42,431 1.5 41.6 188,148 9.7 93.5 766,838 75.0 63.5 489,295 30.6 (3) Net Income from Family Enterprises 45.8 652,961 23.1 53.6 437,669 22.6 27.2 136,141 13.3 36.8 310,517 19.4 (4) Rents and Dividends 26.9 276,672 9.8 25.6 107,593 5.6 8.5 19,259 1.9 16.0 92,982 5.8 (5) Income from Social Security Pensions, etc. 24.6 90,052 3.1 16.9 53,276 2.8 1.6 3,911 0.4 9.7 32,786 2.0 (6) Other Unearned Income F (gifts, Scholarships, etc.) 37.7 99,986 3.5 40.1 37,141 1.9 25.6 14,540 1.4 31.3 37,585 2.3 I (7) Gross Annual Transfer Income 21.9 69,099 2.5 24.4 19,902 1.0 23.5 7,343 0.7 23.3 23,197 1.5 (8) Imputed Rents for Urban Homeowners 22.5 97,145 3.4 44.6 150,032 7.8 - - 1/ - 14.3 52,594 3.3 (9) Annual Value of Durables 91.3 40,463 1.5 92.2 33,841 1.7 73.1 8,936 0.9 81.0 20,955 1.3 Total Annual Household Income - 2,823,011 100.00 - 1,934,893 100.00 - 1,022,597 100.00 - 1,600,743 100.00 Notes: I/ No lmputations were made for rural areas due to the general absence of a housing rental market, and constraints on property sales. Source: 1985 CILSS Survey estimates. - 16 - Based on compositional share estimates, wages form an important component of income in urban areas, followed by income from non-farm family enterprises. Rural incomes are composed primarily of agricultural earnings (75 percent of the total); only 20 percent is derived from non-farm self- employment and wage activities. Overall, nearly a third of household income in the Cote d'Ivoire is derived from wages, slightly under a third from agriculture, roughly a fifth from non-agriculture family enterprises, aSnd the remainder from a variety of other sources (rents, pensions, social security, and so on). It is important to note that some 20 to 25 percent of income in urban areas is derived from indirect or "unearned" sources, while only 5 percent of rural income is obtained from sources not directly related to income generating activities. Clearly rural households are heavily deplendent on earned income and informal support networks to secure and maintain a basic livelihood. According to Table 6, urban households earn nearly two-and-one-half times as much on average as rural households. In Abidjan, for example, average household income is some CFA 2.8 million annually, as compared to CFA 1.9 million in urban areas outside Abidjan, and only CFA 1.0 million in rural areas. Even accounting for cost-of-living differences (see Glewwe, 1987), urban households are significantly better off on average than their rural counterparts. Households, besides the regional classifications, were further classified by the type of productive activities from which they derive income. Specifically, we define five categories: (i) households wholly dependent on wage income (16.7 percent); (ii) households wholly dependent on - 17 - income from agricultural activities (livestock and crop cultivation) (39.6 percent); (iii) households dependent on income from self-employment, but not solely farm income (26.0 percent); (iv) households receiving both wage and self-employment income (16.1 percent); and (v) households receiving no earned income (1.7 percent). Note that the two classification schemes are not independent of one another; for instance, nearly 94 percent of households dependent on earnings from agricultural activities live in rural areas, while over 92 percent of households dependent on wage earnings live in cities. Table 7 shows the distribution of households and average total and adjusted per capita2- income by region and activity classification. In Table 7, some 72 percent of households in Abidjan obtain income from wage activities, and nearly two-thirds of these (48.4 percent overall) receive only wages. In Abidjan, 23.7 percent of households receive only income from self- employment. In contrast, 88.4 percent of rural households are dependent on self-employment income (64.7 percent farm, 23.7 percent mixed farm and non- farm self-employment), 9 percent receive both wages and self-employment income, and only 1.7 percent depend solely on wage activities for earned income. The wealthiest households in Abidjan evidence the most diversified labor portfolios - members work in both the wage and non-wage sectors. However, on a per capita basis, wage dependent households are wealthier in Abidjan; households in the wage/self-employment category tend to be large and 2/ Adult equivalency measures (Glewwe, 1987) are used to construct per capita estimates. Children between 0 and 6 years are assigned a weight of .2, between 7 and 12 years a weight of .3, and between 13 and 17 years a weight of .5. All others are assigned a weight of 1.0. Table 7: Thtal an Adjuated Per pita I b y legon and ketivity Classfication in (Ste d'Iwie Abidjan Other Urban Rural Total Coxmtry Percent Average Adjusted Income Percent Average Adjusted Income Percent Average Adjusted Percent Axerage Adjusted of House- Income Per Capita of House- Inccm Per Capita of House- Income Incone of House- Income Income holds holds holds Per Capita holds Per Capita CFA CFA CFA CFA CFA CFA CFA CFA H Wage Ircame Oily (N=260) 48.4 3,083,596 860,719 25.3 2,862,377 898,785 1.7 1,179,492 346,900 16.7 2,902,273 843,374 Farm hnuce Only (N618) 0.9 n.a. 1/ n.a. 10.5 1,138,950 201,138 64.7 877,722 179,527 39.6 892,290 180,193 Non-farm Self-Eiployment Income, Farm and Non-Farm Self-Paplcyment Income (N=408) 23.7 1,910,783 482,461 34.0 1,494,993 316,385 23.7 1,346,780 260,017 26.0 1,498,418 319,245 Wage and Self- oyment Income (N=251) 23.4 3,551,823 610,597 27.7 2,069,236 348,019 9.0 1,245,948 248,438 16.1 2,264,278 397,481 No Earnied Incme (N=27) 3.6 1,069,084 403,719 2.4 347,181 180,624 0.8 877,851 148,775 1.7 661,032 247,504 Total Country (N1I564) 100.0 2,823,011 688,136 100.0 1,934,893 457,084 100.0 1,022,597 206,848 100.0 1,600,743 362,748 Notes: 1- Too fe entries in cell for reporting. Source: CILSS 1985 Survey tabulations - 19 - evidence relatively high dependency ratios in all regions. Wage dependent households are wealthier according to both criteria in other urban areas. In rural areas, households obtaining income from some form of off-farm activity are clearly better off on average than households solely dependent on farm earnings. These results combined with earlier findings (on income and economic activities) suggest: first, individuals in the wage sector (representing 32.8 percent of households overall) tend to earn significantly more on average than self-employed individuals; and second, there is convincing evidence, however, that many off-farm employment activities, both in urban and rural areas, yield significant income levels. Based on the evidence reported here, it is clear that the self-employed are a heterogeneous group that cannot simply be characterized as marginal, underemployed members of the so-called informal sector (although the characterization undoubtedly applies to some self- employed workers). In Section III it will be shown that self-employed workers are represented across the range of the welfare distribution in the C6te d'Ivoire - they are drawn from some of the poorest and some of the wealthiest households in the country. - 20 - III. INCOME DISTRIBUTION The ability to monitor changes in income distribution is a critical aspect of development planning. Economic and social policy will inevitably affect different groups in different ways; it is often the intention of policy to redress imbalances and shift the burden of economic adjustments from one group to another. Survey data can serve an important function in helping to identify the poor and to assess the impacts of policy interventions. Previous work (Glewwe, 1987) explored the distribution of welfare (that is, consumption expenditures) in the C6te d'Ivoire, and identified important economic and demographic characteristics of the poor (defined variously as the lower 10 percent and lower 30 percent of the welfare distribution). This section analyzes the distribution of income and explores the relationship between household income and consumption expenditures. As expected, income is found to be more variable and highly skewed than consumption. There are a number of possible explanations for this -- exogenous income shocks brought about by such causes as weather and price changes, life cycle effects, and various types of credit or liquidity constraints. In general, total consumption expenditure is assumed to be a better proxy measure of long-term household welfare than annual income., Poverty assessments based on income rather than welfare estimates yield very different results. Table 8 shows household income composition by adult equivalency adjusted per capita income quintiles for Abidjan, other urban areas, and rural areas (quintile 1 represents the poorest households, while quintile 5 represents the wealthiest). There are several important income composition Table 8: Conposition of Household Income In C5te d'lvoire, by Region and Adjusted Per Capita Income Quintile within Each Region 1/ Non-Farm Capital Service Family and Social Other Private Flows Wage Farm Enterprise Interest Security Unearned Transfer Inputed from Total Income Income Income Income Pensions, etc. Income Payments Rent Durables income AbidJan Quintile 1 41.3 -0.7 34.5 2.2 10.2 2.2 3.5 5.2 1.6 100.0 Quintile 2 48.3 0.5 26.1 5.9 5.8 3.3 1.8 6.1 1.6 100.0 Quintile 3 50.9 1.0 16.6 13.4 7.3 1.7 2.9 4.6 1.6 100.0 Quintile 4 52.9 0.2 19.4 11.0 2.9 4.6 3.1 4.4 1.3 100.0 Quintile 5 52.4 2.6 24.8 9.7 1.3 3.6 2.1 2.1 1.4 100.0 Total 51.5 1.5 23.1 9.8 3.1 3.5 2.4 3.4 1.4 100.0 Other Urban Areas Quintile 1 8.7 21.0 24.5 2.1 2.3 1.2 5.9 31.7 2.6 100.0 Quintile 2 24.5 25.5 21.4 5.9 1.6 1.9 1.1 16.1 1.9 100.0 Quintile 3 29.1 12.1 28.8 9.1 4.4 1.2 1.6 12.0 1.7 100.0 Qulntile 4 45.5 8.1 27.7 4.4 3.5 2.0 0.8 6.4 1.6 100.0 Quintile 5 66.7 3.9 16.5 5.1 1.9 2.2 0.4 1.6 1.7 100.0 Total 46.9 9.7 22.6 5.6 2.8 1.9 1.0 7.8 1.7 100.0 Rural Areas Quintile 1 2.1 85.0 6.4 0.6 0.0 0.8 2.3 - _ 2.8 100.0 Quintile 2 1.3 87.0 7.2 0.3 0.2 0.8 1.7 - 1.4 100.0 Quintile 3 1.6 85.8 8.6 0.3 0.4 1.1 1.1 - 1.0 100.0 Quintile 4 4.3 78.7 12.5 1.8 0.4 1.1 0.6 - 0.7 100.0 Quintile 5 10.2 66.8 16.9 2.8 0.4 1.8 0.3 - 0.6 100.0 Total 6.4 75.0 13.3 1.9 0.4 1.4 0.7 - 0.9 100.0 Notes: I/ Quintile 1 Is the lowest, Quintile 5 Is the highest consumption group. 2' No imputations made for rural areas. Source: CILSS Survey estimates. - 22 - characteristics in the C8te d'Ivoire evident from the table. The most important perhaps is that the share of wage income tends to increase with rising per capita income. This effect is most marked for households residing outside Abidjan (8.7 percent of income is derived from wages in the lowest quintile in other urban areas, as compared to 66.7 percent in the highest quintile, and a scant 2.1 percent of income is derived from wages in the lowest rural quintile, as compared to 10.2 percent in the highest quintile), but still apparent within Abidjan (41.3 percent in the lowest quintile, and 52.4 percent in the highest quintile). There is a slight drop in the wage share for households in Abidjan's highest income quintile, primarily due to the increasing importance of rising income from (large) family enterprises. Table 9 shows the composition of household income stratified by adult equivalency adjusted per capita consumption quintiles rather than income quintiles. If consumption is a better measure of long-term welfare, the estimates in this table should be less sensitive to transitory income shifts and should better reflect the relationship between current income and welfare. On the role of wage income: not surprisingly, the effects seen in Table 8 remain and are strengthened by the new groupings; clearly the share of wage income in total income rises as levels of welfare rise, increasing from 40.4 percent of total income in the lowest consumption quintile to 59.5 percent of income in Abidjan's highest quintile, and 23.3 percent rising to 78.3 percent in the highest quintile in other urban areas. Non-farm entrepreneurial earnings are derived from a very heterogeneous set of activities which span the income distribution in the Cate d'Ivoire. For example, in Abidjan they constitute a major source of income - 23 - for the very poor (a 34.5 percent share at the lowest end of the distribution) and for the comparatively rich (24.8 percent of income for the upper 20 percent of the per capita income distribution). This may represent two kinds of family enterprises: (1) small, one or two person operations which yield little cash revenues, and (2) larger enterprises that employ a number of workers outside the households as well as within.31 Analysis of the family enterprise data bears this out; we find more food sellers and petty traders at the lower end of the income distribution, and much of the construction, large traders, and industry at the upper end. Small enterprises operated by poor households tend also to have low capital endowments and limited inventories. The distribution of non-farm family enterprise income is different in regions outside Abidjan, particularly rural areas. There, the share of income from family enterprises clearly increases as income levels increase, indicating the importance of labor diversification and the age-old adage that wealth begets wealth. Farm households that diversify their labor portfolio are clearly better off than those that do not. In urban areas outside Abidjan, family enterprise income constitutes a reasonably small share of total income for the very poor (income quintile 1, with a share of 24.5 percent), a smaller share for the rich (income quintile 5, with a share of 16.5 percent) and a larger share for the middle classes, averaging around 27 or 28 percent of total income. 31 Or, alternatively, the effect of measurement error which causes a substantial understatement of net earnings from family enterprises. Table 9: Composition of Household Income in C6te d'lvolre, by Reg ion and Adjusted Per Capita Expenditure Quintile within each Region 1/ Non-Farm Capital Service Family and Social Other Private Flows Wage Farm Enterprise Interest Security Unearned Transfer Imputed from Total Income Income Income Income Pensions, etc. Income Payments Rent Durables Income Abidjan Quintile 1 40.4 1.0 35.5 7.9 4.6 3.3 1.7 4.5 1.1 100.0 Quintile 2 43.0 0.4 22.7 13.6 10.4 2.2 2.1 4.5 1.1 100.0 Quintile 3 42.3 0.2 35.5 8.2 3.3 1.6 1.6 6.0 1.2 100.0 Quintile 4 55.6 0.5 21.1 10.7 1.6 4.8 1.9 2.4 1.3 100.0 Quintile 5 59.5 3.4 15.3 8.8 1.1 4.1 3.7 2.3 1.9 100.0 Total 51.5 1.5 23.1 9.8 3.1 3.5 2.5 3.4 1.4 100.0 Other Urban Areas Quintile 1 23.3 22.7 34.2 1.6 2.6 1.0 1.0 12.5 1.2 100.0 r. Quintile 2 22.4 24.0 24.2 7.2 3.3 1.5 1.6 14.3 1.5 100.0 4 Quintile 3 33.1 10.1 36.7 4.2 3.8 1.2 0.8 8.4 1.6 100.0 Quintile 4 47.3 7.1 20.9 7.8 3.9 4.0 0.5 7.0 1.5 100.0 Quintile 5 78,3 -1.1 9.1 5.3 0.8 1.2 1.4 2.5 2.5 100.0 Total 46.9 9.7 22.6 5.6 2.8 1.9 1.0 7.8 1.7 100.0 Rural Areas Quintile 1 1.5 82.7 12.9 0.1 0.0 0.6 1.0 - 1.2 100.0 Quintile 2 3.8 74.7 13.1 1.2 0.7 4.5 0.8 - 1.1 100.0 Quintile 3 3.6 76.8 15.5 1.0 0.2 1.0 1.0 - 0.8 100.0 Quintile 4 11.7 72.3 12.4 0.8 0.3 1.0 0.7 - 0.8 100.0 Quintile 5 7.4 73.6 13.0 3.8 0.4 0.7 0.4 - 0.8 100.0 Total 6.4 74.9 13.3 1.9 0.4 1.4 0.7 - 0.9 100.0 Notes: 2I Quintile 1 is the lowest, Quintile 5 is the highest consumption group. - 1 No imputations made for rural areas. Source: CILSS Survey estimates. - 25 - Table 9 suggests a different story for the role of non-farm entrepreneurial income. In Abidjan and other urban areas, the income shares tend to fall with increasing expenditure levels, and are lowest in the highest expenditure quintiles (15.3 percent and 9.1 percent for Abidjan and other urban areas, respectively). Further, in rural areas the distribution of family enterprise income shares is roughly equal across the welfare (consumption) distribution. In all likelihood, this reflects the transitory nature of most kinds of non-wage income in the C6te d'Ivoire. If income from family enterprises has a significant transitory component, then the correlation between consumption (a proxy for permanent income) and family enterprise income would be less than the correlation in a particular time period between household income and family enterprise income. The stability of wage income shares across income and consumption deciles lends credence to this explanation. The share of income derived from agriculture falls with increasing income levels and consumption expenditures in all regions. Farm income also probably has a relatively high transitory component, although perhaps less than non-farm family enterprise income. The flattening found in the distribution of farm income shares across rural consumption quintiles tends to bear this out. As noted earlier, other sources of income constitute a fairly small share of total income in the C6te d'Ivoire. Capital (rents) and interest income are evenly spread throughout mid-to-upper levels of both income and consumption distributions, and transfers and remittances constitute a slightly higher share of income for households at the lower end of the distribution - 26 - living outside of Abidjan. Social security and pension payments, whether grouped by income or consumption quintiles, constitute a larger share of total income for households at the lower end of the relevant distributions. This reflects life cycle differences rather than differences in wealth, as only retired employees (who we have already seen to be in the most wealthy segment of the population) are eligible for social security and pension benefits. Table 10 combines the information in Table 9 across regions and across selected income source categories, showing the price-adjusted 4/ composition of income by per capita expenditure decile for the country as a whole. Based on these estimates, it is clear that the share of wage income in total private income increases with total consumption expenditures for the country as a whole (2.5 percent in the lowest decile to 63.4 percent in the highest); the share of farm income falls steadily (78.0 percent to 4.6 percent); family enterprise income initially rises and then falls; rent and dividend income increases steadily with per capita consumption levels (0.5 percent rising to 9.8 percent of total income); and other income is a relatively small share of the total at the lower end of the welfare distribution, and a fairly constant share (approximately 10 percent) across mid-to-upper ranges. What does all this suggest for poverty and income in the C6te d'Ivoire? Households receiving a large share of total income from wages appear fairly well off, as do households receiving income both from wage 4/ Regional prices indices (Glewwe, 1987) were used to adjust income shares to account for cost-of-living differences between regions. - 27 - Table 10: Composition of Household Income 1/ in Cdte d'Ilvoire by Per Capita Expenditure Decile Wage Agriculture Family Enterprise Rents and Other Total Income Income Income Dividends Income Income Decile 1 2.5 78.0 14.9 0.5 4.1 100.0 Decile 2 8.7 69.1 14.0 2.3 5.9 100.0 Decile 3 5.8 61.1 21.1 1.6 10.3 100.0 Decile 4 10.2 58.5 21.8 0.9 8.5 100.0 Decile 5 21.1 48.8 16.2 4.8 9.2 100.0 Decile 6 22.2 41.2 22.0 3.8 10.7 100.0 Decile 7 22.2 30.0 29.0 6.9 11.9 100.0 Decile 8 29.6 34.4 18.7 6.5 10.7 100.0 Decile 9 43.6 16.7 20.7 7.2 11.8 100.0 Decile 10 63.4 4.6 13.6 8.1 10.2 100.0 Total 32.5 32.7 19.1 9.8 10.1 100.0 Notes: -/ Adjusted for regional price variation as in Glewwe, 1987. Source: CILSS Survey estimates. activities and from some form of self-employment. Of the households in the upper 20 percent of the welfare distribution, 50.8 percent depend entirely on wages, 16.6 percent receive both wages and earnings from self-employment activities, and 19.8 percent earn income from self-employment activities alone, but not exclusively farming (Table 11). Only 10.2 percent of households in the highest welfare quintile receive all earned income from - 28 - farming. In contrast, farm households are among the poorest in the country. Fifty nine percent of households in the bottom 40 percent of the welfare distribution are entirely dependent on agricultural earnings to secure a basic livelihood, and 27.4 percent are wholly dependent on non-farm or mixed ifarm and non-farm self-employment earnings. Less than 13 percent of households in the lower two quintiles receive income from activities outside the so-called family sector. Thus, the majority of the poor are dependent on incomes with a high transitory component; drought or inclement weather can cause a sharp drop in crop income, with concomitant second order effects throughout the economy, depending on linkages. A brief comment is made here on the meaning of a "transitory" eslement in farm and business incomes. In a formal sense, the designations of "transitory" and "permanent" incomes were coined by the permanent incomes theorists headed by Friedman (1957), (see also Mayer, 1972, for a good overview of related theories and relevant empirical work), to differentiate between the behavioral response to long-run, stable components of incomie and more variable components. The marginal propensity to consume out of permanent income is hypothesized to be much greater (in the strictest sense of thes theory, converging to one) than the marginal propensity to consume out of transitory income (likewise in the strictest sense converging to zero). The use of "transitory" here encompasses a broader set of issues than strict adherence to theory would indicate; we assume that income from self- employment or family run enterprises typically has a consumption and an investment component. Farmers accrue farm profits and ultimately plow them - 29 - Table 11: Percentage Distribution of Households by Adjusted Per Capita Expenditure Quintile and Activity Classification In Cate d'ivoire Wage Farm Non-Farm Self- Wage and Self- Income Income Employment Income, Farm Employment Only Only and Non-Farm Self- Income No Earned Total (N=260) (N=618) Employment Income (N=408) (N=251) Income (N=27) (N=1564) Quintile 1 1.3 64.3 24.8 8.9 0.6 100.0 Quintile 2 2.9 52.9 29.9 14.0 0.3 100.0 Quintile 3 9.9 42.7 28.3 17.5 1.6 100.0 Quintile 4 18.2 26.8 28.7 22.9 3.5 100.0 Quintile 5 50.8 10.2 19.8 16.6 2.6 100.0 Total Country 16.6 39.4 26.3 16.0 1.7 100.0 Sources: CILSS Survey estimates. (in a figurative sense) back into the farm or other family businesses through capital and land purchases. One expects similar behavior on the part of small entrepreneurs -- the cash for investment capital must come from somewhere, and credit is difficult to obtain and costly in many LDCs. Thus, income from farm and non-farm self-employment may be less correlated with consumption than wage income for two reasons; (i) inherent income variability, which leads the household to save cash in times of plenty in anticipation of future shortfalls, and possibly to spend more than annual earnings alone might indicate in times of relative scarcity; and (ii) the role of self-employment profits in expanding investment capital and ensuring adequate levels of future liquidity. - 30 - Cumulative Shares of Income and Consumption Expenditures Table 12 which shows the average household income and consumption expenditures by income and consumption deciles and the cumulative share in each decile, indicates that welfare is more evenly distributed in the C6t:e d'Ivoire than prior analyses would lead us to expect. Consider the first panel in the table, which shows adjusted average income and consumption by per capita income deciles. Clearly income is much more unevenly distributed than consumption expenditures, as intuition would suggest. The bottom 20 percent of the per capita income distribution receives some 3.6 percent of income and accounts for 9.6 percent of total consumption expenditures. Income and consumption shares become more equal towards the middle and upper end of the distribution - households in the wealthiest decile receive a third of all income and account for some 20 percent of private consumption. If households are ranked by per capita consumption deciles (the second panel in Table 1L2), we find a surprising degree of equality across the distributions; income is clearly more evenly distributed across expenditure deciles than income deciles. Households in the lower 20 percent of the welfare (consumption) distribution receive some 8.4 percent of total income and consume 7.7 percent of total private consumption. The lower 50 percent of the welfare distribution receive 29.2 percent of total income and account for 24.8 percent of total expenditures. Finally, the upper 10 percent of the welfare distribution receive 22.5 percent of total income and account for 21.9 percent of consumption expenditures in the country, significantly less than the 34.6 percent of income received by the highest income decile. Table 12: Distribution of Adjusted Household Income and Consumption Expenditures in Cote d'lvoire by Adjusted Per Capita Income and Expenditure Deciles 1/ Per Capita Income Deciles Per Capita Expenditure Deciles Average Annual Average Annual Average Annual Cumulative Consumptions Cumulative Average Annual Cumulative Consumption Cumulative Income Share in Expenditures Share in Income Share in Expenditures Share in (CFA) Decile (CFA) Decile (CFA) Decile (CFA) Decile Decile 1 150,054 1.0% 735,856 4.7% 557,849 3.5% 446,351 2.9 Docile 2 416,711 3.6 845,190 10.1 772,855 8.4 749,957 7.7 Decile 3 602,773 7.4 959,596 16.3 941,480 14.4 873,871 13.3 Decile 4 785,366 12.4 1,291,377 24.6 1,191,089 21.9 1,116,142 20.4 Decile 5 966,700 18.5 1,314,656 33.0 1,145,460 29.2 1,244,950 24.8 Decile 6 1,167,061 25.9 1,492,423 42.5 1,320,096 37.6 1,435,611 37.6 Decile 7 1,574,271 35.9 1,567,409 52.6 1,799,729 49.0 1,764,472 48.9 Decile 8 1,877,179 47.7 1,910,363 64.8 2,034,022 61.8 2,025,891 61.9 Decile 9 2,784,798 65.4 2,304,799 79.6 2,478,475 77.5 2,533,641 78.1 Decile 10 5,469,470 100.0 3,182,609 100.0 3,546,717 100.0 3,414,748 100.0 TOTAL 1,578,114 - 1,559,979 - 1,578,114 - 1,559,979 - Notes: I/ Decile 1 is the lowest, Decile 10 is the highest income/expenditure group. Source: CILSS Survey estimates. - 32 - Interestingly, at least some households at the lower end of the expenditure distribution are substantial savers, which could be explained by: (i) basic thriftiness amongst the poor, and a tendency to save against possible future income shortfalls, which suggests that at least some households towards the bottom of the expenditure distribution are there in part through choice rather than pure economic necessity, and (ii) borrowing constraints, which particularly constrain the consumption behavior of thle poor. Note also that households in the lowest income decile evidence substantial dissavings, as one might expect if income has a substantial transitory component. Table 13 shows similar measures to those presented in Table 12, stratified by region and adjusted per capita income and expenditure quintiles. (Note that averages have not been adjusted for regional price differences in Table 13). In these tabulations, the degree of difference in average income for the highest and lowest deciles in Table 12 is to a large extent caused by urban/rural differences in income and consumption levels. For example, rural households in the highest expenditure quintile earn on average CFA 1,791,544 per household annually, while urban households in Abidjan's highest quintile earn on average CFA 4,821,251. This is a marked difference in earnings, and is roughly paralleled by expenditure disparities. While we are reluctant to make definitive statements on the absolute size of the welfare gap between urban and rural areas, the differences in the means are suggestive of substantial differences in urban and rural living standards. Simple comparisons of extremes suggest that while the disparity in income between the wealthiest and poorest groups is greater in Abidjan than in rural communities, Table 13: Compositlon of Household Income and Consumption Expenditures in C6te dlIvoire, by Region and Adjusted Per Capita Income and Expenditure Quintiles within each Region 1/ Adjusted Per Capita Income Quintile Average Cumulative Average Annual Cumulative Average Cumulative Average Annual Cumulative Annual Share Consumptions Share Annual Share Consumptions Share Income (CFA) Income Expenditures (CFA) in Quintiles Income (CFA) Income Expenditures (CFA) in Quintiles Abidjan Quintile 1 637,310 4.5% 1,736,735 12.5% 1,320,606 9.3% 1,502,323 10.8% Quintile 2 1,289,724 13.6 2,227,836 28.5 1,943,546 23.1 2,258,946 27.1 Quintile 3 1,726,047 25.8 2,218,763 44.5 2,354,059 39.7 2,559,170 45.5 Quintile 4 3,192,294 48.3 3,014,164 66.2 3,705,419 65.9 3,353281 69.7 Quintile 5 7,337,055 100.0 4,693,577 100.0 4,821,251 100.0 4,210,141 100.0 Other Urban Areas Quintile 1 442,680 4.6% 1,162,508 12.0% 1,147,011 11.9% 1,005,003 10.4% Quintile 2 1,093,576 15.9 1,578,699 28.3 1,454,764 26.9 1,510,521 25.9 Quintile 3 1,734,408 33.8 1,977,285 48.6 1,865,837 46.2 1,785,870 44.3 Quintile 4 2,562,219 60.3 2,333,991 72.7 2,415,151 71.1 2,383,647 68.9 Quintile 5 3,844,822 100.0 2,649,232 100.0 2,791,699 100.0 3,016,953 100.0 Rural Areas Quintile 1 215,169 4.2% 643,103 12.8% 517,634 10.1% 439,542 8.7% Quintile 2 493,009 13.8 770,470 28.1 833,729 26.4 778,087 24.2 Quintile 3 751,110 28.5 969,810 47.3 920,168 44.4 924,173 42.5 Quintile 4 1,153,562 51.1 1,128,398 69.7 1,051,377 65.0 1,106,589 64.5 Quintile 5 2,503,879 100.0 1,527,391 100.0 1,791,544 100.0 1,791,115 100.0 Notes: 1/ Quintile 1 is the lowest, Quintile 5 is the highest Income/expenditure group. Source: CILSS Survey estimates. - 34 - the disparity in consumption expenditures or welfare is greater in rural areas. A simple count of the households reporting food budget shares greater than 80 percent (a standard poverty indicator) lends further support. ]:n Abidjan, none of the households interviewed reported food budget shares greater than 80 percent of the total value of consumption, and in other urban areas, only 1.2 percent of households reported food budget shares greater than 80 percent. In contrast, over 14 percent of rural households spent more than 80 percent of total outlays on food. The Gini coefficient is a commonly used measure of inequality in income and asset distribution studies. For this study, we used Sen's (1973) definition of the Gini coefficient, which is Gini = - 2 ) (n+l-i)x. n n2piil 1 where n is the sample size, x is the variable of interest, and p is its estimated mean value. A coefficient of zero implies perfect equality while a measure of l implies perfect inequality. As expected, the Gini coefficients for income are substantively higher than similar coefficients for consumption expenditures (Table 14). Further, the Gini coefficients for the whole country are typically higher (particularly in the case of consumption) than region-specific values. This high degree of between-region variation was also identified in Table 13; income (and consumption) may well be more unequal between regions than within regions. Note that the level of income inequality tends to be higher in Abidjan and in rural areas compared to other urban areas. Gini coefficients - 35 - estimated on per capita measures are similar to aggregate measures in Abidjan. However, in rural areas, per capita coefficients indicate a lesser degree of welfare and income equality, while in other urban areas coefficients indicate a greater degree of welfare and income inequality. Table 14: Gini Coefficients for Household Income and Consumption Expenditures by Region and Total Country: C6te d'Ivoire Income-Gini Expenditure-Gini Total Per Capita Total Per Capita Abidjan .536 .523 .354 .344 Other Urban Areas .449 .511 .347 .410 Rural Areas .525 .476 .379 .326 Total Country 1/ .540 .545 .410 .433 Notes: 1/ Adjusted measures are used for country-wide estimates. Source: CILSS Survey estimates. - 36 - IV. COMPOSITION AND DISTRIBUTION OF ASSETS Household income represents returns to labor inputs and private capital. Capital is held in the form of physical assets, financial assets, and human capital endowments, which include education and specific skills (typically represented by work experience). Human capital may also include less tangible items such as access to certain kinds of employment, credit sources, or educational opportunities. Because of the importance of assets in determining income flows, this section describes the extent, composition, and distribution of (measurable) private assets for Ivorian households. The intangibles, while possibly of considerable importance, cannot be measured directly. Composition of Household Assets Physical assets are grouped into three major categories: (i) personal assets, which include all private capital not directly used in market activities, (ii) non-farm family enterprise assets, which include capital used in non-farm self-employment activities, and (iii) farm assets, which include all farm capital and land. Assets held in the form of human capital are discussed later in the section. Personal Assets. Subsumed under this category are jewelry purchases over the past 12 months (admittedly, this is a flow rather than stock value; unfortunately, no measure of the gross value of jewelry owned is availabLe in the CILSS); the present value5/ of all durables owned by the household; the present value of automobiles owned; the estimated value of owned housing 5/ Survey respondents were asked how much they would receive for each durable item owned if they "were to sell it today." - 37 - owned housing stocks 6/ ; and the household's net debt position (money loaned out minus money borrowed) and total cash savings. Country-wide, the value of personal assets held in 1985 was reported to be CFA 937,462; jewelry purchases accounted for 1 percent of the total; durables for 17 percent, automobiles for 16 percent, housing stock for a substantial 47 percent, debts for some -4 percent of the total (overall, households were net debtors); and savings for 23 percent. Not surprisingly, housing accounted for nearly half of the total personal assets held. Non-farm Family Enterprise Assets. These include unsold inventories, buildings and land, business durables such as tools and equipment, and rental property. All values were reported in the questionnaire except rental properties. For these, only an annual flow measure was available (for example, income from property rented out). For purposes of imputing capital values to rental stocks, reported rents were assumed to represent a 25 percent annual return. The value is higher than that used for housing because (i) some of the rental capital may be owned for speculative purposes and yields high returns, and (ii) capital markets are imperfect and access is limited, 61 Data limitations required that we use an indirect method to estimate the value of owned housing stock in Abidjan and other cities. The housing stock in rural areas was attributed a zero asset value primarily due to the lack of housing markets in these areas and even more severe data constraints. In urban areas, we first obtained an imputed annual rental value for the housing stock (see Annex I for details on how this was done), and from this, assuming that rents represent a 12 percent return on housing investments, imputed a capital value. Based on previous work (see citations in Annex I), the 12 percent figure is probably low for poor households (where rent-to-value ratios tend to be on a scale of 3-to-1 or 4-to 1 in many cities of the developing world) and high for wealthier households. - 38 - which means that capital can demand higher rents than under perfect market conditions. In the survey estimates, households own on average CFA 761,827 in non-farm production capital: inventories account for 3 percent; buildings and land used in production for some 7 percent, tools, equipment, and machinery for 41 percent; and rental stocks for the remaining 47 percent. The last: category could have been included above in personal assets; however, it seemed more appropriate to assume that rental stock is owned for directly productive purposes, that is, to generate income, rather than as a personal asset used in the day-to-day maintenance of household activities. Farm Assets. These include land, stored crops 71, livestock, hand tools, and farm equipment. For the country as a whole, the average value of farm assets per household was CFA 3,580,567. Land accounted for the main portion; 95 percent of total farm assets are held in the form of land 8/. The other categories account for roughly equal shares of the remainder, with livestock having the highest share (1.5 percent) and farm equipment the lowest (.7 percent). 7/ Crops in storage were measured in terms of the number of weeks they would feed the household. Value estimates were derived by multiplying the number of weeks by the cost of a week of consumption for the relevant food category as reported in the expenditure section of the questionnaire. 8/ Land was valued according to the household's response to the question "how much could you sell your land for today?" All but a few of the farm households responded to this question. However, as nearly two-thirds of farm households claimed they could not sell their land due to family, social, or cultural constraints, one must wonder how land valuations were made. The maximum "permitted" land value was assumed to be CFA 9,375,000 per hectare. Only a few households reported per hectare land value greater than this, and were therefore set to the limiting value. - 39 - For the country as a whole in 1985, total physical and monetary household assets are valued at CFA 5,264,821 per household, of which personal assets comprise some 18 percent, non-farm business assets some 14 percent, and farm assets the remaining 68 percent. Average household income is CFA 1,600,000 per annum for the total country. Thus, households in the C8te d'Ivoire maintain a stock of assets that is nearly three and one-half times the annual income, primarily tied up in farm land (65 percent), rental capital (7 percent), and a dwelling unit (8 percent). Clearly the distribution of asset holdings will vary by region in the C6te d'Ivoire, depending on the spatial orientation of productive activities. Table 15 shows composition of household assets by region and for the whole country. Averages are computed for total households within appropriate regional categories and for households in the lower 90 percent of the asset distribution; in effect, the wealthiest households (for total asset holdings) are dropped from averages in the second column. According to survey estimates, households in Abidjan hold assets valued at an average of CFA 4,264,392 per household (for the total Abidjan sample) and CFA 2,428,005 per household for those in the lower 90 percent of the asset distribution; thus, the distribution of assets is highly skewed in the capital city. In the asset composition, some 40 percent of the total are personal assets, 8 percent agricultural assets, and the remaining 52 percent non-farm business assets. The share of non-farm business assets drops to 34 percent of the total for households in the lower 90 percent of the asset distribution; the wealthiest households own a significant amount of production capital. Table 15: Composition of iHousehold Assets by Region and Total Country: Cote d'lvoire Abidjan Other Urban Rural Total Country Type of Asset Total Lower 90% Total Lower 90% Total Lower 90% Total Lower 90% Personal Assets (CFA) Annual Jewelry purchases 19,414 13,968 12,849 9,807 5,459 5,164 10,010 7,669 Total personal savings 312,164 123,684 342,964 331,418 125,449 115,521 211,551 161,004 Present value of durables 299,630 183,678 259,056 223,740 73,918 72,013 161,476 122,724 Present value of automobiles 425,428 187,186 164,548 108,815 49,666 37,715 154,365 78,670 Estimated value of housing stock 809,539 746,246 1,205,269 1,335,434 - - 438,564 404,467 l Net debt position * -163,370 -49,569 -64,992 6,674 17,792 23,375 -38,505 7,059 Total 1,702,805 1,205,194 1,964,694 2,015,889 272,283 253,788 937,462 781,594 O Non-Farm Family Enterprise Assets (CFA) Unsoid Inventories 23,484 14,145 54,295 57,671 15,281 13,876 25,322 22,857 Value of buildings and land 30 40 26,434 18,380 85,630 77,050 54,799 51,454 Value of business durables 1,228,032 121,940 164,049 148,964 27,383 17,071 312,981 62,534 Value of rental property 1,064,751 691,192 528,410 290,202 57,965 38,145 372,880 205,131 Total 2,301,631 827,318 773,188 515,207 185,990 146,142 761,827 341,975 Farm Assets (CFA) Estimated value of land holdings 317,136 391,259 1,410,245 1,403,246 5,319,900 5,177,271 3,420,394 3,560,453 Value of stored crops 131 175 9,606 10,234 71,674 71,830 43,201 46,585 Value of large livestock 16,189 61 33,639 37,511 80,142 78,827 56,598 56,460 Value of tools 3,227 3,539 24,183 26,542 51,308 50,959 35,272 37,586 Value of farm equipment 1,240 458 5,635 6,360 41,191 35,766 25,101 23,519 Total 337,923 395,493 1,483,308 1,483,892 5,564,216 5,414,652 3,580,567 3,724,603 Total Household Assets (CFA) 4,264,392 2,428,005 4,221,190 4,014,998 6,022,488 5,814,582 5,264,821 4,848,173 Source: CILSS tabulations. - 41 - Households in other urban areas have assets valued at CFA 4,221,190, which is roughly on a par with per household stocks in Abidjan. Interestingly, deleting households in the upper 10 percent of the asset distribution causes minimal change; the remaining households report holding assets valued at an average of CFA 4,014,998 per household, which suggests that assets are more evenly distributed in other urban areas than in Abidjan. Some 46 percent of total assets are held in the form of personal stocks, 18 percent in non-farm business assets, and the remaining 46 percent in farm capital, primarily land. These shares clearly reflect an increasing orientation towards agricultural activities in lieu of non-farm enterprises as we move from Abidjan to the small and medium-sized cities and towns. Households in rural areas report the highest average assets in the country (CFA 6,022,488). Land is the main component, comprising some 88 percent of the total value. The remainder is split evenly between personal assets and non-farm business assets. Households in urban areas outside Abidjan, as also rural households, do not evidence high skewedness in the distribution of assets. We will return to questions of distribution. There are several cross-regional differences worth noting in Table 15. Most notably, the ratio of asset values to average incomes is lowest in Abidjan (1.51) and clearly highest for rural households (5.69). This is not surprising given the regional dispersion of productive activities and the high degree of land intensiveness of agricultural activities. One must view the capital stock to flow ratio in rural areas with some caution. However, some two-thirds of rural households cannot sell any of their major stock -- land -- which makes it an exceedingly non-liquid asset. Also, there is little - 42 - evidence that land is used as collateral for obtaining credit. This may in part account for some of the high land prices observed in the survey; the scarcity of land offered for sale may artificially inflate market prices. However, rural households may tend to maintain a smaller stock of liquid assets than their urban counterparts. If we define liquid assets as (i) jewelry, (ii) cash savings, (iii) durables and automobiles, (iv) business inventories, and (v) stored crops, then households in Abidjan hold on average CFA 1,080,251 in liquid assets, households in other urban areas hold orLly CFA 843,318, and rural households hold CFA 341,447. These absolute values are somewhat misleading. The ratio of liquid assets to annual income in Abidjan is .38, in comparison to .43 in other urban areas and .33 for rural households. Thus, rural households have high overall asset holdings, but a considerable proportion of these are tied up in very non-liquid stocks. Interestingly, in rural areas, 46 percent of personal assets are in the form of cash savings, the most liquid form of assets, in comparison to only 18 percent in Abidjan and 17 percent in other urban areas. It is often claimed that credit constraints seriously hamper g#rowth in developing countries. Estimates of the average net debt position of Ivorian households lend some support to this view. For example, households in Abidjan are heavily indebted in comparison to their rural counterparts, particularly households in the upper 10 percent of the asset distribution, which indicates that access to credit may be important in building up capital stocks. Further, households in urban areas (including Abidjan) are more probably net debtors than net creditors, clearly obtaining at least some funds from outside the household sector. In comparison, rural households are net - 43 - creditors on average, which suggests that borrowing in rural areas is primarily within the household sector. From survey estimates, households in Abidjan report outstanding debts of CFA 412,712 in contrast to outstanding credits (money owed to them) of CFA 253,094 (they are net debtors); households in other urban areas report outstanding debts of CFA 222,400 and credits of CFA 109,772 (likewise net debtors); and rural households have net debts of CFA 37,067 in comparison to credits of CFA 55,237 -- they are in fact net creditors. (Note that this result may in part be caused by rural sampling biases which tended to oversample the wealthier households.) Some 90 percent of rural households who borrowed money in the past 12 months preceding the survey obtained loans from private individuals, in comparison to only 56 percent of borrowing households in Abidjan. The remainder borrowed from banks, government sources, or other financial institutions. The formal credit market is not extensive or well-developed in rural areas of C8te d'Ivoire. To a great extent, regional classifications serve as proxies for classification by the structure of production. For example, households in rural areas have land assets because they are farmers, not because they live in rural areas per se. Table 16 shows average per household assets for the five categories of households defined by productive activities. The figures in this table clarify what was implied by those in Table 15. Households which sell labor outside the household (that is, do not own the means of production) hold less physical assets overall, and what assets they do own are different than those owned by households who do not sell labor outside. From prior analyses, we know that wage-earning households receive higher incomes and primarily reside in Abidjan. They own substantial Table 16: Composition of Household Assets by Income Source Category in Cote d'ivoire 1 2 3 4 5 Wage Income Farm Income Other Self-Employment Wage and Other No Earned Only (N=260) Only (N=618) Income Only (N=407) Income (N=251) Income (N=27) Personal Assets (CFA) Annual jewelry purchases 22,141 4,718 9,282 12,365 3,417 Total personal savings 295,636 112,257 207,043 389,280 90,307 Durables and automobiles 789,044 96,732 302,709 411,203 85,704 Estimated value of housing stocks 423,328 115,372 625,962 896,583 900,044 Net debt position -298,198 24,943 30,688 -43,566 13,981 Total 1,231,952 354,023 1,175,685 1,665,866 1,093,454 Non-Farm Family Enterprise Assets (CFA) Value of rental property 822,523 140,246 349,199 471,586 807,066 Business assets - _ 1,435,320 120,753 - Total 822,523 140,246 1,784,519 565,425 807,066 Farm Assets (CFA) Estimated value of land holdings 4,019 5,402,820 3,669,498 2,042,241 0.0 Other farm assets 160 251,361 172,660 90,924 684.6 Total 4,179 5,657,181 3,842,158 2,133,164 684.6 Total Household Assets 2,058,655 6,151,451 6,802,361 4,267,638 1,901,205 Source: CILSS tabulations. - 45 - rental properties, durables and automobiles, jewelry, and have higher debts than households in other categories. They also report extensive cash savings, typically in formal savings institutions. In contrast, households which receive earned income only from agricultural activities have most of their asset holdings tied up in land, and own very little else; they are an extreme version of the rural household profile in Table 15. They are net creditors, have some savings (although typically not in formal savings institutions), own some farm equipment and tools, and seldom rent property to others. Households in the third category provide a contrast to "pure" farm households; they are, in combination with households in the fourth category, the C8te d'Ivoire's entrepreneurs or petty capitalists. All households which earn income from non-farm self-employment but do not receive wages are in this category (including those also receiving farm income). Their overall average assets are the highest among the categories; they own substantial amounts of land and substantial production capital for their business. Cash savings are also high in comparison to farm households, and housing investments are extensive within the group. To characterize households in this category simply, they report highly diversified asset portfolios. However, like households receiving only income from farm activities, they are net creditors rather than net debtors. Households in the fourth category are among the wealthiest. These include households who receive both wage income and some form of income from self-employment. Interestingly, households in this category are net debtors, like those who receive only wage income. This suggests some connection between having a job outside the household (which means a steady source of - 46 - income) and borrowing, particularly borrowing from formal credit sources. Households in this category also have large cash savings and are heavily invested in housing stock and durables. Similar to other wage households, they own some rental stock and have both business and farm capital -- in short, a diversified asset portfolio allows them to obtain income from a number of sources. Only 27 households in a sample of 1564 did not receive earned income. Although absolute numbers are too few for generalization, these households appear to own limited assets (with the exception of a dwelling unit and rental stocks), and primarily subsist on rental income and public and private transfer payments. Distribution of Assets Gini coefficients were computed for various categories of physical assets in the C6te d'Ivoire, and for "human assets", defined in terms of- years of formal education completed. Two measures of the household-level stock of human capital are used; (i) aggregate years of education across all household members aged 20 to 60 years old, and (ii) years of education of the most; educated person in the household in the same age group. The first measure treats education as an aggregate stock that is augmented by each year acquired by a household member. For example, a household with three persons each having 2 years of education would have the same aggregate measure as a similarly sized household with one member having six years of education and the other two members having none. The second measure treats education as a sort of household public good which has a particular use within the household production process; any one person can supply all required education inputs. - 47 - Further, the definition implicitly assumes that education is commensurate with managerial ability and does not augment labor inputs directly. The second measure is most appropriate in the analysis of self-producing households, while the first measure might work best in analyzing households who receive the bulk of their income from wage activities. Past work has shown that a year of primary education typically yields a different return than a year of secondary or tertiary education, (for a general review, see Psacharopolous, 1985). Therefore, education was classified into three categories -- primary, secondary, and tertiary.9/ The education variables are measured in terms of years of schooling completed in the relevant category. Table 17 shows Cini coefficients for selected asset categories and education measures by region and country-wide totals. As suggested by averages in Table 15, physical assets are more highly concentrated in Abidjan than in other regions; the Gini coefficient for total assets (excluding land and education) is .820 in Abidjan, .615 in other urban areas, and .706 in rural areas. In contrast and, as expected, because of the greater availability of schools in urban areas, education tends to be less highly concentrated in Abidjan than other regions. The Gini coefficient for the single individual, maximum education variable is only .227 for primary schooling in Abidjan as compared to .715 for rural households, and .515 for 91 The Ivorian education system is similar to the French system. The first 7 years of schooling are considered primary (JE, CP1, CP2, CE1, CE2, CM1, CM2), the next 7 secondary (6E, 5E, 4E, 3E, 2E, 1RE, TER), and the last 8 tertiary (Ul, U2, U3, U4, U5, U6, U7, U8). - 48 - Table 17: Gini Coefficients for Selected Household Assets by Region in CWte d'lvolre Type of Asset Abidjan Other Urban Rural Total Count-ry Personal Assets Total savings .852 .812 .839 .850 Value of durables and automobiles .752 .666 .798 .800 Value of housing .894 .709 - - Non-Farm Family Enterprise Assets Value of rental property .918 .939 .981 .9653 Value of other business assets .988 .938 .988 .987 Farm Assets Total land used .975 .816 .467 .651 Value of other agricultural assets .992 .844 .607 .745 Education Total years of household members - primary .458 .531 .686 .621 - secondary .609 .679 .917 .812 - tertiary .901 .959 - .972 Most educated person, 20-60 years old - primary .227 .387 .715 .543 - secondary .515 .661 .927 .785 - tertiary .890 .959 - .968 Total assets, Excluding Land and Education .820 .615 .706 .793 Source: CILSS tabulations. secondary education in Abidjan as compared to .927 for rural households. The difference in the distribution of physical and human assets in part reflects the higher dependence on wage income in urban areas; many urban households sell labor and skills rather than goods they produce. Returns to education are typically found to be highest in the urban wage sector, and lowest in rural agriculture, although there is still much debate on this subject. Household level production functions described elsewhere (see Kozel, 1987) show significantly higher returns - 49 - to education for wage households than for farm households. However, rural and urban entrepreneurs also eiidenced quite substantial returns to investments in education. The concentration of land holdings (measured in hectares per household) is low in rural areas in the Cote d'Ivoire in comparison to other LDCs. We calculated a Gini coefficient of .467 for rural land, and .607 for other agricultural assets in rural areas. Typical Gini coefficients for land distribution range from a low of .35 or .4 in some Southeast Asian countries to a high of .8 in parts of Latin America. Other Gini coefficients indicate a high degree of concentration of non- farm family enterprise assets and of most personal assets. The distribution of durables is less highly skewed than some other assets, but an average Gini coefficient around .75 (by region) does not indicate a notably even distribution of resources. To summarize the findings in this section: physical assets tend to be highly concentrated in the Cote d'Ivoire, with the possible exception of rural land holdings and other farm capital. In contrast, stocks of human capital (measured in terms of education) are less concentrated, especially in primary schooling in urban areas. However, the distribution of education is uneven in rural areas for secondary and tertiary education. For instance, none of the 900 household sample of rural households reported a member over 20 years old having any tertiary education. Note that education is concentrated in urban areas outside Abidjan, but not nearly to the degree evidenced in the countryside. - 50 - V. SUMMARY AND CONCLUSIONS Households that obtain income from wage activities are typically among the wealthiest in the C6te d'Ivoire, with the exception of a few very successful urban entrepreneurs. In contrast, farm households, particularly those with no source of off-farm income, are among the poorest. As most of the wage employment is in the cities, there is a significant urban/rural income difference in the C6te d'Ivoire. The difference is surprising given the export-crop lead growth strategy pursued by Ivorian policy-makers since independence, and the pervasiveness of coffee and cocoa cultivation in rural areas (nearly two-thirds of farmers grow either or both crops). Households with diversified labor portfolios are generally better off than households that depend on a single source of income (particularly when tied to self- employment activities) to satisfy consumption needs. As expected, consumption expenditures are more evenly distributed than income in the Cote d'Ivoire. Stocks of assets are more unevenly distributed than either income or consumption, with the exception of primary and secondary schooling in urban areas (particularly Abidjan), and rural land holdings. Neither of these results is surprising. Extensive efforts have been made in recent years to increase school enrollment via massive public subsidy programs, and land surpluses still exist in many parts of the country. In contrast to land and "low level" human capital, non-farm production capital is concentrated in the hands of a few urban entrepreneurs. The findings presented in the paper provide useful information about the link between poverty and productive activities in the C6te d'Ivoire. Poverty is a rural phenomenon and an urban phenomenon. In rural areas, the - 51 - poorest households depend primarily on subsistence farming and receive little off-farm income. They are more likely to live in the Savannah region (as opposed to the East Forest or West Forest), which has a lower level of public infrastructure than other regions. Unlike many other developing countries, landlessness and poverty do not seem highly correlated in the C6te d'Ivoire: land is surprisingly evenly distributed among rural households. Efforts to improve the living standards of the rural poor should focus on ways to increase basic services, to expand opportunities for off-farm employment, possibly to improve access to credit, and to increase the export orientation of agricultural activities. In contrast, the urban poor tend to derive the main part of their income from marginal informal sector activities (for example, petty trading and food sales) and low paid wage jobs in the private sector. The only asset which is equitably distributed in urban areas is primary education; physical capital tends to be highly concentrated in the hands of a few households. Interestingly, some 54 percent of households that live in urban areas outside Abidjan do not own the unit in which they reside, while 77 percent of households living in Abidjan are renters, more evidence of the concentration of wealth (as typified by housing units) in the C8te d'Ivoire. Income- oriented poverty alleviation strategies might include policies to: increase school enrollment for children from poor households (at secondary and tertiary levels); expand urban infrastructure; increase productivity in the informal sector; improve access to higher paid jobs in the public sector, and release potential capital constraints by increasing access to sources of credit for small entrepreneurs. - 52 - ANNEX I Imputed Rents for Urban Households In line with previous work (see, for example, Lee and Trost, 1978, Maddala, 1984) selectivity corrected (Heckman, 1979) hedonic rent equations were estimated to impute rental values for owner-occupied housing. The approach is conceptually simple; households who own their dwelling unit receive an annual flow of services from the unit equal to what they wouLd have had to pay to rent it -- in short, homeowners are treated as if they rent their dwelling unit from themselves. Rent imputations were made for homeowners in urban areas (Abidjan and other urban) only. Rural households were excluded primarily for lack of data; some 97 percent of rural households reside in units they own, and information on rural rental markets is too limited to allow any assessment of rent-to- value ratios in these areas. Further, many rural households are not permitted (due to family, ethnic, or other cultural sanctions) to sell land or buildings. The exclusion of rural households from rental assessments does not mean that rural households receive no housing service flows from owned dwelling units, but rather that we lack sufficient information to estimate these service flows with any degree of accuracy. In any case, exclusion of rural households from rent imputations will not affect savings estimates as rent imputations enter both income and consumption valuations. Some 46 percent of households in urban areas outside Abidjan reside in a dwelling unit they own, in comparison to only 23 percent in Abidjan. On the assumption that regionally diverse housing markets may operate along different lines, separate hedonic equations are estimated for (i) Abidjan and - 53 - (ii) other urban areas in aggregate. While it is preferable to further disaggregate other urban estimates, sample size limitations made this impractical. This Annex includes a brief description of (i) the underlying sample selectivity model (knowledge of standard hedonic rent equations is assumed), (ii) estimates for Abidjan and other urban areas of a probit indicator function to predict probabilities of renting and owning, and (iii) selectivity-corrected hedonic rent estimates for each group of households. Derivation of the Basic Model We only observe rental payments for households presently in the rental market; the CILSS data does not include a measure of estimated rent for homeowners, or the present market value of the dwelling unit. We will use the actual rents paid by renters to impute rents for households who own their dwelling units. Therefore, we first estimate an hedonic rent function which relates rents paid to characteristics of the housing unit and the neighborhood for all renters. In general terms, the hedonic function can be represented as R = f(Zl,...,Znlhousehold is a renter) (A1.1) where: R is the market rent, and Z1 -... 'Zn are characteristics of the neighborhood or housing structure. - 54 - In the end what we want is the expected rental value of owner- occupied dwelling units, that is, E[RI(Zl, ...,Z n), household is an owner]. We obtain this measure as follows. Assume all households are either renters or owners (squatters are ignored for the present). Define an indicator variable, S, which takes on a value of one if the household rents its dwelling unit and 0 otherwise. If we assume that rents are some linear function of housing and neighborhood characteristics, then unknown, if 6 = 0 R = t Z + e1 if S = 1 (A1.2) We specify a function to predict whether a household rents or owns its dwelling unit. The function includes the household characteristics and the household members on the right-hand side, and the indicator variable, 6S, on the left-hand side. It is likewise assumed to be linear in parameters. 6 = aX + c2 (A1.3) where: 6 is a 0,1 indicator variable, and X is a vector of independent variables describing household and individual characteristics. We want to obtain an unbiased estimate of rents from equation (A1.2). We cannot simply estimate (A1.2) for renters and use the coefficients - 55 - to impute rents for owners due to classic selectivity problems. Heckman (1979) has shown that an unbiased estimate of rents for the total population can be obtained using the following approach. As before, R = 8Z + e given that E[e2]=O, E[e 2]=a 6 = OX + e2 given that E[e2]=O, E[e22]=0221 f(e 1'62 )is bivariate normal, F(e VE2) is the corresponding cumulative density function, E[e 12 2] a12. R is observed iff £2> -aX, R is not observed if e 2< -aX. We know that the expected value of rents (based on Van der Gaag, 1984) for the total population is: E[RIZI = 8Z (A1.4) However, the expected value of rents for households in the rental market (that is, where the household is a renter rather than owner) is E[RjZ, e2>-aX] = BZ + y(A1) (A1.5) where A1 (typically called the Mills-ratio correction - 56 - factor) is defined as Xi f(-OZ) (Al.6) 1-F(-OZ) a12 (A1.7) a22 and f(.) and F(.) are the normal density and cumulative functions, respectively. Recall that we want to estimate the expected value of rents conditioned on the household being a homeowner, that is, the expected rental value of an owned dwelling unit. Given the above derivation, this is equivalent to E[RIZ, e 2<-ax] .(Al-8) We know that E[RtZI = E[RIZ, e2>-aX]*E[e2>-aX] + E[RIZ, 2<-aX]*E[e2<-WX] (Al.9) From this it follows that E[R|Z, e2<-aX] = 3Z + y(X2) (A1.10) where ?X2 = (Qx) (A1.11) 2 =(_-cX) - 57 - With equation (Al.10) we can compute the expected value of rent, conditioned not only on housing and neighborhood characteristics for homeowners, but also on the likelihood of ownership. The measure of imputed rent used in this study is: the expected rent for the sample of owners rather than the full sample (as is more typically done in income and consumption assessments). The model is estimated in two steps following Maddala (1984). First, estimates of -aX are obtained from a binary probit model using 6 (the decision to own or rent) as the dependent variable. Second, results are then used to estimate A, , which is one of the exogenous variables in the hedonic rent equation. x2 is likewise estimated from the probit equation, and is used with the a's from the hedonic rent equation(s) to impute rents for home- owners, that is, Imputed Rent = ;Z + y (X2) The next section describes the specification of and empirical results from the binary owner/renter probit equation and hedonic rent functions for CILSS sample households. Empirical Results The owner/renter probit model includes a wide array of variables specific to the household. Among these are: (i) employment variables; (ii) measures of housing subsidies; (iii) measures of assets and income; (iv) demographic variables; (v) length of time in present location; and (vi) specific characteristics of the household head. Table I-lA shows means and - 58 - standard deviations of these variables for households in Abidjan and in other urban areas. According to this table, some 11 percent of households in Abidjan and 12 percent in other urban areas receive some form of wage-related housing subsidy. Only 4 percent of dwelling units are used for business purposes in Abidjan as opposed to 11 percent in other urban areas. Also, households in Abidjan have lived in the city (although not necessarily in the same dwelling unit) for an average of 17.63 years, while households in other urban areas have lived in the same place for an average of 13.14 years. Note also that 30 percent of households in Abidjan are not native to the country, as compared to 20 percent in other urban areas. This suggests that there is a significant proportion of non-indigenous people in the Cote d'Ivoire. Households are large in each area; in Abidjan, households have on average 7.37 members, while in other urban areas they have 8.85 members. The hedonic rent equations only include variables that relate to the housing unit or neighborhood. These are (i) descriptions of the physical structure; (ii) availability of public services; and (iii) size of the unit. Table I-1B shows means and standard deviations for exogenous variables included in the hedonic rent equations. From this table, we see that dwelling units in urban areas outside Abidjan are substantially larger than those in Abidjan -- units in Abidjan are on average 54.95 square meters, as compared to 81.37 square meters in other urban areas. Note also that units in Abidjan are far less likely to be occupied by single families, but generally have higher levels of public services (water and sanitation). - 59 - Table I-1A: Means and Standard Deviations of Independent Variables: Indicator Function For Tenure Choice Abidjan Other Urban Areas Standard Standard Mean Deviation Mean Deviation Employee in household? .74 - .58 Public employee in household? .30 - .31 Does household receive housing subsidies? .11 - .12 - Value of subsidies per month (CFA) 5,987 20,796 6,012 18,549 Is dwelling unit used for business? .04 - .11 - Measures of assets and income - total expenditures-/ (CFA/yr) 2,513,928 1,721,709 1,684,050 1,162,245 - present value of business assets (CFA) 664,660 4,378,490 276,114 1,211,956 - income from durables (CFA/yr) 40,897 89,584 33,464 36,382 - value of outstanding debts (CFA) 441,394 2,437,225 219,582 1,228,481 - present value of cash savings (CFA) 322,285 1,085,908 342,494 1,607,442 Years household lived in current place 17.63 11.73 13.14 13.89 Household size 7.37 4.54 8.85 5.79 Characteristics of the household head - female? .12 - .11 - - not Ivorian? .30 - .20 - - age 42.12 11.32 44.81 13.62 - years of education 5.95 5.99 4.29 5.23 Notes: I/ Excludes imputed rents and durables. Source: CILSS tabulations. Table I-2 shows parameter estimates of probit owner/renter indicator functions for households in Abidjan and other urban areas. In general, the results are encouraging and reveal some interesting facets of Ivorian housing - 60 - markets. For households in Abidjan, the probability of renting increases when total expenditures, housing subsidies, and education increase, and decreases when durable flows, outstanding debt, and length of residence increase. Interaction effects (total expenditures with employment status, age, and education of the household head) were generally not significant, with the exception of age*expenditures, which has a negative (and significant) sign. Combining the two expenditure variables (total expenditures, age*total expenditures) tells us that the probability of renting increases as income rises, but at equal levels of income "older" households are more likely to own their dwelling unit more often than households with younger heads. The models for other urban areas are in general similar to those for Abidjan, but with one or two notable exceptions. For instance, households with at least one public employee are more likely to rent a dwelling unit than those without a public employee. Also, large households are less likely to be renters than smaller households, as are female-headed households, and households with older heads. Not surprisingly, non-Ivorian households are significantly more likely to rent a dwelling unit than indigenous households. At least some of the results found in urban areas outside Abidjan reflect the stronger rural orientation of these households -- other "urban" areas range from medium-sized cities to villages. (Note that specifications are not identical between the two areas due to estimation problems with the second step of model estimation -- the hedonic rent functions. Some full(er) specifications caused problems in the estimated ratio of error variances between the two equations in the model.) - 61 - Table l-lB: Means and Standard Deviations of Independent Variables: Hedonic Rent Equation Abidjan Other Urban Areas Standard Standard Mean Deviation Mean Deviation Area (meters2) 54.95 57.53 81.37 67.31 Dwelling unit characteristics - walls of permanent materials? .96 - .80 - - electric lighting? .81 - .75 - - fuel supply gas or electricity? .29 - .13 - - indoor water faucet? .44 - .21 - - flush toilet? .80 - .24 - - inside toilet facilities? .45 - .25 - Structural characteristics - detached, single family home .09 - .28 - - compound, single family occupied .07 - .34 - Rent (CFA/month) 25,961 36,771 22,408 22,215 Source: CILSS tabulations. Table I-3 shows hedonic rent estimates for households in Abidjan and other urban areas. The variable on the left-hand side is cash plus in-kind rental payments measured in CFA per month. A standard linear model is estimated for each subsample. Also, a semi-log model -- the natural log of monthly rent is used as the dependent variable -- is estimated for Abidjan. This latter model was used to make rent imputations for households in Abidjan. Comparisons can be made between the rent equations for Abidjan's households and for households residing in other urban areas, although the models are not symetric so far as variable inclusion is concerned. Note that both models include a measure of the total area of the housing unit (and area squared, to control for non-linearities), structural characteristics - 62 - Table 1-2: Indicator Functions for Choice of Housing Tenure, Renters Versus Owners (all coefficients estimated with respect to the rent alternative) Abidianll Other Urban Areas Employees in household? -.286 (0.75) .332 (1.30) Public employees in household? - .968 (3.29) Does household receive housing subsidies? 1.259 (1.76) Value of subsidies per month (CFA) -.000011 (1.38) Is dwelling unit used for business? .411 (0.75) -.205 (0.07) Measures of assets and income - total expenditures (000 CFA/yr) .000468 (1.79) .00116 (2.26) - present value of business assets (000 CFA) -.0000357 (0.79) - - income from durables (000 CFA/yr) -.00395 (1.37) .0025 (0.50) - value of outstanding debts (000 CFA/yr) -.0000346 (0.64) - - present value of cash savings (000 CFA) .0000249 (0.23) - Years household lived in current place -.012 (1.49) -.013 (1.21) Household size - -.150 (4.91) Characteristics of the household head - female? - -.328 (1.09) - age - -.033 (1.86) - not Ivorian? - 1.198 (4.32) - years of education .031 (1.31) -.556 (0.52) Interaction variables - Employees *total expenditure (000 CFA) .000129 (0.75) - - Age of head total expenditures (000 CFA) -.0000154 (3.88) -.0000173 (1.6) - Age of head education of head - .0017 (0.64) Intercept 1.389 (3.77) 1.482 (1.85) Model statistics Log-likelihood value - at zero -172.3 -209.3 - at convergence -136.5 -103.4 x2 statistic 71.73 211.96 Percent renters 76.0% 50.3% Number of cases 313 302 Notes: '' Asymptotic t-statistics are in parentheses. Source: CILSS tabulations. - 63 - (single family versus non-single family units), some dwelling unit characteristics (materials used in the walls, source of fuel supplies), and "cluster"-specific dummy variables. Models differ in terms of dwelling characteristics -- for Abidjan, a source of lighting dummy is included, while dummy variables relating to water and sanitation were found significant for other urban areas. Table I-3 shows households in Abidjan are willing to pay an average of CFA 309.7 per square-meter per month, in comparison to CFA 337.2 per square-meter per month for urban households outside Abidjan. Note that rents per unit area fall gradually with increasing size of the dwelling unit in other urban areas, but tend to rise with increasing size in Abidjan. Note also that the area-based variables evidence stronger effects in the log-linear model, which suggests that there are substantial non-linearities in the relationship between rents and the size of dwelling units. In Abidjan, households are willing to pay CFA 12,863 per month to obtain walls made of permanent materials (cement, brick, stone, wood, iron), CFA 588.7 per month for electricity, and CFA 19,067 to reside in a unit with gas or electric cooking facilities. In comparison, households in other urban areas are willing to pay only CFA 861.2 per month for walls made of permanent materials, CFA 12,949 for gas or electric cooking facilities, CFA 3,384 for an indoor water faucet, CFA 7,177 for a flush toilet, and CFA 6,955 for inside toilet facilities. Also, households were willing to pay an extra CFA 10,237 per month to live in a single family detached home in Abidjan, and CFA 49,712 to live in a single family compound. In contrast, households in other urban areas were - 64 - Table 1-3: Hedonic Rent Equations Corrected for Sample Selectivity Abidjan I/ 2/ Other Urban AreasY Log (rent) Rent Area (meters2) .023 (7.93) 309.7 (2.9) 337.2 (8.93) Area-squared -.000049 (3.08) .843 (1.5) -.392 (7.37) Dwelling unit characteristics - walls of permanent materials? .416 (2.28) 12,863 (1.9) 861.3 (0.30) - electric lighting? .177 (1.79) 588.7 (1.2) -. - fuel supply gas or electricity? .552 (5.67) 19,067 (5.4) 12,949.0 (4.61) - indoor water faucet? - - 3,384.1 (1.22) - flush toilet? 7,177.2 (1.87) - inside toilet facilities? - - 6,954.7 (2.10) Structural characteristics - detached single-family home -.154 (0.94) 10,237 (1.7) -7,541.0 (3.27) - compound, single family occupied .755 (2.53) 49,712 (4.6) -2,721.5 (1.08) Community Intercepts: Abidjan - Bietry .593 (2.43) 59,226 (6.5) _ - Autre Abobo .282 (1.40) 13,499 (1.8) - Community intercepts: other urban areas - Agnibilekrou - 2,790.5 (0.61) - Man -2,231.2 (0.55) - Bouake Air Force - 8,565.5 (1.94) Intercept 7.988 (38.92) -17,465) (2.4) -1,846.9) (0.57) Lambda .176 (1.05) 13,309 (2.2) 5,270.0 (2.40) Model statistics R2 .645 .696 .782 F-statistic 41.43 51.9 38.16 Number of cases 238 238 152 Mean of dependent variable 9.636 25,961 22,408 Notes: 1/ Dependent variable is the natural logarithm of monthly rent (cash + in-kind) and monthly rent (cash + in-kind). 2/ Asymptotic t-statistics are in parentheses. 3/ Dependent variable is monthly rent (cash + in-kind). Source: CILSS tabulations. - 65 - actually willing to pay less for single family units than multi-family units. However, the signs and magnitude of the parameters must be interpreted in the light of other variables included in the model -- we are, in effect, controlling for different factors in each of the regionally-based models, and these differences may affect the scale and direction of all coefficients. The coefficient for the Mills-ratio correction factor is positive and significant in the linear models for both Abidjan and other urban areas, although not significant in the log-linear model. This means that there is significant correlation between the error term in the probit equation and the error term in the hedonic rent equation (recall that y , the correction coefficient, is equal to a /a 22 from (A1.7)); in short, sample selectivity 12 22 appears to cause problems in imputing rents to homeowners in the C6te d'Ivoire, and should therefore take account of (for example, as we have done here) the imputation procedure. The explanatory power of the model in both regions is good -- for Abidjan, the estimated R2 is .645 and .696 (semi-log and linear, respectively), while for other urban areas the R2 is .782 (linear only). - 66 - ANNEX II Annualized Value of Durables In the CILSS, households were asked to identify the durables they own, when each was purchased, for what price, and the current value of each durable at the time of the survey. This information was used to impute an annualized flow of services obtained from the stock of durables owned by the household. This Annex briefly describes the stock of durables owned by Ivorian households, and how durable flow imputations were made. Table II-1 shows ownership by region for thirteen durable stocks. According to these tabulations, urban households, whether in Abidjan or outside, they are much more likely to own almost any kind of durable good than rural households. Bicycles form the one exception to this rule; some 44.5 percent of rural households report owning a bicycle at the time of the survey, as compared to only 22.6 percent of households living in cities outside Abidjan and 3.6 percent of households in Abidjan. Table II-1 also shows that the most frequently owned household durable is a radio/cassette player -- some 45.8 percent of households in the C6te d'Ivoire report owning at least one. Sixty eight percent of households in Abidjan own a television set, 57.8 percent of households in other urban areas, and 8 percent of rural households, for a national total of 30.7 percent. About half of the urban households also reported owning a fan and a refrigerator; country-wide totals are 22.9 and 24.3 percent, respectively (ownership is rare in rural areas). Sewing machines are fairly common throughout the C8te d'Ivoire; ownership levels are 33.2 percent, 36.4 percent, and 12.7 percent in Abidjan, other urban areas, and rural areas, - 67 - Table II-1: Percentage of Households Owning Durables, By Durable Category and Region I Other Durable Abidjan Urban Areas Rural Areas Total Sewing machine 33.2 36.4 12.7 21.9 Gas stove 37.7 23.8 2.0 14.0 Refrigerator 53.6 50.0 4.7 24.3 Air conditioner 19.2 11.1 0.0 6.4 Fan 49.7 49.4 3.7 22.9 Radio 30.8 28.0 19.6 23.5 Radio/cassette player 55.7 64.2 36.4 45.8 Phonograph 6.3 6.6 3.2 4.5 Stereo equipment 20.9 13.9 0.5 7.6 Television set 68.0 57.8 8.0 30.9 Bicycle 3.6 22.6 44.5 30.7 Mobylette 1.5 21.4 18.5 15.3 Car or truck 22.2 14.8 2.2 9.0 Source: CILSS tabulations. respectively. Note that only 22.2 percent of households in Abidjan own an automobile or truck, as compared to 14.8 percent in other urban areas and 2.2 percent in rural areas. These levels of automobile ownership are typical for sub-Saharan African countries. Table II-2 shows an average estimated annual rate of depreciation for each class of durables included in the CILSS questionnaire. These typically range from around 9 percent (sewing machines and bicycles) to a high of 17 - 68 - percent (fans, radios, air conditioners). The measure is derived as follows: Consider the relationship between the value of a commodity ptLrchased at some time to for price po and the value today, represented by market price Pt. If the durable good was purchased t years ago, Pt = o(1-6)t (A2.1) where: 6 is the rate of depreciation, and all other variables are as previously defined. Solving for 6, equation (A2.1) becomes: 6 = 1 - e gP- logePo)/t) (A2.2) Note that both the purchase price (po) and the estimated price. at the time of the interviews (Pt) is needed to estimate an average rate of depreciation for some particular commodity. Roughly 3 percent of households in the CILSS did not report either or both measures, and were excluded in estimating average depreciation rates (although these households were mot excluded from service flow imputations). Also, households obtaining durables by way of non-market mechanisms (for instance, as gifts) were likewise excluded in estimating depreciation rates. These households were identified by a self-reported zero price at the time of purchase. - 69 - Table II-2: Estimated Depreciation Rates by Type of Durable Durable Depreciation Rate Sewing Machine .091 Gas stove .153 Refrigerator .132 Air conditioner .173 Fan .167 Radio .161 Radio/cassette player .146 Phonograph .142 Stereo equipment .122 Television set .122 Bicycle .092 Mobylette .125 Car or truck .117 Source: CILSS tabulations. Following Diewert (1974), and Deaton (1980), the annual flow of services from a particular durable is defined as V= Pt - p (1-6) (A2.3) - 70 - where: r is the real interest rate or opportunity cost of capital, V is the annualized flow of services from durable stocLcs, Pt+l is the price of the durable at the beginning of next year (that is, at the end of this year), and all other variables are as previously defined. For simplicity, we assume that all price changes (pt-ptl) are caused by stock depreciation. By implication, then, V = p (1 - (1-) (l+r) or, Pt (r+6) (A2.4) t (1+r) In short, the annualized flow of services from durable stocks is equal to the opportunity cost of capital plus the rate of depreciation times the current value of the stocks (or price today), all divided by one plus the opportunity cost of capital. The main problem in measuring the annual flow of services from durables is that we do not know the opportunity cost of capital. Interest rates vary across households in LDC's, with this variation being a function of economic conditions and how well various commodity markets operate. Published market rates may bear little relationship to actual rates. For this study, we simply assumed a zero opportunity cost of capital in computing annualized durable flows. While this is a far from ideal solution, it seems the most practical given the objectives of the research, that is, analyzing private savings behavior in the C6te d'Ivoire. Recall that durable flows enter both - 71 - income and consumption estimates, and are netted out in assessing residual savings rates. Thus for most of the work described here, interest rate assumptions are irrelevant. However, how much impact will the assumption of zero interest rates have on the welfare evaluations presented in the paper? This depends on the values of the interest rates and depreciation rates, that is, on the specific household and the type of commodity under consideration. For example, if the depreciation rate is .08 (roughly the lowest found among the CILSS durable types) and the interest rate is .10, then annualized double, depending on interest rate assumptions. However, in an absolute sense, the differences are not so large, as durable flows constitute only a small proportion of total income, so that even a doubling of the estimated values will have little impact on overall welfare. For this reason, and, given the objectives of the research, the admittedly ad hoc assumption of a zero opportunity cost of capital in estimating the annualized flow of services from durables is deemed acceptable. - 73 - REFERENCES Ainsworth, Martha, and Munoz, Juan. "C8te d'Ivoire Living Standards Study: Design and Implementation", Living Standards Measurement Study Working Paper No. 26, The World Bank, Washington, D.C., 1986. Becker, Gary S. "A Theory of the Allocation of Time", Economic Journal, 75, 1965, pp. 493-517. Evenson, Robert E., and Quinzon, Elisabeth K. "Time Allocation and Home Production in Phillipine Rural Households", Economic Growth Center, Yale University, (mimeo), 1977. Friedman, Milton. A Theory of the Consumption Function, Princeton: Princeton University Press for NBER, 1957. Glewwe, Paul. "The Distribution of Welfare in the Republic of C6te d'Ivoire in 1985", Living Standards Measurement Study Working Paper No. 29, The World Bank, Washington, D.C. 1987. Gronau, Rueben. "Who is the Family's Main Breadwinner -- The Wife's Contribution to Full Income", NBER Working Paper No. 148, September, 1976. Grootaert, Christiaan. "Measuring and Analyzing Levels of Living in Developing Countries: An Annotated Questionnaire", Living Standards Measurement Study Working Paper No. 24, The World Bank, Washington, D.C., 1986. Hecht, Robert M. "The Ivory Coast Economic 'Miracle': What Benefits for Peasant Farmers", Journal of Modern African Studies, March, 1983. Heckman, James J. "Sample Selection Bias as a Specification Error", Econometrica, 47(1), 1979, pp. 153-61. Kimmerling, Baruch. "Subsistence Crops, Cash Crops, and Urbanization: Some Materials from Ghana, Uganda, and the Ivory Coast", Rural Sociology, 36(4), pp. 471-87, 1971. Kozel, Valerie. "Income, Consumption, and Savings Behavior in the C8te d'Ivoire", Ph.D. thesis, Department of Urban Studies and Planning, Massachusetts Institute of Technology, 1987. Kusnic, Michael, and DeVanzo, Julie. "Income Inequality and the Definition of Income: The Case of Malaysia", Rand Corporation Report No. R-2416-AID, June, 1980. Lee, Lung-Fei, and Trost, Robert P. "Estimation of Some Limited Dependent Variable Models With Application to Housing Demand", Journal of Econometrics, 8, pp. 357-382, 1978. - 74 - Maddala, G.S. Limited-dependent and Qualitative Variables in Econometrics, Econometric Society Publication No. 3, Cambridge University Press, Cambridge, 1984. Malpezzi, Stephen, Mayo, Stephen, and Gross, David. "Housing Demand in Developing Countries", World Bank Staff Working Paper No. 733, The World Bank, Washington, D.C., 1985. Mayer, Thomas. "The Propensity to Consume Permanent Income", American Economic Review, 56(5), 1966. Muellbauer, John. "The Measurement of Long-Run Living Standards: An Application and Evaluation of the Permanent Income Hypothesis", (niimeo), July, 1982. Psacharopolous, G. "A Further International Update and Implications", Journal of Human Resources, 20, pp. 583-597, 1985. Ruf, F. "Quelle Intensification en Economie de Plantation Ivoirienne? I - Histoire, Systemes de Production et Politique Agricole", L'Agronomie Tropicale, 39(4), 1984. Singh, R.P., and Asokan, M. "Concepts and Methods for Estimating Incomes in Village Studies in Semi-Arid Tropical India", Andhra Pradesh, ICRISAT, November, 1981. Van der Caag, Jacques. "From Consumption to Consumer Expenditures", (mimeo), 1984. Van der Gaag, Jacques and Wim Vijverberg. "Wage Determinants in C6te d'Ivoire", Living Standards Measurement Study Working Paper No. 33, The World Bank, Washington, D.C., 1987. Visaria, P. "Poverty and Living Standards in Asia: An Overview of the Main Results and Lessons of Selected Household Surveys", Living Standards Measurement Study Working Paper No. 2, The World Bank, Washington, D.C., 1980. World Bank. "The Ivory Coast in Transition: From Structural Adjustment to Self-Sustained Growth", Country Economic Memorandum, The World BaLnk, Washington, D.C., January, 1986. Zartman, I.W. and Delgado, C. The Political Economy of Ivory Coast, a SAIS Study on Africa, New York: Praegen, 1984. Distributors of World Bank Publications ARGBNlTNA FINLAND M1EXICO SPAIN Cai.A-HiMX1~SRI. Aka.-ftrX*knuqppo IalamO Mwadi-PmiaaUbrau, SA, Ca,ana G uco, P.O. Bos 12 y Apaldo Posa 22n-s Caid37 Floaida 165, 4th IoorOfc. 453/465 SFO0101 140llfTalpan. Medx D.F. 2C011 Madrid 1333 Buenors Aim Hela 10 MOROCCO Lftrer. lh ..ImAosa ASDCE AUSTRALIA, PAPUANEWGUINEA, FRANCE Soc#ie d'Ehid-Marketg Marocine Canwl deCent,391 FIll SOLOMON ISLANDS, World BaJk Publliatis 12 me MMrt, d. d'Anfa Wt91 VANUATU, AND WES-ERN SAMOA 66, avenue d'l[& Car b-ma D.A.Bolba&So%r- 75116 Puts SRI LANKA AND THE MALDIVES 11-13tnSa Streaa NETHERLANDS LdkeHouseBwmhbop Mitchram 3132 GERMANY, FEDERAL REPUSUC OF InOr-Publikaales b.v. P.O Box 244 Vidoria UNO-Volag P.O. Box 14 tC0, SirCdttamdaam A. Gadiar PdpeladerAlleeS55 724D RBA Lcshuo M 2wath AUSTRI Dl-S3WlBoami Colasabo I GC-ad and Co. NEW ZEALAND Graben3l GREECE HilblibtaryandlIorenstitonSersce SWEDEN A-10ll Wior NESME Prtvata N Bg Foriqlaliilr 24,1pp,mamassuca 'etPlatia Plasadrs Naw Mrk PAtlos Fadabokdm"ortoa BAHRAIN AMi5-t635 Auddad Rra tan 12Om 16356 Babrain Reamdhnd C Ilam-y S-10 27 SodmthttE Asodati Lid. GUATEMALA NIGERIA P.O. Box 22103 Lo hia dra aa UrvityPress Uminad forax ani MaarmaTcnwm3t7 Cstw Cultual Ptedra Santa ThreeCriBw lCdlngj ae dX Weruvagce:WilhwnsAl 11 call. 6-EOzarma 1 P.Ivat.Mall Bg 5119 Boa 390O4 BANGLADESH Guatemala city badn S-104 25 Stodalato Micro hIduaica Devdopelt Aseatance Sudety MIDAS) HONG KONG, MACAO NORWAY SwITZERLAND Houav Road 16 Ad 20MMLad. Naivesen bdsoa Center Fr klt DRhamondi RAres 6 91.146 PHnce Edwad Road,W. BetneadNarvesensved2 Ubrairie Payat Dhaka 1209 NowlOon P.O. Box 6125 Esaaad rue Gram Hong Kong N-0602 Oslo 6 C-p Adli381 Bmwkdasimac CHIZIlGCaner.11 154NurAhmaedSak HUNGARY OMCAN Chjhtagong 4000 Klt_ra MEMRE hinformaioSvim I'e orwp oris: P.O. BO 139 P.O. Bo 1613, SebAirpot LibrAde Payot 76, KDA. Avenue 13d Budapet 62 Muscat Serv dw AbIrnemmbis Nassa Casepa"Ui23II INDIA PAKISTAN CH 10D2 Lauonnue SELGWUM Allied Publshes Prvate LAd. Mhz Bo.* Agency Publicatos,adesNaboratUrde, 751 Mount Road 65. Sh hrah.eQuaideAaan TANZANIA Av. du Rd 202 Madrs- 6001002 P.O. Box No. 729 Odord Uuivedsty Press 1060 B n1. Laho3 P.O. Box 5299 Issads~~~~~~~~~~~~~~kw ~~~~~DaresSabearn BRAZIL 15jNHerediaHMar PERU Publi-cesoTe-ira laotecadssais Ballard Etate Editloial Dsassola IA THAILAND Ltda. Bnbsy-440D03d Apahado.3i24 Delment ece Rua Pexoto Couide, 209 Lima 3D6 Silm Rod 014095aoPaulo,SP 13/l4AaAli Road Bangkok New DrMl -110 002 PrLIPPINES CANADA Natiasad Boas Store TRINIDAD & TOSAGO, ANTiGUA LeDi*ffuser 17 Chlittalja Aventue 701 RizalA Avenue BARBUDA, BARBADOS, CYP.5,115 irueAmpbre Cabotta-70D072 P.O. BO 1934 DOMINICA. GRENADA, GUYANA. BuesdcavlliU Quoueb MetroM aila JAMAICA, MONTSERRAT, ST. J4O 586 Jayodeva Houmei Buldta XiTTS & NEVIS. ST. LUCIA, Stah AfRoad GadIdnagar POLAND ST.VINCENT &SGRENADINES CHINA BaagaLore-50119 OPPAN pyaternaaa StudtisUuidt ChissH-nsassdal&EconaniPublishng Polac KAitusy i Nuki #9Wt9Street Ho-se 3-5-1129 Kadtiguds CiwRoad 0OD9-1 War.swo C-epe O, DaF.P SL Dsgie Hydabad - SOD 027 Ttinidad. Wesl ltdie Bot9g PORTUGAL Prartha,s Flat. 2ai Flarur Lvraia Postsga TURXCEY COLOMBIA NearThakarelau g.Naangpur 2 RosoCarctno70-74 HasT KitapevY , AS. EnlaceU lid. Aluoedabad-3i0009 120D Llm lbla Cadde, No. 469 Apado Arrou 34V00 Bnep Bogota D.E. Pathiai Hoae SAUDI ARABIA, QATAR Istanbul 16-A AawkSMarg jarlBoalStare COSTA RICA L-ounv-226a0DI PO. Bx 3196 UGANDA Libreria Trepr Riyedhil,71 Uga da Bookshop Calle ll-13 iNDONESIA P.O. Box 7145 Av. Fernandez Gueal Pt hubs. Limited MARB InfmnstioScvi Sw mpla San- JeJL Sasas Ratuloagi 37 Braau*sfkw P.O. Bo.1id AlA alu UNITED ARAB EMIRATES COTE I'iVOLRE ]ikara PUNt Al Dauaa Cnter MElARB Gulf Co. Centre d'Bdln et de Dfhmia Firest Ioor P.O. Boas 6097 Ahicainea(CEDA) IRELAND P.O. ox 715 Sh'a"ah D4 BY. 541 TDCPabiheRirah Abidja04PLaaeau 12 North Fred&x RSydh UNITED KINGDOM Dublin I Haj AbduUllh Abieza Builiding cOriBox3d. CYPRtUS King Kisaed Street P.O. Bas 3 MEAMR E Wo.maaionSev irea ITALY PO. BoX 39 Alton, Hampshire GU34 2PG P.O. Boxs21L °d1a Cammaiosada SaSFi SPA EDgtand Mcewa Vaa Bedetta oEratl, 120/10 CaDENAR Pastale 552 3.3Mcamnt wessaonAwadSte URUGUAY DENMARtK 20125 Plcrace P.O. Box 5978 Iraht.t Nadmsalde[ Ulbra SaoslundsUtterltor Jaddah Sasjose,1116 RooeamsAlA611 JAPAN MrnJoevild' DK-1970 Frederibg C EBtbr Book Samce SINGAPORE. TAIWAN, MYANMAR, 37-3. Hasngo3.C3aae. Bsunkyo-Isa 113 BRUNEI VENEZUELA DOMINICAN REPIUBUC Tksyo ldalorsato Publicationsa [brata det Eote Editsra TaIlr C. po rA PKdvNY eAd Aptdo. 60337 R-teabrcon, a babel Ia Calair 5109 KENYA 02-06 lat Fi. Pei-Pa hasduotri Caraas1060-A Apatado P"l 2190 AhicaBookSariec(BA)Ltd. Bldg SantoDurningo P.O. Box 45245 24N-Induarial Road YUGOSLAVIA Nahm Slagapore 1953 Jugowlveraa KXiga EL SALVADOR YU-11010 Belgrade Tag Repsblike Puodes KOREA, REPUBUC OP SOUTH AFRICA,BOTSWANA Avesida Manue BoriqueArujo J353 Pon Koram Book C}aiiotn ForrsBkgiaaikor EdificioSISA.ler PiO P.O. Box 101, Kwagwhan OKorsUaiveMtyPresMinth Sas Salvador S. Airica P.O. Box 1141 EGYPT, ARAR REUBLiC OF XUWAIT CapeTermB)D Al Abran MEMRB Idermion SeCica Al GAa Street P.O. Beo Si65ro Forr.ubwtio ordea Cdarn inu-tioa Subclpreptlon Serds MALAYSIA P.O. Box 41095 TheMiddleBoatOC5r-er Urdvenatiyd tMlaya Coopestas oig.aB B Osawasti Suret Bkda.hp. Ltindtd al.,oh ,og 2024 Cairo PD. OBX 1127 Jharn Pntai Banu Kuala -=npr LSMS Working Papers (continued) 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 Determninants 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 Cote d'Ivoire No. 44 The Living Standards Survey and Price Policy Reform: A Study of Cocoa and Coffee Production in Coted'Ivoire No. 45 Measuring the Willingness to Pay for Social Services in Developing Countries No. 46 Nonagricultural Family Enterprises in C6te d'Ivoire: A Descriptive Analysis No. 47 The Poor during Adjustment: A Case Study of Cote d'Ivoire No.48 Confronting Poverty in Developing Countries: Definitions, Information, 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 Payfor Education in Developing Countries: Evidence from Rural Peru No. 55 Rigidite des salaires: Donnees microeconomiques et macroeconomiques sur l'ajustement du marc}e 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: Evidence from Two Developing Countries No. 61 Large Sample Distribution of Several Inequality Measures: With Application to COte d'Ivoire No. 62 Testingfor Significance of Poverty Differences: With Application to C6te d'lvoire 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 Cote d'Ivoire No. 67 Does Education Pay in the Labor Market?: The Labor Force Participation, Occupation, and Earnings of Peruvian Women The World Bank Headquarters European Office Tokyo Office N 1818 H Street, N.W. 66, avenue d'1ena Kokusai Building CD 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 Facsimile: (202) 477-6391 Facsimile: (1) 47.20.19.66 Telephone: (3) 214-5001 Telex: wui 64145 WORLDBANK( Telex: 842-620628 Facsimile: (3) 214-3657 RCA 248423 WORLDBK Telex: 781-26838 Cable Address: INTBAFRAD WASHINGTONDC