Report No. 32244-PAK Pakistan Country Gender Assessment Bridging the Gender Gap: Opportunities and Challenges October 2005 Environment & Social Development Sector Unit South Asia Region Document of the World Bank ACKNOWLEDGEMENTS This report was prepared by a team led by Tara Vishwanath, with a core team comprised of Ghazala Mansuri, NisthaSinha, andJennifer Solotaroff. For specific chapters, the report draws heavily on contributionsfrom Jishnu Das (Chapter 3), Monica Das Gupta (Chapter 4), InaamHaq(Chapter 4), HananJacoby (Chapters 2 and 5), andNobuoYoshida (Chapter 3). Acknowledgements are also due to Qazi Azmat Isa for facilitating country counterpart dialogue, as well as coordinatingworkshops andcontributionsfrom experts inthe field. The report benefited from data from the Pakistan Rural Household Survey (2004). The Pakistan Institute of Development Economics (PIDE), Islamabad collectedthe data, collaboratingwith World Bank'sDevelopment EconomicsResearchGroup (DECRG). The team would like to thank Mussarrat Bashir Youssuf and Madiha Afzal for field work, translation andtranscribingofthe qualitativedata. The team acknowledges background papers from Zia Awan, Simi Kamal, RukshandaNaz and Justice (Retd.) MajidaRazviwhich were discussed in a workshop heldinDecember2004 in Islamabad; thanks are also due to the participantsat the workshop for their commentsandinsights. The team thanks the peer reviewers, Karen Mason, ElizabethKing, ChristopherUdry, and Zeba Sathar for insightful comments; and all other participants at the Bankwide review meeting for their suggestions.We also thank all participantsin an e-discussionof an earlier draft. The team is grateful to Richard Ackerman, Jeffrey Racki, Junaid Ahmad, and John Wall for their continuingsupport. Arrangements for missions, workshops and facilitation of contracts were made by Samina Riaz and Mohammad Shafiq. Excellent editorial support was provided by Hedy Sladovich. The team thanks Thelma Rutledge for a superb formatting job. Finally, the team gratefully acknowledges financial support from Consultant TrustFundand BNPPTrust Fund. TABLEOF CONTENTS EXECUTIVE SUMMARY ................................................................................................................. j 1. IMPROVING GENDEREQUALITY INPAKISTAN.................................................................... 1 I. GenderInequalityinPakistan:AnOverview ................................................................................. 4 I1. I11. AddressingKey Gender Gaps......................................................................................................... Gender Assessment Methodology .................................................................................................. 10 15 2. FAMILY LAW AND CUSTOMINPAKISTAN............................................................................. 17 I. Inheritance....................................................................................................................................... 19 I1. Marriage.......................................................................................................................................... I11. Discussionof Resultsand Implicationsfor Policy Interventions................................................... 24 36 3. IMPROVING GIRLS' SCHOOLING............................................................................................... 41 I. Service Delivery inEducation: The Institutional Setup ................................................................. -42 I11. Distance and School Enrollment.,.................................................................................................. I1. Dimensions ofthe Gender Gap inEducation.................................................................................. 44 -48 IV. Will BuildingSchools Increase Enrollment?.................................................................................. 55 V. RaisingLevels of Girls' Schooling: Some Policy Recommendations............................................ 59 4. IMPROVINGHEALTH OUTCOMESFORWOMEN AND CHILDREN.................................. 62 I. Gender-RelatedConstraintstoAccessingHealthCare:Mobility.Decision.Making. andAccess to Information........................................................................................................................... 64 I11. Delivery ofGovernment Health Services inRuralPakistan: The Institutional Setup .................... I1. Women's Health: The Dimensionsofthe Problem ........................................................................ 66 70 V. Policy Recommendations ............................................................................................................... IV. Overcoming Women's Constraints to Accessing Health Services ................................................. 75 82 5. WOMEN AND WORK INPAKISTAN............................................................................................ 86 I1. Constraints on Women's Labor ForceParticipation....................................................................... I. DimensionsofWomen'sParticipationintheLaborForce............................................................. 88 91 I11. The Relationship between Work and Autonomy............................................................................. IV. PromotingWomen's Involvement inthe Labor Force:Recommendations for Public Policy ........98 103 BIBLIOGRAPHY.,........................................................................................................................................ 107 ANNEX TO CHAPTER2............................................................................................................................. 113 ANNEX TO CHAPTER3............................................................................................................................ -118 ANNEX TO CHAPTER5............................................................................................................................. 135 ANNEX TO CHAPTER4............................................................................................................................. 126 LISTOFTABLES. FIGURES. AND BOXES TABLES Table 1.1: Early Age Mortality Rates ......................................................................................................... 6 Table 1.2: Malnutritionand Anemia ........................................................................................................... 7 Table 1.3: Practice of Purdah among Rural Women: PercentPracticing ................................................. 12 Table 2.1: Rural Women's Knowledge and Perceptionsof Their Inheritance Entitlementsby Region .-23 Table 2.2: Extent o f Wife's Control over Dowry ..................................................................................... 32 Table 2.3: Extent of Wife's Control over Bari.......................................................................................... Table 2.4: Size of Marital Transfers ......................................................................................................... 32 33 Table 2.5: Extentto Which Wife's Opinion is Taken into Account inFamily Decisions ........................ Table 3.1: Net School Enrollment Rate, PIHS 2001-02 ........................................................................... 35 44 Table 3.2: Differences inPopulation Size between Villages in Punjab that Receiveda Public School, 1980-2000.......................................................................................................................... 48 Table 3.3 : Access to Schools inRural Pakistan........................................................................................ 49 Table 3-4: Percent of Rural Women Who Support Female Education, by Region................................... 51 Table 3.5: 53 Table 3.6: Women's Perceptions of Safety............................................................................................... Percentageof Primary Reason for Not Educating Girls, by Region........................................ 55 56 Table 4.1: Life Expectancy and Infant Mortality: Comparisons across Regions and over Time ............-62 Table 3.7: Availability of Educated Females............................................................................................ Table 4.2: Under-fiveMortality and Life Expectancy: Comparisons acrossRegions and by Sex...........63 Table 4.4: Percentageof Rural Communities with Primary HealthFacilities within FiveKilometers ....65 Table 4.3: Percentageof Rural Women ReportingRestrictedAccess to Health Facilities ...................... 71 Table 4.5: Rural Communities with a Lady Health Worker, 2001-02 ...................................................... 73 Table 4.6: Lady Health Worker Program: Levels of Responsibilities ...................................................... 74 Table 4.7: Planned Allocation of Lady Health Workers, 2004-05............................................................ 74 Table 4.8: Impact.of Lady Health Workers and Proximity to HealthFacility on Rural 77 Table 5.1: Distribution of Workers inIndustryby Gender....................................................................... Health Center Use ............................................................................................................. 89 Table 5.2a: Rural Labor Force ParticipationRates. 1991-2001 ................................................................. . . 90 90 Table 5.3 : Daily Wage Rates for Agricultural Wage Work inCash Crops, 1991-200 1 .......................... Table 5.2b: Participation in Paid Agricultural Work among Those Working by Province ........................ 91 Table 5.4: Effect of Safety on Participation inPaid Work........................................................................ 92 Table 5.5: Women's Perceptions of the Tradeoff between Status (Practicing Purdah) and Mobility by Percent Interviewed per Region........................................................................................ 93 FIGURES Figure 1.1: RegionalComparisons............................................................................................................ 2 Figure 1.2: Gender Ratio inEnrollments and MillenniumDevelopment Goals Target ........................... 3 Figure 1.3: Sex Ratio Imbalance in Pakistan: Too Few Girls.................................................................. 5 Figure 1.4: Primary School Gross Enrollment Rate. 1991-2001 .............................................................. 8 Figure 1.5: Female Labor Force Participation Ratesby Age and Region................................................. 9 Figure 1.6: Labor Force Participation Rates by Province......................................................................... 9 Figure 1.7: Framework o f Analysis ........................................................................................................ 11 Figure 2.1: Status of Paternal Inheritance among Potential Female Heirs.............................................. 21 Figure 2.2: Age at Betrothal and Nikah .................................................................................................. 25 Figure 2.3: Relation between Spouses .................................................................................................... 26 Figure 2.4: Types of Marriage ................................................................................................................ 27 Figure 3.1: Net Enrollment by Province, 2001-02 .................................................................................. 44 Figure 3.2: Age Pattern o f Enrollment................................................................................................... -45 Figure 3.3: Figure 3.4a: Growth Incidence Curve of Primary School Enrollment Rates, 1990-91and 2001-02 ........45 Enrollment and Household Socioeconomic Status............................................................... 46 Figure 3.4b: Changes inPrimary School Enrollment Rates, 1990-91and 2001-02 ................................. 46 Figure 3.4~: Gender Gap o f Growth Rate of Primary Net Enrollment Rate (NER) ................................. 47 Figure 3.5: Enrollment and School Proximity ........................................................................................ 50 Figure 3.6: Annual Expenditureon Travel to School by Age and Distance to School........................... 51 Figure 3.7: Percentage of Children Who Have Ever Attended School by Settlement............................ 54 Figure4.1: Total Fertility Rate, 1950-2005 ............................................................................................ 63 Figure4.2: Women with More Sons Are More Likely to Use Contraceptives ...................................... -66 Figure 4.3: Gender Difference inProbability of Consulting a Doctor inCase of Illness ....................... 66 Figure4.4: RuralHousehold's Annual Average Medical Expenditure by Age and Sex........................ 67 Figure 4.5a: Percentage of Children Fully Immunizedby Gender, 1990-91 and 2001-02 ....................... 67 Figure 4.5b: Percentage of Children FullyImmunizedby Gender and Type of Figure 4.52: Percentageof Children Fully Immunizedby Province, 2001-02 ......................................... Immunization, 2001-02.................................................................................................. 68 68 Figure 4.6: 69 Percentageof BirthsAssisted by Type of Attendant, 1998-2001......................................... Percentageof Women ReceivingMaternal Health Services................................................ Figure 4.7: 70 Figure 4.8: Percentageof Rural Communities with any Government Primary Health Center or Rural Health Center within Five Kilometers................................................................. -72 Figure 4.9: Utilization of Maternal Health Services: Impact of Proximity to Public Primary Health 75 Figure4.11: Impact of Lady Health Worker Presence.............................................................................. Figure4.10: Presenceof Lady HealthWorker IncreasesChild's Chances of being Immunized.............76 Facility ........................................................................................................................... 76 Figure 4.12: Use of Prenatal Services and Antenatal Tetanus Immunization Status ................................ 78 79 Figure 4.13b: Female Education and Use of Maternal Health Services ..................................................... Figure4.13a: Female Education and ChildhoodImmunization................................................................. 79 Figure 4.14a: Impact of MediaExposure and Female Educationon Use o f Maternal Health Services .....81 Figure 4.14b: Impact of Lady Health Worker Presenceand Female Education on Use of Maternal Health Services .......................................................................................................................... 81 Figure 4.14~:Impact of Lady Health Worker Presenceand Media Exposure on Use of Maternal Health Services.......................................................................................................................... 81 Figure 4.15: Impact of Media Exposure and Distance to Primary Health Center on Use of Maternal Health Services .............................................................................................................. 82 Figure 5.1: 88 Figure 5.2: Impact of Schooling on Women's Labor Force Participation Rate...................................... Labor Force Participation Rates inUrban Areas by Province.............................................. 89 Figure 5.3: Impact of Husband's Schooling on Women's Labor Force Participation Rate.................... 94 Figure 5.4: Impact of Household Socioeconomic Status on Women's Labor Force 94 Figure 5.5: Impact of Schooling on Earnings ......................................................................................... Participation Rate........................................................................................................... 95 Figure 5.6: Impact of Young Children on Women's Labor Force ParticipationRate ............................ 96 Figure 5.7: Percentage o f Rural Women who Report Time Spent Fetching Water and Gathering Firewood............................................................................................................................... 97 BOXES Box 1.2: Customs that Influence Women's Freedom of Movement outside the Household..................... Box 1.1: Attainingthe MillenniumDevelopmentGoals in Pakistan: What Are the Gender Issues............3 12 Box 1.3: Qualitative Study on Gender....................................................................................................... Box 2.1: Exposureto Media Informationon Women's Legal Rights Makes a Difference....................... 14 39 Box 3.1: Rural HouseholdDecisions (not) to Educate Daughters............................................................. 51 Box 3.2: Puberty and Girls' Restricted Mobility May Constrain Their Schooling.................................... 53 Box 4.1: Overcoming Barriers to Accessing Health Services: Insightsfrom Vietnam ............................. 83 Box 4.2: Integrating Centrally-Sponsored Programs with State Programs: The Auxiliary Nurse 84 Box 5.1: Social Perception o f Paid Work by Women............;. ................................................................. Midwife Program in India........................................................................................................ Box 5.2: ImprovingAccess to Basic Services Can RaiseWomen's Labor Force Participation ...............97 -93 EXECUTIVE SUMMARY 1. The quality of life for women and girls in Pakistan has improved in recent decades. School enrollments have risen across all income categories. Child health indicators such as immunization rates and infant mortality have improved for girls and boys. Fertility rates are declining leadingto better health for women. Women's participation in paid labor has increased, particularly in agriculture, and their involvement in the political process has risen thanks to a federal mandate in 2000 to reserve seats for women at selected levels o f government. 2. In certain dimensions, however, large gender inequalities persist. Although more girls are in school, a substantial gender gap in enrollment remains and worsens significantly as girls transition from primary to middle school. Although gender differentials in child immunization have declined, considerable gender differentials persist in other aspects of health care. The use o f reproductive health services i s low, and maternal mortality ratios remain high. The cumulative effects o f this pattern are evident in the high sex ratio o f 108 males per 100 females. In the labor market, lower educational attainment coupled with social norms that restrict mobility confine women to a limited range o f employment opportunities and low wages. The achievement o f the Millennium Development Goals (MDGs) will require significant reductions in these gender gaps. Policy interventions directed specifically at gender inequality in these public domains would allow the government to move decisively toward achieving these goals. 3. This report examines the lack o f economic Figure 1A: Evidence from Rural Sindh and Punjab: opportunities in combination with cultural and Safety Concerns social norms that determine outcomes for women (percent) to a significant degree. One manifestation o f an I `0° 80 2 enormously influential cultural norm i s the n ~ 8 0 1 82 restriction on women's mobility. According to multiple sources of data-including a nationally representative survey and more focused surveys o f rural households in Punjab and Sindh (Figure lA)-the safety o f females in public spaces is a constant worry for urban and rural families alike. A mother may keep her daughter from attending school so that she does not have to walk alone. A woman who needs medical care for herself or her North Punjab' South Punjab1 Sindh 1North Punjab South Punjab 1 ~ Sindh children cannot travel alone to a health center, 11 While wlhin settlerent , While outside senlerent particularly if it i s outside her settlement. A 0 SAFE PUNSAFE common thread emerges in all dimensions, though their sources are numerous and varied, that traces Note: These figures represent the percentage of currently married gender gaps to restrictions on the physical women aged 15-40 who reportedfeeling safe and unsafe within and outside the settlement. mobility o f females and their access to Source: PakistanRural Household Survey, Round2,2004. information: both undercut their ability to acquire key services and pursue life opportunities. Mobility restrictions directly undermine female access to medical care, education, opportunities for paid work, voting and other forms o f political and community participation. The more women are secluded inhouseholds or settlements, the more they lack access to a broad range o f information and are unaware o f their legal rights, the importance o f health maintenance, and the benefits o f participating inthe public sphere. 4. Easing these mobility constraints involves Figure 1B: Evidence from Rural Sindh and inherently gradual processes. Fundamental cultural Punjab: RestrictedMobility shifts require changes inpeople's perceptions o ftheir (percent) environment and, in particular, o f the roles that men Traml time to Basic HealthUnit >=lhour and women can assume. The current restrictions on women's movement outside the home arise from concerns about their security and reputation; as long as these concerns remain, so will the restrictions. By fostering a safer environment-improving women's access to justice, for instance-the government can activate a positive cycle o f increased female participation inthe world beyond the household. The 0Cannot Go Alone Can go alone perception o f a marginally safer environment (both Tramltime to Basic HealthUnit c lhour inside and outside the village) will encourage those individuals who most desire change to respond to government initiatives, and venture beyond traditional restrictions to attend school, earn income, or vote. Along with community awareness o f this marginal change, and the realization that increased mobility does not automatically invoke retaliation, will come further marginal improvement. In this gradual way, the practice o f increased mobility can I Cannot Go Alone Can go alone diffuse to women in other communities ready to Note: Percentageof women aged 15-49. Cross-tabulations from a progress. A logical first step is to increase female question asking respondents if they could travel to the facility alone and whether they needed permission from someone in the enrollments in school, which is likely to unleash a householdto go to the facility. Most women who reportedneeding process driven by women's demand for greater permission also reported that they needed a male household member's permission(father-in-lawor husband). opportunities and involvement inthe public sphere. Source: PakistanRural HouseholdSurvey, Round 1,2001. 5. IfPakistanistoreducegendergapsandachieveitsdevelopmentgoals,policyinterventionswill require a dual focus on near-term and long-term outcomes. In the near term, females need access to basic services and opportunities. In the longer term the economic, cultural, and political environment must sustain improved circumstances for women in health, labor force participation, and other outcomes. Far deeper and more integrated initiatives are needed if long-standing trends in gender inequality are to be reversed, What role does public policy play? In many cases minor changes in laws and institutions can foster greater involvement by women in the public sphere to enable them to pursue activities that further enhance their autonomy and elevate their status. Through an iterative process, these incremental changes contribute to equalizing opportunities for women, altering society's long-standing perceptions, and easing associated constraints. Such changes may encourage parents to educate their daughters, for instance, which will enable future generations o f women to make better health-related and economic decisions within the household, and to participate in political life where they can contribute to further social and legal change. 6. What i s to be done in the meantime, as institutional reforms and economic growth may make limited and slow progress? Active policy measures to promote gender equality in the present are crucial, In particular, near-term approaches must work around existing constraints on women and girls, augmenting their access to basic services, paid work, and opportunities for decision-making in the public sphere. The analysis in this report has incorporated research and insights from scholars and civil society organizations in Pakistan in order to arrive at precisely these types o f near-term approaches. .. 11 7. Fortunately, since the late 1990s, the Government o f Pakistan has nurtured a climate that is conducive to achieving greater gender equality by launching programs designed to increase girls' school enrollments, enhance female access to health care, and facilitate women's participation in the public arena. Pakistan has ratified the Convention on the Elimination o f All Forms o f Discrimination against Women (CEDAW). Specific policies to promote gender equity have been articulated in the National Plan for Development and Empowerment o f Women (2002), National Plan o f Action (1998) and the government's Poverty Reduction Strategy Paper (PRSP).' Prominent civil society organizations have stepped-up efforts to educate the public about the benefits o f reducing gender inequalities; moreover, they are increasingly engaged in dialogue with government ministries about how to incorporate gender issues across all major areas o f development policy. This report identifies policy levers (summarized below) that can augment the ongoing momentum to close gender gaps. It also reveals key obstacles common to gender gaps in all dimensions and points the way forward to enhancing women's opportunities and status in areas where no policy for gender equality currently exists. There are indeed definitive steps the government can take to advance gender equality, the promise implicit in its commitment to achieve the MDGs. Key Themes and PolicyImplications 8. The report focuses on women's disadvantages in family law and inheritance, education, health outcomes, and labor force participation. Brief attention also i s given to women's involvement in other aspects o f public life, such as political participation and community decisionmaking. The analysis uses a mix o f quantitative and qualitative data to identify determinants o f gender gaps in all these themes: quantitative sources include the Pakistan Integrated Household Survey (P1HS)-an urban-ruralhousehold survey representative o f Balochistan, North West Frontier Province (NWFP), Punjab, and Sindh-and two rounds o f the Pakistan Rural Household Survey, 2001 and 2004 (PRHS-I and PRHS 11); qualitative data that comes from interviews with women from rural Punjab and Sindh offers more nuanced explanations of gender gaps' determinants. Background papers by Pakistani scholars and legal experts, commissioned for this report, offer further insights. Each theme thus involves a complex, multi-faceted analysis grounded in absorbable public policy recommendations. Creatingan Enabling Environment to Improve Women's Rights 9. To understandthe condition o f women in a society and their capacity for action, it is important to examine their status both within the domain o f the family and within the larger cultural and sociopolitical context. One aspect of this i s family law which, in Pakistan, i s a mixture o f codified law and customary practices. This report focuses on the rights o f women as articulated within existing family law, and examines the extent to which such laws are enforced, or enforceable in practice, as well as the ways in which traditional institutions interact with family law. Since familial attachments and networks define many aspects o f individual status and rights, the interaction between custom and law withinthe domain o f the family essentially defines the defacto set of opportunities available to women, as well as the barriers they confront in fulfilling other basic needs such as education and health. 10. While some customary practices are often flagrant violations o f state as well as official Islamic law, lacunae in laws can sometimes be filled by customary practices. Ineither case, the potential effect o f custom on women's welfare is quite complex. Inparticular, practices which could easily be perceived as detrimental to women may actually enhance their welfare inthe absence o f enforceable legal institutions. Thus, understanding these institutions and their interaction with the law is an essential step toward identifyingpolicy levers likely to create conditions for the desired outcomes to emerge. ' Government of Pakistan (2003). ... 111 11, The report focuses on two areas o f family law with a substantial potential impact on the welfare of women. The first i s the right to inherit family wealth. Although women in Pakistan have the legal right to inherit, most women with a clear understanding o f their legal entitlement, and whose families have heritable wealth, "voluntarily" relinquish their right infavor o f male relatives, typically a brother (Box 1). By far, the most important reason cited for this is the retention of natal family protection. Most women feel that giving up their legal right to natal family wealth maintains the natal family's obligation to provide financial and physical security in the event o f adverse circumstances. Clearly the presence o f articulated laws, though necessary, may not be sufficient to change behavior. The absence o f complementary formal institutions, inparticular adequate safety nets and access to legal services, not only weakens the enforceability o f such laws; it also reinforces exclusive dependence on family networks Box 1: Rural Women's Voices on Inheritance Ithink,Ishould not claim my share.Iwillnot takeit. My brother is dear to me. Brothers help in the hour of need...IfIwill claim my share, the warmthof our relationship will be affected. Ifeven brothers do not dislike it, their wives and children would definitely feel that. They say that we have taken share from them.Although, it is our right and we should take it but the relationship is affected. -Woman from northern Punjab (Faisalabad), age 30 Ihavenever askedfor my share,nor havethey evergiven it tome.Even ifthere comesatoughtime,I will not askfor my share. IfI askfor my share, others will not say anything, but myfather and brother will become angry with me. -Woman from Sindh(Mirpurkhas),age 24 Ihaveaskedmy brother togive memy share....But hesays thatIam married andit is my husbandls responsibility tofeed me and take care of me. Here most of the womenforego their share; they want to take it but they don 't get it. They can't do anything when they don 't get it; they remain silent about it. -Woman from Sindh (Mirpurkhas),age 22 12. Women's rights under marriage are another domain where numerous laws protecting women have been enacted, many in the earlier part o f the twentieth century. Despite the fact that these laws have been in existence for over half a century, legal scholars and human rights organizations in Pakistan agree that girls and women continue to confront profound disadvantages in marriage and divorce. This report examines laws related to child marriage and legal provisions related to the marital contract. The latter includes the Nikahnama (marriage contract), dowry, the rightto divorce, alimony and Haq Meher. 13. The report finds significant improvementsin the enforcement o f the law related to child marriage. Among rural households, data from Sindh and Punjab suggest that the median age o f marriage is 17, and the mean age is 18. However, close to one-third o f girls still marry before the legal minimumage of 16. Thus, there is significant scope for improving the enforcement o f the child marriage act. Prominent civil society organizations, such as the Aurat Foundation, have stepped in to increase enforcement and have proposedthe redrafting o f laws to further increase the legal minimumage at marriage to 18 years. 14. The Nikahnama i s the centerpiece o f the Muslim Family Law Ordinance o f 1961. By requiring the registration o f a marriage contract the MFLO attempted to make marital practices more transparent and accountable to the law. Although the Nikahnama contains unprecedented protections for women, few o f these provisions are actually enforced. The barriers to enforcement include both a lack o f awareness of the existence o f such laws and customary practices that actively restrict access to the provisions o f the Nikahnama.For example, 75 percent o f interviewed married women had no Nikahnama, and most also were not aware that one was required. Among those who had one, close to three-fourths had no idea what was in the document. Given, then, that only seven percent actually had a Nikahnama and knew what it iv stipulates, the scope o f its protection seems severely limited. Indeed, practically no woman who had a Nikahnama reportedthat it gave her the right o f divorce. The same is true about Haq Meher, a severance clause whereby the husband agrees to pay a pre-specified cash amount to the wife in the event that he initiates divorce. Nominally, Haq Meher appears to be quite important but in the vast majority o f cases, the amounts were far too small to actually provide any viable economic protectionto women. 15. A complex array o f cultural practices around marriage appears to substitute or compensate, at least in part, for poor enforcement o f marital laws. Most women in rural Pakistan marry a close relative (over 78 percent), and the vast majority continue to reside in their natal village and have close daily contact with their natal families. In addition, practices like watta satta (where two different families trade their daughters for marriage) are surprisingly common. More than one in three married women in rural women i s in a watta satta arrangement. While this practice has understandably been a cause for concern for women's rights advocates, since it clearly restricts women's options within marriage, it i s not difficult to see why it continues to thrive. Simply put, the presence o f the daughter's sister-in-law in her natal household dampens the potential for serious mistreatment o f the daughter. Indeed, we find that women who are in a watta satta arrangement report lower levels o f depression and domestic violence and are significantly less likely to experience marital estrangement. Examining customary practice o f this sort is important because it suggests that unless complementary legal reforms are put in place and enforced, customary practices that afford informal protection are likely to persist and would be hard to eliminate. 16. A number o f recommendations on modifications to Family Law from legal scholars in Pakistan resonate with the findings summarized above. Prominent advocates for women's rights in Pakistan have called for increased legal protections related to women's rights under marriage and divorce by amending current provisions in these areas, also recommending specific changes to the Dissolution of Muslim Marriages Act (DMMA) (1939) and to the Nikahnama form to ease restrictions on women's rights to divorce. 17. While modifications to family law constitute clear policy levers for improving the status o f women in the family, equally important is the needto build enabling institutions that allow women to take advantage of available legal protections. Examples include, strict enforcement o f record-keeping o f family events such as marriage, divorce, birth and death which provides women with the crucial evidence needed to seek legal aid and protection; educating women about their rights through intensive outreach and information campaigns regardingchild marriage, dowry, and divorce; and improving access to justice through channels such as free legal aid. Furthermore, to encourage women to actively pursue their rights they must be assured of their physical safety. In this domain, increased provision o f safe houses, crisis centers, and responsive police protection is crucial. Equally, women who move against convention in pursuito ftheir rightsmustbe assured minimumfinancial security through adequate formal safety nets. AddressingMobility Constraints to Enhance Schooling Outcomes 18. Only one of the eight MDGs explicitly promotes gender equality and the empowerment o f women, but gender issues are germane to achieving all o f them. Especially important i s the achievement o f universal education. Education enables women to be more productive both inside and outside the household. A mother's education also has a beneficial impact on family size, the well-being of her children, `and her use of community services. Educated women are also more likely to participate in the political process; illiteracy i s a major obstacle in accessing relevant information and dealing with electoral procedures and political issues. V 19. The report finds that only 58 percent o f primary Figure 2: Percentage of Children Who school-aged boys and 46 percent o f same-aged girls Dropped Out of School were enrolled in primary school in 2001-02. This gap I ' worsens substantially for girls who are 12 and older. Moreover, the rural-urban divide is striking. The gender ' 80 1 gap i s largely a rural phenomenon. In rural areas, girls %60 -1 are less likely to have ever enrolled in school and ifthey tGirls +Boys I , do enroll they are much more likely to drop out (Figure 40 4 2). This leaves little room for inaction if the gender equity goals set out inthe PRSP are to be met. 5 7 9 11 13 15 17 I 20. Girls' education has been the subject o f much Age debate and research in Pakistan, and a number of Source: Based on World Bank staff tabulations using PIHS important demand- and supply-side factors which impact 2001-02 data on rural children. schooling decisions have already been identified. These include demand-side constraints such as family income, parental education, parental attitudes and differences in returns to schooling, as well as supply-side constraints related to school availability and quality. Recent policy initiatives have responded to these findings by introducing important new programs such as stipend schemes and school meal programs. The analysis in this report points to an additional constraint arising from restrictions on the mobility o f girls and young women and argues that if this constraint is not accounted for it is likely to dampen the efficacy o f the aforementioned policy interventions. 21. Analysis results suggest that the practice o f restricted female mobility plays a large role in perpetuating gender gaps in school enrollments. School attendance for girls i s very sensitive to school proximity. Girls are much less likely to attend school unless there i s one available within the settlement they reside in. This sensitivity to school proximity worsens as girls grow into adolescence. Qualitative studies suggest that concerns over safety and norms of female seclusion are the primary factors behind the precipitous drop in enrollment beyond age 12. This concern is also evident in the rising expenditure on transportation to school reportedfor older girls. 22. Decreasing the physical Cost Ofattending school for girls thus i s likely to pay big dividends. H o w can this Figure 3: Average Number of Qualified Women in Rural Communities be achieved? Since large parts o f rural Pakistan are underserved or unserved by schools for girls, school ' 2 5 - construction will continue to be important. However, a more nuanced approach that addresses the needs o f scattered rural population-where dedicated community- 4 ~ level schools are unfeasible-is also required. Even where feasible, however, the construction o f schools is likely to face another important constraint: there simply are not O 5 enough educated women in many Pakistani villages to I staff a school for girls (Figure 3). Government schools I Sindh NWFP Balochistan 1 (and most private schools) for girls require female UCompleted Grade 8 .Completed Grade 10 I prevent educated women from relocating or commuting to grade localities with teaching jobs. Hiring and retaining female 18 and 50 who have completedat least 8 (middle school) or grade lO(high school) teachers thus will continue to be a problem, and ironically Source 'IHS household 2o01-02 data. this problemwill be at its worst inprecisely those areas that are poorly served at present. vi 23. Paradoxically, the construction o f a middle or high school i s not likely to be warranted in every community; however, the absence o f such schools sufficiently close by will hinder the development o f public primary schools, thereby discouraging private primary schools. Breaking this unfortunate cycle will require innovative interventions to ensure girls with access to middle and high schools without having to construct a middle or high school in every village. While cultural constraints on female mobility are not likely to yield to short-run policy levers, the creation o f a cohort o f educated women in every village may be a viable policy interventionon the supply side. Marriage, residence, and migrationpatterns suggest that educated girls frequently remain inthe villages they come from, so this constitutes a potential pool of future teachers for the next generation. 24. Improving schooling outcomes requires a comprehensive strategy that can address the demand- and supply-side constraints noted above. First, given the poor distribution o f schools, there i s a clear need for new school construction to expand access to schools in areas that are currently unserved or underserved. In addition, the conversion o f public primary boys' schools to coeducational institutions is likely to be quite effective in immediately improving primary school access for girls in areas where no primary girls' school exist. Analysis also shows that private primary schools locate only in areas with a sufficient pool of educated women who can be hired as teachers. Thus, public investments in middle and high schools are also likely to have very large payoffs for primary schooling, since they could very quickly produce a pool o f qualified women who could not only staff existing public schools but also serve to pull in new private schools. In addition, it may be possible to immediately alleviate shortages o f teachers by providing residential accommodation with some assured security, as has been attempted in rural Balochistan with some success. Where teacher quality is a concern, mobile teacher training units could also be considered. The success o f this strategy relies, however, on ensuring that young girls who have primary schooling go on to complete middle and high school. This requires appropriate demand-side initiativesthat are sensitive to the mobility constraints highlighted above. 25. Several demand-side initiatives such as the middle school stipend program and the school meal program (Tawana Pakistan) are already underway. Such schemes rely on the idea that low enrollments are primarily due to financial constraints. Without doubt such constraints are likely to be quite important for many rural households. Even where households are concerned about the safety o f young girls, a stipend could allow them to purchase private secure transportation to and from school. Our analysis indicates, however, that safety concerns are not likely to be addressed completely by reliance on existing means o f transportation available in most villages. A rigorous evaluation o f the stipend program i s therefore urgently needed to identify whether the stipend should be pegged to school distance as well as the extent to which uptake o f the program is dampened by safety and mobility concerns. Ifstipends alone are found to be inadequate, several complementary initiatives could be considered to augment their effectiveness. For example, where a school exists within walking distance, trained and licensed chaperones could be used to escort young girls to and from school. Where schools are at a greater distance, as is the case with most middle and high schools, subsidized provision o f secure school transport could also be considered. The feasibility of such complementary initiatives could be assessedas part of an evaluationthough pilot schemes. IncreasingAccess to Health Services to Improve Health Outcomes 26. Three major sets of factors interact to slow the improvement in health outcomes in Pakistan, as in other parts o f South Asia. First, highrates o f poverty make for poor nutrition and health conditions. This is aggravated by neglect of public health and environmental sanitation services, as a consequence o f which peopl-nd poor people in particular, given their living and working conditions-are prone to high levels o f infection. Second, the coverage and quality o f publicly provided health services is poor. Third, gender inequities place constraints on women's and girls' access to health information and vii services. The first two issues are not distinct to women and have been analyzed elsewhere. We focus on the specific constraints to improvingwomen's health outcomes. 27. As with education, mobility restrictions play a large role in perpetuating gender gaps in health outcomes. Further constraints faced by women in seeking health care include limitations on decisionmaking and access to information. If women feel that they or their children need health care, they have to persuade their husbands and/or elders o f this need, obtain permissionto seek care, and find someone to accompany them. They also must be able to access critical information on when to seek medical care and seek information on what health services are supposed to be available to them-not easy when they have limited interaction with the outside world. Women thus face a series o f daunting hurdles to seeking timely health care, even if a well-functioning facility i s available nearby and the household can afford the treatment. 28. What can the government o f Pakistan do to implement its commitment to meetingthe MDGs, two o f which focus on health and one specifically on maternal health? This will require significant effort and investment on multiple fronts. There have been encouraging trends in childhood immunization. Trends inother MDG indicators are less encouraging, however-in particularthat o f a three-quarters reduction in maternal mortality between 1990 and 2015: the maternal mortality ratio in 2000 was only six percent lower than the ratio in 1990. 29. The timeliness and quality o f care during Figure 4: Percentage of Women Receiving pregnancy and childbirth strongly influence maternal Maternal Health Services health, yet the proportion o f women receiving health 70 care during pregnancy and childbirth is low, and has risen slowly and only slightly since the 1980s (Figure 4). Only 35 percent o f women in Pakistan reported receiving prenatal care during their most recent pregnancy, which represents only a 17-percent increase from the late 1980s. The figure for urban areas is 63 Renatal care Births at home Births Postnatalcare Y percent, but only 26 percent in rural areas. Nearly four assisted by out o f five births during 1998-2001 took place at home. TBA The proportion of institutional deliveries rose by only Note: Percentages based on cross-tabulationsfrom the PIHS eight percent from the late 1980s. Levels o f postnatal data. Data refer to use of maternal health services by care are very low, even in urban areas. There are also pregnant women in the three years preceding the PIHS large interprovincial differences in coverage o f these survey, Categories "births at home" "births assisted by trainedbirth assistants" (TBAs) are not mutually exclusive. maternal care services. Punjab i s the best-served 5'ource: PIHS 2001-02household survey data. province, and Balochistan shows the lowest service coverage, followed closely by the NWFP. 30. Gender inequities translate into low investment in the health o f female children and adolescents, Women thus enter childbearing years carrying the burdens o f deprivation during childhood and adolescence; their health reserves are further drained by repeated childbearing and inadequate care during maternity. The resultant cumulative depletion takes its toll in high rates o f maternal morbidity and mortality and inpoor health outcomes for their children. ... V l l l Worker (LHW) program i s effective at increasing the frommdia+No uptake o f some services, such as contraceptive use and L W immunization. Our data confirm the well-established T;O;z;~ schooling. But we also find that information campaigns frommd'a+ increase service utilization, independent of women's No L W schooling, LHW presence, and distance to facility lnformat,On frommdia+ 29% (Figure 5). L W recommendations which can help reducesetsthe gender- 32. From this emerge Of Policy Note: These figures pertainto rural marriedwomen aged 15- 49. The figures refer to the predicted probabilities of using these services from Tables A4.3 and A4.4. In graph A, the relatedconstraints healthoutcomes. ~ i ~ ~ ~ , to improving predicted effect of media exposure is added to the effect of it is critical to mount intensive information campaigns woman's schoolingon the use of these services. Ingraph B, the predicted effect of media exposure is added tothe effect covering a wide range o f health messages. This will help of LHW presence. women access information directly within their homes; moreover, it can also help build community acceptance for paying more attention to women's health needs and thereby reduce the social barriers to women accessing health care independently. Second, the LHW program should be expanded and strengthened. Where such expansion is currently not possible, mobile service provision should be made available in such a way that people know when and where the services can be accessed. At the same time, some changes need to be made in the organization o f the LHW program. We do not find evidence o f hoped for synergies between the LHW program and the health facilities; the presence of a LHW does not seem to stimulate women to use the services o f the health facilities. The report outlines a few simple measures which could help integrate LHWs more with the functioning of the health facilities and thereby create a more coordinated and effective network o f health service delivery to rural households. Drawing Women into the Workforce and Broader Public Arena 33. Women's participation in the labor force not only serves to augment household income and welfare; it also can provide a tremendous impetus to enhanced participation in public affairs. The National Policy for Development and Empowerment o f Women 2002 places particular emphasis on increasing the economic empowerment o f women through their increased participation in the workforce. Nationally representative data indicate that women participate in the labor force at substantially lower rates than men in both urban and rural regions. At the end of the 1990s, only one in four adult women (aged 10 and older) participated in the labor force, a far lower rate than the nearly 70 percent participation rate for men. ix 34. There are also some important sectoral and Figure 6: Labor Force Participation Rates regional variations in female labor force participation by Province (percent) worth noting (Figure 6). First, almost twice as many Punjab women report labor market activity in rural areas as compared to urban areas. Second, a much larger fraction Sindh of women report labor market activity in rural Sindh and NWFP Punjab, as compared to the North West Frontier Province Balochistan and Balochistan. There is no such variation in male All prounces participation rates. Provincial differences in female All prounces NBl participation rates are likely to reflect differential opportunities for agricultural employment. As we discuss All prounces urban below, women in rural areas work predominantly in o Males IFemales 35. Three additional aspects o f female labor market participation are worth noting. First, women work in a much narrower set o f occupations than men. Rural women tend to be concentrated in agriculture, while urban women tend to work predominantly in unskilled service jobs such as personal and household services. Second, the occupations inwhich women are predominantly engaged offer lower wages. Third, these occupations are much more likely to keep women close to or inside the home. Labor force participation by occupation shows that a much higher proportion of urban men than women are engaged in white-collar jobs and within occupations, suggesting further gender segregation by the type of white collar job. A recent report on the need for quotas for women in public sector jobs reports that, despite the existence o f quotas across all cadres, women tend to be concentrated in the education and health departments. This may reflect, in part, a decision by women to remain within the domain o f "socially acceptable" work. Beyond white-collar jobs df the type described above, women's labor market participation in urban areas seems to be concentrated in home-based manufacturing work. One study estimates that since the 1980s, there has been a dramatic increase in the proportion of urban women engaged in informal home-based work, primarily crafts and related occupations. While Pakistani researchers and policymakers have increasingly focused on the plight o f home-based informal workers, especially that o f women home-based workers, initiatives inthis area are hampered by a lack o f data. 36. The analysis in this report shows that among all the possible determining factors, mobility restrictions are perhaps the most important constraint to women's participation in the labor force. Social barriers to women's mobility also shape the sectors and occupations they work in and their choice o f location o f work. In fact, the same cultural restrictions that limit female access to education and health facilities also constrain their opportunities to earn income. 37. As we have discussed above, women's mobility may be limited due to Table 1: Effect of Safety and Participation on Paid Work concerns about their safety as well as Woman's concerns about All paid Paidfarm norms of family honor and seclusion safe work work restrictions. The report shows that Feelsunsafe within settlement Negative Negative impact impact ~ ~ n c e about safety also discourage r n ~ participation in paid work and confine Feelsunsafe outside settlement NO impact NO impact women to jobs within their village or Source: Basedonprobit regressionresultsreportedin Table A5.3 and PRHS 2004. settlement. As Table 1 shows, women who feel unsafe walking within their own settlement are much less likely to work for pay. Analysis also shows that women whose husbands are educated and women who belong to higher-income households are less likely to work. Qualitative data suggests that this may be becausethere is a social stigma attached to women working outside the home. X 38. The limited participation o f women in wage labor is o f particular concern not only because it limits the income-earning opportunities available to households, but also because participation in paid work has important implications for women's autonomy. Many studies have found that working women have greater voice in household decisions. This report identifies an additional aspect o f empowerment through participation in labor markets. Women who work are significantly more likely to participate in community and political activity. Given the increasing role o f local government under the decentralization process, and thus the increased role o f communities in political decision-making, participation in work is perhaps one important avenue through which women's civic participation could be enhanced. 39. The report proposes several ways in which women's labor force participation can be stimulated. Raising women's skill levels, both through formal schooling as well as vocational training, will increase their ability to find employment. Investment in time-saving infrastructure i s also important for encouraging women's participation in work activities. The restrictions on physical mobility and segmentation o f the labor market will continue to circumscribe women's participation, however. Complementary strategies are needed to attract and retain women in the labor force. For example, micro- credit programs could enable women to establish small businesses, and better designed credit and agricultural extension program can benefit women engaged in agricultural work. Given that many women are engaged in home-based work, better regulation as well as the establishment o f appropriate benefits is also likely to be important. Finally affirmative action programs such as quotas or the repealing o f labor laws such as the Factory Law that discriminate against women are likely to be more effective if complemented by efforts to ease mobility constraints and concerns about safety. Conclusion 40. This report identifies two dimensions in which policy must address gender gaps in order to meet Pakistan's development goals. First, because cultural constraints undermine women's access to education, health services, and prospects for earning income, policy requires near-term initiatives that work around these constraints, increasing female acquisition o f basic services and opportunities. Near- term initiatives include augmenting funding o f government programs (such as LHWs and stipend programs to increase girls' attendance) that have been successful in increasing access on a small scale; modifying legislation to further empower women-by expanding their income earning-opportunities and their marital rights, for example; and enforcing these modified as well as existing laws so that women actually will take advantage o f their legal rights. 41. Second, improvements in gender equality will endure only to the degree that formal institutions reinforce them and society accepts them. Policy therefore must incorporate long-term measures to create an environment that enables the reduction o f gender gaps. As discussed earlier, the gradual process o f social change means that many iterations o f change across multiple generations have to occur before conditions improve for women on a wide scale. This slow nature o f widespread improvement does not imply that government can be complacent and wait for a fundamental shift to occur, however. On the contrary, activating and sustaining the virtuous cycle o f improvements requires strategies that reinforce near-term measures. Increasing girls' school enrollments builds future cohorts o f local female teachers and health care providers, but to accomplish this, girls first must feel safe enough to attend schools outside their communities when there are no schools nearby, and women must feel secure enough to work outside their homes, even when this is not the norm. By creating a public arena that is more welcoming to women, policies that initiate improvements infemale human development will achieve the desired results. Once women perceive that there is an arena inwhich they can apply their greater development, their own demand for improved access to services and life opportunities will act as the most effective momentum for fundamental change. The shortest route to improving gender equality i s that which fuels demand for change and accelerates a virtuous cycle o f improvements. xi 1. IMPROVING IN GENDEREQUALITY PAKISTAN Small Steps toDate, Large StridesAhead To have an adequate appreciation of thefar-reaching effects of disparities between women and men, we have to recogn'izethe basicfact that gender inequality is not one affliction, but many,with varying reach on the lives of women and men, and of girls and boys. -Amartya Sen2 1.1, South Asia stands out among all the regions o f the world as a region with a highdegree o f gender inequality-in opportunities, resources, and rewards enjoyed by men and women. Within South Asia, gender disparities in Pakistan are also pronounced: they cut across all classes, sectors, and regions of the country. Although the issues o f gender inequality in Pakistan are well documented, less is known about what drives these gender differences and what policy levers are at hand to effect change. This Gender Assessment describes the multiple dimensions o f tackling these gender inequalities and identifies implementable policies that will most effectively alleviate gender gaps. Achieving this objective requires in-depth understanding o f both economic and non-economic issues. Because existing data provide only a partial grasp o f factors driving gender inequalities, the Gender Assessment has combined data analysis with information on legal, political, and socio-cultural environments. Itis vital to includethe influence of socio-cultural norms on families' reactions to policies and programs, or we risk creating initiatives that are unsuccessful, even ifthey provide all the right economic incentives. 1.2. The recent Policy Research Report on Gender (2001)' extensively analyzed gender issues across the developing world and provided evidence o f the types o f reforms and policies that can promote gender equality. It also stressed that one-size-fits-all policies for promoting gender equality will not work. This Country Gender Assessment (CGA) examines what policies would succeed or fail in the Pakistani context. The CGA builds on Pakistan's Poverty Assessment (2002), which showed that steady economic growth throughout the 1980s and 1990s was not accompanied by commensurate social improvements. With a gross national income (GNI) per capita o f $470 (2003) and 32.6 percent o f the nationalpopulation below the poverty line, Pakistan remains classified as a low-income e ~ o n o m yAt~ the end of the 1990s, . its social indicators were below those o f other developing countries with similar per capita incomes, and they improved more slowly than those of countries with similar growth rates.5 Levels o f illiteracy are among the highest in the region, while school enrollment rates are the lowest. Figure 1.1 reveals large gender gaps in literacy and enrollment relative to the rest of South Asia and to lower-income countries in general. ~~ Based on the text of an inauguration lecture for the new Radcliffe Institute at Harvard University on April 24, 2001. A shortenedversion of this paper was publishedin TheNew Republic on September 17,2001. World Bank (2001). World Bank (2004~). World Bank (2002). 1.3. The economic revival program introduced at the end o f the 1990s is Figure 1.1: RegionalComparisons beginning to reverse the deteriorating macro Fig. 1,la: Regional Comparison: Adult illiteracy rates situation o f a few years ago. Many financial and legislative decision-making I 1 responsibilities have been decentralized to 71 provincial and lower-level local governments, with the goal of improving accountability and 57 3 service delivery. The budget deficit has fallen, inflation has remained below five percent, the Low income current account deficit in the balance o f countnes 48.5 payments has turned into a surplus, and exports have begun to grow again after years o f stagnation. rrr 1.4. All of these developments help create an environment conducive to reducing gender Fig 1 1b RegionalComparison Primary Net EnrollmentRates by gender disparities, and policymakers have committed themselves to a number o f gender specific 120 - I goals in recent years. Pakistan has ratified the Convention on the Elimination o f All Forms of Discrimination against Women (CEDAW). Specific policies to promote gender equity have been articulated in the National Plan for Development and Empowerment o f Women (2002), National Plan o f Action (1998) and Pakistan SouthAsia Low income Sri Lanka the government's Poverty Reduction Strategy countries Paper (PRSP).6 The PRSP is aligned with the Millennium Development Goals (MDGs) and Primary Net Enrollment Rates Boys Primary Net EnrollmentRates Girls ~ ~ identifies gender equality as an explicit goal Note: Illiteracy rates pertainto those 15 andolder. (see Box 1.1). Source: Figures for Pakistan calculated using PIHS 2001. Figures for low- income countries, South Asia, and Sri Lanka are taken from Genderstats World Bank (2004). 1.5, While only one MDG explicitly cites the promotion o f gender equality and the empowerment o f women, gender issues are germane to achieving all eight MDGs. Various dimensions o f the gender gap each have their own effect on the development trajectory, and their effects are also interactive. A primary indicator for eradicating gender disparities is equality in educational opportunity, as measured by school enrollments. A one-percent rise in female school enrollments boosts average GDP levels by 0.37 percent, whereas a one-percent increase in male education has no significant effect; conversely, gender disparity itself significantly lowers levels o f per capita income.' A cross-country study on the impact o f missingthe MDGtarget on gender equality estimates that countries such as Pakistan, which have not achieved the targeted equity in education by 2005, risk losing an average o f 0.4 percentage points in annual economic growth between 2005 and 2015 ifthey remain off track.' Potential negative social effects o f failing to achieve universal education include between 0.1 and 0.6 more births per woman, 2.4 percentage points greater incidence o f underweight children under age five, and up to 32 per 1,000 higher child mortality by 2015. Government o f Pakistan (2003). Klasen (1999) ;Knowles et al., (2002). Abu-Ghaida and Klasen (2004). 2 Box 1.1: Attaining the Millennium Development Goals in Pakistan: What Are the Gender Issues? 1 The government's pledge to achieve the MDGs by 2015 means it is preparedto assume responsibility for ameliorating gender disparities insocial, economic, andpoliticalspheres. While only one of the eight MDGs promotes gender equality and the empowerment of women, gender issues are germane to achieving all eight MDGs. Gender gaps in a country's political, social, cultural, and economic dimensions are intertwined with the country's overall development trajectory. Especially important is the achievementof universal education; because increases in educational attainment are transmitted to subsequent generations, reductions in the gender gap for education are self-perpetuating, Female education also advances other development goals: women are more productive both inside and outside the household; and evidence overwhelmingly shows a mother's educationto have a beneficial impact on family size, the well-being o f her children, and her use of community services.9 Outside the home, women's education raises productivity in both wage employment and agriculture," and enables women to meaningfully participateinthe political process. A primary indicator for eradicating gender disparities is equality in educational opportunity, as measuredby school enrollments. In Pakistan, allocations for basic education have increased, yet public expenditure on educationas apercentageof GDP has remainedlow, comprising about 1.8 percent in 1998." How has public expenditure affected gender gaps in schooling? Analysis of public spending12suggests a significant gender gap among the lower-income quintiles, where girls receive much less in government resources per capita than do boys at both the primary and secondary school levels. Extending this analysis to understandhow girls' enrollment will respond to an increase in public education expenditure also reveals a gender gap among the lower-income quintiles-the marginal impact of an increase in total spending is higher on boys enrollment than on girls'-indicating that boys tend to benefit more than girls from public expenditure on educationat both the primary and secondary 1 e ~ e l s . l ~ Ifthispattern ofpublicexpenditures Figure 1.2: Gender Ratio in Enrollments and MDG Target does not improve, Pakistan will be hard-pressed to attain its goal of 1 0 5 ~ ~ I. ~ ~ x_I ~ ~ ~~~ Ix II gender equality in education by the -6 100% year 2015 (Figure l.2).14 Even 95% though the ratio of the number of ' m f 90% -Actual 85% girls enrolled in primary school e 80% h P R S P Target 1 (regardless of age), relative to the : 75% -MDG Target number of boys enrolled, showed E 70% improvements inthe first half of the 65% 60% 1990s, it has since taken a downward turn. In 2001, the ratio was 74 percent. Pakistan's PRSP aims to reach a ratio Of 82 percent in primary enrollment by 2005, Source: Ratio o f primary Gross Enrollments Rates for 1991, 1995, and 1998taken from PakistanPovertyAssessment(2002). Ratio for 2001basedon World Brnk staff much lower than the MDG target of calculations using PIHS 2001. PRSP target taken from Government of Pakistan 100percent.Ifthe 1990's pattern o f (2003). public expenditureson education persists, achieving any significant improvement in girls' enrollments seems unlikely inthe short-term, or by the MDG target date of 2015. Strauss andThomas (1995). loBehrman and Deolalikar (1988) Quisumbing (1996). World Bank (2002). For comparison, note that in Sri Lanka education spendingwas 5.3 percent of the GDP. Benefit incidence analysis of the impact of public expenditure on education on school enrollments of girls and boys taken from note preparedfor World Bank (2004d) . The expenditures are for 2000-01, while data on school enrollments of girls and boys are from the 1998-99 Pakistan Integrated Household Survey. l3 Under the assumption that public expenditures are proportional to public school enrollments, this exercise of estimating marginal benefit incidence is an approximation of the actual dynamic impact of change in education expenditure on outcomes, 3 1.6. The social justice inherent in promoting gender equality makes its achievement an important aspect of human welfare, intrinsically worth pursuing. As an outcome o f gender inequalities, gender inequities refer to differences not only in men's and women's opportunitjes, but also inwhat Amartya Sen refers to as their capabilities and freedoms (or power). Inequalities in freedoms reflect the fundamental differences between men and women in their ability to achieve their capabilities, ``. the range o f things ., that a person could do and be in her life."'5 The pursuit o f gender equality promotes the fair distribution o f capabilities and freedoms, as well as offer potential economic benefits. 1.7. To reduce gender gaps and reverse long-standing trends in gender inequality will require interventions that promote women's voice and create an environment that fosters women's greater involvement in the public sphere. There have been a number o f efforts in Pakistan to address gender inequality on the legal and political fronts. The reservation o f seats for women in local government and provincial and national assemblies has brought about an unprecedented increase in women's political participation, creating space for women's voice. Government and civil society alike have led efforts to mitigate violence against women and other violations o f the law. In2004, Parliamentpassed a bill against honor killings. Discriminatory laws-prominent among them the Hudood Ordinances-have become the subject o f increasing debate inrecent years. The National Commission on the Status o f Women (NCSW), charged with the mandate o f reviewing existing laws suggesting reforms, has recommended that the Hudood Ordinances be repealed. The government currently is reviewing NCSW's recommendation on the Hudood Ordinances. 1.8. What i s to be done in the meantime, as institutionalreforms and economic growth slowly work to reduce gender inequities? Active policy measures to promote gender equality in the present are crucial. In particular, near-term approaches must work around existing constraints on women and girls, augmenting their access to basic services, paid work, and opportunities for decision-making in the public sphere, Key to this CGA is the finding that in each o f the dimensions analyzed in this study, the persistence o f gender gaps stems from both culturally-based and economic constraints. Restriction o f women's mobility-which varies by region in Pakistan, and in the most conservative form results in female seclusion-is a particularly formidable obstacle to closing gender gaps. Policies that can work around such cultural constraints will both increase service delivery to females and encourage long-term cultural shifts that will reduce these constraints. 1.9. The remainder o f this chapter is devoted to the following: (1) an overview o f gender gaps in Pakistan; (2) a discussion of the methodology and data used in the CGA; and, (3) a description o f the report structure. The overview examines the patterns and trends in indicators that reflect gender disparities arising from the economic and non-economic roles o f men and women, during the decade o f the 1990s (up to the year 2001). Data sources used include the 1998 Census and the 2001-02 round o f the Pakistan Integrated Household Survey (PIHS). I.GENDER INEQUALITY IN PAKISTAN: AN OVERVIEW 1.10. With an area o f 803,940 square kilometers, Pakistan borders India (inthe east and southeast), Iran inthe southwest, Afghanistan (inthe north and northwest), andthe Arabian Seato the south. The country i s made up o f two territories (Islamabad Capital Territory and the Federally Administered Tribal Areas) since it measures the impact of aggregate changes (e.g., change in total enrollment) on different groups (see Lanjouw and Ravallion 1999 for methodological details) using cross-sectionaldata at a certain time. l4 Actual gross enrollments are obtained from various rounds of the Pakistan Integrated Household Survey. The PRSP target is that indicated in Government of Pakistan (2003). Sen (1989). 4 and four provinces (Balochistan, North West Frontier Province, Punjab, and Sindh). The most populous o f these provinces i s Punjab, which is home to roughly one-half the country's total population o f 148.4 million (2003).16 Although Pakistan's official language i s Urdu, a host o f other languages-including Baluchi, English, Pashto, Punjabi, Saraiki and Sindhi-reflect the ethnic diversity o f the population. The largest o f these groups i s Punjabi; Pashto, Sindhi, and Saraiki each account for between 10-15 percent o f the population; and Urdu, Baluchis, and Afghans make up less than 10 percent each. Because the population i s predominantly rural (about 68 percent), nearly one-half the labor force is involved in agriculture, forestry, fishing and hunting. Health 1.11. According to the 1998 Population Figure 1.3: Sex Ratio Imbalance in Pakistan: Too Few Girls Census, the sex ratio (the ratio o f men to women in the population) is 108 males per 100 females." The sex ratio frequently is Number of males per 100females used as an indicator o f gender inequality in 90 95 100 105 110 a society because it reflects gender differentials in mortality.'* A high sex ratio Pakistan indicates premature death o f females, the source o f which could include poor female South Asia access to health inputs or social factors I ~ o wincome countries resulting in sheer neglect. Pakistan's sex ratio o f 108 males per 100 females Sri Lanka indicates excessive female mortality and surpasses even South Asia's already high ratio o f 106 (Figure 1.3). Even compared to other low-income countries, Pakistan's sex Source: Sex ratio for Pakistanbased on 1998Census. Sex ratio for SouthAsia, low-income countries and Sri Lankataken from Gender Stafs (WorldBank) for ratio i s high. the year 2000. 1.12. Amartya Sen brought attention to this simple but powerful sex ratio statistic to calculate the phenomenon he called "missing women."" In a world with no excessive female mortality the sex ratio would be 95 to 98 males per 100 females.20The difference between this benchmark and the actual ratio translates to the number o f missingwomen-that is, women who could have lived but did not because o f premature death. Taking 95 as the benchmark, Pakistan's sex ratio o f 108 implies almost 8 million missingwomen. Sen calculated that more than 100 million women were missing due to the surfeit o f female mortality inparts o f the developing world, most notably South Asia, China, West Africa and parts o f North Africa. Other social scientists have more conservatively estimated the range o f missingwomen to be between 60 and 90 million.21All of these estimates confirm the enormous toll that excess female mortality is exacting on women in these regions o fthe world. l6World Bank (2004~). The sex ratio is defined as the ratio of the number of males to females in the population, and it is calculated as the number of males per 100 females (a ratio of 95 is considered normal) from census data. Census data is the most appropriate data source to use for calculating the sex ratio, as the Census is designedto produce a count of the entire population of the country. However, sex ratio in the population also is frequently estimated using household survey data, which can be problematic because the selection of households in the survey can affect the sex ratio estimated. See Deaton (1998) for a discussion. The sex ratio also can be influenced by sex ratio at birth, migration, and under-enumeration of females. Sen (1990). 2oSen (1990), Coale (1991), Klasen and Wink (2002). Biologically, women tend to have lower death rates than men, so female death rates ideally should be lower than male death rates. Sen (1989), Sen (1990), Coale (1991), Klasen (1994) 5 1.13. Because there is little evidence o f prenatal sex selection in Pakistan, the prevailing sex ratio reflects relatively poor treatment o f girls after birth, rather than female infanticide. This phenomenon has been called "extended infanticide," where girls have an elevated mortality rate in childhood because they may be denied inputslike food, nutrition, and health care.22 1.14. Indeed, female child mortality I Table 1.1: Early Age Mortality Rates exceeds male child mortality in Pakistan. Most recent estimates of childhood Infant Child Under 5 Mortality Mortality Mortality mortality show the female rate between ages 1 and 4 to be 24 per 1,000 births, Male Female Male Female Male Female while the male rate inthis age group is only 1980- 15 per 1,000 births (Table 1.1). Among 1990191 lo2 86 22 37 122 119 infants, male mortality rates exceed those 1997- for females, a pattern consistent with 2000 99 71 15 24 112 93 biologically expected sex-based differences Note: Ratesper 1,000 live births. in rates, In Pakistan, mortality Source: Pakistan DHS 1990-91; Pakistan Reproductive Health and Family rates other than those for very young ages PlanningSurvey, 2000-01, 22Miller (1981). 23Duringthe 1990s, Pakistan's fertility rate began to decline. The total fertility rate (TFR) appears to have declined to about 4.46 by the late 1990s (1998-99 PIHS). This decline has been accompanied by an increase in contraceptive prevalence rates. According to the 1998-99PIHS, the contraceptive prevalencerate was 17 percent, which, though higher than inprevious years, is one ofthe lowest in the region. 24NIPS (2001). 25WHO, UNFPA, and UNICEF estimate(2004). 26Ashford (2002). 27Qureshi, Nazli, and Sumro (2001) provide an overview of results from different surveys over time. Among children aged 0-5 years, the incidence of stunting-which reflects chronic or long-term malnutrition-shows an increasing trend, from almost 43 percent in 1977 (Micro-nutrient Survey) to 50 percent in 1990 (Pakistan Demographic and Health Survey) and 60 percent in '*The 1998-99(Pakistan Socio-Economic Survey). Pakistan Socio-Economic Survey (SES) for 1998-99 shows a higher incidence of malnutrition among boys than among girls (Qureshi, Nazli and Sumro, 2001). Most research on gender disparities in nutritional status based on anthropometric measures tends not to find any statistically significant gender differences. For example, preliminary results from the PRHS (2001) on the incidenceof malnutrition do not show any differences by gender (World Bank 2002). Haddad (1999) outlines three possible reasons for the failure to observe gender differences. First, excessive female child mortality causes the most malnourished girls to drop out of the sample. Second, anthropometric standards are age sensitive, and a high incidence of age 6 prevalent among females than males in each Table 1.2: Malnutrition and Anemia age category (Table 1.2b). Particularly in the 15-24 and 25-44 age groups, there is a a. Malnutrition Among Children Under 5 (percent) clear pattern o f anemia among women. This 1990-94 1998-99 2001-02 high prevalence in childbearing ages is of Rural and Rural and Rural particular concern, since anemia is one Urban Urban cause o f low-birth-weight babies. Male 39.8 44.3 48 Underweight 1.17. Insummary, healthindicators reveal Female 40.5 32 48 a range o f female disadvantage. When the Male 36 64.9 64 various indicators are pieced together to Stunted Female 36.6 53.9 65 explain the existence of excess female Male 13.9 10.6 11 mortality in Pakistan, it appears that gender Wasted Female 13.7 8.1 12 differences in access to preventive and curative medical care could be responsible b. Prevalence of Anemia for the pattern o f gender gaps in health (percent) outcomes observed. Further data analysis is Urban Rural presented inChapter 4. Male 32.5 41.5 5-14 Female 40 42.7 Education Male 15.3 24.6 1.18. The female literacy rate is very low 15-24 Female 33.1 37.5 in Pakistan, since less than one-third (29 Male 8.7 19.5 percent) of adult women (aged 15 and older) 25-44 Female 37.1 37.3 are literate.29The male literacy rate i s higher at 57 percent. Adult women's low literacy Note:Malnutrition levels primarily reflect extremely low Underweight: Low weight for age (2 standard deviationsbelow medianweight for age o freferencepopulation) attainments in female schooling among the Shmted:Low height for age (2 standarddeviations below medianheight for age country's older generations. While both of referencepopulation) male and female literacy rates increased Wasted: Low weight for height (2 standarddeviations below medianweight for heightof referencepopulation) throughout the 1990s, the gender gap in Anemia: low hemoglobincontent) literacy rates did not diminish. Source: 1990-94: National Health Survey of Pakistan (1996), reported in Compendiumof Gender Statistics PakistanSocio-Economic Survey (1998-99). Malnutrition figures for 1998-99 ffom Pakistan Socio-Economic Survey, 1.19. Mirroring the literacy gap are Malnutritionfigures for 2001 from PakistanRural HouseholdSurvey, 2001. gender gaps in school enrollments among children-of school age: girls' enrollment rates are lower than boys' enrollment rates. The primary Gross Enrollment Rate (GER) for girls i s 61 percent. If this rate is adjusted for whether enrollment is age appropriate, then only about 46 percent o f primary school-aged girls are enrolled in primary school. Called the Net Enrollment Rate (NER), this rate is 46 percent for girls and is lower than the GER, suggesting delayed school entry and grade repetition among girls. These patterns are common among boys as well, for whom the GER is 82 percent and, adjusting for age, the NER is 58 percent. The enrollment in grades higher than primary are low for both boys and girls; however, here too there is a gender gap: the net enrollment rate in post-primary grades (grade 6 and beyond) is only 27 percent for girls and about 38 percent for boys.30 misreporting for females could result in a large proportion o f females appearing to be less malnourished than they really are. Third, the anthropometric standards are not gender neutral, since standards differ for females and males. 29Literacy refers to the capability to read and write. 30Netenrollment rates calculated for children aged 11-16comes from World Bank (2003). 7 1.20. The gender gap in primary school Figure 1.4: Primary School GrossEnrollment Rate, enrollments did not narrow appreciably during 1991-2001 the 1990s (Figure 1.4). For most of the decade, the primary GER for girls remained approximately 20 percentage points below that loo1 ~ for boys, except during the start of the 1990s, when the gender gap was slightly higher at 25 percentage points. A noteworthy and worrying trend is that in periods when the gender gap has narrowed, the decreasehasbeen due to a decline 40 I in male enrollment rather than a rise in female 1991 1995-96 1998-99 2001-02 , enrollment. A similar trend is also observed if --tMale +Female NERs are comparedovertime.31 I Source: PIHS roundsfor various years reported nPoverty 1.21. These trends in average net enrollments Assessment, 2002. rates mask the marginal gr&h in enrollments during the 1990s. A growth incidence analysis of enrollments between 1991 and 2001 suggests that changes in primary NER over the last decade were concentrated among the richer groups, who already had high enrollment rates in 1991 (see Chapter 3). There also were interestingvariationsby gender andregion.Inurbanareas, bothboys and girlsbelonging to higher income groups registered growth in enrollments, but rural areas saw a distinct difference between enrollmentgrowthfor boys and girls. While only ruralboys belongingto higher income groups registeredgrowth in enrollments, ruralgirls belongingto bothlower- and upper-incomegroups registered growth. 1.22. The trends andpatterns inprimaryschoolingthroughout the decade and into 2001-02 suggestthat muchwork is neededto meetthe gender equity goals set in the PRSP-namely to reduce the gender gap in primary school enrollment by 2005 (see Figure 1.2 in Box 1.1). A detailed analysis of constraints to girls' schooling is presentedin Chapter 3, Participation in the Labor Force 1.23. As in most developing countries, measuring the extent of female labor force participation is sensitive to the definition of work used and the duration (a week, month, or year) considered. Measurement of male participationin the labor force tends to be less affected by these issues.According to the PakistanIntegratedHouseholdSurvey, which measureswork participationover a reference period that is longer than that used by the Labor Force Survey, 67 percent of males and 25 percent of females were participating in the labor force in 2001-02.32This definition of labor force participation includes both paid and unpaid (family labor) work. Women who participate are much more likely to do so in unpaid work, while men are more likely to participate in paid work. Almost 60 percent of women involvedin the labor force are unpaidworkers. This is a very highrate compared to that of men; among those who participateinthe laborforce, only 19percent of men are unpaid family workers. 3 1 '*Worldforce Bank (2002). Labor participation is measured for those aged 10 or older. Incontrast, most other countries in South Asia calculate labor force participation rates for individuals 15 and older. 8 1.24. The aggregate participation rate of 25 Figure 1.5: Female Labor Force Participation Rates percent for females conceals substantial variation by Age and Region by age and across regions. Figure 1.5 depicts age- specific participation rates for females aged 15- 49. The age-specific participation rates also vary between rural and urban areas, for rural women participate more heavily inthe labor force than do urban women. Older women participate more in the labor force than younger women. Because the average age at marriage is about 22 for females, the age pattern of participation suggests that most 10 women in the labor force are married. This trend 15-19 20-24 25-29 30-34 35-39 40-44 45-49 i s not surprising, given that in Pakistan marriage I i s nearly universal for both men and women over Age the age of 20. In addition, married women are +Urban Rural expected to contribute to their husbands' households, which potentially could explain why Source: World Bank staff calculations usingdata from the PIHS 2001- 1-2 participationrates rise with age. .. 1.25. Female labor participation rates also exhibit considerable variation across provinces Figure 1.6: Labor Force Participation Rates by (Figure 1.6). Women in rural Punjab have the Province highest participation rates, while women in Balochistan have some o f the lowest participationrates in both rural and urban areas. Sindh 70 1.26. In addition to participating in work activities, whether paid or unpaid, it is important NWFP to look at female ownership o f productive assets 63 to better understand women's economic roles. Balochistan Female ownership of an important rural asset, All prownces land, appears to be limited in Pakistan. Data on ownership of assets such as land or access to All projnces mral credit by gender are not readily available from existing data sources. The Pakistan Rural All prownces urban Household Survey (PRHS) in2001has found that women owned only 2.8 percent of plots, despite the fact that 67 percent Of reported that women maintained the right to k m e : World Bank staff calculationsusing PIHS 2001-02 data for individuals aged and older, inheritland. A 1994 survey inruralPunjab found that of the 1,000 households surveyed, only 36 women owned land intheir own names.33 PoliticalParticipation 1.27. Recent legislation mandating reservation of seats for women in local governments, as well as in provincial and national assemblies, has substantially increased women's political representation. As a result of adopting the 33-percent quota mandated by the Local Government Ordinance (200 l), women competed not only for the reserved and open seats on the union, subdistrict (tehsil) and district councils, but also for the posts of Nazims and Naib Nazims. Out of a total of 40,009 seats reserved for women, 33Kazi(1999). 9 36,187 women were elected to various government bodies during the local elections o f 2001 (Islam 2002). A similar reservation/quota system for women's representation exists in the Senate and in the national and provincial assemblies; about 17 percent o f seats in the national assembly are reserved for women, and 18 percent o f seats are reserved for women in provincial assemblies distributed across the four provinces 1.28. The unprecedented number o f women elected to these government bodies following the quota adoption has opened up not only an enormous political space, but also a strategic opportunity for women to set and implement local government agendas. Despite seat reservation, however, political participation problems remain, as several factors continue to constrain women's effective involvement in politics at the local, provincial, and national levels. One such constraint i s related to information: women are significantly less informed than men about political matters, likely because o f their relatively low access to political information. As discussed in Chapter 5, women's political knowledge and involvement also may be inhibited by restricted mobility. 1.29. Although the first tenure o f local government has been characterized by lack o f proper resource allocation to local councils, which has impeded local governments' effectiveness, Pakistan can expect to see strong results from carving out a political space for women in the coming years. Studies from other countries such as India have shown that, over time, communities benefit significantly when women participate in local g ~ v e r n m e n t . ~ ~ 11. GENDER ASSESSMENT METHODOLOGY 1.30. The Gender Assessment has two basic objectives. First, it draws attention to the emerging gender issues in Pakistan. There already exists a large body o f work on Pakistan that seeks to comprehend the underpinnings o f gender inequality. The Gender Assessment synthesizes this work and extends it further with a view to filling knowledge gaps and recommend suitable policy options. The analytical work reported here involved consultation with relevant government institutions and research on a number of important issues-most notably those relating to customary practices and legal matters-by Pakistani experts. This process has yielded a set of policy recommendations that enable Pakistan to move closer to its stated policy objectives on gender. Second, the Gender Assessment aims to expand the Bank's understanding o f gender gaps in Pakistan and to provide a framework for enhancing the effectiveness o f the Bank's efforts to encourage gender mainstreaming in the country. The Analytical Framework 1.31. The Gender Assessment framework centers on the family and the economic, social, cultural, legal and political constraints that families face. This multi-layered structure i s critical to how factors both within and outside the family's control influence gender inequalities (Figure 1.7). Many o f life's most basic decisions are made within the household: families reinforce gender roles, transmit gender norms from one generation to the next, and determine the opportunities available to family members based on their gender. These decisions can magnify or reduce gender gaps. 34Chattopadhyay and Duflo (2001). 10 Figure 1.7: Framework of Analysis 1.32. The conceptual framework recognizes the importance o f factors both inside and outside the family in determining how gender disparities in outcomes arise and are perpetuated. While household socioeconomic status affects gender disparities, gender gaps in outcomes are muted somewhat in higher income households but they persist even in wealthier parts o f Pakistan and in wealthier households. This i s because gender disparities are intimately related to and driven by customs, social norms, and formal structures such as laws and regulations. The gender structures embedded in social and legal institutions affect gender relations and gender outcomes, and thus the scope for policy and action. For example, practices such aspurdah (see Box 1.2 for details) or customs that limit female mobility have an impact on women's access to schools, medical advice and even income-earning opportunities. Male household members, on the other hand, face fewer restrictions on their movement outside the household. These cultural institutions thus establish the incentives, opportunities, and constraints that determine peoples' choices and actions, and they shape power relations within the family and society. 1.33. There are great benefits to using such a framework. By placing the household-and specifically decision-making within the family-at the heart o f the framework, the Gender Assessment can "unpack" the process o f household decision-making for the policymaker. This enables policymakers to select programs and program designs available from national and international experience that are most appropriate for Pakistan, given its current conditions and circumstances, as well as its goals for the future. This framework allows the policymakerto better contemplate and address the following issues: 0 Appropriate policy/program design: Economic incentives alone, such as stipends for girls' education, are not likely to reduce gender gaps. What are the important non-economic factors that program design must incorporate? 0 Appropriate targeting of recipients: Policy will need to specifically target girls and women to directly reduce gender inequities in access to resources and opportunities. Given that prevailing non-economic conditions can make such targeted policy intervention ineffective, how can policy ensure that the targeted delivery of resources translates into equitable access to resources? 0 Appropriate policies to "level theplayingfiela'": Policies such as quotas or reservations may not be the best means of reducing gender gaps. Here, lessons are forthcoming from the experience o f the 33-percent reservation o f seats for women in local government. Even with the introduction of such legislation, seats went vacant in communities where women's mobility and public presence were most restricted. Should the reservation policy be applied in labor force participation to draw women into formal labor markets? The analysis and discussion in subsequent chapters will address such questions and offer recommendations for related policy measures. 11 Box 1.2: Customs that InfluenceWomen's Freedom of Movement Outsidethe Household Norms restricting women's mobility are closely linked to issues ofpurdah and izzat (honor). As reviewed in Mumtaz and Salway (2005), the ideal form of restricted mobility is seclusion of women inside the household. While this is an extreme practice, a more common form is requiring women to seek permission to leave the household and to be accompanied, preferably by a male household member, or at minimum a group of other women or children. Typically understoodto be an Islamic injunction, mobility restrictions are observedinmany regions and religions, as documentedby Mandelbaum(1986) and Jejeebhoy and Sathar (2001). The practice is inextricably linkedto issues of honor inthe society and communitiesobserving it. "Protecting" a woman inthis way safeguardsher honor and, more importantly, preservesthe honor of her family and community. The relationship between the practice of purdah and mobility i s a highly complex. Purdah requires women to thoroughly cover themselves with clothing, although the strictness of practice fluctuates widely throughout Pakistanand across SouthAsia. The most austere form uses the burqa, ablack dress combined with aheadcover, which completely veils the face and body. A less strict version of the dress code requires a scarf to cover the headinpublic. Inaddition to defining customs of dress, purdah has the potential to constrain women's mobility, such as inrural areas, where it can interfere with a woman's ability to perform manual labor, In some cases, however,purdah can even enhance female mobility. According to one woman, age 45, innorthern Punjab: Ihadnomobilityproblems withpurdah....Itensuresrespect. Ifa womanwithpurdahgoesout,she will be respected. Ifa school girl has to go alone and she is wearing a burqa, she will be respected more. Survey data from rural areas of Punjab and Sindh, Table 1.3: Practice of Purdah Among Rural gathered in the recent round of Pakistan Rural Household Women: Percent Practicing Survey (2004), show large variation in the observance of purdah across districts (Table 1.3). The sharpest No Yes differences are between northern Punjab, and southern N.Punjab 48.9 51.1 Punjab and Sindh. While overall observance is higher in S. Punjab (total) 18.5 81.5 Sindh and southern Punjab, the severity of purdah Sindh(tota1) 16.7 83.3 conditional on observance is greater in northern Punjab; TOTAL 24.3 75.7 Le., the proportion of observers with fill body/face Source: Cross-tabulationsfrom PakistanRuralHousehold covering is substantially higher. This is no longer true, Survey (2004). Sources ofData and Information 1.34. Existing survey data has provided much information about gender gaps in outcomes, but more information is necessary to unpack the underlying factors that inhibit efforts to reduce these gaps, Because so many factors originate from socio-cultural contexts, understanding them requires more nuanced and detailed data sources than pre-existing data can accommodate, which has required additional qualitative and quantitative data acquisition and analysis. The analysis also has benefited substantially from insights inpapers by Pakistani specialists on topics rangingfrom gender issues inthe water sector to gender issues in political participation, family law, and access to justice. 1.35. The method o f combining qualitative and quantitative information to learn more about potential implementable policy design was applied in the areas of education and health. Development studies are increasingly adopting this CLquaI-quant''approach as an analytical tool.35 The qual-quant approach i s 35This approach has been adopted in research on poverty where qualitative participatory poverty analysis is combined with data from householdsurveys. See proceedings from a workshop on qualitative and quantitative poverty appraisal, (Cornell University 2001) and Kanbur's discussiontherein. 12 particularly useful in gender analysis because data on social and cultural characteristics-for instance, those reflected in marriage customs-are difficult to measure from quantitative household surveys and to some degree are region specific. Qualitative data on marriage practices have helped improve our understanding o f them, including details on how property rights are ascribed and enforced under different region-based norms. 1.36. Primary household survey data. The Gender Assessment combines several rounds o f the Pakistan Integrated Household Survey (PIHS) and the Pakistan Rural Household Survey (2001) or PRHS- IwiththesecondroundofPakistanRuralHouseholdSurvey(2004)(PRHS-11).PRHS-I1isapaneldata set that follows about 2,000 households from PRHS-I. This second round expanded the community, facility, and household questionnaire to capture, in greater detail, constraints on women's access to schooling, health facilities, credit, markets, institutions, and ownership patterns regarding land and other assets. The PRHS-I1 also provides quantitative data on female mobility and gendered patterns o f inheritance, customary practices related to women, and the incidence o f domestic violence. The second round o f the PRHS has provided a rich information base for analysis, and for identifying the critical constraints that impede progress inthe outcomes summarized inprevious sections. 1.37. Qualitative study on gender. The analysis in the Gender Assessment integrates quantitative information obtained from PIHS and PRHS with qualitative insightsfrom a study carried out in five rural sites in Sindh and Punjab (see Box 1.3 for a description o f the data). The qualitative research also has enabled the assessment o f how women can be effectively integrated into development initiatives that emphasize community involvement and participation-in a context where there are likely to be numerous constraints on female mobility and decision making. 1.38. To offset the shortcomings o f qualitative studies based on a set o f case studies on specific villages or communities, which weakens the credibility and general applicability o f inferences drawn from them, the qualitative study selected sites from a sampling frame for a representative rural household survey (see Box 1.3). In so doing, the qualitative study insights based on a few households in a community can be combined with quantitative data on all the households in that community. The qualitative study also brought to light questions that could be included inthe quantitative household surveys seeking to analyze gender dimensions. The qualitative study was intended to shed light on key gender issues that are difficult to tease out o f quantitative data such as women's participation in community mobilization activities, the interaction o f social and customary practices as reflected in notions o f female honor, restricted mobility, the practice o f purdah, public and private violence against women, inheritance and marriage practices. The study also sought to understand constraints on girls' schooling and women's participation inpolitical decision-making. Policymakers need such insights to understand how programs can be better designed and how women can be better engaged in community decision-making since the devolution reforms and increased political representation o f women in local government. 1.39. Papers fiom Pakistani experts. In a series o f background papers, Pakistani experts have facilitated understanding o f gender inequality with regards to customary practices and legal issues, access to water, and capacity-building o f women representatives. The Gender Assessment derives important lessons from these papers, including the positive and negative effects o f customary practices and formal laws on women. Women in Pakistan are vulnerable, since avenues for obtainingjustice are very limited. In many areas, including family law, women have limited legal protection. Even where laws exist, women's access to the legal system is severely limited by cultural norms that prohibit women's access to public spaces and a general ignorance o f rights and procedures in the justice system. The legal system i s not supportive o f females seeking redress for violations o f their rights.Highcosts and delays in obtaining justice further discourage women from trying to avail themselves o f legal means to protect their rights. For victims o f violence or those fleeing honor killing (the practice o f killing a female family member perceived to have tarnished the family's honor), few legal remedies are available. There also i s a lack o f 13 safe houses for these women and reliable mediation mechanisms. A number o f civil society organizations as well as the government have taken steps to alleviate women's unequal access to justice. Given the magnitude o f the problem, it i s important to find ways to provide legal aid to women. Box 1.3: Qualitative Study on Gender The qualitative survey on gender was undertakenin five rural sites in the provinces of Punjab and Sindh. These sites coveredfive districts, with one site in each of the following districts: Lodhran (southern Punjab), Faisalabad (central Punjab), Chakwal (northern Punjab), Badin (Sindh) and Mirpur Khas (Sindh).Interviews were conducted inthemainvillage andinonesettlement closeto thevillage at eachsite. The study selected sites where both a Rural Support Program(RSP) and a nongovernmentalorganization (NGO) engaged in community mobilization of men and women were active. RSPs invest in the social mobilization of women and men and have formed women's organizations across 80 districts in Pakistan. The site selection enabled the study to assess constraints and benefits associated with women's membership in community organizations set up by the RSPs. Methodology The interview team selected three types of female respondents for the study: married women, Union Councilors representing the Union Council to which the selected site belonged, and the Social Organizer responsible for the RSP's community organization at the site (an employee of the RSP). Inall, 60 women from households, 5 female councilors, and 5 female Social Organizers (SO) were interviewed betweenMay and August 2004. A village-level census already had been conductedinthe five sites as part of a larger quantitative survey of rural households in Sindhand Punjab, enablingthe qualitative study to randomly select 12 married women betweenthe ages of 20 and 45 in each site (6 women were members of community organizations and 6 were not members). The selection process also ensured that at each site, equal numbers of women were selected from the main (or central) village and the settlement. Following the site selection, the team identified and contacted a female councilor representingthe Union Council for that site, as well as the female Social Organizer responsible for that site's community organization, interviewing one of eachper site. Almost all interviews were taped unless respondentswere uncomfortable with taping. Interviewers ensured that respondents interviewed separately and that the interview was conducted in the local language. Interview transcripts subsequentlywere translated into Urdu and English, and then coded with the qualitative data analysis software, NU*DIST, for further analysis. The interviews combined a semi-structured format with open-ended questions, allowing respondents to discuss their views relatedto the interview topics Variation across sites The presence ofthe RSP- that offers menand women the opportunity to mobilize into community organizations- was common to all sites. The five sites also cover a range of regional differences that allow the study to compare and contrast findings from each site. The sites represent a range of agro-climates that affect livelihoods, particularly the nature of women's participation in work activities. For instance, Chakwal in northern Punjab is part of the rainfed (barani) areas where the population relies mainly on rainwater for cultivation, causing agricultural output to vary considerably throughout the year. As a result, men in Chakwal tend to seek employment outside the farm sector, while women have taken over the management of the family farm. In contrastto Chakwal are Lodhran and the Sindh sites that are part of the canal-irrigated areas. Lodhran belongs to the heart of the cotton-growing belt, where agriculture is more market-oriented and cotton-picking offers paid work opportunities for women. Inthe Sindh sites of Badin and MirpurKhas wheat is the major crop followed by cotton. The Punjab sites (main village and settlement) are larger (in terms of number of households) than the Sindh sites. Land ownership rates are low, but range from a high of 52 percent of households in Chakwal to only 15 percentof households inMirpur Khas. 14 1.40. The background papers also address the role o f capacity-building for women leaders as the country prepares for the next round o f local government elections (slated for 2005). The Local Government Ordinance 2001, which provided for 33-percent reservation o f seats for women in all local councils, has created an unprecedented opportunity for women to participate in the country's political process. In the first round o f elections, some 36,187 women were elected to these seats. Most o f the women elected in the 2000-01 elections were new to governmental decision-making and therefore had little knowledge o ftheir rights, roles, and responsibilities as councilors. Increasingthe presence o f women inthe political arena was a necessary first step toward political empowerment, but it is equally important to make women aware o f their roles and responsibilities regarding the local government system ifthey are to make a difference in governmental decision-making. Several government agencies and non- governmental organizations have taken action to build the skills and capacities o f elected councilors. Examples include the Women's Political Participation Project (W3P) o f the Ministry o f Women's Development, Social Welfare and Special Education, and the Citizens' Campaign for Women's Representation in Local Government convened by the Aurat Foundation. These efforts to train women councilors becomes increasingly necessary as the country moves toward the second round o f local government elections. It i s also important to assess these training programs and to distinguishaspects o f training that are working from those that are not. 111. ADDRESSINGKEYGENDERGAPS 1.41. The following chapters offer in-depth findings on gender gaps in the areas o f education, health, and in income-earning activities. Chapter 2 contextualizes the analysis o f constraints faced by women and girls with a discussion o f women's legal entitlements in family law (inheritance, marriage, and divorce). The chapter examines legalentitlements in relationto customary practices revealed by quantitative data in the PRHS-11. This is a useful starting point for the CGA, since gender disparities are shaped by customs, social norms and laws and regulations. This chapter sheds light on the many ways legal and customary institutions impinge upon women's status in the household and in society. It also examines the complex role o f customary practices in the lives o f Pakistani women. While many such practices clearly violate legal provisions and are detrimental to women's welfare, on occasion they compensate for the lacunae in laws or their unenforceability and in so doing provide protection to women. Explaining these legal lacunae and related customary practices provides a foundation for understanding constraints on women's access to opportunities in schooling, health care, labor force participation, and involvement in the public sphere discussed in subsequent chapters. The chapter also discusses policy recommendations made by Pakistani experts on gender issues in areas such as legal provisions, access to justice, and political participation. 1.42. Chapter 3 investigates the gender gap in school enrollment, and identifies constraints that impede both school attendance for girls and female teacher availability. The chapter argues for complementary strategies that can augment ongoing interventions and programs, and address both supply and demand issues. The supply-side strategy emphasizes school proximity and ways to augment the availability of female teachers. The demand-side strategy considers initiatives to improve girls' ability to access schools through non-financial incentives along the lines o f programs currently being implemented in Punjab and Sindh. 1.43, Chapter 4 explores ways to enhance the impact o f policies and programs for improving women's and girls' health. The chapter investigates the determinants o f female health, including the proximity to facilities, outreach programs such as the Lady Health Worker program, and access to health-related information. The effect o f recent policies and programs are examined in depth and further 15 recommendations are made that will likely reduce impediments to women and girls' access to health services. 1.44. Chapter 5 discusses constraints to women's participation in activities in the public sphere. The chapter focuses on factors that shape women's participation in income-earning activities. An understanding o f these factors then frames the discussion o f how women's participation incommunity life can be enhanced. This chapter also analyzes the links between women's participation in income-earning activities, their autonomy, and their visibility in the public sphere that enhances their voice in the community and inpolitical decisionmaking. 1.45. Each o f the chapters highlights policy recommendations that can be implemented in the near term, with the objective of effectively narrowing gender gaps. The report stresses the importance o f learning from existing interventions through rigorous evaluations before scaling up the intervention or introducing new ones. Evaluations can provide information to policymakers to judge which projects should be expanded and guide the scaling-up process. Evaluations also inform policymakers about which aspects o f program design are effective, and which are superfluous. Indeed, each o f the policy recommendations in the areas o f education and health suggest small-scale pilot trials to assess the impact of the interventions. This experience will help assess the applicability of the recommendations on a larger scale. 16 2. FAMILY LAW CUSTOMINPAKISTAN AND The Gender Gap between Policy and Practice TheStateshallprotect the marriage, thefamily, the mother and child. -Article 35 o f the Constitutiono f Pakistan 2.1 To understand the condition o f women in a society, it is important to examine their status both within the domain o f the family as well as within the larger cultural and sociopolitical context, which structures their opportunities and defines their capacity for action. 2.2 This chapter examines ways in which the rights of women are articulated in law and the manner in which customary practices interact with such legal provisions. A major focus o f the chapter is on family law, since familial attachments and networks define many aspects o f individual status and rights in much o f Pakistan and the interaction between custom and law within the domain o f the family essentially defines the de facto set o f opportunities available to women as well as the barriers they confront in fulfilling even basic needs such as education and health. An important message o f this chapter is that customary practices inPakistan play a complex and not entirely detrimental role inthe lives o f women, so understanding these institutions and their interaction with the law i s an essential first step to improving the status o f women inPakistan. 2.2. In the context of family law, the chapter examines legal rights around marriage and the intergenerational transmission o f wealth to women. Even when laws are modified and passed by government, perceptions about these legal arrangements and their enforceability are likely to significantly impact the welfare o f women and can tell us much about their options and choices in other spheres. It is also precisely in these areas that customary practices exert tremendous force. Many practices are flagrant violations o f state as well as official Islamic law and are clearly detrimental to women's welfare. However, some cultural practices protect women in an environment where legal protections are either absent or unenforceable. This is well known in many other contexts where traditional institutions "step into the breach" left vacant by absent or unenforceable legal protections. We examine practices that appear to play this role in rural Pakistan. This i s not to suggest that such customary practices yield optimal outcomes for women, but that given the lacunae in written laws and enforcement capacity some customary practices may actually enhance women's welfare. 2.3. The chapter is organized as follows. Section Ireviews women's legal entitlements in family property and examines patterns o f inheritance in the quantitative data and reviews alternative explanations for the patterns observed. Section I1 describes important customs in Pakistan's marital practices, providing an overview o f women's rights in marriage and divorce. Both sections also review relevant developments in family law. Section I11 presents recommendations regarding legal reform and enforcement o f legal provisions that have been suggested by civil society organizations (CSOs) and legal scholars in Pakistan and contextualizes these recommendations interms o f the chapter's broader themes. 2.4. The chapter relies on three broad sources of information: for law, publications o f Pakistani experts, some o f whom have written background papers for this Assessment; new survey data36;and a qualitative study (described in Box 1.3) in five villages in rural Sindh and P~njab.~'The survey includes 36 The data used here are from the second round of the Pakistan Rural Household Survey (PRHS II), The survey covers 94 2004. villages in Punjab and Sindh (the sample is broadly representative of the provinces) and was designed by Hanan Jacoby and Ghazala Mansuri o f the World Bank's Development Research Group (DECRG). The Pakistan Institute of Development Economics, Islamabad collected the data, collaborating with DECRG. The chapter uses early results from Jacoby and Mansuri's ongoing work. 37 The qualitative study was done as a collaborative piece of work with another ongoing study, an evaluation of a Community Driven Development Project inRural Pakistan. detailed modules on both marriage and inheritance customs, providing a novel view o f cultural practices from a large, representative survey. Survey data on such topics4specially that with any claim to representativeness-have rarely, if ever, been collected in Pakistan. For the purposes o f this chapter, the sample is divided into three regions: northern Punjab (districts Attock, Faisalabad, and Hafizabad), southern Punjab (districts Bahawalpur, Muzaffargarh, and Vehari), and Sindh (districts Badin, Larkana, Mirpurkhas, and Nawabshah). Southern Punjab is generally viewed as culturally and economically closer to Sindh, northern Punjab being on average richer and more developed. As will become apparent, there are strikingcontrasts in customary practices across these regions. 2.5. Legal scholars in Pakistan have frequently noted the lacunae in family law as it pertains to the rights o f females and their position in the family.38 These lacunae, often mistakenly attributed to the tenets o f Islam, result in part from South Asia's history o f neglecting to codify official personal law (Islamic and otherwise), a history with origins in British colonial rule as well as regional variations in customary practice. These scholars note that comprehensive legal reform has repeatedly and intentionally overlooked family law, allowing the entrenchment o f ancient, region-specific tribal practices to avoid offending local interests. This precedent was set in 1772, when the Warren Hastings Plan established a British-style hierarchy o f civil and criminal courts across South Asia, until then governed by Muslim law. In matters regarding family law (such as marriage and inheritance), however, the Hastings Plan adopted indigenous customary norms, an approach that scholars agree was designed to minimize the risk o f local rebellion against Britishrule. 2.6. The Britishreapplied this approach to family law in the 19thcentury, simply compiling the local customs o f South Asia's various regions rather than uniformly codifying family law for the entire subcontinent. Traditional practice was to hold sway in each region, effectively reinforcing the gap between local custom and official law regarding the rights o f women and girls. Even courts were barred from applying formal statutes unless local custom lacked any rule in a particular case or situation. Some o f the time, customary practices prevailed in ways that were deleterious to females, who tended to be defined solely by their family role in a patriarchal society-Le., by their need to be protected and to remain in the private sphere. Though this gap has narrowed since partition (in 1947) with new legislation since then , the discrepancy between family law and customary practices has persisted to some degree, most obviously inthe areas o f inheritance and marriage law. In spite o f legislation designed to codify and clarify females' liberties in Pakistan, cultural norms and religious beliefs routinely override statutory laws, often interfering with women's rights to family inheritance and to protection in marriage and divorce. According to Dr. Shaheen Sardar Ali, "Family law in Pakistan is a mixture o f codified law and customary practices based on religious norms and administered in a secular, procedural framework o f a modern day dispute resolution forum-the judiciary....Culturalnorms and religious rules are just as potent a force, ifnot more, as legislative enactment^."^^ 38Ali (2000); Awan (2005); Mehdi(2002); Razvi (2004). 39Ali (2000: 139). Dr. Ali i s former Minister for Health, Population, Welfare and Women's Development for the Governmentof Pakistan and Former Chairperson of Pakistan's National Commission on the Status of Women. She currently is Professorof Law at the University o f Wanvick, inthe UnitedKingdom. 18 I. INHERITANCE Women's LegalEntitlements to Inheritance 2.7. Women in Pakistan have the legal right to inherit family wealth, yet they rarely exercise this right, Rubya Mehdi, scholar o f gender and property law in Pakistan, points out that although Islamic (Shariah) law and Pakistani state law both entitle women to inherit immovable and movable property (see Annex l), under colonialism and independence alike, the rule in practice has been to deny women's control over their inheritance-of land in particular-and often their entire claim to it.40This i s especially true in rural Pakistan, where the tribal nature o f social organization undermines female inheritance rights. Rather than emphasizing the Islamic concept o f immediate family, inheritance practices emphasize the importance o f keeping property within the larger (tribal) family, which is always headed by men. Another explanation for the low incidence o f female land inheritance is families' routine equation o f dowry (money or property brought by a bride to her husband at marriage) with a share o f inheritance, though this is not legal and has been decried by government officials, activists, and civil society organizations in Paki~tan.~'Because few estates in rural Pakistan include any property other than land, and virtually none consist solely o f non-land assets, only the intergenerational transmission o f land i s considered inthe following review o f inheritance law and subsequent discussion o f analysis results. 2.8. To grasp the complexity o f inheritance practices involving Pakistani women, one must be aware of the legally pluralistic nature o f family law in Pakistan, as well as distinctions between Islamic (Shariah) law, customary law, and state law. Non-Islamic law will not be discussed in this chapter, as laws specifically targeting Hindus, Christians, and other non-Muslim minorities inPakistan typically have fallen outside the scope o f family law since independen~e.~~State law that has codified inheritance rules is based on Shariah law. Due to the powerful force o f customary practice in the country, however, the inheritance rules most often followed are those based on custom, which can differ substantially by region and include those practiced by Muslims and non-Muslims. Because they typically supersede codified state law, these rules are accepted as `customary law' on inheritance, according to Pakistani scholars of gender and Islamic law. Customary law tends to give much less recognitionto women's rightsthan does state law, which in some cases tends to be less generous to women than does formal interpretation o f Shariah law.43 2.9. Based on Islamic law, state law stipulates the share o f women's inheritance to be one-half that o f men in similar relationships to them (e.g., a daughter would inherit one share for every two shares that a son inherits), due to the man's greater responsibility for supporting the family. Shariah law also has distinct provisions for inheritance based on the inheritor's relationship to the deceased: children who inherit along with parents are to receive a greater share than their parents, as a greater share o f the child's life lies in the future than does hisher parent's; widows who have children or filial grandchildren are to receive one-quarter of the inheritance, while widows without any such descendants are to receive one- eighth; and a daughter is to receive one-half her father's property if she has no sisters, while two or more daughters are to receive two-thirds o f all heritable property between them. 40Mehdi (2002: 25-41). See Annex 1for a discussionon changes instate law on Musliminheritance. 4' The Dowry and Bridal Gifis (Restriction) Act of 1976 was the Government of Pakistan's first attempt to constrain the practice of dowry, which is discussed further in Section 3 of this chapter. Please see Annex 2 for more information on civil society organization's efforts to restrict the practice of dowry. 42Non-Muslims have no representation on Pakistan's Commissionon Marriage and Family Laws, which thus tends not to make recommendations for modifying family law regarding non-Muslims, and dispute settlements typically rely on cultural and religious norms specific to the minority groupsimilar to the tradition of customary law practiced among Muslims, For more discussion on this topic, please see Government of Pakistan (1997: 20-21). 43Ali (2000); Mehdi (2002). 19 2.10. State and Shariah laws' clear designation o f female inheritance rights notwithstanding, Pakistani women rarely receive their shares in immovable property, especially in rural areas?4 This occurs is in spite of high courts' attempts to give special consideration to women's interests when hearing disputes over land inheritance, often with the objective o f relieving women from the sense o f obligation to relinquish inheritance to male family members. In several contemporary cases, courts also have emphasized the duty o f brothers to provide their unmarried sister with maintenance, whether she has never been married, widowed or divorced, in addition to any inheritance that is her due-not inexchange for it, as is the common practice.45 According to experts, however, these cases are the rare exceptions: "...because o f the barriers to women inheriting immovable property in traditionally patrilineal communities.. ..[Rlelatively a small number o f cases come to the courts and moreover.. .the superior courts decisions and attitudes are very different from that o f the lower courts, which are more flexible and compromising toward the customary normative orders."46 Results from QuantitativeAnalysis 2.11. Analysis o f the PRHS-I1 survey data corroborates findings from previous studies o f inheritance patterns in Pakistan: when women do inherit property, it typically i s controlled by male heirs due to general powers o f attorney, gift deeds, or voluntary relinquishment o f the property by the female to the male heirs.47 Land remains controlled by male members regardless o f family wealth. Although in principle the family is likely to recognize female members' right to inherit property, it i s rarely the case that the female inheritor retains-that is, inherits and keeps or sells-the property. Women's tendency to retain family property does not vary by family wealth, moreover. The analysis o f regional variation in inheritance patterns suggest that women in Punjab are three to four times more likely than women in Sindh to retain inherited land. 2.12. To ascertain general patterns o f landholding in Pakistan, PRHS-I1 asked all ever married women age 15 and older a set o f questions concerning inheritance from their father, mother, and/or husband, as applicable. Around 45 percent o f women whose fathers had already died reported that their fathers had land or other significant property at time o f death, whereas the deceased mothers o f only 5 percent o f women had left an estate. Given the paucity o f maternal inheritance, we focus here on transfers from the paternal side. Even so, the sample o f potential female heirs is rather selective, though more so in Sindh, with 36 percent o ffather's having heritable wealth, than in southern Punjab (45 percent) and the wealthier northern Punjab (58 percent). The median landholding amount o f fathers at death i s about eight acres in both Sindh and northern Punjab, but less than one-half this amount in southern Punjab. Since few estates inruralPakistan includeany property other than land and virtually none consist solely o f non-land assets, only the intergenerational transmission o f land i s considered inwhat follows, 44Mehdi(2002) http://~akistan.lead.ordmedia/report4jan04.htm. 45A detailed descriptionof the precedent-settingcase of GhulumAli v. Ghulam Sanvar Naqvi and related cases can be found in Mehdi(2002:34-40). 46Mehdi(2002:40). 47Ali (2000); Mehdi(2002) htt~://uakistan.lead.or~/media/reuort4ian04.htni. 20 2.13. Figure 2.1 divides potential I Figure2.1: Status of Paternal Inheritance among - heirs into three categories: (1) those who had a Potential Female Heirs no recognized inheritance rights; (2) those who had inheritance rights but relinquished them (in the vast majority of cases to a brother), or retained them but later returned the land to their natal family; and (3) those i who inherited and either kept the land or sold it. On one level, there is little conceptual difference between categories (1) and (2); a / woman may nominally have the right to I \ -/' inherit a share o f her father's land, but given her standing with regardto her brothers, she i s L I under inexorable pressure to cede her rights to them or to other natal relatives. Looking, Source 2004 Paklstan Rural Household Survey then, at category (3), there is a big difference I across regions. Inherited land was retained by 16 and 19 percent o f women in northern and southern Punjab, respectively, but by fewer than 5 percent o f women in Sindh. 2.14. Multivariate regression analysis shows that the likelihood of inheriting and retaining land does not depend on the size o f the father's estate.48 The same result holds if we consider the share of heritable land received by the woman (which equals zero in most cases). These findings indicate that inheritance practices in rural Pakistan are invariant to family wealth. 2.15. Even once a woman has secured possession o f land inherited from her father, the question remains of how much control she exercises over it. Among the 13 percent o f women who were bona fide heirs, 13 percent o f them subsequently sold their land. Of the remaining women, about 15 percent signed some form of mukhtianaama, a document granting a relative power-of-attorney over the land. Thus, o f the few women hanging on to land inherited from their father most have managed to maintain operational control over it. 2.16. PRHS-I1 includes information on the extent of widow inheritance, but the samples are small. Broadly speaking, the situation o f widows i s similar to that o f daughters; o f the widows whose husbands had estates, only about 16 percent inherited and retained their husbands' land. In most cases where inheritance rightswere relinquished, they were given to a son. Explanationsfor the Gap betweenPolicy and Practicein FemaleInheritance 2.17. Why is land, by far the most important heritable asset in rural Pakistan, generally not transferred to women at the time of the father's (or husband's) death? One explanation, o f course, i s discrimination, Parents favor sons over daughters for many reasons, most of them culturally based. The practice of favoring sons over daughters when bequeathing land is rooted in patriarchal tribal traditions, which inthe past did have a practical function; agricultural land was withheld from daughters to be "kept in the family," using the broader definition o f family to signify the tribe or caste.49 The high incidence of endogamous marriages-in which men and women from the same village are married, discussed later in this chapter-in contemporary Pakistan does not appear to daunt customary practices designed to keep 48 Multivariate regression analysis also shows that while women's education has no significant effect; older women are more likely to inherit and subsequentlyretain land. Older women may be more likely to inherit becausethe property more oftencomes up for inheritancefor women of higher ages than for youngerwomen,whose fathers are more likely to still be living. 49Ali (2000: 114); Mehdi (2002: 35). 21 property under tribal control, preventing the possibility that female family members take land with them whenthey marry andmust move into their in-laws' homes. Eventhough daughters are likely to remain in the village (and caste) when they marry, parents prefer sons over daughters because o f deeply-ingrained beliefs that a son provides resources to family households, whereas a daughter absconds with them. 2.18. Botticini and Siow (2003) suggest an alternative explanation based on efficiency rather than parental preferences. In virilocal (by paternal descent) agrarian societies, sons invariably take over the father's farm and, consequently, need to be given the right incentives to maintain and invest in the farm's assets. Ifdaughters were to share in agricultural land upon the father's death, then brothers would have lower incentives to work hard on the land until their father's death. Given that parents care about daughters' welfare, but prefer not to bequeath wealth to them in the form o f land for this reason, the daughters' inheritance generally will take the form o f dowry and will not include any land. Botticini and Siow present evidence (from medieval Italy) that, once dowry wealth is taken into account, daughters are not discriminated against in favor o f sons in the disposition of parental property. Whether the same applies inrural Pakistan is an issue addressed later inthis chapter. Insightsfrom the Qualitative Data 2.19. Interviews conducted in rural Pakistan (data i s described in Box 1.3) indicate some proximate causes for Pakistaniwomen not claiming their inheritance. A great majority o f the 60 women interviewed expressed concerns that claiming their inheritance would violate custom and incur animosity from the natal family. They also tended to lack information about the exact share o f heritable land to which the law entitles them." Even those who had this information almost invariably gave-or said they would g i v e t h e i r share over to male family members (typically brothers) either because they felt they were abiding by a worthwhile cultural tradition, or because they feared the natal family's (especially brothers') reprisal ifthey took control of the share entitled them. None o f the 60 women interviewed from the five sites in Sindh and Punjab (one of the three in Punjab, Lodhran, lies in southern Punjab) had claimed and retained the land that was their rightto inherit. Only four women (6.7 percent) said they would claim it if they could, but these women either were unable to inherit land (two o f the four were from families with no landholdings) or felt unable to actually obtain the property due to resistance from the natal family. 2.20. Table 2.1 displays percentages o f women who had knowledge about their inheritance rights- among all 60 women, and among women interviewed per region and per site. Although almost all (95 percent) o f the interview subjects were aware o f their right to inherit land, a minority knew the amount to which they were entitled. Of the 24 women interviewed in the two Sindh sites (in Badeen and Mirpur districts), 23 knew of their inheritance right. Only five (22 percent) o f these 23 had an idea o f what amount was their due; moreover, of the two women inBadeen who knew the amount both were under the impression that they were entitled to one-fourth o f the land, instead o f the one-third (technically speaking, one-half o f their brothers' share) assured them by law. 22 Table 2.1: RuralWomen's Knowledge and Perceptionsof Their Inheritance Entitlements by Region (percent) Aware of their Aware of what portion of the Willing to relinquish Region legal right to inheritanceto inherit family property they placate natal property were entitled to inherit family/brothers Total Northern Punjab 100 (24) 45.8 (11) 62.5 (15) 100 (24) Faisalabad 100 (12) 58.3 (7) 75 (9) 100 (12) Talagang 100 (12) 33.3 (4) 50 (6) 100 (12) Southern Punjab (Lodhran) 83.3 (10) 41.7 (5) 75 (9) 100 (12) Sindh 95.8 (23) 20.8 (5) 79.2 (19) 100(24) Badeen 91.7 (11) 16.7 (2) 92 (1 1) 100 (12) Mirpur 100 (12) 25 (3) 67 (8) 100(12) Average of All 95 (57) 35 (21) 71.7 (43) 100(60) Note: Statistics arepercent ofthose interviewed;frequenciesare inparentheses. 2.21. Among all 60 women, those in the two northern Punjab sites (in the Faisalabad and Talagang districts) tended to be most informed about their privileges. All 24 women interviewedfrom northern Punjab were aware of their right to inherit land, though only 11of these 24 knew how muchthey were entitle to inherit (and only four of the 11were from Talagang). InLodhrandistrict, southern Punjab, 10 of the 12 women interviewedknew of their inheritance privileges, but only five knew how much they were allowed to inherit. Interestingly,althoughno one interviewedin Lodhran had accepted or would accept their share-just like in the other four sites-two Lodhranwomen did mentionthe possibility of conditional acceptance, saying that women should accept their portion of land if their brothers are not taking proper care of them. Most women-nearly 72 percent-interviewed in all five sites, when reportingthat they would not consider takingtheir share, expressedwanting to maketheir brothers happy and voiced explicit fears that taking their inheritance would damage their relationshipswith their natal family members, namely their brothers and fathers. A large majority of women expressed these sentiments in every site, with the exception o f those in Talagang, where only one-half(though a large portion,but not a majority)reportedvoluntarily relinquishingtheir shares to brothers. Ithink,Ishouldnotclaimmy share.Iwillnottakeit.My brother isdear tome.Brothershelpinthehour of need...IfIwill claim my share, the warmth of our relationship will be aflected. If even brothersdo not dislike it, their wives and children would definitely feel that. They say that we have taken share3om them.Although, it is our right and we should take it but the relationship is affected. -Woman from northernPunjab (Faisalabad), Age 30 Ihavenever askedfor myshare,nor havetheyevergivenittome.Eveniftherecomesatoughtime,Iwill not askfor my share. rfI askfor my share, others will not say anything, but myfather and brother will become angry with me. -Woman from Sindh(Mirpur Khas), Age 24 The few women interviewedwho wanted to take their share were denied it and felt they hadno recourse in leveragingthe law that entitledthem to it. 23 Ihave askedmy brother to give me my share....But he says thatIam married and it is my husband's responsibility tofeed me and take care of me. Here most of the womenforego their share; they want to take it but they don'tget it. They can't do anything when they don't get it: they remain silent about it. -Woman from Sindh(Mirpur Khas), Age 22 Ifa daughter asksfor her share, she is not considereddecent and nice. My mother askedfor her share so my mamoon (maternal uncle) and hisfamily became very angry. They startedjighting and we separated fiom each other. -Woman from Sindh(MirpurKhas), Age 40 11. MARRIAGE 2.22. Despite landmark laws enacted in the twentieth century to protect women's rights in marriage5' legal scholars and human rights organizations inPakistan agree that girls and women continue to confront profound disadvantages in marriage and divorce.52 As with inheritance practices, cultural constraints on women's rights inmarriage and divorce are grounded not so much inIslamic law itself as inregion-based customs and/or idiosyncratic interpretations o f Islamic law, according to these scholars. Customary practices that discriminate against women have thus been allowed to distort or overlook Muslim and state law on marriage and divorce. As this section will show, however, not all institutions o f marriage currently practiced in Pakistan are necessarily harmfulto women, let alone illegal. 2.23. The following paragraphs address marital customs that are widespread in Pakistan, including child marriage, different marriage types such as first-cousin and watta satta (where two different families trade their daughters for marriage), creation o f the nikahnama (marriage contract), dowry, and endogamy, which has strong implications for women's decisionmaking within the household. Among these common customs only some have a tendency to impinge upon the rights o f females. Two objectives o f the analysis, below, are to (1) better understand which customs tend to disadvantage females; and (2) relay key recommendations-made by Pakistani scholars and civil society organizations-for remedying problems associated with these specific customs. Age at Marriage 2.24. Since 1961, contracting marriage for any female under age 16 has been illegal in Pakistan. The problems associated with child marriage are well known to human development experts and human rights advocates in Pakistan (see Annex 2.2). A minimum age o f 16 was established as part o f a set o f legal initiatives-collectively known as The Muslim Family Law Ordinance (MFLO) o f 1961-to improve the status o f women in Pakistan (please see Annex 2 for elaboration). The 16-year age minimumwas higher than the previous age minimum of 14 for females and 16 for males, set in 1929 under the Child Marriages Restraint Act (CRMA), before which there had been no minimum age for marriage, The MFLO also requires the consent of both male and female parties before the marriage can occur,, an attempt to offset traditional practices that promote early marrying-off o f daughters by natal families, Of course, the definition of "consent" remains open to interpretation. '*Please see Annex 2.2 for a list of lawscited in literatureon family law and women's rights inPakistan, Ali (2000); Mehdi (2002). 24 2.25. Data from the PRHS-I1 are Figure2.2: Age at Betrothal and Nikah ideally suited to analyze issues o f age at marriage and consent. Women S PUNJPB I N were asked both for the age at which they were betrothed and the age at which the marriage ceremony, the nikah, took place. The overall the average age at nikah o f 18 (median 17), and betrothal at age 14 (median 0 P 15) conceals considerable regional e P N variation. In Sindh (see Figure 2.2), the average interval between betrothal r Age at Betrothal and nikah i s nearly five years (median Age at Nkah three years), whereas in both northern 0 and southern Punjab the average 0 10 20 30 40 interval is about two years (median 1 year). By the time they reach age ten, nearly 40 percent o f girls in Sinhh have already been promised in marriage, while the comparative figure in Punjab is just over 10 percent and varies little between north and south. Moreover, a considerable proportion o f girls in Sindh (about 5 percent) appear to have been promised in marriage at or before birth. 2.26. Focusing strictly on prima facie violations o f the MFLO, one finds that 31 percent o f rural women were married before age 16 (10 percent in northern Punjab, 26 percent in southern Punjab, and 43 percent in Sindh). The proportion o f women who were promised in marriage before age 16, however, is much higher-overall 57 percent (32 percent in northern Punjab, 50 percent in southern Punjab, and 73 percent in Sindh). Evidently, the law does not explicitly address child betrothals, since the girl presumably still lives in her natal home while she awaits marriage. As long as the marriage i s ultimately "consensual," the law i s silent. 2.27. This brings us to the issue o f consent. Practically no woman chooses her own husband in rural Pakistan. More precisely, 97 percent o f women report that their parents or other members o f their extended family choose for them, although the woman's opinion was solicited in 12 percent o f these cases. Here again we see a sharp regional contrast, in that the percentage o f woman having had at least some input into the choice o f husband is 27 percent in northern Punjab, 12 percent in southern Punjab, and only 8 percent in Sindh. Women who have a say in the matter tend to be older at the time o f betrothal than those who have no say at all, with a median age o f 17 versus 14. Interestingly, the small minority o f women whose opinions were solicited at the time o f their marriage tended to approve o f their families' choice o f husband. There are practically no cases o f outright disapproval, and most cases o f lukewarm approval or no expressed preference occur when the husband is a blood relative (and thus presumably well-known to the bride in advance o f marriage). Even so, just because a woman approved o f her husband does not mean she necessarily wanted to get married. 2.28. More generally, consent i s a slippery concept. Minimally, it requires that the woman express "no objection" to the identity of her husband and perhaps the timing and conditions o f the marriage. In practice, however, there may be no alternative for a girl but to accept the marriage arranged by her parents; it is a "take-it-or-leave-it" offer in which leaving-it i s simply unthinkable as she has few other 25 options. Thus, while the state can, at least in principle, ensure that minimum age requirements for marriage are met, consent i s far more difficult, ifnot impossible, to verify and enforcees3 2.29. In sum, legal protections regarding girls' marriages appear to have limited scope and effectiveness: girls and sometimes their families are ignorant o f legal restrictions on child marriage; moreover, when their families do know o f these legal restrictions, they can afford to flout them because enforcement is absent. Prominent civil society organizations, such as the Aurat Foundation, have in some cases pursued enforcement. Since its inception in 1986, the Aurat Foundation's Legislative Watch Programme has redrafted several Muslim laws to make them more sensitive to gender issues. Reiterating several recommendations made by the Commission on Marriage and Family Laws, the Legislative Watch Programme has suggested additions and amendments that would make the ChildMarriage Restraints Act more effective.54 Some o f the more important suggestions include changing the definition o f "child" to anyone under age 18 and effectively raising girls' marriage age minimum to 18 years; and increasing punishments for adult males who marry children, as well as for guardians who knowingly allow their children to marry. Of course, the key issue will remain the enforcement o f these restrictions on the ground. Types of Marriage 2.30. As hasjust been noted, marriages inrural Pakistan are almost universally arranged by parents and other relatives with little input from the bride.55 Under these circumstances, the interests o f the family take center stage in the formation o f unions. Examining the types o f customary marriage that exist in rural Pakistan may reveal something o f the nature o fthese familial interests. 2.31. First o f all, PRHS-I1 data indicate that a I remarkable 78 percent o f marriages involve blood Figure 2.3: Relation betweenSpouses 1 I relatives, both paternal and maternal, and another N PUNJAB S PUNJAB 10 percent involve fellow zaaubiradiri (castehribe) members, leaving fewer than 12 percent o f married couples completely unrelated to each other. The patterns are broadly similar across regions, although the preference for paternal relations is notably lower in northern Punjab (Figure 2.3). 2.32. In the vast majority o f cases where the spouses are blood relatives (93 percent), the woman knew her husband for a least a year before marriage and usually for all o f her life. By I I 2004 PakistanFluralmusehoidSurvey contrast, 59 percent of women marrying an unrelated member o f her own zaat met her husband for the first time on their wedding day. This figure rises to 87 percent for women marrying unrelated men outside their zaat. The interval between betrothal and nikah also tends to be larger for marriages among relatives; in Sindh, this gap exceeds five years on average, compared to just over three years for nonblood s3In fact, there appears to be little enforcement of MFLO of any kind. Often family members simply will falsify a girl's age on her marriage certificate to avoid any questions from authorities, who typically will not attempt to ensure the veracity of the informationreported(Ali 2000). 54"Suggested AdditionsiAmendments in Muslim Family Laws: The Child MarriageRestraints Act, 1929 (XIX of 1929)'' Aurat Foundation(variousyears). 55What little input there could be in choice of spouse is constrained by restrictions on girls' mobility and, above all, on her associationwith membersof the oppositesex. Please see Annex 2 for details. 26 relatives. Marriages between relatives thus are arranged further in advance o f the wedding date than those between nonrelatives. 2.33. Marriage between paternal relatives i s said to be a means o f keeping heritable wealth, particularly land, in the paternal lineage. Yet, the mechanism needs clarification. As noted earlier, few women in rural Pakistan directly inherit any property from their father. While women do receive bequests in the form o f dowry, these are typically small amounts relative to the inheritance that their brothers receive, at least for landed households (see below). To be sure, marrying a paternal cousin keeps a woman's assets within her paternal lineage, but this is not a substantial motivation consideringthe sums involved. More salient, given that families indeed care about dividing wealth among their sons and daughters, i s that marrying a paternal cousin enables a woman (and her children) to benefit from her grandfather's estate without actually inheriting it herself; she benefits through her husband's inheritance. As already discussed, in a context where the bulk o f wealth is tied up in land and in which sons take over their fathers' farms, inheritance may be channeled through sons rather than daughters for efficiency reasons. 2.34. Considerations o f property devolution are not the sole, or even an important, explanation for consanguineous marriage in rural Pakistan. The rate o f marriage to paternal relatives among women with landless fathers-in-law i s also high. Indirect inheritance o f land by daughters cannot be a motivation in such cases. Marriage among maternalrelatives is quite prevalent as well. Inthe context o f India, Dyson and Moore (1983) suggest that cross-cousin marriage, in general, enhances the status o f women as compared to marriage into an unrelated household. Women may be better treated, for example, in their cousin's household. Married women in rural Pakistan, unlike their counterparts in much o f India, typically remain in reasonably close contact with their natal families, which may attenuate this particular advantage o f cross-cousin marriage. 2.35. More than one-half o f marriages recorded in the PRHS-I1 data are village endogamous, which is Figure 2.4: Types of Marriage to say that the husband and wife are from the same village. The rate o f village endogamy is 59 percent in Sindh, 53 percent in southern Punjab, and 45 percent in northern Punjab. Not surprisingly, a strong relationship exists between village endogamy --- and consanguinity. Nearly two-thirds (63 percent) o f marriages between relatives occur within the same village, compared to 31 percent o f intra-zaat marriages and 19 percent o f completely unrelated marriages. It is not clear, o f course, whether the desire for consanguinity (or inter-marriage within the zaat for that matter) leads to village endogamy, 2004 Paklslan Rural Houssholj Suwey village endogamy leads to consanguinity, or some thirdmotivation leads to both village endogamy and consanguinity. 2.36. Two unusual forms of marriage in rural Pakistan are, in order o f prominence, watta satta (adlo badlo in Sindhi) and two-sister marriage. Wattasatta is an exchange marriage whereby a woman and her brother (or some other male relative) from one family wed a sister and brother from another family, usually around the same time. The second form involves two sisters marrying two brothers (or other male relatives) from another household. Often (in Sindh at least) these two forms overlap (Figure 2.4). Remarkably, only 31 percent of Sindhi marriages take the "conventional" form, compared to 51 percent o f marriages insouthern Punjab and 68 percent innorthern Punjab. The majority o f wattasatta marriages occur between brother-sister pairs, though somewhat less so in Sindh (65 percent) than in southern and northern Punjab (73 and 80 percent, respectively). The second most popular watta satta arrangement, but 27 still much less prevalent than brother-sister (16 percent o f wattas overall), i s when at least one o f the counterpart couples are uncle and niece. Various other combinations occur as well, but none is very common. 2.37. Watta satta may arise out o f the strong preference for cross-cousin marriage, as just noted. If both brother-sister pairs involved share the same grandfather, then the watta effectively marries two sets of cousins. While the incidence o f watta satta is lowest among women unrelated by blood to their husbands, the rate is still, in an absolute sense, quite highat 25 percent. Wattasatta thus cannot solely be a mechanism for maximizingthe number o f marital links across related families. Wattasatta may also be motivated by a desire to limit dowry (or bari) expenses, which normally must be paid up front. The advantage o f marrying one's daughter into one's son's wife's family i s that there may be no need to exchange dowries at all. This would be desirable when parents are severely cash constrained. We examine this hypothesis empirically below. 2.38. Women's own views provide insights into the roles o f wattasatta and two-sister marriage. When asked whether they would favor their own daughters marrying in a wattasatta arrangement, 70 percent o f current watta satta brides responded strongly in the affirmative, whereas 13 percent said they would strongly disfavor it. By contrast, only 27 percent o f non-watta satta brides would strongly favor the institution for their own daughter, compared to the 44 percent who strongly disfavor. Opinion was less split regardingtwo-sister marriage, with 62 percent o fwomen who are themselves insuch an arrangement strongly favoring it for their own daughters, compared to 43 percent among remaining women. This suggests that women perceive advantages, but also associate a certain degree o f stigma, to these marital institutions. 2.39. What might the advantages be? One way to approach the question i s to ask how a woman's bargaining power i s enhanced either by having her brother married to her husband's sister (watta satta) or by having her own sister as a sister-in-law (two-sister marriage). Inthe case o f watta satta, qualitative data indicate that this arrangement normally involves a tacit agreement among the families for each husband to retaliate in-kind against his wife in case o f the other husband's malfeasance. Thus, by maltreating his wife, a husband risks having his sister maltreated by her husband, a mutual deterrent which raises each wife's bargaining power within their respective marriages. Two-sister marriage may have a similar effect, but through a different mechanism. Two sisters living together in the same household as sisters-in-law are more likely to cooperate than two unrelated sisters-in-law livingunder the same roof. Thus, in case of disagreements with their husbands, each o f the wives in the two-sister marriage has a stronger fallback position (Le., cooperation with each other) than two unrelated wives, This gives them more bargaining power and may consequently lead to better treatment by their husbands and their husbands' families. 2.40. Are women in watta satta or two-sister marriages really better o f f than women in conventional marriages? This is a complicated question, both because o f the difficulty in measuring women's welfare within marriage (see below) and due to the potential selection o f women into these types o f marriage. To the extent that women from poor families, or who are otherwise o f low status, are more likely to enter into such arrangements, they may appear to be worse off; but this mightjust be a selection bias. The PRHS-I1 (2004) asks women several questions about the "quality" o f their marriage, which can be used to construct an index o f marriage quality.56 Regression analysis shows no significant difference in quality of marriage across watta satta, two-sister, and conventional marriages after controlling for the women's age, 56 The questions are: (1) "Does your husband usually spend his free time with you?' (2) "Does your husband show affection toward you?" (3) "Does your husbandrespect you and your wishes?" (4) "How happy are you with your overall relationship with your husband?" Questions (1)-(3) can be answered "frequently", "sometimes", or "never". Question (4) can be answered on a five-point scale. 28 schooling, region o f residence, and husband's land ownership. While this analysis does not necessarily fully surmount the measurement and selection problemsjust mentioned, it does, at the very least, suggest that women in these exotic forms o f marriage are not treated worse than those in conventional marriage, as some Pakistani commentators have implied. 2.41. While more research is needed to uncover the predominant rationale for these pervasive marriage customs, one intriguing possibility emerges from the discussion thus far. In rural Pakistan, a woman would have little, if any, legal recourse in the event o f mistreatment by her husband. It would be practically unheard o f to involve the police, for example, in a domestic abuse case. Perhaps the wife could temporarily returnto her natal family, as explored later in this chapter, but this may entail stigma and shame o f its own. Arrangements like watta satta, two-sister marriage, and even cross-cousin marriage may thus serve, at least partly, to fill a void left by the state. This argument, o f course, presupposes that parents, in arranging their daughter's marriage, are also attentive to her interests-albeit, ifonly to avoid stains onfamily honorthat mightarise from amaritalbreakup. The Marriage Contract (Nikuhnuma) 2.42. The institution o f marriage (nikah) is central to Muslim family law. Scholars agree that the Islamic essence of marriage i s a contract inwhich all parties involved are to honor multiple requisites that serve and protect the various partie~.~'In light o f marital practices that were particularly harmful to females during colonial rule (occurring even in the face o f subcontinentwide legislation to protect the rightso f females inmarriage), the Government o f Pakistan has taken pains to incorporate protections for women and girls into marriage law (see Annex 2) by emphasizing the importance o f the marriage contract (nikahnama)and abiding by its requirements. While these laws have absorbed the sanctity o fthe marriage contract emphasized by Shariah, many are applicable to all citizens o f Pakistan, and not just specific communitiesor religious groups.58 2.43. The nikahnama i s the centerpiece o f the Muslim Family Law Ordinance o f 1961. By requiring the registration o f a marriage contract and that it contain specific protections for both male and female parties, the MFLO attempted to make marital practices more transparent and accountable to the law. The specific protections of women's rights to be incorporated into the nikahnama are detailed in Annex to Chapter 2. 2.44. Although the MFLO's nikahnama contained unprecedented protections for females entering into marriage, these protections have failed to bring sweeping improvements in Pakistan's marital practices. There are two reasons for this. First, as with other aspects o f family law, requirements o f the nikahnama-not to mention o f previous family laws (Le., the Dissolution of Muslim Marriages Act (DMMA) o f 1939 and MFLO o f 1 9 6 l b f t e n are not met by practicing parties, as they conflict with customary practice; moreover, authorities rarely hold parties accountable for fulfilling requirements. Second, according to women's rights advocates inPakistan, the relevant provisions inthe family laws and nikahnama are too weak-e.g., they are too vaguely worded or lack meaningful content-to effectively intervene in customary marital practices that discriminate against women. Civil society organizations and prominent legal scholars in Pakistan stress that the provisions most in need o f change are those related to polygamy and divorce. Customary routines in both areas have undercut women's abilities to extricate themselves fiom marriages inwhich their basic needs and rights are not met. 2.45. Inaccordance with recommendations by the Commissionon the Marriage and Family Laws, The Aurat Foundation's Legislative Watch Program has released several documents recommending changes "Ali(2000);Mannan(1995). Governmentof Pakistan (1997, 19). 29 to marital laws and to the nikahnama form as originally conceived in 1961.59 Many recommendations chairs Pakistan's National Commission on Women, also has urged that divorce law ". ,.provide that a involve easing the requirements for a woman to rightfully seek a divorce.60 Justice Majida Razvi, who husband must intimate pronouncement o f divorce to the competent authority within a limited period e.g., within a week from the date o f pronouncement and violation must entail severe punishment than what is provided in law or a wife should be entitled to give the notice o f talaq [divorce] where a husband fails to do so" because, so far, regarding divorce, "no period has been prescribed within which a husband must inform the competent authority that he has given divorce to hiswife.. ."61 2.46. To further restrict the practice o f polygamy, it also has been recommended that the bridegroom should have to disclose his marital status in the nikahnama, which before only required the bride to disclose her marital status. Justice Majida Razvi writes: ,,.the Nikahnama should have some additional clauses indicating if the bridegroom is already married, has never married or a widower or if marriage has been terminated through divorce in any form the same should be specified. Further ifthere is any existing wife that should also be mentioned, In such a case the particulars o f the existing wife, the name and address should also be mentioned in the Nikahnama. Further in such cases if the bridegroom has taken permission from the first wife or the competent authority and a duly attested document to that effect be produced. Further ifthere are any children from the previous marriage, detailed information in regard to the said children and as to in whose custody children are and who is responsible to maintain them. The same particulars can be mentioned or asked about the wife also.62 2.47. The Legislative Watch Programme also urges a looser definition o f lack o f "equitable treatment" inthe DMh4A's clause granting women the rightto divorce polygamoushusbands; the original provision allows divorce only if the husband, who "has more wives than one, does not treat her equitably in accordance with the injunctions o f the Q ~ r ' a n . ' ' ~It~is important to note that polygamy i s rare in rural Pakistan. Data from the PRHS-I 2001-02 indicates that fewer than 3 percent o f rural households nationwide contain polygamous marriages, although these figures do not account for men who may have other wives living in different households. 2.48. The PRHS-I1 is a rich source o f information on marital contracts. According to these data, three- quarters o f married women do not have a nikahnama (or are not aware that they have one). There is, once again, a striking regional contrast: 69 percent o f northern Punjabi women have a nikahnama and 25 percent o f those in southern Punjab, but only 4 percent o f Sindhis. In addition, 72 percent o f those who report having a nikahnama have never read it or have never had it read to them. Given, then, that only 7 percent actually have a nikahnama and know what is stipulates, the scope o f its protection seems severely limited. Indeed, practically no woman who had a nikahnama reports that it gives her the right o f divorce, though nearly one-third do not have any idea one way or the another. 2.49. Who obtains a nikahnama? Multiple regression analysis shows that, aside from region, the only significant determinant o f having a nikahnama is the woman's education level (controlling for education 59"SuggestedAdditions/Amendments in Muslim Family Laws: Nikahnama Form"; "Suggested Additions/Amendments in Muslim Family Laws: The Dissolution of MuslimMarriages Act, 1939(VI11 of 1939)"; Aurat Foundation& (various years). 6o The Aurat Publication's recommendations for looser grounds-compared to the DMMA's (1939) originally-specified grounds for female-initiated divorce-are discussed inAnnex 3. Razvi (2004: 6). Ibid., p. 7. 63"Suggested AdditiondAmendments inMuslimFamily Laws: The Dissolution of MuslimMarriagesAct, 1939 (VI11of 1939)," Legislative Watch Programme, Aurat Publication and Information Service Foundation, Islamabad. 30 of her father and her spouse's father). Having a primary school education or less nearly doubles the likelihood o f obtaining a nikahnama, and having above a primary school education nearly triples the likelihood. There are two ways o f interpreting this result. First, it is possible that even minimally educated (Le., literate) women are in a stronger position vis-a-vis their husbands' families to demand a nikahnama, compared to uneducated women. An alternative interpretation is that higher female educational attainment i s merely capturing better treatment of girls more generally. That is, families who care more about their daughters will give them more education and, at the same time, will attach greater importance to the nikahnama. Under either interpretation, the nikahnama seems to have positive connotations for women. 2.50. One traditional element o f the marital contract i s the haq meher, a sort o f severance clause whereby the husband agrees to pay a prespecified cash amount to the wife in the event that he initiates divorce. Here, again, we have a custom that appears, at least in principle, to fill a legal lacuna, namely enforcement o f alimony. To this extent, haq meher, negotiated by the parents o f the bride and groom, mitigates arbitrary divorce, or at least protects women against its costs. 2.5 1. Nominally, haq meher appears to be quite important in Sindh, where 53 percent o f women report an amount according to the PRHS-11. This compares to 18 percent o f women innorthernPunjab and only 6 percent o f women in southern Punjab. The amount o f the haq meher, however, i s often so low-500 rupees or less in 30 percent o f cases-to render it a largely symbolic gesture. If we consider only haq meher in excess o f the more respectable figure o f 2,500 rupees, these regional differences are attenuated. Only about 20 percent o f Sindhi women have been promised such a generous haq meher, compared to 9 percent in northern and 3 percent in southern Punjab. Nevertheless, these regional patterns seem to indicate that the nikahnama and haq meher may, to some extent, serve as substitutes for the law in protecting women's interests. Dowry and other MaritalTransfers 2.52. The role o f dowry and other marital transfers i s not well understood, in Pakistan or elsewhere, and remains an active area of economic research. Also lacking i s basic empirical information, such as who actually controls dowry assets after the marriage. This i s a crucial question because, in theory at least, dowry compensates the bride for receiving only one-half (if any) o f her brothers' share o f inheritance. A second important question i s whether dowry demands, as in the case o f India, lead to violence against women. According to some Pakistani experts: The custom o f a dowry puts a premium on the bride. Demands for a substantial dowry are often made before the marriage. Subsequently, the bride i s often humiliated or tortured [by the in-laws] for not bringing the expected amount. At times the bride is subjected to extreme violence. The violence takes many forms like burning, hanging the woman or killing her.64 2.53. Despite the absence of precise figures on the extent o f dowry violence in Pakistan, views such as these have led to the enactment o f the Dowry and Bridal Gifts (Restriction) Act (1976), the Dowry and Bridal Gifts (Restriction) Rules (1976), and the Dowry and Bridal Gifts (Restriction) Amendment Ordinance (1980). The objective o f these statutes i s to restrict dowry and other marriage expenditures, even though such laws are practically impossible to enforce.65 64Government o f Pakistan (1997: 39). For a description of The Dowry and Bridal Gifts (Restriction) Act, 1976,please see Annex 2 of this chapter. 31 PRHS-I1 collects detailed information on transfers to the bride at the time o f her Table2.2: ExtentofWife's Control Over Dowry marriage, Dowry, which comes from the Region Full Partial None Total bride's side, and bari, which comes from the Household Goods groom's side are practically universal in N.Punjab 87.9 5.4 6.7 100 rural Pakistan (at least in Punjab and Sindh). S. Punjab 60.4 19.3 20.3 100 Both transfers are nominally intended to go Sindh 91.7 4.1 4.1 100 to the bride, and the data indicate that, by and large, this is the case in reality. For All 82.5 8.5 9.0 100 each asset transferred, the survey asks Gold women the extent to which it is exclusively N.Punjab 81.0 7.0 12.0 100 hersto dispose o f as she pleases. Tables 2.2 and 2.3 show the responses for what are by S. Punjab 55.6 19.5 24.9 100 far the most commonly given dowry and Sindh 93.2 3.9 3.O 100 bari assets, household goods (clothes, appliances, and utensils) and gold. In the All 79.6 8.9 11.5 100 case o f dowry, only one-fifth o f recipients report less than full control, about equally Table 2.3: Extentof Wife's ControlOver Bari split between partial and no control. Region Full Partial None Total Interestingly, especially in light o f earlier Household Goods findings, Sindh seems to be the most 78.2 8.9 12.9 100 "progressive" region in terms o f security of N.Punjab property rights over dowry assets, whereas S. Punjab 51.8 22.3 25.9 100 southern Punjab is the least secure by a wide Sindh 89.0 3.4 7.6 100 margin, The story is similar for bari assets, although overall the extent o f the wife's All 77.1 9.5 13.5 100 control i s somewhat lower. This perhaps is Gold due to the fact that bari comes from the N.Punjab 67.1 10.7 22.2 100 family o f the groom and the husband and his S. Punjab 46.7 15.9 37.4 100 family feel more entitled to it. Practically Sindh 86.7 6.1 7.2 100 no women in the PRHS-I1 sample report receiving land as part o f her dowry (this is IAll 72.6 9.5 17.9 100 true o f bari as well), a finding consistent with the theory o f dowry outlined above. Bequeathing wealth to daughters in the form o f land creates a disincentive for virilocal sons to maintain and improve the land. According to the theory, a daughter's dowry should largely be inthe form of cash or other assets (e.g., gold, clothes, utensils) whose value does not depend on the work effort o f her brothers. In fact, the only productive asset to feature prominently in dowries is livestock. About 25 percent o f dowries include livestock (buffaloes, most commonly), and they constitute about 9 percent of dowry value on average. But this exception actually proves the rule, as livestock production in rural Pakistan i s principally the responsibility o f women. Consequently, bequeathing wealth to women inthe form o f livestock has no negative incentive effects, 2.55. Giventhe generally high level of female control over dowry (and bari) assetsjust reported, how much wealth transfer to daughters does the typical dowry represent? PRHS-I1 data indicate the median dowry to have a total value of about 10,000 Rs.(Table 2 . 4 t a l m o s t twice as large as median bari. Dowry and bari values are both much higher in wealthier northern Punjab than in southern Punjab and Sindh. The comparison between the value o f the woman's dowry and that o f her brothers' inheritance (calculated by dividing the father's landholdings, valued at median land prices per acre, by the number o f brothers) i s revealing. Among women whose fathers had land, the median dowry i s worth 15,000 Rps., whereas the median value o f brothers' inheritance is worth 100,000 Rps., or more than six times as much, 32 Taken along with the low incidence o f direct inheritance by women, these figures indicate Table 2.4: Size of MaritalTransfers that women from landed households are strongly Median value fRuuees) discriminated against in favor o f their brothers Repion Doww Bari Total when it comes to the disposition o f the patrimony. The situation is different, however, N.Punjab 31,250 16,000 49,500 for women whose fathers do not own land. In S. Punjab 12,000 5,000 18,500 this case, while the median dowry is only worth 7,500 Rps., the brothers are not getting any Sindh 6,350 3,650 11,900 inheritance (although they may be getting inter vivos66transfers in one form or another that are All 10,,200 5,400 17,950 not captured inthe data). 2.56. Multiple regression analysis shows that dowry increases significantly (1) with father's landholdings, education, and whether he holds an important position; (2) with father-in-law's landholdings; (3) if the woman has no sister, but not if she has no brother; and (4) with woman's education, but only above the primary level. Thus, dowry depends strongly on the woman's family wealth relative to the number o f sisters claiming dowry out o f that wealth. This presents an interesting contrast to women's inheritance, the probability o f which, as we have already seen, does not vary by father's wealth. In addition, higher dowries are associated with wealthier husbands, suggesting positive assortative mating on wealth. Finally, the nominal value o f both dowry and bari at the time o f marriage increase significantly with year o f marriage, but the rate o f increase i s modest, on the order o f 2-3 percent per annum. Thus, there is no evidence o f dowry inflation, as in India, with its negative implications for the treatment o f women. 2.57. Regarding marriage type, dowries are not significantly different across watta satta marriages and non-watta satta marriage^.^' Bari expenditures, on the other hand, are substantially lower in watta satta marriages, but given the typical size of bari this difference does not seem to provide a compelling explanation for the highincidence o f exchange marriage inrural Pakistan. 2.58. Finally, there i s the question o f whether demands for dowry or other postmarital transfers are the source o f violence against women. Recent evidence from India (Bloch and Rao 2002) suggests that husbands may use domestic abuse as a way to extort greater dowries from wives' natal families, According to PRHS-I1 data, however, only about 15 percent o f women in rural Pakistan say that their natal families ever provided support to their husband's household. Far and away the most important occasion for such support was to assist in health expenditures, which could have been for the wife herself (with house construction or major repair a distant second). Moreover, only a small minority of respondents, less than 9 percent, report feeling even the mildest pressure to obtain financial support from their natal families. Virtually none said this pressure was intense. Given this evidence, and the highlevel o f control women say they have over dowry assets, there does not appear to be wide scope for dowry- related violence inrural Pakistan. 2.59. Taken together, the findings suggest that laws restricting dowry and other marriage-related transfers, to the extent that they are enforceable, may be counterproductive in Pakistan. First, if we take seriously what women say about their ability to dispose o f dowry assets, then dowry is the principal channel o f female inheritance in rural Pakistan. It i s not clear how cutting o f f this channel will benefit women. Second, in rural Pakistan there i s little evidence o f the kinds of negative consequences o f dowry 66Betweenliving persons "Onlywattasattamarriagesinwhichbothcounterpartcouplesarealreadymarriedareincluded,sincethesecasesaremore likely to be motivatedbyareductionindowry or bari expenses. 33 that have arisen in India. In particular, dowry inflation is minimal and ex-post demands for financial assistance from the natalhousehold are modest. Thus, stamping out highdowries will not obviouslylead to bettertreatment ofwomen. Women's Welfare and Autonomy within Marriage 2.60. It is perhaps fitting to concludethis chapter with an analysis, limitedas it is, of women's welfare within marriage. Implicit in the chapter thus far has been the crucial role of the natal family, especially after marriage. The discussion already has touched upon the high rate of village endogamy in rural Pakistan, but even those figures underestimatethe proximity of natal households. Accordingto PRHS-I1 data, amongthe 10percent ofwomenwhose natal families live ina neighboringvillage, 77 percentreport that they can be visited within the same day (roundtrip), whereas, among the 16 percent (20 percent) of women whose natal families live within (outside) the tehsil, 63 percent (36 percent) are within a one-day visit, All in all, then, 78 percent ofmarriedwomen have easy access to their natalfamiliesinthis sense. 2.61. Ofthe women with easy access, morethan one-halfreportmeeting members of their natal family on a daily basis. Even for those 22 percent without easy access, the majority meet with natal family members several times a year; moreover, 79 percent of women with easy access to their natal homes say that their family would shelter them for a few nights if needbe, a figure that drops off to 61 percent for women whose natalfamilies are not within easy reach. These numbers suggestthat marriedwomen are far from being cut off from their natal home, as is so often the case in India, where marriage is substantially less village endogamousthan in Pakistan. The households least hospitable to their married daughters are found in Sindh, where only about two-thirds of women would be welcomed back home in the event of a marital crisis. Women's expectations of financial support from their natal families show broadlysimilarpatterns. 2.62. When asked whether they had ever temporarily returned to live in their natal home due to an estrangementfrom their husband, 21percent of women answeredaffirmatively. Perhapsnot surprisingly, this percentageis nearly twice as highinPunjab, bothnorthand south, than it is in Sindh. These stays are typically quite short, though, with the modal periodbeing less than a month, and thus probably reflect short-term maritalupheavals rather than long-termbreaches (divorce and separation are exceedingly rare inruralPakistan, withjust over 1percentof women reportingthemselvesas such). Itwould appear, then, that the natal family, both in expectation and in reality, is a safe haven from the tribulations of married life. 2.63. One important indicator of women's welfare within marriage is the degree to which she has a voice in family decisionmaking. The PRHS-I1asks marriedwomen to rank their say in various family decisions according to whether their preferences/opinions were always, mostly, sometimes, rarely, or never taken into consideration. Table 2.5 displays the results for five importantdecisions. One thus sees that 71 percent of rural women always or mostly have a voice in decisions regardingtheir children's schooling, 59 percent on whether to have another child, 56 percent on major consumption expenditures, 26 percent on communityparticipation,andjust 24 percent on working for pay. Surprisingly,given some of the earlier findings, women in Sindh appear to have the greatestvoice in all decisions, except for those regarding community and political participation, in which case Sindh is a distant second to northern Punjab. 34 Table 2.5: Extent to Which Wife's Opinion is Taken into Account in Family Decisions Region Always Mostly Sometimes Rarely Never Total 1, Decisionsregarding children's schooling N.Punjab 40.2 29.0 14.3 3.9 12.6 100.0 S. Punjab 42.8 20.6 17.9 1.9 16.8 100.0 Sindh 62.0 14.6 9.7 2.7 11.1 100.0 Total 51.4 19.7 13.2 2.7 13.1 100.0 2. Whether to have another child N.Punjab 34.1 17.2 11.9 9.1 27.8 100.0 S. Punjab 36.9 15.6 13.8 4.2 29.6 100.0 Sindh 50.8 15.6 7.3 4.5 21.9 100.0 Total 43.3 15.9 10.1 5.4 25.3 100.0 3. Major consumption expenditures N.Punjab 26.5 25.7 16.0 5.3 26.5 100.0 S. Punjab 26.4 10.8 17.0 2.4 43.4 100.0 Sindh 53.5 14.2 7.3 4.8 20.1 100.0 Total 40.2 15.9 11.9 4.3 27.7 100.0 4. Wife's participation incommunity/political activities N.Punjab 27.6 16.0 8.6 6.9 40.9 100.0 S. PUNJAB 5.5 7.2 6.2 2.0 79.2 100.0 Sindh 18.0 5.8 3.3 3.4 69.5 100.0 Total 16.9 8.8 5.4 3.9 65.0 100.0 5. Whether or not wife should work for an income N.Punjab 16.9 5.1 2.8 3.9 71.3 100.0 S. Punjab 16.6 5.1 3.2 1.5 73.7 100.0 Sindh 23.1 4.4 2.7 1.9 67.8 100.0 Total 19.6 4.8 2.9 2.3 70.4 100.0 2.64. Ordered probit regressions o f the determinants o f household decisionmaking authority show that the wife's education matters, but only above the primary level. Women with some education below the secondary level are no more likely to have their opinion taken into consideration in any of the five decisions in Table 4 than women with no education at all. The 7 percent o f women with secondary-level schooling or above, on the other hand, have a significantly greater decisionmaking voice than do unschooled women in all five cases. Once again, however, one must be cautious about interpreting schooling effects in this context. Parents who treat their daughters better in general may provide them with both more education and a more considerate husband. 35 2.65. Two key findings emerge from this analysis. First, the strong tendency toward marital endogamy in rural Pakistan has positive implications for women's welfare. Comparedto India, married women in Pakistan are much more closely connected, geographically and otherwise, with their natal households. This means that, inthe event o f maritaltrouble, women have someone to turnto, or at least some place to go. As a consequence, one can surmise that Pakistani women have, ceteris paribus, higher marital bargaining power than their Indian counterparts. Second, the extent to which women in rural Pakistan take an active role in household decisionmaking depends strongly on the nature o fthe decisions. Women have much greater say in decisions internal to the household, such as those involving children, than they do in decisions on external matters, such as their participation in community activities and in the labor force. This finding confirms the overarching influence o f purdah on the mobility and, ultimately, the autonomy o fwomen inrural Pakistan. 111. DISCUSSIONOFRESULTSAND IMPLICATIONS FORPOLICY INTERVENTIONS 2.66. According to research by highly reputable legal scholars in Pakistan-together with some o f Pakistan's most prominent civil society organizations and human rights activists-it is clear that the lacunae in Pakistan's family law have considerable consequences for females. Using both quantitative and qualitative data, the analysis conducted for this Gender Assessment confirms distinct disadvantages to women and girls that result from certain marital and inheritance customs, many o f which are filling gaps inrelated law and/or enforcement o fthe law. RecommendedModificationsof Family Law 2.67. The analysis corroborates Mehdi's (2002) expert study o f gender and inheritance law inPakistan, which finds that women's tendency to relinquish property inheritance to brothers stems from the expectation that brothers will continue to support sisters into adulthood and after marriage, as women traditionally have had few resources with which to support themselves and limited legislationwith which to protect themselves. Fortunately, by enacting legislationthat protects the rights o f females in marriage and divorce-and now by focusing on increasing opportunities for female education, remunerated labor, and political participation, as the following chapters will show-the Government o f Pakistan is laying groundwork for women's self-empowerment, complementing efforts by prominent civil society organizations to educate the public and foster communities o f support for women. This diminishes the traditional need for women to rely on male members o f the natal family; it also should encourage greater enforcement o f statutes that protect and empower women. 2.68. In terms o f bolstering women's rights to inherit family property, Justice Majida Razvi has recommended.that inheritance law require families to document inheritance rightsthrough the writing o f wills, and that the state more strictly enforce adherence to legal documents: InSection 4 o fthe Family Laws Ordinance, wherein an appeal is lyingbefore the Shariat Appellate Bench o f the Supreme Court, if the same is accepted, then amendment in the law will be needed to make it compulsory for every citizen to make a will inregard to his assets specifically where the issue o f propositus children exists to provide for such children which is allowed under the Islamic laws. In different Islamic countries this mode has been adopted to overcome and solve the problems in regard to children o f a propositus soddaughter. Such will, will not be subjected the limitations which exist under the inheritance law.68 68Razvi(2004). 36 2.69. The empowerment o f females in other family dynamics-such as choice in marriage, age of marriage, control over dowry, and household decisionmaking-also can be bolstered by targeted modifications o f marriage law. These modifications (many o f which have been discussed above) primarily address the three marital customs determined to impinge most on the welfare o f women and girls in Pakistan: child marriage, polygamy, and overly stringent requirements for females to initiate the dissolution o f a marriage. Justice Majida Razvi, The Aurat Foundation's Legislative Watch Programme, and other prominent advocates for women's rights in Pakistan have called for increased legal protections on women's rights in marriage and divorce by amending provisions regarding marriage and divorce in particular family laws. Suggested amendments that would make the ChildMarriage RestraintsAct more effective include: changing the definition o f "child" or "minor" to anyone who i s under 18 years o f age, thus raising the minimum marriage age for girls to 18; increasing the punishment for adult males (18 years and over) who marry a child from a maximum o f one months' imprisonment and/or a fine o f 1,000 rupees (1929 provision) to a maximum o f three years' imprisonment and/or a fine o f 15,000 rupees ifthe bride is between 16 and 18 years o f age, and a maximum of five years' imprisonment and/or fine o f 25,000 rupees if the bride i s under age 16; and increasing the punishment for guardians involved in a child marriage and for whomever knowingly conducts, performs, or directs any child marriage from the 1929 maximum (one month's imprisonment and/or 1,000 rupees) to the recommended punishment for adult males who marry the child (above). 69 2.70. Other Aurat Publications have suggested specific changes to the DMMA (1939) and to the nikahnama form as originally conceived in 1961 in order to ease restrictions on women's rights to divorce. The recommendations primarily involve broadening the grounds on which a woman can rightfully seek and obtain a divorce. The specific recommendations can be found in Annex 3 o f this chapter. Finally, the collective changes recommended by the Legislative Watch Programme and Justice Majida Razvi (2004) to improve legal clauses pertain to polygamy-though it is atypical o f rural Pakistan-and include changes to the nikahnamaform as well as to DMMA provisions (please see Annex 2.3). Buildingthe Infrastructureto EncourageWomen's Pursuitof their Rights 2.71. While modifications to family law constitute clear policy levers for improving the status of women in the family, activists and legal scholars in Pakistan stress the need to reinforce statutory measures by building a community- and law-enforcement infrastructure to allow women to take advantage o f legal protections. Enforce Strict Record-keeping 2.72. Better record-keeping o f family events such as marriage, divorce, birth and death is crucial to helping women seek legal aid and protection. State enforcement therefore must underscore legal recommendations for better record-keeping, such as Razvi's suggestion that "Sub-Section 3 o f Section 5 of the Muslim Family Law Ordinance must be amended to provide that a marriage not solemnized by the Nikah Registrar shall be reported by the person who has solemnized such marriage within specified time.7770 2.73. Legal activist Zia Ahmed Awan, o f the Lawyers for Human Rights and Legal Aid (LHRLA), emphasizes the importance of enforcing legal protections for women at all levels, remarking that the 69"Suggested Additionsihendments in Muslim Family Laws: The Child Marriage Restraints Act, 1929 (XIX of 1929)"; "Suggested AdditiondAmendmentsin MuslimFamilyLaws: NikahnamaForm"; "Suggested Additionsihendments inMuslim FamilyLaws: The Dissolutionof Muslim MarriagesAct, 1939(VI11of 1939)"; Aurat Foundation(various years). 70Razvi(2004). 37 ``...law ministryshouldreviewthependingassignments oflaw commissionandtake immediate action on those that are linked with the protection and promotion o f human rights in the country" and that the government should create mechanisms for strict implementation o f UnitedNations/International/Regional Protocols & Conventions being ratified by the country. Pakistan's Supreme Court, as well, "should take up the issue o f women protection and monitor [the] government's action in this regard.... A Lady Ombudsman should be appointed to check the violence against women and redress their problems with independent judicial powers." At the police level, the police department "should establish monitoring cells at city and provincial level to check the reported cases o f violence against women.. [Tlhe process . o f the investigation o f cases has become outdated and the police department should take measures to adopt latest trends o f investigationand develop a separate team to investigate the matter relating with the crime against women and children." Regarding record-keeping, Awan suggests, "Police should collect national statistics to create a profile o f the women victims o f violence and also the perpetrator^."^' Educate Women about Their Rights andProtections 2.74. H o w to educate girls and women about their rights and sources o f protection i s a critical question in light of low rates o f girls' schooling andpurdah-based restrictions on women's mobility. CSOs such as the Aurat Foundation and Pakistan's Society for the Advancement o f Community, Health, Education and Training (SACHET) have been makingcritical contributions inthis area, usingmedia and community outreach efforts to mount campaigns that disseminate information about female rights regarding child marriage, dowry, and divorce. The beneficial effects o f such information campaigns are inestimable, as evidenced by comments of one southern Punjabi woman (interviewed for the qualitative study), who had left a situation o f domestic violence and then returned to change it for the better after being informed o f her rights in marriage and divorce (see her interview in Box 2.1; her comments are in italics, and the interviewer's are in regular type). 7'Awan (2005: 20). 38 Box 2.1: Exposure to Media Information on Women's Legal Rights Makes a Difference How longhas it beensince you are back?Two months. How are things this time?I said that ifyou hit me now, I will go into court and say he hits me. And ifyou fight with me I will go to court. He said that your family will not let you go and I said I will go alone. It's my right to go. Who told you this? Isaw it on the television. You have one?No, Igo to my mother's house and see it there. They show this on it, say that it's a right of women to do this. How long haveyou been watching television?Forfour-Jive years. Ok, in the CO [local community organization for women], are things like women's right to divorce discussed?Yes. These things are talked about in the CO. [The Social Organizer] came and told us what rights we have. She says askfor your rights. She says it is your choice whether you want to go or not. No one can everforce you. This social organizer of the CO, did you find out more from her or the television? Thatprogram on TV. Whichone?One is that khawateen time, there is this other one, I can'tremember its name. Doesn't matter. Does it come inthe morning or evening?It comes in the mornings. It says in that women have a right to divorce. They askfor more dowry, thosesorts ofprograms. You didn't have a problem that the dowry is too little did you? In our family there is no problem of dowry. Wegive dowry even when we arepoor. They think that itS our daughter, give as much as we can. Now you are happy with him, you want to stay with him?Yes. -Female CO member, Age: 23; Lodhran 2.75. In order to continue progress made with public education initiatives, Zia recommends further action by CSOs, such as launching awareness campaigns-particularly in rural areas-on laws that protect women, addressing gender-based discrimination at grass roots levels, and moving "from a paradigm o f reintegration and rehabilitation to an approach that protects and promotesthe human rightso f women inthe Provide Safe Havens, Free Legal Aid, and other Encouragements to Womento SeekJustice 2.76. A major recommendationo f women's rights advocates is that the government address the lack o f means by which females can access legal support and safety when they boldly go against convention- and often their families-to claim these rights. Zia recommends that a substantial amount o f the budget "be allocated for the development o f women," which would include increasingtheir access to information and providing free legal aid at the district level for cases regarding the rights o f women and girls.73It also includes granting judicial powers and allocating sufficient resources to Pakistan's National Commission on the Status o f Women. L a w enforcement officials also need to be trained about the sensitivities o f gender issues in law, which would be most effective if there were a greater number o f female police 72Awan (2005: 22). 73Awan (2005: 21). 39 officers: "Proper and trained women staff should be appointed at all women police stations to provide community friendly services. Women police station staff should undergo regular sensitization training to deal with the victims as well as general public. The number o f female police staff should be increased."74 2.77. Once women utilize information about their rights and free legal aid, they need to be assured protection against retaliation, such as physical and/or mental abuse by disgruntled husbands, in-laws, or natal family members. According to Awan, "There i s a dire need o f support systems for women victims o f violence and government should immediately establish crisis centershhelters for them."75 Government also should help CSOs develop "unconventional shelters." CSO's also should engage in "strong networking with all service-providing GOs [government organizations] and NGOs [non-government organizations] to help the victims and survivors o f violence" and, to further ensure adequate protectiono f rights, "CSOs should form watch groups to monitor the role o f police and judiciary, especially cases pertinent to violence against women.'"6 With increased availability o f safe houses, crisis centers, and more responsive police protection, women no doubt will feel more assured that pursuing their rights is worth the potential risks. 2.78. As discussed in this chapter, lacunae in the law and their attendant customary practices powerfully shape the landscape o f opportunities for women and girls in Pakistan. The same types of constraints that deter women from claiming their rightsand legal entitlements-such as heritable land, the freedom to not enter or to leave a marriage that is detrimental to their physical and mental well-being, and protection from self-sacrifice in order to protect their family's honor-also inhibittheir progress in other dimensions o f society. As the following chapters will show, female opportunities for education, health care, labor force participation, and involvement inthe public sphere all have the common quality o f being curtailed by limitations that women face in Pakistan. These limitations include restrictions on female mobility, low access to information, and a lack o f leverage in decision-making both within the household and outside it, e.g., inthe political arena. 74Awan (2005: 21). 75Ibid. 76Awan (2005: 22). 40 3. I~~PROVING GIRLS'SCHOOLING Seekknowledge+om the cradle to the grave. -Prophet Mohammed 3.1. At the end o fthe 1990s, one out o f every two children between the ages o f 6 and 10was enrolled in primary school and one out of every three children 11 years old and older was enrolled in middle or high school. The enrollment figures for girls, and specifically girls in rural areas, are much lower. Less than one out o f two girls was enrolled in primary school andjust over one out o f four girls was enrolled inmiddle or high school. These numbers are low in both an absolute sense and when compared to other countries at similar income levels in South Asia. 3.2. L o w educational attainment among women has far-reaching consequences. Education makes women more productive both inside and outside the household.77An educated mother can plan the size o f her family, ensure the well-being o f her children, and make better use o f community services.78Outsidethe home, women's education i s associated with higher productivity in wage employment and in agri~ulture.~~ Educated women are also more likely to participate in the political process; illiteracy i s a major obstacle in accessing relevant information and dealing with electoral procedures and political issues.'' 3.3. The experience o f the 1990s is far from desirable, but some trends provide evidence for cautious optimism. Although overall enrollment levels did not improve, enrollment rates among rural girls rose in both the lowest and the highest expenditure deciles.'l 3.4. Will this pattern of enrollment growth hold inthe coming decade, particularly for rural girls? This chapter examines the experience o f the past two decades and identifies growth levers that have pushedthe process this far, Then, looking at the potential o f these levers to get us to the next stage, we identify key constraints that are likely to emerge over the coming decade, along with an innovative and multi-pronged strategy to obviate them. Ultimately, further improvements in enrollment and retention will rest on successfully pulling in poorer and spatially more isolated children. 3.5. The analysis identified two constraints. First, proximity to schools i s a serious constraint to the enrollment o f girls. Decreasing the physical cost o f attending school for girls is likely to pay big dividends, How can this be achieved? School construction will continue to be important, but a more nuanced approach that addresses the needs of scattered rural population unfeasible -is - where dedicated community-level schools are also required. Second, even where feasible, the construction o f schools i s likely to face another important constraint: there simply are too few educated women in many Pakistani villages to staff a school for girls. Government schools (and most private schools) for girls require female teachers, but significant barriers to female mobility prevent educated women from relocating or commuting to where the jobs are. Hiring and retaining female teachers thus will continue to be a problem, and ironically this problem will be at its worst in precisely those areas which are poorly served at present. 3.6. This is a potential Catch-22 situation.The construction o f a middle or high school is not likely to be warranted inevery community; however, the absence of such schools suflciently close by will hinder the development o f primary schools, and-as we show below-discourage private primary schools. Breaking this unfortunate cycle will require innovative interventions to ensure girls have access to middle and high ''Summersand 77 (1992), Schultz(1989). Straws Thomas (1995). 79Behrmanand Deolalikar (1988), Quisumbing (1996). '!Ziaand Bari (1999) According to the PIHS 1991, only 40 percent o f girls in rural areas age 6-10 were enrolled in a primary school, and only 17 percentof these girls age 11-17 were enrolled in a middle or high school (table 3.1) schools without having to construct a middle or high school in every village. While cultural constraints on female mobility are not likely to yield to short-run policy levers, the creation o f a cohort o f educated women in every village may be a viable policy intervention. Marriage, residence, and migration patterns suggest that educated girls frequently remain in the villages they come from, so this constitutes a potential pool o f future teachers for the next generation. 3.7. Of course, other factors such as income, parental education, parental attitudes and differences in returns to schooling between parents and children also matter.82 These factors also influence the differential schooling choices of parents for their male and female children. Some o f these factors, although well understood in the literature, are unlikely to be amenable to short-run policy levers. Others, particularly financial constraints, are already being addressed by several ongoing income transfer programs via stipends, school meal programs, school books, and uniforms. Our objective is to identify non-economic constraints that impede both school attendance for girls and female teacher availability. We thus argue for complementary interventions to augment the efficacy and uptake o f other ongoing interventions. 3.8. A two-pronged strategy to improve female education is required. The first is a supply-side strategy that decreases the physical cost o f attending schools for girls. This strategy calls for school proximity as a viablepolicy lever that can be used by the government to improve female education. Evidence drawn from current and ongoing work shows that such a policy is likely to yield highdividends. This chapter argues for the introduction o f specific initiatives-beyond school construction-to improve school access. Secondly, the chapter also examines some current demand-side initiatives. While several such initiatives are currently underway, we know little about their impact. Systematic evaluations o f these programs could teach us much about what works and how to design future policy. 3.9. The remainder o f this chapter is structured as follows. Section Idiscusses the institutional arrangements for delivering education services in the country and reviews a number o f the initiatives currently underway to improve schooling outcomes, with particular attention to education outcomes for girls. Section I1examines the gender gap ineducation. Section I11examines the impact o f school proximity on school outcomes. Section IV builds on the analysis o f section I11by examining potential constraints on supply-side initiatives that focus on school construction, private or public, arguing that mobility constraints on women are likely to generate female teacher shortages, which will be more acute in currently underserved areas. Section V examines potential policy alternatives. The data used come from a number of household data sources, the main one being the Pakistan Integrated Household Survey (PIHS) o f 2001-02 and 1991. It also relies on early results from ongoing Preliminary results from the Pakistan Social and Living Standard Measurement Survey (PSLM)/ Core Welfare Indicator Questionnaire (CWIQ) (2004- 05) indicate significant improvements in enrollment and literacy rates. Analysis o f this data is not included inthis report as we do not yet have access to the household-level data. I.SERVICEDELIVERY INEDUCATION: INSTITUTIONAL SETUP THE 3.10. The achievement o f universal primary education, with a particular emphasis on promoting girls' schooling, has been a government priority since the early 1990s. The Education Sector Reform (ESR) action plan (2001-2005) reiterated this. It i s also a core component o f the PRSP education strategy.84 Reflecting this priority, the number of boys' and girls' public primary schools increased throughout the '*Kochar (2001a), Khan (1993), Sathar and Lloyd (1994), Lloyd, Mete and Sathar (2002), World Bank (2001), Pakistan Poverty Assessment (2002), Lokshin and Sawada, Holmes (1999)), Irfan (1985), Sathar and Kazi (1987), Sathar and Lloyd (1994), Ray (2000). 83Jacoby and Mansuri (2005a) using PRHS-I1 and Andrabi, Das and Khwaja (2005). Both studies survey data collected by them. These havebeenappropriatelyreferenced inthe text. 84Government of Pakistan (2003). 42 1990s. By 2000-01, there were 127,709 public primary schools and 12,984 public middle schools in Pakistan.85 3.11. A substantial expansion o f private schools occurred during the 1990s. By 2000 there were about 32,000 private schools inthe country.86According to data from the Private School Census these schools are playing an increasingly important role in primary education, both in absolute terms and relative to public schooling, especially in rural areas where the largest growth in private schools i s taking place.87These for- profit institutions offer mainly co-educational schooling, in sharp contrast to the typical single-sex public school. 3.12. The growth in private schools appears to reflect a rising demand for better quality schools. The PRSP notes that while the expansion o f schools during the 1990s (under the Social Action Plan) increased school quantity, little attention was paid to school quality issues, including teacher availability. The ESR action plan includes a number o f steps to address such quality issues. 3.13. In the post-devolution period, responsibility for the delivery o f education services has shifted to provincial and district governments, which have put in place several programs to improve the delivery o f education.88 The federal government continued to play an important role in setting policy priorities and anchoring federal education programs. The National Education Policy (NEP) emphasizes closing the gender gap in education and improving school quality. The NEP also recognizes the growing importance o f private schools in education and encourages private sector participation. The ESR action plan outlines both short- and long-term strategies for achieving NEP goals, includingthe important "Education for All"`(EFA) goal. 3.14. One focus area o f the ESR and the Compulsory Primary Education Ordinance i s the achievement o f universal primary education. Universal primary education i s to be achieved partly through a change in the mechanism for allocating new schools. Initially schools were allocated to communities on the basis o f p o p ~ l a t i o n ,but gender and need now also play a role in school placement decisions. Our analysis, ~ ~ discussed later in the chapter, shows that this new approach could be quite effective in improving school enrollment and retentionrates for girls. 3.15. The ESR also includes initiatives for public-private partnerships to enhance access to better-quality schools, Some o f the strategies being considered on this front are to transfer the management o f under- utilized public schools to the private sector, to encourage school placement in under-served areas through the provision o f grants and soft loans, and to implement the "adopt a school" program?' 3.16. The implementation strategy for the ESR includes enrollment incentive packages in primary schools, One such innovative program, Tawana Pakistan, is a federally funded school-based meal p r ~ g r a m .This program is implemented through district governments working in close collaboration with ~ ' provincial Education and Health Departments. About 500,000 primary school-age girls (5-9 years) are being targeted under this program for a three-year period (2002-2005). The program, which currently i s being implemented in 20 high-poverty districts all over Pakistan, hopes to reduce the gender gap in school enrollment and improve school retention at the primary-school level. In a typical school, the program 85PakistanEducation and School Atlas, CRPRID, Planning Commission. Andrabi, Das andKhwaja (2002). Private schools continue to be more prevalent inurban areas. The World Bank is supportingprovincial programs in Sindh and Punjab. 89Our analysis suggests that demand-related factors such as village wealth and average education levels also impact school placement, particularly inthe case of girl's schools. 90The World Bank-IMF Joint Staff Assessment of the PRSP noted that this public-private initiative needed to be better planned, including details on how this initiative could be applied elsewhere. 91This project is sponsored by the Ministry of Women Development and Social Welfare (MoWD & SW) and is executed by PakistanBait-ul Maal (PBM) with technical assistance from Aga Khan University. 43 intends to provide food (one cooked meal a day), vitamins, iron, and de-worming medicine to 100 girls, o f which almost two-thirds are not enrolled in school. The non-enrolled girls are invited to join the meal program and get enrolled in the school. The program is currently under implementation; no external evaluation o f the program i s yet available. 3.17. A number o f other enrollment incentive programs also have been introduced in different provinces for middle schools. Middle school stipend programs currently are being implemented in Sindh and Punjab. The Sindh government has an ongoing scholarship program for girls in rural areas, which provides monetary support to girls enrolled inmiddle school. Punjab recently has initiated a stipendprogram for girls enrolled in middle school through the Punjab Education Sector Reform Program. Each girl enrolled in a public sector middle school is given a stipend o f Rs. 200 (less than $4) per month, provided she has an 80- percent attendance rate. The program targets girls in 15 low-literacy districts in the province. About 175,000 female students are being covered by the )I program. Table 3.1: Net School Enrollment Rate, PIHS 2001-02 Secondary (Middle 11. DIMENSIONSOFTHE GENDER IN GAP Primary and High School) EDUCATION (6-10) (II-I6) 3.18.Picture2001-02, only 58 percent o f primary The In in2001-02 Rural school-aged boys and 46 percent o f school-aged girls were enrolled in primary school (Table 3.1). At the Overall 57.9 45.5 37.8 27.1 provincial level participation in primary school was Source: PakistanPovertyUpdate (World Bmk 2003). highest in Punjab and the lowest in Balochistan; the gender gap follows a similar pattern and was smallest in Punjab and largest in NWFP) and Balochistan (Figure 3.1). The rural-urban divide i s striking in Table 3.1. At the primary level, enrollment rates for both boys and girls in urban areas are around 65 percent (boys just above, girls just below). The differential between rural and urban groups for boys at 10 percent is perhaps not very high; the difference is closer to 25 percent for girls, largely due to much lower female enrollment rates in rural regions. For both primary and secondary schooling, the gender gap is largely a rural phenomenon. 3.19. The gender gap seems to arise from both lower initial enrollment and higher drop-out rate for girls. At Figure 3.1: Net Enrollment by Province,2001-02 every age, a higher percentage o f girls in rural regions have never attended school. For boys the percentage Balochistan 1 never enrolled in school declines from 60 percent at age five to 20 percent by age eleven, suggesting that NWFP boys continue to enter school at older ages. For rural Sindh girls, the percentage never enrolled declines from age five to age nine, but remains stagnant at about 50 Punjab percent after age nine, implyingthat girls tend to enter school up to-but not after-age nine (Figure 3.2). Boys Pakistan's education system is thus failing to attract Note These net enrollment ratescoverboth rural andurban areas girls, with a large percentage of rural teenage girls Source Taken from the World Bank (2003) This figure IS based never having enrolled in school. on PakistanIntegratedHouseholdSuwey 3.20. The dropout rate for girls i s particularly high in rural areas. It i s also high for older children and tends to accelerate after age 12, when girls drop out o f school at a much higher rate than boys do (Figure 3.2). This is the age at which children are typically in middle school (classes 6-8). Accelerating dropout rates around adolescence suggests that cultural practices surrounding adolescence and attainment o f puberty 44 may make it difficult for girls to continue attending school upon reaching that age. Inparticular, the norms o f purdah and restricted mobility o f females would be more strictly observed after reaching menarche, rendering access to schools more difficult for girls. We show below that such concerns significantly reduce school attendance for older girls. In such a context, supply-side interventions, such as improvements in school quality, or demand side interventions, such as incentive schemes, are likely to be much more effective ifthey also explicitly address such concerns. We discuss this inmore detail below. Figure 3.2: Age Pattern of Enrollment Percentage Enrolled,Working and Married, Percentage Enrolled, Working and Married, Rural Boys, RHS 2001-02 RuralGirls, RHS 2001-02 100 - 100 801 60 - 10 11 12 13 14 15 16 17 18 Age 10 11 12 13 14 15 16 17 18 +Enrolled -e-- Married Working +Enrolled --EMarried Working ~~ Percent Never Attended School, Rural Children, PercentageDropped Out of School, PlHS 2001-02 Rural Children, PlHS 2001-02 100 - 100 1 80 4 , 80 - I +Girls 60 -+Girls +Boys ~ 40 - -t-Boys 40 - 20 ; 0- 5 7 9 11 13 15 17 Age 5 7 9 11 13 15 17 Age Source: Based on World Bank staff tabulations usingPIHS 2001-02 data on rural children 45 3.2I. Household income remains an important Figure 3.3: Enrollment and household socioeconomicstatus determinant o f school enrollment. Much higher enrollment rates occur among better-off households (Figure 3.3), but the data also reveal an interesting gender pattern: among rural households, the gap in enrollment persists across income (expenditure) groups. This pattern has also been documented in 2 0 previous studies.92 I 1 0 1 0 , , I 1 2 3 4 5 6 7 8 9 1 0 Experience duringthe 1990s P e r C a p i t a E x p e n d i t u r e O C C I e s + U r b a n B o y s - U r b a n G i r l s 3.22. The snapshot in 2001-02 paints a dismal R u r a l B o y s w R u r a l Girls picture for female education, particularly in rural Source: Basedon World Bank staff tabulations usingPIHS 2001-02 areas. But this average picture reveals little about dataon rural andurban children. , marginal growth during the 1990s; namely, which expenditure deciles grew the fastest or growth in girls' schooling relative to growth inboys' schooling 3.23. Figures 3.4a, 3.4b and 3 . 4 ~compare Figure 3.4a: Growth IncidenceCurve of Primary School the PIHS round of 1990-91 and 2001-02. That Enrollment Rates between 1990-91 and 2001-02 is, we match the expenditure decile in 1990-91 urban to the same income expenditure decile in 2 1 2001-02 and compute the average increase in enrollment for this decile, repeating the exercise for each decile. The horizontal axis shows income deciles ranked in order o f increasing income and the vertical axis shows the growth rate. We use two different concepts o f the growth rate. Figure 3.4a shows growth in percentage points and Figure 3.4b shows growth in percentages. That is, if a decile grew from I O to 20 percent in enrollment, it 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 population decileby per caphaconsunption expenditure would show a IO-percentage point increase in -allchildren--------- girls --- enrollment, but a 100 percent increase. Finally, Figure 3 . 4 ~examines the relationship Note: Growth rates of primary school net enrollment rates between 1990-91 between enrollment growth and the gender and 2001-02 are computed for each decile of per capita consumption gap. Instead o f ranking the deciles by expenditure distribution, and are then smoothed bv a non-oarametric method expenditure, we rank them by enrollment (LOWESS). Deciles are computedfor each region-sepuateiy. Source:PIHS 1990-91and2001-02. growth. Thus, the first decile is the Fieure 3.4b: Changes i expenditure decile that showed the smallest n Primary School Enroiment Rates between 1990-91 and 2001-02 enrollment growth during the last decade. urban ral 3 1 3.24. The percentage o f children enrolled -1 nationally has not changed noticeably. This is disturbing given the push for school construction over this same period. In fact, at 95-percent confidence intervals, we cannot reject the hypothesis that there has been no -.l 4 change during the last decade.93Further, the - . 2 1 , I , , , , , , , , , 0 2 4 6 8 1 0 0 2 4 6 8 1 0 0 2 4 6 8 10 populationdecile by per capita consunptlon expendlture 92 Filmer, KingandPritchett (1998). 1 -all children -.------- --- bovsl 93 Sampling errors are very large though. aids A'ote: Growth rates o f primary school net enrollment rates (NER) between 90- 91 4@ 01-02 are computed for each decile of per capita consumption expendibn distribution, and are then smoothed out by a non-parametric method(LOWESS). Deciles are computedfor eachregion separately. Source: PIHS 1990-91 and2001-02. small change over the last decade occurred in the richer groups, where enrollment was already high.But if we look closer we can see considerable regional and gender variation. For urban areas, the pattern o f greater percentage-point increases for higher income groups is replicated across boys' and girls' groups. The differences are stark, with lower enrollment for almost all groups below the median.94For rural areas, there is a dramatic difference between boys and girls. For boys the patterns are similar to those in urban areas, but for girls there i s higher percentage-point growth among the lower and the upper-income deciles. 3.25. Figure 3.4b shows how big this Figure 3.4~:Gender Gap of Growth Rate of Primary Net wP difference is. Since rural girls (particularly in Enrollment Rate (NER) the low-expenditure deciles) had very low national urban rural initial enrollment rates a small change in 2 percentage points can lead to large changes in K W z percentages (moving from 2 percent to 3 ......................................................... .................... ................................................ ...................................................................... percent is a 50-percent increase). The picture in ._ f 0 b urban areas in percentages i s not very different 3 -2 from before; both boys and girls show similar 3 trends, with lower-income deciles -.4 underperforming their higher-income .-OIC counterparts. In rural villages enrollment for -.6 girls from low-income deciles grew close to Fb 0 200 percent during the last decade; at an annualized rate o f growth, this is close to eight I , I , / , I I I I 1 , / / I \ / / I 8 I , , # , , , , I , Mn 41h 6th 8th Max Mn 4th 6th 8th Max Mn 4th 6th 8th Max percent. Over the same period, there i s almost populationdecileordered by growth rate of primry m f o r all children no growth for boys' enrollment in rural areas, Note: Population deciles o f per capita consumption expenditure distribution and the relationship with income i s almost flat. are placed by the growth rate of primary school net enrollment rate for all children, Le., from a decile with the smallest growth rate to one with the 3.26. As might be expected, Figure 3 . 4 ~ largest growth rate. Gender gaps inthe growth rate o f primarynet enrollment rate are computed by subtracting the growth rate of girls from that o f boys, shows a sharp decrease in the gender gap andthen smoothedout by anon-parametric method (LOWESS). among groups with the greatest enrollment Source: PIHS 1990-91and 2001-02. growth. School Placement Appears to Favor Wealthier and More Centrally Located Villages 3.27. The poor overall enrollment growth rate seems at least partly related to the poor distribution o f schools. Earlier studies argue that the placement decision o f public schools was largely independent o f community characteristics. But the PIHS data show that public primary schools for girls a pear to be overwhelmingly located in wealthier and better-located communities (see Table A3.6).9 Since the P placement o f a primary school is unlikely to change village wealth in a few years, we can be reasonably certain that the chain o f causality does not run in the opposite direction. For example Andrabi, Das, and Khwaja (2005) look at villages in Punjab before schools were placed in them between 1980 and 2000. They confirm, in percentage terms, some differences between communities that received girls' primary schools and those that received none, but that these very small differences do not extend to comparisons between villages receiving a primary school and villages that receivedboth a primary and secondary school. Inthe villages that received both types of schools population size played a big role (Table 3.2). This is in line with the stated policies o f school construction during the past two decades: villages had to fulfill two conditions for a school to be constructed: (1) they had to provide land for the school (4 kanals), and (2) they 94 Caution:the poor in 2001-02 may be quite different from the poor in 1990-91, both due to income mobility and migration, and such mobility is likely to be more important in urban areas. For example, a lot o f new migration to cities by the poor, would leave enrollment rates in the lowest deciles unchanged. 95Lloyd, Mete and Sathar (2002). Alderman, Orazem and Paterno (2001). 47 Table 3.2: Differences in Population Size betweenVillages in Punjab that Received a Public School, 1980-2000 Total Total Village classiJication Population Population in 1981 in 1998 No girls' primary or secondary school in2000 1,130 1,703 Receiveda girls primary schoolbetween 1980 and2000 1,131 1,695 Receiveda girls' primary and secondary schoolbetween 1980 and 2000 1,973 2,954 Receiveda girls' secondary school between 1980 and 2000 (Primary pre- 3,420 5,041 existing) Pre-existing girls' primary and secondary school for girls 2,881 4,28 1 Pre-existing primary school for girls, no secondary school in2000 1,676 2,43 1 Source: Andrabi, Das, and Khwaja (2005). had to have a population higher than 500. This suggests that villages without schools (the marginal villages)are smaller, poorer, and probably a lot further from the road. 111. DISTANCE AND SCHOOL ENROLLMENT 3.28. This section examines the impact o f school proximity on enrollment. Systematic evidence from a number o f sources show that school enrollment for girls is highly sensitive to the distance o fthe household from the school. Clearly school distance increases the financial and physical costs o f attendance by increasing transportation costs and commuting time. In the context o f rural Pakistan, however, cultural restrictions on the mobility o f adolescent girls are likely to pose an additional barrier to school enrollment. Distributionof ruralschools 3.29. According to the PIHS, at the end o f 2001-02 about 67 percent o f all rural communities or primary samplingunits (PSUS)'~had a girl's public primary school (compared to 85 percent for boys). This drops to 22 percent for girls' middle schools (26 percent for boys) (Table 3.3). Given the size o f the average PSU (see footnote 21), it is difficult to assess school access at the village level using this data. Even a school "inside a PSU" may be several kilometers from any households resident in that PSU. Ongoing work using PRHS-I1 indicates much poorer school access for girls. Only 46 percent o f sample villages in Sindh and Punjab had a girls' elementary school inside the village. In contrast, 87 percent had a boys' elementary school within the illa age.'^ 96The PSU is the lowest strata in the PIHS. Sample households are randomly selected from each sample PSU. The typical PSU is several times larger than a typical revenue village-but PSU-level data is the only `community' level data available inthe PIHS. 97As part of the PRHS-I1(2004-05) a complete census of all schools was conducted in sample villages across Punjab and Sindh consideredbroadly representative of these provinces. 48 Table 3.3: Access to Schools in RuralPakistan (in percent) Within 1-2 kmsfrom 3-5 kmsfrom >5 kmsfvom community community community community Public primary school for girls 67 7 12 14 Public primary schools for Boys 85 6 5 4 Public middle schools for girls 22 15 23 40 Public middle schools for boys 26 18 27 29 Privateprimary (coeducation) schools 28 14 10 48 Privateprimary (coeducation)schools in rural Punjab 33 20 11 36 Privateprimary (coeducation) schools in ruralNWFP 37 12 13 38 Note:.The table shows, on average, dstance in kilometersfrom community to schools. Source: World Bank staff calculations usingthe rural community survey from PIHS 2001-02 3.30. The 1990s also saw an explosion o f private schools inrural areas. The PIHS data show that by 2000 28 percent o f rural communities in the PIHS sample had a private school. The numbers were particularly high in rural Punjab (33 percent) and NWFP (37 percent, although this also picks up a large number of N G O and trust schools). Furthermore, private schools overwhelmingly locate in villages where there is already a pre-existing primary and/or secondary school. For instance, inthe case o f the PIHS communities, more than 85 percent o fprivate schools were located inPSUs that had a girls' public primary school. 3.3 1, Further evidence that private schools locate only in regions where there are pre-existing public schools is provided by Andrabi and others (2005)' who look at the census o f villages in Punjab by matching data from the Pakistan Census Organization (PCO) with data from the Education Management Information Systems (EMIS). They find that 85 percent o f all private schools in the Punjab were situated invillages that already have a boys' primary school; o f these, 80 percent also have a girls' primary school. In all of Pakistan, enrollment in private schools as a share o f total enrollment i s high in certain districts. In central and north Punjab and the NWFP private school enrollments exceed 20 percent and often go up to 45 percent o f all enrolled children. But in large swathes o f southern Punjab as well as Sindh and Balochistan (with the exception o f Quetta and Karachi), this share is between 0 and 10 percent. 49 3.32. What is clear is that for many Figure 3.5: Enrollment and School Proximity rural children the nearest school i s several kilometers away.98 We examine the A. Enrollment and Household Income impact o f this distance on school Primary School Enrollment Rate vs Income enrollment using PIHS 2001-02. Figure ty to any school / 3.5 looks at school enrollment for different income groups and across PSUs with and without a public school. Not having a public school in the PSU significantly decreases the probability that any child (boy or girl) i s enrolled in school, but the effects are larger for girls compared to boys. Multivariate regression 0- , , , , , , , , , , , 0 2 4 6 6 1 0 0 2 4 6 8 10 IncorreOroup analysis shows that the enrollment probability for a girl in a village with a B. Enrollment Versus Age school is almost 18 percentage points Schooling choice vs age enrOilment V8 r g e drop out rate vs age higher than in a PSU without a school (Table A3.3 and A3.4). There i s a similar (and statistically significant) rise for boys, but the size o f the coefficient is smaller. Figure 3.5a shows that the protective effects o f higher income do not extend to I , , , , , , , girl's schooling; the enrollment gap 6 S 12 15 17 6 9 12 15 17 Age(years) between villages with and without a villages with bch - girls in vlllages with no sch school continues to hold even at high- ___girlsinin boys villageswith Sch --------- boys in Villages with noSch income levels. Note: Upper-left children in each decile of per capita expenditure distribution are classifiedinto four groups in terms of their gender and whether their villages have any school. Enrollment rates of each group are then computed. Upper-right: children in each decile of per capita expenditure distribution are classifiedinto four groups in terms of their gender and whether their villages have aprivale school. Enrollment rates of each group are then computed. Bottom-left: children in each age group are classified into four groups accordingto gender and whether their villages have any primaryisecondary school. Enrollment rates are then computed. Bottom-right: percentage of childrenwho left school after enrolling is computed for the same four groups of childrenas above, Source:PIHS 2001-02. 98A typical revenue village is said to be between two to three square kilometers, 50 Box 3.1: RuralHousehold Decisions (Not) to Educate Daughters Qualitative data on female educationderived from recent interviews with adult females of rural householdscan be found in Box 1.3, Chapter 1. The purpose of the interview questions regarding female education was to explore factors undermining rates of girls' schooling in rural areas of Pakistan-specifically, what is influencing householddecisions not to educate daughters. educatinggirlshadno material benefitbecause Table 3.4: Percent of Rural Women who Support Female even educated girls could not work and Education, by Region contribute financial support to the household(10 Pro- Con- Total percent). Only two women out of 60-3 Region education education (percent) percent-said they would not support educating N. 100 (24) 0 (0) 100% (24) daughters under any circumstances (Table 3.4). s. punjab 100 (12) 0 (0) 100% (12) Both women happened to be from Sindh-one Sindh 92 (22) 8 (2) 100% (24) from Mirpur and one from Badeen. All 97 (58) 3 (2) 100% (60) Note: Frequencies in parentheses. The major constraint against educatinggirls was the lack of anearby school (58 percent ofwomen interviewed). About half of these reportedthe lack of girls' schools to be their village's greatest shortcoming. Sin& hadthe highest incidence (75 percent) ofwomen reporting the constraint of distance, as neither Sindh site had a schoolfor girls; some families inMirpurresortedsendingtheir daughtersto the local boys' primary school. 3.33. Figure 3.6 shows that families report much higher travel expenditures for Figure 3.6: Annual Expenditure on Travel to School girls who are 13 or older and probably by Age and Distance to School enrolled in middle school and beyond- I even after controlling for distance to school. As girls reach middle- school age, 1500 the direct cost o f sending a girl to school seems to increase sharply. The rise in travel expenditures does not appear to occur for boys o f middle-school age. Parents may pay higher transportation costs for middle or high school-age 0-2 5-10 daughters since they want to ensure safe Distance to School2-5 (Round Trip) (Kms) passage. +Age43 Boys +Age43 Girls +Age>=13 Boys U A a e > = 1 3 Girls Source: World Bank staff cross-tabulationsof data on school expendituresreported for currently enrolled children aged6 to 17. 3.34. The analysis also suggests a gender difference inthe way school proximity affects enrollment as the child approaches adolescence (Figure 3.5B and Table A3.2). The sensitivity o f enrollment to school proximity increases sharply for girls age 13 and up. This effect i s entirely absent for boys. Thus, both for primary and secondary education we see a large drop in enrollment in PSUs that do not have schools. The magnitude o f this drop i s larger for girls and particularly for older girls. In the PRHS 2001, the parents' 51 main reason for the child's non-enrollment was, for boys, economic ("school too expensive" cited by 43 percent o f parents), but for girls, while economic motives were cited, the response was more often that they did not "approve" o f their non-enrolled daughters going to school (30 percent) and that the rate o f disapproval o f schooling for non-enrolled sons was lower (7.5 percent). Moreover, this `disapproval' for a girl increases sharply with her age. The qualitative study found similar results (see Box 3.1). Social constraintson mobilitymay be important 3.35. Qualitative data from several studies suggeststhat the way school proximity affects enrollment may be the outcome o f gender differences in adolescent girls' and boys' mobility (Box 3.2). Families clearly feel uncomfortable about sending their children to schools outside their own village-this discomfort is exacerbated in the case o f girls once they cross the age o f menarche. Traveling long distances to attend school clearly entail costs, both direct (such as transportation) and indirect (opportunity cost o f time spent in school, concerns about girls security/reputation). These studies corroborate our own findings that mobility restrictions become a more important factor in parental decisions on schooling once girls reach puberty. " 3.36. A recent survey of adolescents and young adults also documents the gendered nature o f the transition from childhood to adolescence.'00 The study found that as girls transition into adolescence, their mobility and opportunities to interact with the outside world shrinks. For boys, on the other hand, transition to adolescence tends to signify an expansion o f opportunities outside the home. 3.37. These quantitative and qualitative results may be suggestive o f a school location effect on the enrollment of girls, but there are several problems with interpreting these results as the causal impact o f distance on enrollment. First, Ifpublic schools are constructed only in villages with a demonstrated demand for schooling, it comes as no surprise that enrollment is higher in villages with such schools compared to those without. The analysis o f school placement in Section I1 shows that wealthier and better-located villages are more likely to have girls' schools. Second, the size o f the effect found using the PIHS data is likely to be an underestimate since it only indicates whether or not there i s a school inside a PSU. But as footnote 21 notes, the average PSU i s several times larger than a typical revenue village. The PRHS-I1 household data---specifically designed to address these issues- gives us a cleaner and more nuanced picture regarding the impact o f school placement on enrollment. Impact of school locationat the settlement or habitationlevel. 3.38. By and large, public schools are placed in villages on the basis o f village population. A catchment area for a school should be defined by the population served. However, most villages in rural Pakistan are organized around distinct settlements or habitations, which are more organic communities. Typically, a settlement has a distinct name and boundary and i s significantly more zaat/biradari (caste) homogeneous than a revenue village, which is a largely administrative construct. Thus, if culturally based mobility restrictions are important for girls, then crossing settlement boundaries-even inside a revenue village- should impact school enrollment. Specifically, girls in settlements that do not have a school should be less likely to enroll in school. We should not observe a similar impact for boys. Testing the potential for a "crossing boundaries" effect i s the centerpiece o f the empirical analysis usingthe PRHS-I1data."' 99Sathar and others (2003); Mumtaz and Raouf (1996); Khan (2000), Khan (1998). loo Sathar, ul Haque, Faizunissa, Sultana, Lloyd, Diers and Grant (2003) also found this pattern to exist in the Adolescent and Youth Survey (2003) data. This data collection effort included a complete census of all schools in the sample of villages. This included data on all schools inside a revenue village as well as all schools within a two-kilometer walk of the perimeter of each settlement. GPS coordinates are available for both households and schools, so the distance between each household and every local school can be calculated,The 52 Box 3.2: Puberty and Girls' Restricted Mobility May Constrain Their Schooling Puberty i s a physical marker of maturity. The onset of puberty brings with it distinctive gender-defined social trajectories for adolescents in Pakistan, as it does inmany SouthAsian societies.As girls enter puberty they often experience increased enforcement ofpurdah norms and restrictions on their mobility and social interactions. lo' The limits on mobility for young unmarried girls include restrictions on traveling unaccompaniedand the needto obtain permission from a male member of the family to travel outside the home. The qualitative study on gender (Box 1.3) found that distance to school compoundedthe effects of cultural constraints. Interview subjects frequently expressed Table 3.5: Percentages of Primary Reason culturally-based concerns about educating for Not Educating Girls,by Region girls (suchas incurring disapproval from male R~~~~~ Distance Purdah Total villagers or compromising their daughters' honor, especially if she had entered puberty); N.Punjab 46 (11) 12.5 (3) 100 (24) however, few said that purdah and honor '.Punjab 75 (9) 17 (2) lOO(12) concerns would prevent them from educating 62.5 (15) 21 (5) 100(24) daughters-so long as there was a nearby All 58 (35) 16.7 (10) 100(60) school daughters could attend. Less than 17 Note: Frequencies inparentheses. Thebiggestproblem about educatinggirls is that one can go either to thejelds or topick and drop the daughters from/to school. Ifthere is a school in the village, I think everyonewill educate their daughters. -Mains, Mirpur For women and girls, especially those in rural areas, mobility restrictions vary between movements within the community and those outside the community. A qualitative research study of three villages in rural Northern Punjab, undertaken to understand the relationship between women's mobility and their access to health and family planning services, found that unmarried girls were severely constrained in their mobility even within the village (Khan 1998). Girls were restricted from appearing inpublic places insidethe village (shops or bus stops). Their movement outside the village was even more limited. When girls were forbidden to leavethe village, it was becausetravel was perceived as dangerous and often forbidden, due to the potential of sexual harassment at the bus stop or inthe vehicles. 3.39. This analysis looks at whether a child has ever attended school, focusing on children age 12 and older. The data indicate that children continue to enter school at the primary level until age 11. This may itself be due to the fact that school distance dissuades parents from sending very young children to school. Any analysis that includes younger children therefore is likely to include a substantial number of children who will enroll in school in the future, but currently show up as never having attended. Focusing on children age 12 and over allows a cleaner estimate o f the impact o f school location on school attendance. 3.40. Since these children are beyond the primary school entrance age, the measure o f school availability must account for whether or not there was a school at the right level and gender inthe settlement at the time the child was the right age to enter school. Since the census collected data on the year o f establishment o f all schools, such a measure can be constructed; thus, for each child, the researchers construct a variable that identity o f the hamlet in which the household and the school are located is also known. Detailed school characteristics were also collected for each school identified inthe census. lo*Sathar and others (2003); Mumtaz and Raouf (1996); Khan(1998); Khan (2000); Hennink, Rana and Iqbal(2004). 53 measures whether or not there was a primary school serving the right gender inthe settlement in which the child resides when the child was age 11 or younger. A similar measure is created for whether or not there was a middle school serving each gender inthe settlement in 2004. Results show that roughly one-half o f all children in the sample had a primary school serving each gender in the settlement when they were o f primary school age. Predictably, the ratio i s lower for girls and higher for boys, at 50 percent and 58 percent, respectively. 3.41. Among the nearly 2,000 children aged 12 - 17 in the sample, 75 percent o f boys and 52 percent Figure 3.7: Percentageof Children Who Have Ever o f girls had ever attended school. These numbers are Attended School by Settlement consistent with estimates from the PIHS as well as the earlier round of the PRHS.'03 A simple correlation with school availability shows that attendance matters significantly for girls but not for boys. School attendance for girls increases from 44 to 61 percent if there is a school inside the settlement, while the presence o f a school inside the settlement has virtually no impact on school attendance for boys (see Figure 3.7) `04 Girls Boys 3.42. To look at whether this school location nSchool InsideSettlement School Outside Settlement effect survives when accounting for other child, Source: Cross-tabulations from Pakistan Rural Household Survey, household, and community characteristics (such as 2004. the child's ape, distance to the nearest available school, parent's education, and household wealth), PRHS-I1 data was used to conduct a multivariate regression analysis. Since school attendance also can be affected by household preferences, a cleaner test o f the school location effect is to compare school attendance for opposite sex siblings. This washes out the impact o f all household characteristics, like wealth or preferences that may affect attendance decisions for both boys and girls within a household. For within-household comparison o f siblings, the analysis shows that school location has no impact on school attendance decisions for sons, but has a positive and large impact on school attendance decisions for daughters in the same age group. Moreover, even after controlling for attendance due to the existence o f a primary school inside the settlement, the presence o f a middle school insidethe settlement has a further positive impact on attendance that is large and significant. 3.43. These substantial school location effects suggests that any measure to improve school access will yield large returns. Clearly, while building schools-private and public-remains important, building a school in every settlement i s unlikely to occur. Qualitative studies suggest that the real issue is not one o f distance per se but one of safety and o f prevailin cultural norms around the appropriate chaperoning o f young women. Related work usingthe same data,B data from PRHS-I1 (see also Chapter 2), suggests that even adult married women feel significantly less secure when they need to cross the boundaries o f their own settlements-even during the day.*06Only 40 percent o f women reported feeling safe walking alone outside their settlements during the day; however, over 80 percent reported feeling safe walking alone inside their own settlement (Table 3.6). Strategies to reduce the costs (social or financial) o f getting girls to I O 3See also the PakistanPoverty Assessment. IO4 Ongoing work also looks at the impact of the social composition (mainly zaatibiradari) of the community on school enrollment using the notion of "social distance." Similar issues for girls schooling with respect to caste arise in India as well (PROBE Report 1999). I O 5The sample consists of approximately 1,600 currently married (including divorced or separated) women aged 15-40 from 94 villages in Sindh andPunjab. Settlements are distinct habitations or communities within villages. They can be thought of somewhat as more organic entities within the bounds of revenue villages. Most have distinct names and boundaries. The revenue village in contrast is a largely administrative construct. 54 school and back safely may substantially enhance school attendance and retention. We discuss two such strategies inthe final section, which identifies policy initiatives to increase schooling. The first i s the use o f chaperones to accompany girls to school and back. This could be a relatively inexpensive strategy. A second alternative could be to provide subsidized school transportation. Table 3.6: Women's Perceptionsof Safety (in percentages) ~ Feeling of safety when walking alone in the day Province Safe Unsafe Total Whilewithin settlement North Punjab 62.9 37.1 100.0 SouthPunjab 82.0 18.0 100.0 Sindh 90.2 9.8 100.0 Total 81.8 18.2 100.0 While outside settlement North Punjab 33.7 66.3 100.0 South Punjab 54.0 46.0 100.0 Sindh 35.6 64.4 100.0 Total 40.0 60.0 100.0 Note: These figures represent percentageof currently marriedwomen aged 15-40 who reportedfeeling safe and unsafe within and outsidethe settlement. Source: Pakistan Rural HouseholdSurvey -11 (2004). This survey covered rural households in Punjab IV. WILL BUILDING SCHOOLSINCREASE ENROLLMENT? 3.44. The physical and social costs o f going to schools far from the home are a major deterrent to female participation in education. More school construction is not a complete solution to this problem, particularly for girls living in small scattered settlements, as is common in the villages o f Southern Punjab and Sindh. However, buildingmore schools i s likely to be a very important piece o f any strategy to enhance enrollment of both boys and girls. If schools are built, what will it take to make them functional? A bindingconstraint on running schools is the availability o f teachers who will live inthe village where the school is located and that they show up for work on a (somewhat) regular basis. 3.45. The average rural PSU in the 2001-02 PIHS had 1.7 women who had completed class 8 and only 1.1 matriculated women (completed class IO); 47 percent o f all rural PSUs did not have a single woman with an 8* grade education, while 58 percent had no women with a matriculate degree (Table 3.7). Across provinces, Punjab does relatively better (2.1 average and 40 percent reporting zeroes); predictably Balochistan (0.7 grade 8, 77 percent reporting zeroes) and Sindh (0.9 women with grade 8 education and 59 percent reporting zeroes) are much worse. Put simply, most villages have few or no resident women who could be hiredas teachers if a school were to be built. 55 Table 3.7: Availability of Educated Females A. Availability of qualified women major four provinces ~ Completed Grade8 CompletedGrade I O Percentof Average no. of Percentof qualifiedwomen Averageno*Of communitieswith qualified women communities with inacommunity at least one at least one qualified woman inacommunity qualified woman Urban 6.3 92% 5.0 88% Rural 1.7 53% 1.1 42% B. Average number of qualified women in a community CompletedGrade8 CompletedGrade I O Region Urban Rural Urban Rural Punjab 6.5 2.1 5.0 1.3 Sindh 6.6 0.9 5.5 0.1 NWFP 4.2 1.2 3.2 0.9 Balochistan 3.1 0.7 3.1 0.5 C.The share of communitieswith at least one qualified woman CompletedGrade8 Completed Grade I O Region Urban Rural Urban Rural Punjab ' 95 60 94 46 Sindh 87 41 83 35 NWFP 86 46 75 38 Balochistan 77 23 74 17 Note:. "Qualified women" denotes women between ages 18 and 50 who have completed at least grade 8 (middle school) or grade lO(high school). Source: PIHS 2001-02 householddata. The availability of educated women 3.46. There are a couple o f reasons why we should worry about the availability o f educated women, as opposed to educated men as potential teachers. First, all teachers (by explicit rules o f the education department) in government schools for girls must be female. Part o f the reason for this rule is presumably that parents feel more secure in sending their girl children to schools that are staffed by females. Second, in the case o f private schools that are co-educational, there is a strong profit motive that makes it more beneficial to hire female rather than male teachers. This profit motive derives from strong cultural constraints to participation in the labor force and consequently a lower wage for equivalently educated females compared to males. Private schools overwhelmingly hire female teachers because they can pay them (very) little and get away with it; most likely (and we return to this below) they will not set up in villages where they cannot find such women. 3.47. There also are a couple of reasons why we should worry about the availability o f educated women in a particular village rather than in the area surrounding it. If labor markets are well integrated, it should 56 not make a difference whether a sufficient number o f educated women are resident in any given village. As we have discussed above, both quantitative and qualitative studies suggest that security concerns and other cultural constraints make it difficult for rural women to travel outside their own village for a day job. Women rarely relocate for work. Rural-to-rural relocation i s generally small, and when it occurs, it is in response to marriage rather than to employment opportunities. An alternative mechanism through which educated women could disperse to villages with low populations o f educated women i s thus through marriage. However, over one-half o f all marriages in rural Pakistan are village endogamous (within the village),lo7 which limits this possibility. 3.48. Any strategy seeking to expand school access at the primary level, particularly one that seeks to involve the private sector, will require complementary public investments in middle and high schools for girls as well as incentive mechanisms that ease school access. The Social Assistance Program (SAP) during the 1990s was an attempt to bridge this gap. Although this program was heavily criticized and there were severe problems in the field (ghost schools and teachers, and very poor learning), ultimately it did manage to increase the percentage o f educated women in Pakistani villages by a small margin. These women were born in the village where they were educated and were likely to remain in the same village. It is plausible that the dramatic expansion in rural private schools since the mid-1990s was attributable to this development. That first cohort o f educated women i s an important start. Villageswith educated women also haveprivateschools 3.49. Analysis o f PIHS 2001-02 data clearly suggests that there is a causal relationship between villages with education women and private schools (Table A3.7). The supply o f educated women is higher in rural communities with access to a high school for girls. In communities where a public high school for girls i s located five kilometers or less from the community, about 9 percent o f women aged 20-44 have matriculate (class 10) or higher education. In contrast, communities where a girls' high school i s more than five kilometers away, the supply o f educated women i s lower, with only 4 percent o f women aged 20-44 with matriculate or higher schooling. Women with a matriculate degree or more also appear to be more active in the non-agricultural labor force. Among rural women aged 20-44 with matriculate or more, 24 percent report that they are working. Among these working women, more than one-half report teaching as their occupation. 3.50. This analysis is very suggestive, but the PIHS does not allow us to control for the year o f school establishment.'0* Work by Andrabi, Das, and Khwaja (2005) strengthens this argument. They match villages across the census years o f 1980 and 2000, showing that villages receiving a government girls' primary and secondary school during the two intervening decades had increased chances o f attracting a private school by almost 100 percent, compared to villages that received only a government girls' primary school. Furthemore, this study shows no baseline differences between villages that received a primary and secondary school and those that received only a primary school in terms o f educational levels, although those that received both had larger populations (Table 3.2). The authors argue that this result is driven primarily through the larger number o f women with matriculate degrees who then became private school teachers inthe villages that have received both schools. Since female mobility is low, cohorts of educated women needto be built up in everyvillage 3.51. There is little observed migration among educated women in rural Pakistan. Data from the PRHS (2004) shows that marriage is largely village endogamous. Most married women report being born in the ~ lo'Basedon PRHS-11. We do not have data on precisely when these schools were built, but we do know that most private schools were established relatively recently and are likely to post-date the public high school. The new round of data includes a complete census o f all schools ineach sample village and we should be able to answer this question more precisely once that data i s analyzed. 57 village in which they reside and most have natal families in the same village. Among currently married female migrants, almost all (98 percent) report migrating due to marriage or to join a family member after marriage and not in response to employment opportunities. While most rural women aged 20-44 with a matriculate degree or more work in teaching or other formal occupations, the PIHS data do not tell us whether they are employed within the village or outside village. The PRHS (2001) asked women engaged in paid work where they worked. Almost 60 percent of women who were engaged in non-farm work reported working within the village. Giventhis pattern o f marriage and migration, the diffusion o f educated women via marriage i s not likely. Neither can we expect that educated rural women will migrate in response to employment opportunities. 3.52. These facts suggest that the biggest constraint to providing better schooling is the availability o f female teachers and, conversely, alleviating this constraint leads to a private sector response that will plausibly create a virtuous cycle inthe long run.Thus, any strategy, which seeks to expand school access at the primary level, particularly one that seeks to involve the private sector, will require complementary public investments in augmenting school access for girls at the middle and high school levels, We need to create that first cohort o f educated women inevery village. 3.53. To reiterate, public investments in building middle and high schools, where feasible, will have big payoffs; however, it never will be feasible to build schools in every small and isolated settlement. Strategies that reduce the social and financial costs o f attendance thus are likely to have high payoffs, particularly in areas which are relatively remote, relatively poor, and where villages are organized around numerous settlements. On a final note, it also is important to note that there are some important additional constraints to school functioning and quality that will impact school demand. Private schools prefer female teachers and pay them less 3.54. Educated women remain a somewhat captive labor force in the villages where they are educated. This is reflected in their wages to a rather startling degree. Andrabi, Das, and Khwaja (2005)'09report on salary differentials among male and female teachers in public and private schools. An average female teacher in a government school earns a salary o f Rs.5,710 per month-not very different from the earnings of an average male (Rs.6,143). Among private schools, male teachers earn close to Rs.2,000 per month, while females earn only half as much at Rs.1,100. Private schools clearly pay female teachers substantially less than government schools do. Much o f this gender difference in teacher salaries no doubt captures differences in educational attainment and training as well as job tenure. Even when controlling for education, educational training, and experience o f the teacher, the female penalty remains strong in private schools, but vanishes for government schools. Among private schools, female teachers earn on average Rs.650 less per month than their male counterparts, which is close to 33 percent o f the average male wage. When controlling for labor market conditions in the village by looking only at differences within villages, they find no difference in the estimated coefficients, which remains at around Rs.600, for both estimates within villages and within schools. This difference suggests a pure "gender" penalty rather than other variables related to overall labor market conditions. Private schools seemed to have capitalized on this captive labor force since the bulk (over three-fourths) o f teachers in private schools in the sample are female, while only 44 percent o f teachers inpublic schools are female. 3.55. This wage differential extends to other paidwork as well. Multivariate regression analysis usingthe nationally representative PIHS 2001-02 data shows that women earn significantly less than men, even after controlling for the age and education o f the worker. A man with a primary school education is paid an average salary o f about Rs. 2,892 per month, while a woman with a similar educational profile is paid Rs. IO9Using recent datathey collected from over 800 schools (5,000 teachers) in 112villages of Punjab as part ofthe LEAPS study. 58 879. At the secondary-school level, these differences are somewhat attenuated but still large and significant (Rs.3,830 per month for men versus Rs. 1,922 per month for women). When female teachers work in government schools, they are absent more often 3.56. Although women in government schools are paid almost five times more than their private school counterparts, even in the same village, they are absent more often. Andrabi, Das and Khwaja (2005) find that absenteeism across teachers in public and private schools i s 2.6 days during the month (close to 12 percent o f all working days). Teachers in private schools are absent significantly less (1-87days) compared to teachers in government schools (3.18 days). Further, even within government schools, female teachers are the biggest defaulters. An average female teacher in a government school i s absent close to four days a month (close to 20 percent o f all working days) comparedto 2.65 days for her male counterpart"' 3.57. The welfare consequences o f these absences are less certain. InAndrabi, Das and Khwaja's study, a large fraction o f teachers took time o f f due to health-related reasons among members o f their families and other family emergencies. With limited options in terms o f substitute teachers with the same educational background (government female teachers are more educated and better trained than their private school counterparts), the welfare benefits o f (at least) having a teacher in the school 16 days a month may outweigh the costs o f having no one at all. Evenwhen private schools hire teachers, turn-over costs are high 3.58. The LEAPS survey, there were 2,186 teachers in the 311 private schools surveyed, for an average of seven teachers per school. In the two years preceding the survey, close to 500 had left the school, representing a turnover rate o f close to 25 percent every two years. The turnover numbers for government schools were much lower (184) teachers, but still fairly high. Most teachers who left the school in the private sector were women (75 percent), and the lion's share left due to family reasons (marriage, domestic problems, health issues). Such high turnover plays havoc with the runningo f schools, another argument in favor o f creating a reasonably large base o f educated females from which the pool o f teachers can be drawn. v. RAISINGLEVELSGIRLS'SCHOOLING: SOME POLICY RECOMMENDATIONS OF 3.59. Access to schools has to be improved to attract rural female students and retain them in school beyond the primary level. This includes improvements in the quantity and quality o f schools and adapting service delivery to the special needs of female students. Federal and provincial governments can learn much from the experience o f public and private schools, and from parents' responses to them in taking steps to address the large gender gap. Increaseaccess to localpublic schools 3.60. School proximity is critical to girls' enrollment, and a significant proportion o f rural communities do not have a school nearby enough for girls. At the primary school level, the ESR action plan (200 1-2005) was a move in the right direction. This plan aims to expand the number o f schools on the basis o f gender and populationrequirements for schools. 3.6 1. More private primary schools are unlikely to provide a solution to improving female enrollment, given their existing pattern o f expansion. Because private schools require female teachers, they cluster in areas with pre-existing public schools. Because the odds o f attracting a private school increases 'loThese resultswere also confirmedin amultivariate regression analysis. 59 substantially with an increase inthe number o f public girls' schools, without the government taking the first step, the private sector will not move in. To improve the quality and quantity o f primary schooling through the private sector, substantial public investments in middle and high schools are required, particularly in areas which are lagging behind in school availability. Less than a one-quarter o f rural communities have a nearby public middle school for girls, and only 10 percent have a nearby high school for girls. Girls' enrollment at this stage appears to be hampered by parents' concerns about letting adolescent girls attend schools that often are outside the community. 3.62. It is probably not practicalor possible to place a public primary school and middle or high schools inevery rural community. Policymakers should consider cost-effective alternatives that compensate for the constraint o f distance. At the primary level, the present policy o f allowing girls to attend existing boys' schools where no primary schools for girls are available is an effective way to expand school access for girls. Indeed, co-educational private primary schools' tendency to attract young girls shows that parents are not averse to sending girls to co-educational schools at the primary school level. In communities where single-sex schools are far away and girls have to travel alone to the schools, parents may prefer to send their daughters to co-educational institutionsto which siblings can travel together. Attend to specificconstraints: increasenumbersof female teachers 3.63. The lack o f potential female teachers in Pakistan, which government schools require and private schools prefer, is an significant constraint. Schools will continue to have trouble recruiting and retaining female teachers-particularly in the least-served areas-until local cohorts o f female teachers emerge or significant barriers to female mobility that prevent educated women from relocating or commuting to schools that need them are overcome. Because near-term policy levers are unlikely to significantly reduce these barriers, policymakers should focus on interventions that are complementary to those that increase girls' access to schooling. Rural communities need to develop their own cohorts o f educated women in order to compensate for the restrictions on bringing in female teachers from outside the community."' By enhancing local girls' access to all levels o f schooling, these communities can begin building a ready pool from which to draw the teachers o f the next generation. 3.64. The Community Support Program (CSP) in rural Balochistan is an example o f a program that specifically addressed the shortage o f female teachers. Carried out in three divisions, the CSP experiment provided communities with a school in the village, ensured the presence o f a female teacher, and encouraged the communities' involvement in running the school. The CSP set up government-funded community schools. A female teacher was selected through a village education committee set up as part o f the program. Due to the short supply o f educated females, the educational qualifications were relaxed relative to the standard requirement for a government teacher. A woman was eligible to teach if she had a minimum of eight years o f schooling and was a resident of the same village or lived within walking distance o f the village. The teachers were given in-service training to make up for lack o f educational qualification. Evaluation o f the CSP found that the program had a large impact on increasing girls' school enrollment. 11* 3.65. Other approaches to alleviate shortages o f teachers could include drawing from the pool o f educated women from nearby communities. However, this may require assurance o f safe transportation to and from school, or providing residential accommodation with assured security. Enhancedemand-side incentives for girls' education ``IWhile urban teachers can be assigned to teach in rural areas, teacher absenteeism tends to increase with the remoteness o f the school. `I2Kim,Alderman and Orazem(1998). 60 3.66. Several demand-side, initiatives such as the middle school stipend program and the school meal program (Tawana Pakistan) are already underway. Such schemes rely on the idea that low enrollments are primarily due to financial constraints. Without doubt such constraints are likely to be quite important for many rural households. Even where households are concerned about the safety o f young girls, a stipend could allow them to purchase private secure transportation to and from school. Our analysis indicates, however, that safety concerns are not likely to be addressed completely by reliance on existing means o f transportation available inmost villages. 3.67. A rigorous evaluation o f the stipend program is therefore urgently needed to identify whether the stipend should be pegged to school distance as well as the extent to which uptake o f the program is dampened by safety and mobility concerns. If stipends alone are found to be inadequate, several complementary initiatives could be considered to augment their effectiveness. For example, where a school exists within walking distance, trained and licensed chaperones could be used to escort young girls to and from school. Where schools are at a greater distance, as i s the case with most middle and high schools, subsidized provision o f secure school transport could also be considered. The feasibility o f such complementary initiatives could be assessed as part o f an evaluation though pilot schemes. 61 4. 1i"ROVING HEALTH OUTCOMES FORWOMEN AND CHILDREN The State shallprovide basic necessities of life, such asfood, clothing, housing, education and medical relief for all citizens, irrespective of sex, caste, creed or race, as arepermanently or temporarily unable to earn their livelihood on account of injrmity, -Article sickness or unemployment. 38(d) of the Constitution ofPakistan 4.1. Health outcomes have improved significantly in Pakistan Table 4.1: Life Expectancyand Infant Mortality: Comparisonsacross Regionsand Over Time since the early 1950s (Table 4.1). Estimated life expectancy at birth has Life expectancy Infant risen from 43 to 63 years, and infant at birth mortality mortality has fallen from 169 to 79 per 1950-55 2000-05 1950-55 2000-05 1,000 live births. Fertility levels have Pakistan 43.4 62.9 168.6 78.6 fallen by 24 percent between 1975 and Bangladesh 37.5 62.6 200.5 58.8 2000,"3 which itself reduces maternal South-central depletion and mortality and improves Asia 39.4 63.0 187.0 69.1 women's health outcomes. WesternAsia 45.2 67.8 190.6 48.0 Less-developed 4.2. Women's health and regions 40.9 62.8 179.8 62.4 reproductive outcomes have improved Source: PoDulationDivisionof the Deoartmento f Economic and Social Affairs o fthe substantially, as reflected in various UnitedNaGonsSecretariat, WorldPopulation Prospects: The2002Revision and World UrbanizationProspects: The2001 Revision, httu://esa.un.ore/unuD.See also the data indicators. In tandem with fertility from the PakistanDemographicandHealthSurvey andthe PakistanReproductive decline, the gender gap in life HealthandFamilyPlanningSurvey. expectancy has closed, and estimated life expectancy is now higher for females than males. Although the sex ratio has improved over time, it still remains high for the population as whole. Women's age at marriage has also risen sharply, from an estimated 17 years in 1951 to 22 years in 2002.114 Along with greater control over their own reproduction comes greater control over their own lives. 4.3. Improved health and reproductive outcomes have many positive spin-offs for the society as a whole, as well as for the individuals and households concerned. For example, improved health and nutrition is associated with better cognitive development and school performance' l5 and increases the probability o f enrolling in school, especially for girls in Pakistan.'16 School nutrition programs, such as the Tawana program, can help narrow the gender gap in schooling.'17 Better health and nutrition are also associated with higher earnings, as studies inrural Pakistan and elsewhere show."* 'I3NationalInstituteof PopulationStudies (NIPS) (2001) on the Pakistan Reproductiveand Health and Family PlanningSurvey (2000).The total fertility rate (TFR) between 1970-75, based on the 1975 PakistanFertility Survey, was 6.27. The Reproductive Health andFamilyPlanning Survey (2000)estimatedthe TFR during 1997-2000to be 4.8, which representsa 24 percent decline. 'I4Governmentof Pakistan,http://www.mopw.gov.pkipublications/ppla~Chap2,htm, 115Alderman, Behrman,Lavy and Menon (2001); Government of Pakistan(2003); Siddiqi, Haq, Ghaffar, Akhtar, Ali and Larik (2003),Sathar and Casterline (1998); Tinker (1998), World Bank 2002, Miguel and Kremer (2001). Alderman, Behrman, Lavy, and Menon(2001). I" See Chapter 3 for a description of program. 'I8Alderman andothers (1996). 4.4. However, Pakistan's I Table 4.2: Under-five Mortality and LifeExpectancy: demographic indicators still lag behind Comparisons across Regions and by Sex those o f neighboring countries in West Under 5 Life expectancy Asia and South Asia, and in the less- mortality at birth developed world as a whole (Tables 4.1 Male + and 4.2). Female Male Female Male Female Many o f these countries experienced more rapid mortality Pakistan 114 109 119 62.7 63.1 declines over the past half-century, as Bangladesh 79 79 19 61.8 63.4 well as earlier initiation o f fertility South-central Asia 100 98 103 61.6 64.5 decline (Figure 4.1). This partially Western Asia 61 65 56 65.8 70.0 Less-developed explains the observed advantage in regions 94 95 93 61.2 64.6 female estimated life expectancy Source: Population Division of the Department of Economic and Social Affairs of today. l9 the United Nations Secretariaf World Population Prospects: The 2002 Revision and World Urbanization Prospects: The 2001 Revision, http://csa.un.orplunnp. 4.5. Three factors interact to slow the improvement in health outcomes in , Figure 4.1: Total Fertility Rate, 1950-2005 Pakistan, as in some other parts o f South 7 Asia. First, high rates o f poverty make ' : 6.5 1 for poor nutrition and health 6 - ~ 5.5 conditions.'20 This is aggravated by 5 - neglect o f public health and 4.5 - environmental sanitation services. Poor 4 - people in particular - given their living 3.5 - 3 - and working conditions -are frequently ~ 2 5 j exposed t o infection. Second, the coverage and quality o f publicly provided 1950- 1955- 1960- 1965- 1970- 1975- 1980- 1985- 1990- 1995- 2000- 1955 1960 1965 1970 1975 1960 1965 1990 1995 2000 2005 health services i s poor. Third, gender ineauities place constraints on women's +Pakistan +Bangladesh South-central Asia . P ~Western Asia -Less developed regions and girls' to information ' and services. Of Course, the first two Source: Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, World Population Prospects: The 2002 issues are not exclusive to women and Revision and World Urbanization Prospects: The -2001 Revision, have been analyzed elsewhere.I2l Here http:Nesa.un.or~unnn. we focus on the specific constraints to improving women's health outcomes. 4.6. The Pakistan government has consistently demonstrated clear political will to enhance gender equality and improve health outcomes. For example, one o f the objectives o f the National Health Policy i s to promote gender equality in health. Pakistan has committed itself to meeting all eight MDGs, two o f which focus on health (reduce child mortality and improve maternal health).'22 To further these goals, the government will have to act on multiple fronts. Encouragingly, health policy changes introduced since the rnid-l990s, as well as since the introduction ofthe devolution reforms, suggest that improvements are occurring. Outreach related to family planning services has expanded through the private sector as well as through active door-to-door campaigns by public sector personnel, in addition to the fixed-point service delivery already in place. The goal o f reducing population growth is now joined by a greater ' I 9 Female longevity i s higher than male inmost populations: life tables derived from a large number o f countries indicate that at Pakistan's level o f mortality, the life expectancy for women is typically 3.7 years higher than that o f men (Coale and Demeny lifetables). I 2 OOne-third of the population was estimated to be poor at the end o f the 1990s (World Bank 2002). World Bank (2002). 12' For a detailed list of the MDGs go to http://ddp-ext.worldbank.org/ext/MDG/home.do 63 emphasis on providing services to meet women's needs through more integrated functioning o f the Ministries o f Health and o f PopulationWelfare.'23 4.7. We investigate the determinants o f women's health-including the availability and use o f health facilities and programeas well as the effects o f recent health policy changes. Because rural girls and women face more pronounced health disadvantages than their urban counterparts, most o f our analysis focuses on rural areas.124We summarize the causes o f the problem, its dimensions, and the nature o f the available health infrastructure in the public sector, followed by an analysis o f what policies seem to be working, and recommendations regarding any gaps to be filled. This chapter examines the effect of recent policies and programs in depth and, after some assessment, makes recommendations for further action. 4.8. The preceding chapters have discussed the problems that female seclusion and constraints on mobility pose, not only for women's quality o f life, but also for their ability to access public services such as schooling. Here we discuss how these same factors hinder women's access to health services, and how these barriers can be overcome. Our analysis suggests that much can be done to offset the problems o f female seclusion and low education through efforts to increase service outreach and efforts to increase the demand for services through disseminating health information. Pakistan has already begun to put such programs in place successfully, and we offer suggestions to enhance their impact on women's health. I.GENDER-RELATED CONSTRAINTS TO ACCESSINGHEALTH CARE: MOBILITY, DECISION- MAKING, AND ACCESS INFORMATION TO 4.9. Gender inequities restrict women's access to health services in a variety o f ways. Women face social constraints in managing their own health and that o f their children, although they are largely responsible for domestic management o f health: preventing disease by good health and hygiene practices; recognizing illness early and providing home care; seeking medical care when needed; and interpreting and implementing medical instructions. In order to take effective care o f their own and their children's health, women needto be well-informed and to be able to act quickly on their perceptions. 4.10. Chapter 3 shows how women and girls' limited mobility constrains female schooling. A number o f studies on women's access to health services in South Asia emphasize women's restricted mobility as a ~0nstraint.l~~Survey data from rural Pakistan depict the restrictions on travel to health facilities that women face (Table 4.3). The majority o f women report they are unable to a health facility unaccompanied. Indeed, for women and girls, having family members (especially male members) accompany them to health facilities constitutes "social resources"126 that can greatly improve their utilization o f health services. 4.11. A qualitative study on gender conducted inrural areas o f Punjab and Sindh as part of this Gender Assessment (Box 1.3) found that women spontaneously raised concerns about access to health services, even though the study did not directly ask about this. As many as 40 percent o f respondents stated that the primary constraint to accessing health services was their mobility, and fewer (27 percent) stated that the primary constraint was proximity to the facility. Difficulties getting to the health facility included Sathar (2001); Sultan, Cleland, and Ali (2002). This is facilitated by the availability of detailed information inthe PIHS survey, and other surveys. Dysonand Moore (1983); Durrant and Sathar (2000); Schuler, Hashemi,and Riley (1997); Khan (1998); Mumtaz and Salway (2005). Mumtaz and Salway (2005). They analyze how restricted mobility influences women's access to health facilities. They combine a detailed ethnographicstudy from rural Punjab with a nationally representativesurvey data from PakistanFertility and Family Planning Survey (1997). 64 having to be accompanied by the husband or mother-in-law-even if the treatment was for her children and not for herself. Ifthere was an emergency health situation and no one around to accompany them, some women said that as a last resort they were permitted to venture outside the household only if they took along one o f their children as a guarantee o f proper conduct. 4.12. Restrictions on women traveling to a facility do not disappear ifa health facility is Table 4.3: Percentageof Rural Women Reporting nearby, but they do become less stringent with RestrictedAccess to Health Facilities proximity (Table 4.3). A number o f studies from Cannot GoAlone Need Permission Pakistan document similar patterns in women's ~~~~~ltime to health-seeking ability.12' This mirrors the faciliw: lhour >=1 hour lhour >=Ihour negative impact on girls' schooling ifa school i s Hospital 65 78 81 91 located outside the village or settlement boundary: Rural health center 49 74 66 89 rural women and girls face the most stringent ~~~i~ unit 62 health 82 84 93 restrictions on mobility if accessing a health care ~i~~~~~~~71 87 88 94 provider outside the village.'28 Access to doctors Private by unmarried girls tends to be even more hospitaliclinic 49 72 71 88 circumscribed as they cannot go unaccompanied Private doctor 61 87 85 94 by their parents. Moreover, frequent visits to the Hakim 50 60 75 85 doctor by girls has a negatively affect on the Homeopath 27 40 56 80 family's reputation in the community.129 Pharmacy 61 67 78 86 Note: Cross-tabulationsfor women aged 15-29from aquestion asking 4. 3. Another cannot quickly respondenfs ifthey could travel to thefacility alone andwhether they seek health care is that they are typically not neededpermissionfrom someone inthe householdto go tothe facility. Source: PakistanRural HouseholdSurvey (2001). lZ7 "* Sathar and Kazi (1997); Khan, 1998; Mumtaz and Salway (2005). Basedon a qualitative study from three villages inrural Punjab (Khan 1998). Khan(1998). I 3 OKhan(1998). 13' Pakistan ReproductiveHealth and Family Planning Survey 2000-01, National Institute of Population Studies (2001). 13' Munshi and Myaux (1998); Montgomery, Casterline, and Heiland(2001) 65 persuade their husband and/or elders that care is needed, obtain permission to seek care, and find someone to accompany them. They also have to know when to seek medical care and what health services are supposed to be available to them. We discuss below how some o f these hurdles can be overcome to increase effective health-seeking behavior. 11.WOMEN'S HEALTH: DIMENSIONS THE PROBLEM THE OF Early Childhood 4.16. Infant and under-five mortality rates Figure 4.2: Women with More Sons Are More Likelyto remain high in Pakistan, for both boys and girls Use Contraceptives (Tables 4.1 and 4.2). But girls suffer 30% , disproportionately high mortality. Although mortality i s usually higher for males, the reverse holds in Pakistan (Table 4.2). While mortality estimates from national and United Nations sources differ (see Table 1.1, Chapter l), they show similar patterns by gender. 4.17. The biological advantage for female survival is offset by cultural patterns o f son I 0 1 2 3 4 5~ I preference and attendant underinvestment in girls. Nuher Women report a preference for boys over girls, and this is mirrored in actual family-building Source: Cross-tabulationso f PakistanIntegratedHouseholdSurvey 2001-02data for currently marriedwomen aged 15-49 in rural areas, behavior. Women with relatively more daughters who are not currentlypregnant. than sons are more likely to want more children and less likely to practice contraception (Figure 4.2 and Table A4.3). Figure 4.3: Gender Differences in Probability of Consulting a Doctor in Case of Illness 4.18. Son preference is reflected also in gender differentials in child care, which are significant zp 20% - I even controlling for household socioeconomic status, parental education, and distance to health facilities. First, girls' illnesses are significantly less likely to be rep01-ted.l~~Second, those whose illnesses are reported are significantly less likely to be taken for a medical consultation (Figure g 0 % 4 I , , , , , , , , , , , , , , , 4.3). Third, even if girls are taken for a 0 Z b 6 0 \Q \Z \b \6 consultation, less is spent on their medical care Age (Years) than on care for boys (Figure 4.4). Richer &Boy--EGirl households show greater gender discrimination in Source: Pakistan Rural Household Survey 2001. The probability of medical expenditures than do poorer households consulting a doctor is predictedusing regressionresults showninTable (Table A4.1). Another study in Pakistan found A.4.1. that rural households were more likely to consult 133The problems o f under-reporting in self-reported morbidity data are well-documented (see for example Murray and Chen). We find that only 11 percent o f children aged 0-17 years were reported to have been illduring the year preceding the survey, which implies significant under-reporting. However, this does not necessarily affect our analysis, because there is no reason why there should be differential under-reporting by the gender of the child; such a differential should reflect parents' lesser concern about the illness of children of one gender. 66 private doctors (considered to be o f higher Figure 4.4: Rural Households' Annual Average Medical quality) for boys than for girls.'34They also Expenditure by Age and Sex found that the use o f medical care for girls was more sensitive to the price o f services. 1 10000 n l - ~ 8000 1 4.19. Despite these household preferences, government efforts to improve child heath outcomes, as well as gender differentials in child health outcomes, are taking effect. Childhood immunization coverage rose 0-6 7-13 14-17 25-49 significantly during the 1990s (Figure 4.5a), as a result o f the efforts undertaken through the Age Expanded Program on Immunization (EPI). By 2001-02, over one-half o f children aged 12-23 months were "fully immunized," i.e., they had Note The genderchfferencesinexpenditures are statistically received the full course o f recommended significant Source Cross-tabulationsbased on PakistanRural Household Survey vaccinations against tuberculosis, diphtheria, 2001data The figuresrefer to medicalexpendtures reportedfor each householdmemberwho hadbeenillduringthe year and for whom any 4.5b).135 pertussis, tetanus, measles, and polio (Figure medical practitionerwas consulted 4.20. As immunization coverage increased during the 1990s, the gender gap in Figure 4.5a: Percentage of Children Fully Immunized immunization coverage narrowed (Figure by Gender, 1990-91 and 2001-02 4.5a).'36 Substantial gender gaps remain only in rural Sindh and urban NWFP (Figure 4.5~).This Rural 1% i s probably attributable to heightened efforts to 46% make bring free immunization to people's 8 7 Urban 71% ~ doorsteps through health worker outreach, as N 70% well as through immunization camps. Pakistan 51% 53% 4.21. When we control for household 0 % economic status and other factors, we find that Pakistan ~ 7 39% girls continue to be significantly less likely to receive measles and BCG vaccination (Table A4.2), even though these are supposed to be provided free o f charge along with the other Note The percentagesrefer to children aged12-23 monthswho have completed immunizations Polio 3 and DPT 3 refer to lastdose ofthe childhood immunizations. Also, there is still a respectiveimmunizations long way to go to reach the EPI goal o f Source PakistanDemographic and Health Survey 19%-91 and PakistanIntegratedHouseholdSurvey 2001-02 universal immunization. In the rural areas o f Balochistan less than one-quarter of children were fully immunized (Figure 4.5~). 4.22. Levels o f undernutrition are also very high among children, as indicated by their height and weight at given ages. The Pakistan Rural Household Survey 2001 shows that high proportions o f rural children under age five are malnourished, regardless o f gender. Nearly one-half o f these children are 134 Alderman and Gertler (1997). 13' The government-initiated Expanded Program on Immunization (EPI) o f 1982 recommended that all children be immunized against the six preventable childhood diseases by the age o f 12 months. The recommended vaccinations include one dose o f BCG (against tuberculosis), three doses o f DPT (against diphtheria, pertussis, and tetanus), four doses of polio vaccine, and one dose o f measles vaccine. 136 See also Hazarika (2000). 67 underweight, and nearly two-thirds are Figure4.5b: Percentageof ChildrenFullyImmunizedby stunted (Table 1.2). The prevalence o f Gender and Type of Immunization,2001-02 malnutrition is similar for girls and boys. 4.23. Our results are consistent with tvkasles 73% 74% several studies in Pakistan and elsewhere in South Asia, which find significant I 1 r;R% gender differentials in medical care but BCG 85% % not in child nutritional outcomes.'37 Excess female child mortality has also been documented in other parts o f South Polio 3 Asia and East Asia, notably India and China, but there i s no clear evidence that 1 DPT3 son preference in Pakistan has yet manifested in sex-selective abortions or female infanticide.' 38 Women's Health Figure 4.5~:Percentageof ChildrenFully Immunized by Province,2001-02 4.24. Maternal health outcomes are poor inPakistan, as in some other parts o f South Asia. For example, the maternal mortality ratio is estimated to be 500 per 100,000 live births.'39 Given the relatively high number o f births per women, this translates into a lifetime risk of dying o f one in 38-that is, it i s 81% estimated that one o f every 38 women dies due to causes related to childbirth. c z -2 High levels of fertility are a major 66% 4% contributor to poor maternal and child health because repeated childbearing depletes maternal resources and increases 81% the risk o f ill-health for the mother. In Pakistan, the Total Fertility Rate has fallen gradually since the 1990s, but i s still fairly high (Figure 4. l).140 Source: PakistanDemographicand HealthSurvey 1990-91and PakistanIntegrated 4.25. Physical depletion from repeated HouseholdSurvey 2001-02. Note: The percentages refer to children aged 12-23 months who have completed childbearing i s superimposed on a immunizations. Polio 3 and DPT 3 refer to last dose o f the respective immunizations. 13' Chen, Haq, and D'Souza (1981); Das Gupta (1987). Also see Chapter 1. The multivariate regression analysis of nutritional status is available on request. Other studies in Pakistan have obtained the same results: Hazarika (2000), Strauss and Thomas (1995), Behrman (1992). One might expect health disadvantagesto show up as stunting and low weight, but this is not the case in Pakistan. This may be because severely disadvantaged girls drop out of the population, as implied by the recorded excess female child mortality. Anthropometric standards are age sensitive, moreover, so differentials in age misreportingby gender couldresult in girls appearing less malnourishedthan they really are. 13* See Annex 4.2 to this chapter. 13'The estimates in this paragraph are taken from WHO, UNICEF and UNFPA (2004), Table 4 and Annex Table G. Due to the paucity of data, estimates of maternal mortality for most developing countries are subject to a wide confidence interval.However, it seems clear that the ratiosare very high inPakistanby world standards. I4OAccording to the NIPS (2001), which is based on Pakistan Reproductive Health Survey, the TFR at the end of the 1990swas just under five births per woman. See also Sathar and Casterline (1998). 68 population with poor health and nutrition. Women enter their childbearing years bearing the scars o f childhood under-nutrition as well as underinvestment in girls' health during childhood and adolescence. Poor health conditions at the start o f childbearing are further exacerbated by neglect o f women's dietary needs during childbearing: evidence from both Pakistan and Punjab state o f India indicates that women's nutritional intakes do not rise as needed duringpregnancy, and especially duringla~tation.'~' 4.26. Levels o f anemia are high inthe PO ulation as a whole, except among adult urban males, as found in a survey in the early 1990s (Table 1.2).12 There is a sharp gender gap in the prevalence o f anemia among adolescents and adults-perhaps because in additionto the anemia resulting from malnutrition and exposure to diseases such as malaria and intestinal parasites, women have additional demands on their iron supply because o f menstruation and childbearing. This gender gap is sharpest among adult women: 37 percent o f women aged 25-44 in both rural and urban areas were found to be anemic, nearly double the rate for rural males and over four times the rate for urban males. 4.27. Maternal and child health are Figure 4.6: Percentage of Women Receiving Maternal Health - greatly influenced by the quality o f care Services during pregnancy, delivery, and after delivery. This i s especially important under conditions o f repeated childbearing by women who are in poor overall health. The Pakistan Integrated Household Survey (PIHS) 2001-02 data indicate that the proportions o f women receiving good maternal care have risen slowly since the 1980s, but is far from adequate (Figure 4.6, see also Annex to Chapter 4). Only 35 percent o f women in Pakistan reported receiving prenatal care during their most recent pregnancy, which represents only a 17-perCent increase from the late 1980s. Note: These percentages are basedon cross-tabulationsfrom the PIHS data. Data The figure for urban areas is 63 percent, refer to use of maternal health services by pregnant women in the three years but Only 26 percentin areas. During precedingthe PIHS survey. Note that the categories "births at home" and "births their most recent pregnancy, 41 percent assisted by trained birth attendants" are not mutually exclusive. The percentage received Of tetanus toxoid changes in use of maternal health services are calculated by comparing 2001-62 percentages with percentagesfor 1990-91 from the Pakistan Demographic and immunization* This coverage be Health Survey Report (NIPS and Macro International, 1992) , Percentage of women gettingoostnatal care not available for 1990-91. raised, given the simplicity and Source: ?IHS12b01-02 household survey data. effectiveness o f the intervention. Encouragingly, coverage expanded during the 1990s by 40 percent overall, and by 70 percent in rural areas. 14' National Nutrition Survey(2001). Similar findings emerge from IndianPunjab (Das Gupta) 14* 1990-94National Health Survey o f Pakistan. 69 4.28. Nearly four out o f five births in Pakistan Figure 4.7: Percentageof Births Assisted by Type of during 1998-2001 took place at home (Figure Attendant, 1998-2001 4.6). In urban areas, nearly one-half o f deliveries r occurred in institutions, but in rural areas 86 45 percent o f deliveries occurred at home. The proportion o f institutional deliveries rose by only 8 percent from the late 1980s. The risks o f home delivery have potentially been diminished by programs to provide various types o f trained birth attendants. As a result, few births in urban areas o f Pakistan take place without a trained person, and only one-quarter o f births in rural areas are attended solely by family members / neighbors I (Figure 4.7). There have in particular been Urban Rural programs to train traditional midwives (dais) and doctor .TrainedBirthAttendants OTrainedOair trained birth attendants (TBAs). We have no OLady HealthV191tor ONurre Familymember+neighbor , information on the trend in the proportion o f Source: PIHS 2001-02. The data are for births in the three years prior to births attended by trained traditional midwives the survey. (dais), but there has been little overall change in Note: Dais are traditional birth attendants. While PIHS asked about the proportion o f births attended by trained birth whether Lady Health Workers attended any births, no woman reported the presenceof these workers during birth. attendants (TBAs) (Figure 4.7). Levels o f postnatal care are very low, even in urban areas (Figure4.7). 4.29. The PIHS 2001-02 data also indicate large interprovincial differences in coverage o f maternal care services (see Annex to chapter 4). Punjab i s the best served province, and Balochistan shows the lowest service coverage, followed closely by NWFP. Sindh shows the sharpest rural-urban differences, with a relatively well-served urban population (probably because o f Karachi), and a poorly served rural population. There is also some indication that the coverage o f services in Sindh has declined over time. 4.30. Insum, not only are health and nutrition levels low inPakistan, but over and above this, females face additional health disadvantages. They enter their childbearing lives carrying the burdens o f deprivation during childhood and adolescence. Their health reserves are further drained by repeated childbearing and inadequate care during pregnancy, childbirth, and the postnatal period. The resultant cumulative depletion takes its toll in high maternal morbidity and mortality, and in poor health outcomes for their children. 111. DELIVERYOF GOVERNMENT HEALTH SERVICES INRURAL PAKISTAN: THE INSTITUTIONAL SETUP 4.3 1. Health services are provided by two separate ministries in Pakistan: the Ministry o f Health and the Ministry o f Population Welfare. The latter focuses largely on the provision o f family planning and some reproductive health services. The network o f medical services provided by the Ministry o f Health includes dispensaries, Basic Health Units (BHUs), Maternal and Child Health (MCH) centers and Rural Health Centers (RHC). The facilities are linked to Tehsil Headquarters Hospital (THQ) and district headquarter hospitals (DHQ)-the secondary care facilities. The management o f services on the ground was devolved in 2001 from the provincial to the district Departments o f Health (DOH).Each district now has an Executive District Officer o f Health (EDO-H), under whom all the health facilities, including 70 district headquarters hospitals, have been placed. The procurement o f medical supplies also has been devolvedto the district g~vernment.'~~ 4.32. Typically each administrative unit or Union Council (typically covering a population o f about 10,000 individuals) has a BHU where primary health services, including maternal and child health and family planning services, are provided. M C H centers, which are fewer in number, offer midwifery services and are equipped to handle routine deliveries. Rural health centers are fairly large with 20-30 staff and act as referral centers for four to five BHUs and offer limited inpatient services and emergency care. The BHUs and RHCs are primary- or first-level care facilities and are meant to provide all maternal and child health and primary health services. Aside from the network o f medical facilities, there are a number o f national programs. These include disease control services such as the directly observed therapy for tuberculosis (TB DOTS) program.144 4.33. Maternal and child health services provided within this framework o f health facilities include female paramedics such as Lady Health Visitors and Trained BirthAttendants. In addition, the Ministry of Populationruns Family Welfare Centers that provide family planning and reproductive health services. However, access to services are limited by the coverage o f health facilities, as well as by women's mobility constraints Two important national programs seek to overcome these constraints by bringing maternal and child health services to people's doorsteps. These include the Expanded Program on Immunization (EPI) which provides immunization services through clinics and active outreach through immunization camps, and the Lady Health Workers Program (LHW), described below. Coverage and Quality of Health Services 4.34. Only 58 percent o f rural Table 4.4: Percentageof Rural Communities with Primary communities in 2001 had any type o f Health FacilitiesWithin Five Kilometers public primary health care facility within five kilometers (Table 4.4, Figure 4.8). Maternal and Furthermore, 35 percent o f communities Basic health Child Health Family Welfare had neither a nearby public health facility unit Center Center nor a LHW. The distribution o f facilities Punjab 48 18 24 between provinces is uneven. Punjab and Sindh 22 10 13 NWFP are relatively well-served, while NWFP 58 29 30 Sindh and especially Balochistan are Balochistan 17 5 3 poorly served. About one-third of rural Pakistan 43 18 21 communities have a LHW, except in Note: Figures for Pakistanincludethe territories of [spell out both AJK and Balochistan, where only 10 percent of ~ ~ ~ l ; ~ ~ ~ , communities, ~survey, r a l ' ~ ~ ~ ~ ; interprovincial differences in the distribution of health services are broadly consistent with differences in levels o f maternal care received (Figure 4.6). The distribution of rural health centers is more even across provinces, perhaps because these relatively large facilities are located in places where it i s easier to attract staff. '43World Bank (2004a). '44World Bank (2004a). 71 4.35. Even where facilities are Figure 4.8: Percentageof RuralCommunitiesWith Any available, the quality o f services Government PrimaryHealthCenter or RuralHealthCenter Within varies from good to deficient. The FiveKilometers Pakistan Poverty Assessment I outlines a number o f problems that plague the provision o f health and Rural Pakistan education services, as well as other Balochistan social services. Programs often are poorly managed and implemented, NWFP 74% and lack internal accountability as well as social accountability for Sindh quality of services. There i s little political pressure to increase Punjab I service quality, especially since I Any Primary Health Center UMore than one Primary Health Center richer people are able to access Rural Health Center quality services in the private sector. Both health and education Note:Percentages for ruralPakistaninclude the territories of [spell out] A X and FATA. Primaryhealth facilities includeBasic HealthUnits, MaternalandChildHealth Centers, FamilyWelfareCenters, and suffer from low budgetary Rural HealthCenters. Source: PIHS 2001-02 rural communities' survey. allocations, relative to those o f other developing c ~ u n t r i e s .'' ~ ~ 4.36. A variety o f problems render government health facilities o f limited value to potential users.146 Insufficient allocations for non-salary inputs result in widespread shortages o f drugs, supplies, and equipment. A shortage o f female staff affects the ability to provide health care to women. The PIHS 2001-02 found that 40 o f the 100 rural BHUs visited had a sanctioned position for a female doctor, but only three had filled the position. In the case o f midwives or Lady Health Visitors (LHV), 86 BHUs had positions sanctioned, but over one-half of these had not filled the position. Staff absenteeism i s also a problem: a study of rural primary health facilities found about 36 percent o f doctors were absent during normal duty hours.'47 4.37. A social audit of public services carried out in 2002148shows low levels of end-user satisfaction with the quality of services provided in public health facilities. Only 23 percent o f households reported overall satisfaction with these services. The most common reasons for dissatisfaction with services were (1) the health problem not being solved, (2) medicines not available at the facility, and (3) poor quality of services or staff. Only 31 percent of households reported that they usually used government health facilities. 4.38. The combination o f low access to public facilities and poor service quality at these facilities means that the effective availability o f public health care i s very low in many parts o f the country. Many households use private medical services -for example, the PRHS survey shows that two-thirds o f sick children in rural areas received private medical care - but less so for preventive services like immunization. Public facilities are more often used for preventive services such as immunization, which are provided with very active outreach. According to the PIHS survey (2001-02), less than 1 percent o f rural children aged 12-23 months received their most recent immunization from a private facility. In urban areas, this percentage was about 5 percent. In the case o f prenatal care it i s evenly divided: 40 percent o f rural women reported going to a public facility while 37 percent had gone to a private facility, For postnatal care, private facilities are preferred, perhaps because the few women who obtain this care ~ 145 World Bank (2002). 146 Pakistan: Reforming Punjab's Public Finances and Institutions, citedin World Bank (2002). 14' Parvez, Chaudhury,Rehman and Khan (1993) citedinWorld Bank, 1998 (Improving women's health inPakistan) 14* NationalReconstructionBureau(NRB), socialaudit of governanceand delivery of public services, Baseline Survey 2002. 72 are from richer households: 38 percent o f rural women reported receiving postnatal care at a private facility and 28 percent at a public facility. 4.39. The issue, then, i s not merely to increase coverage o f public facilities, but also to improve the actual availability o f services in the existing facilities. In Pakistan, the process o f translating physical proximity into actual proximity and availability i s further complicated by the cultural constraints placed on women. This is discussed below. 4.40. The LHW program (formally called the National Program Table 4.5: Rural Communities with a for Family Planning and Primary Health) seeks to provide active . Lady Health Worker, 2001-02 outreach o f maternal and child health services. LHWs are contract Region Percent workers hired by the program, to serve populations o f about 1,000, They are residents o f the communities they work in, and work out of Punjab 34 their home, which makes it easy for them to reachtheir clients. They Sindh 33 NWFP 38 are young married women aged 20-50 with at least 8 years o f Balochistan 10 schooling. Their status in the community i s enhanced by the fact RuralPakistan 35 that their wages were initially set at a level comparable to that of ;:; Source: PIHS 2001-02 community sun/ey data. primary school teachers, though their real wages have eroded over time. They operate in rural and poor urban areas, and their job i s to ~ ; ' : u ~ ~ ; ~ ~;k,"';;e~ty;;:v;;e onthis deliver preventive and promotive health services to women and their children. By 2001, about one-third o f rural communities had a LHW in the community, but coverage was low inBalochistan(Table 4.5). 4.41. The LHW is expectedto register all the children under five and married women aged 15-49 inher catchment area, and to provide various services to them. These include providing essential drugs for treatment o f minor ailments (such as diarrhea, malaria, acute respiratory tract infection, and intestinal worms); supplyingcontraceptives; and identifyingthose eligible to receive vaccinations and coordinating vaccinator visits to the villages or setting up immunization camps near the villages. Since 2001, more LHWs are being trained to give vaccinations to children and mother^.'^' They are also expected to motivate and refer women to obtain safe motherhood services (prenatal care, safe delivery, and postnatal care). To this end, LHWs are supposed to coordinate with the nearest primary health care facility, traditional birth attendant, or other skilled birth attendant. In addition, they are supposed to organize women's groups and health committees in the community to discuss issues related to better health, hygiene, nutrition, sanitation, and family planning.150 LHWs are supposed to provide hygiene education on drinkingwater and sanitation, advice on child care and nutrition, and growth monitoring o fchildren. 4.42. The LHW program is a national program, centrally funded and directed.151 Policy formulation and operational planning are done at the federal level. Implementation o f this operational plan is the responsibility o f the provincial and district program implementation units. These implementation units are staffed either by health department employees who are on deputation to the LHW program or by contract employees. The LHWs are hired, placed, and supervised by the District Implementation Unitso f the program, with oversight by the Federal and Provincial Implementation Units o f the program (see Table 4.6). 14'Governmentof Pakistan (2004). Is'Governmentof Pakistan(2004). Is'World Bank (2004a). 73 Table 4.6: Lady Health Worker Program: Levels of Responsibilities Government unit Responsibility FederalProgram Implementation Primary healthcare policy formulation, operation planning Unit(Ministry of Health) andbudgeting Provincial Program District LHW allocation, operationalplan implementation, Implementation Units (Provincial payroll Health Department) District ImplementationUnit LHW-primary health care facility allocation, LHW firing, (District Health Office) Lady Health Supervisorhiringifiring, training, operational plan implementation Primary Health Care Facility Selectionof LHW, training, organizing replenishment of supplies,providing meeting point for LHW and Lady Health Supervisor Source: Basedon table reported in Ministry of Health, "National Programmefor Family Planning and Primary Health Care: PromotingHealth; Reducingpoverty" (Governmentof Pakistan2004). 4.43. The LHWs are not accountable directly to the health facilities. A separate cadre o f Lady Health Supervisors (LHS) i s employed by the LHW Program on a contract basis to supervise and monitor the LHWs. A 2001 review'52found the frequency o f supervision was quite high: 70 percent o f LHWs had been supervised in the preceding month and 87 percent inthe two months precedingthe survey. The Lady Health Supervisors report to the District Program ImplementationUnit.Provincial Program Officers from the Provincial Coordinator's office oversee the District and LHW Supervisors. 4.44. At the same time, the LHW has some relationship with the local primary health care facilities. She is attached to the nearest BHUor other public primary health facility, which has a say in her selection and trains her. They are expected to refer patients to these facilities. They visit the facility periodically to collect supplies and to meet with and report to the Lady Health Supervisor. They do not report to the person in charge o f the health facility. The LHW program thus runs through the provincial and district departments o f health, in cooperation with the local health facilities. 4.45. A 2001 review by DFID o f the LHW Table 4.7: Planned Allocation of Lady Health calculated larger impact on health Workers,2004-05 outcomes per unit of cost than comparable alternative Number of Number of services provided through the public primary health districts LHWs facilities. This suggests a high level o f worker Punjab 34 52,381 motivation, since LHWs have no benefits other than Sindh 16 21,225 their salary, and no prospects o f promotion or reward NWFP 24 15,108 for good performance. Also, they face problems o f supply shortages due to budget cuts. Following this Balochistan 26 5,800 review, the program was expanded. By 2004, 70,000 Pakistan 120 100,206 LHWs were working in the field, and further Note: The figures for Pakistan includes the territories of h a d expansion is underway (Table 4.7). Punjab will be the Jammu Kashmir, Federally AdministeredTribal Areas, Northern Areas, and Islamabad. greatest beneficiary o f the projected expansion, while Ofthe planned 100,206 LHWs, 98 percentwere already allocated Balochistan's low coverage receive much less to districts by early 2005, and the rest will be allocatedeffective attention. Part o f this interprovincial inequality can be July 2005. The number of LHWs actually working could be different from the plannedallocationbecause many districts may attributed to differences inthe availability o f qualified not have recndtment. women. Source: National Programme for Family Planning and Primary ~ Is* Oxford Policy Management (2002). Oxford Policy Management (2002). 74 Lady Health Workers are More Likely to be Placed in Communitieswith Girls' Schools 4.46. The strongest determinant o f LHW placement is the availability o f a school for girls in the community (Table A4.5). This i s not a surprising finding, as LHWs are required to have at least middle school education, and the availability o f a school increases the supply o f such women. Thus the incentive programs discussed in Chapter 3 for retaining girls in middle and high school can be expected to have a direct bearing on the possibilities for expanding the coverage o f the LHW program. Since women are much less likely to work outside their own village, it could be difficult to expand the LHW program to areas underserved by girls' schools. 4.47. Given that the LHW program has to find educated women within the catchment area o f a functioning facility, the placement o f LHWs would be expected to be somewhat regressive, as the likelihood of findingthis conjunction o f circumstances is higher inbetter developed areas. The data show that they are indeed more likely to be placed in more developed areas, as indicated by the presence o f drainage in the community (Table A4.5). The DFIDreview also found that LHWs were placed in better- off areas. 154 Our data confirm that LHWs are indeed placed as the program intended, where DOH primary health facilities are available. As o f 2001, only one-half o f rural communities had a public health facility nearby, and only 20 percent had an LHW as well as a public health facility nearby. IV. OVERCOMINGWOMEN'SCONSTRAINTS ACCESSINGHEALTH To SERVICES 4.48. Given the various constraints that women face in meeting their health needs, what can be done to alleviate these problems? The analysis o f data from two surveys conducted in rural Pakistan in 2001- 02-the Pakistan Rural Household Survey and the Pakistan IntegratedHousehold S~rvey'~~-showsmuch can be achieved through fairly simple measures to expand the outreach for services and generate greater demand for them. 4.49. Expanding outreach and generating demand Figure4.9: Utilizationof Maternal HealthServices: for services are, o f course, highly related: if a service Impact of Proximityto Public Primary Health i s provided, people gradually become aware o f its Facility availability and benefits. However, it is possible to accelerate the pace o f demand generation by specific outreach measures, which we discuss next. 8I 40% 1$ P 30% ExpandingService Outreach 20% TheProximity of Public Health Facilities Matters 10% 0% 4.50. Although the general quality o f public health I Tetanus Toxoid Injections Prenatal Serices service delivery is low, certain aspects o f these a No Primaly Health Care Facility Within 5 Kms services appear to work well. People use public sector i Primaly Care Facility Within 5 Kms of Community services if a facility i s close by for several Notes: PIHS 2001-02 data for rural mamiedwomen aged 15-49 who had reproductive health services-immunization for given birth in the three years beforethe survey. The figures refer to statistically significant predicted probabilities from Tables A4.3-A4.5 children and pregnant women, and prenatal consultations (Figure 4.9). Oxford Policy Management(2002). Is'The PRHS 2001 covers the rural areas o f the four provinces and had a detailed module on illness, which we used to analyze determinants of seeking medical treatment for sick children. The PIHS 2001-02 collected data on availability of health facilities and programs, such as the LHW program, so this is used to analyze the determinants of child immunizations, use of maternal health services, and contraceptive use. Although the PIHS was also conducted in urban areas, data on the availability of health facilities and programs were collected only inm a l areas. Both surveys contain household and community-level data. 75 4.51. Having a facility close by, however, does not increase the likelihood o f receiving more skilled attendance at delivery (delivery in an institution or with a TBA in attendance), or postnatal care. More effort is needed to improve the outreach for these services, and perhaps also the demand for them. Use o f postnatal services may be limited by both a low perceived need for these services and .the customary practice o f keeping a woman and her newborn at home for 40 days after birth. 4.52. Having a facility close by significantly increases the probability o f seeking medical care for a sick child (Table A4.1). The further away the health facility is the higher i s the probability o f a child being reported ill-where curative services are less accessible, illnesses may become more severe and therefore more likely to be reported. The success o f the EPI program in closing the gender gap in childhood immunization indicates that more aggressive outreach campaigns may also be successful in reducing gender inequities in accessing medical care for sick children. ThePresenceof Lady Health Workersis Effective at Increasing the Uptakeof Some Services 4.53. Making services available near people's Figure 4.10: Presence of Lady Health Worker homes facilitates the use o f health care services. But Increases Child's Chances of Being Immunized given the mobility constraints that women face in Pakistan, however, having more women involved in 100% - delivering services at the doorstep seems to improve access and outcomes even more. The LHWs provide maternal and child health and contraceptive services at people's doorsteps, and our analysis indicates that this i s quite effective for several purposes. 4.54. The data indicate that L H W s are effective at delivering the main services for which they are responsible: expanding use o f contraception and facilitating the immunization p r ~ g r a r n s ' ~ ~ - r afor t e both childhood immunizations, as well as tetanus immunizations has improved during pregnancy Note The figures referto predictedprobabilitiesfrom Table A4 2 Source. PIHS 2001-02 survey data for childrenaged 12-23 months (Figures 4.10 and 4.11 and Tables A4.2 and A4.3). The presence o f an LHW appears to be more strongly Figure 4.11: Impact of Lady Health Worker Presence associated with immunization uptake than the proximity o f primary health facilities, probably because they guide clients directly to mobile vaccinators. Moreover, there is no gender difference in the probability of a child being immunized if a LHW is present in the community (see Table A4.2). Polio immunization is delivered largely through special outreach efforts and camps, so the effect of LHWs is muted. 4.55. The presence o f an LHW does not, however, significantly increase the probability o f receiving prenatal consultations, skilled attendance at delivery, or postnatal care (Table A4.3). These are not services the L H W is trained to deliver, but she is expected to 156This was also foundby the DFIDreview (Oxford Policy Management2002) 76 refer women to the network o f health facilities to avail o fthemselves of these services. The DFIDreview came to similar conclusions. Expanding Complementarities between the Lady Health Worker Program and Primary Health Care Facilities 4.56. The LHWs and the primary health care facilities are intendedto form a network to provide health services to rural households. Outreach work by LHWs is intended to stimulate families' use o f BHUs and other health facilities to meet their health care needs. However, we find virtually no evidence that the presence o f an LHW significantly stimulates the use o f nearby health facilities for reproductive health services (Table 4.8). Only in the case of postnatal consultations do we find some weak evidence that LHWs increase the use o f nearby government health facilities and reduce the use o f nearby private health facilities, The uptake o f postnatal care is extremely low, however. Inthe case o f contraception, we find that if an LHW i s present in a community, then people are less likely to turn to private health facilities nearby. This is in line with the fact that distributingcontraceptives is one o f the main tasks o fthe LHW. Table 4.8: Impact of Lady Health Workers and Proximity to Health Facility on Rural Health Center Use LHW in community L H W in interacted community Proximity with interacted to public Proximity proximity with primary to private to public proximity LHW health health primary to private presencein care care health care health care community facility facility facility facility Family planning methods Yes No No No Yes (negative) Maternal care: Prenatalcare No Yes No No No Tetanus toxoid immunizations Yes Yes No No No Postnatalcare Birth in medical institution No No No N o No Attendance at delivery No No No No No Childhood immunizations Yes Yes No No No Notes: This table summarizesthe impact of availability of LHWs in the community and the proximity to healthfacilities on households' utilization of various services based on the regression results in the Appendix. These regressions control for characteristics of the households and for the availability of various facilities and LHWs in the community They also control for community-level factors such as electricity, drainage, and distance to the following: tehsil capital, nearestbus stop, market, nearestmotorable rpproach road, public primary school for girls, middle school for girls. 77 4.57. Prenatal services offer a stark example o f this lack Figure 4.12: Use of Prenatal Services and o f synergy. Pregnant women are supposed to receive Antenatal Tetanus Immunization Status tetanus toxoid injections and check ups for signs o f potential complications o f pregnancy. As part o f the 100% - and our data indicate that this is very e f f e ~ t i v e . 'The ~ ~ link I with the DOHfacilities is weak, however: over 40 percent 40% of the women who received a tetanus immunizationreport 20% 4.59. This lack of complementarity between LHWs and health facilities may have little to do with whether the LHWs are referring women to health facilities and motivating them to use these facilities for maternal health services. The LHW program is not designed to overcome mobility constraints on women, which hinder them from going to the centers to avail themseives o f these services. Yet as we see below, if the demand for services increases, people make greater efforts to access services even if facilities are not close by. Also people avoid the health facilities because o f previous experiences with unavailable staff and lack of supplies. L o w service uptake, however, cannot be entirely attributed to the unavailability o f staff and supplies especially since LHWs are placed near the more "functioning" facilities. 4.60. This suggests a deeper issue of poor synergy between the LHW program and the DOHfacilities. LHWs may be working to meet their supervisors' expectations for the vertically-organized program, but not be working to meet ancillary objectives regarding service utilization at facilities run by another department. The LHW's supervision and monitoring process does not hold them accountable for ensuringuptake o f services at health facilities. Is' In some districts, LHWs may be deliveringthese injectionsthemselves. Is* It appears that the monitoringand supervision of LHWs has no checks for whether the women referred to health facilities by the LHW for maternal care services actually go to the facility to avail themselves of the services. The supervisor's checklist includes an assessment of whether the LHW has knowledge and skills for making appropriate referrals. The LHS also reviews records of the referrals made by the LHW. Independent of this monthly supervisionof the LHWs, the primary health facility reports to the district-levelprogram implementationunit, the number of referralsmade by LHWs affiliated with the facility. In this system, therefore, there is very little incentive for LHWs to ensure that women actually use these services. If there were adequateincentives, such as, escortingwomen to healthfacilities, then we would observeLHWs makingefforts to ensureuptake o f services. 78 Demand for Services Female Education Matters 4.6 1. Female education i s widely found to be one o f the most powerful predictors o f maternal and child health outcomes.'59 It is associated with better domestic management o f health. Among other things, educated mothers are more likely to be exposed to information from a wider range o f sources, and to be better able to process the information received. In Brazil, a study analyzing the pathways through which maternal education affects child health found that most o f the correlation between maternal education and child height could be explained by mothers' listening to the radio and watching TV. I 6 O A study from Central Java (Indonesia) found that mother's schooling affects shorter-term measures o f nutritional status mainly through nutritional knowledge.16' 4.62. Education can also enable women to Figure: 4.13a: Female Education and Childhood allocate resources better.'62 for example, in the Immunization World Bank's first community nutritional loan to Indonesia in the 1970s, significant improvements in child nutritional status were found to be related mainly to nutritional e d ~ c a t i 0 n . l ~It~i s also possible that educated women have greater bargaining power, both within the household and in their ability to interact with care providers in obtaining the DPT3 POLIO BCG MEASLES services they seek. OMother No Schooling MotherAttended School 4.63. We find that the schooling o f both parents i s significantly associated with the Figure: 4.13b: Female Education and Use of Maternal probability o f a child being immunized (Figure Health Services 4.13a) but the effect o f mother's education is stronger than that o f the father. The same Birlhin horpitaltclinic applies to a women's probability o f using Postnatal contraceptives and having prenatal consultations Consultation and tetanus immunization during pregnancy Tetanus Toxoid Injections 61% (Figure 4.13b and Tables A 4.2 and A4.3). For postnatal consultations, only the woman's PrenatalConsultation schooling has a significant positive association. I R4Sorne Schooling UNOSchooling Source: The graph is based on data from the PIHS survey (2001-02). In graph A, the percent immunized based on the level of the mother's education is Information Campaigns Also Generate Demand predicted using the regressionresults displayed in Table A4.2 in the Appendix for Services to Chapter 4. In graph B, the percentage of women utilizing maternal health servicesis predictedusingregressionresults displayedinTable A4.3. 4.64. There is considerable evidence that greater exposure to information can "substitute" in some ways for schooling. This has been widely noted, for example, in the spread o f information on contraception and the uptake of contraceptive use.164 Some studies have sought to specifically tease out the extent to which education and information substitute for each other. A study in Brazil found that 159 Strauss and Thomas (1995). I 6 OThomas, Strauss, and Henrique (1990) Webb and Block (2004). Welch (1970). 163 Webb and Block (2004). 164 Rosenzweigand Schultz (1989); Casterline, Sathar, and ul-Haq (2001). 79 schooling and messages gained through community health services acted as substitutes for each other.'65 A study in Morocco found that maternal knowledge and not maternal schooling strongly influences measures o f child's long-term nutritional status (height for age).'66 This study also found that this maternal knowledge inthis setting i s obtained mainly from the media and public service announcements. 4.65. InPakistan, there are a number o f information campaigns on maternal and child health issues.'67 The Ministry o f Population Welfare has a Communication and Advocacy Strategy that focuses on disseminating family planning information, especially to rural areas, youth, and men.168The Ministry o f Health also sponsors health education campaigns that focus on safe m ~ t h e r h o o d .Campaigns on family ' ~ ~ planning are also undertaken by the private sector. The EPI program has had campaigns for childhood immunization and antenatal tetanus immunization. 4.66. We explore how dissemination o f health information can offset the disadvantages of low female education and low coverage o f health facilities. Our survey data include information on whether a woman had been exposed to media messages on hygiene practices, and we use this as an indicator o f women's exposure to health messages from the media. We find that this indicator is positively related to the probability o f using contraception, prenatal consultations, tetanus toxoid immunization, and institutional delivery (Table A4.3). However, there was no significant association with the use o f postnatal consultations. Thomas, Straws, and Henrique (1990) Glewwe (1999). 16'Sathar and Casterline (1998). Government of Pakistan (2003) 16'As part of the Women HealthProject (Government of Pakistan2003) 80 4.67. Does having information reinforce or Figure 4.14a: Impact of Media Exposure and Female substitute for the benefits o f having an Education on Use of Maternal Health Services education? It appears that media exposure to health messages plays a positive role Contraceptive use I independent +om maternal education. The 2% ~ interactions between media exposure and female Il education show that both educated and Tetanus Toxoid Injections ~ uneducated women benefit from media exposure I (Figure 4.14a). For example, among uneducated Prenatalservices women the percent seeking prenatal consultations rose from 20 to 29 percent with media exposure, Among women with some Birth in hospitaVclinic schooling, the percent seeking prenatal 0No inform consultations went up from 51 to 60 percent 0Informationfrom media+ No School with exposure to media messages. No information from media+ School 0Informationfrom media+School 4.68. Since LHWs deliver information along with services to women's doorsteps, they might Figure 4.14b: Impact of Lady Health Worker Presenceand be expected to be more helpful to women with Female Educationon Use of Maternal Health Services no schooling, or to women who may have less I access to information from the media. The analysis suggests that this i s not the case: the ContraceptimUse presence o f an LHW benefits women independently o f their schooling or exposure to 29% media (Figures 4.14b and 4.14~). Note that the I effects are significant only for the services (contraception and immunization) for which the 1 Tetanus Toxoid 38% presence o f a LHW has a significant impact. Immunizations No LHW + No Schooling LHW + No Schooling Figure 4.14~:Impact of Lady Health Worker Presence and Media Exposure on Use of Maternal Health Services Contraceptive use Tetanus Toxoid Injections 6% No information from media+No L W No information from media + LHW 0Informationfrommedia+No LHW 0Informationfrommedia+LHW .Note: These figures pertain to rural married women aged 15-49. The figures refer to the predicted probabilities of using these services from Tables A4.3 and A4.4. In graph A, the predicted effect of media exposure is addedto the effect o f woman's schooling on the use o f these services. In graph B, the predicted effect o f media exposure is added to the effect of LHW presence. B1 4.69. Media exposure has an even more Figure 4.15: Impact of Media Exposure and Distance to striking impact on stimulating the use o f health Primary Health Center on Use of Maternal Health Services facilities. For example, women without a primary health facility close by, the percent with prenatal consultations rose from 19 to 34 Prenatal semces percent with media exposure, and from 29 to 42 I 7 34% 42% percent for those with a facility close by I I (Figure 4.15). In sum, media exposure appears 1 to play a role independent o f other factors Tetanus Toxoid which generate demand for services (in this Injections case female schooling and the presence o f an 51% LHW), as well as the supply o f services. I J Notes: These figures pertainto rural marriedwomen aged 15-49. The figures refer to the predictedprobabilities of using these services from Tables A43 The predicted effect of mediaexposureis addedto the effect of proximity to aprimaly health center (PHC) on the use of these services. V. POLICY RECOMMENDATIONS 4.70. Ifonehadamagicwand, itwouldbepossibletoaddresssimultaneouslyallthemajorfactors underlying poor health outcomes in Pakistan: poverty, exposure to disease, poor public services delivery, low levels o f education, and the constraints women face in caring for themselves and their families. 4.71. Pending larger shifts in policy priorities and their actual implementation, however, much can be done to improve health outcomes. For instance, steps can be taken to expand effective access to public primary health facilities, especially in rural areas, as they are pro-poor in nature. Many communities do not have a facility close by, however. Studies also show that even when facilities are in place, service delivery in many o f these facilities is hindered by staff absenteeism and lack o f supplies and equipment. Here there may be scope for efficiency gains through private contracts to run public facilities. International experience suggests that private-public partnerships can be an effective way to improve public health service deli~ery.'~' 4.72. Improving poor coverage and quality o f public services is critical to improving health outcomes, but it will not eliminate the obstacles faced by women and girls in accessing health services. The discussion in the chapter focused on reducing the gender-related constraints to accessing health services and our findings indicate the effectiveness o f expanding service outreach, information dissemination, and improving the coordination o f services. This leads us to several policy recommendations, 4.73. First, the LHW program should be expanded to areas with relatively low coverage and strengthened, as it appears to be highly effective at delivering basic maternal and child health services to " OWorld Development Report (2003). A successful example of a contracting-out arrangement is the tuberculosis (TB) detection and treatment project in Hyderabad, India (Loevinsohn and Harding 2004). The Government engaged a private non-profit trust working in a poor neighborhood in Hyderabad to deliver TB treatment using the DOTS (directly observed therapy-short course chemotherapy) method. An evaluation of this project found that the private provider was able to achieve a treatment success rate 14-percentagepoints higher than the public sector provider in a nearby area. The private provider was also able to diagnose 21 percent more TB cases per year. The private delivery of services was also found to be more cost-effective. In Pakistan, the Punjab Government contracted an NGO to manage BHUs in Rahim Yar Khan district in 2003. While no evaluation of this project is as yet available, initial assessments suggest that since the NGO took over the management of BHUs in this district the staffing and supply of drugs was greatly enhanced and there was an appreciable increase in the number of outpatient visits (Loevinsohn and Harding 2004). 82 people. We find that having these women deliver services to people's doorsteps considerably enhances the probability o f children being immunized and of women receiving prenatal care and using contraception. The DFIDevaluation (200 1) found the sarne.l7' 4.74. Although distance to primary health care centers i s an important determinant o f the use o f postnatal care, LHWs are at the right place to deliver a range o f such services effectively, and their efforts should be further reinforced through training. Experience in Bangladesh shows that when the demand for contraceptive and maternal health services i s low, much can be achieved by moving from fixed-point service delivery to active outreach through home visits-and when the demand has risen significantly, it i s possible to revert to fixed-point service delivery because women are more likely to seek this services on their own (ICDDR,B, 2003). 4.75. Second, our analysis o f the placement o f LHWs suggests that the supply o f women with middle school or higher education in rural areas may constrain the ability o f the LHW program to expand into underserved areas. Policymakers need to find ways to deliver services to those parts o f the country which, for the foreseeable future, the LHW program will not reach. The LHW program could finance scholarships for girls in middle school in underserved areas with an option to work as an LHW upon completion. In more remote and underdeveloped areas, it may be instructive to follow the example o f Vietnam (Box 4.1). Box 4.1: Overcoming Barriers to AccessingHealth Services: Insights from Vietnam In Vietnam ethnic minorities living in mountainous terrain are frequently underserved with health facilities. These people are often not accustomed to seeking reproductive and child health services on a routine (nonemergency) basis. Under these circumstances,putting outreach staff inthe field to actively identify women who needthese services may miss those who are most socially and financially disadvantaged. To reach these socially and geographically marginalized groups with reproductive and childhealth services, the government tried several pilot initiatives, the most popular and successful of which was to organize campaigns on reproductive and child health services for disadvantaged areas that consisted mobile teams deliver the services, a model that has beentested inmany settings. What distinguished the campaigns in Vietnam from campaigns in others places is that the visits by the mobile teams were preceded by extensive information outreach. This ensured maximum effectiveness of the mobile teams. Communities were informedrepeatedly about when and where the team would arrive (typically at a local market where people tend to congregate), how long the teams would stay, and which services the team would provide and who should seek their services (e.g., pregnant women for antenatal checkups). The mobile team would then spend a few days inthat location to provide the services announced. These campaigns did much to generate demand and increase service utilization inthe most underservedregions o f the country. Provincial staff expressed a high level of satisfaction with this effort because they felt that not only had they raised people's awareness of women's reproductive health needs, but also the campaign had enabledthem to actually respondto increaseddemand by providing free services, including surgery. The campaigns were successful in overcoming a variety of obstacles to better reproductive and child health: , including the following: (1) limited supply of health facilities; (2) limited information about the need for health services; and (3) inability to pay user charges for regular health services. In Pakistan, such a model could be especiallypowerfulbecause of the constraints on women's mobility. Source: World Bank (2004e) "' They also found that the LHW program was more cost-effective than other primary care services. This may be due partly to the fact that they provide both family planning services and child health services, a combination which has been found effective elsewhere(see for example Fauveau 1994, on Bangladesh). 83 4.76. Third, the apparent disjunction betweenthe services o fLHWs and the health facilities needsto be addressed, through measures to increase their mutual support and accountability. There are many potential syngeries and benefits to all from coordinating the LHW program with overall public health care services: the health facilities can provide important technical and logistical support for LHWs, while LHWs can increase facility utilization by referring potential users to them. An example o f coordination between a central program (LHW) and provincial health departments i s the Auxiliary Nurse Midwife (ANM) program inIndia (Box 4.2). InPakistan, the potential for synergy is apparent inthe coordinatic between the EPI program andthe LHW program. Box 4.2: Integrating Centrally-sponsored Programs with State Programs: The Auxiliary Nurse Midwife Program in India Auxiliary Nurse Midwives (ANMs) in India are the outreach staff responsible for maternal and child health services . They are typically posted outsidetheir native village and are subject to transfer. This means that they haveto live and work incommunities often unfamiliar to them, and they have to travel on their own to reach all the villages for which they are responsible.Another problem arises from the fact that they are evaluated largely on their performanceof maternal and child health care tasks, while the communities they serve would like them to meet broader curative serviceneeds. Inmany ways, then program is far from perfect, but it offers insights into how vertical andhorizontal programs can be meshed. The ANMs' salaries are covered by the central government's vertically -organized Family Welfare Pprogram, which also provides and their supplies of contraceptives, folic acid, etc., also are supplied directly by this program. In their work set-up, however, the ANMs are integrated into the regular health services, which are organized and funded by the state governments. They are assignedeither to Primary Health Centers, or (more typically) to their sub-centers. They form part of the staff of the Primary Health Center to which they are attached, and this manifests itself in a high level of integration. They are supervised along with other staff by the medical officer (and health assistant reporting to the medical officer) of the primary health center., and by hisiher health assistants. This takes place at monthly staff meetings at the PHC to evaluate workers' progress and suggest improvements. During monthly staff meetings, the medical officers convey information from their meetings with the district health officer, inform staff about campaigns and surveys proposedby the district-level administration, monitor existing activities, and outline work schedules for the nexl month. ANMs thus participate inregular facility meetings in which the work program of the facility as a whole i s discussed. The support flows in both directions: for example, medical officers are supposed to visit the subcenters periodically and attend clinics organizedthereto examine andtreat difficult cases. The fact of participation in regular meetings and being accountableto the facility head means that the ANMz function as part of the Primary Health Center team. Thus, although the ANMs are part of the centrally-run vertical program for family welfare, andthe health facilities are managedby the state and local governments, the ANMs have a clear incentive to work closely with their health facilities. This assures coordination betweenthe vertical RCHprogram and the health department. Source: Iyer and Jesani(1999). 4.77. In a broader sense, it is important to strengthen the integrated delivery o f maternal, child health, and family planning services. Under the devolution policy, the delivery o f health services is already moving toward this more integrated structure. Population and health services are being decentralized to the provincial level and devolved to the district level. At the federal, provincial, and district levels the Ministries o f Health and Population Welfare are becoming more integrated (Government o f Pakistan, 2003). 4.78. Fourth, it i s critical to have intensive information campaigns covering a wide range o f issues geared toward enhancing people's ability to manage and protect their own health. A large proportion of women have never been to school and/or do not have access to LHWs or health facilities, and our analysis suggests that carefully designed information campaigns can do much to offset the associated 84 disadvantages for women's and children's health. Television reached an estimated 40 percent o f women in rural Pakistan in 2001, and radio reached 36 per~ent."~If, in addition to this, it were possible to use village loudspeakers to communicate a few well-chosen health-related messages, it would be possible to tap further the special advantage o f intensive media campaigns that reach all members o f the community and gradually raise overall awareness on health issues. More local media campaigns can also build community acceptance for paying more attention to women's health needs and reduce the social barriers to women accessing health care independently. For example, loudspeakers in local mosques have been used successfully to alert people when the vaccinator visits the illa age,"^ and their use could be expanded. 4.79. Closing the gaps in health service delivery for women requires action on many fronts. More active coordination o f existing programs that have already yield effective results i s worth considering. The LHW program is an ongoing and expanding program and it should be monitored as changes are introduced to the program to shape its continued success. Furthermore, the design o f national programs and interventions can be guided by lessons learned from many ongoing pilot projects across Pakistan. One such pilot project in the area o f maternal health is the Balochistan Safe Motherhood Initiative, which emerged out o f an operations research study by The Asia Foundation. The Balochistan Safe Motherhood Initiative tested a package o f community-based interventions, including providing health education to women and their husbands, training midwives to recognize and referring high-risk pregnancies, and setting up transport systems to improve access to health care centers. This initiative was successful in significantly lowering perinatal, neonatal, and maternal mortality. Much can be learned, moreover, from the successes with improving immunization coverage in Pakistan since the early 1990s. While logistical and coordination improvements ensured a better and steadier supply o f vaccines, intensive efforts to bring services to people's doorsteps was key to improvements in immunization coverage. The polio immunization campaign, for example, made very successful use o f media and outreach to increase service uptake. 17'Pakistan ReproductiveHealth and Family Planning Survey 2000-01. Oxford Policy Management 2002 85 5. WOMEN'S WORK AND MOVEMENT THE PUBLICSPHERE INTO And their lord hath heard them (andHe sayeth): Lo! Isuffer not the work of any worker, male orfemale, to be lost. Yeprecede onefFom another. -The Qur'an,verse 31285. 5.1. Women in Pakistan tend to be less "visible" with respect to their work outside the home and their contributions to household income, as well as their participation in social and political life. This chapter analyzes the constraints to women's participation in the public sphere. We first focus on labor force participation. This analysis is then used to discuss how women's participation in the political process can be enhanced. 5.2. As the previous chapters demonstrate, women lag behind men in schooling and are restricted in their use o f public services, including health facilities. Not surprisingly, women have significantly lower rates o f labor market participation, and they have much lower rates of participation in other dimensions o f public life, including political activity. 5.3. There are several reasons to link poor human capital outcomes to reduced economic, political, and social productivity among women. First and foremost, low education sets o f f a vicious cycle o f lower attainments for females. Lack o f schooling makes women ill-equipped to obtain higher-skilled and better-paying jobs. These lost opportunities for remuneration make households poorer and less likely to invest inthe education o f all children, especially girls. Research findings indicate that when resources are constrained, parents are more likely to cut back on investments in girls' education than in boys' education. L o w education levels also limit opportunities to obtain information on employment opportunities, as well as the benefits o f community and political participation. 5.4. Another reason to expect low female activity in the public arena is the pervasive nature o f constraints on women's mobility in Pakistan. The same socio-cultural restrictions that curtail female access to education and health facilities circumscribe their opportunities to work and participate in political and community decisionmaking. Indeed, we find that women take up opportunities for paid work only in a very geographically circumscribed manner, limiting themselves to work within their villages, for the most part. 5.5. As with girls attending school, restrictions on women's activity outside the home are rooted in concerns for female safety and family honor. Males in the household may be concerned that women's safety i s at risk. In socially conservative areas, men may also worry about damage to the household's reputation if young women venture out o f doors, particularly to earn money. Such activity can brand the family as low status and imply that the men cannot adequately provide for the economic needs o f the household. In more unequal communities where status hierarchies are quite rigid, such actions can also make women vulnerable to loss o f reputation or honor. Families thus may worry even more about the safety and honor o f young women who leave the household for work than they do about girls who leave the household to attend school. 5.6. The analysis o f labor force participation by women can be summarized as follows. First, mobility restrictions limit women's participation in the labor force. Of course, determinants such as household wealth or the education level o f the woman also play a role. Evidence from Egypt, Turkey, and other countries in the Middle East and North Africa, where women's education levels are high but their participation in work activities remains low, suggests that social barriers to female mobility can significantly stifle women's labor force parti~ipation.'~~Restricted female mobility in Pakistan, particularly as it limits access to services and generates practical problems o f traversing distances, has become a topic o f scrutiny.'75 Second, participation in work, particularly paid work, has important ramifications for women's autonomy. Analysis suggests that women who participate inpaid work are also much more likely to participate in community and political activities. Given the increasing role o f local government under the decentralization process, and thus the increased role o f communities in political decisionmaking, participation in work is perhaps one avenue through which women's civic participation can be enhanced. 5.7. The analysis considers the statistics on women's labor force participation. According to the Pakistan Integrated Household Survey (PIHS) 2001-02, at the end o f the 1990s, only one in four adult women (aged 10 and older) participated in the labor force, a far lower rate than the nearly 70 percent participation rate for men. Women's rate o f labor participation is higher in rural areas (30 percent) and lower in urban areas (15 percent), while male participation rates are close to 70 percent inboth regions. 5.8. We rely on PIHS instead o f the Labor Force Survey (LFS) to analyze labor force participation because o f certain survey procedures with the LFS that may undercount labor force participation by women. The LFS asks about labor force participation duringthe one week preceding the survey and does not pose detailed questions about the type o f activities undertaken. The LFS therefore is less likely to capture labor force participation by women because women are more likely than men to engage in seasonal or unpaid work on the family farm or enterprise. Future LFS rounds may need to take into account longer reference periods to accurately capture women's work (see Annex 5.1 for more details). The PIHS has a longer reference period and asks about labor force participation over the month preceding the survey. Also the gender o f the enumerator and respondent may matter when measuring female participation. For social and cultural reasons that confer negative connotations on female work, a male respondent, such as the household head, may under-report female participation inthe labor force. Having female enumerators, who directly interview women in the household, as does the PIHS, removes this particular bias. 5.9. At the policy level, there appears to be recognition of the needto encourage women's work. The National Policy for Development and Empowerment of Women (2002)' for example, seeks to increase women's capacity to earn a living wage as a means o f enhancing their economic empowerment. When asked, most women say they want to work. The recent survey on Adolescents and Youth in Pakistan (AYP) asked young women who were not working whether they would like to work in the near future if work opportunities were to become available'76:and 77 percent o f females aged 15-19 and 70 percent o f females aged 20-24 reported that they would work if such opportunities arose. The 33-percent reservation o f local government seats for women in the year 2000, moreover, has created a noticeable female presence in the public sector for the first time. Increased participation in all these aspects o f the public arena translates into an opportunity for greater autonomy for women, as well as a broader range of venues in which they can give voice to their concerns and aspirations. This chapter discusses in detail the constraints to women's participation in the labor market and the relationship between work and autonomy. It also offers policy recommendations for encouraging women's participation in the labor force. 5.10. The remainder of the chapter is organized as follows. Section Iuses the latest round o f PIHS data to summarize the dimensions o f women's labor force participation. Section I1discusses the constraints to women's participation in the labor force. Section I11 examines the relationship between labor force participation and women's autonomy, and Section IV offers some recommendations for policy. Our 174Assaad and Arntz (2005) ;World Bank (2004). Sathar andKazi (1997); Khan (1998). 176(Population Council2002): Adolescent and Youth Survey, 2001-02. 87 analysis is based on the data available from the nationally representative PIHS (1991, 2001-02), Pakistan Rural Household Survey (PRHS) 2001 (which provides data that is representative o f rural Pakistan), and PRHS 2004, which provides data representative o f rural Punjab and Sindh. I.DIMENSIONSWOMEN'SPARTICIPATIONINTHELABOR OF FORCE 5.1 1, Analysis of survey data indicates that Figure 5.1: Labor Force Participation Rates in Urban Areas women participate in the labor force at by Province substantially lower rates than men do in both A. UrbanWomen urban and rural regions. Data for 2001-02 1 yields an overall participation rate o f about 25 65% percent for women and 67 percent for men. The Punjab female labor market activity i s different in rural and urban areas. Female rural participation Sindh 67% rates are roughly twice as large as female urban participation rates. There also are significant N W P 62% differences across the provinces in the rural participation rates. A much larger fraction o f 61% women report labor market activity in rural Balochidan Sindh and Punjab, as compared to NWFP and Balochistan. The rate for men i s roughly the All provincesurban 66% same inrural and urban areas. (Figure 5.1). I OMale IFernale 5.12. Provincial differences in rural female participation rates are likely to reflect B. RuralWomen differential opportunities for agricultural employment. As we discuss below, women in Punlab rural areas work predominantly in agriculture, I and Sindh and Punjab constitute the agricultural Sindh 72% heartland o f the country. 5.13. Table 5.1 decomposes labor market NWFP 60% participation by industry. Three aspects o f female labor market participation are evident Balochistan 63% from this table. First, women work in a much narrower set o f occupations than men. Rural 66% women tend to be concentrated in agriculture, All provincesrural while urban women tend to work predominantly in unskilled service jobs such as UMale MFemale personal and household services. Second, the Note: Percentages are calculated for individuals aged 10 and older occupations in which women are predominantly Source: World Bank staff calculations usingdata from PIHS 2001-02. engaged offer lower wages. Third, these occupations are much more likely to keep women close to or inside the home. 88 Table 5.1: Distributionof Workers in Industry by Gender (percent) Rural Urban Male Female Male Female Agriculture 53.8 76.9 5.3 9.4 Other services 13.2 10.6 27.6 57.5 Manufacture 6.6 11.2 20.2 27.1 Construction 10.0 0.2 6.7 0.5 Trade 9.9 1.0 27.1 4.3 Transport 5.6 0.0 11.1 0.8 Source: World Bank staff estimates calculatedusing PIHS 2001-02. 5.14. As elsewhere in the developing world, the Figure 5.2: Impact of Schooling onWomen's analysis obtains an inverted U-shaped relationship Labor Force Participation Rate between women's labor force participation rates and female e d ~ c a t i 0 n . IParticipation rates are ~ ~ 60% 1 l highest among women with no education and ca, m women who have completed secondary school. 40% (Figure 5.2). Women with a primary school .-c 0 m education have the lowest participation rates. .-.oQ20% Education levels also affect the occupations in r am which women work. Women with no schooling i tend to be engaged in unpaid activities, while those 0% with secondary and higher schooling are more No School Primary Secondary Tertiary likely to be engaged in paid work. Women with +Paid Work Urban -2-PaidWorkRural , some schooling may be less likely to engage in +Unpaid Work Urban +Unpaid Work Rural I unpaid work on the family farm because it may not be-attractive to them, while paid work may require Source: Regressionreported in Table A5.1, both based on data from PIHS 2001-02 for aged1549 education beyond the primary level. For instance, to be able to- teach in a primary school, a woman has to have completed at least class 10 (secondary school). 5.15. Breaking down labor force participation by occupation shows that a much higher proportion o f urban men than women are engaged in white-collar jobs, such as clerical and sales professions. While about 18 percent of working women report working inclericaljobs, more than 33 percent o fworking men are engaged in such occupations. Within occupations there is further gender segregation by type o f white collarjob. A recent report on the need for quotas for women in public sector jobs reports that, despite the existence o f quotas across all cadres, women tend to be concentrated in the education and health departments. This may reflect, in part, a decision by women to remain within the domain o f "socially acceptable" work.l7* 5.16. Beyond the white collar jobs o f the type described above, women's labor market participation in urban areas seems to be concentrated in home-based manufacturing One study estimates that since the 1980s, there has been a dramatic increase inthe proportion o f urban women engaged in informal home-based work, primarily crafts and related occupations. Home-based work i s defined by the 177Mammen and Paxson (2000). Note this relationship inthe case o f Thailand and India for example. 17*National Commission on the Status o f Women (2003). 179The informal sector includes much o f the services sector, the construction labor force, women's home-based work, vendors, hawkers, and so on. The 2001-02 LFS reports that a majority o f both working men and women (more than 60 percent) in urban areas were employed inthe informal sector. 89 International Labor Organization (ILO) as work carried out by a person for remuneration in his or her home or in alternative premises o f his or her choice, other than the workplace o f the employer.1s0While Pakistani researchers and policymakers have increasingly focused on the plight o f home-based informal workers, especially that o f women home-based workers, initiatives inthis area are hampered by a lack o f data. Due to the nature o f activities undertaken in this sector, it is difficult for most surveys-including the PIHS-to gauge the extent o f home-based informal work. ChangesinWomen's Work ExperienceDuringthe 1990s 5.17. Whereas female participationrates remained below male participation rates throughout the 1990s, the percentage o f working women in certain occupational categories such as teachers in urban areas and paid agricultural work in rural areas rose. What could be driving the increased female participation in these occupational categories? 5.18. The expansion o f schools- particularly private ones-during the 1980s Table 5.2a: Rural Labor Force Participation Rates, 1991-2001 (in percent) and 1990s improved employment Females Males opportunities for educated women in rural 1991 2001 1991 2001 and urban areas. In 2001-02, slightly more Participation inpaid and 55 57 77 73 than one quarter (27 percent) o f urban unpaid work (inpercent) working women were employed in the Of those participating, 30 27 11 19 professional category.lS1 Since the 1980s percent participating in agricultural wage work there has been an increase in women's O f those participating, participation under this occupational percent in nonagricultural 2 2.4 38 55 category and this increase has been confined wage work mainly to the teaching profession; analysis o f the 2001-02 PIHS finds 21 percent o f Table 5.2b: Participation in Paid Agricultural Work urban working women to be teachers or among Those Working by Province (in percent) teaching associate professionals. In contrast, Females Males only three percent o f menreport being in the 1991 2001 1991 2001 teaching profession. The increased presence Punjab 40 46 14 20 o f women in this category can thus be Sindh 19 26 I 29 attributed to the increase in demand for NWFP 5 2 6 3 Balochistan 3 4 6 18 female teachers from public and private schools alike.Is2 Note: The participation rates are for men and women aged 15-49.Comparing time trends in female and male labor force participation rates requires comparabledata sources over time. As section 5.2 will show, the LFS may not 5.19. In rural areas, between 1991 and capture female participation fully, so other data sources must be used for this 2001, the percentage o f working women exercise. Fortunately, the 1991 PIHS and 2001 round o f the PRHS have comparable questions on labor force participation, as well as similar reference participating in paid agricultural work periods and survey proceduresthat allow us to compare participationrates. increased significantly in Sindh and Punjab Source: World Bank staff calculations usingPIHS 1991and PRHS 2001-02. (Tables 5.2a and 5.2b).Is3 Women in these two provinces mainly work as wage laborers picking cotton during harvest time. It is noteworthy that women continue to dominate this task even though cotton i s a cash crop. Evidence from other parts o f the world suggests that men tend to take over tasks related to cash crops. As tables 5.2a and 5.2b show, the proportion o f working men participating in paid agricultural work also increased between 1991and 2001. Men, however, tend to be involved inharvesting food grains such as wheat and rice. ILO, Convention 177 on Home Work; Haider and Tahir (2002). Pakistan Poverty Update (2004). Women aged 15-59. Kaziand Raza(1991) report a significant increase inprofessionalwomen workers inurban areas during the 1980s. PIHS 1991and PRHS 2001 are used for this analysis. 90 5.20. Government agricultural statistics data on growth in output o f cash crops and Table 5.3: DailyWage Rates for AgriculturalWage Work food grains during the 1990s, shows an in Cash Crops, 1991-2001 increased proportion o f women and men (rupees) participating in paid agricultural work. The Females Males increase in participation may have been 1991 2001 1991 2001 driven to some extent by rising demand for Cotton Picking 20-30 26.82 -- -- men's and women's labor as out ut o f food Sugarcane 25-30 30 40-60 61 grains and cash crops increased. What has harvesting happened to wages as more women and men Note: Participation in agricultural wage work among those who are have entered wage work in rural areas and participating in the labor force. Wages are for Punjab and Sindh. These are nominal wages, not adjusted for inflation. Inflation averaged9.7 percent per output has expanded? A comparison o f annum between 1991 and 2001. No male wages were reported for cotton nominal wages suggests only a modest rise pickingbecausethis is a predominantlyfemale activity in male and female wages. In 1991, for Source: Mansuri (1994) for 1991 wages and PRHS for 2001, and Economic Survey (2003) for inflationrate. example, female daily wage rates for cotton I picking were inthe range of Rs. 20-30 in Sindh and Punjab (Table 5.3), 5.21. The PRHS data also suggest that in 2001, the average female wage rate for the same activity in these provinces was about Rs. 26.82 per day. Given the highrate o f inflation over this periodYig5real wage levels have decreased as female participation in agricultural wage labor in cotton has increased. In sugarcane as well, wages appear to have stagnated (Table 5.3). Throughout the decade o f the 1990s, opportunities for agricultural wage labor expanded in rural areas o f Sindh and Punjab as food grain and cash crop output rose. This expansion may have absorbed increased participation by both men and women. The rise in agricultural output together with an increase in labor force participation appears to have dampened wage increases. 5.22. What i s noteworthy is that the majority o f female paid agricultural workers in rural Sindh and Punjab-almost 80 percent-reportedly work within their village.lg6This would suggest that employers within a village are likely to face little competition for women workers from outside employers, leading to fairly monopsonistic village labor markets. This i s certainly consistent with the wage stagnation observed, despite the increased demand for agricultural wage labor. As shown below, restrictions on women's mobility, particularly those arising from security concerns, indeed structure preferences for the location o f work and are thus likely to pose a significant constraint on rural women's participationin paid work, as well as on the returns to such labor. 11.CONSTRAINTS ONWOMEN'S LABOR FORCE PARTICIPATION 5.23. Evidence presented in previous chapters has shown that restrictions on female mobility significantly constrain female access to schooling and health. The analysis o f female labor force participation demonstrates that restrictions on women's mobility affects their ability to participate in the labor force as well as contributes to the gender gap inwages among those who work. 5.24. Evidence from a number o f qualitative and in-depth quantitative surveys, reviewed in earlier chapters, suggests that it i s difficult for a woman to cross the boundary o f her own village or settlement to undertake work (or even to attend school) in a neighboring village or settlement. For example, data from Agricultural Statistics of Pakistan, Ministry of Food, Agriculture and Livestock, 2000-2001, The output of food grains increased from 19,587,000 tonnes in 1990-91to 25,986,000 tonnes in 2000-01(one metric tonne = 2,204.62 pounds). The output of cashcrops increasedfrom 38,285,000 tonnes in 1990-91to 45,867,000 tonnes in2000-01. The Economic Survey of 2003 reports that inflationaveraged 9.7 percentper year between 1991and 2001 (this is based on the CPI with 1990-91as the base year). Based on 2001 data from PRHS, which askedwomen where they worked: within the village or outside the village. 91 PRHS-I1 clearly shows that girls' enrollment rates were significantly lower if a girl had to cross a settlement boundary to attend the closest school (Chapter 3). A recent qualitative study from Punjab that assessed social barriers to women in rural areas accessing health facilitie~,'~'found that mobility patterns vary depending upon the women's marital, education, and social class status, as well as upon whether they were going someplace within the village or outside it. Women were able to move freely within the village, but they could rarely leave the village alone or without permission and tended to avoid public transport.'" 5.25. The qualitative survey on health covers only three villages in Northern Punjab; therefore, these results may reflect location-specific effects. Data from the PRHS-I, however, is representative o f rural Pakistan in 2001 and covers the four provinces. This survey asked women engaged in paid work about their work location. The patterns reflect mobility restrictions described inthe qualitative study mentioned above: almost 80 percent o f women engaged in agricultural wage labor and about 60 percent o f women engaged in some type o f nonagriculturalwage labor report working within their own villages. The lower percentage o f women in nonagricultural wage labor working within the village could arise from the fact that most o f these women work as teachers or health workers, which may necessitate travel outside the village. Evidence from the qualitative and quantitative studies show that mobility restrictions matter greatly for women's participation in work outside the home. These restrictions therefore can greatly reduce the number o f women who work since they must seekjob opportunities within the village. 5'27* The pRHs-rl surveyprovides a unique opportunity to assess how safety Table 5.4: Effect of Safety on Participation in Paid Work concerns affect women's participation All paid Paidfarm in paid work. Tabulations based on this Woman's concerns about safety work work survey data presented in Table 3.6 Feels unsafewithin settlement Negative Negative (Chapter 3) shows that almost 60 impact impact percent of rural women in Punjab and Feelsunsafe outside settlement No No impact impact Sindh unsafe Outside their Source: Based on probit regression results reported in Table A5.3, settlement, as compared to only 18 PRHS 2004. Is'Khan(1998). See also Mumtaz and Salway (2005) onthis issue. Regressionresults are presentedintable A5.3. 92 Box 5.1: Social Perception of Paid Work by Women Sathar and Kazi's (1997) study from rural Punjab found that husbands tended to underreport their wives' participation in paid work. A quote by a male respondent cited in their report depicts the underlying social stigmaattachedto women's work: The reality is that most women work outside the home in theJields but we want them to stay indoors. Wefeel ashamedwhen our womenwork but they can do so in extreme need. -Malerespondentresidinginbarani(rain-fed) villageinPunjab Women in Pakistan also perceive a tradeoff between the status of a household and the incidence of any female member of that household participating in paid work. Out of 60 women interviewed for a qualitative study (described in Box 1.3) associatedwith this Assessment, 25 (42 percent) linkedstrict purdah, practiced both in terms of strict dress code and being confined to their homes, with people from high-status householdsthat are relatively well off and of relatively high caste. Women from poor and lower-castehouseholds were perceivedas those who could not afford to practice strict purdah because their families' lack of means that necessitatedtheir labor-usually in the fields, though a few were midwives. Table 5.5 shows that while women in more conservative regions tended to perceive the practice of strict purdah as a privilege observed by higher-status households (41.7 percent in Sindh and 66.7 percent in southern Punjab), more women in both parts of Punjab than in Sindh lamented the ways in which purdah (customary mobility restrictions) limited their capacity to work andor join community organizations in which they could meet with other women and receive training in small businessmatters: Iwantedtodosomekindof businessin order to earnfor my houseandto bear all theexpenses myseCf:It was not possible for me, as I had the responsibilities of house and children. To be a woman was also a restriction. In villages, women hesitate to go out. The women, who go outside, are not considered respectable. That's why womenprefer to live inside. -Woman, age 42, from Faisalabad Most surprising was the high number of northern Punjabi women who, like Sindhi women, varied their dress code accordingto distance from home: 75 percentof intervieweesinTalagang and Faisalabad, and 58.3 percent in Sindh, observed strict dress code (Le., wearing a burqah) only during trips to the city or the nearby large town, but dressed more casually in their own villages. In Lodhran (Southern Punjab), the majority practiced similar levels ofpurdah inside and outsidetheir communities. Table 5.5: Women's Perceptions of the Tradeoff between Status (Practicing Purdah) and Mobility by Percent Interviewed Per Region frequencies inparentheses) Percent reporting Percent who purdah restricts perceive purdah their ability to Percent whopracticed to be an indicator work andorjoin stricter purdah in the of wealth andor in community nearby towidcity than in caste-basedstatus activities their village Northern Punjab 29.2 (7) 41.7 (10) 75 (18) SouthernPunjab 66.7 (8) 58.3 (7) 16.7 (2) Sindh 41.7 (10) 33.3 (8) 58.3 (14) All 41.7 (25) 41.7 (25) 56.7 (34) 93 5.28. There i s also a status or class effect o f women's work. In rural areas in particular, the potential costs to women who leave the household-not to mention leaving the community-to perform labor can be prohibitively high. More social than economic, these costs derive from the fact that family honor hinges so greatly on the discretion, concealment, and public perception o f female members who have reached puberty. Consequently, mobility restrictions tend to be more stringent for women belonging to wealthy or higher-class families. A majority o f women interviewed for the qualitative study (see Box 1.3) expressed reservations about working outside the home-even those from poor households in dire need of added income. The major concern was the perceived tradeoff between the incidence o f household females working outside the home and the status o f that household. 5.29. Since working in paid jobs would Figure 5.3: Impact of Husband's Schooling on Women's necessitate leaving the home on a regular basis, Labor Force Participation Rate women belonging to wealthy families are less likely to work. Analysis o f PIHS data confirms this 50% 3 I pattern. Multivariate regression results show that controlling for women's own schooling, the socioeconomic characteristic o f their household and their husband's schooling are important determinants o f their participation inthe labor force. Women whose husbands are educated and women who belong to higher-income households are less likely to undertake work, both paid and unpaid and in both rural and urban areas (Table A5.1 and No School Primary Secondary Tertiary Figures 5.3 and 5.4). This feature o f female Husband's Schooling participation in Pakistan is in contrast with patterns +Urban Rural I observed in other developing countries, like India Source: Regression reported in Table A5.1. based on data from and Thailand, where paid work by women (not PIHS2001-02 for women aged 15-49, unpaid work) rises with household income as well as with husband's education.Ig0 Figure 5.4: Impact of Household SocioeconomicStatus on Women's Labor Force ParticipationRate 5.30. A study of men's and women's work preferences from rural Punjab corroborates the ~ existence o f such a social stigma attached to 1 50%- l women's work (see Box 5.1).19' These patterns also suggest that women in higher-income households perceive less need to work to supplement the household resources than women in the poorer households, who might be pushed into work to support their households. 0% 1 2 3 4 5 5.3 1. Mobility restrictions and the social stigma Per Capita Expenditure Quintile attached to women's work in Pakistan, particularly +Urban --ERural 1 paid work, result in a strongly segmented women's Source: Predicted using regression reported in Table A5.1. Data labor market: female workers do not freely move are from PIHS 2001-02 for women aged 15-49. between markets to exploit work opportunities. The women's labor market is geographically limited to jobs to which women can easily commute, The geographical limitation arises not from a lack o f demand for women's labor in neighboring or faraway locations-indeed, Chapters 3 and 4 demonstrate a desperate need for qualified female workers in girls' "' 190Mammenand Paxson(2000) Sathar and Kazi (1997). 94 schools and health centers that serve women-but from restricted mobility. The evidence regarding increased female participation in paid agricultural work in Sindh and Punjab, discussed in Section I, confirms that female labor participation will increase with the emergence of work opportunities that accommodate the geographical limitations on women's movement. 5.32. Also useful to note i s the fact that women are less likely than men to migrate in response to employment opportunities. InPRHS-I, 98 percent o fthe few currently marriedfemale migrants migrate to marry or to join a family member after marriage, rather than in response to employment opportunities. Men may move their families to their place o f work, but family migration in response to employment opportunities for women i s much less likely. The high incidence of village endogamy also diminishes the need for a female to leave her village or settlement. Most married women report being born inthe village inwhich they reside, and most also have natal families inthe same village. InnorthernPunjab, 45 percent o f women marry within their own village, and the rates are even higher in Sindh and southern Punjab, where 59 and 53 percent o f women are in endogamous marriages, respectively. 5.33. This pattern o f limited mobility- Figure 5.5: Impact of Schooling on Earnings stemming from custom and marital practices- not only limits participation in the labor market 1 L 6400 1 but also constrains returns to such participation. I9 .E4 P 5400 1 As illustrated in the case of private schools in v) 4400 Chapter 3 and agricultural wages inthis chapter, ~ these constraints enable employers to pay ,9 3400 / female workers much less than male workers. Multivariate regression analysis shows a large and'significant male-female earnings gap among salaried jobs: women earn significantly less No School Rimary Secondary Tertiary 1 than men, even after controlling for age and Schooling education of the worker (Figure 5.5). Analysis +Males Females shows that the gender gap in earnings is widest among workers with no schooling and narrows Source: Predicted using regression reported in Table AS.2. Data are from PIHS 2001-02 for men and women aged 20-65 who are engaged as education levels increase (Table A5.2).lg2 The insalariedjobs. gender gap in earnings falls from about Rs 2,000 per month for workers with primary schooling to about Rs 1,385 for workers with education levels above secondary school. (Figure 5.5). Most o f this gender gap (about 70 to 80 percent) can be attributed to women's relatively lower schooling and fewer years o f experience in the labor market; however, a significant portion o f the gender gap remains (20-30 percent). Given the above discussion, some o f this gap is probably reinforced by the lack o f geographical mobility that most women face. 19* Since a very small proportion of female workers engage in paid work, there is a sample selection issue in estimating such a regression equation. The literature is divided over whether or not sample selection correction (using Heckman's method) is critical to such an exercise. The available proceduresfor correcting sample selection require variables that explain selection into paid work, but not earnings. Inthe absence of such identifying variables, the estimation of a sample selection-correctedearnings regressionis likely to produce inconsistent coefficient estimates. 95 5.34. The preceding discussion highlighted the role o f social barriers in Figure 5.6: Impact of Young Children on Women's Labor Force constraining female labor force Participation Rate participation. In addition to these social 25% 1 barriers and low levels of female education, women's work burdens within the home are an important constraint to participation in paid work (see Box 5.2). Women remain responsible for almost all o f the household caretaking activities. They retain primary responsibility for young children, for ill or aged family members, and for all other ' paid work 1unpaid work paid unpaid 1 ~ housework. Inrural areas this often includes URBAN URBAN RURAL RURAL 1 ~ ~ ~ ~ the care o f livestock and the production o f milk and other dairy products for household Source; Regressionreported inTable A5.1,basedon data from PIHS 2001-02 consumption. In households with young for women aged 15-49.Young childrenrefers to childrenagedthree or children, women are much less likely to younger. participate inpaid work. This i s even more the case inurban areas (see Figure 5.6), where children cannot be carried to the fields.'93 5.35. Such constraints promote a vicious cycle o f undercutting female human capital attainment and work experience: by limiting educational attainment among females, there i s no pool o f female workers from which to draw skilled workers, so such jobs do not exist for women; parents, in turn, perceive the lack o f a daughter's potential to earn money from jobs that require education, and thus tend not to educate them. Looking at the brighter side of this challenge, positive steps can reverse this cycle and promote a virtuous circle: improved labor market experience for women will convince parents of the benefits o f educating their daughters, which will create a pool of skilled female workers and promote the absorption of these women in higher-skilled, higher-paying jobs. Likewise, by having a greater sense of their employment potential, women will be encouraged to be more likely to be more active partners in decisions about activities outside the home and their earning potential, which in turn enhances their position in family dynamics. Empirical evidence shows that increases in women's participation in labor market activities tend to be associated with fertility decline (e.g., Lam and Duryea 1999). 96 Box 5.2: Improving Access to Basic Services Can Raise Women's Labor Force Participation In most parts of the developing world, women and girls are mainly responsible for collecting water and gatheringwood for h e l . Inrural areas where basic energy and water infrastructure tends to be poor, women's search for water and wood reduces the amount time they can spend on income-generating activities. Investments in time-saving infrastructure, conservation efforts, and developing markets for energy and water can greatly reduce the time women spend on household maintenancechores, enabling them to participate in earning activities. Figure 5.7: Percentage of Rural Women Who Report Inthe PRHS 2001 survey, 25 percent of women aged 15- , Time Spent Fetching Water and Gathering Firewood I 49 reported fetching water during the week prior to the I survey and about 15 percent reported gathering firewood Balochistan 60% during the same period. Women's involvement in these activities varies by province and access to such resources W P (Figure 5.7). For instance, rural areas of Sin& and Balochistan are prone to water shortages. In the PRHS Sindh 2001 survey nearly 60 percent o f women in rural Punjab Balochistan and 40 percent of women in rural Sindh I reported spendingtime fetching water during the week, in 0 FetchingWater GatherFire\*wd addition to their other chores. The highest participation in firewood collectionwas inrural Balochistan, whereabout Source Cross-tabulation based on Pakistan Rural Household 40 percentofwomen were engaged inthis task. Survey 2001 data This is tabulated for women aged 15-49 The women were asked about their participation in these activities in the week pnor to the survey Kamal's (2005) paper on "Women and Water in Pakistan," written for this Country Gender Assessment, stresses that when access to such basic resources as drinking water worsens, women, not men, bear the higher time costs. She highlights this issue against the backdrop of the alarming depletion of water resources, particularly in Sindh. Based on total availability of water, Pakistanis classified as a water-stressed and water- scarce country since water availability is just above 1,000 cubic meters per capita per year. Given the population growth, water availability is only likely to worsen incoming years, A study using multivariate regression analysis and based on national-level data shows that deterioration inthe access to water inrural Pakistansignificantly raises the amount of time women spend on water collection and also lowers the time they spend on earning activities. Such effects make women's involvement in water and drainage management projects critical. Kamal (2005) discusses the constraints that women face in being zffectively involved in such activities. She describes the experience of a drainage project in rural Sindh (Sanghar District in 1997) in which women participated. As result of this project, the domestic workload for women decreased, while their participation inwork activities increased. Source: World Bank (2001) and Illahiand Gramard(1999) 111.THERELATIONSHIP BETWEENWORKANDAUTONOMY 5.36. The sociological literature on women's autonomy (in the context o f development) tends to link it with the equally slippery concept of empowerment. While specific definitions o f empowerment vary, scholars tend to agree that empowerment for women describes the range of choices and the degree of her control, choice, and power over the decisions she regards as important to her own situation, and to that of her family. Some authors essentially equate autonomy and empowerment, arguing that both grant control over one's own life via family, community, markets, and ~ 0 c i e t y . lOther authors pointedly distinguish ~ ~ the two terms and emphasize that independence is critical to autonomy, whereas individuals may attain `94Jeejeebhoy (2000). 97 empowerment through interdependent mean~.''~ We employ this latter definition, understanding "autonomy" to refer to a woman's ability to make independent decisions and to have control over her immediate decisionmakingenvironment.It is believed that earning an income can enhance a woman's bargaining power within her family, enabling her to participate in decisionmaking. The ability to contribute to the household's earned income also can give a woman an increased sense of her own individuality and well-being, the chance to form and benefitfrom peer relationships,and a "widening of horizon^.""^ Internationalevidence lendscredence to such effects: a woman's own income, be it her own earnings or nonwage income, oftenis foundto be associatedwith her sense of autonomy and involvement in deci~ionmaking."~ 5.37. Based on the above descriptions women's labor force participation, it is clear that the relationship betweenwork and autonomy is highly complex. Women belongingto better-offhouseholds(and likely to be educated) participate very little in paid or unpaid work. These women are likely to have a say in decisionmaking insidethe householddespite the fact that they do not earn an income.'" Indeed, various studies comparing working and nonworking women's ability to influence decisionmakingwithin the family in Pakistan yield different results. Research based on the recent round of the PRHS 2004 finds that women's earned income has no effect on decisions such as a child's education, use of family planning, or the purchase of consumer durables (see Box 5.3, above).'" Other studies find that paidwork i s associated with increased autonomy and participation in household decisionmaking for working women, including decisions regarding their children's education, making large purchases, and using contraceptives.200 A study spanning 10communities across differentagro-ecologicalzones inruralPunjab finds that workingwomen tend to have more say than nonworkingwomen in some aspects of household decisionmaking (such as food preparation) but not inothers (purchase of consumer durables).201Another study based on urban women working in the manufacturingsector (including both home-basedworkers and women working outside the home), finds that despite reports of limited control over their own earnings, working women had greater authority in household decisionmaking than women who did not work.202 PaidWork as a Conduitto Autonomy outside the Household 5.38. These findings indicate that the effect of women's earned income on household decisionmaking allows no predictable pattern. It i s therefore interesting to note that, even though restrictions on women's physical mobility figure so prominently in deterring their paid work opportunities, such restrictions may not limit women's authority in decisions that affect the family. Because much of the sociologicalliteratureregards physicalmobility as a subdomainof empowerment (Malhotra, Schuler, and Boender 2002 ) and equates empowerment with autonomy, it assumes that women whose movementout of doors is restrictedalso lack autonomy-and specifically the ability to make decisions about matters both within and outside the household. The research conducted for this Assessment, however, conveys the importantdistinctionbetweenphysicalmobility and autonomy and, equally important,the distinction betweenautonomy within the householdandautonomy outsidethe household. 5.39. Other aspects of paidwork that set it apart from not working or engaginginunpaidwork involve exposure to the world outside the home. Unpaidwork mostly occurs on the family's farm or enterprise, 195 Malhotr and Mather (1997) ;Govindasamaand Malhotra (1996); Kabeer (1998). `96 Agarwal(l997); Sen (1990b). 19' See Kabeer (2000); Thomas (1990); Schultz (1990). 19* Sathar and Kazi, 2000 `99 Ongoing study by Jacoby and Mansuri, 2005. 2oo Sathar and Kazi (2000); Khattak (2001). 201 Sathar and Kazi (2000). A central idea of the study was that gender norms in Pakistan vary by agro-ecological areas due to differencesinfamilial systems and modes of production. 202 Khattak (2001). 98 so these women rarely leave the household for work-related reasons. In contrast, paid work usually involves leaving home, which provides opportunities for a woman to interact with peers, understand the needs and circumstances o f her community members, and gain a broader perspective on her own needs and circumstance^.^^^ Few Pakistani studies have tested whether women who participate in the labor force also are more likely than nonworking women to participate in community activities. One exception i s the study based on the recent round o f the PRHS 2004 referred to above. As Table 5.6 shows, rural women who engage in paid work are also more likely to be involved in community and political activities.204Since most rural women working for wages are from relatively low-income households with little power inthe community, their involvement in community activity i s o f particular interest. Several o f the women interviewed for the qualitative study (described in Box 1.3) who provide health services mentioned that, though their paid work activities marked them as relatively low-status and poor-even bringing condescension from some o f those who used their services-they also greatly appreciated the regular contact they had with members o f their respective communities. As one woman from Faisalabad remarked: People consider us poor and they come to me when they need me; otherwise if they do not have any work, they will not come. . , .Allah has blessed me. People trust me. Initially, it was a big problem in our village. There was no daai [midwife] who could administer injections. Ican administer an injection to women and can also administer the drip. -Woman who works as a midwife inFaisalabad, Punjab, age 3 1 5.40. Amartya Sen has argued that education and participation in paid work are likely to be important determinants of female voice or agency in a society's decisionmakingprocesses. Indeed, the data support this. Even though earned income may not immediately enhance a married woman's ability to participate in decisionmaking over household matters, we have shown that the increased freedom o f movement granted by labor force participation has a direct effect on participation in other aspects o f public life. The qualitative study also indicates that women engaged in paid work, particularly those who provided services to the community, were remarkably well-informed about the state o f political and economic development in their villages, patterns o f behavior among other villagers, and the nature o f gender relations among their communities. 5.41. Thus, public policies that encourage gender equality in human development outcomes and in access to income-earning opportunities can strengthen women's agency and their capacity to participate in broader community life political decisionmaking. While such policies will take time to raise women's visibility in the public and political spheres, affirmative action such as quotas or reservation o f seats for women could bring rapid results. In Pakistan, reservation o f seats for women in local, provincial, and national governments has increasedwomen's political participation in a relatively short time (since 2001). The unprecedented number o f women elected to district, subdistrict (tehsil) and union councils following the quota adoption opened up an enormous political space, and opportunities for women to make a difference in setting and implementing local government agendas. 5.42. A number of countries including India, the Philippines and Uganda have successfully used affirmative action to increase women's representation in political d e c i ~ i o n m a k i n g . ~But ~ ~ have these short-term measures to empower women actually influenced political decisionmaking? Studies based on 203Sen (1990b) 204This is based on multivariate regressionanalysis. Results are presented in Table A5.4. 2osWorld Bank (2001). 99 the Indian experience show that women representatives have, in fact, made a difference: women leaders are able to influence how resources are utilized for community development. 206 5.43. Whereas the implementation o f the quota system was achieved with few problems2", the actual impact o f drawing women into local government is not yet clear. Women councilors, for instance, lack access to development funds with which they can make meaningful improvements in their communities. The lack o f resources derives mainly from the implementation o f the devolution process and is not limited to female representatives alone; nevertheless, a number o f studies suggest that this intensifies women representatives' sense o f being ineffective leaders.208This sense o f ineffectiveness was echoed by responses from the few women Union Councilors interviewed for the qualitative study. A more problematic constraint is women leaders' lack o f knowledge about political processes. Most o f the women elected in the 2000-01 local government elections were new to governmental decisionmaking and had little knowledge o f their rights, roles, and responsibilities as councilors. Recognizing this, several government agencies and nongovernmental organizations have sponsored skill-building activities to increase the governance capacities o f elected female councilors.209 These efforts have become increasingly critical as the country moves toward local government elections in 2005. In a paper written for this Assessment, Naz (2005) assessed the effectiveness o f these training programs. One o f the main failures o f the training programs has been an inability to teach councilors the basic rules o f functioning in the council, such as how to introduce resolutions in the council, maintain links with the district offices, and address the demands o f voters. 5.44. The female electorate (or constituency), which i s crucial to making women's "voice" louder and more effective in the political arena, also face a number o f gender-specific limitations, including access to information and cultural constraints. First, women are significantly less informed than men about political matters. The qualitative study informing this Assessment (see Box 1.3) reveals that awareness o f local government and the existence o f the quota for women's seats in government i s extremely low. In data available from two Northern Punjab sites, only three out o f 24 women surveyed knew the name o f the Union Nazim; none knew the names o f the Tehsil Nazim; two out o f 24 knew the name o f the District Nazim; none knew the name o f the Chief Minister; only three knew the name o f the Prime Minister, and less than one-half (11 out o f 24) knew the name o f the President o f Pakistan. Women interviewed in southern Punjab and in the two Sindh sites were even less informed about political matters and faced greater restrictions on their ability to vote, mostly due to impositions placed on them from male family members. Out of the 24 women interviewed in Sindh, eight knew the name o f the President, three o f the Prime Minister, and two of the Chief Minister. In Lodhran, southern Punjab, the collective knowledge about political representatives included one woman knowing the name o f the President and two women thinkingthe President was the Prime Minister. 5.45. Second, women's political involvement also is inhibited by restricted mobility and the practice o f purdah, both o f which strengthen local resistance to legal reform. Inone o fthe five interview sites, Badin, none o f the 12 women interviewed had ever voted. Several o f these women reported that they were not allowed to vote by the men in their families: 206 Analysis by Chattopadhyay and Duflo (2001) shows that women heads of local government allocate more resources to developmentof basic infrastructure(water, fuel, and roads) that affect women's work burdens. 207 Women in a number of districts in NWFP were prevented from contesting the elections and voting in the elections. (Aurat Foundation) 208 Naz (2005). '09 Examples include the Women's Political Participation Project (W3P) of the Ministry of Women's Development, Social Welfare and Special Education; and the Citizens' Campaign for Women's Representationin Local Government convened by the Aurat Foundation. 100 No, Ihave never voted Only our men gofor voting; we are not allowed to go. . . . rfwe were allowed to vote, we would have votedfor the samepeople who our men votefor. -Married woman, Badin, Sindh 5.46. The experience reported by women in Mirpurkhas-another district in Sindh Province-was slightly more promising: although only two o f the 12 interview subjects had voted, both had made their own decisions about candidates. One o f these was the Union Councilor, who had contested a reserved seat, In southern Punjab, a majority o f the women (10 out o f 12) had voted in the most recent elections, but for the most part, their decisions were determined by men inthe extended family (usually the head o f the biradari) who told everyone in the extended family whom they should vote for. The majority o f northern Punjabi women had voted in the previous elections, but had also done so based on instructions from their husbands, brothers, or fathers-in-law. InFaisalabad, which had by far the highest voter turnout (75 percent) among women interviewed, only two o f the nine women interviewed who voted did so according to their own choice. Several women in Talagang also expressed the view that for women to engage in any political activity beyond voting-such as participating in political rallies and campaigns, or contesting elections themselves-was socially unacceptable: There is no tradition of women going outfor such purposes except voting. Theyjust go outfor voting. They have no other responsibility in party aflairs or conflicts. Men do not like women to be involved inpolitical gatherings. It is considered against their honor. -Married woman, Talagang, Punjab, age 20 5.47. These views point to the need for more efforts to better train women candidates for local government seats, including knowledge o f the political process, as well as information campaigns that inform women about their right (and obligation) to vote. Any efforts to strengthen and increase female participation in the political process are likely to see better success ifthey also strengthen and support the several NGOs currently working on these issues inthe country. 5.48. Women's limited access to information about the importance o f their political participation has implications for all public policy aimed at increasing gender equality. Data on female political participationhave revealed that women's access to information is extremely limited due to restrictions on female mobility and girls' low levels o f education. Other than hearsay from male family members and children who happen to attend school, women who do not work have minimal contact with the outside world. They obtain most of their information from television, radio, and newspapers. Government and NGO-sponsored media public information programs aimed at educating women about their rights to health care and legal protection have been successful in changing behaviors o f women. (see, for example, the excerpted interview with a woman from southern Punjab, in Chapter 2, Box 2.1). The government's commitment to increase rates of girls' enrollment and schooling no doubt will bring dramatic improvements in women's access to information, including knowledge o f the political process, over the longerterm. . 5.49. For the current generation of adult women, the greatest potential instrument for enhancing women's knowledge and status lies in community organizations for women. Numerous NGOs are working to form and strengthen community-based organizations. Although there i s little evidence to indicate how successful their efforts have been, particularly in mobilizing rural women, a formal evaluation o f one such initiative is currently underway. This evaluation should tell us much more about the role of participatory development efforts in empowering rural women.210Meanwhile, reports from governmental oversight organizations, NGOs, and interviews from the qualitative study (Box 1.3) for this assessment have contributed to a growing pool o f feedback on rural women's experience with community *loA research team led by researchers from the Development Economics Research Group (DECRG) within the bank are currentlyundertakingan evaluation of the National Rural SupportProgram(NRSP). 101 organizations (COs).*'l According to these sources, COS have been successful at involving and mobilizingwomen in rural areas to the extent that these women are (1) made aware of the existence of COSand the opportunitiesthey provide; and (2) allowedby male family membersto participate. Many ruralwomen interviewedfor the qualitativestudy hadnot heardofwomen's COS,evenwhentheir village hadcreatedone, suggesting lack ofsufficientpublic outreach: Nobody ever asked me tojoin the CO or to become a member. If they (had) contacted (me), and had my husband allowed, Iwould havejoined the CO; otherwise, Iwouldnot have. -Woman from Badin, Sindh, age 35 5.50. Ofthe women who hadheardof a CO intheir community, few hadknowledgeabout its functions or benefits or were able tojoin because of householdresponsibilitiesor culturalprohibitionsagainsttheir participationinpublicfora: Ihaveheardthatthey[COS]givecredit. Nooneeverapproachedme. ...Iwasalone in the house and did not get spare time; therefore, I have been restrictedjust to household responsibilities. No other elder, like my mother-in-law or relative lives with me. So Ido not get the time. Otherwise,Iam interested injoining CO. -Mother (age 31) o fthree childrenbelow age 7, Faisalabad, Punjab 5.51. When women participate in COSthe benefits extend to many domains of life. For many, credit schemes offered through COShave provided funds for small household-basedeconomic activities and training for economicallyremunerative work. Some interview subjects also shared that CO membership had increasedtheir self-confidenceas well as their power and status within the household.Many reported a greater sense of freedom due to increasedmobility: Iusedtogofor mytrainingtoattendmyclasseswithburqahandsitthereinthat manner. But my mobility has increased becauseIhave a responsibility now. My knowledge has increased and Ihave gained more conjdence as we meet different people. -Woman's CO President, ruralPakistan, age 29 My husband's attitude has changed a lot since Ibecame a CO member. He is very nice to me now and respects me. -Woman CO member from Badinwhose husbandalso is a member of a CO, age 22 5.52. The qualitativedata exhibits a strong positivecorrelationbetweenwomen's CO membershipand their sense of autonomy and well-beingwithin their households, but it is hardto pin downthe exact cause o f the improvedhousehold status of female CO members. Many o f the women who joined COSwere persuadedor at least allowed to join by husbands who were members of a communityorganizationfor men; in many cases, a husband's membership appeared to be a necessary pre-condition for a wife's participationin a CO far women. Unfortunately the data are unable to tell us whether malemembers let their wives join COSbecause these men were prone to value some degree of autonomy in their wives (making them more likely to be interested in communitydevelopment in the first place) or whether they lettheir wivesjoin COSbecause of learningto value women's autonomy from their own COS. 5.53. An interestingaspect of the qualitativestudy is the resounding emphasis placedby all the women interviewedon their desire to have greater input in decisionmakingthat affects their communities, and their ability to support education for their daughters. Even in regions like Sindh and southern Punjab, *I1The qualitative used inthis gender assessment was co-sponsored by the team conducting the evaluation o f the NationalRural Support Program. See footnote 40. 102 where female literacy and schooling rates tend to be the lowest and female mobility the most restricted, respondents were unanimous intheir support for higher female education and greater involvement in the public sphere as well as within their households. IdidnotbecomeaCOmemberbecauseofpurdah; Idon'tknowwhether otherwomendo not become members because of the same reason. The CO should sensitize our men and explain them to give us fieedom and let us become CO members. Had there been no strictness, we would have become members just like Zareena [another woman in the village whojoined the women's CO]. Wewouldgo out and do the shopping on our own. I think now this spirit of seeing the world has developed in Sindhi women. -Married woman, Mirpurkhas, Sindh Province IV. PROMOTINGWOMEN'S INVOLVEMENT INTHELABOR FORCE: RECOMMENDATIONS FORPUBLICPOLICY 5.54. Effective policy interventions will require comprehensive strategies to encourage women's entry into the labor force and their continued participation. A four-part strategy to address the range of constraints to women's participation, which are multifaceted and mutually reinforcing, involves improving the employability o f female workers, adopting employment programs that speak to the special nature o f labor markets for women, investing in infrastructure that alleviates constraints on the amount o f time women can devote to income-earning activities, and fostering a legal environment that encourages women's labor force participation. Investin FemaleWorkers' Skills 5.55. Education makes workers attractive to employers, and female workers lose out to their male counterparts because of their low schooling on average. Parents will invest intheir daughters' education if returns to schooling in the labor market are clearly visible to them. One way this cycle o f less visible labor market returns and low female education can be broken i s through policies that promote female education (see Chapter 3). 5.56. Another way to improve the employability o f female workers is to provide vocational education and training that builds specific skills for which there i s demand. For example, the Lady Health Worker (LHW) program operating in rural areas and poor urban neighborhoods has been successful in recruiting female health workers to do outreach work. These women have a middle-school level education and no prior exposure to medical education. The LHW program trains the outreach workers to deliver a range o f essential primary health care services to women and their children. Attend to the SpecificNature of Women's Labor Market 5.57. The availability of educated and skilled women alone will not improve employment prospects. The experience from countries in the Middle East and North Africa shows that despite the generally high level o f education obtained by women there, their labor force participation rates continue to remain low because o f mobility restrictions and socio-cultural practices that discourage female participation in Pakistan shares many o f these socio-cultural practices, and this chapter shows how such practices restrict work opportunities locally and segregate men and women into gender-specific jobs. As the 212World Bank, (2004a). 103 experience o f the Middle East and North African countries illustrates, programs mustbe designed to work around the constraints imposed by the nature o fthe labor market for women. 5.58. Affirmative actions such as quotas for women are often introduced to increase the number o f women participating in certain types o f jobs; however, such affirmative action may be difficult to implement given the nature o f the women's labor market. While Pakistan has been successful in increasing the number o f women in local government through a quota system, its experience with the five-percent quota for females in the public sector has shown that such measures by themselves may not be effective in raising the number o f female employees in the public sector. A study conducted by Pakistan's National Commission on the Status o f Women found that a major obstacle to the implementation o f the quota was that women tended to be confined to certain government departments (such as education and health) considered to have appropriate jobs for females. International evidence on the success o fjob quotas to improve women's participation in formal sector jobs suggests that quotas can improve gender equality in employment opportunities if there are no distortions in labor markets. Given the labor market conditions for women, it may be more effective to try complementary strategies. 5.59. One such complementary strategy is to encourage employers (public sector departments) to hire locally to overcome the limited mobility o f female workers. The growing number o f private schools in rural areas have recruited female teachers almost exclusively from local communities, which also ensures less teacher absenteeism. 5.60. Home-based work by women is another labor market where the government can play a facilitating role. It can sponsor entrepreneurship training for women engaged in home-based work, including training in relevant skills and business opportunity identification, as part of a larger package including credit schemes, marketing support, and access to new technology. 5.61. Inthe agricultural sector, knowledge about how to obtain credit and agricultural information will enhance income-producing opportunities for rural women. Because agricultural tasks tend to be gender- specific, the work o f seed preparation, fertilizing, and threshing falls to women, and these activities receive little or no support from agricultural extension workers. Raising and tending livestock also falls primarily to women, and there is inadequate support available to women engaged in these activities to market their products. Technical know-how thus must be provided to women in agriculture, and the design o f agricultural extension work needs to include women's tasks and needs. 5.62. Improving information and documentation systems is another critical labor market policy. Issues with measurement o f women's labor market participation arise from the socio-cultural conditions that shape women's labor market activity. These measurement issues have to be addressed in order to gather accurate and timely information about the economic role o f women. Policy makers at the federal and local government levels need such information to design more effectively targeted social protection programs and labor market policies and programs that deliver training and critical resources such as credit. 5.63. Awareness-building campaigns that promote female work and provide employment-related information have been shown to encourage women's participation in the labor market. In the 1970s, Jordan introduced an awareness campaign to promote women's participation in work. This campaign- which stressed the need for women in the labor force and the benefits that women derive from working- i s reported to have effectively raised female participation rates.*13InPakistan information on employment opportunities, wages, job vacancies and hiring requirements could also be provided through community organizations. World Bank (2004b). 104 Invest in Time-savingInfrastructure to Reduce Women's Time Burden 5.64. Investment in time-saving infrastructure is critical to encouraging women's participation in work activities. Women in rural areas spend a large part o f their time collecting water and fuel wood for the household. These activities leave them less time to participate in income-generating activities, According to Kamal(2005), households in some parts o f rural NWFP could save as much as 1,200 hours per year if water was available within the home. Given that local water sources are becoming depleted, the time burden on women to fulfill these tasks i s likely to increase inthe future. Creatinga Legal Environment that EncouragesWomen's Labor Force Participation 5.65. The legal environment can be modified in two ways to encourage women to join the labor market. First, the government can repeal labor laws that limit an individual's work hours on the basis o f hidher sex. According to the government's periodic report to the Committee on the Elimination o f All Forms o f Discrimination Against Women (CEDAW), a prominent example o f such a law i s TheFactories Act, 1934. Women can work in factories between 6 a.m. and 7 p.m., and in any other establishment between 9 a.m. and 7 p.m., unless the government permits The government i s considering relaxing this law to allow women to work night shifts, thus allowing factories to raise production capacity and international competitiveness, while at the same time dramatically increasing women's opportunities for paid work. International experience shows that young women stand to gain the most from such measures, as they tend to be employed in large numbers in such sectors as apparel and information technology. 5.66. Because many laws limiting women's work hours were created in part to protect the safety o f females, repealin such laws may require complementary legal measures to keep safeguards for working women in The government would need to enact laws-or enforce existing ones-to (1) protect the physical safety o f women who venture outside their homes and/or communities for paid employment (particularly ifthey do so for night shifts); and (2) protect women, who comprise the majority o f available cheap labor, from being exploited in the workplace. To ensure the first type o f protection, companies could offer door-to-door transportation for female workers. Conclusion 5.67. The policies proposed above are targeted toward increasing women's participation in the labor market and expanding the public domain in which they can participate. These policy suggestions range from improving women's skills and education to adopting laws that ease mobility constraints on female participation. Ultimately, returns to education and increased earning capacity will be realized as mobility constraints are eased and women can access work opportunities, wherever they may be. Because these same mobility restrictions impede women's ability to participate inpolitical and community activities, the easing o f such restrictions also will enable full realization o f the government's active encouragement of women in seeking roles in the political realm through seat reservation and other constitutional measures to enhance their participation. 5.68. Realization of such policy initiatives in labor market participation, the political process, and community decisionmaking eventually will foster an environment that naturally reduces the need for mobility constraints on females, which arise out o f safety and family honor concerns, Thanks to devolution and the government's commitment to empowering women, it i s time to focus on initiatives at the local level to further involve women in the development o f their communities. The obvious role for 214CEDAWreport from Pakistan 215World Bank (2004). 105 international donors is to continue to support government initiatives-as well as NGO efforts already on the ground-as they work to increase women's access to information, encourage their active participation in the political process, and enhance their informed involvement in community organizations and decisionmaking. The demand i s there and growing to receive such targeted efforts. 106 BIBLIOGRAPHY Abu-Ghaida, Dina and Stephan Klasen, 2004. "The Costs of Missingthe MillenniumDevelopmentGoal on Gender Equity," World Development, Elsevier, vol. 32(7), pages 1075-1107. Aganval, Bina, 1997, "Bargaining" and gender relations: Within and beyond the household, Feminist Economics, 3( l), 1-51. Alderman, H., J.R. Behrman, V. Lavy and R. Menon, 2001, Child health and school enrollment: A longitudinal analysis, Journal of Human Resources, 36(l),185-205. Alderman, H., P. F. Orazem and E.M.Paterno, 2001, School quality, school cost and the public/private school choices of low income households inPakistan, Journal of Human Resources, 36(2), 304-326. Alderman, H., J.R. Behrman, D. Ross, and R. Sabot, 1996, The returns to endogenous human capital in Pakistan's rural wage labor market, Oxford Bulletinof Economics and Statistics, 58( l), 30-55. Alderman, H. and M. Garcia, 1993, Poverty, household food security and nutrition in rural Pakistan, ResearchReport 96, Washington DC: World Bank. Alderman, Harold and Paul Gertler, 1997, Family resources and gender differences in human capital investments: The demand for children's medical care in Pakistan, in L. Haddad, J. Hoddinott and H. Alderman, eds., Intrahousehold resource allocation in developing countries: Models, methods and policy, Johns Hopkins University Press, Baltimore andLondon. Ali, Shaheen Sardar, 2000, Gender and human rights in Islam and international law: Equal before Allah, Unequalbefore Man? The Hague, Kluwer Law International. Andrabi, T., J. Das and A. Khwaja, 2002, The rise of private schooling in Pakistan: Catering to the urban elite or educating the rural poor? Background Paper for the Pakistan Poverty Assessment, Washington DC: World Bank. Ashford, Lori, 2002, Hidden suffering: Disabilities from pregnancy and childbirth in less developed countries, Policy Brief, Population Reference Bureau, Washington, D.C.. Aurat Foundation, Legislative Watch Programme Publications, various years, Suggested additiondamendments in Muslim Family Laws: The Child Marriage Restraints Act, 1929 (XIX of 1929); The Dissolution of MuslimMarriages Act, 1939 (VI11of 1939); Nikahnama Form. Aurat Publication and Information Services Foundation and Citizen's Campaign for Women's Representation in Local Government, 2001, Gross Violation of Women's Electoral Rights in Swabi, Mardan and Dir,NWFP. Awan, Zia Ahmed, 2004, Violence against women and impediments in access to justice, Background Paperfor the Pakistan Country Gender Assessment, Washington DC: World Bank. Behrman, J. and A. Deolalikar, 1988, Health and Nutrition, in H. Chenery and T.N. Srinivasan (eds.) Handbook of DevelopmentEconomics, Vol. 1, Amsterdam: North Holland. Behrman, J.R., 1992, Intrahouseholdallocation of nutrientsand gender effects: A survey of structural and reduced form estimates, in S. Osmani, ed., Nutrition and Poverty, Oxford: Clarendon Press, World Institute for Development Economics Research (WIDER) series on "Studies in Development Economics", 287-230. Behrman, J.R., 1996, The impact of health and nutrition on education, World Bank Research Observer, 11(1), 23-37. Berhman, J., A. Foster, M.Rosenzweigand P. Vashishtha, 1999, Women's schooling, home teaching and economic growth, Journal of Political Economy, volume 107 (4). Casterline, J. B., Z. Sathar, and M.ul-Haque, 2001, Obstacles to contraceptive use inPakistan: A study in Punjab, Studies in Family Planning, 32,95-110. Cernada, G.P., A.K.U.Rob, S.I. Ameen, and M.S.Ahmad, 1993, A situation analysis of family welfare centers in Pakistan, Islamabad: Population Council. Chattopadhyay, R. and Esther Duflo, 2001, Women as policy makers: Evidence from a India-wide randomized policy experiment, National Bureau of Economic Research (NBER) Working Paper No. W8615. Chaudhury, M.Ghaffar and Zubeda Khan, 1987, Female labour force participation rates in rural Pakistan: Some fundamental explanations and policy implications, Pakistan Development Review, 26 (4), 687 - 696. Chen, L.,E. Haq and S. D'Souza, 1981, Sex bias in the family allocation of food and health care in rural Bangladesh, Population and DevelopmentReview, 7(1), 55-70. Coale, A., 1991, Excess female mortality and the balance of the sexes, Population and Development Review, 17, 5 17-523. Commission of Inquiry for Women, Pakistan, 1997, Report of the Commission of Inquiry for Women, Lahore: Shirkat Gah Women's ResourceCentre. Das Gupta, Monica, 1987, Selective discrimination against female children in rural Punjab, India, Population and Development Review, 13(l), 77-100. Deaton, Angus, 1998, The analysis of household surveys, Johns Hopkins University Press. Deolalikar, 1993, Are there differential returns to schooling by gender? The case o f Indonesian labor markets, Oxford Bulletinof Economics and Statistics, 57, 97-118. Fafchamps, M.and A. Quisumbing, 2001, Social roles, human capital and the intrahousehold division of labor: Evidence from Pakistan, IFPRIworking paper. Fauveau, V., 1994, Matlab: Women, Children and Health, ICDDR,B Special PublicationNo. 35. Federal Bureauo f Statistics (FBS), 2003, Pakistan Labour Force Survey, 2001-02. Filmer, Deon, Elizabeth M.King and Lant Pritchett, 1998, Gender disparity in South Asia: Comparisons between and within countries, Policy Research Working Paper 1867, World Bank, Development ResearchGroup, Washington, D.C. Glewwe, Paul, 1999, Why Does Mother's Schooling Raise Child Health in Developing Countries? Evidence from Morocco, Journal of Human Resources. Greene, M., 1987, Intercensal change and the indirect estimation of mortality: The case of Pakistan, Pakistan DevelopmentReview, 25 (4), 569-582. Government of Pakistan, 2003, Poverty Reduction Strategy Paper. ............................ , 2004, National Programme for Family Planning and Primary Health Care: Promoting health, reducingpoverty. Haddad, Lawrence and J. Hoddinott, 1994, Women's income and boy-girl anthropometric status in Cote d'lvoire, World Development, 24(4), 866-881. Haider, M. and M.Tahir, 2002, Mapping of organizations working for women homeworkers in Pakistan, Aurat Foundation, Lahore, Pakistan. Hamid, S., A micro analysis o f demand side determinants of schooling in urban Pakistan, Pakistan Development Review, 32(4)(part II), 713-723, 108 Hazarika, Gautam, 2000, Gender differences in nutrition and access to health care in Pakistan, Journal of Development Studies, 37(1). Hennink, M., I.Rana and R. Iqbal, 2004, Knowledge of personal and sexual development amongst the young people inPakistan, Opportunities and Choices Working PaperNo. 12. Holmes, J., 1999, Measuring the determinants of school completion in Pakistan: Analysis of censoring and selection bias, Economic Growth Center Discussion Paper No. 794, New Haven, CT. Illahi, Nadeem and Franque Grimard, 2000, Public infrastructure and private costs: water supply and time allocation of womenin rural Pakistan, Economic Development and Cultural Change, Vol. 49, No. 1 pp.45-76. Irfan, Mohammad, 1985, Poverty and household demographic behavior in Pakistan - Insights from PLM survey 1979, Islamabad: PIDE (Studies in Population, Labor Force and Migration, Project Report No. 11). Islam, Nasir, 2002, Local level governance: Devolution and democracy in Pakistan, Paper presented at Workshop on democratization, April 2002, University of BritishColumbia, Vancouver, Canada. Iyer, Aditi and Amar Jesani, 1999, Barriers to the Quality of Care: The Experience of Auxiliary Nurse- Midwives in Rural Maharashtra, in Michael A. Koenig and M.E. Khan (eds.) Improving Quality of Care in India's Family Welfare Programme, Population Council, p.210-237. Jejeebhoy, S. and Z. Sathar, 2001, Women's autonomy in India and Pakistan: The influence of religion and region, Population and Development Review, 27(4). Kabeer, Naila, 2000, Thepower to choose: Bangladeshi women and the labor market decisions in London andDhaka, London andNew York: Verso. Kazi, Shahnaz, 1999, Gender inequalities and development in Pakistan, in Shahmkh Rafi Khan, ed., Ffty years of Pakistan s economy, Oxford University Press. Kazi, S. and B. Raza, 1991, Duality of female employment in Pakistan, 30(4) Part 11, Pakistan Development Review, 733-743. Khan, Ayesha, 1998, Female mobility and social barriers to accessing health and family planning services: A qualitative research study in three Punjabi villages, Islamabad: Ministry for Population Welfare, London School of Tropical Hygiene and Medicine and Department for International Development, BritishGovernment. Khan, Ayesha., 2000, Adolescents and reproductive health in Pakistan: A literature review, Research report No. 11, Islamabad: Population Council and UNFPA. Khan, S. R., 1993, South Asia, Chapter 6 in King and Hill (eds) Women's education in developing countries: Barriers, benefits and policy, Baltimore, Maryland: JHUPress. Khattak, S., 2001, Women work and empowerment, Chapter 6 in Sayeed and Khattak (eds) Women's work and empowerment issues in an era of economic liberalization: A case study of Pakistan's urban manufacturing sector, Pakistan Institute of Labour Education and Research (PILER) and Sustainable Development Policy Institute. Kim, J., H.Alderman and P.F. Orazem, 1998, Can cultural barriers be overcome in girls' schooling? The community support program in rural Balochistan, Working Paper Series on Impact Evaluation of Education Reforms, No. 10, Washington DC: World Bank, Development ResearchGroup, Kim, J., H. Alderman and P.F. Orazem, 1999, Can private school subsidies increase enrollment for the poor? The Quetta Urban Fellowship Program, World Bank Economic Review, 13(3), 443-465. Klasen, S., 1994, Missingwomen reconsidered, World Development, 22(7), 1061 1071. - 109 -----___----1999, Does gender inequality reduce growth and development? Evidence from cross country , regressions, World Bank Policy ResearchReport Working PaperNo. 7, Washington, DC. Klasen, S. and C. Wink, 2002, "Missing women": Revisiting the debate, FeministEconomics, 9,263-299. Knowles, S., P.K. Lorgelly, and P.D. Owens, 2002, Are educational gender gaps a brake on economic development? Some cross country empirical evidence, Oxford Economic Papers, 54, 118-149. Krotki, J., 1985, Reported masculinity ratio in Pakistan, Pakistan Development Review, 24 (Nos. 3 and 4), 267-297. Lanjouw, Peter and Martin Ravallion, 1999, Benefit Incidence and the Timing of Program Capture, World Bank Policy ResearchWorking Paper No. 1956. Lloyd, C., Cem Mete and Z. Sathar, 2002, The effect of gender differences in primary school access, type and quality on the decision to enroll in rural Pakistan, Policy Research Division Working Paper No. 164, Islamabad: PopulationCouncil. Loevinsohn, B. and A. Harding, 2004, Purchasing basic health services inthe community setting, Health, Nutritionand PopulationDiscussionPaper, Washington, DC. Mammen, K. and C. Paxson, 2000, Women's work and economic development, 14 (4), Journal of Economic Perspectives, 141-164. Mannan, M.A. (ed.), 1995, D.F.Mulla's principles of Mahomedan law, Lahore:PLD Publishers. Mansuri, G., 1994, Female labor market participation in Pakistan: A comparative analysis of LSMS data for Pakistan, Working Paper, World Bank. Mehdi, Rubya, 2002, Gender and property law in Pakistan: Resources and discourses. Lahore: Vanguard Books (Pvt) Ltd. Mian, R.M.A., M. Ali, P.A. Ferroni and P. Underwood, 2002, The nutritional status of school-aged children inan urban squatter settlement inPakistan, Pakistan Journal ofNutrition, 1(3), 121-123. Miguel, E. and M. Kremer, 2001, Worms: Education and health externalities in Kenya, NBER Working PaperNo. W8481. Miller,B.D., 1981, Theendangeredsex, CornellUniversityPress, Ithaca, NewYork. Montgomery, M., J. B. Casterline, and F. Heiland, 2001, Social networks and the diffusion of fertility control, Working Paper, Population Council, New York. Mumtaz, K. and F. Raouf, 1996, Woman to woman: Transfer of health and reproductive knowledge, Lahore: Shirkat Gah. Mumtaz, K. and F. Shaheed, 1987, Womenof Pakistan: Twostepsforward one step backward,Zed Press, London. Mumtaz, M. and Uzma Khan, 2000, Literature review on needs assessment among rural women in Pakistan,FirstEdition, Aurat Publicationand Information Service Foundation. Mumtaz, K., and S. Salway, 2005, "Inever go anywhere": Extricating the links between women's mobility and uptake of reproductive health services inPakistan, Social Science and Medicine, 60, 1751- 1765. Munshi, K.and J. Myaux, 2002, Developmentas a process of social change: an applicationto the fertility transition, Working Paper, Brown University, Department o f Economics. National Commission on the Status of Women, 2003, Inquiry report on status of women employment in public sector organizations, Inquiry Report 1. 110 National Institute of Population Studies, 2001, The Pakistan Reproductive Health and Family Planning Survey 2000-01. National Reconstruction Bureau, 2002, Social audit of governance and delivery of public services, Baseline Survey. Naz, Rakshanda, 2005, Training of women councilors, Background Paper for the Pakistan Country Gender Assessment, Washington DC: World Bank. Oxford Policy Management, 2002, Lady Health Worker Programme: External evaluation of the National Programmefor Family Planning and Primary Health Care. Quisumbing, A., 1996, Male-female differences in agricultural productivity: methodological issues and empirical evidence, World Development 24(lo), 1579-1595. Qureshi, Sarfraz Khan, Hina Nazli and Ghulam Yasim Soomro, 2001, Nutritional status in Pakistan, MIMAP Technical Paper Series No. 8, Pakistan Institute of Development Economics, Islamabad, Pakistan. Ray, Ranjan, 2000, Analysis of child labor in Peru and Pakistan: A comparative study, Journal of PopulationEconomics, 13 (l), 3-19. Rami, Majida, 2004, Provisions under family laws and juvenile justice, Background Paper for the Pakistan Country Gender Assessment, Washington DC: World Bank. Reyes, S., 2002, Quotas in Pakistan: A case study, Paper prepared for Regional Workshop on the Implementation of Quotas: Asian Experiences. Sathar, Zeba, 2001, Fertility in Pakistan: Past, present and future, Paper presented at Workshop on prospects of fertility decline inhighfertility countries, UnitedNations Population Division. Sathar, Zeba A. and John B. Casterline, 1998, The Onset of Fertility Transition in Pakistan. Population and Development Review, 24(4),p.773-796. Sathar, Z. and S. Kazi, 1987, Variations in demographic behavior by levels of living in Pakistan, Genus, 43, 113-136. Sathar, Z. and S. Kazi, 1997, Women's autonomy, livelihood and fertility: A study from rural Punjab, Pakistan Insitute o f Development Economics (PIDE). Sathar, Z. and S. Kazi, 2000, Women's autonomy and in the context of rural Pakistan, Pakistan Development Review, 39(2), 89-110. Sathar, Z. and C. Lloyd, 1994, Who gets primary schooling in Pakistan: Inequalities among and within families, Pakistan DevelopmentReview, 33(2), 103-134. Sathar, Z. and M. ul Haque, A. Faizunissa, M. Sultana, C. Lloyd, J. Diers and M. Grant, 2003, Adolescents and youth in Pakistan 2001-02: A nationally representative survey, IslamabadNew York: Population Council. Sawada, Yasuyuki and Michael Lokshin, 1999, Household Schooling Decisions in Rural Pakistan, World Bank Policy ResearchWorking PaperNo. 2541. Sawada, Y. and M. Lokshin, 2001, Household schooling decisions in rural Pakistan, Policy Research Working PaperNo. 2541, Washington DC: World Bank, Development ResearchGroup, Schultz, T. P., 1989, Benefits of educating women, Background Papers Series, Population and Human ResourcesDepartment, Washington DC: World Bank. Sen, A., 1989, Women's survival as a development problem, Bulletinof the American Academy of Arts and Sciences, 43, 14-29. 111 Sen, A,, 1990, More than 100 million women are missing, The New York Review of Books. Siddiqui, Rehana, 2002, Gender dimensions of poverty in Pakistan: An examination based on household survey data, in Christopher Edmonds and Sara Medina (eds) DefiningAgenda for Poverty Reduction, Volume 1, Manila, Phillippines: Asian Development Bank. Siddiqui, Rehana and Rizwana Siddiqui, 1998, "A decomposition of male-female earnings differentials," Pakistan DevelopmentReview, vol. 37 (4). Strauss, J. and D. Thomas, 1995, Human resources: Empirical modeling of household and family decisions, InJ. Behrman and T.N. Srinivasan (eds) Handbook of Development Economics, Volume 3A, Chapter 34. Sultan, M.,J. Cleland, and M.M.Ali, 2002, Assessment of a new approach to family planning services in rural Pakistan, American Journal of Public Health, 92, 1168- 1172. Summers, L., 1992, Investing in all the people, Policy Research Working Papers, Report No. WPS 905, Washington DC: World Bank. Tinker, Anne, 1998, Improvingwomen's health inPakistan, Health, Nutrition and Population Series. Thomas, D., 1990, Intrahousehold resource allocation: An inferential approach, Journal Human Resources,25(4), 635-664. Webb, P. and S. Block, 2004, Nutrition information and formal schooling as inputs to child nutrition, Economic Development and Cultural Change, 52(4). Welch, F., 1970, Education in production, Journal of Political Economy, 78, 37-59. World Bank, 1989, Women in Pakistan: An economic and social strategy, World Bank Country Study, Washington, D.C. -______-------,1993,WorldDevelopmentReport: InvestingInHealth, NewYork, OxfordUniversityPress. -------______-,2001, Engendering development through gender equality in rights, resources, and voice, Policy ResearchReport, Washington DC ---------_____,2002, PovertyinPakistan: Vulnerabilities, social gaps, andruraldynamics, PakistanPoverty Assessment, Washington DC. -__-----------,2003, Poverty and Social Development in Pakistan: an Update usingHousehold Data, Policy Note, SouthAsia Region. _-------------, 2004a, Devolution in Pakistan: Assessments and recommendations for action, Washington, DC. _________----- Gender and development inthe Middle East and North Africa: Women inthe public ,2004b, sphere, MENA Development Report, Washington, DC. -----------__-,2004c, World Development Indicators. ____-__------- , 2004d, Pakistan Public Expenditure Management: Strategic issues and reform agenda, Volume 1, Washington, DC. _------------, 2004e, Vietnam Population and Family Health Project Implementation Completion Report and project supervision documents. Zia, S. and F. Bari, 1999, Baseline Report on Women'sParticipation in Political and Public Life, Aurat Publicationand Information Services Foundation, Lahore. 112 ANNEXES ANNEX 2.1: EVOLUTIONOFINHERITANCE LAWREGARDING WOMEN IN PAKISTAN For most of colonial rule, the customary law denying women's right to inherit overrode the Islamic (Shariat) law stipulating this right. Shortly before partition, the Muslim Personal Law (Shariat) Application Act in 1937 enacted women's right to most forms of inheritance all over India, though its power was limited: it excluded women's right to inherit land, and it also failed to apply the new law to those who had obtained inheritance interests under customary law prior to 1937. According to Mehdi (2002), the law was therefore widely criticized by those who supported women's rightsto property. The establishment of Pakistan brought new legislation, The Punjab Muslim Personal Law (Shariat) Application Act (Ur of I948), which did includeagriculturalland inwomen's inheritancerights. ANNEX 2.2: DETAILS ONMARITALPRACTICESAND MARRIAGELAWINPAKISTAN Effectsof Child Marriage 1, The problems associated with child marriage are well-known to human development experts and human rights advocates in Pakistan. Early marriages choke off the human development potential o f children in myriad ways. This is particularly true for girls, whose early marriage typically means the cessation o f school attendance and markedly increased health risks: girls who become pregnant by age 10-14 are five times more likely to die than women twice their age. Because purdah practices and mobility restrictions on females are severe in parts o f Pakistan, moreover, early marriages are likely to confine females to the marriage household for the great majority o f their lives, compounding their deprivation of education and work opportunities with diminished opportunities for community activities, interactions with their peers, and the development o fmeaningful social relationships.216It i s believedthat the incidence of child marriage remains quite high in contemporary Pakistan; Unicef has estimatedthat 37 percent o f rural children and 32 percent o f urban children are in such marriages (1986-2003).217 These estimates may be misleading, however, as they likely define child marriage as that involving girls under 18 years o f age, in accordance with the international standard o f the UN Convention on the Rightso f the Child. Although Pakistan did ratify the Convention on the Rights o f the Child as early as 1990, there is no such law, as yet, that enforces marriage age minimum o f 18 for girls. Currently, marriage o f girls age 16 and over is not classified as child marriage-and thus is still legal-in Pakistan. Marriage Laws Enacted before and sincePakistan's Independence 2. The government o f Pakistan has taken pains to incorporate protections for women and girls into marriage law by emphasizing the importance o f the marriage contract (nikahnama) and abiding by its requirements. In addition to The Muslim Family Law Ordinance (MFLO) o f 1961, there has been The Dissolution of Muslim Marriages Act (DMMA) o f 1939; the Government Servants (Marriages with Foreign Nationals) Rules (1962); the WestPakistan Family CourtsAct (1964) and West Pakistan Family Courts Rules (1965); the Dowry and Bridal Gifts (Restriction)Act (1976) and the Dowry and Bridal Gifts (Restriction) Rules (1976). Details o f some o fthese laws are discussedbelow. 3. The Child Marriages Restraints Act (CMRA) o f 1929 aimed to ban the practice o f early-age marriage by changing the definition o f "child" to any female under 16 years o f age and any male 18 years o f age, and then outlawing the marriage o f children and prescribing punishments for any male above age 18 who contracts a marriage to a child. 4. Although it predates the creation o f Pakistan by eight years, The Dissolution of Muslim Marriages Act (DMMA) o f 1939 is fundamental to Pakistan's marriage law. The DMMA has been called "one o f the most important pieces o f legislation promulgated in the area o f Islamic family law in the subcontinent" (Ali 2000: 147). When first established, its dual purpose was to clearly lay out the provisions o f Muslim law pertaining to the dissolution o f marriages by women who were married under Muslim law, and to specify the effect that a married Muslim woman's renunciation of Islam has on her marriage tie. A particularly important provision o f the DMh4A (in section 5) was the protection o f the wife's rightto dower in spite o f dissolution o f the marriage. Section 2 (vii) o f the DMMA also granted to a Muslim girl the option o f puberty available to repudiate her marriage-if it occurred while she was a minor-to include a marriage contracted for her by her father or grandfather 216http://www.unfpa.org/swp/2003/englishlch2/index.htm *I7http://www,unicef.orglinfobycountry/paistan_statistics.html 114 5. The Muslim Family Law Ordinance (MFLO) o f 1961 was a response to recommendations by a Commission set up in 1955 after a large portion o f Pakistani society (mostly in the female sector) began agitating for improvements in the status o f women. The express charter o f the Commission was findcreate legal means o f restricting polygamy and granting women more rights o f divorce than they'd had under DMMA. Not until 1961 did the Commission's recommendations take the form o f the MFLO, which contained important provisions to increase women's advantage. For the first time, regulation and formalization o f the process o f divorce was incorporated into nationwide law. The MFLO also restricted polygamy by requiringa husband desirous o f a subsequent marriage to either obtain permission from the existing wife/wives or submit an application to the Arbitration Council. In the event o f the husband contracting such a marriage, the MFLO made him immediately liable to payment o f the dower o f the existing wife/wives. The MFLO also amended the CMRA by increasing the legal age o f marriage from 14 to 16 years for females, and from 18 to 21 years for males. 6. The Dowry and Bridal Gijts (Restriction) Act, 1976, requires that the value o f the bridal dowry and presents given to her by her natal family not exceed 5,000 rupees, though this value excludes money given to the bride at the wedding. The law also requires that all property the bride receives as dowry or bridal gifts is hers without restriction, limits or conditions; the groom also i s denied rights over this property. Furthermore, if the woman dissolves the marriage to which she brought dowry, she has the rightto ask for her dowry to be returned up to three years subsequent to the divorce. The Report of the commission o f Inquiry for Women (1997) has issued recommended modifications to the law, with the intent to more effectively decrease the frequency and size o f dowry. These modifications include expanding the definition o f practicing dowry so that there are penalties for displaying dowry or ostentatiously displaying bridal family wealth in any manner, as well as including a punishment o f three years' imprisonment for mental or physical cruelty inflicted on a wife (plus liability for a fine); and ten years' imprisonment in cases where such cruelty causes a woman to commit suicide. The recommendations also include amendments that empower police to report and investigate cases o f suspicious or unnatural deaths o f women. Customary Practice and Civil Society Responses Related to Dowry, Bride Selling, and Marital ConsentinPakistan 7. A practice relatedto that o fdowry concerns the `selling' or `trading' o fbrides: The ability o f individuals to bypass the law without any fear o f repercussions has also perpetuated customary practices of selling girls into `marriage' in exchange for money, settling disputes with the exchange o f girls known as vani or swara and the use o f girl as compensation for crimes. While the formal laws in Pakistan do not condone these practises, the courts do little to address them, allowing informal justice systems to implement a law o f their own. High levels o f economic hardship and social inequality often lead families to sell their young daughters into marriage as a means o f earning money. Bride prices range from Rs. 80,000 to Rs. 200,000 (1,400-3,500 USD) and younger girls receive higher prices. These sales are not legal and are not done with the consent o f the girl. In some cases the decision is made by one member o fthe family without consulting any other members.218 8. It is not clear how common such phenomena are, since representative data are not available. Nor i s it clear how to draw the line between bride `selling' and the more common practices o f making financial transfers between families at the time o f marriage. At any rate, regardless o f whether a bride is sold, traded, or given freely, the most salient fact from the women's point o f view i s the extent to which her marriage is voluntary. 'I8Report o f the Commissionof Inquiryfor Women, Pakistan, 1997. 115 9. One civil society organization that has devoted considerable resources to education about dowry i s Pakistan's Society for the Advancement of Community, Health, Education and Training (SACHET). InNovember 2001, SACHET launched the Fight against Dowry (FAD), a five-year initiative to educate the public-and Pakistan's youth in particular-about the harm often inflicted on brides due to the practice of dowry. For more information, please see http://www.sachet.org.pk/home/agehi_resource_center/fa~profile-of~roject.asp. 116 ANNEX 2.3: RECOMMENDED CHANGESTO MARRIAGE LAWAND POLICY RELATED TO WOMEN'S RIGHTS INPAKISTAN 1. The specific protections o f women's rightsto be incorporated into the nikahnama are detailed in the Legislative Watch Programme's Aurat Publication that recommends changes in the nikahnama form. These include the following: 0 requirement o f registration within 30 days o fthe solemnization (or performance) o f the Nikah (marriage ceremony) 0 requirement o f the wife's permission ifthe husband i s to take an additional wife/wives 0 requirementthat the bridegroom disclose his marital status, since the bride is required to disclose hers 0 requirementthat the husband disclose any conditions on delegating to his wife, married under Muslimlaw, the righto fdivorce based on grounds provided inthe DissolutionofMuslim Marriages Act (DMMA) o f 1939219 0 Protection o f the marriage tie inthe event that the married Muslimwoman renounces Islam 0 The right to protection o f a woman's dower (the part or interest o f a deceased man's real estate allotted by law to his widow for her lifetime) 0 The requirement that the nikahnama specify the amount o f dower or, ifthe amount is not specified, it i s presumed payable on demand 2. Recommendations also emphasize the importance o f recording the following nikah details: the date on which the marriagewas contracted; the date on which the contract was registered and the amount of registration fee paid; the amount o f the dower; the portion o f the dower to be paid promptly and the portion to be deferred; whether property has been given in lieu o f any part or whole o f the dower (along with its specifications and a value agreed upon by both parties); whether the husband has delegated the right of divorce to his wife and, if so, under what conditions; any restrictions on husband's right of divorce; the husband's marital status; name and address o f the person solemnizing the marriage, along with the name o fthat person's father; signature o fthe bridegroom or his vakil (advocate); signature o fthe bride and her vakil; signatures o f witnesses to the marriage; signatures o f witnesses to the appointment o f the bride and bridegroom's vakil. These grounds include the following: the husband's whereabouts have been unknown for four or more years; the husband has been imprisoned for at least three years; the husbandhas failed to provide for her maintenancefor at least two years; the husband has contravened MFLO provisions intaking a new wife. the husband has neglected to perform his marital obligations for at least three years; the husband's impotency-present at the time of marriage-persists; the husband has been suffering from insanity, leprosy, or a dangerous venereal disease for at least two years; the husbandwas impotent at the time of marriage; the bride was given in marriage before age 15 and repudiated the marriage prior to reaching age 18-provided lack of consummation of the marriage; the husbandtreats her cruelly by conduct that is not limited to physical ill-treatment; the husband associates with women of evil repute or leads an infamous life and/or tries to force his wife to lead an immoral life the husband disposes of or prevents his wife from exercising legal rights over her property; the husbandobstructshis wife inobserving her religious practice or profession; the husband does not treat her equitably to his other wives, ifhe has more than one, in accordancewith the Qur'an's injunctions. 117 ANNEX 3.1: TABLES WITH FULLRESULTS (A3.1-A3.7) Table A3.1: Determinantsof CurrentEnrollment,ChildrenAged 5-19 Rural (N=19910) (1) (2) Child's Characteristics: Girl -0.327** -0.405** (1 if Yes, 0 ifBoy) (3 8.30) (16.90) Age (Years) 0.239** 0.242** (28.80) (29.24) -0.011** -0.011** (29.66) (30.02) Age>= 13 -0.084** -0.008 (1 if Yes, 0 ifNo) (4.96) (0.44) (Age>=13)xGirl -0.174** (9.43) Parents' Characteristics: Mother Ever Attended School 0.182** 0.087** (1 if Yes, 0 ifNo) (12.26) (4.26) Mother Ever Attended xGirl 0.175** (6.01) Father Ever Attended School 0.251** 0.242** (1 if Yes, 0 ifNo) (29.26) (21.41) FatherEver Attended xGirl 0.019 (1.12) Household Characteristics: Share of children aged 0-4 0.001 0.013 (0.06) (0.64) Share of 0-4 childrenxGirl -0.039 (1.25) Main Source of Drinking Water -0.025* * -0.026* * (1 ifLocatedOutside House, 0 if LocatedInside House) (2.62) (2.70) Household Expenditure Quintile (First (Poorest) Quintile is reference category) Second 0.086** 0.088** (7.96) (6.38) Second xGirl -0.002 (0.07) Third 0.136** 0.126** (11.19) (8.04) Third xGirl 0.024 (1.02) Fourth 0.187** 0.178** (13.07) (9.66) Fourth xGirl 0.021 (0.75) 118 Table A3.1: Determinantsof Current Enrollment,ChildrenAged 5-19 (Continued) Fifth 0.289** 0.265** (14.21) (9.80) Fifth xGirl 0.054 (1.38) School Proximity: Primary School Within Village 0.151** 0.101** (1 ifYes, 0 ifNo) (15.15) (8.17) Primary School xGirl 0.142** (7.14) Post Primary School for Girls Within Village 0.061** 0.014 (1 ifYes, 0 ifNo) (5.13) (0.86) Girls' Post Primary School xGirl 0.098** (4.35) Post Primary School for Boys Within Village -0.024* -0.01 1 (1 ifYes, 0 ifNo) (2.34) (0.87) Boys' Post Primary School xGirl -0.034+ (1.69) Notes: Logit Model MarginalEffects. z statistics inparentheses.+ significant at 10%; significant at 5%; * ** significant at 1%. Additional variables not shown in table include community level variables: Total population of community; average community per capita expenditure, distance to: daily market, postoffice, bank, union council; Whether 50 percent of more of the households in village have electricity. Data Source: Pakistan Integrated HouseholdSurvey, 2001-02,Rural Children. The primary school dummy includes public primary school for girls, public primary coeducation school, private school for girls and private coeducation school. A separate dummy for primary school for boys is not included in regression since almost 90 percent of villages that have a primary school for girls or a coeducation private primary school also have a primary school for boys. Thus including a dummy for availability of primary school for girls also captures the availability of primary school for boys. Post-primary school includes middle andhigh school. 119 Table A 3.2: Determinantsof Current Enrollment,ChildrenAged 6-18, Controllingfor Household - FixedEffects (N=ll569) Child's Characteristics: Girl -2.497** (1 ifYes, 0 if Boy) (15.61) Age 1.356** (25.64) -0.064* * (26.33) Age>=l3 0.019 (0.16) (Age>= 13)xGirl -1.091** (9.17) Parents' Characteristics: Mother Ever Attended School 0.124 (1 ifYes, 0 ifNo) (0.43) MotherEver Attended xGirl 0.582** (2.90) FatherEver Attended School 0.112 (1 ifYes, 0 ifNo) (0.52) FatherEver Attended xGirl 0.187 (1.63) Household Characteristics: Share of 0-4 childrenxGir1 -0.642** (2.95) HouseholdExpenditure Quintile (First (Poorest) Quintile is reference category) SecondxGirl 0.134 (0.94) Third xGirl 0.282+ (1.72) Fourth xGirl 0.213 (1.12) Fifth xGirl 0.021 (0.07) School Proximity: Primary School xGirl 0.853** (6.3 1) Girls' Post Primary School xGirl 0.596** (4.07) Boys' PostPrimary School (Girl -0.108 (0.82) Notes: Conditional (Household Fixed Effects) Logit Model Coefficients. z statistics in parentheses. + significant at 10%; significant at 5%; * **significant at 1%. Data Source: Pakistan Integrated Household Survey, 2001-02, Rural Children 120 Table A 3.3: Determinantsof Current Enrollment,Rural Boys Aged 6-18 All 6-10 10-18 10490 4744 5746 Child's Characteristics: Age (Years) 0.288 ** (26.00) -0.013** (26.14) Age>= 13 -0.052+ -0.285** (1 ifYes, 0 ifNo) (1.94) (16.99) Parents' Characteristics: Mother Ever Attended School 0.095** 0.128** 0.064* (1 ifYes, 0 ifNo) (4.49) (3.77) (2.46) Father Ever Attended School 0.255** 0.245** 0.239** (1 ifYes, 0 ifNo) (21.68) (15.01) (15.53) HouseholdCharacteristics: Share of children aged 0-4 0.041 0.009 0.054+ (1.61) (0.30) (1.86) Share of 0-4 children((Age>=13 ) -0.054 (1.25) Main Source of Drinking Water -0.041** -0.025 -0.038* (1 ifLocated OutsideHouse, 0 ifLocated InsideHouse) (3.19) (1.40) (2.25) Household Expenditure Quintile (First (Poorest) Quintile is reference category) Second 0.103** 0.083** 0.105** (7.14) (4.24) (5.27) Third 0.147** 0.136** 0.137** (8.89) (5.82) (6.27) Fourth 0.210** 0.220** 0.180** (10.68) (7.64) (7.11) Fifth 0.307** 0.262** 0.293** (10.74) (5.52) (8.56) Community Characteristics: Primary School Within Village 0.109** 0.099** 0.109** (1 if Yes, 0 ifNo) (8.39) (5.52) (6.26) Post Primary School for Boys Within Village 0.003 -0.008 -0.003 (1 ifYes, 0 ifNo) (0.17) (0.44) (0.16) Boys' PostPrimary School ((Age>=13) -0.025 (1.06) Notes: Logit Model Marginal Effects. z statistics in parentheses.+ significant at 10%;*significant at 5%;**significant at 1%. Additional variables not shown in table include community level variables: Total population of community; average communityper capitaexpenditure, distanceto: daily market, postoffice, bank, union council; Whether 50 percentof more of the householdsin village have electricity. Data Source: PakistanIntegratedHouseholdSurvey, 2001-02, Rural children. 121 Table A3.4: Determinantsof Current Enrollment,RuralGirls Aged 6-18 All 6-10 10-18 9420 4504 4916 Child's Characteristics: Age (Years) 0.155** (14.70) -0.008** (15.67) Age>= 13 -0.105** -0.192** (1 ifYes, 0 ifNo) (4.28) (15.32) Parents' Characteristics: Mother Ever Attended School 0.200** 0.244** 0.153** (1 ifYes, 0 ifNo) (12.10) (8.29) (8.74) Father Ever Attended School 0.201** 0.233** 0.167** (1 ifYes, 0 ifNo) (19.93) (14.13) (13.99) Household Characteristics: Share of children aged 0-4 -0.038+ -0.066* 0.010 (1.81) (2.27) (0.43) Share of 0-4 childrenx(Age>=13 ) 0.038 (0.90) Main Source of Drinking Water -0.008 -0.031+ 0.016 (1 if Located Outside House, 0 if Located Inside House) (0.65) (1.66) (1.13) Household Expenditure Quintile (First (Poorest) Quintile is reference category) Second 0.059** 0.059** 0.067** (4.45) (2.83) (3.97) Third 0.105** 0.116** 0.094** (7.21) (4.9 1) (5.34) Fourth 0.134** 0.113** 0.133** (7.90) (3.93) (6.88) Fifth 0.222** 0.284** 0.170** (9.43) (6.25) (6.91) Community Characteristics: Primary School Within Village 0.184** 0.222** 0.142** (1 ifYes, 0 ifNo) (14.57) (11.16) (8.97) Post Primary School for Girls Within Village 0.042** 0.069** 0.054** (1 ifYes, 0 ifNo) (2.88) (3.23) (3.97) Girls' PostPrimary School x(Age>=13) 0.053* (2.29) Notes: Logit Model Marginal Effects. z statistics in parentheses. + significant at 10%;*significant at 5%;**significant at 1%. Additional variables not shown in table include community level variables: Total population of community; average community per capitaexpenditure, distanceto: daily market, postoffice, bank, union council; Whether 50 percent of more of the households in village have electricity. Data Source: PakistanIntegrated HouseholdSurvey, 2001-02, Rural children. 122 Table A3.5: Choice between public and private primary schools: Nested Multinomial Logit estimates Coefficient Std. Err. Z statistic SchoolChoice: Public or Private: Girlx Private School -0.58 0.091 -6.32 Household's Consumption (excluding expenditureon education) 0.47 0.022 20.85 School in community (1 ifYes, 0 ifNo) 0.85 0.077 11.04 School in community x Girl 0.63 0.112 5.61 Enrollment Decision: Enroll or Not Enroll: Girl -1.04 0.079 -13.08 LR test of homoskedasticity (iv = 1): chi2(2)= 6.85 Prob > chi2 = 0.0326 LR chi2(7) = 4477.247 Log likelihood =-7179.7798 (Prob > chi2 = 0,0000) Notes: Coefficients from Nested MultinomialLogit model. Modelestimatedfor rural children aged 6-10. 123 Table A3.6: Primary schoollocationdecision,RuralPakistan,PIHS 2001-02 Public School Private School Boys' Boys' Girls' Girls' Coeducation Coeducation School School School School School School Logit Fixed Logit Fixed Logit Fixed Effects Effects Effects Logit Logit Logit (1) (2) (3) (4) (5) (6) Bus stop within 1Km of 0.832 0.693 0.590 0.832 0.922 1.160 community (2.84)** (1.89) (2.47) * (2.62)** (2.56) * (2.85) ** Daily Market within 1 1.720 1.733 0.007 0.127 0.738 0.842 KmofCommunity (1.62) (1.57) (0.01) (0.21) (1.74) (1.68) Bank within 1Km of -0.388 -0.199 0.104 -0.047 0.766 0.561 Community (0.68) (0.3 1) (0.22) (0.09) (2.00) * (1.27) Postoffice within 1Km 0.268 0.076 0.855 0.424 1.411 0.738 of Community (0.75) (0.17) (3.29)** (1.28) (5.01)** (2.15)* Union Council within 1 0.061 -0.330 0.602 0.790 0.177 0.301 Km of Community (0.16) (0.74) (2.24)* (2.46)* (0.60) (0.84) Village average per 1.203 6.825 12.793 8.424 14.276 9.389 capita expenditure (0.18) (0.80) (2.40)* (1.14) (2.25)* (1.00) High schools for girls 0.477 0.524 0.618 0.597 1.272 0.973 availablewithin 5km (1.45) (1.28) (2 62) ** * I (1.93) (4.78)* (2.97)** Proportion ofworkers 0.004 0.396 1.070 1.075 1.654 1.441 working inthe non- (0.01) (0.53) (2.27)* (1.79) (2.8 1)** (1.87) agricultural sector in community Proportion of households 1.536 2.639 2.116 2.463 -0.208 0.182 incommunity with at (2.00)* (2.SO) ** (3.72)** (3.23)** (0.29) (0.19) least one adult (age>= 15) with secondary education Community population 0.000 -0.000 0.000 -0.000 0.000 0.000 (0.3 1) (1.65) (3.99)** (0.29) (4.33)* * (0.15) Constant 0.420 -3.672 -6.475 (0.64) (6.28) ** (7.92)* * Number of Communities 575 337 575 465 542 386 Notes: Logit and Conditional Logit Model Coefficients z statistics inparentheses. + significant at 10%; ** . * ~ ~~ significant at 5%; significant at 1%. Village average per capita expenditure divided by 10000. Regression estimated at the level of communities. Fixed effects estimates pertainto district level fixed effects. 124 Table A3.7 Private primary school locationdecision including availability of highschool close to community, Rural Pakistan, PIHS 2001-02 (1) (2) Logit Logit with Fixed (N=532) Effects (N=3 86) Bus stop within 1Kmof community 1.037 1.232 (2.83)** (2.97)** Daily Market within 1 Kmof Community 0.855 0.861 (1.99)* (1.72) Bank within 1Km of Community 0.717 0.558 (1.84) (1.25) Postoffice within 1Km of Community 1.299 0.713 (4.54) ** (2.06)* Union Council within 1Km of Community 0.078 0.202 (0.26) (0.55) Village average per capita expenditure 15.532 9.587 (2.42)* (1.02) High schools for girls available within 5km 1.256 0.922 (4.67)* * (2.79)* * Number of public girls' school available incommunity 0.252 0.226 (2.87) ** (1.99)* Proportion of workers working inthe non-agricultural sector in 1.696 1.442 community (2.82) ** (1.83) Proportion of households in community with at least one adult -0.475 0.134 (age>=l5) with secondaryeducation (0.65) (0.14) Community population 0.000 0.000 (4.43)* * (0.11) Constant -6.896 (8.05)* * Notes: Logit and Conditional Logit Model Coefficients . z statistics in parentheses. + significant at 10%; ** *significant at 5%; significant at 1%. Village average per capita expenditure divided by 10000. Regression estimated at the level of communities. Fixed effects estimates pertainto district level fixed effects. 125 ANNEX 4.1: TABLES WITH FULLRESULTS (A4.1-A4.7) One o f the key factors affecting health examined inthis Chapter is access to health facilities and outreach services. In the regression analysis in this Chapter access to these services i s measured at the community/village level. It i s entirely possible that over time expansion o f health facilities and outreach workers has occurred in some communities and not in others in a way that is closely associated with the level o f development o f that community or village for one. For instance, the government may have made special efforts to recruit outreach workers in communities that are not well connected to main town centers or where health outcomes are poor.220At first glance then it would appear that health outcomes are worse in areas served by LHWs or where facilities are close to the village. To ensure that to the largest extent possible we are indeed estimating the causal impact o f the availability o f health facilities and services we include village characteristics such distance to nearest market and distance o f village from the tehsil headquarters. This issue is common to assessing the impact of availability of facilities at the village level referred to as non- random program placement. A large literature has analyzed the potential solutions to taking into account non- randomexpansionor placementof programs. An overview i s available inThomas and Strauss (1995). 126 Table A 4.1: Determinantsof ProbabilityofFalling111,Consultinga medicalpractitionerand Medicalexpenditures,RuralChildrenaged0-17, PRHS 2001 (1) (2) (31 Probit Marginal MarginalEffects Ordinary Least Effects from Probit with Squares Selection Probability of Probability of Log(Medica1 FallingI11 Consulting Expenditure) if ill Medical practitioner ifI11 Girl -0.242 (4.78)"* -0.032 -0.311 (5.07)** (1.92)+ Child's parent is household 0.028 0.026 -0.070 head (3.45)"" (3.70)** (0.44) Age (years) -0.014 -0.013 0.034 (5.70)* * (5.59)** (0.76) (Age)2 0.001 0.001 -0.00 1 (4.58)* * (4.8 1)** (0.22) Mother Ever Attend School (1 0.012 0.0003 0.371 ifyes) (0.90) (0.03) (1.54) Father Ever Attend School (1 if -0.020 -0.010 0.171 Yes) (2.84)** (1.55) (1.30) Log Per Capita Expenditure of 0.026 0.0021 0.628 household (3.84)** (0.71) (4.88)"" Distance to Health Facility 0.007 -0.0016 -0.030 (4.41)** (2.56)"" (0.98) Distance to Pharmacy 0.007 -0.0008 -0.046 (3.46)** (1.04) (1.26) Distanceto Daily market -0.001 0.0049 0.011 (0.40) (3.68)** (0.36) Community Level: Improper -0.010 -0.0069 -0.081 disposal of waste water (Waste (1.19) (0.92) (0.54) Water thrown into ground) Community Level: Improper 0.024 0.022 -0.263 disposal of garbage (Garbage (3.20)** (3.16)* * (2.04)* thrown into river or ground) Constant 3.153 Observations 7482 7525 Absolute value of z statisiics inparentheses. significantat 10%; + *significant at 5%;**significant at 1%. Notes: Probit model marginal effects in Column 1. Marginal effects from Probit model with selection in Column 2. Regressioncoefficientsfrom Ordinary Least Squares regression.Absolutevalue of z statistics in parentheses. + significant at 10%;* significantat 5%; ** significant at 1%. 127 throughMedia) (1 ifYes) Log (Per Capita -- _- 0.089 0.125 HouseholdExpenditure) (1.80)+ (2.38) * Lady HealthWorker in 0.089 0.038 0.142 0.108 Community (1.82)+ (1.39) (2.86)** (2.21)* GovernmentPrimary 0.082 0.006 0.093 0.137 HealthFacility (BHU, (1.73)+ (0.23) (1.go)+ (2.73)* * MCHCenter, Family Welfare Center) Within 5 Kms Private HealthFacility -0.037 -0.025 -0.030 0.005 Within 5 Krns o f (0.77) (0.94) (0.59) (0.09) ImmunizationCamp held 0.068 0.004 0.0 15 -0.008 within 5 kms of 1I (1.57) I (0.16) I (0.34) I (0.20) Notes: Probit model marginal effects. Regression weighted using household weights. Robust z statistics in parentheses. 10%; * significant at 5%;** + significant at significant at 1%. Additional variables not shown in table include community level variables: Community level variables include dummies for electricity, drainage, distanceto: tehsil capital, nearest bus stop, market, nearest motorable approach road, public primary school for girls, midrlle school for girls. DPT 3 and Polio 3 measure whether final dose of each immunizationreceivedby child or not. Logper capita expenditure only includedfor BCG andMeasles.DPT and Polio are usually suppliedfreeof cost. 128 Woman EverAttended 0.084 0.224 0.207 0.058 0.101 -0.082 School (1 if Yes) (4.07)** (8.49)** (7.10)** (4.33)** (4.89)** (3.50)** HusbandEver Attended 0.033 0.031 0.052 0.009 0.035 -0.030 School (1 if Yes) (2.72)** (1.71)+ (2.71)** (0.89) (2.96)** (1.99)* RatioofNumber of Sons 0.019 alive to Number of Daughters (4.43)** Alive MediaExposure (Heard 0.052 0.060 0.103 0.005 0.05 1 -0.014 Hygienerelatedinformation (2.87)** (2.10)* (3.08)** (0.36) (2.36)* (0.45) through Media) (1 ifYes) Log (Per CapitaHousehold 0.022 0.122 0.117 0.041 0.113 0.030 Expenditure) (1.54) (5.23)** (4.02)** (3.91)** (6.37)** (1.31) Lady HealthWorker in 0.030 0.011 0.064 -0.003 0.009 0.021 Community (2.16)* (0.47) (2.41)* (0.28) (0.56) (0.64) Government PrimaryHealth 0.008 0.052 0.077 -0.004 -0.011 0.040 Facility (BHU, MCHCenter, (0.61) (2.15)* (3.04)** (0.37) (0.63) (1.11) Family WelfareCenter Within 5 Kms) Private HealthFacility 0.007 0.025 0.009 0.014 -0.002 -0.044 Within 5 K m s of Community (0.54) (0.98) (0.33) (1.18) (0.10) (1.23) 129 Table A4.4: Determinantsof use of maternalhealthservices,with interactionterms, Rural Women aged 15-49, PIHS2001-02 (2) (3) (4) (5) (6) Prenatal Tetanus Postnatal Birthina Birth ive use care Toxoid consultations medical assistedby Immunizat institution TBA ions Media Exposure* -0.053 -0.031 -0.083 0.007 0.067 0.012 Woman Attended (1.54) (0.59) (1.11) (0.24) (1.32) (0.18) -0.001 0.016 -0.033 0.045 -0.040 incommuni (0.01) (0.23) (1.66)+ (0.98) (0.69) -0.048 0.010 1I -0.016 II 0.001 II -0.075 School (0.03) (1.11) (0.20) (0.87) (0.03) (1.56) LHW*Log(Per Capita 1I 0.012 0.085 0.164 0.011 0.019 -0.036 (1.73)+ (2.51)* (0.52) (0.63) (0.82) 0.019 0.011 0.008 0.016 -0.058 within 5 kms *Woman (0.33). . (0.39) (0.20) (0.37) (0.38) (1.10) Attended School Government Facility * I -0.014 0.029 0.019 0.044 0.063 0.005 (0.58) (0.35) (1.78)+ (1.60) (0.07) Government -0.052 0.026 0.007 -0.005 -0.001 -0.067 Facility*Log(Per Capita (1.95)+ (0.55) (0.12) (0.24) (0.02) (1.35) Expenditure) Private Facility within 5 0.015 0.016 0.000 0.019 -0.042 -0.014 k m s * Woman Attended (0.45) (0.3 1) (0.00) (0.79) (1.22) (0.26) School Private Facility * LHW -0.046 -0.050 -0.020 -0.048 -0.039 -0.032 (1.92)+ (1.OS) Private Facility*Log(Per 0.006 0.030 Capita Expenditure) (0.21) (0.62) Notes: Probit model marginal effects. Regressionweighted using household weights. Robust z statistics in parentheses. 10%; * significant at 5%; ** +significant at significant at 1%. These interactionterms were estimated as part of a f i l l regression that includedadditional variables not shown in table include woman's age, square of woman's age, whether woman attended school, whether spouse attended school, source o f hygienehealth information (media or family), log of per capita householdexpenditure, dummy variables for: availability o f Lady Health Worker (LHW) within community, availability of nearest primary health care facility within 5 kilometers of community, availability of nearestprivatehealth facility within 5 kilometersof community.Also includedare the following community level variables: Community level variables includedummies for electricity, drainage, distanceto: tehsil capital, nearestbus stop, market, nearest motorable approach road, public primary school for girls, midle school for girls. Full results of the regressions can be obtained onrequest, 130 Table A4.5: Determinantsof VillageLevelPlacementof LHWs,PIHS2001 Basic Health Unit inCommunity 0.166* 0.177* (2.40) (2.44) Nearestbus stop 0-3 Kms from Community 0.085 0.085 (1.OS) (1.01) Nearesttehsil capital 0-3 K m s from Community 0.113 0.106 (1.28) (1.22) Nearestrail station 0-3 Kms from Community -0.014 -0.008 (0.13) (0.08) Community has motorable approachroad 0.101 0.072 (1.20) (0.94) 50 % ofhouseholds incommunity have electricity 0.124+ 0.128+ (1.73) (1.77) Notes: PIHS 2001-02 Data, Community Level Data. Weighted Probit Marginal Effects.Absolute value o f t statistics in parentheses. +significantat10%;* significant at 5%;**significantat 1%. 131 Table A4.6: Percentage of women receiving maternal health services, 2001-02 Source: PIHS 2001-02 householdsurvey data. These percentages are basedon cross-tabulationsfrom the PIHS data. Data refer to use of maternal health services by pregnant women in the 3 years preceding the PIHS survey. Note that the categories"births at home" and "births assisted by TBAs" are not mutually exclusive. The percentagechanges in use of maternalhealth services are calculated by comparing 2001-02 percentageswith percentagesfor 1990-91from the Pakistan Demographic and Health Survey Report (NIPS and Macro International, 1992). Percentageo f women getting postnatal care notavailable for 1990-91. 132 Table A4.7: Percentage of births assisted by type of attendant, 1998-2001 S Note: Dais are traditional birth attendants. 133 ANNEX 4.2: NOTMUCHEVIDENCE OFPRENATAL SEX SELECTIONINPAKISTAN 1. The sex ratio at birth is one statistic that can reveal prenatal sex selection: it i s the ratio o f the number of male births to female births. Biologically, more males are born than females, and normal sex ratios at birth range between 105 and 107 boys per girls. This high male-to-female ratio at birth could be considered an evolutionary adaptation to the fact that females have higher survival probabilities than males. A sex ratio at birth that i s much higher than this biologically expected ratio suggests that female fetuses are being aborted. Societies with a strong preference for sons and with access to prenatal sex detection technology show an alarming rise in sex ratios at birth. Unusually high sex ratios at birth have been reported inparts o f India, China and South Korea. 2. The problem with the sex ratio at birth statistic is that it cannot be computed from a Census. It i s ideally computed from birth registration data, yet reliable and complete data o f this type is difficult to obtain in most developing countries. The data from Pakistan IntegratedHousehold Survey (PIHS) can be used to estimate the sex ratio at birth. Calculated using the 1998 PIHS, the sex ratio at birth shows a national average of 105 (Table A4.8). This is within the normal range and indicates the absence o f pre- natal sex selection in Pakistan. The province-wide sex ratios at birth are well within the normal range, except in Balochistan, where the ratio i s higher than that expected. Since there is little evidence o f sex- selective abortions in Balochistan, this ratio may reflect the underreporting o f female births. Ratios by mother's literacy status show marked variation in Balochistan and NWFP. While literate mothers report ratios inthe expected range or lower, illiterate mothers report fewer female births. Table A4.8: Sex Ratios at Birth (Estimated UsingPIHS 1998Birth HistoryData) Overall Mother Mother (Literate) (Illiterate) Pakistan 105 103 106 Punjab 104 103 104 Sindh 107 106 108 NWFP 105 99 105 Balochistan 109 100 110 3. A telling comparison can be drawn between Punjab and Indian Punjab. Punjab displays a sex ratio at birth o f 104; the considerable ratio o f 124 in Indian Punjab (NFHS-11) results from widespread use of sex-selective abortions. Declining fertility together with strong son preference and rising incomes are largely responsible for the rise in prenatal sex selection in India, China and elsewhere. This combination of factors could well prevail in Pakistan inthe future, as the present trend o f fertility decline continues. 134 (1) (2) (3) (4) (5) (6) (7) All Urban Urban Urban Rural Rural Rural Paid Unpaid Paid Unpaid Work Work Work Work Aee I 0.004 I 0.016 II 0.013 II 0.000 II -0.003 II 0.000 II -0.004 v (1.07) (2.93)** (3.10)** (0.28) (0.58) (0.12) (0.89) Age2 -0.000 -0.000 -0.000 -0.000 0.000 -0.000 0.000 (0.71) (2.41)* (2.48)* (0.43) (0.65) (0.08) (0.79) Education(years) -0.031 -0.015 -0.011 -0.001 -0.053 -0.006 -0.039 (7.64)** (3.96)** (4.27)** (0.78) (6.25)** (1.96)+ (4.68)** Education2 0.003 0.002 0.002 -0.000 0.005 0.001 0.002 (10.96)** (7,94)** (8.45)** (0.11) (6.67)** (4.77)** (2.34)* Husband's education -0.015 -0.010 -0.005 -0.002 -0.022 -0.003 -0.013 (years) (6.11)** (3.43)** (2.22)* (1.75)+ (5.99)** (1.85)+ (3.96)** (Husband's education)2 0.000 -0.000 -0.000 0.000 0.001 0.000 0.000 (0.90) (0.23) (0.45) (0.37) (2.31)* (0.54) (1.15) Secondper capita -0.037 -0.05 1 -0.034 -0.003 -0.03 1 -0.007 -0.015 expenditurequintile (2.33)* (2.53)* (2.63)** (0.61) (1.42) (0.80) (0.81) Thirdper capita -0.059 -0.070 -0.047 -0.006 -0.053 -0.017 -0.023 expenditurequintile (3.56)** (3.54)** (3.63)** (1.12) (2.33)* (2.04)* (1.13) Fourthper capita -0.067 -0.055 -0.035 -0.008 -0.089 -0.026 -0.040 expenditurequintile (3.89)** (2.82)** (2.62)** (1.54) (3.64)** (2.75)** (1.93)+ Fifthper capita -0.124 -0.120 -0.072 -0.012 -0.121 -0.032 -0.069 expenditurequintile (6.65)** (5.58)** (4.96)** (2.11)* (4.47)** (3.56)** (3.12)** Has child aged 3 or -0.032 -0.047 -0.027 -0.007 -0.022 -0.011 -0.005 younger (3.37)** (3.73)** (3.13)** (1.98)* (1.68)+ (2.03)* (0.50) Rural(1 ifyes) 0.153 (10.77)** Punjab 0.003 0.008 -0.015 0.001 0.003 -0.002 -0.057 (0.21) (0.50) (1.49) (0.25) (0.13) (0.21) (2.82)** NWFP -0.110 0.005 -0.037 -0.010 -0.153 -0.065 -0.099 (4.82)** (0.20) (2.49)* (1.92)+ (4.33)** (7.44)** (3.29)** Balochistan -0.134 -0.020 -0.008 -0.008 -0.180 -0.034 -0.155 (4.90)** (0.75) (0.46) (0.95) (4.32)** (2.26)* (4.65)** Baraniagriculture(1 if -0.068 -0.037 -0.010 Yes) (2.08)* (2.44)* (0.34) BaraniPunjab BaraniNWFP BaraniBalochistan , I I I I I I 1 I Observations I 12372 1, 4690 I 4690 4690 7409 7409 7409 stes: PIHS2001Robust z statistics in parentheses. t significant at 10%;*significant at 5%;-** significant at 1% 135 Urban and rural Urban Rural (1) (2) (3) (4) (5) (6) Men Women Men Women Men Women Age (years) 0.078 0.102 0.087 0.126 0.066 0.061 (19.47)** (5.50)** (15.22)** (4.49)** (12.45)** (2.47)* Age2 1 -0.001 -0.001 -0,001 -0.001 -0.001 -0.001 f18.06'1** (18.06)** I (4.68)** f4.68'1** (13.35)** (3.57)** (12.06)** (2.23)* Education Ed 0.058 0.144 0.055 0.139 0.044 0.136 (Years) (55.32)** (33.41)** (39.89)** (25.51)** (27.36)** (13.96)** Constant 6.041 3.886 5.963 3.360 6.294 4.784 (73.50)** (10.40)** (51.49)** (5.97)** (57.47)** (9,57)** Observations 13748 1732 5490 851 8258 881 R-squared 0.20 0.40 0.26 0.44 0.10 0.18 significant at 10%;* significant at 5%; ** significant at 1% 136 TableA5.3: Regressionoffemale work on purdahand safety concerns,PRHS2004 Probit Regression: Any paid work in last year Variable (1) (2) (3) Observespurdah Cyesho) 0.248 _ _ _ 0.143 Purdah index (0-5) -0.036 --- -0.021 Unsafe within settlement --- -0.400* * -0.378** Unsafe outside of settlement --- 0.250** 0.241** Age 0.054 0.053 0.05 Age squared -0.001 -0.00 I -0.00 1 Ever attendedschool 0.009 0.038 0.034 hhlandownership(acres) -0.033 *** -0.032*** -0.032*** S. Punjab 1,157*** 1.191*** 1.157*** Sindh 0.733*** 0.713*** 0.673*** Note:* piO.05;** p