THE COST OF GENDER INEQUALITY NOTES SERIES HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY QUENTIN WODON, ADENIKE ONAGORUWA, CHATA MALÉ, CLAUDIO MONTENEGRO, HOA NGUYEN, AND BÉNÉDICTE DE LA BRIÈRE FEBRUARY 2020 THE COST OF GENDER INEQUALITY NOTES SERIES HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY QUENTIN WODON, ADENIKE ONAGORUWA, CHATA MALÉ, CLAUDIO MONTENEGRO, HOA NGUYEN, AND BÉNÉDICTE DE LA BRIÈRE BACKGROUND TO THIS SERIES Reducing gender inequality makes economic sense substantial gender gaps in earnings, which in turn apart from being the right thing to do. Achieving decrease women’s bargaining power and voice. In gender equality and empowering all women and girls addition, many girls are married or have children is the fifth sustainable development goal and is a top before the age of 18, before they may be physically priority for governments. Countries can achieve this and emotionally ready to become wives and mothers. goal if they take appropriate steps. This note is part Women and girls also face higher risks of gender-based of a series that aims to measure the economic cost of violence in their homes, at work, and in public spaces. gender inequality globally and regionally by examining Their voice and agency is often lower than that of men, the impacts of gender inequality in a wide range of areas whether this is within the household, at work, or in and the costs associated with those impacts. Given that national institutions. This also affects their children. gender inequality affects individuals throughout their For example, children of young and poorly educated life, economic costs are measured in terms of losses in mothers often face higher risks of dying by age five, human capital wealth, as opposed to annual losses in being malnourished, and doing poorly in school. Gross Domestic Product (GDP) or GDP growth. The Fundamentally, gender inequality disempowers women notes also aim to provide a synthesis of the available and girls in ways that deprive them of their basic evidence on successful programs and policies that human rights. contribute to gender equality in multiple areas and achieve the Sustainable Development Goals (SDGs). This lack of opportunities for girls and women entails large economic costs not only for them, but also for While gender parity in basic education has been their households and countries. Achieving gender achieved globally, in many low income countries, girls’ equality would have dramatic benefits for women and educational attainment remains lower than boys at the girls’ welfare and agency. This, in turn, would greatly secondary level and adult women are less literate than benefit their households and communities, and help men. Apart from these gender gaps in educational countries reach their full development potential. It would attainment, discrimination and social norms shape the reduce fertility in countries with high population growth, terms of female labor force participation. Women are as well as reduce under-five mortality and stunting, less likely than men to join the labor force and to work thereby contributing to ushering the demographic for pay. When they do, they are more likely to work transition and the associated benefits from the part-time, in the informal sector, or in occupations that demographic dividend. have lower pay. These disadvantages translate into 1 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 TABLE OF CONTENTS DOMAIN 5: AGENCY, DECISION-MAKING, AND BACKGROUND TO THE SERIES 1 46 SOCIAL CAPITAL KEY RESULTS 3 Women’s Decision-Making 46 Two Pillars of the Gender Equality Agenda 4 Birth Registration 47 Impacts of Gender Inequality by Domain 5 SELECTED POTENTIAL ECONOMIC COSTS 48 Measurement Approach and Baseline Wealth Economic Costs of Gender Inequality 7 48 Estimates Loss in Human Capital Wealth from Gender Summary of Key Findings 9 49 Inequality in Earnings Loss in Human Capital Wealth from Under- Selected Policy Options to Achieve Gender Equality 9 51 five Stunting Conclusion 13 52 Loss in Welfare (Total Wealth Per Capita) from High Population Growth INTRODUCTION 14 FRAMEWORK AND METHODOLOGY 17 Budget Costs from High Population Growth 53 POLICY OPTIONS FOR GENDER EQUALITY 55 DOMAIN 1: LABOR, EARNINGS, AND 19 PRODUCTIVITY INVESTMENTS IN YOUNG CHILDREN 56 Lifetime Earnings and Human Capital Wealth 19 INVESTMENTS IN ADOLESCENT GIRLS 57 Gender Gaps in Human Capital Wealth 20 General Conditions for Schooling and Learning 58 Selected Factors Leading to Gender Gaps in Earnings 24 Interventions to Delay Marriage and Childbearing 60 and Improve Sexual and Reproductive Health DOMAIN 2: CHILD MARRIAGE, EARLY CHILDBEARING, AND EDUCATIONAL 29 INVESTMENTS IN ADULT WOMEN 62 ATTAINMENT Addressing Time Use Constraints 63 Child Marriage 29 Facilitating Access to Productive Assets 64 Early Childbearing 30 Solving Market and Institutional Failures 66 Educational Attainment 33 TARGETING 69 DOMAIN 3: FERTILITY AND POPULATION 37 CONCLUSION 71 GROWTH Total Fertility 37 APPENDIX 1: DATA AND METHODOLOGY 73 Modern Contraceptive Use 39 Data Sources 73 Population Growth 40 Methodology 73 DOMAIN 4: HEALTH, NUTRITION, APPENDIX 2: HUMAN CAPITAL WEALTH 41 76 WELL-BEING, AND VIOLENCE ESTIMATES Children’s Health and Survival 41 REFERENCES 78 Women’s Health and Violence 44 FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 2 KEY RESULTS based on analysis for a core set of 19 countries located mostly in sub-Saharan Africa and South Asia (see Appendix 1 for a Gender inequality remains pervasive worldwide. While in list of those countries and the data used, as well as the reason some countries boys and men may be at a disadvantage in why those countries were selected). some areas, in most countries girls and women continue to bear the brunt of gender inequality. For this reason, The hope is that the associations documented in the study this study focuses on the impacts of gender inequality on help illustrate the wide-ranging potential impacts and cost girls and women. To make the case for more and better of gender inequality, and in this way foster greater policy investments to reduce gender inequality, the study provides mobilization towards achieving gender equality. While the estimates of the impacts and economic costs of gender study pulls together in one place results on potential impacts inequality in five main domains of interest: (1) earnings and costs in many domains, as noted in Box ES.1, the analysis and standards of living; (2) educational attainment, child only provides an order of magnitude of potential impacts marriage and early childbearing; (3) fertility and population and costs, not precise or definitive values. To realize the growth; (4) health, nutrition, well-being, and violence; and economic benefits that could arise from reductions in gender (5) agency, decision-making, and social capital. Multiple inequality, countries will need to make the investments development outcomes affected by gender inequality are necessary to ensure that girls and women get equal considered. For some outcomes, estimates of impacts – or opportunities. Such investments have initial costs, but they rather correlations – are obtained using household survey pay off through higher standards of living and gains in human data for more than 100 countries. For other outcomes that capital wealth leading to long-term growth. may be more salient in developing countries, results are 3 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 BOX ES.1: CONTRIBUTION AND LIMITS OF THE STUDY This study summarizes findings from research on the potential negative impacts of gender inequality on development outcomes and related economic costs. The fact that investing in girls and women is essential for development is not new. This point was made by pioneers such as Boserup (1970) and more recently by a wide range of authors and organizations including – just to cite a few, Klasen and Lamanna (2009), Duflo (2012), World Bank (2012), Agenor and Canuto (2013), Elborgh-Woytek et al. (2013), Cuberes and Teigner (2015), McKinsey Global Institute (2015), Kabeer (2016), International Labour Organization (2018), Ostry et al. (2018), and World Economic Forum (2020). The purpose of this study is to illustrate the potential negative impacts of not investing in girls and women with more recent survey data, new measures, and for a larger set of countries than done so far. By pulling together evidence on the associations between gender inequality and multiple socio-economic domains in many countries, the analysis can help foster greater mobilization for gender equality. The framework for the study follows similar work devoted to the economic impacts of child marriage (Wodon et al., 2017), and the cost of not educating girls (Wodon et al., 2018). This study integrates and updates results from a previous analysis of the cost of gender inequality in earnings (Wodon and de la Brière, 2018). As with any empirical work of this nature, estimates of potential impacts and costs are subject to two important caveats. First, estimates from available observational data do not permit establishing causal relationships. Thus, when referring to potential impacts, the analysis should be taken as only suggestive of what might be achieved with gender equality for girls and women and related policy changes. What is measured are associations between aspects of gender inequality and other development outcomes. For several of the outcomes considered, whether these associations reflect casual relationships can be corroborated by evidence from empirical studies that are able more credibly to establish causality. But for other outcomes, such as impacts on decision making, the ability to engage in altruistic behaviors, or perceptions of well-being, fewer such studies are available. Second, simulations of the benefits of achieving gender equality obtained from the estimates of potential impacts do not account for broader effects in the economy arising from an expansion in opportunities for girls and women. The economics literature suggests that these effects could be sizable. For example, labor market earnings for men may be affected when women enter fields in which they were previously rarely active. Such potential general equilibrium effects are not reflected in this study. Assuming no changes for men in various areas may lead to under- or over-estimation of some of the effects suggested in the study. TWO PILLARS OF THE GENDER EQUALITY have high returns (this argument is not new; see for AGENDA example National Research Council, 2005; Levine et al., 2008; Heckman and Mosso, 2014) There are at • Gender inequality impacts women throughout their least three reasons why investing in adolescent girls – life, but its effects are especially detrimental in and even in younger girls may be especially beneficial. adolescence. The impacts of gender inequality are visible First, earlier investments tend to bear fruits that persist throughout women’s lives, from early childhood to old throughout a woman’s life after the intervention. If a age. This implies that interventions and policies are girl completes her secondary education, this generates necessary to reduce gender inequality throughout the benefits – such as lower fertility and higher labor force life cycle. For example, research suggests that gender participation – for many years afterwards. Second, stereotypes are formed at an early age. Therefore, the cost of interventions in adolescence, or in some finding ways to support changes in attitude at early cases even earlier, tends to be lower than the cost of ages can be highly beneficial later on. At the same time, interventions implemented later in life. This is especially programs aiming to reduce gender inequality through the case for investments in early childhood to prevent investments in adolescent girls are especially likely to some of the impacts of gender inequality on young FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 4 children, boys and girls. Third, at a formative age, expected to be earned by men in the future, women interventions may be more successful in influencing are expected to earn only two thirds of what men earn. values and behaviors. Later in life, it may become There has been only slow progress over the last two more difficult for girls and women to fully benefit from decades towards lower gender inequality in lifetime new opportunities provided to them. For example, earnings as measured through human capital wealth. preventing early marriage may help in strengthening In 1995, women were expected to contribute 58 cents women’s agency within the household. This does not on the dollar in comparison to men. Twenty years mean that new opportunities should not be provided later the proportion was 63 cents. At current rates to and investments made in women in adulthood – of progress (five cents in 20 years), it could take 150 examples of such interventions are provided in this years to reach parity. Two main factors lead women to study. But adolescence is a crucial time during which have lower lifetime earnings than men. First, they have investments in girls may yield the highest returns. lower labor force participation rates and work fewer hours in the labor market than men. Second, they tend hile adopting adequate laws and broad policies is • W to be paid less well when they are in paid employment. a first step, targeted programs are needed in many These factors keep many women in a productivity contexts. Assessments of legal frameworks conducted trap driven by many factors, including social norms by the Women, Business, and the Law program at the relegating them to household care responsibilities World Bank (2020) suggest that countries are making or unpaid work. However, when measuring the some progress, albeit slowly, in adopting adequate returns to educational attainment for women, they laws towards gender equality. However, much more tend to be as large as those observed for men. is needed as laws by themselves are not sufficient. For example, two thirds of all child marriages take • Educational attainment, child marriage, and early place below the minimum age for marriage adopted childbearing. Globally, girls have caught up with boys in by countries in their national legislations. Beyond laws attainment for basic education. Nine in ten girls (89.3 and broad-based policies, targeted interventions are percent) complete their primary education, and three still needed in multiple areas to achieve larger gains in four (76.0 percent) complete their lower secondary towards gender equality. This note focuses especially on education. For boys, the proportions are 89.9 percent two types of programs that may have especially large and 75.3 percent. However, in low income countries, economic benefits: (i) programs helping adolescent girls substantial gender gaps in attainment persist for basic to remain in school (or facilitate the school-to-work education, especially at the secondary level where the transition) and delay marriage and childbearing; and completion rate for girls at 36.9 percent is below that (ii) programs enabling adult women to improve their of boys. Part of this gap is due to persistently high rates economic opportunities. The focus on these two types of child marriage (marrying before the age of 18) and of programs does not mean that other interventions are early childbearing (having a first child before 18) in not needed – simply such programs to address gender many low income countries. In turn, early childbearing inequality are known to be fairly effective and have appears to be mostly due to child marriage in many especially large economic benefits (on girls’ education (but not all) countries. Across two dozen low and and empowerment, see among others Unterhalter et middle income countries for which estimations have al., 2014; Sperling and Winthrop, 2016; Botea et al., been conducted, three fourths of all instances of early 2017, Evans and Yuan, 2019, and Wodon, 2020). childbearing come after (and therefore appear to be due to) child marriage, rather than the other way around IMPACTS OF GENDER INEQUALITY BY DOMAIN (Wodon et al., forthcoming). The prevalence of child marriage has declined substantially in India and South ifetime earnings. The analysis of the impact of gender • L Asia over the last two decades, although it remains inequality on earnings is based on measures of human high at 27.0 percent in 2017 (Le Nestour et al., 2019). capital wealth, which is the value today of the future In sub-Saharan Africa (prevalence at 35.1 percent) earnings of all individuals – men and women – active and in Latin America and the Caribbean (prevalence in the labor force. Globally, for every dollar in earnings at 25.8 percent), much less progress has been 5 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 achieved over time towards reducing child marriage. the same 19 countries, gender equality could help reduce under-five mortality rates by 0.32 percentage • Fertility and population growth. While reducing point, a reduction of slightly more than five percent fertility rates is not be an objective in itself, high rates from base rates. For under-five stunting, the reduction of population growth in low income countries is driven is estimated at 2.1 percentage points on average, or by high fertility come with various consequences – not seven percent from base rates. In other words, while least a stalled demographic dividend and high burdens gender inequality affects under-five mortality and on governments to maintain (let alone increase) public stunting, it is probably not one of their main drivers. investments in children as well as adults. Gender The study also documents impacts of gender inequality inequality has a large impact on fertility and population on other aspects of women’s lives including the risk growth. When girls marry or have children early, they of intimate partner and other forms of violence, tend to have more children over their lifetime. Lack knowledge of HV/AIDS, and whether children are of access to modern contraceptive use, leading to registered at birth. Finally, the study considers the unmet demand for family planning, also contributes issue of violence in and around school, and how various to high fertility rates. Achieving gender equality – for forms of violence may affect boys and girls differently. example by ending child marriage and raising educational attainment for girls in countries where they lag behind • Decision-making and social capital. Gender inequality boys - would change some of the factors that lead to is generally associated with lower levels of decision- high fertility rates. Estimates from regression analysis making for women. A woman’s agency or capacity to for 19 developing countries suggest that achieving exercise choice depends on the enabling environment gender equality could reduce total fertility on average – including policies, regulations, social norms, as well as by 0.70 children per women in those countries. This on access to resources and past achievements. Gender would represent a reduction in total fertility of 13.1 inequality has an impact on resources, for example by percent versus current levels for those countries. contributing to girls’ premature school drop-out and Analysis also suggests an impact of gender inequality lower future earnings. It also affects past achievements on the likelihood of using modern contraception, (as well as capabilities), as is the case when women do although the estimated impacts are smaller: an increase not have access to the same employment or earnings of three percentage points in modern contraceptive as men. Finally, it affects agency by reducing decision- use or 12 percent from the (low) base in the countries. making in the household. Across the same set of 19 Through its potential impact on total fertility, achieving countries as before, achieving gender equality could gender equality would lead to a reduction in annual increase women’s decision-making by 24 points on a rates of population growth. Estimates for a set of scale from zero to 100 (as measured through simulations developing countries suggest an average reduction with an index accounting for individual and joint of the annual rate of population growth of 0.26 decision-making in various areas). This represents an percentage points with gender equality. This estimate increase of almost half from base values of the index. is valid only for those countries and would not extend As another example of impact, the study estimates to developed countries were no large reduction in that achieving gender equality could lead to a small population growth would come from gender equality. increase in birth registrations for children. Finally, the study notes that lack of educational attainment • Health, nutrition, well-being, and violence. By for women is associated with a lower likelihood of weakening conditions for early childhood development, being able to engage in altruistic behaviors, such as gender inequality may have negative impacts on young volunteering, donating to charity and helping strangers. children with lasting negative consequences. The study measures the impact of gender inequality for mothers on the risks of under-five stunting and mortality. The issue is not whether there are differences between boys and girls in those risks. Rather, the focus is on ECONOMIC COSTS OF GENDER INEQUALITY whether gender inequality as it impacts mothers in turns leads to higher risks for both boys and girls. For Estimates of the potential economic costs of gender FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 6 inequality are based on measures of national wealth, inequality are slightly larger in low income countries. which is the assets base that enables countries to produce income (Gross Domestic Product or GDP). A country’s • Lost human capital wealth due to stunting for young wealth includes produced capital (assets such as factories, children (in selected developing countries). Stunting equipment, or infrastructure), natural capital (assets such in early childhood leads to losses in earnings in adult as agricultural land and other renewable and non-renewable life. Estimates from impact evaluations suggest that natural resources), and human capital (present value of the stunted children may loose up to one fourth of their future earnings of the labor force). Human capital accounts expected earnings in adulthood due to stunting in for two thirds of global wealth. If gender equality in earnings early childhood. As gender inequality (experienced were achieved, countries could increase their human capital by mothers) contributes to high stunting rates (for wealth, and thereby their national wealth substantially. By their children) in developing countries, it reduces reducing population growth, countries would also increase expected earnings and thereby human capital wealth their level of national wealth per capita. This would enable for the adult workforce. The economic cost of gender them to strengthen the sustainability of their development inequality due to its impact on stunting for young path. Specifically, key findings on the economic cost of children is estimated at US$71 billion in 2014 for a set gender inequality are as follows: of 17 developing countries with a population of more than two billion people. This is much smaller than the • Lost human capital wealth due to inequality in earnings lost human capital wealth from gender inequality in (across all countries). If women were earning as much earnings, but still substantial for the countries affected, as men, women’s human capital wealth could increase and especially the people affected by losses in earnings by more than half globally (Table ES.1; see Box ES.2 on in adulthood due to stunting during their childhood. the limits of the analysis). Gains would differ between regions and countries, but globally for the 141 countries ost welfare from high population growth (in selected • L included in the analysis, the total gain in human capital developing countries). Women should have agency wealth from gender equality is estimated at US$ 172.3 in terms of the number of children that they have trillion in 2017 or US$24,586 per person. This estimate, over their lifetime. Through child marriage and early which is in 2014 price levels to be comparable to childbearing as well as lower educational attainment estimates for 2014 in Wodon and de la Briere (2018), for girls as compared to boys in many low-income represents about twice the value of GDP globally. countries, gender inequality for girls is associated with Human capital wealth could increase by about one higher fertility and population growth. This reduces fifth globally under gender equality in earnings, leading levels of overall wealth per person in those countries. to substantial gains in global wealth (including natural The gains in wealth per capita that could result from and produced capital). Losses in human capital wealth lower population growth by achieving gender equality due to gender inequality are higher in absolute value and reducing fertility are cumulative over time. If in richer countries because levels of human capital gender equality could be achieved, first year benefits wealth are also higher in those countries. But as a from lower population growth are valued at US$80 proportion of human capital wealth, losses due to gender billion for 16 developing countries with a combined Table ES.1: Human Capital Wealth by Gender and Potential Loss Due to Gender Inequality (US$ of 2014) 1995 2000 2005 2010 2014 2017(*) Human capital wealth per capita, men 56,290 60,940 60,980 62,672 66,832 68,717 Human capital wealth per capita, women 32,584 35,538 36,727 39,498 41,823 42,852 Ratio of women versus men’s human capital 58% 58% 60% 63% 63% 62% Loss as share of baseline human capital 25.9% 25.5% 24.0% 21.6% 21.7% 17.9% Loss as share of baseline total wealth 17.9% 17.8% 16.0% 13.9% 14.0% NA Loss in human capital wealth per capita 23,030 24,603 23,391 22,068 23,620 24,586 Source: Authors. See also Wodon (2018) and Wodon and de la Brière (2018) for estimates up to 2014. Note: (*) Estimates for 2017 are based on projections taking into account GDP growth between 2014 and 2017. The share of human capital in total wealth in 2017 is not provided because projections are not available for total wealth. 7 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 BOX ES.2: THE CHANGING NATURE OF PAID EMPLOYMENT AND OTHER SHIFTS The estimates of the global cost of gender inequality in earnings are based on current conditions, since they rely on estimations of expected future earnings of today’s labor force, with expected earnings measured based on existing household surveys and therefore current conditions. As such, the estimates do not account for potential future shifts, whether those are related to demographic change, the changing nature of work, technological advances, or the potential impact of fragility and conflict among others. For example, men and women are expected to be affected in similar proportions by automation (McKinsey Global Institute, 2019). However, if men are over-represented in emerging technology-related fields with high levels of pay and future opportunities, it could be that gender inequality in earnings may worsen in the future in some countries due to technological change (on how to promote digital jobs for women, see Solutions for Youth Employment, 2018). As another example, although the issue of the potential impact of gender inequality on population growth is considered in this study, the role that population growth will play for human capital wealth through the size of the labor force is not fully taken into account since estimates only consider individuals older than 15. Considering the potential impact of future shifts in the labor market on estimates of gender inequality in earnings is beyond the scope of this study, but those issues could be considered in follow up work. population of 2.3 billion people. Additional benefits more resources could be available to invest in higher would accrue in subsequent years. This is a mechanical quality services. The savings that could be reinvested relationship whereby lower population growth results in higher quality services can be estimated based on in higher GDP per capita, and it does not capture the reduction in the population to be served when additional economic gains from lower fertility. These population growth is itself reduced by achieving gender benefits would rise over time as standards of living equality. For savings related to the provision of public in the countries improve and population grows, education, benefits start to be reaped six years after ultimately representing a substantial share of total gender equality is achieved since this is the time needed gains from gender equality in these high population for fewer children to enter primary school. Savings growth countries. However, while very salient to these are estimated as the reduction in the anticipated countries, from a global perspective, the costs related to cost of reaching universal secondary education by high population growth are substantially smaller because 2030 in 16 countries. The benefits increase over time the countries that would benefit from reductions in and could reach up to $27 billion by 2030 in those population growth have lower levels of wealth than countries. This is an upper bound estimate of potential upper middle and high income countries where impacts savings since countries may not reach universal on population growth would likely be smaller. secondary education by 2030. But this is by no means a negligible amount that could be reinvested in udget costs from high population growth (in selected • B improving the quality of the education being provided. developing countries). By contributing to high fertility and thereby population growth, gender inequality • Difference between developed and developing may contribute to lower quality services provided countries. Thereis a major difference between developed by governments to their population. This is because and developing countries in the economic costs higher population growth may require spreading budget generated by gender inequality. In developed countries, resources more thinly to provide basic education to ever costs related to gaps in labor market earnings tend to larger cohorts of students. While reducing population dominate, given that other costs (as measured in this growth is not an end in itself, if population growth study) tend to be small since the countries have lower were lower by addressing unmet contraception needs levels of population growth and under-five stunting, and empowering women in high fertility contexts, among others. By contrast, in developing countries, FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 8 while costs associated with gender inequality in earnings are also high, costs related to population growth and under-five stunting are far from being negligible, and SELECTED POLICY in some cases may exceed costs related to gender gaps in earnings, at least over a sufficiently long period of OPTIONS TO ACHIEVE time. The implication is that in developed counties interventions to reduce the costs of gender equality GENDER EQUALITY may focus in large part on labor market earnings and other factors affecting income levels. However, in Since gender inequality affects girls and women in virtually developed countries, and especially in low income all aspects of their life, a wide range of interventions to countries, high rates of population growth as well as reduce gender inequality and mitigate its impacts should be poor education, health, and nutrition outcomes must implemented. But to keep the discussion of policy options be tackled as well. This is why in low income countries, manageable, the focus is on three types of investments investments in adolescent girls are so important, along the life cycle: (1) Investments in early childhood including to improve educational attainment, reduce development to reduce the impact of gender inequality on child marriage, and prevent early childbearing. young children, including through the provision of care for very young children; (2) Investments in adolescent girls to SUMMARY OF KEY FINDINGS delay marriage and childbearing while improving education opportunities and reducing fertility; and (3) Investments Table ES.2 provides the main estimated potential impacts in adult women to improve employment and earnings of gender inequality by domain, together with an indication opportunities. This categorization in three buckets is for of country coverage for the estimations. This is done expository purposes. In practice the various types of polices by distinguishing estimates based on global data from matter for all the impacts and economic costs identified in those based on a core set of up to two dozen developing the study. The focus on a subset of the investments needed countries. Potential impacts are summarized by showing to achieve gender equality does not mean that other types of gains from achieving gender equality in comparison to investments are not important or needed. But because these current conditions. It should again be emphasized that what three types of investments are related directly to identified is measured when using regression analysis is associations, impacts and economic costs of gender inequality in this not necessarily causal impacts. In addition, simulations study, these are the investments considered in more detail. of the benefits of achieving gender equality are based on In addition, the study suggests to target high prevalence simple comparative statics – they do not take into account areas for gender inequality or some of its manifestations potential broader effects in the economy arising from, for through interventions and prepare diagnostics and strategies example, an expansion in opportunities for girls and women. to reduce gender inequality. Assuming no changes for men in various areas such as labor force participation and earnings may lead to under- or • Investing in young children. Gender inequality over-estimating some of the benefits from gender equality. manifests itself from early childhood and even before, as Finally, for some indicators, especially in the case of agency in the case of “missing girls” due to parental preferences and decision-making, and social capital and institutions, the for boys. After birth, girls may be at a disadvantage as data pertain to reported behaviors and perceptions, thereby they may not benefit from the same investments as making interpretation more tentative. As a results, estimates boys. In addition, and this is where this study provides of potential impacts and for some of the impacts associated measures of impact, gender inequality as experienced economic costs are not meant to be precise since they by mothers may affect both boys and girls, as is the depend on models and assumptions. But they nonetheless case when early childbearing and low educational demonstrate that the potential economic impacts and costs attainment for mothers lead to higher risk of under- of gender inequality are high not only for girls and women, five mortality and stunting. In order to prevent such but also for their communities and for societies overall. negative impacts, investing in young children is one of the best investments that countries can make. 9 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Table ES.2: Selected Potential Impacts and Costs/Benefits from Gender Equality Domain Coverage Potential Impacts Global Increase in women’s human capital wealth of more than half Gain in women’s labor force participation and full-time work of 20 Earnings and standards of living Global percentage points Global Substantial reduction in poverty from higher earnings and lower fertility Global Elimination of child marriage Educational attainment, child marriage and DCs Reduction in early childbearing by at least three fourths early childbearing Global Gains in educational attainment for girls in low-income countries DCs Reduction in total fertility by 13 percent DCs Increase in contraceptive use by 12 percent Fertility and population growth Reduction in population growth rate by 0.26 percentage point in 16 Global countries Global Improvement in women’s health and psychological well-being DCs Reduction in under-five mortality rate by 5 percent Health, nutrition, well-being, and violence DCs Reduction in under-five stunting rate by 7 percent DCs Increase in women’s knowledge of HIV/AIDS and reduction in violence DCs Increase in women’s decision-making by 45 percent Global Improvement in women’s ability to assess quality of basic services Agency, decision-making, and social capital DCs Increase in likelihood of birth registration by 5 percent Global Increase in women’ reported ability to engage in altruistic behaviors Global Increase in women’s reported ability to rely on friends when in need Global Loss in HC wealth from earnings inequality of US$172 trillion DCs Loss in HC wealth from stunting of US$71 billion in 17 countries Potential economic costs Loss in wealth per capita equivalent to US$80 billion in first year in 16 DCs countries due to high population growth (with cumulative effects over time) DCs Budget costs in education of up to US$27 billion by 2030 in 16 countries Source: Wodon (2018). Note: DCs = Developing countries. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 10 A child’s earliest years present a unique window of (and boys) live or provide modes of transportation opportunity to address inequality (including gender and in some cases boarding to enable them to attend inequality), break the cycle of poverty, and improve schools, especially at the secondary level. Providing a wide range of outcomes later in life. Denboba et al. adequate water, sanitation and hygiene facilities for (2014) suggest a list of 25 interventions considered girls is essential for school infrastructure. Addressing as essential for young children. These interventions prevention and responses to the risk of violence and can be delivered through five integrated packages at sexual harassment either at or en route to school different stages in a child’s life: (i) the family support is also critical for school systems and communities package, which should be provided throughout the to address. It is also essential to ensure that schools ECD period and which includes interventions to provide improve learning outcomes and provide girls (and care options for young children that enable women to boys) with appropriate skills. Among various entry work, (ii) the pregnancy package, (iii) the birth package points that can be used to that end, the following from birth to six months, (iv) the child health and can be mentioned (1) reducing disadvantages that development package, and (v) the preschool package. girls face in remote communities, often due in part Of special importance for the focus of this study are to poor targeting of Government resources; (2) interventions related to planning for family size and creating a more inclusive school culture for girls spacing, given that a woman’s ability to space and limit – including protocols for gender-based violence her pregnancies has a direct impact on her health and prevention and response; (3) providing girls with role well-being as well as on the outcome of each pregnancy, models—including through female teachers; and in addition to the impacts and costs of gender inequality (4) raising the returns to secondary education for through higher fertility as estimated in this study. women through better employment opportunities. This list is not exhaustive and entry points vary nvesting in adolescent girls to delay marriage and • I between countries. In addition, many interventions childbearing while also improving their education that can benefit girls need not be specific to girls. opportunities. Three-pronged strategies are likely to For example, cash transfers for access to schooling be needed to invest in adolescent girls and provide or pedagogical interventions to improve learning may them with better opportunities in life: (1) General benefit boys as well as girls, and in some case may basic conditions must be met for access to education lave larger impacts than girls-specific interventions. and learning; (2) Targeted interventions must be implemented to reach vulnerable girls; and (3) » Implementing interventions to reach especially Efforts must be undertaken to change gender-based vulnerable girls: The focus here is on delaying social norms and gender-based laws, which affect all marriage and early childbearing, while also women but especially adolescent girls (for a review improving knowledge of sexual and reproductive of constraints and promising interventions with a rights and health. The study outlines three types of focus on Africa, see also Chakravarty et al., 2017). interventions that were recently reviewed on the basis of the available evidence from experimental » Ensuring general conditions for access to education and or quasi-experimental studies: (1) There is a need learning. Several reviews discuss the basic general for interventions to expand economic opportunities conditions required for improving girls’ education for adolescent girls who dropped out of school and (Unterhalter et al., 2014; Sperling and Winthrop, who are unlikely to be able to return; (2) Imparting 2016; Evans and Yuan, 2019; Wodon, 2020). Some adolescent girls with life skills and reproductive of the interventions required to ensure these basic health knowledge is also needed, whether girls are conditions are met are likely to be as effective to in school or out of school. Evidence suggests that improve education for girls as interventions targeted safe space clubs where girls may discuss issues of to girls specifically (Evans and Yuan, 2019). In many sexual and reproductive health as well as other countries with a high prevalence of child marriage topics with female mentors may be an effective and low educational attainment for girls, there is a means of achieving this; (3) However, according need to build secondary schools closer to where girls to the literature, the most effective targeted 11 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 interventions to delay marriage and childbearing » R educing, redistributing and recognizing (three Rs) are those that enable girls to remain in school, unpaid work and care: Elson (2017) suggested a especially through incentives offsetting the out- “three Rs” approach to close the gender gap. Time of-pocket and opportunity costs of schooling (see use surveys show that women spend substantially Botea et al., 2017, for a more detailed review, as more time in unpaid home-based work than men, well as Bandiera et al., forthcoming, and Baird and consequently less time in market work. Reducing et al, 2011, as examples of interventions). unpaid work for women would free time for market work or other activities. Various types of policies » Changing gender-based social norms and gender- can help in that regard. This includes providing biased laws: Child marriage, early childbearing, better access to basic infrastructure services (water, low educational attainment for girls, and other electricity, energy) as well as child and elderly care forms of gender inequality are rooted in social services; enhancing women’s mobility through better norms that perpetuate gender discrimination. To and safer modes of transportation and ICT; and tackle this challenge, beyond general conditions expanding programs such as parental leave, flexible that education systems should meet and targeted schedules, and appropriate legislation on retirement interventions to reach vulnerable girls, additional ages while minimizing potential downsides for women community-based interventions that involve all in terms of slower career progression or occupational members of the community may be an effective segregation. Among those interventions, quality means of changing these norms (see for example the care services are especially important as a shift review by Jayachandran, 2019). Such interventions in the proportions of market work, non-market should target men and community leaders and work, and leisure time for women requires a not only women. Finally, adequate laws – for double redistribution of care work, not only within example on the minimum age for marriage without households to adult male members but also between exceptions for parental and judicial consent, but also households and public and private service providers. in many other areas related to work, inheritance, and many other aspects of women’s lives are also » F acilitating access to productive assets: Especially in essential as noted among others in successive low income countries, women’s employment tends Women, Business and the Law reports, but often to be informal and concentrated in agriculture (and not sufficient on their own to achieve change. to some extent services). Women farmers often Processes for registering marriages and births may generate less income than men due to unequal be instrumental in ensuring that legislation related access to inputs and lower returns to these inputs. to the minimum age for marriage are respected. Improving ownership of, secure access to, and control over good quality land requires strengthening nvesting in women to improve employment and • I women’s land rights in legislation and property earnings opportunities. In middle and high income registries (for a review with a focus on Africa, see countries, gender parity has been achieved in O’Sullivan, 2017). Also important is the acquisition educational attainment, with girls even edging boys in of soft technical and managerial skills and access to many countries. Yet this does not mean that women finance among others through micro-credit and the do as well as men in labor markets, as documented by promotion of alternative collateral. Bundled services human capital wealth gaps between men and women. A including (in-kind) capital transfer, asset-specific review of the literature prepared for this study suggests training, technical assistance, stipends for one to that interventions can be implemented in three main two years, and health information and insurance areas: (1) reducing time spent by women in unpaid work as well as life skills training can help push very poor and redistributing care responsibilities within households women out of poverty traps with positive economic and between households and public and private service outcomes and increased savings. High-quality providers; (2) increasing women’s ownership and control business management training of significant duration over productive assets, especially finance; and (3) can benefit female entrepreneurs, as can demand- addressing a variety of market and institutional failures. driven job services tackling barriers to employment. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 12 » Solving market and institutional failures: Both types CONCLUSION of failures can be pervasive with serious implications for gender inequality. Access to information to Gender inequality has negative impacts for girls and women address occupational segregation and pay gaps throughout their lives and these impacts result in large can help improve gender equality. Access to social costs to economies. These impacts were documented in capital (networks, role models, and mentorship) this study in five main areas: (1) earnings and standards also matters. Self-help groups foster increased of living; (2) educational attainment, child marriage and solidarity between peers, independent financial early childbearing; (3) fertility and population growth; decision-making, and greater respect for the (4) health, nutrition, well-being, and violence; and (5) women within their households and communities. agency, decision-making, and social capital. The potential Group approaches may be especially effective, economic costs of gender inequality in terms of lost wealth for instance in agriculture as illustrated by for countries are substantial. But solutions are available to production cooperatives, but also in finance and achieve gender equality. Along a simple life cycle model, entrepreneurship. Another area for reform is the study considered three main types of interventions: legal and fiscal frameworks including labor market (1) Investments in early childhood development to reduce policies ensuring equal opportunities, laws about the impact of gender inequality on young children; (2) access to capital and justice, and policies targeted Investments in adolescent girls to delay marriage and at advancing women to top positions. In OECD childbearing while also improving education opportunities; countries that have more developed tax systems, and (3) Investments in adult women to improve employment policies should avoid penalties for women as “second and earnings opportunities and increase human capital earner”, while earned income tax credits can wealth. In addition, the study also suggested to target provide an income subsidy for low-earner families high prevalence areas for gender inequality or some of its and encourage women to enter the labor force. manifestations through interventions and prepare strategies Finally, ensuring safety and preventing gender- to reduce gender inequality. This is not only the right thing to based violence at home, at work, and in public do, it also makes sense from an economic point of view. spaces is also essential (this requires interventions beyond solving market and institutional failures). • Targeting: Finally, the study makes the case for targeting geographic areas when implementing interventions and preparing cross-sectoral strategies to reduce gender inequality. Ideally, interventions should have universal coverage, but in practice, resources are often limited. Given the importance of achieving change at the community level, including in terms of social norms, targeting interventions in priority areas can help to create tipping points. In addition, although this is a generic point, it is worth noting that preparing a country diagnostic of gender inequality and drafting an evidence-based strategy towards gender equality can help conduct the dialogue needed to achieve consensus and commitment. 13 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 INTRODUCTION The fifth Sustainable Development Goal calls for gender equality and empowering all women and girls (Box 1). Substantial progress has been achieved towards gender Countries have a long way to go to achieve that goal. equality over the last two decades. As just one example, A range of dire statistics exists on aspects of gender according to data from the World Bank’s World inequality ranging from child marriage and female genital Development Indicators, nine in ten girls (89.3 percent) mutilation to legal discrimination, intimate partner and other complete their primary education globally, and three in four forms of violence, lack of political voice, lack of access to (76.0 percent) complete their lower secondary education. contraception and health care, and lack of access to land. As For boys, the proportions are very similar at 89.9 percent noted in Wodon and de la Brière (2018) in the first note in and 75.3 percent. In other words, girls have caught up this series on the cost of gender inequality globally, inequality with boys in educational attainment globally, and in some in earnings between men and women could cost countries countries such as those in the Caribbean girls are often doing up to US$160 trillion in lost human capital wealth measured better than boys. However, in many low income countries, as the present value of the future earnings of the labor force substantial gender gaps persist in educational attainment to using data up to 2014. This note updates those estimates to the detriment of girls, especially at the secondary level where 2017 and extends the analysis of the impacts and economic the completion rate for girls at 36.9 percent is below that of costs of gender inequality beyond labor earnings to a larger boys. And in many other areas, girls and women remain at a set of domains. disadvantage versus boys and men. Often this disadvantage is largest in the poorest countries. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 14 BOX 1: DEFINING GENDER EQUALITY Gender inequality can be defined in various ways. This study follows World Bank (2012, 2016), whereby gender refers to the social, behavioral, and cultural attributes, expectations, and norms associated with being male or female. Gender equality (or inequality) then refers to how these factors determine the way in which women and men relate to each other and to the resulting differences in power between them. This definition is very broad, but guidance on how the concept of gender equality can be operationalized is available from the goals adopted by the international community in 2015 under the Agenda 2030 or Sustainable Development Goals (SDGs). The fifth goal - to achieve gender equality and empower all women and girls - includes six targets: (1) Ending all forms of discrimination against all women and girls everywhere; (2) Eliminating all forms of violence against all women and girls in the public and private spheres, including trafficking and sexual and other types of exploitation; (3) Eliminating all harmful practices, such as child, early and forced marriage and female genital mutilation; (4) Recognizing and valuing unpaid care and domestic work through the provision of public services, infrastructure and social protection policies and the promotion of shared responsibility within the household and the family as nationally appropriate; (5) Ensuring women’s full and effective participation and equal opportunities for leadership at all levels of decision- making in political, economic and public life; and (6) Ensuring universal access to sexual and reproductive health and reproductive rights as agreed in accordance with the Programme of Action of the International Conference on Population and Development and the Beijing Platform for Action and the outcome documents of their review conferences. In addition, the goal calls on member states to (i) Undertake reforms to give women equal rights to economic resources, as well as access to ownership and control over land and other forms of property, financial services, inheritance and natural resources, in accordance with national laws; (ii) Enhance the use of enabling technology, in particular information and communications technology, to promote the empowerment of women; and (iii) Adopt and strengthen sound policies and enforceable legislation for the promotion of gender equality and the empowerment of all women and girls at all levels This study does not consider all the aspects of gender inequality that have been incorporated in the fifth goal under the SDGs as well as other SDGs such as those for education and health, but it considers key dimensions that tend to have large economic costs. This includes gaps by gender in labor market earnings (in part due to gaps in educational attainment but also due to lower labor force participation and lack of full-time work for women in comparison to men), and the effects of gender inequality on population growth, health outcomes including for children of young mothers, and opportunities for adolescent girls. 15 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Gender inequality has a range of potential negative impacts on some of the interventions that could help achieve gender throughout women’s lifetime not only for themselves, but equality, and thereby prevent those losses. The guidance also for their children and households, their communities, provided on programs and policies is not meant to be and societies. The fact that investing in girls and women is exhaustive, and the study does not try to identify some of smart economics is not news. The point has been made in the deep-seated factors that contribute to gender inequality. earlier studies (see Box 2). The contribution of this study is to For example, some governments have adopted national document the potential negative effects of gender inequality strategies to end child marriage and teenage pregnancies. using new measures and with more recent survey data. The Improving girls’ education is also a priority of many hope is that by illustrating the wide-ranging potential impacts governments. Unfortunately, the cultural, economic, and and costs of gender inequality, the analysis will accelerate social conditions that have historically contributed to poor policy mobilization towards achieving equality and providing outcomes for girls remain strong in many countries due to to girls and women the opportunities they deserve. persistent social norms (see Box 2). The primary objective of this study is to estimate some of the economic costs of gender inequality. In addition, the study relies on reviews of the literature to provide guidance BOX 2: SOCIAL NORMS, VOICE, AND AGENCY A large body of work has been conducted on how gender inequality affects development outcomes and how various policies and programs could reduce the magnitude of such inequality - see for example World Bank (2001, 2012, 2016), Klasen and Lamanna (2009), Duflo, (2012), Agenor and Canuto (2013), Elborgh-Woytek et al. (2013), Cuberes and Teigner (2015), McKinsey Global Institute (2015), Kabeer (2016), International Labour Organization (2018), Ostry et al. (2018), and World Economic Forum (2018). Focusing specifically on social norms, voice and agency, Klugman et al. (2014) document constraints facing women and girls worldwide, from high levels of gender- based violence to social norms and laws that curtail their decision-making in multiple areas. The report notes that constraints faced by women and girls stem from their limited endowments (health, education, and assets) and economic opportunities. In addition, social norms about gender roles are also limiting. For example, even when women work outside of the home, they typically remain responsible for housework and child care. Social norms often restrict women’s mobility. They tend to be under-represented in politics and government. Unequal power relationships lead to gender-based violence, and legal discrimination remains pervasive, as is the case when women need their husband’s consent to work. Lack of protection and discrimination under the law interacts with social norms, as is the case when women have limited land rights. Social norms tend to be reinforced by the community where girls and women live, including by teachers and the education system, or employers and the labor market. This is often recognized, but not always acted upon in terms of finding adequate levers to enact change. One of the implications of the deep-seated nature of social norms is that even if some specific manifestations of gender inequality were to be abolished, underlying issues may still remain. For example, even if child marriage were ended, say by managing to keep girls in school until the age of 18, they might marry immediately after reaching 18 and have children soon after that, so that their voice and agency might still be curtailed. This is why beyond tackling specific forms of gender inequality, broader social norms must be addressed. Yet outlining how this can successfully be done is beyond the scope of this note. Other recent studies inventory the evidence on what works to shift gender social norms (see for example Dhar et al., 2018; Marcus, 2018; and Overseas Development Institute, 2015). FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 16 FRAMEWORK AND METHODOLOGY on these four types of costs is driven in part by data availability, but also by an attempt to avoid double counting The framework that guides the analysis is provided in Figure 1 impacts and associated costs. This is done by considering and builds on two recent studies at the World Bank. The first separately the effects of gender inequality first on incomes study focused on the economic impacts of child marriage or expenditures as the numerator of measures of long-term and was implemented jointly with the International Center well-being (human capital wealth per capita), and next on the for Research on Women (Wodon et al., 2017). The second population that benefit from incomes or expenditures as the study considered the cost of not educating girls (Wodon et denominator of measures of well-being (see Appendix 1 for al., 2018). Building on past work, five domains of potential more details). In Figure 1, the framework is presented in terms impacts of gender inequality are considered: (1) fertility of the benefits from gender equality. In this note, we will in and population growth; (2) health, nutrition, and violence; most cases use the language of costs associated with gender (3) educational attainment and child marriage; (4) labor, inequality, but the approach is the same. In order to measure earnings, and productivity; and (5) decision-making and long-term as opposed to annual costs, the valuation of costs social capital. The potential impacts of gender inequality in relies on estimates of the changing wealth of nations, which these domains are estimated using regression analysis and a has a number of advantages as discussed in Appendix 2 (see wide range of datasets (see Appendix 1 for a description of the Hamilton et al., 2018, on estimates of human capital wealth). main datasets and more details on methodology). While the five domains are related to each other in various ways as noted Finally, gender inequality has implications for poverty and in Box 3, for simplicity, key findings are presented in this study inequality. By raising standards of living through higher sequentially for each domain considered individually. earnings and lower population growth, achieving gender equality would lead to reductions in poverty (on the link As done in previous studies on the economic impacts of between fertility and poverty in Africa, see for example child marriage (Wodon et al., 2017) and the cost of not Beegle and Christiaensen, 2019). Furthermore, since girls educating girls (Wodon et al., 2018), four main types of and women from lower socio-economic backgrounds are costs are considered: (i) higher earnings; (ii) higher “welfare” often the most affected by gender inequality, achieving or standards of living due to lower population growth; (iii) equality would also contribute to reducing extreme poverty budget savings or costs; and (iv) other benefits, including and achieving the other targets under the Sustainable in terms of individual feelings and perceptions. The focus Development Goals. Figure 1: Conceptual Framework Associated Losses/Gains 5 Domains of “Impacts” Welfare Gains Development Fertility and population growth GENDER INEQUALITY Outcomes Health, nutrition and violence Earnings Ending Poverty and Achieving the Educational attainment Complex direct and Other Sustainable and child marriage indirect “impacts” Development Goals Budget Labor, Earnings & Productivity Savings Decision-making & Social Capital Other Benefits Source: Wodon and de la Brière (2018). 17 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Several dozen different indicators or outcomes of interest considered as a prima facie evidence of gender inequality are used to capture the potential impacts of gender impacts, statistical comparisons are used. In some cases inequality. Most indicators are objective measures, including however, while differences in outcomes may be small by total fertility rates, women’s earnings, rates of under-five gender, gender inequality may still affect outcomes for both mortality and stunting. A few of the indicators and rates are genders. For example, child marriage and early childbearing perceptions-based or more tentative, such as measures of have negative effects on the children of young mothers, the ability of women to engage in altruistic behaviors. While whether those children are boys or girls. In those cases, the not all indicators may be equally important for development, impact of gender inequality is measured through regression poverty reduction, and shared prosperity, conducting analysis instead of statistical comparisons between boys analysis for a large set of indicators helps to convey the and girls since those comparisons would miss much of the fact that the consequences of gender inequality are truly effects of gender inequality. In other cases, outcomes are pervasive and wide-ranging. observed only for girls and women, either because some outcomes affect women only or mostly (as in the case for One last point on methodology is worth highlighting early childbearing and fertility rates), or because available (see Appendix 1 for details). For some indicators, simple data sources collect information for women but not men statistics are used to measure the impacts of gender (as is often the case for intimate partner violence). In those inequality, while for other indicators, regression analysis is cases too, the potential impact of gender inequality is used. When indicators are available for men and women (or measured through regression analysis instead of statistical boys and girls) and differences in these indicators can be comparisons since such comparisons are not available (see BOX 3: INTERDEPENDENCE BETWEEN DOMAINS AND BENEFITS FROM QUALITATIVE DATA For simplicity, findings on the potential impacts of gender inequality are presented in this study for each domain of potential impact separately. Yet in practice, the various domains are interdependent. To illustrate how this is the case, consider a simple life cycle approach. Social norms may contribute to disadvantage for girls early on, but they emerge in full force in adolescence when in many contexts, girls may have to get married as children if they drop out of school. This contributes to early childbearing and higher total fertility over their lifetime. In turn, having many children may affect women’s ability to participate in the labor market in adulthood, and low educational attainment reduces their earnings when working. This may affect decision-making within the household, voice, and social capital throughout women’s life. Finally, early childbearing, high fertility rate, and income losses also have intergenerational effects, contributing (among others) to higher risks of child mortality and malnutrition for children of poorly educated mothers. The challenges and obstacles faced by girls and women due to gender inequality are multifaceted. They reinforce each other, leading to a diminished ability to break out of patterns of disadvantages. In some cases, interdependence between domains is explicitly modeled. This is the case when considering the potential combined impacts of low educational attainment and child marriage, both markers of gender inequality, on other outcomes. But there are limits to the extent to which the interdependence between domains can be considered without making the analysis overly complex. Qualitative data and narratives are another way to illustrate interdependence between domains. For this reason, selected quotes from qualitative data collected as part of the work program of which this study is part, as well as quotes from a few other existing studies, are provided. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 18 Appendix 1 for more details on the methodology used for DOMAIN 1: LABOR, the analysis). When estimations of the impacts of gender inequality are based on regression analysis, the results EARNINGS, AND are subject to important caveats. First, estimates from available observational data do not permit establishing causal PRODUCTIVITY relationships. Thus, when referring to potential impacts, the analysis should be taken as suggestive of what could be LIFETIME EARNINGS AND HUMAN CAPITAL achieved with gender equality. Second, estimates of potential WEALTH impacts do not account for broader effects in the economy arising for example from better education and employment The first note in this series on the cost of gender inequality opportunities for women. Such effects could happen if the focused on losses in human capital wealth due to gender economy fails to grow at a rate that can generate sufficient inequality in earnings (Wodon and de la Brière, 2018). jobs to absorb the more educated women entering the labor These estimates of potential costs were estimated at market, and/or if the expansion of employment opportunities US$160 trillion in 2014 and are updated in this note to 2017 for women were to negatively affect earnings for men. The based on a simple extrapolation that accounts for recent analysis also does not consider how changes in situations for growth in GDP per capita and population across countries. men could in turn affect women. Beyond earnings, the note covers other areas where gender inequality may have impacts and costs. Before doing so, this Thus, estimates only provide orders of magnitude of potential section first provides the rationale for relying on measures of impacts. They are not meant to be precise nor definitive countries’ wealth, and especially human capital wealth. or ultimate values that account for general equilibrium effects. This approach is by no means perfect, and different Typically, researchers looking at the economic impact of interpretations of what gender equality entails could be gender inequality on development have focused on annual advanced. But the approach is sufficient to provide tentative measures of earnings, national income, or growth in income estimations of the impacts of gender inequality, noting that (e.g. Elborgh-Woytek et al., 2013; Cuberes and Teigner, subsequent work with different assumptions could generate 2015; McKinsey Global Institute, 2015). Many of these different results. In what follows, the analysis of the potential analyses focus on losses in Gross Domestic Product (GDP) impacts of gender inequality on development outcomes is from inequality between women and men in labor markets. presented according to the five domains highlighted in Figure This focus on income is natural since GDP is the standard 1. Thereafter, estimates of economic costs are provided for measure according to which the economic performance some of these potential impacts as shown in Table 1. The of countries is measured. Yet GDP growth is a short-term last few sections of the study are devoted to a discussion of measure of performance, which may be misleading about the selected policy options to achieve gender equality. health of an economy because it does not reflect whether a country is investing in the assets base that will sustain its Table 1: Examples of Cost Estimations Category Estimation Welfare gains Higher wealth per capita due to lower population growth Earnings gains Higher lifetime earnings for women from gender equality in earnings Budget savings Lower public spending for education due to lower population growth Source: Authors 19 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 long-term growth. A country could deplete its natural capital completed by the World Bank on the Changing Wealth of base or fail to invest in its people and still be able generate Nations study (Lange et al., 2018). Building on two previous high rates of GDP growth in the short run, although reports (World Bank, 2006 and 2011), the Changing Wealth probably not in the long-run. Following up on Wodon and of Nations 2018 study covers the period 1995 to 2014. It de la Brière (2018), this note focuses on human capital (the includes not only estimates of produced capital and natural wealth approach) by measuring lifetime losses in earnings, capital, as did previous reports, but also estimates of human instead of annual flows (the GDP approach). More precisely, capital following the approach suggested by Jorgensen and human capital wealth is defined as the present value of the Fraumeni (1992a, 1992b). The analysis is based on data for future earnings of today’s labor force, considering individuals 141 countries accounting for more than 95 percent of the aged 15 and above. world’s population. Estimations of human capital are based on household survey data. It should be noted however that At least three arguments justify using a wealth (stock) there are limitations to the methodology. For example, approach as opposed to a GDP (flow) approach to measure due in part to limits in the data available in many household losses in earnings due to gender inequality. First, using a flow surveys, the estimation of human capital wealth does not approach does not reveal the full magnitude of the losses place a value on non-employment work by women such as in earnings faced by women throughout their working life. household work and child care. Estimates of losses from gender inequality in labor markets based on human capital wealth are substantially larger GENDER GAPS IN HUMAN CAPITAL WEALTH than those based on GDP simply because wealth is larger than GDP. The full magnitude of the losses from gender How large are the potential losses in wealth due to gender inequality appears only when considering human capital inequality in earnings? Global estimates in absolute values wealth or women’s earnings over their lifetime. Second, a will be provided in a subsequent section with other estimates flow approach tends to emphasize losses for individuals at of costs from gender inequality. In this section, estimates the peak of their earnings, since they account for a larger are provided in per capita terms together with a discussion share of the labor earnings in GDP. Again, it seems more of some of the factors leading to losses. Table 2 provides the appropriate to look at individuals’ lifetime earnings to estimates of human capital wealth per capita of men and of better reflect expected losses from gender inequality. This women. The estimates represent the present value today of should give a higher weight to younger individuals than is the expected future earnings of the adult labor force (age 15 the case with the flow approach. Third, and perhaps most and above) divided by the total population in the country. fundamentally, a wealth approach is forward-looking as it emphasizes sustainability since GDP, or more precisely the Consider first estimates for 2014 as provided in Table 2, consumption component of GDP, is essentially is the annual based on Wodon and de la Brière (2018). Globally, human return or income that a country reaps from its wealth, the capital wealth per capita was estimated at US$108,655 assets base that it uses for production. in 2014. Of this amount, US$66,832 was attributed to men’s future earnings, while US$41,823 was attributed Wealth consists of natural capital such as agricultural land, to women’s future earnings. For every dollar in earnings forest, oil, gas and minerals, to give a few examples. It also expected to be contributed by men in the future, women consists of produced capital – think about infrastructure, are expected to contribute only 63 cents. This suggests machinery, factories, or buildings. Finally, wealth consists that women globally will earn over their remaining time in of human capital, such as a well-educated and productive the labor force slightly less than two thirds of what men labor force. These three categories – produced, natural, and can be expected to earn. This represents a high level of human capital, are the main components of the changing gender inequality in expected future earnings. There has wealth of nations, that together with net foreign assets, been some progress over the last two decades towards provide the assets base to produce GDP capita from year lower gender inequality in lifetime earnings as are measured to year. through human capital wealth, but progress has been slow. In 1995, women were expected to contribute 58 cents in Given the advantages of wealth accounting over annual human capital wealth on the dollar in comparison to men. earnings measures to measure losses in earnings due to Twenty years later the proportion was 63 cents. At current gender inequality, we rely in this note on research recently rates of progress (five cents in 20 years), it could take FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 20 almost 150 years to reach parity in contributions to human strong assumption, but it has the merit of simplicity and capital wealth for both men and women. transparency. On a per capita basis, gender inequality could under this assumption lead to a loss in human capital wealth The World Bank is in the process of updating its estimates of US$23,620 per person in 2014 (losses in human capital of the changing wealth of nations, including human capital wealth per capita differ slightly from the gap in human capital wealth, but these estimates are not yet available. For this wealth by gender due to differences in the number of men study, estimates of human capital wealth by gender for 2017 and women). This loss is projected to have increased to are based on projections. The projections rely on gender US$24,586 per person in 2017 due to growth in GDP per ratios observed for 2014 but estimates of total human capital capita and labor earnings in real terms between 2014 and wealth adjusted to take into account real GDP growth as 2017. well as population growth between 2014 and 2017. These projections are tentative, but they are likely to be reasonably As a share of baseline wealth, losses from gender inequality accurate given that levels of GDP per capita across countries in earnings represent 21.7 percent of the baseline human explain close to 95 percent of the variation in the estimates capital wealth in 2014, and 14.0 percent of the baseline total of human capital wealth per capita across countries, as wealth per capita when produced capital, natural capital, mentioned in Appendix 2. Gains in real GDP per capita and net foreign assets are also included in the analysis. The between 2014 and 2017 can thus be used as proxies for gains projected estimate for 2017 is very similar for the share of in human capital wealth per capita, and levels of aggregate baseline human capital wealth (no estimates are provided human capital wealth can be obtained by simply multiplying for the share of total wealth lost in 2017 since estimates projected human capital wealth per capita by a country’s of the changing wealth of nations for that year are not yet population, taking into account population growth between available). These losses as a share of baseline wealth tend to 2014 and 2017. As noted in Box 4, one of the limitations of be slightly lower in 2014 (and 2017) than in 1995. This is in the analysis is that it is based on current conditions, and not part because as just mentioned, there is a slow movement on conditions that may prevail in the future given potential towards more gender equality in earnings in many countries changes in the nature of work, among others (World Bank, over time, which makes the losses due to gender inequality 2019a). slightly smaller. But in addition, human capital in high income countries has been declining slightly in recent years due To compute the potential losses in human capital wealth among others to ageing and a reduction in the share of due to gender inequality, we simply estimate how much labor income in GDP following the great recession. This also more human capital wealth countries would have if women contributed to a small reduction in the losses from gender were earning as much as men. As noted in Box 5, this is a inequality over time as a share of baseline wealth estimates. BOX 4: THE CHANGING NATURE OF WORK AND OTHER SHIFTS The estimates of the global cost of gender inequality in earnings are based on current conditions, since they rely on estimations of expected future earnings of today’s labor force, with expected earnings measured based on existing household surveys and therefore current conditions. As such, the estimates do not account for potential future shifts, whether those are related to demographic change, the changing nature of work, technological advances, or the potential impact of fragility and conflict among others. For example, men and women are expected to be affected in similar proportions by automation (McKinsey Global Institute, 2019). However, if men are over-represented in emerging technology-related fields with high levels of pay and future opportunities, it could be that gender inequality in earnings may worsen in the future in some countries due to technological change (on how to promote digital jobs for women, see Solutions for Youth Employment, 2018). As another example, although the issue of the potential impact of gender inequality on population growth is considered in this study, the role that population growth will play for human capital wealth through the size of the labor force is not fully taken into account since estimates only consider individuals older than 15. Considering the potential impact of future shifts in the labor market on estimates of gender inequality in earnings is beyond the scope of this study, but those issues could be considered in follow up work. 21 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 BOX 5: LIMITATIONS OF THE METHOD USED TO COMPUTE LOSSES IN HUMAN CAPITAL WEALTH The estimation of the losses in human capital wealth due to gender inequality simply assumes that women could work and earn as much as men. While this approach has the merit of being simple and transparent, it does not consider potential effects on men of rising earnings and hours worked for women. We do not account for the fact that men’s earnings may decrease if women become better educated and have access to the same employment opportunities as men (for example, resulting from reductions in occupational segregation). We also assume that women can allocate more time to labor market work without a negative impact on men’s working hours, therefore not considering the possibility of men having to allocate more time to unpaid domestic work or care. Women tend to do most of the domestic work, especially in developing countries. As women work more hours in paid employment, they may have less time for unpaid domestic work, which could affect the number of hours that men may be able to spend in paid employment, depending on options for elderly, child, or other care services available to households. Many other effects could be at work as women catch up with men in earnings. Here, for simplicity, we only compute how much more human capital countries would gain if women had the same lifetime earnings profile as men without any decrease in men’s earnings. In that sense, estimates could be considered an upper bound of the losses from gender inequality in earnings since we do not factor in the potential general equilibrium impact of higher work and earnings for women on men or the labor market more generally. There is evidence that over time, labor market premiums associated with higher levels of educational attainment may be reduced once more workers have those higher levels of education. Angrist (1995) showed that the expansion of access to education in the Palestinian territories led to a reduction in the skills premium. Acemoglu et al. (2004) note that during World War II, higher labor force participation by women depressed wages for low skilled workers. Duflo (2004) suggests similar effects in Indonesia after a large school construction program. If women were to become better educated and if they were to participate more in the labor market, including in sectors traditionally dominated by men, this could reduce expected earnings for men. These are just a few examples of studies suggesting that general equilibrium effects may be large as noted by Acemoglu (2010) (for a recent study on engineers, see Qvist et al., 2016). However, the estimation could also be a lower bound of losses. Indeed, higher earnings for women could lead to more economic activity overall, with positive multiplier effects on the economy and thereby on wages. Furthermore, if systems for the provision of care to family members were expanded, a substantial share of the time now allocated to unpaid care could become paid care work. The literature also suggests that as countries develop and women join the labor market or work longer hours, this may primarily reduce leisure and time spent on domestic chores. Overall, especially through economy-wide multiplier effects, unleashing women’s earnings potential could generate in the medium to long run larger earnings and human capital gains for both men and women than suggested here. We also do not account for intergenerational labor market benefits from unleashing women’s earnings through better education, health, and employment opportunities. The effects for the children of women could be large too, and are not measured here, except in the specific case of reductions in under-five mortality and stunting rates in the section on health costs and benefits. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 22 Table 2: Human Capital Wealth by Gender and Potential Loss Due to Gender Inequality (US$ of 2014) 1995 2000 2005 2010 2014 2017(*) Human capital wealth per capita, men 56,290 60,940 60,980 62,672 66,832 68,717 Human capital wealth per capita, women 32,584 35,538 36,727 39,498 41,823 42,852 Ratio of women versus men’s human capital 58% 58% 60% 63% 63% 62% Loss as share of baseline human capital 25.9% 25.5% 24.0% 21.6% 21.7% 17.9% Loss as share of baseline total wealth 17.9% 17.8% 16.0% 13.9% 14.0% NA Loss in human capital wealth per capita 23,030 24,603 23,391 22,068 23,620 24,586 Source: Wodon (2018); see also Wodon and de la Brière (2018). Note: The loss in human capital wealth per capita is not exactly equal to the difference between human capital wealth per capita estimates for men and women due to differences in the number of men and women. Estimates for 2017 are projections. The loss as share of baseline total wealth is not provided since total wealth is not projected for 2017. Estimates of potential losses in human capital wealth per Caribbean and especially the Middle East and North Africa capita due to gender inequality are provided in Table 3 by in part because of high levels of natural capital from sub- region and income groups. The largest loss in per capita soil assets (especially oil) in that region. The potential loss terms by far is observed for North America, followed by in wealth per capita from the base associated with gender Europe and Central Asia and East Asia and the Pacific. inequality is highest in South Asia, the region with the lowest This is because many of the countries in these regions are initial share of women in human capital. Estimates for 2017 high income or upper middle income, and thereby they are typically slightly higher than for 2014 due to growth in concentrate much of the world’s human capital wealth. But real terms in GDP per capita and resulting labor earnings. the losses in other regions are substantial too, especially in comparison to current levels of human capital wealth and Losses from gender inequality also differ between countries development. For example, in South Asia, the potential ranked by income groups, defined according to the World losses from gender inequality are estimated at US$5,405 Bank classification (low income, lower middle income, upper per capita in 2014, while in sub-Saharan Africa, the estimate middle income, and high income). Among high income is US$2,914 per capita. This is the smallest estimate across countries, a differentiation can be made between OECD regions. However, as a share of initial wealth, the potential and other high-income countries, the latter group including losses from gender inequality in sub-Saharan Africa are several oil-producing countries from the Middle East. In actually larger than the loss in Latin America and the absolute terms, the largest potential losses are observed for Table 3: Potential Loss in Human Capital Wealth Per Capita from Gender Inequality by Region and Income Groups (US$ of 2014) 1995 2000 2005 2010 2014 2017(*) Regions East Asia & Pacific 18,627 18,450 18,663 20,130 23,253 25,367 Europe & Central Asia 39,892 44,511 45,045 46,261 48,884 51,377 Latin America & Caribbean 15,500 11,558 11,945 11,468 10,940 10,545 Middle East & North Africa 9,275 11,261 11,220 11,150 11,757 11,491 North America 146,791 175,923 156,600 126,052 133,299 139,162 South Asia 2,664 3,383 4,374 4,613 5,405 6,394 Sub-Saharan Africa 2,016 1,927 1,435 2,480 2,914 2,906 Income Groups Low income countries 1,335 1,406 1,415 1,675 2,052 2,202 Lower-middle income countries 3,407 3,472 3,958 4,275 4,967 5,653 Upper-middle income countries 6,032 5,764 7,872 9,800 12,067 13,669 High income non-OECD 10,637 14,047 14,378 17,021 18,672 18,627 High income OECD 108,593 121,735 112,859 102,567 108,631 113,674 Source: Wodon (2018); see also Wodon and de la Brière (2018). Note: Estimates for 2017 are projections. 23 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 high income OECD countries and upper-middle income SELECTED FACTORS LEADING TO GENDER GAPS countries (which include China). Absolute losses in human IN EARNINGS capital wealth per capita from gender inequality are much higher in high income than in low income countries simply Two broad factors lead women to have lower lifetime because the levels of wealth on which losses are applied are earnings than men. First, women have lower labor force higher in high income countries. But again, in percentage participation rates and work in paid employment for fewer terms from the base, the picture is different. In 2014, low hours than men. Second, women tend to be paid less income countries are losing 15.1 percent of their base level of when they work compared to men. This is driven in part wealth (including all types of capital) with gender inequality by lower levels of educational attainment and occupational in earnings, which is slightly larger than the increase for the segregation (Das et al. 2019). Differences between men and world, at 14.0 percent in 2014. Note also that losses from women in the returns to education and experience may also gender inequality are lower in proportional terms from the play a role. These factors may keep women in productivity base in high-income non-OECD countries, in part because traps confounded by social norms relegating them to unpaid many of these countries have oil and thereby higher levels of care and informal work. This leads to gender inequality in natural capital. labor income. While documenting in detail the role of these various factors is beyond the scope of this note, pointers on How do these results compare to previous studies? differences in both labor force participation and the returns Comparisons can be made with previous work for both the to education for women and men are worth providing. estimates of (i) gender shares in earnings which are key for the estimation of the losses from gender inequality; and Consider first data on labor force participation. Such data (ii) the aggregate magnitude of the losses associated with are available for men and women from various data sources. gender inequality. Comparisons in terms of the aggregate Two such sources are used in this section: the Gallup magnitude of losses will be done in a subsequent section of World Poll and the GLD database (see Appendix 1 on data this note when presenting aggregate estimates. As to gender sources). While the Gallup World Poll is not a primary shares, previous studies have focused on gender shares in source of data for analysis of labor markets, it is useful to GDP, while we estimate gender shares in human capital complement the GLD database and show that findings with wealth. This difference in focus could lead in differences in both databases are similar. With both data sources, we can estimates. However, given that both approaches are based on look at the potential impact of gender on an individual’s earnings data, they should generate somewhat similar gender employment status, either through statistical comparisons shares. This turns out to be the case. The gender shares of or through regression analysis controlling for other factors GDP reported by the McKinsey Global Institute (2015) are that may affect labor force participation. Table 4 provide broadly similar to ours1. The same conclusion is reached when the estimates, with visualization in Figure 2 for the Gallup comparing globally our estimates of women’s share of human World Poll. The first column of data in the Table provides the capital wealth to estimates of women’s contribution to GDP share of men in the labor force, as well as a decomposition from the World Economic Forum’s Gender Gap Report of that share into various groups. For the Gallup World Poll, (2020).Broadly, there is relatively good alignment, at least at the categories are working full-time or part-time, or being the global and regional levels, suggesting some robustness in unemployed. For the GLD database, the categories are the the estimates (analysis of factors contributing to gender gaps shares working more than 30 hours or less than 30 hours in pay is also available in International Labor Office, 2019)2. and the share of those being unemployed. The next column provides the estimates for women. 1 Our estimate of women’s share of human capital wealth is 38 percent globally in 2014, and McKinsey’s estimate of women’s contribution to GDP is 37 percent. Gender shares are broadly similar at the regional level as well. For East Asia and the Pacific, women’s share of human capital wealth is 35 percent, while McKinsey reports women’s contributions to GDP of 41 percent for China and 34 percent for the rest of the region. In Europe and Central Asia, women’s share of human capital is at 39 percent in this study, versus 38 percent for their share in GDP in Western Europe and 41 percent for Eastern and Central Europe in the McKinsey study. In Latin America and the Caribbean, our share for women is at 44 percent versus 33 percent for McKinsey. In the Middle East and North Africa, we are at 27 percent versus 18 percent for McKinsey. The shares for North America are virtually the same at 41 percent and 40 percent. In South Asia, our share is at 19 percent versus 17 percent for India and 24 percent for other countries in the McKinsey study. Finally, for sub-Saharan Africa, we have the same share for women at 39 percent. 2 As to whether one set of approaches is better than another at the country level to estimate women’s shares of GDP or human capital wealth, this is a question that needs to be investigated further. The results may vary from one country to another depending on the quality of the underlying data. But for broad aggregates as reported here, the underlying shares are fairly similar. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 24 Findings are similar with both databases. The differences and women at 25.8 percent. The corresponding value for between men and women are large, with men much the marginal impact in the regression analysis with the GLD more likely to be in the labor force than women – a well- database is 19.7 percentage points. The same holds for the established fact. Furthermore, among those in the labor likelihood of working full time. Gender effects are thus force, the share working part time or less than 30 hours is clearly at work in decisions to work and how much to work, higher for women than for men. Unemployment rates are even after controlling for other individual characteristics such similar. For this report as well as for a separate study on the as educational attainment. cost of not educating girls (Wodon et al., 2018), the Gallup Poll was also used for a series of indicators –for no other A different story emerges when looking at whether gender indicator were differences between men and women as large gaps in earnings and human capital wealth are affected as those observed for labor force status. substantially by differences in the returns to education for men and women. There is a large literature on the Clearly, differences in labor force participation as well as potential impact of educational attainment on earnings (see between working full time or part time are a leading factor Psacharopoulos and Patrinos, 2018, for a recent review). The affecting differences in human capital wealth by gender. benefits from educational attainment are typically measured When conducting regression analysis for the likelihood of through regression analysis whereby the potential effect on being in the labor force, working full-time (or more than earnings of educational attainment and potential experience 30 hours in the GLD database), working part-time (or (measured as age minus six and the number of years of less than 30 hours), or being unemployed, the impact of schooling) is estimated. In some models, the focus is the the sex variable in the regressions accounts for most of correlation between years of schooling and earnings, and the the difference between men and women (in comparison implicit gain associated with each additional year of schooling. to the impacts of differences between men and women in Other models look at the potential impact on earnings other independent variables). This can be seen in the last of different levels of schooling, such as having a primary, column of the Table that provides the impact at the margin secondary, or tertiary education. Apart from educational of being a woman on labor force status. For example, in the attainment, whether measured through years of schooling or Gallup World Poll, after controlling for other observable in levels, the models may also control for other variables that characteristics that may affect labor force participation, may affect earnings. being a woman is associated with a drop in the likelihood of being in the labor force of 21.4 percent, which is close to For this study, building on Wodon et al. (2018), Montenegro the absolute gap in labor force participation between men and Wodon (2020) estimate the potential impact of Table 4: Differences in Labor Force Participation and Type of Work by Gender, Global Estimates (%) Difference in Shares Share for Men Share for Women Regrssion Estimate (Women Minus Men) Gallup World Poll Share in the labor force 75.3% 49.5% -25.8% -0.21 Share working full-time 57.2% 31.2% -26.0% -0.22 Share working part-time 13.2% 13.4% 0.2% NS Share unemployed (looking for work) 4.9% 5.0% -0.0% 0.00 GLD Surveys Share in the labor force 77.3% 57.6% 19.7% -0.18 Share working more than 30 hours 61.1% 38.8% 22.3% -0.11 Share working less than 30 hours 9.1% 12.3% -3.2% 0.11 Share unemployed (looking for work) 7.1% 6.5% 0.6% 0.02 Source: Authors. Regression analysis based on data from the Gallup World Poll and the GLD database. The regression estimate is the coefficient for the gender dummy variable con- trolling for other factors affecting employment status. Note: Regression estimates reported for the pooled sample that includes data for more than 100 countries with each of the two data sources. NS means that an estimate is not statisti- cally significant at the 10 percent level. 25 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 educational attainment on earnings using a large database and others). These additional controls are limited due to of household and labor surveys available at the World Bank the need to keep comparability across countries. It is worth (see Appendix 1). Models with years of education as well noting that when controls for occupation are used, they may as educational attainment in levels are considered, but we partially net out employment segregation from the measure report below the results from models with levels of education of the gender gap in the returns to schooling. only because this enables us to assess the difference that the quality of the education received makes. As a proxy for Table 5 provides the main results for the models without quality, the authors distinguish for individuals with a primary additional controls apart from educational attainment and education or less whether the individuals declare being experience and its squared value. Average results across literate or not. When educational attainment is measured in countries and surveys are reported first for all surveys, and levels, all individuals with some primary education or primary next for a sample with only the latest survey available for education completed but no education at a higher level each country. Average gains from educational attainment are combined in a single category for primary education, are computed treating all countries equally. In other words, although with differentiation depending on whether they a small country has the same weight as a large one, and poor declare being literate or not. The same is done for secondary and rich countries are also treated equally. Several interesting and tertiary education, but in this case no distinction is findings emerge from the analysis. made between individuals declaring being literate or not because virtually all individuals at those levels declare being First, as expected, the gains in earnings associated with literate. In other words, while the authors do not distinguish higher educational attainment are substantial, especially whether individuals have completed or not a specific cycle of at the secondary and tertiary levels, while they are much study due to data limitations and the fact that the analysis is lower at the primary level. For example, for women with conducted for a large number of countries, they factor in an primary education (partial or completed) who are literate, implicit measure of quality using literacy as a proxy. the average expected gain in earnings versus no education and no literacy (the reference category) is 39.8 percent. By To test for differences in the returns to education by gender, contrast, for women with secondary education, the average the authors estimate models separately for men and women. gain compared to no schooling and no literacy is much larger They also estimate models with and without additional at 129.1 percent. Finally, for women with tertiary education, controls apart from educational attainment and experience. the average gain is at 386.7 percent. Clearly, women with The additional controls considered are location (urban versus higher educational attainment (and literacy in the case of rural) and sector of activity (agriculture, industry, services, primary education) earn more than those with no education FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 26 and no literacy, but the gains start to be much larger at the especially not in a large number of sub-Saharan African secondary and tertiary levels. This is an important finding countries where child marriage affects adolescent girls given that gaps in educational attainment between men and disproportionately, resulting in lower completion rates women are larger, at least in low income countries, at the for girls at the lower and especially upper secondary level secondary level. This also means that for women who do not according to data from Demographic and Health Surveys. In have secondary education, the benefits from the limited addition, in the adult population, women still lag behind men education they have at the primary level are potentially in terms of years of education due to the legacy of the past. much smaller. Note that if the level of earnings for women is lower than for men across all education levels, this is not inconsistent with Second, the marginal gains by level of education are typically equal returns to education, or even returns for women being larger for women than men. The potential beneficial impacts higher than for men. of more years of education are thus slightly higher for women than men. This may be in part because the point Third, quality matters, as proxied by whether individuals of comparison – women with no education at all – have declare being literate or not. When individuals have a primary low earnings, so gains in percentage terms are computed education but are not literate, the gains are virtually non- from a low base. Still, this suggests that differences in existent versus having no education at all. When individuals earnings and human capital between men and women are have a primary education and are literate, the gains are larger, probably not due to any large extent to differences in the and when individuals are literate but do not have a primary returns to education, but rather to differences in labor force education, the gains are almost as large as when they have participation and the type of job held, as well as differences a primary education and are literate. In other words, going in educational attainment. While globally girls have caught to primary school without learning does not generate gains, up with boys in educational attainment at the primary and which helps to emphasize the need for learning apart from lower secondary levels, this is not the case everywhere, and schooling (as advocated in World Bank, 2018b). Figure 2: Differences in Labor Force Participation and Length of Work by Gender, Gallup World Poll (%) 80% 70% 60% 50% 40% 30% 20% 10% 0% Share in the Share working Share working Share being labor force full-time part-time unemployed Men Women Source: Authors. 27 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Table 5: Potential Impact of Educational Attainment on Earnings for Women and Men (Percentage Gains, %) Full Sample Last Year per Country Men Women Men Women No education and illiterate Ref. Ref. Ref. Ref. No education and literate 21.9 30.3 15.8 37.4 Primary education and illiterate 3.6 7.8 6.7 -2.8 Primary education and literate 35.1 39.8 23.6 33.6 Secondary education 101.9 129.1 75.8 108.4 Tertiary education 301.7 386.7 233.2 350.0 Source: Montenegro and Wodon (2020). Regression analysis based on wage earnings data from the World Bank’s GLD database. Note: Reported estimates based on the average value of regression coefficients across counties. The exponential transformation (given that the dependent variable is the logarithm of earnings) is taken for the average coefficient. BOX 6: GENDER INEQUALITY, POVERTY, AND EDUCATION Gender inequality contributes to monetary poverty, a measure of the inadequacy of income to meet basic needs, in several ways. The most important pathways for this impact are related to (1) lower labor earnings for women and their household; and (2) higher dependency ratio and household needs through high fertility. Higher educational attainment for women from gender equality in low income countries not only increases labor income but also results in lower fertility. This study does not provide measures of the potential impact of gender inequality on monetary poverty. These potential effects are likely to be large however, as witnessed in the case of educational attainment by UNESCO (2017). Using data from the Gallup World Poll, Wodon et al. (2018) estimate the potential impact of the level of women’s educational attainment on two types of perceptions: the perceptions of their own standard of living and the perceptions of whether economic conditions are improving or favorable. For example, when women have a secondary education level, they are seven percentage points more likely to state that they have enough money to buy food in comparison to women who have only a primary education or less. With tertiary education, the potential effect for the perceived ability to satisfy food needs is a gain at the margin of 12 percentage points in comparison to a primary education or less. It should be emphasized that individuals with higher levels of educational attainment may have on average higher expectations for their own standards of living. This implies that if the potential impacts of educational attainment on objective standards of living had been measured, impacts would probably have been larger. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 28 DOMAIN 2: CHILD childbearing. The focus in this sub-section is on describing MARRIAGE, EARLY trends in child marriage, while the next sub-section CHILDBEARING, documents the relationship between child marriage and early childbearing. AND EDUCATIONAL Table 6 and Figure 3 provide global and regional trends in ATTAINMENT child marriage from 1990 to 2017. Progress is in the right direction, but slow (Nguyen and Wodon, 2015; Le Nestour CHILD MARRIAGE et al., 2018). The prevalence of child marriage decreased from 24.8 percent in 2000 to 19.1 percent in 2017. The Child marriage, defined as entering into a union (whether decrease in India accounts for a substantial share of the formal or informal) before the age of 18, is a clear global decrease. Prior to 2010, South Asia had the highest manifestation of gender inequality. The practice affects prevalence of child marriage. Today, sub-Saharan Africa has mostly girls and has large negative impacts on a wide range the highest prevalence at 35.1 percent with limited progress of development outcomes (Wodon et al., 2017). It also over time. There is a clear association between country leads in many cases to early childbearing, defined as a girl income levels and the prevalence of child marriage, with having a first child before the age of 18 (for estimates, see much higher child marriage rates in poorer countries. At the Wodon et al., forthcoming; on the impact of child marriage same time, there are large differences between countries on reproductive health, see for example Kamal, 2012, and within income groups, pointing to the role of social norms Onagoruwa and Wodon, 2018a; for a broader review of and policies in influencing child marriage. the negative impacts of child marriage, see Parsons et al., 2015). In simulations of the impact of gender inequality on The number of girls marrying as children peaked globally a range of development outcomes such as fertility, under- at about 13.0 million around 2005. Due to progress in five mortality and stunting, and women’s decision-making, India and other countries, it declined to 10.9 million in achieving gender equality will be assumed to include ending 2017. Because of its population size, South Asia still has the child marriage or early childbearing, depending on the largest number of child marriages (4.1 million in 2017), but indicator being looked at. This also means that no analysis sub-Saharan Africa is not far behind with 3.4 million girls of the correlates of child marriage and early childbearing is marrying as children each year. While the number of girls required for estimating costs, since the assumption is that marrying as children has declined over the last decade in achieving gender equality would end child marriage and early South Asia, it is still increasing in sub-Saharan Africa. Table 6: Trend in the Prevalence of Child Marriage for Girls by Region and Income Group (%) 1995 2000 2005 2010 2014 2017 Regions East Asia & Pacific 9.4 8.9 8.6 7.8 6.5 7.5 Europe & Central Asia 9.4 9.3 9.0 7.3 6.4 5.6 Latin America & Caribbean 23.7 26.3 26.7 27.1 26.2 25.8 Middle East & North Africa 26.6 21.7 17.8 15.7 14.6 13.4 North America 7.2 6.2 6.6 6.2 3.7 2.2 South Asia 58.9 54.7 50.8 46.0 38.0 27.0 Sub-Saharan Africa 44.2 41.6 39.5 39.1 36.8 35.1 Total 24.9 24.7 24.8 23.6 20.5 19.1 Income Groups Low income 31.5 31.2 42.9 42.0 41.0 37.4 Lower middle income 20.2 20.6 9.8 13.2 29.5 24.2 Upper middle income 18.2 21.7 19.2 12.8 9.9 11.9 High income 4.4 3.4 3.7 3.6 2.6 2.1 Total 24.9 24.7 24.8 23.6 20.5 19.1 Source: Le Nestour et al. (2018). Note: Income group classification based on World Bank thresholds. 29 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Figure 3: Trends in the Number of Child Marriages Globally and by Region 12 Number of child marriages (in millions) 8 Sub-Saharan Africa 4 South Asia North America Middle East & North Africa Latin America & Caribbean Europe & Central Asia East Asia & Pacific 0 1990 1995 2000 2005 2010 2017 Source: Le Nestour et al. (2020). Especially in countries where the prevalence of child the policies that could help end child marriage could also be marriage is high, parents often have their daughter marry beneficial for reducing FGM (Wodon et al., 2017; see also early when they are not in school because of a concern Box 7 on the correlates of the transmission of FGM across that she may otherwise engage in sexual activity. In many generations). While this study does not provide estimates of contexts, a pregnancy outside of marriage may lead to the economic costs related to FGM, past work by Taghreed ostracism for the girl, thereby fundamentally affecting et al. (2008) suggests that the costs, while substantial, are her prospects in life. For many parents, the decision to likely to be of a lower order of magnitude than costs related marry their daughter is taken to protect her. For girls to other impacts of gender inequality documented in this themselves, when education and employment opportunities study (for another study on the United Kingdom, see Hex et are limited, staying at home may not be a good option. As al. 2016). will be discussed later, keeping girls in school is key to end child marriage, but this does not mean that other types of EARLY CHILDBEARING interventions and policies – such as setting the minimal legal age for marriage at 18, are not needed. Child marriage is a Early childbearing comes with steep impacts to development deeply rooted social norm (Gemignani and Wodon, 2015). outcomes, not least are the risk to young mothers from The practice needs to be addressed through multifaceted these pregnancies and a higher likelihood of under-five interventions. mortality and stunting for their children. For some girls, having one or more children before the age of 18 may be Before shifting to the discussion of early childbearing, the consequence of child marriage. For others, marriage it is worth noting that while child marriage is commonly may result from an early childbirth or pregnancy. For yet seen as a form of violence against girls, it is not the only others, early childbirths may not be related to child marriage such form of violence. The gender equality goal under the at all. In developed countries, the share of early childbearing SDGs also mentions female genital mutilation (FGM) as a or teenage pregnancies that are due to child marriage are harmful practice. This study does not provide estimates of likely to be low. In the United States, estimates suggest that the economic costs of female genital mutilation as a form 750,000 teenage girls become pregnant every year and four of gender inequality, but such costs could be large for the in five such pregnancies are not wanted. But in developing women who suffered from FGM (Refaei et al., 2016). Many countries, many instances of early childbearing are due to of the deep-seated social norms that lead to the perpetuation child marriage. The share of instances of early childbearing of child marriage also contribute to the perpetuation of FGM likely due to child marriage is estimated for 25 developing from one generation to the next, which implies that some of countries using DHS data (see box 8 for technical details). FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 30 BOX 7: TRANSITION OF FEMALE GENITAL MUTILATION FROM ONE GENERATION TO THE NEXT Female genital mutilation (FGM) continues to affect millions of girls each year. The practice remains common in many countries and can have severe health consequences for girls and their future children. It is also related to deep-seated patterns of gender inequality and gender-based violence. Several factors explain the transmission of the practice from one generation to the next (Onagoruwa and Wodon, 2018b). Community pressure and social norms. Where FGM is a social convention or norm, the social pressure to conform to what others in the community are doing or have been doing, as well as the need to be accepted socially and the fear of being rejected by the community, are strong motivations to perpetuate the practice. The role of perceptions related to sexuality is also prominent in the perpetuation of the practice. FGM has been described as a mechanism to “manage” women’s sexuality, among others to ensure that a girl remains pure and is not tempted to be promiscuous. Community prevalence of FGM has a large effect on individual practice. Thus, there is a key role of local structures of power and authority, including community leaders, religious leaders, circumcisers, and even some medical personnel in perpetuating or helping stop the practice. Mothers’ own past experiences. The risk of cutting for daughters increases by 18 to 40 percentage points if the mother was herself cut. If a mother was married as a child, this also increases the likelihood that her daughter will be cut. In addition to the role that mothers play in the perpetuation of the practice, the mother in law and grandmothers of girls at risk of being circumcised also play a role. Educational attainment. There is a reduction in the risk for daughters to be cut if the mother has at least a secondary education. In the case of support for the continuation of the practice, a secondary or higher level of education for the mother is associated with a reduction in support. Educational attainment for the husband or partner at the secondary level or higher is also associated with a reduction in the risk of cutting for daughters. Better educated fathers are also associated with a reduction in support on the part of mothers and women in general for the practice. This points to the important role that men can play in ending the practice. Wealth and poverty. The risk of daughters being cut and the likelihood for women to support the practice is higher among poorer households. Through impacts on expected earnings and wealth, women’s education especially at the secondary level could again help reduce FGM. Estimates in Table 7 suggest that at the level of mothers, age at first marriage, suggesting less of a direct connection across the developing countries for which estimates were between marriage and sexual activity and early childbearing obtained and without population weights (so that all 25 in some countries than others. One implication is that countries are weighted equally), three in four cases of early while in some countries preventing child marriage could childbearing (75.2 percent) are due to child marriage on drastically reduce early childbearing, in other countries other average. In addition, 84.4 percent of children born of interventions and policies may be needed to delay sexual mothers younger than 18 are attributed to child marriage. debut among girls while also providing access to modern Ending child marriage should therefore have a major impact contraception and ensuring that adolescent girls benefit on reducing early childbirths. This impact would however from programs increasing their sexual and reproductive differ between countries since especially in Latin America health knowledge. Experiences suggest that such programs and parts of sub-Saharan Africa, there is a trend towards implemented in school can have positive effects. earlier sexual activity along with an increase in the average 31 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Table 7: Share of Early Childbearing and Early Childbirths Likely Due to Child Marriage (%) Average share (25 countries) Share of early childbearing likely due to child marriage (for mothers) 75.2 Share of early childbirths likely due to child marriage (for children) 84.4 Source: Wodon et al. (forthcoming). Note: Estimates are based on country-level analysis for 25 developing countries. BOX 8: MEASURING THE SHARE OF EARLY CHILDBIRTHS DUE TO CHILD MARRIAGE To measure the share of early childbirths likely to be due to child marriage (Wodon et al., forthcoming), an upper bound for the share of early childbirths for mothers likely due to child marriage can be defined as one minus the share of mothers who had their first child before the age of 18 but did not marry before 18. A lower bound can be defined by subtracting from the upper bound the share of women who did marry before the age of 18, but had their first child less than nine months after their first marriage, which could suggest that marriage was not the cause of the early childbirth. An intermediate estimate could use a threshold of six months instead of nine months for the comparison of the timing of the first birth and first marriage. Indeed, if a girl/woman does not know for sure that she is pregnant, the pregnancy may not affect the decision to marry. In addition, in some countries, even in traditional contexts, cohabitation and sexual activity is permitted before a formal marriage as long as the marriage has been agreed to. The same approach and definitions can be used when looking at the share of early childbirths as defined at the level of children as opposed to mothers. While this approach has the benefit of being simple, it remains tentative and additional work would be useful to triangulate the results with a more detailed framework, for example based on Bongaarts’ model. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 32 EDUCATIONAL ATTAINMENT attainment persist, especially at the secondary level where the completion rate for girls at 36.9 percent is below that Substantial progress towards gender equality has been of boys. Gender inequality emerges more strongly in achieved in educational attainment over the last two adolescence. Globally, girls were at a substantial disadvantage decades. According to data from the World Bank’s World just under 25 years ago, but have caught up and have a Development Indicators, nine in ten girls (89.3 percent) minor advantage versus boys in completion rates at lower complete their primary education globally, and three in four secondary (Table 8). This is the case in all regions except (70.6 percent) complete their lower secondary education. sub-Saharan Africa, and all income groups except low For boys, the proportions are very similar at 89.9 percent income countries. In sub-Saharan Africa and low income and 75.3 percent. In other words, girls have caught up with countries, while girls have been catching up, large gaps boys in educational attainment globally. However, in low still exist. income countries, substantial gender gaps in educational Table 8: Completion Rates for Lower Secondary Education, Boys and Girls, 1995-2018 (%) 1995 2000 2005 2010 2015 2018 Boys Regions East Asia & Pacific 65.8 77.7 86.7 89.2 84.4 84.2 Europe & Central Asia 80.6 84.4 88.5 91.8 93.3 93.1 Latin America & Caribbean 61.9 67.3 72.4 75.1 76.9 78.1 Middle East & North Africa 62.5 65.2 71.3 70.4 74.9 75.0 North America 98.1 South Asia 54.5 55.7 59.5 68.0 76.9 78.1 Income Groups Low income 25.3 26.7 32.7 40.2 43.8 44.0 Lower middle income 53.5 56.0 60.9 67.7 74.0 74.3 Upper middle income 68.4 78.8 86.9 88.1 83.3 83.5 Middle income 60.4 66.6 72.9 76.1 77.6 77.8 High income 83.7 85.8 89.4 91.9 94.2 94.3 World 60.6 65.6 71.2 74.1 75.4 75.3 Girls Regions East Asia & Pacific 59.3 74.0 86.0 91.5 87.4 86.7 Europe & Central Asia 82.2 85.7 88.1 91.9 93.9 94.0 Latin America & Caribbean 66.6 72.0 77.9 80.9 82.1 83.3 Latin America & Caribbean 54.0 61.3 68.4 70.3 75.3 76.3 North America 94.3 South Asia 37.9 43.1 52.0 65.2 80.6 81.7 Sub-Saharan Africa 22.7 24.4 29.7 36.7 40.4 41.6 Income Groups Low income 16.2 17.7 22.8 31.3 36.3 36.9 Lower middle income 41.5 47.5 56.0 66.1 76.7 77.3 Upper middle income 64.9 77.1 86.8 90.4 85.4 85.2 Middle income 52.4 61.4 70.3 76.1 80.1 80.3 High income 83.9 86.2 89.3 91.6 94.1 94.2 World 53.5 60.6 68.1 72.9 76.1 76.0 Source: Compiled by the authors from the World Bank’s World Development Indicators. Note: Data are not provided for North America. 33 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 While in many regions girls have caught up with boys in girls doing on average worse than boys. Since these gaps educational attainment, for the adult population, and even are strong predictors of later academic choices (including for girls aged 15 to 24, gaps in educational attainment in terms of studying STEM vs non-STEM subjects at the remain, with implications for literacy, especially again in low tertiary level, as noted by Chachashvili et al., 2016) as well income countries. As shown in Table 9, women continue to as labor market outcomes, they have implications for gender lag behind men in literacy, with serious implications for their gap in earnings. Potential drivers of these gaps include family earnings potential in adulthood, as mentioned previously background and social norms, teacher focus and support when discussing gender gaps in earnings and the returns that may differ between genders, and institutional features to education and literacy. In addition, other important reinforcing those outcomes. While most of the existing gender gaps remain. One issue that has drawn attention in research focuses on developed countries, similar or even the literature is the persistent gap in post-primary student wider gaps may be observed in developing countries (e.g., performance (and related choices of study tracks at the Herz and Sperling, 2014). This is one of the areas where secondary level) in mathematics and selected sciences, with further research could shed light on appropriate policies. Table 9: Adult Literacy Rates by Gender, Individuals Age 15-24, 2000-2018 (%) 2000 2010 2018 Men Ages 15-24 Regions East Asia & Pacific 98.1 98.7 98.7 Europe & Central Asia 99.3 99.6 99.7 Latin America & Caribbean 94.6 97.0 98.4 Middle East & North Africa 90.4 91.9 92.0 South Asia 80.2 86.7 91.1 Sub-Saharan Africa 72.9 75.4 79.5 Income Groups Low income 67.5 74.1 78.8 Lower middle income 84.6 88.4 91.3 Upper middle income 97.8 98.3 98.3 Middle income 90.7 92.9 94.1 World 89.8 91.8 92.9 Women Ages 15-24 Regions East Asia & Pacific 86.3 92.0 93.9 Europe & Central Asia 95.4 96.8 97.9 Latin America & Caribbean 88.2 91.1 93.4 Middle East & North Africa 59.1 68.8 72.3 South Asia 44.9 55.7 63.7 Sub-Saharan Africa 47.4 51.0 58.8 Income Groups Low income 43.3 47.5 55.6 Lower middle income 57.2 65.2 70.9 Upper middle income 87.0 92.0 93.7 Middle income 73.2 79.1 82.4 World 75.4 79.9 82.8 Source: Compiled by the authors from the World Bank’s World Development Indicators. Note: estimates for North America and high income countries are not available from the database. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 34 There are strong mutual relationships between girls’ 2008; Nguyen and Wodon, 2014) and that keeping girls education, child marriage, and early childbearing. Pregnancy in secondary school is one of the best strategies to reduce often results in school drop out in low income countries, child marriage and early childbearing (Botea et al., 2017). due to implicit or explicit policies prohibiting pregnant girls Each additional year a girl completes in secondary school to attend and child marriage is almost always incompatible is estimated to reduce the likelihood of marrying as a child with staying in school (Figure 4). Whether marriage on average by 6.1 percentage points across 15 countries drives girls out of school or girls leave for other reasons and the risk of having a first child before age 18 by 5.8 and then get married is difficult to differentiate (Box 9), percentage points (Wodon et al 2018). With several years of but studies suggest that child marriage has a negative education, the reductions in risks of child marriage and early impact on educational attainment (e.g., Field and Ambrus, childbearing are even larger. BOX 9: WHY DO GIRLS DROP OUT OF SCHOOL IN SUB-SAHARAN AFRICA? THE CASE OF NIGER When parents are asked why their daughters dropped out of school, the cost of schooling, early marriages and pregnancies, a lack of learning while in school, and a lack of interest in remaining in school often come up. This also emerges from ethnographic work. Such work in Niger suggests six main obstacles to girls pursuing post-primary education (Perlman et al., 2018a, 2018b). 1. Poor learning outcomes and cost. Rural government schools are so poor in quality and resources that many children graduate from primary school without learning to read. The schools do not charge tuition, but parents complain that the cost of uniforms, transport, lunches and the opportunity costs of losing their daughters’ labor are hardly worth the poor learning outcomes they see. 2. Failure at examinations. Until recently, students could only take the primary school completion exam twice. If they failed, they were ineligible to continue in public education. When girls failed examinations, parents say that they have little choice but to begin looking for a suitable suitor whom their daughter could marry. 3. Lack of nearby secondary schools. Few rural communities have their own secondary school and there are few boarding schools serving communities. Parents must send their children to nearby towns and cover the costs of transportation and room and board. Students stay with relatives or contacts and parents are reluctant to leave their daughters without what they consider proper oversight. 4. Forced withdrawal of married adolescents. Once a girl is married, she is likely to be expelled from school. Husbands show little interest in providing financial support for their adolescent wife’s education especially if they must enroll in a private school. Conversely, the fear of not being allowed to withdraw their daughters from school for marriage is a complaint of some parents. 5. Never enrolling in school or enrolling too late. Some families never enroll girls in school, perhaps in part because parents had no educational opportunities themselves. In some cases, teachers may refuse to enroll children that are considered too old to start primary school. 6. Influence of relatives and demands on first daughters. Extended family members may influence parents on the value of educating girls, not always with positive outcomes. Schooling decisions may also depend on household composition and the activities of other children. Being the first daughter lessens a girl’s chances of going to school as they are expected to help their mother at home during the day. 35 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Figure 4: Typology of Adolescent Girls Aged 15-19 1.8% 15% 30% Married, not in school, any age Married, in school, any age Not married, in school, ages 15-16 Not married, in school, ages 17-19 Not married, not inschool, ages 15-16 21.4% Not married, not in school, ages 17-19 21.1% Source: Authors. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 36 DOMAIN 3: FERTILITY AND distinction is itself often driven by social norms). For this study, we consider the impact of gender inequality on total POPULATION GROWTH fertility defined as the number of live births that women are expected to have over their lifetime. We then extend this TOTAL FERTILITY analysis to look at implications for population growth and standards of living and levels of wealth per capita. Reducing fertility rates may not be an objective in itself, but high rates of population growth driven by high fertility The impact of gender equality is simulated by assuming: come with various consequences – not least is a stalled (1) no child marriage; (2) women having the same demographic dividend and high burdens on governments to education as men; (3) higher earnings which lift maintain (let alone increase) public investments in children households who are in the poorest quintile to the second as well as adults. While gender inequality in earnings affects quintile of well-being, and households in the second levels of human capital wealth, fertility rates affect levels of quintile to the third; (4) a reduction in the spousal age gap wealth per capita through high rates of population growth. (the difference in age between the wife and her husband/ By definition, the rate of growth in total wealth per capita partner) to less than 10 years; (5) women being involved over time is equal to the rate of growth in total wealth in most decisions made in the household; and (6) women minus the rate of growth of the population. A high rate of not accepting wife beating (an outcome assuming to growth of the population makes it more difficult to achieve result from gender equality).Under these assumptions, gains in levels of per capita wealth. Table 10 provides total the impact of gender inequality on fertility is obtained fertility rates as well as population growth rates by region by comparing predicted fertility under gender equality and income group. Especially in sub-Saharan Africa and in and predicted fertility under current conditions for 19 low income countries, fertility rates and population growth countries. For perspective, the impact of gender equality rates remain high. Population growth rates depend in part on is also compared to that of ending child marriage and fertility rates, namely the number of children that women achieving universal secondary education for girls. are expected to have on average over their lifetime (i.e. throughout their childbearing years according to age-specific The estimated impacts of gender inequality are large. On fertility rates). There is a time lag between declines in fertility average across the 19 countries, under the simulations for rates and declines in population growth. This is because apart gender equality, total fertility would be reduced by 0.70 from fertility rates, annual rates of population growth depend children per women towards the end of their reproductive also on the share of the population of childbearing age, life (Table 11). This represents a reduction in fertility of, which may remain large for some time even after fertility on average, 13.1 percent. The largest share of this effect rates have declined. Nevertheless, reducing fertility rates is comes from the impact of child marriage on total fertility. necessary to reduce population growth rates and thereby Ending child marriage by itself could reduce total fertility increase wealth per capita. by 0.5 children per women. Achieving universal secondary education for girls could lead to an even larger reduction in The factors leading to fertility are complex. The proximate total fertility. determinants of fertility include the rate of marriage in the population, the rate of contraceptive use, the rate of abortion, and the average post-partum infecundity duration (Bongaarts model). There is therefore a strong relationship between various aspects of gender inequality and women’s fertility. For example, girls who drop out of school prematurely, marry early, or have children early are likely to have more children over their lifetime. Social norms pertaining to gender roles also lead women to have more children over their lifetime, since they are often seen in traditional societies through their reproductive, as opposed to their productive role (acknowledging that this 37 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Table 10: Total Fertility Rates (Births per Woman) and Population Growth Rates (%), 1995-2016 1995 2000 2005 2010 2015 2017 Total fertility rates (number of births) Regions East Asia & Pacific 1.95 1.79 1.79 1.80 1.81 1.80 Europe & Central Asia 1.65 1.56 1.59 1.73 1.75 1.75 Latin America & Caribbean 2.90 2.62 2.36 2.19 2.09 2.06 Middle East & North Africa 3.91 3.20 2.89 2.88 2.81 2.77 North America 1.94 2.00 2.00 1.90 1.82 1.78 South Asia 3.85 3.46 3.10 2.74 2.49 2.46 Sub-Saharan Africa 6.02 5.77 5.52 5.25 4.92 4.85 Income Groups Low income 6.10 5.80 5.46 5.08 4.71 4.63 Lower middle income 3.79 3.44 3.19 2.96 2.79 2.76 Upper middle income 2.04 1.83 1.81 1.82 1.83 1.83 Middle income countries 2.86 2.60 2.48 2.40 2.34 2.33 High income 1.75 1.74 1.70 1.73 1.69 1.68 World 2.86 2.67 2.57 2.51 2.45 2.44 Population growth rates (%) Regions East Asia & Pacific 1.21 0.94 0.75 0.68 0.68 0.68 Europe & Central Asia 0.19 0.11 0.30 0.39 0.48 0.47 Latin America & Caribbean 1.69 1.45 1.28 1.21 1.08 1.05 Middle East & North Africa 2.73 1.95 2.02 2.15 1.89 1.81 North America 1.16 1.10 0.93 0.86 0.77 0.78 South Asia 2.06 1.87 1.66 1.44 1.30 1.27 Sub-Saharan Africa 2.74 2.67 2.72 2.77 2.74 2.72 Income Groups Low income 2.86 2.73 2.80 2.70 2.58 2.58 Lower middle income 1.94 1.77 1.64 1.55 1.47 1.44 Upper middle income 1.16 0.93 0.76 0.75 0.77 0.77 Middle income countries 1.54 1.35 1.21 1.17 1.14 1.13 High income 0.88 0.63 0.71 0.67 0.61 0.60 World 1.51 1.32 1.25 1.22 1.19 1.18 Source: World Bank’s World Development Indicators. Table 11: Potential Impact of Gender Equality on Women’s Total Fertility Absolute difference Percentage difference (%) Impact from universal secondary education -0.52 -9.61 Impact from achieving gender equality -1.31 -23.14 Impact of early childbearing due to child marriage -0.70 -13.12 Source: Authors, based Onagoruwa and Wodon (2018a). Regression analysis based on DHS data considering the number of children women have towards the end of their reproductive age (women aged 35-49 for sample size reasons). Note: Estimates are based on country-level analysis for 19 developing countries. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 38 Figure 5: Reduction in Total Fertility Under Various Scenarios Impact from achieving gender equality Impact from universal secondary education Impact from ending child marriage 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Source: Authors’ estimation using DHS data. MODERN CONTRACEPTIVE USE As simulated in the regression analysis, the estimated impacts of gender equality on contraceptive use are relatively Unmet demand for family planning remains a widespread small. On average across the 19 countries, under gender issue especially in low income countries. Access to modern equality, modern contraceptive use would increase by 2.6 contraception is a key aspect to ensuring sexual reproductive percentage points (Table 12). This represents an increase health and rights as well as agency for girls and women. The of 12.1 percent given the low use of such methods in those effect of gender inequality is measured using regression countries today. Note that eradication of child marriage plays analysis by assuming (1) no child marriage; (2) women having virtually no role in the impacts on average. Under universal the same education as men; (3) higher earnings which lift secondary education, the gains in contraceptive use would households who are in the poorest quintile to the second be about twice larger than observed under our composite quintile of well-being, and households in the second quintile measure of gender equality. While increasing modern to the third; and (4) women’s decision-making index is contraceptive use will bring down fertility, the pathway increased. As with fertility, we compare predicted modern from gender inequality to use is not as strong as observed contraception use under gender equality to observed use, for other outcomes. benchmarking impacts versus ending child marriage and universal secondary education. Table 12: Potential Impact of Gender Equality on Women’s Use of Modern Contraception Absolute difference Percentage difference (%) Impact from ending child marriage -0.16 0.64 Impact from universal secondary education 5.15 26.57 Impact from achieving gender equality 2.62 12.06 Source: Authors. Regression analysis based on data from Demographic and Health Surveys. Note: Estimates are based on country-level analysis for 19 developing countries. Average potential impacts reported for countries where coefficients for the variables of interest are statistically significant. Figure 6: Increase in Modern Contraceptive Use Under Various Scenarios Impact from achieving gender equality Impact from universal secondary education Impact from ending child marriage -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Source: Authors’ estimation using DHS data. 39 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 POPULATION GROWTH demographic projection models, and scaling those results up to account for the larger potential impact on total fertility Through its potential impact on total fertility, gender rates of gender inequality in comparison to the impact of inequality contributes to population growth. High child marriage. On average across 18 countries, the annual population growth driven by high fertility (and underpinned rate of growth in those countries could be reduced by 0.18 by high unmet needs and lack of female empowerment) percentage point if child marriage and early childbearing can contribute to poverty by delaying the demographic were eliminated (Table 13). In some countries, the potential transition. Building on work on the impact of child marriage effect is larger, as is the case in Niger. Given the comparative and early child-bearing on population growth, estimations potential effects on total fertility of child marriage of the potential impacts of gender inequality are based on documented earlier, a straight extrapolation for a slightly a parametrization of demographic projection simulation different set of countries suggests that the average potential models. The approach consists of reporting results obtained impact of gender inequality on population growth across for child marriage and early childbearing using these these countries could be at about 0.26 percentage point. Table 13: Simulated Potential Impact of Gender Equality on Population Growth Reduction in Annual Rate of Population Growth (Percentage Points) Estimates with demographic projection tools Ending child marriage and early childbearing -0.18 Estimates based on comparative potential impacts on fertility Gender inequality -0.26 Source: Authors. Note: Estimates based on analysis for 18 developing countries with extrapolations for more than 100 countries. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 40 DOMAIN 4: HEALTH, quintile to the second quintile of well-being, and households in the second quintile to the third; (4) the spousal age gap NUTRITION, WELL-BEING, (the difference in age between the wife and her husband/ partner) is less than 10 years; (5) women are involved in most AND VIOLENCE decisions made in the household; (6) women do not accept wife beating; and (7) women do not need to consult others CHILDREN’S HEALTH AND SURVIVAL to access healthcare for themselves. For under-five stunting, the approach is very similar. The impact of gender inequality Early childhood is critical for a child’s development, including on rates of under-five malnutrition and stunting is obtained brain development with lasting consequences in adulthood by comparing predicted rates under current conditions with (Black et al., 2017). As gender inequality affects households, the rates predicted under gender equality. Comparisons are specifically mothers (via early pregnancies, lower decision- again provided, this time with the impact of ending early making, domestic violence and mental health, among childbearing and achieving universal education for girls. others), it may generate spillover effects for children. In harsh conditions, so-called toxic stress responses on the part The estimated impacts of gender inequality are relatively of children can have damaging effects on learning, behavior, small. On average across the 19 countries, under gender and health later in life. Children born of younger mothers equality, under-five mortality rates would be reduced by have higher risks of under-five mortality and malnutrition 0.32 percentage point (Table 15). This represents a reduction than children born of older mothers. A large literature of slightly more than five percent compared to the base shows that such poor childhood conditions result in life-long rates. For under-five stunting, the reduction is estimated consequences on productivity and well-being (Harper et al., at 2.1 percentage points. This represents a reduction of 2003; in the case of the consequences of child marriage, just under seven percent versus the base rates. Note that see for example Wodon, 2016). Table 14 provides trends eliminating early childbearing results in smaller impacts, by regions and income groups in under-five mortality and especially for stunting This is because while the marginal stunting rates. As for many other indicators, rates remain impact of an early childbirth (being born of a mother much higher in sub-Saharan Africa and South Asia, and in younger than 18) on the risks of under-five mortality and low-income countries. stunting is relatively large, only a small share of children are born from mothers younger than 18. In other words, The focus is on measuring the impact of gender inequality for even large effects at the margin do not imply major shifts mothers on under-five mortality and stunting (as a measure nationally. While gender equality would make a difference, of malnutrition). The impact of gender equality on these it would not reduce under-five mortality and stunting two child health outcomes is estimated for 19 countries with dramatically in terms of the share of young children affected. seven indicators for gender equality: (1) no early childbearing; Still, in terms of the number of children affected, achieving (2) women having the same education as men; (3) higher gender equality would improve survival and nutrition for a earnings which lift households who are in the poorest sizeable number of children. 41 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Table 14: Under-five Mortality Rates (per 1,000) and Stunting Rates (%) 1995 2000 2005 2010 2015 2016 2017 Under-five Mortality Rates (per 1,000) Regions East Asia & Pacific 49.1 39.7 29.3 21.9 17.1 16.4 15.8 Europe & Central Asia 27.6 21.7 16.1 12.5 9.9 9.4 9 Latin America & Caribbean 43.1 33 25.7 24.4 18.3 18.3 17.7 Middle East & North Africa 53.1 42.6 34 27.6 24.2 23.6 23.1 North America 9.2 8.3 7.8 7.2 6.7 6.6 6.5 South Asia 111.8 93.8 77.2 62.5 49.3 46.9 44.8 Sub-Saharan Africa 172.3 153.7 126.2 100.9 81.4 78.3 75.5 Income Groups Low income 165.7 143.2 115 93.5 74.4 71.6 69.1 Lower middle income 109.1 94.8 78.8 64.3 52.4 50.4 48.5 Upper middle income 44.6 35.6 26.3 19.2 14.9 14.4 13.7 Middle income countries 84.16 73.41 60.15 48.05 38.91 37.5 36.18 High income 10.5 8.6 7.4 6.5 5.7 5.6 5.4 World 87.1 77.1 63.5 51.5 41.9 40.5 39.1 Under-five Stunting Rates (%) Regions East Asia & Pacific 30 24.6 20 16.2 13.1 12.6 12.2 Europe & Central Asia - - - - - - - Latin America & Caribbean 19.7 16.9 14.4 12.2 10.3 9.9 9.6 Middle East & North Africa 25.6 22.8 20.3 17.9 15.8 15.4 15 North America 3.3 3 2.8 2.6 2.4 2.3 2.3 South Asia 56.2 51.3 46.4 41.5 36.8 35.9 35 Sub-Saharan Africa 45.9 43.2 40.4 37.7 35.1 34.6 34.1 Income Groups Low income 50.6 47 43.4 39.9 36.5 35.9 35.2 Lower middle income 49.9 45.5 41.2 37 33 32.2 31.5 Upper middle income 24.4 18.5 13.8 10.1 7.3 6.9 6.4 Middle income countries 39.4 35.4 31.3 27.3 23.6 23 22.4 High income 3.7 3.4 3.1 2.8 2.6 2.5 2.5 World 35.6 32.6 29.3 26.1 23.2 22.7 22.2 Source: World Bank’s World Development Indicators. Note: Stunting rates are not available for Europe and Central Asia. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 42 Table 15: Potential Impact of Gender Equality on Rates of Under-five Mortality and Stunting Absolute difference Percentage difference (%) Under-five mortality Impact from ending child marriage 0.30 4.83 Impact from universal secondary education 2.23 30.51 Impact from achieving gender equality 0.32 5.36 Under-five stunting Impact from ending child marriage 0.35 0.97 Impact from universal secondary education 12.83 34.89 Impact from achieving gender equality 2.08 6.97 Source: Authors. Regression analysis based on data from Demographic and Health Surveys. Note: Estimates are based on country-level analysis for 19 developing countries. Average potential impacts reported for countries where coefficients for the variables of interest are statistically significant. Figure 7: Reduction in Under-five Mortality Under Various Scenarios Impact from achieving gender equality Impact from universal secondary education Impact from ending child marriage 0.0 0.5 1.5 2.0 2.5 Source: Authors’ estimation using DHS data. Figure 8: Reduction in Under-five Stunting Under Various Scenarios Impact from achieving gender equality Impact from universal secondary education Impact from ending child marriage 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Source: Authors’ estimation using DHS data. 43 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 WOMEN’S HEALTH AND VIOLENCE example in terms of how/where to deliver a baby. Sometimes the husband or partner may make these decisions, or it may Gender inequality may have potential negative impacts on be made by the mother in law in some cultures. The same women’s health for multiple reasons. Giving birth at a very can be said about decisions for antenatal care, which impacts early age raises the risk of maternal mortality (Nove et al., the health and well-being of the mother and the future 2014). A lack of physical maturity when giving birth may lead newborn. A simple measure of gender inequality in health to complications such as obstructed or prolonged labor as is whether women must ask permission to their husband well as fistula. Other health risks related to gender inequality partner simply to seek healthcare for themselves when sick may include malnutrition and depression, as well as negative or injured. As shown in Table 16 based on DHS data for 19 sexual and reproductive behaviors. This last risk can lead not developing countries, in more than 40 percent of cases only to exposure to sexually transmitted infections, but also the decision as to whether women may seek healthcare to lower rates of modern contraceptive use, insufficient birth for themselves is made by their husband or partner alone spacing, unwanted pregnancies, and abortions. instead of the women themselves or by women jointly with their partner (decisions made jointly may not be sub-optimal Gender inequality, for example through low educational versus decisions made by women alone, but when decisions attainment, may also lead to lack of knowledge about are made by partners alone without women being able sexually transmitted diseases such as HIV/AIDS, in part to contribute to decision-making, this can have negative through its effect on educational attainment. The literature consequences for women’s health). also suggests that women’s choices are often constrained, for Table 16: Ability of Women to make Decisions to Seek Healthcare for Themselves Category Share (%) Decision made by women alone 15.98 Decision made by women and husband/partner 40.64 Decision made by husband/partner alone 40.60 Decision made by someone else and other categories 2.79 All categories 100.00 Source: Authors’ estimation using DHS data. Another form of gender inequality leading to health risks another person). While the consequences of some forms for women is gender-based violence (GBV). Violence is of violence may not be highly visible, they are nevertheless ubiquitous at home, in school, at work, and in communities. always damaging. Violence is also often multidimensional, The World Health Organization (1996) defines violence as meaning that individuals are often subjected to multiple “the intentional use of physical force or power, threatened or forms of violence and in multiple locations. actual, against a person or group that results in or has a high likelihood of resulting in injury, death, psychological harm, Violence affects boys as well as girls, and men as well as maldevelopment, or deprivation.” The harm can be actual or women, but GBV is especially severe in many countries threatened. It can lead to injury or death, but also to trauma and often condoned by social norms (Garcia-Moreno et and mental health symptoms. Three main types of violence al., 2005). GBV can lead to negative and at times dramatic are often distinguished: sexual violence (any sexual act, health consequences for women (World Bank, 2012, 2016). intimidation, attempt to obtain a sexual act, unwanted sexual It leads to increased absenteeism at work and limits mobility, comments or advances against another individual using thereby reducing productivity and earnings. It may force girls coercion), emotional or psychological violence (including to drop out of school, and when going to school it puts them verbal and emotional abuse, such as isolating, rejecting, at risk of abuse. It affects agency, including whether women ignoring, insulting, spreading rumors, making up lies, name- can seek care when needed. As is the case for violence more calling, ridiculing, humiliating and threatening), and physical generally, GBV may take many forms, including not only violence (any form of physical aggression with intent to hurt physical and sexual violence, but also emotional, and even FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 44 economic violence, as well as harassment experienced in pocket expenditures, and foregone income for households public and in places of work and education. In the case of is at 3.3 percent of GDP (Duvvury et al., 2013; see also violence against children, estimates suggest that boys are earlier work by Morrison and Orlando, 1999). Other studies slightly more likely to be affected by physical violence, but also find a large economic cost of violence (CARE 2010 for girls are much more likely to be affected by sexual violence. Bangladesh, CEDOVIP 2016 for Uganda, and Vara 2014 For emotional violence, estimates are somewhat similar for Peru), and guidance is available from the literature on between the two groups (the estimates provided in Table 17 what works to prevent GBV (Ellsberg et al., 2014; see also do not include the risk of bullying which is highly pervasive in the resources available under DFID’s What Works Initiative). schools and similar in terms of prevalence for boys and girls; As is the case for many other aspects of gender inequality see Wodon et al., 2020, for details). Beyond the individual discussed in this study, it is also worth noting that there are harm inflicted on women and their families, GBV is a global relationships between issues such as child marriage and the problem with substantial economic costs. While this study lack of education for girls and the risk of violence, especially does not provide estimates of the economic costs of violence for intimate partner violence (IPV). Ending child marriage against women, a growing body of evidence measures these and increasing girls’ schooling could bring with it a reduction costs. In Vietnam, the estimated loss in productivity, out-of- in IPV (Savadogo and Wodon, 2019). Table 17: Likelihood of Children Being Affected by Various Forms of Violence Category Share for Boys (%) Share for Girls (%) Physical violence 58.68 55.49 Emotional violence (not including bullying) 7.64 5.79 Sexual violence 16.99 25.09 Source: Wodon et al. (2020), based on Violence against Children Surveys for eight developing countries. 45 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 DOMAIN 5: AGENCY, and her husband/partner) is less than 10 years; (5) women do not accept wife beating; (6) women are as likely as men to DECISION-MAKING, AND work; and (7) several variables related to women’s decision- making in the village or area where a specific woman lives are SOCIAL CAPITAL assumed to be improved (leave-out-mean variables at the level of the survey’s primary sampling units). Comparisons WOMEN’S DECISION-MAKING are again provided with the potential impacts on the index of decision-making of ending child marriage and achieving Gender inequality is associated with losses in decision- universal secondary education. making (solo or joint) for girls and women. For example, child brides are often vulnerable—they are young, often Table 18 and Figure 9 provide the average impacts observed poorly educated, and from disadvantaged socio-economic across the same set of 19 countries as before in absolute backgrounds (Parsons et al., 2015). When they marry early, terms (increase in the share of women using modern they may fall even more under the control of their husband contraception methods) and proportional terms (increase in and in-laws than would be the case if they had married later. contraceptive use in percentage terms from the base value). This may limit their aspirations, as well as agency (Klugman et The estimated impacts are large. On average across the 19 al., 2014), possibly limiting their decision-making, including countries, under gender equality, women’s decision-making for access to health care during pregnancy and delivery. would increase by close to 24 points. This represents an increase of almost half from the average base index. In this To assess the impact of gender inequality on decision- case, the impacts are larger than those estimated for child making, an index is constructed through principal component marriage and educational attainment for women, in part analysis to reflect four categories of decision-making3. The because community-level factors including those related index considers whether women are able to make decisions to decision-making for women play a large role in affecting either by themselves or jointly with their partner in a number individual-level decisions, and those factors are expected to of different areas. Similar to the analysis for total fertility, improve once gender equality is achieved. contraceptive use, stunting and under-five mortality, the potential impact of gender equality is captured through regression analysis using the following proxies for gender equality: (1) no child marriage; (2) women having the same education as men; (3) higher earnings which lift households who are in the poorest quintile to the second quintile of well- being, and households in the second quintile to the third; (4) the spousal age gap (the difference in age between the wife Table 18: Potential Impact of Gender Inequality on Women’s Decision-making Absolute difference Percentage difference (%) Impact from ending child marriage -0.01 0.21 Impact from universal secondary education 7.15 13.13 Impact from achieving gender equality 23.85 44.89 Source: Authors. Regression analysis based on data from Demographic and Health Surveys. Note: Estimates are based on country-level analysis for 19 developing countries. Average potential impacts reported for countries where coefficients for the variables of interest are statistically significant. 3 First, married women are asked in the surveys about who makes decisions in the household in four areas: health care, household purchases, visits to friends and relatives, and the use of the husband’s earnings. Second, women are asked if they can refuse to have sex with their husband and if they can request their husband to use a condom when having sex. Third, women respond to four different circumstances assessing if a husband is justified in beating their wife in those instances. Finally, women are asked whether getting their husband’s permission to get medical help for themselves is a major problem or not. When joint decision making is included in the potential answers, this is considered as decision-making ability for the woman in the same way as sole decision-making is. The index takes a value between zero and 100. Sensitivity analysis suggest that the results tend not to be affected much whether the index includes all or only part of these decisions – for example, results remain when decisions related to household purchases are not included. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 46 Figure 9: Increase in Decision-making Under Various Scenarios Impact from achieving gender equality Impact from universal secondary education Impact from ending child marriage -5.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Source: Authors’ estimation using DHS data. BIRTH REGISTRATION using the following assumptions based on the correlates used in the regression analysis: (1) child marriage is eliminated; Another indicator related to women’s agency is birth (2) women are assumed to have the same education as men; registrations. The benefits of birth registration are important (3) gender inequality is assumed through higher earnings for children, not only for the exercise of a range of for women to lift households who are in the poorest quintile fundamental rights, but also for being able to assess the age to the second quintile of well-being, and households in the of girls at marriage. How could gender inequality affect birth second quintile to the third; and (4) women are assumed to registration rates? One example relates to countries where be involved in a larger share of decisions within the household laws against child marriage are enforced, but the practice (this is done through leave-out-mean variables at the level of continues nevertheless. In such cases, young mothers the survey’s primary sampling units). might be hesitant to register their children at birth or other members of the household may decide that the child should The estimated impacts are relatively small (Table 19). On not be registered. While it is likely that other factors than average across the 19 countries, under gender equality, the child marriage or early childbearing would have a larger birth registration rate would increase by only two percentage effect on birth registration rates (see for example Wodon points. This represents an increase of about five percent and Yedan, 2019, on Niger), gender inequality could have from the average base value. The impacts are smaller than an effect. The procedure to measure the potential impact those that would result from achieving universal secondary of gender inequality on the likelihood of birth registration education for girls. Figure 10 visualizes the estimates. is again the same as for other DHS-based indicators. The impact of gender equality on birth registration is measured Table 19: Potential Impact of Gender Inequality on Birth Registration Absolute difference Percentage difference (%) Impact from ending child marriage 0.68 1.05 Impact from universal secondary education 11.64 28.89 Impact from achieving gender equality 2.04 4.62 Source: Authors. Regression analysis based on data from Demographic and Health Surveys. Note: Estimates are based on country-level analysis for 19 developing countries. Average potential impacts reported for countries where coefficients for the variables of interest are statistically significant. 47 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Figure 10: Increase in Birth Registration Rate Under Various Scenarios Impact from achieving gender equality Impact from universal secondary education Impact from ending child marriage 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Source: Authors’ estimation using DHS data. BOX 14: ALTRUISTIC BEHAVIORS, FRIENDSHIPS AND SUPPORT NETWORKS, AND SOCIAL INSTITUTIONS Using the Gallup world poll, Wodon et al. (2018) look at the potential impact of women’s educational attainment on (1) whether women made in the past month a monetary contribution to a charity; (2) whether they volunteered their time with any organization in the past month; and (3) whether they helped a stranger or someone they did not know who needed help. A secondary education is associated with an increase in the three behaviors of four to six percentage points versus primary education. For tertiary education, the increase is at 10 to 14 points. Two other questions in the Gallup World Poll relate to whether women are satisfied with their opportunities to make friends, and whether they can rely on these friends when in need. In comparison to women with only a primary education or less, a higher level of education is not associated with an increase in the opportunity to make friends, but it is associated with a higher ability to rely on such friends when in need. Gender inequality could thus reduce altruistic behaviors and social capital through its impact on educational attainment. under-five stunting rates; and (3) Welfare effects from high SELECTED POTENTIAL population growth. While this is clearly not an exhaustive list of impacts, it probably captures some of the largest ECONOMIC COSTS economic costs of gender inequality. The measurement of costs is based on data on the changing wealth of nations and MEASUREMENT APPROACH AND BASELINE specifically human capital wealth (Hamilton et al., 2018). WEALTH ESTIMATES Arguments for using a wealth as opposed to an income approach were provided earlier when discussing the impact of Gender inequality has major potential negative impacts for gender inequality on labor earnings. girls and women, their children and their households, their communities, and societies. Some of these potential impacts Table 20 provides global wealth estimates in constant US have been documented in previous sections. What are the dollars of 2014. As estimated in Lange et al. (2018), global economic costs associated with those potential impacts? In wealth stood at US$1,143 trillion in 2014, suggesting an many cases, this is a hard question to answer, but for a few increase in real terms of 66 percent over 20 year. Human potential impacts, estimations can be provided. The focus in capital wealth was at US$737 trillion. Globally, human capital this section is on three impacts: (1) Lost human capital wealth accounts for more than two thirds of total wealth, versus due to gender inequality in lifetime earnings; (2) Lost human just under one tenth for natural capital and about a quarter capital wealth due to the impact of gender inequality on for produced capital. In per capita terms, total wealth stood FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 48 at US$168,580 per person in 2014 versus US$128,929 because labor earnings as a share of GDP have declined in in 1995. Human capital wealth stood at US$108,654 per OECD countries due to technological change, stagnating person in 2014 versus US$88,874 in 1995. As will be wages, and in some countries a reduction in the share of the shown in subsequent sections of this note, inequality in population in the labor force due to ageing. By contrast, for human capital and total wealth between countries is high. low income and lower middle-income countries, the share of In high income OECD countries, total wealth per capita is human capital wealth in total wealth is increasing as countries above US$700,000, and human capital wealth is at close achieve higher levels of economic development. At lower to US$500,000 per person. This is more than 90 times levels of economic development, natural capital continues the levels in low income countries where human capital to account for a larger share of wealth. Many developing wealth is at US$5,564 per person. In Table 19, estimates countries are experiencing a demographic transition and are also provided for 2017 for human capital wealth based are reaping benefits from the demographic dividend as on projections in order to provide updated values for losses population growth rates slow and the population is becoming in wealth due to gender inequality in earnings for that year. better educated. But this is not the case for all countries, The projected estimates for 2017 are higher than those for especially in sub-Saharan Africa. 2014 due to both growth in real terms in GDP per capita and thereby in labor earnings, and growth in population sizes for LOSS IN HUMAN CAPITAL WEALTH FROM most of the countries. GENDER INEQUALITY IN EARNINGS At the global level, the dynamics of human capital wealth Estimations of human capital wealth are done separately accumulation are driven by shifts in OECD and upper- for men and women. Losses in human capital wealth due middle income countries simply because those countries to gender inequality are calculated by raising earnings for account for 87 percent of global wealth (65 percent for the women to the level of men without losses for men, taking OECD alone). The proportions are even larger for human into account the proportions of men and women in the adult capital wealth. In these countries, the share of human capital population (see Appendix 2 for details). Consider first the wealth in total wealth has fallen slightly in recent years in part estimates for 2014 in Table 21. As mentioned earlier, on a per Table 20: Baseline Estimates of Global Wealth, 1995-2014 1995 2000 2005 2010 2014 2017(*) Total wealth, Trillions, constant 2014 US$ Total wealth 689.9 790.9 889.1 1,024.7 1,143.2 NA Produced capital 164.8 187.9 226.9 269.0 303.5 NA Natural capital 52.5 54.2 70.0 97.2 107.4 NA Human capital 475.6 552.7 595.4 661.1 736.9 781.9 Net foreign assets -2.9 -3.9 -3.3 -2.6 -4.6 NA Population (billions) 5.35 5.73 6.09 6.47 6.78 NA Per capita wealth, constant 2014 US$ Total wealth 128,929 138,064 145,891 158,363 168,580 NA Produced capital 30,793 32,801 37,237 41,570 44,760 NA Natural capital 9,803 9,463 11,487 15,019 15,841 NA Human capital 88,874 96,478 97,707 102,170 108,654 NA Net foreign assets -540 -678 -539 -395 -676 NA Share of total wealth Total wealth 100% 100% 100% 100% 100% NA Produced capital 24% 24% 26% 26% 27% NA Natural capital 8% 7% 8% 9% 9% NA Human capital 69% 70% 67% 65% 64% NA Net foreign assets 0% 0% 0% 0% 0% NA Source: Lange et al. (2018). Note: Estimates for human capital wealth in 2017 are projections. 49 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 capita basis including not only the adult population but also at US$1,143 trillion for the 141 countries included in our children, losses in wealth are estimated at US$23,620 per analysis, while global GDP for those countries is estimated person. Globally, women’s human capital could increase from at US$75 trillion5. Wealth is thus 15 times larger than US$283.6 trillion to US$453.2 trillion with gender equality. GDP. But in proportionate terms, our estimate is more This would be a gain in global wealth from gender equality of conservative. We suggest a loss of 14 percent of baseline US$160.2 trillion in 2014 (about twice the value of global wealth. This is smaller than the loss of 26 percent of GDP GDP). This gain represents 21.7 percent of human capital suggested in the McKinsey Global Institute study. As wealth and 14.0 percent of total wealth including natural and discussed in Wodon (2018), various factors could account produced capital and net foreign assets. for the difference in proportional impacts, including the fact that our estimates of human capital wealth account for the Over time, total wealth lost due to gender inequality labor share in GDP. Still, both types of estimates are only increases from US$123.2 trillion in 1995 to US$160.2 trillion meant to give orders of magnitude of potential losses from in 2014. This increase comes from population growth, as well gender inequality as opposed to precise values. Clearly, the as higher standards of living. But other factors that affect losses from gender inequality are potentially very large. human capital wealth at the country and regional level also play a role. As a share of baseline wealth, losses from gender The largest losses in wealth from gender inequality are inequality tend to be slightly lower in 2014 than in 1995. observed for East Asia and the Pacific, North America, and This is in part because there is a (slow) movement towards Europe and Central Asia, in each case at between US$40 more gender equality in earnings in many countries over trillion and US$50 trillion in 2014. This is because these time, which makes the losses smaller. But in addition, human regions concentrate much of the world’s human capital capital in high income countries has been declining slightly wealth. In per capita terms as well, the losses are larger in in recent years due among others to ageing and a reduction those regions. But losses in other regions are substantial too, in the share of labor income in GDP. This leads to a small including in comparison to current levels of development. In reduction in losses from gender inequality over time as a South Asia, losses from gender inequality are estimated at share of the baseline global wealth. US$9.1 trillion. In sub-Saharan Africa, losses are at US$2.5 trillion or 11.4 percent of the base regional wealth. The Projections for the losses in human capital wealth due to loss in wealth from gender inequality as a share of baseline gender inequality in earnings for 2017 are at US$172.3 wealth is highest in South Asia, the region with the lowest trillion globally, a higher value than the estimate for 2014. As share of women in human capital due to low labor force already mentioned, this results not only from growth in real participation among women. Table 21 also shows that the terms in GDP per capita and thereby in labor earnings, but largest total losses in wealth are observed as expected for also from growth in population sizes for most of the countries. high income OECD countries and upper-middle income These estimates are based on the gender shares in human countries (which include China). Together these two groups capital wealth observed in 2014, but these shares tend to of countries experience a loss of US$140.2 trillion in human change slowly over time, so no major bias is to be expected. capital wealth due to gender inequality. The other countries together lose US$20 trillion in human capital wealth. But How do these results compare to previous studies? The again, in percentage terms from the base, the picture is McKinsey Global Institute (2015) study reports potential different. Low income countries lose 15.1 percent of their gains in GDP from a ‘full potential’ gender equality scenario base level of wealth (including all types of capital) under of US$28 trillion or 26 percent of GDP in 2025 versus a gender inequality, which is slightly larger than the increase ‘business-as-usual’ scenario4. We report losses in human for the world, at 14.0 percent. Note also that losses from capital wealth from gender inequality of US$160 trillion or gender inequality are lower in proportional terms from the 14 percent of our baseline estimate of global wealth in 2014. base in high-income non-OECD countries, in part because Our estimate is larger in absolute value because wealth many of these countries have substantial oil reserves and is larger than GDP. In 2014, global wealth is estimated thereby higher levels of natural capital in their baseline wealth. 4 The McKinsey study also considered a best-in-region scenario in which all countries would match the rate of improvement of the best-performing country in their region. This would add $12 trillion in annual GDP by 2025. 5 Our estimation includes a larger set of countries than included in the McKinsey Global Institute study, although this does not make a very large difference for estimates of global losses given that most of the wealth, especially for human capital wealth, remains concentrated in upper middle income and high-income countries and the fact that these countries are also included for the most part in other studies including that by the McKinsey Global Institute. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 50 Table 21: Aggregate Losses in Wealth from Gender Inequality in Earnings (US$ of 2014, Trillions) 1995 2000 2005 2010 2014 2017(*) Global estimates by gender and gain from gender equality Human capital, men 301.2 349.1 371.6 405.5 453.2 486.2 Human capital, women 174.4 203.6 223.8 255.6 283.6 303.2 Counterfactual human capital, women 297.6 344.5 366.4 398.4 443.8 475.4 Increase in human capital 123.2 140.9 142.6 142.8 160.2 172.3 Loss as share of baseline total wealth 17.9% 17.8% 16.0% 13.9% 14.0% NA Regional estimates of gain from gender equality East Asia & Pacific 34.2 35.8 37.7 42.1 49.9 54.8 Loss as share of baseline total wealth 24.5% 22.1% 20.8% 17.1% 16.6% NA Europe & Central Asia 32.4 36.3 37.2 38.8 41.6 43.6 Loss as share of baseline total wealth 14.3% 14.8% 13.7% 13.0% 13.3% NA Latin America & Caribbean 7.3 5.9 6.5 6.7 6.7 6.7 Loss as share of baseline total wealth 14.3% 10.5% 10.2% 8.8% 7.9% NA Middle East & North Africa 1.6 2.1 2.4 2.7 3.1 3.3 Loss as share of baseline total wealth 10.2% 11.8% 9.9% 7.7% 7.4% NA North America 43.4 55.1 51.3 43.3 47.2 49.9 Loss as share of baseline total wealth 18.8% 19.5% 16.3% 13.3% 13.5% NA South Asia 3.3 4.6 6.5 7.4 9.1 11.4 Loss as share of baseline total wealth 28.8% 32.2% 35.0% 29.4% 29.4% NA Sub-Saharan Africa 1.1 1.1 1.0 1.9 2.5 2.8 Loss as share of baseline total wealth 7.6% 8.8% 6.3% 9.8% 11.4% NA Income groups estimates of gain from gender equality Low income countries 0.4 0.5 0.6 0.8 1.1 1.3 Loss as share of baseline total wealth 11.5% 13.5% 13.8% 14.2% 15.1% NA Lower-middle income countries 6.8 7.6 9.4 11.0 13.5 16.2 Loss as share of baseline total wealth 19.2% 20.7% 20.4% 18.1% 19.1% NA Upper-middle income countries 11.2 11.3 16.1 20.9 26.5 30.4 Loss as share of baseline total wealth 11.8% 10.0% 11.9% 10.4% 10.7% NA High income non-OECD 2.7 3.6 3.8 4.7 5.4 5.4 Loss as share of baseline total wealth 6.5% 8.6% 7.4% 7.1% 7.0% NA High income OECD 102.2 117.9 112.6 105.4 113.7 119.0 Loss as share of baseline total wealth 19.8% 19.8% 17.3% 15.2% 15.3% NA Source: Wodon (2018); see also Wodon and de la Brière (2018). Note: Estimates for human capital wealth in 2017 are projections. LOSS IN HUMAN CAPITAL WEALTH FROM economic costs. Research suggests a loss in productivity UNDER-FIVE STUNTING associated with lower height as an adult (Strauss and Thomas, 1998; Caulfield et al., 2004; Dewey and Begum, 2011). While gender inequality in earnings matters for all countries, Undernutrition could lead to economic losses equivalent to under-five stunting matters more for developing countries, four to 11 percent of Gross Domestic Product in sub-Saharan and especially low income countries where stunting rates are Africa and Asia (Horton and Steckel, 2013). An experiment substantially higher. For stunted children and their families, in Guatemala suggests that children who benefited from the cost of stunting may not be primarily economic. But nutrition supplements were less likely to be stunted and when considering the potential impact on human capital had better cognitive abilities and higher wages in adulthood wealth of stunting due to gender inequality, the focus is on (Hoddinott et al., 2008, 2011, 2013a, 2013b). 51 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 BOX 15: COUNTRY ANALYSIS OF THE COST OF GENDER INEQUALITY: THE CASES OF NIGER, TANZANIA, GUINEA, AND UGANDA Globally, the cost of gender inequality related to under-five stunting and population growth as well as budget savings are much smaller than the cost of gender inequality in earnings. This is because gender inequality in earnings affects women in virtually all countries including upper middle and high income countries, while the effect of reductions in stunting and population growth are much smaller in those countries. Yet in low income countries, impacts through these channels matter substantially. As an example, consider Niger where losses were measured as a share of GDP using similar approaches to those used in this study. Losses due to gender inequality in earnings were estimated at 23 percent of GDP (World Bank, 2018cb). Initially, the value of losses associated with the reduction in population growth is much smaller, but by 2030, these losses are equivalent to about a fifth of the losses due to inequality in earnings. If budget savings for the education sector from lower population growth are included, losses from population growth account for more than a fourth of the losses due to inequality in earnings. Finally, adding losses related to under-five mortality and stunting bring total losses for these channels to almost one third of the value of the losses due to gender inequality in earnings. Similar analysis was conducted for Tanzania (World Bank, 2019b), Uganda (Wodon et al., 2019), and Guinea (World Bank, 2019c) with broadly similar results. Various studies have considered the impacts and economic LOSS IN WELFARE (TOTAL WEALTH PER CAPITA) cost on stunting (Shekhar et al., 2016; Skoufias et al.. 2019; FROM HIGH POPULATION GROWTH and Galasso and Wagstaff 2019). Here, to be consistent with the approach to costing based on human capital wealth, costs Losses in welfare are meant in this study to represent losses are based on losses in wealth from stunting. The estimation in wealth per capita due to high population growth. To considers three variables: (1) the stunting rate for young show how much population growth matters in low income children; (2) the reduction in stunting from that rate that countries, consider Figure 11 which compares growth in could be achieved thanks to gender inequality as measured total wealth between 1995 and 2014 with growth in wealth earlier; and (3) the estimated gain in earnings in adulthood per capita for countries classified by income categories. and thereby human capital wealth that would result given The difference between growth in wealth and in wealth per estimates in the literature of the gains in earnings associated capita is population growth. While total wealth increased in with preventing stunting in early childhood. Essentially most countries over the last two decades, per capita wealth multiplying these variables by prevailing levels of human did not. It grew fastest in middle-income countries, but due capital wealth gives an estimate for each country of the in large part to high rates of population growth, gains were impact of gender inequality on wealth through stunting. smaller in low income countries. The growth in wealth was Based on this approach, the cost of gender inequality due not sufficient in some of these countries to keep up with to its impact on stunting for children under the age of five population growth, making it harder for those countries to is estimated at US$71 billion in 2014 for 17 developing reap the benefits of the demographic dividend. countries with a population of more than two billion people. This is much smaller than lost human capital wealth from Gender inequality has a large potential impact on lifetime gender inequality in earnings, but still substantial for the fertility and population growth, both directly and through countries affected, and especially the individuals affected by a reduction in child marriage and early childbearing. In losses in earnings in adulthood due to stunting in childhood. 16 countries for which simulations were carried out with FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 52 Figure 11: Changes in Total and Per Capita Wealth by Income Status, 1995 to 2014 200% 150% Percent 100% 50% 0% Low income Lower middle Upper middle High income: High income: World income income non-OECD OECD Total Wealth Per capita wealth Source: Lange et al. (2018b). demographic projection tools, the average reduction in countries. However, in comparison to the cost of gender population growth from gender equality was estimated inequality in earnings globally, gains related to population at -0.26 percentage points. The reductions in annual growth are smaller because the countries that would benefit population growth rates are however different depending from substantial reductions in population growth have much on which country is considered. In India, the largest of the lower estimates of total wealth than upper middle and high countries, the reduction was smaller because the country has income countries where impacts on population growth would already gone through much of its demographic transition. be small. Still, while losses from higher population growth For perspective, India’s annual population growth rate is due to gender inequality for women would initially be quite currently at 1.2 percent per year, versus more than three smaller than losses related to women’s earnings, these losses percent per year for some other countries like Niger. How would be far from negligible and would increase over time in much is this worth in terms of human capital wealth per countries with high population growth. capita? In the medium term, since children who would not be born today would be adults only at the end of the period BUDGET COSTS FROM HIGH POPULATION for the simulations (2030), there is no reduction over the GROWTH time horizon in the labor force versus the business as usual scenario. Lower population growth then results in an increase Another benefit from reduced population growth in in human capital wealth per capita since the denominator countries with high fertility rates is reduced pressure for state (population) becomes smaller while the numerator (human budgets to provide services to the population, or an ability capital wealth) does not change (it could actually increase if to provide higher quality services. As an illustration, consider lower fertility rates lead to higher labor force participation results of simulations for education budgets. In the first few by women). years after achieving gender equality, there is no impact on the size of new cohorts of children entering primary If gender equality could be achieved, the first year benefit school, but soon thereafter there is a reduction in cohorts. from lower population growth is valued at US$80 billion This pattern is observed with a lag for secondary schools for 16 developing countries with a population of 2.3 billion too. By reducing the size of new cohorts of children going people. Additional benefits would accrue in subsequent to school, achieving gender equality may provide significant years. Over time, gains would grow rapidly, ultimately savings or enable investments in quality. To estimate those rivalling gains from gender equality in earnings in those savings, assumptions are needed for trends in enrollment and 53 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 completion rates, the efficiency of the education system, increase alongside economic growth in the countries. These recurrent unit costs at various levels of schooling and how three factors are the main reasons why budget savings over these may change over time with economic growth and time become larger. Note that achieving those budget improvements in standards of living (these assumptions savings would entail initial costs since for example to end themselves depend on parameters such as teacher salaries child marriage – one of the components of the simulations and pupil-teacher ratios), and the cost of capital investments on gender equality, interventions are needed to keep girls for school construction. The market share of private schools in school, leading to costs for states and out-of-pocket and also plays a role in the estimation. Fortunately, a costing opportunity costs for parents. These costs would offset model prepared for the 2015 Education for All Global some of the benefits from the reduction in the size of future Monitoring Report can be used for the analysis (Wils, 2015). cohorts of students thanks to lower population growth. The costing model was developed to estimate total costs and external finance needs to reach full primary and secondary The broader message here is that many countries are not education in low- and lower-middle income countries. investing enough in their people. Reducing gender inequality has a role to play in enabling better and more investments Based on simulations for 16 countries, cost savings for in people. Women should have the ability to manage their education budgets from lower population growth under fertility. This is unfortunately often not the case as suggested a scenario of gender equality could reach US$27 billion by unmet needs for family planning. At the national level, by annually by 2030. This would represent 7.4 percent of contributing to high fertility and population growth, gender the expected education sector budgets in those countries inequality may put pressure on the quality of the services by 2030. The budget savings start from a low base and provided by governments to their population. This is because increase over time for three main reasons. First, the impact higher population growth makes it necessary to spread of achieving gender inequality on population growth is small budget resources more thinly, including to provide basic in terms of the reduction in the population of children in education to ever larger cohorts of students. If annual rates age of schooling in the first few years, but rises over time. of population growth were lower, resources might become Second, the counterfactual scenario includes rising budget available to invest in higher quality services. Savings resulting expenditures meant to cover the cost of progressively from lower population growth could then be reinvested in achieving universal secondary education and thereby higher quality services, which themselves would probably progressively reducing the share of children out of school. contribute to reducing gender inequality. Finally, the unit costs for each child to be in school also BOX 16: AVOIDING DOUBLE COUNTING IN ESTIMATING COSTS The four costs highlighted in this section – losses in human capital wealth from gender inequality in earnings, losses due to high levels of population growth due to both welfare losses and budget costs for states of providing basic services, and losses related to stunting do not “double count” as costs as there is no overlap between them. Other types of costs could be documented in future work, and in some cases have been documented in past or on-going work. One example is the costs associated with gender-based violence which affects mostly women. Those cost can represent a large share of GDP, or alternatively human capital wealth. One important point however is that policy makers and advocates need to be careful when combining different costs to avoid double counting. For example, the cost of gender-based violence in terms of lost earnings for women due among others to injuries or other forms of trauma is already implicitly accounted for in the analysis when looking at gender inequality in earnings. By contrast, costs related to health spending because of gender-based violence are not accounted for in our analysis. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 54 is on three types of investments along a simple life cycle POLICY OPTIONS FOR model: (1) Investments in early childhood development to reduce the impact of gender inequality as experienced GENDER EQUALITY by mothers on their young children; (2) Investments in adolescent girls to delay marriage and childbearing while also There is a major difference between developed and improving education opportunities, which would help reduce developing countries in the economic costs generated by population growth; and (3) Investments in adult women gender inequality. In developed countries, costs related to to improve employment and earnings opportunities and gaps in labor market earnings tend to dominate, given that thereby increase human capital wealth. other costs (as measured in this study) tend to be small since the countries have lower levels of population growth and This categorization in three buckets of policies for gender under-five stunting, among others. By contrast, in developing equality corresponds to three distinct periods in life is countries, while costs associated with gender inequality in for expository purposes. In practice the various types of earnings are also high, costs related to population growth and polices have multiple impacts and rationales. For example, under-five stunting are far from being negligible, and in some interventions in early childhood contribute to higher labor cases may exceed costs related to gender gaps in earnings, at productivity later in life, as does educational attainment least over a sufficiently long period of time. The implication is for girls in adolescence (for a more comprehensive look that in developed counties interventions to reduce the costs at policies related to the life cycle in the context of of gender equality may focus in large part on labor market employment and productivity, see World Bank, 2010 earnings and other factors affecting income levels. However, and World Bank, 2019d). Similarly, while some of the in developed countries, and especially in low income interventions in early childhood relate to the availability of countries, high rates of population growth as well as poor family planning with implications for fertility and population education, health, and nutrition outcomes must be tackled growth, delaying child marriage and early childbearing as well. This is why in low income countries, investments also contributes to lower fertility, as do employment in adolescent girls are so important, including to improve opportunities for women among others by shifting the trade- educational attainment, reduce child marriage, and prevent off between so called productive and domestic work. And early childbearing. finally, while interventions in early childhood clearly matter for young children, so do interventions for adolescent girls The main objective of this study is to estimate selected and adult women. economic impacts and costs of gender inequality. But it is also useful to provide guidance on investments that could Focus on a subset of investments to achieve gender help reduce gender inequality. The guidance provided is equality in this study does not mean that other types of not meant to be comprehensive, nor exhaustive. Since investments are not important, nor needed. Many other gender inequality affects girls and women in virtually all types of interventions could be advocated, and a few aspects of their life, a wide range of interventions could be examples of such interventions will be provided through implemented. To keep this section relatively short, the focus boxes in what follows. But because the three types of 55 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 investments highlighted here are related directly to the and cognitive development, are all interrelated (Shonkoff three large economic costs of gender inequality noted et al., 2012). Neurological studies show that synapses earlier, these are the investments considered in this study. (structures that permits neurons in the brain to pass signals In addition, the study calls for targeting high prevalence to other neurons) develop rapidly in the first 1,000 days areas for gender inequality or some of its manifestations of a child’s life. Sensory pathways for vision and hearing when implementing interventions, simply because under develop first, followed by language skills and higher cognitive limited resources prioritization is needed, and for the functions (Nelson, 2000). These synapses form the basis preparation of diagnostic and strategies to achieve gender of cognitive and emotional functioning later in life. As a equality. Those two additional topics are also briefly result, inadequate development, for example due to poor discussed in subsequent sections. nutrition, may not only lead to poor physical growth, but it may also impede brain development, with negative impacts later in life on academic achievement as a student and future INVESTMENTS IN YOUNG productivity as an adult. This provides a strong economic case for implementing nutrition and other interventions such CHILDREN as early stimulation to improve ECD outcomes (Wodon and Shekhar, 2016). In some countries, gender inequality manifests itself from early childhood or even before. Extreme cases include the There is an emerging consensus that investments in ECD issue of “missing girls”, which refers to girls who may never not only should be a priority to enable children to reach be born due to parental preferences for boys. After birth, their full potential, but also have high economic rates of girls may still be at a disadvantage if they do not benefit from returns (Carneiro and Heckman, 2003; Heckman and the same investments and protections as boys. But more Masterov, 2007; Engle et al., 2011; Denboba et al., 2014), importantly for this study, even if differences in indicators particularly when compared to investments made at later such as under-five mortality and stunting tend to be small stages in life. As noted by Black et al. (2017), investing in between boys and girls, the more important issue is that young children is one of the best investments that countries gender inequality affects outcomes for mothers, and this can make. A child’s earliest years present a unique window in turn affects their children’s well-being, whether they are of opportunity to address inequality (including gender boys or girls. In other words, in matters of early childhood inequality), break the cycle of poverty, and improve a wide development, the issue of gender inequality considered here range of outcomes later in life. Brain research suggests is less related to gender gaps in indicators than to how gender the need for holistic approaches to learning, growth, and inequality may affect boys and girls alike. As shown in the development, recognizing that young children’s physical analysis of the impacts of gender inequality, factors such as and intellectual well-being, as well as their socio-emotional high rates of child marriage and early childbearing and low and cognitive development, are interrelated. By the end educational attainment for girls lead their young children to of early childhood, young children should be healthy and be at higher risk, among others, of under-five mortality and well-nourished; securely attached to caregivers; able to stunting. Some of the interventions that can help reduce interact positively with families, teachers and peers; able such risks target young girls and mothers, and they will be to communicate in their native language; and ready to discussed in the next sections. But a range of interventions learn throughout primary school. In contrast, early gaps in can also be implemented at the level of children – both childhood development jeopardize a child’s capacity to reach boys and girls, to ensure that they have a good start in life, these important milestones. Advances in biological and social and to prevent negative effects from gender inequality. In science evidence provide a wealth of resources to inform addition, gender stereotypes are learned at a young age, innovative strategies that promote optimal child growth including through differences in the way boys and girls are and development. Programs that combine services (such as taught (or parented). This may have impacts later in life for a nutrition and psychosocial stimulation) can have especially confidence, persistence, perspectives, and bias. large beneficial impacts and rates of return. Unfortunately, most countries fall short in their delivery of essential services Early childhood development (ECD) is a complex multi- for young children and their families. The challenge is to dimensional process. The various aspects of young children’s develop scalable, cost-effective models for delivering these development – including their physical, socio-emotional. services in low- and middle-income countries. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 56 Several agencies have suggested frameworks to address of gender inequality should be to promote opportunities the needs of young children. A recent series on ECD in for a double redistribution of care work for young the Lancet draws on the concept of nurturing care (Black children, not only within households from female to et al., 2017). UNICEF had long emphasized the need for male members, but also from households to public and multiple interventions for ECD, including basic health, private service providers through the provision of quality nutrition, education, and protection services. The World and affordable care services. Not all women might take Health Organization (WHO) has established guidelines for advantage of such service provision if they prefer to each developmental phase, including pregnancy, postnatal, stay at home to take care of their young children, but baby, infant, and young child health care. The Partnership opportunities to use such services should be provided. for Maternal, Newborn and Child Health has provided policy-makers with specific information on the essential • As noted by Devercelli and Beaton-Day (2020), while health interventions to address the main causes of maternal, childcare is an issue that impacts all working parents, newborn, and child deaths. At the World Bank, Denboba it is particularly important for women’s employment. et al. (2014) suggested a list of 25 interventions considered Lack of affordable childcare often keeps women out as essential for young children. These interventions can of the workforce or prevents them from reentering be delivered through five integrated packages at different the workforce after childbirth. It also limits the quality stages in a child’s life: (i) the family support package, which of employment and income earning opportunities should be provided throughout the ECD period, (ii) the that women can pursue. This can have a wide range of pregnancy package, (iii) the birth package from birth to six negative impacts, including on the family’s economic months, (iv) the child health and development package, and security, gender equality and empowerment, and (v) the preschool package. In reference to these 25 essential business and economic growth. Aside from reducing interventions, two are especially important for this study’s family income, family resources may also be allocated focus on gender inequality: (1) planning for family size and in different ways when women do not have control spacing, and (2) support to families and especially women over their own incomes. Finally, when women exit related to the care economy. the workforce, firms lose valuable employees, resulting in increased costs related to attrition and • Family planning and birth spacing. A woman’s ability to reduced business productivity, as well as missing space and limit her pregnancies has a direct impact on out on the benefits of a more diverse workforce. her health and well-being as well as on the outcome of each pregnancy, and thereby on her children. Planning for family size and spacing allows parents to anticipate INVESTMENTS IN and attain their desired number of children and the desired spacing and timing of their births. This can ADOLESCENT GIRLS be achieved among others through access to modern contraceptive methods and the treatment of infertility, Patterns of gender inequality become salient during but in many settings programs to improve sexual and adolescence, as evidenced by child marriage, early reproductive health knowledge among adolescent girls childbearing, and low educational attainment for girls in and young mothers can also be highly beneficial. comparison to boys in low income countries. Investing in adolescent girls could generate substantial economic benefits • Reducing, redistributing, and recognizing unpaid for three main reasons: (1) Earlier investments tend to have a work and care. Women spend substantially more persistent positive impact throughout women’s lives. If a girl time in unpaid home-based work than men, and completes secondary education and avoids early marriage, consequently less time in market work. A substantial the benefits endure throughout her life; (2) The cost of share of home-based work is allocated to taking care interventions for girls in adolescence or even earlier tends to of young children. Reducing unpaid work for women be lower than the cost of interventions later in women’s life would free time for market work or other activities. cycles; and (3) Interventions targeted at girls at a formative Various types of policies can help in that regard and age may be more successful in influencing values and are discussed below. But within the area of ECD, a key behaviors, not only for the girls directly targeted but for the policy to improve gender equality and reduce the costs community. If women are targeted later in life, returns on 57 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 the investment may be lower, as it will become increasingly attests. Schools also need to provide access to water, difficult for them to fully benefit from new opportunities. latrines, and hygienic facilities, which may be particularly important for adolescent girls. Where schools cannot While interventions for women at a later point in the life be constructed in locations that meet the needs of cycle are also needed, adolescence is a critical period when communities, providing transportation to school is an investments are likely to generate the highest returns. These alternative. Finally, it is essential to ensure that girls returns come not only from the higher earnings that girls will do not suffer physical, sexual, or other harassment at enjoy in adulthood, but also from the reduction in fertility school or while travelling to and from school (see more and population growth that would follow in countries where generally Abramsky et al., 2014, on gender-based population growth remains a barrier to gains in standards of violence and how to reduce it in Uganda, and Mgalla living. To eliminate child marriage and early childbearing and et al., 1998, on a guardian program in primary schools to enable all girls to complete their secondary education and in Tanzania with female teachers elected by colleagues learn the skills they will need while in school, some general and trained as guardians for female students). conditions for schooling and learning must be met. Specific interventions to delay marriage and childbearing and improve • Ensuring that the education system delivers effective sexual and reproductive health knowledge among girls also learning outcomes. In many countries in Africa (Bashir show promise, including for reducing high fertility rates. Both et al., 2018), and more generally in the developing are considered. world (World Bank, 2018b), student learning outcomes, as measured by national and international student GENERAL CONDITIONS FOR SCHOOLING AND assessments, are poor. This needs to be addressed LEARNING through investments to ensure not just greater access but also improved quality. Priorities in this area depend Given that one of the best ways to end child marriage and on each country, but they may include increasing early childbearing is to keep adolescent girls in school, the number of teachers in line with standards and measures are needed to improve access to education emphasizing subject areas with acute shortages (e.g., and learning while in school. Multiple entry points can be mathematics and science). Better in-service teacher considered. Among them are (1) reducing the disadvantages training is often needed, and teacher awards programs confronting girls in remote communities, at times due to can help encourage reductions in gender gaps in school poor targeting of government resources; (2) creating a more performance when such gaps are observed. Providing in- inclusive school culture for girls; (3) providing girls with role service teacher training to challenge gender differences models, such as female teachers; and (4) raising the returns in teacher expectations and establishing teacher to secondary school completion for women through local mentors to support girls can also help. Guidance on employment opportunities. More generally, there is a need these and other teacher policies is available in Beteille to improve basic general conditions in education systems and Evans (2018) and more generally on improving skills especially in low and lower-middle income countries so that in Africa (World Bank, 2019d). all girls remain in school. Several such conditions are worth emphasizing here: nsuring adequate schooling infrastructure.Secondary • E education completion rates for girls are low in many countries in part because there are just not enough secondary schools. Developing a school construction strategy to bring schools closer would help with ensuring girls’ ability to go to school and their safety on the way to school. School construction can reduce transport costs in areas with extremely low schooling density, with particularly positive impacts for girls, as recent evidence from Afghanistan (Burde & Linden, 2013) and Burkina Faso (Kazianga et al., 2013; Sawada et al., 2016) FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 58 • Ensuring the participation of girls. Schooling must be mentors, staff, and parents to improve educational affordable for families. Affordability refers not just to the quality through low-cost educational resources (see direct costs of participation in secondary education, but for example Alcott et al. 2017; Sabates et al. 2018). also to opportunity costs. In many low-income countries, these costs remain high for the poor, especially for nsuring that social norms are progressively changed. • E girls. Fee free public education at the secondary level In addition to policies targeting the education system, can be a major step forward, yet providing secondary broader efforts are also likely to be required to education free of tuition and other direct costs may progressively change social norms that perpetuate not be enough to ensure the participation of all school- gender inequality. Although an extensive discussion of age children, particularly girls (see Koski et al., 2018). issues related to social norms is beyond the scope of this Conditional or unconditional cash transfer programs can study, it must be recognized that child marriage, early help when coverage is sufficient to reach the extreme childbearing, and low educational attainment for girls poor. They can be used to encourage formal schooling are part of deep-seated patterns of gender inequality or to open non-formal pathways to continuing one’s (Klugman at al. 2014). But at the same time, changing education. Also of interest are programs such as those patterns of child marriage, early childbearing, and low run by Camfed to cover the direct and indirect costs educational attainment for girls through a variety of of schooling for girls while also supporting community- policies and incentives may well be one of the best led initiatives to engage parents and train teacher ways to progressively change existing social norms. BOX 19: IMPROVING EDUCATIONAL ATTAINMENT AND LEARNING FOR GIRLS Several reviews consider interventions to improve education for girls and empower them, including Unterhalter et al (2014), Sperling and Winthrop (2015), Botea et al. (2017), Evans and Yuan (2019), and Wodon (2020). For example, Unterhalter et al. (2014) assess the impact of interventions promoting girls’ education specifically through (1) resources (such as cash transfers) and infrastructure; (2) improved institutions responding to student needs; and (3) changed social norms, especially for those affecting the most marginalized. The review summarized the impact of different interventions on three outcomes: participation; learning; and empowerment. For each type of intervention and outcome, the evidence on the likelihood of impact is considered strong, promising, limited, or weak. For participation, the evidence for the impact of cash transfers, information about the potential employment returns to education, and the provision of schools in underserved and unsafe areas is strong. This was also true for a range of interventions related to teacher training, group learning, measures to promote girl-friendly schools, and learning outside the classroom, such as through tutoring. Group learning, programs for learning outside the classroom, and scholarships linked to student performance were also found to have impacts on learning. The evidence for the impact of interventions on empowerment was generally weaker. Evans and Yuan (2019) note in particular that some past efforts to synthesize evidence on how to improve educational outcomes for girls have focused on interventions targeted to girls. However, non-targeted interventions benefitting both girls and boys may also improve girls’ education. Looking at the evidence from a large set of interventions, the authors suggest that to improve both access and learning for girls, girl-targeted interventions may not necessarily deliver better results than interventions that could benefit boys as well as girls and are thus not specifically targeting girls. For example, cash transfer programs may be directed to households as opposed to specifically girls, or interventions for improved pedagogy in the classroom need not necessarily be gender-specific. 59 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 BOX 20: THEORIES OF CHANGE FOR INTERVENTIONS TARGETING ADOLESCENT GIRLS Life skills and SRH knowledge: By increasing knowledge, life skills can raise girls’ awareness of risks associated with becoming pregnant early and increase their desire and ability to avoid early pregnancies through family planning. Through such channels, life skills may lead to better health outcomes for the girls and their children. By increasing girls’ confidence and self-esteem, life skills may also help expand their aspirations, which may heighten their motivation to delay marriage and childbearing. Finally, life skills can increase the communication and decision-making skills of young women and increase their abilities to negotiate their marriage and childbearing preferences. However, while life skills and SRH knowledge may empower girls, they may not be sufficient to delay marriage and childbearing if social norms curtailing agency are not also addressed. Economic opportunities: Programs to increase young women’s earnings may increase their ability to plan and improve their marriage and childbearing decisions in three ways: (1) Improvement in a woman’s ability to make an economic contribution expands her role beyond that of sex and reproduction. This can increase a girl’s desire to delay marriage or childbearing. The transformation of girls from economic liabilities into assets in the eyes of their communities and families can also alleviate the external pressures on them to marry or have children early. (2) The loss in earnings associated with childrearing is an opportunity cost that may increase women’s desire to limit or space births and to exercise reproductive control. (3) Increased earnings may supplement a young woman’s bargaining power within the household and allow her to effectively exercise reproductive control by negotiating delays in sexual debut or marriage, and to better negotiate the terms of sex, such as use of contraceptives. Creating income-generating opportunities for women can also contribute to empowerment by widening a woman’s personal choice and control over SRH outcomes. Incentives for schooling or delaying marriage: In many communities, the economic, cultural, and social environment does not offer adolescent girls viable alternatives to marriage. Once girls drop out of school, possibly because of the school’s poor quality or high cost, parents may find it difficult to identify any option other than marrying off their daughters. In such communities, providing better-quality and affordable primary and secondary education may be one of the best ways to delay marriage and childbearing. Programs to keep girls in school may also lead to tipping points in communities that make it easier for more and more girls to stay in school and thus delay marriage. A few interventions have also aimed to delay marriage by providing financial incentives conditional on not marrying early, with additional schooling often a benefit. Source: Botea et al. (2017). INTERVENTIONS TO DELAY MARRIAGE AND may also be needed so that girls can afford to remain in CHILDBEARING AND IMPROVE SEXUAL AND school, return to school if they dropped out, or expand their REPRODUCTIVE HEALTH livelihood opportunities and that of the household in which they reside if they cannot return to school. To facilitate While it is essential that countries promulgate appropriate selection of such interventions, this section summarizes policies including laws to facilitate the elimination of child international evidence related to three types of interventions marriage and reduce the risk of early childbearing, also for adolescent girls: (1) programs that provide girls with life required are specific strategies and interventions to empower skills and reproductive health knowledge; (2) programs that girls, including through appropriate life skills and knowledge expand girls’ economic opportunities; and (3) programs of sexual and reproductive health. Economic incentives designed to ensure that girls remain in school or that FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 60 enable them to return to school. Each type of program is club). Therefore, it may be best to combine safe spaces however based on a different theory of change (Box 20). with measures to improve livelihood opportunities or The summary of findings from the literature provided here offer incentives to remain in school, which are usually is based on a review of almost 40 such interventions with more effective in delaying marriage and childbearing. robust evaluations by Botea et al. (2017). mployment Opportunities. This category of programs • E To qualify for inclusion in the review, interventions had emphasizes both empowering girls, often through to (1) target girls aged 10–19, either exclusively or as part safe spaces, and providing livelihood opportunities. of a larger group; (2) equip girls with life skills and sexual These programs are particularly appropriate for girls and reproductive health (SRH) knowledge, economic who are not in school and would otherwise have no opportunities, or educational opportunities; (3) demonstrate income-generating skills. Two groups of interventions results in terms of improving the health of young women, are distinguished: (1) livelihood interventions and (2) especially SRH, or delaying marriage or childbearing; and interventions to improve financial literacy and access to (4) have been tested in a developing country, usually in financial services. Impacts in terms of delaying marriage Sub-Saharan Africa but also in other low-income settings and childbearing generally (though not always) tend such as Bangladesh or parts of India (see also Kalamar et to be larger than for the life skills/SRH knowledge al., 2016, for another review of the evidence focusing on programs alone. The programs have some success in programs aiming to delay marriage). Key findings for the increasing earnings, employment, or savings of girls. review are as follows: Several programs also result in increased use of modern contraceptives and improved SRH knowledge, which • Girls’ empowerment. Some interventions emphasize may delay childbearing. Some have also succeeded in empowerment of girls by providing them with life skills delaying age at marriage and reducing teen pregnancies. and SRH knowledge, among others through “safe space For example, the BRAC Uganda Empowerment and clubs”. The clubs convene girls under the guidance of a Livelihoods for Adolescent Girls had the following trusted adult mentor at a specific time and place. The impacts: (1) The likelihood of girls engaging in income- approach was pioneered by BRAC in South Asia and generating activities went up by 32 percent; (2) Among by the Population Council in Africa and Latin America. girls sexually active, self-reported routine condom use The clubs have proven effective when implemented went up by 50 percent; (3) Fertility rates went down well. By combining opportunities to socialize and have by 26 percent; and (4) Reporting of unwanted sex fun with access to mentors, the clubs are attractive to plunged by 76 percent. There were also reductions in girls and offer a platform for other services. They can teenage pregnancies and child marriage, and a shift in be used to provide SRH knowledge and improve life community gender dynamics (Bandiera et al. 2014; skills, including critical thinking and problem solving, Buehren et al. 2016). Overall, adding a livelihood negotiation, and communication (for example within dimension to life skills and SRH knowledge programs a household). The clubs can also boost self-awareness may help delay marriage and childbearing. The focus and self-esteem, and they may also facilitate delivery on economic opportunities may also help to ensure of cognitive skills such as basic literacy and numeracy, the regular participation of girls in the programs. or basic business skills. These programs have helped improve girls’ knowledge of SRH and behaviors. ncentives to Keep Girls in School. This type of • I Outcomes have included increases in girls undergoing interventions aim to keep girls in school, enable them HIV testing or counseling; greater use of modern to return if they have dropped out, or directly delay contraception and family planning; a reduction in the marriage. Quite a few of these interventions have had desire for FGM for daughters; a reduced risk of intimate positive impacts (Kalamar et al. 2016), including for partner violence when programs also reach out to men; delaying marriage or childbearing. While most of the higher self-esteem; and gains in specific skills. However, programs are designed to keep girls in school, some without additional supportive interventions, safe spaces enable girls who dropped out to return to school. Also alone may not delay marriage and childbearing (perhaps effective are conditional cash transfers to incentivize because that may not have been a primary goal for a girls’ schooling, promote health, and support families 61 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 during shocks. Across more than 20 impact evaluations in labor markets and occupational sex segregation, also in of cash transfers programs around Africa (from Burkina part due to social norms. While gender gaps in education Faso to Zimbabwe), all but one showed significantly have been reduced in recent decades, these other factors improved outcomes in education (Evans and Popova, leading to gender gaps in earnings remain prevalent. 2017). Such transfers have been introduced in more than two dozen low-income countries. Cash transfers A woman’s earnings depend on her human capital, work without conditions and income support programs experience (including firm-specific human capital), and have also had positive outcomes, such as reduced child the demand for her labor (which could be gender-biased). labor, expanding schooling, and enhancing childhood Women’s labor force participation is also affected by labor nutrition. While not all programs succeed everywhere, market, fiscal, and family policies as well as employer policies. the evidence is convincing that in comparison to the Across countries, additional factors include political ideology, other two types of programs reviewed above, those religion, and culture, stages in economic development, and focusing directly on schooling for girls, or in some cases industrial mix with different relative demands for female labor using financial incentives to delay marriage, may be in the private informal, private formal, and public sectors. more successful in delaying marriage and childbearing. Harnessing the returns from increased female labor force participation into activities generating more income means • Summary for targeted Interventions. The three levelling the playing field and addressing the reallocation of types of interventions are not an exhaustive list. To time between paid employment and other activities as well improve girls’ educational attainment, additional as persistent and pervasive gender differences in productivity interventions may also be needed. These interventions and earnings across sectors and jobs. were selected because there is evidence that they help improve SRH knowledge and delay child marriage Men’s and women’s jobs differ across sectors, occupations, and early childbearing. The programs are also not types of jobs, and firms as multiple factors lead to gender mutually exclusive; implemented together, they segregation with important implications for gender gaps in can complement each other. While some programs earnings (Goldstein et al., 2019). The World Development are better than others in delaying marriage and Report on gender (World Bank, 2012) posited that these childbearing, all three categories of programs have differences stemmed from three main factors: (i) unequal significant benefits and should be considered when distribution of time use and care responsibilities between formulating a strategy to invest in adolescent girls. men and women and between households and public/private service provision; (ii) unequal access to and control over productive assets (particularly land, credit, insurance and INVESTMENTS IN ADULT savings but also key skills); and (iii) market and institutional failures (access to information and networks, legal and fiscal WOMEN impediments, restrictive social norms). Policies related to these three main factors are outlined in Table 22 from Two main factors lead to gender inequality in human capital Wodon and de la Brière (2018). Differences in these areas wealth. First, men have higher labor force participation rates may affect all women (wage workers, farmers, and self- than women and they tend to work more hours in paid work. employed workers/entrepreneurs) and may reinforce each Women tend to work more hours than men, but a larger other and lead to productivity trap. This is costly not only for share of this effort is dedicated to unpaid work (household women, but also to their household, their community, and chores, care and work on household farms or in household society as estimates of losses in human capital wealth from enterprises), hence they tend to have lower earnings. gender inequality demonstrate. In addition, these differences Second, men tend to earn more than women per hour of represent a serious disincentive to investments in the women work. Despite progress towards equality in educational of tomorrow. attainment between boys and girls, part of the gender wage gap for adults is also due to differences in educational attainment due to deeply entrenched social norms. But other factors also play a role, including gender discrimination FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 62 Table 22: Examples of Interventions to Address Constraints on Women’s Paid Work Constraints/Type of work Wage employees Farmers Entrepreneurs/Self-employed Time use constraints Basic Infrastructure Access to basic infrastructure (cooking energy, water, electricity) Access to safe and affordable transportation Childcare Access to quality, affordable, publicly sponsored or employer-provided childcare Laws & technology Workplace flexibility in-cluding Time saving technology Time saving technology parental leave Access to productive assets Land - Joint titling - Skills Bundled training (technical and managerial) including socio-emotional skills (persis-tence), and asset-specific training Micro-credit (self-employed) In-kind and cash grants Credit (Small & Medium Alternative collateral: moveable assets, - Enterprises) payment history, psychometric tests Digital finance/savings and pay- Direct payments to accounts Individual saving accounts ments systems Other financial services Bundled financial services for risk management including insurance products for business and health needs among others Banking Mobile/web banking and simplification of KYC (Know your customer) rules Market and institutional failures Information Payment transparency Innovations in rural extension. Returns to traditionally Workers’ rights Engagement in value chains male-dominated sectors Social capital Expanding social networks: mentorship and sponsorship, role models Legal frameworks Removing gender differences in business, labor and family laws, enforcing existing laws supporting gender equality Taxation Individual income tax - Differential VAT Social norms Preventing and mitigating gender-based violence Building aspirations and self-confidence Source: Wodon and de la Brière (2018). ADDRESSING TIME USE CONSTRAINTS At home, access to basic infrastructure services (water, electricity, energy), as well as child and elderly care services Virtually every society has a division of labor based on gender can free women’s time. The role of infrastructure in freeing norms – typically with women specializing in reproductive productive time for women has long been recognized work and men in productive work. A recent review of time (Estache and Wodon, 2014). Rural electrification for ex- use surveys (Rubiano and Viollaz, 2018) shows significant ample contributes to women’s economic empowerment differences in the way women and men allocate their daily by increasing the length of the work day, reducing time for time between leisure, unpaid work (household chores and housework and fuel collection, and providing home-based child/elderly care) and market work. Women spend on business opportunities. This is especially the case when average 5 hours in unpaid work and 2.3 hours in market gender biases in the family and local economy are also work while men spend 5 hours in market work and 1.9 addressed, given interdependence in women and men’s time hours in unpaid work. Similar findings have been found in allocation decisions (van de Walle et al., 2013). The same previous work using time use data for sub-Saharan Africa is true for access to water. In Morocco, a project aimed to (Blackden and Wodon, 2006). Recogniz-ing, reducing, reduce the burden of girls traditionally involved in fetching and redistributing unpaid work would thus free a significant water to improve their school attendance. In the project’s amount of time for women to participate in market work. areas, girls’ school attendance increased by 20 percent in four years (World Bank, 2003). 63 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 For child care, Mateo Díaz and Rodriguez-Chamussy At work itself, parental leave, flexible schedules and mode (2016) and Reimo et al. (2017) review some of the evidence of work, and legislation on retirement ages can all make on the impact of providing child care and early education a difference. As noted under the Women, Business, and services. They find that the provision of these services in the Law indicators (World Bank, 2018a), policies that help Latin America increases female employment by 10 to 30 workers balance paid work and family responsibilities include percent. Public provision of affordable and quality child care parental leave (which can be taken by either parent). The is especially important for women’s labor force participation, opportunity for workers to return to their pre-leave work but there is also a role for employer-supported childcare or employer increases labor force participation and helps provided that the costs of provision do not affect negatively workers retain firm-specific human capital. The so-called women’s employment opportunities. Partnerships and father’s quota in Nordic countries provides an incentive for collaboration between the public and private sectors and civil fathers to take their leave or lose it, and to share in the child society organizations can help in this regard (International care. Workplace flexibility, either through part-time work, Finance Corporation, 2017). flexible hours, compressed schedules (“flextime”) or through tele-commuting/home-based work also help workers balance Interventions that make it easier for women to get to work the demands of paid work and family responsibility. For can also be beneficial. While women tend to be responsible both leave and flexible work arrangements, it is important for a disproportionate share of their household’s transport to ensure the participation of both women and men and needs, they tend to have more limited choices for mobility, to calibrate the generosity of leave/flexibility to minimize in terms of mode and distance. A combination of inadequate potential downsides for women in terms of slower career mobility choices (including slower travel options and off- progression or occupational segregation. peak travel when frequencies are low) with more complex travel needs leads to slower travel speed and thereby smaller In many developing countries, flexibility is only available travel distances for women, resulting in limited access to through the informal sector and women tend to be economic opportunities and essential services. Studies in concentrated in those jobs, which are often the only jobs both developing and developed countries show a negative enabling them – at a high cost in foregone income – to correlation between commuting time and women’s balance income-generation and family responsibilities. In participation in the labor force (see for example Black the formal sector, ensuring that women and men can work et al., 2014 for the United States). An increase of one until the same (retirement) age is particularly beneficial for minute in commuting time in metropolitan areas is asso- women who tend to have patchier market work histories ciated with a 0.3 percentage point decline in women’s and shorter employment spells than men, which means labor force participation. that their retirement income is lower. Earlier retirement ages for women can cast an additional penalty as do long Security concerns also affect women’s travel. Policy and vesting periods. program interventions to enhance security through physical infrastructure investments (lighting in stations, design of FACILITATING ACCESS TO PRODUCTIVE ASSETS buses and trains, cameras and alarm systems), developing and testing new security reporting and monitoring tools (with Especially in low income countries, women’s employment mobile technology and witness bystander interventions), is informal, with self-employment being the most common and information measures to foster behavior change (for type of work, and a large share of women still work in the example through education campaigns and increased agricultural sector. Women farmers and entrepreneurs law enforcement) are all positive measures. On-going consistently produce less and generate less income than experiments in several countries (such as Brazil and Pakistan) their male counterparts (World Bank and ONE, 2014, as well as the development of alternative transportation Campos and Gassier, 2017). This reflects both unequal modes (ride-sharing) should shed light on what works and access to inputs and lower returns to these inputs. For what are the constraints. Ride hailing platforms like Didi and female farmers, access to, and control over good quality land Uber also provide opportunities for women’s employment, are especially critical for agricultural investment and rural in terms of flexibility, mobility and personal safety, but household welfare. Yet statutory and customary land tenure discrimination remains (see Accenture and International systems often disadvantage rural women, who are less likely Finance Corporation, 2018). to control land than rural men. Women’s tenure insecurity FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 64 reduces their investments in their land, thus undermining positive outcomes for women across countries, ages and their productivity. Strengthening women’s land rights is key activities. Women still have an unmet demand for those to addressing the issues undermining their productivity. and for entrepreneurs, they help protect specific business For example, Rwanda is making joint ownership the default funds. However, very poor women might be too poor to option in its land titling program, which is associated with save without additional support (Buvinic and O’Donnell, greater productivity (Ali et al., 2014). 2016). Bundled services including a relatively large (in-kind) capital transfer, asset-specific training, technical assistance, Also important is the acquisition of soft technical and a stipend for one to two years, and health information/ managerial skills. For farmers, factors relating to land beyond insurance and life skills training have shown that they can access itself help explain the gender gap. One of these help push very poor women out of poverty traps with positive challenges relates to land size. In Ethiopia and Tanzania, economic outcomes and increased savings. One example is women receive lower returns than men to an extra hectare the BRAC Ultra-poor Graduation program (Banerjee et al., of land. This could be due to lower quality of the land, but it 2015). More generally, innovative approaches such as the could also be due to women’s relative difficulty in managing/ Women Entrepreneurship Finance Initiative can advance hiring farm labor or the application of other inputs across women’s entrepreneurship by increasing access to the larger tracts of land. finance, markets, technology, and networks necessary to start and grow a business. Financial exclusion also remains a barrier for many women farmers and entrepreneurs. Micro-credit by itself is not Acquiring managerial and psychosocial skills is important sufficient for a transformative impact. As women are less for all women, but especially farmers and entrepreneurs. likely to hold titles to their productive assets, they face Women farmers may face additional hurdles than their higher hurdles to secure loans for lack of suitable collateral. male counter-parts in hiring and supervising labor, or in Promising initiatives include the promotion of alternative using inputs such as fertilizers and pesticides correctly. For collateral through moveable asset registries, the use of entrepreneurs, recent evidence points to the importance of payment histories for services such as cell phones, and training combining soft skills (especially for young female psychometric testing to assess lenders’ risk (Buehren et entrepreneurs or in fragile and conflict-affected countries) al., forthcoming). In addition, as women face difficulties to and managerial skills together with grants. This seems to be keep business/farm and household finances separate, health more effective than just providing capital and technical skills. insurance products help to avoid depleting working capital High-quality business management training of significant when responding to family health needs (Campos and duration (6 to 12 weeks) can have positive outcomes for Gassier, 2017). poor female entrepreneurs, with improvements in business practices, leading to increased sales, profits, and survival Given their time constraints, women are also more likely to rates. Demand-driven job services (plus vouchers/subsidies prefer bundled products including insurance and financial to employers and child care/transport stipends for trainees) services (International Finance Corporation et al., 2015). increase economic opportunities of young women, especially Secure (private) individual savings accounts, including in if they tackle discrimination and other barriers in the training the form of commitment accounts and liquid savings, have and work environments. 65 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 SOLVING MARKET AND INSTITUTIONAL Finally, it includes policies targeted at advancing women to FAILURES top positions (such as managerial and board diversity targets). These various laws are expected to positively influence Market failures refer broadly to situations in which markets women’s labor force participation decisions and the type of do not lead to optimal resource allocations. Institutional employment they hold. failures refer to institutions not functioning properly and therefore not achieving their missions. Both types of failures The structure of income tax policy creates a “second earner” can be pervasive with potentially serious implications for penalty if the family is considered the unit of taxation or gender inequality, as a few examples help illustrate. if dependent credits or allowances are eliminated when a spouse enters the labor market (Grown and Valodia, 2010). Access to information to address occupational segregation On the other hand, earned income tax credits provide an and pay gaps can help improve gender equality. Women income subsidy for low-earner families and encourages farmers tend to have less access to information about women in those families to enter the labor force. farming technology and methods as extension services are rarely designed to take their specificities (in terms of Ensuring safety and preventing gender-based violence time availability, types of crops, or access to inputs) into at home, at work, and in public spaces is also essential. account. Enabling women to shift to high value commercial Appropriate laws are still lacking in many countries (Tavares crops shows promise in Africa. Access to information about and Wodon, 2018). There are also potential links between potential returns for women in male-dominated fields can work and gender-based violence. Enhancing women’s labor help female entrepreneurs cross over and shift sectors force participation can promote their empowerment and (Campos et al., 2015), provided they also get support well-being, as well as the welfare of their children (since from male mentors in the field and can withstand sexual mothers often control more spending related to children). harassment and barriers to access credit. However, the empirical relationship between women’s employment and domestic violence is less clear-cut, Access to social capital (networks, role models, and depending on whether husbands perceive their roles as mentorship) also matters. Business associations, networks, breadwinners undermined (especially in case of unem- mentors, and role models hold promise for both women ployment or when the deviance from gender norms is too entrepreneurs and farmers as they complement and strong) and male co-workers perceive potential displacement reinforce the effects of interventions such as business from female employees or female colleagues as “unsuitable”. training, cash transfers and agricultural extension. The The evidence is mixed: non-significant relationship in Jordan complementarity seemingly arises from acquiring both (Lenze and Klasen, 2017), positive in India (Amaral et al., information and social support, although we don’t know 2015 with increases in kidnappings, sexual harassment, whether these measures are similar or work differently. domestic violence and decreases in dowry deaths; Paul, Self-help groups in particular foster increased solidarity 2016), and negative in the United States (Aizer, 2010 with between peers, independent financial decision-making, and the closing in the gender wage gap through exogenous greater respect for the women within their households and changes in labor demand in female-dominated industries). communities (Brody et al., 2015) The direct and indirect costs of gender-based violence to women and their children’s productivity could amount to Another important area for reform is legal and fiscal several percent of global GDP (Hoeffler and Fearon, 2014). frameworks. This includes labor market policies aimed at More rigorous evaluations of the impacts of interventions ensuring equal opportunities in the labor market such as anti- for prevention, deterrence, and mitigation are needed in this discrimination laws and the elimination of laws restricting area to find the approaches that will work best. women’s labor force participation in some sectors. It also includes laws about access to capital and justice, as noted in Women, Business and the Law reports (World Bank, 2018a). FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 66 BOX 21: ENDING INTIMATE PARTNER VIOLENCE: LAWS AND SOCIAL NORMS This study does not provide new empirical results related to violence against women, nor does it provide a systematic review of the evidence on programs to reduce such violence. Still, since violence against women is one of the manifestations of gender inequality, some pointers are worth providing. Are countries protecting women legally against abuse? Analysis based on data from the Women, Business and the Law report conducted for 141 countries suggests that while most countries have laws on domestic violence and sexual harassment, there is still a large gap in overall protection and comprehensive laws (Tavares and Wodon, 2018). One in four countries have yet to adopt any legislation on domestic violence, while for sexual harassment that is the case in over one out of eight countries. Even where laws exist, this does not mean that women are well protected. Typically, existing laws are not comprehensive enough to account for multiple forms of violence (not only physical, but also sexual, economic, and emotional). Furthermore, when laws against sexual harassment exist, they do not cover many settings in which harassment may take place. Between 2013 to 2017, the share of countries with laws on domestic violence increased from 70.9 percent to 75.9 percent, but legal protection remains weak for sexual, economic, and emotional domestic violence. While laws may protect married individuals, they typically do not protect unmarried intimate partners. The share of countries with laws on sexual harassment increased from 83.7 percent to 86.5 percent. But again, not all risks are considered. For example, one in five countries do not have specific laws against sexual harassment in employment, and six in ten countries do not have laws against sexual harassment in education. For sexual harassment in public spaces, only one in five countries has laws. When countries’ populations are accounted for, this translates into large numbers of women unprotected by the law. For example, more than one billion women lack legal protection against sexual violence by an intimate partner or family member, while legal protection from domestic economic violence, which may result in a woman being deprived of the economic means to leave an abusive relationship, is not available for close to 1.4 billion women, with little progress for both measures over time. Estimates of the number of women lacking legal protection against sexual harassment in employment, education, and public places are at 359 million, 1.5 billion, and 2.2 billion respectively. Beyond appropriate laws, interventions to end violence against women must tackle social norms. An emerging evidence base is becoming available6 on what works especially in middle and high-income countries (WHO 2010; Paluck and Ball 2010; Jewkes 2014; Bourey et al., 2015, Tappis et al. 2016), but a few programs in low income countries related to HIV/AIDS and Village Savings and Loan Associations (VSLA) have been evaluated. These programs tend to work through small participatory workshops to challenge existing beliefs, or larger community campaigns to reinforce these efforts using street theatre, discussion groups, cultivation of change agents, and print materials. Some of these programs have resulted in small changes in attitudes/beliefs and reductions in IPV (Heise, 2011; Arango et al., 2014). 6 The authors are grateful to Muthoni Ngatia and Diana Arango for summarizing some of the evidence. 67 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Stepping Stones, a program implemented in more than 40 countries, uses participatory learning to build knowledge, risk awareness, and communication and relationship skills on gender, violence and HIV (World Bank, 2014). Evaluations suggest that men’s self-reported perpetration of physical and/or sexual IPV was lower after participating in the program, although effects waned after two years (Jewkes et al., 2008). In addition, there was no difference in women’s reports of IPV victimization. Self-reporting may have led to measurement issues, for example when awareness of the fact that IPV is unacceptable reduces reporting of IPV perpetration, or when a reduction in the stigma associated with being a victim of IPV increases women’s willingness to report having experienced IPV. The program also has had unintended effects such as more transactional sex with a casual partner and unwanted pregnancies. One explanation could be that behavior change and social norm interventions may inadvertently encourage those farther from the norm to adjust their behavior to match it (Paluck and Ball, 2010). In Rwanda, a VSLA program focusing on women’s economic empowerment engaged men and contributed to more equitable household decision making, increased couples’ communication, and decreased couple conflict, but did not influence IPV rates (Slegh et al., 2013). Studies of other programs including Program H in Brazil and India found some positive impacts, but also no impacts in some cases (Pulerwitz et al., 2010; Barker et al., 2007; Verma et al., 2008). In Cote d’Ivoire, South Africa and Burundi, discussion group-based IPV prevention programs delivered alongside microfinance had mixed results, with few or no gains in Burundi and Cote d’Ivoire (Ferrari and Iyengar, 2010; Hossain et al. 2014) and large gains in South Africa (Kim et al. 2007, Iyengar and Ferrari 2011). These studies evaluated the impact of single-sex discussion groups, while there is an emerging consensus of the need to involve both men and women together to challenge prevailing gender norms. Social norm theory suggests that community and society-wide factors and expectations about others’ beliefs and behaviors may be a binding constraint to reducing IPV, yet few discussion group interventions included such elements. Another example of community-wide interventions is SASA! in Uganda. SASA! means “Now!” in Kiswahili. The program employs multiple strategies to build a critical mass of engaged community members, leaders, and institutions, including local activism, media and advocacy, communication materials, and training. The program’s community engagement and mobilization involves four phases: Start, Awareness, Support, and Action. The content evolves with each phase, with power as a central theme. Results from a randomized controlled trial suggest positive effects after three years, with (i) a reduction in levels of IPV against women of 52 percent; (ii) an increase of 28 percent in the share of women and men finding it is acceptable for women to refuse sex; and (iii) an increase of 50 percent in the share of men and women believing that physical violence against a partner is unacceptable. Finally, beyond programs targeting social norms, it should be noted that conditional and unconditional cash transfers may also have an impact on IPV. Evaluations suggest reductions in reported IPV (Bobonis et al., 2013; Hidrobo and Fernald, 2013; Green et al., 2015; Haushofer and Shapiro, 2016), but in Ecuador a cash transfer programs appears to have led to an increase in IPV (Buller et al., 2016). FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 68 TARGETING Ideally, all girls and women should benefit from interventions Figure 14: Hight Prevalence Areas for Child Marriage aimed at achieving gender equality. At the same time, in among Girls Aged 18-22 in Nigeria practice governments face difficult budget constraints. Given such constraints, it is often better to concentrate available resources in high prevalence areas for gender inequality or a particular manifestation of it such as child marriage (such areas are sometimes called “hotspots”) than to spread resources too widely. The rationale for concentrating resources in a limited number of geographic areas is that this may have a better effect to reach tipping points that can lead to changes in social norms at the community level. To consider an extreme example, consider a government that may have a budget to provide scholarships to 100,000 girls to keep them in schools, which would also help in reducing the prevalence child marriage and early childbearing. One approach would be to sprinkle the resources to a large number of geographic areas, but in that case, while the girls receiving the scholarships would benefit, other girls would probably not benefit. The alternative approach would be to Source: Authors’ estimation using DHS data. target resources to specific areas with low rates of secondary school enrollment and concurrently a high prevalence of child marriage or early childbearing. The girls who would adjacent countries, showing that high prevalence areas often receive the benefits would be better off, but it could be that are cross-border. For example, Hausa communities in the by having many girls in the targeted communities going to North East of Nigeria have high rates of child marriage, but school longer, the dynamics in the communities would also so do Hausa communities in Niger. In such cases, there may change. For example, if many girls remain in school, the be benefits from regional projects, as illustrated at the World pressure to marry girls early would be reduced for parents as Bank by the Sahel Women Empowerment and Demographic the local “marriage market” would have been altered. In other Dividend project. words, concentrating resources in targeted areas of needs can lead to tipping points at the level of communities that may Even though common sense suggests that targeting high then generate benefits for all girls in those communities. prevalence areas may have benefits to reach tipping points, changing social norms is not easy, and our understanding Exploiting the potential benefits of tipping points requires of which approaches may work best is still limited (Marcus, careful planning and attention to local realities. But a first 2018.). It should also be clear that whether strategies basic step is to identify where high prevalence areas are. This targeting social norms are likely to work may depend on step, which would be part of a basic diagnostic of gender local conditions since the role of social norms, for example in inequality in a country as discussed in Box 22, can often be leading to child marriage, is not the same in all communities. undertaken using existing data. As an illustration, Figure 14 For example, research suggests that in Brazil, social norms visualizes the prevalence of child marriage in Nigeria and play an important role in leading to child marriages (Taylor et neighboring countries. Some areas of the country clearly al., 2019), but this is not always the case in Malawi (Steinhaus suffer from higher prevalence rates and lower completion et al., 2019). rates than others. In Figure 14, the analysis for Nigeria is conducted at the level of states, but this type of analysis can be replicated at a lower level of disaggregation with census data. On purpose, the same statistics are provided for 69 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 BOX 22: DIAGNOSTICS AND STRATEGIES FOR GENDER EQUALITY Multiple interventions are required to achieve gender equality. These interventions involve different agencies and Ministries. Effective coordination mechanisms are therefore required. Consider simply the need to ensure that girls remain in school, which can also help in delaying marriage and childbearing. Within the Ministry of Education, several departments are likely to be in charge of different types of programs, so that coordination is required within the Ministry itself. Many countries have several Ministries dealing with education, by level. In addition, other Ministries and agencies have a role to play. Ministries of labor and social protection tend to be in charge of some of the second chance programs for children who dropped out, and they have the main responsibility for the implementation of assistance programs and cash transfers. The provision of basic infrastructure in schools is typically the responsibility of a Ministry of Public Works or its equivalent, but the provision of water, sanitation, and electricity tends to be managed by separate Ministries, one for water and sanitation, and the other for electricity. Health policies, both in early childhood and for adolescent girls in the case of SRH services are managed by Ministries of Health. Apart from these various and possibly other Ministries, such as Ministries of Women’s Affairs and agencies for specific vulnerable groups, the private sector is also a key player for gender equality through both faith-inspired and private secular education providers. Overall, the multitude of organizations with a role to play calls for strategic vision and coordination mechanisms. In practice, the process followed to develop strategies for gender equality is likely to change depending on the country, but core steps can be recommended. Strategies for gender equality could be relatively simple or more sophisticated, depending on country capacity and needs. The strategies could stand by themselves, or they could be embedded in broader development strategies. How the strategies will be prepared and how often they will be updated and revised will also depend on country circumstances. But in general, each strategy should (i) assess major areas where gender inequality is prevalent as well as the main determinants of a lack of gender equality; (ii) set “SMART” targets for gender equality; (iii) prioritize public actions for gender equality; (iv) establish systematic monitoring of trends in gender equality indicators as well as evaluate the impact of government programs and policies for gender equality; and finally (v) ensure that the strategy is developed, implemented, and monitored in a participatory way. Importantly, the gender equality strategy should be grounded in an understanding of the extent, nature, and various challenges to gender equality. Strategies for gender equality should therefore be based on a diagnostic of gender equality issues. This requires appropriate data as well as analysis, and a process to achieve consensus on the findings from the diagnostics. Once a gender inequality diagnostic is available, the next step consists for Ministries and other stakeholders to set targets for gender equality. As noted by Christiaensen et al. (2002), targets may serve as an incentive mechanism that affects behavior in at least three ways: (1) Resource mobilization (targets help in mobilizing human and financial resources in order to achieve certain goals; (2) Resource allocation and consensus building (the process of setting targets helps in revealing priorities and allocating resources); and (3) Accountability (targets provide benchmarks against which the performance of Ministries and other agencies can be judged). Finally, target setting is linked to the government’s budgetary process and fiscal constraints. Attaining targets must not only be technically feasible, but also fiscally feasible, and thus must take into account the budget, so that cost estimates must be prepared. The fiscal feasibility of gender equality targets depends on the government’s capacity for both increasing public spending and enhancing the efficiency of that spending. Again, while this discussion is fairly general and could be applied to many other issues apart from that of gender equality, it is hoped that it provides some pointers on processes to follow to prepare such strategies. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 70 CONCLUSION But they demonstrate that the potential costs of gender inequality are high not only for girls and women, but also for Gender inequality remains pervasive. Women are less their communities and societies. likely than men to join the labor force and to work for pay. When they do, they are more likely to work part-time, in The largest cost of gender inequality by far in Table 23 relates the informal sector, or in occupations that have lower pay. to earnings, simply because gaps in earnings are observed These disadvantages translate into substantial gender gaps in virtually all countries, including upper middle and high in earnings, which in turn decrease women’s bargaining income countries that concentrate the bulk of the world’s power and voice. Many girls are married or have children wealth. In comparison to losses from gender inequality in before the age of 18, before they may be physically and earnings at US$172 trillion globally, losses associated with emotionally ready to become wives and mothers. In part due high population growth (welfare losses and budget costs) and to child marriage, girls’ average educational attainment in stunting for young children are small, essentially because low income countries remains lower than boys, and in most those issues are salient in low and lower-middle income developing countries adult women are less literate than men. countries, and these countries have low levels of wealth. But Women and girls also face risks of gender-based violence in the costs are not small in comparison, for example, to total their homes, at work, and in public spaces. Their voice and annual net official development assistance (ODA) provided agency is often lower than that of men, whether this is within to developing countries. ODA was estimated at US$148 the household, at work, or in national institutions. This also billion in 2016. The losses from gender inequality through affects their children. Children of young and poorly educated population growth and stunting are therefore far from mothers often face higher risks of dying by age five, being negligible in comparison to ODA in the developing world, malnourished, and doing poorly in school. Fundamentally, representing a substantial share of the total losses from gender inequality disempowers women and girls in ways that gender inequality in many of those countries. deprive them of their basic human rights. Apart from measuring selected impacts and costs of The primary objective of this study was to estimate selected gender inequality, the study provided guidance on the impacts and costs of gender inequality. Impacts were investments that could help reduce the cost of gender documented in this study in five main areas: (1) earnings inequality. This guidance is not meant to be comprehensive, and standards of living; (2) educational attainment, child nor exhaustive. Since gender inequality affects girls and marriage and early childbearing; (3) fertility and population women in virtually all aspects of their life, a wide range of growth; (4) health, nutrition, well-being, and violence; and interventions to reduce gender inequality and mitigate its (5) agency, decision-making, and social capital. The potential impacts should be implemented. But to keep the analysis economic costs of gender inequality in terms of lost wealth manageable, the focus was principally on three types of for countries are very large. Table 23 provides a summary investments corresponding to the three types of economic of the estimated potential impacts of gender inequality by costs documented above. Along a simple life cycle model, domain, together with an indication of country coverage the study considers: (1) Investments in early childhood for the estimations. This is done by distinguishing estimates development to reduce the impact of gender inequality based on global data from those based on a core set of up on young children; (2) Investments in adolescent girls to two dozen developing countries (DCs). Potential impacts to delay marriage and childbearing while also improving are summarized by showing gains from achieving gender education opportunities, which would help reduce population equality in comparison to current conditions. It should be growth; and (3) Investments in adult women to improve emphasized that what is measured when using regression employment and earnings opportunities and increase analysis is associations, not necessarily causal impacts. In human capital wealth. Finally, the study suggested to target addition, for some indicators, the data pertain to reported high prevalence areas for gender inequality or some of its behaviors and perceptions, thereby making interpretation particular manifestations through interventions in order to more tentative. Table 23 also summarizes selected monetary reach tipping points on social norms in communities, and to costs from gender inequality. The estimates are only orders prepare diagnostics and strategies to end gender inequality. of magnitude since they depend on models and assumptions. 71 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Table 23: Selected Potential Impacts and Costs/Benefits from Gender Equality Domain Coverage Potential Impacts Global Increase in women’s human capital wealth of more than half Gain in women’s labor force participation and full-time work of 20 percentage Earnings and standards of living Global points Global Substantial reduction in poverty from higher earnings and lower fertility Global Elimination of child marriage Educational attainment, child marriage and DCs Reduction in early childbearing by at least three fourths early childbearing Global Gains in educational attainment for girls in low-income countries DCs Reduction in total fertility by 13 percent Fertility and population growth DCs Increase in contraceptive use by 12 percent Global Reduction in population growth rate by 0.26 percentage point in 16 countries Global Improvement in women’s health and psychological well-being DCs Reduction in under-five mortality rate by 5 percent Health, nutrition, well-being, and violence DCs Reduction in under-five stunting rate by 7 percent DCs Increase in women’s knowledge of HIV/AIDS and reduction in violence DCs Increase in women’s decision-making by 45 percent Global Improvement in women’s ability to assess quality of basic services Agency, decision-making, and social capital DCs Increase in likelihood of birth registration by 5 percent Global Increase in women’ reported ability to engage in altruistic behaviors Global Increase in women’s reported ability to rely on friends when in need Global Loss in HC wealth from earnings inequality of US$172 trillion DCs Loss in HC wealth from stunting of US$71 billion in 17 countries Potential economic costs Loss in wealth per capita equivalent to US$80 billion in first year in 16 countries DCs due to high population growth (with cumulative effects over time) DCs Budget costs in education of up to US$27 billion by 2030 in 16 countries Source: Wodon (2018). Note: DCs = Developing countries. Investments to end gender inequality should not be women. While further work is needed to identify the best based solely on economic considerations. The primary policy options at the country level to improve opportunities motivation for ending gender inequality should be to for girls and women, lessons can be learned from address the substantial risks and suffering affecting girls international experience. Ending gender inequality is not only and women throughout their life. However, demonstrating the right thing to do from a moral and ethical standpoint, it is the magnitude of the impacts and costs of gender equality also a smart investment. provides additional justification for investments in girls and FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 72 APPENDIX 1: DATA AND countries were selected to respond to demand for analytical work on those countries to inform on-going operational METHODOLOGY work at the World Bank. Third, an effort was made to have representations from several regions, and especially from DATA SOURCES sub-Saharan Africa and South Asia where gender inequality tends to be more pronounced in specific areas considered for Three main types of surveys are used for the quantitative this study. analysis. Estimates of gender inequality in earnings are based on nationally representative household and labor The third main source of data is the Gallup World Poll which force surveys from the World Bank’s Global Labor Database covers more than 150 countries. The Poll typically surveys (GLD), previously referred to as the I2D2 database. The 1,000 individuals in each country, using a standard set analysis builds on previous work at the World Bank to of core questions that has been translated into the major measure human capital wealth for 141 countries as part of languages of the respective country. Because the samples at an analysis of the changing wealth of nations (Hamilton et the country level are relatively small, the regression analysis al., 2018). Human capital wealth is defined as the present for this study is conducted with the pooled dataset. While value of the future incomes of the labor force, and it can survey data or specific questions are not available for all years be compared to other sources of wealth such as natural or for all countries, the pooled data set used for the analysis is produced capital. The estimates of human capital wealth have large, with more than 200,000 observations. A total of 114 been disaggregated by gender. When using surveys in the countries are included in the final sample: 10 from East Asia GLD database and estimating human capital wealth, analysis and the Pacific, 40 from Europe and Central Asia, 21 from is conducted for each country separately. Latin America and the Caribbean, four from the Middle East and North Africa, one for North America, seven from South The second main source of data for the estimations is the Asia, and 31 from sub-Saharan Africa. While some regions Demographic and Health Surveys (DHS). Building on have better representation than others, most of the world’s previous work on the economic impacts of child marriage population is included because large counties in terms of and the cost of not educating girls, detailed analysis of population are covered. the correlates of selected development outcomes was implemented with the most recent DHS for 19 developing In addition to relying on surveys, the team conducted countries: Bangladesh, Burkina Faso, Democratic Republic qualitative work on the constraints faced by girls to continue of Congo, Dominican Republic, Egypt, Ethiopia, Guinea, their education, with a focus on sub-Saharan Africa were India, Malawi, Mali, Mozambique, Nepal, Niger, Nigeria, these constraints are most severe. Qualitative data were Pakistan, Republic of Congo, Tanzania, Uganda, and obtained for countries in West Africa, Central Africa, and Zambia. The sample is titled towards sub-Saharan Africa and East Africa. While these data are not used systematically South Asia; Latin America and the Caribbean and the for this note, excerpts from respondents in focus groups or Middle East and North Africa are each represented by in-depth qualitative interviews have been used in background one country. As with surveys from the GLD database, work to illustrate findings that emerge from the quantitative regression analysis is conducted for each country analysis. separately when using DHS data. METHODOLOGY The rationale for the choice of the 19 countries was based on three main considerations. First, many of the countries The study aims to estimate the potential impacts of gender have low levels of educational attainment for girls, high rates inequality on development outcomes and the economic of child marriage and early childbearing, as well as relatively costs associated with some of these potential impacts. As high fertility rates and high rates of under-five mortality and defined in World Bank (2012, 2016), gender refers to the stunting. These were some of the core outcomes for which social, behavioral, and cultural attributes, expectations, and estimations of the potential impact of gender inequality norms associated with being male or female. Gender equality were carried, and it therefore made sense to select countries (or inequality) refers to how these factors determine the way for the estimations where achieving gender inequality was in which women and men relate to each other and to the likely to make a substantial difference. Second, several of the resulting differences in power between them. This definition 73 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 however needs to be operationalized in order to measure the on the risk of being a victim of intimate partner violence is potential impacts of gender inequality on various outcomes. available for women only. As shown in Table 24, regression analysis is used to assess the potential impacts of gender In this study, for some indicators, simple statistics are used inequality on most indicators related to (1) fertility and to measure the potential impacts of gender inequality. This population growth; (2) health, nutrition, well-being, and is done when gender differences in outcomes are prima violence, and (3) agency, decision-making, and social capital. facie evidence of gender inequality, as is the case for most When using regression analysis, simulations rely on proxies indicators related to earnings and labor force participation, for how achieving gender inequality (for example by ending as well as educational attainment, child marriage, and early child marriage and early childbearing, or by ensuring that childbearing. For other indicators, regression analysis is used girls have the same educational attainment as boys) would instead. This is done when gender inequality may affect affect the outcomes of interest. Those proxies are of course outcomes for both genders even if differences in outcomes imperfect and they may not capture the full effect of gender by gender are small. For example, gender inequality affects inequality, but the simulation approach helps in suggesting an the risk of under-5 stunting or under-5 mortality for boys order of magnitude for the potential impacts. and girls alike. Regression analysis is also used when data are only available for women on a specific outcome. For The term ‘potential impact’ is used for simplicity and for the example, in most Demographic and Health Surveys, data study to be readable to non-technical audiences, but one Table 24: Methodology for Measuring the Potential Impacts of Gender Equality on Outcomes Domain Measurement of the Potential Impacts of Gender Inequality Statistical Differences in Regression Analysis Using Proxies for Outcomes by Gender Gender Inequality Earnings and standards of living Education, child marriage, and early childbearing Fertility and population growth Health, nutrition, well-being, and violence Agency, decision-making, and social capital Source: Authors. must be careful about not necessarily inferring causality. To reduce the risk of bias in coefficient estimates, different Estimates of potential impacts are obtained through specifications for the regressions have been used, and we regression analysis based on the identification of indicators typically report results obtained with the largest number of which proxy for gender inequality and controlling for controls. Based on measures of potential impacts, potential other variables that may affect the outcomes of interest. costs associated with selected potential impacts are then Different types of regression techniques are used depending computed. Note that we provide potential cost estimates on the outcomes of interest. What is measured are thus only for a subset of potential impacts. These potential costs statistical associations, and not necessarily impacts as rely on additional assumptions and are thus also tentative. could be observed with randomized control trials or quasi- The estimated costs presented therefore capture only part experimental methods. Said differently, the regression of the total costs. More details on the data sources and analysis provides estimates of potential impacts, but there methodologies used for estimations and how they relate to is always a risk of bias (and in some cases upward bias) in the key findings are available from the authors. measures of the potential impacts being reported due, for example, to the risk of omitted variables bias. When considering economic costs, the analysis focuses on (i) earnings; (ii) “welfare” or standards of living as they relate to FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 74 population growth; (iii) budget savings from lower population services to populations currently not served. While other growth; and (iv) other benefits such as those related to types of costs could be considered, these are some of individual feelings and perceptions for which no economic the largest economic costs of gender inequality, and the costs are computed. The focus on these four types of costs framework allows for avoiding double counting costs (there is is driven in part by data availability and the ability to estimate no overlap between cost categories). approximate costs with a reasonable degree of confidence. The basic idea is to consider human capital wealth per capita Note that considering separately impacts on the nominator as the main indicator of interest. Since human capital wealth and denominator of human capital wealth as described in is based on the earnings of the labor force, gains in human Table 25 does not imply that interaction effects between capital wealth per capita arise when earnings for the labor impacts on various outcomes are not considered. For force increase (larger nominator), or when the size of the example, ending child marriage as one of the benefits of population among which human capital wealth is shared achieving gender inequality has impacts on both earnings decreases (smaller denominator). For earnings, the focus and population growth. The pathways through which these is on (1) potential gains from ending gender gaps in current impacts may be observed are complex, but both impacts are earnings between adult men and women, and (2) potential captured when simulating gender inequality in earnings and gains from lower under-5 stunting rates that lead to higher measuring the potential reduction in population growth. Said productivity in adulthood. When considering population differently, while the study does not consider separately all growth, the focus is on the reduction in the size of the pathways through which gender inequality affects outcomes population in the future that could result from achieving and all the interactions between outcomes, it does capture gender inequality today. Finally, as an illustration of other overall effects in reduced form, either through simple potential effects, the impact of gender inequality on budget statistical comparisons of outcomes by gender or through spending for education is also considered to show that lower regression analysis when simple comparisons of outcomes by population growth could help in achieving savings in the gender do not adequately capture effects. cost of basic service delivery, so that those savings could be reinvested in higher quality services or an expansion in Table 25: Methodology for Estimating Economic Costs Associated with Potential Impacts Domain Measurement of Economic Costs Associated with Impacts Nominator Denominator (Incomes or Expenditure) (Population Size) Higher earnings via gender equality in earnings Higher earnings via reduced under-5 stunting rates Higher welfare via reduced population growth Budget savings via reduced population growth Other benefits without estimates in monetary terms N/A N/A Source: Authors. 75 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 APPENDIX 2: HUMAN The household surveys used for the computation of the earnings profiles—as well as the probability of working—are CAPITAL WEALTH nationally representative. The surveys are in most cases of good quality, but they may still generate estimates ESTIMATES that are not consistent with either the system of national accounts or population data for the countries. Therefore, Human capital wealth is defined as the discounted (present) two adjustments are made. First, to ensure consistency value of future earnings for a country’s labor force. In of the earnings profiles from the surveys with published practice, we estimate how likely it is that various types of data from national accounts, earnings estimates from the individuals will be working, and how much they will earn surveys are adjusted to reflect the share of labor earnings when working. By “various types” of individuals, we mean (including both the employed and the self-employed) in individuals categorized by age, sex, and level of education. GDP as available in the Penn World Tables. Second and Essentially, we use household surveys to construct a dataset separately, the estimations also rely on two variables obtained that captures (1) the probability that individuals are working from data compiled by the United Nations Population depending on their age, sex, and years of education; and Division: (1) population data by age and sex (so that the (2) their likely earnings when working, again, by age, sex data in the household surveys can be better calibrated); and years of schooling. This is done separately for men and and (2) mortality rates by age and gender (so that the women so as to produce estimates of human capital wealth expected years of work can be adjusted, accounting for the by gender. Typically, women’s estimated lifetime earnings are fact that some workers will die before age 65). Again, we significantly lower than for men. adjust data from the surveys to population estimates from the United Nations to ensure that estimates are adequate. Estimates of the likelihood of working for individuals are For individuals in the 15-to-24 age group, the probability of based on observed values in household and labor force remaining in school is also considered. surveys. Estimates of expected earnings are based on Mincerian wage regressions. The regressions are used to Given that the estimation of human capital wealth is based compute expected earnings throughout individuals’ working on Mincerian wage regressions, the measure accounts life, considering their sex, education level, and assumed not only for the number of years of schooling completed experience (computed based on age and the number of years by workers, but also for the earning gains associated with of education completed). Expected earnings are computed schooling (which implicitly factors in the quality of learning in for all individuals in the surveys from age 15 to age 65, noting school), whether individuals work (labor force participation), that some individuals may go to school beyond age 15. The and for how many years they work (accounting for health analysis also considers the life expectancy of the labor force. conditions through life expectancy). Estimations of human In countries with high life expectancy, workers are expected capital wealth are done separately for men and women. This to work until age 65, but in other countries they may not means that once we have estimates of human capital wealth be able to. For simplicity, when estimating the discounted by gender, we can estimate losses in human capital wealth value of future earnings, the same discount factor for future due to gender inequality in earnings in a very simple way. earnings is applied to all countries. If we denote a country’s human capital wealth as measured FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 76 from the expected future earnings of men and women as wealth (both sexes included). The World Bank is in the HCM and HCW, respectively, and the adult population process of updating its estimates of the changing wealth of of men and women by POPM and POPW, the earnings nations, including human capital wealth, but these estimates per adult men and women can be defined as hcM=HCM/ are not yet available. For this study, estimates of human POPM and hcW=HCW/POPW. Under gender equality, capital wealth by gender for 2017 are therefore based on interpreted as ensuring that adult men and women have the projections. The estimates are based on a projection that same future expected earnings, human capital for women takes into account the gender gaps observed in 2014 (these would increase from hcW to hcM. Therefore, the loss in gaps do not change very much from one year to the next), human capital wealth from gender inequality is measured as and growth in real GDP per capita and population between (hcM-hcW)×POPW. Details are provided in Wodon (2018). 2014 and 2017. These projections are tentative, but they are likely to be reasonably accurate given that levels of GDP per Estimates of human capital wealth by gender up to 2014 capita across countries explain close to 95 percent of the are based on Wodon and de la Brière (2018), following variation in the estimates of human capital wealth per capita Hamilton et al. (2018) for estimates of total human capital across countries, as shown in Figure 16. Figure 16: Human Capital Wealth Per Capita and GDP Per Capita 6.5 6.0 5.5 5.0 Human Capital (in Log) 4.5 4.0 3.5 3.0 2.5 2.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 GDP Per Capita (in Log) Source: Hamilton et al. (2018). 77 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 REFERENCES Abramsky, T., K. Devries, L. Kiss, J. Nakuti, N. Kyegombe, E. Starmann, B. Cundill, L. Francisco, D. Kaye, T. Musuya, L. Michau, and C. Watts. 2014. Findings from the SASA! Study: A Cluster Randomized Controlled Trial to Assess the Impact of a Community Mobilization Intervention to Prevent Violence against Women and Reduce HIV Risk in Kampala, Uganda, BMC Medicine 12:122. Accenture and International Finance Corporation. 2018. Driving Toward Equality: Women, Ride-hailing, and the Sharing Economy. Washington, DC: International Finance Corporation. Acemoglu, D. 2010. Theory, General Equilibrium and Political Economy. NBER Working Paper No. 15944. Boston, MA: National Bureau of Economic Research. Acemoglu, D., D. H. Autor, and D. Lyle. 2004. Women, War, and Wages: The Effect of Female Labor Supply on the Wage Structure at Midcentury, Journal of Political Economy 112(3): 497-551. Agénor, P. R. and O. Canuto. 2013. Gender Equality and Economic Growth in Brazil: A Long-Run Analysis, Policy Research Working Paper 6348. Washington, DC: The World Bank. Aizer, A. 2010. The Gender Wage Gap and Domestic Violence. American Economic Review 100 (4) 1847-59. Alcott, B., 2017. Does teacher encouragement influence students’ educational progress? A propensity-score matching analysis. Research in Higher Education, 58(7), pp.773-804. Ali, D. A., K. Deininger, and M. Goldstein. 2014. Environmental and gender impacts of land tenure regularization in Africa: Pilot evidence from Rwanda. Journal of Development Economics 110: 262-75 Amaral, S., S. Bandyopadhyay, and R. Sensarma. 2015. Public Work Programs and Gender-based Violence: The Case of NREGA in India. Birmingham University, Department of Economics Discussion Paper 15-09. Angrist, J. D. 1995. The Economic Returns to Schooling in the West Bank and Gaza, American Economic Review, 85(5): 1065-87. Arango, Diana J et al. 2014. Interventions to Prevent or Reduce Violence against Women and Girls: A Systematic Review of Reviews. Washingtin, DC: The World Bank. Baird, S., C. McIntosh, and B. Özler. 2011. Cash or Condition? Evidence from a Cash Transfer Experiment, The Quarterly Journal of Economics, 126(4): 1709–53. Barker, Gary, Christine Ricardo, Marcos Nascimento, and World Health Organization. 2007. Engaging Men and Boys in Changing Gender- Based Inequity in Health: Evidence from Programme Interventions. Geneva: World Health organization. Bandiera, O., N. Buehren, R. Burgess, M. Goldstein, S. Gulesci, R. Rasul, and M. Sulaiman. Forthcoming. Women’s Empowerment in Action: Evidence from a Randomized Control Trial in Africa, American Economic Journal: Applied Economics. Banerjee, A., D. Karlan, and J. Zinman. 2015. Six randomized evaluations of micro-credit: Introduction and Further Steps. American Economic Journal: Applied Economics 7(1) 1-21. Barageine, J. K., E. Faxelid, J. K. Byamugisha, and B. Rubenson. 2016. ‘As a Man I Felt Small’: A Qualitative Study of Ugandan men’s Experiences of Living with a Wife Suffering from Obstetric Fistula, Culture, Health and Sexuality, 8(4): 481-94. Bashir, S., M. Lockheed, E. Ninan, and J. P. Tan, 2018. Facing Forward: Schooling for Learning in Africa. Washington, DC: The World Bank. Beegle, K. and L. Christiaensen, Editors. 2019. Accelerating Poverty Reduction in Africa. Washington, DC: The World Bank.  Beteille, T., and D. Evans. 2018. Successful Teachers, Successful Students: Recruiting and Supporting Society’s Most Crucial profession. Washington, DC: The World Bank. Black, D., N. Kolesnikova, and L. Taylor. 2014. Why Do So Few Women Work in New York (And So Many in Minneapolis)? Labor Supply of Married Women across U.S. Cities. Journal of Urban Economics 79: 59-71. Black, M. M., S. P. Walker, L. C. H. Fernald, C. T. Andersen, A. M. DiGirolamo, C. Lu, D. C. McCoy, G. Fink, Y. R. Shawar, J. Shiffman, A. E. Devercelli, Q. T. Wodon, E. Vargas-Baron, and S. Grantham-McGregor. 2017, Early Childhood Development Coming of Age: Science through the Life Course, The Lancet, 389 (10064): 77-90. Blackden, M., and Q. Wodon, 2006. Gender, Time Use, and Poverty in sub-Saharan Africa. World Bank Working Paper No. 73, Washington, FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 78 DC: The World Bank. Bobonis, Gustavo J, Melissa González-Brenes, and Roberto Castro. 2013. Public Transfers and Domestic Violence: The Roles of Private Information and Spousal Control. American Economic Journal: Economic Policy 5(1): 179–205. Boserup, E. 1970. The Role of Women in Economic Development, New York: St. Martin’s Press. Botea, I., S. Chakravarty, and S. Haddock, and Q. Wodon. 2017. Interventions Improving Sexual and Reproductive Health Outcomes and Delaying Child Marriage and Childbearing for Adolescent Girls, Ending Child Marriage Notes Series. Washington, DC: The World Bank. Bourey, C., W. Williams, E. E. Bernstein, and R. Stephenson. 2015. Systematic Review of Structural Interventions for Intimate Partner Violence in Low-and Middle-Income Countries: Organizing Evidence for Prevention. BMC Public Health 15(1): 1165. Brody, C., T. de Hoop, M. Vojtkova, R. Warnock, M. Dunbar, P. Murthy, and S. Dworkin. 2015. Economic Self-Help Group Programs for Improving Women’s Empowerment: A Systematic Review. Campbell Systematic Reviews 11, no 19. Buehren, N. 2015. The Cost of the Gender Gap in Agricultural Productivity in Malawi, Tanzania, and Uganda. Mimeo. Washington, D.C.: The World Bank Buller, A. M., M. Hidrobo, A. Peterman, and L. Heise. 2016. The Way to a Man’s Heart Is through His Stomach?: A Mixed Methods Study on Causal Mechanisms through Which Cash and in-Kind Food Transfers Decreased Intimate Partner Violence. BMC Public Health 16(1): 488. Burde, D., and L. L. Linden. 2013. Bringing education to Afghan girls: A randomized controlled trial of village-based schools. American Economic Journal: Applied Economics, 5(3), 27–40. Buvinic, M. and M. O’Donnell. 2016. Revisiting What Works: Women, Economic Empowerment and Smart Design. Washington, DC: CGD, Data 2X, and United Nations Foundation. Campos, F., M. Goldstein, L. McGorman, A. M. Munoz Boudet and O. Pimhidzai. 2015. Breaking the Metal Ceiling: Female Entrepreneurs Who Succeed in Male-dominated Sectors in Uganda. Africa Gender Innovations Lab. Washington, DC: The World Bank. Campos, F. and M. Gassier. 2017. Gender and Entreprise Development in Sub-Saharan Africa. A Review of Constraints and Effective Interventions. World Bank Policy Research Working Paper 8239. Washington, DC: The World Bank. Canning, D., S. Raja, and A. S. Yazbeck. 2015. Africa’s Demographic Transition Dividend or Disaster?, Washington, DC: The World Bank. Carneiro, P. M. and Heckman, J. J., 2003. Human Capital Policy. IZA Discussion Paper No. 821. Bonn: IZA. Chachashvili-Bolotin, S., M. Milner-Bolotin, and S. Lissitsa. 2016. Examination of Factors Predicting Secondary Students’ Interest in Tertiary STEM Education, International Journal of Science Education 38(3): 366-390. Chakravarty, S., S. Das, and J. Vaillant. 2017. Gender and Youth Employment in Sub-Saharan Africa: A Review of Constraints and Effective Interventions. Policy Research Working Paper; No. 8245. Washington, DC: The World Bank. Christiaensen, L., C. Scott, and Q. Wodon. 2002. Development Targets and Costs, in J. Klugman, editor, A Sourcebook for Poverty Reduction Strategies, Volume 1: Core Techniques and Cross-Cutting Issues, Washington, DC: The World Bank. Cuberes, D. and M. Teignier. 2015. How Costly Are Labor Gender Gaps? Estimates for the Balkans and Turkey. World Bank Policy Research Working Paper 7319. Washington, DC: The World Bank. Das, S., E. Carranza, and A. Kotikula. 2019. Addressing Gender-Based Employment Segregation. Jobs Working Paper, Washington, DC: The World Bank. Department of Economic and Social Affairs. 2014. Guidelines for Producing Statistics on Violence against Women— Statistical Surveys, Department of Economic and Social Affairs Statistics Division, New York: United Nations. Denboba, A.D., Sayre, R.K., Wodon, Q.T., Elder, L.K., Rawlings, L.B. and Lombardi, J., 2014. Stepping up early childhood development: investing in young children for high returns. Devercelli, A. and F. Beaton-Day. 2020. Better Jobs and Brighter Futures: Investing in Childcare to Build Human Capital. Washington, DC: The World Bank. Dewey, K.G. and Begum, K., 2011. Long-term consequences of stunting in early life. Maternal & child nutrition, 7, pp.5-18. Dhar, D., T. Jain and S. Jayachandran. 2018. Reshaping Adolescents’ Gender Attitudes: Evidence from a School-Based Experiment in India. NBER Working Paper No. 25331, Boston: National Bureau of Economic Research. Duflo, E. 2004. The Medium Run Effects of Educational Expansion: Evidence from a Large School Construction Program in Indonesia, Journal of Development Economics, 74(1): 163-97. 79 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Duflo, E. 2012. Women Empowerment and Economic Development, Journal of Economic Literature 50(4): 1051–79. Elborgh-Woytek, K., M. Newiak, K. Kochhar, S. Fabrizio, K. Kpodar, P. Wingender, B. Clements, and G. Schwartz. 2013. Women, Work and the Economy: Macroeconomic Gains from Gender Equity. IMF Staff Discussion Note. Washington, DC: International Monetary Fund. Ellsberg, M., D. J Arango, M. Morton, F. Gennari, S. Kiplesund, M. Contreras, and C. Watts. 2014. Prevention of Violence against Women and Girls: What Does the Evidence Say?, The Lancet 385(9977): 1555-66. Elson, D. 2017. Recognize, Reduce, and Redistribute Unpaid Care Work: How to Close the Gender Gap, New Labor Forum 26(2): 52-61. Engle, P.L., Fernald, L.C., Alderman, H., Behrman, J., O’Gara, C., Yousafzai, A., de Mello, M.C., Hidrobo, M., Ulkuer, N., Ertem, I. and Iltus, S., 2011. Strategies for reducing inequalities and improving developmental outcomes for young children in low-income and middle-income countries. The Lancet, 378(9799), pp.1339-1353. Estache, A. and Q. Wodon. 2014. Infrastructure and Poverty in sub-Saharan Africa. New York and London: Palgrave Macmillan. Evans, D. and A. Popova. 2017. Cash Transfers and Temptation Goods. Economic Development and Cultural Change, 65(2), 189–221. Evans, D., and F. Yuan. 2019. What We Learn about Girls’ Education from Interventions that Don’t Target Girls. Mimeo. Washington, DC: The World Bank. Ferrari, Giulia, and Radha Iyengar. 2010. Discussion Sessions Coupled with Microfinancing May Enhance the Roles of Women in Household Decision-Making in Burundi. Centre for Economic Performance, LSE. Field, E. and A. Ambrus. 2008. Early Marriage, Age of Menarche, and Female Schooling Attainment in Bangladesh. Journal of Political Economy 116(5): 881-930. Galasso, E. and Wagstaff, A., 2019. The aggregate income losses from childhood stunting and the returns to a nutrition Intervention aimed at reducing stunting. Economics & Human Biology 34:225-238. García-Moreno, C., H. A. F. M. Jansen, M. Ellsberg, L. Heise, and C. Watts. 2005. WHO Multi-country Study on Women’s Health and Domestic Violence against Women. Geneva: World Health Organization. Gemignani, R., and Q. Wodon, 2015. Child Marriage and Faith Affiliation in sub-Saharan Africa: Stylized Facts and Heterogeneity, Review of Faith & International Affairs, 13(3): 41-47. Goldstein, M., P. Gonzalez Martinez, and S. Papineni. 2019. Tackling the Global Profitarchy : Gender and the Choice of Business Sector. Policy Research Working Paper 8865, Washington, DC: The World Bank. Green, E. P, C. Blattman, J. Jamison, and J. Annan. 2015. Women’s Entrepreneurship and Intimate Partner Violence: A Cluster Randomized Trial of Microenterprise Assistance and Partner Participation in Post-Conflict Uganda. Social Science and Medicine 133: 177–88. Grown, C. and I. Valodia, Editors. 2010. Taxation and Gender Equity. A Comparative Analysis of Direct and Indirect taxes in Developing and Developed Countries. IDRC, Routledge. Hamilton, K., Q. Wodon, D. Barrot and A. Yedan. 2018. Human Capital and the Wealth of Nations: Global Estimates and Trends, in G.-M. Lange, Q. Wodon, and K. Carey, Editors, The Changing Wealth of Nations 2018: Sustainability into the 21st Century, Washington, DC: the World Bank. Harper, C., R. Marcus, and K. Moore. 2003. Enduring Poverty and the Conditions of Childhood: Lifecourse and Intergenerational Poverty Transmissions, World Development 31(3): 535-54. Haushofer, Johannes, and Jeremy Shapiro. 2016. “The Short-Term Impact of Unconditional Cash Transfers to the Poor: Experimental Evidence from Kenya.” The Quarterly Journal of Economics 131(4): 1973–2042. Heckman, J.J. and Masterov, D.V., 2007. The productivity argument for investing in young children. Applied Economic Perspectives and Policy, 29(3), pp.446-493. Heckman, J.J and S. Mosso. 2014. The Economics of Human Development and Social Mobility. Annual Review of Economics 6: 689-733. Heise, Lori. 2011. What Works to Prevent Partner Violence? An Evidence Overview. Working Paper. London: London School of Hygiene and Tropical Medicine. Herz, B., B. K. Herz, and G. B. Sperling. 2004. What Works in Girls’ Education: Evidence and Policies from the Developing World. Washington, DC: Council on Foreign Relations. Hex, N., J. Hanlon, D. Wright, V. Dale, and K. Bloor. 2016. Estimating the Costs of Female Genital Mutilation Services to the NHS, Partnership for Responsive Policy Analysis and Research, University of York and The King’s Fund. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 80 Hidrobo, M. and L. Fernald. 2013. Cash Transfers and Domestic Violence. Journal of Health Economics 32(1): 304–19. Hoddinott, J., J. A. Maluccio, J. R. Behman, R. Flores, and R. Martorell. 2008. Effect of a Nutrition Intervention during Early Childhood on Economic Productivity in Guatemalan Adults. Lancet 371 (9610): 411–16. Hoddinott, J., J. Maluccio, J. R. Behrman,. R. Martorell, P. Melgar, A. R. Quisumbing, M. Ramirez-Zea, R. D. Stein, and K. M. Yount. 2011. The consequences of early childhood growth failure over the life course. IFPRI Discussion Paper 01073. International Food Policy Research Institute, Washington, DC. Hoddinott, J., Alderman, H., Behrman, J. R., Haddad, L. and Horton, S. 2013a. The economic rationale for investing in stunting reduction, Maternal and Child Nutrition 9 (Suppl. 2): 69-82. Hoddinott, J., J. R. Behrman, J. A. Maluccio, P. Melgar, A. R. Quisumbing, M. Ramirez-Zea, A. D. Stein, K. M. Yount, and R. Martorell. 2013. Adult Consequences of Growth Failure in Early Childhood, American Journal of Clinical Nutrition 98(5): 1170-8. Hoeffler, A. and J. Fearon. 2014. Benefits and Costs of the Conflict and Violence Targets for the Post-2015 Development Agenda. Copenhagen Consensus Center. Horton S., and R. Steckel. 2013. Global economic losses attributable to malnutrition 1900-2000 and projections to 2050, in B. Lomborg, Editor, The Economics of Human Challenges, Cambridge, U.K.: Cambridge University Press. Hossain, M. et al. 2014. Men’s and Women’s Experiences of Violence and Traumatic Events in Rural Cote d’Ivoire Before, during and after a Period of Armed Conflict. BMJ Open 4(2): e003644. International Finance Corporation, Axxa and Accenture. 2015. She for Shield: Insure Women to Better Protect All. Washington, DC: International Finance Corporation. International Finance Corporation. 2017. Tackling Childcare: The Business Case for Employer-supported Chidcare. Washington, DC: International Finance Corporation. International Labor Organization. 2018. Global Wage Report 2018/19: What Lies Behind Gender Gaps. Geneva: International Labor Organization. Iyengar, R. and G. Ferrari. 2011. Discussion Sessions Coupled with Microfinancing May Enhance the Role of Women in Household Decision- Making in Burundi. NBER Working Papers Series No. 16902. Boston: National Bureau of Economic Research. Jayachandran, S., 2019. Social Norms as a Barrier to Women’s Employment in Developing Countries. WIDER Working Paper Series 2019-74, Helsinki: World Institute for Development Economic Research. Jewkes, R. 2014. What Works in Preventing Violence Against Women and Girls: Evidence Review of the Effectiveness of Response Mechanisms in Preventing Violence against Women and Girls. London: Department for International Development (DFID). Jewkes, R. et al. 2008. Impact of Stepping Stones on Incidence of HIV and HSV-2 and Sexual Behaviour in Rural South Africa: Cluster Randomised Controlled Trial. BMJ 337: a506. Jorgensen, D.W. and B.M. Fraumeni. 1992a. The Output of Education Sector, in Z. Griliches (ed.). Output Measurement in the Service Sectors. Chicago, IL: University of Chicago Press. Jorgensen, D.W. and B.M. Fraumeni. 1992b. Investment in Education and US Economic Growth. Scandinavian Journal of Economics, 94(Supplement): 51–70.Kabeer, N. 2016. Gender Equality, Economic Growth, and Women’s Agency: The “Endless Variety” and “Monotonous Similarity” of Patriarchal Constraints, Feminist Economics 22 (1), 295-321. Kabeer, N. 2016. Gender Equality, Economic Growth, and Women’s Agency: the “Endless Variety” and “Monotonous Similarity” of Patriarchal Constraints, Feminist Economics, 22(1): 295-321. Kalamar, A. M., S. Lee-Rife, and M. J. Hindin. 2016. Interventions to Prevent Child Marriage among Young People in Low- and Middle- Income Countries: A Systematic Review of the Published and Gray Literature, Journal of Adolescent Health, 59: S16-S21. Kamal, S. M. 2012. Decline in Child Marriage and Changes in its Effect on Reproductive Outcomes in Bangladesh. Journal of Health, Population, and Nutrition 30(3):317–330 Kazianga, H., Levy, D., Linden, L. L., & Sloan, M. 2013. The effects of “girl-friendly” schools: Evidence from the BRIGHT school construction program in Burkina Faso. American Economic Journal: Applied Economics 5(3), 41–62. Kim, Julia C et al. 2007. Understanding the Impact of a Microfinance-Based Intervention on Women’s Empowerment and the Reduction of Intimate Partner Violence in South Africa. American Journal of Public Health 97(10): 1794–1802. 81 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Klasen, S. and F. Lamanna. 2009. The Impact of Gender Inequality in Education and Employment on Economic Growth: New Evidence for a Panel of Countries, Feminist Economics 15(3): 91-132. Klugman, J., L. Hanmer, S. Twigg, T. Hasan, and J. McCleary-Sills. 2014. Voice and Agency: Empowering Women and Girls for Shared Prosperity. Washington, DC: The World Bank. Koski, A., E. C. Strumpf, J. S. Kaufman, J. Frank, J. Heymann, and A. Nandi. 2018. The impact of eliminating primary school tuition fees on child marriage in sub-Saharan Africa: A quasi-experimental evaluation of policy changes in 8 countries. PLoS ONE 13(5): e0197928 Lange, G. M., Q. Wodon, and K. Carey. 2018. The Changing Wealth of Nations 2018: Sustainability into the 21st Century. Washington: The World Bank. Le Nestour, A., O. Fiala, and Q. Wodon. 2019. Global and Regional Trends in Child marriage. Mimeo. London and Washington, DC: Save the Children and The World Bank. Lenze, J. and S. Klasen. 2017. Does Women’s Labor Force Participation Reduce Domestic Violence: Evidence from Jordan. Feminist Economics 23(1): 1-29. Levine, R., C. Lloyd, M. Greene, and C. Grown. 2008. Girls Count: A Global Investment and Action Agenda. Washington, DC: Center for Global Development. Male, C., and Q. Wodon. 2018. Girls’ Education and Child Marriage in West and Central Africa: Trends, Impacts, Costs, and Solutions, Forum for Social Economics, 47(2): 262-74. Marcus, R. 2018. The Norms Factor: Recent Research on Gender, Social Norms, and Women’s Economic Empowerment, London: Overseas Development Institute. Marcus, R., C. Harper, S. Brodbeck, and E. Page. 2015. How Do Gender Norms Change?, Knowledge to Action Resource Series, Lonon: Overseas Development Institute. Mateo Díaz, M. and Rodriguez-Chamussy, L., 2016. Cashing in on education: Women, childcare, and prosperity in Latin America and the Caribbean. The World Bank. McKinsey Global Institute. 2015. The Power of Parity: How Advancing Women’s Equality Can Add US$12 Trillion to Global Growth. London: McKinsey Global Institute. McKinsey Global Institute. 2019. The Future of Women at Work: Transitions in the Age of Automation. London: McKinsey Global Institute. Mgalla, Z., Schapink, D. and Boerma, J.T., 1998. Protecting school girls against sexual exploitation: A guardian programme in Mwanza, Tanzania. Reproductive Health Matters, 6(12), pp.19-30. Ministry of Health, Community Development, Gender, Elderly, and Children. 2017. National Survey on the Drivers and Consequences of Child Marriage in Tanzania. Dar Es Salaam, Tanzania. Montenegro, C. E., and H. A. Patrinos. 2014. Comparable Estimates of Returns to Schooling Around the World, Policy Research Working Paper No. 7020, Washington, DC: The World Bank. Montenegro, C. E., and Q. Wodon. 2020. Measuring the Returns to Quality in Primary Education Globally Using Literacy as a Proxy, Mimeo, Washington, DC: The World Bank. Morrison, A. and M. B. Orlando. 1999. The Socioeconomic Costs of Domestic Violence: Chile and Nicaragua, in A. Morrison and M. L. Biehl, Editors, Too Close to Home: Domestic Violence in the Americas. Washington, D.C.: Inter-American Development Bank. National Research Council. 2005. The Changing Transitions to Adulthood in Developing Countries: Selected Studies. Washington, DC: The National Academies Press. Nayihouba, A. and Q. Wodon. 2018. Gains in Human Capital Wealth: What Growth Models Tell Us, in G.-M. Lange, Q. Wodon, and K. Carey, Editors, The Changing Wealth of Nations 2018: Sustainability into the 21st Century. Washington, DC: The World Bank. Nelson, C.A., 2000. The neurobiological bases of early intervention. Handbook of early childhood intervention, 2, pp.204-227. Nguyen, M. C., and Q. Wodon. 2014. Impact of Child Marriage on Literacy and Educational Attainment in Africa, Background Paper for Fixing the Broken Promise of Education for All. Paris and New York: UNESCO Institute of Statistics and UNICEF. Nguyen, M. C., and Q. Wodon. 2015. Global and Regional Trends in Child Marriage, Review of Faith & International Affairs 13(3): 6-11. Nove, A., Z. Matthews, S. Neal, and A. V. Camacho. 2014. Maternal mortality in adolescents compared with women of other ages: evidence from 144 countries, The Lancet Global Health 2(3): 155-64. O’Sullivan, M. 2017. Gender and Property Rights in Sub-Saharan Africa: A Review of Constraints and Effective Interventions. Policy Research FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 82 Working Paper; No. 8250. Washington, DCL The World Bank. Overseas Development Institute. 2015. How do Gender Norms Change?, London: Overseas Development Institute. Onagoruwa, A., and Q. Wodon. 2018a. Measuring the Impact of Child Marriage on Total Fertility: A Study for Fifteen Countries, Journal of Biosocial Science, 50(5): 626-39. Onagoruwa, A., and Q. Wodon. 2018b. Selected Factors Leading to the Transmission of Female Genital Mutilation across Generations: Quantitative Analysis for Six African Countries, Ending Violence against Women Notes Series, Washington, DC: The World Bank. Ostry, J. D., J. Alvarez, R. Espinoza, and C. Papageorgiou. 2018. Economic Gains from Gender Inclusion: New Mechanisms, New Evidence. IMF Staff Discussion Note SDN/18/06. Washington, DC: International Monetary Fund. Paluck, Elizabeth Levy, and Laurie Ball. 2010. Social Norms Marketing to Reduce Gender Based Violence. IRC Policy Briefcase. Parsons, J., J. Edmeades, A. Kes, S. Petroni, M. Sexton, and Q. Wodon, 2015. Economic Impacts of Child Marriage: A Review of the Literature, The Review of Faith & International Affairs 13(3): 12-22. Paul, S. 2016. Women’s Labour Force Participation and Domestic Violence: Evidence from India. Journal of South Asian Development 11(2). Perlman, D., F. Adamu, and Q. Wodon, 2018a. Why Do Adolescent Girls Drop Out of School in Niger? A Combined Quantitative and Qualitative Analysis, Marchés et Organisations 32(2): 179-94. Perlman, D., F. Adamu, and Q. Wodon, 2018b. Understanding and Ending Child Marriage: Insights from Hausa Communities. Girls’ Education and Child Marriage in West and Central Africa Notes Series, Washington, DC: The World Bank. Plan International República Dominicána. 2017. Niñas esposadas: Caracterización del Matrimonio Infantil Forzado en las provincias de Azua, Barahona, Pedernales, Elías Piña y San Juan. Planteamientos, No2. Psacharopoulos. G., and H. A. Patrinos. 2018. Returns to Investment in Education: A Decennial Review of the Global Literature, Education Economics. forthcoming. Pulerwitz, Julie, Annie Michaelis, Ravi Verma, and Ellen Weiss. 2010. Addressing Gender Dynamics and Engaging Men in HIV Programs: Lessons Learned from Horizons Research. Public Health Reports 125(2): 282–92. Qvist, H.-P. Y., A. Holm, and M. D. Munk. 2016. Demand and Supply Effects and Returns to College Education: Evidence from a Natural Experiment with Engineers in Denmark, CHCP Working Papers 2016-4, London, ON: Department of Economics, University of Western Ontario. Ravallion, M., and Q. Wodon. 2000. Does Child Labor Displace Schooling? Evidence on Behavioral Responses to an Enrollment Subsidy, Economic Journal, 110: C158-75. Refaei, M., S. Aghababaei, A. Pourreza, and S. Z. Masoumi. 2016. Socioeconomic and Reproductive Health Outcomes of Female Genital Mutilation. Archives of Iranian Medicine 19: 805-811. Reimo, A. M. Muñoz and A. Revenga. 2017. What Works for Women’s Work. Mimeo. Washington, DC: The World Bank. Rubiano M. E. and M. Viollaz. 2018. Gender Differences in Time Use: Allocating Time between the Market and the Household. Mimeo. Washington, DC: The World Bank. Savadogo, A. and Q. Wodon. 2018a. Impact of Child Marriage on Women’s Earnings across Multiple Countries. Education Global Practice. Washington, DC: The World Bank. Savadogo, A., and Q. Wodon. 2018.b To What Extent Could Ending Child Marriage Reduce Intimate Partner Violence in sub-Saharan Africa? Education Global Practice. Washington, DC: The World Bank. Sawada, Y., Aida, T., Griffen, A. S., Kozuka, E., Noguchi, H., & Todo, Y. 2016. Election, Implementation, and Social Capital in School-Based Management: Evidence from a Randomized Field Experiment on the COGES Project in Burkina Faso. JICA-RI Working Paper, (120). Shekar, M., Kakietek, J., Dayton Eberwein, J., and Walters, D. 2016. An Investment Framework for Nutrition: Reaching the Global Targets for Stunting, Anemia, Breastfeeding, and Wasting, Washington, DC: The World Bank. Shonkoff, J.P., Garner, A.S., Siegel, B.S., Dobbins, M.I., Earls, M.F., McGuinn, L., Pascoe, J., Wood, D.L., Committee on Psychosocial Aspects of Child and Family Health and Committee on Early Childhood, Adoption, and Dependent Care, 2012. The lifelong effects of early childhood adversity and toxic stress. Pediatrics, 129(1), pp.e232-e246. Skoufias, E., K. Vinha, and R. Sato. 2020. All Hands on Deck: Reducing Stunting through Multisectoral Efforts in Sub-Saharan Africa. Washington, DC: The World Bank. 83 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Slegh, Henny et al. 2013. ‘I Can Do Women’s Work’: Reflections on Engaging Men as Allies in Women’s Economic Empowerment in Rwanda. Gender & Development 21(1): 15–30. Solutions for Youth Employment. 2018. Digital Jobs for Youth: Young Women in the Digital Economy, Washington, DC: The World Bank. Sperling, G.B. and Winthrop, R., 2016. What works in girls’ education: Evidence for the world’s best investment. Brookings Institution Press. Strauss, J. and Thomas, D., 1998. Health, nutrition, and economic development. Journal of economic literature, 36(2), pp.766-817.Solutions for Youth Employment (S4YE). 2018. Digital Jobs for Youth: Young Women in the Digital Economy. Washington, DC: World Bank. Steinhaus, M., L. Hinson, A. Theodore Rizzo, and A. Gregowski, Measuring Social Norms Related to Child Marriage Among Adult Decision-Makers of Young Girls in Phalombe and Thyolo, Malawi, Journal of Adolescent Health 64: S37-S44. Tappis, Hannah, Jeffrey Freeman, Nancy Glass, and Shannon Doocy. 2016. Effectiveness of Interventions, Programs and Strategies for Gender- Based Violence Prevention in Refugee Populations: An Integrative Review. PLoS Currents 8. Tavares, P., and Q. Wodon. 2018. Global and Regional Trends in Women’s Legal Protection against Domestic Violence and Sexual Harassment. Ending Violence against Women Notes Series. Washington, DC: The World Bank. Taghreed A., H. Bathija, D. Bishai, Y.-T. Bonnenfant, M. Darwish, D. Huntingtona, and E. Johansen for the FGM Cost Study Group of the World Health Organization. 2010. Estimating the Obstetric Costs of Female Genital Mutilation in Six African Countries, Bulletin of the World Health Organization, 88: 281–88.Taylor, A. Y., E. Murphy-Graham, J. Van Horn, B. Vaitla, A. Del Valle, and B.Cislaghi. 2019. Child Marriages and Unions in Latin America: Understanding the Roles of Agency and Social Norms, Journal of Adolescent Health 64: S45-S51. Tiwari, S., H. G. Jacoby, and E. Skoufias. 2013. Monsoon Babies: Rain Shocks and Child Nutrition in Nepal. Policy Research Working Paper No. 6395. Washington, DC: The World Bank. UNESCO. 2017. Reducing Global Poverty through Universal Primary and Secondary Education, Policy Paper 32/Fact Sheet 44. Paris: UNESCO. Unterhalter, E., A. North, M. Arnot, C. Lloyd, L. Moletsane, E. Murphy-Graham, J. Parkes, and M. Saito. 2014. Girls’ Education and Gender Equality. London: Department for International Development. Van de Walle, D., M. Ravallion, V. Mendiratta, and G. Koolwal. 2013. Long-term Impacts of Household Electrification in Rural India. World Bank Policy Research WP 6527. Washington, DC: The World Bank. Verma, Ravi et al. 2008. Promoting Gender Equity as a Strategy to Reduce HIV Risk and Gender-Based Violence among Young Men in India. Horizons Final Report. Washington, DC: Population Council. WHO. 2010. Preventing Intimate Partner and Sexual Violence against Women: Taking Action and Generating Evidence. Geneva: World Health Organization. Wils, A. 2015. Reaching education targets in low and lower-middle income countries: Costs and finance gaps to 2030, Background paper prepared for the UNESCO Education for All Global Monitoring Report, Paris: UNESCO. Winkler, A. 2016. Women’s Labor Force Participation. Family-friendly Policies Increase Women’s Labor Force Participation, Benefiting Them, Their Families, and Society at Large. IZA World of Labor Paper 289. Wodon, Q. 2016. Early Childhood Development in the Context of the Family: The Case of Child Marriage, Journal of Human Development and Capabilities, 17(4): 590-98. Wodon, Q. 2020. Estimating the Economic Impacts and Costs of Child Marriage Globally: Methodology and Estimates. Education Global Practice. Washington, DC: The World Bank. Wodon, Q. 2018. What Is the Cost of Gender Inequality in Lost Earnings? Global Estimates Based on the Changing Wealth of Nations. Mimeo. Washington, DC: The World Bank. Wodon, Q. 2018. Education Budget Savings from Ending Child Marriage and Early Childbirths: The Case of Niger, Applied Economics Letters, 25(10): 649–52. Wodon, Q. 2020. Empowering Girls through Education in sub-Saharan Africa, Education Global Practice, Washington, DC: The World Bank. Wodon, Q., and B. de la Brière. 2018, Unrealized Potential: The High Cost of Gender Inequality in Earnings. The Cost of Gender Inequality Notes Series. Washington, DC: The World Bank. Wodon, Q., C. Fèvre, C. Malé, A. Nayihouba, and H. Nguyen. 2020. Ending Violence In and Around Schools: Potential Benefits and Promising Interventions, Washington, DC: The World Bank. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 84 Wodon, Q., C. Male, and A. Onagoruwa. Forthcoming. A Simple Approach to Measuring the Share of Early Childbirths Likely Due to Child Marriage in Developing Countries, Forum for Social Economics, forthcoming. Wodon, Q., C. Male, A. Nayihouba, A. Onagoruwa, A. Savadogo, A. Yedan, J. Edmeades, A. Kes, N. John, L. Murithi, M. Steinhaus, and S. Petroni. 2017. Economic Impacts of Child Marriage: Global Synthesis Report, Washington, DC: The World Bank and ICRW. Wodon, Q., C. Montenegro, H. Nguyen, and A. Onagoruwa. 2018a. Missed Opportunities: The High Cost of Not Educating Girls. The Cost of Not Educating Girls Notes Series, Washington, DC: The World Bank. Wodon, Q., C. Montenegro, H. Nguyen, and A. Onagoruwa. 2018b. Educating Girls and Ending Child Marriage: A Priority for Africa, The Cost of Not Educating Girls Notes Series, Washington, DC: The World Bank. Wodon, Q., C. Nguyen and C. Tsimpo. 2016. Child Marriage, Education, and Agency in Uganda, Feminist Economist, 22(1): 54-79. Wodon, Q. and M. Shekar. 2016. Nutrition, Health, and Early Childhood Development, in B. Pritchard, R. Ortiz, and M. Shekar, editors, The Routledge Handbook of Food and Nutrition Security, Abingdon, UK: Routledge. Wodon, Q., P. Tavares, O. Fiala, A. Le Nestour, and L. Wise. 2017. Child Marriage Laws and their Limitations. Ending Child Marriage Notes Series. London and Washington, D.C.: Save the Children and The World Bank. Wodon, Q., and A. Yedan. 2019. Obstacles to Birth Registration in Niger: Estimates from a Recent Household Survey, Journal of Health, Population, and Nutrition, 38(Suppl 1) 26: 1-11. Wodon, Q., A. Yedan, and E. Leden. 2017. Female Genital Cutting in Egypt: Drivers and Potential Responses, Development in Practice 27(5): 708-18. World Bank. 2001. Engendering Development through Gender Equality in Rights, Resources, and Voice, Washington, DC: The World Bank. World Bank. 2003. Implementation Completion Report on a Loan in the Amount of US$ 10 Million Equivalent to The Kingdom of Morocco for a Rural Water Supply and Sanitation Project, Report No: 25917. World Bank. 2006. Where Is the Wealth of Nations? Measuring Capital for the 21st Century, Washington, DC: The World Bank. World Bank. 2010. Stepping Up Skills for More Jobs and Higher Productivity. Washington, DC: The World Bank. World Bank. 2011. The Changing Wealth of Nations? Measuring Sustainable Development in the New Millennium, Washington, DC: The World Bank. World Bank. 2012. World Development Report 2012: Gender Equality and Development. Washington, DC: The World Bank. World Bank and One. 2014. Leveling the Field: Improving Opportunities for Women Farmers in Africa. Washington, DC: The World Bank and One. World Bank. 2015. Global Monitoring Report 2015/16: Development Goals in an Era of Demographic Change, Washington, DC: The World Bank. World Bank. 2016. Gender Equality, Poverty Reduction, and Inclusive Growth: 2016-2023 Gender Strategy. Washington, DC: The World Bank. World Bank. 2018a. Women, Business, and the Law 2018. Washington, DC: The World Bank. World Bank. 2018b. World Development Report 2018: Learning to Realize Education’s Promise. Washington, DC: The World Bank. World Bank, 2018c. Economic Impacts of Gender Inequality in Niger. Washington, DC: The World Bank. World Bank. 2019a. World Development Report 2019: The Changing Nature of Work. Washington, DC: The World Bank. World Bank, 2019b. Tanzania Economic Update: Human Capital, the Real Wealth of Nations. Washington, DC: The World Bank. World Bank, 2019c. Guinea: The Economic Benefits of a Gender Inclusive Society. Washington, DC: The World Bank. World Bank 2019d. The Skills Balancing Act in Sub-Saharan Africa: Investing in Skills for Productivity, Inclusivity, and Adaptability. The World Bank. World Economic Forum. 2020. The Global Gender Gap Report 2018. Geneva: The World Economic Forum. World Bank. 2020. Women, Business and the Law 2020. Washington, DC: The World Bank. 85 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | FEBRUARY 2020 Recommended citation for this note: Wodon, Q., A. Onagoruwa, C. Malé, C. Montenegro, H. Nguyen, and B. de la Brière. 2020. How Large Is the Gender Dividend? Measuring Selected Impacts and Costs of Gender Inequality. The Cost of Gender Inequality Notes Series. Washington, DC: The World Bank. This note was prepared by a team at the World Bank. The team acknowledges support for this note as part of a work program funded by the Canadian Government and the Children’s Investment Fund Foundation. The authors are especially grateful to Kathleen Beegle, as well as Sameera Al Tuwaijri, Niklas Buehren, Quy-Toan Do, Roberta Gatti, Caren Grown, Lucia Hanmer, and Oni Lusk-Stover for valuable comments. Muthoni Ngatia and Diana Arango contributed to the identification of interventions to prevent violence against women. Caren Grown provided strategic guidance. The authors are also grateful to Stefano Mocci and Meskerem Mulatu for their support, and Sherryl Silverman for support for communications and dissemination. The team is also grateful to colleagues from Global Affairs Canada, and in particular Candice Dandurand, Kayla Richardson, and Alexandra Stefanopoulos for their support. The findings, interpretations, and conclusions expressed in this note are entirely those of the authors and should not be attributed in any manner to the World Bank, its affiliated organizations or members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this note should take into account its provisional character. The World Bank does not guarantee the accuracy of the data included in this work. Information contained in this note may be freely reproduced, published or otherwise used for noncommercial purposes without permission from the World Bank. However, the World Bank requests that the original study be cited as the source. © 2020 The World Bank, Washington, DC 20433. FEBRUARY 2020 | HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY | 86 THE COST OF GENDER INEQUALITY NOTES SERIES: HOW LARGE IS THE GENDER DIVIDEND? MEASURING SELECTED IMPACTS AND COSTS OF GENDER INEQUALITY