64665 2012 world development report GENDER EQUALITY AND DEVELOPMENT 2012 world development report Gender Equality and Development 2012 world development report Gender Equality and Development © 2011 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org All rights reserved 1 2 3 4 14 13 12 11 This volume is a product of the staff of the International Bank for Reconstruction and Devel- opment / The World Bank. The �ndings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Dan- vers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright .com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Of�ce of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. Softcover ISSN: 0163-5085 ISBN: 978-0-8213-8810-5 eISBN: 978-0-8213-8812-9 DOI: 10.1596/978-0-8213-8810-5 Hardcover ISSN: 0163-5085 ISBN: 978-0-8213-8825-9 DOI: 10.1596/978-0-8213-8825-9 Cover photo: Arne Hoel, World Bank Photo credits: Overview/World Bank, Part I/National Geographic, Part II/Kiet Vo, Part III/National Geographic Cover design: Critical Stages Figures design and infographics: Design Symphony, Cymetrics, Harkness Design, and Naylor Design For the �rts time, the World Development Report is published with a companino mobile app for the iPad. Key features include: access content from the WDR 2012 in multiple ways; browse by key messages; browse and search the report by topic, region, and keyboard; access the report overview and key messages document, both available in 7 languages; share and save features; and view tabular data from the report. For more information, visit bit.ly/ wdr2012app. Contents Foreword xiii Acknowledgments xv Abbreviations and data notes xvii Main messages xx Overview 2 Why does gender equality matter for development? 2 What does this Report do? 6 Where has there been the most progress in gender equality? 8 Where have gender inequalities persisted and why? 13 What is to be done? 22 The political economy of reforms for gender equality 35 A global agenda for greater gender equality 36 Notes 38 References 40 Introduction: A guide to the Report 46 Gender equality and development: Why do the links matter? 46 What does this Report do? 48 Navigating this Report: A roadmap 50 Notes 51 References 51 Part I Taking stock of gender equality 54 1 A wave of progress 56 Times are changing? 56 Rising global consensus for women’s rights 57 Better outcomes for women in many domains 59 Change begets change 66 Notes 69 References 69 v vi CONTENTS 2 The persistence of gender inequality 72 Severely disadvantaged populations 73 “Sticky� domains, despite economic progress 76 Reversals 85 “Sticky� gets “stickier� 87 Notes 88 References 89 Spread 1 Women’s pathways to empowerment: Do all roads lead to Rome? 94 Notes 97 References 97 Part II What has driven progress? What impedes it? 98 Explaining the framework 99 Applying the framework 101 Notes 102 References 102 3 Education and health: Where do gender differences really matter? 104 Endowments matter 105 Education 106 Health 117 Technical Annex 3.1 Computing the flow of missing girls at birth and excess female mortality after birth 139 Chapter summary: In reducing gender gaps in education and health, tremendous progress has been made where lifting a single barrier—in households, markets, or institutions—is suf�cient to improve outcomes. Progress has been slower either where multiple barriers need to be lifted at the same time or where a single point of entry produces bottlenecks 141 Notes 142 References 143 4 Promoting women’s agency 150 Women’s agency matters 151 Economic growth can promote women’s agency but has limited impact 152 Rights and their effective implementation shape women’s choices and voices 157 Social norms prevent—or promote—gains in women’s agency 168 Women’s collective agency can shape institutions, markets, and social norms 176 Chapter summary: Women continue to have less capacity than men to exercise agency 181 Contents vii Notes 182 References 184 Spread 2 The decline of the breadwinner: Men in the 21st century 194 Note 196 5 Gender differences in employment and why they matter 198 Understanding gender differences in productivity and earnings 201 What explains employment segregation by gender? A �rst look 210 Gender, time use, and employment segregation 215 Gender differences in access to productive inputs and employment segregation 224 Gender impacts of “aggregate� market and institutional failures 230 Breaking out of the productivity trap: How and why to do it 236 Chapter summary: Persistent employment segregation by gender traps women in low-productivity, low-paying jobs 239 Notes 240 References 242 6 Globalization’s impact on gender equality: What’s happened and what’s needed 254 The world is becoming more integrated—Recent trends and facts 255 Trade openness and ICTs have increased women’s access to economic opportunities 255 Adapt or miss the boat 264 Globalization could also promote more egalitarian gender roles and norms 267 Old problems, emerging risks 269 Is the glass half full or half empty? The need for public action 271 Chapter summary: Globalization has the potential to contribute to greater gender equality 271 Notes 272 References 273 Spread 3 Changing ages, changing bodies, changing times—Adolescent boys and girls 280 Note 283 Part III The role of and potential for public action 284 Choosing the right policies 285 Enabling policy implementation 285 The global agenda for action 286 7 Public action for gender equality 288 Policies to reduce gaps in health and education 289 Policies to improve economic opportunities 296 viii CONTENTS Policies to improve women’s agency 305 Avoiding the reproduction of gender inequalities across generations for adolescents and young adults 314 Making gender-smart policies: Focusing “gender mainstreaming� 317 Wanted: Better evidence 320 Notes 321 References 323 8 The political economy of gender reform 330 Informal institutions—Social networks as agents of change 332 Inclusive markets 340 Bringing gender into formal institutions and policies 345 Seizing windows of opportunities 348 Pathways to change 350 Notes 354 References 355 9 A global agenda for greater gender equality 360 Rationale for and focus of a global agenda 360 What to do and how to do it 362 Notes 370 References 371 Bibliographical Note 373 Background Papers and Notes 377 Selected Indicators 381 Selected World Development Indicators 389 Index 411 Boxes 1 What do we mean by gender equality? 4 7 Intervening early to overcome future labor market failures— 2 The Millennium Development Goals recognize the intrinsic The Adolescent Girls Initiative 34 and instrumental value of gender equality 4 0.1 Problems with estimating the effect of gender equality on 3 How women and men de�ne gender in the 21st growth 49 century 7 4 What do we mean by markets, formal institutions, and 1.1 Gender and the Millennium Development Goals 58 informal social institutions? 8 2.1 The many faces of climate change 86 5 Reducing maternal mortality—What works? Look at Malaysia and Sri Lanka 25 3.1 Adult mortality risks: Who are the outliers? 119 6 Catalyzing female employment in Jordan 29 3.2 Four Africas 135 Contents ix 4.1 Pensions—Coverage, amounts, and survivor bene�ts are 6.3 Occupational tasks and skill requirements—Getting the important for women’s autonomy 156 terms right 259 4.2 Property in marriage (and divorce) 162 6.4 Leveraging mobile and ICT technology to improve access to 4.3 Widows risk losing their assets but might gain some services 264 freedom 163 6.5 Globalization and working conditions—Some progress, but 4.4 Legal pluralism and its prevalence 165 more needs to be done 267 4.5 What does it mean to be a “good wife� and a “good 7.1 Improving water supply: Dakar and Phnom Penh 291 husband�? 172 7.2 Reducing maternal mortality: What Malaysia and Sri Lanka 4.6 Masculinity and its impact on roles, preferences, and have done 295 behaviors 173 7.3 Protecting men and women and boys and girls from income 4.7 Why do social norms persist? 174 shocks 296 4.8 How stereotypes influence performance 175 7.4 Catalyzing female employment in Jordan 301 5.1 Closing the access gap—Recent advances in female labor 7.5 Innovative approaches to expanding access to �nance for force participation 199 women and entrepreneurs 303 5.2 Women in the boardroom 204 7.6 Including women’s voice in peace and postconflict reconstruction processes 308 5.3 Gender discrimination in hiring? Evidence from employment audit studies 205 8.1 Georgia—Evolving gender roles in a new society 332 5.4 What do we mean by employment segregation by 8.2 Feminism in perspective 334 gender? 206 8.3 Competing interests—Caste, ethnic, and religious politics 5.5 Good jobs and bad jobs: What are they and who does and gender 335 them? 211 8.4 More women in public of�ce—The Namibian Women’s 5.6 The seeds of segregation are planted early—How gender Manifesto Network 335 differences in education trajectories shape employment segregation 216 8.5 Differences among women about their right to vote— The case of Switzerland 336 5.7 Overview of data used in analyzing gender differences in time use patterns 218 8.6 Domestic workers in Brazil 337 5.8 What did you do all day? Perceptions on time use patterns of 8.7 How popular culture can change social attitudes 339 the opposite sex 221 8.8 Four good practices for greater gender diversity 342 5.9 Gender of the household head versus household 8.9 Land titling in Peru—Using a gender lens for a gender- composition: What matters most for policy? 225 neutral program 346 5.10 Family formation and public sector employment in 8.10 Gender machineries in practice 347 Egypt 232 8.11 Courts and constitutional challenges in Uganda’s divorce 5.11 The business case for gender equality 238 law 348 8.12 Fiji: International norms as a driver of gender equality in 6.1 A job today or a better job tomorrow—The impact of family law 349 increased access to economic opportunities on women’s human capital 258 8.13 Changing social norms from the bottom up 352 6.2 The impact of globalization on men (and women) in 8.14 Tunisia—Women’s voice and women’s rights 353 developed countries 259 8.15 Sweden—Encouraging an involved fatherhood 353 Figures 1 Gender outcomes result from interactions between 5 Female labor force participation has increased over time at households, markets, and institutions 9 all income levels 12 2 Across the world, women are having fewer children 9 6 Low-income countries lag behind in realizing progress in 3 Gender parity in enrollments at lower levels has been female school enrollment 13 achieved in much of the world, but tertiary enrollments are 7 Female disadvantage within countries is more marked at very low and favor women 10 low incomes 14 4 Using the framework to explain progress in 8 Women and men work in different sectors 16 education 11 9 Explaining persistent segregation and earnings gaps 18 x CONTENTS 10 Across the world, women spend more hours per day on care 3.1 Gender parity in enrollments at lower levels has been and housework than men 19 achieved in much of the world, but tertiary enrollments are 11 Gender differences in agricultural productivity disappear very low and favor women 107 when access to and use of productive inputs are taken into 3.2 In most countries with moderate or high total inequality in account 20 educational outcomes, less than one-�fth of inequality stems from gender 108 B0.1 GDP per capita and gender equality are positively 3.3 What explains progress in school enrollments? 109 correlated 49 3.4 Free primary education reduced gender gaps in 1.1 Gender parity in enrollments at lower levels has been enrollments 110 achieved in much of the world, and tertiary enrollments 3.5 Cross-country differences in mean scores on the 2009 PISA now favor women 61 dwarf gender differences within countries 114 1.2 Gender explains little of the inequality in education 3.6 Adult and child mortality in Sub-Saharan Africa 118 participation for children 12–15 years old 63 3.7 Adult mortality: Over time and by sex 119 1.3 Women are living longer than men 64 3.8 Income growth did not reduce excess female mortality during 1.4 What took the United States 100 years took India 40 and 1990–2008 in low- and middle-income countries 123 the Islamic Republic of Iran 10 64 3.9 Why are so many girls missing at birth? 124 1.5 Gender explains little of the inequality in use of preventive 3.10 There is little or no gender disadvantage in vaccination rates, health services 65 nutrition outcomes, or use of health services when a child 1.6 The gender gap in labor force participation narrowed falls sick 126 between 1980 and 2008 66 Small differences do not explain the variation in the fraction 1.7 Across countries, at every income level, female labor force of excess deaths across countries 126 participation increased between 1980 and 2008 66 3.11 Men and women, boys and girls, are treated the same when 1.8 Who agrees that a university education is more important for they visit health facilities 127 a boy than for a girl? 68 3.12 Levels of excess female childhood mortality in high-income 1.9 Who agrees that when jobs are scarce, men should have more countries in the early 1900s were similar to those of low- and right to a job than women? 68 middle-income countries today . . . 128 . . . and the excess female mortality declined with reduction in 2.1 Female enrollments remain strikingly low in some overall childhood mortality 128 countries 73 3.13 Maternal mortality ratios declined steeply in selected 2.2 In some countries, female disadvantage augments at lower countries during 1930–60 129 incomes . . . 74 3.14 High income countries today had excess female mortality 2.3 . . . yet in others, at low levels of wealth girls stay longer in at the reproductive ages during the �rst half of the school than boys 75 20th century . . . 130 2.4 At low incomes, fertility rates remain high—And the poorer . . . and the excess mortality at all income levels declines with the country, the larger the gap between rich and poor 76 reductions in maternal mortality 130 2.5 Maternal mortality in many developing countries is similar to 3.15 What explains excess mortality among girls and women in that in Sweden before 1900 78 the reproductive ages? 131 2.6 Women are more likely than men to work in the informal 3.16 Excess female mortality by age in four countries with high sector 79 HIV prevalence 132 2.7 Women and men work in different sectors (and different 3.17 In some countries, there is excess male mortality 133 occupations) 80 3A.1 Sex ratio and age-speci�c mortality, 2008 139 2.8 Across the world, women spend more hours each day on housework and care than men . . . and men spend more time 3A.2 Excess female mortality globally at each age in 2008 using various reference groups 140 in market activities 81 2.9 Who controls women’s own income? 82 4.1 Witnessing violence as a child is associated with perpetrating 2.10 Perceptions in many nations are that wife-beating is violence as an adult 152 justi�able 83 4.2 Limited progress in women’s agency is explained by mutually 2.11 There is great heterogeneity in rates of domestic violence reinforcing constraints in markets, formal institutions, and reported across nations 84 informal institutions 153 2.12 Men are perceived as better political leaders than 4.3 Richer women marry later 154 women 85 4.4 Women’s control is greater in wealthier households 155 Contents xi 4.5 Form of acquisition of land by gender in six Latin American 5.9 Women bear the brunt of housework and care while men are countries 155 mostly responsible for market work 219 4.6 Working outside the home broadens men’s and women’s 5.10 Convergence in gender time use patterns is stronger for networks (almost) equally 157 market work than for housework and care work 220 4.7 In most countries, access to contraceptives is a less signi�cant 5.11 In Mexico and Thailand, married women are more likely to constraint than lack of knowledge and opposition to move between inactivity and informal self-employment than contraception 158 men and single women 222 4.8 Progress on inheritance is faster for daughters than for 5.12 Female-headed households are less likely to own and operate widows 160 land than male-headed households 226 4.9 Few women seek services in case of domestic violence 168 5.13 Female-headed households in rural areas are less likely 4.10 Education dampens normative constraints more than than male-headed households to have received credit in the income 169 last 12 months 227 4.11 Despite differences in the age of marriage, many girls still 5.14 Access to productive inputs and markets is lower among marry before the age of 18 170 female-headed households than among male-headed 4.12 Reasons why victims of violence do not seek help 171 households 228 4.13 Even in 2010, women ministers were twice as likely to hold a 5.15 Mutually reinforcing market and institutional constraints are social portfolio than an economic one 177 the main reason why women appear to be in a productivity trap 237 4.14 Women’s voice in society is limited by social norms on women’s roles and abilities and by formal B5.1.1 Participation rates—Converging 200 institutions 178 B5.8.1 Understanding the amounts of time the opposite sex spends 4.15 Perceptions on leadership skills are still very prevalent, and on nontraditional male/female activities and leisure 221 less educated cohorts are more biased 179 6.1 Global trade has grown rapidly since 1990 255 4.16 Women are much less likely to belong to a political party 6.2 Cell phone and Internet access has increased signi�cantly in than men 179 both developed and developing countries 256 B4.1.1 Sources of income for China’s elderly, 2005 156 6.3 Economic opportunities have changed 257 B4.3.1 Elderly women are more likely to live alone and elderly men 6.3a Female (and male) employment in the manufacturing and with their spouses 163 service sectors has grown faster in developing countries, B4.3.2 Husband’s family receives the majority of his assets in most reflecting the broader changes in the global distribution of countries 163 production and labor 257 5.1 There are systematic gender differences in earnings 202 6.3b . . . and increases in female employment levels (but not male) between 1995 and 2005 were correlated with increases in 5.2 Women are overrepresented among wage and unpaid family international trade 257 workers 207 6.4 The United States experienced a dramatic increase in brain 5.3 Gender differences in agricultural productivity diminish requirements and a decline in brawn requirements between considerably when access to and use of productive inputs are 1950 and 2005 260 taken into account 208 6.5 Men and especially women in Brazil, India, Mexico, and 5.4 Differences in productivity between female and male Thailand have experienced an increase in brain requirements entrepreneurs are dwarfed by differences in productivity and a decline in brawn requirements over the past 15 between formal and informal entrepreneurs 208 years 261 5.5 Economic development is positively correlated with the share of female workers in wage employment and negatively 6.6 In Africa, women are less likely than men to own or use a correlated with the share of women in unpaid work, self- cell phone 262 employment, and entrepreneurship 212 6.7 Differences in Internet access and use between developed 5.6 Tanzania and Brazil illustrate how employment patterns by and developing countries are still very large, and gender gender change with economic growth 213 gaps are signi�cant in some developed and developing countries 263 5.7 Industry and occupational segregation patterns are common across countries with very different levels of economic 6.8 Telework has grown rapidly in recent years, particularly development and aggregate sectoral distributions of among female workers 265 employment 214 6.9 The share of female employment varies signi�cantly across 5.8 Access to economic opportunities and the resulting industries 266 segregation in employment are the product of households, 6.10 The number of countries that have rati�ed CEDAW has risen markets, and institutions, and their interactions 217 in all regions to reach 187, of 193, in 2011 266 xii CONTENTS 7.1 Reducing gaps in endowments 290 8.6 Progress toward increasing women’s rights is 7.2 Improving economic opportunities 296 clustered around major international human rights conferences 350 7.3 Improving women’s agency 306 S1.1 Main factors in moving up the ladder 94 8.1 Social actors and their interactions shape the role of markets, formal and informal institutions in advancing gender S1.2 Ladder shares now and 10 years ago in Bukoba, equality 331 Tanzania (urban) 96 8.2 Social networks can engage public opinion, mobilize support, S1.3 Ladder shares now and 10 years ago in Dhamar, Republic of and inspire change 333 Yemen (rural) 96 8.3 Men around the world support women’s rights and S2.1 Factors that explain gains in power 195 policies 338 S2.2 Factors that explain losses of power 196 8.4 Economic and political economy considerations have prodded �rms to promote gender equality S3.1 Characteristics of good girls/boys and bad girls/boys 281 policies 341 S3.2 Adolescents opinions on women’s roles 283 8.5 State action is central for the design and adoption of gender-progressive policies 345 P2.1 Gender outcomes result from interactions between households, markets, and institutions 101 Maps 1 Earnings gaps between women and men 17 B4.2.1 Women in different parts of the world have different control over assets—Which matters in case of divorce or 0.1 Economies where qualitative assessments were the husband’s death 162 conducted 50 5.1 Gender differences in occupation and industry of 3.1 In China and India, the number of girls missing at birth employment account for a large fraction of the gender gap remains high, and parts of Africa experienced large increases after accounting for individual characteristics 209 in excess female mortality during 1990–2008 122 B5.1.1 Female labor force participation—Some high rates and 4.1 In Sub-Saharan Africa, customary laws are formally some low 199 recognized in most countries, and at times are discriminatory 166 Tables 1 Almost 4 million missing women each year 15 3.2 Skewed sex ratios at birth and excess female mortality 2 The agenda for global action at a glance 38 persist across the world, leading to females missing at birth and excess female mortality during childhood and the 2.1 Missing girls at birth increased between 1990 and 2008 in reproductive years 121 India and China, as did excess female mortality in adulthood in Sub-Saharan Africa 77 5.1 Female farmers have lower average productivity than male farmers 202 3.1 Gender segregation in �eld of study: In most countries, women dominate health and education studies and men 9.1 The agenda for global action at a glance 361 dominate engineering and sciences 115 Foreword The lives of girls and women have changed dramatically over the past quarter century. Today, more girls and women are literate than ever before, and in a third of developing countries, there are more girls in school than boys. Women now make up over 40 percent of the global labor force. Moreover, women live longer than men in all regions of the world. The pace of change has been astonishing—indeed, in many developing countries, they have been faster than the equivalent changes in developed countries: What took the United States 40 years to achieve in increasing girls’ school enrollment has taken Morocco just a decade. In some areas, however, progress toward gender equality has been limited—even in devel- oped countries. Girls and women who are poor, live in remote areas, are disabled, or belong to minority groups continue to lag behind. Too many girls and women are still dying in child- hood and in the reproductive ages. Women still fall behind in earnings and productivity, and in the strength of their voices in society. In some areas, such as education, there is now a gen- der gap to the disadvantage of men and boys. The main message of this year’s World Development Report: Gender Equality and Develop- ment is that these patterns of progress and persistence in gender equality matter, both for development outcomes and policy making. They matter because gender equality is a core development objective in its own right. But greater gender equality is also smart economics, enhancing productivity and improving other development outcomes, including prospects for the next generation and for the quality of societal policies and institutions. Economic develop- ment is not enough to shrink all gender disparities—corrective policies that focus on persist- ing gender gaps are essential. This Report points to four priority areas for policy going forward. First, reducing gender gaps in human capital—speci�cally those that address female mortality and education. Sec- ond, closing gender gaps in access to economic opportunities, earnings, and productivity. Third, shrinking gender differences in voice and agency within society. Fourth, limiting the reproduction of gender inequality across generations. These are all areas where higher incomes by themselves do little to reduce gender gaps, but focused policies can have a real impact. Public actions need to address the underlying determinants of gender gaps in each pri- ority area—in some cases, improving service delivery (especially for clean water, sanitation, and maternal care), for others, tackling constraints that originate in the workings of mar- kets and institutions to limit progress (for example, in reducing gender gaps in earnings and productivity). Development partners can complement public action. In each of the four priority areas, efforts need more funding (particularly to support the poorest countries as they address female mortality and gender gaps in education); better gender-disaggregated data; more experimentation and systematic evaluation; and broader partnerships that include the private sector, development agencies, and civil society organizations. xiii xiv F O R E WO R D Gender equality is at the heart of development. It’s the right development objective, and it’s smart economic policy. The World Development Report 2012 can help both countries and international partners think through and integrate a focus on gender equality into develop- ment policy making and programming. Robert B. Zoellick President The World Bank Group Acknowledgments This Report has been prepared by a core team led by Ana Revenga and Sudhir Shetty, and comprising Luis Benveniste, Aline Coudouel, Jishnu Das, Markus Goldstein, Ana María Muñoz Boudet, and Carolina Sánchez-Páramo. Research assistance was provided by Rabia Ali, María Inés Berniell, Rita Costa, Nina Rosas, and Lucía Solbes Castro. The multi- country qualitative assessment was coordinated by Patti L. Petesch and Carolyn Turk. Extensive and valuable contributions were made by Andre Croppenstedt, Malcolm Ehrenpreis, Rebekka Grun, Mary Hallward-Driemeier, Tazeen Hasan, Karla Hoff, Ghazala Mansuri, Claudio E. Montenegro, and Bob Rijkers. The World Development Report 2012 is co-sponsored by the Development Econom- ics Vice-Presidency (DEC) and the Poverty Reduction and Economic Management Vice- Presidency (PREM). The work was conducted under the joint guidance of Justin Yifu Lin in DEC and Otaviano Canuto dos Santos Filho in PREM. Ann E. Harrison and the DEC team and Mayra Buvinic and the PREM Gender (PRMGE) team provided valuable guidance and contributions at various stages of the production of this report. A panel of advisers comprising Bina Agarwal, Ragui Assad, Anne Case, Alison Evans, Raquel Fernández, Naila Kabeer, Ravi Kanbur, Santiago Levy, and Germano Mwabu provided excel- lent advice. Valuable comments and contributions were provided by Kathleen Beegle, Laura Chioda, Louise Cord, Maria Correia, Monica Das Gupta, Shantayanan Devarajan, Marianne Fay, Francisco H. G. Ferreira, Ariel Fiszbein, Indermit Gill, Alejandro Hoyos, Emmanuel Jime- nez, Elizabeth King, Andrew Mason, William Maloney, Ambar Narayan, Pierella Paci, Tara Vishwanath, and Michael Walton. Many others inside and outside the World Bank contrib- uted with valuable comments and input (their names are listed in the Bibliographical Note). World Bank President Robert B. Zoellick and Managing Directors Sri Mulyani Indrawati, Mahmoud Mohieldin, and Ngozi Okonjo-Iweala provided invaluable guidance and advice. The team bene�ted greatly from many consultations, meetings, and regional workshops held locally and in-country. These discussions included policy makers, civil society represen- tatives, academics, and development partners from Benin, Bolivia, Burkina Faso, Burundi, the Caribbean nations, the Central African Republic, Chile, Colombia, the Dominican Repub- lic, Georgia, Guatemala, India, Indonesia, Jordan, Kenya, Kuwait, Lebanon, Mali, Mexico, Morocco, Panama, Paraguay, Rwanda, Senegal, the Slovak Republic, South Africa, Sudan, Tanzania, Thailand, Togo, Turkey, Uganda, Uruguay, Vietnam, Zambia, and Zimbabwe. Consultations were also held at different stages of report preparation with representatives from multilateral and bilateral partners, including the Australian Agency for International Development (AUSAID), the Canadian International Development Agency (CIDA), the Inter-American Commission of Women-Organization of American States (CIM-OAS), the Danish International Development Agency (DANIDA), the U.K. Department for International Development (DFID), the Food and Agriculture Organization (FAO), the Japan International Cooperation Agency (JICA), the International Labour Organization (ILO), the Ministry for Foreign Affairs of Finland, MCC, NORAD, the Organisation for Economic Co-operation xv xvi AC K N OW L E D G M E N T S and Development-Development Assistance Committee (OECD-DAC) Gendernet, the Swiss Agency for Development and Cooperation (SDC), the Swedish International Development Cooperation Agency (SIDA), UN Women, the United Nations Children’s Fund (UNICEF), the United States Agency for International Development (USAID), and the United Nations Economic and Social Council (ECOSOC) 55th Commission on the Status of Women. The team would like to acknowledge the generous support of the Government of Nor- way through its Royal Ministry of Foreign Affairs, SDC, AUSAID, CIDA, the Government of Sweden through its Ministry for Foreign Affairs, the multi-donor Knowledge for Change Program (KCP), the Nike Foundation, the World Bank Nordic Trust Fund, and Fast Track Initiative Education Program Development Fund; as well as the in-kind support from JICA, DFID, and OECD. The team wishes to acknowledge the excellent support of the WDR production team com- prising Rebecca Sugui, Cecile Wodon, and Mihaela Stangu, and of the resource management team of Sonia Joseph and Evangeline Santo Domingo. We thank also Ivar Cederholm, Vivian Hon, Jimmy Olazo, and Irina Sergeyeva for their constant support. Other valuable assistance was provided by Gytis Kanchas and Nacer Mohamed Megherbi. Vamsee Krishna Kanchi, Swati P. Mishra, Merrell Tuck-Primdahl, and Roula Yazigi assisted the team with the website and communications. Bruce Ross-Larson was the principal editor. The Development Data Group contributed to the data appendix and was responsible for the Selected World Development Indicators. Design Symphony contributed to the design. The Of�ce of the Publisher and GSDTR pro- vided excellent publishing, translation, and dissemination services, with special thanks to Mary Fisk, Stephen McGroarty, Nancy Lammers, Santiago Pombo-Bejarano, Denise Bergeron, Rick Ludwick, Cecile Jannotin, Hector Hernaez, and Bouchra Belfqih for their contributions. Abbreviations and data notes ABBREVIATIONS AIDS acquired immunode�ciency syndrome ALMPs active labor market policies ANC African National Congress APEC Asia-Paci�c Economic Cooperation ART antiretroviral therapy ASEAN Association of Southeast Asian Nations ATM automated teller machine AUSAID Australian Agency for International Development BPO business process outsourcing CARICOM Caribbean Community CCT conditional cash transfer CEDAW Convention on the Elimination of All Forms of Discrimination against Women CGAP Consultative Group to Assist the Poor CIDA Canadian International Development Agency CIM-OAS Inter-American Commission of Women (Organization of American States) CWDI Corporate Women Directors International DANIDA Danish International Development Agency DFCU Development Finance Company of Uganda DFID United Kingdom Department for International Development EAP East Asia and Paci�c Region ECA Europe and Central Asia Region ECD early child development ECOSOC Economic and social council (United Nations) EdAttain Education Attainment and Enrollment around the World database EFM excess female mortality EU European Union EU-SILC European Union Statistics on Income and Living Conditions FAO Food and Agriculture Organization FDI foreign direct investment FENATRAD Federação Nacional dos Trabalhadores Domésticos (National Federation of Domestic Workers) FGC female genital cutting FHHH female-headed households FINCA Foundation for International Community Assistance FLFPR female labor force participation rate FPE free primary education xvii xviii A B B R E V I AT I O N S A N D DATA N O T E S GBA Global Banking Alliance for Women GDP gross domestic product GEME Gender Equity Model Egypt HDI Human Development Index HIV human immunode�ciency virus I2D2 International Income Distribution Database ICRW International Center for the Research on Women ICT information and communications technology IFC International Finance Corporation ILO International Labor Organization ITES information technology enabled service ITU International Telecommunications Union JICA Japan International Cooperation Agency KCP Knowledge for Change Program LABORSTA International Labour Organization Bureau of Statistics database LAC Latin America and the Caribbean Region LFPRs labor force participation rates MCC Millennium Challenge Corporation MDGs Millennium Development Goals MNA Middle East and North Africa Region MHHH male-headed households MMR maternal mortality rate MTUS Multinational Time Use Study NAALC North American Agreement on Labor Cooperation NAFTA North American Free Trade Agreement NGO nongovernmental organization NHO Næringslivets Hovedorganisasjon (Confederation of Norwegian Enterprises) NORAD Norwegian Agency for Development Cooperation NSS national statistical systems OECD Organisation for Economic Co-operation and Development OECD-DAC Gendernet Development Assistance Committee’s Network on Gender Equality of the Organisation for Economic Co-operation and Development PEKKA Pemberdayaan Perempuan Kepala Keluarga (Female-Headed Households Empowerment Program) PETT Proyecto Especial de Titulación de Tierras (Special Land Titling Project) PISA Program for International Student Assessment PROBECAT Programa de Becas de Capacitación para Trabajadores (Labor Retraining Scholarship Program) REFLEX Research into Employment and Professional Flexibility database RIGA Rural Income Generating Activities database ROSCAs rotating savings and credit associations SADC Southern African Development Community SAR South Asia Region SDC Swiss Agency for Development and Cooperation SERNAM Servicio Nacional de la Mujer (National Women’s Service) SEWA Self-employed Women’s Association SIDA Swedish International Development Cooperation Agency SMEs small and medium enterprises Abbreviations and data notes xix SSA Sub-Saharan Africa Region UIS UNESCO Institute for Statistics UNAIDS Joint United Nations Program on HIV/AIDS UNDP United Nations Development Programme UNESCO United Nations Educational, Scienti�c and Cultural Organization UN-HABITAT United Nations Human Settlements Program UNICEF United Nations Children’s Fund UNIFEM United Nations Development Fund for Women UNRISD United Nations Research Institute for Social Development UN WOMEN United Nations Entity for Gender Equality and the Empowerment of Women USAID United States Agency for International Development WHO World Health Organization WINGOs women’s international nongovernmental organizations DATA NOTES The countries included in regional and income groupings in this Report are listed in the Classi�cation of Economies table at the end of the Selected World Development Indicators. Income classi�cations are based on GNP per capita; thresholds for income classi�cations in this edition may be found in the Introduction to Selected World Development Indicators. Group averages reported in the �gures and tables are unweighted averages of the countries in the group, unless noted to the contrary. The use of the word countries to refer to economies implies no judgment by the World Bank about the legal or other status of a territory. The term developing countries includes low- and middle-income economies and thus may include economies in transition from central planning, as a matter of convenience. The term advanced countries may be used as a matter of convenience to denote high-income economies. Note: Dollar �gures are current U.S. dollars, unless otherwise speci�ed. Billion means 1,000 million; trillion means 1,000 billion. Main Messages of the World Development Report 2012 GENDER EQUALITY MATTERS Empowering women as economic, political, FOR DEVELOPMENT and social actors can change policy choices and make institutions more representative of Gender equality is a core development objec- a range of voices. In India, giving power to tive in its own right. It is also smart economics. women at the local level led to increases in Greater gender equality can enhance produc- the provision of public goods, such as water tivity, improve development outcomes for the and sanitation, which mattered more to next generation, and make institutions more women. representative. • Productivity gains. Women now represent DEVELOPMENT HAS CLOSED SOME 40 percent of the global labor force, 43 per- cent of the world’s agricultural labor force, GENDER GAPS . . . and more than half the world’s university The disadvantages faced by women and girls students. Productivity will be raised if their that have shrunk most rapidly over the past skills and talents are used more fully. For ex- quarter century include: ample, if women farmers were to have the same access as men to fertilizers and other • Educational enrollment. Gender gaps in pri- inputs, maize yields would increase by almost mary education have closed in almost all one-sixth in Malawi and Ghana. And elimi- countries. In secondary education, these gaps nating barriers that discriminate against are closing rapidly and have reversed in many women working in certain sectors or occupa- countries, especially in Latin America, the tions could increase labor productivity by as Caribbean, and East Asia—but it is now much as 25 percent in some countries. boys and young men who are disadvantaged. Among developing countries, girls now out- • Improved outcomes for the next generation. number boys in secondary schools in 45 Greater control over household resources by countries and there are more young women women can enhance countries’ growth pros- than men in universities in 60 countries. pects by changing spending patterns in ways that bene�t children. And improvements in • Life expectancy. Since 1980, women are liv- women’s education and health have been ing longer than men in all parts of the world. linked to better outcomes for their children And, in low-income countries, women now in countries as varied as Brazil, Nepal, Paki- live 20 years longer on average than they did stan, and Senegal. in 1960. • More representative decision making. Gen- • Labor force participation. Over half a billion der equality matters for society more broadly. women have joined the world’s labor force Main Messages of the World Development Report 2012 xxi over the last 30 years as women’s partici- UNDERSTANDING PROGRESS pation in paid work has risen in most of AND PERSISTENCE the developing world. An important rea- son has been the unprecedented reduc- Income growth by itself does not deliver tion in fertility in developing countries greater gender equality on all fronts. Indeed, as diverse as Bangladesh, Colombia, and where gender gaps have closed quickly, it is the Islamic Republic of Iran. because of how markets and institutions— formal and informal—have functioned and evolved, how growth has played out, and . . . BUT OTHER GAPS PERSIST how all these factors have interacted through household decisions. For example, in educa- Gender disparities still remain in many areas, tion, income growth (by loosening bud- and even in rich countries. The most persis- get constraints), markets (by opening new tent and egregious gaps include: employment opportunities for women), and formal institutions (by expanding schools • Excess deaths of girls and women. Fe- and lowering costs) have all come together males are more likely to die, relative to to influence household decisions in favor of males, in many low- and middle-income educating girls and young women across a countries than their counterparts in rich broad range of countries. countries. These deaths are estimated at about 3.9 million women and girls under Gender gaps persist where girls and women the age of 60 each year. About two-�fths face other disadvantages. For poor women in of them are never born, one-sixth die in poor places, sizable gender gaps remain. And early childhood, and over one-third die in these disparities are even larger when pov- their reproductive years. And this number erty combines with other forms of exclusion, is growing in Sub-Saharan Africa, espe- such as remoteness, ethnicity, and disability. cially in childhood and the reproductive For ethnic minority women in Vietnam, for years and in the countries hardest hit by instance, more than 60 percent of childbirths the HIV/AIDS epidemic. occur without prenatal care—twice as many • Disparities in girls’ schooling. Despite as for the majority Kinh women. the overall progress, primary and second- ary school enrollments for girls remain Markets, institutions, and households can much lower than for boys for disadvan- also combine to limit progress. Gender gaps taged populations in many Sub-Saharan in productivity and earnings, for exam- countries and some parts of South Asia. ple, are pervasive. And they are driven by deep-seated gender differences in time use • Unequal access to economic opportuni- (reflecting social norms about house and ties. Women are more likely than men to care work), in rights of ownership and con- work as unpaid family laborers or in the trol over land and assets, and in the workings informal sector. Women farmers tend to of markets and formal institutions, which farm smaller plots and less pro�table work in ways that disadvantage women. crops than men. Women entrepreneurs operate in smaller �rms and less pro�t- Globalization can help. In today’s globalized able sectors. As a result, women every- world, forces such as trade openness and the where tend to earn less than men. spread of cheaper information and com- • Differences in voice in households and munication technologies have the potential in society. In many countries, women— to reduce gender disparities by connecting especially poor women—have less say women to markets and economic opportu- over decisions and less control over re- nities, reshaping attitudes and norms among sources in their households. And in most women and men about gender relations, countries, women participate less in for- and encouraging countries to promote gen- mal politics than men and are under- der equality. But their impact will be muted represented in its upper echelons. without effective domestic public action. xxii WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 PRIORITIES FOR DOMESTIC and young women when poverty, ethnic- POLICY ACTION ity, or geography excludes them, and to reach boys where gender disadvantages Policy makers in developing countries will have reversed. Cash transfers conditioned need to focus on those gender gaps where on school attendance are often effective in the payoffs for development are potentially reaching these groups. Pakistan has used the largest, higher incomes by themselves do such transfers to get girls from poor fami- little to reduce these gaps, and a reorienta- lies to school, while Jamaica has relied on tion of policies would yield the greatest ben- them to keep at-risk boys in school. e�t. These priorities are: • To narrow disparities between women • Addressing excess deaths of girls and and men in earnings and productivity, women and eliminating gender disadvan- a combination of policies is needed to tage in education where these remain address the various constraints that dis- entrenched. proportionately affect women’s access to economic opportunities. Depending on • Closing differences in access to economic context, these include measures to: opportunities and the ensuing earnings Lift women’s time constraints, by pro- and productivity gaps between women viding child care as with Colombia’s and men. subsidized day-care programs for work- • Shrinking gender differences in voice ing mothers, and improving infrastruc- within households and societies. ture as with South Africa’s rural electri- �cation program. • Limiting the reproduction of gender in- equality across generations. Improve women’s access to productive resources, especially to land as was done Focused and sustained domestic pub- in Ethiopia by granting joint land titles lic action is essential to bring about gender to wives and husbands, and to credit as equality. And to be effective, these policies will in Bangladesh. need to target the root causes of gender gaps. Tackle information problems and in- In some areas, as with maternal mortality, stitutional biases that work against governments will need to address the single women. These include the use of quo- binding constraint to progress (weak service tas or job placement programs as is be- delivery institutions). In others, as with dif- ing done in Jordan, or reforming gen- ferential access to economic opportunities, der biases in service delivery institutions policies will be needed that tackle the multi- as was done for agricultural extension ple constraints that come from the workings through women’s self-help groups in of markets and institutions to limit progress. the Indian state of Orissa. In these cases, policy makers will need to pri- oritize these constraints and address them • To diminish gender differences in house- simultaneously or sequentially. hold and societal voice, policies need to address the combined influence of social • To reduce excess deaths of girls and norms and beliefs, women’s access to eco- women in infancy, early childhood, and nomic opportunities, the legal framework, the reproductive years, policy action to and women’s education and skills: improve the delivery of services (especially To equalize voice within households, of clean water, sanitation, and maternal measures that increase women’s control care) is of primary importance. Vietnam over household resources and laws that has been able to reduce excess mortality enhance the ability of women to accu- among young girls by expanding access to mulate assets, especially by strengthen- clean water and sanitation. And Turkey ing their property rights, are of particu- has reduced maternal mortality through lar importance. Recent reforms of improved health care delivery and a focus family law in Morocco that equalized on expectant mothers. the ownership rights of husbands and • To shrink persisting educational gaps, wives over property acquired during policies need to improve access for girls marriage are an example. Main Messages of the World Development Report 2012 xxiii To increase women’s voice in society, complementing these efforts in each of these policies include quotas on political rep- four priority areas and, more generally, in resentation, as has been done by many supporting evidence-based public action countries across the world, and mea- through better data, impact evaluation, and sures to foster and train future women learning. leaders and involve women more in • In some areas, as with educational gender groups such as trade unions and pro- gaps, this will require adjusting current fessional associations. support, such as ensuring that the Educa- • To limit the reproduction of gender in- tion for All Fast Track Initiative reaches equality across generations, it is impor- disadvantaged girls and boys, or sustain- tant to reach adolescents and young adults ing existing efforts, as with partnerships because this is the age when they make de- focused on adolescent girls. cisions that determine their acquisition of • In other areas, it will demand new or skills, future health, economic prospects, additional action on multiple fronts— and aspirations. Interventions, therefore, some combination of more funding, co- need to focus on: ordinated efforts to foster innovation and Building human and social capital learning, and more effective partnerships. as cash transfer programs have done The funding should be directed partic- in Malawi, and improving informa- ularly to supporting the poorest coun- tion about returns to education and tries in reducing excess deaths of girls health education programs, which has and women (through investments in kept boys in school in the Dominican clean water and sanitation and mater- Republic; nal health services) and removing per- Facilitating the transition from school sistent gender gaps in education. to work with job and life skills training More support is needed especially to programs as in Uganda; and improve the availability of gender- Shifting aspirations as with exposure to disaggregated data and to foster more role models such as woman political experimentation and systematic evalu- leaders in India who challenge prevail- ation of mechanisms to improve wom- ing social norms. en’s access to markets, services, and justice. The partnerships should extend be- THE ROLE OF THE yond governments and development INTERNATIONAL COMMUNITY agencies to include the private sector, civil society organizations, and aca- While domestic policy action is crucial, the demic institutions in developing and international community can play a role in rich countries. World Development Report 2012: Gender Equality and Development Overview Baruani is reflecting on how women’s and men’s lives have changed over the past decade in Ijuhanyondo—a village in Tanzania. “Ten years back was terrible,� she recalls. “Women were very be- hind. They used to be only at home doing housework. But now, they are in businesses, they are in poli- tics.� Others hold similar views. “We do not depend a lot on men as it used to be,� says Agnetha. “We have some cash for ourselves, and this assists us in being free from men and to some extent controlling our lives.� In addition to managing their businesses, the women now make up half the members of the street committee that runs the village. Despite these positive changes, many challenges continue to weigh on women’s daily lives. Fewer than half the homes in the village have piped water. Even more dif�cult, Tungise and other women of the village still fear violence by their partners: “When they are drunk, they can begin beating up women and children in the house. The worst bit of it is forcing sex with you.� Although legally women can inherit land or a house, tradition prevails. “Yes, women can inherit property,� says Flora, the executive secretary of the street committee. “In fact, in the will the father is supposed to give each son and daughter some- thing, and nowadays the law is strict, equally. But still, men give to their sons and argue that women have the property of where they are married.� Dodoma Rural Community Report, from “De�ning Gender in the 21st Century: Talking with Women and Men around the World: A Multi-Country Qualitative Study of Gender and Economic Choice� (World Bank 2011) WHY DOES GENDER EQUALITY the law in such areas as property ownership, MATTER FOR DEVELOPMENT? inheritance, and marriage. In all, 136 countries now have explicit guarantees for the equality of The story of Ijuhanyondo village in Tanzania all citizens and nondiscrimination between men mirrors the evolution of gender equality across and women in their constitutions. the world over the past quarter century. Although Progress has not come easily. And it has not many women continue to struggle with gender- come evenly to all countries or to all women—or based disadvantages in their daily lives, things across all dimensions of gender equality. The have changed for the better—and at a pace that likelihood of women dying during childbirth would have been unthinkable even two decades in Sub-Saharan Africa and parts of South Asia ago. Women have made unprecedented gains in is still comparable to that in Northern Europe rights, in education and health, and in access to in the 19th century. A wealthy urban child in jobs and livelihoods. More countries than ever Nigeria—boy or girl—averages around 10 years guarantee women and men equal rights under of schooling, while poor rural Hausa girls aver- Overview 3 age fewer than six months. The rate at which advancement of women and has been rati�ed to women die relative to men is higher in low- and date by 187 countries. middle-income countries compared with their high-income counterparts, especially in the crit- Gender equality matters for ical years of infancy and early childhood and development—It is smart economics in the reproductive period. Divorce or widow- Gender equality matters also as an instrument hood causes many women to become landless for development. As this Report shows, gender and lose their assets. Women continue to clus- equality is smart economics: it can enhance eco- ter in sectors and occupations characterized as nomic ef�ciency and improve other develop- “female�—many of them lower paying. Women ment outcomes in three ways. First, removing are also more likely to be the victims of violence barriers that prevent women from having the at home and suffer more severe injuries. And same access as men to education, economic op- almost everywhere women’s representation in portunities, and productive inputs can generate politics and in senior managerial positions in broad productivity gains—gains all the more business remains far lower than men’s. important in a more competitive and globalized Do these patterns of gender inequality– world. Second, improving women’s absolute and in human and physical capital endowments, relative status feeds many other development in economic opportunities, and in the ability outcomes, including those for their children. to make choices to achieve desired outcomes Third, leveling the playing �eld—where women (agency)—matter, particularly those that per- and men have equal chances to become socially sist even as the development process unfolds? and politically active, make decisions, and shape This World Development Report (WDR) ar- policies—is likely to lead over time to more rep- gues that they do for two reasons. First, gender resentative, and more inclusive, institutions and equality matters intrinsically, because the abil- policy choices and thus to a better development ity to live the life of one’s own choosing and path. Consider each in turn. be spared from absolute deprivation is a basic human right and should be equal for everyone, Misallocating women’s skills and talent independent of whether one is male or female. comes at a high (and rising) economic cost Second, gender equality matters instrumentally, Gender equality can have large impacts on pro- because greater gender equality contributes to ductivity. Women now represent more than 40 economic ef�ciency and the achievement of percent of the global labor force, 43 percent of other key development outcomes. the agricultural workforce, and more than half of the world’s university students. For an econ- Gender equality matters in its own right omy to be functioning at its potential, women’s Following Amartya Sen, we see development skills and talents should be engaged in activities as a process of expanding freedoms equally for that make the best use of those abilities. But, all people.1 In this view of development, gen- as the stories of many women illustrate, this der equality is a core objective in itself (box is not always the case. When women’s labor is 1). So, just as development means less income underused or misallocated—because they face poverty or better access to justice, it should also discrimination in markets or societal institu- mean fewer gaps in well-being between males tions that prevents them from completing their and females. This viewpoint is also evident in education, entering certain occupations, and the international development community’s earning the same incomes as men—economic recognition that women’s empowerment and losses are the result. When women farmers gender equality are development objectives in lack security of land tenure, as they do in many their own right, as embodied in Millennium countries, especially in Africa, the result is lower Development Goals 3 and 5 (box 2). It is seen access to credit and inputs and to inef�cient as well in the adoption and widespread rati�ca- land use, reducing yields. Discrimination in tion of the Convention on the Elimination of credit markets and other gender inequalities in All Forms of Discrimination against Women access to productive inputs also make it more (CEDAW). Adopted by the United Nations dif�cult for female-headed �rms to be as pro- General Assembly in 1979, the convention es- ductive and pro�table as male-headed ones. tablished a comprehensive framework for the And, when women are excluded from manage- 4 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 BOX 1 What do we mean by gender equality? Gender refers to the social, behavioral, and cultural attributes, that arise from circumstances beyond the control of individuals and expectations, and norms associated with being a woman or a man. those that stem from differences in preferences and choices. A sub- Gender equality refers to how these aspects determine how women stantial body of research documents such male-female differences in and men relate to each other and to the resulting differences in risk aversion, social preferences, and attitudes about competition. It power between them. follows that if men and women differ, on average, in attitudes, prefer- This Report focuses on three key dimensions of gender equality ences, and choices, then not all observed differences in outcomes identified by men and women from Afghanistan to Poland to South can be attributed to differences in opportunities. Africa, as well as by researchers: the accumulation of endowments Those who argue for equality of outcomes argue that differ- (education, health, and physical assets); the use of those endow- ences in preferences and attitudes are largely “learned� and not ments to take up economic opportunities and generate incomes; inherent—that is, they are the result of culture and environment and the application of those endowments to take actions, or agency, that lead men and women to internalize social norms and expec- affecting individual and household well-being. These are aspects of tations. Persistent differences in power and status between men equality where shortfalls of choice are reflected in shortfalls of wel- and women can become internalized in aspirations, behaviors, and fare. They matter in and of themselves. But they are also closely preferences that perpetuate the inequalities. So, it is difficult to interlinked. define equality of opportunity without also considering how actual Gender inequality is both similar to and different from inequal- outcomes are distributed. Only by attempting to equalize out- ity based on other attributes such as race or ethnicity. Three differ- comes can one break the vicious circle of low aspirations and low ences are of particular relevance to the analysis of gender equality. opportunity. First, the welfare of women and men living in the same household Despite this debate, it is difficult in practice to measure oppor- is difficult to measure separately, a problem that is compounded by tunities separately from outcomes. Indeed, equality of opportuni- the paucity of data on outcomes in the household. Second, prefer- ties and equality of outcomes are tightly linked both in theory and ences, needs, and constraints can differ systematically between in measurement. For this reason, the Report takes a pragmatic men and women, reflecting both biological factors and “learned� approach, focusing on both outcomes and opportunities in relation social behaviors. Third, gender cuts across distinctions of income to endowments, agency, and access to economic activities. Follow- and class. These characteristics raise the question whether gender ing Sen, we also believe that while people may disagree in what is equality should be measured as equality of outcomes or equality of just or fair, they will agree on eliminating what are “outrageously opportunity. The economic and philosophical literature on this unjust arrangements.� In other words, while it may be difficult issue is divided. to define whether gender equality is about outcomes or opportuni- Those who defend framing gender equality as equality of oppor- ties, most will agree that gross manifestations of gender inequality tunity argue that it allows one to distinguish between inequalities should be eliminated. Sources: Booth and Nolen 2009; Croson and Gneezy 2009; Gneezy, Leonard, and List 2009; Kabeer 1996; Sen 1999; World Bank 2011. BOX 2 The Millennium Development Goals recognize the intrinsic and instrumental value of gender equality The 2010 Millennium Development Goal (MDG) Sum- equality and women’s empowerment are develop- mit concluded with the adoption of a global action ment objectives in their own right (MDG 3 and 5), as plan to achieve the eight goals by 2015. The summit well as serving as critical channels for achieving the also adopted a resolution calling for action to ensure other MDGs and reducing income and non-income gender parity in education and health, economic poverty. Gender equality and women’s empower- opportunities, and decision making through gender ment help to promote universal primary education mainstreaming in development policy making. The (MDG 2), reduce under-five mortality (MDG 4), resolution and the action plan reflect the belief of the improve maternal health (MDG 5), and reduce the international development community that gender likelihood of contracting HIV/AIDS (MDG 6). Source: WDR 2012 team. ment positions, managers are less skilled on function, is large: ensuring that women farmers average, reducing the pace of innovation and have the same access as men to fertilizer and other technology adoption.2 agricultural inputs would increase maize yields The direct payoff to correcting these failures, by 11 to 16 percent in Malawi and by 17 percent many rooted in how markets and institutions in Ghana.3 Improving women’s property rights Overview 5 in Burkina Faso would increase total household of countries (such as Bangladesh, Brazil, Côte agricultural production by about 6 percent, with d’Ivoire, Mexico, South Africa, and the United no additional resources—simply by reallocat- Kingdom) shows that increasing the share of ing resources (fertilizer and labor) from men to household income controlled by women, either women.4 The Food and Agriculture Organiza- through their own earnings or cash transfers, tion (FAO) estimates that equalizing access to changes spending in ways that bene�t chil- productive resources between female and male dren.10 In Ghana, the share of assets and the farmers could increase agricultural output in share of land owned by women are positively developing countries by as much as 2.5 to 4 per- associated with higher food expenditures.11 In cent.5 Eliminating barriers that prevent women Brazil, women’s own nonlabor income has a from working in certain occupations or sectors positive impact on the height of their daugh- would have similar positive effects, reducing ters.12 In China, increasing adult female income the productivity gap between male and female by 10 percent of the average household income workers by one-third to one-half (chapter 5) and increased the fraction of surviving girls by 1 per- increasing output per worker by 3 to 25 percent centage point and increased years of schooling across a range of countries.6 But achieving these for both boys and girls. In contrast, a compa- gains will not occur automatically as countries rable increase in male income reduced survival get richer: multiple and sometimes reinforcing rates and educational attainment for girls, with barriers to gender equality can get in the way. no impact on boys.13 In India, a woman’s higher These productivity gains are likely to be even earned income increases her children’s years of larger in a more integrated world where ef�ciency schooling.14 in the use of resources is essential to a country’s Improvements in women’s own education competitiveness and growth. Indeed, recent and health also have positive impacts on these work shows that gender inequality has become and other outcomes for their children. Better more costly for most countries in a world of open nutritional status of mothers has been associ- trade.7 Gender inequality diminishes a country’s ated with better child health and survival.15 ability to compete internationally—particularly And women’s education has been positively if the country specializes in exporting goods and linked to a range of health bene�ts for chil- services for which men and women workers are dren—from higher immunization rates to bet- equally well suited. Industries that rely more on ter nutrition to lower child mortality. Mothers’ female labor expand more in countries where (and fathers’) schooling has been positively women are more equal.8 The relationship also linked to children’s educational attainment goes the other way: countries with an advantage across a broad set of countries; in Pakistan, in making products that rely more on women’s children whose mothers have even a single year labor also have become more gender equal.9 of education spend one extra hour studying at And in countries and regions with rapidly aging home every day and report higher test scores.16 populations, like China and Europe and Central Women’s lack of agency—as seen in domes- Asia, encouraging women to enter and remain tic violence—has consequences for their chil- in the labor force can help dampen the adverse dren’s cognitive behaviors and health as adults. impact of shrinking working-age populations. Medical research from developed countries has So, in a globalized world, countries that reduce established a link between exposure to domes- gender-based inequalities, especially in second- tic violence as a child and health problems as ary and tertiary education and in economic par- an adult—men and women who experienced ticipation, will have a clear advantage over those violence in the home as children are two to that delay action (chapter 6). three times more likely to suffer from cancer, a stroke, or cardiovascular problems, and �ve Women’s endowments, agency, and to ten times more likely to use alcohol or ille- opportunities shape those of the next gal drugs than those who did not.17 Numerous generation studies also document how experiencing vio- Greater control over household resources by lence between parents as a child is a risk factor women leads to more investment in children’s for women experiencing violence from their human capital, with dynamic positive effects own partners as adults, and for men perpetrat- on economic growth. Evidence from a range ing violence against their partners.18 6 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Increasing women’s individual and collective to systematically favor the interests of those with agency leads to better outcomes, institutions, more influence. Institutional constraints and and policy choices market failures that feed gender inequalities are Agency is about one’s ability to make choices— less likely to be addressed and corrected, leading and to transform them into desired actions and to their persistence. As highlighted in the World outcomes. Across all countries and cultures, Development Report 2006: Equity and Develop- there are differences between men’s and women’s ment, an “inequality trap� may thus emerge, ability to make these choices, usually to women’s preventing generations of women from getting disadvantage. These gendered differences mat- educated and taking up economic opportuni- ter for women’s well-being but also for a whole ties on a par with men, reducing their ability to set of outcomes for their families and for soci- make informed choices and to realize their po- ety in general. Women’s agency influences their tential as individuals.23 ability to build their human capital and take up economic opportunities. In Bangladesh, women with greater control over health care and house- WHAT DOES THIS REPORT DO? hold purchases have higher nutritional status. Women’s agency also matters for the welfare of This Report focuses on the economics of gender their children. In Mexico, the daughters (but equality and development. It uses economic the- not the sons) of women with more control ory to understand what drives differences in key over household decisions work fewer hours on aspects of welfare between men and women— household tasks. education and health, access to economic op- Women’s collective agency can be transfor- portunities and productive resources, and the mative for society. It can shape the institutions, ability to make effective choices and take ac- markets, and social norms that limit their indi- tion. And it uses the same economic lens to vidual agency and opportunities. Empowering explore what policy interventions and broader women as political and social actors can change societal action can be taken to reduce these policy choices and make institutions more rep- gender differences and improve development resentative of a range of voices. Female suffrage outcomes generally. The Report does not limit in the United States led policy makers to turn itself to economic outcomes—indeed, it devotes their attention to child and maternal health and roughly equal attention to human endowments, helped lower infant mortality by 8 to 15 per- economic opportunities, and women’s agency, cent.19 In India, giving power to women at the signaling the importance of all three interrelated local level (through political quotas) led to in- aspects in human welfare. Nor does it ignore the creases in the provision of public goods (both central role of social and political institutions, female-preferred ones such as water and sani- whether formal or informal, in determining tation and male-preferred goods such as irri- gender outcomes. But in its framing of the issues gation and schools) and reduced corruption.20 and in the evidence it brings to the case for gen- Bribes paid by men and women in villages with der equality, it draws heavily on the economic a female leader were 2.7 to 3.2 percentage points literature on gender. less than in villages with a male leader.21 In India We adopt this approach for four reasons. First, and Nepal, giving women a bigger say in man- it provides valuable insights into how key gender aging forests signi�cantly improved conserva- outcomes emerge and evolve as the development tion outcomes.22 Women’s greater public voice process unfolds, as well as how the role and ef- not only bene�ts women and children but can fectiveness of policy influence these outcomes. also bene�t men. In many rich countries, greater Second, it builds on a tradition of World Bank female participation in economic activity has work on the economics of gender (most notably, combined with their increased representation the Engendering Development report24) and on in political leadership to reshape social views on the institution’s strongest areas of expertise and balancing work and family life in general and to specialization. Third, there are signi�cant data pass more family-friendly labor legislation. and knowledge gaps that we can help �ll in this Conversely, when women and men do not area. Fourth, while the Report often arrives at di- have equal chances to be socially and politically agnoses similar to those of other approaches, it active and to influence laws, politics, and policy provides different insights into the policy levers making, institutions and policies are more likely that can be used in support of gender equality. Overview 7 The Report focuses largely on inequalities affecting women, dwelling on ones likely to be BOX 3 How women and men define gender in the reproduced and passed on to the next genera- 21st century tion. But it also focuses on inequalities affecting men, while recognizing that most of these male To inform this Report the World Bank conducted new field research in 19 coun- inequalities affect fewer realms of welfare. tries in all regions to gain a first-hand look at how men and women experience We adopt an empirical approach, prefer- gender in their everyday lives. ring rigorous and evidence-based analysis and Women and men from all age groups, incomes, and locations see educa- highlighting causality where feasible. For this, tion, the ownership of assets, access to economic opportunities, and opportunities to earn an income as the keys to improving their well-being and we draw on a large and growing body of quan- that of their families. In 500 focus groups, researchers identified women’s and titative gender research, complemented by new men’s roles and responsibilities in private and public spheres—with women’s analysis, particularly on time use, domestic tasks being largely associated with family care and home production, and violence, mortality risks, and inputs into agri- men’s with income generation and decision making. But differences across culture and entrepreneurship. We also draw on generations clearly show that these roles are being redefined in a world that new qualitative �eld research with more than offers new opportunities and demands for both men and women. 4,000 men and women in 98 communities from The findings also show that old problems persist in new settings even as new challenges are emerging. Many groups face pervasive disadvantages—for 19 developing countries, exploring how gender them, change remains an aspiration for future generations but not a reality in affects their everyday lives and their aspirations, their everyday lives. education, job choices, decision making, and other aspects of well-being (box 3).25 Source: World Bank 2011. A global report like this one cannot provide Note: The exercise was conducted with men and women of different age groups in 98 communities (about in-depth analysis of speci�c country circum- 4,000 individuals) in Latin America (Dominican Republic and Peru), Europe and Central Asia (Moldova, Po- land, and Serbia), Africa (Burkina Faso, Liberia, Sudan, South Africa, and Tanzania), South Asia (Afghanistan, stances. Nor can it cover all relevant dimensions Bhutan, and India), the Middle East (West Bank and Gaza and the Republic of Yemen), East Asia (Indonesia of gender equality. Instead, it proposes a con- and Vietnam), and the Paci�c Islands (Fiji and Papua New Guinea). ceptual framework to explain gender inequal- ity and recommend public action, which can be adapted as necessary to speci�c countries, Drawing on past and recent work on gender issues, and sectors. It then illustrates the use of and development within the World Bank26 and this framework by focusing on aspects of gen- elsewhere, the Report posits that gender out- der equality where there has been most progress comes can be understood through the responses worldwide (education, fertility, life expectancy, of households to the functioning and structure labor force participation, and the extension of of markets and institutions, both formal and in- legal rights) and where there has been little or formal. Families decide how many children to very slow change (excess female mortality, seg- have and when, how much to spend on educa- regation in economic activity, gaps in earnings, tion and health for daughters and sons, how to responsibility for house and care work, asset allocate different tasks (inside and outside the ownership, and women’s agency in private and household), and other matters that influence public spheres). gender outcomes. “ I believe that a woman must be educated and must work in order to prove herself in society and to be a better mother. Young woman in Rafah city, West Bank and Gaza Women should work. Why should I stay at home if I can work outside? I should also earn income and my people and myself should enjoy the money I can make. Those days where our mothers were to ask for money from our fathers, even for simple things like underwear are gone: we need our own money and this means that we should work. � Young woman in Bukoba municipality, Tanzania 8 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 (box 4). Voice and bargaining power of house- BOX 4 What do we mean by markets, formal institutions, hold members are de�ned by a range of factors, and informal social institutions? including their ownership of and control over resources, their ability to leave the household (exit options), and social norms. In this way, Markets—a variety of arrangements that allow buyers and sellers to exchange household decision making, markets, formal (the rights over) any type of goods and services subject to a set of rules. Markets allow for any item that is exchanged to be evaluated and priced. Markets can be institutions, and informal institutions combine influenced and shaped by formal and informal institutions. and interact to determine gender-related out- comes (�gure 1). Formal institutions—all aspects that pertain to the functioning of the state, The bene�ts of economic development (the including laws, regulatory frameworks, and mechanisms for the delivery of ser- vices that the state provides (such as judicial services, police services, basic infra- combination of higher incomes and better ser- structure, health, and education). vice delivery institutions) on gender outcomes can be seen clearly through this framework as Informal social institutions—the mechanisms, rules, and procedures that shape emerging from the workings of households, social interactions but do not pertain to the functioning of the state. In this Report, the focus is on gender roles, beliefs, social norms, and social networks. markets, and institutions and their interactions. Gender roles provide guides to normative behaviors for each sex within certain These impacts are illustrated in �gure 1 by the social contexts. Roles gain power as they are learned through socialization, elab- “growth� arrow that turns the gears in the di- orated in cultural products, and enacted in daily life. The repeated experience of rection of greater gender equality. The impact of performing gender roles affects widely shared beliefs about men’s and women’s more gender equality on growth is in turn cap- attributes and one’s own sense of identity. Social norms refer to patterns of tured by the “gender equality� arrow that flows behavior that flow from socially shared beliefs and are enforced by informal social back into higher growth. sanctions. These can affect household bargaining in many ways: they set limits on what can be bargained about; they can be a determinant of or constraint to bargaining power; they can affect how bargaining is conducted; and they them- selves can be subject to bargaining and can change. Social networks refer to the WHERE HAS THERE BEEN THE MOST system of social relationships and bonds of cooperation for mutual benefit that PROGRESS IN GENDER EQUALITY? shape one’s opportunities, information, social norms, and perceptions. For women and girls in developing countries, Sources: Agarwal 1994, 1997; Fehr, Fischbacher, and Gätcher 2002; Kabeer 1999; Sen 1990. much has changed for the better in the past quarter century. Take female life expectancy at They make these choices on the basis of the birth. It increased dramatically in developing preferences, incentives, and constraints of dif- countries (by 20 to 25 years in most regions in ferent family members, and in relation to their the past 50 years) to reach 71 years globally in relative voice and bargaining power. Preferences 2007 (compared with 67 for men), and women are shaped by gender roles, social norms, and now outlive men in every region of the world. social networks (which we group under the label The changes were much faster than when to- informal institutions). Incentives are largely in- day’s rich countries were poorer. It took more fluenced by markets (including the markets for than 100 years for the number of children born labor, credit, land, and goods), which determine to a woman in the United States to decline from the returns to household decisions and invest- 6 to 3; the same decline took just over 35 years ments. Constraints arise from the interplay of in India and less than 20 in Iran (�gure 2). The formal institutions (comprising all that pertain same patterns can be seen in primary education. to the functioning of the state) and markets but It took the United States 40 years (from 1870 also reflect the influence of informal institutions until 1910) to increase enrollments among girls aged 6 to 12 years from 57 percent to 88 percent; Morocco achieved a similar increase for this age “ I think women should go out as well to look for a job because men are failing to get jobs; for women it is easier because they have more options. group in just over a decade (from 58 percent in 1997 to 88 percent in 2008). Girls’ education � Progress in closing gender gaps in education has Young man in rural Ngonyameni, been steady and sustained at all levels—primary, South Africa secondary, and tertiary. In many countries, and especially for higher education, these gaps are Overview 9 FIGURE 1 Gender outcomes result from interactions between households, markets, and institutions R EQUALIT NDE Y po GE lic ies S UTION MAL INSTIT INFOR ECONOMIC OPPORTUNITIES TS HOUSEHOLDS RKE MA AGENCY ENDOWMENTS FOR MAL INST ITUT ION S GR OW TH Source: WDR 2012 team. now reversing, with boys and young men at a FIGURE 2 Across the world, women relative disadvantage. Two-thirds of all coun- are having fewer children tries have reached gender parity in primary education enrollments, while in over one-third, How fast can fertility decline? girls signi�cantly outnumber boys in secondary Iran, education (�gure 3). Even in regions with the Islamic Rep. largest remaining gender gaps—South Asia and Bangladesh Sub-Saharan Africa (particularly West Africa)— there have been considerable gains. And in a Morocco striking reversal of historical patterns, more Zimbabwe women than men now attend universities, with women’s tertiary enrollment across the globe Colombia having risen more than sevenfold since 1970 India (fourfold for men). Yet while boy disadvantage United is slowly emerging in some places, girl disadvan- States tage where it exists tends to emerge earlier in life 0 20 40 60 80 100 and is deeper. number of years for the total fertility rate to fall from more than 6 children Women’s market work to less than 3 Women’s labor force participation has grown in the past 30 years as expanding economic op- Source: www.gapminder.org portunities have drawn many female workers 10 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 FIGURE 3 Gender parity in enrollments at lower levels has been achieved in much of the world, but tertiary enrollments are very low and favor women Primary education Secondary education Tertiary education 100 100 100 men Some African Enrollment gaps dwarf school enrollment, women, gross, % disadvantaged countries lag behind gender gaps school enrollment, girls, net, % school enrollment, girls, net, % 80 80 80 60 60 60 40 40 40 Women are more likely 20 20 20 to participate girls girls than men disadvantaged disadvantaged 0 0 0 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 school enrollment, boys, net % school enrollment, boys, net % school enrollment, men, gross % East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa South Asia Sub-Saharan Africa High-income countries Source: WDR 2012 team estimates based on World Development Indicators. Note: The 45° line in each �gure above shows gender parity in enrollments. Any point above the 45° line implies that more women are enrolled relative to men. into the market. Between 1980 and 2008, the tures change so that activities in which men gender gap in participation narrowed from 32 no longer have an advantage become more percentage points to 26 percentage points. By prominent. This shift opens new opportunities 2008, women represented more than 40 percent for women’s employment, and households re- of the global labor force. Large increases in par- spond to these signals by educating daughters. ticipation in countries that started with very low Richer countries can also invest in more ac- rates (mainly in Latin America and the Carib- cessible education systems by building schools bean and to a lesser extent in the Middle East and hiring teachers. When combined with bet- and North Africa) combined with small declines ter incentive and accountability systems, these in countries that started with very high rates inputs help deliver better and cheaper services, (mainly in Eastern Europe and Central Asia) lowering the costs of access to households and mean that rates have converged across regions, increasing their use. Where all these factors although signi�cant differences remain. Female have worked together, the gaps have closed rap- labor force participation is lowest in the Mid- idly, as in Morocco. dle East and Northern Africa (26 percent) and But even if bottlenecks appear in any one South Asia (35 percent) and highest in East Asia of these channels—pro-boy preferences within and the Paci�c (64 percent) and Sub-Saharan households or inadequacies in the provision of Africa (61 percent). education or slow growth or limits on women’s employment opportunities—the other channels What explains progress? still have allowed progress in educating girls. Pol- Where gaps have closed quickly, it has been a icies targeted to getting children to school, such result of how markets and institutions have as the conditional cash transfers used in more functioned and evolved, how growth has played than 30 countries worldwide (many explicitly out, and how all these factors have interacted targeting girls, as in Bangladesh and Cambodia), through household decisions. For education, have also helped. These forces are illustrated in consider each in turn. Higher incomes allow �gure 4 by the (green) gears representing house- families that had previously only sent their sons holds, formal institutions, and markets all mov- to school to now send their daughters as well. ing in ways that narrow educational gender gaps As countries get richer, their economic struc- (“oiled� by supportive policies). Overview 11 FIGURE 4 Using the framework to explain progress in education R EQUALIT ND E Y po GE lic ies NS ITUTIO MAL INST INFOR ECONOMIC OPPORTUNITIES TS ns RKE r HOUSEHOLDS MA g retu i n n More stable eas ucatio Incr ed incomes AGENCY ENDOWMENTS to FOR MA Low L INSTI er d TUT and irect, ION opp indi S ortu rect, cost nity s GR OW TH Source: WDR 2012 team. The interactions between households, mar- has increased sharply over time. So, at every level kets and institutions can also explain the pattern of per capita income, more women are now en- and pace of female labor force participation. A gaged in economic activity outside the home woman’s decision to work outside the home re- than ever before. sponds both to changes in her own wages and There are two main reasons why gains in to changes in her household income. As low- some domains of gender equality in many de- income countries grow richer, women partici- veloping countries came faster than they did for pate less in market work because their household today’s rich countries when they were at com- incomes also rise. Over time, their education lev- parable incomes. First, the incomes of many els also increase as formal institutions respond. developing countries have grown faster. Since Rising incomes also lead to later marriage and 1950, 13 developing countries have grown at childbearing and lower fertility. These factors an average of 7 percent a year for more than 25 all bring women back into the labor force. In 10 years or longer—a pace unprecedented before Latin American countries, almost two-thirds of the latter half of the 20th century.28 Second, the the increase in women’s labor force participa- various domains of gender outcomes are inter- tion in the past two decades can be attributed related. So, improvements in one have spurred to more education and to changes in family advances in others. The decline in fertility that formation (later marriage and lower fertility).27 has come with higher incomes has helped lower These different impacts of income growth and the number of deaths associated with maternal rising women’s wages lead to a U-shaped pat- mortality. And bearing fewer children has given tern of female labor force participation across women more time to invest in acquiring hu- countries (�gure 5). But notably, since 1980, the man capital and to participate in the economy. female participation rate at each level of income Forward-looking parents have responded to the 12 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 from educating girls and from women working FIGURE 5 Female labor force participation has increased over become evident. Consider the notable advances time at all income levels in gender equality in two very different coun- tries: Bangladesh and Colombia. • In the four decades since Bangladesh gained female labor force participation rate, % 80 independence, the average number of chil- 70 2008 dren a woman will have during her lifetime fell from almost 7 to just over 2. School en- rollment among girls rose from a third in 60 1991 to 56 percent in 2005. And just in the latter part of the 1990s, labor force participa- 50 tion for young women more than doubled. 1980 • In Colombia, the average number of children 40 a woman will bear dropped from 3.2 to 2.4 4 6 8 10 12 between the mid-1980s and 2005. Women log, GDP per capita (constant 2000 US$) also reversed the education gap and now have higher completion rates than men for primary, secondary, and even tertiary educa- Source: WDR 2012 team calculations based on International Labor Organization 2010 (130 countries). tion. And the country has the steepest in- crease in women’s labor force participation in the region, giving it one of the highest par- expanded employment opportunities by in- ticipation rates in Latin America. Women creasing their educational investments in their there are well represented in managerial po- daughters. These better educated girls are more sitions and in �nance—the glass ceilings no- likely to work when they become older, have toriously hard to break through even in many fewer children, and exercise more voice in their rich countries. households—feeding the cycle of change. So, the progress in fertility, the gains in education, The problem of severely disadvantaged the gains in women’s agency, and the shifts of populations women to market work are not only related but The combined forces of markets, service deliv- also mutually reinforcing. Public policies have ery institutions, and income growth that have themselves played a role, because the big push contributed to closing gender gaps in education, for universal education of the past decade has fertility, and labor force participation for many helped get all children to school. women have not worked for everyone. For poor The main lesson: when market signals, for- women and for women in very poor places, mal institutions, and income growth all come sizable gender gaps remain. And these gaps are together to support investments in women, gen- even worse where poverty combines with other der equality can and does improve very quickly. factors of exclusion—such as ethnicity, caste, And these improvements can occur even when remoteness, race, disability, or sexual orienta- informal institutions, such as social norms tion. Even in education, where gaps have nar- about what is “appropriate� for girls and boys rowed in most countries, girls’ enrollment in or women and men, may themselves take time primary and secondary school has improved to adapt. This is not to say that social norms little in many Sub-Saharan countries and have not been important in determining these some parts of South Asia. School enrollments outcomes. The differences across countries and for girls in Mali are comparable to those in among regions within countries both in clos- the United States in 1810, and the situation ing gender gaps in educational attainment and in Ethiopia and Pakistan is not much better levels of women’s labor force participation high- (�gure 6). And in many countries, gender dis- light their influence. But the fast pace of change parities remain large only for those who are in education and even in labor force participa- poor. In both India and Pakistan, while boys tion almost everywhere shows how these norms and girls from the top income quantile (�fth) adapt quite quickly as the economic returns participate in school at similar rates, there is a Overview 13 gender gap of almost �ve years in the bottom income quantile (�gure 7). FIGURE 6 Low-income countries lag behind in realizing Beyond the poor, gender gaps remain par- progress in female school enrollment ticularly large for groups for whom ethnicity, geographical distance, and other factors (such 100 Egypt, Arab Rep. as disability or sexual orientation) compound 90 gender inequality. Almost two-thirds of out-of- Bangladesh school girls globally belong to ethnic minority female school enrollment, ages 5–19, (%) 80 groups in their own countries.29 The illiteracy Mozambique Tajikistan rate among indigenous women in Guatemala 70 stands at 60 percent, 20 points above indige- Nigeria Niger Côte d’Ivoire nous men and twice the rate of nonindigenous 60 women.30 Ethiopia 50 For these severely disadvantaged groups— Burkina Faso Mali which can be pockets of disadvantage or en- 40 tire swaths of countries or regions—none of Pakistan the forces that favor educating girls and young 30 women are working. So, the growth in aggregate United States income may not be broad-based enough to ben- 20 in 1900 e�t poor households. Market signals are muted 10 because economic opportunities for women do not expand much or because other barriers— 0 such as exclusion caused by ethnicity, race, or 17 0 17 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 20 0 20 0 05 7 8 9 0 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 17 caste—get in the way of accessing those oppor- year tunities. And service delivery is often riddled United States with problems because poverty, distance, and trend line (United States, 1850–2000) discrimination mean that these groups do not Source: WDR 2012 team estimates based on U.S. Census and the International Income Distribution see an expansion of schools and teachers. This Database (I2D2). does not mean that the channels that have fa- Note: Values between 1760 and 1840 are based on female school enrollment trending between 1850 vored girls’ education elsewhere will not work and 2000. for these groups. It means that efforts need to be redoubled to ensure that the essential build- ity for house and care work, gaps in asset owner- ing blocks for progress (broad-based income ship, and constraints to women’s agency in both growth, expanding employment opportunities the private and public spheres. Progress in these for women, and effective service delivery) are domains is dif�cult to see, despite greater pros- in place. And these efforts may need to be com- perity in many parts of the world. Indeed, many bined with complementary interventions that of these gender disparities remain salient even address speci�c disadvantages that compound among the richest countries. gender inequality (chapter 7). Gender disparities persist in these “sticky� domains for three main reasons. First, there may only be a single institutional or policy WHERE HAVE GENDER INEQUALITIES “�x,� which can be dif�cult and easily blocked. PERSISTED AND WHY? We illustrate this problem with excess female mortality. Second, disparities persist when mul- By contrast to areas that have seen good prog- tiple reinforcing constraints combine to block ress, change has come slowly or not at all for progress. We use disparities in the economic many women and girls in many other dimen- sphere (the persistence of gender earnings gaps sions of gender equality. Health disadvantages and gender segregation in employment) and in that show up in the excess relative mortality of agency (differences in societal voice and house- girls and women fall into this category. So do hold decision making) to illustrate this prob- other persistent gender disparities, including lem. Third, gender differences are particularly segregation in economic activity, gender gaps in persistent when rooted in deeply entrenched earnings, male-female differences in responsibil- gender roles and social norms—such as those 14 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 FIGURE 7 Female disadvantage within countries is more marked at low incomes Benin Congo, Dem. Rep. The Gambia 10 10 10 median grade attained, median grade attained, median grade attained, ages 15–19 ages 15–19 ages 15–19 5 5 5 0 0 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 income quintile income quintile income quintile India Pakistan Togo 10 10 10 median grade attained, median grade attained, median grade attained, ages 15–19 ages 15–19 ages 15–19 5 5 5 0 0 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 income quintile income quintile income quintile girls boys Source: WDR 2012 team estimates based on EdAttain. about who is responsible for care and house- below the age of 60. About two-�fths of them work in the home, and what is “acceptable� for are never born, one-�fth goes missing in infancy women and men to study, do, and aspire to. and childhood, and the remaining two-�fths do And these gaps tend to be reproduced across so between the ages of 15 and 59 (table 1). generations. Consider each in turn. Growth does not make the problem disap- pear. Between 1990 and 2008, the number of Higher mortality of girls and women missing girls at birth and excess female mortal- The rate at which girls and women die relative to ity after birth did not change much; declines in men is higher in low- and middle-income coun- infancy and childhood were offset by dramatic tries than in high-income countries. To quantify increases in Sub-Saharan Africa in the repro- this excess female mortality (“missing� girls and ductive ages. Part of the increase is because women) and identify the ages at which it occurs, populations increased. But, unlike Asia, where this Report estimated the number of excess fe- the population-adjusted missing women fell male deaths at every age and for every country in every country (dramatically in Bangladesh, in 1990, 2000, and 2008.31 Excess female deaths Indonesia, and Vietnam), most Sub-Saharan in a given year represent women who would not countries saw little change in the new millen- have died in the previous year if they had lived nium. And in the countries hardest hit by the in a high-income country, after accounting for HIV/AIDS epidemic, things got much worse. the overall health environment of the country The Report’s analysis helps explain these they live in. Globally, excess female mortality patterns. Depending on the period in the life after birth and “missing� girls at birth account cycle, girls and women are missing for dif- every year for an estimated 3.9 million women ferent reasons. Missing girls at birth reflect Overview 15 TA B L E 1 Almost 4 million missing women each year Excess female deaths in the world, by age and region, 1990 and 2008 (thousands) Total women girls at birth girls under 5 girls 5–14 women 15–49 women 50–59 under 60 1990 2008 1990 2008 1990 2008 1990 2008 1990 2008 1990 2008 China 890 1,092 259 71 21 5 208 56 92 30 1,470 1,254 India 265 257 428 251 94 45 388 228 81 75 1,255 856 Sub-Saharan Africa 42 53 183 203 61 77 302 751 50 99 639 1,182 High HIV-prevalence countries 0 0 6 39 5 18 38 328 4 31 53 416 Low HIV-prevalence countries 42 53 177 163 57 59 264 423 46 68 586 766 South Asia (excluding India) 0 1 99 72 32 20 176 161 37 51 346 305 East Asia and Pacific (excluding China) 3 4 14 7 14 9 137 113 48 46 216 179 Middle East and North Africa 5 6 13 7 4 1 43 24 15 15 80 52 Europe and Central Asia 7 14 3 1 0 0 12 4 4 3 27 23 Latin America and the Caribbean 0 0 11 5 3 1 20 10 17 17 51 33 Total 1,212 1,427 1,010 617 230 158 1,286 1,347 343 334 4,082 3,882 Source: WDR 2012 team estimates based on data from the World Health Organization 2010 and United Nations Department of Economic and Social Affairs 2009. Note: Totals do not necessarily add up due to rounding. overt discrimination in the household, result- South Asia. High maternal mortality is the main ing from the combination of strong preferences contributor to excess female mortality in the re- for sons combined with declining fertility and productive years. In Afghanistan, Chad, Guinea- the spread of technologies that allow parents to Bissau, Liberia, Mali, Niger, Sierra Leone, and know the sex before birth.32 This is a particular Somalia, at least 1 of every 25 women will die issue in China and North India (although now from complications of childbirth or pregnancy. spreading to other parts of India), but it is also And a much larger fraction will suffer long-term visible in parts of the Caucasus and the Western health consequences from giving birth.33 Balkans. Progress in reducing maternal mortality has Missing girls during infancy and early child- not been commensurate with income growth. hood cannot be explained by a preference for In India, despite stellar economic growth in re- sons alone, although discrimination against girls cent years, maternal mortality is almost six times may contribute to it. It is a result not so much of the rate in Sri Lanka. In the past two decades, discrimination as of poor institutions that force only 90 countries experienced a decline of 40 households to choose among many bad options, percent or more in the maternal mortality ra- particularly regarding water and sanitation. tio, while 23 countries showed an increase. The Markets and households cannot compensate for main problem is, again, that households are be- these poor services. ing asked to make many decisions in the face of Missing women in the reproductive ages re- bad options—a result of multiple service deliv- flect two main factors. First, stubbornly high ery failures. In many parts of the world, this sit- rates of maternal mortality persist, especially in uation is reinforced by social norms that influ- much of Sub-Saharan Africa and some parts of ence household behavior and make it dif�cult 16 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 for women to get maternal health care quickly in the quality of institutions—in the provi- enough even where it is available. And high fer- sion of clean water, sanitation, and maternal tility, partly reflecting low incomes, compounds health care. Because there is only a single point the problem in parts of Sub-Saharan Africa. of entry—through better institutions—for ad- Second, the impacts of the HIV/AIDS pan- dressing female mortality, solving the problem is demic on the mortality of women in many East- hard—much harder than getting girls to school. ern and Southern African countries have been But for any basic notions of human justice, the dramatic. The reason for the greater prevalence global development community must make ad- of HIV/AIDS among women relative to men is dressing this problem a priority. their greater susceptibility and the greater like- lihood that their sexual partners are older and Gender segregation in economic activity thus more likely than younger men to have HIV. and earnings gaps In addition, countries that have had a low-lying Although women have entered the labor force civil conflict (such as Democratic Republic of in large numbers across much of the developing Congo) have also seen an increase in the num- world in the past quarter century, this increased ber of “missing� women. This is in contrast to participation has not translated into equal em- other countries that have had outright wars— ployment opportunities or equal earnings for like Eritrea, where men who went “missing� in men and women. Women and men tend to the years of war increased. work in very different parts of the “economic An examination of the historical experience space,� with little change over time, even in of northern and western European countries high-income countries. In almost all countries, and the United States reveals that similar pat- women are more likely than men to engage in terns of excess female mortality in infancy and low-productivity activities. They are also more the reproductive years existed there but disap- likely to be in wage or unpaid family employ- peared between 1900 and 1950. These reductions ment or work in the informal wage sector. In occurred primarily because of improvements agriculture, especially in Africa, women operate smaller plots of land and farm less remunerative crops. As entrepreneurs, they tend to manage smaller �rms and concentrate in less-pro�table FIGURE 8 Women and men work in different sectors sectors. And in formal employment, they con- centrate in “female� occupations and sectors distribution of female / male employment across sectors (�gure 8). These patterns of gender segregation 31% Communication Services 16% in economic activity change with economic de- velopment but do not disappear. 21% Retail, Hotels, and Restaurants 17% As a result of these differences in where 13% Manufacturing 12% women and men work, gender gaps in earn- ings and productivity persist across all forms of 4% Finance and Business 4% economic activity—in agriculture, in wage em- ployment, and in entrepreneurship (map 1). In 0.5% Electricity, Gas and Steam, and Water 1% almost all countries, women in manufacturing 0.5% Mining 2% earn less than men. In agriculture, farms oper- ated by women on average have lower yields 2% Transport and Telecommunications 7% than those operated by men, even for men and women in the same households and for men 27% Agriculture, Hunting, etc. 29% and women cultivating the same crops.34 Fe- 1% Construction 11% male entrepreneurs are also less productive than male entrepreneurs.35 In urban areas in Eastern 100% All Sectors / All Occupations 100% Europe and Central Asia, Latin America, and Sub-Saharan Africa, the value added per worker is lower in �rms managed by women than in Source: WDR 2012 team estimates based on International Labour Organization 2010 (77 countries). those managed by men.36 For �rms operating in Note: Totals do not necessarily add due to rounding. rural Bangladesh, Ethiopia, Indonesia, and Sri Overview 17 MAP 1 Earnings gaps between women and men (female earnings relative to $1 of male earnings) Germany 62¢ Iceland 69¢ Georgia 60¢ India 64¢ A.R. of Egypt 82¢ Bangladesh 12¢ Mexico 80¢ Benin 80¢ Nigeria 60¢ Ethiopia 34¢ Sri Lanka 50¢ Malawi 90¢ Salaried Workers Farmers Entrepreneurs Sources: Data for Benin come from Kinkingninhoun-Mêdagbé and others 2010; for Malawi from Gilbert, Sakala, and Benson 2002; for Nigeria from Oladeebo and Fajuyigbe 2007; for Bangladesh, Ethiopia, and Sri Lanka from Costa and Rijkers 2011; and for Egypt, Georgia, Germany, Iceland, India, and Mexico from LABORSTA, International Labour Organization. Lanka, the differences in pro�tability are signi�- disproportionate responsibility for housework cant between female-owned and male-owned and care, while men are responsible mostly for businesses.37 market work (�gure 10). When all activities are So, what explains this persistent gender seg- added up, women typically work more hours regation in economic activity and the resulting than men, with consequences for their leisure gaps in earnings? The Report argues that gender and well-being. And everywhere they devote differences in time use, in access to assets and more time each day to care and housework credit, and in treatment by markets and formal than their male partners: differences range institutions (including the legal and regula- from one to three hours more for housework, tory framework) all play a role in constraining two to ten times the time for care (of children, women’s opportunities. These constraints are elderly, and the sick), and one to four hours shown in �gure 9 as wedges blocking progress less for market activities. Even as women take toward greater gender equality. Income growth up a bigger share of market work, they remain has some influence in shifting these patterns but largely responsible for care and housework. does not eliminate them. The mutually reinforc- And these patterns are only accentuated after ing interactions between these different factors marriage and childbearing. make the problem particularly dif�cult to break. A second factor driving segregation in em- Consider each in turn. ployment and earnings gaps is differences in The differing amounts of time that men and human and physical endowments (including women allocate to care and related household access to assets and credit). Despite increases in work are one factor driving segregation and women’s education, there are still differences in the consequent earnings gaps. In most coun- human capital between women and men. These tries, irrespective of income, women bear a include a gap in years of schooling among older 18 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 FIGURE 9 Explaining persistent segregation and earnings gaps S UTION INSTIT INFO RMAL on norms Social rket work a care/m ECONOMIC TS OPPORTUNITIES RKE ess to MA l acc and HOUSEHOLDS erentia it/l Differential d Diff or/cre , and allocation of lab rkets s ma twork time/resources ne AGENCY ENDOWMENTS FOR MAL Bias INST ed l ITUT aws ION and /regula S GRO infr limi tion astr ted s, uctu re W H T Source: WDR 2012 team. cohorts as well as differences in what women large. Data for 16 countries in �ve developing and men choose to study in younger cohorts— regions indicate female-headed households are differences that affect employment segregation, less likely to own and less likely to farm land.38 especially in countries where most young people More generally, where evidence is available go to college. In agriculture and entrepreneur- for all farmers, women seldom own the land ship, large and signi�cant gender disparities in they farm. For example, in Brazil, women own access to inputs (including land and credit) and as little as 11 percent of land. And their land- in asset ownership are at the root of the gender holdings are systematically smaller than those productivity gap. Indeed, yield differences for owned by men. In Kenya, women account for 5 female and male farmers disappear altogether percent of registered landholders nationally.39 when access to productive inputs is taken into And in Ghana, the mean value of men’s land- account (�gure 11). Differences in access to in- holdings is three times that of women’s land- puts may be further compounded by differences holdings.40 Similarly large gaps are observed in in the availability of “market time,� as noted use of fertilizers and improved seed varieties in above, which can make the same investment agriculture, and in access to and use of credit less productive for women than for men. Jointly, among entrepreneurs. these constraints mean that women entrepre- Third, market failures and institutional con- neurs and farmers are often restricted to busi- straints also play a role. Labor markets often do nesses and activities that are less pro�table and not work well for women, especially if their pres- less likely to expand. ence is limited in some sectors or occupations. How big are gender differences in access to When few women are employed, employers may assets (especially land), credit, and other in- hold discriminatory beliefs about women’s pro- puts? A variety of data sources suggests they are ductivity or suitability as workers—these beliefs Overview 19 FIGURE 10 Across the world, women spend more hours per day on care and housework than men FOR SALE Market activities Housework Child care Pakistan 0.6 4.7 5.5 2.5 1.2 0.2 Cambodia 2.7 3.8 4.4 3.3 0.9 0.1 South Africa 2.1 3.8 4.2 1.8 0.5 0.0 Bulgaria 2.9 3.9 4.7 2.6 0.4 0.1 Sweden 3.2 4.6 3.2 2.3 0.6 0.3 Italy 2.1 4.8 4.9 1.4 0.6 0.2 women = 12 hours men Source: Berniell and Sanchez-Páramo 2011. can persist if there are no mechanisms in place measures, prevent women from entering some to correct them. Access to information about sectors or occupations. jobs, and support for promotions and advance- In sum, whether women are farmers, en- ment, often occur in gendered networks, hurt- trepreneurs, or workers, many are caught in a ing women trying to enter a male-dominated productivity trap: working hard on an uneven �eld (or equally hurting men trying to enter a playing �eld with unequal access to productive female-dominated one, such as nursing). And inputs. This trap imposes signi�cant costs on sometimes, legal barriers, framed as protective women’s welfare and economic opportunities 20 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 of married women in India are not involved in FIGURE 11 Gender differences in agricultural productivity spending decisions, even about their own in- disappear when access to and use of productive comes. Even in an upper-middle-income coun- inputs are taken into account try like Turkey, more than a quarter of married women in the lowest income quantile lack con- In Malawi, women’s agricultural trol over their earned income.41 Women’s ability productivity is 13.5% lower than men’s to own, control, and dispose of property still dif- this difference disappears when women and men have equal access to inputs –13.5% Malawi (National) 0.6% fers from that of men—sometimes legally, often in practice. And again, these patterns change –40% Nigeria (Osun State) 0% only slowly as countries grow richer. A clear manifestation of the lack of agency –21% Benin (Central) 0% is domestic violence. Violence is the opposite –17% of freedom—an extreme form of coercion that Ghana (National) 0% by de�nition negates agency. Women are at far –26% greater risk of violence by an intimate partner or Ethiopia (Central Highlands) –25% someone they know than from violence by other –7.7% people. And women are more likely than men to Kenya (Subnational) 12.5% be killed, seriously injured, or victims of sexual Kenya (Western), 2008 –19% violence by intimate partners.42 The prevalence 0% of domestic violence varies greatly across coun- Kenya (Western), 1976 –4% tries, with no clear relationship to incomes; 6.6% while incidence tends to rise with socioeconomic –60 –50 –40 –30 –20 –10 0 10 20 deprivation, violence knows no boundaries. In percent some middle-income nations, such as Brazil average gender gap (Sao Paolo and Pernambuco region) and Serbia gender gap with equal access to inputs (Belgrade), women report that the incidence of physical violence by intimate partners is as high as 25 percent.43 In Peru (Cusco), almost 50 Source: Alene and others 2008; Gilbert, Sakala, and Benson 2002; Kinkingninhoun-Mêdagbé and others 2010; Moock 1976; Oladeebo and Fajuyigbe 2007; Saito, Mekonnen, and Spurling 1994; Vargas Hill and percent of women are victims of severe physical Vigneri 2009. violence during their lifetime, and in Ethiopia (Butajira), 54 percent of women reported being today—and serious disincentives to invest in the subject to physical or sexual abuse by an inti- women of tomorrow. mate partner in the past 12 months. 44 Multiple factors are at work behind these Less voice in societal and household large gaps in women’s voice. In society, low decision making representation can be self-perpetuating, with In much of the world, women have less input women unable to convey their ability to lead. than men in decision making in their house- So, in politics, voters will not be able to judge holds, in their communities, and in their soci- accurately the capacity of a woman leader. And eties. Consider women’s underrepresentation in women’s entry may be limited by societal beliefs formal politics, especially in its upper reaches. that being engaged in politics is a masculine Fewer than one-�fth of all cabinet positions is activity or that women are less effective leaders held by women. And women’s lack of represen- than men—beliefs that are hard to break until tation extends to the judiciary and labor unions. a critical mass of women rises to political lead- These patterns do not change much as countries ership. Different responsibilities for care work get richer. The share of women parliamentari- also mean that women lack the flexibility or the ans increased only from 10 percent to 17 percent time to invest as heavily as men in participating between 1995 and 2009. in political institutions. The lack of networks for Whether and how much voice a woman has women also makes it more dif�cult for them to in household decision making over patterns of ascend to positions of authority in political par- spending, including spending on children, are ties or labor unions. important markers of her agency. As many as a In the household, two important determi- third of married women in Malawi and a �fth nants of a woman’s voice are her income and Overview 21 her control over household assets. Economic pro-male bias in India and Ethiopia but not in growth can improve the material conditions for Peru and Vietnam. exercising agency, with women generally having A growing body of research also suggests that more voice in wealthier households. But higher attitudes about women in the family and the household incomes alone are not enough to workplace are transmitted across generations. eliminate the lower capacity of women to ex- When women do not work outside the home, ercise agency. What matters are a woman’s own their daughters are also less likely to do so as income and assets as well as her ability to leave adults, and their sons are less likely to marry the household, all of which increase her bargain- women who work outside the home.48 Young ing power and ability to influence household men and young women also tend to study in choices. In India, owning property substantially very different �elds—with women favoring enhances women’s voice in the household on education and the humanities, while men fa- various matters and reduces her risk of domes- vor engineering, agriculture, and sciences—in tic violence.45 Similarly, as women’s shares in ways that are unrelated to abilities (chapter 3) household earnings increase in Colombia and yet repeat themselves over generations and do South Africa, so does their control over key not go away as incomes rise. Evidence also sug- household decisions. There is also evidence of a gests that domestic violence witnessed as a child relationship between women’s assets, earnings, is repeated in adulthood.49 Women in Haiti who and shares of household income, and the inci- had witnessed domestic violence were more dence of domestic violence.46 likely to report being the victims of physical or But if women’s earnings are limited by mal- sexual violence.50 The pattern is similar in Cam- functioning markets or other gender-differenti- bodia and Mexico. 51 And men in South Africa ated barriers to economic opportunities and as- who reported witnessing violence between their set ownership, women’s voice in the household parents were signi�cantly more likely to report will remain muted. Reinforcing these market perpetrating physical violence themselves.52 and legal influences are social norms dictating Norms may be learned in the household, but that men, not women, make the major decisions they are often reinforced by market signals and in households. institutions, which are gender biased in many aspects. For example, gender differences in the Reproducing gender inequality across responsibility for house and care work, as just generations discussed, are rooted in gender roles but strength- Perhaps the “stickiest� aspect of gender out- ened by discrimination in labor markets and by comes is the way patterns of gender inequal- a lack of child-care services. At the root of gen- ity are reproduced over time. Part of this dered patterns of what men and women study is persistence is rooted in slow-moving social a combination of factors that feed into household norms and how they affect what happens in decisions (norms about what is appropriate for the household. Women and men internalize girls and boys), institutions (gendered education social norms and expectations in ways that af- systems), and markets (gendered networks and fect not only their own aspirations, behaviors, occupational segregation). For domestic violence, and preferences but also those of their chil- empirical work �nds signi�cant explanatory dren. The Young Lives study looked at educa- power at the individual, household, and commu- tional aspirations and noncognitive skills of nity levels, reinforced by social perceptions and boys and girls at ages 8, 12, and 15 for 12,000 institutional failures (including a lack of protec- children in Ethiopia, Andhra Pradesh in India, tive laws and services or their poor enforcement Peru, and Vietnam.47 Parental aspirations for and delivery).53 the education of their children were biased to- ward boys in Ethiopia and India by the age of What can we learn from the persistence 12 and toward girls in Peru and Vietnam. By of all these gender gaps? the age of 15, these biases had been transmit- Markets and institutions (formal and informal) ted to children, with clearly higher educational can work against greater gender equality— aspirations among boys in Ethiopia and India, in ways that are often mutually reinforcing. and among girls in Vietnam. And by age 15, Sometimes service delivery institutions fail, as measures of agency or ef�cacy showed a strong for young girls and women during childbirth. 22 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Other times markets do not work well, with other domains of gender equality, such as occu- results that are worse for women, as illustrated pational segregation and many manifestations by evidence of discrimination in both labor and of women’s agency, income growth and better credit markets. Often reinforcing these mar- service delivery are far less effective in unlocking ket failures, however, are formal institutions the often multiple and reinforcing constraints that treat women and men differently. Laws that underlie persistent gender gaps. and regulations can constrain women’s agency The new forces of globalization can reduce and opportunities more than those of men, as many of these gaps. First, trade openness and when women and men have different owner- the diffusion of new information and com- ship rights, or when restrictions are placed on munication technologies have translated into hours and sectors of work for women but not more jobs and stronger connections to markets for men. Where credit and labor markets already for women, increasing their access to economic discriminate, such unequal laws and regulations opportunities and contributing to their eco- can accentuate the problem. Unequal treatment nomic empowerment. Second, urbanization may also manifest itself more indirectly through and greater access to information have allowed biased service delivery, as is the case for agricul- many in developing countries to learn about life ture extension services. Here, institutional bias and mores in other parts of the world, including and market structure (with women underrepre- the role of women, possibly affecting attitudes sented in nonfood crops that are often the target and behaviors. Third, the incentives for public of extension services) reinforce and even deepen action for gender equality are stronger than ever inequalities. because the rising global consensus on the in- All institutions (formal and informal) have trinsic importance of women’s economic, social, considerable inertia. They tend to reflect the in- and political empowerment means that gen- terests of those who wield more power and in- der inequality hurts a country’s international fluence, and they are dif�cult to change without standing. But this potential of globalization some form of collective agency or voice.54 Social will not be realized without effective domestic norms can be especially slow to change: norms public action to close remaining gender gaps in that may have served a purpose at one point in endowments, agency, and access to economic time, but are no longer useful, may endure sim- opportunities. ply because of custom or because a social pen- So, what should governments in developing countries do to foster greater gender equality? alty is associated with being the �rst to break the What areas of gender inequality should they norm, or because the norm bene�ts a dominant focus on? Should they start with interventions group in society (in this case, men). The norm’s in education and health, or should they focus persistence can perpetuate gender inequalities on access to economic opportunities or agency? long after its original rationale has disappeared. What combination of policies should they im- In sum, gender-differentiated market fail- plement, and in what sequence? At �rst blush, ures, institutional constraints, and persistent these questions can appear overwhelming be- social norms often combine to reinforce gender cause of the multiplicity of priority areas and inequalities and make improving gender equal- the number of available policy instruments. ity much more complex. When there are multi- This Report shows how better analysis can help ple constraints, they all need to be addressed. reduce the complexity of policy choice and de- sign in several ways. The starting point is to determine which as- WHAT IS TO BE DONE? pects of gender inequality should be of highest priority for policy going forward. Three criteria Nothing is automatic about the growth and matter in this regard: development process that delivers greater gen- der equality on all fronts. Part of the reason is • First, which gender gaps are most signi�cant that higher incomes and better delivery of ser- for enhancing welfare and sustaining devel- vices by the state help reduce gender gaps only opment? So, where are the likely payoffs for in some domains. And even in these domains, development from addressing gender dis- the improvements do not reach all women. In parities likely to be the largest? Overview 23 • Second, which of these gaps persist even as Policies to reduce gender gaps in countries get richer? So, where do higher human capital endowments incomes by themselves do little to reduce (health and education) disparities? Addressing gender gaps in human capital en- • Third, for which of these priority areas has dowments—excess female mortality at speci�c there been insuf�cient or misplaced atten- periods of the life cycle and pockets of gender tion? So, where would a reorientation of pol- disadvantage in education—requires �xing the icies yield the greatest bene�ts? institutions that deliver public services. Providing basic services in a timely manner to expectant Applying these criteria, we conclude that four mothers and improving the availability of clean areas should be of the highest priority for policy water and sanitation to households will go a long makers: way to closing the gender gaps in excess mortal- • Reducing gender gaps in human capital en- ity. Education services need to focus on improv- dowments (addressing excess female mortal- ing access for the signi�cant population groups ity and eliminating pockets of gender disad- that are currently disadvantaged by poverty, eth- vantage in education where they persist) nicity, caste, race, or geography. Such a focus will help address the “gender inequality traps� that • Closing earnings and productivity gaps be- affect the poor and excluded in society. tween women and men These solutions can come from either the • Shrinking gender differences in voice demand or the supply side, but they cannot be gender blind. On the contrary, they must factor • Limiting the reproduction of gender inequality in explicitly, both for design and implementa- over time, whether it is through endowments, tion, the drivers of gender inequality that cause economic opportunities, or agency the gender gaps in health and education out- Obviously, not all these priorities apply to all comes to persist. And they must bring into the countries. And speci�c country characteristics process of policy design and implementation will dictate how corrective policies will need to the voices of those that the policy is trying to be customized. reach—excluded women and girls, and the men Our analysis also emphasizes that, in choos- and boys who live with them. ing and designing policies, it is necessary to tar- get the determinants of the gender gaps of con- Reducing excess female mortality cern, not the outcomes. The framework in Part The main determinants of excess female mor- 2 of the Report helps highlight these underlying tality in different periods of the life cycle have causes, and shows how they emerge from the little to do with how quickly countries grow. workings of markets and institutions and their They result from household preferences and interactions with each other and with house- from failures in the ways that markets and in- holds. In other words, the framework identi�es stitutions function. The entry points for policy what problem needs to be solved and whether are dictated by which of these influences is most interventions should target markets, formal in- binding in each period. stitutions, informal institutions, or some com- Skewed sex ratios at birth is a problem in a bination of the three. few parts of the world, including China, parts of Having identi�ed the underlying causes of India, and parts of the Caucasus and the Western the gender gap of concern, the Report draws on Balkans. The underlying cause is son preference the experience with policy interventions across among households, which has been exacerbated a broad range of countries to provide guid- in some of these places by rapid income growth. ance on speci�c interventions that could work Higher incomes have increased access to ultra- in different settings. It also looks at the political sound technologies that assist in sex selection at economy of reforms and emphasizes that policy birth. So, policies need to work on two fronts. design and implementation must be attuned First, laws need to be enacted and enforced to to countries’ institutional, social, political, and deal with the abuse of sex selection technologies, as cultural environments and to the societal actors has been done in China and India. But experi- involved. ence shows that enforcement is dif�cult, if not 24 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 impossible, to achieve without imposing dra- ef�ciently while ensuring that services remain conian restrictions that are not feasible in most accessible and affordable for poor people. societies and that raise other ethical concerns. The solution will depend on the setting, but a And the dif�culty in enforcing such restrictions few elements are critical. is exacerbated because these problems are worse • Appropriate regulations that recognize the among the wealthy in these societies. rationale for government intervention. A second, and more promising, approach is to enhance household perceptions of the value of • An adequate structure of incentives for pro- daughters. Expanding economic opportunities viders to make them more accountable to for young women, including those in the labor policy makers. market, is one way of doing this, and it can work • Measures to strengthen accountability of in conjunction with the process of develop- both providers and policy makers to service ment to reverse son preference. Just look at the users. Republic of Korea, one of the few cases where excess female mortality at birth was reversed in In urban areas, providing clean water will a short period.55 And the process can be comple- require an emphasis on improving the struc- mented by providing �nancial incentives to par- ture of contracts and, in some circumstances, ents to have daughters (such as the “Apni Beti greater involvement of the private sector. In Apni Dhan� program in some Indian states) and Manila, such reforms yielded large impacts: wa- supporting media campaigns to change societal ter supply coverage expanded from 67 percent ideas about gender equality. in 1997 to 99 percent in 2009 and brought ef- In infancy and early childhood, excess �ciency gains through lower water losses and mortality of girls is not rooted in households operating costs. In low-income settings, where or markets—although both can worsen it. It �nancing options and the capacity of public in- is rooted in the failure of institutions to pro- stitutions may be more limited, charging small vide clean water, sanitation, waste disposal, and amounts for services, relying on independent drainage. Countries with high female mortality providers, and �nding ways of making provid- in infancy are those where the burden of infec- ers more accountable to users can help, even in tious diseases remains high. Today’s rich coun- smaller urban areas—this was the path followed tries eliminated their excess female mortality of by Cambodia. In rural areas, local governments young girls by improving access to clean water can improve community systems, as Uganda has and sanitation in the early part of the 20th cen- done by collecting an additional small tax and tury. And developing countries that have ex- placing it in a fund administered by the district perienced large drops in excess girl mortality council to pay for major water repairs. during the past two decades, like Bangladesh, For sanitation in urban settings, there usu- China, and Vietnam, have done the same. So, ally is enough demand for improvements so for the “missing� girls in Sub-Saharan Africa long as individuals and communities can cap- to “reappear,� countries must invest in similar ture the bene�ts of investing in the facilities. So, systems and provide adequate water, sanita- the solution is to strengthen property rights and tion, and waste disposal services to their whole recognize informal settlements, thus stimulating population and not just to the better-off. While demand while ensuring that communities have these services will bene�t all young children, access to independent providers. In rural areas young girls will bene�t more due to the reduc- and less dense urban settings, the priority for tion in infectious diseases. improving sanitation is to change behavior, raise How exactly should countries do this? If the awareness, and boost demand, through commu- experience of today’s rich countries is a guide, nity peer pressure and information campaigns part of the solution is to provide clean water at as in some communities in Cambodia, Indone- the point of use through piped delivery. Other so- sia, and Vietnam by appealing to people’s sense lutions, like water treatment at the source, are of community responsibility. less effective in reducing diarrhea morbidity Increasing the coverage of piped water and because of the potential for recontamination.56 sanitation is expensive, so signi�cant funding— The problem is then to design an institutional likely external—will be needed in poor coun- framework that expands access to clean water tries. A recent analysis of infrastructure funding Overview 25 needs for Sub-Saharan Africa concluded that vices need special attention. One way to help is additional spending on clean water and sani- to provide poor women with cash transfers con- tation would need to be about one and a half ditional on their seeking maternal care. An ex- times current levels—over $11 billion annu- ample is India’s Janani Suraksha Yojana, where ally—to improve access signi�cantly.57 How- such transfers increased the uptake of assisted ever, as documented in chapter 3, the return to deliveries in the presence of a skilled attendant these investments taking account of mortality by around 36 percent.60 declines is very large. Fourth, efforts to reduce maternal mortality In the reproductive years, maternal mortal- need to go beyond improving health systems ity remains especially high in Sub-Saharan Af- and services and work across sectors. The suc- rica and parts of Asia. The main reason is a fail- cesses of Malaysia and Sri Lanka in addressing ure of the institutions that deliver medical care maternal mortality early in their development and services to expectant mothers. While norms illustrate this point (box 5). Fairly small invest- that delay women from getting prompt medi- ments in infrastructure (rural roads) and in cal help during childbirth and high fertility may women’s education, combined with training be contributing factors in some places, solving maternal health providers and building hospi- the problem, as with providing clean water and tals dramatically reduced maternal mortality.61 sanitation, requires �xing the institutions that Fifth, it is essential that the political pro�le of deliver these services. the problem be raised. Turkey illustrates what is This �x will require providing more resources possible in this regard. Turkey’s maternal mor- to frontline service providers and ensuring that the tality rate in 2000 was 70 per 100,000 live births. entire system of maternal care works: A new government capitalized on the political First, the quality of the people in deliver- ing the chain of services needs to be upgraded. While additional health workers, especially BOX 5 Reducing maternal mortality—What works? skilled birth attendants, will be a continuing Look at Malaysia and Sri Lanka need, coverage can be increased in underserved areas by drawing in community-level providers Improving the delivery of maternal care is hard, but it can be done—even at and the private sector. relatively low incomes, as Sri Lanka and Malaysia show. From more than 2,000 Second, those providing maternal health ser- per 100,000 births in the 1930s, the maternal mortality ratio in Sri Lanka fell to vices have to be more responsive to expectant about 1,000 by 1947, and then halved to less than 500 in the next three years. By 1996, it had fallen to 24. In Malaysia, it halved from 534 over the seven years from mothers. One way is to make service providers 1950 to 1957. Then, with a halving every decade or so, it came down to 19 by more accountable to them. Getting information 1997. to users—for example, on service standards, To overcome the range of institutional obstacles that hampers the effective quality of services and policies to improve workings of health systems, Sri Lanka and Malaysia adopted integrated and them—can help but needs to be combined with phased approaches. And they did this with modest total public expenditures some way for users to act on that information. on health—1.8 percent of gross domestic product, on average, since the 1950s. In Uganda, community-based monitoring im- Health programs in both countries exploited synergistic interactions of health care with basic education, water and sanitation, malaria control, and integrated proved both the quality and quantity of primary rural development—including building rural roads, which helped deal with health care services.58 Another route to account- obstetric emergencies. Financial, geographic, and cultural barriers to maternal ability is to ensure that citizens are able to hold care were addressed by ensuring a front line of competent, professional mid- their political representatives responsible for the wives widely available in rural areas, providing them with a steady supply of failures; the politicians, in turn, need to exercise drugs and equipment, linking them to back-up services, and improving com- more effective control over the service provid- munication and transportation. Simultaneously, facilities were strengthened to ers. The power of this mechanism is evident in provide obstetric care and deal with complications. Better organizational man- agement improved the supervision and accountability of providers. Area- Peru, where improving maternal health required specific mortality data were provided through monitoring systems so that extending coverage, giving the right incentives empowered communities could hold political leaders accountable, and to service providers, and having citizens’ voices national and subnational actors were forced to recognize the unacceptability loud enough to be heard by policy makers.59 of every maternal death. Finally, both countries were strongly committed to Professional attention for deliveries rose from 58 improving the status of women: women gained voting rights before or soon percent of births in 2000 to 71 percent in 2004. after national independence, and female education received special attention. Third, the �nancial constraints that poor women face in accessing maternal health ser- Source: Pathmanathan and others 2003. 26 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 support that brought it to power and, in 2003, concentrated in lower productivity activities, undertook a Health Transformation Program, self-employment, and the informal sector. Even emphasizing institutional reform, client respon- in the formal wage sector, they cluster in certain siveness, and a focus on underserved areas. The occupations and industries, usually lower pay- budget allocated to primary health care and pre- ing. These differences remain even as countries vention in underserved areas rose by 58 percent, get richer. air ambulances were put in service for remote Three factors drive these patterns. First, populations, the health workforce was redis- women and men have very different responsi- tributed for better coverage in poor areas, and bilities for care and housework, and as a result conditional cash transfers encouraged pregnant very different patterns of time use, which im- mothers to use prebirth hostels and deliver in pinge directly on choices of employment and public hospitals. By 2009, the maternal mortal- economic activity. Second, women and men ity rate had fallen to 19.8.62 face differential access to productive inputs and often differential treatment by markets and in- Providing education to severely stitutions. Third, these mutually reinforcing disadvantaged populations constraints can generate a “female productivity Even as gender gaps in educational enroll- trap.� Policies thus need to target these underly- ment shrink nationally, they remain for poor ing factors. Because multiple factors may be at people and for those disadvantaged by other play more often than not, effective policy inter- circumstances—remoteness, ethnicity, caste, ventions may need to target several of them— race, or disability. To reach these groups, pol- either simultaneously or sequentially. icy makers can build on experience and evi- dence from Cambodia, Colombia, Honduras, Releasing women’s time Mexico, Nicaragua, Pakistan, and Turkey. A Gender differences in access to economic op- range of options includes remedies for both sup- portunities are driven in part by differences in ply (such as building more schools in remote ar- time use that result from deep-rooted norms for eas and recruiting local teachers) and demand the distribution of responsibility for care and (such as cash transfers conditioned on girls’ school housework. Addressing these binding norms attendance). A key to designing cost-effective and releasing women’s time means paying more interventions is the availability and the cost of attention to three types of policies: child care and collecting information about local character- parental leave policies; improvements in infra- istics and conditions. Where relatively little is structure services; and policies that reduce trans- known, less locally customized policies, such as action costs associated with accessing markets. cash transfers conditioned on sending daugh- Policies such as subsidies to or public provi- ters to school, may be more effective in reducing sion of child care can compensate women for gender disparities. The transfers have had posi- the costs they incur within the home from tive impacts on enrollments in both middle- engaging in market work. Child care can be income and lower-income settings, especially provided either directly by the state (including for increasing enrollments among groups with local governments) or through the private sec- low enrollments to start (such as pockets of tor, possibly with public subsidies and regula- severely disadvantaged populations).63 Having tion. Among developing countries, child-care gained wide political acceptance because of their policies have been used in some middle-income ef�cacy, such policies are being implemented in Latin American countries. Examples include more than 30 countries. publicly provided or subsidized day care such as Estancias Infantiles in Mexico, Hogares Co- Policies to improve women’s economic munitarios in Colombia, and similar programs opportunities in Argentina and Brazil. The evidence from Across the world, women and men access eco- these countries as well as from rich countries nomic opportunities—whether in wage em- (mainly in northern and western Europe) that ployment, agriculture, or in entrepreneurship— have similar schemes is that they increase the in fundamentally different ways. Women tend number of hours worked by women as well as to occupy very different parts of the economic lead them to work more in formal employment. space from men and are disproportionately In lower-income countries, child-care solutions Overview 27 are particularly needed for women employed working outside the home, making it easier to in the informal sector and rural women. In In- manage the multiple burdens of house, care, dia, the nongovernmental organization (NGO) and market work. And information and com- Mobile Creches is experimenting with differ- munication technologies can help reduce both ent models for providing child-care services for the time and mobility constraints that women women employed in the rural informal sector face in accessing markets and participating in and on public works programs. Similar efforts market work. Mobile banking programs, such have been undertaken in the Indian state of as M-PESA in Kenya, are allowing women to Gujarat by the Self Employed Women’s Asso- process small �nancial and banking transac- ciation, which has set up day-care centers for tions more effectively and promoting savings, the 0–6-year-old children of its members. Other which is especially bene�cial to small women options for publicly provided day care are ei- entrepreneurs. In India, a program run by an ther to lengthen the school day (particularly NGO, the Foundation of Occupational Devel- at grades where attendance is only for half of a opment, organized groups of women to focus day) or to lower the age at which children enter on marketing, provided them with access to the school system. cell phones and the Internet, thus helping them Parental leave policies have been tried mainly market their products directly and increase their in rich countries—and these typically take the pro�t margins.64 form of maternity leave. While these policies have increased women’s labor force participa- Closing gaps in access to assets and inputs tion in these countries, their applicability in de- Female farmers and entrepreneurs have less ac- veloping countries may be more limited. First, cess to land than their male counterparts. Simi- they can be used only in the formal sector, which larly, both the demand for and use of credit are typically represents only a fraction of employ- lower among female farmers and entrepreneurs ment in emerging and low-income economies. than among their male counterparts. These dif- Second, they can actually make it less attractive ferences are rooted in failures of markets and in- for employers to hire women of child-bearing stitutions and in their interactions with house- age unless the maternity leave is publicly �- hold responses. For example, accessing credit nanced. Policies that provide both paternity and often requires collateral, preferably land or im- maternity leave and make the former mandatory mobile assets. Women are thus at a disadvantage (as in Iceland, Norway, and Sweden) have the because they have lower or less secure access to advantage of not being biased against women land and are disproportionately employed in the while also helping to shift the underlying norms service sector where capitalization is lower and about care. But this approach may be �nancially output is often intangible. These forces may be beyond the �scal capacity of many developing further reinforced by gender-based preferences economies. in the households that can lead to unequal re- Improvements in infrastructure services— source allocations (of land, for example) to male especially water and electricity—can help free and female members. up women’s time spent on domestic and care Policies need to focus on these underlying work. Electri�cation in rural South Africa, for determinants of differential access—leveling instance, has increased women’s labor force par- the institutional playing �eld by strengthening ticipation by about 9 percent; in Bangladesh, it women’s ownership rights, correcting biases in has led to more leisure time for women. In Paki- service delivery institutions, and improving the stan, putting water sources closer to the home functioning of credit markets. was associated with increased time allocated to Strengthening women’s land and ownership market work. Other studies show no impact rights can help female farmers and entrepreneurs. on market work but noticeable impacts on lei- The main constraint that needs to be addressed sure time, which also increase women’s welfare is the restriction on women’s ability to own and (chapter 7). inherit assets and to control resources. Experi- Interventions can also focus on reducing the ence from India and Mexico shows that equal- (time) transaction costs associated with access to izing provisions of inheritance laws between markets. Better and more effective transport op- women and men increases asset ownership by tions can reduce the time costs associated with women. Discriminatory land laws, which lie 28 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 at the root of agricultural productivity gaps in such as Grameen Bank in Bangladesh and many countries, also need to be reformed to FINCA in Peru. Microcredit has now evolved provide, at least, for joint ownership in marriage, beyond group lending to such schemes as Banco increasing women’s ability to use land in access- Sol in Bolivia and Bank Rakyat Indonesia that ing economic opportunities. An even better way offer larger individual loans and rely on repay- to secure married women’s land rights (espe- ment incentives rather than peer monitoring. cially in the case of divorce or death of a spouse) Lack of access to formal credit can also be sur- is mandatory joint land titling. In two regions mounted through �nancial innovation and by in Ethiopia where land certi�cation involved the adapting a credit model that addresses the needs issuance of joint titles to both spouses, women’s of small businesses, as Access Bank in Nigeria, names appeared on more than 80 percent of all DFCU in Uganda, and Sero Lease and Finance titles, four times the 20 percent in the region in Tanzania have done. Recognizing that women where the certi�cate was issued only in the name are less likely to have established credit records of the household head.65 than men, and lower asset bases on which to Correcting biases in service delivery institu- draw for collateral, these large commercial tions such as the workings of government land banks partnered with the International Finance distribution and registration schemes and agri- Corporation to develop new instruments to cultural extension agencies can improve wom- support and extend credit services to female- en’s access to economic opportunity in many owned businesses and female entrepreneurs. In- countries. Redressing these biases requires ac- terventions included developing new products tions on several fronts. First, service providers such as loans that are collateralized with equip- need to target women explicitly and addition- ment or based on cash flow as well as training ally. For example, land redistribution programs for the staff of �nancial institutions and strate- that target the head of the household will not gic support to help banks increase their number serve women well. Instead, governments can put of female clients. Initial experience with these in place mandatory joint titling on redistributed interventions shows an increase in the shares of land that is coupled with gender sensitization female clients using �nancial services and taking policies and more female representation on local out larger loans with better-than-average repay- land boards. Second, women can be given some ment (chapter 7). power within the service delivery organization, including in setting priorities. For agricultural Addressing discrimination in labor markets extension, for example, women could be put In wage employment, the underrepresentation in decision-making positions at the Ministry of women in certain sectors or occupations can of Agriculture. Third, technology can be used feed discriminatory beliefs among employers to expand the reach of services, as was done in (or reinforce preexisting beliefs) that women are Kenya for agricultural extension through the not suitable workers or good candidates for em- use of call centers. Fourth, improved monitor- ployment. The importance of networks (often ing can make the problem visible. Finally, the gendered) in job search and professional pro- female users of the service should be provided motion can further reinforce women’s exclusion information on the level of service they are due. from certain jobs, positions, sectors, or occupa- This step can be aided by building the collective tions. Breaking this information problem and element of demand—for example, supporting expanding networks can be addressed through women’s farmer organizations. three main types of policies: active labor market Improving the functioning of credit markets by policies; af�rmative action programs; and group addressing the information problems caused by formation and mentoring interventions. lack of experience with women borrowers can Active labor market policies combine training, help address productivity gaps between women placement, and other support to enable women and men in agriculture and entrepreneurship. to enter or reenter the workforce. Although Microcredit schemes have been the most com- these policies are not typically motivated by the mon way of addressing these problems, by help- goal of narrowing gender wage gaps, evidence ing women access small-scale credit and build a from Argentina, Colombia, and Peru indicates track record of borrower performance. Typically, that they can increase women’s employment these take the form of group lending schemes and earnings in the formal sector by allowing Overview 29 BOX 6 Catalyzing female employment in Jordan Despite growing education levels, labor force par- service, resume writing, interviewing, and positive ticipation rates for women in the Middle East and thinking. North Africa remain very low. In Jordan, only 17 There appears to be strong demand for these percent of 20- to 45-year-old women work, com- policies. Despite low employment rates, the major- pared with 77 percent of men. This labor force ity of recent female graduates want to work: 93 per- participation gap also holds among the more cent say they plan to work after they graduate, and educated; among community college graduates 91 percent say they would like to work outside the it starts immediately on graduation. These low house after they are married. Of those invited to employment rates make it difficult for new gradu- attend the training courses, 62 percent completed ates seeking to enter the labor market. With few them, with unmarried women more likely to attend. women employed, young women lack role models Those who began the courses gave them positive to follow into employment as well as the network reviews, claiming the courses had given them much connections to help them find jobs. Employers, more confidence to begin searching for jobs. Four lacking experience with working women, may be months into the wage subsidy program, about a reluctant to hire women if they believe women are third of those using vouchers had found a job. less committed to staying employed. Early results from a midline evaluation suggest The Jordan New Opportunities for Women that job vouchers have significant employment (Jordan NOW) program is a pilot to rigorously eval- effects: employment rates among graduates who uate the effectiveness of two policies: short-term received vouchers alone or vouchers plus training wage subsidies, and employability skills training. are between 55 to 57 percent compared with 17 to Short-term wage subsidies give firms an incen- 19 percent among those who received training tive to take a chance on hiring young female grad- alone or received neither training nor vouchers. In uates and an opportunity to overcome stereotypes all groups, employment effects are higher for by directly observing young women working for unmarried women. Financial empowerment (mea- them. The subsidies can also give young women sured as the proportion of women who have their more confidence to search for work and to own money and can decide how to use it) also approach employers. In the pilot, each voucher increased significantly for all who received either has a value equal to the minimum wage for six vouchers or training or both. Follow-up surveys months. will determine whether these employment effects Employability skills training augments the of job vouchers are sustained in the longer term technical skills that graduates learn in community and will also focus on other measures of empower- college with the practical skills to find and suc- ment and changes in attitudes. The surveys will ceed in employment. Many employers say recent also allow further investigation of the link between graduates lack these interpersonal skills and other marriage and work, given the early findings that basic job skills. In the pilot, students received 45 married women are less likely to attend the train- hours of instruction in team building, communica- ing, less likely to use the vouchers, and less likely to tions, presentations, business writing, customer be employed. Source: WDR 2012 team. participants to better communicate their abili- rules, careful monitoring of impacts, and cred- ties to employers. A similar program currently ible sanctions for noncompliance are essential.66 under implementation and evaluation in Jordan Where such programs have been implemented, shows promising signs of success (box 6). the effect has clearly been to redistribute wage Af�rmative action is another way to over- employment from men to women. And, while come information failures. The goal is to push the economic ef�ciency of such policies is still de- women’s participation in wage employment to bated, the most comprehensive evidence (from a “critical threshold� (often argued to be about long experience in the United States) points to 30 percent), where information failures and little or no adverse ef�ciency effects.67 This ex- networks no longer bind. Experience (mainly perience and that in other countries also show from rich countries) shows that af�rmative ac- that any potential negative ef�ciency impacts tion works best if it is mandatory. Af�rmative can be addressed by ensuring that the af�rma- action also can be implemented through public tive action programs are temporary and are re- sector employment and contracting, but clear moved once women’s representation reaches the 30 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 needed critical mass. In the absence of explicit Because these constraints resemble the ones af�rmative action policies, female employment that limit women’s prospects in labor markets, in the public sector in fairly large numbers can the policy solutions are similar. Quotas and have a demonstration effect. In rich countries, other types of af�rmative action have promoted public sector growth has been important in in- women’s political representation at various lev- tegrating women into the labor markets.68 els of politics. Such measures range from vol- Supporting the creation of women’s networks untary commitments by political parties to in- can be effective where gendered networks hinder clude women candidates on their electoral lists women workers, farmers, or entrepreneurs. Such to specifying shares of legislative seats reserved interventions work best when they combine the for women. Which option is best for a country building of social capital and networks with the depends on its political system. For example, provision of training, information, and mentor- reserving individual seats for women will not ing. One example is the Jordan NOW program work in proportional representation systems, described in box 6. Another, more established whereas voluntary party quotas may work when example, is the Self Employed Women’s Asso- parties have strong leadership and internal dis- cipline. Whatever the system, its design and en- ciation in India, which has evolved into an ef- forcement are critical. In Spain, where positions fective organization representing the interests of on the ballot for the Senate were in alphabetical a large number of informal sector workers and order, parties tended to choose women with last entrepreneurs, providing extensive information, names that landed lower in the ballot and who support, and training services to its members. thus were less likely to win a seat. 70 Removing discriminatory treatment in la- Broader tensions also need to be acknowl- bor laws and regulations can promote women’s edged and taken into account if quotas are used economic opportunities. Among these laws and to increase women’s political representation. regulations, the priority should be to revisit Mandatory quotas involve the state circum- the limits (including outright bans) on part- scribing part of the democratic process, so this time work in many countries. Because women distortion has to be balanced against the need to provide a disproportionate share of household redress persistent inequalities. One option, taken and care work, such restrictions end up limit- by local governments in India, is to implement ing work options for women much more than quotas on a rolling basis—with a different set of for men. Relaxing these prohibitions would seats chosen for reservation in different elections give women more opportunities for paid em- over time. And as with all af�rmative action, it ployment. In Argentina, removing the ban on helps to specify a clear goal or time period up part-time contracts in the formal sector led to front. The structure of the reservation also mat- a signi�cant shift of women with children from ters. Designating particular seats for women part-time work in the informal sector to part- runs the risk of creating “token� women’s seats. time contracts in the formal sector.69 Quotas have increased female representa- tion. In Mexico, candidate quotas increased the Policies to shrink differences in voice share of women in parliament from 16 percent to more than 22 percent. Reserved seats in Mo- Increasing women’s societal voice rocco increased the proportion of women in Women generally have less voice than men parliament from less than 1 percent to almost 11 both in society and in households. At the soci- percent. Quotas in local governments in India etal level, income growth does little to reduce also showed that such measures can change un- these gaps. Norms that dictate that politics is for derlying beliefs among voters about the ef�cacy men; beliefs that women make worse leaders, of women politicians, even in a short period, which are fed in part by the low participation and increase the proportion of women elected of women in politics; norms around care and to these positions even after the quotas are no housework, which limit the time available to longer in place.71 women to participate in formal political institu- Af�rmative action in the political realm tions; and the gendered networks within politics needs to be complemented by measures that that work against women all matter more than increase women’s voice in other societal institu- income growth. tions, such as trade unions, corporations, the ju- Overview 31 diciary, and professional associations. This can work, and jobs requiring higher skills.72 Mo- be done through quotas as well as by mentoring rocco also reformed its family laws in the 1990s, schemes, women’s networks, and skills develop- and in 2004 the new Family Code completely ment in these realms targeted to women. Collec- eliminated references to the husband as the head tive action by women’s groups can be particu- of the household. larly effective in this, as with the Self Employed Greater effort is also needed to make these Women’s Association in India. More generally, rights more effective and justice systems more because women tend to be better represented responsive to women’s needs. Interventions are in less formal organizations, laws and regula- needed on both the supply side and the de- tions should ensure a level playing �eld for such mand side. Greater capacity of the institutions organizations. that apply the laws, more accountability in the justice system to promote predictable outcomes Increasing women’s voice within households in line with the law, and procedures to promote The muted voice of women within their house- women’s access to justice and women’s repre- holds reflects the combined influence of their sentation in judicial institutions are critical on more limited access to economic opportunities, the supply side. Also important are mechanisms the nature of social norms, the legal framework, for the implementation of laws. Evidence from and the enforcement of laws. Key determinants Ethiopia illustrates how the procedures around of control over household resources are ac- mandatory joint land titling helped promote cess to economic opportunities and the legal women’s rights over land. Women’s demand for framework—particularly rights over property enforcement of their rights can be promoted by and those that determine access to assets. For broadening literacy, increasing the accessibility domestic violence, social norms and the content of legal aid services, and reducing costs of legal and enforcement of laws are important. And for procedures. And data have to be collected and fertility, norms, bargaining power, and service made public so that the problems of women’s delivery are critical. access to justice are made more visible. Increasing control over household resources Reducing domestic violence Thus, the most promising policies to increase Reducing domestic violence requires action women’s voice in households center on reform- on multiple fronts. The goal is to prevent vio- ing the legal framework so that women are not lence before it happens. The �rst step is to en- disadvantaged in controlling household assets act laws that de�ne different types of violence and expanding their economic opportunities. against women, prescribe mandates and duties For the legal framework, land laws and aspects for enforcement and investigation, raise societal of family law that govern marriage, divorce, awareness, and signal a government’s commit- and disposal of property are particularly im- ment. These laws must be put in place in coun- portant. A cross-cutting issue applies to the tries that lack them, especially in the Middle many countries where multiple legal systems East and North Africa, South Asia, and Sub- exist. Reconciling these systems, which may in- Saharan Africa. And countries with such laws on clude customary and religious law, is a priority, the books need to make them more speci�c and especially to ensure that all laws are consistent actionable. with a country’s constitution. Kenya made such A second step is to shift norms and behaviors changes in its recent constitutional reforms. around domestic violence to emphasize preven- Although reforms in these areas are po- tion. Education and awareness programs, such litically and socially complex and depend very as Soul City in South Africa, can change norms much on country context, experience shows that about domestic violence among both men and change is possible. Ethiopia reformed its family women. Increasing women’s bargaining power law in 2000 by eliminating the husband’s ability in their households—by improving women’s to deny permission for his wife to work outside economic opportunities and enhancing their the home and requiring both spouses to agree in control over resources and their ability to leave administering family property. The �rst phase marriages—also can change behavior. But in- of these changes shifted women’s economic ac- creasing women’s bargaining power can also tivities toward work outside the home, full-time risk increasing the likelihood of violence in the 32 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 short term. So, speci�c mitigation measures may contraception. Contraception uptake is higher be needed. when husbands are included in family planning Third, when violence does occur, victims education, as was shown in Bangladesh74 and in need timely and effective assistance ranging Ethiopia. 75 from the police and judiciary to health and The second is increasing the quality of fam- social services, as with support integrated in ily planning services. Improvements in this re- Malaysia’s one-stop crisis centers at govern- gard need to focus on three areas. First, a suf- ment hospitals. Service providers—police and �cient range of contraceptive options needs to judiciary, health, and social services—need to be available. Second, adequate information on target women explicitly and additionally. Tar- the available options, their side effects, and the geting women also requires bringing services advantages and limitations of different methods closer to women to deal with time and mobility needs to be given to women so they can make constraints—for example, by providing com- an informed decision. Third, services need to munity paralegals and mobile legal aid clinics be provided in a manner that protects the in- that enable women to use the justice system. dividual’s or the couple’s privacy. This will re- In many contexts, bringing services closer to quire training health care providers in protocols individuals (demand) can be combined with designed speci�cally for family planning. Recent increasing the awareness of service delivery or- experience in Zambia shows that very different ganizations, particularly management, about outcomes regarding fertility and contraception gender issues (supply). PEKKA Women’s Le- can result depending on whether women are ap- gal Empowerment in Indonesia trains village proached individually and in private or together paralegals, with a focus on domestic violence with their partners. and family law. Another way of improving women’s access to justice is to increase the share Policies to prevent the reproduction of of women in the judicial and police forces re- gender inequality across generations sponsible for addressing domestic violence. The reproduction of speci�c gender inequali- The Indian state of Tamil Nadu introduced ties across generations gives rise to “gender in- 188 all-women police units to cover both rural equality traps,� which are likely to most affect and urban areas and to focus on crimes against the poor and excluded in society. Women’s lack women. These units increased women’s com- of political voice means that the market and in- fort in approaching the police, including mak- stitutional failures feeding gender inequality are ing reports of domestic abuse.73 unlikely to be corrected. Income growth alone does little to address the processes that underlie Increasing control over fertility these persistent gaps. The previous sections dealt Increasing women’s control over their fertility with policies to address three of these gaps that requires actions in several areas. Availability of reproduce over generations—reaching pockets family planning services is still limited in parts of remaining disadvantage in education, in- of the world. In some cases the underserved creasing women’s voice and participation in so- population covers entire countries, but more cietal institutions, and increasing women’s voice often these women live in speci�c geographic within households. Here, we address measures areas within countries —usually rural areas—or to address the gender inequalities in human are poor. For these groups, improved delivery of capital, opportunities, and aspirations that are family planning services is a priority. set early in life. Control over fertility decisions—the num- Decisions in adolescence can shape skill ac- ber and spacing of children—goes beyond is- quisition, health outcomes, and economic op- sues related to provision of reproductive health portunities. Adolescence is also a period when services, so two other policy areas need to be one’s lifelong aspirations are molded, and when addressed. The �rst is to boost women’s ability social norms and perceptions start to bind for within the household to voice their preferences re- boys and girls. Horizons for girls often shrink, garding number and spacing of children. As dis- especially for poor girls or girls in rural areas cussed earlier, access to economic opportunities, where distance and norms for mobility can control over assets, and appropriate laws help. So be a signi�cant constraint. Empowering ado- does educating men on the bene�ts and use of lescents to make better choices for themselves Overview 33 can make a big difference to their lives, to their and reported sexual behavior in boys.83 For families, to their communities, and, as future adolescents, the promotion of contraception, workers and citizens, to society more broadly. when combined with education interventions Interventions need to build human and social and skill building, and appropriately targeted capital; facilitate the transition from school to to cultural and social settings, can be effective work; and increase their aspirations and agency. in reducing unintended pregnancies.84 Such a Efforts to influence and reduce risky behavior program for adolescent girls in Uganda resulted are also important. in a signi�cant increase in condom use and a Scholarships and conditional cash trans- lower number of children among participants.85 fers can increase school attendance and reduce Sometimes, economic empowerment alone can dropout rates for adolescents, especially girls. have a marked impact. A recent evaluation of a These positive impacts are well documented youth job training program in the Dominican in Latin America in countries such as Colom- Republic that included life skills training plus an bia, Ecuador, Mexico, and Nicaragua.76 More apprenticeship showed a signi�cant reduction recently, evidence from Africa is beginning to in pregnancies among participants.86 show similar results. In Malawi, fairly small Exposure to female role models whose posi- cash transfers to girls increased enrollment and tions of leadership or power contradict stereo- reduced dropout rates.77 Moreover, while these types of women’s role can reduce the intergener- transfers were aimed at education, they had ational transmission of gender norms. A study of bene�ts in other realms, such as reducing HIV political reservation for women in India showed infections. Other tools can also be brought to that teenage girls who have repeated exposure to bear to help girls stay in school. Providing them women leaders are more likely to express aspira- with information on the returns to schooling tions that challenge traditional norms, such as is one such tool: for example, in Madagascar, a desire to marry later, have fewer children, and information on earnings for primary school obtain jobs requiring higher education.87 In- completers provided to boys and girls as well as creased economic opportunities for young girls their parents increased attendance rates by 3.5 can also change their own and their communi- percentage points.78 In the Dominican Republic, ties’ perceptions of gender roles for adolescent a similar effort to provide accurate information girls. A study of a program in Delhi that linked on actual returns to education to boys also had communities to recruiters for high-paying tele- a positive impact.79 Other evidence suggests that phone work found that these communities were incentives (prospects for winning a scholarship more likely to have lower expectations of dowry or direct payments for performance) can affect and to �nd it acceptable for women to live alone children’s own perception of their abilities and before marriage and to work before and after can improve test scores.80 marriage or childbirth.88 Vocational training targeted speci�cally to youth increased both the likelihood of employ- Making other policies “gender-smart� ment and the earnings for young women in Understanding how gender factors into the Colombia and Peru.81 In Kenya, providing in- workings of households, markets, and institu- formation to young girls about the relative re- tions can matter for policies even when improv- turns to vocational training in male or female- ing gender equality is not the main objective. dominated industries increased girls’ enrollment Why? Because gender-differentiated failures in in trade school courses that prepared them for markets, gender biases in institutions, and the typically male-dominated trades that yielded way gender relations play out in the household higher returns.82 The Adolescent Girls Initiative all affect (and sometimes constrain) the behav- aims to evaluate a range of these interventions, ior of both men and women. These changes in including skills training and mentoring, in a behavior can affect how men and women re- number of low- and middle-income countries spond to policies. Failing to take them into ac- (box 7). count can thus mean the policy will have unin- Health education programs have proved to be tended consequences or simply not work. effective in reducing risky behaviors. A program Take relations within the household. They in rural Tanzania substantially improved sexual clearly affect how the household responds to knowledge, attitudes, condom use in both sexes, policy—sometimes with unintended conse- 34 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 BOX 7 Intervening early to overcome future labor market failures—The Adolescent Girls Initiative The Adolescent Girls Initiative, a public-private trades attractive to women, with some trained as partnership, promotes the transition of adoles- electricians, masons, and mobile phone repair cent girls from school to productive employment technicians. through innovative interventions that are tested Lessons from implementation highlight girls’ and then scaled up or replicated if successful.a distinctive social capital needs, which must be Under way in Afghanistan, Jordan, Lao People’s addressed to facilitate the uptake of economic Democratic Republic, Liberia, Nepal, and South opportunities among often vulnerable and iso- Sudan (and soon in Rwanda), the initiative targets lated young females. Trainers in the Liberia pilot about 20,000 adolescent girls and young women formed girls into teams of three or four, who made ages 16–24. public commitments to support one another, both Interventions range from business develop- inside and outside the classroom, throughout the ment skills training and services to technical and training. The positive peer pressure helped keep vocational training, targeting skills in high demand. attendance rates high, with almost 95 percent In all projects, girls receive life-skills training to completing the training, and addressed the vari- address the most important barriers to their eco- ety of educational levels among participants. nomic independence. Each country intervention is Another promising innovation from the Libe- tailored to the local context and specific needs of ria pilot was a formal savings account at a local adolescent girls. Because the evidence on what bank for all participating girls, with an initial works is thin, rigorous impact evaluation is a part deposit of $5. The savings accounts not only of the initiative. enabled the girls to practice their financial literacy The skills training aims to equip girls with tech- skills beyond the classroom but built trust with nical skills with proven demand in the local labor formal financial institutions, and girls expressed market. In all pilots, training providers are asked to satisfaction with being connected to the modern conduct market assessments before selecting economy for the first time. trades for which training is developed and offered. Job fairs were organized to market the pro- While the focus is on matching skills to the mar- gram to potential employers interested in placing ket, the results in many cases challenge norms for girls in internships or jobs. Private sector human gender-appropriate occupations. resource and career development specialists met In Liberia, participating young women are with trainees individually to impart their knowl- offered six months of training for jobs in house edge about the industry, coach them on profes- painting, professional driving, and security guard sionalism in the workplace, and give constructive services. In Nepal, they are offered three months feedback on the skills demonstrated. These one- of occupational skills training followed by a man- on-one meetings offered girls the opportunity to datory skills test and three-month job placement. build networks and to tap into industry-specific The focus is on identifying nonstereotypical information crucial for new job entrants. a. Current donors to the Adolescent Girls Initiative include the Nike Foundation, a main partner of the initiative, as well as the governments of Australia, Denmark, Norway, Sweden, and the United Kingdom. The Bank’s Gender Action Plan also supports the initiative’s country projects. Pledges to the initiative stand at $22 million. quences. For example, many conditional cash of these issues for mothers and families (Co- transfer programs initially targeted women be- lombia, Peru), or even dedicated social workers cause—cognizant of how women were likely to (Chile). spend money differently from men—it seemed So, how can considerations of gender in- a way to get more of the transfer spent on chil- equalities and their underlying determinants dren’s endowments. But the transfers changed be integrated into broader policy and program bargaining power within the household and, in design? The analytical framework in this Report some cases such as Mexico, resulted in short- provides a guide. First, what happens inside the term increases in domestic violence. While this household shapes the impact of policies. An ex- effect can disappear or change in nature in the ample comes from Papua New Guinea. Gender longer term, a number of later transfer pro- roles in harvesting oil palm have men climbing grams included conditions to discourage do- the trees and harvesting the fruits, while women mestic violence (Brazil), training and awareness collect the fruit that has fallen on the ground. Overview 35 The oil palm industry came to the realization THE POLITICAL ECONOMY OF that 60–70 percent of the fruit on the ground REFORMS FOR GENDER EQUALITY was not being collected. They tried multiple initiatives designed to deal with constraints Well-designed public policies to address speci�c women faced, including giving the women spe- market failures and institutional or normative cial nets to use, and timing the collection to constraints can support signi�cant advances in deal with women’s care duties. None of these gender equality. But the choice of policies and worked. Finally, the Mama Lus Frut scheme was their implementation does not occur in a vac- introduced whereby women received their own uum. Policies must be attuned to countries’ in- harvest record cards and were paid directly into stitutional, social, and political environment and their personal bank accounts. Yields increased to the societal actors involved. It is important to signi�cantly, as did female participation in oil understand how reforms actually take place and palm harvesting. what factors allow them to be sustained so that Second, many non-gender-focused policies they produce change. and programs can bene�t from taking into ac- Two characteristics of gender reform pro- count women’s underrepresentation in markets, cesses are worth noting. First, as with all reforms, sectors, or occupations, a situation that can cause they redistribute resources and power between information problems not only for women but groups in society, including between men and also for those who seek to employ them, lend to women. Even when policies to advance gender them, or provide them with services. One exam- equality are well chosen and enhance economic ple of how to take women’s underrepresentation ef�ciency, some groups may lose as a result. into account is Ecuador’s program to expand Second, such reforms often confront power- credit bureau databases to include micro�nance. ful societal norms and beliefs regarding gender This intervention will help micro�nance insti- roles. Each of these features means there will tutions make better lending decisions, indepen- likely be opposition, and managing this push- dent of to whom they are lending. And because back is the key to successful reforms. micro�nance clients are predominantly women, A range of countries—rich and develop- it will also help them access a broader range of ing—show that several aspects of the political �nancial services. economy of reforms are especially relevant to Third, policy design should seek to level the gender equality. First, reforms are most likely to playing �eld for women and men, especially succeed when support for them is broad-based. where laws and regulations treat them dif- It is essential, then, to build coalitions that mo- ferently and where systems enforce laws and bilize around the reforms. These coalitions can regulations differentially, even when nomi- include nonstate actors such as political parties, nally equal. Looking for and �xing this type of trade unions, civic organizations and associa- discrimination when revising laws and regula- tions, and the private sector. Women’s groups in tions or enforcement mechanisms can provide particular have been a driving force for greater an opportunity to improve gender equality as a gender equality in labor legislation and family secondary bene�t. Take the case of taxes, which law. For instance, women workers in the infor- can explicitly discriminate against women mal sector have challenged their employers and when women face different rates for the same sometimes the state through such organiza- income as men. For example, in Morocco, tions as the Self Employed Women’s Association the tax allowance for children is allocated to in India and Nijera Kori in Bangladesh. These men, so men face a lower tax burden. Women groups have provided voice for women and cre- receive this allowance only if they prove that ated space for public action to counter the resis- their husband and children depend �nancially tance to reform. on them. This design is neither ef�cient not Women and men are partners in improving gender equitable. gender equality. While most initiatives that call When these considerations are factored into on men to support gender equality are still small, general policy design, the policies are more signs point to broader engagement in many areas likely to reach their intended objectives, and it and growing male support for women’s rights becomes easier for policy makers to tweak them in many developing countries. For instance, the to improve gender equality in the process. Rwanda Men’s Resource Center engages men 36 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 and boys in combating gender-based violence. such “incremental� reforms may not be enough In surveys of male attitudes toward gender to overcome the path dependence and institu- equality in Brazil, Chile, Croatia, and Mexico, tional rigidities that result in persistent gender adult men overwhelmingly express the view that inequality. Bold government action with “trans- “men do not lose out when women’s rights are formative� reforms may be necessary to alter promoted.�89 Even in India, where men in this social dynamics and move to a more equitable survey were less supportive of gender equality equilibrium. In choosing between incremen- overall, their support for some policies, such as tal and transformative policies as part of gen- quotas in universities and places in government, der reforms, the challenge for policy makers is was strong. to balance the pace of change with the risks of Second, �rms—big and small—can articulate reversal. Incremental policies will bring about a compelling business case for gender equality. In change only slowly. But transformative policies a fast-changing global economy, the demand for can risk a backlash. A way forward is to be se- skills has swelled, encouraging �rms to expand lective in implementing transformative policies their talent pool. Businesses have sought not and ensure that adequate attention is paid to only to attract and recruit female talent but also their implementation. to retain it through measures to facilitate work- life balance. Firms know that a diversity of opin- ions can enrich decision making and stimulate A GLOBAL AGENDA FOR GREATER ingenuity. And gender equality per se has grown GENDER EQUALITY to be a desirable trait that customers and inves- Domestic action is central to reducing inequali- tors look for. Corporate social responsibility is ties. Global action—by governments, people an avenue for �rms to enhance competitiveness and organizations in developed and developing through product differentiation and capture the countries, and by international institutions— loyalty of women’s growing market power. cannot substitute for equitable and ef�cient Third, shocks and exogenous changes can domestic policies and institutions. But it can present policy makers with windows of oppor- enhance the scope for and impact of domestic tunity to launch reforms that can improve gen- policies. And it can influence whether global der outcomes. Such windows sometimes stem integration and the opportunities it brings— from unpredictable circumstances, such as a through information, mobility, and technolo- national disaster. In 1998 the disaster that fol- gy—lead to greater gender equality and better lowed Hurricane Mitch in Nicaragua facilitated lives for all women, or just for some. a dialogue on domestic violence. A national Global action should focus on areas where campaign and the enactment of legislation fol- gender gaps are most signi�cant both intrinsi- lowed. Other windows come from shifts in the cally and in terms of their potential develop- political or economic landscape. The changes in ment payoff—and where growth alone cannot Spain during the democratic transition in the solve the issues. In other words, international late 1970s were particularly dramatic for fam- action should focus on complementing country ily law and reproductive rights. And still oth- efforts along the four priority areas identi�ed in ers emerge from the advocacy of transnational this Report: agencies and role modeling in the global agenda. For example, the monitoring and advocacy by • Reducing excess female mortality and closing Colombian local women’s groups in CEDAW education gaps where they remain informed the expansion of reproductive health • Improving access to economic opportunities guarantees in the Colombian Constitution, ad- for women opted in 1990, and facilitated greater access to contraception. • Increasing women’s voice and agency in the Finally, there are multiple paths to reform. household and in society Often governments follow societal cues in pacing • Limiting the reproduction of gender inequal- and pushing reforms. When policy formulation ity across generations and implementation follow cues from ongoing shifts in markets and social norms, convergence In addition, there is one cross-cutting prior- and alignment can fuel sustainable change. But ity: supporting evidence-based public action Overview 37 through better data, better knowledge genera- and across borders. Such partnerships could tion and sharing, and better learning. be built among those in the international de- The motivation for an agenda for global ac- velopment community around funding issues, tion is threefold. First, progress on some fronts with academia and think tanks for the purpose requires channeling more resources from rich of experimentation and learning, and, more to developing countries (for example, to cre- broadly, with the private sector to promote ac- ate greater equity in human endowments or to cess to economic opportunities. Together, these tackle the root causes of excess female mortal- partnerships could support countries in lever- ity around the world). Second, effective action aging the resources and information needed to sometimes hinges on the production of a pub- successfully promote gender equality in today’s lic good, such as the generation of new (global) globalized world. information or knowledge. And third, when the The relative importance of these activi- impact of a particular policy cuts across borders, ties will obviously vary across countries. Table coordination among a large number of coun- 2 provides a bird’s eye view of the proposed tries and institutions can enhance its effective- agenda for global action (described in more de- ness, not least by building momentum and pres- tail in chapter 9). Areas marked with a check are sure for action at the domestic level. those where new or additional action is needed Based on these criteria, initiatives included or where a refocus of existing initiatives is called in the proposed agenda for global action can be for. Of course, there are also important ongoing grouped into three types of activities: efforts in the areas not marked with checks—for instance, innovation around the delivery models Providing �nancial support. Improvements for the prevention of HIV/AIDS, or partnerships in the delivery of clean water and sanitation or focused on adolescents. In these latter areas, the better health services, such as the ones needed focus should be on sustaining ongoing efforts and to bring down excess female mortality among partnerships, and meeting prior commitments. girls and mothers in the developing world, will Finally, the framework and analysis presented require signi�cant resources—often beyond the in the Report provide four general principles for means of individual governments, particularly policy and program design, which can enhance those of relatively poor countries. The interna- the impact and effectiveness of global action tional development community can �nancially across all priority areas. These principles are: support countries willing and able to undertake • Comprehensive gender diagnostics as a pre- such reforms in a coordinated manner through condition for policy and program design. Gen- speci�c initiatives or funding facilities to ensure der disparities persist for multiple reasons: maximum impact and minimize duplication. there may be a single institutional or policy “�x� that is dif�cult and easily blocked; there Fostering innovation and learning. While a may be multiple reinforcing constraints in great deal has been learned about what works markets, formal institutions, and households and what does not when it comes to promot- that combine to block progress; or they may ing greater gender equality, the truth remains be deeply rooted in gender roles or social that progress is often held back by the lack of norms that evolve only slowly. Effective pol- data or adequate solutions to the most “sticky� icy design requires a good understanding of problems. That is the case, for instance, regard- which of these situations prevails in a partic- ing gender differences in time use patterns and ular context, and of where and what the the norms around care they stem from. The de- binding constraints are. To be useful, this di- velopment community could promote innova- agnostic must drill down into what happens tion and learning through experimentation and in households, markets, and formal institu- evaluation in ways that pay attention to results tions, their interactions, and how they are and process, as well as to context, and thus facili- shaped by social norms. tate a scaling-up of successful experiences. • Targeting determinants versus targeting out- Leveraging effective partnerships. As chapter comes. In choosing and designing policies, 8 makes clear, successful reform often requires it is necessary to target the market and in- coalitions or partnerships that can act within stitutional constraints that generate existing 38 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 TA B L E 2 The agenda for global action at a glance Directions for the global development community New/additional initiatives Providing �nancial Fostering innovation Priority area Leveraging partnerships that need support support and learning Increasing access to education √ √ among disadvantaged groups Increasing access to √ √ clean water Closing gender gaps in human endowments Increasing access to √ √ √ specialized maternal services Strengthening support for prevention and treatment of √ √ HIV/AIDS Increasing access to child Promoting women’s access to care and early childhood √ √ economic opportunities development Investing in rural women √ √ Increasing women’s access to √ Closing gender gaps in voice the justice system and agency Shifting norms regarding √ √ violence against women Preventing intergenerational Investing in adolescent girls reproduction of gender √ and boys inequality Generating new information √ √ Supporting evidence-based public action Facilitating knowledge sharing √ and learning Source: WDR 2012 team. gender gaps, rather than the outcomes them- tures, and as a result so do household and selves. These constraints may be multiple and individual behaviors. This implies that the even outside the immediate domain where same policy can have very different results the outcome is observed. depending on the contexts—or, as the dis- cussion in chapter 8 makes clear, that there • “Upstreaming� and strategic mainstreaming. are multiple paths to reform. Because gender gaps are often the result of multiple and mutually reinforcing constraints, effective action may require coordinated NOTES multisectoral interventions, or sequential in- terventions. And in many instances, such 1. Sen 1999. 2. Esteve-Volart and Bagues 2010. interventions can take the form of general 3. Gilbert, Sakala, and Benson 2002; Vargas Hill and policies that are made “gender smart� by in- Vigneri 2009. corporating gender-related issues into their 4. Udry 1996. design and implementation. To maximize 5. FAO, IFAD, and ILO 2010. impact, it is thus necessary for gender issues 6. Cuberes and Teignier Baqué 2011; Hurst and oth- to be upstreamed from speci�c sector prod- ers 2011. ucts and projects to country and sector pro- 7. Do, Levchenko, and Raddatz 2011. grams. That will allow for more strategic 8. Do, Levchenko, and Raddatz 2011. gender mainstreaming. 9. Do, Levchenko, and Raddatz 2011. 10. Haddad, Hoddinott, and Alderman 1997; Katz • No one size �ts all. The nature, structure, and and Chamorro 2003; Duflo 2003; Thomas 1990; functioning of markets and institutions vary Hoddinott and Haddad 1995; Lundberg, Pol- widely across countries, as do norms and cul- lak, and Wales 1997; Quisumbing and Maluccio Overview 39 2000; Attanasio and Lechene 2002; Rubalcava, 48. Fernández and Fogli 2009; Fogli and Veldkamp, Teruel, and Thomas 2009; Doss 2006; Schady and forthcoming; Farré and Vella 2007. Rosero 2008. 49. Agarwal and Panda 2007. 11. Doss 2006. 50. Gage 2005. 12. Thomas 1990. 51. Yount and Carrera 2006; Castro, Casique, and 13. Qian 2008. Brindis 2008. 14. Luke and Munshi 2011. 52. Abrahams and others 2009. 15. Thomas, Strauss, and Henriques 1990; Allendorf 53. Kishor and Johnson 2004. 2007. 54. World Bank 2005. 16. Andrabi, Das, and Khwaja 2011; Dumas and 55. Chung and Das Gupta 2007. Lambert 2011. 56. Waddington and others 2009. 17. Felitti and others 1998; McEwen 1999. 57. Chioda, Garcia-Verdú, and Muñoz Boudet 2011. 18. Kishor and Johnson 2004; Jeyaseelan and others 58. Björkman and Svensson 2009. 2007; Hindin, Kishor, and Ansara 2008; Koenig 59. Cotlear 2006. and others 2006; Martin and others 2002. 60. Lim and others 2010. Janani Suraksha Yojana 19. Miller 2008. also had signi�cant impacts on perinatal and 20. Beaman and others, forthcoming; Chattopad- neonatal deaths, which declined by 3.7 deaths per hyay and Duflo 2004. 1,000 pregnancies and by 2.5 deaths per 1,000 21. Beaman and others, forthcoming. live births, respectively. The study was unable to 22. Agarwal 2010a; Agarwal 2010b. detect an effect on maternal mortality; however, 23. World Bank 2005. perhaps because maternal death is a relatively 24 World Bank 2001. rare event and the sample size of the study was 25. World Bank 2011. big enough only to detect very large effects. 26. See World Bank (2001) and also World Bank 61. Prata and others 2010; WHO and others 2010. (2011b), from which we draw for the conceptual 62. Baris, Mollahaliloglu, and Sabahattin 2011. framework. 63. Fiszbein and others 2009. 27. Chioda, Garcia-Verdú, and Muñoz Boudet 2011. 64. FAO 2003. 28. World Bank 2008. 65. Deininger, Ali, and Zevenbergen 2008. 29. Lewis and Lockheed 2006. 66. Leonard 1989, Holzer and Neumark 2000. 30. Chioda, Garcia-Verdú, and Muñoz Boudet 2011. 67. Holzer and Neumark 2000. 31. For a detailed description of the methodology, 68. Gornick and Jacobs 1998; OECD 1993; Schmidt which builds on Anderson and Ray (2010), see 1993. chapter 3. 69. Bosch and Maloney 2010. 32. The problem of many missing girls was �rst doc- 70. Esteve-Volart and Bagues 2010. umented by Sen (1992), Coale (1984), and Das 71. Quotas are implemented on a rotating basis Gupta (1987). across localities. 33. WHO, UNICEF, UNFPA, and World Bank 2010. 72. Gajigo and Hallward-Driemeier 2011. 34. FAO 2011. 73. Natarajan 2005. 35. For the purpose of the discussion in this chap- 74. Barker and Ricardo 2005. ter, the term “entrepreneur� refers to individuals 75. Terefe and Larson 1993. who are self-employed with no employees own 76. See Rawlings and Rubio (2003) for Mexico and account workers and with employees employers. Nicaragua, Barrera-Osorio and Linden (2009) 36. Sabarwal, Terrell, and Bardasi 2009; Bruhn 2009; for Colombia, and Schady and Araujo (2006) for Hallward-Driemeier 2011. Ecuador. 37. Costa and Rijkers 2011. 77. Baird and others 2009. 38. FAO 2011. 78. Nguyen 2008. 39. Nyamu-Musembi 2002. 79. Jensen 2010. 40. Deere and Doss 2006. 80. Angrist and Lavy 2009; Kremer, Miguel, and 41. Team estimates based on ICF Macro 2010. Thornton 2009. 42. Reed and others 2010. 81. Attanasio, Kugler, and Meghir 2008; Ñopo, Robles, 43. WHO 2005. and Saavedra 2007; Hjort and others 2010. 44. United Nations Department of Economic and 82. Hjort and others 2010. Social Affairs 2010. 83. Ross and others 2007. 45. Agarwal and Panda 2007. 84. Gilliam 2010; Bearinger and others 2007. 46. Pronyk and others 2006; ICRW 2006; Swamina- 85. Bandiera and others 2011. than, Walker, and Rugadya 2008. 86. Martinez and others 2011. 47. Dercon and Singh 2011. 87. Beaman and others 2009. 40 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 88. Jensen 2010. Working Paper Series 13931, National Bureau of 89. Barker and others 2011. Economic Research, Cambridge, MA. Attanasio, Orazio, and Valérie Lechene. 2002. “Tests of Income Pooling in Household Decisions.� Re- REFERENCES view of Economic Dynamics 5 (4): 720–48. Baird, Sarah, Ephraim Chirwa, Craig McIntosh, and The word processed describes informally repro- Berk Özler. 2009. “The Short-term Impacts of a duced works that may not be commonly avail- Schooling Conditional Cash Transfer Program able through libraries. on the Sexual Behavior of Young Women.� Policy Research Working Paper Series 5089, World Bank, Abrahams, Naeemah, Rachel Jewkes, Lorna J. Martin, Washington, DC. Shanaaz Mathews, Lisa Vetten, and Carl Lombard. Bandiera, Oriana, Niklas Buehren, Robin Burguess, 2009. “Mortality of Women from Intimate Partner Markus Goldstein, Selim Gulesci, Imran Rasul, Violence in South Africa: A National Epidemiolog- and Munshi Sulaiman. 2011. “Economic Em- ical Study.� Violence and Victims 24 (4): 546–56. powerment of Female Adolescents: Evidence Agarwal, Bina. 1994. A Field of One’s Own: Gender from Uganda.� Presentation to the American Agri- and Land Rights in South Asia. Cambridge, U.K.: cultural Association, Pittsburgh, PA. Cambridge University Press. Baris, Enis, Salih Mollahaliloglu, and Aydin Sabahat- ———. 1997. “ ‘Bargaining’ and Gender Relations: tin. 2011. “Healthcare in Turkey: From Laggard Within and Beyond the Household.� Feminist Eco- to Leader.� British Medical Journal 342 (c7456): nomics 3 (1): 1–51. 579–82. ———. 2010a. “Does Women’s Proportional Barker, Gary, Manuel Contreras, Brian Heilman, Ajay Strength Affect Their Participation? Governing Singh, Ravi Verman, and Marcos Nascimento. Local Forests in South Asia.� World Development 2011. “Evolving Men: Initial Results from the 38 (1): 98–112. International Men and Gender Equality Survey ———. 2010b. Gender and Green Governance: The (IMAGES).� International Center for Research on Political Economy of Women’s Presence Within and Women and Instituto Promundo, Washington, Beyond Community Forestry. New York: Oxford DC. University Press. Barker, Gary, and Christine Ricardo. 2005. “Young Agarwal, Bina, and Pradeep Panda. 2007. “Toward Men and the Construction of Masculinity in Sub- Freedom from Domestic Violence: The Neglected Saharan Africa: Implications for HIV/AIDS, Con- Obvious.� Journal of Human Development and Ca- flict, and Violence.� Social Development Papers, pabilities 8 (3): 359–88. World Bank, Washington, DC. Alene, Arega D., Victor M. Manyong, Gospel O. Barrera-Osorio, Felipe, and Leigh L. Linden. 2009. Omanya, Hodeba D. Mignouna, Mpoko Bokanga, “The Use and Misuse of Computers in Educa- and George D. Odhiambo. 2008. “Economic Ef�- tion: Evidence from a Randomized Experiment in ciency and Supply Response of Women as Farm Colombia.� Policy Research Working Paper Series Managers: Comparative Evidence from Western 4836, World Bank, Washington, DC. Kenya.� World Development 36 (7): 1247–60. Beaman, Lori, Raghabendra Chattopadhyay, Esther Allendorf, Keera. 2007. “Do Women’s Land Rights Duflo, Rohini Pande, and Petia Topalova. 2009. Promote Empowerment and Child Health in Ne- “Powerful Women: Does Exposure Reduce Bias?� pal?� World Development 35 (11): 1975–88. Quarterly Journal of Economics 124 (4): 1497–540. Anderson, Siwan, and Debraj Ray. 2010. “Missing Beaman, Lori, Esther Duflo, Rohini Pande, and Petia Women: Age and Disease.� Review of Economic Topalova. Forthcoming. “Political Reservation and Studies 77 (4): 1262–300. Substantive Representation: Evidence from Indian Andrabi, Tahir, Jishnu Das, and Asim Ijaz Khwaja. Village Councils.� In India Policy Forum, 2010, ed. 2011. “Students Today, Teachers Tomorrow. Iden- Suman Bery, Barry Bosworth, and Arvind Pana- tifying Constraints on the Provision of Educa- gariya. Brookings Institution Press and the Na- tion.� Policy Research Working Paper Series 5674, tional Council of Applied Economic Research, World Bank, Washington, DC. Washington, DC, and New Delhi. Angrist, Joshua, and Victor Lavy. 2009. “The Effects of Bearinger, Linda H., Renee E. Sieving, Jane Ferguson, High Stakes High School Achievements Awards: and Vinit Sharma. 2007. “Global Perspectives on Evidence from a Randomized Trial.� American the Sexual and Reproductive Health of Adoles- Economic Review 99 (4): 1384–414. cents: Patterns, Prevention, and Potential.� Lancet Attanasio, Orazio, Adriana Kugler, and Costas Meghir. 369 (9568): 1220–31. 2008. “Training Disadvantaged Youth in Latin Berniell, Maria Inés, and Carolina Sánchez-Páramo. America: Evidence from a Randomized Trial.� 2011. “Overview of Time Use Data Used for the Overview 41 Analysis of Gender Differences in Time Use Pat- Deere, Carmen Diana, and Cheryl R. Doss. 2006. terns.� Background paper for the WDR 2012. “Gender and the Distribution of Wealth in Devel- Björkman, Martina, and Jacob Svensson. 2009. “Power oping Countries.� Research Paper Series 2006/115, to the People: Evidence from a Randomized Field United Nations University and World Institute for Experiment on Community-Based Monitoring in Development Economics Research, Helsinki. Uganda.� Quarterly Journal of Economics 124 (2): Deininger, Klaus, Daniel Ayalew Ali, Holden T. Stein, 735–69. and Jaap Zevenbergen. 2008. “Rural Land Certi- Booth, Alison L., and Patrick J. Nolen. 2009. “Gen- �cation in Ethiopia: Process, Initial Impact, and der Differences in Risk Behaviour: Does Nurture Implications for Other African Countries.� World Matter?� Centre for Economy Policy Research, Development 36 (10): 1786–812. London. Dercon, Stefan, and Abhijeet Singh. 2011. “From Nu- Bosch, Mariano, and William F. Maloney. 2010. trition to Aspirations and Self-Ef�cacy: Gender “Comparative Analysis of Labor Market Dynam- Bias over Time among Children in Four Coun- ics Using Markov Processes: An Application to tries.� Oxford University, Oxford, U.K. Processed. Informality.� Labour Economics 17 (4): 621–31. Do, Quy-Toan, Andrei Levchenko, and Claudio Bruhn, Miriam. 2009. “Female-Owned Firms in Latin Raddatz. 2011. “Engendering Trade.� Background America. Characteristics, Performance, and Ob- paper for the WDR 2012. stacles to Growth.� Policy Research Working Paper Doss, Cheryl R. 2006. “The Effects of Intrahousehold Series 5122, World Bank, Washington, DC. Property Ownership on Expenditure Patterns Castro, Roberto, Irene. Casique, and Claire D. Brin- in Ghana.� Journal of African Economies 15 (1): dis. 2008. “Empowerment and Physical Violence 149–80. throughout Women’s Reproductive Life in Mex- Duflo, Esther. 2003. “Grandmothers and Grand- ico.� Violence Against Women 14 (6): 655–77. daughters: Old-Age Pensions and Intrahousehold Chattopadhyay, Raghabendra, and Esther Duflo. Allocation in South Africa.� World Bank Economic 2004. “Women as Policy Makers: Evidence from a Review 17 (1): 1–25. Randomized Policy Experiment in India.� Econo- Dumas, Christelle, and Sylvie Lambert. 2011. “Edu- metrica 72 (5): 1409–43. cational Achievement and Socio-Economic Back- Chioda, Laura, with Rodrigo Garcia-Verdú, and Ana ground: Causality and Mechanisms in Senegal.� María Muñoz Boudet. 2011. Work and Family: Journal of African Economies 20 (1): 1–26. Latin American Women in Search of a New Balance. Esteve-Volart, Berta, and Manuel F. Bagues. 2010. “Are Of�ce of the Chief Economist and Poverty Gender Women Pawns in the Political Game? Evidence Group, LAC. Washington, DC: World Bank. from Elections to the Spanish Senate.� Working Chung, Woojin, and Monica Das Gupta. 2007. “The Paper Series 2009–30, Fundación de Estudios de Decline of Son Preference in South Korea: The Economía Aplicada, Madrid. Roles of Development and Public Policy.� Popula- Farré, Lídia, and Francis Vella. 2007. “The Intergen- tion and Development Review 33 (4): 757–83. erational Transmission of Gender Role Attitudes Coale, Ansley J. 1984. “Rapid Population Change in and Its Implications for Female Labor Force Par- China, 1952–1982.� Report 27, Committee on ticipation.� Discussion Paper Series 2802, Institute Population and Demography, National Academies for the Study of Labor, Bonn. Press, Washington, DC. Fehr, Ernst, Urs Fischbacher, and Simon Gächter. Costa, Rita, and Bob Rijkers. 2011. “Gender and Rural 2002. “Strong Reciprocity, Human Cooperation Non-Farm Entrepreneurship.� Background paper and the Enforcement of Social Norms.� Human for the WDR 2012. Nature 13 (2002): 1–25. Cotlear, Daniel, ed. 2006. “A New Social Contract for Felitti, Vincent D., Robert F. Anda, Dale D. Norden- Peru: An Agenda for Improving Education, Health berg, David F. Williamson, Alison M. Spitz, Valerie Care, and the Social Safety Net.� Country Study, Edwards, Mary P. Koss, and James S. Marks. 1998. World Bank, Washington, DC. “Relationship of Childhood Abuse and House- Croson, Rachel, and Uri Gneezy. 2009. “Gender Dif- hold Dysfunction to Many of the Leading Causes ferences in Preferences.� Journal of Economic Lit- of Death in Adults: The Adverse Childhood Expe- erature 47 (2): 448–74. riences (ACE) Study.� American Journal of Preven- Cuberes, David, and Marc Teignier Baqué. 2011. tive Medicine 14 (4): 245–58. “Gender Inequality and Economic Growth.� Fernández, Raquel, and Alessandra Fogli. 2009. “Cul- Background paper for the WDR 2012. ture: An Empirical Investigation of Beliefs, Work, Das Gupta, Monica. 1987. “Selective Discrimination and Fertility.� American Economic Journal: Macro- against Female Children in Rural Punjab, In- economics 1 (1): 146–77. dia.� Population and Development Review 13 (1): Fiszbein, Ariel, Norbert Schady, Francisco H. G. 77–100. Ferreira, Margaret Grosch, Nial Kelleher, Pedro 42 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Olinto, and Emmanuel Skou�as. 2009. World Bank Hjort, Jonas, Michael Kremer, Isaac Mbiti, and Ed- Policy Research Report: Conditional Cash Transfers: ward Miguel. 2010. “Vocational Education Vouch- Reducing Present and Future Poverty. Washington, ers and Labor Market Returns: A Randomized DC: World Bank. Evaluation among Kenyan Youth.� Harvard Uni- Fogli, Alessandra, and Laura Veldkamp. Forth- versity and Southern Methodist University, Berke- coming. “Nature or Nurture? Learning and the ley, CA. Processed. Geography of Female Labor Force Participation.� Hoddinott, John, and Lawrence Haddad. 1995. “Does Econometrica. Female Income Share Influence Household Ex- FAO (Food and Agriculture Organization). 2003. penditures? Evidence from Côte D’Ivoire.� Oxford “Revisiting the ‘Magic Box.’ � Case Studies in Lo- Bulletin of Economics and Statistics 57 (1): 77–96. cal Appropriation of Information and Communi- Holzer, Harry J., and David Neumark. 2000. “Assess- cation Technologies, FAO, Rome. ing Af�rmative Action.� Journal of Economic Lit- ———. 2011. “The State of Food and Agriculture erature 38 (3): 483–568. 2010–11. Women in Agriculture: Closing the Gen- Hurst, Erik, Chang-Tai Hsieh, Charles Jones, and der Gap for Development.� FAO, Rome. Peter Klenow. 2011. “The Allocation of Talent FAO (Food and Agriculture Organization), IFAD and Economic Growth.� Chicago Booth, Chicago. (International Fund for Agricultural Develop- Processed. ment), and ILO (International Labour Of�ce). ICF Macro. 2010. “Demographic and Health Surveys.� 2010. “Gender Dimensions of Agricultural and Measure DHS, ICF Macro, Calverton, MD. Rural Employment: Differentiated Pathways out ILO (International Labour Organization). 2010. “Key of Poverty. Status, Trends and Gaps.� FAO, IFAD, Indicators of the Labour Market.� ILO, Geneva. and ILO, Rome. ICRW (International Center for Research on Gage, Anastasia J. 2005. “Women’s Experience of In- Women). 2006. Property Ownership & Inheritance timate Partner Violence in Haiti.� Social Science & Rights of Women for Social Protection: The South Medicine 61 (2): 343–64. Asia Experience. Washington, DC: ICRW. Gajigo, Ousman, and Mary Hallward-Driemeier. Iyer, Lakshmi, Anandi Mani, Prachi Mishra, and 2011. “Constraints and Opportunities for New Petia Topalova. 2010. “Political Representation Entrepreneurs in Africa.� World Bank, Washing- and Crime: Evidence from India’s Panchayati Raj.� ton, DC. Processed. International Monetary Fund, Washington, DC. Gilbert, Robert A., Webster D. Sakala, and Todd D. Processed. Benson. 2002. “Gender Analysis of a Nationwide Jensen, Robert. 2010. “Economic Opportunities and Cropping System Trial Survey in Malawi.� African Gender Differences in Human Capital: Experi- Studies Quarterly 6 (1). mental Evidence for India.� Working Paper Series Gilliam, Melissa L. 2010. “Interventions for Prevent- ing Unintended Pregnancies among Adolescents.� 16021, National Bureau of Economic Research, Obstetrics and Gynecology 115 (1): 171–72. Cambridge, MA. Gneezy, Uri, Kenneth L. Leonard, and John A. List. Jeyaseelan, L., Shuba Kumar, Nithya Neelakantan, 2009. “Gender Differences in Competition: Evi- Abraham Peedicayil, Rajamohanam Pillai, and dence from a Matrilineal and a Patriarchal Soci- Nata Duvvury. 2007. “Physical Spousal Violence ety.� Econometrica 77 (5): 1637–64. against Women in India: Some Risk Factors.� Jour- Gornick, Janet C., and Jerry A. Jacobs. 1998. “Gen- nal of Biosocial Science 39 (5): 657–70. der, the Welfare State, and Public Employment: Kabeer, Nalia. 1996. “Agency, Well-Being & Inequal- A Comparative Study of Seven Industrialized ity: Reflections on the Gender Dimensions of Pov- Countries.� American Sociological Review 63 (5): erty.� IDS Bulletin 27 (1): 11–21. 688–710. ———. 1999. “Resources, Agency, Achievements: Haddad, Lawrence, John Hoddinott, and Harold Al- Reflections on the Measurement of Women’s derman. 1997. Intrahousehold Resource Allocation Empowerment.� Development and Change 30 (3): in Developing Countries: Models, Methods, and 35–64. Policy. Baltimore: Johns Hopkins University. Katz, Elizabeth, and Juan Sebastian Chamorro. 2003. Hallward-Driemeier, Mary. 2011. “Improving the “Gender, Land Rights, and the Household Econ- Legal Investment Climate for Women in Sub- omy in Rural Nicaragua and Honduras.� Paper Saharan Africa.� World Bank, Washington, DC. presented at the Annual Conference of the Latin Hindin, Michelle J., Sunita Kishor, and Donna L. American and Caribbean Economics Association, Ansara. 2008. “Intimate Partner Violence among Puebla, Mexico, October 9. Couples in 10 DHS Countries: Predictors and Kinkingninhoun-Mêdagbé, Florent M., Aliou Diagne, Health Outcomes.� DHS Analytical Studies 18, Franklin Simtowe, A�avi R. Agboh-Noameshie, U.S. Agency for Iinternational Development, and Patrice Y. Adégbola. 2010. “Gender Discrimi- Washington, DC. nation and Its Impact on Income, Productivity, Overview 43 and Technical Ef�ciency: Evidence from Benin.� Nguyen, Trang. 2008. “Information, Role Models Agriculture and Human Values 27 (1): 57–69. and Perceived Returns to Education: Experimen- Kishor, Sunita, and Kiersten Johnson. 2004. Pro�ling tal Evidence from Madagascar.� Working Paper, Domestic Violence: A Multi-Country Study. Calver- Massachusetts Institute of Technology, Cam- ton, MD: ORC Macro. bridge, MA. Koenig, Michael A., Rob Stephenson, Saifuddin Ñopo, Hugo, Miguel Robles, and Jaime Saavedra. Ahmed, Shireen J. Jejeebhoy, and Jacquelyn 2007. “Occupational Training to Reduce Gender Campbell. 2006. “Individual and Contextual De- Segregation: The Impacts of ProJoven.� Working terminants of Domestic Violence in Northern Paper Series 623, Inter-American Development India.� American Journal of Public Health 96 (1): Bank Research Department, Washington, DC. 132–38. Nyamu-Musembi, Celestine. 2002. “Are Local Norms Kremer, Michael, Edward Miguel, and Rebecca and Processes Fences or Pathways? The Example Thornton. 2009. “Incentives to Learn.� Review of of Women’s Property Rights.� In Cultural Trans- Economics and Statistics 91 (3): 437–56. formations and Human Rights in Africa, ed. Ab- Leonard, Jonathan S. 1989. “Women and Af�rmative dullahi A. An-Na’im. London: Zed Books. Action.� Journal of Economic Perspectives 3 (1): Oladeebo, J. O., and A. A. Fajuyigbe. 2007. “Technical 61–75. Ef�ciency of Men and Women Upland Rice Farm- Lewis, Maureen A., and Marlaine E. Lockheed. 2006. ers in Osun State, Nigeria.� Journal of Human Ecol- Inexcusable Absence: Why 60 Million Girls Aren’t in ogy 22 (2): 93–100. School and What to Do About It. Washington, DC: OECD (Organisation for Economic Cooperation and Center for Global Development. Development). 1993. Private Pay for Public Work. Lim, Stephen S., Lalit Dandona, Joseph A. Hoisington, Performance-Related Pay for Public Sector Manag- Spencer L. James, Margaret C. Hogan, and Em- ers. Paris: OECD. manuela Gakidou. 2010. “India’s Janani Suraksha Pathmanathan, Indra, Jerker Liljestrand, Jo M. Mar- Yojana, A Conditional Cash Transfer Programme tins, Lalini C. Rajapaksa, Craig Lissner, Amalia to Increase Births in Health Facilities: An Impact de Silva, Swarna Selvaraju, and Prabha Joginder Evaluation.� Lancet 375 (9730): 2009–23. Singh. 2003. “Investing in Maternal Health Learn- Luke, Nancy, and Kaivan Munshi. 2011. “Women as ing from Malaysia and Sri Lanka.� World Bank, Agents of Change: Female Income and Mobility Washington, DC. in India.� Journal of Development Economics 94 Prata, Ndola, Paige Passano, Amita Sreenivas, and (1): 1–17. Caitlin Elisabeth Gerdts. 2010. “Maternal Mortal- Lundberg, Shelly J., Robert A. Pollak, and Terence J. ity in Developing Countries: Challenges in Scaling Wales. 1997. “Do Husbands and Wives Pool Their Up Priority Interventions.� Women’s Health 6 (2): Resources? Evidence from the United Kingdom Child Bene�t.� Journal of Human Resources 32 (3): 311–27. 463–80. Pronyk, Paul M., James R. Hargreaves, Julia C. Kim, Martin, Sandra L., Kathryn E. Moracco, Julian Garro, Linda A. Morison, Godfrey Phetla, Charlotte Amy Ong Tsui, Lawrence L. Kupper, Jennifer L. Watts, Joanna Busza, and John D. H. Porter. 2006. Chase, and Jacquelyn C. Campbell. 2002. “Do- “Effect of a Structural Intervention for the Pre- mestic Violence across Generations: Findings vention of Intimate-partner Violence and HIV in from Northern India.� International Journal of Rural South Africa: A Cluster Randomized Trial.� Epidemiology 31 (3): 560–72. Lancet 2368 (9551): 1973–83. Martinez, Sebastian, and others. 2011. “Hard Skills Qian, Nancy. 2008. “Missing Women and the Price of or Soft Skills.� Presentation to the World Bank, Tea in China: The Effect of Sex-Speci�c Earnings Washington, DC. on Sex Imbalance.� Quarterly Journal of Economics McEwen, Bruce S. 1999. “Stress and Hippocampal 123 (3): 1251–85. Plasticity.� Annual Review of Neuroscience 22 (1): Quisumbing, Agnes R., and John A. Maluccio. 2000. 105–22. “Intrahousehold Allocation and Gender Rela- Miller, Grant. 2008. “Women’s Suffrage, Political tions: New Empirical Evidence from Four De- Responsiveness, and Child Survival in American veloping Countries.� Discussion Paper 84, Food History.� Quarterly Journal of Economics 123 (3): Consumption and Nutrition Division, Interna- 1287–327. tional Food Policy Research Institute, Washing- Moock, Peter R. 1976. “The Ef�ciency of Women as ton, DC. Farm Managers: Kenya.� American Journal of Agri- Rawlings, Laura, B., and Gloria M. Rubio. 2003. “Eval- cultural Economics 58 (5): 831–5. uating the Impact of Conditional Cash Transfer Natarajan, Mangai. 2005. “Status of Women Police in Programs: Lessons from Latin America.� Policy Asia: An Agenda for Future Research.� Journal for Research Working Paper Series 3119, World Bank, Women and Policing 17: 45–47. Washington, DC. 44 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Reed, Elizabeth, Anita Raj, Elizabeth Miller, and Husbands Make a Difference?� American Journal Jay G. Silverman. 2010. “Losing the ‘Gender’ in of Public Health 83 (11): 1567–71. Gender-Based Violence: The Missteps of Research Thomas, Duncan. 1990. “Intra-Household Resource on Dating and Intimate Partner Violence.� Vio- Allocation: An Inferential Approach.� Journal of lence Against Women 16 (3): 348–54. Human Resources 25 (4): 635–64. Ross, David A., John Changalucha, Angela I. N. Obasi, Thomas, Duncan, John Strauss, and Maria-Helena Jim Todd, Mary L. Plummer, Bernadette Cleophas- Henriques. 1990. “Child Survival, Height for Age, Mazige, Alessandra Anemona, Dean Everett, and Household Characteristics in Brazil.� Journal Helen A. Weiss, David C. Mabey, Heiner of Development Economics 33 (2): 197–234. Grosskurth, and Richard J. Hayes. 2007. “Biologi- Udry, Christopher. 1996. “Gender, Agricultural Pro- cal and Behavioural Impact of an Adolescent Sex- duction, and the Theory of the Household.� Jour- ual Health Intervention in Tanzania: A Commu- nal of Political Economy 104 (5): 1010–46. nity-Randomized Trial.� AIDS 21 (14): 1943–55. United Nations Department of Economic and Social Rubalcava, Luis, Graciela Teruel, and Duncan Affairs. 2009, “World Population Prospects 2009.� Thomas. 2009. “Investments, Time Preferences, United Nations, New York. and Public Transfers Paid to Women.� Economic ———. 2010. “The World’s Women 2010: Trends Development and Cultural Change 57 (3): 507–38. and Statistics.� United Nations, New York. Sabarwal, Shwetlena, Katherine Terrell, and Elena Vargas Hill, Ruth, and Marcella Vigneri. 2009. “Main- Bardasi. 2009. “How Do Female Entrepreneurs streaming Gender Sensitivity in Cash Crop Perform? Evidence from Three Developing Re- Markets Supply Chains.� International Food Pol- gions.� World Bank, Washington, DC. Processed. icy Research Institute, Washington, DC. Saito, Katrine A., Hailu Mekonnen, and Daphne Waddington, Hugh, Birte Snilstveit, Howard White, Spurling. 1994. “Raising the Productivity of and Lorna Fewtrell. 2009. “Water, Sanitation Women Farmers in Sub-Saharan Africa.� Africa and Hygiene Interventions to Combat Childhood Technical Department Discussion Paper Series Diarrhoea in Developing Countries.� Interna- 230, World Bank, Washington, DC. tional Initiative for Impact Evaluation 31E, Syn- Schady, Norbert, and Maria Caridad Araujo. 2006. thetic Review, New Delhi, London, and Washing- “Cash Transfers, Conditions, School Enrollment, ton, DC. and Child Work: Evidence from a Randomized WHO (World Health Organization). 2005. “WHO Experiment in Ecuador.� Policy Research Working Multi-country Study on Women’s Health and Paper Series 3930, World Bank, Washington, DC. Domestic Violence against Women: Initial Results Schady, Norbert, and José Rosero. 2008. “Are Cash on Prevalence, Health Outcomes and Women’s Transfers Made to Women Spent Like Other Responses.� WHO, Geneva. Sources of Income?� Economics Letters 101 (3): WHO, UNICEF (United Nations Children Fund), 246–48. UNFPA (United Nations Population Fund), and Schmidt, Manfred G. 1993. “Gendered Labour Force World Bank. 2010. “Trends in Maternal Mortality: Participation.� In Families of Nations: Patterns of 1990 to 2008.� WHO, Washington, DC. Public Policy in Western Democracies, ed. Frances World Bank. 2001. “Engendering Development— G. Castles. Dartmouth Publishing Company, Al- Through Gender Equality, Resources, and Voice. dershot, U.K., and Brook�eld, VT. Policy Research Report, World Bank, Washington, Sen, Amartya. 1990. “Gender and Cooperative DC.� Conflict.� In Persistent Inequalities: Women and ———-. 2005. World Development Report 2006: Eq- Development, ed. Irene Tinker. Oxford, U.K.: Ox- uity and Development. New York: Oxford Univer- ford University Press. sity Press. ———. 1992. “Missing Women.� British Medical ———. 2008. Growth Report. Strategies for Sustained Journal 304: (6827): 587–8. Growth and Inclusive Development. Washington, ———. 1999. Development as Freedom. New York: DC: World Bank. Knopf. ———. 2011. “De�ning Gender in the 21st Century: Swaminathan, Hema, Cherryl Walker, and Margaret Talking with Women and Men around the World, A. Rugadya, eds. 2008. Women’s Property Rights, A Multi-Country Qualitative Study of Gender HIV and AIDS, and Domestic Violence: Research and Economic Choice.� World Bank, Washington, Findings from Two Rural Districts in South Africa DC. and Uganda. Cape Town: HSRC Press. Yount, Kathryn M., and Jennifer S. Carrera. 2006. Terefe, Almas, and Charles P. Larson. 1993. “Modern “Domestic Violence against Married Women in Contraception Use in Ethiopia: Does Involving Cambodia.� Social Forces 85 (1): 355–87. A guide to the Report G ender refers to socially constructed develop? We argue in this Report that the links and learned female and male roles, between gender equality and development go behaviors, and expectations. All cul- both ways and that each direction of this re- tures interpret and translate the bio- lationship matters for policy making. Higher logical differences between men and women incomes and improved service delivery—both into beliefs about what behaviors and activities essential elements of broad-based economic de- are appropriate for each gender as well as their velopment—contribute to greater gender equal- rights, resources, and power. For example, most ity. That is why the rise in global prosperity in societies give the primary responsibility for the the past quarter century has seen the unprece- care of infants and young children to women dented narrowing of gender gaps on many edu- and girls, and that for military service and na- cation and health outcomes as well as in labor tional security to men. Gender thus shapes one’s market opportunities. More women than men life chances and one’s role in the home, in soci- now attend universities across the world. And ety, and in the economy. women make up over 40 percent of the world’s Played out over the life cycles of individuals, labor force. gender can translate into inequalities—in hu- But not all gender gaps have shrunk or are man capital, economic opportunities, citizen- shrinking with rising incomes. Poor girls and ship, and political participation. It determines those who live in remote areas or belong to ex- the way households allocate resources to sons cluded groups are far less likely to attend pri- and daughters, through decisions about boys’ mary and secondary school than boys in the or girls’ education or about where they work, same circumstances. Compared with their high- with sons typically working on the farm and in income counterparts, women and girls in low- other market work while daughters work in the and middle-income countries die at higher rates home and care-giving activities. By the time girls relative to men, especially in the critical years and boys become adults and form households, of infancy and early childhood, and in their women typically have fewer years of education reproductive years. Women continue to clus- than men (although this is changing rapidly), ter in sectors and occupations characterized as work longer hours but fewer in the labor force, “female�—many of them lower paying. Women earn lower wages, and have less say in their com- are more likely to be the victims of violence at munities and societies. home and to suffer more severe injuries. And al- most everywhere, the representation of women in politics and in senior managerial positions in GENDER EQUALITY AND business remains far lower than that of men. DEVELOPMENT: WHY DO THE Understanding which of these gaps respond LINKS MATTER? to economic development and why they do so is relevant to policy because it helps shine the light Why do these gender differences matter for de- on the gender gaps that need attention. The dis- velopment, and how do they evolve as countries parities between women and men or girls and A guide to the Report 47 boys that shrink as countries get richer—differ- The Food and Agriculture Organization ences in access to education, for example—need (FAO) estimates that equalizing access to pro- less policy attention through a gender lens than ductive resources for female and male farmers those that are more persistent, such as differ- could increase agricultural output in develop- ences in wages, agricultural productivity, and ing countries by as much as 2.5 to 4 percent.1 societal voice. Eliminating barriers preventing women from The reverse relationship—from gender equal- entering certain sectors or occupations would ity to development—also matters for policy have similar positive effects, increasing output for two reasons. First, gender equality matters per worker by 13 to 25 percent.2 in its own right, because the ability to live the These gains are large in the 21st century’s life of one’s own choosing and be spared from integrated and competitive world, where even absolute deprivation is a basic human right, modest improvements in the ef�ciency of re- to be enjoyed by everyone, whether one is male source use can have signi�cant effects on growth. or female. Because development is a process In a world of open trade, gender inequality has of expanding freedoms equally for all people, become more costly because it diminishes a gender equality is a core objective in itself. country’s ability to compete internationally— Just as lower income poverty or greater access particularly if the country specializes in export- to justice is part of development, so too is the ing goods and services that men and women narrowing of gaps in well-being between males workers are equally well suited to produce.3 and females. Industries that rely more on female labor ex- Second, greater gender equality can enhance pand more in countries where women are more economic ef�ciency and improve other devel- equal.4 In a globalized world, then, countries opment outcomes. Evidence from a (slowly) that reduce gender-based inequalities, especially growing set of microeconomic studies points to in secondary and tertiary education and in eco- three main channels for greater gender equality nomic participation, will have a clear advantage to promote growth: over those that delay action. • Reducing barriers to more ef�cient alloca- The rapid aging of the world’s population tion of women’s skills and talents can gener- implies that fewer workers will be supporting ate large (and growing) productivity gains. growing numbers of elderly in the decades to come, unless labor force participation increases • Improving women’s endowments, opportu- signi�cantly among groups that participate less nities, and agency can shape more positive today—mainly women. For instance, in Europe outcomes for the next generation. an anticipated shortfall of 24 million workers • Increasing women’s individual and collective by 2040 could be reduced to 3 million if female agency produces better outcomes, institu- participation rates rise to those of men.5 Nor is tions, and policy choices. the problem limited to rich countries. In devel- oping countries and regions with rapidly aging Misallocating women’s skills and talent demographic structures, like China or Eastern comes at a large (and rising) economic cost Europe, encouraging women to enter and re- Gender equality can have large impacts on pro- main in the labor force can dampen the impact ductivity, especially with women now represent- of shrinking working-age populations. ing larger shares of the world’s workforce and university graduates. For countries to be per- Women’s endowments, opportunities, and forming at their potential, the skills and talents agency shape those of the next generation of these women should be applied to activities Women’s economic empowerment and greater that make the best use of those abilities. But this control over resources also increase investments is not always the case. Women’s labor is too of- in children’s health, education, and nutrition, ten underused or misallocated—because they boosting future economic growth. Evidence face discrimination in markets or societal insti- from a range of countries (Bangladesh, Brazil, tutions that prevents them from having access Côte d’Ivoire, Mexico, South Africa, and the to productive inputs and credit, entering certain United Kingdom, among others) shows that occupations, and earning the same incomes as increasing the share of household income that men. The consequence: economic losses. women control, either through their own earn- 48 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 ings or cash transfers, changes spending in ways the provision of public goods (both female- that bene�t children.6 preferred goods such as water and sanitation, Improvements in women’s health and edu- and male-preferred goods such as irrigation cation also bene�t the next generation. Better and schools). It also reduced corruption and nutritional status and higher education levels of increased reported crimes against women along mothers are associated with better child health with arrests for those crimes.11 outcomes—from immunization rates to nutri- Several studies have also examined the tion to child mortality.7 And mothers’ school- relationship between gender equality and ing is positively linked to children’s educational economic growth at an aggregate level us- attainment across a broad set of countries. In ing cross-country data. Because the links be- Pakistan, children whose mothers have even a tween growth and gender equality go in both single year of education spend one extra hour directions, the results of this work are more studying at home every day and receive higher dif�cult to interpret than those of microeco- test scores.8 nomic studies (box 0.1). But combined, the Women’s lack of agency—evident in domes- two strands of research provide considerable tic violence—has consequences for their chil- evidence that gender equality and growth are dren’s cognitive behavior and health as adults. correlated and that gender equality matters Medical research from developed countries has for many aspects of growth. Even so, as we ar- established a link between exposure to domes- gue in the box, more careful work is needed, tic violence in childhood and health problems especially microeconomic analysis that can in adulthood. Numerous studies also document establish causal relationships between gender how witnessing violence between one’s parents equality and growth and, in so doing, high- as a child increases the likelihood that women light the channels for policy to reduce gender experience violence from their own partners as gaps and increase economic growth. adults, and that men perpetrate violence against their partners.9 WHAT DOES THIS REPORT DO? Increasing women’s individual and collective agency produces better This World Development Report focuses on the outcomes, institutions, and policy choices economics of gender equality and development. Across countries and cultures, men and women It uses an economic lens to understand what un- differ in agency—that is, their ability to make derlies and drives differences between men and choices that lead to desired outcomes—with women in key determinants of welfare—in hu- women usually at a disadvantage. When women man capital endowments such as education and and men do not have equal chances to be socially health, in access to economic opportunities and and politically active—and to influence laws, productive resources, and in the ability to make politics, and policy making—institutions and choices and take action, or agency. It uses the policies are more likely to systematically favor same approach to explore which policy inter- the interests of those with more influence. So, ventions and broader societal action can reduce the institutional constraints and market failures these gender differences and improve outcomes. that feed gender inequalities are less likely to be The Report does not limit itself to economic addressed and corrected, perpetuating gender outcomes, however—indeed, it devotes roughly inequality over generations. equal attention to human capital endowments, Women’s collective agency can be trans- to economic opportunities, and to agency, sig- formative for society as a whole. Empowering naling that all three are important and interre- women as political and social actors can change lated in determining welfare. Nor does it ignore policy choices and make institutions more rep- the central role of social and political institu- resentative of a range of voices. Female suffrage tions, whether formal or informal, in determin- in the United States led policy makers to turn ing gender outcomes. their attention to child and maternal health In examining gender gaps and their pat- and helped lower infant mortality by 8–15 per- terns across countries and over time, we adopt cent.10 In India, giving power to women at the a strongly empirical approach, preferring rigor- local level (through political quotas) increased ous and evidence-based analysis, and highlight- A guide to the Report 49 B OX 0.1 Problems with estimating the effect of gender equality on growth One thing is clear. Income and gender equality are positively corre- delivery that accompanies economic development can increase lated. Box figure 0.1 shows this correlation for one measure, the gender equality, as in education (chapter 3). World Economic Forum’s Index of Economic Participation and 2. Other things are going on that could drive changes in both gender Opportunity, which measures male-female differences in labor force equality and growth. Consider investments to improve health. participation, wages, income, political participation, and number of Studies of deworming and nutrition interventions show that technical workers. they benefit girls more than boys. Each intervention improves B OX F I G U R E 0. 1 GDP per capita and gender equality are positively gender equality and human capital, each of which is indepen- correlated dently important for growth. Now consider institutional change: institutions can boost growth, and some kinds of institutions can 12 improve gender equality. Suppose that a country is improving its legal system by expanding the reach and efficiency of the courts per capita GDP in 2009 (in logs) and police. That expansion will contribute to growth (as con- 10 tracts are better enforced) and could contribute to gender equal- ity (by making it easier for women to use justice systems). In the data, we observe a correlation between gender equality and 8 growth, but this relationship is not causal—the underlying cause is a change in institutions. So the correlation between gender quality and growth may actually be capturing the relationship 6 with a third factor that is causing changes in both gender equal- ity and growth. 4 3. The relationship is not robust. Empirical work shows that the cor- 0.2 0.4 0.6 0.8 1 relation between gender equality and growth is quite sensitive to time periods and countries. Work on the relationship between Index of Economic Participation and Opportunity female education and growth, for example, shows that results are Sources: World Development Indicators and World Economic Forum 2010. quite sensitive to how the analysis is done. Indeed, changes in how this relationship is measured can cause the effect of female What is not clear is whether this correlation is capturing the education on growth to change from negative to positive. effect of growth on gender equality or the effect of gender equal- In sum, the relationship between gender equality and growth is ity on growth. In reality, it is probably capturing some of both, but not only complex—it clearly goes in both directions. Broad-brush, we cannot tell which relationship matters more from cross-country cross-country studies cannot tell the magnitude of this relationship, data.a nor can they provide significant insights into what drives this rela- 1. There are good reasons for the relationship to go in both directions. tionship. Careful microeconomic work (some exists, but more is First, with changes in technology (changing with economic needed) can provide more definitive evidence—both on the impor- development), the relative return to manual skills has declined as tance of gender equality for growth and on where policy interven- that for cognitive skills has increased. Second, the better service tions are needed. Sources: Alesina, Giuliano, and Nunn forthcoming; Kremer, Miguel, and Thornton 2004; Maluccio and others 2009; Munshi and Rosenzweig 2005; Qian 2008. a. For a more detailed discussion of these reasons, see Bandiera and Natraj 2011 and Cuberes and Teignier 2011. ing causality where feasible. For this, we draw on ing, and other aspects of well-being (see box 3 in a large and growing body of quantitative gen- the overview for details).12 der research, complemented by new analysis— A global report like this one cannot attempt particularly on time use; domestic violence; to provide in-depth analysis of speci�c country mortality risks; and earnings and productivity circumstances. Nor can it cover all relevant di- in labor markets, agriculture, and entrepreneur- mensions of gender equality. Instead, the Report ship. We also draw on new qualitative �eld re- proposes a conceptual framework to explain search, which involved more than 4,000 men gender inequality and recommend public poli- and women in 98 communities from 19 de- cies, which can be adapted as necessary to spe- veloping countries (map 0.1), exploring how ci�c countries, issues, and sectors. Building on gender affects their daily lives as well as their as- earlier Bank work on gender and development pirations, education, job choices, decision mak- (most notably on Engendering Development), 50 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 M A P 0.1 Economies where qualitative assessments were conducted Poland Moldova Serbia Afghanistan West Bank and Gaza Bhutan Dominican Vietnam Republic Burkina Faso Sudan (North Yemen, India Sudan) (Orissa, Rep. Andhra Liberia Pradesh) Papua New Guinea Tanzania Indonesia (Jakarta, Banten, Peru Sumatera Barat) Fiji South Africa (Kwazulu-Natal) locations where WDR qualitative assessments were carried out Source: WDR 2012 team. the framework posits that gender outcomes can NAVIGATING THIS REPORT: be understood through the responses of house- A ROADMAP holds to the functioning and structure of mar- kets and institutions, both formal and informal. The Report has nine chapters in three parts. The Report then illustrates the use of this frame- Part 1—Taking stock of gender equality—presents work by focusing on aspects of gender equality the facts that will then provide the foundation where there has been most progress worldwide for the rest of the Report. It combines existing (education, life expectancy, labor force par- and new data to document changes in key di- ticipation, and the extension of legal rights to mensions of gender equality over the past quar- women) and those where there has been little or ter century and across regions and countries. Its very slow change (excess female mortality, seg- main message is that very rapid and, in some regation in economic activity, gaps in earnings, cases, unprecedented progress has been made in responsibility for house and care work, asset some dimensions of gender equality (chapter 1), ownership, and women’s agency in the private but that it has not reached all women or been and public spheres). uniform across all dimensions of gender equal- This empirical approach helps establish the ity (chapter 2). link between analysis and policy choice. It em- The contrast between the patterns and trends phasizes that the design of policies targeted at described in the �rst two chapters of the Report reducing speci�c gender gaps needs to take into prompts one to ask what explains the progress account what happens in the household, as well or lack of it. Part 2—What has driven progress? as the functioning and structure of markets What impedes it?—constitutes the analytical core and formal and informal institutions—and the of the Report. It presents the conceptual frame- interactions between all of them. Through the work and uses it to examine the factors that have analysis of alternative policies, it shows that, fostered change and the constraints that have when these aspects are not considered, the in- slowed progress. The analysis focuses on gender tended results of policy interventions can be differences in education and health (chapter 3), muted or even contrary to what is expected. agency (chapter 4), and access to economic op- A guide to the Report 51 portunities (chapter 5)—discussing the roles 10. Miller 2008. of economic growth, households, markets, and 11. Beaman and others, forthcoming; Chattopadhyay institutions in determining outcomes in these and Duflo 2004; Iyer and others 2010. three spheres. Part 2 concludes with a discus- 12. World Bank 2011. sion of the impact of globalization on gender inequality, paying attention to the opportuni- ties and challenges created by new economic REFERENCES and social trends (chapter 6). The analysis in The word processed describes informally reproduced these four chapters leads to the identi�cation of works that may not be commonly available through four priority areas for action: reducing gender libraries. gaps in human capital endowments, promoting higher access to economic opportunities among Alesina, Alberto, Paola Giuliano, and Nathan Nunn F. women, closing gender gaps in household and 2011. “Fertility and the Plough.� American Eco- societal voice, and limiting the intergenerational nomic Review Papers and Proceedings 101 (3): reproduction of gender inequality. 499–503. Part 3—The role and potential for public Allendorf, Keera. 2007. “Do Women’s Land Rights action—presents policy recommendations, ex- Promote Empowerment and Child Health in amines the political economy of reforms for Nepal?� World Development 35 (11): 1975–88. Andrabi, Tahir, Jishnu Das, and Asim Ijaz Khwaja. gender equality, and proposes a global agenda 2008. “A Dime a Day: The Possibilities and Lim- for action. The discussion starts with a detailed its of Private Schooling in Pakistan.� Comparative description of policy options addressing the Education Review 52 (3): 329–55. four priority areas, complemented with con- Attanasio, Orazio, and Valérie Lechene. 2002. “Tests crete illustrations of successful interventions in of Income Pooling in Household Decisions.� Re- different contexts (chapter 7). An examination view of Economic Dynamics 5 (4): 720–48. of the political economy of gender reforms fol- Bandiera, Oriana, and Ashwini Natraj. 2011. “Does lows, with an emphasis on the issues that dis- Gender Inequality Hinder Growth? The Evidence tinguish reform in this area from other types and Its Policy Implications.� Background paper of redistributive or equality-enhancing reforms for the WDR 2012. (chapter 8). Global action on gender equality Beaman, Lori, Esther Duflo, Rohini Pande, and Petia should focus on complementing country ef- Topalova. Forthcoming. “Political Reservation and Substantive Representation: Evidence from Indian forts on the four priority areas identi�ed in the Village Councils.� In India Policy Forum, 2010, ed. Report (chapter 9). Suman Bery, Barry Bosworth, and Arvind Pana- gariya. Brookings Institution Press and The Na- NOTES tional Council of Applied Economics Research, Washington, DC, and New Delhi. 1. FAO, IFAD, and ILO 2010. Chattopadhyay, Raghabendra, and Esther Duflo. 2. Cuberes and Teignier Baqué 2011; Hurst and oth- 2004. “Women as Policy Makers: Evidence from a ers 2011. Randomized Policy Experiment in India.� Econo- 3. Do, Levchenko, and Raddatz 2011. metrica 72 (5): 1409–43. 4. Ibid. Cuberes, David, and Marc Teignier Baqué. 2011. 5. McKinsey & Company Inc. 2007. “Gender Inequality and Economic Growth.� 6. Haddad, Hoddinott, and Alderman 1997; Katz Background paper for the WDR 2012. and Chamorro 2003; Duflo 2003; Thomas 1990; Do, Quy-Toan, Andrei Levchenko, and Claudio Hoddinott and Haddad 1995; Lundberg, Pol- Raddatz. 2011. “Engendering Trade.� Background lak, and Wales 1997; Quisumbing and Maluccio paper for the WDR 2012. 2000; Attanasio and Lechene 2002; Rubalcava, Doss, Cheryl R. 2006. “The Effects of Intrahousehold Teruel, and Thomas 2009; Doss 2006; Schady and Property Ownership on Expenditure Patterns Rosero 2008. in Ghana.� Journal of African Economies 15 (1): 7. Thomas, Strauss, and Henriques 1990; Allendorf 149–80. 2007. Duflo, Esther. 2003. “Grandmothers and Grand- 8. Andrabi, Das, and Khwaja 2008; Dumas and daughters: Old-Age Pensions and Intrahousehold Lambert 2010. Allocation in South Africa.� World Bank Economic 9. Kishor and Johnson 2004; Jeyaseelan and others Review 17 (1): 1–25. 2007; Hindin, Kishor, and Ansara 2008; Koenig Dumas, Christelle, and Sylvie Lambert. 2011. “Edu- and others 2006; Martin and others 2002. cational Achievement and Socio-Economic Back- 52 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 ground: Causality and Mechanisms in Senegal.� Child Bene�t.� Journal of Human Resources 32 (3): Journal of African Economies 20 (1): 1–26. 463–80. FAO (Food and Agriculture Organization of the Maluccio, John A., John Hoddinott, Jere R. Behrman, United Nations), IFAD (International Fund for Reynaldo Martorell, Agnes R. Quisumbing, and Agricultural Development), and ILO (Interna- Aryeh D. Stein. 2009. “The Impact of Improving tional Labour Of�ce). 2010. “Gender Dimensions Nutrition during Early Childhood on Education of Agricultural and Rural Employment: Differ- among Guatemalan Adults.� Economic Journal 119 entiated Pathways Out Of Poverty. Status, Trends (537): 734–63. and Gaps.� FAO, IFAD, and ILO, Rome. Martin, Sandra L., Kathryn E. Moracco, Julian Garro, Haddad, Lawrence, John Hoddinott, and Harold Al- Amy Ong Tsui, Lawrence L. Kupper, Jennifer L. derman. 1997. Intrahousehold Resource Allocation Chase, and Jacquelyn C. Campbell. 2002. “Domes- in Developing Countries: Models, Methods, and tic Violence across Generations: Findings from Policy. Baltimore, MD: Johns Hopkins University. Northern India.� International Journal of Epidemi- Hindin, Michelle J., Sunita Kishor, and Donna L. ology 31 (3): 560–72. Ansara. 2008. “Intimate Partner Violence among McKinsey & Company Inc. 2007. “Women Mat- Couples in 10 DHS Countries: Predictors and ter. Gender Diversity, A Corporate Performance Health Outcomes.� DHS Analytical Studies 18, Driver.� McKinsey & Company Inc., London. U.S. Agency for International Development, Miller, Grant. 2008. “Women’s Suffrage, Political Washington, DC. Responsiveness, and Child Survival in American Hoddinott, John, and Lawrence Haddad. 1995. “Does History.� Quarterly Journal of Economics 123 (3): Female Income Share Influence Household Ex- 1287–327. penditures? Evidence from Côte D’Ivoire.� Oxford Munshi, Kaivan, and Mark Rosenzweig. 2006. “Tra- Bulletin of Economics and Statistics 57 (1): 77–96. ditional Institutions Meet the Modern World: Hurst, Erik, Chang-Tai Hsieh, Charles Jones, and Pe- Caste, Gender and Schooling Choice in a Global- izing Economy.� American Economic Review 96 ter Klenow. 2011. “The Allocation of Talent and (4): 1225–52. Economic Growth.� Chicago Booth, Chicago, IL. Qian, Nancy. 2008. “Missing Women and The Price of Processed. Tea in China: The Effect of Sex-Speci�c Earnings Iyer, Lakshmi, Anandi Mani, Prachi Mishra, and Pe- on Sex Imbalance.� Quarterly Journal of Economics tia Topalova. 2010. “Political Representation and 123 (3): 1251–85. Crime: Evidence from India’s Panchayati Raj.� Quisumbing, Agnes R., and John A. Maluccio. 2000. International Monetary Fund, Washington, DC. “Intrahousehold Allocation and Gender Relations: Processed. New Empirical Evidence from Four Developing Jeyaseelan, L., Shuba Kumar, Nithya Neelakantan, Countries.� Discussion Paper 84, Food Consump- Abraham Peedicayil, Rajamohanam Pillai, and tion and Nutrition Division, International Food Nata Duvvury. 2007. “Physical Spousal Violence Policy Researcy Institute, Washington, DC. against Women in India: Some Risk Factors.� Jour- Rubalcava, Luis, Graciela Teruel, and Duncan Thomas. nal of Biosocial Science 39 (5): 657–70. 2009. “Investments, Time Preferences, and Public Katz, Elizabeth G., and Juan Sebastian Chamorro. Transfers Paid to Women.� Economic Development 2003. “Gender, Land Rights, and the Household and Cultural Change 57 (3): 507–38. Economy in Rural Nicaragua and Honduras.� Schady, Norbert, and José Rosero. 2008. “Are Cash Paper presented at the Annual Conference of the Transfers Made to Women Spent Like Other Latin American and Caribbean Economics Asso- Sources of Income?� Economics Letters 101 (3): ciation, Puebla, Mexico, October 9. 246–48. Kishor, Sunita, and Kiersten Johnson. 2004. Pro�l- Thomas, Duncan. 1990. “Intra-Household Resource ing Domestic Violence—A Multi-Country Study. Allocation: An Inferential Approach.� Journal of Calverton, MD: ORC Macro. Human Resources 25 (4): 635–64. Koenig, Michael A., Rob Stephenson, Saifuddin Thomas, Duncan, John Strauss, and Maria-Helena Ahmed, Shireen J. Jejeebhoy, and Jacquelyn Camp- Henriques. 1990. “Child Survival, Height for Age, bell. 2006. “Individual and Contextual Determi- and Household Characteristics in Brazil.� Journal nants of Domestic Violence in Northern India.� of Development Economics 33 (2): 197–234. American Journal of Public Health 96 (1): 132–38. World Bank. 2011. “De�ning Gender in the 21st Kremer, Michael, Edward Miguel, and Rebecca Century: Talking with Women and Men around Thornton. 2009. “Incentives to Learn.� Review of the World, A Multi-Country Qualitative Study Economics and Statistics 91 (3): 437–56. of Gender and Economic Choice.� World Bank, Lundberg, Shelly J., Robert A. Pollak, and Terence J. Washington, DC. Wales. 1997. “Do Husbands and Wives Pool Their World Economic Forum. 2010. “Global Gender Gap Resources? Evidence from the United Kingdom Report 2010.� World Economic Forum, Japan. PART I Taking stock of gender equality W omen’s lives have improved greatly over the past de- cades. Enjoying ever higher education, women have greater control over their life choices. They use those choices to participate more in the labor force; have fewer children; diversify their time beyond housework and child care; and shape potential contribution to the well-being of the household in relation to other household members—and on views of gen- der roles and women’s preferences and needs. Agency is the process by which women and men make choices and translate them into desired outcomes. It has many their communities, economies, and societies. And the pace of dimensions. This Report focuses on �ve outcomes closely as- change for many women in the developing world has accelerated. sociated with women’s ability to make choices: control over But progress has not been uniform. A vast chasm persists to- resources, decision making over family formation, control day in more than one way. There are women in every region of over movement, freedom from violence, and the ability to the world for whom life has not changed much. They still battle have a voice in society. many of the issues that women before them had to face. Some In these three domains, shortfalls of choice are reflected in girls still cannot go to school on a par with boys. They may not shortfalls of welfare. These three domains matter in and of inherit assets from their parents. And they have lower legal and themselves. But they are also closely interlinked. Agency al- social status. Even in Sweden and the United States, change is still lows women to build their endowments. Endowments shape hard to come by in many facets of life and the economy. Women access to economic opportunities and incomes. The ability of are paid less and are still employed disproportionately in nurs- women to access economic opportunities and earn income ing and teaching, for example, while men dominate engineering can influence their agency. And so on. and construction. Only 28 chief executive of�cers of the Fortune Chapter 1 documents the unprecedented gains women 1000 companies are women. have made in rights, in human capital endowments, and in Why so much progress in some areas and so little in others? access to economic opportunities. Most countries today have Part 1 provides a factual foundation for the rest of the Report by explicit guarantees in their constitutions for the equality of all assembling existing and new data to map out patterns and trends citizens and for nondiscrimination between men and women. in gender equality over time and across regions and countries. It Not only have things changed for the better; changes are also takes stock of the changes in gender outcomes across the world happening much faster than when today’s rich countries were in recent decades. Many of these changes have occurred because much poorer. of the wave of global prosperity that has swept across much of Chapter 2, by contrast, shows that things have not changed the developing world. Yet this prosperity has not worked for for all women or in all aspects of gender equality. First, for poor many women around the world, and in some respects not at all. women and for women in poor places, sizable gender gaps re- The focus here is on three key domains of gender equality: main, even in education enrollments and fertility, where global the accumulation of endowments, the use of those endowments progress has been great. For the wealthiest people across coun- to take up economic opportunities and generate incomes, and the tries, there is little gender disadvantage, but it is large for severely application of those endowments to take actions (agency) affect- disadvantaged populations at the bottom of the income distribu- ing individual and household well-being. tion. Ethnicity, distance, disability, or sexual orientation, among Endowments encompass education, health, land, and other other factors, further compound gender inequality. Second, in assets such as �nancial resources that women and men accu- some domains of gender equality there has been very little—or mulate during their lifetimes. Many, such as basic education, are very slow—change, even as countries get richer. These “sticky� amassed at early stages in life, reflecting mostly the decisions of domains include excess female mortality in key periods of the parents. Others, such as material assets, are acquired later through life cycle and occupational differences in the economic sphere. such mechanisms as productive labor and inheritance rights. In many areas of women’s agency, including political voice and Access to economic opportunities determines how endow- representation, differences between men and women remain en- ments and time generate income and consumption—key dimen- trenched, even in very rich countries and despite nearly a century sions of well-being. Decisions about time allocations between of women’s activism. Third, systemic shocks, such as droughts home and workplace, productive activity, and leisure take place or economic downturns, adversely affect males and females, in the household. They depend on the value placed on women’s and their precise impacts depend on the context and the shock. 55 CHAPTER 1 A wave of progress D espite the hardships many women endure in their daily lives, things have changed for the better—and at a speed that would not have been expected even two decades ago. In four major “ Let our daughters go to school and let them get good jobs. The moment they will be independent from men in thinking and earning, then they will areas—women’s rights, education, health, and � labor force outcomes—the gains in the second have very good lives. half of the 20th century were large and fast in Adult woman, Tanzania many parts of the world. Improvements that took 100 years in wealthier countries took just 40 years in some low- and middle-income coun- dramatically. For younger cohorts, the gender tries. Change has also been accelerating, with gap in primary education enrollments has prac- gender equality gains in every decade building tically disappeared, and the gains in second- on gains from the decade before. ary and higher education have been enormous. Women are living longer and healthier lives in much of the world, in part because lower fer- TIMES ARE CHANGING? tility has reduced their risk in childbirth. And they are participating more than ever in market This chapter traces where progress has been work. Economic growth has driven much of the made on gender equality and how. It starts with progress, through higher household incomes, the evolution of women’s rights and the �ght better service delivery, and new labor market op- for equality under the law. Equality of rights portunities for women. But it has not been the matters because a lack of rights can constrain only factor—the association between economic the choices available to women in many aspects growth and better outcomes for women has been of life. Achieving them in today’s high-income neither automatic nor uniform across countries. countries took considerable time. In contrast, Changes in one domain of gender equality gains under the law have occurred much faster have fostered change in others, influencing the in the developing countries, aided by a rising next generation, reinforcing the whole process. global consensus supporting formal rights and For example, the expansion of economic op- guarantees of equality for women. portunities for women in service industries in In tandem with these gains in formal rights, Bangladesh and India has boosted school en- low- and middle-income countries have seen rollments for girls, which feeds into higher labor unprecedented gains in outcomes for women, force participation and better educational out- both in absolute terms and relative to men. More comes for the next generation.1 women are literate and educated than ever be- This is not to say that all problems have been fore, and the education gap with men has shrunk solved or that progress was easy. Indeed, chapter A wave of progress 57 2 looks at countries and population groups with with progress monitored independently by the continuing gender disadvantages as well as at Committee on the Elimination of Discrimina- the many facets of life where gender imbalances tion against Women. remain serious, even crippling. Delineating the Despite being known in the 1980s as the areas of change provides a foundation for un- “Cinderella treaty� for its vagueness of language derstanding the constraints to gender equal- and weak monitoring and sanctions, CEDAW ity—especially where they remain pervasive and has promoted legislative and administrative deep. And this understanding can help in setting change.2 In 1998, it influenced Turkey’s domes- priorities for policy and public action. tic violence act (Law No. 4320 Family Protection Law). Turkey’s Constitutional Court also an- nulled requirements that husbands give permis- RISING GLOBAL CONSENSUS sion for a wife’s professional activities, making extensive references to CEDAW.3 Australia’s Sex FOR WOMEN’S RIGHTS Discrimination Act 1984 draws on CEDAW to The past three decades have seen great progress prohibit discrimination in public life on the ba- in securing women’s formal rights and consti- sis of sex, marital status, pregnancy, or potential tutional guarantees of equality. The Convention pregnancy. on the Elimination of All Forms of Discrimi- In 2003, the African Union adopted the Pro- nation against Women (CEDAW), which the tocol to the African Charter on Human and Peo- ples’ Rights on the Rights of Women in Africa. United Nations General Assembly adopted in Better known as the Maputo Protocol, it asserts 1979, established a comprehensive framework women’s rights to take part in the political pro- for the advancement of women. Ten years later, cess, to enjoy social and political equality with almost 100 nations across all continents had rat- men, and to control their reproductive health. i�ed this international bill of rights for women. Article 5 refers to the “elimination of harmful And today, the number of signatories has almost practices,� including ending polygamous mar- doubled, to 187 countries. riage and female genital cutting. Of 53 African More than 30 years in the making, CEDAW countries, 46 signed the protocol, and by Febru- is the most widely supported international hu- ary 2011, 30 countries had rati�ed it.4 man rights treaty and the primary international Under the auspices of the Organization of vehicle for monitoring and advocating gender American States, all Latin American coun- equality. De�ning what constitutes discrimina- tries signed in 1994 the Belém do Pará Inter- tion against women and setting an agenda for na- American Convention on the Prevention, Pun- tional action, CEDAW is particularly concerned ishment and Eradication of Violence against with civil rights, the legal status of women, re- Women. Since then, 28 nations have enacted productive roles and rights, and the impact of laws with sanctions against domestic abuse. cultural factors on gender relations and on bar- These different international legal frame- riers to the advancement of women. It is the only works reflect the rising global consensus on human rights treaty to explicitly address deci- equal rights for men and women—a consensus sions about family planning and family forma- that did not emerge overnight but rather evolved tion. Countries ratifying CEDAW are required to from a long, slow struggle for equal rights for ensure that domestic legislation complies with it, women that started in the advanced economies as early as the 18th century and continued in de- veloping countries in the second half of the 20th century, reinforced by the emphasis on gender “ I know that [women] have many rights: I can remember the right to education, the prohibition of the excision, and the prohibition to forced equality in the Millennium Development Goals (MDGs) (box 1.1). Equal rights in the advanced economies—A long time coming marriage. � Adult woman, Burkina Faso Women’s circumstances in the 18th century were very different than they are today. In 1789, the French revolution asserted that men are “born 58 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 The march toward equal property and suf- B OX 1.1 Gender and the Millennium Development Goals frage rights has been slow and long. Only in 1857 did the British Parliament pass the Matri- The 2010 MDG Summit concluded with a global action plan to achieve the monial Causes Act, allowing married women to eight MDGs by 2015. It also adopted a resolution calling for action to ensure inherit property and take court action on their gender parity in education and health, economic opportunities, and decision own behalf. And not until 1882 did the Married making at all levels through gender mainstreaming in the formulation and Women’s Property Act recognize a husband and implementation of development policies. The resolution and the action plan a wife as two separate legal entities, conferring reflect the belief of the international development community that gender to wives the right to buy, own, and sell property equality and women’s empowerment are development objectives in their own right (MDG 3), as well as critical channels for the achievement of the other separately. Suffrage was not universal until 1928, MDGs. Gender equality and women’s empowerment help to promote universal when, as a result of the Representation of the primary education (MDG 2), reduce under-five mortality (MDG 4), improve People Act, women over age 21 received the vote maternal mortality (MDG 5), and reduce the likelihood of contracting HIV/AIDS on equal terms as men. The story is similar in (MDG 6). Scandinavia: Norway, for example, provided full The 2010 resolution also stresses that achieving the MDGs will require coor- economic rights to women in 1888 and suffrage dinated interventions that target women and other vulnerable groups across rights in 1913. sectors: In the United States, New York was the �rst • Taking action to improve the number and active participation of women state to pass, in 1848, a Married Women’s Prop- in all political and economic decision-making processes, including invest- erty Act. Wives’ rights to earnings and property ing in women’s leadership in local decision-making structures and creating an even playing field for men and women in political and government gradually spread to other states over the follow- institutions ing half century. Political voice was longer in • Expanding access to financial services for the poor, especially women coming. A proposed constitutional amendment • Investing in infrastructure and labor-saving technologies, especially in rural guaranteeing women’s right to vote was intro- areas, that benefit women and girls by reducing their domestic burdens duced in the U.S. Senate in 1878, but it did not • Promoting and protecting women’s equal access to housing, property, and receive a full vote until 1887, only to be voted land, including rights to inheritance. down. Three more decades elapsed before the 19th amendment to the constitution guarantee- Source: WDR 2012 team. ing universal suffrage was rati�ed in 1920. The struggle against discrimination in other domains, such as labor and family law, picked and remain free and equal in rights� universally, up momentum in the second half of the 20th but the Declaration of the Rights of Man and of century. the Citizen did not include women, and a year In the United States, until the passage of Title later, the National Assembly chose not to extend VII of the Civil Rights Act of 1964, women could civil and political rights to women. The legal legally be passed over for promotions in the system in the British colonies, based on English workplace. Married women needed the consent common law, is another case. As Sir William of their husbands to obtain a loan. And marital Blackstone summarized in his Commentaries on rape was not recognized as a criminal act.5 Until the Laws of England in 1765: the 1980s, female flight attendants were required to be single when they were hired and could be By marriage, the husband and wife are �red if they married. one person in law: that is, the very being In Germany in the early 1950s, women or legal existence of the woman is sus- could be dismissed from the civil service when pended during the marriage, or at least is they married. And through 1977, they of�cially incorporated and consolidated into that needed their husbands’ permission to work. of the husband; under whose wing, pro- Until reuni�cation with East Germany in 1990, tection, and cover, she performs every- children of single mothers were assigned a legal thing; and is therefore called in our guardian.6 law-french a femme-couvert. For this rea- Japan’s Equal Employment Opportunity Act son, a man cannot grant anything to his of 1985 obliged employers merely to endeavor wife, or enter into covenant with her: for to treat men and women equally during job the grant would be to suppose her separate recruitment, assignment, and promotion. The existence. mandate for equal treatment came about in A wave of progress 59 1997. The �rst domestic violence law was passed natory legal provisions. In 2005, the Kenyan in 2001. Court of Appeal held that there was no reason- able basis for drawing a distinction between Progress has been faster in low- and sons and daughters in determining inheritance. middle-income countries In 2001, the Tanzanian High Court held that a Progress has been most notable for political widow is entitled to administer the estate on be- rights, tied to a change in the concept of citi- half of her children. In both cases, principles of zenship. National franchise movements gave equality and nondiscrimination prevailed. shape to a more inclusive paradigm of the nation-state in the �rst half of the 20th cen- tury. Until then, citizenship had long been con- BETTER OUTCOMES FOR strued as “male.� Extending suffrage in already WOMEN IN MANY DOMAINS established nation-states involved local social movements and social networks rede�ning citi- The march for women’s rights has gone hand in zenship only after a lengthy renegotiation of do- hand with better outcomes for many women— mestic political power. In contrast, new nations both in absolute terms and relative to men. Dur- emerged into a “new world order.� National and ing the past quarter century, sustained growth in international organizations embraced a gender- many countries has reduced disparities on some neutral model of citizenship, with women fully dimensions of gender equality. And the pace of accepted as persons capable of autonomous change in these outcomes has been much faster decisions.7 Only three countries that became in today’s low- and middle-income countries independent in the 1900s (Austria, Ireland, and than it was in high-income countries. That can be Libya) extended suffrage to men before women. seen in indicators as varied as fertility, female ed- But Switzerland did not break with tradition ucation and literacy, and female labor force par- and extend the franchise to women until 1971. ticipation. In most countries where broad-based Among the latest countries to give women the income growth has combined with better insti- right to vote, Bhutan changed the practice of tutions for service delivery and more economic casting one vote per household and adopted opportunities for women, the improvements in women’s full suffrage in 2008. Today, only Saudi these indicators have been dramatic—and in Arabia restricts the franchise to men and re- some cases at rates never before witnessed. moving this restriction for municipal elections Moreover, they occurred along some dimen- is under consideration. sions even in the face of social turmoil or sig- Similar progress has been made in women’s ni�cant institutional challenges. Consider Ban- rights beyond full suffrage. In the Philippines, gladesh, Colombia, and the Islamic Republic of sweeping legislative changes in the 1980s and Iran. 1990s recognized gender equality across a wide Starting from a low base, the Bangladesh array of domains. The 1987 constitution rein- economy has almost tripled since 1980. The forced earlier constitutions by giving added Bangladesh constitution, adopted in 1972, guar- emphasis to the notion of gender equality. The anteed equal rights to all citizens, regardless of Comprehensive Agrarian Reform Law of 1988 gender, religion, or other social divisions, and assured equal rights to ownership of land. And reserved 15 parliamentary seats for women, a 1989 act amended the Labor Code to protect later increased to 30. In 1975, the government women from discrimination in hiring and pay. reserved 10 percent of public jobs for women Similarly, in 2004, Morocco overhauled its fam- and created a special Ministry of Women’s Af- ily code to promote greater equality between fairs. Outcomes for women also improved dra- women and men in multiple spheres. matically on various fronts: The rati�cation of CEDAW and other inter- • From 1971 to 2009, the total fertility rate— national treaties established a comprehensive the number of children a woman is expected framework to promote equality for women. to have through her reproductive years—de- These treaties spurred further progress toward clined from 6.9 children to 2.3. securing formal rights in other domains of women’s lives, in large part by facilitating new • Between 1991 and 2005, the number of girls legislation or promoting the repeal of discrimi- in school increased from 1.1 million to 4 mil- 60 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 lion. Female enrollment climbed from 33 per- The Islamic Republic of Iran’s economy has cent of the total to 56 percent, with somewhat almost doubled since 1980. And human devel- smaller increases among girls from the poor- opment outcomes among Iranian women have est two quintiles. consistently improved along some key dimen- sions in the aftermath of the Islamic revolution: • The labor force participation of young wom- en (ages 20–24) increased almost two and a • From 1979 to 2009, the Islamic Republic of half times over 1995–2000. Although overall Iran saw the world’s fastest decline in fertil- labor force participation remains low, the ity—from 6.9 children to 1.8 (below replace- expansion of employment opportunities for ment). young women (linked to the growth of the • The female-to-male ratio in primary school garment industry, health services, and social is the world’s highest, with 1.2 girls enrolled work) has increased girls’ school enrollment for every boy. The number of women in sec- and lifted social restrictions on female mobil- ondary school as a percentage of the eligible ity, allowing for a visible feminization of pub- age group more than doubled from 30 per- lic spaces (chapter 6).8 cent to 81 percent, and in 2009, more than The economy of Colombia, an upper-middle- half of all Iranian university students, 68 per- income country in Latin America, has expanded cent of the students in science, and 28 per- over one and a half times since 1980. Following cent in engineering were women. upheavals and a recession in 1999, it stabilized • Women make up 30 percent of the Iranian after 2002. Colombia has also long been beset by labor force today, with the percentage of eco- violence and the illegal drug trade. One remark- nomically active women having increased able feature is the improvement in women’s sta- from 20 percent in 1986 to 31 percent in tus in the past 25 years: 2008. • From 1986 to 2005, the total fertility rate Each of these three societies has faced some dropped from 3.2 children to 2.4. circumstances commonly viewed as constrain- • Women reversed the education gap and now ing gender equality. Yet in all of them, income have higher completion rates than men for growth, better institutions for service delivery, primary, secondary, and even tertiary educa- and new market opportunities for women have tion. The last is particularly striking given contributed to greater gender equality in health, that in 1984 almost twice as many men rela- education, and labor market outcomes even as tive to women were college educated. women in these countries continue to face sig- ni�cant challenges in other aspects of their lives. • In 1980, the labor force participation of Co- lombian females in the 13 largest cities was More girls in school the second lowest in the region (above only More women are literate than ever before. Be- Costa Rica), but by 2004, it was the second tween 1950 and 2010, the average schooling for highest, next to Uruguay. Remarkably, the women over age 15 in low-income countries in- largest increases were among women with creased from 1.5 years to 6.5. Compare that with children ages 0–6—women least likely to an increase from 2.6 years to 7.6 for men—and work in most countries. And the representa- with current averages in adult high-income pop- tion of women is high in managerial posi- ulations of 10.9 years for women and 11.2 years tions and in �nance—the glass ceilings noto- for men.9 Because the adult population includes riously hard to break in many advanced older people who do not increase their educa- economies. “ Child marriages have stopped. Girls are being sent to school. Even the poorest of us are sending our daughters to school. � Adult man, India A wave of progress 61 F I G U R E 1.1 Gender parity in enrollments at lower levels has been achieved in much of the world, and tertiary enrollments now favor women Primary education Secondary education Tertiary education 100 100 100 men Some African Enrollment gaps dwarf school enrollment, women, gross, % disadvantaged countries lag behind gender gaps school enrollment, girls, net, % school enrollment, girls, net, % 80 80 80 60 60 60 40 40 40 Women are more likely 20 20 20 to participate girls girls than men disadvantaged disadvantaged 0 0 0 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 school enrollment, boys, net % school enrollment, boys, net % school enrollment, men, gross % East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa South Asia Sub-Saharan Africa High-income countries Source: WDR 2012 team estimates based on World Development Indicators. Note: The 45° line in each �gure above shows gender parity in enrollments. Any point above the 45° line implies that more women are enrolled relative to men. tion, these rising averages reflect even greater boys were in the minority in a wide range of na- changes among young cohorts. In the United tions including Bangladesh, Brazil, Honduras, States, it took 40 years, from 1870 to 1910, for Lesotho, Malaysia, Mongolia, and South Africa. the share of 6–12-year-old girls in school to in- Tertiary enrollment growth is stronger for crease from 57 percent to 88 percent. Morocco women than for men across the world. The did the same in 11 years from 1997 to 2008. number of male tertiary students globally more These younger cohorts in all world regions than quadrupled, from 17.7 million to 77.8 have experienced steady and sustained improve- million between 1970 and 2008, but the num- ment toward greater gender equality in primary ber of female tertiary students rose more than education. In the past decade, female enroll- sevenfold, from 10.8 million to 80.9 million, ments have grown faster than male enrollments overtaking men. Female tertiary enrollment in the Middle East and North Africa, South Asia, rates in 2008 lagged behind in only 36 develop- and Sub-Saharan Africa. Gender parity has been ing countries of 96 with data (see �gure 1.1). reached in 117 of 173 countries with data (�g- In Tunisia, 59 percent of the 351,000 students ure 1.1). Even in regions with the largest gender enrolled in university in 2008 were women. As gaps—South Asia and Sub-Saharan Africa (par- chapter 6 shows, this increase in female enroll- ticularly West Africa)—gains have been con- ment is consistent with an increasing demand siderable. In 2008, in Sub-Saharan Africa, there for “brain� rather than “brawn� jobs in a glo- were about 91 girls for every 100 boys in primary balizing world. school, up from 85 girls in 1999; in South Asia, Although boys are more likely than girls to the ratio was 95 girls for every 100 boys. be enrolled in primary school, girls make better The patterns are similar in secondary edu- progress—lower repetition and lower dropout cation, with one notable difference. In roughly rates—than boys in all developing regions.10 one-third of developing countries (45), girls According to international standardized stu- outnumbered boys in secondary education in dent achievement tests, girls tend to outperform 2008 (see �gure 1.1). Although the female gen- boys in language skills, while boys tend to have der gap tends to be higher in poorer countries, a smaller advantage over girls in mathematics. 62 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 The 2009 Program for International Student Gender now explains very little of the re- Assessment shows that 15-year-old girls in all maining inequality in school enrollment (�gure participating countries performed better than 1.2). In a large number of countries, a decompo- boys in a reading test. sition of school enrollments suggests that wealth Economic growth has lowered the barriers is the constraining factor for most, and in only to school entry for millions of boys and girls a very limited number will a narrow focus on throughout the world and reduced gender in- gender (rather than poverty) reduce inequalities equality in schooling, particularly as countries further (chapter 3).16 move from lower to middle and higher in- comes.11 First, as countries prosper economi- Healthier lives cally, the supply of services by government or The second half of the 20th century also saw other service providers increases. Second, rising large improvements in men’s and women’s incomes erode the need for families to differen- health. Life expectancy at birth most clearly re- tiate educational investments across children, flects improvements in health in populations on the basis of gender, birth order, or any other across the world. The average number of years reason, as they face less stringent budgets.12 women could expect to live rose from 54 (51 for Because more girls than boys had been out of men) in 1960 to 71 (67 for men) in 2008. This school, overall improvements in enrollments period also saw the world’s fastest ever decline tend to reduce gender differences. Third, as in fertility—from an average of about 5 births growth opens new employment opportunities per woman in 1960 to 2.5 in 2008, lowering for women in sectors that demand a certain level the number of deaths associated with maternal of skills, such as light manufacturing or services, mortality. And bearing fewer children has given incentives for parents to invest in their daugh- women more time to invest in acquiring human ters’ education increases, because that education capital and to participate in market work. now yields a greater return. In most world regions, life expectancy for Eliminating school fees has had a similar ef- both men and women has consistently risen, fect in increasing overall enrollments and reduc- with women on average living longer than men. ing gender differences. The free primary educa- The gap between male and female life expec- tion programs across Sub-Saharan Africa have tancy, while still rising in some regions, stabilized had an overwhelming response. In their �rst in others. On average, life expectancy at birth for year, student enrollments climbed 68 percent in females in low-income countries rose from 48 Malawi and Uganda and 22 percent in Kenya.13 years in 1960 to 69 years in 2008, and for males, States have also mandated and enforced par- from 46 years to 65. Mirroring the worldwide in- ticipation in schooling through compulsory ed- crease in life expectancy, every region except Sub- ucation laws. Mass education systems expanded Saharan Africa added between 20 and 25 years quickly after World War II, and universal edu- of life between 1960 and today (�gure 1.3). And cation laws can now be found in almost all na- since 1980, every region has had a female advan- tions.14 Such laws, usually combined with large tage in life expectancy. infrastructure and human resource investments But there have been notable reversals. In to enhance service delivery, have brought more Eastern Europe and Central Asia, women’s ad- children into school throughout the world. In vantage in life expectancy increased partly be- 1997, Turkey sought to increase educational cause of a sharp increase in male mortality, opportunities for children ages 11–13 years, with the differences apparently increasing over particularly rural girls, by expanding manda- time (see �gure 1.3 for Ukraine). In some Sub- tory education from �ve years to eight. With the Saharan countries, the ravaging effects of AIDS, launch of the Basic Education Program, enroll- especially for women, are evident. Since 1990, ments jumped by 1.5 million children. Net en- gains in female life expectancy relative to men rollment rates, on the decline between 1991 and have shrunk (see �gure 1.3 for Botswana). 1997, then rose from 86 percent in 1997 to 96 Increases in female life expectancy have been percent in 2002. Gains for rural girls were par- driven in part by a signi�cant decline in the risk ticularly impressive, jumping 160 percent in the of mortality during one of the most dangerous program’s �rst year alone in the nine provinces periods in a woman’s life—the early reproductive (of 81) with the greatest gender disparity.15 years and the experience of childbirth. First, the A wave of progress 63 F I G U R E 1.2 Gender explains little of the inequality in education participation for children 12–15 years old Low inequality Moderate inequality High inequality 100 contribution to total inequality, % 80 60 40 20 0 Moldova (2005) Congo, Dem. Rep. (2007) Jordan (2009) Niger (2006) Turkey (2003) Swaziland (2006–07) Cameroon (2004) Uganda (2009–10) Congo, Rep. (Brazzaville) (2009) Albania (2008–09) Guyana (2005) Namibia (2006–07) Colombia (2005) Lesotho (2009–10) Nepal (2006) Ukraine (2007) Armenia (2005) Maldives (2009) Zambia (2007) Tanzania (2007–08) Haiti (2005–06) Vietnam (2002) Ghana (2008) Peru (2004–08) Bolivia (2008) Mozambique (2003) Madagascar (2008–09) Liberia (2009) Burkina Faso (2003) Ethiopia (2005) Sierra Leone (2009) Guinea (2005) Chad (2004) Zimbabwe (2005–06) Malawi (2004) Egypt, Arab Rep. (2008) Senegal (2008–09) Mali (2006) Côte d’Ivoire (2005) Azerbaijan (2006) Benin (2006) Dominican Republic (2007) Cambodia (2005) Kenya (2008–09) India (2005–06) Nigeria (2008) Morocco (2003–04) gender wealth other factors Source: WDR 2012 team estimates based on Demographic and Health Surveys. Note: The measure of inequality refers to the percentage of total opportunities that must be reallocated to ensure that all the circumstances groups have the same average cover- age rate. Low inequality is between 0.3 percent and 2.1 percent, moderate inequality is between 2.3 percent and 6.4 percent, and high inequality is between 6.5 percent and 26.7 percent. Results are sorted by size of gender contribution to total inequality. risk of death per birth declined. During 1990– States more than 100 years took India 40 (�gure 2008, 147 countries experienced declines in the 1.4). Similarly, in Morocco, the fertility rate fell maternal mortality ratio, 90 with a decline of 40 from 4 children per woman to 2.5 between 1992 percent or more.17 The Middle East and North and 2004. Africa had the largest decline (59 percent), fol- On various other aspects of health status and lowed by East Asia and Paci�c (56 percent) and health care, differences by sex are small. In many South Asia (53 percent). Second, the exposure to low-income countries, the proportion of chil- the risk of death was lower because of dramatic dren stunted, wasted, or underweight remains declines in fertility rates all over the world. With high, but girls are no worse off than boys. In fact, women choosing to have fewer children, the data from the Demographic and Health Surveys lifetime risk of death from maternal causes de- show that boys are at a slight disadvantage.18 In clined, even where the risk of death during each Brazil, Côte d’Ivoire, and Vietnam, men’s and birth changed little. women’s heights are increasing at almost identi- In most developing countries, fertility rates cal rates, while in Ghana women’s heights have fell sharply in a fairly short period. These de- increased more rapidly than those of men.19 In clines were much faster than earlier declines in many countries, children’s and adults’ anthro- today’s rich countries. In the United States, fer- pometric outcomes do not allow them to reach tility rates fell gradually in the 1800s through their full potential, but an individual’s sex is not 1940, increased during the baby boom, and the main culprit. North Indian states are a no- then leveled off at just above replacement. In table exception; women have grown taller at a India, fertility was high and stable through 1960 much slower rate than men, and girls’ anthro- and then sharply declined from 6 births per pometric outcomes remain worse than boys— woman to 2.3 by 2009. What took the United both in levels and in changes over time.20 64 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 F I G U R E 1.3 Women are living longer than men East Asia and Pacific Europe and Central Asia Latin America and the Caribbean 80 80 women, Ukraine 80 life expectancy at birth life expectancy at birth life expectancy at birth 70 70 70 60 60 men, Ukraine 60 50 50 50 40 40 40 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 years years years women men women men women men Middle East and North Africa South Asia Sub-Saharan Africa 80 80 80 life expectancy at birth life expectancy at birth life expectancy at birth 70 70 70 women, Botswana 60 60 60 men, Botswana 50 50 50 40 40 40 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010 years years years women men women men women men Source: WDR 2012 team estimates based on World Development Indicators. Similarly, there is little evidence of systematic F I G U R E 1 .4 What took the United gender discrimination in the use of health ser- States 100 years took vices or in health spending. Out-of-pocket India 40 and the Islamic spending on health in the 1990s was higher for Republic of Iran 10 women than for men in Brazil, the Dominican Republic, Paraguay, and Peru.21 Evidence from How fast can fertility decline? South Africa reveals the same pro-female pat- Iran, tern,22 as does that for lower income countries. Islamic Rep. In the Arab Republic of Egypt, more was spent Bangladesh per capita on outpatient services for females (68 Morocco Egyptian pounds a year) than on male health Zimbabwe care (58).23 The gender difference in amounts Colombia spent on inpatient services also favored females but much less. In Ghana, females absorbed more India of a health subsidy than males did (56 percent United States of overall health spending in 1992). Evidence from India, Indonesia, and Kenya tells a similar 0 20 40 60 80 100 number of years for the total fertility story.24 rate to fall from more than 6 children For preventive health services such as vacci- to less than 3 nation, poverty rather than gender appears to be the major constraining factor (�gure 1.5).25 As Source: http://www.gapminder.org. with education enrollments, a decomposition A wave of progress 65 FIGURE 1.5 Gender explains little of the inequality in use of preventive health services Low inequality Moderate inequality High inequality 100 contribution to total inequality, % 80 60 40 20 0 Egypt, Arab Rep. (2008) Senegal (2008–09) Bangladesh (2007) Honduras (2005–06) Maldives (2009) Bolivia (2008) Rwanda (2007) Sierra Leone (2008) Uganda (2009–10) Morocco (2003–04) Zambia (2007) Ghana (2008) Jordan (2009) Cambodia (2005) Albania (2008–09) Indonesia (2007) Malawi (2004) Philippines (2008) Swaziland (2006–07) Colomba (2005) Cameroon (2004) Namibia (2006–07) Kenya (2008–09) Peru (2004–08) Lesotho (2009–10) Modova (2005) Dominican Rep. (2007) Burkina Faso (2003) Nepal (2006) Mali (2006) Nigeria (2008) Armenia (2005) Congo, Rep. (Brazzaville) (2009) Congo, Dem. Rep. (2007) Niger (2006) Guinea (2005) Turkey (2003) Ethiopia (2005) Mozambique (2003) Benin (2006) Liberia (2009) Madagascar (2008-09) Zimbabwe (2005–06) India (2005–06) Azerbaijan (2005) Haiti (2005–06) gender wealth other factors Source: WDR 2012 team estimates based on Demographic and Health Surveys. Note: The measure of inequality refers to the percentage of total opportunities that must be reallocated to ensure that all the circumstances groups have the same average cover- age rate. Low inequality is between 1.0 percent and 3.2 percent, moderate inequality is between 3.4 percent and 5.2 percent, and high inequality is between 5.2 percent and 22.0 percent. Results are sorted by size of gender contribution to total inequality. suggests that only a handful of countries have Asia and the Paci�c, Europe and Central Asia, high inequality in measles vaccinations (other and Latin America and the Caribbean, while vaccines are even more universal) and where more than 60 percent of women remain eco- gender is a major contributing circumstance. nomically inactive in South Asia and the Mid- dle East and North Africa. More women participate in market work Labor force participation increased markedly Female labor force participation has grown since for women with more education, but declined 1960, dramatically in some regions. Expanding among women ages 15–24, who have remained economic opportunities have drawn large num- in school longer, slowing the growth in overall bers of new female workers into the market. Be- participation since 1990. tween 1980 and 2008, the global rate of female Around the world, for very poor countries, labor force participation increased from 50.2 female labor force participation is high, reflect- to 51.8 percent while the male rate fell slightly ing a large labor-intensive agricultural sector from 82.0 to 77.7 percent. So the gender gap and signi�cant numbers of poor households.27 narrowed from 32 percentage points in 1980 to In this situation, women are willing to enter 26 percentage points in 2008 (�gure 1.6).26 the labor force even at fairly low wages because Driving the convergence across countries unearned incomes are also low. As per capita are large increases in participation in coun- incomes rise, unearned income rises (through tries that started with very low rates (primarily higher male wages and earnings), and these in Latin American and the Caribbean and in higher incomes are typically associated with the Middle East and North Africa) combined women withdrawing from the labor market. with small declines in countries that started Social barriers against women entering the paid with very high rates (primarily in Eastern Eu- labor force also regain prominence, and their rope and Central Asia). Participation rates now participation rates fall. But as countries con- exceed 50 percent in Sub-Saharan Africa, East tinue to develop, further increases in women’s 66 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 estimated on the basis of data from 130 coun- F I G U R E 1.6 The gender gap in labor force participation tries around 1980 and 2008, highlights these narrowed between 1980 and 2008 two features (�gure 1.7). First, there is a clear U-shape relationship across countries at differ- 0 ent incomes for both periods. Second, the par- changes positive ticipation rate associated with each level of de- toward women velopment increases over time—the U-shaped female-male participation gap, 2008, % –20 curve moves upward as time passes. So at each level of income, more women were participating in market work in 2008 than in 1980. Changes in education and family struc- –40 ture have also driven women’s employment over time. Delays in the timing of marriage and childbearing as well as lower fertility have –60 brought more women into the labor market, ranging from as little as 7 percent in Bolivia to about 30 percent in Argentina and Colom- bia, and contributing an average of 21 percent –80 overall to the rise in female labor force par- –80 –60 –40 –20 0 ticipation across 10 Latin American countries. female-male participation gap, 1980, % An additional 42 percent of the rise in female East Asia and Pacific South Asia labor force participation in Latin America can Europe and Central Asia Sub-Saharan Africa be attributed to more education—in Panama, Latin America and the Caribbean high-income countries education accounts for as much as 81 percent. Middle East and North Africa Urbanization, household technology, and the Source: WDR 2012 team estimates based on International Labour Organization 2010. sectoral structure of the economy appear to Note: The 45° line in the �gure above shows parity in the values on the vertical and horizontal axis. matter much less in bringing women into the labor force.28 F I G U R E 1.7 Across countries, at every income level, female CHANGE BEGETS CHANGE labor force participation increased between 1980 and 2008 A notable transformation has taken place around the world with impressive gains in women’s rights, educational attainment, health female labor force participation rate, % 80 outcomes, and labor force participation. Positive 70 2008 feedback loops between gains in these different areas explain why change has been so quick in the developing world. 60 Improvements in one area (higher educa- tion) can drive changes in other areas (such as 50 lower fertility and higher labor force participa- tion). Similarly, better labor opportunities can 1980 in turn induce more investments in education 40 and in women’s health for the next generation, 4 6 8 10 12 while equal rights can underpin progress on all log, GDP per capita (constant 2000 US$) fronts. Conversely, a lack of improvement in, say, rights can hold back improvements in women’s Source: WDR 2012 team estimates based on International Labour Organization 2010. access to market work, and failures to redress gender gaps can impair a whole range of health education and wages move them back into the outcomes for women. Understanding these labor market. feedback loops is important for policy design. The relationship between economic devel- So is understanding the constraints to progress, opment and female labor force participation, and whether they are rooted in what happens in A wave of progress 67 households, in how markets operate, or in insti- Perhaps the clearest demonstration of the tutions (formal and informal). link comes from declines in maternal mortal- ity. Life expectancy rose sharply in Sri Lanka New labor market opportunities can between 1946 and 1953 as a result of declines spur investments in education and health in maternal mortality ratios, from 180 per 1,000 for girls live births to 50. Given that women had 5 chil- How much parents invest in their children’s dren on average, 1 in 10 women died giving education is partly determined by the returns to birth before these declines—a huge risk. Using that education. Early studies showed that new variation in the timing of the declines in differ- agricultural technologies that favored women’s ent parts of the country, one study shows that production increased girls’ enrollment.29 A new the overall declines in maternal mortality ra- generation of studies extends these insights in a tio boosted female life expectancy by 1.5 years, globalizing economy. female literacy by 1 percentage point, and female For instance, the rise of outsourcing in In- education by 0.17 years.32 dia offers new opportunities for women in the Given that a reduction in the maternal mor- wage sector and increases parental investments tality ratio also reduces maternal morbidity, it in girls’ education.30 Recruitment services that increases the ability of women to participate in informed families about new employment op- the labor force. Evidence from the United States portunities for Indian women increased the shows that in 1920 one in six women suffered chances of girls ages 5–15 years to be in school from a long-term disability incurred in child- by 3 to 5 percentage points but had no effect on birth. A sharp decline in maternal mortality boys.31 The girls also had higher body mass in- resulting from the discovery of sulfa drugs in dex (a measure of health) and were 10 percent 1936 (and their immediate widespread use) more likely to be employed in wage work. Per- ceived improvements in the likelihood of a job went hand in hand with dramatic improve- triggered investments in human capital for girls ments in health for women after childbirth. even when there were no changes in other po- Improvements in the conditions of childbirth tential limiting factors, such as poverty, cost, or were the biggest force behind the rise in married distance to school. Evidence of greater returns women’s labor force participation in the United was enough to stimulate greater human capital States between 1920 and 1950.33 accumulation. Who would have thought that the fastest It has often been posited that cultural and way to increase female labor force participa- social norms (or “informal institutions� in the tion in the United States at the beginning of Report’s framework) “hold back� human capi- the 20th century would be to reduce maternal tal investments. So, many policy efforts try to mortality? Households, markets, social norms, change the status quo by trying to nudge norms. and formal institutions are inextricably con- The results here present an alternative route— nected, and the key is to �nd ways to stimulate expand economic opportunities, and human progress in all domains. For reducing mater- capital investments in girls will increase. Mar- nal mortality, chapter 3 shows that income or kets can affect private household decisions, even household actions are less powerful than effec- with slow-moving social norms. tive institutions—and public investments are critical. Investing in the future A similar consideration—linking parental in- vestments in education and returns down the line—underpins the relationship between health and schooling. At its starkest, lowering the risk of death should lead to greater human capital accumulation during childhood. Put simply, the “ What are your hopes for your daughter’s future? She must be bright and intelligent, educated, and look after this community. For my sons, they must be educated to take ownership of land, longer you get a payout from your investment, build permanent houses, and develop the more attractive it is to make that investment. this community. If the risk of dying from childbirth is high, par- ents factor in this risk and reduce investments in daughters. � Adult woman, Papua New Guinea 68 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Choosing differently have voted on the measure in their cloak Recent �ndings suggest that women’s rights and rooms, it would have been killed as em- agency play a role seeing that those public invest- phatically as it was �nally passed out in ments are made. In a world where women care the open. 34 about different things from men (and women do appear to care for children more than men Growth in public spending led to the scal- do), it may be that when women have more ing up of intensive door-to-door hygiene cam- voice, they can drive institutional investments paigns and to a sharp decline in child mortality in a way that favors children. So, when women by 8 to 15 percent. Roughly 20,000 child deaths have more rights in the political arena, does the were averted every year because women won the nature of public investment change? Yes. right to vote. A variant of these broad results Recall that in the United States women won has also been documented in Switzerland, with voting rights state by state over the 19th century the additional twist that women seem to have until they were federally mandated by constitu- become more supportive of health and welfare tional amendment in 1920. Public health spend- expenditures over time than men.35 For Europe, ing increased dramatically as women won the the results suggest that female suffrage increased right to vote. The Sheppard-Towner Maternity spending on health, education, and welfare ex- and Infancy Protection Act of 1921 provided penditures over time.36 federal funding for maternity and child care. According to one observer, Intergenerational cycles The links across sectors also play out over time. Indeed, fear of being punished at the polls Recent studies show that, in England and the by American women, not conviction of the United States, more maternal education leads to bill’s necessity, seems to have motivated a host of better outcomes for children—better Congress to vote for it. As one senator ad- education and better health. For children ages mitted to a reporter from the Ladies Home 7–8, an additional year of mother’s schooling Journal, if the members of Congress could increases the child’s performance on a standard- F I G U R E 1 .8 Who agrees that a FIGURE 1.9 Who agrees that when university education jobs are scarce, men is more important for a should have more right to boy than for a girl? a job than women? 50 80 changes positive changes positive toward women toward women Turkey 40 Georgia Ukraine 60 Poland Georgia Turkey Russia Federation % agree, 1994–99 % agree, 1994–99 Japan Romania India 30 Mexico Russia Federation India Peru Poland China China Romania Brazil Chile 40 Ukraine Slovenia Brazil Bulgaria Argentina South Africa 20 Bulgaria Chile UruguayJapan Spain South Africa Slovenia Switzerland Australia Spain Mexico United States Great Britain Finland Germany Argentina Uruguay Norway Colombia Germany 20 Peru United States 10 Australia Sweden Norway Finland New Zealand New Zealand Sweden 0 0 0 10 20 30 40 50 0 20 40 60 80 % agree, 2005–07 % agree, 2005–07 Source: WDR 2012 team estimates based on World Values Surveys, Source: WDR 2012 team estimates based on World Values Surveys, 1994–99 and 2005–07 waves. 1994–99 and 2005–07 waves. Note: The 45° line in the �gure above shows parity in the values on Note: The 45° line in the �gure above shows parity in the values on the vertical and horizontal axis. the vertical and horizontal axis. A wave of progress 69 ized math test and reduces the incidence of be- NOTES havioral problems.37 Similar results document 1. Kingdon and Theopold 2008; Pitt, Rosenzweig, the link between maternal education and child and Hassan 2010. education in low-income countries, where even 2. Chinkin 2010. a single year of maternal education can make 3. Acar 2000. a large difference. Children of mothers with a 4. http://www.au.int/en/sites/default/files/999 single year of education in Pakistan spent an ad- Rights_of_Women.pdf. ditional hour every day studying at home and 5. Zaher 2002. reported higher test scores.38 6. Bennhold 2010. The intergenerational impact of female edu- 7. Ramirez, Soysal, and Shanahan 1997. cation is not restricted to the home. Contempo- 8. Hossain 2011. raneous and intertemporal links came together 9. Barro and Lee 2010. in the rise of private schooling in Pakistan, 10. Grant and Behrman 2010. which exploded after 1990.39 A curious feature 11. Dollar and Gatti 1999. of the expansion was that these private schools 12. Filmer 1999. were overwhelmingly located in villages where a 13. Avenstrup, Liang, and Nelleman 2004. public girls’ secondary school had existed a de- 14. Meyer, Ramirez, and Soysal 1992. cade earlier. By establishing secondary schools 15. Dulger 2004. for girls in rural areas, the government created 16. Hoyos and Narayan 2011. a cohort of teachers who could then teach chil- 17. WHO and others 2010. 18. United Nations 2010. dren at the primary level in the next generation. 19. Schultz 2005. The students of today became the teachers of to- 20. Deaton 2008. morrow. An institutional improvement (public 21. Casas, Dachs, and Bambas 2001. secondary schools for girls) enabled a household 22. Irving and Kingdon 2008. response (more girls with secondary education) 23. Egypt Ministry of Health Department of Plan- that then played out in a change in the market ning and Harvard School of Public Health–Data (private schools and more female employment for Decision Making 1998. opportunities) one generation later. 24. Sen, Asha, and Östlin 2002; Lee 2009; Demery and Gaddis 2009. Changes in attitudes 25. Hoyos and Narayan 2011. Upholding rights also has a ripple effect in so- 26. International Labor Organization 2010. cial mores and attitudes. The World Values Sur- 27. Goldin 1995; Mammen and Paxson 2000. vey provides a window on how social percep- 28. Chioda, Garcia-Verdú, and Muñoz Boudet 2011. tions have shifted. Traditionally, social attitudes 29. Bardhan 1974; Rosenzweig and Schultz 1982; toward women have given primacy to their do- Foster and Rosenzweig 1999. mestic roles as mothers and homemakers. Pro- 30. Oster and Millet 2010. gressively, social norms appear to be shifting to 31. Jensen 2010. 32. Jayachandran and Lleras-Muney 2009. be more accepting of women as social actors— 33. Albanesi and Olivetti 2009; Albanesi and Olivetti across a large number of countries with data, 2010. even if signi�cant differences remain across 34. As cited in Miller 2008. countries. 35. Funk and Gathmann 2007. Women are increasingly considered to have 36. Aidt, Dutta, and Loukoianova 2006. equal standing with men in access to tertiary 37. Currie and Moretti 2003; Carneiro, Meghir, and education and participation in the labor force. Parey 2007. In most countries with data, there has been a 38. Andrabi, Das, and Khwaja 2002. signi�cant reduction in the share of people who 39. Andrabi, Das, and Khwaja 2011. believe that “university is more important for men than for women� and “men should have more right to a job than women when jobs are REFERENCES scarce� (�gures 1.8 and 1.9). In Turkey, the share of individuals concurring that higher education The word processed describes informally reproduced is preferable for men fell from 34 to 20 percent works that may not be commonly available through in 10 years. libraries. 70 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Acar, Feride. 2000. “Turkey.� In The First CEDAW ed. Meena Shivdas and Sarah Coleman. London: Impact Study—Final Report, ed. Marilou McPhe- Commonwealth Secretariat. dran, Susan Bazilli, Moana Erickson, and Andrew Chioda, Laura, with Rodrigo Garcia-Verdú, and Ana Byrnes. Toronto, Ontario: International Women’s María Muñoz Boudet. 2011. Work and Family: Right Project, Centre for Feminist Research, and Latin American Women in Search of a New Bal- York University. ance. Washington, DC: World Bank, Of�ce of the Aidt, Toke S., Jayasri Dutta, and Elena Loukoianova. Chief Economist (LCRCE) and the Poverty and 2006. “Democracy Comes to Europe: Franchise Gender Group (LCSPP), Latin America and the Expansion and Fiscal Outcomes 1830–1938.� Eu- Caribbean Region (LAC). ropean Economic Review 50 (2): 249–83. Currie, Janet, and Enrico Moretti. 2003. “Mother’s Albanesi, Stefania, and Claudia Olivetti. 2009. “Gen- Education and the Intergenerational Transmis- der Roles and Medical Progress.� Working Paper sion of Human Capital: Evidence from College Series 14873, National Bureau of Economic Re- Openings.� Quarterly Journal of Economics 118 search, Cambridge, MA. (4): 1495–532. ———. 2010. “Maternal Health and the Baby Boom.� Deaton, Angus. 2008. “Height, Health, and Inequal- National Bureau of Economic Research Paper ity: The Distribution of Adult Heights in India.� Series 16146, Cambridge, MA. American Economic Review 98 (2): 468–74. Andrabi, Tahir, Jishnu Das, and Asim Ijaz Khwaja. Demery, Lionel, and Isis Gaddis. 2009. Social Spend- 2002. “The Rise of Private Schooling in Pakistan: ing, Poverty and Gender Equality in Kenya: A Ben- Catering to the Urban Elite or Educating the e�t Incidence Analysis. Nairobi: Deutsche Gesell- Rural Poor?� Background Paper for the Pakistan schaft für Technische Zusammenarbeit (GTZ). Poverty Assessment, World Bank, Washington, Dollar, David, and Roberta Gatti. 1999. “Gender In- DC. Processed. equality, Income, and Growth: Are Good Times ———. 2011. “Students Today, Teachers Tomorrow: Good for Women?� Policy Research Report on Identifying Constraints on the Provision of Educa- Gender and Development, Working Paper Series tion.� Policy Research Working Paper Series 5674, 1, World Bank, Washington, DC. World Bank, Washington, DC. Dulger, Ilhan. 2004. “Turkey: Rapid Coverage for Avenstrup, Roger, Xiaoyan Liang, and Soren Nelle- Compulsory Education—The 1997 Basic Educa- mann. 2004. Kenya, Lesotho, Malawi and Uganda: tion Program.� Paper presented at the Scaling Up Universal Primary Education and Poverty Reduc- Poverty Reduction: A Global Learning Process tion. Report 30765. Washington, DC: World Bank. and Conference, Shanghai, China, May 25. Bardhan, Pranab K. 1974. “On Life and Death Ques- Egypt Ministry of Health Department of Planning tions.� Economic and Political Weekly 9 (32/34): and Harvard School of Public Health–Data for 1293–304. Decision Making. 1998. Health Care Utilization Barro, Robert J., and Jong-Wha Lee. 2010. “A New and Expenditures in the Arab Republic of Egypt. Data Set of Educational Attainment in the World, Cairo and Boston: Egypt Ministry of Health De- 1950–2010.� Working Paper Series 15902, National partment of Planning and Harvard School of Bureau of Economic Research, Cambridge, MA. Public Health–Data for Decision Making. Bennhold, Katrin. 2010. “20 Years after Fall of Wall, Filmer, Deon. 1999. “The Structure of Social Dis- Women of Former East Germany Thrive.� The In- parities in Education: Gender and Wealth.� Policy ternational Herald Tribune, October 6. Research Report on Gender and Development, Carneiro, Pedro, Costas Meghir, and Matthias Parey. Working Paper Series 5, World Bank, Washing- 2007. “Maternal Education, Home Environments ton, DC. and the Development of Children and Adoles- Foster, Andrew D., and Mark Rosenzweig. 1999. cents.� IZA Discussion Paper Series 3072, Institute “Missing Women, The Marriage Market and for the Study of Labor, Bonn. Economic Growth.� Working Paper Series 49, Casas, Antonio Juan, J. Norberto W. Dachs, and Alex- Stanford Center for International Development, andra Bambas. 2001. “Health Disparities in Latin Stanford, CA. America and the Caribbean: the Role of Social and Funk, Patricia, and Christina Gathmann. 2007. “What Economic Determinants.� Pan American Health Women Want: Suffrage, Gender Gaps in Voter Organization, Equity and Health: Views from the Preferences and Government Expenditures.� Uni- Pan American Sanitary Bureau, Occasional Publi- versitat Pompeu Fabra, Barcelona; Stanford Uni- cation 8, Washington, DC. versity, Stanford, CA. Processed. Chinkin, Christine. 2010. “Thoughts on the UN Goldin, Claudia. 1995. “The U-Shaped Female La- Convention on the Elimination of All Forms of bor Force Function in Economic Development Discrimination against Women (CEDAW).� In and Economic History.� In Investment in Women’s Without Prejudice: CEDAW and the Determination Human Capital, ed. T. Paul Schultz. Chicago, IL: of Women’s Rights in a Legal and Cultural Context, University of Chicago Press. A wave of progress 71 Grant, Monica J., and Jere R. Behrman. 2010. “Gen- Education, 1870–1980.� Sociology of Education 65 der Gaps in Educational Attainment in Less De- (2): 128–49. veloped Countries.� Population and Development Miller, Grant. 2008. “Women’s Suffrage, Political Review 36 (1): 71–89. Responsiveness, and Child Survival in American Hossain, Naomi. 2011. “Exports, Equity and Em- History.� Quarterly Journal of Economics 123 (3): powerment: The Effects of Readymade Garments 1287–327. Manufacturing Employment on Gender Equality Oster, Emily, and M. Bryce Millet. 2010. “Do Call in Bangladesh.� Background paper for the WDR Centers Promote School Enrollment? Evidence 2012. from India.� Working Paper Series 15922, National Hoyos, Alejandro, and Ambar Narayan. 2011. “In- Bureau of Economic Research, Cambridge, MA. equalities of Opportunities among Children: Pitt, Mark M., Mark R. Rosenzweig, and Nazmul How Much Does Gender Matter.� Background Hassan. 2010. “Human Capital Investment and note for the WDR 2012. the Gender Division of Labor in a Brawn-Based ILO (International Labor Organization). 2010. “Key Economy.� Economic Growth Center Discussion Indicators of the Labour Market.� ILO, Geneva. Paper Series 989, Yale University, New Haven, CT. Irving, Margaret, and Geeta Gandhi Kingdon. 2008. Ramirez, Francisco O., Yasemin Soysal, and Suzanne “Gender Patterns in Household Health Expendi- Shanahan. 1997. “The Changing Logic of Po- ture Allocation: A Study of South Africa.� CSAE litical Citizenship: Cross-National Acquisition of Working Paper Series 2008–32, Centre for the Women’s Suffrage Rights, 1890 to 1990.� American Study of African Economies, Oxford University, Sociological Review 62 (5): 735–45. Oxford, U.K. Rosenzweig, Mark R., and Theodore Paul Schultz. Jayachandran, Seema, and Adriana Lleras-Muney. 1982. “Market Opportunities, Genetic Endow- 2009. “Life Expectancy and Human Capital In- ments and the Intrafamily Distribution of Re- vestments: Evidence from Maternal Mortality sources: Child Survival in Rural India.� American Declines.� Quarterly Journal of Economics 124 (1): Economic Review 72 (4): 803–15. 349–97. Schultz, T. Paul. 2005. “Productive Bene�ts of Health: Jensen, Robert. 2010. “Economic Opportunities and Evidence from Low-income Countries.� In Health Gender Differences in Human Capital: Experi- and Economic Growth: Findings and Policy Im- mental Evidence for India.� Working Paper Series plications, ed. Guillem López-Casanovas, Berta 16021, National Bureau of Economic Research, Rivera, and Luis Currais. Cambridge, MA: MIT Cambridge, MA. Press. Kingdon, Geeta Gandhi, and Nicolas Theopold. 2008. Sen, Gita, George Asha, and Piroska Östlin, eds. 2002. “Do Returns to Education Matter to Schooling Engendering International Health: The Challenge Participation? Evidence from India.� Education of Equity. Cambridge, MA: MIT Press. Economics 16 (4): 329–50. United Nations 2010. The World’s Women 2010: Lee, Jinhyun. 2009. “Estimating Individual Non- Trends and Statistics. New York: UN. linear Health Expenditure and Income Relation- WHO (World Health Organization), UNICEF ships using Household Data: The Indonesian (United Nations Children’s Fund), UNFPA SUSENAS Case.� School of Economics and Fi- (United Nations Population Fund), and World nance, University of St. Andrews, St. Andrews, Bank. 2010. Trends in Maternal Mortality: 1990 Fife, U.K. Processed. to 2008. New York and Washington, DC: WHO, Mammen, Kristin, and Christina Paxson. 2000. UNICEF, UNFPA, and World Bank. “Women’s Work and Economic Development.� Zaher, Claudia. 2002. “When a Woman’s Marital Journal of Economic Perspectives 14 (4): 141–64. Status Determined Her Legal Status: A Research Meyer, John W., Francisco O. Ramirez, and Yasemin Guide on the Common Law Doctrine of Cover- Nuhoglu Soysal. 1992. “World Expansion of Mass ture.� Law Library Journal 94 (3): 459–87. CHAPTER 2 The persistence of gender inequality T hings have changed for the better, but “ not for all women and not in all do- mains of gender equality. Progress has We are not educated. We don’t been slow and limited for women in know anything and do not know very poor countries, for those who are poor, how we can be able to change even amid greater wealth, and for those who face our life and obtain power and other forms of exclusion because of their caste, � freedom. disability, location, ethnicity, or sexual orienta- tion. Whether for comparisons between men Adult woman, rural Afghanistan and women in the same country, or absolute comparisons of women across countries, the progress in some domains is tempered by the so- bering realities that many women face in others. resentation in policy making remains far lower Consider the likelihood of women dying dur- than men’s. Or that domestic violence contin- ing childbirth in Sub-Saharan Africa, which is ues to exact a heavy toll on women around the still comparable to the rate in Northern Europe world—regardless of individual or country in- during the 19th century. Or the difference in come, women continue to be the primary vic- school enrollments in Nigeria, where a wealthy tims of violence at home and to suffer more urban child averages around 10 years of school- severe injuries. ing, while poor rural Hausa girls average less Where is progress absent? than 6 months.1 Or the fact that women remain severely disadvantaged in their control over re- • Severely disadvantaged populations. Across sources and assets in the household. Or that the and within countries, gender gaps widen at earnings differentials between men and women lower incomes, and, in the poorest economies, have not changed much. Or that women’s rep- gender gaps are larger. The bene�ts of eco- “ There is no such thing as equality between men and women in this community, maybe in towns and urban areas, but not here! A man is always above the woman. Adult man, rural South Africa Women just started entering society, so the man is still trying to maintain his control. � Adult man, West Bank and Gaza The persistence of gender inequality 73 nomic growth have not accrued equally to all “ men and all women for some parts of society. Household poverty can mute the impact of Money. And someone who would national development, and the differences are convince my dad to let me continue often compounded by other means of social my education. . . . If I had enough exclusion, such as geography and ethnicity. money, I would enroll somewhere as a part-time student. You have to pay for • “Sticky� domains. Improvements in some do- every exam. mains of gender equality—such as those re- lated to occupational differences or partici- pation in policy making—are bound by constraints that do not shift with economic � Young woman, Serbia growth and development. Gender disparities school. The Republic of Yemen has one of the endure even in high-income economies de- world’s largest gender disparities in net enroll- spite the large gains in women’s civil and ment rates, and progress has been dif�cult to economic rights in the past century. These sustain.2 School enrollments for girls 5–19 years outcomes are the result of slow-moving in- old in Mali are equivalent to those in the United stitutional dynamics and deep structural fac- States around 1810 (�gure 2.1). tors that growth alone cannot address. Gender disparities have also lingered among • Reversals. External shocks—sometimes eco- groups that have not bene�ted from growth nomic, sometimes political, sometimes insti- within countries: income poverty widens gen- tutional—can erase hard-earned gains. In der gaps. While the educational attainment of some instances, improvements in gender equality have been reversed in the face of un- expected shocks that revealed or worsened FIGURE 2.1 Female enrollments remain strikingly low in some institutional or market failures. The shocks countries affect both males and females, but multiple factors shape their impact on gender differen- 100 Egypt, Arab Rep. tials—among them, the source and type of 90 shock, economic and institutional structures, Bangladesh female school enrollment, ages 5–19, (%) and social norms. Even when shocks do not 80 Mozambique have differential gender impacts, the absolute Tajikistan welfare losses for both men (and boys) and 70 women (and girls) can be substantial. In par- Nigeria Niger Côte d’Ivoire 60 ticular, adverse circumstances early in life, as in the critical �rst three years, can have irre- Ethiopia 50 versible long-term effects. Burkina Faso Mali 40 Pakistan SEVERELY DISADVANTAGED 30 United States POPULATIONS 20 in 1900 While much of the world has reduced gender 10 gaps in health and education, conditions for women in some low-income countries have not 0 improved much. In many South Asian and Sub- 17 0 17 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 20 0 20 0 05 7 8 9 0 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 17 Saharan countries, girls’ enrollments in primary year and secondary education have progressed little. United States In Eritrea, the female primary net enrollment trend line (United States, 1850–2000) rate rose from a very low base of 16 percent in Source: WDR 2012 team estimates based on U.S. Census and the International Income Distribution Data- 1990 to just 36 percent in 2008. In Afghanistan, base (I2D2). Chad, and the Central African Republic, there Note: Values between 1760 and 1840 are based on female school enrollment trending between 1850 and are fewer than 70 girls per 100 boys in primary 2000. 74 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 F I G U R E 2.2 In some countries, female disadvantage augments at lower incomes … Benin Congo, Dem. Rep. The Gambia 10 10 10 median grade attained, median grade attained, median grade attained, ages 15–19 ages 15–19 ages 15–19 5 5 5 0 0 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 income quintile income quintile income quintile India Pakistan Togo 10 10 10 median grade attained, median grade attained, median grade attained, ages 15–19 ages 15–19 ages 15–19 5 5 5 0 0 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 income quintile income quintile income quintile girls boys Source: WDR 2012 team estimates based on EdAttain. wealthy boys and girls is very similar, gender tends to be small—as in Uzbekistan or Viet- inequalities are intensi�ed among the poor. In nam—gender differences also tend to be small. India, the median boy and girl ages 15–19 in the The disadvantage for girls tends to be more wealthiest �fth of the population reach grade 10, pronounced and to emerge earlier than for boys. but the median boy in the bottom �fth reaches A girl in the poorest �fth of the population in only grade 6, and the median girl only grade 1. the Democratic Republic of Congo studies three Across countries there is little gender disadvan- fewer years than a poor boy. And a cumulative tage for the wealthiest: households in the top in- gender bias against girls builds over the edu- come quintile tend to achieve full gender parity cational life cycle. In 2008, there were only 66 in education. female tertiary students for every 100 male stu- Poor girls face a signi�cant schooling dis- dents in Sub-Saharan Africa and 76 in South advantage in much of Africa and South Asia, a Asia.3 Sub-Saharan Africa is the only region disadvantage that increases at lower incomes, as where growth in male tertiary enrollment has in Benin, the Democratic Republic of Congo, outpaced female enrollment growth, especially The Gambia, and Togo (�gure 2.2). Yet the for doctoral degrees. opposite can be observed elsewhere—in Ban- The gaps between rich and poor are the same gladesh, Brazil, the Dominican Republic, the in health. Lower fertility rates imply that fewer Philippines, and the República Bolivariana de women are exposed to the risk of childbirth, Venezuela, girls at low levels of wealth tend to and reductions in parity (the number of times stay longer in school than boys (�gure 2.3). Re- a woman has given birth) and the age-structure gardless of whether the gender gap favors boys of births have accounted for a sizable fraction or girls at low household incomes, in countries of declines in the lifetime risk of death from where the difference between rich and poor maternal causes.4 Although fertility rates have The persistence of gender inequality 75 F I G U R E 2.3 … yet in others, at low levels of wealth girls stay longer in school than boys Bangladesh Brazil Dominican Republic 10 10 10 median grade attained, median grade attained, median grade attained, 8 8 8 ages 15–19 ages 15–19 ages 15–19 6 6 6 4 4 4 2 2 2 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 income quintile income quintile income quintile Philippines Venezuela, RB 10 10 median grade attained, median grade attained, 8 8 ages 15–19 ages 15–19 6 6 4 4 2 2 1 2 3 4 5 1 2 3 4 5 income quintile income quintile girls boys Source: WDR 2012 team estimates based on EdAttain. dropped dramatically in all regions since 1960, ethnic minorities are poorer and less urban than they have been rising in many Sub-Saharan the general population. An estimated two-thirds countries. In Nigeria, the total fertility rate rose of girls out of school globally belong to ethnic from 4.7 children in 1999 to 5.7 in 2008.5 minorities in their countries.6 In Guatemala, the As in education, household wealth makes a illiteracy rate among indigenous women stands difference. In all countries, fertility rates for the at 60 percent, 20 percentage points above indig- poor are higher than for the rich (�gure 2.4). enous males and twice that of nonindigenous fe- Yet at low fertility (typically in richer coun- males.7 For ethnic minorities in Vietnam, more tries), the differences between the bottom and than 60 percent of childbirths take place with- the top quintiles tend to be small—on the order out prenatal care, twice the rate for the majority of 0.5 to 1 live birth. At higher fertility (usually Kinh. More urban ethnic minority groups and in poor countries), the differences widen. In groups not concentrated in poor regions tend to Zambia, the average fertility of a woman in the experience smaller differences with the major- poorest quintile is 8.5 children (the highest in ity populations. In China, rural ethnic minority the world), but for a woman in the richest �fth, groups have less access to education and health it is just over 3. than the more urban Han, Hui, and Manchu, In addition to household wealth, ethnicity but the school enrollment and health insurance and geography are important for understand- gaps narrow in urban areas.8 ing and addressing gender inequality. Even in Other factors of exclusion, such as caste, dis- countries that have grown rapidly, poor and ability, or sexual orientation, also tend to com- ethnic minority women tend to bene�t far less pound disadvantages in ways that affect devel- than their richer and ethnic majority counter- opment outcomes. More research is needed to parts. So, wide gender disparities endure. Many better understand these links. 76 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 control over resources, women’s political voice, F I G U R E 2.4 At low incomes, fertility rates remain high—And or the incidence of domestic violence. the poorer the country, the larger the gap between In some cases, individual preferences, mar- rich and poor ket failures, institutional constraints, and so- cial norms continue to reinforce gender gaps Niger Zambia: Women in despite economic progress. Income growth Mali the poorest quintile may also have unexpected adverse effects on Congo, Dem. Rep. bear more than Chad 8 children on gender equality through new gendered prefer- Zambia average; whereas ences. In other cases, development outcomes Malawi in the richest Liberia have not always reflected extensive formal quintile, the Angola average drops gains in securing equal rights. Despite notable Guinea Tanzania to 3.5 children. improvement in expanding legal guarantees to Nigeria women and men alike, slow implementation Rwanda has impeded a move into gender parity. Social Ethiopia Sierra Leone norms continue to bind to varying degrees in Cameroon all nations, and a chasm remains between the- Senegal Madagascar ory and practice. Congo, Rep. (Brazzaville) Economic growth can even temporarily ag- Kenya Pakistan gravate gender differentials in some countries. Ghana In China, new opportunities for rural industrial Haiti Swaziland wage work led families initially to favor the ju- Zimbabwe nior secondary education of males, considerably Jordan widening the gender gap in the 1980s.9 But as Namibia Lesotho the economy continued to grow, girls rapidly Cambodia caught up with boys in the 1990s. Sub-Saharan Philippines Honduras Africa is in the midst of a signi�cant expansion Nepal of secondary education. As in China, more Af- Egypt, Arab Rep. Bangladesh rican boys than girls attended secondary school India between 1999 and 2008, deepening the gender Indonesia Colombia gaps. In 2008, there were 79 girls for every 100 Dominican Republic boys, down from 83 girls per 100 boys in 1999. Azerbaijan United States Indeed, girls face signi�cant barriers to second- Australia ary school entry, but enrollment rates tend to be Armenia low all around. Moldova Ukraine In some cases, these gaps work against boys. Everywhere in the world, repetition and, to a 0 1 2 3 4 5 6 7 8 lesser extent, dropout rates are higher among average number of children boys than among girls. Some upper-middle- income and advanced economies are concerned richest quintile average poorest quintile fertility rate fertility rate fertility rate about male underachievement in education— girls outperforming boys academically. In the United States and Israel, girls obtain better grades Source: WDR 2012 team estimates based on Demographic and Health Surveys. in all major school subjects, including math and science.10 In France, women are the majority in enrollments at the elite Grandes Ecoles de Com- “STICKY� DOMAINS, DESPITE merce (business schools).11 Male underperfor- ECONOMIC PROGRESS mance in higher education usually is not rooted in social exclusion, but men can also be subject In two areas, income growth has brought only to cultural norms that steer them away from modest and gradual progress toward gender academic achievement. Identifying education as equality in most developing countries: female primarily a “female� endeavor means that young mortality and access to economic opportunities. men in several Caribbean nations, such as Dom- And gender gaps have not narrowed in women’s inica and Jamaica, withdraw from school.12 The persistence of gender inequality 77 Missing girls at birth and excess “ � female mortality My sister gave birth when she was 14 Sex ratios at birth and mortality across coun- and died. tries in 1990, 2000, and 2008 reveal continuing Young woman, the Republic of Yemen disadvantages for women in many low- and middle-income countries (and disadvantages for men in some regions for speci�c reasons). First, the problem of skewed sex-ratios at in China, India, and Sub-Saharan Africa. In birth in China and India (and in some coun- other countries—notably some post-transi- tries in the Caucasus and the Western Bal- tion economies—excess male mortality has kans) remains unresolved (table 2.1). Popula- become serious. tion estimates suggest that an additional 1.4 Over the past three decades, some aspects million girls would have been born (mostly in of the problem remained the same, while oth- China and India) if sex ratios at birth in these ers changed dramatically. Skewed sex ratios at countries resembled those found worldwide. birth were identi�ed in the early 1990s,14 and Second, compared with developed economies, as prenatal sex determination spreads and fer- the rate at which women die relative to men in tility declines, the problem has become worse. low- and middle-income countries is higher Excess female mortality is slowly shifting from in many regions of the world.13 Overall, miss- early childhood in South Asia to adulthood in ing girls at birth and excess female mortality Sub-Saharan Africa, declining in all low-income under age 60 totaled an estimated 3.9 million countries except in Sub-Saharan Africa (see women in 2008—85 percent of them were chapter 3). TA B L E 2.1 Missing girls at birth increased between 1990 and 2008 in India and China, as did excess female mortality in adulthood in Sub-Saharan Africa Excess female deaths in the world, by age and region, 1990 and 2008 (thousands) Total women girls at birth girls under 5 girls 5–14 women 15–49 women 50–59 under 60 1990 2008 1990 2008 1990 2008 1990 2008 1990 2008 1990 2008 China 890 1,092 259 71 21 5 208 56 92 30 1,470 1,254 India 265 257 428 251 94 45 388 228 81 75 1,255 856 Sub-Saharan Africa 42 53 183 203 61 77 302 751 50 99 639 1,182 High HIV-prevalence countries 0 0 6 39 5 18 38 328 4 31 53 416 Low HIV-prevalence countries 42 53 177 163 57 59 264 423 46 68 586 766 South Asia (excluding India) 0 1 99 72 32 20 176 161 37 51 346 305 East Asia and Pacific (excluding China) 3 4 14 7 14 9 137 113 48 46 216 179 Middle East and North Africa 5 6 13 7 4 1 43 24 15 15 80 52 Europe and Central Asia 7 14 3 1 0 0 12 4 4 3 27 23 Latin America and the Caribbean 0 0 11 5 3 1 20 10 17 17 51 33 Total 1,212 1,427 1,010 617 230 158 1,286 1,347 343 334 4,082 3,882 Source: WDR 2012 team estimates based on World Health Organization 2010 and United Nations Department of Economic and Social Affairs 2009. Note: Totals do not necessarily add up because of rounding. 78 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 tries, with only 1,900 maternal deaths.17 One of F I G U R E 2.5 Maternal mortality in many developing countries every 10 women in Afghanistan and 1 of every is similar to that in Sweden before 1900 14 in Somalia and Chad die from maternal causes, and a much larger fraction suffer long- 1,600 term health issues stemming from complica- Afghanistan tions during and after childbirth. 1,400 Progress in maternal mortality has not kept maternal mortality ratio, per 100,000 live births up with GDP growth. During 2000–08, the econ- 1,200 Chad omies of Chad and Tanzania grew at impressive annual rates of 9.4 percent and 7 percent, but maternal mortality declined by a mere 8 percent 1,000 Liberia Lib i (to 1,200 per 100,000 live births) in Chad and by 14 percent (to 790) in Tanzania. South Africa sulfa drugs introduced 800 Tanzania in 1930s and large-scale large scale grew at a modest 4 percent annually during the Sub-Saharan institutionalization same period, and maternal mortality increased Africa t of births by 8 percent to 410 per 100,000 births—a mani- Bangladesh 600 Cameroon Cambodia Burkina festation of the HIV/AIDS epidemic. Since 1990, both India and Equatorial Guinea had declines Indonesia Malawi 400 of 41 percent in their maternal mortality ratios, n C B op ean ar a d Ce aste e C eric b an which fell to similar levels in 2008, but the two ib th m e d A India countries had radically different growth trajec- an atin l A ur 200 sia ra E L nt rn tories—a mere 3 percent a year in Equatorial nd Dominican Republic la E n China Guinea compared with a solid 8 percent in In- Po 0 r p Europepe dia. Driving the high maternal mortality rates 17 0 17 0 17 0 17 0 17 0 17 0 17 0 17 0 17 0 17 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 18 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 19 0 20 0 00 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 16 year in many countries are poor obstetric health ser- Sweden vices and high fertility rates. Income growth and trend line (Sweden, 1750–1850) changes in household behavior alone appear insuf�cient to reduce maternal mortality; pub- Source: WDR 2012 team estimates based on World Health Organization and others 2010. lic investments are key to improving maternal health care services. The disadvantage against unborn girls is What accounts for these patterns? Chapter 3 widespread in many parts of Asia and in some provides a deeper discussion, but two issues are countries in the Caucasus (such as Armenia and highlighted here: maternal mortality and the Azerbaijan), where the intersection of a prefer- preference for sons. The female disadvantage in ence for sons, declining fertility, and new tech- mortality during the reproductive ages is in part nology increases the missing girls at birth. In driven by the risk of death in pregnancy and China and India, sex ratios at birth point to a childbirth and associated long-term disabilities.15 heavily skewed pattern in favor of boys. Where Although maternal mortality ratios have fallen parents continue to favor sons over daughters, by 34 percent since 1990, they remain high in a gender bias in sex-selective abortions, female many parts of the world: Sub-Saharan Africa had infanticide, and neglect is believed to account the highest ratio in 2008 at 640 maternal deaths for millions of missing girls at birth. In 2008 per 100,000 live births, followed by South Asia alone, an estimated 1 million girls in China and (280), Oceania (230), and Southeast Asia (160).16 250,000 girls in India were missing at birth.18 Bangladesh, Cambodia, India, and Indonesia The abuse of new technologies for sex-selective have maternal mortality ratios comparable to abortions—such as cheap mobile ultrasound Sweden’s around 1900, and Afghanistan’s is simi- clinics—accounted for much of this shortfall, lar to Sweden’s in the 17th century (�gure 2.5). despite laws against such practices in many These high mortality ratios translate into nations, such as India and China.19 Economic large absolute numbers of maternal deaths, es- prosperity will continue to increase amniocen- pecially where fertility rates remain high. In tesis and ultrasound services throughout the de- 2008, there were 63,000 maternal deaths in India veloping world, possibly enabling the diffusion and 203,000 (more than half of the total) in Sub- of sex-selective abortions where son-preferences Saharan Africa, in stark contrast to rich coun- exist.20 The persistence of gender inequality 79 These preferences do not appear to change easily. Even among later children of South and FIGURE 2.6 Women are more likely Southeast Asian immigrants in Canada, the than men to work in the United Kingdom, and the United States, the informal sector share of male births remains unusually high.21 This does not imply that change is impos- sible: The Republic of Korea’s male-female sex Nepal Bolivia ratio under age �ve was once the highest in Egypt, Arab Rep. Asia, but it peaked in the mid-1990s and then Peru reversed—a link to societal shifts in normative Indonesia values stemming from industrialization and Brazil urbanization.22 Colombia Argentina In a smaller set of countries, there are also India missing men. In Eritrea in the 1990s, a large South Africa number of young men went missing due to con- Mexico flict. In some countries in Latin America and the Russian Federation Caribbean, violence may have contributed to ex- Thailand Ukraine cess deaths among young males. In Eastern Eu- Turkey rope and Central Asia, a much larger number of Macau SAR, China men are missing in middle age, and this excess Kyrgyz Republic male mortality has been linked to the preva- Kenya lence of types of conduct deemed more socially 0 10 20 30 40 50 60 70 80 90 100 acceptable among men, such as alcohol use and informal sector employment other risky behavior. as a % of total employment women men “ Men and women are not paid the same daily wages. If men get Nu 200, then women get only Nu 150 for doing the same work. It is not fair or Source: WDR 2012 team estimates based on International Labour Organization 2010. Note: Most recent year available between 1999 and 2006. formal businesses, often out of their homes. Of just. � Different work, less pay Adult woman, Bhutan industrial homeworkers in some developing countries, such as Chile and Thailand, 80 per- cent are women.24 Because of the nature of these businesses, female owners are concentrated in the smallest �rms—smaller in employees, sales, Men and women work in different industries costs, and the value of physical capital. They also and occupations in developed and developing have lower pro�ts than male-owned �rms. In nations. But as chapter 1 showed, although more Latin America, average pro�ts are between 15 women are working outside the home in almost and 20 percent of a standard deviation lower for all countries, they are clustered into selected female than for male-owned �rms.25 parts of the “economic space,� with little change Third, even within the formal and informal over time, even in high-income countries. Three sectors, women and men choose very different markers of the segregated workspace are partic- jobs (�gure 2.7). Women are more likely to be ularly striking. First, women are more likely to in communal and public services, retail services, engage in low productivity activities than men and trade. Men are overrepresented in danger- and to work in the informal sector (�gure 2.6). ous professions—such as mining, construction, Women are more likely to be wage workers and transport, and heavy manufacturing—with high unpaid family workers than men, to have less occupational injury rates in poor and rich coun- mobility between the formal and informal sec- tries alike. The burdens of defense and maintain- tors, and to transition more between the infor- ing public order also fall heavily on men. These mal sector and being out of the labor force.23 patterns are similar across countries and regions, Second, among the self-employed, women and, if anything, they are accentuated at higher outside agriculture tend to operate small in- incomes. 80 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Housework and care are still a F I G U R E 2.7 Women and men work in different sectors (and woman’s domain different occupations) One domain where gender differences appear to be particularly persistent is the allocation of time distribution of female / male employment across sectors to housework and care. Over time and across 31 % Communication Services 16 % countries, irrespective of income, women bear disproportionate responsibility for housework 21 % Retail, Hotels, and Restaurants 17% and care, while men are mostly responsible for 13 % market work. These differences, deeply rooted Manufacturing 12 % in gender roles, reduce women’s leisure, wel- 4% Finance and Business 4% fare, and well-being. An immediate outcome of these different domestic responsibilities is that 0.5 % Electricity, Gas and Steam, and Water 1% men and women have very different patterns 0.5 % Mining 2% of time use and different amounts of leisure. These patterns have implications for women’s 2% Transport and Telecommunications 7% ability to invest in education (chapter 3), their 27 % Agriculture, Hunting, etc. 29 % agency (chapter 4), and their ability to take up economic opportunities (chapter 5), and to 1% Construction 11 % participate more broadly in economic, political, and social life (chapters 4 and 6). 100% All Sectors / All Occupations 100% In six countries—with widely different in- comes, economic structures, and social norms— Source: WDR 2012 team estimates based on data from LABORSTA Labor Statistics Database, International the patterns are remarkably similar (�gure 2.8). Labour Organization. Everywhere, women devote 1 to 3 hours more a Note: Totals do not necessarily add up due to rounding. day to housework than men; 2 to 10 times the amount of time a day to care (of children, el- Such gender-differentiated patterns con- derly, and the sick), and 1 to 4 hours less a day to tribute to the persistence of sizable gender gaps market activities. These are averages for all men in earnings. Differences in average wages by and women, and the differences are accentuated gender range from 20 percent in Mozambique with family formation. As chapter 5 describes, and Pakistan to more than 80 percent in Côte marriage signi�cantly increases the time de- d’Ivoire, Jordan, Latvia, and the Slovak Repub- voted to housework for women but not for men. lic. The gaps are slowly diminishing, partly be- Children signi�cantly increase the time spent on cause of improvements in education among care by both men and women but more so for women relative to men, differences in the con- women. centration of women in some sectors and oc- Time use for women and men converge as cupations, and shifts in work experience pat- income and education increase, mainly because terns and career interruptions linked to greater women become more like men (increasing their control over fertility.26 hours devoted to market work and decreasing the “ Women need more free time, women are more tired than men. . . . They take care of the house, of the children. While men, they are the entire day at work and don’t have to take care of the house. And if the woman has a job also, then she gets even more tired. Adult woman, Moldova These days, for a woman to be rated as a ‘good wife’ one has to be a superwoman, working very hard both at home and in the office, ful�lling every demand of your family members as if we don’t have any right to enjoy. � Adult woman, Bhutan The persistence of gender inequality 81 F I G U R E 2.8 Across the world, women spend more hours each day on housework and care than men … and men spend more time in market activities FOR SALE Market activities Housework Child care Pakistan 0.6 4.7 5.5 2.5 1.2 0.2 Cambodia 2.7 3.8 4.4 3.3 0.9 0.1 South Africa 2.1 3.8 4.2 1.8 0.5 0.0 Bulgaria 2.9 3.9 4.7 2.6 0.4 0.1 Sweden 3.2 4.6 3.2 2.3 0.6 0.3 Italy 2.1 4.8 4.9 1.4 0.6 0.2 women = 12 hours men Source: Berniell and Sánchez-Páramo 2011. hours to housework and care), not because men income or family spending reflects their con- take up more housework and care (chapter 5). trol over their own lives and their immediate environment. Second, trends in domestic vio- Less voice and less power lence capture intrahousehold gender dynamics Some dimensions of gender equality where and asymmetric power relations between men progress has been slowest fall in the domain of and women. Third, patterns in political voice women’s agency. Consider three aspects. First, can measure inclusiveness in decision making, women’s ability to make decisions about earned exercise of leadership, and access to power. 82 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 in some developing countries, particularly in F I G U R E 2.9 Who controls women’s own income? Sub-Saharan Africa and Asia, are not involved in household decisions about spending their Ukraine personal earned income. As many as 34 percent Colombia of married women in Malawi and 28 percent of Honduras Moldova women in the Democratic Republic of Congo are Peru not involved in decisions about spending their Haiti Niger earnings. And 18 percent of married women in Indonesia Bolivia India and 14 percent in Nepal are largely silent Benin on how their earned money is spent.27 Morocco Madagascar Husbands have more control over their Philippines wives’ earning at lower incomes. In Turkey, only Swaziland Cambodia 2 percent of married women in the richest �fth Congo, Rep. (Brazzaville) of the population have no control over earned Dominican Republic Egypt, Arab Rep. cash income, a proportion that swells to 28 per- Burkina Faso cent in the poorest �fth. In Malawi, 13 percent Ethiopia Zimbabwe of married women in the richest �fth have no Armenia control, compared with 46 percent in the poor- Ghana Chad est �fth (�gure 2.9). Azerbaijan Less control over resources and spending is Mali Guinea partly a reflection of large differences between Kenya men and women in the assets they own. Assets Cameroon Lesotho are typically inherited, acquired at marriage, or Turkey Namibia accumulated over the lifetime through earnings Bangladesh and saving. As shown above and explored fur- Nepal Uganda ther in chapter 5, women typically earn less than Nigeria men, particularly when aggregated over the life India Mozambique cycle. This disparity directly affects their ability Zambia to save, irrespective of male-female differences Rwanda Liberia in savings behavior. And as chapter 4 explores, Sierra Leone inheritance and property rights often apply dif- Congo, Dem. Rep. Malawi ferently to men and women so that gender dis- parities in access to physical capital and assets 0 5 10 15 20 25 30 35 40 45 remain large and signi�cant. Land makes up the % of women not involved in decisions about their own income largest share of household assets, particularly women in richest quintile for the poorest and rural households.28 Women women in poorest quintile own as little as 11 percent of land in Brazil and all women, average 27 percent in Paraguay. And their holdings are smaller than those of men. In Kenya, as little as 5 Source: WDR 2012 team estimates based on Demographic and Health Surveys. percent of registered landholders are women.29 In Ghana, the mean size of men’s landholdings was three times that of women’s.30 Less control over resources In many countries, land ownership remains Many women have no say over household �- restricted to men only, both by tradition and by nances, even their own earnings. The Demo- law. In most African countries and about half graphic and Health Surveys show that women of Asian countries, customary and statutory “ Some working women don’t even know how much they get paid for their job because their husbands cash their salary for them. � Adult man, West Bank and Gaza The persistence of gender inequality 83 laws disadvantage women in land ownership. “ According to customary law in some parts of I think that women are now a problem: they Africa, women cannot acquire land titles with- get money and they no longer listen to us. out a husband’s authorization.31 Marriage is the So, if you want to continue being a man in the most common avenue for women to gain ac- cess to land. But husbands usually own it, while house, you need to bring the discipline. You wives only have claim to its use. While property must beat her up, and if any child intervenes, rights for women have slowly begun to improve you also beat them. Then they all fear and � in some countries, legislation has often proved respect you. insuf�cient to change observed practices. Adult man, Tanzania More vulnerable to violence at home Physical, sexual, and psychological violence food and 59 percent for arguing with their hus- against women is endemic across the world. bands (�gure 2.10). A flagrant violation of basic human rights The prevalence of domestic violence varies and fundamental freedoms, violence can take greatly across rich and poor countries. Physical many forms. International statistics are not al- ways comparable, yet incontrovertible evidence shows that violence against women is a global FIGURE 2.10 Perceptions in many nations are that wife- concern.32 beating is justi�able Women are at far greater risk than men of violence by an intimate partner or somebody they know than from violence by other people. Ukraine Peru According to South African data, for example, Dominican Republic teachers were the most common perpetrators Philippines Honduras of the rape of girls under age 15 (one-third of Bolivia Moldova cases).33 About 50 percent of female homicides Armenia Swaziland in South Africa were perpetuated by an intimate Nepal partner. The mortality rate from intimate part- Malawi Indonesia ner violence there is estimated at 8.8 per 100,000 Bangladesh Haiti women.34 Overall, women are more likely than Madagascar men to be killed, seriously injured, or victims of Namibia Ghana sexual violence from intimate partners.35 Turkey Nigeria The number of countries with laws regulat- Benin ing intimate partner violence has risen. In 2006, India Zimbabwe 60 countries had speci�c legislation to address Rwanda Lesotho domestic violence, up from 45 in 2003, and 89 Azerbaijan had some form of legal provision. Many of these Egypt, Arab Rep. Kenya are higher-income countries; most developing Mozambique Cameroon countries with laws against intimate partner vi- Liberia olence are in Southeast Asia and Latin America. Zambia Morocco Yet in many nations, violence against women Sierra Leone Congo, Rep. (Brazzaville) is perceived as acceptable or justi�able. On aver- Niger age, 29 percent of women in countries with data Uganda Burkina Faso concurred that wife beating was justi�ed for ar- Mali Congo, Dem. Rep. guing with the husband, 25 percent for refusing Ethiopia to have sex, and 21 percent for burning food. In Guinea Guinea, 60 percent of women found it permis- 0 10 20 30 40 50 60 70 80 sible to be beaten for refusing to have sex with percentage of women who agree their spouses. In Ethiopia, 81 percent of women Rationale for beating say that it is justi�ed for a husband to beat his argues with husband burns food refuses to have sex respondents agree with at least one wife for at least one of the reasons listed in the Demographic and Health Surveys; 61 percent reported violence to be appropriate for burning Source: WDR 2012 team estimates based on Demographic and Health Surveys. 84 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Men can also be victims. Domestic violence F I G U R E 2 .1 1 There is great against them is more circumscribed than against heterogeneity in rates women, in incidence, nature, and severity. Ac- of domestic violence cording to the National Crime Victimization reported across nations Survey in the United States, intimate partner violence affected 4.1 females per 1,000—more Brazil (Pernambuco) than half a million women—and 0.9 males per Thailand (Nakhonsawan) 1,000—117,000 men—in 2009. So men were a Tanzania (Mbeya) �fth as likely to be victims of domestic abuse rural Bangladesh (Matlab) as women. In England and Wales, about two Peru (Cusco) in �ve domestic violence victims between 2004 Ethiopia (Butajira) Samoa and 2009 were men. Based on data from the Japan (Yokohama) British Crime Survey, about 4.0 percent of men Serbia (Belgrade) (4.8 percent of women) reported suffering part- Brazil (Sao Paulo) ner abuse in the past 12 months, an estimated 600,000 male victims.38 urban Namibia (Windhoek) Thailand (Bangkok) Tanzania (Dar es Salaam) Less likely to hold political of�ce Peru (Lima) Few nations have legal restrictions for women to Bangladesh (Dhaka) run for public of�ce, yet the number of women 0 10 20 30 40 50 60 70 holding parliamentary seats is very low, and % of women having suffered progress in the last 15 years has been slow. In physical or sexual violence 1995, women accounted for about 10 percent lifetime of members of the lower or single houses of na- last 12 months tional parliaments, and in 2009, 17 percent.39 In Africa and most of Asia, the number of female Source: World Health Organization 2005. parliamentarians more than doubled. Also dur- ing the last 15 years, the number of countries with at least 30 percent women as parliamen- or sexual abuse by an intimate partner within tarians rose from 5 to 23—including 7 from the last 12 months was most prevalent in Ethio- Sub-Saharan Africa as well as Argentina, Cuba, pia (Butajira) and Peru (Cusco), involving 54 Finland, Iceland, the Netherlands, and Sweden. and 34 percent of women respectively. At the Rwanda’s parliament has 56 percent women, up other extreme, Japan (Yokohama) and Serbia from 17 percent in 1995. East Asia registered the (Belgrade) are below the 4 percent mark (�g- least progress, and the number of women par- ure 2.11). In many instances, the violence can liamentarians in particular is low in the Paci�c be grave. In Peru (Cusco), almost 50 percent of Islands. women are victims of severe physical violence Although men and women are equally likely during their lifetime.36 And even with low in- to exercise their political voice by voting, men cidence, the numbers are unacceptably high. A are often perceived to be superior in exercising 3 percent domestic abuse incidence rate for Po- political power. Responses to the World Values land is equivalent to 534,000 women in one year, Surveys over several years point to a general or 1,463 new women a day.37 positive evolution of views on gender equality Domestic violence knows no boundaries, in politics in the last decade (�gure 2.12). But but incidence rates tend to rise with socioeco- people continue to view men as “better� politi- nomic deprivation. Reported across all eco- cal and economic leaders than women. nomic groupings, it is usually most prevalent And men have better chances than women of among economically disadvantaged women. winning an election. The likelihood of a female Women’s low social and economic status can be candidate winning a parliamentary seat over a both a cause and a consequence of domestic vi- man is estimated to be 0.87 (with 1 signifying olence. Low educational attainment, economic that men and women are equally likely to suc- duress, and substance abuse are among the many ceed in an electoral contest), with considerable compounding factors for abusive behavior. variation across countries. Women have greater The persistence of gender inequality 85 percent in 1998. Higher female participation in F I G U R E 2.12 Men are perceived as cabinet positions can be observed in every re- better political leaders gion, especially in Western Europe, Southern than women Africa, and Latin America and the Caribbean. In 1998, women occupied more than 20 percent Georgia of ministerial posts in only 13 countries, but 63 India countries 10 years later. In 2008, Chile, Finland, Russian Federation France, Grenada, Norway, South Africa, Spain, Turkey Sweden, and Switzerland had cabinets with Romania more than 40 percent female ministers.41 Ukraine China South Africa Chile REVERSALS Bulgaria Japan The gender disadvantages discussed in this Poland chapter form the backbone for the analysis Brazil in the remainder of the report—and the poli- Slovenia cies advanced to mitigate them. In contrast to Mexico these facets of gender relations, where there is Uruguay often a clear pathway for analysis, the gendered Peru impact of external shocks, which can gener- Sweden ate large losses in welfare and well-being, de- 0 10 20 30 40 50 60 70 80 90 100 pends on many speci�c circumstances. In some % of people who believe that cases, men or boys are worse affected; in others, men make better political leaders than women do women and girls. The impact of shocks can be highest value is past’s value and the lowest is today’s value temporary, with large losses in welfare after the shock followed by subsequent catch-up.42 But highest value is today’s value and the lowest is past’s value some shocks, especially when they hit early in life (as in the critical �rst three years) can also Source: WDR 2012 team estimates based on World Values Surveys, waves 1994–97 and 2005–07. have irreversible consequences. Outlined here is the multifaceted nature of external shocks and their impacts. The message: protection against chances of prevailing over men in Africa. In con- shocks should be a key part of any development trast, in Asia and high-income countries, wom- policy, and whether a gendered lens is appropri- en’s chances drop considerably. Women are least ate depends on context. likely to win elections in the Paci�c Islands and Whether the source is �nancial, political, or Latin America and the Caribbean.40 natural (box 2.1), shocks and hazards can affect Quotas and reservations have helped expand men and women differently, a function of their female parliamentary representation. Ninety distinct social roles and status. First, market fail- countries have some quota mechanism for par- ures, institutional constraints, and social norms liamentary representation, whether in seat reser- can amplify or mute gender differences in the vation, candidate quota legislation, or voluntary impact of shocks. The mechanisms that produce political party quotas. Sixteen countries, all in these outcomes are multiple. For example, the Africa and Asia, explicitly reserve parliamentary higher mortality rates for girls and women in seats for women. In others, such as Finland, there the 2004 Indian Ocean Tsunami have been re- has been little change. For instance, 38.5 percent lated to their more limited mobility caused by of newly elected members of parliament in Fin- restrictive clothing and caring for small chil- land in 1991 were women, and by 2011, this dren.43 Second, those failures, constraints, and share had increased marginally to 42.5 percent. norms can amplify or mute gender differences in Women’s participation in cabinets, regard- the vulnerability to shocks. The fact that women less of the structure and size, has also lagged. tend to own and control fewer assets than men, The proportion of women among ministers for example, can make them more vulnerable to was on average 17 percent in 2008, up from 8 expected income shocks. 86 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 economic contraction is 7.4 deaths per 1,000 for B OX 2.1 The many faces of climate change girls, �ve times the 1.5 for boys.45 With proper nourishment, older children and adults can usu- Climate change results in more frequent droughts and floods and more vari- ally compensate for nutritional de�cits during a able rainfall. A rising fraction of the world population is affected by climatic shock. shocks and natural disasters as a result of their greater frequencies and larger In contrast, the impact of economic crises numbers of people in disaster-prone areas. Cold days, cold nights, and frosts on associated risky behavior, morbidity, and have become less frequent, while the frequency and intensity of heat waves mortality tends to be greater for men. The sud- have increased. Both floods and droughts are occurring more often. The interi- den political and economic transformation in ors of continents have tended to dry out despite an overall increase in total precipitation. Globally, precipitation has increased, with the water cycle sped Eastern Europe fueled a sharp and unexpected up by warmer temperatures, even as the Sahel and Mediterranean regions have drop in male life expectancy. In many countries more frequent and more intense droughts. Heavy rainfall and floods have in the region, particularly the Russian Federa- become more common, and there is evidence that the intensity of storms and tion, men bear a greater share of the burden of tropical cyclones has increased. ill health. Premature male mortality has been Women appear more vulnerable in the face of natural disasters, with the overwhelmingly concentrated in the unmar- impacts strongly linked to poverty. A recent study of 141 countries found that ried population. Women outlive men by nine more women than men die from natural hazards. Where the socioeconomic status of women is high, men and women die in roughly equal numbers during years on average—a gap larger than in the rest and after natural hazards, whereas more women than men die (or die at a of world. The rise in male mortality is partly younger age) where the socioeconomic status of women is low. Women and related to increased risky behavior, including children are more likely to die than men during disasters. The largest numbers smoking and alcoholism.46 In Russia, a recent of fatalities during the Asian Tsunami were women and children under age 15. survey shows that 19 percent of men, but only By contrast, 54 percent of those who died in Nicaragua as a direct result from 1 percent of women, were classi�ed as problem Hurricane Mitch in 1998 were male. drinkers.47 Stress owing to the absence of eco- Erratic weather can also affect agricultural productivity, which can reduce the income and food of households. The reductions in food availability may not nomic activities (challenging traditional gender affect all household members equally. And temperature and precipitation fluc- roles of men as breadwinners) and weak family tuations may affect the prevalence of vector-borne, water-borne, and water- or social support networks are linked to the de- washed diseases, as well as determine heat or cold stress. clines in male health.48 Men and women may be affected differently by changing weather. House- When families experience an income shock, hold evidence from rural India and Mexico suggests that this may be indeed girls’ education suffers more than boys’ in some the case, but the impact and direction depend on the climatic shock and envi- countries but not in others. Girls in Turkey were ronmental context. In some locations in Mexico, rural girls can have lower height-for-age than boys after a positive rainfall shock or a negative tempera- more likely than boys to drop out of school in ture shock. Yet girls in high-altitude areas have higher height-for-age than boys response to lower household budgets.49 And in as a result of warmer weather. Indonesia, girls were more likely to be pulled out in response to crop losses.50 Vulnerability to ex- Sources: Jacoby, Sa�r, and Skou�as 2011; Neumayer and Plümper 2007; Skou�as, Vinha, and Conroy 2011; ternal shocks is particularly important because World Bank 2009. interruptions in schooling can increase the risk of dropping out, and lags and delays in school progression can have a permanent impact on Two areas where shocks can generate signi�- overall grade attainment. In Ethiopia, girls ages cant reversals are in education and health out- 7−14 are 69 percentage points more likely to be comes and in access to economic opportunities. in school if they attended school in the previ- ous period, and boys 21 percentage points more Reversals in education and health likely.51 The health of infant girls tends to fare worse But boys may also be pulled out of school as a result of negative income shocks. Under- during an economic shock, usually to bolster nutrition during gestation or infancy and de- household �nances. When low-skilled work op- clines in health-care-seeking behavior increase portunities are available, boys more than girls mortality and morbidity risks in later life. In are very likely to be used to complement dwin- India, the mortality of girls rises signi�cantly as dling family income. While Ethiopian boys have a result of macroeconomic crises, but boys ap- generally enjoyed greater access to schooling, pear to be better protected.44 A study of 59 de- in times of economic crisis they have also been veloping countries suggests similar results. The the �rst to be withdrawn from school to work.52 average increase in infant mortality during an And in Côte d’Ivoire, while enrollments of both The persistence of gender inequality 87 boys and girls fell in response to drought, boys’ bor force, while younger, more educated women enrollments fell more (chapter 3). from wealthier households often exit the labor Income shocks may have mixed effects in market in response to economic crises. The im- relation to endowment accumulation. In some pact of crises on women’s labor force participa- instances (mainly in middle- and high-income tion has often been the strongest in the informal economies), income shocks can actually bring or unregulated sectors of the economy, which boys and girls back into school. Wage reductions more readily absorb additional women in petty or poor labor market conditions in a crisis lower commerce or domestic service.58 the opportunity cost of schooling, inducing The direction of labor market impacts and households to keep their children in school—es- their gender differences depend on the nature pecially boys who are more likely to be engaged of the macroeconomic shock. Export-oriented in wage labor. In Nicaragua, the school participa- industries, such as light manufacturing, were tion of rural boys increased 15 percentage points the initial casualties in the most recent �nancial after a sudden drop in coffee prices around crisis. Higher female participation rates in these 2000–02.53 In Argentina, the deterioration in �elds led to strong �rst-round negative em- employment rates during the deep 1998–2002 ployment effects for women. But lower female �nancial crisis also increased the probability of participation rates in sectors that shrank in the boys attending school.54 crisis, such as construction, or industries depen- dent on external demand, such as tourism, im- Reversals in access to economic plied that the aggregate employment effects for opportunities women relative to men were muted once these Shocks can bring about reversals in economic second-round effects are taken into account. opportunities for both men and women. In Latvia, Moldova, Montenegro, and Ukraine, Women tend to hold more precarious jobs, op- men tended to lose their jobs more than women. erate smaller and less capitalized �rms, and be The sectors most affected by the crisis in those engaged in more vulnerable economic activities countries—such as construction and manufac- than men, suggesting that they would be more turing—tend to be male-dominated. likely to be affected by economic shocks. But the Similarly, during the Asian crisis of 1997, evidence does not support this oft-held view. female employment was not the hardest hit. In the recent �nancial crisis, there were no Women in East Asian nations were dispropor- common patterns by gender and no evidence tionately employed in �rms more resilient to the that women were more affected than men.55 Ev- crisis.59 But the gender earnings gap increased, idence from 41 middle-income countries sug- particularly in larger �rms. In other words, gests that the main impact was on the quality women’s smaller net employment impact came of employment (number of hours worked and at the expense of a larger reduction in their wages), rather than on the number of jobs.56 earnings. Gender impacts vary signi�cantly by country and defy simple generalizations.57 Both labor market entry (added workers) “STICKY� GETS “STICKIER� and exit (discouraged workers) during crises might operate simultaneously, affecting differ- Chapter 1 noted that changes are intercon- ent groups of women differently. Women from nected. Progress in one dimension of gender low-income households typically enter the la- equality can multiply the effects on another dimension. The same applies to an absence of change. A lack of progress in one dimension can compound the negative effects in another “ The �nancial situation in Moldova is very bad. I think that women should stay home and take care of the family. dimension. Gender differences can thus endure, bound together by many layers of constraints that reinforce one other. Breaking this impasse requires action on various strands of this web of � Adult man, Moldova persistent inequality. Maternal education is inversely correlated with infant and child mortality in developing 88 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 countries. In Mozambique, low maternal educa- The clustering of men and women in differ- tion is a strong predictor of low health service ent occupations and sectors begins earlier, in the use and child malnutrition.60 And stunted chil- educational system. While female participation dren—before 24 months of age—have poorer is increasing across all �elds of study as more psychomotor skills and lower cognitive achieve- women enter tertiary education, segregation by ment. Undernourished boys and girls are also area of specialization remains. Male bias is most less likely to be enrolled in school or enter school marked in engineering, manufacturing, and late, sustaining the cycle of deprivation.61 construction. In about two-thirds of the world’s Poor health of girls and women spills over countries, men also outnumber women in sci- into the next generation. Both contemporary ence. But in nine-tenths of the world, women and childhood health of the mother matter outnumber men in education; humanities and for the health of the next generation. Maternal arts; social sciences, business, and law; and well-being—measured through short stature, health and welfare. low body mass index, and anemia—affects Educational segregation by specialization size at birth, survival, and child growth. So un- does not go away—and even appears to in- derinvesting in the health of girls and women crease—with economic development. Cam- contributes to child mortality and intergenera- bodia, Lao People’s Democratic Republic, Mo- tional cycles of poor health among survivors.62 rocco, and Namibia are among the countries Girls who are born small and do not catch up with the least gender segregation by study areas, in growth fail to attain the height predicted by though men are more likely to obtain a tertiary their genetic potential. Their reduced uterine degree. Among Organisation for Economic and ovarian size implies lower birth weights for Co-operation and Development (OECD) coun- their offspring, engendering a new intergenera- tries, Turkey has the least gender segregation in tional cycle of deprivation.63 tertiary �elds of study, while Croatia, Finland, Lack of property ownership and control mat- Japan, and Lithuania have the most. In Norway ters for women’s agency. Assets are an important and Denmark, women make up two-thirds of tertiary enrollments, but only a third of science element to boost voice and bargaining power students is female.68 in household decision making, access to capi- These are just several examples, among tal, and overall economic independence. Nepali many, of how constraints in one aspect of gen- women who do not own land have less say in der equality can hold back progress on other di- household decision making than women with mensions, causing gender inequality to persist. land.64 In Colombia, a lack of property or so- This persistence comes with large economic, cial assets constrains women in negotiating for social, and political costs. Part 2 of the Report the right to work, controlling their own income, analyzes the foundations of these persistent gen- moving freely, and contesting domestic abuse.65 der disparities, rooting them in the interactions In Kerala, India, women’s independent owner- between households, markets, and formal and ship of immovable property is a signi�cant pre- informal institutions. dictor of long-term physical and psychological domestic violence, over and above the effects of other factors. The odds of being beaten if a NOTES woman owns both a house and land are a twen- tieth of those when she owns neither.66 1. UNESCO 2010. Cash employment is also strongly associated 2. Yuki and others 2011. with women’s empowerment. Not earning a 3. UNESCO Institute for Statistics 2010. cash income is most consistently associated with 4. Berry 1977; Högberg and Wall 1986. married women not making decisions—on top- 5. USAID 2006. 6. Lewis and Lockheed 2006. ics such as their health care, large household 7. Chioda, Garcia-Verdú, and Muñoz Boudet 2011. purchases, purchases for daily needs, and visits 8. Hannum and Wang 2010. to family or friends. Higher household wealth 9. Hannum 2005. by itself does not consistently enhance the likeli- 10. Perkins and others 2004. hood that women will make decisions alone or 11. Vincent-Lancrin 2008. jointly in most countries.67 12. Jha and Kelleher 2006. The persistence of gender inequality 89 13. See chapter 3, technical annex. 63. Ibañez and others 2000. 14. Sen 1999. 64. Allendorf 2007. 15. AbouZahr 2003. 65. Friedemann-Sánchez 2006. 16. WHO and others 2010. 66. Agarwal and Panda 2007. 17. Ibid. 67. Kishor and Subaiya 2008. 18. The estimate could be affected where births of 68. UNESCO 2009. girls go unreported. 19. Jha and others 2006. 20. Guilmoto 2009. REFERENCES 21. Abrevaya 2009; Almond, Edlund, and Milligan 2009; Dubuc and Coleman 2007. The word processed describes informally reproduced 22. Chung and Das Gupta 2007. works that may not be commonly available through 23. Bosch and Maloney 2010; Gong and van Soest libraries. 2002. 24. Beneria 2001. AbouZahr, Carla. 2003. “Global Burden of Maternal 25. Bruhn 2009; Mead and Liedholm 1998. Death and Disability.� British Medical Bulletin 67 26. Chioda, Garcia-Verdú, and Muñoz Boudet 2011. (1): 1–11. 27. UN DESA 2010. Abrahams, Naeemah, Rachel Jewkes, Lorna J. Martin, 28. Banerjee and Duflo 2007. Shanaaz Mathews, Lisa Vetten, and Carl Lombard. 29. Nyamu-Musembi 2002. 2009. “Mortality of Women From Intimate Part- 30. Deere and Doss 2006. ner Violence in South Africa: A National Epide- 31. Katz and Chamorro 2003. miological Study.� Violence and Victims 24 (4): 32. In addition, the absence of longitudinal data 546–56. prevents analyzing how trends have evolved over Abrevaya, Jason. 2009. “Are There Missing Girls in the time. United States? Evidence from Birth Data.� Applied 33. Jewkes and others 2002. Economics 1 (2): 1–34. 34. Abrahams and others 2009. Agarwal, Bina, and Pradeep Panda. 2007. “Toward 35. Reed and others 2010. Freedom from Domestic Violence: The Neglected 36. United Nations Department of Economic and Obvious.� Journal of Human Development and Ca- Social Affairs 2010. pabilities 8 (3): 359–88. 37. UN DESA 2010. Allendorf, Keera. 2007. “Do Women’s Land Rights 38. Home Of�ce British Government 2009. Promote Empowerment and Child Health in Ne- 39. UN DESA 2010. pal?� World Development 35 (11): 1975–88. 40. Ibid. Almond, Douglas, Lena Edlund, and Kevin Milligan. 41. Ibid. 2009. “O Sister, Where Art Thou? The Role of Son 42. Dercon 2011. Preference and Sex Choice: Evidence from Immi- 43. Nishikiori and others 2006. grants to Canada.� Working Paper Series 15391, 44. Bhalotra 2010. Nartional Bureau of Economic Research, Cam- 45. Baird, Friedman, and Schady 2007. bridge, MA. 46. World Bank 2007. Baird, Sarah, Jed Friedman, and Norbert Schady. 2007. 47. Bobrova and others 2010. “Aggregate Income Shocks and Infant Mortality in 48. Doyal 2000. the Developing World.� Review of Economics and 49. Tansel 2002. Statistics 93 (3): 847–56. 50. Cameron and Worswick 2001. Banerjee, Abhijit Vinayak, and Esther Duflo. 2007. 51. Mani, Hoddinott, and Strauss 2010. “The Economic Lives of the Poor.� Journal of Eco- 52. Rose and Al-Samarrai 2001. nomic Perspectives 21 (1): 141–67. 53. Maluccio 2005. Beneria, Lourdes. 2001. Changing Employment Pat- 54. López Bóo 2010. terns and the Informalization of Jobs: General 55. Habib and others 2010. Trends and Gender Dimensions. Geneva: Interna- 56. Khanna, Newhouse, and Paci 2010. tional Labor Of�ce. 57. Cho and Newhouse 2011; Turk and Mason 2009. Berniell, María Inés, and Carolina Sánchez-Páramo. 58. Sabarwal, Sinha, and Buvinic 2010. 2011. “Overview of Time Use Data Used for the 59. Hallward-Driemeier, Rijkers, and Waxman 2010. Analysis of Gender Differences in Time Use Pat- 60. Lindelow 2008. terns.� Background paper for the WDR 2012. 61. Naudeau and others 2011; Victora and others Berry, Linda G. 1977. “Age and Parity Influences on 2008. Maternal Mortality: United States, 1919–1969.� 62. Bhalotra and Rawlings 2011. Demography 14 (3): 297–310. 90 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Bhalotra, Sonia. 2010. “Fatal Fluctuations? Cyclicality Gong, Xiaodong, and Arthur van Soest. 2002. “Wage in Infant Mortality in India.� Journal of Develop- Differentials and Mobility in the Urban Labor ment Economics 93 (1): 7–19. Market: A Panel Data Analysis for Mexico.� Labour Bhalotra, Sonia, and Samantha B. Rawlings. 2011. “In- Economics 9 (4): 513–29. tergenerational Persistence in Health in Develop- Guilmoto, Christophe Z. 2009. “The Sex Ratio Transi- ing Countries: The Penalty of Gender Inequality?� tion in Asia.� Population and Development Review Journal of Public Economics, 95(3–4): 286–99. 35 (3): 519–49. Bobrova, Natalia, Robert West, Darya Malyutina, Habib, Bilal, Ambar Narayan, Sergio Olivieri, and Sophia Malyutina, and Martin Bobak. 2010. “Gen- Carolina Sanchez-Paramo. 2010. “Assessing Ex- der Differences in Drinking Practices in Middle ante the Poverty and Distributional Impact of the Aged and Older Russians.� Alcohol and Alcoholism Global Crisis in a Developing Country: A Micro- 45 (6): 573–80. simulation Approach with Application to Bangla- Bosch, Mariano, and William F. Maloney. 2010. desh.� Policy Research Working Paper Series 5238, “Comparative Analysis of Labor Market Dynam- World Bank, Washington, DC. ics Using Markov Processes: An Application to In- Hallward-Driemeier, Mary, Bob Rijkers, and Andrew formality.� Labour Economics 17 (4): 621–31. Waxman. 2011. “Ladies First? Firm-level Evidence Bruhn, Miriam. 2009. “Female-Owned Firms in Latin on the Labor Impacts of the East Asian Crisis.� America. Characteristics, Performance, and Ob- World Bank, Washington, DC. Processed. stacles to Growth.� Policy Research Working Paper Hannum, Emily. 2005. “Market Transition, Educa- Series 5122, World Bank, Washington, DC. tional Disparities, and Family Strategies in Rural Cameron, Lisa A., and Christopher Worswick. 2001. China: New Evidence on Gender Strati�cation “Education Expenditure Responses to Crop Loss and Development.� Demography 42 (2): 275–99. in Indonesia: A Gender Bias.� Economic Develop- Hannum, Emily, and Meiyan Wang. 2010. “China: A ment and Cultural Change 49 (2): 351–63. Case Study in Rapid Poverty Reduction.� In “In- digenous Peoples, Poverty and Development,� Chioda, Laura, Rodrigo Garcia-Verdú, and Ana María ed. Gillette Hall and Harry Patrinos, World Bank, Muñoz Boudet. 2011. Work and Family: Latin Washington, DC. Processed. American Women in Search of a New Balance. Högberg, Ulf, and Stig Wall. 1986. “Age and Parity as Washington, DC: World Bank. Determinants of Maternal Mortality—Impact of Cho, Yoonyoung, and David Newhouse. 2011. “How Their Shifting Distribution among Parturients in Did the Great Recession Affect Different Types Sweden from 1781 to 1980.� Bulletin of the World of Workers? Evidence from 17 Middle-Income Health Organization 64 (1): 85–91. Countries.� IZA Discussion Paper Series 5681, In- Home Of�ce British Government. 2009. “Crime in stitute for the Study of Labor, Bonn. England and Wales 2008/2009, Volume 1—Find- Chung, Woojin, and Monica Das Gupta. 2007. “The ings from the British Crime Survey and Police Re- Decline of Son Preference in South Korea: The corded Crime.� In Home Of�ce Statistical Bulletin, Roles of Development and Public Policy.� Popula- ed. A. Walker, J. Flatley, C. Kershaw, and D. Moon. tion and Development Review 33 (4): 757–83. London: Home Of�ce British Government. Deere, Carmen Diana, and Cheryl R. Doss. 2006. Ibañez, Lourdes, Neus Potau, Goya Enriquez, and “Gender and the Distribution of Wealth in Devel- Francis de Zegher. 2000. “Reduced Uterine and oping Countries.� Research Paper Series 2006/115, Ovarian Size in Adolescent Girls Born Small for United Nations University–World Institute for Gestational Age.� Pediatric Research 47 (5): 575–77. Development Economics Research, Helsinki. ILO (International Labor Organization). 2010. “Key Dercon, Stefan. 2011. “Young Lives: Are There Gender Indicators of the Labour Market.� ILO, Geneva. Differences in Children’s Outcomes?� Background Jacoby, Hanan G., Abla Sa�r, and Emmanuel Skou�as. paper for the WDR 2012. 2011. “Climate Variability and Child Health in In- Doyal, Lesley. 2000. “Gender Equity in Health: De- dia.� World Bank, Washington, DC. Processed. bates and Dilemmas.� Social Science & Medicine Jewkes, Rachel, Jonathan Levin, Nolwazi Mbananga, 51 (6): 931–9. and Debbie Bradshaw. 2002. “Rape of Girls in Dubuc, Sylvie, and David Coleman. 2007. “An In- South Africa.� Lancet 359 (9303): 319–20. crease in the Sex Ratio of Births to India-Born Jha, Jyotsna, and Fatimah Kelleher. 2006. Boys’ Un- Mothers in England and Wales: Evidence for Sex- derachievement in Education: An Exploration Selective Abortion.� Population and Development in Selected Commonwealth Countries. London: Review 33 (2): 383–400. Commonwealth Secretariat, Gender Section, and Friedemann-Sánchez, Greta. 2006. “Assets in Intra- Commonwealth of Learning. household Bargaining among Women Workers in Jha, Prabhat, Rajesh Kumar, Priya Vasa, Neeraj Dhin- Colombia’s Cut-Flower Industry.� Feminist Eco- gra, Deva Thiruchelvam, and Rahim Moineddin. nomics 12 (1–2): 247–69. 2006. “Low Male-to-Female Sex Ratio of Children The persistence of gender inequality 91 Born in India: National Survey of 1.1 Million Nishikiori, Nobuyuki, Tomoko Abe, Dehiwala G. M. Households.� Lancet 367 (9506): 211–18. Costa, Samath D. Dharmaratne, Osamu Kuni, and Katz, Elizabeth G., and Juan Sebastian Chamorro. Kazuhiko Moji. 2006. “Who Died as a Result of 2003. “Gender, Land Rights, and the Household the Tsunami?: Risk Factors of Mortality among Economy in Rural Nicaragua and Honduras.� Internally Displaced Persons in Sri Lanka: A Ret- Paper presented at the Annual Conference of the rospective Cohort Analysis.� BMC Public Health 6 Latin American and Caribbean Economics Asso- (1): 73–81. ciation, Puebla, Mexico, October 9. Nyamu-Musembi, Celestine. 2002. “Are Local Norms Khanna, Gaurav, David Newhouse, and Pierella Paci. and Processes Fences or Pathways? The Example 2010. “Fewer Jobs or Smaller Paychecks? Labor of Women’s Property Rights in Rural Kenya.� In Market Impacts of the Recent Crisis in Middle- Cultural Transformations and Human Rights in Income Countries.� Poverty Reduction and Net- Africa, ed. Abdullahi A. An-Na’im. London: Zed Books. work Economic Management, Economic Premise, Parish, William L., Tianfu Wang, Edward O. Lau- April 2010, No. 11, World Bank, Washington, DC. mann, Suiming Pan, and Ye Luo. 2004. “Intimate Kishor, Sunita, and Lekha Subaiya. 2008. “Understand- Partner Violence in China: National Prevalence, ing Women’s Empowerment: A Comparative Risk Factors and Associated Health Problems.� Analysis of Demographic and Health Surveys International Family Planning Perspectives 30 (4): (DHS) Data.� DHS Comparative Reports 20, 174–81. MacroInternational, Calverton, MD. Perkins, Robert, Brian Kleiner, Stephen Roey, and Lewis, Maureen A., and Marlaine E. Lockheed. 2006. Janis Brown. 2004. The High School Transcript Inexcusable Absence: Why 60 Million Girls Aren’t in Study: A Decade of Change in Curricula and School and What to Do About It. Washington, DC: Achievement, 1990–2000. Washington, DC: Na- Center for Global Development. tional Center for Education Statistics, U.S. De- Lindelow, Magnus. 2008. “Health as a Family Mat- partment of Education. ter: Do Intra-household Education Externalities Reed, Elizabeth, Anita Raj, Elizabeth Miller, and Matter for Maternal and Child Health?� Journal of Jay G. Silverman. 2010. “Losing the ‘Gender’ in Development Studies 44 (4): 562–85. Gender-Based Violence: The Missteps of Research López Bóo, Florencia. 2010. “In School or at Work? on Dating and Intimate Partner Violence.� Vio- Evidence from a Crisis.� IZA Discussion Paper Se- lence against Women 16 (3): 348–54. ries 4692, Institute for Labor Studies, Bonn. Rose, Pauline, and Samer Al-Samarrai. 2001. “House- Maluccio, John A. 2005. “Coping with the ‘Coffee Cri- hold Constraints on Schooling by Gender: Empir- sis’ in Central America: The Role of the Nicara- ical Evidence from Ethiopia.� Comparative Educa- guan Red de Protección Social.� FCND Discussion tion Review 45 (1): 36–63. Paper 188, International Food Policy Research Sabarwal, Shwetlena, Nishta Sinha, and Mayra Bu- Institute, Food Consumption and Nutrition Divi- vinic. 2010. “The Global Financial Crisis: As- sion, Washington, DC. sessing Vulnerability for Women and Children.� Mani, Subha, John Hoddinott, and John Strauss. World Bank, Washington, DC. Processed. 2010. “Long-Term Impact of Investments in Early Sen, Amartya. 1999. Development as Freedom. New Schooling: Empirical Evidence from Rural Ethio- York: Alfred A. Knopf. Skou�as, Emmanuel, Katja Vinha, and Hector V. pia.� Discussion Paper Series 00981, International Conroy. 2011. “The Impacts of Climate Variabil- Food Policy Research Institute, Washington, DC. ity on Welfare in Rural Mexico.� Policy Research Mead, Donald C., and Carl Liedholm. 1998. “The Working Paper Series 5555, World Bank, Wash- Dynamics of Micro and Small Enterprises in De- ington, DC. veloping Countries.� World Development 26 (1): Tansel, Aysit. 2002. “Determinants of School Attain- 61–74. ment of Boys and Girls in Turkey: Individual, Naudeau, Sophie, Naoko Kataoka, Alexandria Val- Household and Community Factors.� Economics erio, Michelle J. Neuman, and Leslie Kennedy of Education Review 21 (5): 455–70. Elder. 2011. Investing in Young Children: An Early Turk, Carolyn, and Andrew Mason. 2009. “Impacts of Childhood Development Guide for Policy Dialogue the Economic Crisis in East Asia: Findings from and Project Preparation. Washington, DC: World Qualitative Monitoring in Five Countries.� Pa- Bank. per presented at the Third China-ASEAN Forum Neumayer, Eric, and Thomas Plümper. 2007. “The on Social Development and Poverty Reduction, Gendered Nature of Natural Disasters: The Impact Fourth ASEAN+3 High-level Seminar on Poverty of Catastrophic Events on the Gender Gap in Life Reduction, and Asia-wide Regional High-level Expectancy, 1981–2002.� Annals of the Association Meeting on the Impact of the Global Economic of American Geographers 97 (3): 551–66. Slowdown on Poverty and Sustainable Develop- 92 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 ment in Asia and the Paci�c Meeting, Hanoi, Sep- going Trend.� In Higher Education to 2030 Volume tember 28. 1: Demography, ed. Organisation for Economic UNESCO (United Nations Educational, Scienti�c, Co-operation and Development (OECD). Paris: and Cultural Organization). 2009. Global Educa- OECD Publishing. tion Digest 2009: Comparing Education Statistics World Bank. 2007. Global Monitoring Report 2007: Across the World. Paris: UNESCO. Millenium Development Goals, Confronting the ——–—. 2010. EFA Global Monitoring Report: Reach- Challenges of Gender Equality and Fragile States. ing the Marginalized. Oxford, U.K.: Oxford Uni- Washington, DC: World Bank. versity Press for UNESCO. ——–—. 2009. World Development Report 2010: De- UNESCO Institute for Statistics. 2010. Global Educa- velopment and Climate Change. Washington, DC: tion Digest 2010: Comparing Education Statistics World Bank. across the World. Special Focus on Gender. Mon- WHO (World Health Organization). 2005. WHO treal: UNESCO Institute for Statistics. Multi-country Study on Women’s Health and Do- UN DESA (United Nations Department of Economic mestic Violence against Women: Initial Results on and Social Affairs). 2009. World Population Pros- Prevalence, Health Outcomes and Women’s Re- pects 2009. New York: United Nations. sponses. Geneva: WHO. ——–—. 2010. The World’s Women 2010: Trends and ——–—. 2010. World Health Statistics 2010. Geneva: Statistics. New York: United Nations. WHO. USAID (U.S. Agency for International Develop- WHO (World Health Organization), UNICEF ment). 2006. Women’s Lives and Experiences: (United Nations Children’s Fund), UNFPA Changes in the Past Ten Years. Washington, DC: (United Nations Population Fund), and World USAID. Bank. 2010. Trends in Maternal Mortality: 1990 to Victora, Cesar, Linda Adair, Caroline Fall, Pedro Hal- 2008. Geneva: WHO. lal, Reynaldo Martorell, Linda Richter, and Harsh- Yuki, Takako, Keiko Mizuno, Keiichi Ogawa, and pal Sachdev. 2008. “Maternal and Child Under- Mihoko Sakai. 2011. “Promoting Gender Parity, nutrition: Consequences for Adult Health and Lessons from Yemen: A JICA Technical Coop- Human Capital.� Lancet 371 (9609): 340–57. eration Project in Basic Education.� Japan Inter- Vincent-Lancrin, Stéphan. 2008. “The Reversal of national Cooperation Agency Research Institute. Gender Inequalities in Higher Education: An On- Background paper for the WDR 2012. 94 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 SPREAD 1 Women’s pathways to empowerment: Do all roads lead to Rome? W hat gives women power? Conversations with almost 2,000 women across 19 countries SPREAD FIGURE 1.1 Main factors in moving show that they depend on a combination of up the ladder factors to feel empowered.1 “Increased con�dence to man- age the house independently,� “more communication with occupational and neighbors and community members,� the ability to “go out economic change of house to do marketing, shopping, and other household financial management work such as paying electricity and water bills,� “increased control of �nancial transactions in and outside the house,� education and training and “husband’s support and permission to go outside of the social networks house� were the main answers of women in Bhubane- individual swar, India, when asked to describe what it meant to be characteristics/behaviors powerful and free. Similarly, rural women from Paro, marital or family Bhutan, associated gains in power with education, status change spouse’s and family members’ support, and hard work— psychological traits but also with education programs for women who have marital relations missed school and with role models such as elected fe- male community leaders who “have helped women think social assistance better� and female small business owners who have pros- pered and boosted the con�dence of the women in their leadership community. illegal activities The characteristics of a powerful woman that come up most often are related to generating and managing 0 10 20 30 40 50 60 70 80 90 100 income, followed closely by acquiring an education, and % of mentions by communities then by personal traits and access to social networks urban women rural women (spread �gure 1.1). It is also clear that no single factor can explain changes Source: WDR 2012 team calculations, based on “De�ning Gender in the 21st in empowerment. Any one factor may be present for Century: A Multi-Country Assessment� (dataset). many women with different levels of power and may even determine gains or losses in power, depending on ent steps on a �ctional “ladder of power and freedom�— other factors operating in women’s lives. For example, with the top step for women with the most power and the changes in marital and family conditions create oppor- bottom for those with the least. They were also asked to tunities for some women when “the husband supports his repeat this ranking to reflect where the women would wife� (Papua New Guinea), or if they “get a good and un- have been on the ladder 10 years ago. In 79 percent of derstanding husband who can allow her to do business and cases, women saw a dramatic upward movement in the engage in educational activities� (Tanzania). And even a past 10 years (almost 20 percentage points larger than divorce can be positive. “Divorce can free a woman from men’s perception of their gains in power in the same pe- a lot of strains and she’ll become stronger,� recognized riod). But that was not so in all cases. A community in women in West Bank and Gaza. But for other women rural South Africa saw 80 percent of its women as being the same process can have the opposite effect: “a woman at the bottom of the ladder, “All of us here are struggling, can fall [can lose power] if she loses her husband, her chil- so we have little power, and we are not free to do what we dren, or the support of her parents� (Burkina Faso). “If you want to do because we do not have money,� explained one have three children and your husband dies, a single income woman. They mentioned not having savings and having would not be suf�cient� (Peru). “[A] divorce when the man dif�culty purchasing basic goods, “What can they possibly leaves the wife it’s even worse than death for her� (Poland). save, because whatever little money they have they spend on Women’s pathways to empowerment are determined food. It is very dif�cult to think about savings if you hardly by different combinations of factors. To trace such path- get money and you are always hungry because the little you ways, women in each country were asked to place 100 might get you want it to make your children happy at least representative women from their community on differ- for that day,� said another woman. They also pointed to Women’s pathways to empowerment: Do all roads lead to Rome? 95 the daunting number of people suffering from HIV/ freedom in relation to other constraints and barriers that AIDS. A powerless woman “is often sick, her health is un- women were facing. stable, and she cannot even access health facilities because A higher national HDI or a low prevalence of domes- the clinic is very far and she does not have money for trans- tic violence was suf�cient to explain women’s empower- port,� and “her husband is likely to be sick.� ment gains in half the communities across all countries. In the Dominican Republic’s capital city, Santo Do- Either factor by itself counters negative conditions, such mingo, women reported fast upward movement on as restricted mobility or lack of jobs. their ladders thanks to two factors: “now women study For rural women, participation in social networks— more and work more.� In Afghanistan’s Jabal Saraj, where organizations and networks for women in the commu- women placed 60 of the 100 women on the top step, nity, their relevance in the community, and the presence twice the number of 10 years before, a larger combina- of female leaders—was a key factor. Higher social capital tion of conditions was identi�ed: “In the past, women and network presence countered obstacles like domestic did just home chores like cooking at home and warming violence in 25 percent of the communities. Social capital the oven, but now there are possibilities such as gas and is the only factor that allows women to feel empowered electricity.� “Now some women have jobs out of the house even when facing high levels of domestic violence in their and most of the girls are going to school.� And women have communities. In its absence, women have to increase “participated in election as candidates for provincial coun- their agency on many aspects—freedom of movement, cil and others.� control over family formation, and control of assets—to Each community had its own stories to explain counter the disempowering force of domestic violence. changes in women’s power, but many elements were the Restrictions in any of the other agency conditions are less same from community to community. To understand binding if domestic violence incidence is low. the main commonalities and combinations of factors Urban women depend as much on the local structure driving female empowerment, a comparative qualitative of opportunities—availability of jobs for women and analysis combining dimensions of agency with the struc- a dynamic labor market—as they do on social capital. ture of opportunities in the community and the national In fact, when both come together, lack of control over human development level was conducted.2 The variables family formation or high incidence of domestic violence included: becomes less of an obstacle. When social capital is not strong, the relevance of the local structure of oppor- • Dimensions of agency. Women’s control over assets, tunities increases, but it needs to be paired with other control over family formation, freedom from domes- positive gains in agency such as increased control over tic violence, freedom of physical mobility, and bridg- assets or freedom of movement and of violence threats ing social capital—from community networks to for women to move up their power ladder. family support and friends. These different combinations show that pathways • Speci�c characteristics of the community environment may vary, but some combinations drive women’s gains or structure of opportunities. Informal institutions faster and better. The effect of any factor is likely to de- (level of pressure to conform to gender norms and pend on the con�guration of other factors—the role of positive/negative vision of gender norms); formal in- economic opportunities will depend on each woman’s stitutions (presence of services in the community ability to move freely as well as on asset ownership and such as transport, schools, health, electricity, and wa- social capital. ter); and economic opportunities for women and What do these pathways look like? Two examples markets (availability of jobs and share of women from two communities: working in the community). “A woman who is powerful is called omukazi [powerful woman]. I think most of us here are powerful women,� says • General national context for human development— Joyce in Bukoba, Tanzania, after acknowledging that the measured by the country’s score on the United Na- lives of women in her community have prospered, “Yes, tions Development Programme’s Human Develop- women have always moved up. I was married, and I really ment Index (HDI). suffered with my husband. When I left him, it is when I Various combinations of these factors were tested started doing my things and I am now very �ne: I can get to distinguish the necessary conditions for women’s what I want; I can do what I want; I take my children to empowerment in each case, the common explanations school� (spread �gure 1.2). across countries and cultures, and the factors that were For women in Bukoba, social capital has been the key suf�cient by themselves to explain gains in power and element. The community has a good array of organiza- 96 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 S P R E A D F I G U R E 1 .2 Ladder shares now and SPREAD FIGURE 1.3 Ladder shares now 10 years ago in Bukoba, and 10 years ago in Tanzania (urban) Dhamar, Republic of Yemen (rural) 100 100 90 90 80 80 70 70 60 percent 60 percent 50 50 40 40 30 30 20 20 10 10 0 0 women now women 10 men now men 10 women now women 10 men now men 10 years ago years ago years ago years ago top step middle step bottom step top step middle step bottom step Source: WDR 2012 team calculations, based on “De�ning Gender in the 21st Source: WDR 2012 team calculations, based on “De�ning Gender in the 21st Century: A Multi-Country Assessment� (dataset). Century: A Multi-Country Assessment� (dataset). tions helping women, and 33 percent of the local council by their husband every day and they are there. When you members are female. Women recognize the value of their talk to them, they say they are married and they cannot networks: “We have these self-help groups, we meet there separate. These women will never climb the ladder; they and talk about so many things that happen in our lives. You will stay at the bottom.� get advice from your fellow woman on how to deal with In rural Dhamar in the Republic of Yemen, women men who disturb you in your family or how to do business.� also see themselves moving up despite low economic “Once you get the right group and listen to what they say, participation and education in their community (spread you get up.� Friends also have a central role: “These are the �gure 1.3). Like the Bukoba women, their pathway in- people you count on as woman. They will give you advice. cludes social capital—in this case in the shape of informal They form part of the groups you belong to. And when you networks—as well as some gains in education, all in an have quarrels with your husband, they will be the �rst ones environment with too few opportunities and some mo- to come and assist. If you have small children, they will al- bility restrictions. “Men can �nish their education; men ways keep an eye on your children.� have the freedom to go out and to learn [but] women visit But social capital is not enough to increase women’s each other in their free time and chat,� said Fatima and empowerment. It needs to be combined with a good Ghalya when comparing their happiness with men’s. In structure of opportunities and services, which women in their community, a powerful woman would have many Bukoba also seem to have: “You can go to the factory and acquaintances and friends, while a woman with little ask for different kinds of jobs, such as sweeping, cleaning power “is the woman who doesn’t have influence among dishes or the compound, and even becoming their agent to Dhamar’s women.� buy and bring �sh to the factory.� Favorable laws are also Women in Dhamar see two factors as the most press- in place: “It is easy because the law is clear. No one loses as ing: having an education, and being able to move more such: you want to divorce, you divide the property, and each freely. Having freedom means having the ability to move one goes his or her way.� within the village. But most women cannot travel out- Women in Bukoba also recognize the barriers that side the village without appropriate companionship: prevent other women in the community from gaining “A woman cannot work outside the village unless she has power and freedom: “The moment you know that you can Mahram (male legal guardian) with her.� Gaining mobil- do things by yourself and not depend on a man is the mo- ity will allow them to �nish their education and those ment you begin going up. I see some women being beaten who dropped out of school to resume it. “If there is Women’s pathways to empowerment: Do all roads lead Overview to Rome? 97 transportation, they will let me learn, and I can become a tries. For further information, the assessment methodology teacher,� says a young woman. Job aspirations are linked can be found at http://www.worldbank.org/wdr2012. to mobility restrictions: “Work opportunities are limited 2. Fuzzy-set qualitative comparative analysis (fs/QCA). For inside the village except teaching, and recruitment for men references, see Ragin (2008) and Ragin (2000). The tech- nique allows for testing models of different pathways to in agricultural work.� Although these Yemeni women achieve an end, in this case, the levels of empowerment experience many dif�culties, they nonetheless see im- reported by the women in the various community groups. provements in their power and freedom over the previ- Given the nature of qualitative data—textual and repre- ous generations. sentative of individuals’ voices, perceptions, and experi- ences—comparing across countries and communities is done by measuring different degrees in the cases that �t NOTES each model (membership degrees). 1. The study economies include Afghanistan, Bhutan, Burkina Faso, the Dominican Republic, Fiji, India, Indonesia, Libe- ria, Moldova, North Sudan, Peru, Papua New Guinea, Po- REFERENCES land, Serbia, South Africa, Tanzania, Vietnam, West Bank and Gaza, and the Republic of Yemen. The focus groups in- Ragin, Charles. 2000. Fuzzy Set Social Science. Chicago: Chicago cluded male adults, female adults, male youth, female youth, University Press. male adolescents, and female adolescents; the adolescent ———. 2008. Redesigning Social Inquiry: Fuzzy Sets and Be- groups were conducted only in a subset of 8 of the 19 coun- yond. Chicago: Chicago University Press. PART II What has driven progress? What impedes it? P art 2 of this World Development Report explains why progress toward gender equality has been made in some areas but not in others. For this purpose, we use a conceptual framework positing that households, markets, and institutions (both formal and informal) and the interac- institutions to determine gender outcomes, and how policy can affect these interactions and ultimately gender outcomes. Understanding how households make decisions Households make decisions about how many children to have tions among them shape the relationship between economic and when to have them, how much to spend on education development and gender equality and that, in doing so, they and health for daughters and sons, how to allocate different ultimately determine gender outcomes (see overview box 4). tasks (inside and outside the household), and other matters This framework builds on earlier work within the World Bank that determine gender outcomes. These choices are made on and elsewhere. the basis of the preferences, decision-making (or bargaining) We apply this framework to analyze the evolution of gender power, and incentives and constraints of different household differences in endowments (chapter 3), agency (chapter 4), and members. access to economic opportunities (chapter 5) as well as the im- Preferences can be innate or shaped by gender roles, so- pact of globalization on gender equality, with a focus on access cial norms, and social networks (grouped under the label to economic opportunities (chapter 6). informal institutions). And they may be shared by or dif- This empirical approach helps establish the link between fer across individuals within the household. For instance, analysis and policy choices. It emphasizes that the design of evaluations of transfer programs, such as pensions and con- policies to reduce speci�c gender gaps needs to take into account ditional cash transfers, show that spending decisions differ what happens in the household, in the functioning and structure depending on whether money is given to women or men of markets, and in formal and informal institutions as well as in within the household, suggesting differences in men’s and the interactions between all of them. When these aspects are not women’s preferences.1 considered, the intended results of policy interventions can be Bargaining capacity is determined by two distinct sources muted or even contrary to expectations. of individual power: economic and social. Economic bargain- ing power depends primarily on the wealth and assets each individual controls and his or her contribution to total house- EXPLAINING THE FRAMEWORK hold income, while social bargaining power results primarily from formal and informal institutions. For instance, a wom- The proposed framework builds on three premises. First, the an’s capacity to decide how to allocate her time across various household is not a unitary block with a common set of pref- activities in and outside the household may be weakened by erences and goals. Instead, it is made up of different members her lack of asset ownership (low economic bargaining power) with their own preferences, needs, and objectives, as well as dif- or by strong social proscriptions against female work outside ferent abilities to influence decision making in the household. the home (low social bargaining power). Second, markets and institutions affect the relationship between In addition, economic and social bargaining power is in- economic development and gender equality both directly and fluenced by an individual’s capacity to take advantage of op- indirectly (through their impact on household decisions). Third, portunities outside the household and to bear the costs of markets and institutions are not static but are shaped and con- leaving the household (“exit�). For example, where custody of ditioned by society (understood as the sum of individuals and children in the case of divorce is based primarily on �nancial households). And it is precisely this process that allows markets capacity to care for the children, women’s capacity (and will- and institutions, including social norms and values, to evolve ingness) to use the threat of exit as a bargaining tool may be over time in response to policy interventions or exogenous limited by their own incomes and assets. changes brought about by, say, globalization. Finally, incentives and constraints are largely influenced Building on these three ideas, the framework captures how by markets and institutions. Markets determine the returns households make decisions, how they interact with markets and to household decisions and investments and thus provide 99 100 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 incentives for the allocation of limited resources to compet- But markets and institutions can also dampen the impacts ing ends. For instance, the returns to education in the labor of growth for many gender outcomes, which means that some market—both in better employment opportunities and higher gender biases do not disappear with economic development— wages—will influence decisions to send a girl or boy to school. the “sticky domains� discussed in Part 1. Persistent differences Similarly, the level of female wages will influence whether in the jobs that men and women perform provide a good ex- women will devote more time to market work, at the expense of ample. These differences are rooted in the ways markets and housework and care. institutions work and channel men and women into different Constraints arise from markets, institutions, and their inter- segments of the economic space. These mutually reinforcing play. Discrimination against women in the labor market, for ex- market and institutional barriers, including those related to ample, limits the number and types of jobs that women can take gender roles and norms, mute the impact of economic devel- on. And traditional norms for women’s role in the economic opment on the incentives and constraints for women’s access sphere may reinforce the discrimination. Similarly, the availabil- to economic opportunities. ity of an adequate school can affect the decision to send girls What about the impact of gender equality on economic de- (and boys) to school. velopment? As argued in this Report, greater gender equality enhances economic ef�ciency and improves other develop- Understanding the relationship between economic ment outcomes through three main channels: productivity development and gender equality gains associated with better use of existing resources, im- The relationship between gender equality and economic de- provements in outcomes for the next generation, and more velopment (the combination of higher incomes and better representative institutions and policy making. When aligned, service delivery) is a two-way relationship. In the framework households, markets, and institutions can support and here, households, markets, and institutions and their interac- strengthen these positive links. For example, well-functioning tions shape this relationship and ultimately determine gender and affordable education and health systems can ease restric- outcomes. tions on access to human capital among women—especially Start with the impact of economic development on gender where distance and mobility are factors and where the private equality. Higher incomes and better service delivery can increase sector can step in to �ll gaps in public sector provision. Simi- gender equality, acting through markets and institutions in ways larly, if labor markets work well, educated women will enter that affect household decisions. Education and health provide a the labor force and contribute their talents and skills. good illustration. But in some cases, market and institutional constraints can lead to inef�cient gender outcomes both in the household and • First, to the extent that aggregate economic growth translates outside it, impeding economic development. For instance, in- into higher household incomes, constraints on access to edu- secure land property rights among female farmers in Africa cation and some health services are relaxed. So, fewer house- lead to lower than optimal use of fertilizer and other produc- holds will have to choose between girls and boys and women tive inputs and shorter than optimal fallowing times, reducing and men in providing or accessing these services. agricultural productivity. And when employment segregation • Second, the changes in market structures and signals that excludes women from management positions, the average tal- typically accompany growth (such as the expansion of man- ent of managers is lower, reducing technology adoption and ufacturing and service activities) and encourage the greater innovation. participation of women in the wage labor market also work We refer to these situations as market failures and institu- to reduce these gaps by placing a greater value on girls’ edu- tional constraints. Market failures occur with discrimination, cation and by empowering women in household decision information problems, or limitations in the type and nature making. of contracts available.2 Institutional constraints, by contrast, arise from legal restrictions, customary practices, social norms, • Third, economic development means the improved delivery or other formal or informal institutional arrangements that of public services such as health and education. And these in- result in failures to determine and enforce rights. Correcting stitutional improvements lower their costs to households, al- these market failures and institutional constraints can yield lowing them to use more of them—for males and females— substantial productivity gains and broader economic bene�ts. with attendant improvements in gender outcomes. And in a more competitive and integrated world, even modest The combination of higher incomes, higher returns to hu- improvements in how ef�ciently resources are used can have man capital investments, and better access to services strength- signi�cant effects on a country’s competitiveness and growth. ens the incentives for and lowers the constraints to investing in These ideas are illustrated in our framework graph by the girls and women’s human capital and is then likely to lead to interconnected gears representing markets, formal and infor- a reduction in gender gaps in educational attainment and in mal institutions, and households (�gure P2.1). Propelled by health outcomes such as life expectancy. economic growth, the gears representing markets and insti- What has driven progress? What impedes it? 101 F I G U R E P 2.1 Gender outcomes result from interactions between households, markets, and institutions R EQUALIT ND E Y po GE lic ies S UTION MAL INSTIT INFOR ECONOMIC OPPORTUNITIES TS RKE HOUSEHOLDS MA AGENCY ENDOWMENTS FOR MAL INST ITUT ION S GR OW TH Source: WDR 2012 team. tutions turn, moving the household gear. The combined move- also command higher returns, increase their productivity, or ment of these gears ultimately triggers the turning of the gear both—and ultimately earn a higher income. representing gender outcomes, thus increasing gender equality. An individual or a household has little capacity to change One can also imagine the movement going in the opposite di- the way markets and social institutions function, but soci- rection—that is, with the gender equality gear moving those rep- ety does (as the sum of individuals and households). And it resenting households, markets, and institutions and ultimately is precisely this capacity that allows markets and institutions generating higher economic growth. In this setting, market and to evolve over time as a result of collective action, policy in- institutional constraints can be thought of as wedges that prevent terventions, or outside changes brought about by, say, expo- one or more gears from turning, thereby weakening both direc- sure to norms, values, and practices in other societies through tions of the two-way relationship between economic growth and global integration. gender equality. Throughout the Report we use the color green to indicate when markets, formal, and informal institutions are functioning to strengthen the relationships between economic APPLYING THE FRAMEWORK development and gender equality, and the color red to indicate when they are acting as constraints on these links. The four chapters in Part 2 use this framework to address two Although we have described the framework as a snapshot at questions. What explains progress or the lack of progress in one point in time, we also use it to look at changes over time. closing gender gaps in endowments, agency, and access to Allocation decisions and agency formation in the household economic opportunities? And how has globalization affected take place in a dynamic context, with outcomes today affecting gender equality? The answers to these questions point to the both decision making and returns in the future. For example, as role for policies and their likely impacts. individuals work and earn income, they can accumulate assets, Applying this framework produces important insights. affecting their bargaining power and thus future household al- Progress has come fastest in areas where market and insti- location decisions. In addition, greater access to assets may mean tutional forces have aligned to provide strong incentives that the household or some individuals in the household can for households and societies to invest equally in men and 102 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 women—reversing any previous disadvantages that girls and generational reproduction of gender inequality. In these areas, women may have faced. In these cases, progress has been pro- progress has been slow and often unresponsive to economic pelled by forces acting through various points of entry— development alone. markets, formal institutions, and informal institutions. This pro- So, in these areas, the role of policy is twofold. It can cess has been aided where economic growth has been sustained strengthen the relationship between gender equality and eco- and broad-based, as in education, where gender gaps have closed nomic growth, and it can address the market failures and in- very rapidly (chapter 3). stitutional constraints that stall progress. The �rst idea is illus- In contrast, progress has stalled—even with economic trated by the drops of oil that ease the turning of the market growth—where market and institutional constraints reinforce and institutions gears. The second is equivalent to removing inequalities between women and men in endowments, agency, wedges that stop some gears from moving. A more detailed and access to economic opportunities. This situation can arise discussion of the rationale for and role of public action is in when constraints in one of the three spheres through which eco- Part 3. nomic development affects households—markets, formal insti- tutions, and informal institutions—are particularly binding, or when constraints in all three spheres are mutually reinforcing. NOTES In the �rst case, there is a single entry point for policy action in that the most binding constraint has to be removed for im- 1. Duflo 2003; Lundberg, Pollak, and Wales 1997. provements in gender outcomes to occur, as illustrated by the 2. Information problems include information asymmetries and ex- discussion of excess female mortality in chapter 3. In the second ternalities. With information asymmetries, market participants case, improvements in gender equality will require coordinated are unable to gather or have differential access to necessary in- interventions in a number of spheres, as illustrated by the dis- formation. Externalities imply that the behavior of some mar- ket participants affects the outcomes of others in the same—or cussion of women’s agency and access to economic opportuni- other—market. ties in chapters 4 and 5. Against this background, globalization has operated through markets and institutions to lift some of the constraints to gen- der equality in human capital, agency, and access to economic REFERENCES opportunities. But the women most affected by existing con- Duflo, Esther. 2003. “Grandmothers and Granddaughters: Old-Age straints risk being left behind in the absence of public action Pensions and Intrahousehold Allocation in South Africa.� World (chapter 6). Bank Economic Review 17 (1): 1–25. Based on the analysis in chapters 3–6, four priority areas for Lundberg, Shelly J., Robert A. Pollak, and Terence J. Wales. 1997. “Do public action are identi�ed: reducing gender gaps in human Husbands and Wives Pool Their Resources? Evidence from the capital, promoting women’s access to economic opportunities, United Kingdom Child Bene�t.� Journal of Human Resources 32 closing gender gaps in voice and agency, and limiting the inter- (3): 463–80. CHAPTER 3 Education and health: Where do gender differences really matter? I nvestments in health and education— Overall, progress has been slower where human capital endowments—shape the multiple barriers among households, markets, ability of men and women to reach their and institutions need to be lifted at once or full potential in society. The right mix of where there is only a single effective point of en- such investments allows people to live longer, try for progress. Consider each in turn. healthier, and more productive lives. System- Although girls participate equally (or more) atic differences in investments between males at all education levels, the educational streams and females, independent of their underlying they choose are remarkably different and stable causes, adversely affect individual outcomes in across countries at very different incomes. Men childhood and adulthood and those of the next continue to study engineering while women generation. Left uncorrected, these differences continue to learn how to be teachers. While part translate into large costs for societies. of the problem lies in the educational system, Where do gender differences arise in human these patterns are reinforced by gender norms capital endowments, how are they reduced, in households and markets. Some gender norms and when do they persist? Global comparisons relate to care in the household (overwhelmingly of participation in education and mortality risks provided by women) and its implications for among women and men show that progress has the kinds of jobs that women choose. Others been tremendous where lifting a single bar- have to do with the continuing “stickiness� in rier is suf�cient—for households, markets, or employers’ attitudes toward family formation institutions. and childbearing. Equal gender participation in Consider the increasing participation of different �elds of studies requires simultaneous women at all educational levels. Gaps in primary changes among households, markets, and insti- enrollments have closed, and in secondary and tutions. That has not happened so far. tertiary enrollments, new gaps are emerging— Things can also get stuck where there is only for boys. College enrollments increased sevenfold one point of entry: households or markets or for women over the past three decades, fourfold institutions. for men. The reason is that interventions tar- Health disadvantages for women fall in this geted at any one of households or markets or in- category. Consider girls missing at birth (a de�- stitutions have all increased enrollments. Where cit of female births relative to male ones) and all three drivers have come together, change has excess female mortality after birth (women and accelerated. Conversely, where all three have not girls who would not have died in the previous changed, progress has stalled. Further reducing year had they been living in a high-income girls’ disadvantages in educational participation country after accounting for the overall mortal- requires sharpening the focus on severely disad- ity of the country they live in). vantaged populations, for whom all the drivers of Globally, girls missing at birth and deaths progress are missing. from excess female mortality after birth add up Education and health: Where do gender differences really matter? 105 to 6 million women a year, 3.9 million below the “ age of 60. Of the 6 million, one-�fth is never . . . [we want] sons. Everybody does. They can work born, one-tenth dies in early childhood, one- and earn. Girls have to marry. �fth in the reproductive years, and two-�fths at Adolescent girls, rural India older ages. Because those who died in the earlier ages also had the longest to live, they account for the bulk of lost women years around the world. In-laws decide how many children to have. If the The problem is not getting any better. In 1990, first child is a girl, then they will ask to have more � missing girls and excess female mortality below children until a son is born. the age of 60 accounted for 4 million women a Young woman, urban India year; in 2008, the number was 3.9 million, fueled by a near doubling in Sub-Saharan Africa. Unlike Asia, where population-adjusted excess female women, which arise from poor institutions, mortality fell in every country (dramatically in excess male mortality is often tied to behavior Bangladesh, Indonesia, and Vietnam), most Sub- deemed more socially acceptable among men, Saharan African countries saw little change in the such as smoking, heavy drinking, and engaging new millennium. And in the countries hardest in risky activities. hit by the HIV/AIDS epidemic, things got worse. Understanding both sorts of mortality risks In South Africa, excess female deaths increased and what can be done about them is critical from (virtually) zero between the ages of 10 and for any notion of human justice. Solving excess 50 in 1990 to 74,000 every year by 2008. female mortality requires institutional changes; Comparisons with Europe, where excess fe- solving missing girls at birth and excess male male mortality existed during the early 20th mortality delves fundamentally into social norms century but disappeared by 1950, suggest that and household behavior—both of which are the patterns today reflect a combination of overt more dif�cult to understand and tackle. discrimination before birth, and poor institu- tions combined with the burden of HIV/AIDS after birth. More than 1.3 million girls are not ENDOWMENTS MATTER born in China and India every year because of overt discrimination and the spread of ultra- Education and health investments have a huge sound technologies that allow households to impact on the ability of individuals—whether determine the sex of the fetus before birth. In- men or women—to function and reach their formal institutions that generate a preference potential in society. For both boys and girls, for sons are the primary bottleneck. childhood investments in health affect outcomes After birth, overt discrimination plays a throughout the course of life. Low birth weights smaller role. Instead, poor institutions of public and childhood exposure to disease have been health and service delivery exact a heavy burden linked to lower cognitive development, school- on girls and women. In parts of Sub-Saharan ing attainment, and learning in adolescence. Africa, HIV/AIDS risks have compounded the Less healthy children are at an elevated risk of problem. But where countries have improved becoming less healthy adults.2 Poorer health basic institutions of public health and service outcomes in adulthood in turn affect economic delivery, excess female mortality has declined. outcomes, reflected in health-related absences Even in countries where historical studies iden- from the labor force and lower work hours and ti�ed discrimination against girls (like Bangla- earnings.3 desh), better public health measures improved life-chances for both boys and girls, but more so for girls.1 Clearly, excess female mortality is not a prob- lem in all countries. In the Russian Federation and some other post-transition countries, mor- tality risks have increased for both sexes—but “ Educated women do not sit around and wait for men to provide for them. They do not need a man to buy things for them. � particularly for men. In these contexts, there is excess male mortality relative to high-income Young woman, urban South Africa countries today. Unlike mortality risks among 106 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Similarly, as chapter 5 shows, investments in run implications for society. If more boys than education determine women’s ability to earn girls are born, eventually many men will be un- higher wages and to own and operate produc- able to �nd wives. Recent research suggests that tive farms and �rms. On average, differences in such a “marriage squeeze� is already well under education explain a signi�cant fraction of the way in China and India.12 variation in wages and incomes among adults. These basic themes—educating girls and In both high- and low-income countries, gen- women, improving health outcomes in child- der differences in education have contributed hood, lowering the risks of giving birth, and ad- signi�cantly to the productivity and wage gap dressing skewed sex ratios at birth (the latter two between men and women. leading to excess female mortality and missing Health and education investments in women girls at birth)—play out consistently in the rest are also special in three ways. First, in their roles of the chapter. The focus throughout is twofold: as mothers, educated women pass on the ben- �rst, separating the problems that will likely e�ts of higher education to their children. Chil- diminish with income growth from those that dren born to more educated mothers are less will remain “sticky,� and, second, understanding likely to die in infancy and more likely to have how and where policy can be effective through higher birth weights and be immunized.4 Evi- the framework of this Report—the interactions dence from the United States suggests that some between households, markets, and institutions. of the pathways linking maternal education to child health include lower parity, higher use of prenatal care, and lower smoking rates.5 In Tai- EDUCATION wan, China, the increase in schooling associated with the education reform of 1968 saved almost A decade into the new millennium, there are 1 infant life for every 1,000 live births, reducing many reasons to feel optimistic about the state infant mortality by about 11 percent.6 In Paki- of women’s education around the world. Prog- stan, even a single year of maternal education ress has been remarkable, and many of the gaps leads to children studying an additional hour at salient in the 20th century have closed. Today, home and to higher test scores.7 girls and boys around the world participate Second, women face particular risks during equally in primary and secondary education. In pregnancy and childbirth: 1 of every 11 women tertiary education, a clear bias is emerging that in Afghanistan and 1 of every 29 in Angola favors women—with enrollments increasing dies during childbirth.8 Compare that with 1 of faster for women than for men. every 11,400 in Sweden. As this chapter shows, These gains have been possible because, for maternal mortality and excess female mortality enrollments, lifting single barriers, whether stem- in the reproductive ages are closely linked. But ming from households, markets, or institutions, high maternal mortality rates also have implica- has been suf�cient. These multiple entry points tions for educational investments and the ability have allowed policies to circumvent bottlenecks of women to participate in society. As the risk arising from adverse preferences, low returns to of dying in childbirth declines, educational in- female education, or poor institutions. vestments increase (and more so for girls).9 In But the optimism has to be tempered on Sri Lanka, reductions in the maternal mortality three fronts. First, in some countries and in ratio increased female literacy by 1 percentage point.10 And because reductions in maternal mortality ratios also reduce maternal morbid- ity (in the United States in 1920, one of every six women suffered from a long-term disability stemming from giving birth), improvements in the conditions of childbirth can drive increases “ It is good to go to school because as a woman if you are an educated person, then men won’t take advantage of you. So it is important that you go to in the labor force participation rate of married school. women.11 Third, overt discrimination that leads to male-biased sex ratios at birth can have long- � Young woman, urban Liberia Education and health: Where do gender differences really matter? 107 some populations within countries girls are The good news still the last to enroll and the �rst to drop out Most countries around the world have at- in dif�cult times. These severely disadvantaged tained gender parity in primary education, populations face a host of different problems, with an equal number of boys and girls in and the “best� solution to their problems will be school. Among children currently not attend- context speci�c. Second, children in low-income ing primary school, 53 percent are girls, with a countries learn far less than their high-income concentration of gender disadvantage in some counterparts. Low learning affects both boys African countries, including Benin, Chad, Niger, and girls, and gender differences are small. In a and Togo. But, even in these countries, progress globalizing world, poor skills will dramatically has been substantial: in Sub-Saharan Africa the affect the future outcomes of all children. number of girls for every 100 boys in primary Third, women and men continue to choose very different �elds of study in secondary and school increased from 85 in 1999 to 91 in 2008. tertiary education. These patterns of stream di- Moving from primary to tertiary enrollment vergence are similar across poor and rich coun- shows three patterns (�gure 3.1). First, most tries, suggesting that increases in enrollments children participate in primary schooling, but and learning are necessary, but not suf�cient, secondary enrollments range from very low to to even the playing �eld in later life. The �elds very high across countries; again, some coun- of study that men and women choose feed into tries in Sub-Saharan Africa stand out for their their occupational choices, which in turn affect particularly low rates of participation. In ter- the wages they earn throughout their adult lives. tiary education, low participation is the norm There has been less success in addressing stream in developing countries. Therefore, increases in divergence because many barriers need to be secondary (and tertiary) enrollment for both lifted at the same time—households, markets, boys and girls are necessary in several countries. and institutions need to change simultaneously, Second, at low overall levels of secondary enroll- through complex polices that act on multiple ment, girls are less likely to be in school, while fronts. at high levels the pattern reverses with the bias F I G U R E 3.1 Gender parity in enrollments at lower levels has been achieved in much of the world, but tertiary enrollments are very low and favor women Primary education Secondary education Tertiary education 100 100 100 men Some African Enrollment gaps dwarf school enrollment, women, gross, % disadvantaged countries lag behind gender gaps school enrollment, girls, net, % school enrollment, girls, net, % 80 80 80 60 60 60 40 40 40 Women are more likely 20 20 20 to participate girls girls than men disadvantaged disadvantaged 0 0 0 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 school enrollment, boys, net % school enrollment, boys, net % school enrollment, men, gross % East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa South Asia Sub-Saharan Africa High-income countries Source: WDR 2012 team estimates based on World Development Indicators. Note: The 45° line in each �gure above shows gender parity in enrollments. Any point above the 45° line implies that more women are enrolled relative to men. 108 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 now against boys. The number of countries with all the rich, but only 50 percent of the poor, are girls disadvantaged in secondary education is enrolled, and enrollment is no different for boys similar to the number with boys disadvantaged. and girls. In Country B, all rich and poor boys, Third, in tertiary education, girls are more likely but only 50 percent of rich and poor girls, are in to participate than boys—a difference that in- school. Both countries have equal total inequal- creases with overall participation rates. Between ity in the educational system, but the patterns 1970 and 2008, the number of female tertiary are very different. Decomposing total inequality students increased more than sevenfold (from in these countries would show that inequality is 10.8 million to 80.9 million), compared with a generated entirely by wealth (with all differences fourfold increase among males. across wealth groups) in Country A but entirely While these results are positive, they illus- by gender in Country B. trate disparities by gender only. An alternative Repeating this exercise across many coun- question is whether there are other dimensions tries represented in the Demographic and of disadvantage, and if so, what is the relative Health Surveys shows that in most of them, the weight of gender versus (say) poverty in the situation is similar to Country A—with differ- production of inequality in schooling partici- ences in wealth accounting for most educational pation? Decomposing overall inequality in the inequality—not to Country B (�gure 3.2). Pov- educational system into four components— erty rather than gender feeds overall educational location, parental education, wealth, and gen- inequalities in most of the world. In fact, even der—helps answer this question.13 in countries with high total inequality (coun- Suppose that in two countries there is an equal tries where the differences in school enrollment number of rich and poor households with boys between advantaged and disadvantaged groups and girls in every income group. In Country A, are high), gender accounts for at most 38 per- F I G U R E 3.2 In most countries with moderate or high total inequality in educational outcomes, less than one-�fth of inequality stems from gender Inequality in school attendance among children 12–15 years old Low inequality Moderate inequality High inequality 100 contribution to total inequality, % 80 60 40 20 0 Moldova (2005) Congo, Dem. Rep. (2007) Jordan (2009) Niger (2006) Swaziland (2006–07) Cameroon (2004) Turkey (2003) Albania (2008–09) Namibia (2006–07) Colombia (2005) Lesotho (2009–10) Nepal (2006) Chad (2004) Uganda (2009–10) Congo, Rep. (Brazzaville) (2009) Guyana (2005) Ukraine (2007) Armenia (2005) Maldives (2009) Zambia (2007) Tanzania (2007–08) Haiti (2005–06) Vietnam (2002) Ghana (2008) Peru (2004–08) Bolivia (2008) Mozambique (2003) Zimbabwe (2005–06) Malawi (2004) Egypt, Arab Rep. (2008) Madagascar (2008–09) Liberia (2009) Burkina Faso (2003) Ethiopia (2005) Senegal (2008–09) Sierra Leone (2009) Guinea (2005) Mali (2006) Côte d’Ivoire (2005) Azerbaijan (2006) Benin (2006) Dominican Republic (2007) Cambodia (2005) Kenya (2008–09) India (2005–06) Nigeria (2008) Morocco (2003–04) gender wealth other factors Source: WDR 2012 team estimates based on Demographic Health Surveys in various countries during 2002–10. Note: The measure of inequality refers to the percentage of total opportunities that must be reallocated to ensure that all possible combinations of circumstances have the same average enrollment. Low inequality is between 0.3 percent and 2.1 percent, moderate inequality is between 2.3 percent and 6.4 percent, and high inequality is between 6.5 per- cent and 26.7 percent. Results are sorted by size of gender contribution to total inequality. Education and health: Where do gender differences really matter? 109 cent of overall inequality; in contrast, poverty stitutional constraints or increasing household frequently accounts for 50 percent or more of incomes is suf�cient to increase female partici- the total. Almost all countries where gender in- pation in education (�gure 3.3). When all three equality is a problem, and where total inequality have happened simultaneously, change has been is high, are in Africa—with India and Turkey the even faster. Take each in turn. only exceptions. This basic description of progress on educa- Changing returns. Starting from the early 1980s, tion shows that change has come to every coun- empirical evidence emerged that when returns try and region, and that in most countries, the to women’s education increased, so did parental remaining inequalities are concentrated around investments in the schooling of girls. These early poverty and other circumstances (notably, rural studies showed that changes in the agriculture or urban residence) rather than gender. A global technology that increased the returns to female focus on inequality would thus imply that, in education also led parents to invest more in girls’ most contexts, redistributive efforts should now schooling. A new generation of work brings to- be directed to poverty. gether globalization and returns to education in the context of changing technologies. What explains the progress? The rise of outsourcing in India is offer- One key message of this Report is that progress ing new work opportunities—particularly for has come in areas where lifting a single bar- women. The opening of a new information- rier is suf�cient. Consistent with this message, technology-enabled service (ITES) center, for studies now show that increasing the returns to example, increased the number of children educational investment in girls or removing in- enrolled in a primary school by 5.7 percent, F I G U R E 3.3 What explains progress in school enrollments? R EQUALIT ND E Y po GE lic ies S UTION MAL INSTIT INFOR ECONOMIC OPPORTUNITIES TS ns RKE r HOUSEHOLDS MA g retu n asi cation More stable e Incr edu incomes AGENCY ENDOWMENTS to FOR MA Low L INSTI er d TUT and irect, ION opp indi S ortu rect, cost nity s GR OW TH Source: WDR 2012 team. 110 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 F I G U R E 3.4 Free primary education reduced gender gaps in enrollments a. Malawi, by bringing more girls into school b. Lesotho, by bringing more boys into school 105 210 190 100 gross enrollment, grade 1 gender parity index 170 95 150 90 130 85 110 free primary education free primary education 80 90 1990 1993 1996 1999 2002 2005 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 girls boys Source: WDR 2012 team based on Edstats. with the increase driven primarily by higher time to evolve. But even where returns have been enrollments in English-language schools.14 This low, changes in the structure of formal institu- increased enrollment was equally large for boys tions have increased educational attainment. and girls, reflecting very local information mar- These pathways are perhaps best illustrated by kets on the returns to education. looking at changes in the price of schooling— Similarly, business process outsourcing where the costs incurred could be direct (fees (BPO) opportunities affect education of and uniforms), indirect (distance to school), or women.15 In randomly selected villages in In- result from forgone opportunities (wages that dia, three years of BPO recruiting services were children could earn outside school). provided to women primarily to increase aware- Reductions in schooling fees erode the need ness and information about the employment for families to differentiate educational invest- opportunities. Given that the intervention was ments across children. The free primary edu- at the level of the village, the study found large cation programs launched across Sub-Saharan effects—three years later, girls ages 5–15 in the Africa, for instance, increased student enroll- villages that received the intervention were 3–5 ments 68 percent in the �rst year in Malawi and percentage points more likely to be in school, Uganda and 22 percent in Kenya. The aboli- had a higher body mass index (a measure of tion of education levies contributed to bring- health), and were 10 percent more likely to be ing more girls relative to boys into school and employed in wage work. Human capital invest- reducing the existing gender gap in primary ments for boys did not change. The interven- education in Malawi (�gure 3.4). In Lesotho, tion did not change either structures within the the government launched a similar program household (for instance, the bargaining power phased in yearly beginning in the �rst grade, of the mother) or the way schools functioned. and participation jumped 75 percent (see �gure Information about market returns alone suf- 3.4).16 In contrast to Malawi, boys in Lesotho �ced to increase female enrollment and improve have been historically less likely than girls to outcomes for girls. be enrolled in primary school, particularly in the higher grades. Free primary education sup- Changing institutional constraints. If female en- ported a signi�cant influx of overage boys into rollment responded only to increasing returns, the educational system. progress would have been slower—returns can Reductions in the distance to school have be notoriously slow to move and can take a long also helped. In Pakistan, adult women who grew Education and health: Where do gender differences really matter? 111 up in a village with a school by the time they Changing household constraints. Higher and were seven years old reported higher primary more stable sources of household income have schooling than those whose villages received a helped bring girls into school even when returns school after they were seven or those whose vil- and salient institutional features remained un- lages never received a school.17 In Afghan com- changed. Household income has been tied to munities randomly selected to receive a school, greater enrollments for children—more so for enrollment of girls increased by 15 percentage girls—and increases in maternal income have a points more than that of boys—over and above greater impact on girls’ schooling than increases a 50 percent increase in attendance for both in paternal income.27 When households face sexes.18 After six months, the girls in the villages a sudden drop in income, perhaps because of with such schools also reported far higher test poor harvests, they immediately reduce invest- scores. ments in schooling. Whether these reductions The price of schooling also reflects opportu- affect boys more than girls depends on the un- nity costs stemming from forgone child labor, derlying labor market conditions. either in the formal labor market or at home In villages affected by droughts in Côte in household chores. Higher children’s employ- d’Ivoire in 1986 and 1987, school enrollment fell ment tends to be associated with lower school 14 percentage points for boys and 11 percentage attendance.19 Employment rates for 14–16-year- points for girls.28 During the same period, en- old boys and girls in urban Brazil increased as la- rollment increased 5 percentage points for boys bor markets improved, and children were more and 10 percentage points for girls in villages not likely to leave school as local labor market con- affected by droughts. Girls in primary and sec- ditions became more favorable.20 For Ethiopia, ondary schools in Turkey were more likely than an increase of 10 percentage points in the adult boys to drop out in the face of household bud- employment rate generated a 10–25 percentage get constraints.29 Similar results are found in point increase in the probability of �nding em- countries ranging from Ethiopia (1996–2000) ployment for youth ages 10–24.21 This effect was to Indonesia (1993).30 In higher-income coun- stronger for youth who never attended school. tries, by contrast, a reduction in job opportuni- On average, boys are more likely to engage in ties for school-age children brought about by agricultural or other paid or unpaid productive economic contractions may support investments work: the opportunity cost of education is of- in schooling, as work opportunities for children ten higher for males than females, leading more dry up.31 Typical results from Latin America— boys out of education and into nondomestic notably Argentina, Mexico, and Nicaragua—all child labor.22 show that reductions in labor market opportu- The flip side of greater male involvement in nities increased enrollments in school for boys the formal labor market is that girls bear a larger relative to girls.32 share of domestic labor. Households rely heavily Unsurprisingly, programs that provide in- on children, especially girls, for natural resource come to households and help them weather collection and caring for family members. Ma- economic downturns keep children in school. lawian girls ages 6–14 spend 21 hours a week on Perhaps the most convincing evidence on this domestic work, while boys spend 13½ hours; front comes from studies of conditional cash in rural Benin, girls spend 1 hour a day collect- transfers—cash given to households only if their ing water compared with 25 minutes for boys.23 children attend school for a minimum number In the Arab Republic of Egypt, a 10 percentage of days. Giving the household the ability to pro- point increase in the probability of domestic tect educational investments in children when work—measured by household access to piped times are bad is precisely one of the roles these water, piped sewerage, and garbage collection— transfers were intended to play, and the evidence reduces the marginal probability of girls’ school- shows that they work.33 ing by 6 percentage points.24 In Peru, in-house water supply has a signi�cant impact on the The bad news grade-for-age of girls but not of boys.25 And in Despite signi�cant progress, gender disparities Kenya, simulation models suggest that reducing in education have not been entirely erased. Girls the distance to a source of water by 2 kilometers in many regions of the world continue to face would increase overall enrollment and attain- severe disadvantages in primary and second- ment twice as much for girls as for boys.26 ary school enrollment, and across the board, 112 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 children in low-income countries learn little in adults surveyed had completed primary edu- school. As countries grow richer, the problems cation. There were very few schools, and grade of enrollment and learning become less salient, �ve completion was low; often more children but girls and boys continue to choose very dif- were out of school than in school. Attendance ferent �elds of study in secondary and tertiary rates varied among these villages, ranging from education. These choices have repercussions for 10 percent to 60 percent for girls. Many out-of- the occupations they choose and the wages they school girls were caring for siblings or working earn. Consider each in turn. in the �elds. While most members of these com- munities suffered from hunger and malnutri- Severely disadvantaged populations in tion, girls were particularly disadvantaged—the primary and secondary enrollment last to eat, they did not receive nutritious food, In speci�c regions, missing economic drivers such as meat.36 of female education combined with other area- Of 945,000 children ages 6–14 not studying at speci�c ecologies—such as poor safety, scattered any school in Turkey in 2006, 194,000 said they populations, and linguistic differences—give could not afford school expenses, while 22,000 rise to gender disadvantages, which mostly, but had to work and thus could not study. Among not always, work against girls. These ecologies these children, thousands are seasonal workers. are likely very different, and policies that �nely At the crucial 14–15 age range, when children target local problems will vary across severely typically transition from primary to second- disadvantaged populations. Six examples clarify ary school, children in advantaged groups (in the problem for different populations. households with fewer children living in urban Afghanistan and Pakistan are two of the few areas of better-off regions) had 100 percent en- countries where female enrollment remains low rollment compared with 10 percent for those in even at the primary level. It is widely believed disadvantaged groups (in households with more that this is because households discriminate children, living in worse-off regions). against girls in their schooling decisions. Yet, In recent years, Jamaican boys have under- new evidence suggests that families are as eager achieved in school, starting in the early years to send their girls to school as boys when the and increasing in secondary and tertiary edu- school is close by, but are more reluctant to en- cation. With declines in boys’ participation in roll their girls in schools that are farther from secondary schooling, the gender parity ratio in their houses. In Pakistan, a half-kilometer in- secondary education in 2008 was 1.04 in favor crease in the distance to school decreased female of girls, with boys twice as likely as girls to re- enrollment by 20 percentage points.34 Among peat a grade. Apart from technical vocational families living next to a school, girls are as likely subjects and physics, girls outperform boys in as boys to go to school in both Afghanistan and the Caribbean Secondary Education Certi�cate Pakistan. Part of the “distance penalty� for girls examinations, with 30 percent of girls passing could reflect safety concerns in crossing settle- �ve or more subjects compared with only 16 ment boundaries within the same village.35 In percent of boys.37 A recent program identi�es this severely disadvantaged population, solving four key challenges in boys’ development38: low the problem of distance to school for female self-esteem among young boys, violence and a enrollment, rather than tackling any innate dis- lack of discipline, masculine identities that drive crimination at the household level, will yield boys and young men away from better academic large dividends. performance, and limited opportunities for jobs The population in the highlands of northeast after graduation. Cambodia is scattered in remote and small vil- As a region, Sub-Saharan Africa stands out lages. Lack of access to land, religious suppres- in the low participation of females in schooling. sion, and limited learning and use of Khmer The disadvantage has narrowed dramatically (the national language) all marginalize these between 1990 and 2008, with the ratio of female communities and de facto restrict access to edu- to-male primary completion increasing from cation. The indigenous Kreung and Tampuen 0.78 to 0.91. Yet girls remain at a signi�cant communities in 2001 had high child mortality disadvantage in Central and West Africa, where (twice the national average for children under only 8 girls complete primary school for every �ve) and low literacy rates—only 5 of 1,970 10 boys.39 Take Burkina Faso. Estimates suggest Education and health: Where do gender differences really matter? 113 that three-�fths of the population live on less (through satellite schooling facilities) and Af- than $1 a day and more than four-�fths live in ghanistan (through community schools) helps rural areas, many surviving on subsistence agri- reduce the costs of travel and again brings girls cultural activities. Of every 1,000 children born, into school.45 A comparison of the character- 207 will die before age �ve—the ninth-highest istics of children in and out of school in Sub- child mortality rate in the world. Schooling—at Saharan Africa suggests that in 11 countries out- a net enrollment rate of 42 percent for boys and of-school children are very similar to enrolled 29 percent for girls—is among the lowest in the children and that policies that have been effec- world. Not only are schools distant and dif�cult tive in expanding enrollments in the past will to access, there often is insuf�cient room for bring these children in.46 children who do enroll, and high out-of-pocket The precise policy to be followed depends on expenses further discourage participation.40 the context and how much is known about it. The problems in these six severely disad- For instance, �nancial transfers to households vantaged populations are very different—from conditional on school attendance bring girls in. distance in Pakistan and Afghanistan, to poor But if the purpose of the transfer is purely to economic opportunities in Cambodia, to high increase educational participation (these trans- costs and low income in Turkey, to violence and fers also bene�t poor households directly), they masculine identities in Jamaica, to low physi- are expensive tools, because transfers also reach cal access and overall poverty in Burkina Faso. households that would have sent their children One way forward is to develop context-speci�c to school even without the added incentive—in strategies that address the speci�c issues. Com- most of these countries, the cost of the trans- munity schools in Afghanistan have reduced the fer per additional child enrolled is close to the distances girls must travel, and in villages where country’s per capita GDP. But numerous condi- these schools have been built, female disadvan- tional cash transfer programs are producing tages in enrollment have vanished.41 Turkey has results across countries, suggesting that if spe- conducted large campaigns to promote the en- ci�c policies are hard to design, a uniform rollment of girls in school, some targeting dis- “second-best� solution—conditional transfers advantaged regions, such as “Father, send me to households—could work just as well but cost to school� and “Girls, off to school.� Similarly, more. The problems of severely disadvantaged Jamaica is involving fathers in schooling and populations for education could, in part, be making the curriculum more boy-friendly, and solved by getting more money to households— a school-feeding program in Burkina Faso has provided that adequate educational facilities increased boys’ and girls’ enrollment by 5–6 per- exist. centage points.42 But in each of these regions, alleviating insti- Poor learning for girls and boys tutional or household constraints also helps— In addition to the problems of severely disad- whether by increasing supply through more vantaged populations, a second issue, common school construction or by increasing demand to many low-income countries, is poor learn- through easing households’ �nancial constraints. ing. Children in low-income countries typi- Pakistan, Turkey, Cambodia, and Jamaica all cally learn less and more slowly relative to their have programs that give cash to households if high-income counterparts. Although there are they send their children (in some cases, speci�- small differences across boys and girls (where cally girls) to school, and these have increased these exist, boys tend to do better at mathemat- enrollments for targeted children. The increases ics and girls at reading), the gender difference have been fairly large: 10 percentage points for is dominated by the difference across countries primary-age girls in Pakistan, 11 percentage (�gure 3.5). points for secondary-age girls in Turkey, 30 to To see how big these differences are, look at 43 percentage points for girls transitioning from the raw numbers. Only 27 percent of children primary to secondary schools in Cambodia, and ages 10 and 11 in India can read a simple pas- 0.5 days a month in Jamaica.43 Preliminary re- sage, do a simple division problem, tell the time, sults from a �nancial transfer pilot in Burkina and handle money. This low learning is not an Faso suggest similar results.44 Similarly, school Indian problem; it recurs in nearly all low- and construction in countries like Burkina Faso middle-income countries. For the developing 114 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 F I G U R E 3.5 Cross-country differences in mean scores on the 2009 PISA dwarf gender differences within countries a. Mathematics b. Literacy Kyrgyz Rep. Kyrgyz Republic Panama Azerbaijan Peru Peru Qatar Panama Indonesia Qatar Tunisia Albania Albania Kazakhstan Colombia Argentina Brazil Indonesia Jordan Tunisia Argentina Jordan Montenegro Montenegro Kazakhstan Brazil Trinidad Colombia Mexico Trinidad Thailand Thailand Chile Romania Uruguay Mexico Romania Uruguay Bulgaria Bulgaria Azerbaijan Serbia Serbia Chile Turkey Russian Federation Israel Turkey Croatia Lithuania Greece Austria Russian Federation Luxembourg Lithuania Israel Latvia Croatia Italy Slovak Republic Spain Czech Republic Portugal Spain Ireland Greece United States Slovenia Luxembourg Latvia Hungary Italy United Kingdom Macao SAR, China Czech Republic Portugal Sweden Hungary Poland United Kingdom Austria Denmark Slovak Republic France France Ireland Norway Germany Slovenia Sweden Denmark Liechtenstein Iceland United States Estonia Iceland Germany Poland Australia Switzerland Belgium Estonia New Zealand Norway Macao SAR, China Belgium Netherlands Netherlands Canada Australia Japan Japan Switzerland New Zealand Liechtenstein Canada Finland Singapore Korea, Rep. Hong Kong SAR, China Hong Kong SAR, China Finland Singapore Korea, Rep. 300 350 400 450 500 550 boys 300 350 400 450 500 550 girls Level 1 Level 2 Level 3 Level 4 Level 1 Level 2 Level 3 Level 4 mean scores mean scores Source: WDR 2012 team based on Edstats. Note: The highest level is 6. In mathematics, at level 1, students answer clearly de�ned questions involving familiar contexts; at level 3, students execute clearly de�ned procedures; at level 5, students develop and work with models for complex situations. In literacy, at level 1 students are capable of completing only the least complex reading tasks; at level 4, students demonstrate an accurate understanding of long or complex texts whose content or form may be unfamiliar; at level 6, students make multiple inferences, comparisons, and contrasts, which are both detailed and precise. Education and health: Where do gender differences really matter? 115 countries as a whole, 21.3 percent of 15-year-old A REFLEX (Research into Employment and children tested by the Program for International Professional Flexibility) study uncovers similar Student Assessment (PISA) could not achieve patterns. For example, Italian men and women level 1 pro�ciency in mathematics—the most are equally represented in about half the �elds basic skills. In Argentina, the �gure was 64 per- of study, but large gender disparities exist in the cent, in Brazil 72.5 percent, in Indonesia 65 per- other half.48 Women are more likely to obtain a cent, and in Thailand 53 percent. In a globalizing degree in education and humanities, and men world, the top performers also matter. Just over in engineering, architecture, and agricultural 13 percent of children in developed countries and veterinary science. The gender distribution perform at competency level 5 or above. Com- of graduates in social sciences, business, law, sci- pare that with 1 percent or less in Argentina, ence and mathematics, health, and social ser- Brazil, and Indonesia. Clearly, �xing poor learn- vices corresponds to that of the population of ing is imperative for both boys and girls. tertiary graduates. The sharp divergence in �elds of study does The problem of stream divergence not reflect the capabilities of men and women As countries grow richer and systems of service in different subjects. There is no systematic evi- delivery improve, enrollment de�cits for se- dence of large gender differences in average or verely disadvantaged populations and the over- subject-related ability at the secondary level. arching issue of poor learning may become less Test scores from standardized secondary school of a problem. But the playing �eld will still not graduation tests are similar for men and women be level for women and men. Signi�cant and in Indonesia but are slightly higher for women persistent gaps remain in the �elds of study that with a college education in both Indonesia and women and men choose as part of their formal Italy.49 What matters instead is stronger sorting education, and the patterns of these choices are on ability among men, combined with signi�cant very similar in rich and poor countries. gender differences in attitudes. In the REFLEX As with enrollment and learning, these study, male top performers on the secondary choices matter because, as chapter 5 shows, they graduation tests were 10 percent more likely to translate into gender differences in employment choose a male-dominated �eld than other males, and ultimately into differences in productivity while the impact of test scores on choice was in- and earnings: gender differences in occupation signi�cant for female-dominated and neutral and sector of employment account for 10–50 �elds and among female top performers. More- percent of the observed wage gap in 33 low- and over, “choosing a demanding/prestigious �eld middle-income countries (of 53 with data). As countries grow richer, gender disparities may TA B L E 3 . 1 Gender segregation in �eld of study: In most countries, shift from enrollments and learning to segrega- women dominate health and education studies and tion in �elds of study. So, policy attention may men dominate engineering and sciences have to shift accordingly. These gender differences in education tra- Fraction of countries where the Number of jectories emerge early and grow larger as young field of study is countries men and women acquire more education. At Female Male dominated dominated Neutral the secondary level, women are more likely Field of study % % % than men to choose general education and less Agriculture 3 74 22 89 likely to choose vocational education. In 63 per- cent of countries (109 of 172), the fraction of Education 84 6 10 97 women enrolled in general secondary educa- Engineering, manufacturing, and construction 0 100 0 97 tion is higher than that of women enrolled in vocational secondary education.47 At the ter- Health and welfare 82 4 13 97 tiary level, these differences magnify. Across the Arts and humanities 55 6 39 96 world, women are overrepresented in education Science 13 68 20 96 and health; equally represented in social sci- Services 21 59 21 87 ences, business, and law; and underrepresented Social sciences; business in engineering, manufacturing, construction, and law 23 16 61 97 and science (table 3.1). Source: WDR 2012 team estimates based on data from UNESCO Institute for Satistics. 116 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 of study� signi�cantly increased the probability even when she brings home all the income—as of enrolling in a male-dominated �eld for men in Ghana (see �gure 5.10 in chapter 5). Reduc- but not for women, and it decreased the prob- ing the time allocated to care at home will go ability of enrolling in a female-dominated �eld only so far if schooling reinforces gender norms for both. Countries pay a heavy cost when the about what girls should study and parental as- average quality of every �eld is reduced because pirations feed into how much children want to of the mismatch between training and ability. study and what they want be when they “grow Stream divergence is dif�cult to address up.� And unless �rms are willing to experiment precisely because it requires policies that act with hiring women in male-dominated occu- on households, markets, and institutions pations and vice-versa, shifting the allocation simultaneously. of tasks at home or changing the �eld of study Part of the problem lies in the educational is not going to help (it may even hurt). The system, which generates expectations about school system needs to say that it is acceptable what girls and boys are “supposed� to study. For for a man to be a nurse and a woman to be an example, some English subject textbooks cur- engineer. Firms need to be willing to hire male rently in use in Australia and Hong Kong SAR, nurses and female engineers. And tasks at home China, tend to depict women in a limited range need to be allocated according to individuals’ of social roles and present stereotyped images time constraints and capabilities, not gender of women as weaker and operating primarily in norms. Unless all three happen simultaneously, domestic domains.50 change will be hard. Part of the problem lies with informal in- stitutions that influence household aspirations. From education to health The Young Lives study looked at educational Despite dramatic improvements in educational aspirations and noncognitive skills of boys and participation, much remains to be done for se- girls at ages 8, 12, and 15 for 12,000 children verely disadvantaged populations around the across Ethiopia, Andhra Pradesh in India, Peru, world. Poor learning affects both boys and girls and Vietnam. Parental aspirations for the educa- and hampers the future ability of young popula- tion of their children were biased toward boys tions to participate in an increasingly globalized in Ethiopia and India by the age of 12 and to- world where, as chapter 6 shows, jobs are shifting ward girls in Peru and Vietnam. By the age of from those based on “brawn� to jobs based on 15, these biases had been transmitted to chil- “brain.� And girls and boys systematically choose dren, with clearly higher educational aspirations different �elds of study in all countries; these are shown among boys in Ethiopia and India and choices that shape later life employment choices among girls in Vietnam. Also by age 15, mea- and hence wages. The framework of households, sures of agency or ef�cacy showed a strong male markets, and institutions helps illustrate that bias in India and Ethiopia but not in Peru or progress has been rapid where improvements in Vietnam. Asked “when you are about 20 years any one has helped circumvent potential bottle- old, what job would you like to be doing,� 31 necks in another; progress has been slower where percent of girls and 11 percent of boys in India all three need to move together. chose “teacher,� whereas 35 percent of boys and Health issues, by contrast, are different. 9 percent of girls wanted to enroll in university. First, unlike education, where biological differ- In Peru, 21 percent of boys (5 percent of girls) ences may play a smaller role, women and men wanted to be engineers, while in Vietnam there are intrinsically different physically and in the were few notable differences.51 health risks they face. Given the same inputs, Part of the problem lies with markets and girls and boys may achieve similar educational �rms that have been unwilling to experiment outcomes, but because of biological differences, with new forms of flexible production and em- the same health inputs may result in very dif- ployment that support family formation and ferent health outcomes. Any analysis of health childbearing, even in countries where other issues needs to account for these fundamental gender differences are notably smaller. Reduc- differences. Second, health outcomes reflect a ing stereotypes in education can go only so far if type of irreversibility that is different from that there is no maternal leave down the line and the in education outcomes. True, health and edu- woman has to do 90 percent of the housework, cation investments during childhood and their Education and health: Where do gender differences really matter? 117 timing will irreversibly affect cognitive develop- many Sub-Saharan African countries the pat- ment and learning outcomes throughout the tern is reversed, and differences are increasing, course of life. But a teacher absent from school as are overall adult mortality risks for both on any given day harms learning in a completely sexes. Worsening female mortality rates are different way from a doctor who happens to particularly notable in the HIV/AIDS-afflicted be absent from a facility at the time a woman countries, but even in Central and West Africa, goes into labor, a situation that can turn life- where HIV/AIDS prevalence rates are lower, threatening within minutes. So, formal service mortality risks are getting worse. Sub-Saharan delivery institutions will naturally play a larger Africa is the only region in the world where rel- role in health, and for some health issues, they ative mortality risks are worsening for women. will be the primary bottleneck. Fourth, male mortality risks have increased in many post-transition countries, reflecting par- ticular types of behavior and health risks that HEALTH appear to have worsened over the last three decades. Gender disadvantages in health can arise in both The �ndings—interpreted within the frame- sickness and death. Yet because women and men work of households, markets, and institutions— are biologically different, ascribing gender differ- yield sharp policy conclusions. Depending ences in mortality and morbidity to biological where they are in the life cycle, women and men differences is fraught with conceptual dangers. If face disadvantages for different reasons. Miss- women live longer than men (which they do in ing girls at birth arise from household discrimi- most countries52), is it because they are biologi- nation. Any solution to this problem has to cally stronger or because there is discrimination come through household decision-making pro- against men? Further, biological differences may cesses. These processes can be manipulated still be malleable: a biological predisposition through markets and institutions, but markets may be easy to �x, much like a pair of glasses will and institutions alone will not do the trick. Af- �x genetically poor eyesight. But when men and ter birth, although discrimination remains sa- women are biologically susceptible to different lient in some countries, in many other countries diseases (breast or prostate cancer), how can we high female mortality reflects poorly perform- judge whether one is more crippling than the ing institutions of service delivery. Improving other? Cutting across the conceptual issues are institutions is the key to reducing female mor- poor data: in many countries, the information tality. Even in populations with discrimination, on morbidity is sparse and of uneven quality. better institutions can help reduce the adverse More troubling, regions with good data may impacts of differential treatment. High male be precisely those where gender differences are mortality usually reflects types of behavior that smaller, leading to misplaced policy priorities. are socially deemed more acceptable among To present a global picture, this chapter men. Because there is a single point of entry for adopts a reductionist approach, focused entirely each of these problems, solving them will be on sex ratios at birth and mortality after birth. hard. But for any notion of human justice, it is If the disadvantages thus uncovered are small, imperative. it would be a mistake to argue that gender dis- advantages in health are small—they could well The facts on dying and death in emerge in comparisons of morbidity. But four low-income countries �ndings suggest that mortality disadvantages Some issues that this chapter highlights are bet- are not small. ter known than others: skewed sex ratios at birth First, the well-known problem of skewed in North India and China, high female mortal- sex ratios at birth in some countries remains ity in infancy and early childhood in South unresolved. Second, compared with developed Asia, and rising male mortality in some post- economies, the rates at which women die rela- transition countries have all received attention tive to men are systematically higher in many in the past decade. Less well known is that excess low- and middle-income countries around female mortality is a continuing phenomenon the world. Third, while men die more than beyond childhood and a growing problem in women at all ages in developed economies, in Sub-Saharan Africa. Simple comparisons of 118 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Saharan Africa and other countries, although F I G U R E 3.6 Adult and child mortality in Sub-Saharan Africa the rate of decline has been slower in the former. 6,000 • Adult mortality rates in other countries have under-5 mortality, deaths per 1,000 live births remained roughly stable over the past 25 adult deaths per 100,000 adult population 200 5,000 years, but in Sub-Saharan Africa, they dou- bled between 1980 and 2000. 150 4,000 • A large portion of this increase in Sub- Saharan Africa is attributable to HIV/AIDS, 3,000 with adult mortality rates in high HIV- 100 prevalence countries reaching more than half 2,000 the levels seen in the years of the genocides in Rwanda and Cambodia—but on a sustained 50 1,000 and rising basis. • Particularly surprising is the fact that adult 0 0 mortality did not decrease, and actually in- 1975–70 1980–84 1985–89 1990–94 1995–99 2000–04 creased, in several countries in Sub-Saharan period Africa with low HIV/AIDS prevalence, par- non-Sub-Saharan Africa, under-5 ticularly those in Central and West Africa. non-Sub-Saharan Africa, adult Figure 3.7 uses World Health Organization Sub-Saharan Africa, lower HIV, under-5 Sub-Saharan Africa, lower HIV, adult (WHO) estimates for all countries between Sub-Saharan Africa, higher HIV, under-5 1990 and 2008, to show how the relative rates Sub-Saharan Africa, higher HIV, adult of adult mortality for women and men have changed over this period. Countries below the solid maroon line saw a worsening of rela- Source: de Walque and Filmer 2011. Adult mortality rates are based on sibling rosters from 83 Demographic and Health Surveys collected over time from 46 countries, and are estimated between 1975 and the most tive mortality risks for women, and those to current period, typically 2000; they show the likelihood of dying in every �ve-year period of the data. the right of the dashed black line saw a wors- ening of adult mortality (so countries in the lower right quadrant, for instance, saw a wors- male and female mortality risks over time help ening of both). Although the numbers are make that point. not strictly comparable to those in �gure 3.6 Of every 1,000 adults between the ages of (partly because of different time periods, and 15 and 60 in the rich countries, somewhere partly because of the measure of adult mortal- between 56 (Iceland) and 107 (United States) ity), the broad story remains similar (also see men and women will die each year.53 In India, box 3.1). that number rises to 213 (in China to 113). In In most countries, adult mortality risks de- Central and West Africa, adult mortality rates clined. In the HIV-affected Sub-Saharan African are higher, routinely exceeding 300 and in many countries (those with a prevalence above 5 per- countries 400. Compare that with conflict coun- cent in 2008), mortality risks are getting worse, tries like Iraq (285) and Afghanistan (479). And and relatively more women are dying than men. in HIV/AIDS-affected countries, the numbers Surprisingly, a large number of African coun- rise to between 481 (Malawi) and 772 (Zimba- tries saw very small improvements in mortality, bwe). In their mortality risks, these countries are with greater improvements for men; over this worse than Afghanistan (and far worse than Iraq period, almost no country in Africa saw rela- or Pakistan). tive declines in mortality risks for women. In Comparisons over time highlight the dra- contrast, the majority of countries around the matic difference between Sub-Saharan Africa world experienced declines in adult mortality and other regions of the world (�gure 3.6). Here and relative improvements for women. Less sur- are the patterns54: prisingly, the other main country grouping that • Infant and early childhood mortality (under- stands out prominently in �gure 3.7 consists of �ve mortality) has declined in both Sub- some Eastern Europe and Central Asian coun- Education and health: Where do gender differences really matter? 119 tries, where again mortality risks have gotten worse, but more so for males.55 FIGURE 3.7 Adult mortality: Over time and by sex In 2008, the 14 countries with the highest adult mortality risk for women (in descend- 2 Declining adult mortality, especially ing order) were Zimbabwe, Lesotho, Swaziland, sex ratio of mortality, 2008, relative to 1990 for women Zambia, South Africa, Malawi, the Central Af- Maldives rican Republic, Mozambique, Tanzania, Chad, 1.5 India Jamaica Uganda, Cameroon, Burundi, and Nigeria. Af- China Iraq ghanistan comes in at number 15, and Pakistan Bangladesh Iran,Islamic Rep. at number 64. For child mortality (under �ve, 1 per 1,000 births), the worst places for girls (in Lesotho Kenya Swaziland descending order) were Afghanistan, Angola, Mozambique Botswana Eritrea South Africa Zimbabwe Chad, Somalia, Mali, the Democratic Republic Liberia .5 of the Congo, Nigeria, Sierra Leone, Guinea- Increasing adult mortality, Bissau, the Central African Republic, Burkina Tonga with disproportionately higher increased Faso, Niger, Burundi, Equatorial Guinea, and mortality for women 0 Liberia. This basic description highlights the ap- 0 1 2 3 adult mortality ratio, 2008, relative to 1990 proach in the rest of the chapter. We will show East Asia and the Pacific Europe and Central Asia that the focus on female mortality is slowly Latin America and the Caribbean Middle East and North Africa shifting from childhood to adulthood and Organisation for Economic Co-operation South Asia from South Asia to Sub-Saharan Africa, while and Development countries Sub-Saharan Africa: high HIV prevalence the problem of missing girls at birth remains Sub-Saharan Africa: low HIV prevalence rooted in India and China. To do so, the argu- Source: World Health Organization 2010. ment triangulates by looking at every country Note: Adult mortality is expressed as the probability of death between the ages of 15 and 60. The sex ratio over time, by examining the historical context of mortality is male adult mortality divided by female adult mortality. BOX 3.1 Adult mortality risks: Who are the outliers? Several countries in figure 3.7 highlight particular women, who benefited from the country’s focus stories, help motivate the analysis in the remainder on maternal mortality and safe motherhood. The of this chapter, and remind us why any particular chapter highlights the fundamental role of mater- summary of the data can be problematic, requiring nal mortality and related health issues in contribut- country-by-country analysis. ing to female mortality in the adult population. In Eritrea and Liberia the cessation of conflicts Tonga saw large increases in relative female around 1990 reduced mortality risks for men (and mortality risks. The problems that this chapter somewhat for women). The drop in male mortality focuses on are absent in Tonga: It has high immu- rates, however, worsened the relative mortality risk nization rates among children, very few infant for women. A good thing for both men and women, deaths, and no maternal deaths in 2008. But Tonga this shows how a misinterpretation can be avoided also has severe problems with noncommunicable by comparing mortality with a single reference for diseases and one of the world’s highest diabetes all countries, instead of within countries and over rates: 75 percent of women are obese relative to time (as the figure implicitly does). 56  percent of men. Heart attacks accounted for Iraq and Jamaica saw large increases in relative 48 percent of all deaths in 2006. Although the chap- male mortality risks. In both countries, crime and ter only touches on morbidity and mortality caused violence are taking men’s lives and increasing over- by noncommunicable diseases, Tonga reminds us all mortality risks. The link between conflict, violent that the problems today may well be different from crime, and male mortality is taken up in the chap- those tomorrow and that every region will have ter’s discussion of male mortality and specific issues its own specifics that a global report cannot ade- in Sub-Saharan Africa. quately address.a Between 1990 and 2008, overall mortality risks declined dramatically in Maldives, especially for a. Somanathan and Hafez 2010. 120 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 of the now-rich countries, and by ensuring that periods in a woman’s life after birth. But because the facts and interpretations are robust to alter- women under 60 years also have the longest to native data sources. At the end, “stress-testing� live, they account for 81 percent of the annual yields robust conclusions that point to the need years lost around the world to excess female mor- for fundamental institutional reform, better tality. Excess male mortality accounts for 1 mil- provision of public goods such as clean water lion men a year, primarily concentrated in some and sanitation, and a continuation of the war post-transition countries (more than half) and against HIV/AIDS. some Latin American countries. Because of the greater life-years lost to mortality before 60 and Missing girls at birth and excess female because of greater sensitivity to the choice of the mortality post-birth reference group in the older years (see technical Ideally, analysis of mortality risks by sex and age annex), the focus of this chapter is on mortality would look at the relative age-mortality pro�les risks below age 60, particularly in the three criti- of women and men across countries and over cal periods for women—at birth, in infancy and time. But that is hard to do because it requires early childhood, and in the reproductive years. comparing different age-mortality functions, While missing girls at birth are indeed con- especially dif�cult when mortality risks across centrated in India and China, consistent with comparison groups cross (perhaps multiple the earlier discussion, excess female mortality times) at different ages. To summarize this com- after birth is highest in Sub-Saharan Africa, the plex data in a readily understandable manner, only region where the numbers are going up the chapter computes two measures. over time (table 3.2 and map 3.1). These three Missing girls at birth are estimated through population groupings—China (with a popula- comparisons of the sex ratio at birth in coun- tion of 1.3 billion), India (1.15 billion), and Sub- tries around the world with those in compa- Saharan Africa (0.8 billion)—together account rable populations with no discrimination.56 for 87 percent of the world’s missing girls and It also computes excess female (male) mortality excess female mortality. by comparing the mortality risks of women rel- But the age pro�les are very different. In ative to men in every country and every age with China, most excess female mortality is at birth. those seen in developed economies today—the In India, missing girls at birth and excess female “reference population.�57 mortality in early childhood and in the repro- This excess mortality measure is computed ductive years each account for roughly a third. for all countries around the world at three points In Sub-Saharan Africa, excess female mortality in time—1990, 2000, and 2008. To understand in the reproductive years accounts for 78 per- what may drive these mortality risks, the same cent in the HIV/AIDS countries and 55 percent measure is computed for 13 developed coun- in the low-HIV countries. Sub-Saharan Africa is tries historically—in some cases going back to the only region in the world where the numbers 1800. Changes in the relative mortality pro�le increased between 1990 and 2008—both abso- by age for developed countries affect the com- lutely (from 0.6 million a year to 1.1 million) putation of excess mortality in other countries. and as a fraction of the female population. So, to better interpret patterns across countries At the outset, we rule out one explanation and over time, the chapter always maintains the for excess female mortality and missing girls same “reference� for all computations. Assump- at birth—lack of income growth (�gure 3.8). tions built into this particular summary of the There is a strong relationship between income mortality data are discussed in the technical an- and excess female mortality—Sweden, unsur- nex to this chapter. prisingly, has lower excess female mortality than These two computations suggest that miss- Cameroon, but there is little or no relationship ing girls at birth and excess female mortality af- between the change in excess female mortality ter birth add up to more than 6 million women between 1990 and 2008 and economic growth a year. Of these, 23 percent are never born, 10 in the same period. Some countries that have percent are missing in early childhood (under grown (Angola and South Africa) have seen little �ve years), 21 percent in the reproductive years change or a worsening in excess female mortal- (15–49 years), and 38 percent in the age 60 and ity; others with less growth (Nepal) have seen a older group. These are the three most dangerous dramatic decline. The lack of a relationship be- Education and health: Where do gender differences really matter? 121 TA B L E 3.2 Skewed sex ratios at birth and excess female mortality persist across the world, leading to females missing at birth and excess female mortality during childhood and the reproductive years Missing girls at birth and excess female deaths (in thousands) Total women girls at birth girls under 5 girls 5–14 women 15–49 women 50–59 under 60 1990 2008 1990 2008 1990 2008 1990 2008 1990 2008 1990 2008 China 890 1,092 259 71 21 5 208 56 92 30 1,470 1,254 India 265 257 428 251 94 45 388 228 81 75 1,255 856 Sub-Saharan Africa 42 53 183 203 61 77 302 751 50 99 639 1,182 High HIV-prevalence countries 0 0 6 39 5 18 38 328 4 31 53 416 Low HIV-prevalence countries 42 53 177 163 57 59 264 423 46 68 586 766 South Asia (excluding India) 0 1 99 72 32 20 176 161 37 51 346 305 East Asia and Pacific (excluding China) 3 4 14 7 14 9 137 113 48 46 216 179 Middle East and North Africa 5 6 13 7 4 1 43 24 15 15 80 52 Europe and Central Asia 7 14 3 1 0 0 12 4 4 3 27 23 Latin America and the Caribbean 0 0 11 5 3 1 20 10 17 17 51 33 Total 1,212 1,427 1,010 617 230 158 1,286 1,347 343 334 4,082 3,882 Source: WDR 2012 team estimates based on data from the World Health Organization 2010 and United Nations Department of Economic and Social Affairs 2009. Note: Totals do not necessarily add up due to rounding. tween gender disadvantages in mortality and in- “ come growth is consistent with a large literature that comes to the same conclusion.58 The firstborn must be male. After that it’s all the same. � To examine why there are missing girls at birth and excess female and male mortality and what Young man, rural Serbia can be done about it, we need alternative expla- nations. Given the age-grouping of disadvan- tage around birth, infancy, and the reproductive years, each of which may have different causes, A deadly combination of three factors led to the chapter develops the arguments in turn. increasing numbers of unborn girls in the late 20th century (�gure 3.9). First, fertility started Girls missing at birth—The India-China dropping as female education and the returns to problem it in the labor force increased; in China, the one- The problem of many missing girls was �rst child policy reduced fertility. Second, ultrasound documented, separately, by Coale, Das Gupta, became widely available (allowing for prenatal and Sen.59 Subsequent studies con�rm the geo- sex determination), starting from big cities and graphical variation within India and China—in moving to small towns and rural areas. Third, India initially in the northern belt but gradually the preference for sons remained unchanged— spreading south, and in China a gradual spread- families now want two children, but they want ing inland from the eastern coast.60 at least one son. 122 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 M A P 3.1 In China and India, the number of girls missing at birth remains high, and parts of Africa experienced large increases in excess female mortality during 1990–2008 Countries with the highest fractions India of unborn females in 2008 Serbia Georgia Azerbaijan Armenia China UNBORN FEMALES IN 2008 0 5,000 10,000 15,000 PER 100,000 FEMALE BIRTHS Unborn female per 100,000 female births 0–1,000 1,000–2,000 2,000–5,000 5,000–15,000 No data MISSING AT BIRTH Reference countries Countries with the highest fractions of women missing in 2008 Central African Republic Swaziland Zimbabwe 500 550 600 650 EXCESS FEMALE DEATHS Excess female deaths per 100,000 female population IN 2008 PER 100,000 FEMALE POPULATION 0–100 100–300 300–650 No data EXCESS FEMALE DEATHS AFTER BIRTH Reference countries Countries with Countries with largest declines largest increases Bangladesh South Africa Bolivia Swaziland CHANGE IN EXCESS FEMALE Nepal Zimbabwe DEATHS PER 100,000 FEMALE POPULATION, 1990–2008 -200 0 200 400 -300–0 Change in excess female deaths per 100,000 female population 0–300 300–600 No data CHANGE IN EXCESS FEMALE DEATHS Reference countries Source: WDR 2012 team estimates based on data from World Health Organization 2010 and United Nations Department of Social and Economic Affairs 2009. Education and health: Where do gender differences really matter? 123 F I G U R E 3.8 Income growth did not reduce excess female mortality during 1990–2008 in low- and middle- income countries Did income growth reduce excess female mortality during excess female deaths per 100,000 female population 1990−2008 in low- and middle-income countries? change in excess female deaths per 100,000 400 4,000 300 There is no relationship In South Africa and Angola, between income growth income increased—and so and excess mortality did excess female deaths 200 in either age group 300 3,000 female population GDP per capita 100 In Nepal, income increased less 200 dramatically, but 2,000 0 excess deaths declined –100 1,000 100 –200 0 0 –.2 0 .2 .4 .6 .8 1990 2000 2008 income growth, log left axis right axis excess female mortality, ages 0–4 excess female mortality income, Angola excess female mortality, ages 15–49 0–59 years, Angola excess female mortality income, Nepal 0–59 years, Nepal excess female mortality income, South Africa 0–59 years, South Africa Source: Staff calculations based on data from World Development Indicators 2009; World Health Organization 2010; and United Nations, Department of Economic and Social Af- fairs, Population Division 2009. Note: Growth is de�ned as changes in log GDP per capita during 1990–2008. Missing women numbers are expressed as a fraction of female population in relevant age group. To see how these factors play out, think of even when the �rst child is a boy, but when the an earlier time when every household had four �rst child is a girl, the second child is much more children. In this scenario, if the �rst child was a likely to be a boy. This phenomenon has been girl, the likelihood would still be high that one demonstrated in China, the Republic of Korea, of the remaining children would be a boy. But if and India; for India, it is particularly strong for households had only two children, and the �rst educated Hindu women in northern India.61 In one was a girl, there was an even chance of hav- addition, antenatal investments, like inoculation ing yet another girl—rather than the son they against tetanus, were higher when women were wanted. As ultrasound became available, so did pregnant with boys rather than girls—and even the “solution�—if the unborn child was a girl, higher when the �rst child was female.62 the parents could abort the child and try again. These last results hold not only for north- Unfortunately, this is precisely what the data ern India but also for Bangladesh, China, and indicate. In the absence of sex-selective abor- Pakistan, suggesting that forms of disadvantage tion, the odds are even that the second child will against unborn girls may be widespread across be a boy or a girl independent of the sex of the South and East Asia.63 And they may be very �rst. The data show, however, that the probabil- hard to change: as in their own countries, Chi- ity of the second child being a boy or a girl is nese and Indians living in the United States show “ Sons would ensure continuation of their family, would take care of parents at old age and perform funeral rites when parents die. � Young woman, rural India 124 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 F I G U R E 3.9 Why are so many girls missing at birth? IONS TITUT AL INS INFORM ce eferen Son pr ECONOMIC OPPORTUNITIES HOUSEHOLDS MARKETS Prenatal sex selection and lower fertility ENDOWMENTS AGENCY Unborn Girls FOR MAL INST ITUT ION S Source: WDR 2012 team. very similar patterns of sex selection in �rst and ilies with one son. If the �rst child is a daughter, second births.64 the family will have another child, stopping if it Not only does son preference affect the gen- is a son and carrying on if it is a daughter. The der composition of birth, it may also affect the result? Girls will disproportionately have many way children already born are treated. One argu- more siblings than boys.66 ment demonstrates the nuanced links between One result of son preference is that girl chil- son preference and gender disadvantages for al- dren receive less nutrition than boys in northern ready born children.65 India (but not worldwide, as nutritional dif- Take again the simple framework of son pref- ferences show). Given the repercussions of son erence and instead of �xing fertility, let fertility preference on the different number of siblings itself be a matter of choice. In some families, the for boys and girls, could it be that some share of �rst child will be a son. If their “stopping� rule is the female disadvantage stems from the number the birth of a son, these will be single-child fam- of siblings rather than from overt discrimina- tion? As it turns out, once family size is appro- priately controlled for, there is no female disad- vantage in nutrition. That is, the entire observed “ Lhamo was told that if she gave birth to a daughter again, he was going to leave her once and for all. Lhamo was soon conceived with her third child. He would often beat her, but she was difference in nutrition between boys and girls is attributable to the difference in the numbers of their siblings. Gender disadvantage in the un- born child—manifest in fertility behavior—has no further impacts on the nutritional outcomes still willing to take the chance of giving Dorji a of the born child once fertility behavior is con- son so that their lives would come back to trolled for. normalcy. � Adult woman, Bhutan In a similar fashion, families may wait less time to have a second child following the birth of a girl than the birth of a boy. The desire to have Education and health: Where do gender differences really matter? 125 impact of kinship structures on the value placed “ on girls has also been advanced as the dominant Nowadays, men and women are equal. hypothesis for why excess female mortality is But according to tradition, we still seen in some societies but not others.72 prefer having a son to carry on the Is such discrimination against girls a wide- family line. � spread pattern linked to excess female mortality Young man, rural Vietnam in early childhood? Perhaps, as was discovered in Bangladesh in the early 1970s and India in the 1980s, girls are less likely to be vaccinated, less likely to be given medical care, and less likely a son following the birth of a girl may result in to receive nutrition at home.73 But a compari- disadvantages for the born female, if the mother son of countries around the world shows two reduces breast-feeding to increase fertility and things. First, these differences are small or non- hasten conception.67 The fundamental insight existent to begin with (in Sub-Saharan Africa, from these �ndings is that preferences over the for instance, it is the boys who suffer nutritional unborn child drive gender disadvantage—po- deprivation). Second, there is little association tentially manifest in children already born. between excess mortality among girls and disad- The intersection between son preference, vantages in vaccination, differential use of med- declining fertility, and new technologies has ical care, or differences in childhood nutrition added to the number of girls missing at birth as measured through the heights and weights of and may well disadvantage children already boys and girls (�gure 3.10). Nor is there any as- born through the number of siblings and the sociation with female labor force participation. timing of births. Changes in informal institu- These patterns too reflect similar �ndings in the tions and, through them, household behavior previous literature.74 are key to resolving this problem. And it can be Even if households treat boys and girls simi- done. Korea, where the male-to-female ratio at larly, it could be the case that health care pro- birth �rst increased sharply and then declined, viders might discriminate against girls. To as- suggests that broad normative changes across sess this possibility, researchers observed more society brought about by industrialization and than 30,000 interactions between doctors and urbanization can ultimately return sex ratios at patients in seven countries around the world— birth to normal ranges.68 Afghanistan, Burkina Faso, India, Mozambique, Paraguay, Rwanda, and Uganda—and recorded Excess female mortality in early childhood the time spent, questions asked, and examina- What causes excess mortality among girls dur- tions completed, all markers of medical care ing infancy and early childhood? One possible that correlate with overall quality of care. Sur- explanation that has received a lot of attention prisingly, the main �nding was that girls and is discrimination by parents toward girls. Cer- boys are treated very similarly once they are tainly, in parts of the world like Afghanistan, taken to health care facilities. There were no China, northern India, and Pakistan, such dis- differences between boys and girls (or between crimination is a serious problem. Studies have women and men) (�gure 3.11). In all seven shown delays in seeking medical care and lower countries, doctors spent the same time, asked expenditures for girls, and in the 1990s, even in a the same questions, and completed the same period of sharp economic growth, anthropomet- number of examinations regardless of the sex ric outcomes for boys improved faster than for of the patient.75 girls.69 An economic rationale for such discrimi- If neither households nor providers discrim- nation is a link to the structure of returns—for inate against girls, what is the source of excess instance, in districts within India, where the soil female mortality in early childhood? Historical is amenable to higher female labor use in agri- forensics provide the key to this puzzle. In the culture, excess mortality among females is lower early 20th century, European countries faced than the average in India.70 Higher women’s the same patterns—no girls missing at birth but wages are associated with greater female mobil- high excess mortality among girls in early child- ity and authority, while higher male wages have hood. Between 1900 and 1930, the excess mortal- the opposite effect.71 Beyond economics, the ity vanished in almost all of them (�gure 3.12). 126 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 domestic hygiene practices. In the United States, F I G U R E 3.10 There is little or no gender disadvantage in clean water and sanitation accounted for the en- vaccination rates, nutrition outcomes, or use of tire decline in infant mortality during this time health services when a child falls sick and in the disappearance of excess female mor- tality in infancy.76 Respiratory illness: Epidemiological changes caused by these Polio vaccination Taken to a health facility public health investments explain both declining 100 100 Boys disadvantaged Boys disadvantaged infant mortality and the disappearance of excess 80 80 female mortality in infancy and early childhood. girls, % girls, % 60 60 40 40 Between the early 1900s and 1930, the share of 20 20 infectious diseases as a cause of death declined, 0 Girls disadvantaged 0 Girls disadvantaged increasing the share of perinatal and congenital 0 20 40 60 80 100 0 20 40 60 80 100 factors.77 Although girls were (and are) more boys, % boys, % robust than boys for both infectious diseases and perinatal conditions, they are even more ro- Stunting prevalence Measles vaccination bust than boys for perinatal conditions relative Girls disadvantaged 100 Boys disadvantaged to infectious diseases. Boys always had a disad- 80 80 vantage in mortality, and as infectious diseases girls, % girls, % 60 60 declined, their disadvantage increased. Today, a 40 40 20 20 high burden of infectious disease in countries 0 Boys disadvantaged 0 Girls disadvantaged where poor public health systems do not pro- 0 20 40 60 80 100 0 20 40 60 80 100 vide clean water, sanitation, waste disposal and boys, % boys, % drainage, is part of the reason for higher relative female mortality risks in early childhood com- pared with the rich countries. Small differences do not explain the variation If this institutional-biological hypothesis in the fraction of excess deaths across countries were true today, one would expect to see less excess female mortality in countries with lower 60 The coefficient b1 on the explanatory variable in the regression: Excess deaths = b1 x +b2 is infant mortality. And that is indeed the case: the close to zero and almost always insignificant. stunting relationship between excess mortality for girls change in excess female deaths per 100,000 female population 40 and overall infant mortality is exactly the same in 2000 as it was in 1900 for the European coun- measles polio tries (see �gure 3.12). Bangladesh, China, and health facility 20 use in case of vaccination vaccination Vietnam, which have managed to reduce overall respiratory illness infant mortality through clean water and better sanitation, have also reduced the excess female 0 mortality in infancy and early childhood. But in female labor force much of West Africa, there has been less focus participation on clean water and sanitation: between 1990 –20 and 2005, the fraction of urban households with estimated regression coefficient 95% confidence piped water actually declined from 50 percent to on explanatory variable X interval 39 percent in 32 African countries. Not surpris- ingly, countries like Burkina Faso and Nigeria Source: Staff calculations based on data from the World Health Organization 2010; United Nations, Depart- ment of Economic and Social Affairs, Population Division 2009; World Development Indicators 2009; and Demographic Health Surveys 1985–2008. Note: Data for vaccination coverage and health service use were pooled over the years 1990–2008. This sharp decline—after virtually no change during the entire 19th century—was coinciden- “ Before pollution, we could cook and drink the river water. Now oil palm pollution has spoiled our river . . . we have no choice but we have to drink tal with large investments in public health, nota- the water from the river. bly clean water and sanitation (broadly de�ned to include waste disposal, drainage, toilets, and vector control), along with outreach to improve � Adult woman, rural Papua New Guinea Education and health: Where do gender differences really matter? 127 F I G U R E 3.11 Men and women, boys and girls, are treated the same when they visit health facilities Average period of time spent with practitioner at each visit and questions asked and examinations conducted by practitioner Average time spent % of essential questions % of essential examinations a. Children with patient (minutes) asked by practitioner conducted by practitioner 7 59 Afghanistan No data available 6 58 17 29 47 Burkina Faso 18 28 50 7 No data available No data available Mozambique 9 19 20 Rwanda No data available 25 27 7 71 48 Tanzania 45 6 69 83 Uganda No data available No data available 85 Average time spent % of essential questions % of essential examinations b. Adults with patient (minutes) asked by practitioner conducted by practitioner 7 88 86 Afghanistan 6 80 87 17 33 55 Burkina Faso 16.5 37 61 17 57 Mozambique No data available 15 61 13 14 Rwanda No data available 14 13 9 77 49 Tanzania 79 53 8 17 85 43 Uganda 14 86 42 female male Source: Ali and others 2011. Note: There were no signi�cant differences between boys (men) and girls (women) conditional on disease. have seen a much slower decline in early child- services that governments in most European hood excess female mortality. countries provided in the early part of the 20th Bringing down mortality risks for boys and century. Reducing the burden of infectious dis- girls in low-income countries today is largely a eases will produce declines in child mortality, question of providing the basic public health more for girls than boys. 128 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Excess mortality in adulthood—Women F I G U R E 3.12 Levels of excess female childhood mortality in Excess female mortality also affects women ages high-income countries in the early 1900s were 15–60, particularly women in their reproductive similar to those of low- and middle-income years (ages 15–49) who live in low- and middle- countries today . . . income countries. In this age group, excess female mortality has declined in absolute numbers and as a proportion of population in every region of excess female deaths per 100,000 female population, 1,500 the world except Sub-Saharan Africa, which di- Excess female mortality in childhood vides into two regions: the HIV/AIDS–affected disappeared in Europe countries, where excess female mortality has 1,000 between 1900 and 1930 increased even as a fraction of population; and those countries, mainly in Central and Western ages 1–4 Africa, where HIV/AIDS is less of a problem and where excess female mortality has declined, 500 albeit slowly. Two mechanisms drive excess mortality in the reproductive years—maternal mortality and morbidity related to childbirth, and HIV/AIDS. Maternal mortality is funda- 0 mentally different from excess female mortality at other ages in that, to reduce it, societies must 1900 1920 1940 1960 1980 2000 focus on an intrinsically female condition and Belgium Spain Italy speci�cally on improving the maternal health- England and Wales Switzerland Netherlands care system. Throughout this Report, “maternal France Denmark Portugal mortality� implies not only death during child- Japan Finland Sweden birth but also concurrent morbidities brought Norway on by the experience of pregnancy and child- birth. These include severe anemia (and its rela- tionship with malaria) and obstetric �stula.78 . . . and the excess female mortality declined with reduction in As in early childhood, adult women in some overall childhood mortality populations around the world experience sig- ni�cant discrimination in health expenditures 1,500 and health-seeking behavior. But, again as with excess deaths per 100,000 female population early childhood, this discrimination does not appear to be systematic. In countries ranging 1,000 from India to Egypt to South Africa, a small bias favors women in overall health expenditure and sometimes in use of the health system (chap- 500 ter 1). Clinical observations of practice in seven countries were also unable to uncover differ- ences in the way men and women were treated The two lines overlap by health providers (see �gure 3.11). While these 0 The relationship between excess �ndings could still be consistent with discrimi- female deaths and overall childhood nation, due to biological differences and differ- mortality is the same today as it ent health needs, they also suggest that evidence –500 was in the early 1900s of such discrimination will be dif�cult to �nd. 0 5 10 15 20 In contrast, the two issues discussed next— childhood mortality rate maternal mortality and HIV/AIDs—have very high-income countries, 1900–30 clear and obvious pathways to women’s health, low- and middle-income countries, 2008 particularly to their mortality. Maternal mortality Sources: WDR 2012 team estimates based on data from World Health Organization 2010; United Nations Department of Economic and Social Affairs 2009; and Human Mortality Database 2011, Max Planck Insti- In high-income countries, there were a total of tute for Demographic Research, and University of Califormia, Berkeley. about 1,900 maternal deaths in 2008. In India, Education and health: Where do gender differences really matter? 129 there were 63,000, and in Sub-Saharan Africa, 203,000 (56.7 percent of the global total). One FIGURE 3.13 Maternal mortality ratios declined steeply in of every 14 women in Somalia and Chad will die selected countries during 1930–60 from causes related to childbirth. As a propor- tion of all births, more women die in childbirth large declines 800 in the maternal in India today than did in Sweden at the begin- mortality rate ning of the 1900s—and in Liberia today than in between 1930–60 maternal deaths per 100,000 births Sweden in the 17th century.79 500 Between 1930 and 1960, the maternal mor- tality ratio—the risk of death for every birth— fell signi�cantly in developed countries (�gure 400 3.13). The ratio began to drop sharply in the late 1930s in most countries, driven in part by the introduction of sulfa drugs in 1936 and by an in- 200 crease in the number of institutional births with better care.80 The ratios then converged strongly across countries in the 1940s and 1950s—most 0 countries reached modern levels in the early 1900 1920 1940 1960 1980 2000 1960s; Italy, Japan, and Portugal reached those year levels in the mid-1970s. Declines were sharper in the Anglo-Saxon countries relative to the Belgium Spain Italy Nordics, which already had low maternal mor- England and Wales Switzerland Netherlands tality rates in 1935. France Denmark Portugal Japan Finland Sweden The United States stands out as the coun- Norway try with the highest maternal mortality ratio for 1900–30, but like the others, sharp declines Source: Albanesi and Olivetti 2010. began around the mid-1930s and fell to cur- rent levels by 1960. These declines were brought about largely by simultaneous improvements in a spike downward coinciding with World War I the medical system at the point of delivery and and a peak in 1918 with the flu epidemic) and in services to pregnant women, and by shifts in then declined sharply to zero between 1930 and expectations of where to deliver––from home to 1960.82 The late declines are precisely for coun- hospital.81 tries—Italy, Japan, and Portugal—where mater- The patterns are fully reflected in changes nal mortality rate declines occurred latest. in excess female mortality in the reproductive- For all countries in 1990–2008 and for high- age groups for selected countries (�gure 3.14). income countries with historical data, the basic For these countries, excess female mortality in pattern remains similar with higher maternal adulthood remained fairly high until 1930 (with mortality ratios associated with greater excess “ Nobody in this village has access to drinking water. People bring water from a spring and a water pool that are at a 100–1,000 meter distance from the village. Those also dry out during some seasons and the children and especially the girls spend more than five hours daily to bring water. . . . Between the neighboring villages, sometimes violence occurs over drinking water and residential places because there is not enough water for people and they don’t have shelters and they make homes for themselves in the desert. And they sometimes fight each other over this. � Adult man, rural Afghanistan 130 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 female mortality in adulthood. Reducing mater- F I G U R E 3.14 High income countries today had excess nal mortality rates is thus critical for reducing female mortality at the reproductive ages excess female mortality in adulthood. This can during the �rst half of the 20th century . . . be done in one of two ways—reducing fertil- Excess mortality among women in 13 European ity so that women are less exposed to the risk countries, 1900–2000 of death (including the risk from unsafe abor- tions) or reducing the maternal mortality ratio 1,000 excess female (the risk of death for every birth). mortality in the Take each in turn. When fertility rates are reproductive ages disappeared in high, reductions in the rate will reduce the risk 800 Europe between of dying from causes related to childbirth. Ma- 1930 and 1960 excess female deaths per 100,000 of female population, ages 15–49 ternal mortality risks depend on the age of the mother (slightly higher in young ages than the 600 average across age groups and then increasing in older ages in a “J� shape); parity, or number of children borne by a woman (the �rst preg- 400 nancy and higher parities increase mortality risks to the mother); independent cohort ef- fects (mothers born in cohorts that smoked 200 more, for instance, would have higher risks); World and time effects (later decades imply better War I medical care). In rich countries, changes in ma- 0 ternal age and shifting parity distributions ac- 1900 1920 1940 1960 1980 2000 counted for 18 percent of the decline in mater- Belgium Spain Italy nal mortality.83 Studies in a limited number of England and Wales Switzerland Netherlands low-income countries (typically subnational) France Denmark Portugal report similar estimated reductions of around Japan Finland Sweden 25 percent, using models that eliminate births Norway after parity �ve and those outside the safest ages for birth (20 to 39). 84 So, changes in the age and parity structure of birth in high fertil- . . . and the excess mortality at all income levels declines with ity contexts could reduce maternal mortality reductions in maternal mortality by 20 percent or so. 1,000 The studies from low-income countries were for periods when total fertility rates The relationship between excess were high—more than 6 births per woman in excess female deaths per 100,000 female population aged 15–49 800 female mortality at 15–49 and Bangladesh, for example. Today, fertility rates maternal mortality has not changed much over time are that high in just six countries—Afghani- stan, Chad, Niger, Somalia, Timor-Leste, and 600 Uganda. Forty countries have fertility rates higher than 4, and all but Afghanistan, Gua- 400 temala, Papua New Guinea, Timor-Leste, and the Republic of Yemen are in Sub-Saharan Af- rica. For the most part, however, fertility has 200 declined dramatically over the past 30 years in most low- and middle-income countries. 0 In countries with fertility at 3 or less, further 0 200 400 600 shifts in the age and parity distributions may maternal deaths per 100,000 live births not have large impacts on the maternal mor- tality ratio. Because family planning policies account for 10–15 percent of the reductions Sources: WDR 2012 team estimates based on data from World Health Organization 2010; United Nations Department of Economic and Social Affairs 2009; Human Mortality Database 2011, Max Planck Institute for when starting from initially high levels, these Demographic Research, and University of California, Berkeley; and Albanesi and Olivetti 2010. policies could continue to help in selected Education and health: Where do gender differences really matter? 131 F I G U R E 3.15 What explains excess mortality among girls and women in the reproductive ages? S UTION INSTIT INFORMAL ECONOMIC OPPORTUNITIES MARKETS HOUSEHOLDS ENDOWMENTS AGENCY Excess female deaths FOR MAL Imp INST rove ITUT and i ION hum nfrastr S an r uctu eso re urce s Source: WDR 2012 team. countries but are unlikely to substantially Women are biologically 1.2 times more likely reduce the number of women who die from to acquire the virus because women’s bodies are childbirth every year.85 The best way to reduce more susceptible to infection than men. Sexu- the unacceptably high number of women who ally active young women, whose bodies are still die from childbirth (and related causes) every developing, may be especially vulnerable. Sexu- year in low-income countries will likely be re- ally transmitted diseases (such as herpes simplex gion speci�c and will depend both on fertility virus type 2) that affect men and women differ- reductions and speci�c policies to reduce the ently also contribute to the greater susceptibility maternal mortality ratio (�gure 3.15). of women to HIV infection.90 With respect to behavior, women date and marry men who are HIV/AIDS a few years older, which also contributes to a dif- In addition to maternal mortality, the HIV/ ferential age-gradient among men and women AIDS epidemic is contributing to excess female in HIV infection rates. mortality in Africa (�gure 3.16). In Sub-Saharan Without treatment, HIV infection develops Africa, women account for 60 percent of all into AIDS and, after 7–10 years, death. But the adult HIV infections,86 with the gender gap in HIV/AIDS link to excess female mortality in prevalence largest for younger adults. The ratio Africa is not relevant for all countries in the of female to male prevalence for 15–24 year olds region, particularly those in Central and West is 2.4 across Sub-Saharan Africa.87 Comparisons Africa. It is concentrated among the set of high- of age-infection pro�les for women and men prevalence countries in Southern Africa and show that after age 34, HIV prevalence rates are parts of East Africa, which bear a dispropor- similar for men and women.88 tionate share of the burden of AIDS in Africa as Biology and behavior have both contributed well as globally (box 3.2). to greater prevalence of HIV among women— Not only has HIV/AIDS hit women the hard- referred to as the “feminization of AIDS.�89 est, but coping with the crisis has had system- 132 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 F I G U R E 3.16 Excess female mortality by age in four countries with high HIV prevalence a. Botswana b. Kenya excess female deaths, thousands excess female deaths, thousands .8 10 .6 5 .4 0 .2 0 –5 0 20 40 60 0 20 40 60 age, years age, years c. Uganda d. South Africa 10 20 excess female deaths, thousands excess female deaths, thousands 15 5 10 0 5 0 –5 –5 0 20 40 60 0 20 40 60 age, years age, years 1990 2000 2008 Source: WDR 2012 team estimates based on data from World Health Organization 2010 and United Nations, Department of Economic and Social Affairs 2009. wide impacts on the delivery of health services. coverage is an extraordinary achievement: as re- Prenatal care, care during birth, and children’s cently as 2003, only a few privileged HIV/AIDS vaccination rates have suffered where HIV rates patients had access to ART in Africa. And it will are the highest in Sub-Saharan Africa.91 reduce the number of deaths from HIV/AIDS— Encouragingly, these patterns are now chang- and decrease female mortality rates in adult- ing, and so will excess female mortality, both as hood. Botswana, Kenya, and Uganda—which life-prolonging treatment becomes available have high ART coverage rates—experienced a and as incidence rates change. As of end-2009, reduction in excess female mortality between an estimated 5.2 million people in low- and 2000 and 2008 (although the levels are still sig- middle-income countries were receiving anti- ni�cantly above those in 1990). In contrast, with retroviral therapy (ART) and in Sub-Saharan a high HIV prevalence, large population, and Africa, nearly 37 percent of people in need of slower expansion of ART, South Africa saw a treatment could obtain those life-saving medi- steady increase in excess female mortality from cines.92 In the countries with the highest preva- 1990 to 2000 and a further increase to 2008. lence, treatment coverage varies: 83 percent in Botswana, 48 percent in Malawi, 36 percent in Excess mortality in adulthood—Men South Africa, 43 percent in Uganda, 68 percent In some countries, the analysis illustrates pat- in Zambia, and 34 percent in Zimbabwe. This terns of excess male mortality. In the formerly Education and health: Where do gender differences really matter? 133 F I G U R E 3.17 In some countries, there is excess male mortality Eritrea Mexico 4 excess female deaths, thousands excess female deaths, thousands 0 2 –2 0 –4 –2 –6 0 20 40 60 0 20 40 60 age, years age, years Russian Federation Thailand 0 5 excess female deaths, thousands excess female deaths, thousands –20 0 –40 –5 –60 –10 0 20 40 60 0 20 40 60 age, years age, years 1990 2000 2008 Source: WDR 2012 team estimates based on data from World Health Organization 2010 and United Nations Department of Economic and Social Affairs 2009. Note: A value on the y axis less than zero implies excess male mortality. For example, at age 20, about 5,000 excess male deaths occurred in Eritrea in 1990. socialist countries, in countries and regions with Outside situations of warfare, violence re- high rates of violent crime or periods of war and mains gendered: men stand charged for 80–90 conflict, and in those experiencing localized epi- percent of all violent crimes in Australia, Europe demics of HIV in parts of the male population, and the United States, and the higher prevalence men die at a signi�cantly higher rate relative to of male criminal behavior translates into higher men in high-income countries, after accounting incarceration rates for men.94 More male deaths for overall mortality conditions in their country result from homicide than from armed conflict. of residence. Of all violent deaths globally in 2000, just un- First, men are the immediate victims of der a third were victims of homicide, the vast armed conflict. A study of conflicts in 13 coun- majority among men. Latin America had the tries during 1955–2002 found that 81 percent highest homicide rate (27.5 per 100,000), more of violent war deaths were among males.93 An than three times the rate reported in any other example is Eritrea, where a 30-year war of inde- region. Mexico, in particular, as a substantial pendence ended in 1991. Figure 3.17 illustrates transit and production hub for cocaine, mari- the signi�cant excess male mortality that oc- juana, and other illicit drugs, experienced most curred there in the last full year of the war. of the region’s drug-related violence.95 134 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 Most victims of homicide in Latin America Higher male mortality, whether from sub- are young males from low socioeconomic back- stance use, violent crime, or risky sexual behav- grounds, and rates peak among males between ior, is not a feature of these low- and middle- the ages of 15 and 24 years.96 In addition to income settings alone. These behaviors also its relation with drugs and turf battles among prevail among men across the industrialized competitive cartels, violence in Latin America world, where women live on average 5 to 10 has also been linked to ideals about masculin- years longer than men. This female advantage in ity that force men to confront others when chal- longevity is not purely genetic (and where ge- lenged. Studies of youth violence in República netic factors play a role, they continue to be mal- Bolivariana de Venezuela, for example, show the leable). Studies show that differences in mortal- importance among males of ‘‘earning’’ respect ity are sensitive to environmental factors. Men in front of others, and acts of violence are one smoke more than women, have poorer diets, way to achieve this status.97 Excess male mor- and internalize stress differently. Improvements tality in Latin American countries is precisely in along any of these dimensions would narrow the 20–40 age group where such behavior be- the gender gap in life expectancy. comes salient. This �nding raises a more pertinent question: Second, the formerly socialist countries are the What are the factors that mediate men’s risk- location of a stark male disadvantage in health. taking, substance use, and increased health risks? In addition to the economic and social upheavals Do societies condone these behaviors among the populations of many of these countries en- men, because they reflect underlying ideals about dured, they also experienced sharply rising death “masculinity� or “manliness,� at least up to a rates—the only region other than Sub-Saharan certain point? A growing body of literature sug- Africa to do so during the last three decades. In gests this might indeed be the case, and at least Russia, male life expectancy at birth declined by in the high-income countries health systems have 6.6 years between 1989 and 1994 (from 64.2 to started to emphasize both “behavior� change and 57.6 years), and prime age men were hit hardest. institutional improvements. Although policy le- Studies have identi�ed increased alcohol con- vers to effect behavioral change are multifaceted sumption and psychosocial stress, likely brought and dif�cult to pinpoint (after all, they require on by changes in the economic environment and a shift from understanding the body to under- weakening social safety nets, as the primary fac- standing the brain), the success of antismoking tors causing the spike in mortality among men.98 campaigns in the West suggests that this challenge More broadly, deaths and disability rates related is now being addressed around the world.104 to alcohol and substance abuse are higher for men than for women across the world.99 Poor institutions and bad default options In Thailand, male mortality in the late 1980s To repeat: missing girls at birth and excess fe- rose as the �rst wave of the HIV epidemic struck male mortality below the age of 60 account for intravenous drug users, mostly men. Because the nearly 3.9 million women a year. Excess male epidemic diffused into the general heterosexual mortality, primarily resulting from conflict and population much later, from commercial sex risky behavior, also leads to unnecessary deaths. workers to their clients, and then from clients to There has been little change in the numbers in their spouses, the peak in mortality among men the past three decades, and in several countries occurred earlier and reached a higher level than in Sub-Saharan Africa, mortality risks are wors- among women.100 In addition, proportionately ening. The solutions for reducing female mor- more women (53 percent) received antiretrovi- tality risks after birth are largely institutional: ral therapy than men, who typically presented clean water and sanitation for infancy and early with a more advanced stage of the disease than childhood, and better care for expectant moth- women.101 Studies note the role of social norms ers, and reductions in HIV/AIDS and improved that condoned risky sexual behavior among family planning services in some countries for men, for whom visits to commercial sex work- the reproductive years. ers before and after marriage, often in the com- Skeptics may question why a century of in- pany of peers, were widely accepted.102 Women, creased medical knowledge and better medical by contrast, were expected to be abstinent before care has not reduced the salience of the institu- marriage and faithful afterward.103 tional provision of clean water and sanitation or Education and health: Where do gender differences really matter? 135 B OX 3.2 Four Africas Between 1980 and 2000, school enrollments in most of Sub- are between 150 and 200 per 1,000 live births, and while school Saharan Africa increased and mortality in early childhood decreased enrollments have increased in recent years, they are still low. (albeit more slowly than in other low-income countries), but mor- tality risks for adults increased. Data from the Demographic and HIV/AIDS Africa: The third Africa consists of countries with high Health Surveys show that the increases were largest among men HIV/AIDS prevalence. At the end of 2009, in countries such as with less than primary education—although mortality increased Botswana, Lesotho, Swaziland, South Africa, Zambia, and Zim- for both men and women and for those with and without primary babwe, about one in six to one in four adults between the ages schooling.a Ten years later, after a period of rapid growth in many of 15 and 49 were living with HIV/AIDS. Mortality risks during early African countries, where do mortality risks stand? And how does childhood, school enrollments, and fertility rates are similar to Sub-Saharan Africa compare with countries in South Asia, such as those in India today. In 1990, mortality profiles for men and women Afghanistan, India, and Pakistan, usually thought of as places with in Botswana and South Africa were similar to those in high-income high gender discrimination? There are now four Africas, each of countries today. But by 2000, mortality risks increased in adult- which has a different effect on women’s ability to acquire and enjoy hood, more so for women. In 1980, mortality risks were higher for a healthy life: progressive countries throughout the continent, more educated men, while women with more schooling and both where education levels are high and mortality risks low; the HIV/ men and women in urban areas had a mortality advantage. By AIDS Africa, primarily in the South; conflict countries like Eritrea and 2000–04, less educated men and women had higher mortality risks Liberia; and, curiously, West and Central Africa. than their more educated counterparts, and mortality rates among both men and women in rural areas were equal to those of their Progressive Africa. Countries such as Ethiopia, Ghana, Madagas- urban counterparts. car, and Togo have largely escaped the HIV/AIDS epidemic. Mortal- ity rates of children under age five are around 100 per 1,000 live “ births (under 76 in Ghana). Excess female mortality after birth is lower, and school enrollments are relatively high. Reductions in fer- The worst abuse happening today is when tility rates have decreased exposure to mortality risks during child- a man infects a woman with HIV/AIDS . . . birth; total fertility rates are now between 4 and 5. In health and Unprotected sex causes fights between a education, these countries look like Pakistan but with somewhat higher enrollments at the primary school level. Fertility rates are man and a woman, because a woman would still higher than in India (2.7) and Pakistan (3.9), as are under-five say she wants to use condom but a man � mortality rates. would refuse. Adult woman, South Africa Conflict Africa: Sub-Saharan Africa has experienced two types of conflicts over the past three decades. During the 1980s and 1990s, outright war in countries like Eritrea and Liberia claimed the lives of many young men. Except for periodic flare-ups, these are decreas- Central and West Africa: The real puzzles in Sub-Saharan Africa ing over time. Yet the effects last. In Bargblor Town in Liberia, no one are the Central and West African countries, including Burkina Faso, has access to electricity, piped water, public stand pipes, or a sew- Chad, Mali, Niger, and Nigeria, among others (Somalia is very similar age system. There is no public transport, and the nearest hospital to these countries). Except Ghana and Senegal, many of these coun- takes approximately 3½ hours to reach by foot. In an emergency, tries have seen either no change or a worsening of overall mortality women report having to run or walk to the hospital. Children who risks during this time. In these countries, mortality risks for women die in the hospital are carried home to their village to be buried. In have systematically increased. Women with more than primary other countries, widespread civil conflict continues to extract a schooling have seen the greatest increase in the risk of dying, heavy toll among women. A recent study suggests that in the Dem- although even in 2000, urban and educated women still had lower ocratic Republic of Congo, 29 of every 1,000 women were raped mortality than other groups. Today, Burkina Faso, the Central Afri- between 2006 and 2007—58 times the annual rate in the United can Republic, Chad, Mali, Niger, and Nigeria look very much like States.b Excess female mortality increased between 1980 and 2008. Afghanistan in their mortality risks, fertility rates, and girls’ school- Total fertility rates hover between 5 and 6, under-five mortality rates ing. In these countries, mortality under the age of five ranges from 170 to 220 (Afghanistan is higher, at 257), total fertility rates range from 4.5 to above 7 (Afghanistan is 6.6), and adult mortality risks are virtually the same as those in Afghanistan. Their enrollments in pri- “ � Men are just like us, we all go through this hell together. Adult woman, Liberia mary and secondary school also mirror the Afghan data. These findings are puzzling because the countries in this group have little in common. Some are landlocked, some coastal; some are anglophone, others francophone; some fast-growing, others slow-growing, and some have seen conflict and some have not. The 136 WO R L D D E V E LO P M E N T R E P O RT 2 0 1 2 B OX 3.2 Four Africas (continued) one thing they do have in common is that women’s livelihoods, the continent, particularly Central and West Africa. Of all papers already weak, are deteriorating. published in the top 202 economics journals between 1985 and 2004, 149 papers were on Pakistan and 1,093 on India—but there The four Africas highlight the dramatic effects of HIV/AIDS and were no papers on the Central African Republic, 1 on Chad, 14 on conflict—and the fact that poor institutions and service delivery Benin, 2 on Guinea Bissau, and 20 on Niger. Only for Burkina Faso can harm women just as much, or even more, than outright gender (47) and Nigeria (148) do the numbers start picking up. Before decid- discrimination. In determining where the international community ing what to do, the global community should seek to understand should focus, it is worth pointing out just how little is known about what is going on.c a. de Walque and Filmer 2011. b. Peterman, Palermo, and Bredenkamp 2011; Shannon 2010. c. Das and others 2009. better maternal care. Even without clean water, there—that baby is burning with a fever. parents now know to boil water before giving I walked all the way yesterday to a nearby it to children. Even when children get diarrhea, town to get a pill. I spent whatever money most parents know how to treat them. House- I had earned in the town yesterday and holds can always take pregnant mothers to hos- trudged back late at night, but the pill has pitals. These kinds of private solutions run into made no difference.� I touched the baby’s two problems. First, private actions do not take forehead—it was burning and he was suck- into account the fact that sick children infect ing at the breast desperately but the mother other children as well. These “externalities� for did not seem to have much milk. I asked the infectious diseases generate well known under- mother if she had eaten anything. Now the investments in private behavior. others joined in the conversation and said Second, while private actions work well when that they were waiting for some of the men there are few key choices to be made, reality can to return. There was no food in the house. be very different. Poor people around the world They would cook something if the men are forced to make many, many choices—most managed to earn some rice or some coarse of them bad—every time a child falls sick: where grain. Concerned about the baby’s condi- do they get �rewood to boil the water, where tion, I asked if they knew about oral rehy- should they get sugar, should they take the child dration solution. They did not know any- to the doctor who could be �ve hours away, how thing about it—the ANM (Auxiliary Nurse much will the doctor charge, should the mother Midwife) never came up to the hamlet.. . . . wait for the husband if the child needs to be car- The Sarpanch (headman) who was ac- ried? Each choice can have devastating conse- companying me was getting quite defensive. quences if things don’t pan out. He said if these people do not come down, if Poor people everywhere have to choose and they do not tell us what troubles them, how make many decisions about many things that can we help them? I asked the Sarpanch to richer people take for granted every day. When explain to the young mother and the older institutions are bad, so are people’s default woman sitting next to her how important it choices—and “free to choose� becomes “forced was for the baby to receive fluids. In his dia- to choose.� Under these circumstances, many ill- lect, he began to explain. “You have some nesses and many life choices create excess female salt in the house, don’t you? Well take this mortality. much sugar, put in this much water and To illustrate these points, look at this example boil it and then put a pinch of salt in it and of providing oral rehydration solution to a sick squeeze a few drops of lemon. No sugar in child in a poor village: the house? Yes, but go down to someone on the lower hamlet—they may not give The tall man pointed to the young woman you sugar if it is for yourself, but if you suckling her baby and said, “Look, look say that it is needed to save the life of the Education and health: Where do gender differences really matter? 137 child, they will surely give you a �stful.� The the doctor was called, by the time the doctor woman nodded. “Where will you get the could be found, by the time the ambulance water from?� I asked. Now a new problem went to �nd the doctor, by the time the doc- arose, for the nearest pump was not work- tor came, by the time the husband went out ing. They were all drinking water from a to buy drugs, IV set, drip, and bottle of ether, stream nearby that was stagnant and dirty. by the time the husband went round to look The Sarpanch told the baby’s grandfather for blood bags all round town, by the time that the water must be boiled and cooled. the husband found one and by the time the I was beginning to see the hopelessness of husband begged the pharmacist to reduce the situation. No sugar, no source of clean the prices since he had already spent all his drinking water, and a shortage of fuel. But money on the swabs, dressings, drugs, and the man who was inebriated again got ag- fluids, by the time the hematologist was gressive, “Whatever you say, we will not go called, by the time the hematologist came to anyone’s door to beg.� The women were and took blood from the poor tired hus- listening more intently and I thought they band . . . by the time the day and night intended to follow it up. “But do not just nurses changed duty, by the time the day feed it to him all at once,� I said, “give it in and night doctors changed duty, by the time small sips.� (How shall I demonstrate that?) the midwife came again, by the time the The woman took a leaf, folded it in a kind doctor came, by the time the t’s had been of spoon and said, “like this?� The Sarpanch properly crossed and all the i’s dotted and promised to help by getting a packet of oral the husband signed the consent form, the rehydration solution.105 woman died. Today the husband wanted to sell the drugs and other things they never This is not an isolated story. Take maternal used to be able to carry the body of his wife mortality. Technically, much maternal mortal- back to their village but he could never trace ity can be reduced if treatment is prompt and [the body] again in the hospital.� 106 adequate. Yet that is easier said than done. A pathbreaking study from Ghana, Nigeria, and Every step of the way to delivering a baby Sierra Leone illustrates that even when women safely is fraught with problems, beginning with are taken to hospitals, delays in receiving care the recognition of danger signs.107 Even when can lead to devastating outcomes: danger signs are recognized, families “are also aware that there is not much the medical facil- Today, Mary, the lady who helps us in the ity can do for her when there is no trained doc- house, came late to work. I told her off for tor or nurse-midwife, when blood shortages being late and asked why. She said that one are regular, and when equipment is frequently of her townswomen . . . had died in the hos- broken. People do not bother to seek care when pital while giving birth to a baby. This was they know that they probably will not be cured,