University of Southern California Dissertations and Theses

Demand -side obstacles to girls' education in developing countries

(USC Thesis Other)

Demand -side obstacles to girls' education in developing countries

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DEMAND-SIDE OBSTACLES TO GIRLS’ EDUCATION IN DEVELOPING COUNTRIES by Jane Arnold Lincove A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY PUBLIC ADM1NSTRAT10N August 2005 Copyright 2005 Jane Arnold Lincove Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3196845 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3196845 Copyright 2006 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A cknow ledgem ents This dissertation is dedicated to my husband Evan for endless hours o f listening, reading, critiquing, and motivating. Special thanks to the faculty o f the School o f Policy, Planning, and Development at USC — especially the members o f my dissertation committee, Elizabeth Graddy and Delores Conway, and chair Gary Painter. Special thanks to my family and friends, including the members o f SPPD’s “Dissertation Club” — Jun Delfin and Kyu-Nahm Jun. Also thanks to those who attended conference presentations o f individual chapters at 2004 APPAM, 2005 SPSA, and 2005 ASPA conferences. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T ab le o f C ontents Acknowledgem ents ii List of Tables iv List o f Figures v Abstract vi Introduction 1 Chapter 1 Increasing the D em and for Girls’ Education: An Exam ination o f Policy and Context Variables in D eveloping Countries Summary 5 Text 6 Notes 42 Works Cited 44 Chapter 2 Growth, Girls’ Education, and Fem ale Labor: A Longitudinal Analysis Summary 49 Text 50 Notes 89 Works Cited 90 Chapter 3 Equity and Governance in Girls’ Education Summary 94 Text 95 Notes 132 Works Cited 133 Bibliography 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV L ist o f T ables Tables in Chapter 1 Table_______________________ Title______________________ Page 1 Policy and Context Variables 16 2 Countries in Regression Analysis 21 3 Mean Values for Dependent and Independent 23 Variables 4 Regression Results for Female Secondary 27 Enrollment 5 Regression Results for Total Fertility Rate 31 6 Regression Results for Child Mortality Rate 32 7 Correlations 34 8 Structural Equation Model Regression Results 35 9 Decomposition o f Effects for Total Fertility 36 Rate 10 Decomposition o f Effects for Child Mortality 37 Rate Tables in Chapter 2 Table Title Page 11 Countries in Longitudinal Regression Analysis by 67 Region 12 Variable Means by Income 68 13 Variable Means by Girls’ Secondary Enrollment 69 14 Variable Means by Majority Religion 70 15 Countries in Cross-Sectional Analysis by Region 79 16 Cross-Sectional Regression Results 80 17 Cross-Sectional and Longitudinal Regression 82 Results, Reduced Form Model 18 Cross-Sectional and Longitudinal Regression 83 Results, Full Model Tables in Chapter 3 Table Title Page 19 World Bank Education Projects 2002-2004 109-110 20 Gender Equity Measures in Countries with 113 Education Projects 21 Governance Strategies in Education 114 22 Governance Strategies in Countries with Low 121 Gender Equity Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V L ist o f F igures Figures in Chapter 1 Figure_______________________T itle______________________ Page 1 Theoretical Model 8 2 Scatter Plots o f Policy Variables & Female 24 Education 3 Scatter Plots o f Context Variables & Female 25 Education Figure Figures in Chapter 2 T itle Page N o. 4 Female Economic Participation and National Wealth by Country 51 5 Female Economic Participation and Female Secondary Enrollment 52 6 Economic Participation and National Wealth Over Time 75 7 Female Economic Participation and Female Secondary Enrollment Over Time 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. v i A bstract Girls’ education has attracted the attention o f researchers and practitioners in international development thanks to a growing body o f evidence suggesting that female literacy facilitates development. Efforts to expand the supply o f schooling for girls have resulted in increases in girls’ enrollment, but gender gaps in educational access persist in many developing countries. This dissertation addresses demand-side obstacles to girls’ education. The demand for girls’ education is typically lower than boys’ due to family poverty, girls’ domestic and wage labor, and labor markets that do not reward female education. In addition, social stigmas against female participation can create additional costs o f sending daughters to school. Demand-side obstacles mean that even as supply increases, girls may not access education, despite significant social benefits. Chapter 1 examines policy and context variables that influence the demand for girls’ education. Four theoretical influences on demand — wealth, opportunity costs, rates o f return, and preferences — are empirically tested using international data from developing countries. The results o f structural equation modeling procedures suggest that economic growth accompanied by policies that promote girls’ education will improve development outcomes. Chapter 2 examines the relationship between female labor force participation and female schooling over time. Longitudinal analysis o f international data using fixed-effects models suggests that economic growth can induce women to leave the labor market, but education can increase female labor force participation over time. Chapter 3 examines the implementation o f girls’ education Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. projects by the World Bank focusing on the dual goals o f increasing girls’ access while reforming governance to improve cost-effectiveness and transparency. Governance strategies o f privatization, decentralization, and citizen participation are evaluated in the context o f girls’ education and gender equity. The results o f these studies reinforce the importance o f girls’ education as a development goal and highlight the importance o f governance strategies that protect equity while improving efficiency. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 IN T R O D U C T IO N The World Bank’s website has this to say about girls’ education as a development strategy: It has recognized the striking body o f empirical evidence that demonstrates strong benefits o f girls’ education which span across a wide range o f areas including maternal and child health, social stability, environmental benefits and economic growth. Girls’ education and the promotion o f gender equality in education are critical to development, and policies and actions that do not address gender disparities miss critical development opportunities. (World Bank, 2001) Unicef reports that: In country after country, educating girls yields spectacular social benefits for the current generation and those to come. An educated girl tends to marry later and have fewer children. The children she does have will be more likely to survive; they will be better nourished and better educated. She will be more productive at home and better paid in the workplace. She will be better able to protect herself against H IV /A ID S and to assume a more active role in social, economic and political decision-making throughout her life. (UNICEF, 2005) With this level o f support from the development community, girls’ education is at the top o f the international agenda for human and economic development. Efforts to expand girls’ education have resulted in impressive increases in girls’ access to schooling. However, significant obstacles to girls’ education still exist on the demand side. Girls in developing countries have high opportunity costs o f attending school due Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 to economic the need for school-age daughters to provide wage labor, domestic labor, or child care. In addition, in societies where women have low labor market returns to education, parents have few incentives to invest in daughters. Added to these economic issues are potential cultural obstacles to educating girls’ in countries where patriarchal traditions persist. The result is that many countries continue to see significant gaps in male and female education despite supply-side policies. This dissertation addresses demand-side obstacles to girls’ education from three different perspectives. Chapter 1 seeks to identify the relative importance o f four theoretical influences on demand for education: wealth, opportunity costs, rates o f return, and parental preferences. Surprisingly, the results indicate that wealth, opportunity costs, and short-time rates o f return influence girls’ access to schooling, however long-term prospects for labor market participation do not. The results o f structural equation modeling suggest that economic growth will lead to greater social benefits if growth is accompanied by policies that provide public goods that increase the demand for girls’ schooling. Specifically, policies that reduce the burden o f domestic labor and improve access to health care can boost development outcomes through girls’ education. Chapter 2 focuses specifically on the unlikely finding that female labor market prospects are unrelated to girls’ education. Previous researchers observed a U-shaped female labor force participation curve where women leave the labor market at early stages o f development and female education and return when a white collar sector develops to reward female literacy. Theoretically, this means that countries undergoing Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 growth and expansion o f female education may see initial declines in the size o f the female labor market. This study uses longitudinal data to determine if countries can be expected to travel this development path over time. Results o f fixed-effect regressions suggest that there is an initial negative effect o f economic growth on female labor force participation. However, this negative effect is countered by the positive effect o f girls’ education. Therefore, girls’ education is needed to maintain and improve the female labor supply throughout development. Given the positive evidence from Chapter 1 and 2, Chapter 3 approaches the question o f implementation o f girls’ education programs by the World Bank, focusing on achieving a balance between equity goals and efficiency. The World Bank and other organizations worry that inefficient institutions o f governance will undermine the effectiveness o f development aid and impose institutional reforms as a component o f lending. World Bank projects musts address the dual goals o f equity and effective governance. Chapter 3 looks at implementation o f governance strategies in girls’ education programs at the World Bank, first through theoretical analysis o f governance and equity in the education sector, and then through empirical analysis of the content o f education projects. The results suggest that the use o f governance strategies varies with awareness o f equity issues. Specifically, privatization and decentralization to increase supply can also increase gender gaps. This effect can be mediated by creating financial incentives for local communities to increase girls’ education. Overall the results o f these three studies suggest that while girls’ education is improving throughout the world, there is still significant work to do to fully understand Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 and address the demand side o f the equation. While girls’ continue to face high opportunity costs o f education and low labor market prospects, aid organization such as the World Bank are increasingly concerned with efficiency rather than equity. Since girls’ education can accelerate other development goals this is one area w here equity and efficiency are intrinsically linked and must be balanced. The key for future research is to identify ways to promote program effectiveness and efficiency in use o f resources without undermining equity goals and recognizing that progress may be slow and incremental. It is my hope that the studies in this dissertation will contribute to this knowledge. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 Chapter 1 In creasin g the D em an d for G irls’ E ducation: An E xam ination o f P olicy and C ontext V ariables in D ev elo p in g C ountries Sum m ary This study uses international data to test an integrated model o f the demand for girls’ education and its effect on development outcomes. The framework integrates four theoretical approaches to the demand for education and isolates policy levers from the broader economic, political, and social context. The results suggest that economic growth increases girls’ access to education. However, government and aid agencies can further increase the demand for education through policies that reduce opportunity costs of girls’ domestic work and increase returns to education by improving girls’ health. Contrary to human capital theory, the results suggest that is it not necessary to improve labor market access for women in order to increase school enrollment for girls. Structual equation results reveal that much o f the relationship between economic growth and development outcomes, such as fertility and child mortality, is explained by girls’ education and the provision o f public goods such as clean water and health care. This study confirms that increasing girls’ education requires attention to demand, as well as supply. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 1 INTRO DUCTIO N Girls’ education has become a popular development strategy, thanks to a growing body o f theory and evidence linking female schooling and literacy with broader development goals including reduced fertility and improved child health. Increasing girls’ education is one the United Nations’ Millennium Development Goals and is considered to be a pivotal strategy to jump-start economic growth (United Nations, 2000). The World Bank proposes girls’ education as a policy goal that is humanitarian, democratic, and cost effective and is therefore both politically and economically attractive (Summers, 1994). Efforts to increase the supply o f education for girls have resulted in impressive increases in school enrollment and female literacy, but many developing countries have experienced significant demand-side obstacles to full female school enrollment (Hill & King, 1993; Filmer, 2004). There is a fundamental tension in the literature on education and economic growth about how to overcome the low demand for girls’ education. An economic growth approach argues that the demand for education will increase as wealth increases (Goldin, 1995; Bils & Klenow, 1998). The human capital approach argues that investments will increase when labor markets offer higher wages for women (Schultz, T.W., 1960; Becker, 1981; Ram, 1982; Schultz, T.P., 1988; Galor & Weil, 1996). O ther theories focus on the high opportunity costs o f girls’ schooling (Smock, 1981; Heward, 1999; McMahon, 1999). Sociological and demographic theories argue that family structure, religious beliefs, and sex stereotypes perpetuate a cycle o f underinvestment in girls, including educational investments (Cleland, 1985; Boserup, 1995; Jejeebhoy, 1995). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 Importantly, these competing theories have conflicting implications for the types o f policies that will facilitate increases in girls’ education. This paper contributes to the literature on girls’ education and development by examining the relative importance o f these four theoretical influences on demand, and the subsequent effect on development outcomes, using international data. Regression analysis is used to test the relative effect o f each theoretical influence and to identify areas where public policy can facilitate an increase in demand. Then a structural equation model is developed to test the indirect effect o f demand-side variables on the important development outcomes o f fertility and child health through their effect on girls’ education. Decomposition o f structural effects enables an examination o f how economic growth combined with development policies can influence development outcomes both directly and indirectly by increasing the demand for girls’ education. The results support theoretical assertions that wealth, opportunity costs, and short-term returns to education influence the demand for girls’ education, but potential labor market returns do not. Much o f the correlation between economic growth and girls’ education comes through the provision o f public goods such as clean water and health care. The results confirm that girls’ education is associated with significant reductions in child mortality and fertility, indicating that efforts to increase the demand for girls’ education have significant benefits for future generations 2 THEORETICAL M ODEL The literature on girls’ education suggests that family decisions concerning girls’ school enrollment and its influence on social welfare are conditioned on a combination Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 o f economic, political, and social factors. Following Becker’s model o f educational investments, factors that influence consumption o f education can be divided into supply and demand (Becker, 1975). O n the supply side are the quality and quantity o f school slots available to girls through public and private schools. Demand-side influences are wealth, opportunity costs o f schooling, parental preferences, and — because educational consumption represents an investment in future productivity — rates o f return to educational investments. Integrated models o f family consumption o f education include all four elements o f demand (Buchmann, 2000; Bowman, 1984; Hill & King, 1993). However, there is no clear consensus about the relative importance o f each. Figure 1 illustrates an integrated theoretical model o f household decision-making concerning girls’ education. Figure 1. Theoretical M odel Developm ent Outcomes Population Health Gifts1 Education SUPPLY-SIDE Quality o f education (F) Quantity o f education (P) D EM A N D -SID E Wealth •poverty (C) •distribution o f income (C) Opportunity Costs •Domestic labor (P) •Investment in sons (C) Returns to Education •Health care (F) •Labor market (C) Parental Preferences •Religion (C) •Political tights (C) ♦Nuclear family structure (C,P) Each theoretical approach to educational consumption suggests a different role for public policy. Some approaches stress the importance o f the broad economic, social, or political context, which can only be changed incrementally over time. O ther Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 approaches point to specific policy changes, such as the provision o f health care or public goods that can increase girls’ access to schooling in the short-run. The demand- side factors in Figure 1 are labeled as either policy or context variables to identify potential strategies for public policy to facilitate an increase in demand for girls’ education.1 The first influence on demand is family wealth. A resource-dilution approach to family decision-making argues that large families have limited resources so each child receives only a small am ount o f educational investments (Parish & Willis, 1993; Lloyd & Gage-Brandon, 1994). If education is a normal good, increases in income should result in greater consumption o f schooling for boys and girls (Goldin, 1995). In addition to having greater ability to purchase education, Bils and Klenow (1998) argue that as families gain wealth, they are also able to transmit a higher quality o f intergenerational learning. The implication o f this approach for development policy is that increases in family wealth through economic growth are sufficient to increase girls’ access to schooling. There is some empirical evidence that both girls and boys do receive more education as incomes increase (Chemichovsky, 1985; Buchmann, 2000). However, there is also evidence from East Asia and South America that females do not benefit from economic growth as much as males, because gender inequality persists as wealth increases. Even as women gain more education through economic growth, the gaps between female and male achievement and opportunities can grow (Greenhalgh, 1985; Parish & Willis, 1993; Raynolds, 1998). Smock (1981) suggests that income and political Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 inequality may contribute to continued underinvestment in girls as wealth increases by reducing women’s capacity to compete for scarce resources. The second theoretical influence on demand for girls’ education is the opportunity costs o f girls’ schooling. A family-economy approach to family consumption argues that in some circumstances, wages from child labor are preferred over school attendance (Schultz, T.P., 1988). Opportunity costs o f educating daughters include both foregone wages and contributions to domestic work and child care (Appiah & McMahon, 2002). Case studies suggest that school-aged girls are frequently absent from school or unable to complete homework due to domestic responsibilities, and many teen-age girls withdraw from school entirely to help their mothers with rigorous domestic chores (Bellew & King, 1993; Bendera, 1999; Rose & Tembon, 1999; Wynd, 1999). While the domestic role o f women is socially determined, the implication o f family-economy theories is that the demand for girls’ education can be increased through the provision o f public goods that improve domestic production, such as clean water, electricity, and modem communications. Thus opportunity costs o f girls’ schooling may be reduced through effective development policies that invest in infrastructure and public goods. The third influence on demand is the rate o f return to investments in human capital for girls. A household-production approach suggests that educational investments depend primarily on long-term returns to education from future labor market prospects. Women’s wages are typically lower than men’s due to time spent out o f the labor market for pregnancy and child care, sex-segmented labor markets, and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wage discrimination (Schultz, T.W., 1960; Ram, 1982; Becker, 1985). The implication is that family investments in girls’ education will only increase when women’s access to labor markets improves (Galor & Weil, 1996). This leads some authors to recommend labor market policy as a strategy to increase girls’ education (World Bank, 1995; Buchmann, 2000). O ther authors doubt that labor market policy can change employer preferences as long as low average female education reinforces negative stereotypes o f women (Lloyd & Nicmi, 1979). An additional concern is whether parents can expect to benefit from a daughter’s education. It is typical for a daughter’s future earnings to go to her husband’s family, while sons continue to support their parents throughout adulthood (Greenhalgh, 1985). Because female labor depends on the cultural roles o f women, the labor market is considered a context variable in this analysis rather than immediate target for policy. In societies with low female labor force participation, parents may base educational consumption decisions on more short-term rates o f return, such as the probability that girls will succeed in school. In a context where females have a lower economic value and low autonomy, daughters typically receive less nutrition and have less access to health care than sons (Sen, 1990; Higgins & Alderman, 1997). Poor health reduces the likelihood that girls will benefit from schooling because unhealthy girls are unable to pay attention in class and to compete academically with boys. Case studies in developing countries suggest that parents do forego investments in girls’ education because they assume girls will fail academically, regardless o f future labor market prospects (Bunwaree, 1999; Rose & Tembon, 1999). Public policies can increase the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 short-run returns to girls’ education by providing an adequate health care system that compensates for incentives for parents to under invest in the health o f daughters. The final theoretical influence on demand is parental preferences. Cultural approaches to girls’ education argue that underinvestment in girls’ education results from social and political arrangements that reduce the economic incentive to invest in daughters (Caldwell, 1982; Boserup, 1990). These theorists are less optimistic about the potential for public policy to increase the demand for education, because social institutions can persist even when economic incentives change (Bardhan, 1988). Hill and King (1993) model cultural influences as “psychic costs” o f educating daughters, which can damage families through lost reputation and lost opportunities for social and economic interaction. Specific cultural influences cited in the literature as reducing the demand for girls’ education include religious beliefs that stress female purity, political institutions that limit female participation in public life, and traditional patriarchal family structure. Specifically, fundamentalist Islamic cultures are reported to be more resistant to girls’ education, because religious practices may include efforts to shield females from modem attitudes and values (Csapo, 1981; Smock, 1981). A lack o f political rights for women reduces public attention to female education and reduces parents’ incentive to educate daughters as a valued ethic o f democratic citizenship (Meyer, Ramirez & Soysal, 1992; Fierlbeck, 1997). Traditional patriarchal family structures are also associated with male authority. The transition to modem, nuclear families, beginning with modernized farming, gives women greater autonomy and access to resources, which increase Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 investments in daughters (Mason, 1993). Urbanization and mechanization are expected to facilitate this process through the shift from family production, where power is centralized under patriarchs, to wage labor where nuclear families are financially independent (Caldwell, 1982; Boserup, 1990). Although this transition can take several generations, governments and aid agencies may be able to facilitate this transition through policies that modernize farming and means o f production. The second stage o f the model in Figure 1 depicts the influence o f girls’ education on development outcomes. Based on extensive historical evidence, increases in girls’ education in developing countries are expected to lead to further development outcomes including reduced fertility and improved child health (Cleland, 1985; Hill & King, 1993; Jejeebhoy, 1995; McMahon, 1999). Education influences the behavior o f mothers by instilling modem attitudes and values, knowledge o f hygiene, and the ability to use family resources more efficiently (Kasarda, Billy & West, 1986; Mason, 1993). Educated women also have higher opportunity costs o f motherhood because they make higher wages (Bulatao, 1984). Importantly, the process through which women’s education translates into reduced fertility and improved child health is mediated by social and cultural influences that determine the costs and benefits o f children to educated and uneducated parents (Caldwell, 1978; Mason, 1993; Jejeebhoy, 1995). This study integrates these four theoretical influences on demand into a more comprehensive model o f educational consumption and its influence on development outcomes. The integrated, two-stage model in Figure 1 is tested using an international data set o f developing countries. Unlike previous studies o f girls’ education, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 theoretical framework applied here identifies the role for public policy implied by each theoretical approach and tests policy levers in the context o f development. Structural equation models provide new insight into the relationship among correlated components o f development including economic growth, urbanization, and access to public goods in the relationship between girls’ education and economic development. 3 METHODOLOGY Data & Variables To test the theoretical model depicted in Figure 1, country-level data were gathered from multiple international data sets. Data on education and economic conditions come primarily from the World Development Indicators (WDI) published by the World Bank (World Bank, 2002). Data on female labor market conditions come from the United Nations Women’s Indicators and Statistics (WISTAT) (United Nations, 1999). Data on the majority religion o f countries are from the Adherents Organization, a web research service that tracks international issues o f religious freedom and practice. Published membership and adherent statistics are referenced and aggregated to the national level for world religions (Adherents, 2004). Finally, data concerning political rights come from the Freedom House Freedom in the World Survey for 2000 (Freedom House, 2004). The survey ranks political rights and civil liberties on a scale from one to seven. The variables selected to operationalize the underlying theoretical concepts in Figure 1 are listed in Table 1. Wealth is measured by two economic context variables — G D P per capita and the G IN I index o f income distribution.2 Opportunity costs o f girls’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 education come primarily from domestic labor, which varies with access to public goods that relieve the burden o f housework. This effect is tested with a policy variable measuring the percent o f households with access to clean water. Short-term rates o f returns are tested with a policy variable measuring access to public health by the number o f doctors per 1,000 people. Long-term rates o f return are tested with a context variable measuring female economic participation (FEP). This measure is restricted to economically active women ages 20-24, who are less likely to be married and are thus more likely to pass wages back to parents as a return to childhood investments in education. Parental preferences are represented by context variables measuring urbanization, political freedom, and Islamic beliefs. Urbanization is measured as the percent o f the population living in cities. Political freedom is represented by the Freedom House index measuring civil liberties on a scale o f one to seven.3 The civil liberties index includes measures o f adherence to the rule o f law, fundamental rights, access to information through a free press, and equality o f opportunity (Freedom House, 2004). Islamic beliefs are measured by a dummy variable equal to one if the majority o f the population identifies itself as Islamic. The degree o f modem production, an important influence on nuclear family structure, is represented by a policy variable measuring the number o f tractors per 1,000 hectares o f arable land. Dependent variables measure girls’ education, fertility, and child mortality. Girls’ education is measured as female gross enrollment rates (GER) in secondary school. Secondary school enrollment is used because the literature suggests that fertility declines occur only when women gain access to secondary education (Gille, 1985; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 Jejeebhoy, 1995; Lam & Duryea, 1999).4 Fertility is measured by the total fertility rate (TFR), which is equal to the number o f births per women o f childbearing age in 2000. Health is measured by the child mortality rate (CMR), which is equal to the number o f deaths o f children ages 0-5 per 1,000 live births in 2000. Table 1. Policy & Context Variables T heoretical Concept V ariable N a m e M easurem ent Source T yp e W ealth G D P p er capita G D P p er capita + W D I context G IN I G IN I index W D I context Opportunity Costs clean w ater % households w ith access to an im proved w ater source W D I policy Rates o f Return doctors num ber o f doctors per 1,000 people W D I policy FE P female econom ic participation (ages 30-24) WISTAT context Parental Preferences Islamic majority Islamic A dherents context civil liberties Freedom H ouse scale o f civil liberties Freedom H ouse context urban % population living in urban areas W D I context tractors num ber o f tractors per 1,000 hectares o f farmland W D I policy + G D P per capita is measured in 1995 US dollars Regression M odels To gain a better understanding o f the relative effect o f policy and context variables, the first model includes only context variables in a prediction o f girls’ secondary enrollment. The reduced form estimates female gross enrollment rates as a function o f context only. The relationship between context and female secondary enrollment is tested as a simple OLS model. GER =ot + Pi log(GDP per capita) + GINI + p3 FEP + p4 urban (1) + 3s Islamic + pg civil liberties + e Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 The second model adds demand-side policy variables to determine w hether policies have an effect on educational consumption that is independent o f the economic, social, and political context: GER = a + Pi log(GDP per capita) + £ $ 2 G IN I + £ 8 3 FEP + p4 urban (2) + S5 Islamic + £ $ 6 civil liberties + £ 8 7 doctors + £ 8 g tractors + £ 8 9 dean water + e Inclusion o f a variable representing the supply o f education is needed to fully test the theory o f education consumption (Becker, 1975; Bowman, 1984). However, testing ibis effect is problematic due to endogeneity o f enrollment and supply variables. Enrollment may increase because o f increased supply o f education, but variables reflecting supply, such as public expenditures and teacher-student ratios, also respond to increases in enrollment. Measures o f the quantity and quality o f schooling in international datasets do not include an appropriate instrument to apply here. Recognizing this limitation, a third model is estimated adding a variable for the supply o f education. Supply is tested by a W DI variable measuring public education funding as a percent o f G D P per capita. GER =0 + pi log(GDP pet capita) + 82 GINI + £ 8 3 FEP + £ 8 4 urban (3) + p5 Islamic + , 3s civil liberties+ £ 8 7 doctors + £ 3g tractors + £ 8 9 dean water + pio education funding + e The second stage o f regressions predicts the effect o f female enrollment on total fertility rates (TFR) and child mortality rates (CMR). The reduced form equation assumes that social benefits are a function o f female secondary enrollment and the economic, social, and political context. Here, it is important to identify whose education matters most. It is possible that school enrollment during the mother’s Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 generation is important if education affects a woman’s preferences for family size and the opportunity costs o f child care. It is also possible that fertility decisions are based on the current costs and benefits o f children, which are affected by the likelihood that children will attend and complete school. First, the effect o f mother’s education is tested using a lagged variable for female secondary gross enrollment rates in 1985 — fifteen years before the outcome measures. The reduced models controls only for context variables and female secondary enrollment: TFR/CM R = at + ft G E R 1985 + p2log(GDP per capita) + p3 GINI (4) + ^4 FEP + Ps urban + Pe Islamic + 3- civil liberties + e The full model tests whether the policy variables have a direct effect on fertility and child mortality in addition to the indirect effect through female secondary enrollment. The OLS estimations for the full model are: TFR/CMR = a + pi G ER 1985 + p2 log(GDP per capita) + S3 GINI (5) + f$ 4 FEP + Ss urban + pglslamic + p7 civil liberties + , 3g doctors + pj tractors + Pm clean water + e Finally, in Model 6, current female enrollment in 2000 is added to the above model to determine if opportunities for daughters to enroll in school affect fertility and health in addition to mother’s education. TFR/CM R = a + Pi GERiggs + P2 GER2000 + p3 log(GDP per capita) (6) + p4 GINI + p5 FEP + Pg urban + p7 Islamic + px civil liberties + P9 doctors + P10 tractors + Pn dean water + e Structural Equation M odeling The regression models predict development outcomes controlling for policy and context. Regression analysis in based in the assumption that right-side variables are independent. This is not a realistic assumption for the right-side variables in this study. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 In international development, policy and context variables are correlated, and the influence o f these correlations is important to policy-makers. For example, the economic growth is associated with greater availability o f health care and thus influences girls’ education both directly and indirectly. These complex relationships can be examined using structural equation modeling. Structural equation modeling tests causal relationships by modeling the direct effect of independent variables on outcomes, as well as the indirect effect through intervening and correlated variables. Effects can be divided in three types: 1) direct effects o f an independent variable on a dependent variable, 2) indirect effects o f an independent variable through an intervening variable, or 3) correlation effects through a second independent variable that has direct or indirect effects on the dependent variable. If the estimation is a good fit, the sum o f these estimated effects will recreate the correlation between the independent and dependent variables (Knoke, Bohmstedt & Mee, 2002). Thus, this method can separate the correlation between two variables into the direct effect and the effects that occur indirectly through intervening or correlated variables. In this case fertility and child mortality are the dependent variables, female gross enrollment is the intervening variable, and context and policy variables are correlated variables that influence both girls’ education and development outcomes.5 Therefore, the lagged female enrollment variable cannot be included as an intervening variable, because current policies cannot cause past school enrollment. Using only current female secondary enrollment, the model here tests the causal hypothesis that female access to education influences fertility and child mortality by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 0 altering the costs and benefits o f children (Bulatao, 1984). The formal representation o f the model is: Q' v = ^ P 9 = 1 The bivariate correlation Q between variables i and j is estimated as the sum o f the paths from variable i to q, multiplied by the correlation between variable j and q. The path P is a partial correlation that can be estimated with OLS regression on standardized variables. All variables were standardized with a mean o f zero and standard deviation o f one. The coefficients therefore cannot be interpreted in terms o f absolute effect sizes, but they can be compared directly to one another. 4 RESULTS Countries in Analysis Full data on school enrollment, policy variables, and context variables are available for a subset o f 81 o f the 151 low and middle income countries in the WDI data sets. Low and middle income countries were selected so the sample would reflect countries at different levels o f development. Historical data on female secondary enrollment in 1985 is missing for eight o f these countries, so models that include the lagged enrollment variable have only 73 observations. The countries included in analysis are listed in Table 2. The sample includes all regions o f the world and provides a reasonably good geographic sample o f all developing countries. However, the sample contains a greater proportion o f low income countries compared to middle income countries than the larger population. Because there are data collection problems in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 1 countries with political instability or oppressive regimes, the 81 countries here may reflect a greater level o f political stability and freedom on average than countries that were excluded due to missing data. Table 2. Countries in R egression Analysis Africa South A m erica M iddle E ast & . E urope & Botswana Bolivia N orth Africa Central A sia Burkina Faso Brazil Algeria Belarus + Burundi Chile Egypt Georgia Cameroon Colombia Iran Hungary Central African Rep. Ecuador Jordan T .ithuania + Cote D ’Ivoire Guyana Morocco Romania Ethiopia Paraguay Tunisia Russian Federation The Gambia Peru Yemen + Slovak Republic + Ghana Uruguay Tajikistan + Guinea Venezuela Turkey Guinea-Bissau Ukraine + Kenya Uzbekistan Lesotho Central A m erica & A sia & the Pacific Madagascar Caribbean Bangladesh Malawi Costa Rica Cambodia Mali Dominican Republic China Mauritania El Salvador Tndia Mozambique Guatemala Indonesia Namibia + Honduras Lao PDR Niger Jamaica Mongolia Nigeria Mexico Nepal Rwanda Nicaragua Pakistan Senegal Panama Papua N ew Guinea Sierra Leone Trinidad & Tobago Philippines South Africa + Sri Lanka Tanzania Thailand Uganda Vietnam Zambia Zimbabwe + Data on lagged female secondary enrollment is missing. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 2 Descriptive Statistics Table 3 displays descriptive statistics for each variable. To place countries in the dataset in a development context, mean values for O ECD countries are also displayed. Descriptive statistics show that developing countries lag behind O EC D countries in female schooling and development outcomes. In the countries included in the analysis, mean female secondary enrollment was only 34.3 percent in 1985, and increased to 51.8 percent in 2000. The mean fertility rate o f the dataset was 3.65 births per woman, and fertility ranged up to 7.22 births per women. The mean child mortality rate is 90.21 deaths per 1,000 births, compared with only 6.34 in O ECD countries. The economic and political context is also very different in the countries in the dataset versus O EC D countries. Annual G D P per capita in the developing countries averaged only $1,444 in 1995 US dollars, compared with over $29,000 in O ECD countries. The countries in the dataset had a higher mean level o f income inequality and lower mean level o f urbanization than the O ECD means. The countries in the dataset also had a lower mean level o f civil liberties. The countries in the dataset varied in access to public goods, but in general, lagged far behind O ECD countries. In the countries included in this analysis, the number o f doctors ranged from only 0.03 per 1,000 people up to 4.43 per 1,000 people. Access to clean water averaged 75 percent in the developing countries, but was as low as 24 percent. The extent o f modem agriculture was very low in the developing countries. O ECD countries had an average o f 18.16 tractors per 1,000 acres o f farmland, compared with a mean o f only 0.92 tractors per 1,000 acres in developing Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 countries. Public funding for education in developing countries ranged from less than one percent o f the G D P per capita up to more than 10 percent. Table 3. M ean Values for D ependent and Independent Variables Variable M inim um M axim um M ean (sd) O E C D m ean total fertility rate (TFR) 1.08 7.22 3.65 (1.56) 1.63 child mortality rate (CMR) 9.74 316 90.21 (70.95) 6.34 female secondary enrollment (GER) 1985 2.3% 99.8% 34.3% (27.0) 92.8% female secondary enrollment (GER) 2000 4.9% 100% 51.8% (27.0) 100% urban 6.2% 91.9% 45.5% (20.9) 78.4% G DP per capita $115 $6,104 $1,444 (1,484) $29,208 G IN I 24.4 70.7 44.0 (10.0) 32.1 Islamic 0 1 0.23 (0.42) 0 female economic participation (ages 20-24) 30.3% 90.8% 62.0% (15.2) 70.9% education as % G DP per capita 0.7% 10.3% 4.1% (2.1) n /a civil liberties + 2 7 4.07 (1.34) 7 doctors per 1,000 people 0.03 4.43 0.99 (1.18) 3.04 tractors per 1,000 hectares 0.003 4.11 0.92 (1.06) 18.16 clean water 24% 100% 75.2% (18.4) 99.3% + Civil liberties are measured on a scale o f 1-7, with 7 reflecting the highest level o f freedom. Scatter Plots Figures 2 and 3 display two-by-two scatter plots o f the relationships between the policy variables and female secondary enrollment and between context variables and female secondary enrollment, respectively, for the developing countries. As expected, there is a positive relationship between access to clean water and tractors and girls’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 school enrollment. The relationship between doctors and school enrollment is positive but also appears to be nonlinear, with diminishing returns to additional doctors after a certain level o f public health is reached. There is also a clear positive relationship between wealth and girls’ secondary enrollment. The relationship between urbanization and income distribution is less clear. There appears to be a weak positive relationship between urbanization and school enrollment, and no discemable relationship between the G IN I index and school enrollment. There is also no clear relationship between female economic participation and female schooling. Figure 2. Scatter Plots o f Policy Variables & Female Education Clean Water E l™ • » : "U B M * * B t tMf 2 3 4 0 30 ifo 12} D octors T Q M ® Mt“ kO V ,S D I “ 7E M pB8 “ to -W* „ 1 35 2 2 5 3 35 A 45 ftj^saaEpa I.GOGpetjie Tractors ' “ 'ml*' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 Figure 3. Scatter Plots o f Context Variables & Fem ale Education Urbanization W ealth F em ale E c o n o m ic Participation G IN I Index Predicting Girls’ Enrollment Model 1 predicts female secondary enrollment as a function o f context variables only. The results are displayed in Table 4, Column 1. Based on the apparent nonlinear relationship between doctors and school enrollment (see Figure 2), a quadratic term for the number o f doctors per 1,000 people was added to this and subsequent models. The coefficients are in the direction predicted by theory. Both urbanization (p<0.05) and G D P per capita (p<0.01) have a positive, significant association with female secondary enrollment. The G IN I index has a significant, negative coefficient (p<0.05), indicating that greater economic inequality reduces girls’ access to schooling. An Islamic majority reduces female school enrollment, but the coefficient is not statistically significant. This Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 suggests that Islamic countries do not necessarily have a strong stigma against educating girls. The coefficient on civil liberties is positive, but this effect is also not statistically significant. These results suggest that the economic context may be more important than political or Islamic institutions in determining girls’ access to schooling. Model 2 adds policy variables that are predicted to increase the demand for female education.6 These results are displayed in Table 4, Column 2. A one percent increase in access to clean water results in a 0.40 percent increase in female secondary enrollment (p<0.01). The effect o f doctors is much larger. The addition o f one doctor per 1,000 people results in a 22.4 percent increase in female secondary enrollment (p<0.01). The quadratic term for doctors is significant and negative (p<0.05), indicating that there are eventually decreasing returns to additional access to health care. The coefficient on the number o f tractors is also positive but not significant. It is possible that there is a long lag time before agricultural technology influences family structure and opportunities for girls. In the short-run, it is possible that the technology in farming has a greater influence on men than women.' Model 3 adds the effect o f the supply o f education. This variable is significant and positive (p<0.01) as expected, indicating that consumption o f education increases with public funding for education. This provides support for Becker’s theory o f educational consumption, but because o f endogeneity problems, the effect size should not influence interpretation. The addition o f the supply variable has almost no effect on the other coefficients, suggesting that the endogeneity of supply o f education and school enrollment does not bias the prediction o f the coefficients for demand-side variables. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 Table 4. R egression Results for Fem ale Secondary Enrollm ent (GER) D epend en t Variable: fem ale secondary enrollm ent C O L U M N 1 M odel 1 C O L U M N 2 M odel 2 C O L U M N 3 M odel 3 Coeff. t-ratio Coeff. t-ratio Coeff. t-ratio Constant -64.649 -2.51** -71.274 -3.18*** -68.034 -3.17*** C ont ext log (GDP per capita) 16.593 4 9g*** 9.162 2.89*** 7.670 2.50** G IN I index -0.594 -2.46** 0.057 0.26 -0.746 -0.34 female economic participation (20-24) 0.154 0.93 0.159 1.07 0.204 1.42 urban 0.376 2.31** -0.080 -0.49 0.002 0.01 Islamic -5.035 -0.87 -2.502 -0.50 -2.595 0.55 civil liberties 1.244 0.62 1.275 0.77 1.79 1.13 P ol i ci es doctors 22.519 3 44*** 25.076 3.97*** (doctors)2 -3.444 -2.40** -4.318 -3.08*** tractors 3.489 1.54 2.543 1.17 clean water 0.404 2.98*** 0.325 2.45** Suppl y o f Educat i on education spending as % o f G D P per capita 2.560 282*** R-squared 0.653 0.783 0.805 N o. o f Observations 81 81 81 ***p<0.01, **p<0.05, *p<0.10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 Importantly, the addition o f the demand-side policy variables also affects the coefficients for the context variables. Controlling for the policies that influence demand, coefficients for urbanization and the G IN I index are not significantly different from zero. G D P per capita remains significant (p<0.01), but the effect size is approximately cut in half. These results suggest that the relationship between some contextual variables and school enrollment is spurious. The influence o f urbanization and the distribution o f wealth on girls’ school enrollment may come through the provision o f public goods that often accompanies economic growth and urban migration. Predicting Fertility & Child M ortality The results o f regressions on fertility and child mortality are displayed in Tables 5 and 6, respectively. This second stage o f regression analysis enables us to see how public policies influence development outcomes through girls’ education. Model 4 tests the influence o f context and enrollment on fertility and child mortality to determine if the effect o f girls’ education on positive externalities is independent o f the economic, social, and political context. Model 4 predicts fertility and child mortality based only the educational access o f the mother using a lagged school enrollment variable. Controlling for context variables, lagged secondary enrollment has a negative effect on fertility (p<0.01) and child mortality (p<0.01). G DP per capita also has a significant, negative effect on fertility (p<0.01) and child mortality (p<0.01). The G IN I index has a significant positive effect on fertility (p<0.10) and child mortality (p<0.10). Surprisingly, the coefficient on civil liberties is positive for fertility and mortality, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 9 although only the coefficient on fertility is statistically significant (p<0.10). It is unclear why an increase in civil liberties would increase family size, and this result conflicts with theoretical assertions that democracy is associated with declines in fertility. Model 5 adds demand-side policy variables. The num ber o f doctors (p<0.05) and percent o f population with clean water (p<0.10) have a significant negative effect on child mortality, independent o f the lagged effect o f girls’ secondary education. With the addition o f the policy variables, G D P per capita is the only context variable that significantly reduces child mortality (p<0.01). Investments in child health appear to increase when mothers have access to education, as well as w hen family wealth increases and access to health care improves. O f the policy variables, only access to clean water has a significant negative effect on fertility (p<0.10). Both G D P per capita (p<0.01) and the G IN I index (p<0.10) continue to have a significant effect on fertility. These results suggest that fertility decisions respond primarily to parents’ wealth — as wealth increases, family size decreases. In addition, improved domestic conditions may lead to reduced family size, possibly because children are not needed to assist mothers at home. Model 6 attempts to separate the effects o f education during the mother’s generation and education during the current generation by adding an independent variable for girls’ secondary enrollment in 2000. Girls’ secondary enrollment in both 1985 (p<0.01) and 2000 (p<0.10) have independent, significant negative effects on fertility. This suggests that the mother’s education, as well as the daughters’ educational prospects influence fertility decisions. In the estimation o f child mortality, the coefficient on girls’ secondary enrollment in 1985 is significantly different from zero Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 (p<0.10), but the coefficient on girls’ secondary enrollment in 2000 is not. This suggests that improvements in child health may come primarily through educated mothers’ increased knowledge o f hygiene and increased willingness to seek medical care (Jejeebhoy, 1995). Importantly, the number o f physicians also has a significant negative effect on child mortality, indicating the combination o f educated mothers and availability o f health care can reduce child mortality. Placing these results into the theoretical framework, they suggest that wealth has positive effects on the demand for girls’ education. However, the effect o f wealth on girls’ enrollment is significantly reduced with the addition o f the policy variables in Model 2. This suggests that much o f the correlation between economic growth and expanded educational opportunities for girls comes through a reduction in opportunity costs from improved domestic technology and increased short-term returns to girls’ education through access to health care. Surprisingly, female labor market participation is not a significant predictor o f girls’ education. Parents may be more concerned with whether girls will be competitive in school than with daughters’ prospects for employment. Islamic religion and political structure appear to be less important in determining girls’ schooling than the economic factors o f wealth, opportunity costs, and short-term returns to education. The effect o f modernization through urbanization and modem farming is also not significant. The economic context influences fertility and child mortality directly, but also through the indirect effect o f female education, which reduces fertility and improves child health. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 Table 5. Regression R esults for T otal Fertility Rate (TFR) D ep en d en t Variable: total fertility rate Fertility C olum n 1 M odel 4A C olum n 2 M odel 5A C olum n 3 M odel 6A Coeff. t-ratio Coeff. t-ratio C o e ff t-ratio Constant 8.304 7.76*** 8.843 7.978 6.21*** Femal e Educat i on G ER 1985 -0.030 -6.89*** -0.030 -5.32*** -0.024 -3.76*** G ER 2000 -0.011 -1.73* C ont ext log (GDP per capita) -0.759 -5.65*** -0.664 -4.13*** -0.556 -3 27*** G INI index 0.021 1.88* 0.020 1.73* 0.022 1.91* FEP (ages 20-24) -0.002 -0.24 -0.007 -0.89 -0.004 -0.58 urban 0.002 0.27 0.000 0.01 -0.001 -0.12 Islamic 0.188 0.81 0.234 0.98 0.246 1.05 civil liberties 0.140 1.75* 0.166 1.99* 0.161 1.96* Pol i ci es doctors -0.011 -0.03 0.128 0.37 (doctors)2 0.020 0.27 -0.012 -0.15 tractors 0.021 0.18 0.029 0.24 clean water -0.012 -1.75* -0.007 -1.06 R-squared 0.807 0.817 0.826 N o. o f Observations 73 73 73 ***p<0.01, **p<0.05, *p<0.10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6. R egression R esults for Child M ortality Rate (CMR) 3 2 D epend en t Variable: child mortality rate C hild Mortality C olum n 1 M odel 4B C olum n 2 M odel 5B Colum n 3 M odel 6B Coeff. t-ratio Coeff. t-ratio C o eff t-ratio Constant 324.238 50.57*** 356.370 5 93*** 323.540 4 93*** Fem al e Educat i on G ER 1985 -0.975 -4.12*** -0.786 -2.72*** -0.580 -1.74* G ER 2000 -0.402 -1.19 C ont ext log (GDP per capita) -47.654 -6.53*** -39.390 -4.80*** -35.546 -4.04*** G IN I index 1.190 1.92* 0.918 1.54 0.983 1.65 FEP (ages 20-24) 0.486 1.34 0.060 0.16 0.145 0.37 urban 0.471 1.16 0.691 1.61 0.657 1.53 Islamic 13.878 1.11 12.915 1.06 13.332 1.10 civil liberties 3.764 0.86 4.373 1.02 4.200 0.99 Pol i ci es doctors -48.134 -2.73*** -43.205 -2.39** (doctors)2 11.123 2.87*** 9.989 2.51** tractors 4.743 0.77 5.017 0.81 clean water -0.677 -1.97* -0.524 -1.43 R-squared 0.756 0.796 0.801 N o. o f Observations 73 73 73 ***p<0.01, **p<0.05, *p<0.10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 Structural Equation M odeling The regression results reveal that part o f the effect o f economic growth and the full effect o f urbanization on development outcomes disappears in models that control for policy variables. While regression analysis can control for independent variables, in the context o f economic development, we are also interested in the relationships between independent variables and particularly the relationship between the context o f growth and public policies that prom ote social benefits. While clean water and improved health care often accompany economic growth, this is not necessarily the case, because provision o f public goods requires investments by governments or aid agencies. The structural equation models were developed to decompose the correlation between economic growth, education, and development outcomes into the direct effect o f increases in wealth and the indirect effects o f the policy variables. Correlations are displayed in Table 7, and the standardized regression results are displayed in Table 8. The correlations show that girls’ secondary enrollment is positively related to G D P per capita, urbanization, doctors, tractors, and clean water. Fertility and child mortality are negatively correlated with female secondary enrollment, G D P per capita, urbanization, doctors, tractors, and clean water, and positively correlated with inequality in the distribution. However, strong correlations between the independent variables suggest that these relationships are complex. Standardized regression results indicate that G D P per capita, doctors, tractors, and clean water have a positive, direct effect on girls’ secondary enrollment. In turn, G D P per capita and girls’ secondary enrollment have a significant, negative effect on fertility and child mortality. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 4 Significant coefficients on G IN I index indicate that increases in income inequality raise fertility and child mortality. These results generally confirm the results o f regression analysis, except that the num ber o f tractors was not significant in regression analysis, and it is here, probably because standardized coefficients are more sensitive to small differences in ibis variable. Table 7. Correlations Variables fertility child mortality female GER % urban lo g (GDP per capita) GINI index FEP doctors tractors clean water fertility 1.00* child mortality 0.87* 1.00* fem ale G E R + -0.84* -0.81* 1.00* % uib an -0.65* -0.63* 0.67* 1.00* lo g (G D P per capita) -0.72* -0.79* 0.77* 0.77* 1.00* G IN I index 0.28* 0.18 -0.11* 0.11 0.08 1.00* F E P + + 0.04 0.21 -0.14 -0.39* -0.30* -0.29* 1.00* doctors -0.72* -0.61* 0.69* 0.62* 0.50* -0.30* 0.06 1.00* tractors -0.60* -0.54* 0.61* 0.42* 0.53* -0.27* -0.04 0.52* 1.00* clean water -0.62* -0.67* 0.68* 0.58* 0.66* -0.08 -0.35* 0.47* 0.42* 1.00* + gross female enrollment in secondary school in 2000 + + percent o f wom en age 20-24 who are employed or unemployed * correlation is significant at 5 percent level or better (p<0.05) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 Table 8. Structural Equation M odel R egression R esults Stage 1 Stage 2 Colum n 1 C olum n 2 C olum n 3 D ep en d en t Variable G E R 2000 fertility child m ortality Std. Std. Std. T-ratio CocfT. * A 'fduO Coeff.* Coeff.* Femal e Educat i on G ER 2000 -0.463 -4 25*** -0.378 -3.24*** C ont ext urban -0.031 -0.12 -0.108 - 1.02 0.119 1.05 log (GDP per capita) 0.392 3.71*** -0.208 -1.94* -0.480 -4.18*** G IN I index 0.035 0.51 0.180 2.83*** 0.157 2.30** FEP (ages 20-24) 0.059 0.81 -0.082 -1.20 0.086 1.19 P ol i ci es doctors 0.319 3.45*** -0.132 -1.42 -0.132 -1.33 tractors 0.151 2.01** -0.025 -0.35 0.040 0.52 clean water 0.243 2.94*** -0.047 -0.58 -0.075 -0.86 R-squared 0.760 0.794 0.765 N o. o f 81 81 81 Observations *** p<0.01, ** p<0.05, * p<0.10 + Variables are standardized to mean=0, sd = l. The decomposition o f effects, displayed in Tables 9 and 10, provides insight into how much o f the relationship between context variables and development outcomes is due to correlations with the provision of public goods. The total effects show that the models are able to predict the true correlations within one ten-thousandth for each variable. The results indicate that a large portion o f the relationship between Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 urbanization and development outcomes is due to the correlation between urbanization and other variables. The direct effect o f urbanization on child mortality is positive and the correlation effects are negative. This suggests that declines in child mortality do not occur because o f urbanization itself, but because cities offer greater access to doctors and clean water. Average wealth, measured by G D P per capita, has a direct negative effect on child mortality and fertility and also reduces child mortality and fertility indirectly by increasing girls’ education. However, approximately one-third o f the correlation between G D P per capita and fertility and child mortality is explained by the correlation effects o f doctors, tractors, and clean water. Table 9. D ecom position o f Effects for Total Fertility Rate (TFR) Independent Variables D irect effect on fertility Indirect effect through secondary enrollm ent Correlation effects T otal estim ated correlation A ctual correlation C ont act urban -0.108 0.006 -0.546 -0.648 -0.648 log (GDP per capita) -0.208 -0.182 -0.328 -0.718 -0.718 G IN I index 0.180 -0.016 0.111 0.275 0.275 FEP (ages 20-24) -0.082 -0.027 0.152 0.043 0.043 P ol i ci es doctors -0.132 -0.148 -0.437 -0.716 -0.716 tractors -0.025 -0.070 -0.503 -0.599 -0.599 clean water error -0.047 0.490 -0.113 0.453 -0.462 -0.622 -0.622 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 Table 10. D ecom position o f Effects for Child M ortality Rate (CMR) Independent Variables D irect effect on child mortality Indirect effect through secondary enrollm ent Correlation effects T o ta l estim ated correlation Actual correlation C ont ext urban 0.119 0.005 -0.754 -0.630 -0.630 log (GDP per capita) -0.481 -0.148 -0.157 -0.786 -0.786 G IN I index 0.157 -0.013 0.038 0.182 0.182 FEP (ages 20-24) 0.086 -0.022 0.142 0.201 0.201 P ol i ci es doctors -0.132 -0.121 -0.361 -0.614 -0.614 tractors 0.040 -0.057 -0.527 -0.544 -0.544 clean water -0.075 -0.092 -0.502 -0.668 -0.668 error 0.490 0.485 The structual equation results suggest that the transition from increased wealth and urbanization to improved development outcomes is at least partially contingent on corresponding implementation o f policies to increase access to health care, clean water, and other public goods. While, these policies have direct effects child mortality but not fertility, they also have an indirect effect on these outcomes by increasing girls’ access to schooling, which in turn accelerates development outcomes. 5 DISCUSSION This study confirms that wealth, opportunity costs, and short-term rates of return increase investments in girls’ education, which in turn lead to positive development outcomes. Resource-dilution theories assume that wealth will increase Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 educational investments. The results here consistently show that girls’ access to education also depends on policies that address opportunity costs and rates o f return by providing clean water and health care. Importantly, the effects o f public policies on girls’ education are independent o f Islamic religion and political context Because the later result was unexpected, alternative measures for political freedom were tested with similar results.8 This runs contrary the assertion that parents in some cultures are unwilling to educate girls even if the economic costs are low. In her research on Pakistan, Heward suggests that traditional gender roles affect parents’ preferences for the quality o f schooling, but it is simply not true that parents are unwilling to education girls: Contrary to the long-accepted view that parents are unwilling to educate their daughters, it is now widely understood that most parents do want to educate their daughters. It is poor provision o f suitable schools that prevents them from sending their daughters to school. (Heward, 1999: 210-211) The results o f this study suggest that in addition to supply-side reforms to make schools more appropriate for girls, demand-side policies are important to promote girls’ education in countries with traditional gender roles. An important implication is that education policy should be considered in the economic context o f the costs o f girls’ schooling, and that efforts to increase or improve supply should be coupled with efforts to identify and reduce the opportunity costs o f attending school. The strong negative effect o f girls’ schooling on fertility rates and child mortality supports the assertion that girls’ education is important for long-term economic development. Even if women’s labor force participation is limited, the social benefits Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 o f girls’ education, confirmed in this study, justify significant public investment in girls’ education. The effects o f the lagged education variable also indicate that it may take a generation to realize the full benefits o f investments in girls’ schooling, particularly concerning child health and mortality. The importance o f girls’ education also suggests that policies to increase demand have additional payoffs beyond the direct social benefits. The structural equation modeling results here suggest that an important way get these benefits is to implement demand-side policies that increase girls’ access to schooling. A third unexpected finding is that the relationship between women’s labor market participation and access to education is weak in developing countries. Contrary to the household production models, female labor market participation does not significantly increase girls’ school enrollment. Labor market opportunities as they are now may not influence the demand for girls’ education if jobs for women do not offer rewards for education. Typically, labor markets in developing countries are segmented so only males can access jobs in engineering, law, and other fields that reward education (Ram, 1982). Alternately, evidence shows that educated women actually leave the labor market during the early stages o f development and return only when a significant clerical sector has been established (Goldin, 1995). It has also been suggested that initially girls’ education is intended to increase home productivity rather than workforce productivity (Lam & Duryea, 1999). Finally, it is possible that the measure o f female labor force participation used here does not adequately reflect the relationship between education and employment. A better measure might be the percent o f women working Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 in management and other upper-level professions, but currently this data is not available for m ost developing countries. 6 CONCLUSION This study is the first o f its kind to test multiple theoretical influences on the demand for girls’ education using international data. The results add to understanding o f how international development — including economic growth and increased access to public goods — enables girls to access education within the institutional constraints o f poverty, religion, and limits on political freedom. By identifying a role for public policy, the results here contribute to an understanding o f how policies work within the development context. The results indicate that while economic growth does increase female enrollment, policies that reduce opportunity costs and increase the short-term returns to girls’ schooling can also significantly increase enrollment. The implication for policy makers is that greater gains in girls’ schooling will occur when economic growth is accompanied by access to public goods and public health and when attention is paid to the demand for education, as well as the supply. Reducing the opportunity costs of education and helping parents invest equal resources in daughters and sons can facilitate universal education. Policy-makers tend to segment education policy as a specific area that deals primarily with the quantity and quality o f schooling. A more effective approach to education policy in developing countries will integrate implementation o f traditional education policies with broad efforts to address the need for girls’ education. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 Expanding the scope o f education policy to include factors such as home economics and child health can remove obstacles to schooling that are created by the realities o f daily life for families in developing countries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 2 NO TES 1 This approach follows the models o f Subbarao and Raney (1993) and Kaufman and Wang (1995), which isolate policy variables in studies o f the effect o f girls’ education on fertility and child health and the effect o f macroeconomic policy on education initiatives, respectively. 2 G D P is expressed in constant 1995 US dollars. 3Freedom House’s index gives a rating o f 1 to the “m ost free” countries and 7 to the “least free” countries. To facilitate interpretation o f regression results, the index was reversed so the most-free countries receive the highest ranking. This reversed variable was also used by McMahon (1999) in his study o f the social benefits o f education. 4 Some researchers argue that girls’ primary education actually increases fertility by teaching basic health and hygiene that can reduce infant mortality. Thus, Gille (1985) warns that from a demographic perspective, “a little education can be a dangerous thing.” Secondary education is needed to reduce fertility because school enrollment prevents early marriage and completion o f secondary school may create employment opportunities that further delay marriage and childbirth (Jejeebhoy, 1995). 5 Similar uses o f the application structural equations to theoretical issues in economic development include Loebner & Driver (1973), focusing on fertility and birth control in India, Nassirpour (1985), focusing on fertility and oil revenue in Iran, and Stinner (1979), focusing on modernization and family structure in the Philippines. These studies investigate the effect o f economic growth and correlated policy factors on development outcomes. 6 Because these development policies are expected to be correlated with the context o f development, tests were conducted to ensure that multicoll ineanty was not a problem. The variation influence factor (VIF) statistics are displayed below. Variable VIF log (GDP per capita) 4.94 G INI 1.70 female economic participation, ages 20-24 1.83 urbanization 4.58 Islamic 1.53 civil liberties 1.72 doctors per 1,000 people 3.39 tractors per 1,000 hectares 2.14 % population with clean water 2.57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 ' Regressions were run with male secondary enrollment as the dependent variable, and here the number o f tractors was positive and significant. It may take another generation before advances in production influence the nuclear family structure as hypothesized by Boserup (1990). 8 Alternative measures o f political context tested included Freedom House’s index o f political rights and the year women gained the right to vote. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 Works Cited Adherents Organization. (2004). Majority Religions, h ttp ://www.adherents.com. Appiah, E. N., & McMahon, W. W. (2002). The social outcomes o f education and feedbacks on growth in Africa. The Journal of Development Studies, 38(4), 27-68. Bardhan, P. (1988). Alternative approaches to development economics. In H. Chenery & T. N . Srinivasan (Eds.), Handbook of Development Economics (Vol. 1). Amsterdam: Elsevier Science Publishers, B.V. Becker, G. (1975). Human Capital New York: National Bureau o f Econom ic Research. Becker, G. (1981). ^4 Treatise on the Family. Cambridge, MA: Harvard University Press. Becker, G. (1985). Human capital, effort, and the sexual division o f labor. Journal of Labor Economics, 3(1), S33-S58. Bellew, R. T., & King, E. M. (1993). Educating Women: Lessons from Experience. In E. M. King & M. A. Hill (Eds.), Women's Education in Developing Countries. Baltimore, MD: The Johns Hopkins University Press. Bendera, S. (1999). Promoting education for girls in Tanzania. In C. Heward & S. Bunwaree (Eds.), Gender, Education < & ' Development (pp. 117-132). London: Zed Books Ltd. Bils, M., & Klenow, P. J. (1998). Does schooling cause growth or the other way around? N BER Working Paper Series (No. 6393). Cambridge, MA: National Bureau of Economic Research. Boserup, E. (1990). Economic change and the roles o f women. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Boserup, E. (1995). Obstacles to advancement o f women during development. In T. P. Schultz (Ed.), Investment in Women's Human C apital Chicago: University o f Chicago Press. Bowman, M. J. (1984). An integrated framework for analysis o f the spread o f schooling in less developed countries. Comparative Education, 28(4), 563-583. Buchmann, C. (2000). Family structure, parental perceptions, and child labor in Kenya: W hat factors determine who is enrolled in school? Social Forces, 78(4), 1349- 1378. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 Bulatao, R. A. (1984). Reducing Fertility in Developing Countries: A Review of Determinants and Policy Levers. World Bank Staff Working Papers (No. 680). Washington, DC: The World Bank. Bunwaree, S. (1999). Gender inequality: The Mauritian experience. In C. Heward & S. Bunwaree (Eds.), Gender, Education <&Development 135-154). London: Zed Books Ltd. Caldwell, J. C. (1978). A theory o f fertility: From high plateau to destabilization. Population and Development Review, 4(4), 553-577. Caldwell, J. C. (1982). The transition from familial to labour market production and the social implications. In E. A. Hammel (Ed.), Theory of Fertility Decline. London: Academic Press, Inc. Chemichovsky, D. (1985). Socioeconomic and demographic aspects o f school enrollment and attendance in rural Botswana. Economic Development and Cultural Change, 33(2), 319-332. Cleland, J. (1985). Marital fertility decline in developing countries: Theories and the evidence. In J. Cleland & J. Hobcraft (Eds.), Reproduct ive Change in Developing Countries: Evidence from the World Fertility Survey. Oxford: Oxford University Press. Csapo, M. (1981). Religious, social and economic factors hindering the education o f girls in Northern Nigeria. Comparative Education, 17(3), 311-319. Fierlbeck, K. (1997). Getting representation tight for women in development: Accountability, consent and the articulation o f women's interests. In A. M. Goetz (Ed.), Getting Institutions Right fo r Women in Development. New York: St. Martins Press, Inc. Filmer, D. (2004). If You Build It, Will They Come? School availability and school enrollment in 21 poor countries. World Bank Policy Research Working Papers (No. 3340). Washington, DC: The World Bank. Freedom House. (2004). Freedom in the World, http://www.ffeedomhouse.org. Galor, O., & Weil, D. N . (1996). The gender gap, fertility, and growth. The American Economic Renew, 86(3), 374-387. Gille, H. (1985). Policy implications. In J. Cleland & J. Hobcraft (Eds.), Reproduct ive Change in Developi ng Countri es: Evidence from the World Fertility Survey. Oxford: Oxford University Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 Goldin, C. (1995). The U-shaped female labor force function in economic development and economic history. In T. P. Schultz (Ed.), Investment in Women' s Human Capital. Chicago: University o f Chicago Press. Greenhalgh, S. (1985). Sexual stratification: The other side o f "growth with equity" in East Asia. Population and Development Review, 11(2), 265-314. Heward, C. (1999). Closing the gender gap? The informal sector in Pakistan. In C. Reward & S. Bunwaree (Eds.), Gender, Education & Development (pp. 203-217). London: Zed Books Ltd. Higgins, P. A., & Alderman, H. (1997). Labor and women's nutrition: The impact o f work effort and fertility on nutritional status in Ghana. The Journal o f Human Resources, 32(3), 577-595. Hill, M. A., & King, E. M. (1993). Women's education in developing countries: An overview. In E. M. King & M. A. Hill (Eds.), Women' s Education in Developing Countries. Baltimore, MD: The Johns Hopkins University Press. Jejeebhoy, S. R. (1995). Women's Education, Autonomy, and Reproductive Behavior: Experience from Developing Countries. Oxford: Clarendon Press. Kasarda, J. D., Billy, J. O. G., & West, K (1986). Status Enhancement and Tertility: Reproduct ive Responses to Social M obility and Educational Opportunity. Orlando, EL: Academic Press, Inc. Kaufman, D., & Wang, Y. (1995). Macroeconomic policies and project performance in the social sectors: A model o f human capital production and evidence from 1 .DC's. World Development, 23(5), 751-765. Knoke, D., Bohmstedt, G. W., & Mee, A. P. (2002). Statisticsfor Social D ata Analysis. Itasca, Illinois: F.E. Peacock Publishers. Lam, D., & Duryea, S. (1999). Effects o f schooling on fertility, labor supply, and investments in children, with evidence from Brazil. The Journal o f Human Resources, 34(1), 160-192. Lloyd, C., & Niemi, B. T. (1979). The Economics of Sex Differentials. New York: Columbia University Press. Lloyd, C., & Gage-Brandon, A. J. (1994). High fertility and chldren's schooling in Ghana: Sex differences in parental contributions and educational outcomes. Population Studies, 48(2), 293-306. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 Loebner, H., & Driver, E. D. (1973). Differential fertility in central India: A path analysis. Demography, 10(3), 329-350. Mason, K. O . (1993). The impact o f women's position on demographic change during the course o f development. In N . Federici, K. O. Mason & S. Sogner (Eds.), Women's Position and Demographic Change. Oxford: Clarendon Press. McMahon, W. (1999). Education and Development Measuring the Social Benefits. Oxford: Oxford University Press. Meyer, J. W., Ramirez, F. O., & Soysal, Y. N. (1992). World expansion o f mass education, 1870-1980. Sociology of Education, 65(2), 128-149. Nassirpour, M. (1985). The effect o f oil revenue on the fertility pattern in Iran. The Journal of Marriage and the Eamily, 47(3), 785-796. Parish, W. L., & Willis, R. J. (1993). Daughters, education and family budgets: Taiwan experiences. The Journal of Human Resources, 28(4), 863-898. Ram, R. (1982). Sex differences in the labor market outcomes o f education. In G. P. Kelly & C. M. Elliott (Eds.), Women's Education in the Third World: Comparative Perspectives. Albany: State University o f New York Press. Raynolds, L. T. (1998). Harnessing women's work: Restructuring agricultural and industrial labor forces in the Dominican Republic. Economic Geography, 74(2), 149-169. Rose, P., & Tembon, M. (1999). Girls and schooling in Ethiopia. In C. Heward & S. Bunwaree (Eds.), Gender, Education & Development (pp. 85-100). London: Zed Books Ltd. Schultz, T. W. (1960). Capital formation by education. The Journal of Political Economy, 68(6), 571-583. Schultz, T. P. (1988). Education investments and returns. In H. Chenery & T. N. Srinivasan (Eds.), Handbook of Development Economics (Vol. 1). Amsterdam: Elsevier Science Publishers, B.V. Sen, A K. (1990). Gender and cooperative conflicts. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Smock, A. C. (1981). Women's Education in Developing Countries: Opportunities and Outcomes. New York: Praeger Publishers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 8 Stinner, W. P. (1979). Modernization and family extension in the Philippines: A social demographic analysis. The Journal of Marriage and the Family, 42,161-168. Subbarao, K., & Raney, L. (1993). Social gains from female education: A cross-national study. World Bank Discussion Papers. Washington, DC: The World Bank. United Nations. (1999). WIST AT: Women's I indicators Database. New York: United Nations Publications. United Nations. (2000). United National Millennium Declaration, http://www.developmentgoals.org. World Bank. (1995). Toward G ender Equality: The Role for Public Policy. Washington, DC: World Bank. World Bank. (2002). World Development Indicators. Washington, DC: World Bank. Wynd, S. (1999). Education, schooling and fertility in Niger. In C. Heward & S. Bunwaree (Eds.), Gender, Education < & Development (pp. 101-116). London: Zed Books Ltd. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 4 Works Cited Adherents Organization. (2004). Majority Religions, h ttp ://www.adherents.com. Appiah, E. N., & McMahon, W. W. (2002). The social outcomes o f education and feedbacks on growth in Africa. The Journal of Development Studies, 38(4), 27-68. Bardhan, P. (1988). Alternative approaches to development economics. In H. Chenery & T. N. Srinivasan (Eds.), Handbook of Development Economics (Vol. 1). Amsterdam: Elsevier Science Publishers, B.V. Becker, G. (1975). Human CapitaL New York: National Bureau o f Economic Research. Becker, G. (1981). A Treatise on the l'amily. Cambridge, M A: Harvard University Press. Becker, G. (1985). Human capital, effort, and the sexual division o f labor. Journal of Tabor Economics, 3(1), S33-S58. Bellew, R. T., & King, E. M. (1993). Educating Women: Lessons from Experience. In E. M. King & M. A. Hill (Eds.), Women's Education in Developing Countries. Baltimore, MD: The Johns Hopkins University Press. Bendera, S. (1999). Promoting education for girls in Tanzania. In C. Heward & S. Bunwaree (Eds.), Gender, Education < & Development (pp. 117-132). London: Zed Books Ltd. Bils, M., & Klenow, P. J. (1998). Does schooling cause growth or the other way around? N BER Working Paper Series (No. 6393). Cambridge, MA: National Bureau o f Economic Research. Boserup, E. (1990). Economic change and the roles o f women. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Boserup, E. (1995). Obstacles to advancement o f women during development. In T. P. Schultz (Ed.), Investment in Women's Human CapitaL Chicago: University of Chicago Press. Bowman, M. J. (1984). An integrated framework for analysis o f the spread o f schooling in less developed countries. Comparative Education, 28(4), 563-583. Buchmann, C. (2000). Family structure, parental perceptions, and child labor in Kenya: W hat factors determine who is enrolled in school? Social Forces, 78(4), 1349- 1378. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 5 Bulatao, R. A. (1984). Reducing Fertility in Developing Countries: A Review o f Determinants and Policy Levers. World Bank Staff Working Papers (No. 680). Washington, DC: The World Bank. Bunwaree, S. (1999). Gender inequality. The Mauritian experience. In C. Heward & S. Bunwaree (Eds.), Gender, Education < & Development (pp. 135-154). London: Zed Books Ltd. Caldwell, J. C. (1978). A theory o f fertility: From high plateau to destabilization. Population and Development Review, 4(4), 553-577. Caldwell, J. C. (1982). The transition from familial to labour market production and the social implications. In E. A. I Iammel (Ed.), Theory of Fertility Decline. London: Academic Press, Inc. Chemichovsky, D. (1985). Socioeconomic and demographic aspects o f school enrollment and attendance in rural Botswana. Economic Development and Cultural Change, 33(2), 319-332. Cleland, J. (1985). Marital fertility decline in developing countries: Theories and the evidence. In J. Cleland & J. Hobcraft (Eds.), Reproductive Change in Developing Countries: Evidence from the World Fertility Survey. Oxford: Oxford University Press. Csapo, M. (1981). Religious, social and economic factors hindering the education o f girls in Northern Nigeria. Comparative Education, 17(3), 311-319. Fierlbeck, K. (1997). Getting representation right for women in development: Accountability, consent and the articulation o f women's interests. In A. M. Goetz (Ed.), Getting Institutions Right for Women in Development. New York: St. Martins Press, Inc. Filmer, D. (2004). If You Build It, Will They Come? School availability and school enrollment in 21 poor countries. World Bank Policy Research Working Papers (No. 3340). Washington, DC: The World Bank. Freedom House. (2004). Freedom in the World, http://www.freedomhouse.org. Galor, O., & Weil, D. N . (1996). The gender gap, fertility, and growth. The American Economic Review, 86(3), 374-387. Gille, H. (1985). Policy implications. In J. Cleland & J. Hobcraft (Eds.), Reproductive Change in Developing Countries: Evidence from the World FertiEty Survty. Oxford: Oxford University Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 Goldin, C. (1995). The U-shaped female labor force function in economic development and economic history. In T. P. Schultz (Ed.), Investment in Women's Human CapitaL Chicago: University o f Chicago Press. Greenhalgh, S. (1985). Sexual stratification: The other side o f "growth with equity" in East Asia. Population and Development Review, 11(2), 265-314. Heward, C. (1999). Closing the gender gap? The informal sector in Pakistan. In C. Heward & S. Bunwaree (Eds.), Gender, education < & Development (pp. 203-217). London: Zed Books Ltd. Higgins, P. A., & Alderman, H. (1997). Labor and women's nutrition: The impact o f work effort and fertility on nutritional status in Ghana. The Journal of Human Resources, 32(3), 577-595. Hill, M. A., & King, E. M. (1993). Women's education in developing countries: An overview. In E. M. King & M. A. Hill (Eds.), Women's education in Developing Countries. Baltimore, MD: The Johns Hopkins University Press. Jejeebhoy, S. R. (1995). Women's education, Autonomy, and Reproductive Behaiior experience from Developing Countries. Oxford: Clarendon Press. Kasarda, J. D., Billy, J. O. G., & West, K. (1986). Status enhancement and Fertility: Reproductive Responses to Social Mobility and educational Opportunity. Orlando, FL: Academic Press, Inc. Kaufman, D., & Wang, Y. (1995). Macroeconomic policies and project performance in the social sectors: A model o f human capital production and evidence from LDC's. World Development, 23(5), 751-765. Knoke, D., Bohmstedt, G. W., & Mee, A. P. (2002). Statistics for Social Data Analysis. Itasca, Illinois: F.E. Peacock Publishers. Lam, D., & Duryea, S. (1999). Effects o f schooling on fertility, labor supply, and investments in children, with evidence from Brazil. The Journal of Human Resources, 34(1), 160-192. Lloyd, C., & Niemi, B. T. (1979). The economics of Sex Differentials. New York: Columbia University Press. Lloyd, C., & Gage-Brandon, A. J. (1994). High fertility and chldren's schooling in Ghana: Sex differences in parental contributions and educational outcomes. Papulation Studies, 48(2), 293-306. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 7 Loebner, H., & Driver, E. D. (1973). Differential fertility in central India: A path analysis. Demography, 10(3), 329-350. Mason, K. O. (1993). The impact o f women's position on demographic change during the course o f development. In N. Federici, K. O. Mason & S. Sogner (Eds.), Women's " P osition and Demographic Change. Oxford: Clarendon Press. McMahon, W. (1999). Education and Development: Measuring the Social Benefits. Oxford: Oxford University Press. Meyer, J. W., Ramirez, F. O., & Soysal, Y. N. (1992). World expansion o f mass education, 1870-1980. Sociology of Education, 65(2), 128-149. Nassirpour, M. (1985). The effect o f oil revenue on the fertility pattern in Iran. The Journal of Marriage and the Family, 47(3), 785-796. Parish, W. L., & Willis, R. J. (1993). Daughters, education and family budgets: Taiwan experiences. The Journal of Human Resources, 28(4), 863-898. Ram, R. (1982). Sex differences in the labor market outcomes o f education. In G. P. Kelly & C. M. Elliott (Eds.), Women's Education in the Third World: Comparative Perspectives. Albany: State University o f New York Press. Raynolds, L. T. (1998). Harnessing women's work: Restructuring agricultural and industrial labor forces in the Dominican Republic. Economic Geography, 74(2), 149-169. Rose, P., & Tembon, M. (1999). Girls and schooling in Ethiopia. In C. Heward & S. Bunwaree (Eds.), Gender, Education < & Development (pp. 85-100). London: Zed Books Ltd. Schultz, T. W. (1960). Capital formation by education. The Journal of Political Economy, 68(6), 571-583. Schultz, T. P. (1988). Education investments and returns. In H. Chenery & T. N. Srinivasan (Eds.), Handbook of Development Economics (Vol. 1). Amsterdam: Elsevier Science Publishers, B.V. Sen, A K. (1990). Gender and cooperative conflicts. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Smock, A. C. (1981). Women's Education in Developing Countries: Opportunities and Outcomes. New York: Praeger Publishers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 8 8tinner, W. P. (1979). Modernization and family extension in the Philippines: A social demographic analysis. The Journal of Marriage and theFamily, 42, 161-168. Subbarao, K., & Raney, L. (1993). Social gains from female education: A cross-national study. World Bank Discussion Papers. Washington, DC: The World Bank. United Nations. (1999). WISTAT: Women's Iindicators Database. New York: United Nations Publications. United Nations. (2000). United National Millennium Declaration, h ttp ://www.developmentgoals.org. World Bank. (1995). Toward Gender Equality: The Role for Public Policy. Washington, DC: World Bank. World Bank. (2002). World Development Indicators. Washington, DC: World Bank. Wynd, S. (1999). Education, schooling, and fertility in Niger. In C. Heward & S. Bunwaree (Eds.), Gender, Education < & Development (pp. 101-116). London: Zed Books Ltd. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Chapter 2 G rowth, G irls’ E d u cation , and F em ale Labor: A L ongitudinal A nalysis Sum m ary In Chapter 1, regression results showed no relationship between labor force expectations and girls’ access to secondary school. This study investigates further this unexpected finding. Despite theoretical links between girls’ education and female labor force participation, researchers have had difficulty identifying this relationship empirically. At first it was hypothesized that cultural factors such as religion and social structure were more important determinants o f female labor force participation than wealth or access to education. Eventually, empirical work showed a U-shaped relationship between wealth and female labor force participation. Although this relationship was identified using cross-sectional data, theorists speculated that women in developing countries leave the labor market with initial growth and return in later stages o f development. This study uses longitudinal data from 141 countries to compare the U-shaped labor force theory with the development path o f countries over time, controlling for education, labor market structure, and cultural factors. The results of fixed-effects time series models confirm the U-shaped female labor force participation curve. However, the initial negative effect o f growth is much smaller than previously estimated, and the negative effect o f growth is mediated by the positive effect o f increases in girls’ access to secondary school. The policy implication is that provision o f girls’ education can increase female labor supply as development progresses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 1 INTRO DUCTIO N Economic growth is expected to increase human capital in developing countries through increased investments in education and expanded labor market opportunities. For women in developing countries, where women have historically had low social and economic status, economic growth is expected to bring new opportunities to attend school and participate in labor markets. However, identifying the empirical relationship between economic growth and female employment has been a challenge for researchers. Unable to identify a linear relationship between economic growth and female employment, researchers hypothesized that female labor force participation is a function o f cultural variables such as social mobility (Semyonov, 1980) or religion (Youssef, 1974), rather than economic variables or the level o f schooling. Eventually empirical work by Pampel and Tanaka (1986), confirmed by Goldin (1994; 1995), identified a non-linear, U-shaped relationship between economic development and female labor force participation. Cross-sectional data suggest that female employment is relatively high in both low-income and high-income countries, with a large dip in middle-income countries. Although these studies used cross-sectional data, the implication is that as growth progresses in less developed countries, female labor force participation will decline with initial growth and then increase at later stages of development. Goldin (1994) provides historical evidence female labor force participation in the United States followed a U-shaped path, but with only cross-sectional evidence, it is unclear if today’s developing countries are following a similar path. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 The U-shaped female labor force participation curve is apparent in Figure 4, which plots the log o f G D P per capita against female economic participation in 2000 for more than 150 countries along with a fitted, quadratic regression line. Countries with low and high incomes have high female economic participation, while middle- income countries have the lowest female economic participation. However, many individual countries fall far from the fitted line, and the relationship does not appear to be very strong. Figure 4. Fem ale Econom ic Participation and N ational W ealth by Country O O o < x > C P © c a S a. o ■ e « s £2. O o oo o o o o 1 3 0. o tz o o O O «i £ OO o°o o oo c 1 0 . 10 tog(GDP per capita) Year; 1990 Sources; W orld Bank (2000); International Labor Organization (2000) The relationship between female labor force participation and economic growth is complicated by the intervening role o f female education. Theoretically, education should create labor force opportunities for women (Becker, 1985). Figure 5 illustrates the relationship between female economic participation and female secondary school enrollment. The fitted quadratic regression line provides evidence that a U-shaped Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 2 relationship also exists between these two variables. Importantly, countries with the lowest female economic participation do not have the lowest female education. O n the contrary, female economic participation appears to be relatively high in countries where fewer than 20 percent o f women access secondary school. Figure 5. Fem ale Econom ic Participation and Female Secondary Enrollment 60 f 40.' o k . o c © o © £ 2 0. £ £ O O ° o O 60 100 120 temate secondary enrofimerrt Source: W orld Bank (2000); IL O (2000) The U-shaped relationships described above have profound implications for development policy. If cross-sectional evidence describes a typical development path, periods o f economic growth may lead unexpectedly to reductions in female labor force participation. More disturbingly, investments in girls’ education that are intended to boost development outcomes, may lead to further reductions in the size o f the labor force. Only longitudinal analysis can determine if reductions in female labor force participation are a typical response to increases in wealth and female education. This study contributes to an understanding o f the relationship between female labor force participation and economic developing by integrating theories o f education, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 labor, and growth in empirical analysis o f international time-series data. First, previous cross-sectional models with a new consideration o f the time-lagged relation between education and work force participation. Next, time-series models are developed to determine if the relationship in cross-sectional results reflect a development path over time. The results support the existence o f a U-shaped relationship between wealth and female labor force participation. However, controlling for country-specific fixed- effects, the negative effect o f initial economic growth is small, and female schooling is only positively related to female labor force participation over time. If expansion o f female education accompanies economic growth, developing countries will probably see little or no decline in female labor force participation as growth progresses. 2 THEORETICAL BACKGROUND Theoretical support for a U-shaped female labor force participation curve comes primarily from Boserup’s (1970; 1990) theory o f women in development. In early stages o f development, economies shift from family agricultural enterprises to industrial wage labor. While women participate widely in family-based agricultural production, they may be unable to participate in industrial labor and industrial farming due to the physical demands or this type o f work, gender discrimination, and the domestic demands o f large families. Thus, mechanized farming and industrial expansion in early stages o f development cause women to leave the formal economic sector and focus on domestic production. A t later stages o f development, expansion o f the labor market to include a service sector creates new opportunities for women and female labor force employment increases, under the important condition that women Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 gain access to schooling. Pampel and Tanaka (1986) observed that these two key stages in development could create a U-shaped female labor force participation curve. Using pooled international data from 1965 and 1970, they demonstrate with OLS regression that a quadratic relationship between economic development, measured by energy use per capita, and female economic participation is significant, controlling for female access to education relative to males, fertility, occupational segregation, and state political institutions. Placing the U-shaped curve in the context o f family consumption decisions, Goldin (1994; 1995) hypothesized that a social stigma prevents women from working in industrial jobs when economies initially shift from agriculture to mechanized production. The stigma dictates that a husband should not allow his wife to engage in heavy labor. With increased wages for the husband, the wife is able to stay home and substitute home production for labor market participation. The upward portion o f the U-shaped curve occurs with a second shift in labor market structure away from industry to white-collar clerical jobs. White-collar jobs are less subject to the stigma against women’s work and offer wages sufficient to overcome the preference for women to stay at home. Using OLS regression, Goldin also finds a quadratic relationship between economic growth, measured by GDP per capita, and female economic participation using 1990 data. A shortcoming o f Goldin’s approach is that we must assume that the social stigma is a global norm that transcends cultural differences. An alternative is offered by Gal or & Weil (1996) who simply assume that women are disadvantaged because industrial employment requires superior physical strength. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 Implicitly, the theory o f the U-shaped curve depends on women gaining access to education in order to benefit the second shift away from heavy industry to white collar employment. Therefore, it is im portant to understand how female schooling changes during times o f economic growth and how this effect influences the relationship between economic growth and female labor force participation. Female schooling in developing countries has historically been low (Hill & King, 1993). Human capital theory holds that families are more likely to invest in education for daughters when there is the potential for future labor market returns (Schultz, 1960; Becker, 1975). When women focus on domestic production, mothers will educate their daughters in domestic skills through informal education at home, while sons receive labor market preparation through more formal schooling (Becker, 1985). A lack o f jobs that reward female education can create a cycle where girls receive little schooling and women are not able to enter educated professions (Lloyd & Niemi, 1979; Ram, 1982). From a human capital perspective, we would expect gains in female schooling to be accompanied by labor market opportunities for women. There is significant evidence that economic growth is associated with expanded educational opportunities for girls, regardless o f female labor force participation. Girls’ school enrollment and completion rates have increased significantly in the last thirty years, with many countries shrinking or closing gender gaps in educational access (Hill & King, 1993). A t the same time, labor market segmentation and wage discrimination by gender persists (Celle de Bowman, 2000; Tzannotos, 2003). Youssef (1974) identifies large increases in female education even in Middle Eastern countries where Islamic culture dictates that women Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 do not work outside the home. She argues that cultural objections to women’s economic activity are much stronger than obstacles to girls’ education. If female schooling increases during economic growth but female labor force participation decreases or remains very low, an alternative to the human capital approach is needed. Theorists offer three possible explanations for the independent increase in girls’ education. First, education for both sons and daughters is a normal good, with investments increasing as wealth increases, regardless o f future returns to schooling (Goldin, 1994; Galor & Weil, 1996). Second, countries may invest broadly in education for girls and boys to maintain political stability and support democratic institutions (Schultz, 1960; Wise & Darling-Hammond, 1983). Third, home production, rather than labor market productivity, may be the target o f female schooling in some countries. Case studies reveal school curricula and instructional techniques that reinforce women’s domestic role and inferiority to men (Smock, 1981; Rose & Tembon, 1999; Wynd, 1999). At the same time, schooling offers girls’ access to literacy, analytical thinking, knowledge o f hygiene and medication, money management skills, and other basic skills that have been shown to improve domestic conditions by reducing infant and child mortality and increasing a mother’s capacity to educate her children at home 0ejeebhoy, 1995). Theoretical development and empirical work provides strong support for the third hypothesis, which also supports a U-shaped relationship between education and female labor market participation. Lam and Duryea (1999) argue that a woman’s schooling has value for both household and labor market productivity. A t home, a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 mother’s education increases the quality o f children through improved health and education. With initial levels o f schooling women’s reservation wage rises faster than labor market wages if families get greater benefits from the production o f high quality children than from women’s wages. These authors hypothesize that with secondary schooling, women’s market wages exceed the reservation wage, and women enter the labor market. This theory implies that the relationship between female schooling and labor force participation is U-shaped. With initial levels o f schooling, women in less developed countries may leave the labor market to specialize in home production. With secondary schooling and above, women reenter the labor market when wages reward educational investments. Additional support for this theoretical approach comes from demographic research, which demonstrates that the primary education o f mothers can improve child health, but only secondary education will create opportunity costs o f child birth through labor market opportunities (Gillc, 1985; Mason, 1993; Jejeebhoy, 1995). Lam and Duryea find some empirical support for this hypothesis using household-level data in Brazil. Brazilian women with primary schooling have dramatically lower birthrates than uneducated women, but do not participate in the labor market at high rates despite increasing wages. With a secondary education, Brazilian women have both low fertility and high labor force participation. Behrman et al. (1999) found education in India increased for both males and females during the Green Revolution, a time when the labor market returns to human capital increased significantly for men but not at all for women. The authors speculate that women’s Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 home productivity in educating sons was valued to such an extent that female education was also an im portant com ponent o f economic growth despite women’s exclusion from labor markets. A similar study o f human capital returns in Pakistan also found that women’s education can accrue as benefits to male relatives even when women are excluded from the labor force (Fafchamps & Quisumbing, 1999). Case study evidence from Mexico also suggests that highly educated women marry highly educated men, which allows them to stay out o f a workforce where women’s opportunities and wages are limited by sex discrimination (Smock, 1981). This connection between women’s schooling and home productivity provides insight into a clear motivation to educate girls that is independent o f labor force returns to female education. The effects o f wealth and education on female labor force participation also depend on the structure o f labor markets and the nature women’s participation. Theories o f the U-shaped female labor force participation curve identify two key labor market transitions: first, from agricultural to industrial labor and second, from industrial to service labor. Goldin (1994) offers one empirical test o f the effect o f labor market structure that suggests that the combination o f female education and a white collar sector that is open to women increases female labor force participation. Beyond this initial test, there is surprisingly little quantitative work on the role o f labor market structure in female employment. There is, however, a growing body o f case-study evidence that conflicts with the theoretical assertion that women are excluded from the labor market by an expansion o f industrial labor. Standing (1989) argues that the switch from unionized male labor to lower-cost women’s labor is driving much o f the increase Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 in export production in developing countries. If this is true, increased female labor force participation is a result o f either new blue collar jobs in production or the displacement o f male workers in existing production jobs. Evidence from the Central America supports the assertion that growing female labor participation is not necessarily the result o f a new clerical sector. Instead, labor market segmentation and discrimination restrict women to low-paying, unprotected factory work as economic growth progresses (Raynolds, 1998). A final influence is women’s access to economic markets, which may be culturally determined. Both Islam and Catholicism are associated in theory with lower female labor force participation (Smock, 1981; Youssef, 1974) or delegation o f women to specific employment sectors (Nassar, 2003). Economic freedom to participate in labor markets may also be associated with women’s political freedom (Sen, 1996). Therefore, a more complete model o f female labor force participation includes economic growth and female education, as well as controls for labor market structure, religion, and women’s political rights. 3 M ETHODOLOGY Data & Variables The quality o f international longitudinal data has improved considerably to enable longitudinal studies o f international development. For this study, data were compiled from the World Bank’s World Development Indicators (WDI) (World Bank, 2002), United Nations Women’s Indicator Statistics (WISTAT) (United Nations, 1999), and the International Labour Organization’s (ILO) employment statistics (International Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 Labour Organisation, 2000). Full data are available for 141 developed and developing countries from 1970 to 2000. Because o f the intervals at which the ILO published employment data, longitudinal observations are available for the years 1990, 1995, and 2000. This is not a sufficient time-frame to see an individual country transition from low- to high-income, but individual countries at different stages o f development should experience the decreases and increases in female labor force participation described by theory. Selecting appropriate variables to operationalize the theoretical concepts o f labor force participation, education, and economic growth is an important issue that has been addressed in detail by previous authors (Pampel & Tanaka, 1986; Goldin, 1994; Goldin, 1995). This study attempts to build on previous discussions o f these issues. Female labor force participation is measured using the ILO’s definition o f economic participation. An individual is considered economically active if she is employed or seeking employment in the formal economy. This measure can be particularly problematic for wom en because they may not be counted as employed if they participate in family agricultural production (Pampel & Tanaka, 1986) or small family enterprises (Heward, 1999). The shortcomings o f ILO labor force data are balanced by the benefit o f the survey’s impressive international coverage. This study uses ILO measures o f female economic participation with the caveat that women’s labor force participation may be underreported, particularly in countries with a large informal sector. A second issue is selecting the appropriate group o f women for analysis. W omen’s labor force participation is interrupted by pregnancy and child care (Becker, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 1975). A reduction in female labor force participation is expected among women o f childbearing age. These issues are discussed in detail by Goldin (1994), who determines that labor force participation o f women after their childbearing years is the most appropriate measure o f long-term change because these women are more likely to hold permanent, stable positions. Wherever possible, this study uses disaggregated female economic participation for ages 45-59 as the primary dependent variable. However, longitudinal analysis uses overall female economic participation for ages 18-59, because there is not sufficient lagged data on the education o f women in the older age group. The first key independent variable is wealth. Like Goldin, this study uses the natural log o f G D P per capita to measure wealth. This variable was taken from the WDI which measure G D P per capita in constant 1995 US dollars. Pampel and Tanaka focus their study on the theoretical affects o f development and industrial structure rather than growth, using energy use per capita rather than income per capita. Energy use per capita combines the effects o f wealth and industrial development into one measure. For this study, industrial structure is operationalized separately from wealth in order to test the theoretical assertions concerning labor market structure. The labor market transitions identified by theory are represented by variables from the WDI measuring percent o f total employment in industry jobs and percent o f total employment in service jobs. The second key dependent variable is female education. Based on theories o f home productivity and labor market wages, female labor force participation is expected to decrease with primary schooling and increase with secondary schooling. There is no Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 2 way to identify increased secondary enrollment independent o f increased primary enrollment. This study uses gross female enrollment in secondary school as the single measure o f female education. For female secondary enrollment to increase, primary completion must also increase so this variable measures primary completion plus access to secondary education.1 It is the education o f women currently in the workforce that should be related to female labor market prospects. Unlike previous studies, this study uses lagged female secondary enrollment to reflect educational access during the time when the target age group o f women would have been in secondary school. For women aged 45-59 enrollment is lagged 30 years, and for women aged 18-59, enrollment is lagged 15 years. Additional independent variables measure the social context to reflect women’s access to labor markets. Religious norms are expected to discourage female labor force participation most strongly in Islamic countries, where female participation in economic markets may be discouraged. Catholic countries are expected to discourage female labor force participation to a lesser extent due to the focus on women’s domestic activity and high fertility (Youssef, 1974; Smock, 1981; Pillai & Wang, 1999). Goldin addresses this issue by excluding the Middle East from her analysis. For this study, controls are added in the form o f two dummy variables reflecting if a country has a majority Islamic or a majority Catholic citizens. These data are from the Adherents Organization, which documents international religious membership and participation (Adherents, 2004). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 In addition to religion, women’s political participation is expected to be associated with greater access to economic markets and thus greater female labor force participation. Data concerning political rights come from the Freedom House Freedom in the World Survey for 2000 (Freedom House, 2004). Freedom House gathers data on political rights and civil liberties from multiple sources including news reports, non governmental organizations, academic studies, and cite visits. These data include a ranking o f civil liberties on a scale from one to seven. Freedom House assigns a ranking o f one for the “most-free” countries and seven for the “least-free” countries. To facilitate interpretation o f regression coefficients, these rankings were reversed to assign higher values to countries with greater freedom. 2 Cross-Sectional M odels The first set o f estimations uses ordinary least squares (OLS) regression to identify the relationship between growth, education, and female economic participation using cross-sectional data from the year 2000. The first model tests female economic participation (FEP) as a function o f GD P per capita and female gross enrollment rates in secondary school (GER). Following Goldin, the dependent variables is economic participation rate o f women ages 45-59 who are above childbearing age. Secondary gross enrollment rates are lagged 30 years to reflect access to education for this group o f women. A U-shaped relationship for both G D P per capita and female secondary enrollment is tested by including quadratic terms in the regression. The U-shape would be confirmed by a negative coefficient for the linear term and a positive coefficient for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 the quadratic term. The full model tests the effect o f wealth and education, while controlling for the context o f development. Control variables includes dummy variables for majority Islamic and Catholic, the index o f civil liberties, and two measures o f labor market structure. FEP(4 5 .5 9 ) = a + Pi ln(GDP per capita) + P2 ln(GDP per capita)2 + p3 GER(t-3 0 ) (1) + P 4 GER2 ( t - 3 0 ) + e FEP( 4 5 - 5 9 ) = a + pi ln(GDP per capita) + p 2 ln(GDP per capita)2 + P 3 G E R <, 3 0 ; (2) + P 4 GER2( t -jq i + P s industry + p6 service + p v Islamic + P s Catholic + p 9 civil liberties + e Longitudinal M odels The cross-sectional models provide a snapshot o f the cross-country relationship between variables, but they do not explain how female labor force participation has changed over time. Longitudinal models provide insight into changes in variables across time. Longitudinal data are not available for a sufficient timeframe to use the economic participation rates o f older wom en and 30-year lagged education variables. Instead, female economic participation for ages 18-59 is the used as the dependent variable, and education is lagged 15 years to reflect the average educational access o f this group o f women. Religion variables are excluded from the longitudinal models because they do not vary across time. First, OLS models were run using the pooled, longitudinal data. The simple and full models are: FEP = o t + Pi ln(GDP per capita) + P 2 ln(GDP per capita)2 + Pj GER/m s; (3) + 3 i GER2 ( t 1 5 ) + e FEP = a + Pi ln(GDP per capita) + P2 ln(GDP per capita)2 + P3 GER(t-i5) (4) + Pt GER2 (, i5 ) + ps industry + p6 service + p ? civil liberties + e Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 The OLS estimation o f time series data introduces two problems. Including multiple observations from countries across time introduces bias from country-specific effects that do not vary across time. It is highly likely that unmeasured country-specific effects, such as cultural norms and attitudes towards women, influence economic growth and female access to schooling, as well as the female economic participation. The second problem is that longitudinal data introduces serial correlation because errors may be correlated within and across countries across time. The first problem is solved by using a fixed-effects model, which controls for country specific effects by estimating: Ja = + p’(xj) + % Where i identifies an individual country and o c , is the unobservable fixed effects for the rth county. The fixed-effect model can be rewritten in terms o f deviations from the group mean as: 7it-y. = P’& t- x .) + %t-% O r in terms o f the group mean as: v , = < * i+ + e , Where i • is the mean within i. Thus, the intercept in the fixed-effects model reflects the unobservable country-specific effects, while the coefficients measure the effect o f changes in the group means across time. In this case, the fixed-effect models that mirror the OLS models above are: FEP; = a, + Pi [ln(GDP per capita); ] + fa [ln(GDP per capita)2; ]-!- fh [GER; ] (5) + 3.i [GER2;] + e; FEP, = a, + 3i pn(GDP per capita); ] + $2 fln(GDP per capita)2 ; ] + fa [GER; ] (6) + fa [GER2 ;] +^s [industry;]-!- fa [service,] + fa [civil liberties;]-!- e; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 The second problem with pooled time-series data is the potential for serial correlation across time. To correct for serial correlation, the fixed-effects model is estimated using a generalized least squares estimation. This estimation reduces bias by adjusting for correlations across errors within groups with a maximum likelihood estimator (Greene, 1997). 4 RESULTS Summary Statistics Table 11 displays the 141 countries included in longitudinal analysis. Table 12 lists means and standard errors for all variables. Countries are identified as low, middle, or high income based on the World Bank’s income categorizations w ith income expressed in 1995 US dollars. Low income countries have a mean G D P per capita of $404, middle income countries have a mean G D P per capita o f $2,864, and upper income countries have a mean G D P per capita o f $24,840. Mean values for female economic participation variables support the hypothesis o f the U-shaped relationship between wealth and female economic participation. For women ages 45-59, low-income countries have a mean economic participation rate o f 69.1 percent, middle income countries have a mean rate o f 45.1 percent, and high-income countries have a rate o f 53.8 percent. For women ages 18-59 the effect remains, but is less pronounced. The mean economic participation rate o f women ages 18-59 is 40.3 percent in low-income countries, 31.7 percent in middle-income countries, and 39.1 percent in high-income countries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Table 11. Countries in Longitudinal R egression Analysis by Region Sub Saharan Africa South A m erica E urope E a st A sia (n=39) a 1 1 & I I ja. > = 2 5 ) Angola Argentina Albania Australia Benin Bolivia Austria Bangladesh Botswana Brazil Azerbaijan Brunei Burkina Faso Chile Belgium Cambodia Burundi Colombia Bulgaria China Cameroon Ecuador Croatia Fiji Cape Verde Guyana Czech Republic H ong Kong Central African Paraguay Denmark India Repub. Chad Peru Estonia Indonesia Comoros Uruguay Finland Japan Dem . Rep. o f Congo Venezuela France Republic o f Korea Rep. o f Congo Georgia Lao PD R Cote D ’Ivoire M iddle E a s t/ Greece Malaysia Ethiopia N orth Africa Hungary Maldives The Gambia ( « = « ) Iceland Mongolia Ghana Algeria Ireland Nepal Guinea Bahrain Italy N ew Zealand Guinea-Bissau Cyprus Kazakhstan Pakistan Kenya Egypt Kyrgyz Republic Papua N ew Guinea Lesotho Islamic Rep. o f Iran Luxembourg Philippines Liberia Israel FYR Macedonia Singapore Madagascar Jordan Malta Solomon Islands Malawi Kuwait Moldova Sri Lanka Mali Lebanon Netherlands Thailand Mauritania Morocco Norway Vietnam Mauritius Oman Poland Mozambique Saudi Arabia Portugal North A m erica/ Niger Syrian Arab Republic Romania Central Am erica Nigeria Tunisia Russian Federation (n=16) Rwanda United Arab Emirates Spain The Bahamas Senegal Sweden Barbados Sierra Leone Switzerland Belize Sudan Turkey Canada Swaziland United Kingdom Costa Rica Tanzania Uzbekistan Dominican Republic Togo E l Salvador Uganda Guatemala Zambia Haiti Zimbabwe Honduras Jamaica Mexico Nicaragua Panama Trinidad and Tobago United States Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 Table 12. Variable M eans by Incom e (standard errors in parentheses) V ariable L o w In c o m e M id d le In c o m e H ig h In c o m e n = 5 3 n = 5 4 n = 3 4 F em ale e co n o m ic p articip atio n (% o f to tal fem ale p o p u latio n ) ages 45-59 (WISTA'I) 69.1 (1.81) 45.1 (2.44) 53.8 (3.41) all women (WDI) 40.3 (1.05) 31.7 (1.53) 39.1 (1.53) F em ale seco n d ary sch o o l en ro llm en t (g ross % ) 1970 6.4 28.1 63.5 (1.89) (2.71) (3.70) 1985 24.4 53.3 88.4 (3.97) (3.44) (2.45) 2000 29.5 74.3 108.3 (3.09) (2.41) (4.43) L ab o r m a rk e t stru c tu re (% o f to ta l em p lo y m en t) % agriculture 62.3 (3.05) 19.7 (1.78) 4.93 (0.67) % industry 11 .1 (1.12) 26.6 (0.94) 27.9 (1.54) % service 24.9 (2.08) 53.0 (1-82) 66.9 (1.62) W ealth (US $) G D P per capita $404 (27.85) $2,864 (239.91) $24,840 (1968.26) P o litical F reed o m civil liberties 3.4 (0.15) 4.5 (0.20) 6.0 (0.26) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 Table 13. Variable M eans by Girls’ Secondary School Enrollment (standard errors in parentheses) Variable L ess than 50% n=50 50% to 80% n = 38 Greater than 80% n= 53 Fem ale eco n o m ic participation (% o f total fem ale population) ages 45-59 (WISTAT) all women (WDI) 6 6.7 (2.35) 38.4 (1.26) 4 2 .7 (3.00) 30.8 (1.82) 56.6 (2.48) 39.4 (1 .2 2 ) F em ale secondary school enrollm ent (gross %) 1970 1985 2 0 0 0 5.4 (0 .6 8 ) 15.3 (1.73) 23.2 (1.81) 2 7 .0 (3.22) 51.0 (3.57) 69.1 (1.33) 58.3 (3.54) 84.4 (2.79) 103.2 (2.95) Labor m arket structure (% o f total em ploym ent) % agriculture % industry % service 60.6 (3.57) 12.4 (1.35) 26.5 (2.34) 2 3 .2 (2.89) 23.9 (1.19) 51.3 (2.84) 11.7 (1.75) 27.3 (1.28) 60.1 (1.91) W ealth (US $) G D P per capita 1522 (72.50) $4,276 (1022.23) $15,700 (1969.56) political freedom civil liberties 3.4 (0.16) 4.2 (0.24) 5.7 (0 .2 1 ) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 Table 14. Variable M eans by Majority R eligion (standard errors in parentheses) Variable Islam ic C atholic O th e r/N o n=31 n=41 Majority R eligion n = 69 Fem ale e con om ic participation (% o f total fem ale population) ages 45-59 (WtSTAT) 4 8.0 (4.06) 47.6 (2.73) 65.5 (1.91) all women (WDI) 3 0.9 (1.98) 33.0 (1.37) 4 1.5 (0.99) F em ale secondary sch ool enrollm ent (gross %) 1970 13.2 36.1 3 1.4 (2.79) (4.08) (4.17) 1985 39.9 54.7 53.6 (5.67) (4.58) (4.62) 3000 5 6.2 73.6 65.2 (5.50) (5.13) (5.13) Labor m arket structure (% o f total em ploym ent) % agriculture 37.6 (4.87) 21.5 (3.64) 36.0 (3.77) % industry 2 2 .8 (2.72) 23.5 (1.25) 18.9 (1 .2 2 ) % service 3 8.2 54.6 44.0 (3.16) (2.76) (2 .8 6 ) W ealth (US $) G D P per capita $3,230 (881.19) $7,855 (1868.05) $8,674 (1562.56) Political Freedom civil liberties 3.0 (0.17) 5.2 (0 .2 2 ) 4.7 (0 .2 0 ) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 The mean values o f female secondary enrollment indicate that female access to education is much greater in wealthier countries. In 1970, female enrollment in secondary school averaged only 6.4 percent in low-income countries. Educational access improved to 24.4 percent in 1985, and 29.5 in 2000, but this remains low compared to wealthier countries. Female secondary enrollment in middle income countries has increased more substantially from a mean o f 28.1 percent in 1970, to 53.3 percent in 1985, and 72.9 percent in 2000. Secondary enrollment in upper income countries increased from a mean o f 63.5 percent in 1970, to 88.4 percent in 1985, and to full enrollment by 2000. As predicted by theory, labor markets structures emphasize agriculture employment in low-income countries, and industrial and service employment in middle and high-income countries. Low-income countries have a mean o f more than 60 percent o f employees in agriculture and only 24.9 percent in service. Upper income countries have a mean o f more than 60 percent o f employees in service and less than 5 percent in agriculture. The mean percent o f employment in industry increases from 11.1 percent in low-income countries, to 26.6 percent in middle-income countries, and 27.9 percent in high-income countries. Middle and upper-income countries have similar levels o f employment in industry, but upper-income countries have more service jobs and far less agriculture. The civil liberties index suggests that freedom increases with income. Upper- income countries have the highest average level of civil liberties with a mean of 6.0 on Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 2 the 7-point Freedom House scale. Middle-income countries have a mean civil liberties index o f 4.5, and low-income countries have a mean o f only 3.4. Table 13 displays mean values by the level o f female secondary enrollment in 2000. The first group includes countries with less than 50 percent gross enrollment. The second group includes countries with 50 to 80 percent gross enrollment. The final group includes countries with greater than 80 percent gross enrollment. A U-shaped relationship between education and labor market attachment is suggested by the mean values for female economic participation by enrollment level. For women ages 45-59, countries with low female schooling have a mean female economic participation o f 66.7 percent. Mean economic participation falls to 42.7 percent in countries in the middle, and increases to 56.6 percent in countries high female schooling. A similar but less dramatic effect occurs for women ages 18-59. Countries with high female secondary enrollment in 2000 have also experienced the m ost dramatic growth in female enrollments since 1970. While countries with low enrollment in 2000 increased from an average o f only 5.4 percent in 1970 to 23.2 percent in 2000, countries with high enrollment increased from 58.3 percent in 1970 to more than 100 percent in 2000. Countries in the middle group also had impressive increases from only 27 percent in 1970 to almost 70 percent in 2000. Female secondary enrollment is also associated with development. Countries with low female secondary enrollment are primarily agricultural with a small industrial sector and mid-size service sector. Countries in the middle have less agriculture, more industry, and more than 50 percent o f employees in the service sector. Countries with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 high female secondary enrollment have very litde agriculture, a mid-size industrial sector, and a large service sector. Income in countries with low female enrollment is very low, averaging only $522 in G D P per capita. Countries in the middle have a higher average G D P per capita o f $4,276. Countries with high female enrollment also have high incomes, with an average G D P per capita o f $15,700. Civil liberties are also highest in countries with high female education and lowest in countries with low female education. Table 14 displays means by religion. In 31 countries, the majority identify as Islamic. In 41 countries, the majority identify as Catholic. In the remaining 69 countries, the majority identifies as other religions or there is no majority religion. Islamic countries have the lowest levels o f female economic participation at all ages, but the differences are not as dramatic as we might expect from anecdotal evidence. For women o f all ages, mean economic participation rates are 30.9 percent in Islamic countries, 33 percent in Catholic countries, and 41.5 percent in other countries. Female secondary enrollment is consistently highest in Catholic countries in 1970, 1985, and 2000. Islamic countries continue to have low f emale secondary enrollment with average enrollment o f 56.2 percent in 2000, compared with 73.6 percent in Catholic countries, and 65.2 percent in other countries. Income and labor markets also differ by religion. Mean G D P per capita is only $3,220 in Islamic countries, compared with $7,855 in Catholic countries, and $8,674 in other countries. The labor market structure in Catholic countries may be the most amenable to educated, female employment as these countries have more than 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 4 percent o f employment in service industries. Islamic and other countries rely more heavily on agricultural employment. Overall, women are statistically m ost disadvantaged in Islamic countries in terms o f income, educational access, civil liberties, and an emphasis on industrial and agricultural employment. Women in Catholic countries enjoy the highest level o f education and civil liberties, but women in countries with other or no majority religion enjoy the highest income levels. Tim e Series Graphs The time-series graphs in Figure 6 illustrate the changing relationship between female labor force participation and economic growth over time. If low-incomes countries were moving from high labor force participation to low labor force participation as wealth increases, we would expect the curve to become more pronounced as economic growth professes, but Figure 3 illustrates that the U-shaped relationship is weakening over time. Figure 7 illustrates that a similar change is taking place in the relationship between female labor force participation and female education. Once again, the U- shaped relationship is very strong in 1970, but almost non-existent by 2000. Figures 6 and 7 provide evidence that the cross-sectional relationship between growth, education, and female labor force participation does not reflect the longitudinal path that developing countries are experiencing. The disintegration o f the U-shaped curves suggests that three is not one clear path through which development influences female economic participation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 Figure 6. Fem ale E conom ic Participation and N ational Wealth over Tim e 1970 60, S O - ? 40 § 8 2 0 E & 0 1980 logCGCPpertafrta) I S t e S % i 1990 2000 O J o O <P o o'-'o© S ■ a | © o o o <*> d 5 i fcgfGOP per capita) O o E § 8 E * log(GDP per capte) Sources: W orld Rank (2000); International L abor O rganization (2000) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 Figure 7. Fem ale Econom ic Participation and Fem ale Secondary Enrollment over Tim e 1970 1980 s s f Indesand syisiiA iieit 1990 2000 i i s i I e g 8 4 1 e £ lertate secondary enrollment S -a f & o £ g 8 " B E £ temsle secorrtary enrollment Sources: W orld B ank (2000); International Labor O rganization (2000) C ross-Sectional Regression Results The cross-sectional regression models were tested using data from the year 2000. The dependent variable is the rate o f economic participation o f women ages 45- 59. Female secondary enrollment is lagged 30 years to reflect educational access for this group. Because 1970 enrollment data is not available for all countries, the cross- sectional regressions include a subset o f 119 countries. Countries included in these regression and listed in Table 15, and regression results are displayed in Table 16. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 The results for Model 1 the support hypothesized U-shaped relationships for both G D P per capita and secondary school enrollment and the coefficients are statistically significant. Both schooling and wealth have independent U-shaped effects on FEP. The coefficients on the simple terms are negative and significant, and coefficients for quadratic terms are positive and significant. Model 2 adds controls for labor market structure, religion, and political freedom. With these added control variables, the U-shaped relationship between wealth and female economic participation remains, although the quadratic term is no longer statistically significant. The U-shaped relationship between female schooling and female economic participation also remains, and both coefficients are statistically significant. As predicted by Goldin, the percent o f employment in the industrial sector has a negative effect on female economic participation (p<0.10). Contrary to theoretical predictions, the coefficient on the percent o f employment in the service sector is also negative, but not significant. The variables reflecting social and cultural context are all significant in the directions predicted by theory. The excluded group in Model 2 is countries that are neither majority Islamic nor majority Catholic. The significant negative coefficients on the religion dummy variables indicate that female labor force participation is approximately 9.1 percent lower in Islamic countries (p<0.05) and 8.6 percent lower in Catholic countries (p<0.01) compared with countries with other or no majority religion. Political freedom is positively associated with female economic participation. An increase o f one point on the Freedom House scale of civil liberties, is associated with an increase o f approximately 3.5 percent in female economic participation (p<0.01). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 These results provide limited support some for theories behind the U-shaped female labor force participation curve. In cross-sectional analysis female economic participation appears to decrease with increases in both income and female schooling and to increase when income and female schooling progress to higher levels, as seen in developed countries. The significance o f female schooling, independent o f wealth, suggests that schooling has as independent effect on women’s economic participation and it not simply the correlation between wealth and education that influences the home productivity effect described by Lam and Duryea (1999). The expansion o f industry is associated with lower female employment, as predicted by Goldin, but the expansion o f the service sector is not associated higher female employment in this estimation. What appears to be more important is the prevalent religious context. A strong presence o f Catholic or Islamic norms lowers female labor force participation by almost 10 percent. It is notable that the addition o f these control variables has a greater effect on the coefficient on the wealth variables than the coefficient on the female school enrollment variables. This may support Youssef s hypothesis that religious obstacles to women’s participation in economic markets are more difficult to overcome than religious obstacles to girls’ education. Finally, these regressions confirm that political freedom is positively associated with women’s access to economic markets through employment. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 Table 15. Countries in Cross-Sectional Analysis by R egion Sub Saharan Africa South A m erica E urope E a st Asia (n=39) <n=11) (o=24) (n=20) Angola Argentina Albania Australia Benin Bolivia Austria Cambodia Botswana Brazil Belgium China Burkina Faso Chile Bulgaria Fiji Burundi Colombia D enm ark H ong Kong Cameroon Ecuador Finland India Cape Verde Guyana France Indonesia Central African Rep. Paraguay Greece Japan Chad Peru Hungary Republic o f Korea Comoros Uruguay Iceland Lao P D R Dem . Rep. o f C ongo Venezuela Ireland Malaysia Rep. o f Congo Italy M ongolia C ote D ’Ivoire M iddle E a s t/ Luxembourg N epal Ethiopia N orth Africa Malta N ew Zealand T h e Gambia (•>=14) N etherlands Pakistan G hana Algeria Norway Papua N ew Guinea G uinea Bahrain Poland Philippines Guinea-Bissau Egypt Portugal Singapore Kenya Islamic Rep. o f Iran Romania Sri Lanka Lesotho Israel Spain Thailand Liberia Jordan Sweden M adagascar Kuwait Switzerland N orth A m erica/ Malawi Lebanon Turkey Central Am erica Mali M orocco U nited K ingdom (0=11) Mauritania O m an Barbados Mauritius Saudi Arabia Canada Mozambique Syrian Arab Republic C osta Rica N iger Tunisia E l Salvador Nigeria U nited Arab Em irates Guatem ala Rwanda H onduras Senegal Jam aica Sierra Leone M exico Sudan Nicaragua Swaziland Panam a Tanzania Trinidad and Tobago Togo Uganda Zam bia Zim babwe Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 16. Cross-Sectional R egression R esults D e p e n d e n t V ariable M o d e l 1 M o d el 2 female economic participation, coeff. t-ratio coeff. t-ratio age 45-59 Constant 20 8 .4 1 7 5 63*** 147.752 4.17*** W ealth log (GDP per capita) -3 4 .8 2 4 -3 49*** -1 9 .5 5 7 - 2 .0 1 ** [log (G D P p er capita)] — squared 1.824 2 3 9 *** 0 .9 8 9 1.65 F em ale E d u ca tio n secondary enrollm ent, , ^ -0 .4 9 4 -2 o i* * -0 .5 0 0 -2.06** (secondary enroHm ent)^.^ — 0.0 0 9 3 4 5 *** 0.0 0 8 2 9 9 *** squared L a b o r M ark et S tructure % industrial sector -0 .3 1 4 -1.76* % service sector -0 .1 4 6 -1 .3 8 R eligion majority Islamic -9 .1 0 2 -2.18** m ajority Catholic -8 .6 3 5 - 2 3 4 *** P o litical F reed o m civil liberties 3.519 2 83*** R-squared 0.423 0.5 5 6 N o. o f Observations 119 119 *p<0.10, **p<0.05, ***p<0.01 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 Longitudinal Regression R esults Longitudinal analysis can determine if this cross-sectional relationship is also important for individual countries over time. The dependent variable is female economic participation for ages 18-59, and secondary school enrollment is lagged 15 years. The data set includes 141 countries, listed in Table 11, with a mean o f 2.8 observations per country. Tables 17 and 18 display the results for the cross-sectional, OLS longitudinal, and fixed-effects models. Results for the reduced-form model are displayed in Table 17. First, the cross- sectional models are replicated using economic participation o f women ages 18-59 as the dependent variable. These results closely resemble the results above for women ages 45-59. The U-shaped relationship between GDP per capita and female economic participation is confirmed, as well as the U-shaped relationship between secondary enrollment and female economic participation. Next, the OLS model is replicated with longitudinal data from 1990, 1995, and 2000. These results also suggest independent U- shaped relationships exist between both G D P per capita and female secondary enrollment and female economic participation. When the longitudinal model is estimated as a fixed-effect model, the reduced-form fixed-effects model confirms the predicted U-shaped relationship between G D P per capita and female economic participation. However, the negative coefficient on G D P per capita is much smaller in the cross-sectional model than the coefficient in the longitudinal OLS model. This suggests that the OLS models overestimate the reduction in female labor force participation due to increases in national wealth. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 2 Table 17. C ross-Sectional and Longitudinal R egression R esults, Reduced Form M odel M o d e l 1 D e p e n d e n t V ariab le O L S c ro ss-sectio n al O L S L o n g itu d in al F ixed -E ffects female econom ic participation coeff. t-ratio coeff. t-ratio c o e ff t-ratio Constant 90 .3 5 3 4 57*** 111.231 9 23*** 33 .2 3 7 4 9 4 *** W ealth log (G D P p er capita) -1 2 .5 2 5 -2.32** -1 8 .6 9 6 -5.37*** -5 .0 0 9 -2.36*** [log (G D P p er capita)]- squared 0 .7 2 7 2.18** 1.096 5 3 7 *** 0.578 ^ -|4*** F e m ale E d u c a tio n secondary enrollm ent (t-15 y ears) -0 .3 2 7 -2 .8 8 *** -0 .2 5 2 -3 77*** 0.146 6.35*** secondary enrollm ent (t-15 years) — squared 0 .0 0 4 3 0.0 0 3 5 90*** -0 .0 0 0 4 -2.29** R-squared 0.271 0.2 9 7 0.478 N o. o f Observations** 141 389 3 8 9 N o. o f Countries 141 141 141 *p<0.10, **p<0.05, ***p<0.01 + R-squared for fixed-effects m odels is within country variation + + M ean observations p e r country is 2-8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 Table 18. Cross-Sectional and Longitudinal R egression R esults, Full M odel M o d el 2 D e p e n d e n t V ariable O L S cro ss-sectio n al O L S L o n g itu d in al F ix ed -E ffects female economic participation coeff. t-ratio coeff. t-ratio coeff. t-ratio Constant W ealth log (GD P per capita) [log (G D P per capita)]- squared F em ale E d u ca tio n secondary enrollm ent (t-15 years) secondary enrollm ent (t-15 years) — squared L a b o r M ark et S tru ctu re % industrial sector % service sector Political R ig h ts civil liberties 77.43 3.98*** -9 .9 7 4 -1.88* 0.5 8 9 1.81* -0 .2 2 8 -2.00** 0.003 3.04*** -0 .0 7 5 -0 .8 6 -0 .1 4 2 -2.48** 1.832 3.19*** 9 4 .2 6 8 7 .9 6 *** -1 4 .6 5 4 -4.53*** 0 .8 9 0 4.49*** -0 .1 6 6 -2.53** 0.0 0 3 4.61*** -0.071 -1.31 -0 .1 8 0 -5.11*** 1.624 4.81*** 2 9 .3 6 2 4.28*** -3 .8 4 8 -2.04** 0.4 6 3 3.28*** 0.1 4 8 6.46*** -0 .0 0 0 6 -2.85*** -0 .0 4 2 -1 .0 2 0.0 8 3 2.63*** -0 .1 0 2 -0 .8 5 R-squared 0.350 0.382 0.509 N o. o f Observations ‘ 1 141 389 3 8 9 N o. o f Countries 141 141 141 *p< 0.10, **p<0.05, ***p<0.01 + R-squared for fixed-effects models is within country variation + + M ean observations per country is 2.8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 More importantly, the U-shaped relationship between female secondary enrollment and female economic participation disappears in the fixed-effects model. The direction o f the coefficients on female secondary enrollment and its quadratic term is reversed: the coefficient on secondary enrollment is now positive and significant (p<0.01), while the coefficient on the quadratic enrollment term is negative and significant, but also very small (p<0.05). This relationship remains in the fixed-effects estimation o f the full model that controls for labor market structure and civil liberties. These results indicate that the negative relationship between female schooling and female economic participation in OLS models is due to bias in the estimation and does not reflect the true effect within countries over time. Instead, female schooling appears to have a positive effect on female labor force participation. The fixed-effects estimation o f the full model also differs from the OLS models in the effects o f labor market structure. In the OLS models, both industrial employment and service employment have negative coefficients, although only the coefficient on service employment is statistically significant. This runs contrary to theories o f economic growth that associate industry with decreased female employment and growth o f the service sector with increased female employment. In the fixed- model, the results for labor market structure do support the predictions o f theory. Industrial employment has a negative effect on female economic participation and service employment has a positive effect on female employment, although only the latter effect is statistically significant (p<0.01). The insignificance o f industrial employment in the fixed-effects models may reflect the increased inclusion o f women in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 export production in many developing countries, as described by Standing (1989) and Raynolds (1998). Finally, the coefficients on civil liberties are positive and significant in the OLS models and negative and not significant in the fixed-effects model. This suggests that the positive effect o f civil liberties on female economic participation may be part o f country-specific effects. The civil liberties index does not directly measure free access to economic markets as much as political access through free press, transparent government, and civil society. It is likely that women’s access to economic markets is correlated with a political ideology that promotes civil liberties, but that an increase in civic participation does not necessary mean that labor markets are open to women. 5 DISCUSSION As labor markets in developing countries change, it is important to understand the changing roles o f women in labor markets. It is also important to predict the potential effects o f investments in female schooling on labor markets. From previous cross-sectional research, we might expect female workforce participation to fall initial periods o f economic growth. The cross-sectional results in this study support this expectation, as well as the expectation that women may leave the labor force as a result o f early increases in access to schooling. Cross-sectional analysis, confirmed in this paper, can lead to the misleading conclusion that early stages o f economic development lead women to focus on home production, while men specialize in industrial labor market production. This is close to the expectation of Becker’s human capital Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 approach, which predicts that an efficient household division o f labor is one where men specialize in labor market production, while women specialize in domestic production. However, longitudinal analysis with fixed-effects models suggests that while a U-shaped female labor force participation curve exists in a snapshot, developing countries do not necessarily experience a decline in female labor force participation as national incomes and female education increase. Although the U-shaped relationship between economic growth and female economic participation remains in the fixed- effects models, the magnitude o f the decline in female economic participation as incomes rise is much smaller. At the same time, female access to schooling has only positive effects on female economic participation without an initial period where women leave the labor market to focus on domestic production. Countries where women have historically specialized in domestic production may see more educated women out o f the labor market, but this does not mean that increases in schooling will result in fewer women in the labor market overall. With the addition o f the positive effect o f female education, which typically increases with economic growth, many countries may see little or no drop in female economic participation as wealth increases. The results o f this study also provide insight into the relationship between labor market structure and female labor participation. Industrial employment does not have a statistically significant negative effect on female economic participation in any o f the longitudinal models. This conflicts with the assumption that industrial employment is too physically demanding for women, and supports the assertion that female employment is integral to industrial production o f exports. The second labor market Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 shift to white collar service jobs does bring women into the labor market, as predicted by Boserup, Goldin, and others. 6 CONCLUSION Historically, men and women have not had equal access to education. A positive result o f economic growth has been increased access schooling for girls, with complex implications for women and markets. Researchers have typically had difficulty empirically linking educational investments in men with labor market productivity and economic growth (Pritchett, 1996; Griliches, 1997; Gundlach, 2001). The issue is even more complicated for women, as women divide their time between childcare and work throughout their lives. In addition, women’s access to labor markets can be limited by cultural constraints, which further complicates our expectations about the changing labor market behavior in the face o f economic growth. Theories o f economic growth have recognized the importance o f the initial endowment o f physical capital, as well as human capital, in the rate o f economic growth (Barro, 1991; Mankiw, Romer & Weil, 1992). The longitudinal results here indicate that both initial endowments o f human capital for women and the initial level o f female access to labor markets may also be relevant in how growth influences women’s contributions to the labor market. Endowment effects concerning female education may be a function many factors including labor market structure, egalitarian culture, and family structure. The response o f women to increases in wealth and education is an important factor is public policy for developing countries. Particularly, it is important to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 understand how female education has external benefits for families — such as reduced fertility and improved child health — and labor markets — through expansion o f the skilled workforce. The results o f this study suggest that initial economic growth is accompanied by a decline in female economic participation. However, longitudinal models that correct for country-specific effects suggest that this decline is not as big as previously thought. In addition, longitudinal analysis shows that female schooling can increase female economic participation. These results suggest that countries with different initial levels o f female schooling and female labor force participation will travel different paths to development. The U-shaped curve o f female labor force participation may disappear over time as more and more countries take a non-traditional path to development. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 NOTES 1 Pampel & Tanaka argue that the ratio o f females to males in secondary is the correct measure for estimation because it measures female skills relative to males. However, high female enrollment compared to males can result from civil wars and other domestic problems that take boys out o f school, so absolute female enrollment is preferred for this study. 2 This strategy is also used by McMahon (1999) who tests the influence of education on social outcomes, including the spread o f democracy. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 WORKS CITED Adherents Organization. (2004). Majority Religions, http://www.adherents.com. Barro, R. J. (1991). Economic growth in a cross section o f countries. The Quarterly Journal of Economics, 106(2), 407-443. Becker, G. (1975). Human Capital New York: National Bureau o f Economic Research. Becker, G. (1985). Human capital, effort, and the sexual division o f labor. Journal of Labor Economics, 3(1), S33-S58. Behrman, J. R., Foster, A. D., Rosenzweig, M. R., & Vashishtha, P. (1999). Women's schooling, home teaching, and economic growth. The Journal of Political Economy, 107(4), 682-714. Boserup, E. (1970). Women's Role in Economic Development. In. New York: St. Martin's Press. Boserup, E. (1990). Economic change and the roles o f women. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Celle de Bowman, O. (2000). Peruvian female industrialists and the globalization project: Deindustrialization and women's independence. Gender and Society, 14(4), 540-559. Fafchamps, M., & Quisumbing, A. R. (1999). Human capital, productivity, and labor allocation in rural Pakistan. The Journal of Human Resources, 34(2), 369-406. Freedom House. (2004). Freedom in the World, http://www.freedomhouse.org Galor, O., & Weil, D. N. (1996). The gender gap, fertility, and growth. The American Economic Review, 86(3), 374-387. Gille, H. (1985). Policy implications. In J. Cleland & J. Hobcraft (Eds.), Reproductive Change in Developing Countries: Evidence from the World Fertility Survey. Oxford: Oxford University Press. Goldin, C. (1994). The U-shaped female labor force function in economic development and economic history. N BER World Papers (No. 4707). Cambridge, MA: National Bureau o f Economic Research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 Goldin, C. (1995). The U-shaped female labor force function in economic development and economic history. In T. P. Schultz (Ed.), Investment in Women's Human Capital. Chicago: University o f Chicago Press. Greene, W. H. (1997). Econometric Analysis (3rd ed.). New Jersey: Prentice-Hall, Inc. Griliches, Z. (1997). Education, human capital, and growth: A personal perspective. Journal of Labor Economics, 15(1), S330-334. Gundlach, E. (2001). Education and economic development: An empirical perspective. Journal of Economic Developmentff), 37-59. Heward, C. (1999). Closing the gender gap? The informal sector in Pakistan. In C. Heward & S. Bunwaree (Eds.), Gender, Education & Development (pp. 203-217). London: Zed Books Ltd. Hill, M. A , & King, E. M. (1993). Women's education in developing countries: An overview. In E. M. King & M. A. Hill (Eds.), Women's Education in Developing Countries. Baltimore, MD: The Johns Hopkins University Press. International Labour Organisation. (2000). Yearbook of Labour Statistics. Geneva: International Labour Office. Jejeebhoy, S. R. (1995). Women's Education, Autonomy, and Reproductive Behavior. Experience from Developing Countries. Oxford: Clarendon Press. Lam, D., & Duryea, S. (1999). Effects o f schooling on fertility, labor supply, and investments in children, with evidence from Brazil. The Journal of Human Resources, 34(1), 160-192. Lloyd, C., & Niemi, B. T. (1979). The Economics of Sex Differentials. New York: Columbia University Press. Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A contribution to the empirics o f economic growth. The Quarterly Journal of Economics, 107(2), 407-437. Mason, K. O. (1993). The impact o f women's position on demographic change during the course o f development. In N. Federici, K. O. Mason & S. Sogner (Eds.), Women's Position and Demographic Change. Oxford: Clarendon Press. Nassar, H. (2003). Egypt: Structural adjustment and women's employment. In E. A. Doumato & M. P. Posusney (Eds.), Women and Globalisation in the Arab Middle East. Boulder, CO: Lynne Rienner Publishers, Inc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 2 Pampel, F. C., & Tanaka, K. (1986). Economic development and female labor force participation: A reconsideration. Social Forces, 64(3), 599-619. Pillai, V. K., & Wang, G.-z. (1999). Women's Reproductive Rights in Developing Countries. Brookfield, Vermont: Ashgate Publishing Company. Pritchett, L. (1996). Where Has All the Education Gone? World Bank Policy Research Working Papers (No. 1581). Washington, D.C.: The Work Bank. Ram, R (1982). Sex differences in the labor market outcomes o f education. In G. P. Kelly & C. M. Elliott (Eds.), Women's Education in the Third World: Comparative Perspectives. Albany: State University o f New York Press. Raynolds, L. T. (1998). Harnessing women's work: Restructuring agricultural and industrial labor forces in the Dominican Republic. Economic Geography, 74(2), 149-169. Rose, P., & Tembon, M. (1999). Girls and schooling in Ethiopia. In C. Heward & S. Bunwaree (Eds.), Gender, Education Development (pp. 85-100). London: Zed Books Ltd. Schultz, T. W. (1960). Capital formation by education. The Journal ofPoliticalEconomy, 68(6), 571-583. Semyonov, M. (1980). The social context o f women's labor force participation: A comparative analysis. The American Journal of Sociology, 86(3), 534-550. Sen, G. (1996). Gender, markets and states: A selective review and research agenda. World Development, 24(5), 821-829. Smock, A C. (1981). Women’ s Education in Developing Countries: Opportunities and Outcomes. New York: Praeger Publishers. Standing, G. (1989). Global feminization through flexible labor. World Development, 17(7), 1077-1095. Tzannatos, Z., & Kaur, I. (2003). Women in the MEN A labor market. In E. A. Doumato & M. P. Posusney (Eds.), Women and Globalisation in the Arab Middle East. Boulder, CO: Lynne Rienner Publishers, Inc. United Nations. (1999). WISTAT: Women's Indicators Database. New York: United Nations Publications. World Bank. (2002). World Development Indicators. Washington, DC: World Bank. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 Wynd, S. (1999). Education, schooling, and fertility in Niger. In C. Heward & S. Bunwaree (Eds.), Gender, Education & Development (pp. 101-116). London: Zed Books Ltd. Youssef, N. H. (1974). Women and Work in Developing Societies. Westport, CT: Greenwood Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 4 Chapter 3 E q uity & G overnance in Girls’ E d ucation Sum m ary This chapter focuses on implementation o f education projects for girls in developing countries. Developing countries face the challenge o f expanding education with limited resources. At the same time, promoting gender equity is vital to achieving basic development goals. Appropriate governance strategies are needed to provide education in a way that is cost-effective and inclusive. Privatization, decentralization, and citizen participation are three strategies that are promoted by the World Bank to control costs and improve transparency in education. Because the demand for girls’ education is low, strategies that depend on local resources and participation can increase inequities in education. Theory suggests that the central government should maintain authority to enforce equity requirements as schooling expands. Empirical analysis o f World Bank education projects reveals that most countries with low girls’ education are aware of the importance o f girls’ education. However, governance strategies vary with a level awareness o f gender issues and the demand side o f education consumption. Implications for expansion of girls’ education are discussed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 1 INTRODUCTION A central challenge for international development projects is to balance efficiency and equity goals. D uring the 1990’s the World Bank and International Monetary Fund attempted to facilitate economic growth through Structural Adjustment Programs (SAPs) which focused on reducing the size and scope o f government. One strategy to reduce government spending was to charge fees for attendance at public school. Critics argue that this and other components o f SAPs had profound effects on equity (Bayliss & Fine, 1998; Khan, 2002; Sender, 2002; Stiglitz, 2002). Gender equity was a specific concern, and empirical evidence suggests that girls’ were less likely to attend school when fees were charged (Nasssar, 2003). This can be a serious setback for developing countries, as female education is linked to positive development outcomes including reduced fertility, improved health outcomes, and increased participation o f women in economic and political life. In the late 1990’s, the World Bank adopted a new development strategy called the Comprehensive Development Framework (CDF). The CDF focuses on more flexible implementation o f governance strategies that are effective within the local context (Hanna & Agarwala, 2002). Governance strategies advocated by the CD F including privatization, decentralization, and citizen participation to facilitate markets, increase government responsiveness, and increase transparency. Unlike SAPs, CDF strategies are implemented as an integrated component o f World Bank programs, including education projects. Theoretical development in governance and gender suggests that the governance strategies promoted by the CDF can also have profound Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 implications for equity. This chapter examines the implementation o f privatization, decentralization, and citizen participation in World Bank education projects to determine the potential equity consequences. Cost-effectiveness is particularly important in the provision o f education. To support economic growth and development, governments struggle to expand school enrollment in order to build human capital, political stability, and nationhood. Strategies to expand enrollment and improve quality, while reducing costs, involve replacing direct provision by the central government with private markets (Colclough, 1996) or increasing community participation in decentralized education systems (Parry, 1997). These strategies depend on families and communities to contribute to the costs o f education, thus enabling educational expansion with limited allocations o f national resources. Proponents o f cost-sharing strategies argue that efficiency is increased because individuals and communities also benefit directly from formal schooling (Rondinelli et al., 1983). Critics argue that it is a grave mistake to assume that the individual or local demand for education is strong enough to support universal education in poor and rural areas (Jones, 1998; Maclure, 1994). Poverty, high opportunity costs, and low returns to education mean that the parents’ ability to pay for schooling is far below the broad social benefits o f education (Maclure, 1994). Furthermore, individual demand and community support for education can be lower for certain groups o f students, resulting in severe inequities. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 Gender inequities in education are particularly im portant because female education is associated with positive development outcomes including reduced fertility, reduced child mortality, and increased female participation (Schultz, 2002). If governance strategies to promote efficiency in education result in reduced female enrollment, these long-run goals o f economic development will not be achieved (Abu- Ghaida & Klasen, 2004). Historically, girls’ access to education has been far below that o f boys’ (Hill & King, 1993). While some o f the disparity can be blamed on a low supply o f school slots for girls, research has shown that demand-side obstacles to girls’ education also keep girls from enrolling in or completing school (Filmcr, 2004) . Several social and economic factors contribute to low consumption o f girls’ schooling including high opportunity costs, low returns in the labor market, and social stigmas against female literacy (Bowman, 1984; Buchmann, 2000; Hill & King, 1993). Because boys and girls face different costs and benefits o f education, governance strategies that promote cost- savings can have different effects for males and females. As a result, institutions designed to promote efficiency based solely on the costs and benefits to one group, or on the average costs and benefits o f both groups, can add to inequities in education. These inequities, in turn, can undermine basic development goals by reducing the educational attainment o f women. In short, the high costs and low individual benefits that create a reduced demand for girls’ education in developing countries must be addressed when designing institutions o f governance. While strategies that promote efficiency can contribute to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 economic development by helping to ensure that education funds are invested wisely, promoting female education is equally im portant for economic development. Thus, governance strategies in education must take into account issues o f gender in order to achieve underlying development goals. This paper examines the extent to which governance in the education sector responds to the differences in demand for boys and girls. First, theoretical implications o f three key governance strategies — privatization, decentralization, and citizen participation — are explored. Then, World Bank education projects are analyzed to see how governance strategies are being implemented in the education sector by an international organization that supports the dual goals o f girls’ education and effective governance. The next section takes a closer look at three education projects in countries where girls’ education is low. Finally, the implications for the expansion o f girls’ education are discussed. This paper contributes to an understanding o f governance and development by examining privatization, decentralization, and citizen participation in the context o f demand-side obstacles to girls’ education. These theoretical observations are applied to empirical analysis o f the strategies implemented in recent World Bank education projects. The results suggest that an awareness o f gender equity and demand-side obstacles leads to use o f governance strategies that reduce the direct cost o f schooling for girls. 1 lowevcr, the broad use o f decentralization and privatization without special consideration o f the effects on girls’ enrollment has the potential to further increase gender gaps education. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 2 THEORETICAL BACKGROUND Governance & Gender A key criticism o f past development strategies is that inequities are intensified when institutional arrangements ignore the context of women in development (Boserup, 1970). Neoclassical economics looks at the household as a single decision-making unit (Becker, 1975). More recent work looks at resource allocation within families and finds that women’s contributions to the family are often undervalued because women’ s domestic work is not purchased in markets (G. Sen, 1996). Women and girls often receive fewer family resources including health care, nutrition, and education (A. Sen, 1990; World Bank, 1995). As a result, women and girls are often more dependent on social policy and government provision o f education, health care, and jobs (Assad, 2003). Under the structural adjustment programs (SAPs) that were required by the International Monetary Fund (IMF) and World Bank in the 1990’s, a central governance strategy was to reduce government spending. Many countries responded by reducing public services including health care and education. Empirical evidence suggests that women were relatively worse off under structural adjustment than men. Women experienced higher unemployment in countries where female employment is concentrated in the public sector (Assaad, 2003; Nassar, 2003). Women were also harmed by the additional child care burden that resulted from the removal o f education and health services (Olmstead, 2003). Finally, gender gaps in education widened as a result o f cuts in public funding for education (Stromquist, 1999). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 0 From an economic perspective, SAPs and other governance strategies affect men and women differently because they face different costs and benefits from public goods (Agarwal, 2001). There are several reasons offered for these differences. First, women’ s costs and benefits differ from men’s because female domestic labor has no market value, making women dependent on families and governments for support (G. Sen, 1996). Women’s domestic role also affects the way they value common pool resources such as water and firewood, so women are affected differently by institutional arrangements that govern the use o f these resources (Zwaarteveen, 1997). Women also participate in different industries than men creating different consequences of government regulation and macroeconomic changes (Ram, 1982; Assad, 2003; Raynolds, 1998). Women’s costs and benefits also depend on legal protection from sexual exploitation in public and economic life (Sender, 2002), and the extension o f equal rights o f property ownership and inheritance (Meinzen-Dick et al., 1997). Finally, women’s roles create high costs o f participation in politics and public life (Weinberger & Jutting, 2001). As a result, the calculus o f female use o f public goods, responses to government policy, and participation in politics is very different than men’s. Within the education sector, gender differences in costs and benefits arise from different social roles and different levels o f political and economic participation. Girls typically have higher opportunity costs o f schooling because families rely on school- aged daughters to provide domestic or wage labor (MacMahon, 1999; Schultz, 1988; Smock, 1981). Girls also have lower individual returns to educational investments due to segmented labor markets, employer assumptions about women’s skills, and sexual Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 discrimination (Ram, 1982). If the benefits o f female education accrue in terms o f greater household productivity (Lam & Duryea, 1999), these benefits have no clear market value. Parents also have different cost and benefit calculations for sons and daughters because a daughter’s future productivity typically accrues to her husband’s family, while a son’s future productivity can support parents into old age (Greenhalgh, 1985). Finally, social stigmas against educating daughters may influence demand, if there are social or reputation costs associated with sending daughters to school (Buchmann, 2000; Hill & King, 1993). The consequence o f greater costs and lower benefits o f girls’ education is that die demand for girls’ education is lower than the demand for boys’ education. To increase girls’ access, education policy needs to encompass strategies that promote equity and inclusion through increased demand. Demand-side interventions involve reducing the costs o f schooling, easing the burden o f domestic labor, reducing the incidence o f child labor, and eliminating social stigmas against educating daughters (Winter & Macina, 1999). Governance in the Education Sector This section addresses the implications o f selected governance strategies for gender equity in education. Educational governance involves the institutions for policy making, financing, and administration o f the schools. In order to expand education, developing countries must find ways to finance increasing enrollments with limited resources. The primary motivation for governance reform in education is to provide high quality education at the lowest cost to the greatest num ber o f students. To achieve Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 2 basic development goals, this needs to be achieved in a context that also promotes girls’ education and removes demand-side obstacles. Three governance strategies to achieve this goal include decentralization, privatization, and citizen participation. While these strategies are discussed one at a time, it is recognized that appropriate institutions vary by context, and individual countries will develop unique combinations o f privatization, decentralization, and participation (Parry, 1997). Privatization Because education is central to economic development and nation-building publicly-provided education is widely promoted. However, there are several strategies to introduce quasi-market competition into public education systems (Gordon & Whitty, 1997; Jones, 1992). The first is to charge student fees to recover the costs of schooling. Fees can then be used to expand education by building new schools. The economic rationale behind fees is that free education results in oversupply when students attend for free, even if their marginal benefit is less than the marginal cost paid by the government. Fees recover costs from students with a high willingness to pay for schooling and eliminate the waste o f oversupply (Thobani, 1984; Tan, Lee, & Mingat, 1984; Mingat & Tan, 1986). This approach has a negative effect on gender equity because students with a low demand are m ost likely to exit schools when fees are charged (Colclough, 1996). In addition, individual revealed preference for education does not consider the social benefits (Maclure, 1994). If the individual benefits of education for girls are lower than boys, school fees will contribute to greater gender inequity and the social benefits o f Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 female education will be lost. Empirical evidence from countries that implemented fees as a part o f SAPs in the 1990’s suggests that school fees reduce girls’ enrollment more than boys’ (Assad, 2003; Stromquist, 1999). A second privatization strategy is to support private education as a substitute for public education. Private schools can improve educational quality by competing with public schools for students and may also satisfy excess demand in countries with resource constraints. Colclough (1996) argues that cost-saving are often less than predicted because private schools are typically subsidized to some extent by education ministries’ work in curriculum, inspection, and teacher training. Because private education markets are diverse, it is difficult to predict the effect on educational equity (Colclough, 1996). In some countries, charitable or missionary schools provide high-quality' education to students who are excluded from public schools due to social or financial circumstances. In other countries, private education is elitist, and high tuition costs contribute to existing inequities. Private education markets also mean that governments relinquish control over education goals including socialization to a common culture and curricula, labor market training, and social equity (Wise & Gage-Brandon, 1983; Pierson, 1988). This can have a negative effect on gender equity if the government cannot require private schools to provide equal access to girls. A third approach to privatization involves introducing competition into public education by allowing schools to compete for students or funding. Schools are rewarded for desirable outcomes — such as student achievement, enrolling underserved Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 populations, or attracting high quality students — creating an incentive for improved quality. The specific measures o f achievement that are rewarded are chosen by political processes, so accountability systems typically make schools accountable to specific populations (Levin, 1974). Tanking performance and funding can have equity implications if low-performing schools are unable to compete for funds and quality at these schools continues to decrease. The effects on equity could be positive or negative (Lee & Wong, 2004; Powers, 2004). If funding is tied specifically to efforts to expand girls’ education, competitive funding processes could create an incentive to improve equity. If funding is tied to other outcomes, such as achievement test scores, it could create a disincentive to enroll girls’ if there is a perception that girls’ average achievement is lower than boys’. Decentralization Decentralization in education involves devolving authority from the central government to lower levels. Decentralizing education is intended to improve cost- effectiveness by increasing responsiveness to local needs, improving economies o f scale by allowing central education ministries to focus on policy and planning, and fostering greater participation at the local level (Maclure, 1994). Parry (1997) identifies several types o f decentralization that can be combined and mixed depending on national goals and needs and local capacity to govern. First, authority can be devolved to different levels including regional offices o f the central government, local education authorities, local governments, or school sites. Second, there are different types o f authority that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 can be devolved including administrative control o f school management, fiscal control o f expenditures or revenues, and political control o f policy and planning. The degree and type o f educational decentralization depends on context. Central governments that need to reduce costs may delegate greater fiscal control to local governments, even allowing them the power to raise revenue to support education (Parry, 1997). Governments that w ant control over curriculum and access will retain a great deal o f policy and planning authority (Pierson, 1998). Many authors warn that decentralization o f education can be dangerous in relatively new countries where political institutions are unstable and the central government needs to cultivate a sense o f unity and nationhood (Jackson, 2004; Lockhheed & Vers poor, 1991; Blunt & Turner, 2005). Both fiscal and administrative decentralization can have profound effects on educational equity. Critics observe that decentralization m ust include protections for the most vulnerable (Blunt & Tuner, 2005). Decentralization in practice is a political process, which means local control is often granted to political elites. If local elites discriminate against women or do not gain from female education, it is unlikely that decentralization o f education will improve gender equity (Crook & Manor, 2002). Local contributions o f resources will also depend on the willingness and ability o f local communities to pay for girls’ education, which may be very low if the private benefits are small (Maclure, 1994). With a low demand for girls’ education and low political power for women, it is unlikely that decentralization will lead to new resources for girls’ education, and inequities are likely to increase. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 Citizen Participation Citizen participation can serve multiple governance functions from assuring compliance with national goals, to enlisting contributions from the community, to building political capacity and strengthening democracy (Cornwall, 2003). Proponents o f citizen participation also argue that it also improves efficiency by preventing corruption from the bottom up (Haarhuis & Leeuw, 2004) and achieving low-cost voluntary compliance (Krishna, 2003). Participation in education is expected to increase transparency to improve school quality, gain support o f local communities to increase school resources, and improve student outcomes by increasing parent involvement in learning. It can include parent-teacher associations, community oversight boards, democratically elected principals, and many other forums o f public input. In theory, citizen participation gives voice to all stakeholders, but empirical research illustrates that often participation processes give voice only to the m ost vocal (Cornwall, 1996). The extent to which participation increases gender equity will be limited in communities where women have limited political rights (Fierlbeck, 1997). Women are often excluded from formal and informal social networks that facilitate political participation, which means that women’s issues are often not discussed (Silvey 2003). Weinberger and Jutting (2001) observe that women, and particularly working mothers, have higher costs o f participation than men making them less likely to attend public meetings. To address the limited o f participation o f women, authors argue that institutions need to be designed to encourage and accommodate female participation (Kardam, 1997; Fierlbeck, 1997; Cornwall, 2003). The degree to which citizen Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 participation influences girls’ education may depend on the relative costs and benefits o f participation for women, the political rights o f women, and the structure o f participatory processes. If participation is controlled by local elites, it is not guaranteed that girls’ education will be a salient issue. 3 EMPIRICAL EVIDENCE This section examines implementation o f governance strategies and girls’ education as components o f World Bank education projects. With its Comprehensive Development Framework (CDF) the World Bank has dedicated itself to a focus on effective governance as a means to economic development (Wolfensohn, 1998). The World Bank favors governance strategies that promote a reduced role for government in markets, decentralization to promote local ownership and responsiveness to local needs, and citizen participation to prom ote transparency and reduce corruption (Khan, 2002). The World Bank also embraces the goal o f “Education for All” and asserts that “girls’ education and the promotion o f gender equality in education are critical to development, and policies and actions that do not address gender disparities miss critical development opportunities” (World Bank, 2001). The objective o f this empirical analysis is to explore how the goals o f governance reform and gender equity are addressed in World Bank education projects and to determine if governance strategies in countries with low girls’ education are implemented in ways that will promote girls’ enrollment. Data on World Bank projects comes from publicly available Project Information Documents (PID) (World Bank, 2005). PID s summarize the development context, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 project goals, and implementation strategies for projects in the education sector. Although PID s are prepared by each recipient country, the content is negotiated with World Bank staff, so a PID for an approved project should reflect the preferences o f the World Bank, as perceived by the recipient country (Jones, 1992). It is important to note that PID s summarize the intent o f projects before they are implemented and are not evaluations o f completed projects. It is unknown if projects are implemented as they are described in PIDs. Therefore, PID s should be interpreted only as statements of intent. The World Bank project database includes a total o f 95 projects from 62 countries under the theme o f Education for All and approved in the years 2002,2003, and 2004. During this period, governance reform and girls’ education were both stated goals o f the World Bank. O f these projects, 40 were general poverty reduction projects, structural adjustment credits, or emergency support for an economic crisis. These 40 projects were excluded from further analysis because less than half the hands were targeted for the education sector, and very few details were offered about how education hinds would be spent In the remaining 55 projects, the majority o f hinds are targeted for the education sector, and PIDs included detailed information about needs, objectives, and implementation strategies to improve education. The PIDs for these 55 education projects were subjected to content analysis to identify governance strategies and strategies to promote girls’ education, with particular attention to demand-side obstacles. Table 19 lists the 56 education projects included in this analysis. These projects were approved for implementation in 48 countries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 Table 19. World Bank Education Projects 2002-2004 C ountry P ro ject Afghanistan E ducation Q uality Im provem ent Program Armenia E ducation Q uality & R elevance Azerbaijan E ducation Sector D evelop m en t P roject Bangladesh Program m atic E ducation Sector A djustm ent Credit Bangladesh R eaching O ut o f S ch ool Children Project Bangladesh Primary E ducation D evelop m en t P roject II Bangladesh Fem ale Secondary S ch ool A ssistance P roject Bhutan E ducation D evelop m en t Project Brazil Pernam buco Q uality E nhancem ent in Basic E ducation Brazil F U N D E SC O L A 3 Burkina F aso B asic E ducation Sector Project Chad E ducation Sector R eform Program Chile L ifelong I earning and Training China B asic Education /W estern Provinces C olom bia E ducation Q uality Im provem ent Republic o f C ongo Support to Basic E ducation Project D om inican Republic Early Childhood E ducation Project Eritrea E ducation Sector Im provem ent Project Ghana E ducation Sector P roject Tndia Elem entary E ducation Iraq Em ergency T extbook P rovision Project Jamaica Second R eform o f Secondary E ducation P roject Jordan E ducation Reform fo r the K now ledge E conom y K enya Free Primary E ducation Support Project K osovo E ducation Participation Im provem ent Project Kyrgyz Republic Investm ent Loan Laos Second E ducation D evelop m en t Project L esotho Second E ducation Sector D evelopm ent P roject T .ithnania Education Im provem ent Project M acedonia E ducation M odernization M aldives Integrated H um an D evelopm ent P roject M exico Basic E ducation D evelopm ent Phase III M exico Basic E ducation D evelopm ent Phase II Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 0 C ountry P ro ject M orocco Alpha M aroc M ozam bique H igher E d u cation P roject N epal E ducation fo r AH Project N epal C om m unity S ch ool Support P roject Nicaragua E ducation P roject N iger B asic E du cation P roject N igeria U niversal B asic E ducation Project Pakistan N ational E du cation A ssessm ent System Pakistan Punjab E du cation R eform A djustm ent Credit Paraguay Secondary E du cation R eform Peru Rural E du cation P roject Romania Rural E du cation Sierra L eone R ehabilitation o f B asic E ducation S t K itts & St. N evis O E C S E d u cation D evelopm ent P roject S t Lucia E ducation D evelop m en t P roject Tajikistan E ducation M odernization Project Tanzania Secondary E ducation D evelopm ent Program Turkey B asic E du cation II Uruguay T hird B asic E ducation Q uality Im provem ent P roject V ietnam Prim ary E du cation for D isadvantaged Children Y em en B asic E du cation D evelop m en t Program Yugoslavia Serbia E ducation Im provem ent P roject Source: W odd Bank (2005) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l M easures o f Gender Equity Developing countries differ significantly in the levels o f girls’ education provided and the levels o f gender equity in education. There are several indicators that can be used to measure the need for girls’ education in developing countries. In this analysis, girls’ absolute access to education is measured by gross enrollment rates (GER) in secondary school. The gross enrollment rate measures the percentage o f girls enrolled in school as a percentage o f the population o f girls in the appropriate age group.1 Secondary education rates are used to measure need because researchers believe that only secondary education brings about the social and economic changes necessary to link education with development goals such as reduced fertility, improved health, and female labor force participation. There are concerns that primary education only may increase fertility, because girls’ gain knowledge o f health and hygiene but do not gain opportunities to work in the labor market (Gille, 1986). A second measure o f need is girls’ relative access to education compared to boys. This is measured by gender parity, which is the ratio o f girls’ enrollment to boys’ enrollment in secondary school. A ratio o f one means perfect gender parity. The third measure o f need is the Gender Development Index (GDI), which measures female social and economic status compared to males. The G D I is an index o f female development outcomes — such as life expectancy, education, and income — in relation to the level o f male development outcomes. The G D I does no t directly measure education, but measures need as a function o f the development outcomes that are expected to improve as female education increases. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 2 For this study, school enrollment data come from the 2000 World Development Indicators (World Bank, 2001), and the G D I comes from the 2002 United N ation’s Human Development Report (UNDP, 2002). Additional data from the World Development Indicators and United Nations Women’s Indicator Statistics (WISTAT) (UN, 2000) are used to describe the status o f women and girls’ education in countries selected for further analysis. Table 20 lists the indicators o f need for each o f the 48 countries with an education project The data show that the need for initiatives to improve girls’ education varies across countries. Female secondary school enrollment ranges from only 4.9 percent to over 100 percent, with a mean o f 53 percent. Education parity ranges from 0.64 to 1.31, with a mean o f 0.94. Education parity values above 1.0 reflect female secondary enrollment that exceeds male secondary enrollment. This can occur if the demand for boys’ education is low, or more often because a civil war takes school- aged boys out o f the classroom and into the military (Rose & Tembon, 1999). The G D I ranges from 0.278 up to 0.841 with a mean o f 0.604, indicating that women’s development outcomes also vary across countries. The highest G D I o f 0.841 in Lithuania represents an international ranking o f 36th among the 144 developed and developing countries that received a G D I rank in 2002. The lowest G D I o f 0.278 in Niger represents a ranking o f 144th. This is the lowest G D I among all countries. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 Table 20. Gender Equity M easures in Countries with Education Projects Country G ender D evelop m en t In dex G ender Parity in Secondary E ducation F em ale E nrollm ent in Secondary School* A fghanistan Armenia .752 1.08 76.0% Azerbaijan 80.3% Bangladesh .499 1.05 45.9% Bhutan Brazil .768 1.10 113.8% Burkina Faso .291 0.64 8.0% Chad .368 5.1% C hile .830 0.75 64.3% China .741 59.5% Colom bia .770 1.10 73.2% Republic o f C ongo .488 0.83 38.1% D om inican R epublic .728 1.26 66.5% Eritrea .431 0.67 22 .6% Ghana .5 6 4 0.81 32.5% India .572 40.1% Iraq 29.1% Jamaica .762 1.04 84.8% Jordan .734 89.0% K enya .486 0.91 29.2% K osovo Kyrgyz Republic 1.00 85.8% Laos .528 0.72 31.3% L esotho .483 1.20 35.8% T rthnania .841 0.99 94.8% M acedonia 0.97 82.9% M aldives 1.07 57.2% M exico .792 1.05 77.3% M orocco .604 35.0% M ozam bique .339 0.64 9.3% N epal .484 0.75 43.2% Nicaragua .660 1.18 58.4% N iger .278 0.65 4.9% N igeria .458 27.3% Pakistan .471 0 .67 19.3% Paraguay .736 1.03 60.8% Peru .736 78.3% Romania .775 1.01 82.9% Sierra L eone 0.82 23.8% S t K itts & S t N evis S t Lucia 1.31 100.4% Tajikistan .668 0.83 71.2% Tanzania .401 0.81 5.2% Turkey .746 47.7% Uruguay .829 1.14 104.5% V ietnam .689 0.91 64.0% Y em en .436 25.4% Y ugoslavia 1.01 87.9% M E A N S .604 0.94 53.0% Sources: U N D P (2005); W orld Bank (2002) + G ross enrollm ent is the percentage o f girls o f secondary sch ool age enrollm ent in secondary school. G ross enrollm ent can exceed 100 percent if older gid s are enrolled in secondary sch ool, typically because o f grade repetition. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 Governance Strategies This section reviews the governance strategies described in PID s with a focus on the implications for girls’ education. Table 21 summarizes the use o f privatization, decentralization, and citizen participation in World Bank projects. O f the 55 education projects, 51 include at least one o f these strategies, and most include a combinations o f strategies. Decentralization was the m ost frequent strategy, cited in 79 percent o f projects. Citizen participation was cited in 46 percent o f projects, and privatization was cited in 34 percent o f projects. Table 21. Governance Strategies in Education G overnance Strategy Activities privatization decentralization civil society m arket-based strategies including: students fees private sch ools com petition in contracting com petition fo r textb ook purchases com petition betw een sch ools for grants authority devolved to low er levels: lo ca l governm ent local sch o o l councils sch o o l adm inistrators adm inistrative decentralization: curriculum sch o o l im provem ent plans construction plans fiscal decentralization budgeting authority raising revenue c o st sharing w ith central governm ent parent-teacher associations com m unity councils school-com m unity partnerships Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 Privatization O f the 55 education projects, 19 include some form o f privatization or market- based strategies. There were no projects that included instituting education fees. On the contrary, m ost projects focus on reducing costs to students through free tuition, stipends, or scholarships. Although it was infrequent, some countries plan to use public funds to support private schools through tuition reimbursements for parents or the direct public purchase o f slots in private schools. A second approach was to expand private education markets by removing barriers to entry. Private schools are unlikely to contribute to gender equity, because they attract students with the highest demand. More frequently, projects introduce competition into education indirectly, such as through competitive bidding for school construction or textbook production. One project included a competitive process to select a university to become the national teacher training institute. Public-private partnerships were also cited, but there were no specific examples o f w hat these partnerships would like look beyond an effort to include business in the education sector. These strategies have some potential to reduce costs and improve the quality o f education. There can be indirect equity effects if cost savings are reinvested in other strategies that increase the demand for girls’ education, but there is no indication that these strategies will directly influence equity one way or the other. A more promising strategy, cited in approximately 14 percent o f projects, is to link school funding with school performance through competitive grant processes. In m ost cases, these grants were targeted for schools that illustrate both the need to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 expand enrollment o f an underserved population and the development o f an innovative project. Competitive grants are the only market-based strategy that can be tied directly to equity goals. However, implementation o f a competitive grant process also raises concerns about the equity effects on schools that fail to win grant funding. These schools, although they serve communities that may have severe education inequities, will lose out on funding opportunities due to an inability to compete. For girls’ education, it is promising that World Bank projects appear to promote reducing, rather than increasing, education fees. This should help to overcome some o f the economic obstacles to girls’ enrollment, although high opportunity costs and other indirect costs may still keep girls’ out o f school. Expanding private schools can increase supply, but it is unlikely to provide additional slots for girls if there is higher demand for slots for boys. The effect o f competitive grant processes is ambiguous. Girls could suffer if schools are rewarded for high achievement, and boys’ average achievement higher than girls’. Girls’ education could increase if schools are held accountable to gender equity goals. Decentralization Forty-four education projects include some form o f decentralization, making this the m ost frequently cited governance strategy. Authority for school management is delegated to different levels including local governments, state education councils, local school boards, school management teams, and school principals. School-based management, which involves delegating budgeting authority to committees o f stakeholders at school sites, is implemented in several projects. In some cases, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 decentralized authorities are required to be democratically elected or to include public participation, but in most cases the institutional design o f local authorities is not specified. The types o f decisions that are delegated to the local level include school management, budgeting, school construction, textbook acquisition, and the authority to design local initiatives to improve quality. In several cases, schools sites are required to create individual action plans to improve enrollment and quality. A number o f strategies for fiscal decentralization strategies are also cited. Strategies for reallocating school funds include allocating construction funds to the local level and allowing schools to determine their own budgets based on per student funding from the national education ministry. In a small number o f cases, local communities are expected to finance school construction. Primarily, strategies to decentralize financing focus on the authority to allocate resources received from the central government and do not require localities to raise revenue independently. N o projects include plans to cut off central funding in favor o f local revenue sources. These decentralization strategies can have mixed results for educational equity. If local authorities are more inclusive or more concerned about girls’ education than the central government, decentralization could increase girls’ education. Because many o f the social benefits o f girls’ education accrue at the national level, it is unlikely that gender equity will be implemented from the bottom-up. This can be accomplished if localities are held accountable for providing girls’ education under a decentralized education system. It is promising for equity that most countries intend to continue to fund education from the national level. The success o f national initiatives to achieve Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 equity in a decentralized system may depend on the ability o f the central bureaucracy to enforce incentives. Citizen Participation Twenty-six projects include citizen participation strategies. The m ost common strategy is to develop community-school partnerships where stakeholders participate in school oversight or management. Parent teacher associations are also cited as a strategy to promote participation at school sites. A small number o f projects also include assistance to strengthen the capacity o f community members to participate in school management. O ther strategies include parent training, outreach to parents, and parent participation in specific decision such as textbook selection. It is difficult to predict the influence o f these strategies on equity because it is unclear what types o f decisions stakeholders will be involved in. If participation is at the school level, parents o f enrolled children are more likely to participate than parents o f girls w ho do not attend school. This group o f parents will be more interested in the quality o f education than expanding the quantity. Parent outreach activities are intended to increase the demand for education by providing information about the benefits of education. However, outreach efforts do not change the economic costs o f sending girls to schools. The issue o f obstacles to women’s participation is not addressed in the PIDs. Overall, it is unlikely that participation can increase the demand for girls’ education as long as community-wide demand for girls’ education is low and women’s participation is not encouraged. In this case, the preferences expressed in participatory decision Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 making and planning will reflect a low demand for girls’ education and will not raise this issue as a priority (Macclure, 1994). This does not mean that participation in general is bad for gender equity. If parents recognize the need for girls’ education, it is more likely to be addressed in a participatory institutional context than in a context where decision making is open only to elites, but this is dependent on issues o f openness and inclusion that are not addressed in the World Bank PIDs. Dem and-Side Strategies The first step in addressing girls’ education is to set a goal to increase access or improve gender equity. Clearly setting a development goal initiates a concrete process o f goal achievement that can guide planning and implementation (Colclough, 2005). While 33 o f the 56 education projects state that gender equity or girls’ education is an im portant issue, only 24 back up this statement with a clear goal. O f the 24 projects that set a goal to increase girls education, all but one include demand-side interventions for girls. The most frequent strategy to increase demand is to provide schooling at a lower cost for girls. Free tuition reduces the direct costs o f schooling. Projects also include assistance with indirect costs and opportunity costs through free supplies and uniforms and additional cash stipends specifically for girls. Cash stipends should be particularly effective because they compensate parents for at least part o f the foregone benefits o f girls’ domestic or wage labor while they attend school. In addition, this is a benefit that parents can only get if they enroll daughters, so it can equalize the demand for girls’ and boys’ education. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 0 The second strategy is to alter parental preferences through outreach and public information campaigns about the benefits o f girls’ education. Public outreach can be effective if parents have incomplete or incorrect information about the benefits o f girls’ education. However, this only applies to the private benefits o f education which may, in reality, be very low for girls. Knowledge o f the social benefits o f girls’ education may not increase girls’ enrollment if die economic obstacles mean that parents cannot afford to enroll daughters. Com bining Governance & D em and-side Interventions As is apparent in Table 20, not all countries with World Bank education projects suffer from underprovision o f girls’ education. Four for the 48 countries have achieved female secondary enrollment o f 100 percent, and in 17 countries female secondary enrollment is equal to or greater than male enrollment. This section focuses on education projects in countries with the greatest need for girls’ education to determine if goals and strategies are responsive to equity problems. The first objective is to determine if countries with the greatest need for girls’ education include demand-side interventions. The second objective is to examine the combination o f governance and girls’ education initiatives in countries with the greatest need. Table 22 displays the frequency o f strategies for countries in the lowest quartile by each o f the three measures o f need. Looking at attention to gender issues, 80 percent o f projects in countries with low GDI have a gender equity goal and demand- side interventions for girls. Countries with low gender parity and low female secondary enrollment are less likely to include a gender equity goal and demand-side interventions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 Attention to gender issues in education appears to respond more to the status o f women than to current access to girls’ education. Almost half o f countries with low female secondary enrollment have projects that do not include demand-side interventions. Table 22 also illustrates that almost all countries with a high need for girls’ education are implementing some form o f education decentralization. Privatization is implemented less frequently, in 40 percent o f countries with low G D I and gender parity, and 54 percent o f countries with low secondary enrollment. Citizen participation is implemented in 46 percent o f countries with low secondary enrollment, 40 percent o f countries with low G D I, and 20 percent o f countries with low gender parity. Table 22. Governance Strategies in Countries with Low Gender Equity L o w est Q u artile P rivatization D e ce n tr a liz a tio n C itizen P a rticip a tio n E q u ity G oal D em a n d -S id e In terv en tio n G D I (n=10) 40% 100 % 40% 80% 80% gender pari tv (n=10) 40% 90% 20 % 60% 60% secondary enrollm ent (n= 13) 54% 92% 46% 62% 54% Sources: U N D P (2005); W orld Bank (2002) Case Studies The section takes a closer look at the projects in three countries with the a high need for girls’ education projects: Niger, Afghanistan, and Tanzania. These countries share a need for girls’ education but differ in the selection o f governance strategies and acknowledgement o f girls’ education as a policy issue and development goal. The brief descriptions below are not meant to give a detailed analysis o f female education and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 2 governance strategies in general. Instead, these descriptions illustrate how awareness o f gender equity influences the choice o f governance strategies as expressed in PIDs. Exam ple #1: N iger’s Basic Education Project O f the 48 countries with World Bank education projects, Niger may have the greatest need for girls’ education. It has the lowest G D I, lowest female secondary enrollment, and third lowest education parity. Girls’ primary school enrollment was only 28.6 percent in 2000, with secondary enrollment o f only 4.9 percent. Girls’ secondary enrollment is only 65 percent o f boys’. With a G D P per capita o f only $200 in 2000, Niger faces many development obstacles that can be reduced through female education including a fertility rate o f 7.2 births per woman and a child mortality rate o f 270 deaths per 1,000 live births. Sixty-two percent o f teenage girls are employed, suggesting that wage labor prevents girls from attending secondary school. Niger’s Basic Education Project received $30 million in World Bank support in 2003. The 1 ’ID explicitly identifies gender equity as a problem in Niger, identifying inequities both in enrollment and resource allocation. The PID also identifies demand- side obstacles for boys and girls including poverty, child labor, and perceptions that schooling will no t improve labor market outcomes. Increasing female enrollment is identified as a specific project goal, and this goal is backed up with specific strategies. The objectives o f this projects are to expand basic education by increasing enrollment and improving management. Supply-side interventions include building schools, using double shifts to maximize use o f existing buildings, and restructuring teacher contracts to improve incentives. The primary demand-side strategy is to lower Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 the direct costs o f education for girls by providing free textbooks and supplies. There is also an initiative to increase community awareness o f the importance o f girls’ education, which is supported by a project to gather statistics on female enrollment. The PID states that governance strategies are designed to reduce costs to enable expansion in supply. Specifically, decentralization involves clarifying the roles o f different institutional levels including school management committees, local entities, regional offices, and the central education ministry. The process o f decentralization appears to be at a very early stage, so activities focus on developing skills and defining roles. There are no specifics in the PID about how gender equity will be supported in decentralization. The project also includes supporting the development o f private schools in higher education and in urban areas. Limiting private school development to urban areas reflects an awareness o f the low demand for education in rural areas, particularly for girls, which is explicitly stated in the PID. This limited privatization is less likely to have a negative effect on gender equity than broader forms o f privatization. Example #2: Afghanistan’s Education Quality Improvement Program Due to its recent political history, complete data on Afghanistan’s education system and development outcomes are not available, but it is widely known that formal girls’ education was severely limited under Taliban rule. Girls’ primary enrollment in 2000 was reported to be less than one percent. The fertility rate in 2000 was 6.84 births per woman and the child mortality rate was 257 deaths per 1,000 live births. Afghanistan’s Education Quality Improvement Project received $35 million in World Bank funds in 2004. The PID identifies an undersupply o f education as the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 primary problem in Afghanistan. It is explicitly stated in the PID that the education ministry assumes there is a large unmet demand for education. Low girls’ enrollment is identified as a particular problem that is attributed to o f a low supply o f girls’ schools and female teachers. Demand-side obstacles to education are not mentioned or addressed. Responding to the identified supply-side problems, Afghanistan’s education project focuses on supply-side interventions only. The project addresses supply through funds for school construction and teacher training. To address gender equity, girls’ schools and coed schools are given priority in school construction funding. Control is decentralized to the community level where school management committees are granted authority over school improvement and finances. This project also involves citizen participation through involvement o f the community and N G O ’s in school planning. Thus, responsibility for ensuring that there is sufficient girls’ education is delegated to the local level where communities may or may not support educational equity. However, the central government does provide a financial incentive to increase the supply o f girls’ education through infrastructure funding. Exam ple #3: Tanzania’s Secondary Education Developm ent Program O f the countries with World Bank education projects, Tanzania has the fifth lowest gender development index, third lowest female secondary enrollment, and tenth lowest gender parity. Girls’ school enrollment is 63 percent in primary school, and drops to only 5 percent in secondary school. More than 70 percent o f teenage girls’ are employed in Tanzania, suggesting that wage labor creates opportunity costs for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 schooling. Boys’ secondary enrollment is slightly higher than girls, but still only reached 6.4 percent in 2000. With a 2000 G D P per capita o f $190, Tanzania faces a fertility rate o f 5.3 births per woman and a child mortality rate o f 165 deaths per 1,000 live births. Tanzania received $150 million in World Bank funding in 2004 for the Secondary Education Development Program. The project’s goals are to increase secondary enrollment, improve learning outcomes, and improve public administration o f secondary education. O n the supply side, this project includes initiatives to recruit, train, and retrain teachers and to improve the quality o f curricula, textbooks, and testing. Governance strategies include privatization and decentralization. Barriers to private schools will be removed, and competitive grants will be offered to both community and private schools to expand secondary enrollment. Importantly, competitive grants are matching grants, which require contributions o f funds at the local level. Decentralization includes devolution o f administrative and management authority, but not fiscal decentralization, to lower levels within the education sector. To address the low demand for education, scholarships will be provided to cover school fees for students from poor families. Tanzania’s PID does not mention girls’ education or gender equity as an issue and includes no interventions to increase girls’ education specifically. With secondary enrollment below 10 percent, Tanzania clearly has a priority to increase enrollment for all populations. However, the selected governance strategies may result in more inequity as enrollment increases. Expansion o f private schools that charge fees will increase enrollment only for populations with high demand. Matching grants for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 educational expansion favor communities that have high wealth and high demand for education. Decentralization o f administration is also dangerous if there are no mandates for equity. This project includes no components for participation or inclusion in decentralized decision-making, making the process subject to control by elites. Because there is no m ention o f gender equity in this project description, it is unclear if any attempts are being made to balance equity with efficiency as the education sector expands. 4 DISCUSSION This paper approaches governance in education from the perspective o f the importance o f increasing girls’ education. If cultural and economic factors lead to a lower demand for girls’ education, these factors must be considered in the design o f governance institutions. Governance strategies designed to promote efficiency can ignore the consequences for gender equity, because costs can be recovered by providing education only to individuals with the greatest demand. The considerable social benefits o f girls’ education, combined with the low private benefits, mean that consideration o f only private demand will result in underprovision o f education for girls and continued or increased gender inequity. A review o f theories o f governance and equity illustrates that privatization, decentralization, and citizen participation can increase inequity if there are no safeguards to protect girls and other vulnerable groups. Empirical review o f W orld Bank projects reveals that countries are using a combination o f privatization, decentralization, and citizen participation to reform governance in education with differing implications for girls. Importantly, many World Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 Bank projects include no strategies to promote demand. Some countries do address demand by reducing the direct and indirect costs o f schooling. Assistance for girls is provided through free tuition, textbooks, and supplies. Taking this approach a step further, some projects include stipends for families that enroll girls in school to overcome some o f the high opportunity costs that result from dependence on daughters’ domestic and wage labor. A second demand-side strategy is to provide outreach to the community about the importance o f girls’ education. This approach can overcome social obstacles, but does no t address the economic costs associated with sending daughters to school. Outreach efforts will be less likely to influence girls’ enrollment when the economic obstacles are significant. The World Bank education projects also include several governance strategies that can contribute to girls’ education. The first is to reduce the use o f school fees. When given a choice between paying fees for sons or daughters, poor families will typically benefit more from educating sons. Eliminating fees can only help gender equity. Girls’ education is also being promoted by tying school funding to achievement o f equity goals. This creates a clear incentive for communities to increase opportunities for girls to attend school. Ideally, schools would also be accountable for girls’ achievement to ensure that the quality o f girls’ education is high. A third strategy that can support girls’ education is the involvement o f the broader community is school planning. Greater inclusion o f women in political life can expand awareness o f the importance o f girls’ education. A caveat here is that women’s political empowerment must be a component o f participation and processes must be designed to reduce Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 participation costs for women. While many World Bank projects include citizen or community participation, greater attention to female participation is needed to promote gender equity. O ther World Bank strategies can have a negative effect on girls’ education and gender equity. Depending on the private sector to increase the supply o f school slots is likely to increase education for boys, but not for girls. With no guarantees o f democracy or participation in school governance, plans to decentralize school management may also increase gaps in education. If local elites control decentralized school management and have no incentive to increase girls’ education, there will be no benefits for girls. Similarly, decentralized school financing can reduce resources for girls’ education if local demand is low. It is promising that only a small number o f World Bank projects include requirements for local revenue to support schools. Finally, girls’ education will no t be helped by citizen participation strategies that only include parents and teachers o f current students. While these individuals have an interest in improving school quality and reducing inefficiencies, they do not speak for students who are unable to attend school. Broader community participation is a preferred strategy to promote equity. The three projects reviewed in greater detail provide a small snapshot o f the combination o f strategies being implemented in countries with low female education and high inequity. These three examples demonstrate how girls’ education is approached very differently in different contexts. The project in Niger reflects an open acknowledgement o f gender inequities in enrollment and resource allocation. These Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 inequities are addressed directly with efforts to reduce to costs o f schooling for girls. Promotion o f private schools is limited to higher education and urban areas, based on a recognition that private primary and secondary schools can harm equity. However, efforts to decentralize the education system are not discussed in the context of protecting and promoting equity. This suggests that some but not all o f Niger’s governance strategies are integrated with equity goals. The project in Afghanistan includes recognition o f gender inequity, but attributes the problem entirely to a lack o f supply. Control over school management is placed almost entirely in the hands o f local communities, which may or may not have a large demand for girls’ education. The central government uses school financing to create an incentive to build school for girls. However, if the demand for girls’ education is indeed low, this incentive may be ineffective. The final project in Tanzania reflects no recognition o f gender inequity or the need for girls’ education. The recommended governance strategies also ignore obstacles to girls’ education and could lead to greater gender gaps as supply increases. These strategies include promoting private secondary schools that charge fees and decentralizing authority with no provisions for inclusion or democratic decision-making. These three examples illustrate how gender issues can be applied to educational governance and also how ignorance o f gender issue leads to the application o f governance strategies that can increase inequities. Based on the Tanzanian example, there is still work to be done to increase awareness of girls’ education and gender equity in development. There is also work to be done to align governance strategies with equity goals. It appears that decentralization in particular is being implemented without Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 safeguards to support vulnerable groups. Allocating too much authority to the local level before there is a sufficient supply o f education could result in an education system designed only for boys. Similarly, depending on private schools to increase supply could also exclude girls from a growing education system. 5 CONCLUSION It is a challenge for developing countries to ensure equity while creating a system o f national education that is cost-effective and responsive to local needs. Attention to equity means making education accessible to groups that have historically displayed a low demand. In many developing countries, the demand for girls’ education is much lower than boys. This means that education policy, including the governance o f institutions in the education sector, must be designed to promote girls’ education. Unfortunately, several strategies that can improve efficiency in education can also undermine equity. In particular, privatization and decentralization transfer costs to parents and communities, potentially increasing gender gaps. In the long-run these policies will hurt economic development because female literacy is so important to basic development outcomes. Evidence from World Bank sponsored education projects suggests that while strategies to promote demand for girls’ education are being promoting, these strategies are often implemented in a governance environment that ignores gender issues. It is very promising that countries are attempting to reduce the costs o f public schools and to create a more inclusive environment for decision-making. However, decentralization Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 and privatization continue in other forms. Safeguards are necessary to ensure that these strategies do not limit opportunities for girls’ to attend school. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 1 Gross enrollment rates can exceed 100 percent if girls who are older than secondary school age are still enrolled in secondary school. This typically occurs because o f high rates o f grade repetition or efforts to enroll out-of-school youth. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 Works Cited Abu-Ghaida, D . and S. Klasen (2004). "The costs o f missing the millennium development goal on gender equity." W orld Development 32(7): 1075-1107. Agarwal, B. (2001). "Participatory exclusions, community forestry, and gender: An analysis for South Asia and a conceptual framework." World Development 29(10): 1623-1648. Assaad, R. (2003). Gender and employment: Egypt in comparative perspective. Women and Globalisation in the A rab Middle E ast. E. A. Doumato and M. P. Posusney. Boulder, CO: Lynne Rienner Publishers, Inc. Bayliss, K. and B. Fine (1998). "Beyond Bureaucrats in Business: A critical review o f the World Bank approach to privatization and public sector reform." Journal of International Development 10(7): 841-55. Becker, G. (1975). Human C apital New York: National Bureau o f Economic Research. Blunt, P. and M. Turner (2005). "Decentralisation, democracy, & development in a post-conflict society: Commune councils in Cambodia." Public Administration & Development 25: 75-87. Bowman, M. J. (1984). "An integrated framework for analysis o f the spread o f schooling in less developed countries." Comparative Education 28(4): 563-83. Buchmann, C. (2000). "Family structure, parental perceptions, and child labor in Kenya: What factors determine who is enrolled in school?" Social Forces 78(4): 1349- 1378. Colclough, C. (1996). "Education and the market: Which parts o f the neoliberal solution are correct?" World Development 24(4): 589-610. Colclough, C. (2005). "Rights, goals, and targets: How do those for education add up?" Journal o f International Development 17: 101-111. Cornwall, A. (2003). "Whose voices? Whose choices? Reflection on gender and participatory development." World Development 31(8): 1325-1342. Crook, R. and J. Manor (2002). Democratic decentralization: Essential elements. Making Develoment Work: Development Eeaming in a World o f Poverty & Wealth. N. Hanna and R. Picciotto. New Brunswick: Transaction Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Fierlbeck, K. (1997). Getting representation right for women in development Accountability, consent and the articulation o f women's interests. Getting Institutions Bight fo r Women in Development. A. M. Goetz. New York: St. Martins Press, Inc. Filmer, D. (2004). If You Build It, Will They Come? School availability and school enrollment in 21 poor countries. Policy Research Working Papers. T. W. Bank. Washington, DC: The World Bank. Gordon, L. and G. Whitty (1997). "Giving the 'hidden hand’ a helping hand? The rhetoric and reality o f neoliberal education reform in England and New Zealand." Comparative Education 33(3): 453-467. Greenhalgh, S. (1985). "Sexual stratification: The other side o f "growth with equity" in East Asia." Papulation and Development Bevieiv 11(2): 265-314. Haarhuis, C. M. and F. L. Leeuw (2004). "Fighting governmental corruption: The new World Bank programme evaluated." Journal o f International Development 16(547- 561). Hanna, N . and R Agarwala (2002). Toward a comprehensive development strategy. Making Development Work: Development Learning in a W orld of Poverty & Wealth. N . H anna and R. Picciotto. New Brunswick: Transaction Publishers. Hill, M. A. and E. M. King (1993). Women's education in developing countries: An overview. Women' s Education in Developing Countries. E. M. King and M. A. Hill. Baltimore, MD: The Johns Hopkins University Press. Jackson, P. (2004). "Ethnicity, decentralisation, and the fissile state in Georgia." Public Administration & Development 24(75-86). Jones, P. W. (1992). World Bank Fi nancing o f Education: Fending, Learning and Development. London: Roudedge. Jones, P. W. (1998). "Globalisation and internationalism: Democratic prospects for World education." Comparative Education 34(2): 143-155. Kardam, N . (1997). Making development organizations accountable: The organizational, political and cognitive contexts. Getting Institutions Bight for Women in Development. A. M. Goetz. New York: St. Martins Press, Inc. Khan, M. H. (2002). Corruption and governance in early capitalism: World Bank strategies and their limitations. Reinvent ing t he World Bank. J. R. Pincus and J. Winters, A. Ithaca: Cornell University Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 Krishna, A. (2003). "Partnerships between local government and community-based organisations: Exploring the scope for synergy." Public Administration & Development 23: 361-371. Lam, D. and S. Duryea (1999). "Effects o f schooling on fertility, labor supply, and investments in children, with evidence from Brazil." The Journal o f Human Resources 34(1): 160-192. Lee, J. and K. R. Wong (2004). "The impact o f accountability on racial and socioeconomic equity: Considering both school resources and achievement outcomes." American Educational Research Journal41 (4): 797-832. Levin, H. M. (1974). "A conceptual framework for accountability in education." The School Review 82(3): 363-391. Lockheed, M. E. and A. M. Verspoor (1991). Improving Primary Education in Development Countries. Oxford: Oxford University Press. Maclure, R. (1994). "Misplaced assumptions o f decentralization and participation in rural communities: Primary school reform in Burkina Faso." Comparative Education 30(3): 239-254. McMahon, W. (1999). Education and Development: Measuring the Social Benefits. Oxford: Oxford University Press. Meinzen-Dick, R. S., L. R. Brown, et al. (1997). "Gender, property rights, and natural resources." World Development 25(8): 1303-1315. Mingat, A. and J. P. Tan (1986). "Expanding education through user charges: What can be achieved in Malawi and other LDCs?" Economics of Education Review 5(8): 273- 286. Nassar, H. (2003). Egypt: Structural adjustment and women's employment. Women and Globalisation in the A rab M iddle E ast. E. A Doumato and M. P. Posusney. Boulder, CO: Lynne Rienner Publishers, Inc. Olmstead, J. (2003). Reexamining the fertility puzzle in MENA. Women and Globalisation in the A rab M iddle E ast. E. A. Doumato and M. P. Posusney. Boulder, CO: Lynne Rienner Publishers, Inc. Parry, T. R. (1997). "Achieving balance in decentralization: A case study of education decentralization in Chile.” W orld Development 25(2): 211-225. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Pierson, C. (1998). "The new governance o f education: The conservative and education 1988-1997." Oxford Redew o f Education 24(1): 131-142. Powers, J. M. (2004). "High-stakes accountability and equity: Using evidence from California's Public Schools Accountability A ct to address the issues in Williams v. State o f California." American Educational Research Journal41(4): 763-796. Ram, R. (1982). Sex differences in the labor market outcomes o f education. Women' s Education in the Third World: Comparative Perspectives. G. P. Kelly and C. M. Elliott. Albany, State University o f N ew York Press. Raynolds, L. T. (1998). "Harnessing women's work: Restructuring agricultural and industrial labor forces in the Dominican Republic." Economic Geography 74(2): 149-169. Rondinelli, D. A., J. R. Nellis, et al. (1983). Decentralization in developing countries: A review o f recent experience. World Bank Staff Working Papers. Washington, DC: The World Bank. Schultz, T. P. (1988). Education investments and returns. Handbook o f Development Economics. Volume 1. H. Chenery and T. N. Srinivasan. Amsterdam: Elsevier Science Publishers. Schultz, T. P. (2002). "Why governments should invest more to educate girls." World Development 30(2): 207-225. Sen, A. K. (1990). Gender and cooperative conflicts. Persistent Inequalities. I. Tinker. New York: Oxford University Press. Sen, G. (1996). "Gender, markets and states: A selective review and research agenda." World Development 24(5): 821-829. Sender, J. (2002). Reassesing the roles o f the World Bank in Sub-Saharan Africa. Reinvent ing the World Bank. J. R. Pincus and J. Winters, A. Ithaca: Cornell University Press. Silvey, R. (2003). "Engendering social capital: Women workers and rural-urban networks in Indonesia's crisis." World Development 31(5): 865-879. Smock, A. C. (1981). Women’ s Education in Developing Countries: Opportunities and Out comes. New York: Praeger Publishers. Stiglitz, J. E. (2002). Ghbali^ation and Its Discontents. New York: W.W. Norton & Company. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 Stromquist, N . P. (1999). The impact o f structural adjustment programmes in Africa and Latin America. Gender, Education and Development: Beyond Access to Empowerment . C. Heward and S. Bunwaree. London: Zed Books Ltd. Tan, J. P., K. H. Lee, et al. (1984). User charges for education: The ability and willingness to pay in Malawi. World Bank Staff Working Paper No. 661. Washington, DC: The World Bank. Thobani, M. (1984). "Charging user fees for social services: Education in Malawi." Comparative Education Remew 28(3): 402-423. UN (1999). WISTAT: Women's Indicators Database. New York: United Nations Publications. UNDP (2005). Human Development Reports, http://hdr.undp.org. Weinberger, K. an d j. P. Jutting (2001). "Women’ s participation in local organizations: Conditions and constraints." W orld Development 29(8): 1391-1404. Winter, C. and R. Macina (1999). Girls'Education: World Bank Support t hrough the International Development Association. Wise, A. E. and L. Darling-Hammond (1983). "Educational vouchers: Regulating their efficiency and effectiveness." Educational Researcher 12(9): 9-12 18. Wolfensohn, J. D. (1998). A Proposal for a Comprehensive Development Framework: A Discussion D raft Washington, DC: The World Bank. World Bank, The (1995). Toward Gender Equality: The Role fo r Public Polity. Washington, DC: World Bank. World Bank, The (2001). Overview o f Girls' Education, http://www l.worldbank.org / education/ girlseducation/ World Bank, The (2002). World Development Indicators. Washington, DC, World Bank. World Bank, The (2005). Projects and Operations, h ttp :// www l.worldbank.org/ education/projects/ Zwarteveen, M. Z. (1997). "Water. From basic need to commodity: A discussion on gender and water rights in the context of irrigation." W orld Development 25(8): 1335-1349. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 F u ll B ibliography Abu-Ghaida, D. and S. Klasen (2004). "The costs o f missing the millennium development goal on gender equity." W orld Development 32(7): 1075-1107. Adherents Organization. (2004). Majority Religions, h ttp ://www.adherents.com. Agarwal, B. (2001). "Participatory exclusions, community forestry, and gender: An analysis for South Asia and a conceptual framework." World Development 29(10): 1623-1648. Appiah, E. N., & McMahon, W. W. (2002). The social outcomes o f education and feedbacks on growth in Africa. The Journal o f Development Studies, 38(4), 27-68. Assaad, R. (2003). Gender and employment: Egypt in comparative perspective. Women and Globalisation in the A rab M iddle E ast. E. A. Doumato and M. P. Posusney. Boulder, CO: Lynne Rienner Publishers, Inc. Bardhan, P. (1988). Alternative approaches to development economics. In H. Chenery & T. N . Srinivasan (Eds.), Handbook of Development Economics (Vol. 1). Amsterdam: Elsevier Science Publishers, B.V. Bayliss, K. and B. Fine (1998). "Beyond Bureaucrats in Business: A critical review o f the World Bank approach to privatization and public sector reform ." Journal of International Development 10(7): 841-55. Barro, R. J. (1991). Economic growth in a cross section o f countries. The Quarterly Journal ofEconomics, 106(2), 407-443. Becker, G. (1975). Human Capital. New York: National Bureau o f Economic Research. Becker, G. (1981). A Treatise on the Family. Cambridge, MA: Harvard University Press. Becker, G. (1985). Human capital, effort, and the sexual division o f labor. Journal of Labor Economics, 3(1), S33-S58. Behrman, J. R-, Foster, A. D., Rosenzweig, M. R., & Vashishtha, P. (1999). Women's schooling, home teaching, and economic growth. The Journal o f Polilical Economy, 107(4), 682-714. Bellew, R. T., & King, E. M. (1993). Educating Women: Lessons from Experience. In E. M. King & M. A. Hill (Eds.), Women' s Education in Developi ng Countries. Baltimore, MD: The Johns Hopkins University Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 Bendera, S. (1999). Promoting education for girls in Tanzania. In C. Heward & S. Bumvaree (Eds.), Gender, Education & Development (pp. 117-132). London: Zed Books Ltd. Bils, M., & Klenow, P. J. (1998). Does schooling cause growth or the other way around? NBER Working Paper Series (No. 6393). Cambridge, MA: National Bureau o f Economic Research. Blunt, P. and M. Turner (2005). "Decentralisation, democracy, & development in a post-conflict society: Commune councils in Cambodia." Public Administration & Development 25: 75-87. Boserup, E. (1970). Women's Pole in Economic Development. In. New York: St. Martin's Press. Boserup, E. (1990). Economic change and the roles o f women. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Boserup, E. (1995). Obstacles to advancement o f women during development. In T. P. Schultz (Ed.), Investment in Women' s Human Capital. Chicago: University o f Chicago Press. Bowman, M. J. (1984). An integrated framework for analysis o f the spread o f schooling in less developed countries. Comparative Education, 28(4), 563-583. Buchmann, C. (2000). Family structure, parental perceptions, and child labor in Kenya: W hat factors determine who is enrolled in school? Social Forces, 78(4), 1349- 1378. Bulatao, R. A . (1984). Reducing Fertility in Developing Countries: A Review o f Determinants and Policy Levers. World Bank Staff Working Papers (No. 680). Washington, DC: The World Bank. Bumvaree, S. (1999). Gender inequality: The Mauritian experience. In C. Heward & S. Bunwaree (Eds.), Gender, Education Development (pp. 135-154). London: Zed Books Ltd. Caldwell, J. C. (1978). A theory o f fertility: From high plateau to destabilization. Population and Development Review, 4(4), 553-577. Caldwell, J. C. (1982). The transition from familial to labour market production and the social implications. In E. A. Hammel (Ed.), Theory of Fertility Decline. London: Academic Press, Inc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 Celle de Bowman, O. (2000). Peruvian female industrialists and the globalization project: Deindustrialization and women's independence. Gender and Society, 14(4), 540-559. Chemichovsky, D. (1985). Socioeconomic and demographic aspects o f school enrollment and attendance in rural Botswana, liconomic Development and Cultural Change, 33(2), 319-332. Cleland, J. (1985). Marital fertility decline in developing countries: Theories and the evidence. In J. Cleland & J. Hobcraft (Eds.), Reproduct ive Change in Developing Count ri es: Cadence from the W orld Fertility Survey. Oxford: O xford University Press. Colclough, C. (1996). "Education and the market: Which parts o f the neoliberal solution are correct?" W orld Development 24(4): 589-610. Colclough, C. (2005). "Rights, goals, and targets: How do those for education add up?" Journal of International Development 17: 101-111. Cornwall, A. (2003). "Whose voices? Whose choices? Reflection on gender and participatory development." World Development 31 (8): 1325-1342. Crook, R. and J. Manor (2002). Democratic decentralization: Essential elements. Making Develoment Work: Development Learning in a World of Poverty sir Wealth. N. Hanna and R. Picciotto. New Brunswick Transaction Press. Csapo, M. (1981). Religious, social and economic factors hindering the education o f girls in Northern Nigeria. Comparative education, 17(3), 311-319. Fafchamps, M., & Quisumbing, A. R. (1999). Human capital, productivity, and labor allocation in rural Pakistan. The Journal o f Human Resources, 34(2), 369-406. Fierlbeck, K. (1997). Getting representation right for women in development: Accountability, consent and the articulation o f women's interests. In A. M. Goetz (Ed.), Getting Institutions Rightfor Women in Development. New York: St. Martins Press, Inc. Filmer, D . (2004). If You Build It, Will They Come? School availability and school enrollment in 21 poor countries. World Bank Policy Research Working Papers (No. 3340). Washington, DC: The World Bank. Freedom House. (2004). Freedom in the W orld http://www.freedomhouse.org. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 Galor, O., & Weil, D . N . (1996). The gender gap, fertility, and growth. The American Economic Review, 86(3), 374-387. Gille, H. (1985). Policy implications. In J. Cleland & J. H obcraft (Eds.), Reproduct ive Change in Developing Countries: Evidence from the World Fertility Survey. Oxford: Oxford University Press. Goldin, C. (1994). The U-shaped female labor force function in economic development and economic history. NI3ER World Papers (No. 4707). Cambridge, MA: National Bureau o f Economic Research. Goldin, C. (1995). The U-shaped female labor force function in economic development and economic history. In T. P. Schultz (Ed.), Investment in Women's Human Capital. Chicago: University o f Chicago Press. Gordon, L. and G. Whitty (1997). "Giving the 'hidden hand' a helping hand? The rhetoric and reality o f neoliberal education reform in England and New Zealand." Comparative Education 33(3): 453-467. Greene, W. H. (1997). Econometric Analysis (3rd ed.). New Jersey: Prentice-Hall, Inc. Greenhalgh, S. (1985). "Sexual stratification: The other side o f "growth with equity" in East Asia." Population and Development Review 11(2): 265-314. Griliches, Z. (1997). Education, human capital, and growth: A personal perspective. Journal o f Labor Economics, 15(1), S330-334. Gundlach, E. (2001). Education and economic development: An empirical perspective. Journal o f Economic Devebpmentff), 37-59. Haarhuis, C. M. and F. L. Leeuw (2004). " Fighting governmental corruption: The new World Bank programme evaluated." Journal of International Development 16(547- 561). Hanna, N . and R. Agarwala (2002). Toward a comprehensive development strategy. M aking Development Work: Development Teaming in a W orld of Poverty Wealth. N. H anna and R. Picciotto. New Brunswick; Transaction Publishers. Heward, C. (1999). Closing the gender gap? The informal sector in Pakistan. In C. Heward & S. Bunwaree (Eds.), Gender, Education & Development (pp. 203-217). London: Zed Books Ltd. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 Higgins, P. A., & Alderman, H. (1997). Labor and women's nutrition: The impact o f work effort and fertility on nutritional status in Ghana. The Journal o f Human Resources, 32(3), 577-595. Hill, M. A., & King, E. M. (1993). W omen's education in developing countries: An overview. In E. M. King & M. A. Hill (Eds.), Women' s Education in Developing Countries. Baltimore, M l): The Johns Hopkins University Press. International Labour Organisation. (2000). Yearbook of Labour SM isties. Geneva: International Labour Office. Jackson, P. (2004). "Ethnicity, decentralisation, and the fissile state in Georgia." Public Administration & Development 24(75-86). Jejeebhoy, S. R. (1995). Women' s Education, Autonomy, and Reproduct ive Behavior. Experience from Developing Countries. Oxford: Clarendon Press. Jones, P. W. (1992). W orld Bank Fi nancing of Education: Fending Learning, and Development. London: Routledge. Jones, P. W. (1998). "Globalisation and internationalism: Democratic prospects for World education." Comparative Education 34(2): 143-155. Kardam, N . (1997). Making development organizations accountable: The organizational, political and cognitive contexts. Getting Institutions Rightfor Women in Development. A. M. Goetz. New York: St. Martins Press, Inc. Kasarda, J. D., Billy, J. O. G., & West, K. (1986). Status Enhancement and Fertility: Reproduct ive Responses to Social M obility and Educational Opportunity. Orlando, FL: Academic Press, Inc. Kaufman, D., & Wang, Y. (1995). Macroeconomic policies and project performance in the social sectors: A model o f human capital production and evidence from LDC's. World Development, 23(5), 751-765. Khan, M. H. (2002). Corruption and governance in early capitalism: World Bank strategies and their limitations. Reinventing the World Bank. J. R. Pincus and J. Winters, A. Ithaca: Cornell University Press. Knoke, D., Bohmstedt, G. W., & Mee, A. P. (2002). Statistics fo r Social D ata Analysis. Itasca, Illinois: F.E. Peacock Publishers. Krishna, A. (2003). "Partnerships between local government and com m unity-based organisations: Exploring the scope for synergy." Public Administration < & Development‘23: 361-371. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 Lam, D., & Duryea, S. (1999). Effects o f schooling on fertility, labor supply, and investments in children, with evidence from Brazil. The Journal of Human Resources, 34(1), 160-192. Lee, J. and K. R. W ong (2004). "The impact o f accountability on racial and socioeconomic equity. Considering both school resources and achievement outcomes." American Educational Research Journal41(4): 797-832. Levin, H. M. (1974). "A conceptual framework for accountability in education." The School Review 82(3): 363-391. Lloyd, C., & Niemi, B. T. (1979). The Economics of Sex Differentials. New York: Columbia University Press. Lloyd, C., & Gage-Brandon, A. J. (1994). 1 Iigh fertility and chldren's schooling in Ghana: Sex differences in parental contributions and educational outcomes. Population Studies, 48(2), 293-306. Lockheed, M. E. and A. M. Verspoor (1991). Improving Primary Education in Development Countries. Oxford: Oxford University Press. Loebner, H., & Driver, E. D. (1973). Differential fertility in central India: A path analysis. Demography, 10(3), 329-350. Maclure, R. (1994). "Misplaced assumptions o f decentralization and participation in rural communities: Primary school reform in Burkina Faso." Comparative Education 30(3): 239-254. Mankiw, N. G., Romer, D., & Weil, D . N. (1992). A contribution to the empirics o f economic growth. The Quarterly Journal ofEconomics, 107(2), 407-437. Mason, K. O. (1993). The impact o f women's position on demographic change during the course o f development. In N. Federici, K. O. Mason & S. Sogner (Eds.), Women' s Position and Demographic Change. Oxford: Clarendon Press. McMahon, W. (1999). Education and Development: Measuring the Social Ben fits. Oxford: Oxford University Press. Meinzen-Dick, R. S., L. R. Brown, et al. (1997). "Gender, property rights, and natural resources." World Development 25(8): 1303-1315. Meyer, J. W., Ramirez, F. O., & Soysal, Y. N. (1992). World expansion o f mass education, 1870-1980. Soci ology of Education, 65(2), 128-149. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Mingat, A. and J. P. Tan (1986). "Expanding education through user charges: What can be achieved in Malawi and other LDCs?" Economics of Education Review 5(8): 273- 286. Nassar, H. (2003). Egypt: Structural adjustment and women's employment. Women and Globalisation in the A rab M iddle E ast. E. A. Doumato and M. P. Posusney. Boulder, CO: Lynne Rienner Publishers, Inc. Nassirpour, M. (1985). The effect o f oil revenue on the fertility pattern in Iran. The Journal o f Marriage and the Family, 47(3), 785-796. Olmstcad, J. (2003). Reexamining the fertility puzzle in MENA. Women and Globalisation in the A rab Middle E ast. E. A. Doumato and M. P. Posusney. Boulder, CO: Lynne Rienner Publishers, Inc. Pampel, F. C., & Tanaka, K. (1986). Economic development and female labor force participation: A reconsideration. Social Forces, 64(3), 599-619. Parish, W. L., & Willis, R. J. (1993). Daughters, education and family budgets: Taiwan experiences. The Journal o f Human Resources, 28(4), 863-898. Parry, T. R. (1997). "Achieving balance in decentralization: A case study o f education decentralization in Chile." W orld Development 25(2): 211-225. Pierson, C. (1998). "The new governance o f education: The conservative and education 1988-1997." Oxford Review o f Education 24(1): 131-142. Pillai, Y. K., & Wang, G.-z. (1999). Women' s Reproduct ive Rights in Developing Countries. Brookfield, Vermont: Ashgate Publishing Company. Powers, J. M. (2004). "High-stakes accountability and equity: Using evidence from California's Public Schools Accountability Act to address the issues in Williams v. State o f California." American Educational Research Journal41(4): 763-796. Pritchett, L. (1996). Where Has All the Education Gone? World Bank Policy Research Working Papers (No. 1581). Washington, D.C.: The W ork Bank. Ram, R. (1982). Sex differences in the labor market outcomes o f education. In G. P. Kelly & C. M. Elliott (Eds.), Women' s Education in the Third World: Comparative Perspecti ves. Albany: State University o f New York Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Raynolds, L. T. (1998). Harnessing women's work: Restructuring agricultural and industrial labor forces in the Dominican Republic. Economic Geography, 74(2), 149-169. Rondinelli, D . A., J. R. Nellis, et al. (1983). Decentralization in developing countries: A review o f recent experience. World Bank Staff Working Papers. Washington, DC: The World Bank. Rose, P., & Tembon, M. (1999). Girls and schooling in Ethiopia. In C. Heward & S. Bunwaree (Eds.), Gender, Education Development (pp. 85-100). London: Zed Books Ltd. Schultz, T. W. (1960). Capital formation by education. The Journal of Political Economy, 68(6), 571-583. Schultz, T. P. (1988). Education investments and returns. In H. Chenery & T. N. Srinivasan (Eds.), Handbook o f Development Economics (Vol. 1). Amsterdam: Elsevier Science Publishers, B.Y. Schultz, T. P. (2002). "Why governments should invest more to educate girls." World Development 30(2): 207-225. Sen, A. K. (1990). Gender and cooperative conflicts. In I. Tinker (Ed.), Persistent Inequalities. New York: Oxford University Press. Sen, G. (1996). "Gender, markets and states: A selective review and research agenda." World Development 24(5): 821-829. Semyonov, M. (1980). The social context o f women's labor force participation: A comparative analysis. The American Journal o f Soci ol ogy, 86(3), 534-550. Sender, J. (2002). Reassesing the roles o f the World Bank in Sub-Saharan Africa. Reinvent ing the World Bank. J. R. Pincus and J. Winters, A. Ithaca: Cornell University Press. Silvey, R. (2003). "Engendering social capital: Women workers and rural-urban networks in Indonesia's crisis.” World Development 31 (5): 865-879. Smock, A C. (1981). Women's Education in Developing Countries: Opportunities and Out comes. New York: Praeger Publishers. Standing, G. (1989). Global feminization through flexible labor. World Development, 17(7), 1077-1095. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 Stiglitz, J. E. (2002). Globalisation and Its Discontents. New York: W.W. Norton & Company. Stinner, W. P. (1979). Modernization and family extension in the Philippines: A social demographic analysis. The Journal o f Marriage and the Family, 42, 161-168. Stromquist, N. P. (1999). The impact o f structural adjustment programmes in Africa and Latin America. Gender, Education and Development: Beyond Access to Empowerment . C. Heward and S. Bunwaree. London: Z ed Books Ltd. Subbarao, K-, & Raney, L. (1993). Social gains from female education: A cross-national study. World Bank Discussion Papers. Washington, DC: The World Bank. Tan, J. P., K. H. Lee, et al. (1984). User charges for education: The ability and willingness to pay in Malawi. World Bank Staff W orking Paper No. 661. Washington, DC: The World Bank. Thobani, M. (1984). "Charging user fees for social services: Education in Malawi." Comparative Education Review 28(3): 402-423. Tzannatos, Z., & Kaur, I. (2003). Women in the M ENA labor market. In E. A. Doumato & M. P. Posusney (Eds.), Women and Globalisation in the A rab Middle East. Boulder, CO: Lynne Rienner Publishers, Inc. United Nations. (1999). WISTAT: Women's Iindicators Database. New York: United Nations Publications. United Nations. (2000). United National Millennium Declaration, http://www.developmentgoals.org. UNDP (2005). Human Development Reports, http://hdr.undp.org. Weinberger, K. and J. P. Jutting (2001). "Women's participation in local organizations: Conditions and constraints." World Development 29(8): 1391-1404. Winter, C. and R. Macina (1999). Girls'Education: W orld Bank Support t hrough the International Development Association. Wise, A. E. and L. Darling-1 lammond (1983). "Educational vouchers: Regulating their efficiency and effectiveness." Educational Researcher' 12(9): 9-12 18. Wolfensohn, J. D. (1998). A Proposal for a Comprehensive Development Framework: A Discussion Draft. Washington, DC: The World Bank. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 World Bank. (1995). Toward Gender Equality: The Role for Public Policy. Washington, DC: World Bank. World Bank, The (2001). Overview o f Girls’ Education, http://w w w l .worldbank.org / education/ girlseducation/ World Bank, The (2002). W orld Development Indicators. Washington, DC, World Bank. World Bank, The (2005). Projects and Operations, http ://w w w l.worldbank.org/ education/projects/ Wynd, S. (1999). Education, schooling, and fertility in Niger. In C. Heward & S. Bunwaree (Elds.), Gender, Education < & Development (pp. 101-116). London: Zed Books Ltd. Youssef, N. H. (1974). Women and W ork in Developing Societies. Westport, CT: Greenwood Press. Zwarteveen, M. Z. (1997). "Water: From basic need to commodity: A discussion on gender and water rights in the context of irrigation." World Development 25(8): 1335-1349. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Hospital conversions: The California experience

PDF

Business political influence on agency regulatory decision making: The efficacy of corporate political activity in Taiwan's CATV license awarding

PDF

Biopharmaceutical strategic alliances: Interorganizational dynamics and factors influencing FDA regulatory outcomes

PDF

Electric system restructuring and system reliability

PDF

An empirical analysis of foreign aid and the provision of international public goods

PDF

Assessment of prognostic comorbidity in hospital outcomes research: Is there a role for outpatient pharmacy data?

PDF

Globalization and the decline of the welfare state in less developed countries

PDF

Dialectical feminism and contradictory economic systems: Campesina daily life from Somoza to Aleman

PDF

Analysis of the role of government in Taiwan's industrialization and economic development

PDF

A methodology to identify high -risk patients with diabetes in the California Medicaid populations (Medi -Cal)

PDF

IT -enabled dynamic capabilities in new product development: Building a competitive advantage in turbulent environments

PDF

Before and after reform: The effects of structural adjustment on women's employment in Latin America

PDF

Environmental constraints, governance patterns, and organizational responsiveness: A study of representation in nonprofit organizations

PDF

An analysis of health risk selection and quality of care under Medicare fee -for -service and Medicare managed care health care systems

PDF

Dependence on foreign labor, quality of education and unemployment in the GCC countries: In search of solutions

PDF

Deviant cosmopolitanism: Transgressive globalization and traveling citizenship

PDF

Greater than the sum of her parts: A multi -axis analysis of Black women and political representation

PDF

Immigrants and housing attainment: Different forms of residential assimilation

PDF

Economic transition, micro- and small -scale enterprise in Zimbabwe, and the significance of participatory development to institutional economics

PDF

Does autonomy matter? Investigating research -based practices in charter and other public schools

Asset Metadata

Creator Lincove, Jane Arnold (author)

Core Title Demand -side obstacles to girls' education in developing countries

School Graduate School

Degree Doctor of Philosophy

Degree Program Public Administration

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,Political Science, public administration,women's studies

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Painter, Gary D. (committee chair), Conway, Delores (committee member), Graddy, Elizabeth (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-447287

Unique identifier UC11340347

Identifier 3196845.pdf (filename),usctheses-c16-447287 (legacy record id)

Legacy Identifier 3196845.pdf

Dmrecord 447287

Document Type Dissertation

Rights Lincove, Jane Arnold

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA