Women's work, education, and family welfare in Peru

-C,fid)v J- KC WDP-1 16 FILE copy World Bank Discussion Papers Women's Work, Educatlon, and Family Welfare in Peru Barbara K. Herz and Shahidur R. Khandker, editors F*ILEoni  Recent World Bank Discussion Papers No. 58 Making the Poor Creditworthy: A Case Study of the Integrated Rural Development Program in India. Robert Pulley No. 59 Improving Family Planning, Health, and Nutrition Outreach in India: Experiencefrom Some World Bank-Assisted Programs. Richard Heaver No. 60 Fighting Malnutrition: Evaluation of Brazilian Food and Nutrition Programs. Philip Musgrove No. 61 Staying in the Loop: International Alliancesfor Sharing Technology. Ashoka Mody No. 62 Do Caribbean Exporters Pay Higher Freight Costs? AlexanderJ. Yeats No. 63 Developing Economies in Transition. Volume I: General Topics. F. Desmond McCarthy, editor No. 64 Developing Economies in Transition. Volume HI: Country Studies. F. Desmond McCarthy, editor No. 65 Developing Economies in Transition. Volume III: Country Studies. F. Desmond McCarthy, editor No. 66 Illustrative Effects of Voluntary Debt and Debt Service Reduction Operations. Ruben Lamdany andJohn M. Underwood No. 67 Deregulation of Shipping: What Is to Be Learnedfrom Chile. Esra Bennathan with Luis Escobar and George Panagakos No. 68 Public Sector Pay and Employment Reform: A Review of World Bank Experience. Barbara Nunberg No. 69 A Multilevel Model of School Effectiveness in a Developing Country. Marlaine E. Lockheed and Nicholas T. Longford No. 70 User Groups as Producers in Partidpatory Afforestation Strategies. Michael M. Cernea No. 71 How Adjustment Programs Can Help the Poor: The World Bank's Experience. Helena Ribe, Soniya Carvalho, Robert Liebenthal, Peter Nicholas, and Elaine Zuckerman No. 72 Export Catalysts in Low-Income Countries: A Review of Eleven Success Stories. Yung Whee Rhee and Therese Belot No. 73 Information Systems and Basic Statistics in Sub-Saharan Africa: A Review and Strategyfor Improvement. Ramesh Chander No. 74 Costs and Benefits of Rent Control in Kumnasi, Ghana. Stephen Malpezzi, A. Graham Tipple, and Kenneth G. Willis No. 75 Ecuador's Amazon Region: Development Issues and Options. James F. Hicks, Herman E. Daly, Shelton H. Davis, and Maria de Lourdes de Freitas [Also available in Spanish (75S)] No. 76 Debt Equity Conversion Analysis: A Case Study of the Philippine Program.John D. ShiUing, Anthony Toft, and Woonki Sung No. 77 Higher Education in Latin America: Issues of Effidcency and Equity. Donald R. Winkler No. 78 The Greenhouse Effea: Implicationsfor Economic Development. Erik Arrhenius and Thomas W. Waltz No. 79 Analyzing Taxes on Business Income with the Marginal Effective Tax Rate Model. David Dunn and Anthony Pellechio No. 80 Environmental Management in Development: The Evolution of Paradigms. Michael E. Colby No. 81 Latin America's Banking Systems in the 1980s: A Cross Country Comparison. Felipe Morris, Mark Dorfinan, Jose Pedro Ortiz, and others. No. 82 Wy Educational Polides Can Fail: An Ovenriew of Selected African Experiences. George Psacharopoulos No. 83 Comparative African Experiences in Implementing Educational Polides. John Craig No. 84 Implementing Educational Polides in Ethiopia. Fassil R. Kiros (Continued on the inside back cover.)  1 16 c World Bank Discussion Papers Women's Work, Education, and Family Welfare in Peru Barbara K. Herz and Shahidur R. Khandker, editors . ! The World Bank Washington, D.C.  Copyright C 1991 The Intemnational 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 February 1991 Discussion Papers present results of country analysis or research that is circulated to encourage discussion and comment within the development community. To present these results with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. Any maps that accompany the text have been prepared solely for the convenience of readers; the designations and presentation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries concerning the legal status of any country, territory, city, or area or of the authorities thereof or concerning the delimitation of its boundaries or its national affiliation. The material in this publication is copyrighted. R'vt-qests 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, widhout 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 tide hist (with full ordering information) and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Publications Sales Unit, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'1ena, 75116 Paris, France. ISSN: 0259-210X Barbara K. Herz is chief of, and Shahidur R. Khandker a research economist in, the Women in Development Division of the World Bank's Population and Human Resources Department. Library of Congress Cataloging-in-Publication Data Women's work, education, and family welfare in Peru / Barbara K. Herz, Shahidur R. Khandker. p. cm. -- (World Bank discusslon papers ; 116) ISBN 0-8213-1774-1 1. Women in development--Peru. 2. Women--Peru--Economic conditions. 3. Women--Peru--Social conditions. I. Herz, Barbara Knapp. II. Khandker, Shahidur R. III. Series. HQ1240.5.P4W66 1991 305.42'0985--dc2O 91-6831 CIP  PREFACE Based on evidence gradually emerging from the developing world, the World Bank believes that the main way to help women, and thereby contribute to poverty reduction, less environmental stress, and other development objectives, is to enable vomen to raise their own productivity and income. During the U.N. Decade for Women (1975-85), efforts were made to increase public awareness about the difficulties women face and to institute policies promoting expanded opportunities for women. But rigorous research remained scarce on how to do that and on what precisely the benefits would be. The World Bank has therefore begun to prepare some 23 country-level assessments and action plans and undertaken several research efforts using household-level sample surveys to demonstrate what can be done to improve opportunities for women and how that will contribute to development. Country assessments have been published for Bangladesh, Kenya, and Pakistan, and one on India is forthcoming. This report on Peru is the first of the new wave of research to reach fruition. It does not answer all questions, of course, but sheds some light on what women's situation is, how it compares to that of men, what can be done to help, and what the results are likely to be. * This report refers to Intis at the official exchange rate of June, 1985 when one U.S. dollar equaled about 11 Intis.  ACKNOWLEDGEMENTS This discussion paper comprises several studies prepared by different authors based on the Peruvian Living Standard Survey (PLSS) data. Each paper was reviewed in the Bank through seminars. Barbara Herz directed the identification of policy issues, Shahid Khandker provided technical supervision and both the editors drafted the summary. The editors would like to thank the authors of each chapter. They also acknowledge their indebtedness to T. Paul Schultz, George Psacharopoulos and Karen Cavanaugh who made extensive and valuable suggestions, and to Jacques van der Gaag for giving advice and access to PLSS data. The editors take responsibility for any shortcomings that remain. The editors also thank Elinor Berg for editorial assistance and Andrew Danz and Benjamin Patterson for the overall production of this paper. The authors are grateful to the Ministry of Foreign Affairs, Norway, for support for the research carried out in this volume.  Contents Summary and Introduction Barbara K. Herz and Shahidur R. Khandker . . . . . . . . . . . . vii Chapter 1. The Extent and Impact of Women's Contribution in Peru, a Descriptive Analysis Marcia Schafgans .1... . . . . . . . . . . . . . . . . . . . . I Participation in the Labor Force . . . . . . . . . . . . . . . . 2 Socio-economic Factors and the Composition of the Labor Force . . 8 Contribution to Family Welfare . . . . . . . . . . . . . . . . . 16 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Chapter 2. Labor Market Participation, Returns to Education, and Male-Female Wage Differences in Peru Shahidur R. Khandker . . . . . . . . . . . . . . . . . . . . . . 37 Model Specification and Estimation Strategy . . . . . . . . . . . 38 Data Characteristics .... . . . . . ..... . . . . . . . . 47 The Results .... . . . . . . . ...... . . . . . . . . . . 49 Discussion .... . . . . . . . ...... . . . . . . . . . . 61 Appendices .... . . . . . . . ...... . . . . . . . . . . 64 Chapter 3. Modeling Economic Behavior in the Informal Urban Retail Sector in Peru J. Barry Smith and Morton Stelcner ... . . . . . . . . . . . . 67 Some Stylized Facts on the Informal Sector . . . . . . . . . . . 69 Model Formulation and Concepts ... . . . ..... . . . . . . 73 Description of the Data and Variables . . . . . . . . . . . . . . 79 The Empirical Model. . . . . . ... . . . . . . . . . . . . . . . 84 Empirical Findings and Interpretation . . . . . . . . . . . . . . 90 Policy Implications . . . . . . . . . . . . . . . . . . . . . . . 118 Appendices .... . . . . . . . ...... . . . . . . . . . . 120 Chapter 4. Household Production, Time Allocation, and Walfare in Peru John Dagsvik and Rolf Aaberge . . . . . . . . . . . . . . . . . . 129 Labor Market Activity, Income Formation and Welfare . . . . . . . 131 The Econometric Framework ... . . . . . ..... . . . . . . . 142 Summary Statistics and Parameter Estimates . . . . . . . . . . . 148 Policy Simulation Results for Lima . . . . . . . . . . . . . . . 152 Conclusion .... . . . . . . . . . . . ...... . . . . . . 159 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . 161  Chapter 5. Fertility Determinants in Peru, a Quantity-Quality Analysis Marcia Schafgans . . . . . . . . . . . . .. . . . . . 171 The Economics of Fertility . . . . . . . . . . . . .. . . . . . 172 Empirical Results .... . . . . . . . . . . . . . . . . . . . . 178 Simulation and Policy Implications . . . . . . . . . . . . 191 Conclusions . . . . .. . . . . . . . . . . . . . 192 Appendices . . . . . . . . . .... . . . . . . . . # . . . 195 Chapter 6. Gains in the Education of Peruvian Women, 1940 to 1980 Elizabeth 1M. King and Rosemary Bellew . . . . . . . . . . . . . . 205 Trends in Education .... . . . . . . .... . . . . . 205 A Household Model of Education with Gender Differences . . . . . 208 Empirical Results from the Adult Sample . . . . . . . . . . . . . 213 Results from the Youth Sample ... . . . . . . ...... . . . 219 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Appendices .... . . . . . . . . . . . . . . . . ...... . 229 Chapter 7. Does the Structure of Production Affect Demand for Schooling in Peru? Indermit Gill .... . . . . . . . . . . . . . . . . . . . . . . 233 The Theoretical Framework ... . . . . . . . . . . . . . . * . . 235 Some Theoretical Extensions ... * ... ..... . .... . 239 Household Level Empirical Evidence . . . . . . . . . . . . . . . 241 Province Level Empirical Evidence ... . . . . . . . . . . . . . 255 Conclusions and Policy Implications .. ..... . . . . . . . . 266 Appendices .... . . . . . . . . . . . . . . . . . . . . . . . 267 Bibliography ......... . . . . .. . . . . . . . . . . . 269  SUMMARY AND INTRODUCTION Barbara K. Herz and Shahidur R. Khandker This report examines ways of improving women's productivity and education and the consequences for development in Peru. It finds that women account for about 39 percent of family income in Peru. They carry the main responsibility for child care and heavily influence family decisions on children's education and family size. Improving opportunities for women can thus be a means to foster economic and social development as well as an end in itself. The main way to expand women's opportunities is through human capital investments, notably education beyond the primary level. This will increase women's earning capacity and broaden their labor force participation -- and thereby promote economic growth, family welfare, and slower population growth. The report's findings are based on econometric analysis of the household survey data from the Peruvian Living Standards Survey (PLSS) conducted in 1985-86. The PLSS is a national probability sample of 5,100 families and 26,000 individuals. Women's Contributions to Development About 45 percent of Peru's total wage and self-employed labor force are women. Women contribute to GNP and earn income in a variety of ways. Some 57 percent of Peruvian women engage in economic activities - about 10 percent work in the wage sector, 30 percent as farmers, and 15 percent in the informal sector. Women's economic participation varies by where they live. In urban areas, some 14 percent work in the wage sector, while 10 percent are farmers and 19 percent work in informal activities. By contrast, in rural areas, only 4 percent work for wages, and 10 percent are self-employed in informal activities, while 56 percent are farmers. Women's employment options and productivity in Peru as elsewhere depend heavily on education, but Peruvian women have about five years of schooling, on average, compared to seven years for men. While only 8 percent of men did not attend school, 25 percent of women never enrolled. The gender differences in educational attainment are more pronounced in rural areas than in urban areas and among the children of the poor. As women gain education and income, couples tend to opt for smaller families. Women's income and education matter in this regard more than men's because women tend to spend more time with children and, of course, bear the children. Educated mothers who can earn a substantial income prefer smaller families partly to make time for their income earning activities and for other interests. As the opportunity cost of the mother's time rises (as measured by the wage she could earn), her preference for a smaller family becomes stronger. Women with a secondary education have, on average, 3 children, while women with no education have 6 children. The relationship stands up when more sophisticated analysis takes account of other influences on family size. Parental education also influences children's schooling in different ways. Father's education usually matters more for boys than for girls, but. the mother's education increases girls' school enrollment as much as 40 percent more than the father's education.  viii Raising Women's Income and Productivity What are the most effective ways to increase women's income and productivity? The principal way, this research finds, is to educate women, especially at the secondary level. Education increases women's wage income at least as much as it increases men's. At the secondary level the returns to schooling for women are 15 percent compared to 9 percent for men. Primary education alone increases women's productivity in the informal activities. Education helps women break into the wage labor force and get better paying jobs afterwards in a variety of activities. Of course, women respond to offers of higher wages, but higher wages are justified and thus more likely to occur as women's productivity increases, and education is the main route to higher productivity. Yet human capital variables, notably education and experience, explain only one third of the male-female differences in wages in the labor market. Other constraints on productivity as yet to be identified or barriers to women's labor force participation presumably account for the remaining differences. Providing credit to the self-employed woman working in Peru's vast and growing informal sector has a high pay-off. Poor women who operate small retail businesses can earn an 18 percent rate of return on credit funds provided to finance working capital. Women entrepreneurs do respond to economic incentives but tend to earn less than men. They are more concentrated in labor- intensive fields, accounting for some 70 percent of the work force in retail and textile activities which are the most labor-intensive activities in Peru. Women entrepreneurs have only one fourth of male entrepreneurs' fixed capital investment. Female entrepreneurs are also less educated than male entrepreneurs: Female retailers, on average, have 6 years of schooling compared to 8 years of schooling for male retailers. These differential factor endowments account for much of the male-female differences in earnings in informal activities. In the country as a whole, women are more concentrated in services, while men predominate in industry. Of the employed female labor force, 72 percent of women are in the service sector, while only 15 percent are in industry. By contrast, 52 percent of the male labor force works in services and 24 percent in industry. Expanding services thus tends to increase women's employment opportunities and so raises the economic returns to female education more than expansion in industry, which tends to favor men. Whether this reflects different preferences, deficits in female education, or labor market barriers is not clear. The Female Education Paradox Even though female education has substantial economic returns and broader social benefits (for children's health and schooling and lower fertility), girls in Peru continue to receive less education than boys. In rural areas the "gender gap" in school enrollment is apparent even at the primary level, and nationally it is striking at the secondary level and beyond. The proportion of the school-aged children enrolled in secondary school is 9 percent higher for boys than for girls in Peru as a whole and in rural areas it is 17 percent. The data do not permit detailed analysis of the reasons for this gender gap. It is reasonable, however, to suppose that parents may be reluctant to  ix incur the costs of educating girls because the social benefits accrue so far in the future and do not affect the parents much. The girls themselves, their children, and the society in which they live will reap the main benefits. Parents may also believe girls are less likely than boys to succeed in finding a good job once educated, because of discrimination reflecting traditional expectations about women's proper occupations. Moreover, traditional culture may favor the education of boys, and parents may have more concern for girls' physical safety or social reputation, particularly as the girls grow older. While further research would help to diagnose the reasons for the gender gap in education, it is possible now to identify some policy measures that do help close the gap. The more educated parents are, the more willing they are to educate their daughters as well as their sons. Maternal education, as noted earlier, is especially important in this regard. The question then is how to achieve a generation of educated mothers. This report suggests four main policy measures. First, simply increasing the number of school places helps. Even though parents may prefer to educate sons before daughters, once most boys are in school, the additional places go to girls. Second, locating the schools closer to children's homes facilitates enrollment of all children but especially girls. Third, improving the quality of education is particularly important to increase girls' schooling. As quality improves (measured by the supply of text books or by the number of school teachers), parents become more willing to educate their girls, even more than their boys. In the 1960s Peru undertook a major initiative to expand the educational system, increasing school places as well as improving quality. The quality and quantity improvements helped increase girls enrollments and reduced the gender gap but were not enough to bridge the gender gap, especially in rural areas. Thus, additional steps such as scholarship for girls at the secondary level may be helpful. Since female education has more impact than male education on children's schooling and family size and these benefits exceed the private costs to parents, a special effort to educate girls is warranted on economic grounds. Finally, efforts to improve the earning capability of women may build parental willingness to educate their daughters, as parents see the economic returns to female education increase. In this regard, measures to assure women's equal access to the wage labor force may be helpful, as will measures to boost the earning capacity of the vast numbers of women employed in the informal sector. Promoting Female Entrepreneurship The importance of Peru's large and growing informal sector is well documented, and many Peruvians will doubtless continue to depend on it for their livelihood for some time to come. The informal sector can also serve as a stepping stone to more productive activities in the formal sector. Measures to strengthen the informal sector should therefore be undertaken in parallel with measures to open up women's opportunities in the wage labor market. Evidence reported in this report suggests that women's entrepreneurial activities are constrained by the scarcity of capital. While capital markets may not be designed deliberately to exclude women, that may be the practical result if women lack the collateral, financial experience, and education to cope with the formal credit system. This is particularly likely to be the case with poor women. Measures to target credit for the poor, male or female, may help them shift into more productive lines of work. Further research is needed on how to extend  x credit more effectively to the poor, but experience in several countries suggests that special measures may be needed to reach women, because of gender-specific barriers rooted in childbearing or tradition.' It bears emphasizing that such credit need not and probably should not be subsidized below normal commercial interest rates. The poor already pay exorbitant rates for whatever credit they can find in the informal sector, the record of subsidized credit programs is generally dismal, and the main need is for reliable and continued access to capital which can best be assured through credit programs that are financially sustainable. Easing other constraints to women entrepreneurial ability such as lack of skills and training is also worth considering. Policy makers may develop professional skills development policies for women. The Underlying Model and Research Methods The report's data analysis is based on econometric analysis of a household economic framework that suggests how to improve women's education and productivity and what the consequences of doing so will be. The contribution of women to development is assessed in the context of the family because women must balance the time they spend producing goods and services at home and in the marketplace with the demands of childbearing. An understanding of family economics is thus essential for analyzing household resource allocation by gender and may explain the gender gap in education and productivity. It may also help to show which government policies can encourage families to invest in women as well as men. Two competing family models may be used--the usual "unified" and the more recent "bargaining" family model, both in the tradition of neoclassical consumer theory (Schultz 1989). The "unified" model assumes that family members agree on objectives and pool their resources to maximize a common family welfare function. (Alternatively, one dominant family member may impose his or her own preferences). This model assumes that market goods do not yield utility directly but require time to be spent to yield utility (Becker 1965). Thus, quantities consumed of any good depend on the opportunity cost of an individual's time, which reflects his or her market wage. Market wages depend in turn on education and market opportunities. Although education can increase women's productivity at home and in the market, the increase is thought to be more pronounced in the marketplace. Higher market wages for women can raise the relative opportunity cost of producing children and encourage families to invest more in the child's human capital .2 Evidence also indicates that gains in women's education and income promote an intrahousehold resource allocation that is more equal between males and females (World Bank 1989). In contrast, the key message of bargaining models (McElroy and Horney 1981; Manser and Brown 1980) is that people within families do not agree on 'Women with young children may have more trouble leaving home to seek credit or training. 2Women's education is the most important determinant of family welfare, in terms of fertility reduction, children's health, school and occupational promotion (Schultz 1989).  xi objectives, and women may influence family decisionmaking more as they gain control over resources. These models, which assume that individual members of a family have conflicting preferences, make it possible to see how individuals use their resources to benefit the family (or the reverse) (Schultz 1989). In some cultures women control unearned income and hence affect family's resource allocation (Schultz 1989; Duncan 1989). In others women's wages influence intrahousehold resource allocation (Rosenzweig and Schultz 1982). In a sense, the unified model is just a special case of the bargaining model, where people agree or where one bargainer dominates. The unified household model has weaker data requirements. It can rely mostly on household-level information, while the bargaining family model requires more detailed individual-level information. The choice of model is therefore often driven by data availability. Collection of individual-level information is not an easy task and is seldom attempted. Even collecting data at the household level is difficult. The Peruvian Living Standards Survey (PLSS) provides data primarily on households. It supplies some data on individuals but not enough to permit using the "individualistic" family model. The report therefore relies on the unified family model to document the relationships among women's work, education, and family welfare in Peru, but it does permit some indirect inferences that hint at underlying bargaining. The data, which were collected by Peruvian Instituto Nacional de Estadistica and the World Bank in 1985-86, contain information on labor force participation and wages, income from various sources including agriculture, marital status, fertility, consumption, savings and credit. Because data on unearned income and assets were collected only for households, not for individuals, we were unable to examine the impact of individually owned assets on family decisions. Nor were we able to measure the extent and impact of gender bias in the allocation of resources for household consumption. While data on households often have information on individual labor supply and earnings, they do not cover individual consumption. Although attempts have been made to identify the extent of gender bias in household consumption by using aggregate family consumption data (Deaton 1988), this approach has serious limitations.3 The report uses the individual and household information to explain gender differences in time allocation, school investment, productivity in formal and informal activities, and their effects on household outcomes such as fertility and school enrollment of children. Unfortunately, data on family planning services and contraceptive use are also missing from this sample survey. Thus, no attempt could be made in the report to understand how the provision of family planning and female education may reinforce each other or how family planning 3The main criticism of adult-equivalent scales as proposed by Deaton (1988) and others is that one cannot separate the factors reflecting household technology ("needs") from those governing the intrahousehold distribution rule ("wants"). See Gronau (1989) for details.  xii alone influences fertility.' Similarly, because of a lack of data on agricultural extension and limitations on farm-level production and input use data, no attempt could be made to determine whether or how women's farm productivity is improved when women have better access to extension, credit and other public services. Outline of the Report's Chapters The report has seven chapters. A brief discussion of the contents and findings of each of the remaining chapters follows next. Schafgans (Chapter One) gives an overview of men's and women's labor force participation. She discusses possible associations between a woman's wage labor force participation and her observed fertility, education, marital status, spouse's economic activities, children's school enrollment and household characteristics. Schafgans also analyzes school enrollment by gender to identify male-female differences in school enrollment that may affect wage labor force participation and other outcomes. Schafgans shows that 57 percent of women are economically active. Women are mostly self-employed, while men are mostly employed in the wage sector. Women's predominance in the self-employed sector reflects their lower education and household responsibilities. Recognizing self-employed women as a legitimate part of the labor force and their needs for improved productivity would clarify the potential benefits to be gained by improving women's productivity. Schafgans finds that education increases women's wage labor force participation at the expense of self-employed work in agriculture. Educated women also have fewer and better educated children. Moreover, the households headed by women have, on average, lower incomes and per capita expenditure than the households headed by men. Note, however, that both types of households have equal per capita food consumption. Khandker (Chapter Two) analyzes wages and wage labor force participation to see what explains gender-related differences. In Peru as a whole women earn about half of men's wages. The wage labor market participation rate for women is 13 percent compared to 35 percent for men. Khandker estimates a wage equation to explain wages and, using the wage estimates, calculates the private returns to schooling for men and women. He also examines whether these returns influence school enrollment of boys and girls. He deals with two econometric issues that are often ignored in research on wage estimation. Sample selection bias may occur if one only considers wage workers, excluding those who are self-employed, because this procedure ignores reasons that may lead people to choose to stay out of the wage labor force. A better method is to estimate a wage equation along with an equation that predicts whether or not an individual will participate in the wage labor market at all (by using at least one other variable not in the wage function to identify the labor force equation). Khandker estimates the wage function with such a sample-selection correction and 4However, as information is available on community distance to family planning center for rural areas, an analysis is sought to explain the relative impact of education and the distance to family planning center on fertility behavior.  xiii compares the estimates of schooling returns to those obtained without a sample- selection correction. He also deals with the unobserved variable bias that may arise if unobserved characteristics correlated with years of education also influence wages. He uses a household fixed-effect model to correct the wage gap and school-return estimates for unobserved household characteristics. Khandker finds that the human capital model explains only about a third of male-female wage gap, although it substantially accounts for differences in men's and women's wage labor market participation. Thus, more education means women are more likely to work for wages. In estimating economic returns to education, correction for sample selection and unobserved household characteristics is found important for both women and men. This indicates that women as well as men who participate in the labor market are not representative of all women and men. Even after correcting the estimates for possible sample selection and unobserved variable bias, the economic returns to education are higher for women than for men in Peru. Another finding is that private schools are more effective than public schools in raising individual productivity. This implies that gender differences in productivity cannot be removed by educating girls in public schools unless government makes public schools more effective. Yet another finding of his chapter is that an extensive school system and well developed labor markets can help reduce the "gender gap" in school investment decisions of the parents. Smith and Stelcner (Chapter Three) analyze women's participation and productivity in retail trade, which accounts for 46 percent of informal activities in Peru. They measure productivity of labor as the marginal revenue product of a unit of labor. Since retail businesses can be identified as either male- or female-owned or jointly owned, the study compares the major constraints to raising the productivity of male and female retailers. Smith and Stelcner obtain reasonable quantitative assessments of the relative productivity of men and women and explain gender-related differences in retail productivity. Using an econometric model of retail trade, they show the relative contribution of labor and other inputs. Retail trade is also subject to a selection procedure. The buyers' selection depends on many factors including the unobservable "sales effort" of a retailer. Smith and Stelener develop a nonlinear revenue function to show the probability of a retailer's sales to potential buyers and the expected price per sale. The probability of sales depends on such characteristics as the retailer's education, experience, and capital as well as the factors that characterize the market forces, including the retailers' market outlets. Smith and Stelcner show that primary education improves retail productivity but endowments of capital, expenses and labor have even greater influence on productivity variations than education or ownership of the firm - female, male, or mixed. Firms with smaller endowments typically have the higher productivity. They find no major differences in economic behavior between enterprises of male-only, female-only, and mixed units. Differences in the resource endowments of the firm lead firms to behave differently. The female- only enterprises often have less capital than their male-only counterparts. Simulation results indicate that both men and women entrepreneurs in different income groups are equally productive if their education is raised to the same level. When credit is given to finance working capital needs, the returns to  xiv working capital are lower for low-income female enterprises than for male enterprises of similar income class. This finding seems counterintuitive--if women have less capital, they should get a higher return to capital. One possible interpretation of this puzzle is that women entrepreneurs from the bottom income group may lack complementary inputs such as skills and training to effectively utilize the loan. However, for an equal amount of credit, male and female entrepreneurs of higher income groups do not behave differently. Dagsvik and Aaberge (Chapter Four) analyze family behavior in household production, time allocation, and family welfare. They present the data analysis in three sections. First, they discuss the relationship between labor market participation and income inequality. They consider entrepreneurial income and wage earnings of household members (male, female and children) and the resulting economic contribution to family income. Second, they estimate an interdependent utility (that is, structural) model with household- and individual-specific information to explain household production, labor supply and consumption behavior. Third, they simulate the possible effects of an increase in wages and education on family welfare, measured by per capita income or expenditure. They also examine the possible reductions in income inequality because of changes in the education and wages of men and women. Dagsvik and Aaberge indicate that family income would rise significantly as a result of increases in women's wages and education. Thus, a 20 percent increase in women's wages from its mean of 5.3 Intis in Lima increases women's labor force participation by about 4 percent, their total wage earnings by 25 percent, and women's share of family cash income by 5 percent. A similar percent increase in men's wages from its mean level of 6.4 Intis raises men's share of family income by 14 percent. Compared to women with no education, women with a least nine years of schooling have 22 percent greater labor market participation. The corresponding increases in women's wage earnings and share of family income, respectively, are 43 percent and 8 percent. A similar increase in men's education raises men's share in family income by 11 percent. On balance, improving women's education has a larger effect than improving wages in increasing women's contribution to family income. Note, however, that as education is the key to improving individual productivity, improving women's education will help increase both their productivity as well as share in the family's cash income. Dagsvik and Aaberge's simulation study also indicates that incomes are so unequally distributed that raising wages and education of men and women will not do much to reduce income inequality. Peru needs other policy measures such as assets redistribution to reduce its severe income inequality. Schafgans (Chapter Five) examines the impact of women's and men's wages and education on the demand for a particular number of children as compared to the quality of children measured as schooling per child. She investigates the extent of a trade-off between the number of children and schooling per child that exists in Peru. The presence of such a trade-off is emphasized in the literature as an important factor in slowing population growth and promoting family welfare. Within the neoclassical household model framework, Schafgans estimates the reduced-form equations for the number of children and the schooling per child as functions of household income (alternatively, father's and mother's earned income), mother's age and education, other household characteristics, and community characteristics. The household income or father's income can have  xv either a positive or negative effect on the number of children but should have a positive effect on schooling per child. But the mother's wage measures the opportunity cost of her time. Since women spend more time with children, the mother's wage should have a negative effect on family size and a positive effect on child quality. Similarly, the mother's education should have a negative effect on the demand for children and a positive effect on child quality as the opportunity cost of women's time in the home rises and as education promotes women's wage labor force participation and builds interest in modern and more effective contraceptives. In implementing the model Schafgans uses both a linear and a discrete choice model to explain fertility behavior in Peru. She uses a linear and a household-fixed effect model to explain schooling decisions. Her results indicate that women's education increases children's schooling and reduces the quantity of children. When women earn higher wages, they demand fewer children and may demand more children's education. However, the father's wage has more influence than the mother's on children's education. In contrast, it is the mother's wage and education that influence family size more. Her study also shows that the distance to the family planning center in rural areas does not significantly affect fertility decisions, but the lack of further information on the nature and quality of family planning services prevents any solid inferences. Schafgans' analysis also clearly indicates that there is a gender preference (for sons) in parental investment in schooling. She did not pursue, however, whether father's or mother's preferences differ between bouts and girls in terms of schooling. King and Bellew (Chapter Six) discuss the determinants of individuals' school attainment and enrollment over time and the role of government policies in closing the gender gap. They ask: How rapidly has the expansion of public education changed schooling attainment of boys and girls? Have additional opportunities for education been equitably distributed between men and women? What other aspects of public education policies influence the gender gap in education? What factors beside government policies explain variations in the levels of education between men and women? King and Bellew use the neo-classical household model to explain variations in school attainment, incorporating both the direct and indirect costs of schooling as well as individual learning ability. Why do boys and girls attain different levels of schooling even if they have the same learning ability? The authors suggest two possible explanations: Parents may have different preferences for boys' and girls' schooling; and labor markets may reward the education of boys and girls differently. Beyond this, parental preferences can interact with market forces. For example, a rise in female wages and new work opportunities can increase the returns to women's education, which may then encourage parents to educate daughters' even if they feel a systematic bias against daughters. The speed and magnitude of the response depend on the availability of jobs for girls as well as the price and income elasticities of their demand for education. Community characteristics such as school availability and school quality can also influence parents' investment on education. King and Bellew find that both mother's and father's education heavily influence children's school enrollment. However, they find that the father's education influences boys' enrollment more than girls', while the  xvi mother's education influences girls' enrollment more than boys'. Also improving school quality proxied by the supply of text books, the number of grades, and the number of teachers increases girls' school enrollment more than boys'. Government policies aimed at increasing access to schooling have largely reduced the male-female gap at the primary school level but failed to do so at the secondary and postsecondary levels. King and Bellew's analysis indicates that the relative effect of parental education also differs in school attainment of daughters and sons. In the adult sample, for sons' education, father's education has twice as large an effect as the mother's education, while for daughters' education, mother's education has a larger influence. Similarly, in the youth sample, the mother's education has a stronger effect on the daughters' education attainment. "Quality variables" such as the supply of text books and the number of teachers are particularly important in persuading parents to educate their daughters. Perhaps parents' demand for daughters' education is more price elastic than their demand for boys' education. Gill (Chapter Seven) also helps explain why parents invest less in daughters' education than in sons'. He develops an intrahousehold resource allocation model linked with the market demand for educated workers to see whether household demand for schooling depends on the jobs available in the community. He assumes that parents consider their own expectation about future labor activities when making schooling decisions for their children. He hypothesizes that parents want more education for their children if education- intensive sectors predominate in the local economy. He takes the shares of regional GDP generated in two education-intensive sectors -- services and industry -- to represent the demand for educated workers. He relates schooling attainment to shares of services and industry and looks for differences for women and men. Gill finds that expansion in services or industry leads to an improvement in schooling levels of both boys and girls. However, expanding the service sector encourages more investment for girls, while industry encourages more for boys. Thus, a faster increase in women's human capital accumulation in comparison with men's can be associated with an increase in the relative share of the services in GDP. But expanding services may not be as productive for growth as it is for reducing male-female school gap. We are thus back to the classical dilemma: equity and growth may not be jointly maximized. Yet when human capital is a major constraint on a country's overall economic and social progress, reducing the male-female gap in human capital can have a large pay- off for development. Women are not only half the country's labor force but carry the main responsibility for children's care. In summary, the findings demonstrate important relationships among women's education (especially at the secondary level), work choices, productivity and income, family welfare, and family size. Education improves women's labor force participation and productivity. It also raises children's human capital investments (especially for girls) and reduces family size. By improving economic productivity, education thus helps women in Peru to increase their share of family income and hence their contribution to development.  CHAPTER 1 A COMPARISON OF MEN AND WOMEN IN THE LABOR FORCE IN PERU* Marcia Schafgans 1. Introduction This chapter uses household survey data to compare the proportion of men and women in the labor force in Peru and to examine the hours spent on the job and in the home. It also explores the associations among education, employment, marital status, number of children, and school enrollment to evaluate the contribution of women to family welfare. The Peru Living Standards Survey, on which this analysis is based, was developed by the World Bank, the National Institute of Statistics in Peru, and the Central Bank of Peru to measure the major aspects of economic well-being at the household and community level. Two questionnaires were used: the first provides information on income, consumption, employment, schooling, health, fertility, housing, migration, and savings. These data were complemented by a survey of rural communities that gathered information on prices, transportation, communication, and public services. The survey was conducted between June 1985 and July 1986. The sample of 5,120 households (26,000 individuals) interviewed reflects the distribution of the population living in urban and rural areas (towns with fewer than 2,000 inhabitants) and in natural regions (for a detailed description see Grootaert and Arriagada 1986). All monetary values are in June 1985 prices, at an exchange rate of about 10 intis to the U.S. dollar. For this study the population is grouped into three categories: metropolitan Lima, other urban areas, and rural areas. The number of women in the work force in Latin America has been rising faster than the number of men. From 1970-80, the average annual rate of growth was 2.5 percent for men and 5.1 percent for women. In Peru women accounted for about 45 percent of the labor force in 1985-86, up from 24 percent in 1980 (IDB 1987). Most of these women are self-employed. Except in Lima, most men are also self-employed, although women make up the larger proportion of workers in this sector, a result of lower education as well as household responsibilities. The type of activities women pursue differs considerably from the type that men select. Women are poorly represented in higher-paid jobs, and their wages in the formal sector are generally lower than the wages of men. As this analysis is purely descriptive, it gives only an indication of the possible reasons behind the differences in the contributions of men and women. Further * Marcia Schafgans is a graduate student at Yale University and is a consultant to the Women in Development Division of the World Bank's Population and Human Resources Department. The author is grateful to Jorge Castillo-Trentin, Barbara Herz, Shahid Khandker, Beth King, Jacques van der Gaag, and Morton Stelcner for helpful suggestions, Ben Patterson for assistance with the graphs, and Eleanor Berg for her editorial assistance.  research is required to shed light on some of the issues highlighted here. For instance, how is an individual's work status affected by socioeconomic factors? What explains the participation of the individual in a particular sector? How can we explain differences in wages or incomes in formal and informal activities? We need to know more about the endowment in human capital, and the returns to education. Do parents make a trade-off between the number and quality of children? How do improvements in productivity affect the distribution of household income? These are important issues which are considered in the chapters of this report. Men have higher rates of participation in the work force than women, and spend more hours on the job. Women devote more time to the household, and their jobs show a high level of compatibility with these household responsibilities, particularly in rural areas. Households headed by men are better off than those headed by women. But improvements in the economic status of women, for example shifts from agricultural work to the wage sector (or even to nonagricultural self-employment), reduces the difference in the per capita consumption expenditures of male- and female-headed households. Furthermore, this improved economic status has a positive effect on the children's welfare by increasing school enrollment, especially in rural areas. In general girls benefit more than boys from an improvement in the mother's education or employment. 1.1 Participation in the Labor Force Buvinic and others (1983) point out that the division of labor in third world countries typically "assigns women to labor-intensive production, and the division of labor within the market restricts women to work characterized by low technology, inefficient production and marginal wages." Peru is no exception. Women make up about 45 percent of Peru's labor force.' Fifty-seven percent of women versus 71 percent of men are working.2 The highest participation rates are in rural areas: 71 percent of all women and 80 percent of all men. Women in Lima record the lowest participation rates, with 46 percent in the labor force. The lowest rate for men (62 percent) is in other urban areas. The smallest difference is in rural areas; the largest difference is in Lima. Most working women are self-employed. Employed men generally are also self-employed, while in Lima most work for wages. But the proportion of women who are self-employed is larger than that of men. In Lima, rural regions, and other urban areas, 55, 94, and 78 percent respectively of working women are I A description of the data and employment definitions used in this analysis is given in appendix 1. 2 Table Al shows the distribution of the labor force calculated by dividing the number of employees in a given economic activity by the potential labor force. Figure 1 shows the employment figures in a stacked-bar graph, in which the labor force participation rates are indicated by the total length of each bar.  Figure 1: Labor force status of men and women, by region 100% 80% - 60% I-Ue~os ........ Self - .riow 40% a00ulctA ...........PIivate secto H ~~~~M Pubic sector 20% 0% Men Women Men Women Men Women Lima Other urban areas Rural Note: Main occupation last 7 days. self-employed, while men in these regions record rates of 35, 80, and 54 percent respectively. In the wage sector, proportionately more women than men work in the public sector (see table Al). In Lima and other urban areas most workers are paid; in rural areas most of the work is unpaid. In all regions, a higher proportion of men than women (65 percent compared to 41 percent) report receiving wages (both salaried workers and self-employed individuals).3 Stelcner (1988) discusses the nonagricultural occupations of self- employed workers. He shows that the types of activities women pursue differ considerably from those of men (see table A2). In retail food and textiles, women account for about three-fourths of the workers; in retail nonfood and food processing, 60 to 70 percent; in personal services, about half, and in the remaining sectors (construction, transportation, primary industries (fishing, hunting, forestry, and mining), and other manufacturing) women account for only a small proportion of the workers. The same differences are also apparent in the wage sector, indicating that women are poorly represented in the higher- paid jobs. Figure 2 and table A3 show the labor force participation rates of men and women by age cohort. Participation rates for all age groups are higher 3 A clear definition of paid self-employment, however, is not available.  - 4 - Figure 2: Labor force participation of men and women, by age cohort Lima 100% 80% 60% 40% 20% 0% 6-14 16-19 20-29 30-39 40-49 60-69 60- Other urban areas 100% 80% 60% 40% 20% 0% 8-14 16-19 20-29 30-39 40-49 60-59 80. Rural 100% 60% 60% 40% 20% 0% 6-14 16-19 20-29 30-39 40-49 50-69 60. Men M Women  in rural than in urban areas. There are more men than women in the labor force, although there is little difference in the rates of participation in rural areas. Participation in the economy declines with age, although the drop in the participation of women more than 40 years old is more pronounced in Lima. 1.1.1 Hours Worked in Main and Secondary Jobs. Table 1 shows that men work more hours in their main job than women. In all regions men and women in the wage sector work longer hours (43 hours and 38 hours respectively) than their self- employed counterparts (38 hours and 28 hours); and women in the private sector work significantly more hours than women in the public sector. Table 1: Average hours worked in the main job Males Females Lima Other urban Rural Lima Other urban Rural Sector areas areas Labor force 50.0 38.7 37.5 28.0 28.7 29.8 (0.5) (0.5) (0.3) (0.6) (0.6) (0.3) Wage workers 43.9 42.9 41.9 36.0 36.3 40.6 (0.6) (0.7) (0.7) (0.8) (1.0) (1.4) Public sector 42.1 41.1 37.4 29.9 30.9 29.6 (1.0) (0.9) (2.0) (1.1) (1.3) (4.4) Private sector 44.7 44.0 42.5 39.1 40.5 41.9 (0.7) (0.9) (0.8) (1.0) (1.5) (1.4) Self-employed 41.2 38.6 37.0 26.0 28.1 29.3 (1.0) (0.8) (0.4) (0.9) (0.7) (0.3) Agriculture 18.0 27.3 36.7 7.9 14.7 29.3 (2.9) (1.3) (0.4) (0.4) (0.6) (0.3) Nonagriculture 43.8 42.6 40.3 33.7 36.0 29.5 (1.0) (0.9) (1.4) (1.1) (0.9) (0.9) Note: Standard errors in parentheses. Table A4 shows average hours worked in the main job by age cohort. In Lima and other urban areas, 6- to 14-year-olds work an average 17 hours a week; in rural areas they work an average 25 hours a week. The number of hours worked initially increases with age, although the rise is less pronounced for women. The reduction in the number of hours worked occurs in urban areas, primarily in the wage and self-employed nonagricultural sectors.  - 6 - Table 2 shows the distribution of hours worked in second jobs. Thirteen percent of women and 12 percent of men have second jobs. This "moonlighting" is more common in rural areas, where 18 percent of women and 15 percent of men have a second job, than in Lima, where only 7 percent of women and 9 percent of men have additional jobs. Employed men spend an average 13 hours a week at their second jobs, significantly higher than women (except in rural areas). Women in Lima and other urban areas spend nine hours a week on average in a second job. Table 2: Average hours worked at second jobs Males Females Lima Other urban Rural Lima Other urban Rural Sector areas areas Employed 12.4 12.1 13.4 8.1 9.5 13.6 (0.6) (0.6) (0.3) (0.4) (0.4) (0.3) Wage workers 15.0 14.4 15.5 11.5 8.7 20.1 (1.3) (1.6) (0.7) (1.6) (1.4) (2.3) Self-employed 11.6 11.7 12.7 7.6 9.5 13.3 (0.7) (0.6) (0.4) (0.4) (0.4) (0.3) Agriculture 6.7 10.9 13.2 6.2 8.2 12.8 (0.9) (0.8) (0.5) (0.3) (0.3) (0.6) Nonagriculture 13.3 12.2 12.1 9.5 11.4 13.6 (0.9) (0.8) (0.5) (0.9) (0.8) (0.4) Note: Standard errors in parentheses. 1.1.2 Hours Worked in the Household Table 3 shows the average number of hours worked at home by men and women in the labor force. There were no significant differences across regions in hours worked at home (unconnected with business). Women aged 20 to 59 devote most of their time to the family regardless of their status in the labor force. They spend an average 30 hours a week working at home, or four times more than men in that age-group. Self-employed women spend more hours in the household than women working for wages. And women who work on family-owned farms spend more hours on household activities than women in nonagricultural activities. Self-employed women spend five times more hours than self-employed men on household chores. Young women also spend considerable time working in the home. Girls 6 to 14 years old work an average 12 hours a week (compared to 8 hours a week for boys); girls 15 to 20 years old spend an average of 21 hours a week (compared to 7 hours a week for boys).  - 7 - Table 3: Average hours worked at home by gender Age cohort Women Men Sector 6-14 15-19 20-59 60+ 6-14 15-19 20-59 60+ Labor force 12.3 20.7 30.5 26.4 8.2 7.0 6.2 7.8 (0.30) (0.59) (0.31) (0.79) (0.22) (0.28) (0.12) (0.37) Wage workers 7.5 10.5 21.1 25.3 7.1 4.9 5.5 5.3 (1.49) (1.02) (0.58) (3.69) (0.77) (0.36) (0.17) (0.66) Self-employed 12.5 22.4 33.1 26.5 8.3 7.7 6.6 8.4 (0.31) (0.64) (0.35) (0.81) (0.22) (0.37) (0.17) (0.42) AgricuLture 12.7 23.3 35.1 28.0 8.2 7.7 6.9 8.7 (0.33) (0.78) (0.45) (0.94) (0.24) (0.43) (0.22) (0.51) Nonagriculture 11.2 20.7 30.3 23.1 8.8 7.8 6.2 7.7 (0.81) (1.09) (0.54) (1.53) (0.71) (0.74) (0.27) (0.72) Note: Standard errors in parentheses. 1.1.3 Distribution of Earnings in the Formal Sector Table 4 shows average hourly wages of men and women in the formal sector. Services offer high wages for men and women; agricultural work is the most poorly paid. Men receive higher wages than women for all activities except manufacturing. Even holding constant for educational attainment, men receive higher wages than women. This indicates that women are working in lower salaried jobs than men, since differences in wages for similar jobs tend to diminish with increased education. Table 4: Average hourly earnings in the formal sector, by educational attainment (in intis June 1985) ALL Primary Secondary Postsecondary education education education Formal sector Men Women Men Women Men Women Men Women Observations 2,431 947 806 220 1,029 422 504 256 ALL 6.18 5.09 3.74 1.80 5.74 4.83 11.19 8.26 (0.19) (0.21) (0.15) (0.16) (0.29) (0.25) (0.54) (0.56) Commerce 5.59 3.54 3.19 2.66 4.93 3.48 9.61 4.75 (0.43) (0.24) (0.40) (0.24) (0.41) (0.29) (1.41) (0.83) Services 7.34 5.94 4.10 2.66 7.31 5.01 8.60 9.20 (0.44) (0.33) (0.23) (0.24) (0.91) (0.33) (0.37) (0.74) Manufacturing: Nontextiles 5.81 5.54 4.58 3.79 4.48 6.42 12.67 n.a. (0.37) (0.95) (0.45) (1.12) (0.30) (1.34) (1.62) Textiles 4.53 3.39 3.19 1.34 4.29 3.26 n.a. n.a. (0.63) (0.40) (0.65) (0.19) (0.47) (0.43) AgricuLture 2.57 1.75 2.24 2.81 3.62 2.14 n.a. n.a. (0.18) (0.10) (0.11) (0.36) (0.75) (0.14) Note: Standard errors in parentheses. n.a.= not availabLe or too few observations  - 8 - 2. Socioeconomic Factors and the Composition of the Labor Force A number of socioeconomic factors may influence the composition of the labor force including education, marital status, economic contribution, and number of children. 2.1 Education Table 5 shows the rates of participation in the labor force for men and women aged 20 to 59, by the level of education.4 The share of employed men and women in the wage sector increases with education at the expense of the self- employed workers in agriculture. Education does not appear to have a linear or even monotonic effect on the participation of women in the labor force. This finding appears to be in line with other studies on labor in Latin America (King 1989).5 Table 5: Labor force participation by education Peru Lima Other urban areas Rural areas Men Women Men Women Men Women Men Women Labor force participation rates Education: None 96.6 83.7 n.a. 66.7 94.1 75.3 97.2 86.4 Primary 97.2 76.2 96.3 64.3 97.3 70.3 97.3 85.8 Secondary 95.4 62.9 95.9 62.1 93.8 61.6 96.8 69.1 Postsecondary 97.9 76.4 99.1 75.7 96.2 77.3 100.0 74.1 Employment status:" Wage workers Education: None 26.5 4.6 n.a. 10.2 46.7 5.0 22.9 4.2 Primary 33.4 10.8 54.5 25.6 47.8 9.1 23.3 6.5 Secondary 57.5 40.2 72.0 55.3 55.4 33.2 34.4 11.8 Postsecondary 64.9 71.8 69.0 71.0 62.8 74.4 56.4 60.0 Employment status: Self-employed in agriculture Education: None 66.1 75.3 n.a. 32.7 16.7 42.0 73.8 82.4 Primary 46.9 52.9 3.5 23.1 12.3 29.4 69.6 75.5 Secondary 14.4 20.7 0.5 12.2 5.1 17.5 52.3 53.7 Postsecondary 6.6 7.1 0.0 8.0 5.0 5.0 36.4 15.0 ' As percentage of total employment n.a. = not available or insufficient observations Figure 3 and table AS show the average years of schooling of men and women by status in the labor force. In all regions the most educated men and women work in the formal sector; the least educated are self-employed farmers. 4 The figures for men in Lima with no education have been omitted because there were too few observations. 5 "In Santiago, Chile, Castafieda (1986) found that ... women with no schooling and those with more than thirteen years of education tend to have larger probabilities of participating in the labor market than women with primary or secondary education."  - 9 - Figure 3: Years of school for men and women by work status and age cohort Lima Average yeara 14 12- 10 8 4 Men Women 2 - 15- 20- 30- 40- 0- 60. 16- 20-0-3 40- 60- 6 19 29 $9 49 69 19 29 39 49 60 Age cohort Other urban areas Av rage year. 14 12- 10 8 --- - Wage worker 8 \ -- - - - Self-enrployed, agriculture 4 \ - - Self-enpbyed, nonagr4Wbture 2 Men Women -= 16- 20- 30- 40- 60- 60- 15- 20- 30- 40- 50- 80- 10 20 39 49 69 19 29 39 49 69 Age cohort Rural Average years 14 12 - Men 10 - Women 4- 2- o 16- 20- 30- 40- 60- 60. 16- 20- 30- 40- 50-69 30. 19 29 39 49 69 10 29 39 49 Age cohort  - 10 - Self-employed women are usually less educated than men; women in the formal sector are usually better educated. Older workers tend to have less education than younger ones, indicating an increased investment in education (King 1989). 2.2 Marital Status Figure 4 and table A6 show the labor force participation rates of women aged 20 to 59 by marital status. Sixty-seven percent of the women in Peru are married (or cohabiting), 14 percent are separated, and 19 percent are single. Marital status has little effect on women' s participation in the work force in rural areas, where most of the women are working. Separated women have the highest rates of participation because of their increased economic responsibility. In Lima fewer married women are working (about 61 percent), compared to 66 percent in other urban areas, and 84 percent in rural areas, an indication that the traditional male-female division of labor is sharper in Lima than elsewhere.6 Even so, the data show the importance of the dual role of women in the economy, a finding consistent with other societies (Buvinic 1983). Figure 4: Women in the labor force, by marital status 100% 80% Ma 7rl 1 Elps 9igl barwnpky-SngMM,redSpa in m UnlfempIcy.d 60% rae rated rat dnw CB~ Seff-enVoyd, 40% M Wage woraer 20% .. 0%marrled Sops- Single married sepa- single married Seap- Single rated rate d rated Lima Other urban areas Rural 6 In Lima 99 percent of married men participate in the labor force, in other urban areas 97 percent and in rural areas 98 percent.  - 11 - Marriage tends to move women out of the wage sector. In Lima 75 percent of single employed women work for wages compared to only 37 percent of married women. In other urban areas the figures are 56 and 19 percent respectively, and in rural areas 13 and 5 percent. Divorced or widowed women tend to move back to the wage sector. Forty-seven percent of separated women in Lima who are employed work for wages, higher than the 29 percent in other urban areas, and the 14 percent recorded in rural areas. More self-employed separated women than married women work in the nonagricultural sector. 2.3 Economic Activities of Wives and Husbands Figure 5 and table A7 compare the labor market decisions of wives and husbands.7 In Lima and in other urban areas, women are least likely to be in the labor force if their husbands are wage workers. In rural areas the wives of men in the nonagricultural activities record the lowest rates. Women whose husbands work in the wage sector are proportionally better represented in the wage sector than other women (43 percent in Lima, 35 percent in other urban Figure 5: Labor force status of married women, by husband's labor force status 100% 80% i status: wife 80% Lima Other urban areas Rural~~=1uranioy 7 efn ienVloyedi 40% 20% mWg re 0% ML W.G. .1-. .-.. W.G. 6.1.. .1*~W~6 .11*G8.1* Lima Other urban areas Rural Labor force status: husband 7" Men in agricultural self-employment activities in Lima are not discuss,ed since there are so few in this particular group.  - 12 - areas, and 12 percent in rural areas). There is a high correlation between wives and husbands working on family farms or in self-employed nonagricultural activities. This correlation parallels a high correlation in education. Sixty- six percent of women with a primary school education are married to men with the same education. The figures for women with secondary education are 54 percent, and for a postsecondary education, 68 percent (see table A8). 2.4 Child Care As shown in table A9, women in Peru (15 to 49 years old) have an average of 4.26 children (3.46 in urban areas and 5.24 in rural areas). Among women aged 45 to 49 only 1.5 percent are childless. The number of children born is inversely proportional to the mother's education. For instance, women in urban areas aged 35 to 39 with no education have on average 6 children, as shown in table A10. The number of children for women in the same age group with a primary education drops to 4.4, to 3.3 with a secondary education, and to 2.4 for postsecondary. Women in rural areas with no schooling bear 6.9 children, and the figure falls to 6.3 with a primary education, 3.8 with a secondary diploma, and three with a postsecondary education. One manifestation of this burden is that women with young children are less likely to go to work. But unlike women in industrial societies where there is a strong trade-off between market work and child care, poor women in developing countries tend to sacrifice leisure time instead, since they are still responsible for the household (Buvinic 1983). Moreover, most women choose an occupation that is compatible with child care and household responsibilities. These are often low-paid informal activities. Because childbearing interrupts the career of women in the formal sector, more of them work at jobs that require lower qualifications and pay lower wages (Gronau 1988). Using the number of children under age six as an indication of the amount of child care required, figure 6 and table All show the status of women -o7ith and without dependent children. As this number increases women tend to drop c:ut of the labor force, which implies that some trade-off exists between market w ork and child care, although the trade-off is lower in rural than in urban a reas. Women in Lima without children or with only one child tend to work iiI the wage sector, while women with two or more children are likely to be self- exiployed, again, because of the compatibility between such work and care of children. The trade-off in the number of hours women work is shown in table 6. 2.'5 School Enrollment In many developing countries girls over 10 years old are often taken out of school because of financial restrictions, bad job prospects, or household responsibilities. This assumption is confirmed by Table 7, which shows that aanr,ollment rates are about the same for boys and girls for the younger group, uvitlh 92 percent of the boys and 89 percent of the girls enrolled in school. The diflference increases, however, for the older age cohort, with 72 percent of the bc)yn and only 64 percent of the girls enrolled in school.  - 13 - Figure 6: Labor force status of women, by number of children under 6 years 100% 80% 80.4 ! | J - [:E Unenmloyed 80% M F Self-employed. nonagriouture IM: ~ ~ Se-errTployed, 40% agricultu or more or more or more Lima Other urban areas Rural There are also important differences in enrollment by region. A ttendance in Lima and other urban areas is higher than in rural areas, particularly for the older age cohorts, indicating that access to secondary .. . . . ... Table 6: Average female weekly hours of work by number of dependent children Lime Other urban Rural Number of Children areas None 34.9 34.1 33.6 (0.6) (0.8) (0.4) One 33.4 31.2 32.2 (2.6) (1.4) (0.6) Two 25.6 29.3 30.4 (4.0) (4.4) (0.8) Three or more 26.1 27.4 28.1 (32.5) (16.8) (1.7) Note: Standard deviation in parentheses  - 14 - Table 7: School enrollment Peru Lima Other Urban Rural Age Areas 6 to 14 years Boys 92.4 98.4 97.3 86.8 Girls 89.1 98.3 96.4 80.5 15 to 19 years Boys 71.5 83.9 83.3 55.6 Girls 63.6 84.1 78.3 38.6 education is more equitable in urban areas.8 And the gap between the enrollment of boys and girls is lowest in Lima and highest in rural areas. The gap is more profound for 15-to 19-year-olds, pointing to the fact that girls drop out of school earlier than boys. As expected, school enrollment reduces the rates of participation in the work force (see figures 7, 8, and table A12), for all groups except boys 6 to 14 years old in rural areas.9 In these regions 54 percent of the boys who attend school are employed (mostly on family farms), compared to 48 percent of nonenrolled boys. As they move out of school, rural boys start working forwages, although girls usually remain self-employed. In Lima, where there is a wider job market, students 15 to 19 years old work for wages. 3. Contribution to Family Welfare Table 8 shows that per capita household expenditures are significantly higher in households headed by men (see Rosenhouse (1989)) than in those headed by women.'