World Bank Group Pakistan@100 From Poverty to Equity Policy Note March 2019 Silvia Redaelli Table of Contents Abstract ......................................................................................................................................... 1 Chapter 1: Growth, Poverty Reduction and Inequality: What Do We Know? .................... 2 Chapter 2: Economic and Social Mobility in Pakistan ............................................................ 9 Chapter 3: Inequality of Opportunities in Pakistan ................................................................ 13 Chapter 4: Aspiration Gaps and Intergenerational Transmission of Gender Inequality .. 21 Chapter 5: Poverty and Conflict in Pakistan .......................................................................... 25 Policy Discussion ......................................................................................................................... 29 References .................................................................................................................................. 37 i List of Figures Figure 1: GDP per capita growth and poverty, 2001-15 ................................................................................................................ 3 Figure 2: Poverty trends, by province.................................................................................................................................................... 5 Figure 3: Urban-rural poverty gap, by province................................................................................................................................ 6 Figure 4: Growth incidence curves .......................................................................................................................................................... 8 Figure 5: Trends in shared prosperity ................................................................................................................................................... 8 Figure 6. Absolute and relative mobility .............................................................................................................................................11 Figure 7: Coverage and HOI trends, national level, 2004-14 .....................................................................................................15 Figure 8: Contribution of circumstances to inequality of opportunity, 2006-14 .............................................................17 Figure 9: HOI, by province, 2014............................................................................................................................................................18 Figure 10: Contribution of gender to inequality of educational opportunities across provinces, 2014 ................18 Figure 11: District-level HOI, 2004-14 ................................................................................................................................................19 Figure 12: Trends in security incidents, 1988-2017 .....................................................................................................................25 Figure 13: Trend in security incidents, 2013-17.............................................................................................................................26 Figure 14: M&E system and the policy cycle.....................................................................................................................................35 List of Tables Table 1: Opportunities considered in the HOI analysis................................................................................................................15 Table 2: International comparison of access to opportunities .................................................................................................16 Table 3: Parents aspirations for children ...........................................................................................................................................23 Table 4: Adolescents’ own aspirations ................................................................................................................................................23 List of Boxes Box 1. Data and Poverty Estimation Methodology .......................................................................................................................... 4 Box 2. Inequality Measurement With Household Survey Data .................................................................................................. 7 Box 3. Absolute And Relative IGM ........................................................................................................................................................10 Box 4. HOI Methodology............................................................................................................................................................................14 Box 5. Aspirations And Outcomes.........................................................................................................................................................21 Box 6. Conflict and Poverty In Pakistan .............................................................................................................................................27 Box 7. Climate Change And Conflict .....................................................................................................................................................28 Box 8. Fundamental Principles Of Official Statistics .....................................................................................................................31 Box 9. Affirmative Action: The Experience Of The United States, South Africa, And India .........................................33 Box 10. Interventions Addressing Gender Norms .........................................................................................................................34 ii Glossary BISP Benazir Income Support Program CBN Cost Of Basic Needs CPI Consumer Price Index ECE Economic Commission For Europe FATA Federally Administered Tribal Areas FEI Food Energy Intake GBV Gender-Based Violence GDP Gross Domestic Product HIES Household Integrated Survey HOI Human Opportunity Index ICRW International Centre On Research On Women IFPRI International Food Policy Research Institute IGM Intergenerational Mobility KP Khyber Pakhtunkhwa LEAPS Learning And Educational Achievement In Punjab Schools M&E Monitoring And Evaluation NFC National Finance Commission NGO Nongovernmental Organization NISP National Internal Security Policy PBS Pakistan Bureau Of Statistics PIHS Pakistan Integrated Survey PIPS Pakistan Institute For Peace Studies PKR Pakistani Rupee PRHS Pakistan Rural Household Survey PRSP Poverty Reduction Strategies Paper PSLM Pakistan Social And Living Standards Measurement SC/ST Scheduled Caste And Scheduled Tribes iii Abstract Since 2001, economic development has contributed to lifting more than 32 million people out of poverty and the percentage of Pakistanis who are poor has begun to fall sustainably over time. Meanwhile, in line with global trends, the focus of the policy debate has shifted from poverty toward inequality. While monetary inequality—at least the part that can be measured with existing data—has not deteriorated over time, inequalities across and within regions in Pakistan persist, and there are signs that prosperity is not being shared by the poorest segments of the population. More worrying still is the existence of inequality traps—in which exclusion is transmitted across generations—and inequality of opportunities across space and between girls and boys. This points to a deficiency of equity and justice in the process of socioeconomic development which, if left unattended, could harm both growth and social stability in the long run. The objective of this note is to contribute to the understanding of inequality in Pakistan, taking advantage of the available household survey data. The report first reviews recent trends in monetary poverty and inequality, and then provides new evidence on the extent of socioeconomic mobility and equality of opportunities in the country. The note also explores how existing inequalities grounded in social norms, including gender norms arising from entrenched patriarchy, affect individuals’ capacity to aspire for a better, or simply different, future, and how geographical inequality in development outcomes might be correlated to conflict and violence. The emerging message is clear. If Pakistan wants to achieve its ambitious target of becoming one of the top 10 economies in the world by 2047, equity should be put at the center of the development process. 1 Chapter 1: Growth, Poverty Reduction and Inequality: What Do We Know? Sustained economic growth is a necessary, albeit insufficient, driver of poverty reduction. While there is abundant evidence that economic growth has been vital in lifting 1.1 billion people worldwide out of poverty since 1990 (World Bank, 2016a), the effect of growth on poverty reduction is not determined a priori and depends critically on the pace of growth and how its benefits are shared within a society. In the case of Pakistan, historical analysis of the link between growth and poverty reduction is hampered by the lack of consistent poverty time-series data until 2001. Nonetheless, several studies note that the link between growth and poverty reduction has historically been weak, with performance in terms of poverty reduction varying over time. During the 1960s, strong economic growth had an ambiguous impact on poverty. According to some scholars, growth contributed to a decline in both urban and rural poverty (Naseem, 1973), while according to others the period was characterized—at least in the general perception—by rising poverty and widening inequalities (Gazdar, 1999). Policy changes during the 1970s dampened growth, but despite lower growth than in the previous decade, the perception was that the welfare of the poorest improved (Gazdar, 1999). In the 1980s, improved economic policies and more favorable external factors, such as opportunities for migration in the Middle East, led to an uptick in growth, which averaged 6 percent annually, and a substantial decline in poverty. 1 The trend somehow reversed during the 1990s, when large fluctuations in growth rates were matched by high consumption volatility and no progress in poverty reduction. On balance, poverty remained almost unchanged between 1991 and 1999 (World Bank, 2002). Between 2001 and 2015, while growth continued to fluctuate, in part due to the negative impact of natural disasters and macroeconomic shocks, the period—in contrast to the previous decade—was characterized by an uninterrupted decline in poverty. In 2001, using the current official poverty line (Box 1), the poverty headcount rate in Pakistan was as high as 64.3 percent. After 14 years, in 2015, poverty had more than halved, reaching 24.3 percent. Most importantly, poverty declined consistently over time despite swings in economic growth (Figure 1). In each period for which Household Integrated Survey (HIES) data are available, the poverty rate is lower than the rate estimated in the previous survey round. While the role of different factors in driving the pace of poverty reduction is not yet clear, largely due to data limitations, it is plausible that economic growth, an increase in international migration, and the expansion of social protection and pro-poor spending under the government’s Poverty Reduction Strategies (PRSP-I and PRSP-II)2 all contributed to poverty reduction, along with urbanization and growth of the (informal) off-farm economy. 1 The evidence of a decline in poverty during the 1970s and 1980s is corroborated by the increase in real wages among unskilled construction workers in urban areas and agriculture workers (World Bank, 1995). 2 Poverty Reduction Strategy Paper I and II, 2003/2008. PRSP Secretariat, Ministry of Finance, Government of Pakistan. 2 Figure 1. GDP Per Capita Growth and Poverty, 2001-15 9% 70 8% 60 Poverty Headcount (%) 7% Growth (Real GDP) 50 6% 5% 40 4% 30 3% 20 2% 1% 10 0% 0 Growth (Real GDP) Poverty Headcount Source: World Bank staff calculations. 3 Box 1. Data and Poverty Estimation Methodology Poverty trends presented in this note are based on household survey data from the HIES, which provides information on socioeconomic indicators at both the national and provincial levels, including an urban/rural breakdown. The survey has been conducted by the Pakistan Bureau of Statistics (PBS) more or less regularly since 1963, although comparability over time is limited as the survey design (instruments and sample) changed to respond to evolving data requirements. In 1998–99, the HIES was integrated with the Pakistan Integrated Survey (PIHS) and conducted without major revisions since 2001–02. Eight rounds of data are used for the analysis, namely 2001–02, 2004–05, 2005–06, 2007–08, 2010–11, 2011–12, 2013–14, and 2015–16. Pakistan’s first official poverty line was set in 2001 at PKR 723.40 per adult equivalent per month. The poverty line was estimated using the Food Energy Intake (FEI) methodology and a minimum threshold level of 2,350 kcal per day per adult equivalent. According to this poverty line, 34.7 percent of the population was registered as poor in 2001–02, a proportion that progressively declined to 9.3 percent in 2013–14. Pakistan’s success in reducing poverty encouraged the government to raise the bar and estimate a more ambitious and inclusive poverty line reflecting the evolution of consumption patterns in the society. The new poverty line was estimated using HIES 2013–14 data and was set at PKR 3,030.32 rupees per adult equivalent per month in 2013–14 prices. In line with international best practice, the poverty line was estimated following the Cost of Basic Needs (CBN) method. The new line sets a higher bar and provides a more inclusive view of those that are considered disadvantaged in Pakistan today. In absolute numbers and based on population estimates of 180 to 200 million in 2013–14, the new poverty line allows 6.8 to 7.6 million households (53 to 59 million people) to be classified as poor. This demonstrates the government’s commitment to reaching low-income households through its policies and interventions. Backcasting this new poverty line to 2001–02, using the Consumer Price Index (CPI), shows that the headcount rate using this new higher line would have been 64.3 percent in 2001–02—almost double the rate seen using the old poverty line. However, the trends over time remain the same. 80 64.3 Poverty Headcount 60 51.7 50.4 44.1 34.7 36.8 36.3 40 29.5 23.3 21.9 16.5 12.2 20 12.1 9.3 0 2001 2004 2005 2007 2010 2011 2013 Old Poverty Line New Poverty Line However, the pace of poverty reduction has not been uniform throughout Pakistan. Between 2001 and 2015, poverty in urban areas declined at an annualized rate of 9 percent, compared with 6 percent in rural areas. In 2015, rural poverty was more than twice as high as poverty in urban areas 3 and, despite a decline in the share of rural population, rural areas still account for four out of five poor individuals—the same share as at the beginning of the century. A substantial stagnation in the urban/rural profile was matched by a general trend in poverty convergence at the provincial level (Figure 2). Khyber Pakhtunkhwa (KP) and Punjab were particularly successful at continuously reducing poverty between 2001 and 2015. Progress was 3In 2015, the poverty headcount in urban areas was 12.5 percent, against 30.7 percent in rural areas. Corresponding poverty rates in 2001 were 50.0 percent and 70.2 percent, respectively. 4 particularly impressive in KP, where poverty declined by 56 percentage points, allowing the province to graduate from being the poorest in 2001 to the least poor, together with Punjab, in 2015. 4 Poverty trends in Sindh and Balochistan followed a similar pattern but, while Sindh halved poverty over the period, progress was less strong in Balochistan. In 2015, Balochistan had by far the highest poverty rate (42.2 percent), twice as high as Punjab and KP. Figure 2. Poverty Trends, by Province 80 60 40 Percent Punjab 20 Sindh KPK 0 Balochistan Source: World Bank staff calculations. An equalizing trend also occurred within provinces, as poorer districts experienced a larger reduction in poverty. Higher rates of urbanization were in general associated with poverty reduction, although pockets of severe poverty also persist within districts that contain urban centers. Decomposition of inequality in district poverty between and across provinces shows that the variation in district poverty between provinces is larger than the variation within each province (Mansuri et al., 2018). At the same time, urban-rural inequalities remain prominent also within provinces. With the unique exception of KP, the poverty gap between urban and rural areas increased in all provinces (Figure 3), notably in Sindh and Balochistan. With a rural poverty headcount close to 50 percent, Sindh and Balochistan have the highest rural poverty rates among Pakistan’s provinces and the highest urban-rural poverty gaps. 5 4 In 2015, the poverty headcount rate in KP was 18.0 percent, against 20.9 percent in Punjab. However, the difference between KP and Punjab’s poverty headcount is not statistically significant. 5 In 2015, the poverty gap between rural and urban areas was 33.6 percentage points in Sindh and 21.8 percentage points in Balochistan. Corresponding figures in Punjab and KP were 16.3 and 9.8 percentage points, respectively. 5 Figure 3. Urban-Rural Poverty Gap, by Province 40 35 30 Percentage points 25 20 15 10 5 0 Punjab Sindh KPK Balochistan 2001-02 2015-16 Source: World Bank staff calculations. A mixed picture emerges from an analysis of the latest welfare trends. While Pakistan clearly made substantial progress in lifting about 32 million people out of poverty between 2001 and 2015, progress has been heterogeneous and substantial inequality persists, especially between Pakistan’s cities and its rural areas. Besides a higher poverty risk, rural households also face a disadvantage on virtually all aspects of service delivery, from education and health to access to key utilities, water and sanitation, and infrastructure, 6 with severe consequences in terms of human capital accumulation and long-term poverty reduction. Due to this disparity in outcomes, the progress in poverty reduction has been overshadowed by a general perception of worsening well-being and rising inequality. Measuring economic inequality accurately in Pakistan is complicated by a lack of comprehensive administrative data on income or wealth. Household survey data suffer from well-known shortcomings when it comes to inequality measurement and can only provide limited evidence (Box 2). Consumption inequality, as measured by the Gini coefficient, increased only marginally, from 27.5 in 2001 to 30.3 in 2015. A similar picture emerges when looking at income inequality, with the Gini coefficient increasing by just over 1 percentage point, from 37.5 in 2001 to 38.7 in 2015. Nationwide, the top 10 percent of the population consumes on average three times more than the bottom 10 percent and their incomes are five times as large, with no substantial changes over the period. 6Net enrolment in primary and middle school in rural areas shows a 13 and 14 percentage point deficit, respectively. These gaps rise to 17 and 18 percentage points among girls. The female literacy rate at 32 percent is half that of urban areas (HIICS 2015–16). Similar gaps exist in health (Mansuri et al. 2018). Rural toddlers are 8.5 percentage points less likely to have adequate immunization by age 3. Rural women are 10 percentage points less likely to receive prenatal care and 12 percentage points less likely to get postnatal care. They are also 28 percentage points less likely to give birth in a facility or hospital. Rural households are also far less likely to have access to key utilities (Household Integrated Income and Consumption Survey (HIICS 2015–16). They are 13 percentage points less likely to have an electricity connection and 64 percentage points less likely to have a natural gas connection. 6 Box 2. Inequality Measurement with Household Survey Data Household surveys are well known for failing to provide an accurate account of incomes or consumption of the richest members of society. Several reasons explain this problem. First, nationally representative household surveys typically focus on measuring consumption of goods that are more frequently purchased by the majority of the population, if not by its poorest segments. As such, household surveys might fail to capture consumption of top-end items purchased by the richest segments of the population. Second, the richest households tend to have higher nonresponse rates in household surveys and, if interviewed, might be more inclined to underreport their incomes. The extent of the problem varies by country and can depend on a range of factors, from perceptions of safety to survey implementation protocols. In the specific case of Pakistan, several indicators point to substantial top-truncation leading to an underestimation of top incomes. According to the HIES 2013–14, the top 1 percent of households consume on average PKR 158,200 per month and report monthly average income of PKR 383,213—values that general perceptions suggest are being grossly underestimated. The highest household monthly income and consumption recorded in the survey is PKR 4,278,000 and PKR 1,014,844, respectively. The extent of underreporting can also be gauged by assessing discrepancies between consumption measures obtained from household surveys and national accounts. In the case of Pakistan, the gap between per capita household consumption as measured in HIES 2013–14 and the household final consumption expenditure recorded in national accounts is between 40 and 50 percent of per capita GDP. A similar indication emerges when comparing purchases of motor vehicles (both new and second hand) recorded in the latest HICS 2015–16 in Punjab and changes of property/vehicle acquisitions registered in Punjab’s administrative records. The extent of underreporting is significant, with household survey data accounting for only 43 percent of registered car/motorbikes. A more indicative picture emerges when looking at the evolution of consumption growth at different percentiles of the distribution (growth incidence curve) and at trends in shared prosperity—measured as the average consumption growth of the poorest 40 percent of the population. As shown in Figure 4, except for the period between 2005 and 2010, consumption growth was stronger at the higher end of the distribution. The average consumption growth of the bottom 40 percent declined progressively, from 3.7 percent annually between 2001 and 2005 to 3 percent annually between 2005 and 2010, and to only 2.2 percent annually between 2010 and 2013 (Figure 5). Between 2010 and 2015, not only did consumption of the poorest 40 percent grow less than previously recorded, but it also grew 1 percentage point less than the average of the population, resulting in an increase in inequality. 7 The decline in shared prosperity and the uptick in inequality is a worrisome indication of the fact that economic progress is increasingly not being shared with the poorest segments of Pakistan’s population, potentially explaining mounting concerns over growing inequalities. 7 Between 2010 and 2015, the Gini coefficient increased from 27.7 to 30.3. 7 Figure 4. Growth Incidence Curves 8 GIC (2001/02-2005/06) 6 Annualized growth rate 4 GIC (2010/11-2015/16) 2 GIC (2005/06-2010/11) 0 5 15 25 35 45 55 65 75 85 95 Percentiles of per-adult equivalent expenditure Source: World Bank staff calculations. Figure 5. Trends in Shared Prosperity 6 5 Growth rate, percent 4 3 2 1 0 2001-02/ 2005-06/ 2010-11/ 2001-02/ 2005-06 2010-11 2015-16 2015-16 Annualized Growth Rate Mean Consumption Bottom 40 Annualized Growth Rate Mean Consumption Source: World Bank staff calculations. 8 Chapter 2: Economic and Social Mobility in Pakistan Actual and perceived inequality become amplified in societies that lack socioeconomic mobility. Inequality and intergenerational socioeconomic mobility are closely intertwined. From an intergenerational perspective, when children can aspire to achieve levels of education, jobs, and living standards that are materially better than the levels enjoyed by their parents, inequality begins to decline over time. However, when the capacity to climb the social ladder is predetermined by the lottery of birth, inequality persists over time, with negative consequences on equity and dynamic efficiency. A society that lacks mobility is unable to mobilize all available talents, leading to lower and less inclusive growth, and ultimately to higher inequality in the long run. Pakistan has made substantial progress in terms of poverty reduction and overall improvement in standards of living over the past decades. The process of economic development in Pakistan has led to absolute upward mobility across generations. Children and youth living in today’s Pakistan experience a much higher quality of life than their grandparents. If we look at the period from the early 1970s to the present, welfare as measured by GDP per capita (calculated in 2010 US$) increased 2.5 times from US$453 to US$1,178. Social indicators have also improved significantly. For example, a child born in Pakistan today can expect to live, on average, more than a decade longer than a child born two generations ago and achieve higher educational levels. Nonetheless, the picture is less positive when considering intergenerational mobility (IGM) from a relative perspective, that is, the extent to which an individual’s position on the economic ladder within a society is independent of the position of the individual’s parents (Box 3). A recent report (World Bank, 2018b) analyzing trends in IGM in education across 146 countries indicates that Pakistan ranks among the worst performing countries in absolute educational mobility, defined as the share of adults that are more educated than their parents, and relative mobility, defined as the correlation between individuals’ education and that of their parents. Pakistan also ranks among the 10 worst performing countries when looking at the share of individuals in the 1980s’ generation who made it to the top quartile of education out of all those born to parents with education in the bottom half of their generation. Ideally, if one’s ability to obtain an education did not depend on how well educated one’s parents are, the share would be 25 percent. In the case of Pakistan, only 9.4 percent of individuals born in the bottom half make it to the top compared with a median of 15 percent among developing economies. Compared with other South Asian countries, Pakistan is doing marginally better than India (8.9 percent) and Bangladesh (8.6 percent), but worse than Nepal (11.4 percent), Afghanistan (12.3 percent), Sri Lanka (15.9 percent), and the Maldives (24.8 percent). 9 Box 3. Absolute and Relative IGM Socioeconomic mobility has been interpreted in several ways in the economic and sociological literature, including mobility within and between generations, and as mobility in incomes, educational attainment, and occupation. This note focuses on IGM—both absolute and relative. To illustrate the two concepts of IGM used here, it is helpful to imagine two generations of adults standing on different rungs of the same economic ladder, where the rungs indicate one’s economic success relative to everyone else based on, for example, lifetime income (see the figure below). Absolute upward IGM measures the extent to which the current generation has managed to climb up the ladder relative to the previous generation, or the extent to which the rungs occupied by the current generation are higher than the rungs occupied by the previous generation, that is, the parents of the current generation. Relative IGM is the extent to which every individual’s position on the economic ladder is independent of the position of the individual’s parents. If an individual reaches a rung of the ladder among peers that is different from what the individual’s parents occupied among parents of the peers, then there has been relative mobility. While the two concepts are related, one may exist without the other. If all individuals in a generation climb two rungs relative to their parents without passing or being passed by anyone else in that generation, then there is absolute IGM, but not necessarily any relative IGM. Conversely, a society may exhibit high relative IGM, but no absolute IGM if all individuals in the current generation are on rungs that are different from the rungs occupied by their parents, while the current generation as a whole occupies the same rungs of the ladder as the previous generation. In this case, the standard of living of the society overall has not improved. Relative IGM is consistent with the interpretation of mobility as origin independence applied in an intergenerational context. In a society with high relative IGM, the lifetime incomes of individuals are influenced less by the origin of the individuals, that is, their parents. Origin independence is closely related to the concept of equality of opportunity (Roemer, 1998). Indeed, relative mobility may be synonymous with equality of opportunity, if the latter is defined in terms of a single circumstance, such as parental earnings or parental educational attainment. Source: World Bank 2018b. 10 Figure 6. Absolute and Relative Mobility 1.00 0.80 Absolute upward mobility 0.60 Pakistan 0.40 0.20 0.00 0.00 0.20 0.40 0.60 0.80 1.00 Inter-generational persistence Note: Higher intergenerational persistence indicates lower relative mobility. Source: World Bank 2018b A recent study provides additional evidence of limited mobility and inequality traps in rural Pakistan (Mansuri and Shrestha, 2018). Taking advantage of a 24-year household panel, 8 the study explores the impact of village-level inequality in land ownership on the intragenerational and intergenerational economic mobility of poor and vulnerable households. Intragenerational mobility is measured in terms of the odds of an initially transiently or chronically poor household escaping poverty, 9 while intergenerational economic mobility is measured by looking at the extent to which the educational attainment and labor market activity of children represents a movement up the economic ladder relative to their parents. Land ownership patterns in rural Punjab and Sindh have remained remarkably stable despite attempted land reforms in the 1970s. This long-term stability allows the concentration of land wealth at the village level to be treated as plausibly exogenous to individual households, and thus useful for understanding the causal impact of initial wealth on intragenerational and intergenerational socioeconomic mobility. 10 First, the analysis confirms the pattern of poverty reduction observed at the national level. Despite high levels of poverty incidence at baseline, most of the households that were either chronically or transiently poor in 1991 had managed to escape poverty by 2010. The analysis also corroborates the low level of educational mobility across generations in Pakistan. In 2010, only 40 8 The panel compiles different sets of household panels using multiple rounds of the Pakistan Rural Household Survey (PRHS). The first rounds were conducted by the International Food Policy Research Institute (IFPRI) in 1986 to 1991. The households surveyed in these rounds were included (along with their splits) in a larger rural survey conducted in 2001, 2004, and 2010. The paper focuses on 115 villages from 10 districts located in Punjab and Sindh, for which data are available across all rounds. 9 Households are defined as chronically poor if their per capita monthly consumption during the five annual survey rounds between 1986 and 1991 is always below the CPI adjusted poverty line, whereas, transiently poor households are those for whom per capita monthly consumption is, on average, at or above the CPI adjusted poverty line but falls below the poverty line at least once in the 5-year period. 10 Robustness tests also indicate that results are not driven by panel attrition. 11 percent of household heads had completed more years of schooling than their fathers, despite the fact that four out of five individuals in their fathers’ generation had no schooling at all. Most importantly, the study provides evidence that land inequality constrains both economic and educational mobility. Chronically poor households residing in more unequal villages are significantly less likely to have escaped poverty over time. 11 Similarly, upward education mobility is significantly lower in villages where land distribution is more concentrated. 12 The channels through which existing inequalities affect socioeconomic mobility are multiple (Rama et al., 2015). Inequality of outcomes can constrain the capacity of households at the bottom of the distribution to access finance and borrow to accumulate human and physical capital, which in turn may perpetuate inequality. Inequality can also affect the provision of public goods. At the same time, in an unequal society, the rich can use their power to pull resources for public goods they value, with potential positive benefits for the poor. However, a high degree of inequality makes it more tempting for the rich and the upper middle class to opt out of public services because they can afford better private alternatives. In this case, the quality of public services is bound to deteriorate, hence harming the poor. The analysis in the Mansuri and Shrestha paper shows that the accumulation of assets by households differs significantly between the most equal villages and the most unequal ones, with poorer households being more likely to accumulate livestock in villages with a lower level of land inequality. The paper also shows that inequality affects the provision of public goods. Villages where land inequality is higher enjoyed greater access to services such as electricity, drainage, and access to public transport, but did significantly worse on the quality of teachers in public schools. 11 Chronically poor households residing in villages that are in the top decile of land inequality are 6.2 percentage points more likely to still be poor in 2010, compared with chronically poor households in middle-decile villages. Alternatively, increasing land Gini from 0 to 1 decreases the likelihood of chronically poor households escaping poverty in 2010 by 67.3 percentage points. 12 Conditional on fathers having no schooling, the likelihood that the household head has completed at least lower secondary education is 8.2 percentage points smaller in top-decile villages compared to the likelihood of a similar transition in middle- decile villages. The size of this effect is also not trivial, amounting to 55 percent decrease in upward mobility compared with the mean (15 percent). 12 Chapter 3: Inequality of Opportunities in Pakistan A focus on monetary outcomes alone provides only a partial picture of the extent of inequality within a society. Most importantly, it does not provide any insight on the fairness of the process through which such inequalities come into existence. In fact, while concern is mounting about growing inequality and its impact on social cohesion, evidence from recent studies in behavioral economics also shows that fairness of economic process is of greater concern for people than economic inequality itself (Starmans, Sheskin, and Bloom, 2017). Tolerance for inequality varies over time and across countries. In general, public support tends to be greater for inequalities that stem from differences in effort and ability, and in contexts characterized by high, either actual or perceived, social mobility (Hirschman and Rothschild, 1973). On the other hand, inequality that is predetermined by the lottery of birth and that perpetuates itself across generations tends to be considered inherently unfair. Similarly, policies aimed at ensuring equal opportunities for children of all socioeconomic backgrounds receive greater support than those aimed at equalizing outcomes. However, while the latter could have implications on investments and efficiency, leveling the playing field is not only necessary to ensure equity and justice but is also instrumental in putting talent to full use and achieving sustainable growth and development (World Bank, 2005). Basic opportunities, such as completing primary school on time, continuing studies after primary, and having clean water to drink, and clean and adequate sanitation facilities and electricity at home, appear to be far from universal and unevenly distributed in Pakistan. Consider for example the case of Shaheen, a girl from Washuk, one of the poorest districts in rural Balochistan, whose parents are illiterate, and that of Amir, a boy from Lahore city, whose parents are educated. Would they have equal odds of becoming doctors, lawyers, politicians, or successful entrepreneurs? The answer is no. Understanding the extent to which circumstances at birth affect opportunities and whether the ‘opportunity gap’ is improving over time is therefore important to promote justice and avoid wasted human potential. The Human Opportunity Index (HOI) was developed to capture the extent to which opportunities are evenly distributed within a society and are not predetermined by circumstances at birth. Identifying opportunities that might affect outcomes in life is of course a daunting challenge, not just theoretically, but also in terms of data requirements. Following a growing body of international literature, opportunities for children are defined as access to a set of basic goods and services in education, health, and infrastructure (electricity, water and sanitation) that are deemed necessary for an individual to realize her/his human potential later in life (de Barros et al., 2009). A key feature of the HOI is that it considers both the overall coverage rates of these services and how equally the coverage is distributed, by measuring the extent to which those without coverage are concentrated in groups with particular circumstances, such as economic status, gender, parental education, or ethnicity (Box 4). From a policy perspective, the HOI is a useful measurement tool. If the overall objective is not only to widen coverage but also to promote the equality of basic opportunities for children, then public policy should be oriented toward directing marginal investments to increase basic opportunities for the most disadvantaged groups. 13 Box 4. HOI Methodology The HOI measures the availability of services that are necessary to progress in life, adjusted by how unequally the services are distributed among different groups in the population. It shows how children’s access to services necessary for a productive life, such as education or clean water (also termed as opportunities), is influenced by circumstances out of their control, such as gender, caste, parental income, or place of birth. The formula to compute this index is HOI = ̅ (1 − ), where is the average coverage of an opportunity and is the Dissimilarity Index (henceforth, D-index) that measures inequality in opportunity due to circumstances that are exogenous. D is defined by 1 = � |, � | − 2 =1 where is the coverage rate for group k (where each group is defined by a set of circumstances unique to that group), and is the share of group k in total population. The D-index is then the weighted average of absolute differences of group-specific access rates () from the overall average access (or coverage) rate (). In practical terms, the D-index reflects the amount by which the coverage of an opportunity needs to be discounted to arrive at the HOI. By construction, the D-index varies between 0 and 1. A value of 0 indicates that access rates for all groups considered are the same, while positive values indicate that certain groups of individuals have a lower probability of access to the service considered. As an example, consider two countries, A and B, and consider a basic opportunity such as access to primary education. Suppose in both countries, 50 percent of all children go to school. From the perspective of overall coverage, both countries look alike. Now suppose that in country A, no girl attends school, but in country B, 50 percent of both girls and boys attend school. The HOI discounts the coverage rate of 50 percent in country A because access is more unequal. For country B, there is no discounting because there is no gender inequality, making the HOI 50 percent, or equal to the coverage. Because country B has a higher HOI, it is more equal than country A, even though the enrolment rate is the same in both countries. Over time, changes in the HOI of a country can vary due to an increase in provision of services to all (‘scale effect’) by distributing services more fairly (‘equalization effect’) across groups, or by a change in the composition of the circumstance groups (‘composition effect’). Source: Paes de Barros et al., 2009. The main objective of HOI analysis in Pakistan is to assess whether improvement in access (coverage) that has occurred over time has contributed to equalizing opportunities for Pakistani children whose circumstances at birth differ in terms of gender, education of the household head, composition of the household (household size), and geographical area of residence (urban or rural area, province). 13 To this end, data from the 2004–05 and 2014–15 Pakistan Social and Living Standards Measurement (PSLM) surveys have been used. 14 The set of opportunities considered in the analysis spans education outcomes and access to basic household infrastructure (Table 1). 13 Unfortunately, data do not allow to account for ethnicity, religion, or caste, which could have provided further insights on the extent of horizontal inequalities and exclusion. 14 These two surveys were chosen because they allow the analysis up to the district level and provide observations over roughly a decade, which should be long enough to detect progress. 14 Table 1. Opportunities Considered in the HOI Analysis Category Opportunity Definition Basic Improved water Household has access to improved water sources (piped infrastructure water, hand pump, motorized pump/tube well, covered well, or filtration plant) in the dwelling (children ages 0-14 years) Improved sanitation Household has access to improved sanitation (that is, flush toilets connected to public sewerage or septic tank/pit) in the dwelling (children ages 0-14 years) Electricity Household has access to electricity (that is, uses electricity for home lighting) in the dwelling (children ages 0-14 years) Education Attending primary Child is currently attending primary school (children ages 6- school 10 years in Class 1 to Class 5) Attending middle Child is currently attending middle school (children ages 11- school 13 years in Class 6 to Class 8) Finished primary Child completed at least five years of education (children school on time ages 10-12 years completed Class 5) As shown in Figure 7, access to basic opportunities for Pakistani children is far from universal, particularly when looking at education opportunities and access to improved sanitation. Finishing primary school is the opportunity in which Pakistan performs the worst, followed by middle-school attendance and improved sanitation. Compared with other countries in South Asia, the Middle East, North Africa and Latin America, children in Pakistan are particularly deprived in terms of education opportunities (Table 2). Equally worrisome is that progress over time has been marginal, both in terms of coverage and the HOI. Marginal gains are noticeable only for opportunities related to access to electricity and improved sanitation. Figure 7. Coverage and HOI Trends, National Level, 2004-14 100 Panel a: Coverage Panel b: HOI Electricity 80 Improved Water Finished Primary… Attending Primary Attending Middle… 60 School Attending Primary… Improved 2014 Sanitation Electricity 40 Attending Improved Sanitation Middle School 20 Finished Improved Water Primary 0 20 40 60 80 100 School 0 2014 2004 0 20 40 2004 60 80 100 Source: World Bank staff calculations based on PSLM 2006 and 2014. 15 Table 2. International Comparison of Access to Opportunities Finishing Access to Access to Access to Attending primary electricity water sanitation school school Coverag Coverag Coverag Coverag Coverag HOI HOI HOI HOI HOI e e e e e Pakistan 90.4 92.8 84. 88.5 42. 52.1 61. 67.0 21. 25.2 3 6 0 2 Banglades 40.0 50.0 97. 97.0 38. 45.0 — — 45. 53.8 h 0 0 7 Bhutan 63.0 72.0 95. 96.0 29. 38.0 — — 27. 36.4 0 0 7 India 60.0 68.0 80. 83.0 27. 36.0 — — 58. 67.2 0 0 1 Maldives 100. 100.0 78. 86.0 93. 94.0 — — 94. 95.5 0 0 0 2 Nepal 69.0 75.0 86. 88.0 32. 37.0 — — 52. 60.1 0 0 2 Sri Lanka 81.0 85.0 86. 88.0 87. 90.0 — — 96. 97.6 0 0 6 LAC 90.9 93.9 62. 69.2 65. 73.1 94. 95.5 72. 77.0 5 4 2 0 Brazil 99.2 99.5 88. 93.0 89. 92.6 98. 98.4 65. 71.7 9 3 0 1 Mexico 98.6 99.1 55. 66.0 84. 89.2 94. 93.1 91. 93.1 6 0 5 1 Egypt — — 92. 94.2 48. 59.3 93. 95.2 80. 83.4 3 5 2 2 Jordan — — 63. 66.6 43. 53.4 85. 87.5 99. 99.3 2 6 9 1 Morocco — — 32. 50.2 26. 46.5 76. 82.8 41. 49.8 1 9 5 0 Iraq — — 76. 84.4 14. 23.4 76. 80.1 76. 80.6 4 7 3 2 Tunisia — — 71. 82.1 36. 53.6 93. 94.7 68. 73.7 6 7 0 9 West Bank — — 90. 93.3 48. 56.3 97. 98.3 88. 90.0 and Gaza 7 7 9 3 Note: LAC = Latin America and the Caribbean. Further analysis shows that changes in coverage between 2004 and 2014 accounted for about half of the progress in the HOI. The equalization component was relatively stronger for access to basic infrastructure (sanitation, water, and electricity), while composition effects 15 were more relevant in the case of education opportunities. Overall, household heads’ education and location are the most important circumstances that determine inequality in access to education and 15 See Box 4. 16 infrastructure services. As shown in Figure 8, differences in where children are born (province and urban/rural location) are more relevant for inequality of opportunities in basic infrastructure, while the education of the household head is the most relevant circumstance for inequality in education opportunities. The strong association between household head characteristics and education opportunities suggests an intergenerational transmission of disadvantage, which contributes to the perpetuation of poverty and inequality over time. Remarkably, the relative contribution of circumstances in explaining inequality in access to education and basic infrastructure has not varied significantly over time (Figure 8). The notable exception to this is the growing significance of belonging to a certain province (relative to urban or rural residence) in accounting for inequality in access to water and sanitation. 16 Gender is particularly relevant for primary school attendance, where it accounts for an increasing share of inequality of opportunity over time. 17 Figure 8. Contribution of Circumstances to Inequality of Opportunity, 2006-14 100 5.0 9.9 9.0 9.6 15.8 17.5 17.7 15.1 21.8 80 25.7 48.2 23.8 26.9 26.9 18.7 19.4 67.7 60 81.1 41.5 53.1 41.4 40 48.2 56.6 29.3 48.3 55.2 55.6 56.7 20.6 36.2 20 35.1 8.6 29.1 21.2 10.8 7.6 13.5 15.9 9.6 6.1 6.2 0 2004 2014 2004 2014 2004 2014 2004 2014 2004 2014 2004 2014 Improved Water Improved Electricity Attending Primary Attending Middle Finished Primary Sanitation School School School Gender HH Size Head's education Urban or rural Province Source: World Bank staff calculations based on PSLM 2004 and 2014. Note: The contribution of inherited circumstance to inequality of opportunities is computed applying Shapley decomposition on the D-index. The HOI analysis reinforces evidence of sharp inequalities across provinces. Inequality of opportunities is highest in Sindh and Balochistan (Figure 9). Not only do these two provinces have the lowest level of coverage, they also have the highest inequality in the distribution of access based on circumstances at birth. Across all provinces, location (urban or rural) remains the most important circumstance affecting access to basic household infrastructure. In Punjab and Sindh, the 16This finding is in line with evidence of different pace of rural poverty reduction across provinces presented in Section 1. 17Gender contribution to inequality of education opportunities is higher in Pakistan than in Latin America and the Caribbean and most of other Middle Eastern and North African countries. Compared with other countries in South Asia, the contribution of gender to inequality of opportunity in primary school attendance in Pakistan is second only to estimates for Afghanistan (Rama et al., 2015). 17 education of the household head is the most significant determinant of access to educational opportunities. In contrast, gender is by far the most important in KP and Balochistan, where girls’ education suffers the most because of social norms (Figure 10). 18 Figure 9. HOI, by Province, 2014 Improved Water 100 80 Finished 60 Improved Primary 40 Sanitation School 20 0 Attending Balochistan Middle Electricity School KP Punjab Attending Primary Sindh School Source: World Bank staff calculations based on PSLM 2014. Figure 10. Contribution of Gender to Inequality of Educational Opportunities Across Provinces, 2014 70 60 50 40 Percent 30 20 10 0 Punjab Sindh KP Balochistan Attending Primary School Attending Middle School Finished Primary School Source: World Bank staff calculations based on PSLM 2004 and 2014. Equality of opportunities varies substantially across districts, along with district poverty rates. In general, across all opportunities considered, poverty and equality of opportunity are 18In KP, gender accounts for 59, 49, and 22 percent of inequality of opportunities in primary school attendance, middle school attendance, and primary school completion, respectively. Corresponding numbers in Balochistan are 34, 26, and 31 percent. 18 negatively correlated, and poorer districts tend to have lower HOI scores. Between 2004 and 2014, the HOI improved in most of the districts, across all opportunities considered (Figure 11). Analysis of district-level HOI distribution shows a declining trend in inequality, as measured by the Gini coefficient, between 2004 and 2014. This equalizing trend is driven by the performance of Punjab and KP, while Sindh and Balochistan are once again lagging. Looking at the distribution of district HOI in 2014, inequality of education opportunities across districts within a province is higher than inequality between provinces, while the opposite holds true for inequality in access to basic infrastructure. Figure 11. District-level HOI, 2004-14 Attending Primary School Attending Middle School Finished Primary School 100 50 2014 hoi 0 Electricty Improved Sanitation Improved Water 100 50 0 0 50 100 0 50 100 0 50 100 2004 hoi Graphs by opportunity Source: World Bank staff calculations based on PSLM 2004 and 2014. The picture that emerges from this analysis is that of a country in which basic opportunities in life are largely predetermined by inherited circumstances. Progress over time has been marginal and mostly achieved by expanding coverage, with limited progress on equalizing opportunities. Even more concerning is the fact that, due to data limitations, these HOI estimates are likely to only provide a lower bound for the true extent of inequality of opportunities. The inclusion of other inherited circumstances, such as belonging to a religious minority or to a particular caste or ethnic group, could further lower HOI measures. Similarly, the analysis does not consider how circumstances at birth might affect the ‘quality’ of the opportunities considered. For example, the quality of education services is likely to vary considerably across population subgroups defined by circumstances at birth. Evidence collected under the Learning and Educational Achievement in Punjab Schools (LEAPS) study shows substantial gaps in quality of education between public and 19 relatively less affordable private schools. 19 Not only access to good quality of education might depend on the welfare status of the household but—given uneven geographical distribution of private schools—being born in a particular location could also influence children’s learning outcomes, with long-lasting implications in terms of opportunities later in life. Similarly, the quality of basic infrastructure is substantially affected by location. As discussed in a recent study (Mansuri et al., 2018), access to improved water and sanitation varies widely, together with its quality. For example, 58 percent of households connected to piped water have more than 6 hours of water a day in Punjab, compared with just 7 percent in Sindh and 2 percent in Balochistan. In general, access is much higher in urban areas, notably in provincial capitals, which benefit from most of public resource allocation. From 2009 to 2014, the total per capita allocation for Lahore was almost 18 times higher than the average of all other Punjab districts combined. In KP and Balochistan, the gap was slightly smaller (9 to 18 times higher) but still large. The gap was largest in Sindh, with Karachi receiving almost 100 times more in per capita terms than other districts in the province. The relationship between district poverty and allocations also remains weak, even if provincial capitals are excluded. In Punjab and Sindh, in particular, poorer districts tend to receive smaller allocations than richer ones. 19 Children in private schools score significantly higher than those in government schools, even when they are from the same village: by the time children in private schools are in Class 3, they are 1.5 to 2.5 years ahead of government school students in terms of learning (Andrabi et al., 2008). 20 Chapter 4: Aspiration Gaps and Intergenerational Transmission of Gender Inequality Inequality of outcomes and aspirations critically affect future-oriented behavior, notably effort made and investment decisions. Consider again Shaheen and Amir, and suppose they both dream of becoming doctors and one day being able to save lives. To realize this dream, they will pursue their studies with dedication and effort. In an ideal world with perfect capital markets and gender-neutral social norms, Shaheen’s and Amir’s parents will support their children’s dream and invest in long years of education, knowing that, one day, a doctor’s earnings will repay for the investment made. But what is the impact of existing social, economic, and gender inequalities on aspirations? In other words, what if Shaheen who is a girl from a poor household in rural Balochistan does not even dream of becoming a doctor because of the circumstances she was born in? A growing body of evidence suggests that existing inequalities can lower the aspirations of individuals in disadvantaged groups and contribute to the transmission of privilege and disadvantage across generations. Moreover, aspirations interact with social hierarchies and norms, reinforcing their impact on outcomes and reproducing them over time. Box 5. Aspirations and Outcomes In his influential theory on aspirations, Appadurai (2004) refers to an aspirations window, or the set of similar individuals whose lives and achievements help form one’s future goals. The aspirations window is shaped by multiple reference groups, including fellow family members, peers, neighbors, and others with whom one interacts on a regular basis. However, to the extent that the composition of these reference groups is influenced by the individual’s socioeconomic status and by social hierarchies and norms, the differences in the capacity to aspire reflect and reinforce existing inequalities. Parental aspirations have been found to play a crucial role in shaping children’s educational outcomes, either directly through parental investments, or indirectly, by affecting children’s aspirations and effort. For example, in Andhra Pradesh, India, parental aspirations have been associated with greater parental investment in education and a higher probability that a child is enrolled in private school and will exhibit better educational achievement at age 15 (Galab et al., 2013; Serneels and Dercon, 2014). Evidence also suggests that children of a low socioeconomic status might have lower aspirations, and that rigid social hierarchies may contribute to or compound the effects of such an aspirations gap. An experiment in India shows that providing cues to one’s place in the caste order influences the ability of low-caste boys to learn and the willingness of high-caste boys to expend effort (Hoff and Pandey, 2014). Research in Pakistan suggests that long-standing social hierarchies discourage school enrolment among children, especially girls, in lower- status social groups (Jacoby and Mansuri, 2015). Parental aspirations have been found to favor boys in Ethiopia and India, and girls in Vietnam, biases that are mirrored in the aspirations of boys and girls in these countries (Dercon and Singh, 2013). In Pakistan, the interplay between aspirations and social norms in perpetuating disadvantage over time is particularly evident in the context of gender inequalities. Gender equality and women’s participation in all spheres of public life are enshrined in the Constitution of Pakistan. Nonetheless, despite several initiatives to safeguard the interests of women by successive governments, women’s empowerment indicators continue to fall short of 21 projections. Pakistan ranks 143 out of 144 countries on the World Economic Forum’s Global Gender Gap Index. Female literacy is as low as 48 percent, 25 percentage points lower than male literacy. 20 Female labor force participation, while it almost doubled from 13.3 percent in 1992 to 25 percent in 2014, remains one of the lowest, not just in South Asia but globally (World Bank, 2018a). Underlying these dismal statistics is the rigidity of the form of patriarchy that women and men experience in Pakistan. Scholars over the years have documented the existence of ‘classic patriarchy’ in South Asia (Kandiyoti, 1998). Under this form of patriarchy, men are considered not only to be superior to women in all aspects of life, but also control women throughout their life cycle: as daughters, wives, mothers, and mothers-in-law. This system provides incentives to devalue women and girls, whose agency is thereby severely limited (Solotaroff and Pande, 2014). In Pakistan, for instance, despite improvements in gender equality in socioeconomic indicators, 21 women and girls remain severely restricted in their choices for mobility, education, marriage, and employment, and are exposed to gender-based violence (GBV). 22 Patriarchal social norms create and perpetuate gender inequality, not only in agency and outcomes but also in aspirations for the future, as evidenced by a recent study. As part of the evaluation of a participatory development program, 23 mothers and fathers were asked about aspirations for their unmarried sons and daughters in terms of education, employment, marriage, and fertility. Similar questions were also asked directly to adolescent girls and boys in the households to explore any mismatch in aspirations, and the evolution of aspirations over time and across generations. As shown in Table 3, parents have greater aspirations for their sons than their daughters. On average, mothers aspire for 3 years more education for sons than for daughters and hope to marry their daughters 3 years earlier but are much less inclined to let them have a say in the choice of their spouse, compared with sons. Almost all mothers would allow their sons to work, work for a nongovernmental organization (NGO), or run for an election, whereas they do not share the same aspirations for their daughters. Only 51 percent say their daughter would be allowed to work. Interestingly, fathers have higher educational aspirations for their children than mothers. A higher 20 Based on PSLM 2014-15 data. 21 Evidence suggests that improvements in gender equality in socioeconomic indicators may, at least in the short run, exacerbate this situation of rigid patriarchal control. Klugman et al. (2014) show that when endowments (such as health and education) and economic opportunities become more equitable, social norms can be particularly hard to dislodge and can impose limitations on other gender-equitable outcomes or overburden women with increased expectations of responsibilities and labor. 22 Data on myriad forms of violence against women and girls in Pakistan are unreliable and not up to date. However, it is known that such violence is very high. For instance, one in three ever-married women (ages 15-49 years) has experienced physical and emotional spousal violence in their lives (Pakistan Demographic and Health Survey 2012-13); other studies estimate these figures to be much higher. 23 As part of a randomized trial of a participatory development program, the authors surveyed 139 rural villages across five districts in Pakistan (Bahawalpur, Mianwali, Hyderabad, Tando Muhammad Khan, and Nowshera). From each village, 24 households were randomly selected and interviewed at baseline in 2010, midline in 2013, and endline in 2016. The analysis presented in this note covers only control villages that did not receive the intervention (44 in total). It excludes Nowshera, where the endline survey could not be done in 2016 due to conflict. It thus looks at the status of aspirations in the absence of any intervention (Gine and Mansuri, 2018). 22 proportion of fathers (56 percent) would allow their daughters to work and, on average, fathers aspire for one extra year of education for their children than mothers. Table 3. Parents Aspirations for Children Mothers’ aspirations Fathers’ aspirations for: for: Daughters Sons Daughters Sons Preferred years of education 7.28 10.15 8.51 11.74 Expected age at marriage 21.05 23.70 20.64 23.32 Child will have some say in choice of spouse (%) 13.10 51.30 2.60 40.00 Preferred number of children 4.37 4.75 3.75 4.05 Prefer more male children (%) 20.50 34.50 31.40 35.50 Allowed to work (%) 50.50 97.30 56.40 99.40 Allowed to work for NGO (%) 33.50 89.20 35.80 92.80 Allowed to contest local elections as candidates 29.60 81.90 22.40 67.10 (%) Source: Gine and Mansuri 2018. The gender aspiration gap is confirmed in adolescents’ responses (Table 4). When asked about aspirations for their own education, the number of years that girls say that they would like to study is about 2 years less than what is aspired for by their male siblings, but higher than what is expected of them by their mothers or fathers. Meanwhile, while the desire to work is nearly universal among boys, only half of the adolescent girls express a desire to work either before or after marriage, in line, albeit slightly less, with their parents’ aspirations. Similarly, adolescent boys are much more likely than girls to say they would work for a local NGO (84 vs. 26 percent) or consider running for office (60 vs. 11 percent). Table 4. Adolescents’ Own Aspirations Female Male Preferred years of education 9.16 10.98 Child's preferred age of marriage 22.17 22.83 Will choose spouse themselves (%) 4.70 20.00 Preferred number of children 3.46 4.25 Prefer more male children (%) 17.00 34.30 Wants to work (before marriage) (%) 51.90 97.90 Wants to work (after marriage) (%) 47.20 97.10 Wants to work for NGO (%) 26.40 84.30 Would like to run for elections (%) 11.30 60.00 Source: Gine and Mansuri, 2018. Adolescents were also asked what they think their parents would want for them in terms of marriage outcomes. Girls want to get married at around 22 years, but they think their parents will marry them off at the age of 20. Indeed, girls are accurate at predicting their parents, who say their daughters will be married before the age of 21, on average. 23 Results from the analysis also confirm that parents’ and adolescents’ aspirations are closely tied to individual and household characteristics, such as poverty and education. Mothers and fathers with some formal education desire significantly more education for their children, although they maintain higher aspirations for the education of boys than girls. Parents’ education also affects their daughters’ aspirations. Girls whose parents have some formal education aspire to achieve the same level of education as their male siblings. The impact of parental education on employment aspirations for girls is also strong, particularly in the case of maternal education. A higher share of mothers with some education would let their daughters work, and a higher share of girls whose mothers have some education are willing to participate in the labor market, either before or after marriage. The impact of parental education on aspirations and on the intergenerational transmission of gender norms reinforces the evidence discussed in previous sections. The analysis also suggests the importance of considering the role played by social norms and aspirations to improve the effectiveness of social policies aimed at breaking the cycle of inequality. 24 Chapter 5: Poverty and Conflict in Pakistan The relationship between poverty, inequality and conflict is the object of long-standing debates across the social sciences.24 While few doubts remain that armed violence adversely affects economic development and the living conditions of individuals and communities, the question of whether poverty and inequality can cause conflict and violence remains unresolved due to the complexity of factors at play. Several scholars have argued that persistent inequality among and between socioeconomic groups may increase social discontent and, eventually, the propensity of individuals and groups to engage in crime and political violence. 25 Others posit that it is the relative deprivation and the exclusion of segments of society from sharing the benefits of economic progress that fuel discontent and create a fertile ground for conflict to flourish (Hirschman and Rothschild, 1973). Since its inception, Pakistan has been plagued by violence and conflict. Armed forces have been involved in fighting home-grown, as well as foreign, terrorists in the north-western region, while terrorist and insurgent groups have repeatedly attacked civilian and non- civilian targets in many areas in the country. Since Pakistan joined the global War on Terror in 2001, violence in the country has increased dramatically, peaking in 2009 when the army intensified operations to clear militants from Federally Administered Tribal Areas (FATA) (Figure 12. Between 2001 and 2011, conflict claimed the lives of 35,000 people—a situation that many observers have likened to a civil war (Lieven, 2011). According to government estimates, the direct and indirect cost incurred by Pakistan due to incidents of terrorism during the past 16 years amounted to US$123.13 billion (Pakistan Economic Survey 2016-17). Figure 12. Trends in Security Incidents, 1988-2017 4000 3500 3000 2500 2000 1500 1000 500 0 BFRS PIPS Source: World Bank staff calculations based on BFRS and Pakistan Institute for Peace Studies (PIPS) data. 24 See Ray and Esteban (2017) for a critical review of the literature. 25 See discussion in Boix (2003, 2015) and Fajuzylber et al. (1998). 25 According to the latest available data, 26 the number of security incidents has progressively declined over the past 5 years, mostly due to the normalization of army operations in FATA and consequent sharp decline in security incidents involving militants. The decline in violence related to nationalist insurgents and of violence-related ethnic, sectarian, and communal disputes was less sharp, with Balochistan taking the lead as the most conflict-affected province of Pakistan (Figure 13). In 2017, about 30 percent of incidents were related to nationalist or ethnic/sectarian/communal disputes, and 42 percent of incidents occurred in Balochistan against 37 percent in FATA and KP combined. Figure 13. Trend in Security Incidents, 2013-17 3000 3000 2500 2500 2000 2000 1500 1500 1000 1000 500 500 0 0 2013 2014 2015 2016 2017 2013 2014 2015 2016 2017 Criminal Militants Balochistan FATA KP Nationalist insurgents Ethnic/Communal/Sectarian Punjab Sindh Border clash Source: World Bank staff calculations based on BFRS and PIPS data. Recent changes in the profile and geography of violence put issues related to social cohesion, poverty, and inequality at the forefront of the debate on addressing conflict and violence in Pakistan (Box 6). The 2018 National Internal Security Policy (NISP) well reflects the complexity of the security challenge and in fact envisions a “peaceful, democratic and inclusive society forged by the promotion of the rule of law, inclusive growth, political stability, and respect for diversity,” and identifies in “providing social justice” a cornerstone in the quest for internal stability, which is defined “not simply as absence of violence, but as the presence of conditions necessary for enhancing social cohesion and resilience.” 26Data provided by the PIPS. The dataset compiles information from various sources including newspapers, magazines, journals, field sources, and screening of official record. More than 30 English and Urdu dailies, magazines, and journals, both national and regional, and various television news channels are monitored to update the database and archives. The dataset spans the period 2006-07 to the present. 26 Box 6. Conflict and Poverty in Pakistan Research on the relationship between poverty, inequality and violent conflict in Pakistan has been mostly qualitative, and the results are often contrasting. Some studies suggest that poverty and various forms of inequality contributed to feelings of estrangement and resentment among the worse-off, and to weakening of the social contract, which resulted in increased support for insurgents, religious militancy, and radicalization. Other studies cast doubt on these findings, providing evidence from small-sample studies on the education profile of militants and returning jihadis, or by analyzing support for militant organizations through public opinion surveys or constituency-level election results. Zaidi (2010) conducted a small-sample survey throughout Pakistan to understand the drivers of radicalization. Zaidi’s results suggest that poverty may promote radicalization in Pakistani society. Azam and Aftab (2009) argue that the inability of the state and society to address various forms of inequality in Pakistan has led to feelings of estrangement and resentment among the worse-off, who—the study argues—may be more inclined toward supporting insurgencies. Malik (2009) presents evidence on how socioeconomic deprivation, horizontal inequalities, and the weakening of the social contract between the state and citizens may have contributed to mobilizing support for religious militancy and radicalization in Pakistan. Other studies provide contradictory evidence. Abbas (2007) surveyed 517 men who had left Pakistan to support the jihad in Afghanistan and were being held in jails in Pakistan’s KP province following their reentry into Pakistan after 2001. From his small-sample study, Abbas concluded that the jihadis were mostly educated in public schools, had some form of employment, and were drawn from mainstream Pakistani society. Similarly, in her study of families of 141 militants who died in conflicts in Afghanistan and the Kashmir Valley, Fair (2007; 2008) also found that more than 90 percent of militants had attended public schools (as opposed to madrassahs) and about half of them had some form of employment before their departure to Afghanistan. Fair et al. (2012) conducted a large-scale public opinion survey in Pakistan to consider the relationship between support for militant organizations and religious views, and found that neither religious practice nor support for political Islam is related to support for militant groups. More recently, Blair et al. (2013) estimated the causal effects of feelings of relative poverty and perceptions of violence have on individual support for militant groups using an original, large-scale survey experiment. Results show that feelings of relative poverty are negatively related to support for militants. This dislike is strongest among the urban poor, particularly those living in the most violent urban districts. In addition, Bullock, Imai and Shapiro (2011) show that individuals’ level of education and income have an insignificant impact on the probability that individuals support militant groups in Pakistan, once the empirical analysis takes into consideration regional differences in support levels. In a recent study, Justino et al. (2018) empirically test the effect of poverty on the incidence and intensity of political violence in Pakistan, using comprehensive district-level panel data constructed by collating welfare data at the district level obtained from PSLM surveys and data on different manifestations of political violence (BFRS) over the period 2004-10. The empirical results show overall a lack of any systematic relationship between poverty levels, and the incidence and intensity of political violence across districts in Pakistan, suggesting the complexity of factors at play. We find a small and only marginally significant negative effect of poverty in some specifications, meaning that for the large part of the models considered, and contrary to general views, higher poverty districts are less affected by violence. The only exception is the case of Punjab, where results show a marginally positive relationship between poverty and the number of terrorism events and the number of dead in guerrilla attacks on security services. As discussed in previous sections, inequalities between provinces, between urban and rural areas, and between lagging and leading districts within provinces, are on the rise and will likely continue to deepen if left ungoverned. Underlying urbanization, forces increasing competition on scarce natural resources (notably water) due to population growth and governance 27 failures, and the compounding effect of climate change 27 (Box 7) are all likely to put further stress on Pakistan’s resilience to conflict and violence. However, recognizing the importance of a more equitable and inclusive process of development not just to achieve sustainable growth but also for peace and security provides an opportunity for action moving forward. Box 7. Climate Change and Conflict A growing literature links changes in temperature and precipitation patterns to increasing crime rates, civil conflict, intergroup riots, migration, and mortality. It is possible that these could be triggered by monetary effects such as negative rain shocks that lower income which, in turn, increases the likelihood of crime, violence, and civil conflict. Through a meta-analysis of the literature, Hsiang, Burke, and Miguel (2013) find strong causal evidence linking climatic events to human conflict across a range of spatial and temporal scales and across all major regions of the world. They find that the magnitude of climate’s influence is substantial: for each 1°C increase in long-term average temperatures or 1 standard deviation increase in extreme rainfall, interpersonal violence rises by 4 percent and the frequency of intergroup conflict rises by 14 percent. Because locations throughout the inhabited world are expected to warm 2°C to 4°C by 2050, the researchers argue that the amplified rates of human conflict could represent a large and critical impact of anthropogenic climate change. The relationship between climate change and conflict is an active area of research, and several groups are currently trying to reproduce the results of Hsiang, Burke, and Miguel (2013). Analysis of crime, agriculture, and weather data from India from 1971 to 2000 shows that drought and heat exert a strong effect on virtually all types of crimes, with the effect on property crimes being greater than violent crimes (Blakeslee and Fishman, 2017). This relationship is relatively stable over three decades of economic development. They also find the effects of income shocks on crime are highly non-symmetric: although negative agriculture shocks consistently lead to increases in crime, positive agriculture shocks do not result in a decline in crime. The researchers conclude that despite the effects that accompany economic growth—higher incomes, greater access to consumption smoothing instruments, and reduced susceptibility of agriculture to climatic variability—there is little evidence that crime has become less responsive to extreme rainfall than it was before the improvements. Source: Mani et al., 2018. 27According to Mani et al. (2018), about 49 million people in Pakistan today (or more than 25 percent of the country’s population) live in locations that would become climate vulnerability hotspots by 2050 under intense climate change. The most economically productive areas will be the most affected: Sindh Province emerges as the most vulnerable hotspot in Pakistan, followed by Punjab. Overall, the authors report that increase in temperatures and changes in precipitation patterns will have a negative impact on living standards in Pakistan. Consumption expenditures are expected to drop by 2.0 to 2.9 percent, depending on the magnitude of climate change, and an additional 5.7 to 21.4 million people could be pushed into poverty by 2050. 28 Policy Discussion The government of Pakistan has been implementing its Poverty Reduction Strategies (PRSP-I and PRSP-II) since 2001. Poverty reduction lies at the center of both policy frameworks. In PRSP-I, poverty reduction was to be achieved by accelerating growth and by targeting social assistance to protect the poor and the vulnerable against negative shocks. 28 PRSP-II broadened the previous strategy with greater attention to pro-poor growth, more focus on employment generation and labor-intensive sectors, and a reorganization of the social protection system around the then newly introduced Benazir Income Support Program (BISP). A policy focus on poverty may have contributed, together with economic development, to the progress achieved since 2001. Nonetheless, the evidence discussed in this note suggests that prosperity is increasingly not being shared equally in the population, and that a sizeable portion of Pakistan’s immense human capital potential is being wasted due to the lack of economic mobility, inequality of opportunities, and aspiration gaps. The Pakistan Vision 2025 paper recognizes the importance of a development process based on inclusiveness and social justice. Quoting the words of the Minister of Planning and Development at the time, Prof. Ahsan Iqbal, “[…] development has to be of people, for people and by people.” The strategy articulated in the first pillar “Putting people first: developing human and social capital,” while calling for an increase in public spending on human capital, falls short of recognizing the need for leveling the playing field to ensure opportunities are equalized early in the lives of Pakistani children. The inclusion of vulnerable segments of society should also go beyond devoting additional resources to the social protection system and focus on equalizing opportunities. It is currently estimated that 22 million children are out of school, of which 5 million are of primary school age. Many, if not the majority, are similar to Shaheen, born to parents with poor educational backgrounds and living in the most deprived areas of the country. Shaheen could have been a doctor, but she will not due to the lack of aspirations and opportunities. Shaheen will be a mother one day and her children will likely follow the same path. Breaking inequality traps requires not only an increase in the resources devoted to human development but also stronger efforts to target lagging areas and marginalized segments of the population. Improving opportunities to those who are disadvantaged might have higher marginal cost, both in monetary and political terms compared with a ‘business-as-usual’ scenario. However, it is critical to making Pakistani society more inclusive and to making sure that all young men and women of today will have equal chances to realize their potential and contribute positively to Pakistan’s socioeconomic development for the decades to come. Bringing Pakistan on the path to a more inclusive development process requires a comprehensive strategy, which can be articulated in four main pillars: (i) mobilize resources by 28Spending on the 17 pro-poor programs identified in the PRSP increased from PKR 1,300 per capita in 2001, just before the launch of the Poverty Reduction Strategy, to PKR 3,600 per capita in 2007–08. As a percentage of GDP, PRSP expenditure increased from 2.9 to 5.5 percent between 2001 and 2007/08. The share of PRSP programs in overall spending rose from 16 to 25 percent over the same period (World Bank, 2008). 29 increasing efficiency and equity in revenues collection and public spending; (ii) improve data quality and availability to support evidence-based policymaking; (iii) invest in policies aimed at equalizing opportunities; and (iv) track the impact of policies and adjust for better outcomes. Mobilize resources by increasing efficiency and equity in revenues collection and public spending Creating the fiscal space for increasing investments in human capital is a key priority. Efforts to increase revenue mobilization and efficiency of public spending should go hand in hand with improving fairness in the overall system. On the revenue-collection side, expanding the tax base should be achieved at the same time as reducing inequities (such as exemptions, preferential treatments to some sectors, and outdated property valuation tables) and the overall progressiveness of the tax system. In terms of public spending, a greater focus should be devoted to transitioning from a system of universal subsidies, which is inherently regressive, to one in which transfers are targeted to vulnerable segments of the population only. Such a policy shift will not only save Pakistan’s government a substantial amount of resources, but it will also reduce inefficiencies and distortions that curb investment and growth, encourage efficient and sustainable use of resources, and discourage unproductive rent-seeking behaviors. Increasing progressivity of spending should go hand in hand with correcting imbalances between capital and recurrent expenditures, improving operational efficiency, and strengthening accountability to ensure higher spending translates into better services for the poor and, ultimately, better outcomes on the ground. Efforts to mobilize resources and improve the system’s progressivity should be anchored on solid analytical foundations to prioritize areas of intervention, and assess the scope and impact of proposed reforms. Improve data quality and availability to support evidence-based policymaking Access to accurate, reliable, and timely statistics is vital for policymakers, who require a solid base of evidence to support the design and evaluation of policies. First, promoting social and economic inclusion of all the people of Pakistan will require mainstreaming data collection on topics related to gender, disability, ethnicity, religion, and nationality in Pakistan’s social statistics. This will allow for a comprehensive profile of identity-based vulnerabilities. Second, the capacity of PBS should be strengthened, and its mandate should be enhanced to support greater data integration. In particular, PBS should play a central role in coordinating data compilation across line ministries, government entities, and different government layers in the production of official statistics29 to ensure a better flow of information and ultimately better management of resources and monitoring of outcomes. Lastly, a comprehensive and modern Statistics Law 30 should be enacted to regulate PBS’s mandate and to guide the implementation of the fundamental principles of professional 29 This coordination role should include the setting of statistical standards, the adoption of agreed classification schemes, the proper deployment of statistical tools and analysis and, most importantly, the flow of data to PBS for final compilation. 30 Current legislation dates to the Statistics Act of 1975. More recent reforms, such as the General Statistics Reorganization Act of 2011, address organizational reforms, notably the creation of PBS—without providing a framework for the improvement and modernization of Pakistan’s statistical system. 30 independence, impartiality, accountability, and transparency about methods of collection, compilation, and dissemination of official statistics (Box 8). Box 8. Fundamental Principles of Official Statistics The Fundamental Principles of Official Statistics were first developed in 1991 by the Conference of European Statisticians and subsequently adopted at the ministerial level by the Economic Commission for Europe of the United Nations (ECE) in 1992. These principles, 10 in total, articulate professional and scientific standards essential to ensure that national statistical systems produce appropriate and reliable data. After a consultation process, the same set of principles was adopted by the United Nations Statistical Commission at its Special Session of April 11 to 15, 1994, as the United Nations Fundamental Principles of Official Statistics. More recently, the United Nations General Assembly, in its resolution 68/261 of January 29, 2014, endorsed the Fundamental Principles of Official Statistics, recognizing the importance of good statistics for decision-making in a democratic society and acknowledging that the principles were still as relevant today as they had been in the past. Principle 1. Official statistics provide an indispensable element in the information system of a democratic society, serving the government, the economy, and the public with data about the economic, demographic, social, and environmental situation. To this end, official statistics that meet the test of practical utility are to be compiled and made available on an impartial basis by official statistical agencies to honor citizens’ entitlement to public information. Principle 2. To retain trust in official statistics, the statistical agencies need to decide according to strictly professional considerations, including scientific principles and professional ethics, on the methods and procedures for the collection, processing, storage, and presentation of statistical data. Principle 3. To facilitate a correct interpretation of the data, the statistical agencies are to present information according to scientific standards on the sources, methods, and procedures of the statistics. Principle 4. The statistical agencies are entitled to comment on erroneous interpretation and misuse of statistics. Principle 5. Data for statistical purposes may be drawn from all types of sources, be they statistical surveys or administrative records. Statistical agencies are to choose the source with regard to quality, timeliness, costs, and the burden on respondents. Principle 6. Individual data collected by statistical agencies for statistical compilation, whether they refer to natural or legal persons, are to be strictly confidential and used exclusively for statistical purposes. Principle 7. The laws, regulations, and measures under which the statistical systems operate are to be made public. Principle 8. Coordination among statistical agencies within countries is essential to achieve consistency and efficiency in the statistical system. Principle 9. The use by statistical agencies in each country of international concepts, classifications, and methods promotes the consistency and efficiency of statistical systems at all official levels. Principle 10. Bilateral and multilateral cooperation in statistics contributes to the improvement of systems of official statistics in all countries. Source: www.unstats.un.org. Invest in policies aimed at equalizing opportunities Better data and additional resources should be used in the design and implementation of policies directly aimed at reducing spatial and social inequalities, that is, inequality across geographical areas (provinces, districts, urban/rural areas, and so on) and equalizing opportunities to the advantage of poor and marginalized segments of Pakistan’s society. Curbing spatial and social inequalities will be crucial in strengthening the social contract and ensuring a more sustainable, peaceful, and inclusive process of development. 31 Reducing spatial inequalities will require improving mechanisms that form the basis of intragovernmental transfers—federal/province/district. This could entail making development gaps more explicit in the allocation formula used for the National Finance Commission (NFC) award of federal divisible pool transfers to provinces, and improving accountability by providing adequate incentives for performance. 31 Similarly, development gaps and performance could also be formally introduced in the allocation of resources from provinces to districts. Reforms aimed at anchoring resource allocation to local development needs should be complemented by efforts to address capacity and accountability constraints that limit the effective use of public resources at the local level, particularly in lagging areas. 32 In this context, greater attention should be devoted to tailoring the modes of service delivery to the specific needs of communities in lagging areas by, for example, combining interventions aimed at addressing quantity and quality constraints on the supply side (building/improving facilities, training teachers or health workers, and so on) with others addressing constraints on the demand side (conditional cash transfers, information/outreach campaigns, and so on). Tackling social inequalities will require challenging existing structures of power and social norms on which exclusion and inequalities based on gender, ethnicity, religion, sexual orientation, disability, or other group characteristics are grounded and transmitted from one generation to the next. To this end, legislative interventions should focus on strengthening protection against discrimination and gender-based violence and on promoting greater participation of women and minority groups in the political and economic development of the country by means of affirmative action (Box 9). 31 For the first time in 2009, the 7th NFC Award for federal divisible pool transfers to provinces expanded the resource allocation formula beyond population. In particular, the formula allocated 82 percent of resources based on population, 10.3 percent based on poverty, 5 percent based on provincial tax effort, and 2.7 percent based on the inverse of provincial population density (Shah, 2012). This formula implies a share of 51.74 percent to Punjab, 24.55 percent to Sindh, 14.62 percent to KP, and 9.09 percent to Balochistan from the divisible pool. No explicit formula has yet been used to allocate resources from provinces to districts. Debate is currently ongoing in Punjab and KP. 32 As discussed in the ‘Governance’ policy note, these objectives would require strengthening the technical competencies of the public administration and strengthening local government systems. 32 Box 9. Affirmative Action: The Experience of The United States, South Africa, And India In many countries, systematic disadvantage based on gender, race, religion, sexual orientation, and disability has been tackled with affirmative action. Affirmative action is a type of anti-discriminatory measure that requires organizations and agencies to take proactive steps to remove discriminatory gaps in outcomes (such as education and employment) for marginalized groups. Affirmative action can range from interventions focusing on outcomes—such as quotas, court- imposed remedies for discriminatory practices, and legislative requirements on government contractors—to “positive action” to ensure minorities’ or women’s equal opportunities such as targeted mobilization campaigns. In the United States, the Civil Rights Act of 1964 introduced the principle of ‘affirmative action’ into the political, judicial, and administrative spheres of society. The program does not establish quotas but rather encompasses a large number of activities designed to improve the presence of minorities and women in the workforce and higher education institutions and government contracting. Studies find positive impact on increasing female and black male employment, increased minority groups’ representation in higher education institutions, and increased government contracts for minority and women-owned businesses. While evidence remains inconclusive, most of the studies looking at efficiency implications find either increased or small negative efficiency consequences (Holzer and Neumark, 2000). In South Africa, affirmative action was introduced in 1998 under the Employment Equity Act to eliminate unfair discrimination and “take positive or affirmative measures to attract, develop, and retain individuals from previously disadvantaged groups: Blacks (including African, Colored and Indians), women and people with disabilities” (Act 55, 1998). Under the Act, employers are requested to consult with their employees and representative trade unions to undertake audits of employment policies and practices in the workplace, profile the workforce, and identify any underrepresentation of designated groups. Based on such audits, employers are requested to prepare employment equity plans, proceed with their implementation, and publicly report on the progress. The Department of Labor has the right to send inspectors to compare the situation on site against reports filed and issue sanctions. In 2004, the government introduced a Code of Good Practice and scorecards to be used for government procurement, public-private partnerships, and the sale of state-owned enterprises. Studies found a positive, albeit marginal, impact of affirmative action interventions in South Africa, where discrimination drivers strongly affect pre-labor market outcomes (Burger and Jaffa, 2010). In India, affirmative action—the ‘reservation system’—is based on quotas targeted toward Scheduled Caste and Scheduled Tribes (SC/ST) groups. According to the reservation system, 22.5 percent of all government jobs, seats in educational institutions that have complete or partial government funding and electoral constituencies at all levels of government are reserved for SC and ST persons. Application of quotas has succeeded in ensuring SC/ST representation in the political sphere, while their implementation in government jobs and educational institutions has not been as effective in practice, due to the lack of a monitoring agency and of penalties for loopholes used to circumvent quota restrictions (Deshpande, 2005). Evidence also suggests that affluent families among the preferred groups have obtained most of the benefits of job reservation (Sowell, 2004), and that reservation by itself has not been able to bring about the total social transformation envisaged when quotas were mandated in the Constitution. Policies and programs aimed at fighting poverty and equalizing opportunities ex ante should also be designed to accompany the inclusion of the poorest segments of the population. These interventions should go hand in hand with strengthening access and quality of service delivery in lagging areas, such as conditional cash transfers to support: school enrolment, maternal health, and early childhood nutrition programs; awareness campaigns to strengthen accountability in service delivery and support human capital investments; and labor market interventions to support employability and income generation (training on soft skills, training cum literacy, placement services, and so on). Lastly, efforts to reduce inequality and equalize opportunities should be reinforced by programs aimed at promoting an equitable and inclusive evolution of social norms. A growing body of evidence suggests, for example, that school-based programs aimed at strengthening life skills of adolescent girls, or the provision of safe spaces for girls along with access to peer networks could 33 promote more gender-equitable attitudes and support female empowerment. Programs engaging young men and boys through group education sessions, positive role models of masculinity, and mass-media campaigns have also shown encouraging positive effect on reducing intimate partner violence and support for inequitable gender norms (Box 10). Box 10. Interventions Addressing Gender Norms A change in social norms is often hard to measure. Still, there is a growing body of evidence on programs to address social norms regarding gender. Life skills in school-based projects. Life skills programs for adolescent girls display some evidence of changed behaviors and attitudes that are more gender equitable. Life skills are usually combined with a package of interventions (vocational training, knowledge on reproductive health, nutrition, effects of child marriage and violence, and sports). For instance, a program in India trained teachers to educate young girls regarding healthy ways to respond to real-life situations (Jaya, Dhillon and Kumar, 2014). In addition, the program organized advocacy sessions to sensitize school principals and parents. It was found that girls in program schools were better able to challenge gender stereotypes. Another program (also in India) combined class life skills modules with extracurricular activities and street theater to expose young girls and boys to more gender-equitable attitudes. The Girl Shield Program in Pakistan (Croll, 2007) is another example of educating girls about their rights, enhancing girls’ ability to negotiate these rights with their parents, as well as changed parental and community perception of these girls. The Kishori Kontha Program in Bangladesh had a similar program design and similar results although they focused on out-of-school centers rather than classrooms (Scales et al., 2013). A similar program called Kishoro Bikashi in Nepal reported increased confidence among girls to make decisions about their own lives, including marriage. Female mentors and safe spaces for young girls. The provision of safe spaces for girls and access to peer networks have been shown to reduce the risks of early marriage in Ethiopia and Uganda (World Bank, 2014) and in Bangladesh (International Center on Research on Women [ICRW], 2013). Programs such as the Safe and Smart Savings Products for Vulnerable Adolescent Girls (Austrian and Muthengi, 2013) aimed to support economic and social empowerment of young girls and included weekly meetings with adult female mentors and creation of safe spaces for girls supplemented with financial literacy and support in opening a savings account. The evaluation concluded that girls who had been exposed to safe spaces combined with other training were less likely to report that they feared being raped or report being teased by people of the opposite sex. However, economic empowerment training alone led to an increase in reported sexual harassment and violence. A program in India (Baker et al., 2009) provided girls with a safe space to discuss their right to refuse early marriage, but it also taught girls specific skills to negotiate with their parents to delay their own marriage. As a result of the intervention, girls felt empowered to discuss early marriage with their parents, negotiate their marriage terms, and also speak about other taboo issues such as menstruation and sexuality. Self-help groups and peer networks. India has been home to both government and nongovernment interventions on women’s self-help groups. Self-help groups are usually economic groups of rural women that receive collective finance and enterprise and/or livelihoods. Brody et al. (2016) conducted a mixed-methods systematic review of self-help groups and found that there is evidence that economic self-help groups have positive effects on women’s economic and political empowerment. In addition, such groups also have an impact on women’s family size decision-making power and social mobility. Programs with a training component (such as financial, business, or life skills training) have a larger effect than programs that do not involve training. The review identified that some mechanisms that facilitate empowerment are gaining financial skills, the capability to speak in front of others, access to decision making in the household, and also the experience of support and solidarity from other group members. An evaluation of a program called ‘Do Kadam Barabari Ki Ore’ in Bihar, India, found that when women in self-help groups attended ‘gender transformative group learning sessions’, they upheld gender egalitarian roles (including those within marital relationships). Regarding the importance of peer networks, a program in India’s largest women’s bank (SEWA Bank) offered two days of business counseling for women and a random subsample was invited to attend with a friend. An evaluation (Field et al., 2016) showed that the intervention significantly increased participants’ business activity but only if they were trained with a friend. Those trained with a friend were more likely to have taken business loans, were less likely to be housewives, and reported increased business activity and higher household income. 34 Engaging men and boys. Impact evaluations among men and boys have demonstrated that addressing gender inequality and harmful gender stereotypes among young boys and men can change attitudes around intimate partner violence, and also lead to more gender-equitable attitudes and expectations among men (Solotaroff and Pande, 2014). Interventions typically involve group education sessions, positive role models of masculinity, and mass-media campaigns. The Yaari Dosti (Verma et al., 2008) program in India was a program for young men in urban slums who were at a higher risk of HIV. The program attempted to change harmful beliefs and attitudes around masculinity. It addressed sexual and reproductive health, violence (against women and men), substance use, and fatherhood. Participants reported less support for inequitable gender norms than control groups and self-reported a decrease in intimate partner violence. Some programs use sports as a vehicle to connect to young men, offer role models of healthy masculinity, and teach them how to control aggression and harmful masculinity norms and aggression toward young girls. For instance, a Mumbai-based program (Das et al., 2012) engaged cricket coaches and mentors to raise awareness about what constitutes disrespectful behavior and teaches skills for being an effective bystander when witnessing harmful behavior. Others focusing on norm change among boys are school-based and mixed gender. Yet another program in Brazil, Indonesia, Rwanda, and South Africa (Heilman et al., 2017) focuses on group education for young men to have positive, more gender-equitable fatherhood and to give young men tools to advocate for their own health and the health of their partners and children. Track the impact of policies and adjust for better outcomes Providing more and better services to support more inclusive growth under a tight fiscal environment requires boosting government performance and creating policies and programs that are as effective and efficient as possible. To this end, a monitoring and evaluation (M&E) system should be developed to track inputs and measure the results in terms of outputs, outcomes, and impact of government policies, programs, and projects at the national, sectoral, and subnational levels. A strong M&E system should build on a modernized statistical system that integrates and harmonizes data across administrative sources and different data providers, to ultimately produce information and services for both citizens and policy makers. Figure 14. M&E system and the policy cycle Review policy Review the past planning and budget period Develop policy and Evaluate and audit begin planning the efficiency and effectiveness of activities Set objectives, policies, strategies, and priorities Monitor activities and account for Prepare budget expenses Allocate resources Implement and manage planned activities Source: Lopez-Acevedo, Krause, and Mackay, 2012. 35 As shown by international evidence, a strong M&E system contributes to sound government at different stages of the policy cycle (Figure 14) (Lopez-Acevedo, Krause and Mackay, 2012). The early stages of the policy process—analyzing and developing government policy and planning priorities and strategies—benefit from evidence of what has or has not worked in the past (evidence-based policymaking). Information on the performance of existing government programs and on the expected performance of new programs is important for the next stage of the policy cycle—the allocation of resources in the budget. M&E information, especially evaluation findings that explain past performance, helps guide government decisions so that the most cost- effective collection of policies and programs can be adopted in the annual budget. At the next stage of the policy cycle—the implementation and management of activities funded by the budget—M&E helps managers to monitor their activities, including government service delivery and staff management, so that they learn quickly what is working and what is not. Evaluations or reviews can identify the reasons for this good or bad performance. This is the learning function of M&E, often referred to as results-based management. The final stages of the policy cycle include accountability relationships. M&E reveals the extent to which the government has achieved its objectives and thus provides the evidence needed to ensure strong government accountability to outside actors such as parliament and civil society, and within government between sector ministries and central ministries, among agencies and their sector ministry, and among ministers, managers, and staff. Strong accountability can provide the incentives necessary to improve performance. 36 References Abbas, Sohail. 2007. Probing the Jihadi Mindset. Islamabad: National Book Foundation. Andrabi, Tahir, Jishnu Das, Asim Ijaz Khwaja, Tara Vishwanath, and Trista Zajonc. 2008. Pakistan - Learning and Educational Achievements in Punjab Schools (LEAPS): Insights to Inform the Education Policy Debate. Washington, DC: World Bank. Appadurai, Arjun. 2004. “The Capacity to Aspire: Culture and the Terms of Recognition.” In Culture and Public Action, edited by V. Rao and M. Walton, 59–84. Palo Alto, California: Stanford University Press. Austrian, Karen, and Eunice Muthengi. 2013. Safe and Smart Savings Products for Vulnerable Adolescent Girls in Kenya and Uganda: Evaluation Report. Nairobi: Population Council. Azam, Muhammad and Safiya Aftab. 2009. “Inequality and Militant Threat in Pakistan.” Conflict and Peace Studies 2 (2): 33. Baker, Amber, Mutsumi Nakagami, Tara Noronha, Katherine Potaski, and Emily Puckart. 2009. A Qualitative Assessment of Girls Gaining Ground: Working towards Female Empowerment in Maharashtra, India. Maharashtra, India: Bhavishya Alliance. Blair, Graeme, C. Christine Fair, Neil Malhotra, and Jacob N. Shapiro. 2013. “Poverty and Support for Militant Policies: Evidence from Pakistan.” American Journal of Political Science 57 (1): 30– 48. Blakeslee, David S., and Ram Fishman. 2017. “Weather Shocks, Agriculture, and Crime: Evidence from India.” Journal of Human Resources 52 (3). Boix, Charles. 2003. Democracy and Redistribution. Cambridge: Cambridge University Press. Boix, Charles. 2015. Political Order and Inequality. Cambridge: Cambridge University Press. Brody, Carinne, Thomas de Hoop, Martina Vojtkova, Ruby Warnock, Megan Dunbar, Padmini Murthy, and Shari L. Dworkin. 2016. Economic Self-help Group Programs for Improving Women’s Empowerment: A Systematic Review. London: International Initiative for Impact Evaluation Bullock, Will, Kosuke Imai, and Jacob N. Shapiro. 2011. “Statistical Analysis of Endorsement Experiments: Measuring Support for Militant Groups in Pakistan.” Political Analysis 19 (4): 363–384. Burger, Rudolf and Rachel Jaffa. 2010. “Affirmative Action in South Africa: An Empirical Assessment of the Impact on Labor Market Outcomes.” CRISE Working Paper 76. 37 Croll Elizabeth. 2007. “The Girl Shield Project Pakistan: Assessment Report.” School of African and Oriental Studies, University of London, London. Das, Madhumita, Sancheeta Ghosh, Elizabeth F. Miller, Brian O’Conner and Ravi Verma. 2012. Engaging Coaches and Athletes in Fostering Gender Equity: Findings from the Parivartan Program in Mumbai, India. New Delhi: ICRW and Futures without Violence. Dercon, Stefan, and Abhijeet Singh. 2013. “From Nutrition to Aspirations and Self-efficacy: Gender Bias over Time among Children in Four Countries.” World Development 45: 31–50. Deshpande, Ashwini. 2005. “Affirmative Action in India and the United States.” Background paper for the World Development Report 2006. Fair, Christine C. 2007. “Who Are Pakistan's Militants and Their Families?" Terrorism and Political Violence 20 (1): 49–65. Fair, Christine C. 2008. Madrassah Challenge: Militancy and Religious Education in Pakistan. Washington, DC: United States Institute of Peace Press Books. Fair, Christine C., Neil Malhotra, and Jacob N. Shapiro. 2012. “Faith or Doctrine? Religion and Support for Political Violence in Pakistan.” Public Opinion Quarterly 76 (4): 688–720. Field, Erica, Seema Jayachandran, Rohini Pande, and Natalia Rigol. 2016. “Friendship at Work: Can Peer Effects Catalyze Female Entrepreneurship?” American Economic Journal: Economic Policy 8 (2): 125–153. Galab, Shaik, Uma Vennam, Anuradha Komanduri, Liza Benny, and Andreas Georgiadis. 2013. “The Impact of Parental Aspirations on Private School Enrolment: Evidence from Andhra Pradesh, India.” Young Lives Working Paper 97, Oxford Department of International Development (ODID), University of Oxford. Gazdar, Haris. 1999. “Review of Pakistan Poverty Data.” Monograph Series #9. Sustainable Development Policy Institute. Gine, Xavier, and Ghazala Mansuri. 2018. “A Seat at the Table: Experimental Evidence on Gender Inclusion and the Status of Women in Rural Pakistan.” Mimeo, World Bank. Heilman, Brian, Ruti Levtov, Nikki van der Gaag, Alexa Hassink, and Gary Barker. 2017. State of the World’s Fathers: Time for Action. Washington, DC: Promundo, Sonke Gender Justice, Save the Children, and MenEngage Alliance. Hirschman, Albert and Michael Rothschild. 1973. “The Changing Tolerance for Income Inequality in the Course of Economic Development.” Quarterly Journal of Economics 87 (4): 544–566. Hsiang, Solomon M., Marshall Burke, and Edward Miguel. 2013. “Quantifying the Influence of Climate on Human Conflict.” Science 341. 38 Hoff, Karla, and Priyanka Pandey. 2014. “Making Up People—the Effect of Identity on Performance in a Modernizing Society.” Journal of Development Economics 106: 118–131. Holzer, Harry, and David Neumark. 2000. “Assessing Affirmative Action.” Journal of Economic Literature 38: 483–568. ICRW, 2013. “Asia Child Marriage Initiative: Summary of Research in Bangladesh, India and Nepal. International Center on Research on Women. New Delhi. Jacoby, Hanan G., and Ghazala Mansuri. 2015. “Crossing Boundaries: How Social Hierarchy Impedes Economic Mobility.” Journal of Economic Behavior & Organization 117: 135–154. Jaya, Jaya, Poonam Dhillon, and Sashi Kumar. 2014. “Challenging Gender Stereotypes and Sexual Harassment in Schools: Evidence from Adolescent Education Programs in India.” Paper presented at the Population Association of America 2014 Annual Meeting, Boston, MA, May 1–3. Justino, Patricia, Silvia Redaelli, Zareen Aslam, and Nathan Reilly. 2018. Poverty, Inequality and Violent Conflict in Pakistan. Washington, DC: World Bank. Kandiyoti, Deniz. 1998. “Bargaining with Patriarchy.” Gender and Society 2 (3): 274–90. Klugman, Jeni, Lucia Hanmer, Sarah Twigg, Tazeen Hasan, Jennifer McCleary-Sills, and Julieth Santamaria. 2014. Voice and Agency: Empowering Women and Girls for Shared Prosperity. Washington, DC: World Bank Group. Lieven, Anatol. 2011. Pakistan, a Hard Country. New York: Allen Lane. Lopez-Acevedo, Gladys, Philipp Krause, and Keith Mackay. 2012. Building Better Policies: The Nuts and Bolts of Monitoring and Evaluation Systems. World Bank Training Series. Washington, DC: World Bank. Malik, Sadia M. 2009. “Horizontal Inequalities and Violent Conflict in Pakistan: Is There a Link?” Economic and Political Weekly 44 (34): 21–24. Mani, Muthukumara, Sushenjit Bandyopadhyay, Shun Chonabayashi, Anil Markandya, and Thomas Mosier. 2018. South Asia’s Hotspots: The Impact of Temperature and Precipitation Changes on Living Standards. Washington, DC: World Bank. Mansuri, Ghazala, Farhanullah Sami, Dung Thi Thy Doan, Priyanka Pandey, and Muhammad Ali. 2018. When Water Becomes a Hazard: A Diagnostic Report on the State of Water, Sanitation and Poverty in Pakistan and its Impact on Child Stunting. Washington, DC: World Bank. Mansuri, Ghazala, and Slesh Shrestha. 2018. Wealth Inequality, Poverty and Mobility: Evidence from a Panel Survey in Rural Pakistan. Mimeo, World Bank. 39 Naseem, Syed M. 1973. “Mass Poverty in Pakistan: Some Preliminary Findings.” Pakistan Development Review 12 (4): 317–360. Paes de Barros, Ricardo; Ferreira, Francisco H.G.; Molinas Vega, Jose R.; Saavedra Chanduvi, Jaime. 2009. Measuring Inequality of Opportunities in Latin America and the Caribbean. Washington, DC: World Bank; New York: Palgrave Macmillan. Rama, Martin; Béteille, Tara; Li, Yue; Mitra, Pradeep K.; Newman, John Lincoln. 2015. Addressing Inequality in South Asia. South Asia Development Forum. World Bank Group, Washington, DC. © World Bank. Ray, Debraj, and Joan M. Esteban. 2017. “Conflict and Development.” Annual Review of Economics 9: 263–293. Roemer, John. 1998. Equality of Opportunity. Cambridge, MA: Harvard University Press. Scales, Peter, Peter Benson, Larry Dershem, K. Fraher, Raphael Makonnen, Shahana Nazreen, Amy Syvertsen and Sarah Titus. 2013. “Building Developmental Assets to Empower Adolescent Girls in Rural Bangladesh: Evaluation of Project Kishoree Kontha.” Journal of Research on Adolescence 23 (1): 171–84. Serneels, Pieter, and Stefan Dercon. 2014. “Aspirations, Poverty and Education: Evidence from India.” Young Lives Working Paper 125, Oxford Department of International Development (ODID), University of Oxford. Shah, Anwar. 2012. Making Federalism Work—the 18th Constitutional Amendment. World Bank Policy Paper Series on Pakistan PK 03/12. Washington, DC: World Bank Group. Solotaroff, Jennifer L., and Rohini Prabha Pande. 2014. Violence against Women and Girls: Lessons from South Asia. South Asia Development Forum. Washington, DC: World Bank Group. Sowell, Thomas. 2004. Affirmative Action around the World: An Empirical Study. New Haven: Yale University Press. Starmans, Christina, Mark Sheskin, and Paul Bloom. 2017. “Why People Prefer Unequal Societies.” Nature Human Behavior 1. Verma, Ravi, Julie Pulerwitz, Vaishali Sharma Mahendra, Sujata Khandekar, Ajay K. Singh, Sourav S. Das, Sunil Mehra, Anita Nura, and Gary Barker. 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. World Bank. 1995. Pakistan Poverty Assessment. Report 14397-PAK. Washington, DC: World Bank. World Bank. 2002. Pakistan Poverty Assessment. Poverty in Pakistan: Vulnerabilities, Social Gaps and Rural Dynamics. Report 24296-PAK. Washington, DC: World Bank. 40 World Bank. 2005. World Development Report 2006: Equity and Development. Washington, DC: World Bank. World Bank. 2008. Pakistan Tax Policy Report. Tapping Tax Bases for Development. World Bank. 2014. “Adolescent Girls Initiative: Status of Pilot Implementation” World Bank Group, Washington DC. World Bank. 2016a. Taking on Inequality. Poverty and Shared Prosperity 2016. Washington, DC: World Bank. World Bank. 2016b. Seeking Opportunities for All: 2016 Humanity Opportunity Index for Latin America and the Caribbean (English). Washington, DC: World Bank Group. World Bank. 2018a. Female Labor Force Participation in Pakistan: What Do We Know? Washington, DC: World Bank. World Bank. 2018b. Fair Progress? Economic Mobility across Generations around the World. Washington, DC: World Bank. Zaidi, Manzar. 2010. “A Link Between Poverty & Radicalization in Pakistan.” Pakistan Institute for Peace Studies. 41