Report No. 66421-PK Pakistan Towards an Integrated National Safety Net System Assisting Poor and Vulnerable Households: An Analysis of Pakistan’s Main Cash Transfer Program January 24, 2013 Human Development Sector South Asia Region Document of the World Bank Document of the World Bank CURRENCY EQUIVALENT (Exchange Rage Effective as of January 24, 2013) Currency Unit = Pakistan Rupees US$1 = 97 PKR FISCAL YEAR July 1 – June 30 ABBREVIATIONS AND ACRONYMS ADB Asian Development Bank AJK Azad Jammu and Kashmir AKRSP Aga Khan Rural Support Program BISP Benazir Income Support Programme CCSSC Cabinet Committee for Social Sector Coordination CCT Conditional Cash Transfer CNIC Computerized National Identity Card CSP Child Support Programme CZF Central Zakat Fund DFID Department for International Development (UK) DGSE Directorate General of Special Education EOBI Employees Old Age Benefit Institution ESSI Employees Social Security Institution FATA Federally Administered Tribal Areas FSP Food Support Program GDP Gross Domestic Product GNI Gross National Income GoP Government of Pakistan GPF General Provident Fund ICT Islamabad Capital Territory IDP Internally Displaced Person IFA Individual Financial Assistance ILO International Labor Organization KP Khyber Pakhtunkhwa LFS Labor Force Survey LZC Local Zakat Committee MNA Member of National Assembly ii  MPA Member of Provincial Assembly MTDF Medium Term Development Framework NADRA National Database and Registration Authority NCCWD National Commission for Child Welfare and Development NCRCL National Centres for Rehabilitation of Child Labour NCRDP National Council for the Rehabilitation of Disabled Persons NCSW National Council of Social Welfare NGO Non-Governmental Organization NPO Non-Profit Organization NSPS National Social Protection Strategy NSS National Savings Scheme NTD National Trust for the Disabled NTN National Tax Number PBM Pakistan Bait-ul-Mal PBS Pakistan Bureau of Statistics PCO Pakistan Census Organization PESSI Punjab Employees Social Security Institution PGPIF Punjab General Provident Investment Fund PIB Provincial Investment Bonds PIHS Pakistan Integrated Household Survey PKR Pakistan Rupee PPAF Pakistan Poverty Alleviation Fund PPF Punjab Pension Fund PRSP Poverty Reduction Strategy Paper PSLM Pakistan Social & Living Standards Measurement Survey RSP Rural Support Programs RSPN Rural Support Program Network PSU Primary Sampling Unit SECP Securities and Exchange Commission of Pakistan SESSI Sindh Employees Social Security Institution TFC Term Finance Certificates TOR Terms of Reference TUS Time Use Survey TVO Trust for Voluntary Organizations VPS Voluntary Pension Scheme WWB Workers Welfare Board WWF Workers Welfare Fund Vice President Isabel Guerrero Country Director Rachid Benmessaoud Sector Director Jesko S. Hentschel Sector Manager Pablo Gottret Task Team Leaders Xiaohui Hou and Cem Mete iii CONTENTS Contents .................................................................................................................................................... iv Acknowledgments ..................................................................................................................................... v Executive Summary ................................................................................................................................. vii Chapter 1: Introduction .................................................................................................................................... 1 I. Pakistan’s Vision for an Integrated Social Safety Net System within a Social Protection Framework ............................................................................................................................. 1 II. A Concise Description of the Current Social Safety Net System .......................................... 2 III. Government’s Strategy to Achieve its Longer Term Social Protection Goals ...................... 3 Chapter 2: Poverty and Households’ Vulnerability to Food Crisis and Other Shocks .............................. 7 I. Introduction ........................................................................................................................... 7 II. Growth, Inequality, and Poverty in Pakistan ......................................................................... 9 III. Food Prices, Other Shocks and Coping Mechanisms .......................................................... 12 IV. The Impact of the 2008 Food Price Crisis ........................................................................... 18 Chapter 3. Pakistan’s Main Unconditional Cash Transfer Program, BISP, and the Protection of the Poor ................................................................................................................................................................... 25 I. Introduction ......................................................................................................................... 25 II. Original BISP Design and Its Performance ......................................................................... 28 III. The New BISP Design ......................................................................................................... 34 IV. Challenges in Safety Net Program Implementation in Militancy Affected Areas ............... 40 Appendix 1 Safety Net Programs in Pakistan: An Overview ................................................................. 42 Appendix 2 Social Safety Net Expenditures and Wheat Subsidy Program ............................................. 58 Appendix 3 Supplementary Tables for Chapter 2 ................................................................................... 62 Bibliography ............................................................................................................................................ 67 iv ACKNOWLEDGMENTS The Government of Pakistan, through the Benazir Income Support Programme (BISP), requested the World Bank to assess the effectiveness of Pakistan’s safety nets in assisting the poorest and most vulnerable households. This report provides an analysis of poverty and vulnerability in Pakistan and documents the effectiveness of BISP in assisting the poor in this context. These findings have already informed the design of BISP, both for improving the targeting performance of the program and also other aspects of program implementation including payment delivery. This report was written by a team led by Xiaohui Hou (Senior Economist, WBIHS) and Cem Mete (Lead Economist, SASSP). The team consisted of Andrea Vermehren (Lead Social Protection Specialist, SASSP), Iftikhar Malik (Senior Social Protection Specialist, SASSP), Kinnon Scott (Senior Economist, DECPI), Amjad Zafar Khan (Social Protection Specialist, SASSP), Iffat Idris (Consultant, SASSP), Madiha Afzal (Consultant, SASSP), Shumaila Rifaqat (Consultant, SASSP), Farrah Ilyas (Consultant, SASSP), Alison Barret (Consultant, SASSP), Khalid Mahmud (Consultant, SASSP), Seo Yeon Hong (Consultant, SASSP), Jing Zhang (Consultant, SASSP), Stefania Cnobloch (Consultant, SASSP), Denis Nikitin (Consultant, SASSP), Ximing Wu (Consultant, SASSP), Mohammad Aslam Malik (Program Assistant) and Izabela Chmielewska (Team Assistant). This work was carried out under the direction of Pablo Gottret (Sector Manager, SASSP) and Mansoora Rashid (Sector Manager, LCSHS). The team collaborated with DFID on the design and funding of various surveys, and with UNICEF and the World Food Program for Damage Needs Assessment following militancy and floods crises. The peer reviewers were Philip O’Keefe (Lead Economist, EASHS), Qaiser Khan (Lead Economist, AFTSP) and Aleksandra Posarac (Lead Economist, HDNSP). The Government of Pakistan provided invaluable assistance during this process. Intense collaboration with the Benazir Income Support Program—especially for the fielding of the 2009 BISP beneficiary survey—enabled timely inputs into the national rollout of the poverty scorecard instrument. The team also thanks the Pakistan Bureau of Statistics’ staff for their close collaboration in designing and fielding the panel component of the Pakistan Social & Living Standards Measurement Survey (PSLM), which enabled an analysis of poverty dynamics in Pakistan. This report benefited significantly from the work carried out by Pakistani researchers. In particular, the Bank team and the PRSP Secretariat of the Ministry of Finance issued a call for papers in September 2009, asking Pakistani researchers to submit proposals on informal sector work, poverty and social protection, time poverty among employed individuals, and child/youth work. The following researchers produced papers using Time Use Survey data: Dr. Naveeda Salam (Planning Commission), Dr. Ghulam Mohammad Arif (Pakistan Institute of Development Economics), Maqsood Sadiq Tabassum (Center for Research on Economic and Social Transformation Pakistan), Mohammad Yasin Janjua (Center for Research on Economic and Social Transformation Pakistan), Deborah Budlender (Community Agency for Social Enquiry), Sana Khalid (Consultant), Najam us Saqib, Dr. Nuzhat Ahmad (Applied Economics Research Centre) and Asma Bashir. Preliminary findings of this report were presented at the June 2011 Workshop on Social Protection Policy Choices in Pakistan, organized by the Planning Commission and UNICEF. During October 2011 the team carried out consultations on the draft report by presenting the key findings at the Planning v Commission, Pakistan Institute of Development Economics and BISP. Government officials from all four provinces as well as development partners were consulted via separate workshops a at the Islamabad Serena and World Bank office respectively. This version of the report incorporates feedback and suggestions received during this process. The initial findings from the Time Use studies were discussed with relevant government departments (including the Planning Commission and the sectoral Ministry representatives) during a 2010 consultation workshop organized by the PRSP Secretariat of the Ministry of Finance. The results of process evaluation, spot checks, and beneficiary assessments undertaken as part of Benazir Income Support Program implementation were also utilized in the report. We extend our appreciation to UK Department of International Development (DfID), the Trust Fund for Environmentally & Socially Sustainable Development (TFESSD) and the Gender Action Plan (GAP) Trust Fund for providing complementary financial assistance in completing this task. vi EXECUTIVE SUMMARY 1. Well functioning social safety net programs are key components of successful economic growth and poverty reduction strategies as they redistribute income to the poorest and most vulnerable, which has short as well as long term effects on poverty and inequality. This report shows that poor Pakistani households have benefitted from cash transfers by utilizing them to fulfill basic needs, and thereby creating a foundation for further improvement of living conditions. 2. Regular and predictable cash transfers allow poor households to smooth household expenditures, improving their ability to manage risks. Cash transfers also help households to invest in children’s human capital, enabling them to take advantage of sizable returns to (per year of) schooling at 7 percent for males and 9 percent for females in Pakistan. More broadly, cash transfers help in remedying credit market failures, allowing households to take up investment opportunities that they would otherwise miss. At the community level, cash transfers facilitate the functioning of markets after a crisis. Objective and transparent programs are also likely to have a positive impact on social cohesion. Finally, the presence of an effective social safety net allows policy makers to avoid suboptimal economic policies (for example certain price subsidies), which may interfere with the functioning of markets. 3. In 2007, the World Bank produced the report “Social Protection in Pakistan: Managing Household Risks and Vulnerability�. This document gave a detailed review of social protection mechanisms in Pakistan and made recommendations for their improvement, notably to increase coverage and enhance effectiveness. During the same year, the Government of Pakistan endorsed the Social Protection Strategy to Reach the Poor and the Vulnerable, which recognized helping the poor and vulnerable segments of the society through social safety nets as a key objective of the Poverty Reduction Strategy, and highlighted the need to reform ad hoc, reactive and fragmented sectoral interventions. These documents were invaluable in guiding the Government’s social safety net reform agenda that followed. 4. After 2007 the Government of Pakistan significantly increased its spending on social safety nets, while at the same time transitioning to a new institutional framework, which is meant to improve the coverage, targeting effectiveness and adequacy of social safety net programs. However, since 2007, Pakistan has also endured three major crises. First, the effects of the global economic downturn resulted in rising food inflation and reduced demand for Pakistani exports. Second, the security environment deteriorated rapidly due to the militancy, especially in Khyber Pakhtunkhwa (KP) and the Federally Administered Tribal Areas (FATA). Finally, during June-July 2010, torrential rains released a wave of floods, stretching the entire length of the country; followed by the August 2011 floods that primarily affected Sindh and Balochistan. In this context, this report analyzes the nature of poverty and vulnerability and assesses the effectiveness of Pakistan’s main cash transfer program in assisting the poor. vii I. Poverty and vulnerability in Pakistan 5. The share of the Pakistani population that lives in poverty declined from 30.6 percent in 2001/02 to 22.3 percent in 2005/06 and to 17.2 percent in 2007/08. 1 Between 2008 and 2010, the welfare distribution was dynamic across all expenditure deciles, particularly within the lower welfare distribution. About 50 percent of the households below the poverty line in 2008 had moved out of poverty by 2010, but a similar number of households fell into poverty. During this period, households that made an upward transition still ended up hovering near the poverty line. Likewise, the risk of households falling into poverty during this two-year period was much higher for those households in the bottom 50 percent of the 2008 household consumption distribution. 6. The key determinants of moving into and out of poverty are access to agricultural land, the household head’s education level, urban/rural residence, and household size. All of these indicators are good predictors of a household’s likelihood of being poor at any point in time. Access to agricultural land seems to be especially valuable in reducing the likelihood of rural non-poor households becoming poor over time. 7. What makes Pakistan different from many other developing countries is not so much the prevalence of poverty, but the fact that Pakistani households have to cope with a wide variety of crises. These crises include the global food-price increases, floods, and militancy in parts of the country. In addition to these external shocks, other burdens include household-level events such as illness or death in the family. This report reveals that welfare losses from the crises were significant in Pakistan, which deepened the gap between poor and non-poor, and further increased inequality among the provinces. Welfare loss is accompanied by a significant reduction in caloric intake. Households cope with crises by switching to lower-quality/cheaper food, reducing the quantity of food intake, decreasing the non-food expenditures, spending savings or foregoing investments, and even decreasing education-related expenditures. 8. A household’s likelihood of being exposed to a shock—as well as how the affected households deal with a shock and the shock’s impact on household welfare—varies widely by household characteristics, as well as by the type of shock. For example, poor households are more likely to experience shocks in general, and their recovery from a shock is slower than non-poor households. A much higher percentage of households in the bottom quartile reduced the quantity of food intake than households in the top quartile, and the reductions in education expenditures were more significant for the poor households. Poverty is not the only factor that influenced how households were affected by shocks, however. Food price increases had the greatest impact on urban households, which was probably due to their more limited access to agricultural land and thus subsistence agriculture option. Other shocks did not have the same visible impact on urban households. Among idiosyncratic shocks, health shocks were by far the most prevalent in Pakistan, and for many affected households, lead to poverty. 1 Based on World Bank staff calculations using PSLM 2007/08. The Government of Pakistan has not endorsed or published poverty estimates since 2005/06. viii II. Are Pakistan’s safety net programs equipped to assist the poor and the vulnerable? 9. Until the introduction of the Benazir Income Support Program (BISP), Pakistan’s safety net system was comprised of two cash transfer programs: Zakat, and the Food Support Program administered by the Bait-ul Mal. Both programs were weakly targeted to the poor: only 46 percent of total Bait-ul-Mal expenditures (and 43 percent of total Zakat expenditures) reached the poorest 40 percent of the population. During the second half of 2008, the Government of Pakistan launched BISP as its main national social safety net program. The short-term objective of the program was to cushion the adverse impact of food, fuel and financial crises on the poor, but its broader objective is to meet the redistributive goals of the country by providing a minimum income support package to the chronic poor and those who are highly vulnerable to future shocks. The program provides cash transfers of Rs. 1,000 [$10.5] per month to eligible families. As a result of the expansion of this program, the federal government’s social safety nets spending increased from 0.3 percent of GDP (a very low amount compared to other developing countries) in 2003/04 to 0.9 percent of GDP including selected subsidies meant to assist the poor. 10. When BISP was rolled out during the third quarter of 2008, the main mechanism for targeting of the program was through the distribution of forms to parliamentarians. The parliamentarians, in turn, encouraged their constituents to apply for benefits. However, recognizing the need for a more objective and transparent targeting method, the government adopted a poverty scorecard for improving the targeting performance of the program, as well as establishing a legal and institutional framework to ensure sound management, governance, monitoring, and administration of the program. The transition from the parliamentarian based beneficiary identification to the poverty scorecard-based beneficiary identification system first took place in 15 districts and was completed nationwide by the summer of 2011. 11. The original beneficiary selection approach that relied on parliamentarians seems to have made the most of available, easily observable information that the parliamentarians and their contacts had. In particular, 35 percent of parliamentarian-selected beneficiary households were female headed. The ineligibility filters, utilizing the National Database and Registration Authority (NADRA), turned out to be helpful in this targeting process. However, this system suffered from critical shortcomings that prevented it from being a long-term solution. In particular, many of the parliamentarian-selected beneficiaries who had high poverty scores also ranked high in terms of political connections and political participation, implying that social networks mattered significantly in the selection process. In addition to concerns about objectivity and transparency, there were other problems with the original system, such as difficulties in establishing a grievance mechanism for a subjective and heterogeneously implemented targeting mechanism. 12. The parliamentarian based targeting instrument was pro-poor but it did not adequately capture the poorest. While the performance of the program compared well to the Food Support Programs’ targeting performance (which has since been discontinued), it is also clear that the decision to move to a poverty scorecard based targeting approach (as well as various institutional reforms associated with it) were highly appropriate. This objective targeting tool and experimentation with alternative payment approaches represent a significant improvement over the original BISP design, together with ongoing efforts to improve grievance redressal procedures. ix III. The Way Forward: Policy Options 13. With BISP (household) poverty survey having been completed nationwide, Pakistan has a credible poverty database: a significant accomplishment for a developing country. The Government of Pakistan, or indeed any of the Provincial Governments, can use the poverty scorecard database housed in BISP to target programs—including those linked to health, education or employment services—to the poorest segments of the population. Indeed, BISP has already shared the database with several partners in order to improve the designs of such complementary programs. The next steps in the further modernization of the sector include the following: 14. Using available resources effectively by establishing central bodies with responsibility for social protection in provincial governments. 1. Developing a National Beneficiary Registry While BISP is the leading federal safety net authority (as provided in the BISP Act 2010), at the provincial level, the Departments of Social Welfare’s weak capacity is a major challenge in the coordination of cross-sectoral interventions. To ensure better integration, coordination, and monitoring of various social programs, a National Beneficiary Registry (NBR) system should be developed by capitalizing on the BISP poverty database, defining the custodian authority at the national level to manage the updating of the NBR, and actively share data with federal and provincial government departments guided by data sharing protocols. 2. Updating the National Social Protection Strategy (NSPS) The federal government formulated the NSPS in 2007. However, since then the institutional architecture of social protection has changed radically through the launching of a number of new initiatives, BISP being the largest one. The updating of the NSPS (led by the Planning Commission) has progressed at a much slower rate than expected. Even though the provincial governments are included in the national strategy consultations, none of them have a clear social protection strategy. Yet the provinces have greater responsibilities in the sector since the 18th constitutional amendment was enacted and they are already spending resources on various sporadic activities. 3. Reducing Untargeted Price Subsidies The price subsidies on wheat, power, etc., should be gradually diminished as the cash transfers to the poor are channeled through the targeted safety net programs. Federal as well as provincial governments interfere with the markets through price subsidies. Such subsidies are substantial and weakly targeted to the poor, although due to the complexity of the price subsidy mechanisms, (for example the utility store arrangement) it is difficult to quantify the full extent of inefficiencies associated with the supply chain. 15. Building on the cash transfer program platform by implementing graduation programs. With the nationwide implementation of proxy means test based identification of poor households, and experimentation with alternative payment systems, Pakistan is positioned to build on the main cash transfer platform. Implementing graduation programs will help some poor households exit poverty in the short term and reduce the likelihood of intergenerational transfer of poverty by investing in the human capital of poor children. Such graduation programs include cash transfers to support children’s nutrition x and preventative health care (immunizations, etc.); cash transfers to increase the school enrolments and attendance of poor children; and vocational training courses to improve the skills of adult household members. 1. Defining the Roles of Federal and Provincial Governments in Implementing Graduation Programs The priorities and design parameters for graduation programs that build on the basic federal cash transfer program should be determined at the provincial level through the joint work of federal and provincial authorities. If the provinces fund the graduation programs, they should also determine the program mix, coverage, and benefit amounts. Nevertheless, the federal government may choose to provide incentives for certain programs beyond the base unconditional cash transfers, by co-financing programs that may have large positive externalities. Such federal co-financing would need to be accompanied with a program-outcome monitoring function. 2. Expanding Conditional Cash Transfers in the Education Sector Poverty targeting of cash transfers to poor families with primary school age children, paired with a co-responsibility of school enrolment and attendance, would significantly increase school enrollments of children given (i) the strikingly high drop-out rates in primary schools, particularly for children from poor families; and (ii) positive impact evaluation results of a pilot intervention described by this report. This effort would be complementary to the nationwide BISP cash transfers for the poorest, and act as a long-term graduation strategy for these families by increasing their investment in human development. 3. Piloting Conditional Cash Transfers in the Health Sector While Pakistan has plenty of experience with linking cash grants to education services, there is no such experience with linking grants to maternal or child health/nutrition services. To explore this idea, which could potentially benefit the nutrition and health outcomes of mothers and children in Pakistan, a detailed survey of the available services needs to be carried out to identify the sector’s supply- side constraints. Such a survey would provide basic information on the extent to which a CCT with health/nutrition-related co-responsibilities could be effectively instituted in this context. 4. Empowering Women through Incorporating Appropriate Design Features in Safety Net Programs The education and health interventions discussed here would be even more effective by incorporating design features to help empower women. Consistent with international evidence, this report shows that in Pakistan, women’s empowerment is linked (among other things) to better schooling and nutrition outcomes for children. The Government of Pakistan has strongly emphasized this issue in their programs to date. BISP payments are made directly to qualifying women, and due to this program, the percentage of females with NADRA IDs increased significantly, which facilitates their access to various services, including the ability to open a bank account. xi 16. Integrating the disaster response mechanisms and safety net institutions so that in times of disasters a national safety net mechanism can be used to determine the scale of poor population affected and to deliver immediate financial assistance 1. Setting-up/Identifying the Institutions to Operationalize the Emergency Action Plan Pakistan is prone to natural disasters and has had to weather frequent militancy and political crises. Despite the fact that, at times, NADRA has assumed the responsibility to deliver cash transfers to Internally Displaced Persons (IDPs) affected by a crisis, no institutional mechanism is yet in place to cope with the typical post-disaster scenario of greatly expanded numbers of people in need of cash or food transfers. A recently prepared action plan to respond to emergencies, which has been approved by the Prime Minister, needs to be operationalized by allocating specific roles and responsibilities to relevant institutions for the delivery of post-crisis assistance to the affected population. 2. Utilizing the BISP Poverty Database in Determining the Needs of Different Crisis Affected Areas, as stipulated by the Emergency Action Plan Crisis response needs to take into account the prevalence of poverty and demographic characteristics of affected areas. Depending on the nature of the shock, urban/rural and other geographic features also need to be considered. For example, this report finds that the urban households were much more negatively affected by the food price increases compared to rural households. The BISP database would be a valuable asset to enable and supplement regional targeting in the future. 17. Responding to household-level crises by insuring households against the negative effects of catastrophic illness. 1. Increasing the Coverage of Social Security Programs In middle-income and high-income countries, social security programs serve to assist households in times of crisis. But in Pakistan, as is the case in many low-income countries, the coverage of social security programs remains low. The prevalence of a large informal sector is one reason for this: assistance from Employees Social Security Institution (ESSI), Employees Old Age Benefit Institution (EOBI) and Workers Welfare Fund (WWB) are confined to registered workers in the formal sector. Even within the formal sector, the availability of social security benefits varies. As a result, a gradual increase in the coverage of social security programs that include informal sector workers emerges as a priority. 2. Piloting Programs to Insure Poor Households against Catastrophic Illness At the household level, micro-crises frequently occur. In particular, the deterioration of an adult’s health often has disastrous effects on household wealth. Analysis of longitudinal household survey data reveals that in the Pakistani context, insurance against catastrophic illness would significantly improve the living conditions of poor households and prevent middle-income households from falling into poverty. A first step would be to pilot one or more types of health insurance programs among the poorest households (since these are most often severely affected by catastrophic illnesses). The eligible households could be identified from the BISP database. One option could be to collect premiums from those whose poverty scores are above the eligibility threshold for a fully xii subsidized program, but below the BISP cutoff point. Small-scale pilot initiatives, drawing on international experience, could lead to the identification of a suitable, cost-effective solution for Pakistan. 3. Piloting Interventions to Encourage Preventive Health Care Behavior At the same time, efforts should be put into preventing health shocks by promoting behavior change and practices that can promote good health. One example would be to raise hygiene awareness – thereby reducing incidence of diarrhea, malaria and other diseases. Such interventions can be carried out with relatively few resources. Hence, despite being untargeted, they can still be very cost-effective with regard to improving health outcomes across the population. 18. Instituting regular monitoring mechanisms and improving program design over time by incorporating lessons learned from the impact-evaluation studies 1. Undertaking Third-Party Evaluations The designs of Pakistani cash transfer programs have benefited significantly from monitoring and evaluation activities. The continuation of these independent monitoring and evaluation efforts is critical to ensure the credibility of the safety net programs and to improve their effectiveness continuously by taking into account changing socioeconomic conditions. Third party spot-checks, process-evaluations and impact evaluations will continue to be critical as the social safety net system evolves. 2. Improving the Designs of the Surveys that PBS Undertakes Data for regular and timely updates of the poverty situation, as well as tracking the targeting performance of various programs can be generated by adding a “social protection transfers module� to the Pakistan Social & Living Standards Measurement Survey (PSLM). The 2008-2010 PSLM Panel Survey already employs such a module, which can be incorporated into the standard PSLM. Rotating modules on selected topics—such as skills or disabilities—can be considered for the PSLM, as well as the Labor Force Survey (LFS). To better capture trends in the non-income dimensions of poverty, time use surveys can be carried out in regular (e.g., five year) intervals, building on the PBS’s first experience in 2007 of undertaking such an exercise. There are also strong payoffs to regularly undertaking school and health facility surveys (or censuses collecting GPS location information), which to date have been carried out in an ad hoc manner. xiii CHAPTER 1: INTRODUCTION I. PAKISTAN’S VISION FOR AN INTEGRATED SOCIAL SAFETY NET SYSTEM WITHIN A SOCIAL PROTECTION FRAMEWORK 1.1. The vision of Pakistan’s Social Protection Strategy to Reach the Poor and Vulnerable (2007) is “to develop an integrated and comprehensive social protection system, covering all the population, but especially the poorest and the most vulnerable�. Consistent with this vision, the goals of the strategy are identified as (i) to support chronically poor households and protect them against destitution, food insecurity, exploitation and social exclusion; (ii) to protect poor and vulnerable households from the impacts of adverse shocks to their consumption and well-being that, if not mitigated, would push non- poor households into poverty, and poor households into deeper poverty; and (iii) to promote investment in human and physical assets, including health, nutrition, and education, by poor households capable of ensuring their resilience in the medium run and of interrupting the intergenerational cycle of poverty. 1.2. This vision is fully consistent with the World Bank’s Social Protection and Labor Strategy for 2012-2022, which recognizes that effective social protection and labor policies and programs help create the resilience, equity and access to jobs and opportunities essential to save lives, reduce poverty and promote sustainable growth. The overarching goal of this strategy is to help developing countries improve resilience, equity and opportunity. In this context, resilience refers to preventing drops in well- being from a range of shocks and key sources of resilience are social insurance programs. Equity refers to protection against destitution. Social assistance programs — including cash transfers and targeted food assistance — serve an especially important role in supporting this goal. Opportunity refers to promotion of better health, education, skills development and helping men and women to secure better jobs. Interventions that support this goal include conditional cash transfers to incentivize health and education investments and labor market programs that provide unemployment benefits, build skills and enhance workers’ productivity and employability. 1.3. The focus of this report is Pakistan’s main safety net program: the Benazir Income Support Program (BISP), which is only one part of the broader poverty reduction and social protection interventions. In particular, social protection includes labor policy, contributory social insurance, and social care services as well as safety nets. This chapter briefly summarizes the Government’s broader social protection goals and priority areas of intervention, followed by an analysis of poverty and vulnerability in Pakistan (Chapter 2) and an assessment of how BISP served to assist the poor and vulnerable (Chapter 3). 1 II. A CONCISE DESCRIPTION OF THE CURRENT SOCIAL SAFETY NET SYSTEM 1.4 Pakistan’s social protection system comprises safety nets, social security and employment promotion and protection. Annex 1 provides a more comprehensive overview of safety net programs, which includes the Benazir Income Support Program (BISP) as the main cash transfer program. 1.5 The BISP was established in 2008. The initial aim of BISP was to help the ‘poorest of the poor’ through unconditional cash transfers. It was initiated against a backdrop of high inflation, particularly food price increases, and the serious negative impact of this on the purchasing power of the poorest sections of society. PKR 34 billion was allocated for BISP in the 2008/09 budget, equivalent to 0.3 percent of GDP. In that year, PKR 14 billion was disbursed to 1.76 million beneficiaries in the form of cash grants of PKR 1,000 per month. The program gradually aims to reach 15 percent of the total population. 1.6 Now that the unconditional cash transfer system is being implemented nationwide, the Government is expanding the coverage of its safety net by introducing a conditional cash transfer program to improve human capital through education. Two key factors motivated the decision to move towards a supplementary conditional cash transfer benefit. First, the BISP database revealed that 71 percent of beneficiaries’ children between ages 5 and 12 did not attend school. Second, the impact evaluation results from a pilot CCT program in Pakistan showed that small amounts of additional (conditional) cash transfers can lead to significant increases in primary school enrollments and attendance in this context. 1.7 Prior to the launch of BISP, main social safety net programs in Pakistan were Zakat and Pakistan Bait-ul-Maal (PBM). Both of these programs were characterized with weak targeting performance, low coverage and benefit adequacy, as shown later in this report. The PBM’s main cash transfer in tervention, the Food Support Programme (FSP) was discontinued in 2008/09 following the launch of BISP. Zakat continues to operate through private financing focusing exclusively on needy Sunni Muslims, at a small scale. 1.8 The Government has also employed cash transfer programs in response to emergencies, most recently to assist families who were affected by the July/August 2010 floods (which covered over 100,000 square km, resulting in 1,980 deaths and 2,946 injured). In this case, the Federal Government worked with Provincial Governments and the National Database Registration Authority (NADRA) to initiate the Citizen’s Damage Compensation Program: a cash grant program for flood affected families. 1.9 In addition to the cash transfer programs, other small, scattered programs provide social welfare and care services to persons with disabilities, child laborers, and others. Despite the consolidation of cash-transfer programs through BISP in recent years, the responsibility for social protection programs overall remains dispersed among a large number of departments and attached bodies. 2 III. GOVERNMENT’S STRATEGY TO ACHIEVE ITS LONGER TERM SOCIAL PROTECTION GOALS 1.10 In order to achieve the National Social Protection Strategy goals discussed in section 1, the Government has identified priority areas of intervention in the same document. These are (i) reaching the poorest; (ii) addressing the needs of vulnerable children; (iii) addressing the needs of vulnerable and poor women; (iv) providing employment and income earning opportunities; (v) protection against environmental and natural disaster; (vi) improving access to social care services and shelter; (vii) protection against health shocks; and (viii) protection for workers in the formal and informal labor markets. 1.11 In particular, the strategy emphasizes the need to expand the coverage of cash transfers through a combination of conditional (to increase school enrollments and to reduce malnutrition) and unconditional programs. Furthermore, the strategy proposes consideration of pilot interventions that would combine cash transfers with skills-development interventions aimed at “graduating� the poor. 1.12 Through BISP, the Government has achieved a number of its stated objectives, which can be viewed as the first phase of a comprehensive sectoral reform agenda. Namely, chapter 3 of this report shows that there have been considerable improvements in reaching the poorest individuals, and addressing the needs of the vulnerable women and children. Similarly, basic mechanisms are now in place to respond to emergencies, although clearly more needs to be done to improve coordination among responsible federal and provincial agencies. 1.13 The second phase of the reform agenda would build on the significant achievements of the last few years through (i) the design and implementation of programs to improve the productivity and employment of workers; (ii) enabling poor’s access to affordable insurance; and (iii) improvements in the quality and availability of social care services. This agenda will have to be carried out in a dynamic political, economic and social environment, however. First, there is a need to consider the social protection sector implications of the 18th constitutional amendment, which provided more resources and responsibilities to the provinces for public service delivery. Second, there is a need to consider the sustainability of planned interventions. 18th Constitutional Amendment and the Social Protection Sector in Pakistan 1.14 Passage of the 18th Constitutional Amendment led to 22 subjects being removed from the Concurrent List, i.e. they are the sole responsibility of the provinces. As a result of the Amendment, the federal Ministry of Zakat and Ushr, as well as the Ministry of Social Welfare and Special Education have been disbanded. The Ministry of Labor & Manpower, which the Workers Welfare Fund falls under, has also been devolved. 1.15 The increased provincial share in the new National Finance Commission (NFC) Award – up from 46 percent to 57 percent2 – means that in the short term resource availability for social protection is less critical than the capacity to spend those funds. However, funding will be a critical issue for those 2 Under the 7th National Finance Commission Award collection charges of the federal government have been cut from 5 percent to 1 percent, thereby increasing the size of the divisible pool. The provincial share of the divisible pool went up from 46.25 percent to 56 percent for 2010-11 and thereafter to 57.25 percent. 3 programs that depend on resource mobilization rather than budgetary allocations, and in which resources have to date been centrally pooled and then distributed among the provinces. This includes Zakat and the Workers Welfare Boards.3 1.16 Even before the 18th Amendment, the responsibility for the delivery of basic social and economic services of direct relevance in the daily lives of the general public were already mandated to sub-national governments. 4 Indeed, a number of different provincial government departments and agencies have already been managing programs which would be covered under the rubric of social safety nets. The portfolio of their respective schemes is characterized by a fair degree of overlap and duplication, which require both consolidation and promotion of core competencies. The 18th Amendment has amplified the pressure on provincial governments to adopt such an agenda, to streamline administrative procedures and improve effectiveness of program implementation. 1.17 Even though coordination of programs within each province is likely to improve as a result of the th 18 amendment, in the absence of a central, coordinating body, coordination among provinces might deteriorate. Moreover, what remains to be decided is who will support/compensate for the weaker capacities and finances in the smaller provinces. As needs are greatest in those provinces, mechanisms will be required to facilitate the provinces with weak capacity to take on devolved responsibilities so that the population there are not further disadvantaged. 1.18 As a result, the Federal Government continues to have a critical role in ensuring Pakistan’s long term poverty reduction and social protection goals are met. There are strong payoffs to operating a base safety net program centrally, to ensure that regardless of where a poor household may reside in within Pakistan, they have the same chance of qualifying for basic cash transfer assistance. Similarly, BISP in its current form would also enable poor children to participate in primary education, which is compulsory by law. 1.19 The BISP poverty database would also allow each province to address its needs and priorities. For example one province may want to operate with a more generous poverty-score-based eligibility criteria covering the "extra households" from their resources; another province may run a skills program for a subset of BISP eligible households; yet another one may provide additional health insurance benefits to that group. Such an integrated database would enable provinces minimize unintentional overlaps and duplication of interventions. Considering the externalities to education and health investments, the federal government may also encourage the use of this database by providing matching grants for selected provincial grants that serve to achieve national human development goals. The Federal Government may also become increasingly active in organizing third-party monitoring and evaluation of the programs that utilize the poverty scorecard database. This information would then be used to help determine which programs the Federal Government would co-finance by matching grants. 3 Both derive their income from contributions, for the former Zakat contributions and for the latter social security contributions. The vastly differing capacities of the provinces to generate these resources will impact their program portfolio after devolution. If resource mobilization and distribution is devolved, Sindh will gain significantly, while Balochistan and KP will struggle to maintain current programs. 4 Excluding utilities like electricity, gas and telecommunications, which continue to be managed and regulated by Federally administered agencies. 4 Sustainability of Social Protection Sector Spending 1.20 We discuss the fiscal space for the funding of safety net programs in Pakistan by considering (i) macroeconomic conditions including economic growth and increase in overall government revenue; (ii) a re-prioritization of social safety net programs within the government budget; (iii) social safety net grants and foreign aid; and (iv) increases in the efficiency of existing safety net programs.5 Appendix 2 provides key expenditure statistics. A key insight that emerges from this exercise is that there is an enormous opportunity to increase the fiscal space for targeted social safety net programs through re-prioritization of expenditures away from untargeted and inefficient subsidy programs (including the wheat subsidy and electricity subsidy). Another promising avenue to pursue would be to improve the efficiency and integration of the fragmented safety net programs. In contrast, there is little potential to increase fiscal space significantly by relying on economic growth. There might be some possibility for fiscal space improvement through grants and foreign aid, but unpredictability of such assistance makes it a less promising option. 1.21 The fragile macroeconomic situation of Pakistan creates a fiscal challenge to expand the safety net programs. Pakistan’s tax to GDP ratio is less than 9 percent, which imposes significant resource constraints on public service delivery in general and social safety net financing in particular. At the outset of 2007-08, the macroeconomic condition of the country deteriorated rapidly, owing to the sharp rise in international oil and food prices (specifically wheat), in combination with policy inaction and internal political turmoil. Both fiscal and current account deficits increased significantly. In the absence of adequate remedial policy measures to address the imbalances—in particular, not passing on the international price increases to domestic consumers—the economy began to adjust through a slowdown in growth and rising inflation. Moreover, in July 2010, Pakistan experienced unprecedented flooding nationally, affecting approximately 20 million people, damaging or destroying more than 1.9 million homes, and displacing about 7.5 million people. This flooding had a substantial impact on the pace of real gross domestic product (GDP) growth, the rate of inflation, the size of the fiscal deficit, and the balance of payments. The government debt to GDP remains considerably high at around 56.8 percent. The economy is greatly vulnerable to external shocks due to its heavy dependence on oil and textile prices; the former constituting 31 percent of total imports and the latter representing 49 percent of total exports. Pakistan’s real GDP is expected to grow by 3.5 percent in 2013. However, the projections for consumer prices and the balance on current account remain grim, whereby the former is predicted to rise overtime, by 12.5 percent in 2013 and 14 percent in 2017, and in terms of the latter a deficit is likely to persist.6 1.22 Social safety net programs in Pakistan were not a priority of the government until recently, with woefully low allocations compared with many other developing nations where safety net spending comprises of 1 to 2 percent of GDP. The safety net expenditure has increased over the 2006-11 period, with a 133 percentage change between 2007-08. In the following years, it continued to show an increasing trend in terms of expenditure on safety nets as a percentage of GDP, reaching 0.6 percent by 2010-11 (about 0.9 percent of GDP including international assistance). The initial increase in the safety nets spending is due to a rise in social security and welfare spending. After 2008, much of the increase is due 5 This discussion of the safety nets fiscal space adopts Heller’s (2006) framework and builds on the application by Tandon and Cheryl in health sector (2010). 6 http://www.imf.org/external/country/PAK/rr/2012/031512.pdf 5 to the government’s flagship safety net program of BISP. There have also been significant increases in spending on natural calamities and disasters, particularly as a result of the cyclone in 2007 and the worst flooding in its history in 2010. 1.23 There is a significant opportunity for re-prioritization to generate additional fiscal space for safety nets, specifically by further reducing subsidies. The subsidies have been quite prominent in budgetary allocations with a substantial rise in 2007-08, owing to the fuel, food, and financial crisis. Nevertheless, a declining trend is observed after 2009, in line with the government policy of phasing out subsidies and replacing them with direct cash transfer programs. If the government moves away from untargeted electricity and wheat subsidy programs in favor of targeted cash transfer programs such as BISP, the scale of targeted social safety net interventions can be increased significantly. 1.24 Grants and loans from the international organizations and development agencies are another potential course to increase fiscal space of safety net programs. The main international partners providing assistance to Pakistan in social safety nets have been WB, ADB, DFID and USAID. Percentage of total lending towards the safety net program has increased significantly for almost all donors in 2010 and 2011 after the food crisis and launch of the BISP. However, along with exploring this avenue for creating more fiscal space for safety nets and seeking additional foreign aid in this regard, it is important that Pakistan balances its debt ratio to avoid long-term aid-dependence. 1.25 There is also significant room to improve the efficiency of safety net programs to increase effective fiscal space for safety net programs. This can be achieved by using the BISP poverty score database to integrate various social safety net programs at both federal level and provincial level. An increased reliance on information technology for program management and payment delivery would also lead to administration efficiency and transparency. 6 CHAPTER 2: POVERTY AND HOUSEHOLDS’ VULNERABILITY TO FOOD CRISIS AND OTHER SHOCKS 2.1 Pakistani households are exposed to a wide variety of crises and shocks. These crises include external shocks such as global food-price increases, floods, and militancy in parts of the country, as well as household-level events such as illness or death in the family. The causes and consequences of these shocks are closely linked with households’ vulnerability and movement in and out of poverty. 2.2 The objective of this chapter is to describe the poverty trends in Pakistan, and to illustrate households’ vulnerability to both idiosyncratic and aggregate shocks, with a focus on the 2008 food crisis. This chapter first reviews the inequality and poverty trends at the macro level, and then moves to the micro level and focuses on moving in and out of poverty, as well as vulnerability to various shocks and the coping mechanisms. The special focus on food crisis is because of the historical breadth and depth of the 2008 food crisis, in which, the price of food staples increased to almost record highs: the global price of wheat rose by 136 percent and rice by 217 percent from 2006 to 2008 (World Bank, 2010a) while in Pakistan the price of wheat increased by 106 percent, and prices of other staple foods rose between 20 percent and 120 percent at the end of 2008 from 2005. 2.3 In analyzing the impact of food crisis and other shocks, this chapter not only uses the classic measurements of poverty (per adult equivalent expenditure), but also examines access to education and nutrition (measured by per adult equivalent caloric intake). The analysis is based mainly on the 2008- 2010 PSLM Panel Survey (see Box 2.1), taking advantage of its panel nature to document poverty dynamics. 2.4 The PSLM data reveal that the welfare distribution was dynamic across all expenditure deciles between 2008 and 2010, particularly within the lower end of the welfare distribution. About 50 percent of the households below the poverty line in 2008 had moved out of poverty by 2010, but a similar number of households fell into poverty. During this period, households that made an upward transition still ended up hovering near the poverty line. Likewise, the risk of households falling into poverty during this two-year period was much higher for those households in the bottom 50 percent of the 2008 household consumption distribution. A household’s likelihood of being exposed to a shock—as well as how the affected households deal with a shock and the shock’s impact on household welfare—varies widely by household characteristics, as well as by the type of shock. 2.5 The welfare losses from crises were also significant in Pakistan, deepening the gap between poor and non-poor and further increasing inequality among the provinces. Welfare loss is accompanied by a significant reduction in caloric intake. Households cope with crises by switching to lower-quality/cheaper food, reducing the quantity of food intake, decreasing the non-food expenditures, spending savings or foregoing investments, and even decreasing education-related expenditures. 7 2.6 Documenting such vulnerability to different shocks and its relation to poverty are crucial for designing effective poverty alleviation policies and programs. In addition, understanding this interrelationship is necessary to comprehend the breadth and depth of the effects of shocks, as well as its heterogeneous effects on different segments of the population. With this information, policy makers will be able to design more appropriate and relevant programs to help mitigate the shocks’ impact on the poor. Information on the heterogeneous effects of shocks on different groups of households and individuals can also help to adjust or refine programs’ targeting mechanisms. 2.7 The chapter is organized as follows. Section II describes poverty and inequality trends in Pakistan. Section III presents households’ self-reported experiences of shocks and the corresponding coping mechanisms. Finally, Section IV illustrates the negative effects of different crises on household welfare. Box 2.1: Survey Data Used in This Chapter Box 1. PSLM 2008-1010 Panel Survey PSLM 2008-2010 Panel Survey: The PSLM Panel survey was carried out by the Pakistan Bureau of Statistics with technical support from the World Bank. Households surveyed during the January-March and April-June quarters of the PSLM survey during the 2007-08 round were re-interviewed. All urban and rural areas of all four provinces were included in the survey, with the exception of military restricted areas. The 2007/08 survey has a sample size of 15,512 , consisting of households from 1,113 Primary Sampling Units (PSUs). A two-stage stratified sample design was adopted for the survey. In the 2010 round, 7,726 households were visited and data was collected for 6,954 households. The households that split within the same PSU were also included in the follow-up survey. However, households which had shifted out of sample Primary Sampling Units(PSU) are not tracked (643 households), and 129 households covered in the sample PSU in 2008 refused to participate in the 2010 survey. The survey instruments used in the PSLM 2010 are almost exactly the same as those for the PSLM 2007-2008, including the standard modules of household demographic information, education, employment, asset and consumption, etc. To gather more information on experiences of food price shocks and other shocks, specific modules on shocks and coping mechanisms were introduced in the 2010 survey. 8 II. GROWTH, INEQUALITY, AND POVERTY IN PAKISTAN Inequality trends 2.8 Inequality in Pakistan remained stable over the past decade. The Gini coefficients 7 of consumption grew a little during the rapid economic expansion from 2001/02 to 2005/06, and began to decline slowly in 2005/06. Thus, the economic expansion is relatively distributionally neutral: both poor and wealthy households enjoyed the benefits of growth and suffered during the economic downturn. However, from 2008 to 2010, the Gini coefficient slightly decreased in urban areas, but increased in rural areas (Table 2.1). Table 2.1: Gini Coefficient between 1998 and 2010, by Urban and Rural Pakistan Urban Rural 1998-99 0.301 0.357 0.253 2001-02 0.276 0.321 0.239 2004-05 0.298 0.341 0.203 2005-06 0.302 0.351 0.247 2007-08 0.290 0.325 0.253 2008 0.309 0.353 0.254 2010 0.308 0.337 0.267 Source: PSLM data.The Gini coefficients in 2008 and 2010 are restricted to the panel sample, conducted in two quarters of the year. 2.9 The level of inequality varies by province and is highest in Sindh, followed by Punjab, KPK, and Balochistan. In all four provinces, inequality is higher in urban areas than in rural ones. The difference between urban and rural areas is largest in Sindh province (Figure 2.1). Figure 2.1: Gini Index across Regions, both Urban and Rural Sindh Punjab 0.393 0.406 0.412 0.392 0.348 0.335 0.348 0.355 0.257 0.272 0.276 0.289 0.23 0.222 0.22 0.211 2001-02 2004-05 2005-06 2007-08 2001-02 2004-05 2005-06 2007-08 7 The Gini coefficient is a measure of the inequality of a distribution, a value of 0 expressing total equality and a value of 1 maximum inequality. 9 KPK Balochistan 0.355 0.349 0.304 0.323 0.26 0.278 0.276 0.278 0.301 0.23 0.231 0.248 0.216 0.219 0.225 0.195 2001-02 2004-05 2005-06 2007-08 2001-02 2004-05 2005-06 2007-08 Data source: PSLM2001/02. PSLM 2004/05, PSLM2005/06, and PSLM 2007/08. The darker color refers to urban areas and the light color refers to rural areas. Poverty dynamics 2.10 Before the recent series of economic and natural shocks, the poverty reduction was significant. The share of the Pakistani population that lives in poverty declined from 30.6 percent in 2001/02 to 22.3 percent in 2005/06. This progress is consistent with other poverty indicators, such as depth of poverty and severity of poverty. In terms of urban and rural distribution, urban areas experienced a sharper reduction in poverty than rural areas in all measures of poverty between 2001/02 and 2005/06 (Table 2.2). Table 2.2: Poverty Rates in Pakistan, 1989/99 to 2005/06 (percent) 1998/99 2001/02 2004/05 2005/06 Pakistan 30.6 34.5 23.9 22.3 Urban 20.9 22.7 14.9 13.1 Rural 34.7 39.3 28.1 27 Source: Poverty Reduction Strategy Paper II (2009). 2.11 More recently, however, Pakistan experienced its severest economic downturn in decades in the midst of global food crisis and financial crisis. The extent of the economic downturns’ impact on the poverty rate is still under debate, since the country context seems to be highly relevant. This chapter uses 2008-2010 PSLM Panel Survey to examine the poverty dynamics in this two-year span. Since there is no official poverty line published, this chapter uses the 20 percentile mark for the analysis. This mark is chosen because the latest published poverty rate is around 20 percent. The weakness of using the 20 percent rate is that it is associated with a fixed percent of the population, so the prediction of the size of population moving in or out of a fixed poverty line is unknown. However, this approach does illustrate the dynamics of households’ relative welfare status. 2.12 Between 2008 and 2010, the welfare distribution was dynamic across all expenditure deciles, particularly within the lower welfare distribution. Figure 2.2 shows the movement of households into and out of each expenditure decile. Appendix 3 Table A.1 provides more detailed information. The most stable households are those at or above the 90th percentile in per capita expenditure; 58.2 percent 10 remained at or above the 90th percentile in both 2008 and 2010. Around 39 percent of households in the lowest 10 percent in 2008 stayed at or below that level in 2010, while 61.3 percent transitioned to the higher decile groups; 20.4 percent of households between the 10th and 20th percentile remained in the same decile in 2010, while 22.7 percent moved down to the lowest decile. This evidence clearly shows that Pakistani households are highly vulnerable to moving down in the welfare distribution. Figure 2.2: The Transition Dynamics of Adult Equivalent Expenditure by Deciles 234 5 6 10 1 7 8 9 10 9 12 3 4 5 6 7 8 9 10 Per Capita Expenditure Decile (2008) 8 1 2 3 4 5 6 7 8 9 10 7 1 2 3 4 5 6 7 8 9 10 6 1 2 3 4 5 6 7 8 9 10 5 1 2 3 4 5 6 7 8 9 10 4 1 2 3 4 5 6 7 8 910 3 1 2 3 4 5 6 7 9 8 10 2 1 2 3 4 5 6 7 10 89 1 1 2 3 4 5 10 6 789 0% 20% 40% 60% 80% 100% Data Source: PSLM 2008-10. The vertical axis is decile of per adult equivalent expenditure in 2008, and each number in the horizontal rows stands for the decile of per capita expenditure in 2010. Dark grey cells represent households that remained in the same decile; dotted grey cells represent those households that moved to a lower decile in 2010; white cells represent households that transitioned to a higher decile. 2.13 Relative to the 20 percent mark, the size of the transition in-and-out of poverty is quite large. About 70 percent of households at or above the 20th percentile in 2008 remained above that level in 2010, while about 10.5 percent of households that fell below the 20th percentile remained there (see details in Appendix 3 Table A.2). This implies that about 50 percent of the households below the poverty line moved out of poverty and a similar number fell into poverty. The transition dynamics around the poverty line is quite large, indicating that households’ welfare status around the poverty line is volatile. 2.14 There is a large variation across provinces in both distribution and transition of poverty. KPK province has the lowest percentage of the population with household expenditures below the 20th percentile, while Balochistan has the highest percentage of the population with household expenditures at or below the 20th percentile. In Punjab and Sindh provinces approximately 20 percent of households spend at or below the 20th percentile. Balochistan has the highest percentage of households transitioning in as well as transitioning out of the bottom 20 percent, indicating that poverty is most dynamic in this province. On the other hand, Sindh has a relatively high percentage of households moving into poverty, but a relatively low percentage transitioning out of poverty. 11 Figure 2.3: Per Adult Equivalent Household Expenditures, by Province 0.4 0.2 0 Punjab Sindh KPK Balochistan Percentage of households at or below the 20 percentile line in 2010 Percentage of households above 20th percentile in 2008 transitioning into the bottom 20th percentile in 2010 Percentage of households below 20th percentile in 2008 transitioning out of the bottom 20th percentile in 2010 Data source: 2008-2010 PSLM Panel Survey. 2.15 The probability of being poor is higher for those households without access to agricultural lands, whose heads of households are not highly educated, those residing in rural areas, and households that are larger. Table 2.3 and Appendix 3 Table A.3 list the factors that explain per adult equivalent expenditure transitions into and out of poverty. Transitions into and out of poverty are less likely to happen as the household’s per adult equivalent expenditures move away from the 20th percentile in 2008. Table 2.3: Key Determinants of Moving into and out of Poverty Determinants 1. Access to agricultural land Having access to agricultural land decreases the probability of transitioning into the bottom 20 percent, but it is not correlated with transitioning out of the bottom 20 percent. 2. Education of household head A household with a better-educated household head is less likely to transition into the bottom 20 percent. 3. Urban/Rural Households Rural households were more likely to transition into the bottom 20 percent. 4. Household size A larger household size increases the likelihood of falling below the bottom 20 percent line. III. FOOD PRICES, OTHER SHOCKS AND COPING MECHANISMS 2.16 Understanding and measuring households’ vulnerability to becoming poor is necessary for designing effective safety net programs. Ideally, such a vulnerability measure would capture the multidimensional aspects of poverty, including both monetary and nonmonetary measures. Monetary indicators of poverty and living conditions include income, consumption, and assets. Nonmonetary indicators include malnutrition, access to health, education, and basic services; and perceptions of poverty 12 or deprivation. The section below uses both monetary and non-monetary indicators to analyze the impact of food price increases and other shocks on Pakistani households. The perception of food price increase 2.17 The majority of households surveyed responded that they were highly affected by the food price crisis. Households’ perception of the food price shock in the 2008-2010 PSLM Panel Survey is measured on a scale of 1 to 5, with 1 being not affected and 5 being severely affected. In the survey, 35.1 percent reported being highly affected and 16.6 percent reported being severely affected. Around 26 percent said they were moderately affected, 18.3 percent reported being mildly affected, and only 4.2 percent stated that they were not affected. 2.18 The timing of the price increases is consistent with households’ self-reported difficulty buying food. Households responded that the end of 2008 (Oct.–Dec., 2008) was the most difficult time to purchase either wheat or rice. In 2009, the situation became less difficult, although the percentage of households that purchased grains from utility stores did not change much over time. Figure 2.4: Self-reported Difficulty of Purchasing Wheat and the Incidence of Having Purchased at the Utility Stores 15 13 Percent (%) 11 9 7 5 Difficulty buying Purchased in utility store Data source: 2008-2010 PSLM Panel Survey. 2.19 Poorer households rated the food price increase to be more severe than nonpoor households. The higher the per capita expenditure of a household, the less severe the food price shock was perceived to be. In other words, poor households reported experiencing the food price shocks more severely than wealthier households (see Figure 2.5). Regardless of household location, the differences of experiencing the shock of the food price crisis are similar between poor and non-poor households. Nevertheless, a greater proportion of rural households than urban households responded that they were highly (36 percent) and severely (17.6 percent) affected by the food price crisis. However, this descriptive analysis does not control for the poverty status. Thus, the urban rural difference may be due to the incidence of the poverty being higher in the rural areas than in the urban areas. 13 Figure 2.5: Perceived Severity of the Food Price Increase by Household Per Capita Expenditure 5 4 3 2 1 7 8 9 10 Log per capita expenditure in 2008 bandwidth = .8 Data source: 2008-2010 PSLM Panel Survey. 2.20 A number of factors contribute to the perceived severity of a food price crisis (Table 2.4): x Having access to agricultural land and ownership of livestock (such as buffalo) reduces the perceived severity of the food price shock. x Larger households perceived a less severe price shock. x Households with more educated heads perceived a less severe price shock than others. x Female headed households perceived the price shock as less severe than male headed households. x Households in Punjab province perceived the food price shock to be more severe than households in other provinces. Table 2.4: Determinants of the Severity of the Perceived Food Price Shock Dependent Variable: self-reported food price shock (from not affected [1] to severely affected [5]) Log PCE -0.626*** Having access to agricultural land -0.272*** Own Cattle -0.178 Own Buffalo -0.240** # of Household members -0.039*** HH head: Educ Cl.1-5 -0.209*** HH head: Educ Cl.6-8 -0.243*** HH head: Educ Cl.9-10 -0.420*** HH head: Educ Cl.11 & beyond -0.739*** HH head: female -0.440*** HH head: work 0.181** HH head: age -0.006*** Sindh -0.425*** KPK -0.707*** Balochistan -0.565* Rural 0.032 Note: standard error is clustered at the district level. Baseline household characteristics (2008) are used to identify the determinants of experiencing shocks. 14 Other (self-reported) shocks 2.21 Pakistani households are also vulnerable to other aggregate and idiosyncratic shocks. In the 2008- 2010 PSLM Panel Survey, households were asked to report their experiences in the past two years. On average, 9.6 percent of households reported experiencing at least one aggregate shock, such as a drought, flood, earthquake, landslide, crop disease, livestock epidemic, fire, and/or conflict/displacement; 18.7 percent of households reported having experienced a health shock such as illness or death of a household member; and, 6.1 percent of households reported having experienced at least one idiosyncratic shock other than a health shock, such as theft, unemployment, divorce, reduced remittances, and/or violence. The top five sources of household income shocks, in descending order, are a sudden health problem or accident, chronic or long-term sickness, drought, theft/fraud, and lack or loss of employment. 2.22 The self-reported experience of shocks reflects the impact of shocks only to some degree, as these are perceived measures. In theory, an aggregated shock affects all households in a region but wealthy and poor households usually perceive such shocks differently. Thus, an analysis of the relationship between a household’s poverty status and its self-reported experience reveals the perceived impact of shocks (correlated with actual impact) to households along the welfare distribution. 2.23 As a household’s per capita expenditure increases, it is less likely to experience an aggregate shock. Figure 1.7 shows that in 2008, at lower levels of household per capita expenditure, the incidence of shocks is much higher. This implies that poor households are more likely to be affected by aggregate shocks, and that they have no or very limited power to alleviate the impact of aggregate shocks. 2.24 Health shock is the most prevalent idiosyncratic shock in Pakistan. A health shock can lead to catastrophic health expenditures and loss of income. The incidence of health shocks is quite high for poor and lower-middle-income households, as illustrated by Figure 2.6. When asked to rank the number one reason for income shocks in the household, households ranked health-related episodes number one and two. This finding suggests that social protection or health insurance programs that aim to protect the poor and vulnerable from catastrophic spending on health have the potential to provide significant relief to the most vulnerable households. Figure 2.6: The Probability of a Health, Aggregate, and Other Idiosyncratic Shock, by Household Per Capita Expenditure in 2008 .2 .15 .1 .05 0 7 8 9 10 Log per capita expenditure in 2008 Health Shock Other Idiosyncratic Shock Aggregated Shock Data Source: 2008-2010 PSLM Panel Survey. 15 2.25 In addition to per capita expenditure, other factors also contributed to determining the household response to health and other shocks (see Appendix 3 Table A.5). Most households responded to food price increases by switching to lower-quality foods (77 percent), 67.2 percent of households reduced the size of food portions, 29.7 percent lowered nonfood expenditures, and 23.4 percent reported dipping into their savings. Reported Coping Mechanisms 2.26 To deal with other income shocks, many households changed their food purchasing patterns and their food intake behaviors. The respondents to the 2008-2010 PSLM Panel Survey ranked changes in food behavior as the number one way to cope with general income shocks. Food behavior changes include switching to lower-quality foods and reducing food intake, as well as buying food as needed (rather than in bulk). In descending order, the top six ways in which households cope with income shocks are: switching to lower-quality foods (48.05 percent), reducing the quantity of food intake (47.10 percent), borrowing money from relatives/friends (36.5 percent), buying food as needed (34.8 percent) instead of purchasing in bulk, comparison shopping (30.75 percent), and spending savings or investments (27.6 percent). 2.27 Almost 18 percent of households report coping with income shocks by reducing education-related expenditures without taking children out of school. Only 1 percent of income-shock affected households reported taking children out of school, and 2.6 percent reported sending children to work. Based on these self-reported measures, it seems that most households do not take their children out of school as a response to income shocks, at least not in the short term. Two cautionary remarks are in order here. First, it is possible that some parents cite sickness or poor school performance as a reason for a child dropping out of school, when an income shock is the real reason. Second, primary school enrollment rates in Pakistan are already very low, and low rates are highly correlated with poverty. 2.28 The coping mechanisms are different along the welfare distribution (See Figure 2.7). Households in the lowest quartile were more likely to reduce their quantity of food intake, while more households in the top quartile were more likely to spend less on nonfood items to maintain their usual level of food consumption. This implies that many poor households must cut expenditures on necessities, while wealthier households have the luxury to reduce discretionary spending. Figure 2.7: Coping Mechanisms by Per Capita Expenditure 1 .8 .6 .4 .2 7 8 9 10 Log per capita expenditure in 2008 Reduce Quantity of Food Intake Reduce Non-Food Exp. Data Source: 2008-2010 PSLM Panel Survey. The vertical line refers to 20th percentile in the log per capital expenditure distribution. 16 2.29 In response to the food price crisis, many households are forced to purchase and consume lower quality and cheaper foods to meet their basic needs. Most households (77 percent) indicated that they switched to lower-quality foods as a coping mechanism, 67.2 percent of households responded by reducing the size of food portions, 29.7 percent said they spent less on non-food items, and 23.4 percent reported using their savings to compensate for the rise in food prices. Coping strategies are largely the same in urban and rural areas (see Figure 2.8). Figure 2.8: Top Five Strategies for Coping with Food Price Increases, Urban and Rural 75.6 77.8 70.4 Percentage (%) 61.9 29.2 30.0 26.5 18.4 16.1 18.7 Switch to lower Reduce quantity Decrease non- Spend savings Decrease quality or of food intake food or investments education- cheaper food expenditures related expenses Urban Rural Data Source: 2008-2010 PSLM Panel Survey. 2.30 Yet, coping strategies differ by level of household income. A higher percentage of households in the bottom quartile reduced the quantity of food intake (79.3 percent) than households in the top quartile (48.7 percent). A reduction in food intake could have devastating effects on human development, particularly for children. The existence of a safety net during a food crisis helps poor households meet their basic needs. Although the bottom income quartile spends more of their savings (25 percent) than the top quartile (17.6 percent), this does not mean that they have more savings. Actually, they might have used all their potential savings for economic production on food purchases to meet their basic needs. Figure 2.9: Top Five Strategies for Coping with a Food Price Crisis, by Per Capita Expenditure 78.6 79.3 70.3 Percentage (%) 48.7 34.5 26.9 25.2 17.6 18.1 15.8 Switch to lower Reduce quantity Decrease non- Spend savings Decrease quality or of food intake food or investments education- cheaper food expenditures related expenses Bottom Quartile Top Quartile Data source: 2008-2010 PSLM Panel Survey. 17 IV. THE IMPACT OF THE 2008 FOOD PRICE CRISIS 2.31 Higher food prices had a negative impact on poverty and human development in many developing countries. Because of the 2008 food price crisis, a significant portion of the 2.3 billion people living on less than $2 a day became poorer, and another 100 million fell into poverty (Ivanic and Martin, 2008). In addition to the direct impact on monetary-based poverty measures, rising food prices can worsen nutrition and reduce households’ utilization of education and health services due to a fall in their real income, which further depletes the productive assets of the poor. There is clear evidence from past crises that children suffer long-term health consequences from short-term shocks (Alderman, Hoddinott, and Kinsey, 2006; Friedman and Sturdy, 2010). Although the evidence is mixed, food price increases may reduce educational attainment because of the opportunity cost of keeping children in school in some developing countries (Ferreira & Schady, 2009). Food price increases may also force poor households with low coping capacity (because of inadequate credit markets, for example) to invest less in their livelihoods or sell their productive assets (Carter and others, 2004; Lokshin and Rvallion, 2000; Cater and others, 2007). The sections below examine the impact of the 2008 food crisis on budget share, calorie intake8, and children’s education. The Impact of the 2008 food crisis on total food expenditure and budget share 2.32 Pakistani households spent most of their food budgets on dairy and wheat products. On average, households spent 16.5 percent of their total expenditures on dairy and 13 percent on wheat. Other important food items include spices (9.15 percent, including sugar), meat (7.17 percent), vegetables (6.98 percent), and cooking oil (6.79 percent). The budget share of each food category varies across income levels. The poor spend proportionally more on wheat, oil, spices, and vegetables, but less on meat, milk, and fruit. Wheat consumption accounts for about 20 percent of the total household expenditure for the poorest 10 percent, but only 4.7 percent for the top 10 percent of households. The top 10 percent of households spent a significant share on nonfood items: 43 percent compared to 22 percent for the average household. 2.33 Global food prices negatively affected Pakistani households throughout 2008. Compensating variation is used to document the magnitude of the food price crisis. This refers to the additional expenditure necessary to maintain household welfare at the pre-crisis level (Friedman and Levinsohn, 2002). Such expenditures began rising dramatically early in 2008, and had doubled by the end of the year. This shows that the ex-ante determined welfare loss depends exactly on when the effect is measured. 2.34 The food price crisis disproportionately affected the poor over time. Poor and nonpoor welfare losses were similar in July 2007, but the gap widened as months passed, and by June 2008, the difference between poor and nonpoor became almost 10 percent of an average household’s total expenditure (see Figure 2.10). The gap between the poor and nonpoor grew over the course of the crisis, indicating that the effects of the food price crisis on income distribution could be underestimated if compensating variations are measured at only one point in time. 8 In this report, we did not difference caloric intake or caloric availability. The measure of calories comes directly form household survey and converted from quantity of food consumed. 18 Figure 2.10: Welfare Losses during the Crisis Year, Poor and Nonpoor (measured as a percentage of household’s total expenditures) 70 % of Household's Total Expenditure 60 50 40 30 20 10 0 08 0 7 09 0 7 10 0 7 11 0 7 12 0 7 01 0 7 02 0 8 03 0 8 04 0 8 05 0 8 06 0 8 07 0 8 8 09 0 8 10 0 8 11 0 8 12 0 8 8 00 00 20 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 ,2 , 07 08 All Households Non Poor Poor Data Source: PSLM 2005-2006, PSLM 2007-2008. 2.35 Large differences exist across the provinces in compensating variation to the food crisis in Pakistan, especially between Punjab and Balochistan. As Figure 2.11 indicates, households in Balochistan were the most adversely affected over the whole period. KPK was hit hardest at the end of the year. The difference in compensating variation between Punjab and Balochistan was around 17 percent of total household expenditure in June 2008. The compensating variation for households in Punjab is estimated as 37 percent of total household expenditure and in Balochistan 54 percent. This gap lasts until the end of 2008. Figure 2.11: Welfare Losses during the Crisis Year, by Region (measured as a percentage of household’s total expenditures) % of Household's Total Expenditure 70 60 50 Punjab 40 Sindh 30 KPK Balochistan 20 10 0 07, 2007 08, 2007 09, 2007 10, 2007 11, 2007 12, 2007 01, 2008 02, 2008 03, 2008 04, 2008 05, 2008 06, 2008 07, 2008 08, 2008 09, 2008 10, 2008 11, 2008 12, 2008 Data Source: PSLM 2007-2008. 2.36 The relative difference between location and income level is the same when substitution effects and food self-production are considered in the analyses. Because the prices of goods that households relied on went up substantially, they substituted cheaper goods for more expensive goods; thus, the compensating variation is smaller when the substitution effect lessens the negative welfare effects, but the overall effects were still quite large. 2.37 In summary, the global rise in food prices had a deep impact on Pakistani households throughout 2008. Compensating variation estimates suggest that the average household would need 38 percent of its 19 total pre-crisis expenditure to maintain pre-crisis consumption levels. The more precise estimation of welfare loss depends on when the loss is measured. The impact of the food price crisis rose dramatically throughout 2008 and the impact estimated at the end of 2008 was twice as high as measured at the start of the year. The wheat price increase affected poor households more than nonpoor households, and the gap between the poor and nonpoor grew over the course of the crisis. The regional variation in welfare loss over the crisis period is quite significant. Households in Balochistan were the most adversely affected, followed by those in Khyber Pakhtunkhwa (KPK). Impact of the 2008 food crisis on caloric intake 2.38 There is a positive relationship between per capita household expenditure and per capita food caloric intake in Pakistan. Figure 2.12 clearly shows that per capita caloric intake increases as households spend more per capita. The median caloric intake in 2006, 2008, and 2010 was 2,373, 2,287, and 2,258 calories per adult equivalent, respectively. Per capita calorie availability in 2006 was higher than in 2008, along the welfare distribution, implying that the food crisis negatively affected caloric intake during 2007/08. 2.39 The impact of the 2008 food crisis on total caloric intake seems to be more resilient for the poor and near poor than for nonpoor. Figure 2.12 also shows that the average total caloric intake for the poor is lower in 2010 than in 2008, while total caloric intake for the nonpoor rebounded from 2008 to 2010. This in part is due to the wheat price in Pakistan being kept high, even though the international wheat price had dropped; it could also be a result of the more permanent and damaging effects of the food crisis on poor households. For example, the poor might have to deplete their savings or even sell some productive assets to cope with the immediate effects of the food crisis. Figure 2.12: Relationship between Food Caloric Intake and Log Per Capita Expenditures 4000 3500 3000 2500 2000 2 2006 2008 2010 1500 6 7 8 9 10 Log per capita expenditure Poverty line in 2008 2006 2008 2010 Data Source: PSLM 2007-2008. The vertical line refers to 20th percentile in log per capital expenditure distribution. 2.40 A large food price increase has negative effects on caloric intake. As early as the last quarter of 2007, caloric intake decreased (from the pre-crisis period), and again by 8 percent during April–June 2008 (see Figure 2.13 and Appendix 3 Table A.6). This finding is consistent with the observation that the price of wheat reached its peak in June 2008. 20 Figure 2.13: Median Total Caloric IntakeTrends in 2005/06 and 2007/08, by Months 2200 2150 2100 2050 2005-06 2000 2007-08 1950 1900 1850 1800 r ry ne y ch r ril st r er ay be be be ar Ap ua gu ob Ju ar M m nu m em M br Au ct ve ce Ja Fe O pt No De Se Data Source: PSLM 2005-2006 and PSLM 2007-2008. 2.41 There is little change in caloric intake from purchased foods, but dramatic decreases in self- produced foods, despite higher prices in the first two quarters of 2008. Caloric intake from food purchases was lower in the first quarter of 2008, but not in any other quarter. The caloric intake from self-produced foods, however, decreased dramatically. In the first and second quarters of 2008, households reported about 30 percent fewer calories from self-production than in the equivalent period in 2006. The probability of caloric consumption from self-production did not change—households were no less likely to consume calories from home production during the crisis quarters. This might be because some households preferred selling their produced foods to gain income from higher food prices. Figure 2.14: Median Caloric intake from Purchased and Home-Produced Food (2005/06 and 2007/08) Calories from purchased food, per capita Calories from home production, per capita 2150 1200 2100 1000 2050 2000 800 1950 1900 600 1850 1800 400 1750 200 1700 1650 0 Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun 2005-06 2007-08 2005-06 2007-08 Data Source: PSLM 2005-2006 and PSLM2007-2008. 2.42 The gap between the poor and nonpoor per capita caloric intake widened from 2008 to 2010. Figure 2.15 shows that wealthier households in 2010 were better off than in 2008, in terms of per capita caloric availability, but poor households experienced a slight decrease in caloric intake from 2008 to 2010. This is the same irrespective of the self-reported experience of food price increases. The implication is that recovery from the highest food prices in June 2008 took longer for the poor than for the nonpoor. 21 Figure 2.15: Caloric Intake and Perceptions of the Food Price Crisis Affected by food price crisis Not affected by food price crisis 1000 1500 2000 2500 3000 3500 4000 4500 1000 1500 2000 2500 3000 3500 4000 4500 Per Capita Caloric Availability 7 8 9 10 7 8 9 10 Log per capita expenditure Log per capita expenditure 2008 2010 2008 2010 Note: Self-reported food shock is defined as the households reported having experienced food shock more than mildly. Note the vertical line refers to 20th percentile in log per capital expenditure distribution and the horizontal line is the median per capita food caloric availability. 2.43 The negative effects of crisis on urban households are also seen in their caloric consumption patterns. Urban households consumed about 5 percent fewer calories on average. This decline appears to be a direct result of rising food prices, which led to smaller purchases of various categories of food. 2.44 Analyzing the changes in caloric intake by different food sources explains the protective role of self-production—especially for rural households. Although caloric intake declined for rural households, the calories they consumed through market purchases of food were largely unchanged over the crisis period. This suggests that rural households did not experience as severe a fall in real income as urban households. At the same time, calories obtained through self-production declined significantly, most likely because of a shift to greater market activity. This may have been due to rural households’ efforts to capitalize on the relative price rise of agricultural goods by selling their self-produced agricultural products. Impacts on child schooling and child labor 2.45 In addition to budget share and caloric availability, food price shocks also affect other dimensions of human development. One of the most important is child labor and schooling. One mechanism of possible causality is the direct cost of education; another is the opportunity cost of keeping children in school rather than working. Thus, in some cases, children may be pulled from school and put to work to augment the household income. Once children drop out of school and fall behind their peers, it is often very difficult for them to return to school and resume making progress at an appropriate grade level for their age. This subsection analyzes the correlation between children’s education and child labor in relation to a self-perceived food price shock. The purpose is not to establish the causality between food price increases and child labor, since self-reported shocks are endogenous, but to use the panel survey to illustrate the correlation between food price increases and children’s schooling, and the patterns of transitioning in and out of school with respect to a food price shock. 2.46 There is a clear difference between the poor and nonpoor households with respect to the effect of self-reported food price shocks on school enrollment. The school enrollment rate is higher for nonpoor 22 households, irrespective of whether the households reported being affected by the food price shock; however, the school enrollment rate for poor households decreased significantly in 2010, from 2008, for households that reported being affected by the food price shock. This is not the case for households that did not report a food price shock. The result illustrates the potential negative effect of a food price increase on children’s education. Figure 2.16: Probability of Attending School and the Perceived Impact of Food Price Increases Affected by food shock Not affected by food shock 1 1 .8 .8 schooling probability .6 .6 .4 .4 .2 .2 0 0 7 8 9 10 7 8 9 10 Poverty line Log 2008 expenditure incapita per Poverty line Log incapita 2008expenditure per 2008 2010 2008 2010 Data Source: 2008-2010 PSLM Panel Survey. The vertical line refers to 20th percentile in log per capital expenditure distribution. 2.47 Self-reported food shock increases the probability that children drop out of school between 2008 and 2010, the magnitude of the effect being larger for poor households. Self-reported food shock is also correlated with lower spending on education, particularly for girls. The reduction in spending occurred mainly in households above the poverty line. The higher the household income, the larger the gap between the households that reported a food price shock and those that did not. This is probably because school expenditures are already at the minimum for those households at the lower end of the income distribution, and there is no room for further reduction. 2.48 The reduction in education expenditure is larger for girls than for boys (Figure 2.17). The regression in Appendix 3 Table A.7 supports this finding, after controlling for other covariates. Figure 2.17: Food Shocks and Schooling Expenditures on Children Aged 5 to 14, by Gender Boys Girls 15000 15000 changes in schooling expenditure 10000 10000 5000 5000 0 0 7 8 9 7 8 9 Log per capita expenditure Log per capita expenditure no food shock food shock no food shock food shock Data Source: 2008-2010 PSLM Panel Survey. The vertical line refers to 20th percentile in log per capital expenditure distribution. 23 The impact of a health shock 2.49 Health shocks negatively affected caloric intake of Pakistani households, but only for the poor. Household health shocks usually involve catastrophic expenditures. Some households experience full or partial loss of earnings when a household member becomes ill and is unable to work, which directly affects the budget share available for food purchases and caloric consumption. Although health shocks are reported as experiences that have occurred in the past two years, and caloric intake is reported by month, the negative impact of a health shock on caloric intake is quite significant, as illustrated by Figure 2.18. This effect is particularly large for the poor and the near poor. However, the relationship between health shocks and child schooling is not statistically significant at the 95 percent level. Figure 2.18: Caloric Intake and Self-Reported Health Shock .8 Poverty line .6 .4 .2 0 -.2 7 8 9 Log per capita expenditure No Health Shock Health Shock Data Source: 2008-2010 PSLM Panel Survey. The vertical line refers to 20th percentile in log per capital expenditure distribution. 24 CHAPTER 3. PAKISTAN’S MAIN UNCONDITIONAL CASH TRANSFER PROGRAM, BISP, AND THE PROTECTION OF THE POOR 3.1 Prior to the sectoral reform process that started in 2007, publicly financed social protection programs in Pakistan were limited in adequacy, targeting efficiency, and their ability to respond to shocks. Pakistan’s safety net system comprised of two cash transfer programs, namely, the Food Support Program administered by the Bait-ul Maal and Zakat, covering approximately 3.2 million households in 2006/07. Both programs were weakly targeted to the poor: only 46 percent of total expenditures of Bait-ul-Mal (and 43 percent of total Zakat expenditures) reached the poorest 40 percent of the population. A variety of factors contributed to the weak performance of safety net programs in Pakistan, including the prevalence of small, fragmented and duplicative programs; low spending levels resulting in insufficient coverage and low benefit adequacy; governance challenges resulting in infrequent and irregular payments to beneficiaries; and lack of transparent eligibility criteria (World Bank, 2007). 3.2 These shortcomings were acknowledged by Pakistan’s first Poverty Reduction Strategy Paper, which called for a review of safety net programs. In turn, the Social Protection Strategy to Reach the Poor and the Vulnerable (endorsed by the cabinet in 2007) highlighted the need to reform ad hoc, reactive and fragmented sectoral interventions. With the introduction of BISP, the Government introduced a new institutional framework for safety net program implementation, by establishing an independent safety net agency to provide a uniform platform to implement rationalized safety net programs; the development of administrative policy guidelines for program implementation; the rationalization of overlapping safety net programs to improve fiscal and administrative efficiency; and, the development of graduation and exit strategies to facilitate households’ movement out of poverty. In this context, Bait-ul-Mal’s Food Support Program was discontinued. While Punjab province experimented with a food support program launched at around the same time with BISP, it was also discontinued in 2010. Zakat continues to function but its coverage remains low (Appendix 1). 3.3 The Government also aimed to improve the targeting efficiency of the new safety net program by adopting a proxy means test based beneficiary selection (poverty scorecard); and by establishing a national targeting system through appointing separate agencies for data collection and eligibility determination. In order to improve program administration further, pilots were carried out to evaluate the effectiveness of alternative payment systems (smartcard and mobile phone-based payments). Work has also started on the design of graduation programs to build on the main unconditional cash transfer. 3.4 Since BISP emerged as Pakistan’s main unconditional cash transfer program, unlike the broader coverage of the World Bank’s 2007 review of safety net programs in Pakistan, this chapter will focus solely on the BISP experience (see Box 3.1). This chapter provides an assessment of the positive and negative aspects of both the “original� BISP program—which relied on parliamentarian based beneficiary identification alongside exclusion criteria applied through the NADRA database—and the current/new BISP program, where eligibility is determined through an objective proxy means test (poverty 25 scorecard).9 The findings reported in this chapter are primarily based on the analysis of two household surveys: the 2009 BISP Beneficiary Survey and the 2008-2010 PSLM Panel Survey. This information is supplemented with insights from the 2010 World Food Program Survey data, qualitative assessments based on focus group discussions (OPM 2009), as well as the process evaluation and spot checks activities undertaken as part of the roll-out of the poverty scorecard-based targeting (Box 3.2). 3.5 This assessment of the BISP experience suggests that the program was able to respond quickly to the urgent needs of the poor and vulnerable population by making use of available information and tools. While imperfect, the initial phase of BISP targeting was in fact comparable, if not superior, to the performance of (then) existing social safety net programs in Pakistan. However, the household survey data also reveals that the decision to quickly move to a poverty scorecard-based targeting approach (as well as various institutional reforms described below) was also appropriate, since this objective targeting tool and experimentation with alternative payment approaches represent significant improvements over the original BISP design, alongside ongoing efforts to improve grievance redress procedures. 3.6 The objectives of this chapter are to describe the background and rationale for establishing the main unconditional cash transfer program in Pakistan: BISP; describe the technical and implementation aspects of establishing a national targeting system that BISP has adopted, and; discuss a broader range of vulnerabilities that prevail in Pakistan—which might not be captured by asset or consumption-based proxies of welfare—and implications for BISP as it continues to evolve. Section II assesses the performance of the original/parliamentarian-based BISP design. The new BISP design is discussed next. The final section turns to a broader discussion of vulnerabilities in Pakistan and implications for BISP highlighting areas (such as crisis response) where BISP has not taken the lead to date recognizing the rapidly evolving nature of the program and institutional capacity constraints. Box 3.1: Key Milestones in the Evolution of BISP to Date The government established BISP as its main safety net agency in FY08/09 to provide basic income support for the poor to address chronic poverty. In 2010, Pakistan’s Parliament unanimously approved the BISP Law, endorsing the institution as the country’s national safety net authority under the Prime Minister’s office. In July 2011, the beneficiaries identified by legislators during the first phase of BISP, prior to the PSC survey, were phased out and replaced by PSC targeted families. To date 2.7 million families have been paid. BISP has adopted a public private partnerships approach working with a variety of agencies including National Data Registry Authority (NADRA), the Post Office, commercial banks, NGOs and private sector institutions to operate the program. A basic monthly cash transfer payment of PKR 1,000 is paid to women in the identified poor families, based on their Computerized National Identity Card (CNIC). Since the launch of BISP three years ago, more than 15 million female citizens have registered for CNICs, largely due to their interaction with BISP, providing these women the possibility to vote, to have access to banking and other services. The budget allocation for BISP for the financial year 2012/13 is Rs70 billion, to provide cash assistance to 5.5 million families, which constitute roughly 18 percent of the population. 9 In December 2008, the Government of Pakistan decided to improve the existing targeting mechanism of BISP through the adoption of a poverty scorecard (a Proxy Means Test) used for the identification of beneficiaries. The pilot scorecard roll-out started during April 2009 in 15 districts. The Government has completed the national roll-out by the Summer of 2011. 26 Box 3.2: Brief Description of the Available Household Survey Data 2009 BISP Beneficiary Survey: The BISP Beneficiary Survey was conducted in 15 randomly selected districts of Pakistan, during the months of August and September 2009. The sample beneficiaries were selected from the list of the parliamentarian identified beneficiaries provided by the National Database and Registration Authority (NADRA). The sample was drawn from the NADRA computer records in April 2009. At that time, there were 1,660,584 beneficiaries in the record, comprising 46.23 percent from Punjab, 26.04 from Sindh, 18.35 percent from NWFP, 6.08 percent from Balochistan, 2.38 percent from FATA, 0.73 percent from Islamabad, and 0.14 percent from the Northern Areas (Gilgit-Baltistan). The study used the probability proportional to size method to obtain a sample of 2,500 households. For this purpose, a randomly selected sample of 5,680 beneficiaries’ names, addresses, NIC numbers, and names of spouses/fathers was provided by NADRA for the 15 selected Tehsils in the 13 districts of the four Provinces, including Islamabad from the capital territory, and Muzaffarabad from Azad Jammu and Kashmir. The total number of Districts/Tehsils was purposively decided based on an eventually representative sample size to capture the large diversity in levels of development across districts. The 13 distri cts from the four provinces were then allocated to roughly represent the relative share of the beneficiaries from each Province in the NADRA records on the day of sampling. As a result, seven districts in Punjab, three in Sindh, two in KP and one in Balochistan were selected. Having selected the districts, one tehsil in each district was drawn randomly. The survey consisted of six main modules to collect the necessary information: household roster; disability; household consumption; poverty scorecard; BISP awareness/processes; and use of BISP benefits. 2010 World Food Program Survey: The survey was carried out in February/ March 2010, to assess the food security situation in the militancy-crisis affected areas of KP and FATA. Six of seven agencies in FATA were covered (Mohmand, Bajaur, Kurram, South Waziristan, North Waziristan, Khyber, with the exception of Orakzai), as well as nine districts in KP (Lower Dir, Buner, Shangla, Swat, Tank, Dera Ismail Khan, Hangu, Swabi, Mardan). This area hosts a population of 13.1 million, which is around 8 percent of Pakistan’s population. A two-stage cluster sampling method was adopted to preserve strata representativeness; yet data are not representative at the district level. Four strata have been constructed by grouping the districts/ agencies, according to their shared history in regard to the militancy crisis. The sample consists of 1,800 households, with an even representation across strata. 2008-2010 PSLM Panel Survey: Description provided by Box 2.1 in Chapter 2. 27 II. ORIGINAL BISP DESIGN AND ITS PERFORMANCE Original BISP design: Making the most of available resources to address an urgent need 3.7 During the second half of 2008, the Government of Pakistan launched BISP as its main national social safety net program. A key motivation for establishing the new program was the recognition that existing publicly financed social protection programs in Pakistan are limited in adequacy, targeting efficiency, and ability to respond to shocks. 3.8 The short-term objective of BISP was to cushion the adverse impact of food, fuel, and financial crises on the poor, but its broader objective is to meet the redistributive goals of the country by providing a minimum income support package to the chronic poor, and those who are highly vulnerable to future shocks. The introduction of BISP as the Government’s main national cash -transfer program is consistent with the National Social Protection Strategy’s objective of creating a national safety net program to provide minimum income support to the chronic and transient poor, which could also serve as a common platform for other safety net programs. The program provides cash transfers of Rs. 1,000 [$12] per month to the eligible families and assisted 2,275,200 families in 2010. The budget allocation of Rs. 34 billion, doubled the Federal Government’s social safety nets spending from 0.3 percent in 2003/04 to 0.6 percent of GDP in 2008/09. The total safety net spending reached about 0.9 percent of GDP in 2009/10. 3.9 To quickly launch the program, parliamentarians (members of the National Assembly and Senators) were each asked to identify 8,000 beneficiary households on a prescribed form, on which names, national ID card, and household income information was collected. The resultant information was processed through ineligibility filters at NADRA. These ineligibility filters were: any family members employed by the government/ semi-government/authority/ department or armed forces of Pakistan; or drawing a pension from the government/semi government/authority/department or armed forces of Pakistan; or receiving any post-retirement benefits from any government department/ agency; or owns agricultural land more than three acres or residential house/ plot of more than eighty square yards (three marlas); or receiving income support from any other program like Punjab Food Support Scheme;10 or possesses a machine readable passport; or possesses a National Identity Card for Overseas Pakistanis (NICOP). How did the original BISP design perform in reaching the poorest households? 3.10 The original BISP design was implemented quite rapidly nationwide, but questions remained about the targeting performance of the parliamentarian-based selection of beneficiaries. This was primarily due to (i) heterogeneity among parliamentarians when it comes to their ability or willingness to reach out to and identify the poorest and most vulnerable households, and (ii) each parliamentarian receiving a fixed number of forms regardless of the prevalence of poverty in their constituencies. Furthermore, even if the parliamentarian based selection of beneficiaries may have performed adequately as a short-term stop-gap measure, the lack of an objective selection criteria and transparency concerns threatened the long-term sustainability of the initial BISP design. 10 The Punjab Food Support Program was discontinued in 2010. 28 3.11 One reason for documenting the performance of the parliamentarian based targeting in a timely manner was to better organize the transition from the original system to poverty scorecard based system. For this purpose, key questions not only included the overall targeting performance of the program, but also what percent of existing beneficiaries would still qualify for benefits under the poverty scorecard based targeting. The Pakistan experience would also be relevant for other countries to deal with crises situations when more objective and transparent systems are not readily available. 3.12 Beneficiary identification by parliamentarians has been pro-poor. Even though the original program’s performance was worse than what would happen under the poverty scorecard based targeting, it is better than the PBM Food Support Program in terms of percentage of benefits distributed to the poorest 40 percent of the population (see Figure 3.1). However, the poverty scorecard is clearly superior in reaching the poor, with over 75 percent of benefits reaching the poorest 40 percent of the population (no targeting instrument can perfectly approximate household consumption due to significant seasonal and yearly variations in household consumption documented by Chapter 2). Also, under the poverty scorecard-based targeting, a negligible share of households from the wealthiest 20 percent of the population would be eligible for benefits, while under the original BISP targeting almost 10 percent of the benefits were distributed to this group and this statistic is close to 20 percent for PBM Food Support Program and Zakat. Figure 3.1: Distribution of Benefits (Targeting Accuracy) By quintiles of household consumption per adult equivalent Data sources: BISP statistics are derived from the PSLM 2008-10 panel data and Bait-ul-Mal and Zakat statistics are from the 2007 World Bank 50 40 30 20 10 0 Q1 Q2 Q3 Q4 Q5 BISP (original targeting) BISP (poverty scorecard targeting) Bait-ul-mal:Food Support Program Zakat 3.13 In addition to analyzing the distribution of benefits to households with different consumption patterns, it is also of interest to document the distribution of poverty scores of the original BISP beneficiaries. An analysis of the distribution of poverty scores of first-phase beneficiaries would provide further insights on a system that was rapidly put in place to address an urgent need. 3.14 At around a poverty score eligibility cutoff point of 16 , about 20 percent of (original) BISP beneficiaries still qualify for benefits under the new system (see Figure 3. 2). A significant percentage of (original) BISP beneficiaries have poverty scores that are well above the national average of 15.6 in 2007/08. Having said that, on average, parliamentarian selected BISP beneficiaries do have lower poverty scores than the rest for the eligibility thresholds considered 29 by the Government. For instance, 24.4 percent of (original) BISP beneficiary households fall below the 17.5 cutoff, compared to 15.6 percent nationwide (based on PSLM 2007/08). Figure 3.2: Distribution of Poverty Scores among Original/Parliamentarian-Identified BISP Beneficiaries Source: Authors’ calculations based on the BISP 2009 Beneficiary Survey. 3.15 The poverty scorecard, of course, is not a perfect instrument for identifying the poor because it considers a limited set of characteristics. Parliamentarians may have taken into account other vulnerabilities in identifying prospective beneficiaries, such as disability or absence of a male household head, which are not included in the poverty score calculation. 11 Consequently, in theory, households ineligible for the program based on their poverty score might have been assessed as eligible by the parliamentarians because they have a disabled member or they are headed by a female. Here we examine the extent to which such considerations have guided the parliamentarian based beneficiary identification. Did the original BISP targeting focus extensively on the female headed households or the disabled? 3.16 About 35 percent of (original) BISP beneficiaries are female-headed households, and thus, parliamentarian-based beneficiary identification has clearly prioritized female-headed households. The proportion of female-headed households in the population is less than 9 percent in Pakistan (PSLM 2007- 08).12 Providing safety net benefits to female-headed households may be justified on the grounds of their special vulnerability (perhaps to unexpected shocks), even if they are relatively well-off in terms of 11 The reason for the exclusion of disability status is due to the well-known difficulties involved in assessing and verifying disability status. However, the BISP has allowed households with one or more disabled members to appeal for eligibility: upon verification, such households with a poverty score of 20 or less are then qualified to receive benefits. Household head’s gender is also not considered in the poverty score calculation because this indicator is not correlated with poverty, which could be due to the remittances received (in some cases from the spouse of female headed household working abroad), or because of household formation decisions (some of the poorest and most vulnerable females may not be in a position to head their own households). 12 The differences by urban/rural residence are negligible. 30 household possessions, schooling attainment, land and livestock ownership. Nevertheless, it is important to point out that in Pakistan female-headed households are not necessarily poorer. Figure 3.3 reveals that the poverty scores of female-headed households also tend to be higher among (original) BISP beneficiaries, revealing that some of the female-headed households were selected into the program despite being relatively wealthy. Figure 3.3: Distribution of Poverty Scores among Original/Parliamentarian-Identified BISP Beneficiaries (by Gender of Household Head) Source: Authors’ calculations based on the 2009 BISP Beneficiary Survey. 3.17 In contrast, there is no evidence that suggests that parliamentarian based beneficiary selection favored households with disabled members. The lack of a reliable estimate of national prevalence of disability makes it difficult to assess whether BISP’s parliamentarian targeting favors the disabled (in which case the disability rate among the BISP participants would be higher than the national rate). One of the few recent studies on the topic, Khan (2009), reports a 5.8 percent restricted functional limitation13 rate in KP, and 4.7 percent in AJK, but both of these areas were heavily affected by the 2005 earthquake, and these rates are likely to be above the national average. In comparison, the 3.5 percent restricted functional limitation rate among the members of BISP beneficiary households is probably close to the national average, suggesting that there was no emphasis on targeting households with disabled members. Also, households with a disabled member are evenly distributed across PMT-derived welfare levels in the 2009 BISP Beneficiary Survey. 13 A respondent is considered to have restricted functional limitation if he or she has a lot of difficulty or is unable to perform at all in any of the functional domains: seeing, hearing, mobility (walking or climbing stairs), remembering or concentrating. 31 To what extent did political connections help in becoming a beneficiary under the original BISP targeting approach? 3.18 Providing safety net benefits to female-headed households and households with a disabled member may be justified on the grounds of their special vulnerability, even if they fare relatively well in terms of the poverty score. Indeed, as discussed previously, BISP considers a more generous eligibility threshold (a poverty score of 20 or less) for households with at least one disabled member. However, other characteristics, such as political connections, should not influence the selection into an objective social safety net program. 3.19 Two definitions of political connectedness of a household are considered: (a) an index of political participation, a continuous variable running from 0 to 1, and (b) a discrete variable indicating contact with an official. The index of political participation is an unweighted average of responses to the following questions: How often do you vote in elections? How often do you discuss politics with friends, relatives, or fellow workers? How often do you try to persuade your friend, relatives or fellow workers to share your view? How often do you work for a party or a candidate? How often do you go to any political meetings, rallies or similar events in support of a particular candidate? How often do you watch any programs about political campaigns on television? How often do you listen to any speeches or discussions about political campaigns on the radio? Respondents’ answers, ranging from 1 to 4, are further transformed to the scale between 0 and 1, where 0 indicates that the respondent never participates in any of the listed activities and 1 indicates that the respondent often participates in all of them. The indicator of contact with an official takes the value of 1 if the respondent had direct contact or contact through an assistant with the following officials: (a) a Member of the National Assembly (MNA), or a senator; (b) Member of Provincial Assembly (MPA), or a Member of Legislative Assembly (MLA); (c) any other politician or civil servant to express one’s views. Otherwise, the indicator of contact with an official takes the values of 0. 3.20 An analysis of who among the original BISP beneficiaries would still qualify under the poverty scorecard based targeting reveals that those with strong political involvement or those who had a contact with a politician or civil servant were less likely to qualify under the poverty scorecard. In fact, the effects are so large that almost all of those with highest poverty scores are those with political connections.14 The significance of political variables need not necessarily signal corruption issues, however. Rather, this measure reflects a host of ways in which political involvement and contact with officials may improve one’s chances of learning about the program, as well as knowing how to go about obtaining, filling and submitting the relevant forms. Yet political connections naturally detract from the fairness and equity of the process of safety net benefit distribution. 14 The probability of an existing (parliamentarian selected) beneficiary’s qualification under the new proxy means test based system declines by about 28 percentage points as the political participation index increases from 0 to 1 (well-connected), after taking into account the effects of the gender of the household head, share of elderly household members, presence of a disabled individual in the household, urban/rural residence, parliamentarian’s gender, age and previous occupation. In the specificati on that leaves out the parliamentarian characteristics, the magnitude of the political participation index effect increases to 38 percentage points at the same direction. 32 What was the role played by informal contacts during application process, compared to formal application mechanisms? 3.21 Informal contacts—friends, family and “non-elected local influential persons,� 15 —played a central role at all stages of the original/parliamentarian based BISP application process, from learning about the program to submitting applications. The vast majority of the original BISP program participants learnt about the program from friends and family (49 percent), and the local influential persons (46 percent). Mass media ads reached a relatively small proportion of the population: about 8 percent of original program beneficiaries learned about the program from TV ads, 3 percent learned from radio, and 4 percent from the newspapers. 3.22 Informal contacts also figured prominently at all stages of the application process of the original BISP design. Nearly two-thirds of the original BISP participants approached a local influential person, a friend, or a family member to access BISP. Fifty-eight percent obtained BISP forms through family, friends or local influential persons, and nearly one-half of households had their forms verified by such an informal party, and ultimately filed the BISP application through them. Contrary to the expectation that mosques and health centers could be used to make BISP more accessible, they have played a nearly negligible role in disseminating information about BISP, or in the application process. 3.23 The original BISP application process has been largely free of charge, despite the fact that the application process offered scope for rent seeking. Only 5 percent of the surveyed BISP participants reported making any kind of payment during the application process, with most payments approximately 50 rupees or less. The few beneficiaries did so to obtain the BISP application form (2 percent), and for assistance in filling the BISP form (3 percent of those who needed help); with only 1 percent of respondents reported paying for form verification. What were the key issues encountered during payment delivery? 3.24 In contrast to the topics covered above, many of the issues encountered during payment delivery of the original / parliamentarian-based BISP would also apply to the new system that relies on the poverty scorecard for beneficiary identification. BISP is currently experimenting with alternative payment mechanisms based on smartcard and mobile-phones. 3.25 Under the current arrangements, BISP payments are supposed to be delivered “at the door of the beneficiary� by the postman to minimize the indirect costs (cost of traveling to a remote location to collect the payment) to the beneficiaries. Yet, the BISP beneficiary survey data reveals that in 42 percent of cases the payments were received at a post office. Over half of beneficiaries in rural areas had to travel two or more kilometers to collect their BISP payments. 3.26 A more serious concern has to do with the prevalence of baksheesh payments, or tips for delivering the BISP cash transfers. Unlike payments during the application process, baksheesh payments were found to be widespread at the time of payment collection. One-third of households reported paying baksheesh at the time of collection of their payments. The incidence is higher in urban areas than in rural areas (39 and 27 percent respectively). A slightly greater proportion of beneficiaries paid baksheesh when it was delivered by the postman (32 percent of cases), as opposed to collecting at the post office (28 15 The local influential persons are non-elected de facto community leaders, such as village heads or large landowners. 33 percent of cases). Seventy-eight percent of baksheesh payments were equal to or less than 100 rupees, however, in approximately one-fourth of cases—most of them in rural areas—exceeded 100 rupees. In relative terms, baksheesh payments account for approximately 3 percent of the total BISP transfers received by households. How did the beneficiaries spend the cash transfers? 3.27 Beneficiaries report spending BISP assistance on the most basic and pressing needs. About 32 percent of the cash transfers were spent on basic food items, followed by medical expenses (23 percent), debt payments (10 percent), and clothes (7 percent). Spending on children's education accounts for only 5 percent of the transferred amount in the overall sample, although urban households spent a greater proportion of their benefits on education than rural households (9 percent versus 4 percent). About 3 percent of the transfers went to savings. 3.28 These spending trends on food, medical expenditures, and clothes require little explanation since they are commonly observed in other developing countries as well. However, some background information on debt payments can be informative in the Pakistan context. In the absence of timely cash transfers, some households are able to borrow money, but not always with favorable terms and over 27 percent of Pakistani households report being indebted. In cases where households are not able to borrow, or when they are able to borrow only in harsh terms (which might include bonded-child-labor arrangements), their inability to borrow against future income creates inefficiencies, in terms of households’ ability to smooth consumption, but also to invest in the human capital of children. A larger proportion of illiterate and eligible households are indebted than their literate or non-eligible counterparts; and shock-affected households are more likely to be indebted. III. THE NEW BISP DESIGN Key design changes 3.29 After the launch of the original BISP to address an immediate need, the Government sought to reform the program. Certain aspects of the program remained the same, including the delivery of the payment to adult females, and the requirement for the eligible females to have NADRA IDs to receive payments. Both of these features of the program are believed to be associated with positive externalities. The delivery of payments to adult females are shown to have led to increased investments in children’s human capital in other countries, also perhaps positively influencing female empowerment (both these possibilities will be investigated in the Pakistan context through the ongoing BISP impact evaluation exercise). The program also increased the share of poor households that have NADRA IDs, which is a requirement for opening a bank account, getting a driver’s license, mobile phone etc. 3.30 One set of changes aimed to improve the targeting efficiency of the program by adopting a proxy means test based beneficiary selection (or poverty scorecard); establishing a national targeting system through appointing separate agencies for data collection and eligibility determination; and adopting a sound policy on transitioning from the old to the new targeting system. The Government also introduced a new institutional framework for safety net program implementation, by establishing an independent safety net agency to provide a uniform platform to implement rationalized safety net programs; the development 34 of administrative policy guidelines for program implementation; the rationalization of overlapping safety net programs to improve fiscal and administrative efficiency; and, the development of graduation and exit strategies to facilitate households’ movement out of poverty. Yet another set of changes are still in the works, including the piloting of alternative payment systems (smartcard and mobile phone-based payment) and also design of graduation programs to build on the main unconditional cash transfer. 3.31 The Government considered alternative approaches and adopted an objective targeting instrument—the poverty scorecard—to deliver program benefits to the poor and vulnerable (Box 3.3). The poverty scorecard is a proxy means test (PMT)—a targeting tool that is used in many countries to identify the poor for provision of social assistance. The poverty scorecard, or PMT, includes a limited number of indicators that correlate well with poverty along with a simple scoring system to help identify the program beneficiaries. The poverty scorecard-based targeting is an improvement over the previous system in a number of ways: (i) a transparent and fair enrollment process that allows all to apply for it; (ii) objective and transparent criteria for determining the eligibility; (iii) adoption of verification tools to reduce exclusion and inclusion errors; (iv) gains in targeting efficiency shown previously; as well as, (v) provision of a welfare ranking of households that can be used for supplementary (graduation) programs that may use different eligibility cutoff points. 3.32 The poverty scorecard relies on a smaller set of easily verifiable household and individual characteristics to serve as proxies for household per capita consumption. The list of characteristics for poverty scoring has been identified by prior analysis (Vishwanath et al, 2009). Although not as accurate as a full consumption aggregate to measure poverty, the poverty score requires considerably less information, which is easy to collect and verify. Furthermore, the poverty score is less susceptible to the short-term fluctuations in welfare since it relies on household characteristics, which are considerably more static (for example, number of dependents, size of land owned, housing characteristics and selected household assets).16 16 Ownership of certain key assets affects the poverty score, and insofar as the beneficiaries spend their BISP payments on these assets, the poverty score may reflect the effect of transfers already received. However, this is an unlikely scenario since our findings indicate that a very small proportion of the BISP payments are spent on assets. 35 Box 3.3: Alternative Approaches to Targeting the Poor Means Tests: A verified means test is usually regarded as the gold standard of targeting. It seeks to collect (nearly) complete information on households’ income and/or wealth and verifies the information collected against independe nt sources. Where suitable databases exist and interagency cooperation can be obtained, information may be verified by cross-linking the registries of, say, the welfare agency, property registrars, tax authorities, social security agencies, and the like. When this is not possible, households may be asked to submit copies of records of transactions, such as pay stubs, utility bills, or tax payments. In the best of cases, means testing can be extremely accurate. However, means tests work best in situations of high levels of literacy, low informal sector activity and strong documentation of economic transactions. Proxy Means Tests: Proxy means tests generate a score for applicant households based on fairly easy-to-observe household characteristics, such as the location and quality of household’s dwelling, its ownership of durable goods, its demographic structure, and the education and possibly occupations of its adult members. The indicators used to calculate this score and their weights are derived from statistical analysis (usually regression analysis or principal components analysis) of data from detailed household surveys, of a sort too costly to be carried out for all applicants of large programs. The advantage of proxy means testing is that it requires less information than true means testing, yet is objective. Community-Based Targeting: Community-based targeting uses a group of community members or leaders whose principal functions in the community are not related to the transfer program to decide who in the community should benefit. The advantage of community-based targeting is that it relies on local information on individual circumstances, which may be more accurate and less costly to collect than using other methods. Geographic Targeting: With geographic targeting, location determines eligibility for benefits. People who live in the designated areas are eligible and those who live elsewhere are not. Few programs target only on the basis of geography, but many programs use geographic targeting in conjunction with other targeting methods, especially when programs are not fully funded. In such cases poverty maps can be used to focus the program in only some areas of the country or to allocate spaces in the program among subnational jurisdictions. The advantage of geographic targeting is that it is administratively simple, requiring none of the machinery for individual assessment programs described above. Demographic Targeting: The usual and simple forms of demographic targeting are based on age, with child allowances and social pensions being the most common. Part of the rationale is that individuals may be particularly vulnerable in childhood and old age. The logic is somewhat clouded by the reality that most individuals, certainly children, and in most countries the large majority of elderly as well, live in households with several individuals and generally one or more income earners. Because households tend to pool their resources, at least in part, many children and elderly do not live in poverty even though they do not generate income for themselves. Demographic targeting is obviously administratively simple. Self Targeting: Self-targeted programs are technically open to everyone, but are designed in such a way that take-up is expected to be much higher among the poor than the non-poor or the level of benefits is expected to be higher among the poor. One of the most common applications of self-targeting in social assistance is the use of low wages in public works programs to induce participation only by the poor. The administrative costs of the targeting are quite low, although administering public works programs is not simple. Source: For more detailed information including a comprehensive discussion of the pros and cons of the targeting options summarized here, see For Protection and Promotion, The World Bank 2008. 36 Coverage and targeting implications of the new design 3.33 Under the eligibility cut-off point of 16.17, about 18 percent of the population (corresponding to 4.8 million families) would be covered by the program. The key factors that influenced the Government’s decision on the cutoff point included the prevalence of poverty in Pakistan (which is believed to be around 20 percent), the targeting performance of the poverty scorecard instrument, and the availability of budget (see Table 3.1). Table 3.1: The Coverage and Budgetary Requirements of BISP under Different Poverty Score Eligibility Cut-off Points Coverage (% Families) Budget Implications Family (Rs. Billion) Cutoff scores % of population Rs. 1000/m Percent Million 15.79 12.9 4.7 17.0 56.0 16.04 13.4 4.8 17.6 58.1 16.10 13.7 4.9 17.9 59.3 16.17 13.8 5.0 18.0 59.6 16.65 14.6 5.3 19.0 63.2 17.45 16.3 5.9 21.0 70.5 18.28 18.0 6.5 23.0 78.2 19.04 19.9 7.2 25.0 86.0 19.90 21.6 7.8 27.0 93.6 3.34 As previously discussed, the adoption of the poverty scorecard as a targeting instrument represents a significant improvement in terms of the share of benefits that reach the poorest segments of the country, when compared not only with the previous approaches tried in Pakistan (see Figure 3.1). 3.35 Similarly, the poverty scorecard based targeting in Pakistan compares well to many other developing countries, with about 47 percent of the funds expected to be allocated to the poorest quintile of the population (see Figure 3.1). Even though some countries such as Bulgaria do achieve significantly better performance, it is important to recognize that in countries where poverty is in low single digits, it is much more important for safety net programs to be able to converge on that small group. In the Pakistani context, however, a simple and objective targeting tool that allocates a large share of the funds (over 75 percent) to the lowest two quintiles emerges as an extremely valuable tool especially because many households in the low quintiles move in-and-out of poverty over time (as shown by Chapter 2). 3.36 However, one can think of at least three situations where the provinces or external development partners might want to utilize the BISP database with different eligibility cutoff points. The first has to do with provision of assistance during crisis. While targeted relief can be provided to the IDPs etc., there are also concerns that such categorical targeting does not necessarily capture the poorest households. Poverty scorecard data collected prior to a crisis can be utilized to provide supplementary assistance to those who qualify under the BISP eligibility cut-off point, and depending on the nature of the crisis to those who have poverty scores that are slightly above this eligibility threshold. Second, there may be payoffs to operating with more generous safety net program eligibility criteria in areas that experience longer-term 37 challenges. Militancy affected areas of KP and FATA represent one such situation. Finally, certain human development programs may have to set more stringent eligibility criteria (focusing on a subset of BISP beneficiaries), or more generous eligibility criteria (operating under a different cutoff point), depending on the sector and development objective. Administrative challenges faced during test phase of poverty scorecard data collection 3.37 The poverty scorecard data collection in 15 districts was accompanied by third party process evaluation and spot checks. The scale of the effort has been significant, due to the Government’s decision to collect poverty scorecard information utilizing a census approach (as opposed to accepting applications through targeting centers) to reduce the likelihood of excluding the poorest and most vulnerable groups. Key lessons learned from these assessments were utilized to inform the nationwide rollout of the new targeting approach. Here, we summarize key lessons learned. Key findings from process evaluation of the test phase 3.38 The poverty scorecard data collection for the 15 districts was carried out under BISP management, through three partner organizations: Rural Support Program Network (RSPN), Pakistan Poverty Alleviation Fund (PPAF), and Pakistan Census Organization (PCO). 3.39 Process evaluation took place during poverty scorecard data collection — albeit at a smaller scale than first envisaged — providing weekly inputs to the BISP management and the three partner organizations. These three partner organizations represent different models of organization with their own strengths and weaknesses. During actual data collection, a significant amount of heterogeneity prevailed among the way partner organizations carried out their tasks. Despite the detailed targeting manual that was made available to the partner organizations, these documents were not always sufficient to anticipate actual ground conditions and provide ready answers to problems and situations that arise in the field. Another factor that led to the high degree of heterogeneity in the interpretation of the Terms of References (TORs) and targeting manual, both across and within partner organizations, was that real-time course correction mechanisms were employed only exceptionally. 3.40 None of the three organizations was experienced at carrying out a survey of the type that was required by the test phase. Even the PCO’s own experience was limited to a simpler form. While some of the RSPN local partners had some experience of implementing scorecard surveys, this is a relatively recent development, and its results have not been scrutinized by third parties. The PPAF and its partners were also new entrants into the field. All three partner organizations emphasized the use of local personnel for enumeration and field level supervision, but their management approaches were different. The PCO’s case was unique amongst the three partner organizations; as a public agency working with public servants, it was restricted in terms of options for the structure of the remuneration package. Both the RSPN and the PPAF attempted to link remuneration to performance, or in the PPAF’s case, to encourage its sub-partner organizations to do so. 3.41 Lack of awareness in the communities also led to negative impact on the quality of enumeration. Additionally, lack of awareness regarding the objectives may have led to faulty or incomplete information being provided by the survey respondents. 38 3.42 Even though arrangements were made with NADRA to increase the share of adults with NADRA IDs prior to the poverty scorecard data collection, in practice, there were a number of difficulties. These ranged from NADRA teams not showing up, to NADRA teams visiting an area but without necessary information exchange. 3.43 Other issues emerged during piloting in the main urban test site. First, it was a challenge to incorporate unplanned areas (slums and katchi abadis). There was lack of clarity on the inclusion of these areas, and further, it was difficult for teams to address them, as they were not necessarily included on formal maps. Similarly, illegal construction on government lands (e.g., alongside railroad tracks) was not covered in the guidelines. Other exclusion issues related to lack of willingness on part of the POs/ enumerators to cover minority groups, and the exclusion of domestic staff (servants/maids). In the absence of a clear and effectively designed and executed public information campaign, enumerators also encountered resistance on the part of some households to participate in the survey. Key findings from spot checks of the test phase 3.44 The spot checks took place after the poverty scorecard data collection by the partner organizations, to verify the quality of information collected during the test phase. For this purpose, 13,000 households were re-interviewed in six districts. 3.45 On the positive side, when households were matched, it was clear that the accuracy of data collection was pretty good with strong overlaps in household characteristics, including the schooling attainment, share of dependents etc. Similarly, information collected on 14 out of 19 household possessions were similar both during original data collection and during spot checks for the smaller sample. The household possessions that diverged between the two data collection efforts were air cooler, cooking stove, buffalo, cow and goat. The explanation for the first two items might have to do with the presence of similar items in the possessions list: namely, air conditioner and cooking range. It is not clear why the reporting on livestock diverges. In addition to the quality issues during interviewing, other potential reasons for the divergence in results may be low literacy level of population; time lag between the survey implementation and spot checks (varying from 11 weeks to 21 weeks); change in respondent, the extent to which female enumerators were employed (more during the spot checks); and greater awareness amongst people regarding eligibility conditions of the BISP program. 3.46 On the negative side, about 38 percent of households covered by the spot checks were not covered by the test phase data collection. The percentage of households that are not covered did not vary much by partner organization. One reason for such a low coverage during the test phase may have to do with the incentives provided to the partner organizations (in particular, the estimated number of households in each region), and also the inadequate maps used by the survey teams. These identified issues were addressed during the national rollout of the program. Emerging findings from a qualitative assessment on the perceptions of BISP 3.47 A quantitative evaluation of BISP is underway, but since it will take some time to document the impact evaluation results, here we summarize insights from focus group discussions carried out in four districts, one from each province (see OPM 2010 for more information on methodology). Discussions in eight communities that participated in poverty-scorecard implementation suggest that the BISP cash 39 transfer has made a significant impact on the recipients’ ability to deal with shocks. The participants also indicated that the BISP transfers help to relieve tensions within households by improving relations between the recipients and their husbands. The recipients thought the poverty scorecard based selection was carried out fairly and universally. Interestingly, CNIC processing (a requirement for receiving BISP benefits) was not given much prominence by the respondents. The key bottleneck that the respondents highlighted was irregularities in the distribution of benefits through the postal service, which is consistent with the BISP beneficiary survey results. In order to minimize the irregularities in the distribution of benefits, BISP is in the process of rolling-out debit-card/smart-card based payments. These indications from the focus group discussions will need to be further verified through the analysis of household survey data collected as part of quantitative BISP impact evaluation. IV. THE CHALLENGES OF SAFETY NET PROGRAM IMPLEMENTATION IN MILITANCY AFFECTED AREAS 3.48 There are strong arguments for BISP to operate a base income program utilizing the same criteria nationwide. Among other things, this approach would help ensure that a household living in similar conditions anywhere in Pakistan would have roughly the same probability of qualifying for this safety net program. In other words, by design, more funding would go to areas where poverty is more prevalent. This approach also gives a sense of fairness, which is important for the long-term sustainability of a national safety net program. 3.49 However, the living conditions in militancy-affected areas are significantly worse than in the rest of Pakistan, since dismal security situation has implications for the availability and quality of services as well as for the functioning of markets. Thus there may be a need to consider the design and implementation of (complementary) safety net programs in selected regions. The distribution of poverty score is shown by Figure 3.4, separately for Pakistan as a whole, for KP, and for militancy-affected areas. Two key messages emerge. First, the distribution of households is shifted to the left in militancy affected areas (i.e., there is more extreme poverty). Second, (asset) poverty has increased in 2008/09, compared to 2007/08. Figure 3.4: Distribution of Poverty Score in Militancy Crisis Areas, in Comparison to Khyber Pakhtoon Khwa (KPK) and Pakistan .04 .05 .04 .03 .03 density .02 .02 .01 PovScore = 16.17 PovScore = 16.17 .01 0 0 0 20 40 60 0 20 40 60 Poverty score Poverty score Pakistan KPK Militancy crisis areas Pakistan KPK Militancy crisis areas Source: 2007-2008 PSLM Source: 2008-2009 PSLM 3.50 Militancy-affected areas are also significantly different from the rest of the population in terms of indebtedness. As discussed previously, nationwide, over 27 percent of households report being indebted. 40 In militancy-affected areas, this percentage is even higher at 45 percent. The likelihood of being in debt — for those in the militancy-affected areas who would qualify for BISP assistance under the poverty scorecard targeting mechanism—is even higher at 54 percent. 3.51 The reasons surveyed households provide for debts are illuminating, in that they are predominantly to cover very basic requirements: 44 percent of households turn to debt to buy food, 28 percent to cover health expenditures, 7 percent to repair their house, 5 percent to cover routine expenses (transportation, clothes, shoes, and school-related spending), only 7 percent for productive investments (renting land, buying/feeding animals, buying agricultural assets), 3 percent to pay for ceremonies, and 6 percent for other purposes. Figure 3.5 also indicates that debt payments are among the main difficulties faced by 16 percent of households in the crisis affected areas. 3.52 Two main difficulties faced by households in militancy-affected areas are common to the rest of Pakistan: food price increases and health expenditures. In addition to the debt issue discussed above , a sizable share of the population are affected by the militancy directly, with over 16 percent of households mentioning high levels of insecurity, almost 8 percent mentioning strict rules imposed by the militants, 28 percent mentioning curfews/bans, 12 percent mentioning fighting and military attacks, 8 percent mentioning displacement, and 6 percent mentioning their house being damaged during fighting. In summary, all these factors make the case for adopting an eligibility cutoff point that is more generous for the militancy-affected areas, supplemented with additional incentives to enable human capital investments. Figure 3.5: Main Difficulties Faced by Households in Militancy Affected Areas Source: Background paper by Pop (2010), utilizing WFP Survey data. 41 APPENDIX 1 SAFETY NET PROGRAMS IN PAKISTAN: AN OVERVIEW I. Introduction 1.1 This section summarizes the evolution of social safety nets in Pakistan over the past few years. It begins with an assessment of government commitments to social protection, and of the main social safety net programs prior to the Benazir Income Support Programme (BISP), namely Zakat and Pakistan Bait-ul-Maal. It also details social welfare and other safety net initiatives by federal and provincial governments. We then outline the implementation arrangements for these programs and the main problems and challenges faced. This provides the context for the major reforms in social protection undertaken in 2007-8, notably the launch of BISP. 1.2 Before focusing on safety net programs, it would be useful to have an indication of the full range of social protection mechanisms in the country. Social protection mechanisms in Pakistan are both formal and informal. Formal mechanisms include programs by federal and provincial governments for cash transfers, education stipends, and food support; pension schemes and other benefits for workers in the formal sector (both government and private sector funded); labor policies and employment generation programs; child protection and disaster management initiatives. 1.3 Social protection provision in Pakistan is not confined to government. Several NGOs and civil society organizations are engaged in similar endeavors, providing assistance to the poor. Key organizations include Edhi Foundation, the Aga Khan Rural Support Program (AKRSP), Sind Institute of Urology and Transplantation, Orangi Pilot Project, and SUNGI. Set up in 1948, the Edhi Foundation is now the largest welfare organization in Pakistan. It provides a range of relief services to the poor: soup kitchens, medical treatment in its many clinics and hospitals, homes for orphans and the handicapped, women’s shelters, legal and other support for prisoners, emergency disaster relief and so on. 1.4 Despite the plethora of government and civil society schemes, large numbers of Pakistanis depend primarily on informal safety nets of private charity and support from family, i.e., biraderi (extended family/kinship groups) and community networks. The extended family system is strong in Pakistan, particularly in rural areas. When facing hard times, biraderi and similar social networks can be an important source of support. Access to these networks is, by definition, limited to members of a particular tribe, caste or ethnic group. 1.5 Finally, the poor resort to a range of ‘self-help’ coping mechanisms: sale of assets (jewelry, land, livestock); taking loans (usually at exorbitant rates); pulling children out of school to work; cutting back on expenses (food, education, healthcare); and turning to illegal activities for income generation. While these mechanisms may provide short-term relief, they entail the erosion of assets (including education), and thus in the long-term may trap people deeper in poverty. 42 II. Policy Commitments 1.6 Pakistan has always had a strong policy commitment to social protection. Article 38 [(d) and (e)] of the Constitution stipulates: The State shall provide for all persons employed in the service of Pakistan or otherwise, social security by compulsory social insurance or other means; provide basic necessities of life such as food, clothing, housing, education and medical relief, for all such citizens, irrespective of sex, creed, caste, or race, as are permanently or temporarily unable to earn their livelihood on account of infirmity, sickness or unemployment; reduce disparity in the income and earnings of individuals. As in other developing countries, establishing the right to social protection by means of a constitutional decree may not translate into actual social protection cover, due to competing demands over scarce resources. 1.7 Under MDG 1, the Government of Pakistan has committed to halve between 1990 and 2015, both the proportion of people whose daily income is less than one dollar, and the proportion who suffer from hunger. Key policy documents further recognize and prioritize social protection. The 2003 Poverty Reduction Strategy Paper (PRSP) had social protection as the fourth of its core pillars (‘Targeting the Poor and Vulnerable’). It provided for targeted interventions to address poverty and generate income and employment. The 2005 Medium-Term Development Framework (MTDF) specifically recognized ‘social protection as a means of strengthening poor people’s capabilities to mitigate and manage risk and vulnerability, and thus, to impact positively on underlying poverty and inequity’. 1.8 None of the four provincial governments have a specific social protection policy. The Government of Punjab’s White Paper on the Budget 2010/11 describes the vision of the provincial government in relation to social protection as to: ‘Strive to attain the ultimate objective of welfare state according to collective social and ethical aspirations of people free from exploitation, deprivation and discrimination.’ Specific objectives are to ‘create welfare facilities and healthy living opportunities for vulnerable groups,’ promote gender mainstreaming, promote skills development for income generation, and tackle problems of drug abuse and child begging. 1.9 The KP Social Welfare Department describes social welfare as ‘concerned with the well-being and uplift of the community at large and the vulnerable groups in particular.’ The focus of its programs is on ‘the neglected, disadvantaged, under privileged and exploited section of our society.’ The Sindh Social Welfare Department describes its vision as to: ‘Provide enabling environments and tangible opportunities through policies, programmes and projects that would contribute in promotion of social progress and socio-economic uplift by addressing the needs of the marginalized and vulnerable segments of the society .’ The Balochistan Department of Social Welfare, Special Education and Literacy/Non-Formal Education describes its vision as: ‘To provide better facilities to socially disadvantaged people and to empower women.’ 1.10 Prior to the launch of BISP, the two main social safety net programs in Pakistan were Zakat and Pakistan Bait-ul-Mal (PBM). While BISP has replaced them as the country’s main social safety net program, both are on-going although PBM’s food support program has been discontinued. 43 III. Zakat 1.11 The Zakat system, introduced in 1980 by President Zia Ul Haq, is based on the Islamic injunction for Muslims who can afford it, to pay 2.5 percent of their savings to help the poor. Zakat funds are used exclusively to help needy Sunni Muslims. 1.12 The largest share (60 percent) of regular Zakat funds goes to the Guzara Allowance, monthly payments of PKR 500 to poor households. Permanent rehabilitation grants were discontinued in 2008/09. The principle behind Zakat disbursal is to accommodate the maximum number of people, rather than to concentrate resources on the few. Education stipends are paid to pupils/students from primary to university level; there is separate provision for students of deeni madaris (religious schools). Zakat beneficiaries are eligible for help with health care expenses in government facilities; the ceiling is PKR 2,000 for outpatients and PKR 3,000 for inpatients. Unmarried women can receive marriage grants of up to PKR 10,000. Zakat funds are also used to help the needy in emergency/disaster situations. In 2010/11 PKR 100 million was distributed to internally displaced persons (IDPs) and PKR 4 billion among the four provinces to help flood affectees. Table A1 summarizes resource allocations among the various Zakat programs. Table A2: Allocation of Zakat Funds # SECTORAL HEAD ALLOCATION ALLOCATION FOR 2009-10 FOR 2010-11 PKR million A. REGULAR ZAKAT BUDGET 1. Guzara Allowance 60% 2,225.429 2,225.429 2. Education Stipends 18% 667.629 667.629 3. Stipends to Students of Deeni Madaris 8% 296.723 296.723 4. Health Care 6% 222.542 222.542 5. Marriage Assistance to Unmarried Women 8% 296.726 296.726 Sub-Total: 3,709.049 3,709.049 B. OTHER ZAKAT PROGRAMS 6. Educational Stipends (Technical) 1628.920 1,228.920 7. Eid Grants 185.455 185.455 8. Leprosy Patients 1.500 1.500 9. National Level Health Institutions 932.142 828.450 44 10. Model Deeni Madaris 12.000 12.000 11. i. Natural Calamities 200.000 500.000 ii. Man-made (a) IDPs 300.000 -- (b) Affectees of Terrorism 54.017 -- iii. Hepatitis C 100.000 100.000 12. Administrative Expenditure 432.071 543.269 13. Amount reserved for enhancement 278.804 Nil 14. Media Campaign 72.000 -- 15. Supplementary Funds for Dist. Zakat 8.578 10.785 Committees Sub-Total: 4,205.487 3,410.379 Grand Total (A+B) 7,914.536 7,119.428 Source: Ministry of Zakat website. 1.13 Zakat funds are collected from bank savings accounts, fixed deposit certificates, government securities and other sources, as well as voluntary contributions. These are deposited in a Central Zakat Fund (CZF). Accumulated reserves in the CZF make up the shortfall in collections to fund Zakat programs. In November 2010, reserves were approximately PKR 10 billion. Once CZF reserves are depleted, Zakat programs will be entirely dependent on contributions, the reduced funding will likely lead to a cut in programs, unless alternative funding sources can be found. 1.14 Funds are transferred from the CZC to provincial departments in two installments, on July 1 st and January 1st of each financial year. Zakat funds are distributed among the four provinces based on population. The Northern Areas and Islamabad Capital Territory each receive a lump sum of less than 1 percent. The allocation of provincial Zakat funds to different programs is carried out in accordance with shares decided federally, i.e., 60 percent on Guzara allowance, 18 percent on education stipends, 8 percent for Deeni Madaris students, 6 percent for health care, and 8 percent for marriage assistance. 1.15 In terms of geographic distribution, Provincial Zakat Councils distribute funds to districts based on population; District Zakat Committees then make allocations to Local Zakat Committees based on needs. Committee members are supposed to be respected local people; they serve on a voluntary basis so administrative costs are low. However, there is no objective targeting mechanism. Furthermore, monitoring (to assess continued need and eligibility) relies solely on committee members’ local knowledge. There is no automated information management system; records are at best kept on paper. 45 1.16 Punjab receives 57.36 percent of Zakat funds; the allocation for 2010/11 was PKR 2.04 billion. There are around 25,000 Local Zakat Committees in Punjab, the heaviest concentration in the country. People seeking Zakat support apply for this to the LZC; criteria for eligibility include a monthly income of less than PKR 670 (based on the old poverty line) and not receiving support from any other source (e.g. Bait-ul-Mal). 1.17 Khyber Pakhtunkhwa receives 13.82 percent of Zakat funds; the allocation for 2010/11 was PKR 705 million. Beneficiaries are decided by the LZC members based on their local knowledge: the aim is to reach the ‘poorest of the poor.’ However, a 2006 Asian Development Bank (ADB) study found that ‘the names of 42 percent of Zakat-receiver households were included in the list (of recipients) on the recommendation of local councilors or other influential persons, such as local landlords, religious leaders or relatives of members of Zakat Committees (Arif 2006). Once applicants are accepted as ‘Mustahiqeen’ they tend to retain that status indefinitely. In FY 2009/10 400,786 benefited from Zakat programs, with maximum recipients of Guzara Allowance (105,457) followed by Education Stipends (154,101). 1.18 Sindh receives 23.71 percent of Zakat funds, with the allocation for 2010/11 approximately PKR 1 billion. The total number of beneficiaries in the province in 2009/10 was 349,240. Karachi City District had the highest number of beneficiaries (88,849), followed by Dadu (19,270), and Ghotki (18,321). The lowest number of beneficiaries was in Jamshoro (2,812). Most records are kept manually at district and provincial levels, but some information is automated, and there are plans to automate the entire system, including records of beneficiaries, Zakat collection and distribution. There is no sharing of records/databases with other social safety net programs. 1.19 Balochistan currently receives 5.11 percent of provincial allocations from the Central Zakat Fund; based on the Punjab allocation (57.36 percent; PKR 2.04 billion), the 2010/11 allocation for Balochistan was around PKR 182 million. On the ground distribution of Zakat assistance is the responsibility of Local Zakat Committees; each member chooses ten Mustahiqeen. Review of beneficiaries is carried out every six months. IV. Pakistan Bait-ul-Mal 1.20 The Pakistan Bait-ul-Mal (PBM) was established in 1992 to provide support to the destitute—widows, orphans, the elderly, the disabled and particularly those ineligible for Zakat support (e.g. minorities). It covers all parts of Pakistan, including AJK, FATA and the Northern Areas. The organizational set-up comprises a head office, five regional offices and 144 district (field) offices, making for a total work force of 2,200. 1.21 The PBM is funded entirely by the federal government. For many years, the biggest PBM program, both in terms of resources and beneficiaries, was the Food Support Programme (FSP). This was discontinued in 2008/09 because of concerns over targeting, as well as the launch of the Benazir Income Support Programme (BISP). The Tawana Pakistan initiative, aimed at providing cooked meals to school-going children, was also disbanded, mostly for logistical reasons (the need to arrange cooking facilities, etc.), but also because of targeting concerns. The major current PBM programs are summarized in Table A2. 46 Table A2: Allocation of Resources for Bait-ul-Mal Programs Program Resource Allocations FY 2010-11 (PKR million) Child Support Programme 120 Individual Financial Assistance 692 Miscellaneous Projects 500 NGOs 60 Administrative Expenses 627 TOTAL 2,000 Source: PBM Head Office, Islamabad 1.22 The Child Support Programme (CSP) aims to increase primary school enrolment and reduce drop-out rates by tying cash transfers to education. Families with one school-going child are paid PKR 300/month and those with more than one receive PKR 600/month. CSP was launched in 2006/07 in three pilot districts (Bhakkar, Punjab; Tharparkar, Sindh; and Kohistan, KP). In 2008-09, the program was scaled up to eight additional districts: Ghanche (GB), Kharan (Balochistan), Rawalpindi (Punjab), Nawab Shah (Sindh), Quetta (Balochistan), Multan (Punjab), Muzafarabad (AJK), and Abbotabad (KP). 1.23 CSP targets former FSP beneficiaries, but applies additional ‘filters,’ considering the poverty status of the household as well as the presence of primary school age children. Targeting is thus less discretionary than under FSP, and there are provisions to monitor continued eligibility and impact. A rapid impact assessment one year into the pilot scheme found that the compliance rate was good and drop-out rates among beneficiaries were low. CSP is being expanded to a further eight districts, with planned coverage of 40,000 households and 80,000 children. The Government funds CSP, and the World Bank provided technical support to set up the pilot and carry out its evaluation. 1.24 With the closure of FSP, Individual Financial Assistance (IFA) accounts for the biggest share of PBM expenditure. IFA is provided largely for medical treatment and for higher education. Money for medical treatment is paid directly to government hospitals; the maximum is PKR 400,000/case, but there is provision for greater amounts. Similarly, funds for higher education are paid directly to government colleges. Since 1992, a total of PKR 4,990 million has been disbursed as IFA among approximately 443,000 beneficiaries. 1.25 Significant PBM projects include National Centres for Rehabilitation of Child Labour (NCRCL), Pakistan Sweet Homes and Vocational Dastkari Schools. NCRCL are designed to wean children aged 5-6 away from hazardous labor through provision of free education, clothing, footwear and stipends to children, and a monthly subsistence allowance to the parents. The Sweet Homes are for orphans, and one of its objectives is to counter the appeal of madrassahs; children are given shelter, food, clothing and free education. The Dastkari Schools train women in tailoring, embroidery and other marketable skills, thereby enhancing income generation and supporting their empowerment. 47 1.26 In addition to its own projects, PBM funds a number of NGOs to provide institutional support to people with disabilities, orphans and the destitute. It also provides financial support to orphans and carries out innovative pilot rehabilitation projects. PKR 56.587 million was disbursed to NGOs in 2008/09 for 35 projects benefiting 115,130 people. Finally, PBM also provides free lungars (meals), health facilities, distribution of food and cooking utensils, and is giving similar support to flood affectees. Total spending on assistance to IDPs in 2008/09 was PKR 522 million. 1.27 With respect to provincial implementation, Bait-ul-Mal programs in Punjab have been suspended for the past three years because of political differences over the constitution of District PBM Committees. The tenure of the previous Committees expired in October 2008, and thereafter, new ones have not been notified. The Child Support Programme in Punjab is being implemented from Islamabad. 1.28 The program size of PBM in Khyber Pakhtunkhwa expanded marginally between FY 2003/04 and FY 2006/07, but there was an approximately 71 percent increase in expenditures in FY 2007/08, owing to the assistance program for IDPs. Other PBM initiatives in KP include 25 NCRCL, 3 Sweet Homes (orphanages) and IFA approximately PKR 95 million. PBM’s operations in Sindh have faced cuts in allocations driven by the overall reduction in government allocation to PBM: 38 percent reduction between 2007/08 and 2008/09). This was partly due to the closure of the Food Support Program, and also to the establishment of BISP as the nation’s main safety net program. PBM initiatives in Sindh include the CSP program in two districts (Tharparkar, and Nawab Shah), 36 NCRCL, 3 Sweet Homes (orphanages), and IFA disbursement to the tune of PKR 63 million. V. Social Welfare 1.29 In addition to the two cash transfer programs, Zakat and Bait-ul-Maal, diverse social welfare initiatives are being implemented at federal and provincial level. It should be noted that the description of federal initiatives below predates the 18th Constitutional Amendment; many of these have since/will be devolved to the provinces. 1.30 Social welfare programs at federal level were being implemented by a number of affiliated government bodies: the Ministry of Social Welfare and Special Education, and falling under it, the Directorate General of Special Education (DGSE), National Commission for Child Welfare and Development (NCCWD), National Council for the Rehabilitation of Disabled Persons (NCRDP), National Council of Social Welfare (NCSW), National Trust for the Disabled (NTD), Pakistan Bait-ul-Mal (PBM) and Trust for Voluntary Organizations (TVO). 1.31 The Ministry of Social Welfare and Special Education, which has been disbanded, was implementing a number of programs geared towards developing community organizations to engage in social welfare and development activities; providing medical services to poor patients in federally administered hospitals in a number of major cities and the Northern Areas; giving training to women in market-related skills; and organizing training courses for NGOs and CBOs. Smaller schemes include a pilot school social work project in Islamabad providing counseling and other support to students, a Model Child Welfare Centre in Islamabad 48 providing maternal and child health services, a program to give non-formal education to child domestic laborers, and a pilot ambulance scheme in rural areas of Lahore. 1.32 The Ministry also had a number of programs to strengthen/regulate NGOs: the District-Based NGOs Management System, a National Capacity Building Programme for NGOs Regulatory Authorities, and a project to certify non-profit organizations (NPOs) in collaboration with the Pakistan Centre for Philanthropy. The Ministry’s Planning Implementation and Monitoring Unit carried out 37 development projects in 2008/09. Details of budget allocations and beneficiaries are given in Table A3. 49 Table A3: Summary of Ministry of Social Welfare Programs 2008/09 Budget for FY 2008/09 No. of Beneficiaries Program/Project (PKR million) in 2008/09 Community Development Programme 44.302 32,125 Social Services (Medical) Centres/Projects 68.838 54,211 Social Welfare Training Institute, Islamabad 5.283 276 Women Welfare and Development Centre, 3.942 1,200 Islamabad Pilot School Social Work Centre, Islamabad 1.757 2,962 Model Child Welfare Centre, Hummuk, 2.923 2,686 Islamabad District-Based NGOs Management System 4.000 National Capacity Building Programme for 172 NGOs Regulatory Authorities Planning Monitoring and Implementing Unit 84.553 Integrated Social Development Projects 1.920 4,301 Child Domestic Labour Basic Education 6.550 763 Enabling Programme Non-Profit Organization Programme 74.118 1,952 trainees People’s Rural Health Ambulance Services 31.00 (Pilot Project), Lahore Source: Ministry of Social Welfare Yearbook 2008-09. 1.33 The Directorate General of Special Education runs 128 Special Education Centres around the country, as well as the National Institute of Special Education, National Library and Resource Centre, three service centers for Vocational Rehabilitation and Employment of Disabled Persons, and the Special Education Complex Peshawar. It also manages the National Braille Press and implements various development projects in the area of disability, e.g., promoting inclusive education. 1.34 Key activities of the National Commission for Child Welfare and Development (NCCWD) include the establishment of a Child Protection Management Information System; conduct of national communication campaigns on child abuse; and promotion of child protection legislation and policies; it also runs the National Child Protection Centre in Islamabad. The budget allocation for NCCWD in 50 2008/09 was PKR 7.040 million. Medical Assessment Boards constituted under the National Council for the Rehabilitation of Disabled Persons (NCRDP) assess people with disabilities, and recommend suitable rehabilitative measures. NCRDP also ensures observance of the 2 percent quota for employment of disabled persons in the public and private sectors. The National Trust for the Disabled (NTD) has a complementary mandate to ensure effective planning, implementation and coordination of services for assessment, education, care, job placement and rehabilitation of disabled persons. NT D’s main work is running three model Special Education Complexes at Karachi, Mianwali and Naushero Feroze. 1.35 The National Council of Social Welfare (NCSW) is mandated to foster the growth and development of voluntary social welfare agencies/services in the country. It provides technical and financial assistance to NGOs, and monitors and evaluates their programs. The NCSW budget allocation for 2008/09 was PKR 48.843 million, of which PKR 28.976 million was the development budget, and PKR 19.876 million the non-development. Alongside primary health care, education and safe drinking water, its focus is on poverty alleviation and rehabilitation of the disabled. 1.36 Social welfare has now been fully devolved to the provinces. However, even prior to the 18th Constitutional Amendment, all provinces implemented social welfare programs. 1.37 In Punjab, unlike at federal level, there are separate Departments for Social Welfare and for Special Education. The main initiatives/programs being carried out by the Punjab Social Welfare Department are: awareness-raising about the Convention on the Rights of the Child (CRC), with support from UNICEF; three orphanages; one disabled home in Lahore; Dar-ul-Aman (women’s shelters) in all districts of Punjab except Nankana; three old-age homes in Lahore, Rawalpindi and Multan; and industrial homes (vocational training centers) for women in 34 districts. Further initiatives to help the disabled are carried out by the Department of Special Education. The Social Welfare Department was also heavily involved in provision of relief to flood affectees. 1.38 The KP Department of Social Welfare and Women’s Development carries out a range of programs for beggars, orphans, destitute women and disabled children. It also provides stipends and monthly allowances to senior citizens and unemployed graduates. Activities during 2009/10 included special education for 40 disabled children; shelter and skills provided to 120 orphans; 1,260 poor women given vocational training; skills development for 150 male beggars; rehabilitation aids provided to 1,235 disabled people. During FY 2010/11 the Department planned to set up Social Welfare Complexes in Swabi, Charsadda and Karak Districts; 25 Dastkari (vocational) centers across the province; and a senior citizens home in Peshawar. Senior citizens’ allowance is being given to 1,800 elderly people and monthly stipends to 1,200 postgraduate students. 1.39 The Sindh Department of Social Welfare provides a number of services for the disabled, orphans and women. These include five day-centers for the physically handicapped, and those with multiple handicaps, where they can acquire skills. The Department also runs an orphanage for girls in Karachi and for boys in Hyderabad. There are four Dar-ul-Amans, with another one underway in Benazirabad. Women’s welfare centers provide literacy, Quranic reading and vocational skills; these are located in most talukas. Finally, the Social Welfare Department conducts periodic awareness-raising campaigns on issues like child immunization and hygiene. 51 1.40 The Balochistan Directorate of Social Welfare and Special Education has a number of programs to help ‘deprived people.’ These include provision of shelters for women who are without care or whose cases are being processed in the courts, rehabilitation programs for drug addicts and residential and educational facilities for deaf, blind and other disabled children. However, all these facilities are located in Quetta. VI. Other Safety Net Initiatives 1.41 Use of subsidies has been widespread in Pakistan: for wheat, gas, sugar, fertilizer and so on. While most of these are untargeted subsidies, a number of subsidy initiatives are being implemented specifically geared at the poor. The Government of the Punjab has been particularly active in this regard. Its Sasti Roti (‘cheap bread’) scheme set up tandoors (ovens) to give subsidized rotis to the poor. The Sasti Roti tandoors were initially located in all areas, including affluent areas, leading to ‘across the board’ purchase of subsidized rotis. Later tandoors were confined to the poorest areas; which is likely to have led to an improvement in targeting. Total spending on the scheme over three years was around PKR 12 billion; it was stopped in 2010 and funds diverted to help flood affectees. 1.42 As discussed in chapter 2, health shocks are the most common sort of shock experienced by households in Pakistan and have a significant negative impact on the poor, in particular. To help families cope with health shocks, the Government of the Punjab provides free medicines and disposable items to all patients in District and Tehsil Headquarter Hospitals (DHQs and THQs). It has resulted in a significant increase in the health budget: the annual expenditure on medicines is around PKR 6 billion. The subsidized treatment is targeted in the sense that the poor tend to use government hospitals. However, resources are limited and at the facility level, the final choice of patients to help is made by hospital superintendents/doctors. Dialysis treatment is also free in government facilities in Punjab. The cost of this program is around PKR 500 million per year. 1.43 The second major sector on which the Punjab Government focuses on is education. Spending on education stipends amounted to PKR 1,140 million in 2009/10, and the allocation for 2010/11 was PKR 1,200 million. The stipends go to female secondary school students (in 15 districts of Punjab); and for the payment of fees of poor deserving students, particularly for high school and higher education (colleges, universities). 1.44 The Government of KP is implementing the Bacha Khan Poverty Alleviation Programme (BKPAP), with the help of Sarhad Rural Support Program, in four districts at a cost of PKR 1,500 million. BKPAP has three components: social mobilization, livelihood strengthening and social protection. The focus under the latter is on provision of micro-insurance for poor households to enable smooth household consumption in the event of social/health related shocks, technical vocational skills focused on basic education, and health/hygiene. A community investment fund is also being set up as a social security measure. This is a revolving welfare fund managed by the communities themselves. 1.45 All four provinces also implement other safety net/social protection initiatives such as provision of micro-finance and skills development. The vast majority of these are small-scale and coordination between different programs is limited. 52 VII. Implementation Arrangements 1.46 There is no single agency responsible for social protection in Pakistan. As indicated in the introduction, social protection programs vary greatly in nature and scope—cash transfers, medical assistance, pensions, public works programs, social welfare initiatives targeting specific groups and so on. Implementation of these diverse programs is divided among a plethora of agencies. 1.47 At the federal level, the key entities involved were the Ministry of Social Welfare and Special Education and Ministry of Zakat and Ushr (both now disbanded), and the Ministry of Labour, Manpower and Overseas Pakistanis (to be devolved), Finance Division and Planning Commission. The Ministry of Social Welfare implemented a number of social welfare programs directly, and oversaw the functioning of Pakistan Bait-ul-Mal. The Ministry of Zakat and Ushr was responsible for Zakat programs. The Workers Welfare Fund falls under the Ministry of Labour and Manpower. The Finance Division has an obvious role in resource allocation for the various public sector programs; it also manages the federal civil servants pension scheme and regulates those of autonomous bodies. The Planning Commission sets the overall vision and direction of government policies, including for social protection, and monitors implementation. VIII. Benazir Income Support Programme 1.48 In 2008, as a key element of the National Social Protection Strategy, the federal government launched the Benazir Income Support Programme (BISP). The initial aim of BISP was to help the ‘poorest of the poor’ through unconditional cash transfers. It was initiated against a backdrop of high inflation, particularly food inflation, and the serious negative impact of this on the purchasing power of the poorest sections of society. PKR 34 billion was allocated for BISP in the 2008/09 budget, equivalent to 0.3 percent of GDP. In that year, PKR 14 billion was disbursed to 1.76 million beneficiaries in the form of cash grants of PKR 1,000 per month. Over the next five years, the program aims to reach 15 percent of the total population. 1.49 Initially, due to the absence of poverty data, BISP targeting was carried out through parliamentarians. Applicants completed simple forms and submitted these to their members of parliament (MNAs and MPAs); parliamentarians forwarded the forms to the National Database and Registration Authority (NADRA). NADRA was assigned the task of data entry and verification, and applying the eligibility criteria.17 In April 2009, this beneficiary selection through parliamentarians was stopped and a more rigorous and objective poverty scorecard targeting system was introduced. 1.50 The BISP comprehensive poverty survey was initiated in 2009. It was first conducted in fifteen pilot districts and, following refinements, expanded nationwide. The national rollout process has been completed by the summer of 2011. BISP covers all four provinces, as well as AJK, Gilgit-Baltistan, FATA and the Islamabad Capital Territory (ICT). The survey collects data on some 23 variables, including household members’ characteristics such as gender, age, and schooling attainment, as well as ownership of selected household assets, land and livestock. Based on these criteria, a poverty score is 17 The eligibility criteria under the parliamentarian based targeting system consisted of 6 indicators: monthly income less than PKR 5,000; no family member in government service; possession of no land or less than 3 acres of agricultural land or up to 3 marlas residential property; should not be beneficiary of other support programs; should not have a bank account; and should not possess a passport or an Overseas Pakistani ID Card. 53 calculated for each household; those falling below the cut-off score are deemed eligible for BISP cash grants. Based on their poverty scores, payments to some initial beneficiaries (identified by MNAs/MPAs) were stopped. BISP has provisions for people to appeal if they are deemed ineligible. 1.51 The goal of the poverty survey, in addition to improving BISP targeting, is to generate a national poverty database, which details the spatial distribution of poverty in the country and identifies the most vulnerable population groups. This database is being made available to other programs/organizations, and thus, has utility beyond BISP. Currently the mechanism and frequency of updating the poverty scorecard database is not defined, however. 1.52 Eligible families are paid money orders through Pakistan Post at their doorstep; grants of PKR 2,000 are distributed every two months. Payments are made to the female head of the household, a measure designed to promote women’s empowerment. The use of diverse implementing partners—for data collection, processing, payment and third party monitoring—is to act as a check against misuse of funds. 1.53 BISP is also involved in the design and delivery of a number of poverty exit strategies. For instance, one member of each beneficiary household will be provided vocational/technical training to enable them to generate income for themselves, although this program is not functioning yet. This is to be followed by provision of micro-credit under the ‘Waseela-e-Haq’ scheme to the families selected through a computerized drawing with cash awards of PKR 300,000. This is a one-time payment that has to be used for income generation. As of November 2010, 6,933 people had received grants of PKR 300,000 under the scheme. Thirdly, BISP is developing plans for health and life insurance for the poor. This would likely entail BISP paying for health and life insurance premiums to provide coverage to poor families. 1.54 BISP has also funded emergency relief provision for IDPs in FATA and Swat, and for earthquake affectees in Balochistan. Between December 2008 and August 2009, PKR 34 million was allocated for Balochistan earthquake victims, and almost PKR 1 billion for IDPs. A further PKR 104 million has been provided to bomb blast (terrorism) victims and their families. Table A4: BISP Resource Allocations and Beneficiaries Amount Disbursed No. of Scheme Reporting Period Beneficiaries (PKR million) Cash transfers to poor Oct. 2008 – June 2010 42,400 2,275,206 Waseela-e-Haq Oct 2009 – Nov 2010 0.3 each 6,933 Terrorism (bomb blast) Mar 2009 – to date 104 1,166 victims IDPs Swat/Malakand Apr 2009 – Mar 2010 950 318,126 IDPs Bajaur Dec 2008 – Jun 2009 28 3,965 54 Earthquake affectees Dec 2008 – Aug 2009 34 3,729 Balochistan Source: BISP website. IX. Social Protection Spending 1.55 The strengthened commitment to social protection in Pakistan has translated into increased resource allocations, as seen in Table 2.4. Social security and welfare spending was equivalent to 0.05 percent of GDP in 2006-07; however, it rose from PKR 4,513 million in 2006-07 to PKR 18,942 million in 2007-08, and PKR 51,788 million in 2009-10. Much of the most recent increase in the last two years is due to the new Benazir Income Support Programme (BISP). 1.56 Table A5 also shows significant increases in spending on natural calamities and disasters. This is not surprising given that Pakistan has been hit by multiple large and smaller-scale disasters in recent years, including earthquakes, cyclones and flooding. Since July 2010, Pakistan has seen the worst flooding in its history, so spending on natural calamities can be expected to be vastly higher in the current financial year. Table A5: Spending on Social Protection in Pakistan 2006-07 to 2009-10 2006-07 2007-08 2008-09 2009-10 Category Expenditure Expenditure Expenditure Expenditure (PKR % GDP (PKR % GDP (PKR % GDP (PKR million) million) million) million) Social Security and 4,513 0.05 18,942 0.18 29,129 0.22 51,788 Welfare18 Food Support 3,458 0.04 4,371 0.04 12,420 0.092 0 Program Low cost 299 0.003 597 0.01 583 0.007 858 Housing Natural Calamities 5,008 0.06 7,728 0.07 10,083 0.076 11,007 and Disasters Source: PRSP II Annual Progress Reports FY 2006/07, FY 2007/08, FY 2008/09, and PRSP II Annual Budgetary Expenditures 2009/10. 18 Social Security & Welfare includes Benazir Income Support Programme, Pakistan Bait ul Mal, and EOBI. 55 1.57 Provincial allocations are smaller, but they are expected to increase over time as social protection responsibilities are devolved after the 18th Constitutional Amendment. The Government of Punjab allocated PKR 900 million for the social protection sector in the 2010/11 development budget, but this amount does not include all the initiatives by diverse entities falling under the rubric of social protection. The Government of KP allocated PKR 205.112 million for ‘Social Welfare and Women’s Development’ in the 2010/11 Annual Development Programme, but again this figure does not include all social protection initiatives. Balochistan allocated PKR 0.782 billion for social protection in 2010/11, out of a total revenue expenditure of PKR 83.444 billion. X. Recent Developments Emergency Response 1.58 Pakistan has been hit by a number of major natural and other disasters in the past few years. Military operations against armed groups in the north of the country (notably FATA, parts of KP) led to civilian casualties, thousands of people being displaced from their homes, and destruction of infrastructure and property. This was followed by the 2010 floods disaster which affected 20 million people across Pakistan, stretching from FATA and KP in the north, right down to Sindh. It caused widespread destruction of houses, and damage to education and health facilities, power, transport and communication infrastructure, crops and livestock. In August 2011 parts of Sindh were again inundated – in many cases the same areas as were flooded in 2010. While a smaller-scale and more static event (the 2010 floods situation kept changing as the floodwater moved downstream), this has nonetheless caused huge loss of property, livestock and infrastructure. 1.59 In the wake of all these disasters two critical challenges arise: how to provide immediate relief to affected families, and how to help them rebuild? These questions are particularly pertinent because, generally, it was the poorest households who were worst affected. In the 2010 pre-flood scenario, the highest population concentration in all affected areas was around the poverty line. This implies that the floods pushed those already poor deeper into poverty, and pushed those just above the poverty line (the vulnerable) into poverty. 1.60 The implications of this for safety nets are two-fold: a) the impact of anti-poverty measures on the already poor will be reduced, and the number of chronically poor (permanently dependent on transfers) could expand; b) as the vulnerable fall into poverty, the total number of people in need of assistance will increase (albeit often temporarily). If this increased demand for safety nets is not met, households will be forced to adopt coping strategies such as reducing consumption, selling assets, pulling children out of education, putting them to work and so on—all factors that inhibit their long-term chances of exiting poverty. 1.61 In response to the 2010 floods Zakat, Pakistan Bait-ul-Mal and other safety net providers distributed relief goods to IDPs, but did not expand their regular programs. Instead, the government launched the Watan Card scheme. Flood affectees had to register for this card, which entitled them to a one-off cash grant of PKR 20,000 to help their rehabilitation, followed by a second installment of PKR 20,000 to a targeted group within the population of first tranche beneficiaries. Targeting was based on affected communities in Punjab, Sindh and Balochistan (at least 40 or 50 percent of a community had to 56 be calamity-affected for the community to be eligible); whereas in KP targeting was based on flood affected households (based on the findings of a Rapid Household Survey). The former was quicker, but excluded flood-affected households in communities not designated as ‘calamity affected’; these were reached in the KP targeting system. 1.62 Registration was carried out with support from NADRA, and was dependent on possession of Computerized National Identity Cards (CNIC). Payment was through bank ATM machines. An evaluation of the scheme reported that NADRA made an ‘extraordinary effort’ to issue CNICs to those without them – meaning very few people were excluded for not possessing a CNIC. It also stated that people had found the Watan card easy to use and generally did not face problems accessing payment. Moreover, many were considering turning their accounts into proper bank accounts (OPM 2011). 1.63 In response to the 2011 floods a similar relief mechanism is being used to that deployed after the 2010 floods. Affected households in Sindh are issued with a Pakistan Card – as of early October 2011, 110,000 cards have been issued – which entitles holders to a cash grant of PKR 20,000, paid in two equal installments. Validation of beneficiaries is carried out through the Provincial Disaster Management Authority and NADRA, which also ensures transparency (prevention of identity fraud) in cash disbursement. 1.64 Watan Card grants and relief supplies can bring some immediate relief to affected families, but many have lost their sources of livelihood and assets and need long-term support. Recovery and reconstruction strategies focusing on restoring services and infrastructure will lead to employment opportunities for local communities. However, all these initiatives further highlight the need for an effective safety net mechanism with the capacity to immediately respond to the needs of the affected population by instantly producing lists of affectees and mobilizing cash transfers for income support in the short-term, and in the medium- to long-term, coordinating livelihood support programs to facilitate the affected population’s efforts to get back on its feet. 1.65 The BISP poverty scorecard database could potentially be used to identify those in obvious need of help in the wake of a disaster such as the 2010 and 2011 floods, i.e. the poor and vulnerable. But additional mechanisms will be needed to rapidly identify those in need of support among the non-poor, e.g. those who have lost their homes and livestock. The National Disaster Management Authority is the agency with the primary mandate for this work. A further area that has to be addressed is payment mechanisms. The Watan card system proved effective in flood-affected areas in 2010, but there are some parts of the country (e.g. Kohistan) where it cannot be used. These are often the most disadvantaged areas, so finding suitable payment mechanisms to reach these people has to be a priority. Overall, institutional mechanisms for social protection within the NDMA need to be built up, and coordination needs to be strengthened between the NDMA, BISP and other safety net programs so as to ensure an effective post-disaster response. 57 APPENDIX 2 SOCIAL SAFETY NET EXPENDITURES AND WHEAT SUBSIDY PROGRAM I. GDP Growth Projections (Appendix Table A5 provides Social Protection spending figures) Table A6: IMF Projections on GDP Growth and Consumer Prices Actual Projections 2008 2009 2010 2011 2012 2013 2017 Real GDP % change 3.7 1.7 3.8 2.4 3.4 3.5 3.5 Consumer Prices 10.8 17.6 10.1 13.7 12 12.5 14 Balance on current account -8.5 -5.7 -2.2 0.2 -1.9 -2.1 -3.7 Source: World Economic Outlook, International Monetary Fund Part 2. Wheat Subsidy Program 2.1 Wheat is one of the most important food commodities in Pakistan, accounting for more than 55 percent of total caloric consumption on average and much higher for the poorest population. This has been a strong justification for the government to intervene in the wheat market by providing wheat subsidies, with the main objectives of a) supporting farmer incomes (to incentivize increase in productivity and output) by setting the procurement price before the sowing season, and b) improving household food security through provision of consistent supply of affordable wheat flour (to stabilize domestic market price) through mainly provision of procured wheat at subsidized prices to flour mills, and at a smaller scale to Utility Stores Corporation (USC). Figures A1 and A2 provide a visual description of each of these channels. 2.2 The procurement (support) price is determined by the federal government based on the cost of production studies undertaken by Agricultural Prices Commission (APCOM), and is announced before the sowing season (September) to enable the farmers to respond to the price incentives. The federal government in consultation with the Provincial Food Departments (PFDs) sets overall procurement targets, which have been fairly constant over the years amounting to around 50-55% of the marketable surplus (20% of total production) in the country19. The PFDs20 and Pakistan Agricultural Services and Storage Corporation (PASSCO) (on behalf of federal government) procure the wheat during harvest season (April-May). The PFDs with the assistance of executive authorities at the district level, set up procurement centers, the locations of which are advertised21. The procurement takes place in a mutual manner, in which the farmers bring their wheat to procurement centers and get reimbursed for the 19 In general, 60% of the wheat produced stays in the farms for own consumption. Around 40% of the harvest results in marketable surplus (which amounts to 8-10 mmt) of which the government procures about 50-55% (around 5 mmt). Punjab is the only net-surplus province while KP and Balochistan are deficit provinces. 20 The Provincial Food Departments borrow from commercial banks under the government guarantee to undertake the procurement during harvest season. 21 PASSCO also has its own procurement centers. 58 transportation costs, or else the district food controller and his/her officers visit the farmers and procure the wheat at the farm. Since this procurement process has been implemented for many years, the procurement officers have extensive knowledge of the farms and their production. Inter-provincial (even inter-district) restrictions to wheat movement may be instituted to achieve procurement targets. The level of procurement price in relation to market price (which is determined by the size of the harvest) determines whether the target will be reached. The procurement process continues until a deadline set by the provincial government. In years where the target is not reached, the government (i.e. Trading Corporation of Pakistan- TCP) imports wheat. 2.3 The PFDs are responsible for transporting the wheat from the procurement centers and for storing and issuing the wheat. PASSCO also stores the wheat it procures for the federally administered areas. During the immediate post-harvest period of June-September, the private sector supplies the wheat market entirely. The procured wheat is not released until September when supplies in the market get low (stress period). The imported wheat is also routed through the PFDs for release. Figure A1 Release of Wheat to Flour Mills Figure A2 Release of Wheat to Utility Stores 59 2.4 Similar to the procurement price, the issue/release price is determined again by the federal government in consultation with the provincial governments. The issue price does not cover all the costs of procurement, storage, release (i.e. incidentals). In fact, in the past there have been instances when the issue price is lower than the procurement price. All the procurement costs, storage/transportation costs and issuing costs as well the discrepancy between the procurement price and the issue price (if any) are borne by the provincial governments in provinces and the federal government in the federally administered areas. In the case of imported wheat, the direct wheat subsidy (the difference between the import price and the lower issue price) is initially borne by the federal government but recovered later from the provinces based on the wheat issued to them. 2.5 The federal wheat subsidy has been on a declining path since 2010. During the period 2008-09 and 2009-10, it increased by 31 percent in nominal terms, owing to the food crisis, however; it remained a constant 0.2 percent of the GDP. In the following years, the government reduced it significantly, represented by a total decrease of 95 percent from 2008-12, assuming only 0.005 percent of the GDP by 2011-12. This was the result of abolishing by 2011-12, the subsidy to USC for sale of flour; payment of subsidy to PASSCO on cost differential for sale of wheat; and reimbursement of cost differential of imported wheat to TCP, and only contributing towards the sale of wheat in Gilgit Agency and FATA, as shown in Table A3. 2.6 When it comes to provincial wheat subsidies, Punjab has been allocating the most for wheat subsidies, amounting to Rs. 30,000 million in 2011-12, as compared to Rs. 2,500 million in KP, Rs. 1,500 million in Sindh, and none in Balochistan. In terms of percentage of current revenue expenditure allocated towards subsidies, Punjab has exhibited a rising trend touching the highest 6.9 percent in 2011- 12; however, both Sindh and KP have depicted an overall declining trend, and Balochistan has gradually reduced subsidies to zero (Figure A3). Figure A3: Percentage of Current Revenue Expenditure towards Provincial Wheat Subsidies 8.0% 7.0% 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% Sindh Punjab KP Balochistan 2008-09 2009-10 2010-11 2011-12 2.7 The government operates fiscally expensive and inefficient wheat procurement and distribution scheme, by providing an untargeted subsidy to the entire population, whereby most of the benefits are reaped by the wheat flour millers and traders. This has also resulted in a significant excess capacity of 60 around 300 percent in the wheat milling industry22, while crowding out private sector participation in wheat marketing. The rapid increase in world prices accompanied with sluggish rise of domestic prices lead to a substantial subsidy on imports. In 2007-08, government spent Rs.55 billion (USD 750 million) on wheat imports and distribution of subsidized flour through USC without explicit targeting and limited geographic coverage. In addition, fertilizer subsidies in this regard accounted for a large fiscal burden due to the increased world market prices. Domestic prices are almost 60 percent higher as compared to pre- crisis levels, which substantially impacted the poor and the vulnerable. The government’s limited agricultural policymaking capacity has led to regulated prices which have at times given wrong signals in the market. This is reflected in the2008-09 period when the government delayed the announcement of raising the procurement price to $240/Mt in 2008 from $163 in 2007, thus leaving the wheat harvest unaffected. Despite raising it further to $300/Mt in 2009, the government could not achieve the objective of encouraging farmers to grow more and sell domestically, as by that time international wheat prices had declined to much lower levels.23 2.8 Further, government’s interventions in the procurement, handling, marketing and storage of wheat through provincial food departments and PASSCO, widened the gap between the import and export parity prices, particularly during September 2007and March 2008, in which it released 4.3 million Mt of wheat into the domestic market at subsidized rates. In July 2008, the domestic wheat price was 21 percent and 36 percent lower than the export and import parity prices respectively, which provided a strong incentive for smuggling to the neighboring countries, where higher wheat prices prevailed, such as Afghanistan and Iran – approximately 1.5 million Mt of wheat was smuggled.24 Later on, a ban on exports was imposed, which kept the domestic wheat prices below the export parity levels, thus encouraging hoarding, due to the expectation of a rise in future prices. 22 http://www.pakissan.com/english/issues/wheat.policy.problems.and.issues.shtml 23 Food Prices Increases in South Asia- National Responses and Regional Dimensions, World Bank 24 Ibid 61 APPENDIX 3 SUPPLEMENTARY TABLES FOR CHAPTER 2 Table A1: Transitions, 2008–2010 Decile in 2010 Decile in 2008 1 2 3 4 5 6 7 8 9 10 1 38.7 18.6 13.3 10.4 7.5 5.1 3.3 1.7 1.3 0.1 2 22.7 20.4 15.5 13.9 11.4 6.2 5.6 2.5 1.2 0.7 3 15.2 17.6 15.0 14.4 12.6 10.1 8.2 3.9 2.0 1.0 4 9.5 14.6 17.6 14.0 12.7 11.1 9.5 5.8 4.0 1.2 5 6.2 13.0 12.6 14.6 14.3 12.4 11.0 8.4 5.6 1.9 6 4.2 7.5 11.5 11.8 12.1 16.5 15.3 11.0 7.1 3.0 7 1.7 5.6 8.5 10.1 14.0 14.1 13.3 15.9 11.0 5.8 8 1.3 1.4 4.8 7.8 8.1 14.6 15.7 20.9 17.0 8.4 9 0.1 1.2 1.2 2.3 6.6 7.4 13.0 21.1 28.0 19.2 10 0.1 0.4 0.1 0.7 0.9 2.8 4.9 9.0 22.8 58.2 Note: Each cell stands for the percentage of the population in the specific decile in 2008 (row), falling into the certain decile in 2010 (column). 62 Table A2: Movement in per Adult Equivalent Expenditure, 2008–2010 2010 2008 Above 20 percentile line At or below the 20 percentile line Above 20 percentile line 70.0 % 10.0% At or below 20 percentile line 9.5 % 10.5% Note: Households with per adult equivalent household expenditure below the 20th percentile in Pakistan were defined as being in the bottom 20 percent. Table A3. Proportional Price Change (%), 2005/06 to June 2008 Commodity Price Change (%) Standard deviation Dairy products 42.01 (13.33) Meat 24.52 (21.95) Fruits 42.11 (53.49) Vegetables 9.38 (19.03) Spices 13.10 (13.89) Wheat 79.74 (26.20) Rice 173.19 (64.72) Cereals 5.99 (16.98) Pulses 43.24 (33.29) Oil 86.88 (27.46) Other foods 12.14 (17.32) Nonfood 16.55 (8.45) 63 Table A4: Determinants of Health Shocks Factor Log PCE -0.066 Asset index quintile-2 -0.085 Asset index quintile-3 -0.118 Asset index quintile-4 0.025 Asset index quintile-5 -0.213 #HH mem(female 0~5 yrs) -0.069 #HH mem(male 0~5 yrs) 0.018 #HH mem(female 5~15 yrs) -0.011 #HH mem(male 5~15 yrs) -0.018 #HH mem(female 15~55 yrs) 0.059 #HH mem(male 15~55 yrs) 0.043 # HH mem(female 55+ yrs) 0.195** # HH mem(male 55+ yrs) 0.240** HH head: Educ Cl.1-5 -0.033 HH head: Educ Cl.6-8 -0.147 HH head: Educ Cl.9-10 -0.024 HH head: Educ Cl.11 & beyond -0.1 HH head: female -0.207 HH head: Having worked at least one hour past month -0.208* HH head: Age -0.002 Sindh -0.309* KPK 0.138 Balochistan -0.126 Rural -0.047 Constant -0.703 Note: Standard error is clustered at the district level. Baseline household characteristics (2008) are used to identify the determinants of experiencing shocks. 64 Table A5: Effect of the Food Price Shock on Caloric Availability Log (Caloric Log (Caloric Pr (Caloric Availability Log (Total Caloric Availability from Availability from Self- from Self- Availability) Purchased Food) Production) Production>0), Probit Year 2007, 2008 0.042 ** (0.008) 0.034 *** (0.011) 0.225 *** (0.051) -0.037 *** (0.012) Quarter 2 (Oct–Dec) 0.030 ** (0.008) 0.030 *** (0.011) 0.025 (0.050) -0.037 *** (0.012) Quarter 3 (Jan–Mar) 0.030 ** (0.008) 0.057 *** (0.011) 0.059 (0.051) -0.037 *** (0.012) Quarter 4 (Apr–Jun) 0.027 ** (0.008) 0.062 *** (0.011) 0.082 (0.053) -0.038 *** (0.012) Year 07,08 X Quarter 2 -0.044 *** (0.011) -0.020 (0.016) -0.141 * (0.073) -0.027 (0.017) Year 07,08 X Quarter 3 -0.075 *** (0.011) -0.034 ** (0.016) -0.308 *** (0.074) -0.014 (0.017) Year 07,08 X Quarter 4 -0.078 *** (0.011) -0.018 (0.016) -0.307 *** (0.077) -0.022 (0.017) Sample Size 30,823 30,803 12,779 30,827 65 Table A6: Perceived Food Price Shock and Schooling Expenditure, by Gender Boys Girls I II III IV Perceived Food Price Shock -447.477 105.835 -795.734*** -249.476 [287.738] [503.343] [245.887] [425.651] PCE 2nd tercile in 2008 x Food Shock -68.364 6.688 [700.929] [604.791] PCE 3rd tercile in 2008 x Food Shock -1,561.758** -1,574.796*** [700.162] [594.272] PCE 2nd tercile in 2008 150.892 211.731 87.983 84.813 [289.934] [629.179] [245.275] [543.170] PCE 3rd tercile in 2008 1,724.205*** 2,898.675*** 1,410.247*** 2,595.600*** [322.354] [626.567] [274.203] [531.572] Observations 3,005 3,005 2,834 2,834 Note: Household characteristics and regional dummies are controlled for, but for brevity are not shown. 66 BIBLIOGRAPHY Alderman, H. J. Hoddinott, et al. 2006. "Long Term Consequences of Early Childhood Malnutrition.� Oxford Economic Papers 58 (3): 450-474. Arif, G. M. 2006. ‘Poverty Targeting in Pakistan: The case of Zakat and the Lady Health Worker Program.� Poverty Strategies in Asia. ADB Institute. Cited in: Khyber Pakhtunkhwa Social Protection & Socio-Economic Indicators, Govt. of KP/UNICEF, 2010. Ashraf, T. 1996. "Maternal Mortality: A Four Year Review." Journal of the College of Physicians and Surgeons of Pakistan 6 (3): 159-161. Carter, M. R., P. D. Little, et al. 2007. "Poverty Traps and Natural Disasters in Honduras and Ethiopia." World Development. Chaudhury, N. and D. Parajuli. 2006. “Conditional Cash Transfer and Female Schooling: Impact of the Female School Stipend Program on Public School Enrolments in Punjab, Pakistan.� World Bank Policy Research Working Paper (WPS4102). Fauveau, V., D. R. Sherratt, et al. 2008. "Human Resources for Maternal Health: Multi-purpose or Specialists?" Human Resources for Health 6: 21. Ferreira, F. H. G. and N. Schady. 2008. "Aggregate Economic Shocks, Child Schooling and Child Health." World Bank Policy Research Working Paper (WPS4701). Friedman, J. and J. Levinsohn. 2002. "The Distributional Impacts of Indonesia's Financial Crisis on Household Welfare: A “Rapid Response� Methodology." The World Bank Economic Review 16 (3): 397-423. Friedman, J. and J. Sturdy. 2011. "The Influence of Economic Crisis on Early Childhood Development: A Review of Pathways and Measured Impact." Human Development Perspectives, World Bank. Food and Agriculture Organization of the United Nations (FAO). 2009. "The State of Food Insecurity in the World, Economic crises-impacts and lessons learned." Rome: FAO, UN. Gitter, S. R., and B. L. Barham. 2008. "Women's Power, Conditional Cash Transfers, and Schooling in Nicaragua." World Bank Economic Review 22(2): 271-290. GHK. 2010. Process Evaluation of Benazir Income Support Program Scorecard Based Poverty Targeting under Test Phase. Grosh, M., C. del Ninno, E. Tesliuc and A. Ouerghi. 2008. “For Protection and Promotion.� The Design and Implementation of Effective Safety Nets. Washington, D.C.: World Bank. Government of Pakistan. 2010. “Pakistan Millennium Development Goals Report 2010: Development Amidst Crisis.� Center for Poverty Reduction and Social Policy Development. Planning Commission, GoP. 67 Government of Pakistan. 2009. Poverty Reduction Strategy Paper II. Finance Division, GoP: 7. Hanlon, J., A. Barrientos and D. Hulme. 2010. Just Give Money to the Poor: The Development Revolution from the Global South. Sterling, VA: Kumarian Press. Hoope-Bender, P. T., J. Liljestrand, et al. 2006. "Human Resources and Access to Maternal Health Care." International Journal of Gynecology and Obstetrics 94: 226-233. Hou X. 2011 “Women’s Decision-Making Power and Human Development in Islamic Culture- Evidence from Pakistan.� World Bank Policy Research Working Paper Series (WPS 5830). Hou, X. and N. Ma. 2011. “Empowering Women: The Effect of Women’s Decision-Making Power on Reproductive Health Services Uptake—Evidence from Pakistan.� World Bank Policy Research Working Paper Series (WPS5543). Khan, Mansoor Hasan. 2009. “Functional Limitation and Participation Barriers in Earthquake Affected Areas of AJK and NWFP.� Pakistan Poverty Alleviation Fund. Rouche, Shaun. 2010. “Wheat Price Jump: A Supply Shock that Should Unwind.� IMF Survey Magazine. Available: http://www.imf.org/external/pubs/ft/survey/so/2010/new090110a.htm. Innovative Development Strategies. 2010. The BISP Spot Check Phase 1 Report. Ivanic, M. and W. Martin. 2008. "Implications of Higher Global Food Prices for Poverty in Low-income Countries." Agricultural Economics 39: 405-416. Lokshin, M. and M. Ravallion. 2000. "Welfare Impacts of Russia's 1998 Financial Crisis and the Response of the Public Safety Net." Economics of Transition 8 (2): 269-295. Mete, C. Cynthia L. and Naveeda S. 2011. “Learning Versus Working: Factors Affecting Adolescent Time Allocation in Pakistan.� World Bank Manuscript. Nikitin, Denis. 2011. “Trends in Labor Force Participation, Employment and Earnings in Pakistan during 2001 to 2007.� World Bank Background Paper. Oxford Policy Management (OPM). April 2011. Evaluating Implementation of Pakistan’s Citizens’ Damage Compensation Programme (Phase I) Final Report. Oxford Policy Management (OPM). December 2010. Benazir Income Support Programme Evaluation Qualitative Baseline Report. Pop, L. 2010. “Covariate Shocks and Vulnerabilities in KP and FATA.� World Bank Background Paper. Rana, R. 1992. "Induced Abortion and its Complications: A Common Problem in Pakistan." Pakistan Journal of Obstetrics and Gynecology 5 (1): 53-59. Rizvi, N., and S. Nishtar. 2008. "Pakistan's Health Policy: Appropriateness and Relevance to Women's Health Needs." Health Policy 88(2-3): 269-281. 68 Schultz, T. 1990. "Testing the Neoclassical of Family Labor and Fertility." Journal of Human Resources 25: 599-634. Thomas, D. 1990. "Intra-household Allocation: An Inferential Approach." Journal of Human Resources 25: 635-664. Vishwanath, T. Hou X. and Yoshida N. 2009. “Poverty Scorecard for Pakistan: A Recommended Approach for Targeting the Poor.� World Bank Manuscript. World Bank. 2004. “Program Document for Education Sector Adjustment Credit for the Government of Punjab.� IDA Program Document (27528-PK). ---. 2007. "Social Protection in Pakistan: Managing Household Risks and Vulnerability." World Bank Report (35472-PK). ---. (2010 b). "Food Price Increases in South Asia, National Responses and Regional Dimensions." World Bank Report (56086). ---. (2010 a). “Food Price Watch.� PREM World Bank Publication Series (53160). World Food Program. 2010. Food Security and Market Assessment in Crisis Affected Areas of NWFP and FATA. 69 IBRD 33460R PA K I S TA N SELECTED CITIES AND TOWNS MAIN ROADS PROVINCE CAPITALS RAILROADS NATIONAL CAPITAL PROVINCE BOUNDARIES RIVERS INTERNATIONAL BOUNDARIES 65E 70E TAJIKISTAN 75E CHI CH I N A Ka h ra K2 K us NORTHERN k o (Mt. Godwin- r a (8,611 m) Austen) d u ARE AS AREAS m PAKISTAN i n Chitral I nd s u Ra H ng KHYBER e 35N Approximate Line of Control PAKHTUNKHWA Saidu To Muzaffarabad Kargil Kabul Srinagar Peshawar Jammu ISLAMABAD and Kashmir Rawalpindi Kahat FED. CAPITAL AFGH AF ANI G H AN I S TAN Bannu TERRITORY ISLAMABAD lum Jhe Gujrat Gujranwala To Mandi ab D.I. Khan n To Che Kandahar Zhob Lahore Faisalabad vi Ra Chaman Zhob PUNJAB To To Bhatinda Ludhiana Sahlwal 30N Quetta D.G. Khan Multan tlej 30N Su us Ind To Kerman Bahawalpur Nok Kundi B A L O C H I S TA N Surab rt se ISLAMIC D e REPUBLIC INDIA r OF To Ranipur a h ng e Panjgur IRAN Khash T a nR i s hka kra Ma Moro Ma l To Turbat C e ntr a Bela SINDH Kerman To Jodhpur Gwadar Pasni Hyderabad 25N 25N Karachi Thatta Badin us Ind This map was produced by Rann of Kutch the Map Design Unit of The World Bank. The boundaries, Arabian Sea colors, denominations and 0 50 100 150 200 Kilometers any other information shown on this map do not imply, on the part of The World Bank Group, any judgment on the 0 50 100 150 Miles legal status of any territory, or any endorsement or acceptance of such boundaries. 65E 70E 75E NOVEMBER 2010