WOMEN IN THE WORKFORCE IN QUETTA: RESULTS FROM THE QUETTA URBAN HOUSEHOLD SURVEY Pakistan Gender and Social Inclusion Platform & Pakistan Poverty and Equity Program August 2022 Abstract. Pakistan’s female labor force participation smaller and not always significant. When production (FLFP) remains low by regional and global standards. of goods for family use is included in the accounting of Furthermore, data show major disparities between female employment, FLFP increases by a small margin rural and urban FLFP, with the latter being signifi- (17.6 percent). The employment profile of women in cantly lower. This note contains analysis of women’s urban Quetta shows that they are mostly employed in labor market outcomes as reported by women in the low-value-added activities—mainly manufacturing, city of Quetta, Balochistan Province, using data from as garment and handicraft workers—and display the World Bank’s 2021 Quetta Urban Household Sur- a higher prevalence of own-account, informal, and vey (QUHS).1 The multipurpose QUHS (as well as a sim- home-based work. Men, on the contrary, are mostly ilar survey conducted in urban Peshawar in 2020) at- paid employees. Moreover, women’s jobs are in line tempts to improve measurement of FLFP by collecting with socially accepted occupations, likely a function information on labor market outcomes directly from of how easily these jobs can be done from home. Over all working-age household members. Hence, in terms 78 percent of employed women in urban Quetta are of implementation, it differs from standard labor force home-based workers who work on their own account surveys (LFSs) in Pakistan that use one or two proxy and have low chances of upward mobility. These re- respondents to report for other household members. It sults are in line with figures for urban Peshawar and also increases the number of questions that directly list urban Pakistan. all possible forms of female employment. Finally, it al- lows for a more comprehensive definition of employment The observed low level of FLFP and the nature of the by accounting for production of goods for family use. profile of female employment in urban Pakistan are explained by factors such as low human capital endow- Self-reporting of labor status in the QUHS allows for ment, lack of agency in various aspects of life, limited mo- more accurate estimates of FLFP in urban Pakistan bility, safety concerns, deep-rooted patriarchal norms, compared to what is reported in the most recent LFS. stereotypes about women’s role in the household, time According to benchmarks from different subsamples devoted to unpaid care and household work, and lack of in the 2020–21 LFS, the FLFP estimates from the QUHS information about labor market opportunities. always yield higher rates of FLFP (16.1 percent). The difference between the FLFP estimates in the two sur- JEL Codes: J13 J16 J18 J21 J22 J24 J46 O17 veys is significant at the 99 percent level. On the con- trary, in the case of men, the difference in estimates Keywords: gender, labor market participation, em- of labor force participation between the surveys is ployment, social norms, household surveys, Pakistan 1 This note was prepared in collaboration between the Social Sustainability and Inclusion (SSI) Global Practice, and the Poverty and Equity (POV) Global Practice (GP) in the World Bank. The leading author for the note was Paola Buitrago (consultant, POV GP), under the guidance of Moritz Meyer (senior economist, POV GP); Maria Beatriz Orlando (lead social development specialist, SSI GP); and Uzma Quresh (senior social development specialist, SSI GP). The effort was supported by Saleha Waqar and Noor Rahman (consultants, SSI GP). The team is grateful for comments received from Najy Benhassine (country director, Pakistan), Silvia Redaelli (senior economist, POV GP), Ana Maria Munoz Boudet (senior social scientist, POV GP), Aliya H. Khan (labor economist and former professor, Quaid-i-Azam University), and Mariam Mohsin (lecturer, Pakistan Institute of Development Economics). The note also benefited from comments by Maria Qazi and Ahmad Durrani (consultants, SSI GP). 1 Introduction conditions, labor market participation before and after the COVID-19 pandemic, safety, sexual harassment, aspi- Pakistan’s female labor force participation, particu- rations and values, and many others. The QUHS follows a larly in urban areas, remains one of the lowest in the similar methodology as (and was informed by) previous world, not just in South Asia. According to the ILOSTAT work conducted in urban Peshawar in 2020 (see Mancini database, in 2019 only nine countries had lower female 2021). Both the Quetta and Peshawar surveys differ from labor force participation (FLFP) rates than Pakistan, standard LFSs in Pakistan in terms of implementation. where the rate was 22.6 percent. Official figures from the They collect information on labor directly from all work- labor force survey (LFS) indicate that FLFP fell about 2 ing-age household members, whereas the LFSs use one percentage points between 2014 and 2018. Moreover, or two proxy respondents to report for other household urban FLFP in Pakistan has remained consistently low members. since 2005, at around 10 percent (Amir et al. 2018; Cho and Majoka 2020). The aim of this report is to present the main findings from QUHS 2021 and address the following questions: Is FLFP The World Bank’s Women in the Workforce study in in urban Pakistan truly as low as it appears (section 3)? Pakistan (started in 2019 and ongoing) is a multimethod Why does FLFP remain low (section 4)? What are the study to investigate urban FLFP and gain a nuanced un- barriers to women’s work (section 5)? What are the char- derstanding of the patterns of and constraints on wom- acteristics and quality of the jobs held by women and to en’s work. The qualitative component of the study ana- what extent do these differ from men’s jobs (section 6)? lyzed the labor market experiences of women in Quetta, Finally, what can be done to promote greater FLFP (sec- Peshawar, Lahore, and Karachi. The findings were used tion 6)? The report also includes a brief description of the to design the 2021 Quetta Urban Household Survey methodological innovations in the QUHS (section 2) and (QUHS 2021), a multipurpose household survey to col- a special analysis on how the COVID-19 pandemic affects lect information on a range of themes, including living women’s work (box 4). BOX 1. QUETTA AT A GLANCE Quetta is the largest city and provincial capital of Balochistan Province. Balochistan is the poorest province in Pakistan, with a poverty rate of 42.7 percent, followed by Khyber Pakhtunkhwa (KP) Province with 29.5 percent. At the same time, Balochistan accounts for only 12 percent of Pakistan’s poor.2 Balochistan hosts the sec- ond-largest share of Afghan refugees living in Pakistan (22.8 percent), after KP (58.1 percent), making Quetta the second-most important urban center of settlement for these populations (UNHCR Data Portal). According to the 2017 Census, Quetta District has a population of 2.3 million and Quetta City has 1 million peo- ple. Located in the northern part of Balochistan, close to the Durand Line, the de facto Pakistan–Afghanistan border, Quetta has served as a trade center between the two countries. The population’s ethnicity is mainly Pashtun, followed by Brahui- and Balochi-speaking populations. There is also significant representation of the Hazara community, which settled in Balochistan after migrating from central Afghanistan, mainly in the nineteenth century. The city has been adversely affected by incidents of terrorism and conflict due to security concerns. Further, violent insurgent groups have disproportionately targeted the Hazara community, which is largely of Shiite creation of settlements faith. This threat has resulted in the segregation of living communities, leading to the ​ at the outskirts of Quetta dedicated to the Hazara population. The social fabric is largely dominated by the importance of family ties and tribal kinship. Family members, mainly men, dictate women’s choices. As per Pashtun customs, women are closely associated with family honor, so their actions and movements are systematically controlled (Paterson 2008; Sanauddin 2015). Be- cause of conservative social and family norms that limit women’s mobility outside the home or local commu- nity, a significant share of employed women in Balochistan are home-based workers (HBWs) in the informal economy (USAID 2012). 2 World Bank estimates based on the 2018–19 Household Integrated Economic Survey. 2 Control over women can translate into gender-based violence (GBV). Regional data on violence against women (Pakistan Demographic and Health Survey 2017–18) suggests that women in Balochistan experience the second-highest rate of GBV (48 percent) in Pakistan after women in former federally administered tribal areas (56 percent). Along with high incidence of GBV, there is wide acceptance of GBV, especially among women. As much as 52 percent of women and 31 percent men in Balochistan agree that wife beating is justified under specific circumstances. Further, 20 percent of women in Balochistan report that their husbands exert marital control over their actions, and 49 percent report having experienced incidences of spousal violence. In addi- tion, only 69 percent of women report having control over their earnings, the lowest rate reported across all of Pakistan. These patterns indicate limited involvement of women in matters both within and outside the home, which are governed by restrictive patriarchal gender norms. In Pakistan, many women manufacture embroidered products since it builds on traditional skills and gives them the opportunity to work from home without violating social norms. Balochistan has a rich tradition of embroidery, and many women, while homebound, work in the embellished garment sector (USAID 2012). Survey Methodology: What Is New in to the COVID-19 pandemic. The survey design is informed the Quetta Urban Household Survey? by and follows a similar methodology to previous research conducted by the World Bank in urban Peshawar. The QUHS is a multipurpose household survey de- Fieldwork took place between November 2020 and signed to reach a statistically representative sample to March 2021. Each respondent provided informed con- study the welfare of the city’s Pakistani host commu- sent. Data were collected on paper via separate ques- nity and Afghan refugee populations. The survey ques- tionnaires for men and women. Census blocks were used tionnaire includes a range of themes, such as water and as primary sampling units (PSUs) (see appendix A.1 sanitation, urban poverty, labor market participation and for details on sampling). The sample was drawn at the economic empowerment, women’s status and gender in- household level, with a final sample of 2,406 households equality (including, but not limited to, sexual harassment covering a total of 18,255 individuals. Data were collected and perceptions of safety), domestic and international mi- from working-age (15–64) men and women separately gration, and individual aspirations. Due to the timing of the (see descriptive statistics at the individual level in table 2 survey, QUHS 2021 includes a series of questions related and at the household level in table A1 in appendix A).  TABLE 1. QUHS 2021 TECHNICAL DETAILS Characteristic Description Data collection Fieldwork period Data collection (including pilot) from November 2020 to March 2021 Mode of data capture Paper-assisted personal interviewing (PAPI) with separate questionnaires for men and women Sampling Sampling frame 2017 Census Primary sampling units 220 Final sample size (households) 2,406 (including 671 Afghan refugee households) Sample composition Individuals 18,255 Males 9,414 Females 8,841 Working-age men (15–64) 5,227 Working-age women (15–64) 4,829 Afghan refugees 5,331 Working-age (15–64) Afghan refugees 2,638 3 With a focus on labor market outcomes, QUHS 2021 ence or, if they did not tell anyone, why they did not do so. attempts to improve measurement of FLFP by im- The module includes an additional layer of consent above plementing three design features. First, it asks each that required for participation in the survey overall. Enu- woman of working age directly about her labor market merators were trained to report whether female respon- engagement. Second, it increases the number of ques- dents had full privacy or seemed visibly uncomfortable tions that directly list all possible forms of employment when answering questions in this module. to account for whether women (a) engage in wage, sal- ary, or other paid work; (b) help with the paid job of a Unless otherwise specified, all figures in this report refer family member; (c) work in a nonfarm family business to the population in urban Quetta, including the Pakistani that they manage or another family member manages; or host community and the Afghan refugees. (See box 3 (d) work in family farming, livestock, or fishing. Finally, for relevant findings about the labor profile of Afghan for those working in family farming, the survey allows for refugee women.) accounting of goods produced for family consumption by differentiating between products intended for sale and Measuring Women’s Work those for family use. Among the many questions raised by the available The module on sexual harassment is structured as estimates of FLFP in urban Pakistan is if the mea- a series of questions that capture information on surement is gender neutral and should be taken at different kinds of harassment. The first questions ask face value. The literature warns of potential downward whether respondents have ever experienced various biases affecting the measurement of FLFP, especially in forms of sexual harassment. Follow-up questions ask low-income countries. One factor that may contribute about whom (if anyone) they spoke to about their experi- to underestimation of FLFP is widespread use of proxy QUHS 2021 is designed to obtain a representative sample of Quetta City’s population.  TABLE 2. DESCRIPTIVE STATISTICS FOR WORKING-AGE WOMEN AND MEN PARTICIPANTS IN QUHS 2021 Variable Women (%) Men (%) Age 15–18 14.0 15.9 19–24 21.1 20.8 25–44 45.8 43.3 45–64 19.1 20.1 Afghan refugees 12.1 11.5 Marital status Married 62.4 54.2 Single 30.3 41.7 Divorced/separated 0.4 0.2 Widow/widower 3.5 0.5 Engaged/promised 3.4 3.5 Age at marriage (years) 19.7 24.3 Literate 50.1 81.5 Education No schooling 47.0 17.6 Incomplete primary 5.9 5.2 Completed primary 19.1 27.4 Completed secondary (grade 10 or vocational) 11.8 19.7 Completed upper secondary (grade 12) 7.6 12.7 Completed tertiary 8.5 17.3 Completed upper tertiary 0.1 0.2 4 TABLE 2. DESCRIPTIVE STATISTICS FOR WORKING-AGE WOMEN AND MEN PARTICIPANTS IN QUHS 2021 (CONTINUED) Variable Women (%) Men (%) Employment status Employed 15.6 69.8 Unemployed 0.5 2.4 Out of labor force 83.9 27.8 Number of cohabiting children 0–5 1.2 1.2 Number of cohabiting children 0–14 3.3 3.3 Adequate food consumption (reported by head of household) a 73.1 72.9 Access to cell phone 46.6 89.4 Access to internet 32.4 65.8 Language (spoken by head of household) Pashto 42.6 44.6 Brahvi 17.3 17.5 Urdu 10.8 10.9 Balochi 6.8 6.2 Hazargi 6.5 5.0 Punjabi 5.8 5.5 Other 10.2 10.3 a. Food adequacy takes a value of 1 if the male primary respondent considers the household’s food consumption adequate or better. b. Access to cell phone includes both owning a phone and access through a spouse, brother, or friend. c. Access to internet includes both at home and through other means. respondents in household survey labor modules.3 The of goods and services primarily for the market (Interna- male household head, who usually reports for other tional Conference of Labour Statisticians 2013)—leaves household members, may not be adequately informed out many productive activities typically carried out by about women’s economic activities or may fail to report women, such as activities related to production of goods them due to implicit bias regarding women’s work. A for family consumption or playing a supportive role (of- typical example is a woman who is unpaid but is sup- ten unpaid) in family businesses. porting a family business managed by a male household member. Empirical evidence on the use of proxies in Self-reported labor status in the QUHS allows for labor modules provides mixed results and is limited in more accurate estimates of FLFP in urban Pakistan covering different cultural contexts (Ambler et al. 2021; than data from the latest LFS. These results are con- Bardasi et al. 2011; Benes and Walsh 2018; Desiere and sistent with similar work conducted in Peshawar. Fig- Costa 2018; Dillon et al. 2012). In general, researchers ures 1 and 2 and table 3 present FLFP rates calculated have found that questionnaires that directly elicit all from the QUHS and from recent LFSs relevant for bench- possible forms of labor market engagement constituting marking purposes. In all cases, the FLFP measured using employment through separate questions achieve a more the QUHS is higher than the estimates from the LFSs. precise measurement of FLFP. But the discussion of mea- According to QUHS 2021, self-reported FLFP in urban surement issues within the literature on women’s work Quetta is 16.1 percent. In contrast, the FLFP estimate goes far beyond data collection into the definition of for Quetta District in the latest LFS (2020–21) is 6.5 work itself. Even when recorded without error, the stan- percent.4 The FLFP rate estimate under QUHS 2021 dard concept of work—which focuses on the production represents a 9.6 percentage-point increase relative to 3 Most household and labor force surveys do not expressly require each household member to answer directly for themselves. Given the time constraints and difficulties in having all members present during the interview, the questionnaire is administered to only one or two respondents, who in most cases are the household head and the spouse. 4 The universe for the Labour Force Survey consists of all urban and rural areas in the four provinces of Pakistan and Islamabad excluding military restricted areas. The population of excluded areas constitutes about 1 percent of the total population (Pakistan Bureau of Statistics 2022). 5 Self-reported working status allows for more accurate estimates of FLFP.  FIGURE 1. URBAN LABOR FORCE PARTICIPATION RATE FIGURE 2. URBAN LABOR FORCE PARTICIPATION RATE, OF WOMEN, 2020–21 MEN, 2020–21 Share of working-age women in the labor force (%) Share of working-age men in the labor force (%) Women Men 100 100 74.9 72.2 72.6 80 80 60 60 40 40 16.1 17.6 6.5 20 20 0 0 Labor Force Survey Quetta Survey Quetta Survey, Labor Force Survey Quetta Survey Quetta Survey, 2020/21, Quetta 2021 including work for own 2020/21, Quetta 2021 including work for own District consumption* District consumption* Note: All estimates refer to individuals ages 15–64. LFS estimates refer to Quetta District in 2020–21. Quetta Survey estimates refer to urban Quetta in 2020–21. Bars indicate a 95 percent confidence interval. The difference between the LFS and QUHS estimates is statistically significant at any conventional level for women only; it is not significant in the case of men. The difference between the extended and traditional definitions of LFP under the QUHS is significant at the 99 percent level for women and at the 90 percent level for men. *The labor force estimate has been extended to include subsistence agriculture. TABLE 3. LFP RATES FOR QUETTA CITY, QUETTA DISTRICT, AND URBAN BALOCHISTAN Women Men Data source/regional unit LFP (%) diff (QUHS – LFS) Obs. LFP (%) diff (QUHS – LFS) Obs. LFS 2020–21, Quetta District 6.5 9.6*** 1,165 74.9 –2.7 1,394 LFS 2020–21, urban Quetta District 2.4 13.7***   502 75.4 –3.2   591 LFS 2020–21, urban Balochistan 5.4 10.7*** 3,347 77.6 –5.4*** 3,921 LFS 2017–18, urban Balochistan 8.6 7.5*** 3,273 79.7 –7.5*** 3,705 QUHS 2021, Quetta City 16.1 4,733 72.2 5,113 Note: All estimates refer to individuals ages 15–64. About 28 percent of the population in Balochistan Province are urban, and the largest share live in Quetta District (29 percent), followed by Kech District (9 percent) (Population Census 2017). In this regard, the urban Balochistan estimate under the LFS is a good benchmark to compare with the estimates under the QUHS. *p < 0.1  **p < 0.05  ***p < 0.01 the LFS (figure 1). Furthermore, the difference between by international standards, they suggest that the meth- the two estimates (QUHS and LFS) is significant at the ods for measuring women’s work through survey data 99 percent level.5 These findings are in line with earlier should be improved. work in urban Peshawar (Peshawar Urban Household Survey 2019) showing that self-reported FLFP is 4 per- Adopting a more comprehensive definition of em- centage points higher than the LFS 2017–18 estimate ployment that includes production of goods for fam- for urban KP Province (Mancini 2021).6 Finally, in the ily use increases FLFP by a small margin. Extending case of men, the difference between the QUHS and LFS the concept of employment to include people engaged in rates is small and not always significant, indicating that production of agricultural goods for own consumption self-reporting likely leads to better estimates for women. (subsistence agriculture) generates a more comprehen- While the FLFP estimates under the QUHS are still low sive estimate of labor force participation (LFP). While 5 The P value of a two-sample t-test for the difference of the two estimates being zero is 0.00. Note that the comparison refers to different populations (Quetta District versus Quetta City). In addition, it may be muddled by confounding factors. Only experimental evidence can definitively pin down the size of respondent bias in this context. 6 The LFS 2017–18 is representative at the province level only. 6 men’s LFP does not change under the more comprehen- home. Qualitative research pertaining to FLFP in urban sive definition of employment, women’s LFP increases areas indicates that the traditional honor culture also in- from 16.1 to 17.6 percent (figure 1). This small difference fluences the sectors in which women seek employment is justified by the urban context. Nonetheless, in the case and creates a barrier to exploring jobs beyond those con- of women, the difference is statistically significant at any sidered socially acceptable for women. Hence, women in conventional level. The results are also in line with similar Pakistan are frequently engaged in home-based work or work conducted in Peshawar. in the education sector, and jobs for women in trade, food services, construction, transport, communications, and The following sections of this note are based on the tra- hospitality are virtually nonexistent. ditional definition of employment (and LFP),7 which ex- cludes production of goods for family consumption. Women have reported facing restrictions from male fam- ily members when they expressed interest in unconven- Social Norms and FLFP tional job roles, and men have opined that workplaces where the sexes mix freely are in defiance of local norms Research highlights a myriad of interconnected fac- (World Bank 2019). For jobs outside the home, women tors that greatly limit FLFP in Pakistan, including may have to restrict their job search to proximal employ- social and cultural restrictions on women’s mobility, ers or locations where it is convenient for male household safety concerns, rigid gender role ideologies, and members to accompany them. These trends are observed the association of women with family honor. As ex- and confirmed in the data from QUHS 2021 as well as in plained in multiple studies, an honor culture is strongly previous work in Peshawar. linked with social image or reputation—that is, repre- sentation of self in the eyes of others. For instance, An- Women’s employment remains limited mainly to the jum, Kessler, and Aziz (2019) have termed Pakistan as household setting due to mobility restrictions and having an “honor culture.” In patriarchal societies like the burden of having the sole responsibility for care Pakistan, in order to control women’s behavior and, and housework. In Pakistan, as in other parts of South hence, protect their honor, men often limit women from Asia, social norms around division of labor at home are leaving the home and require women in their families relatively inflexible. Women tend to perform most house- (or clans) to limit their connections to the outside world. hold and care work. Using the Pakistan Time Use Survey When women go out, they must be chaperoned and ap- 2007,8 Field and Vyborny (2015, cited by Tanaka and propriately garbed. Within an honor culture, women are Muzones 2016) found that women who are out of the typically expected to display shyness in their demeanor, labor force still spend many hours each day working on avoid eye contact with men, refrain from loud speech or household chores, and employed women, on average, laughter (especially in the presence of men), and limit spend more time per day on household and care work their interactions and conversations with males outside than employed men. The latter finding could be, in part, their family to necessary topics. This results in restrained because employed men typically work longer hours for a speech and movement for women (Sanauddin 2015), an wage than employed women. But it is also possible that effect that is significantly pronounced in Quetta. employed women spend fewer hours earning a wage be- cause they must juggle their time between market work Women typically abide by the honor code and are heav- and household work. Indeed, when women are asked in ily influenced by it in terms of their decision-making, LFSs why they are not available for work, the majority say mobility, and interaction with spaces outside the home. they have home responsibilities that prevent them from Any violation of the code leads to severe repercussions. working (Field and Vyborny 2015). By restricting women’s mobility and access to the public sphere, the honor code has a profound impact on the ex- Analysis of mobility from the Pakistan Time Use Survey tent and quality of women’s LFP. Asadullah and Wahhaj 2007 highlights wide mobility gaps between men and (2016) found that community norms such as the practice women across the country. For instance, women in Pa- of purdah have a negative effect on women’s participation kistan (age 11 or above) are about 16 times more likely in paid work. Since women often cannot leave home, they than men to remain at home and not report any trip in seek employment opportunities that can be managed at the past day. This disparity increases into adulthood and 7 Persons in employment are defined as all those of working age who, during a short reference period, were engaged in any activity to produce goods or provide services for pay or profit. They comprise (a) employed persons “at work” (people who worked in a job for at least one hour, including sporadic/casual work, but excluding people who exclusively work in subsistence agriculture) and (b) employed persons “not at work” due to temporary absence from a job or working-time arrangements (such as shift work, flextime, and compensatory leave for overtime) (International Conference of Labour Statisticians 2013). 8 See https://catalog.ihsn.org//catalog/3537/download/49900. 7 marriage. On average, women make half as many daily These mechanisms are hard to quantify, but they trips as men (2.8 and 5.4, respectively), with the widest play an important role in determining women’s gender gaps in work- and socio-cultural-related trips. representation in the workforce. In addition to socio­ Compared to men’s trips, women’s are 46 percent shorter demographic variables, the instrument developed for in duration, indicating that they are constrained in trav- the QUHS includes a set of questions aimed at eliciting eling outside their village or immediate neighborhood social and cultural norms. Table 4 presents descriptive (Adeel and Yeh 2018; Adeel, Yeh, and Zhang 2013). statistics on selected relevant characteristics of women in and out of the labor force. Results show that women International evidence supports the finding that percep- in the labor force are more likely to live in smaller tions around women’s roles as homemakers show a neg- households than women outside the labor force. As ative relationship with FLFP, suggesting that traditional expected, the presence of young children (ages 0–5) gender roles within the household also play a role (World in the household is negatively associated with FLFP. Bank 2022). This relationship appears to hold across These results reflect the role of childcare and house- South Asian countries. Analysis of World Values Survey hold work in women’s decision/ability to work. In re- data from Bangladesh, India, and Pakistan across multi- gard to education, women with tertiary education are ple periods and cohorts indicate a clear negative associ- more likely to be in the labor force than women with ation between women’s employment rate and gender-­ less education. However, women with no education based attitudes about men having a greater right to jobs (less than primary) are also overrepresented among when they are scarce (World Bank 2022). those in the labor force. Factors associated with female labor force participation  TABLE 4. DESCRIPTIVE STATISTICS FOR FLFP OF WOMEN 15–64 ABLE TO WORK FLFP = 1 (n = 928) FLFP = 0 (n = 3,786) Full sample (n = 4,810) Variable Mean SD Mean SD Mean SD Age 32.35 11.25 31.62 12.56 31.69 12.37 Afghan refugee (dummy) 0.21 0.41 0.10 0.31 0.12 0.33 Married (dummy) 0.64 0.48 0.63 0.48 0.63 0.48 Education (completed grades) Less than primary 0.62 0.49 0.52 0.50 0.53 0.50 Primary 0.13 0.33 0.20 0.40 0.19 0.39 Secondary 0.14 0.35 0.20 0.40 0.19 0.39 Tertiary or above 0.11 0.32 0.08 0.27 0.09 0.28 Relationship to household head Spouse 0.48 0.50 0.42 0.49 0.43 0.49 Daughter 0.23 0.42 0.28 0.45 0.27 0.45 Daughter-in-law 0.07 0.25 0.15 0.35 0.13 0.34 Other 0.22 0.42 0.15 0.36 0.17 0.37 Household composition (number of members in age/sex group) 8.93 4.49 9.61 6.05 9.52 5.85 0–5 1.20 1.40 1.24 1.50 1.23 1.49 6–14 2.04 1.77 2.03 2.00 2.04 1.97 15–24 2.13 1.89 2.31 2.06 2.28 2.03 25–44, females 1.23 1.04 1.13 1.01 1.15 1.02 25–44, males 0.97 0.98 1.21 1.20 1.17 1.17 45–64 0.91 0.85 1.13 0.90 1.09 0.89 65+ 0.34 0.60 0.30 0.59 0.31 0.59 Nuclear family (dummy) 0.45 0.50 0.43 0.49 0.43 0.50 8 TABLE 4. DESCRIPTIVE STATISTICS FOR FLFP OF WOMEN 15–64 ABLE TO WORK (CONTINUED) FLFP = 1 (n = 928) FLFP = 0 (n = 3,786) Full sample (n = 4,810) Variable Mean SD Mean SD Mean SD Education of household head (completed grades) Less than primary 0.44 0.50 0.28 0.45 0.31 0.46 Primary 0.18 0.39 0.21 0.41 0.20 0.40 Secondary 0.16 0.36 0.28 0.45 0.26 0.44 Tertiary or above 0.22 0.41 0.23 0.42 0.23 0.42 Food adequacy (dummy)a 0.64 0.48 0.75 0.43 0.73 0.44 Asset score b –0.57 1.94 0.26 1.61 0.12 1.69 Access to cell phone (dummy) c 0.49 0.50 0.46 0.50 0.47 0.50 Access to internet (dummy)d 0.33 0.47 0.32 0.47 0.32 0.47 Purdah (dummy) 0.97 0.16 0.99 0.12 0.98 0.13 Feels safe outside own neighborhood (dummy) 0.46 0.50 0.46 0.50 0.46 0.50 Experience of sexual harassment (dummy) 0.37 0.48 0.26 0.44 0.27 0.45 Involvement in decision-making e Work inside home 0.21 0.41 0.16 0.36 0.17 0.37 Work outside home 0.20 0.40 0.15 0.35 0.16 0.36 Community activity 0.21 0.41 0.15 0.35 0.16 0.36 Political activity 0.20 0.40 0.16 0.36 0.17 0.37 Shopping 0.51 0.50 0.51 0.50 0.51 0.50 Education 0.30 0.46 0.23 0.42 0.24 0.43 Marriage 0.04 0.18 0.04 0.20 0.04 0.19 Health 0.38 0.49 0.33 0.47 0.34 0.47 Beliefs in support of women’s work 0.97 0.17 0.89 0.31 0.90 0.29 Number of patriarchal norms the male household head agrees with (0–5) 2.98 0.94 2.89 0.98 2.90 0.97 Note: The sample includes all working-age women (15–64 years old). “Able to work” refers to women who are not in school/training and not ill/injured/ disabled. The variable for FLFP has 96 missing values. See detailed results (average marginal effects) from the probit FLFP equations in appendix B. a. Food adequacy takes a value of 1 if the male primary respondent considers the household’s food consumption adequate or better. Asset index estimates follow a similar methodology to that of the Demographic and Health Surveys (DHSs). The minimum value is –4.78, and the b.  maximum is 6.91. c. Access to cell phone includes both owning a cell phone and accessing one through a spouse, brother, or friend. d. Access to internet includes access both at home and through other means. Involvement in decision-making takes a value of 1 if a woman is included in the decision-making, whether she makes decision alone or together with e.  a partner. As discussed in the next sections, these findings are con- Furthermore, according to a probit model showing con- sistent with the female labor force in urban Quetta having ditional correlations for women’s participation in the low education in general and the fact that socially accept- labor force, the addition of controls for social norms to able jobs for women are predominantly low skilled and low a baseline specification for demographic characteristics value-added. In addition, women in the labor force are more does not affect the size or sign of the coefficients of these likely to live in poorer households, as measured by a lower characteristics (marital status, household composition, average for the food adequacy dummy variable and a lower education) (see detailed specifications and results in score on the asset index, compared to their peers outside appendix B). This implies that beliefs, norms, and edu- the labor force. This suggests that women take up employ- cation, for example, are systematically linked. Prevail- ment due to necessity. Also, not surprisingly, a positive atti- ing norms influence decisions about women receiving tude toward women’s work and perceived involvement in education and therefore have a strong impact on labor women’s decision to work inside or outside the home, as market outcomes. These results make the case for going well as involvement in decisions regarding women’s com- beyond the regression setting to better understand the munity and political activity, are positively associated with role of culture in influencing women’s representation in the probability of participating in the labor market. the workforce. 9 What Constrains Women’s Work? more, 52 percent of women have either incomplete pri- mary education or have never attended school (versus The previous section briefly discussed the intercon- 23 percent of men). The share of women who have com- nected factors associated with women’s persistent low pleted at least upper secondary school is only 16 percent labor market engagement in Pakistan, including social (versus 30 percent of men). The low level of women’s hu- and cultural restrictions on women’s mobility, safety man capital endowment is reflected in the overall lower concerns, rigid gender role ideologies, and the notion of rate of FLFP but also in the highly skewed education pro- honor. Based on the data collected through the various file of working women. As shown in figure 3, women with modules in the QUHS, this section presents the analysis of postsecondary education are a minority (8 percent) but and main findings on the barriers to women’s LFP and the are slightly overrepresented among employed women constraints employed women face in advancing in their (11 percent).9 Women with less than primary education job. This section focuses on the following constraints: low are also overrepresented among the employed, which is human capital endowment, limited agency, patriarchal linked to the fact that, along with the female labor force norms and traditional gender roles, the gender gap in in urban Quetta having low education in general, the so- care and household work, limited outside mobility, con- cially acceptable jobs performed by women in Quetta are cerns over safety in public spaces, and women’s sources predominantly low skilled and low value-added (see the of information about jobs and household welfare. The next section). analysis suggests that all of these play an important role in determining whether women can work for pay, what jobs In addition, women in younger cohorts attain, on aver- they can do, and how they can perform within these jobs. age, more education than those in older cohorts. While 13.3 percent of women aged 45–64 have achieved sec- Human Capital Endowment ondary education or above, this figure is more than double among women aged 15–24 (32.6 percent). On Women’s human capital endowment in Quetta is the contrary, the difference between younger and older low, reflected in the overall FLFP rate and the highly male cohorts is only 4 percentage points (42.3 and 46.6 skewed educational profile of employed women. Edu- percent, respectively), suggesting educational achieve- cational attainment in urban Quetta is low in general, but ment among men is consistent across age groups. This it is strikingly low among women. Whereas 80 percent of is particularly relevant in the context of the COVID-19 working-age men are literate (can read and write), only pandemic, when school closures imposed an extra 50 percent of working-age women are literate. Further- burden for girls and younger women in school. For in- The low level of women’s human capital is reflected in the skewed educational profile of working women.  FIGURE 3. EDUCATIONAL ATTAINMENT (COMPLETED GRADES) BY SEX AND WORKING STATUS Share of working-age adults (%) Women Men 70 62% Below primary 60 Completed primary 51% 52% Completed secondary 50 Completed upper secondary Completed postsecondary (tertiary) 40 31% 30 27% 24%26% 23% 20% 20% 20% 21% 20 19% 19% 19% 17% 16% 13% 11% 13% 12% 11% 12% 13% 10 7% 7% 8% 8% 7% 8% 0 Employed Not employed All working-age Employed Not employed All working-age Note: All estimates refer to individuals ages 15–64. Below primary = did not complete grade 5. Completed primary = completed at least grade 5 but not grade 12. Completed lower secondary = completed at least grade 10 but not grade 12 (may include vocational diploma obtained after middle or metric school). Completed upper secondary = completed at least grade 12 but not the second year of university (may include vocational diploma obtained after grade 12). Completed postsecondary = completed at least the second year of university. 9 No similar pattern emerges among men. Men with lower levels of education are relatively more likely to participate in the labor market and to be employed. 10 stance, recent qualitative work on changing household strictly individual matters—for instance, those regarding dynamics in response to mandated COVID-19 school one’s own political participation or access to health care. closures in Punjab concludes that re-enrollment of girls is particularly challenging given their increased Beliefs on whether women should work for pay vary load of household tasks, loss of learning, and lack of depending on educational attainment, the respon- engagement with educational TV programming (Malik dent’s age, household composition, and other char- et al. 2022). acteristics. Given the prominent role of the husband in making decisions regarding a spouse’s labor market en- Agency gagement, it is worth noting that 19 percent of men in Quetta believe women should never work for pay (in Pe- Women’s lack of agency encompasses all aspects of shawar, this figure is 25 percent). However, the share of life and contributes to low FLFP. Low levels of FLFP women who believe that women should not work under contrast with women’s beliefs concerning work for pay. any circumstance is 8 percent (in Peshawar, this figure is Overall, 90 percent of working-age women believe that 13 percent). This signals greater acceptability of female women should work for pay (compared to 76 percent of work among women themselves. A closer look shows men). But only 6.6 percent of women are able to decide that men and women with lower secondary education autonomously whether they can work for pay outside the or greater are more likely to accept female work. Simi- home. When it comes to the decision to work inside the larly, younger cohorts of men and women (ages 15–18) home, this number rises to just 7.9 percent. Table 5 shows express more acceptability of paid work among women. that most women indicate their husband or father is the In the case of men, those living in households with chil- primary decision-maker about whether they can work for dren ages 0–5 are less likely to agree with women work- pay and whether from home or not. Furthermore, women ing than those in households with no young children. In are often excluded from decision-making about even the case of women, there is no difference in the share who Women’s lack of agency encompasses all aspects of life.  TABLE 5. DECISION-MAKERS ABOUT ASPECTS OF WOMEN’S LIVES Mother/ Father/ Parents/ Other family You and QUHS item You (%) Spouse (%) mother-in-law father-in-law parents-in-law members spouse (%) (%) (%) (%) (%) Who mainly decides … if you can work outside your 6.6 48.5 9.0 4.3 17.4 9.4 4.9 house for pay? if you can work inside your 7.9 47.2 9.0 4.2 17.6 9.4 4.8 house for pay? whether you can participate 7.5 47.3 9.0 4.1 17.7 9.5 4.9 in political activities? about buying goods like 39.7 24.3 11.4 6.0 9.8 6.2 2.7 clothes/shoes for yourself? to start or continue your 12.4 41.2 12.0 3.9 15.5 11.0 4.0 education? to whom and when you 2.9 3.7 1.1 2.7 16.3 70.1 3.2 should be married? to seek professional medical 19.0 33.8 15.2 5.5 13.2 9.9 3.6 treatment? to seek professional medical help if you think you have 17.3 37.6 16.7 4.7 11.7 8.5 3.5 COVID-19? Married women only whether you should have 3.0 45.3 45.4 1.2 3.4 1.0 0.8 more children? whether to use birth control? 3.2 45.4 44.8 1.3 3.5 1.1 0.8 whether to buy or sell 4.7 64.0 13.5 1.6 7.7 4.4 4.2 goods? 11 Patriarchal norms impact the decision to send girls to school, creating a vicious cycle threatening women’s empowerment.  FIGURE 4. REASONS FOR NEVER HAVING ATTENDED SCHOOL Share of working-age women (%) Elders/parents/brother/husband did not approve 30.1 Schooling is not common in community 26.7 Girls/women do not need formal schooling 11.3 School too expensive 6.9 Had to work at home 6.6 Schooling is not approved by local influential leaders 3.6 School too far away/not available 3.4 Formal education not useful 3.1 Other reason (specify) 2.4 Illness in the family 1.2 Marriage 1.1 0 5 10 15 20 25 30 35 Note: Graph shows reasons with a share of 1 percent or more only. agrees with women working based on the presence of When it comes to starting/continuing their education, young children. According to the asset index, men living about 57 percent of women say that their father, hus- in wealthier households are in greater agreement with band, or father-in-law is the main decision-maker about women being able to work, whereas no major difference whether they will pursue education. Furthermore, among is found across wealth quintiles in the case of women. working-age women who have never attended school Lastly, employed men are slightly less supportive of fe- (about half of all women aged 15–64), the vast majority male work than nonworking men. cite a reason for never having done so related to patriar- chal norms. Around 72 percent of these women say that Norms a male relative, their husband, or (to a lesser degree) an influential leader did not approve; that schooling is not Patriarchal norms affect not only the decision (and common in their community; or that schooling is not ability) to work but also the decision to send girls to perceived as important for girls/women (figure 4). This school, which creates a vicious cycle, as education pattern creates a vicious cycle in which women do not re- increases the likelihood women will engage in work. ceive education, which prevents them from accessing bet- Most men in Quetta subscribe to deep-rooted patriarchal norms.  FIGURE 5. AGREEMENT WITH TRADITIONAL GENDER ROLES AMONG MEN Share of primary male respondents (%) Men are better at starting businesses than women. 93.3% It is better if a man earns and a woman takes care of the 82.1% home and children. If parents are in need, daughters/daughters-in-law should take more 81.6% caring responsibilities than sons/sons-in-law. Mothers should take more childcare responsibilities of 78.6% children than fathers. If a woman earns more than her husband, it is 64.1% likely to cause problems. A woman should do most of the household chores even if the 52.6% husband is not working. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Note: The set of questions on agreement with traditional gender roles was applied to male primary respondents (ages 20–60) only. 12 Childcare responsibilities impact women’s ability to work.  FIGURE 6. DISTRIBUTION OF WORKING-AGE WOMEN ACROSS HOUSEHOLD TYPE Share of working-age women (%) 100% 21.1 22.6 30.6 80% Couple with children and other family members at home (extended) 60% Couple with children, no others at home (nuclear) Couple without children 68.1 66.2 Single with children 40% 55.9 Single-person, no-children households 20% 0% Employed women Nonemployed women All working-age women Note: Single-person households, single parents with children, and couples without children may have other relatives at home. ter and/or more productive jobs and ultimately resigns Unpaid care and household work them to low earnings. Hence, the opportunity cost of stay- ing home is low, which discourages them from joining the It is well established globally that women perform labor market. In addition, women with no education are more unpaid care and household work than men, less likely to be employed than women with some educa- but this disparity is greatest in Pakistan. According tion (employment is 12.2 percent among those without to UN Women (2019), Pakistani women spend 11 addi- schooling, compared to 19.7 percent among those with tional hours on unpaid household chores and caregiving some schooling). for every hour spent by men on the same activities. While there are no time use data for Quetta, evidence for urban Traditional gender roles assign women a range of Peshawar indicates that men spend virtually no time housework and care responsibilities, which inhibit on house or care work, while women spend on average their ability to work. As shown in figure 5, most men in 5.3 hours per day on this kind of work, an amount that Quetta subscribe to beliefs that women’s rightful role is decreases only slightly when they are employed. Fur- taking care of the home and children, indicating that patri- thermore, in a recent survey on the COVID-19 pandemic archal gender norms are deeply rooted. While QUHS 2021 covering urban areas in Pakistan (Taş et al. 2021),11 more asks primary male respondents only about agreement women than men report an increase in unpaid work after with traditional gender roles, data for urban Peshawar the COVID-19 pandemic began regardless of employment indicate that women also identify themselves with such status, though the gender gap is largest between work- roles. More than 80 percent of women agree that mothers ing men and women. In line with this evidence, data from should take more childcare responsibilities than fathers, QUHS 2021 show that employed women work, on aver- and more than 90 percent agree that it is better if men age, one hour per day less than they did before February earn money and women do housework. Not surprisingly, 2020 (6.4 and 7.7 hours, respectively), likely because of employed women in Quetta are more likely to be found in increased housework and care responsibilities.12 extended-family households10 than nonemployed women because other family members might provide support Mobility with childcare and housework responsibilities, therefore enabling them to work (figure 6). Similarly, nonemployed Women, including those who are employed, spend women are more likely to be found in nuclear households, very little time outside the home and are usually ac- where childcare responsibilities are more likely to fall ex- companied when they do. Table 6 shows that women clusively on them and therefore limit their ability to work. left their home no more than three times during the 10 Extended-family households refer to a couple with children and other family members at home. Nuclear households refer to a couple with children only. 11 The authors used the administrative database of Pakistan’s largest online job platform and an online COVID-19 survey. They collected information about the socioeconomic status and coping strategies of job seekers and employers. 12 The data show no difference in average hours worked per day among employed men (9.6 before and after the COVID-19 pandemic began). 13 week before COVID-19–related lockdowns took place, shown in the next section), as this allows families to keep with employed women being slightly more mobile than up with prevailing social norms. Not surprisingly, among nonemployed women. Women mostly leave their home employed women, only 12 percent indicate leaving their to go to another house (82 percent); less often, they go to home to go to work. shops to buy groceries/clothes (45 percent), visit a clinic or health worker (38 percent), or go to social events (26 Safety percent). The overwhelming majority observe purdah (table 7), and most are usually accompanied when leav- Safety concerns when going out can further limit ing their home (85 percent, or 80 percent among em- women’s mobility and, therefore, are extremely im- ployed women). The most cited companions are other portant for female empowerment and FLFP. Women women, the husband, or a child. Notably, these results tend to feel safe within the bounds of their limited move- were observed among all working-age women regardless ments, but there is evidence that they would feel much of their working status, which implies that the prevailing less comfortable expanding their mobility. According to social norms and values are equally binding for working QUHS 2021, as many as 27.2 percent of all women report and nonworking women. This is consistent with the idea having experienced some form of sexual harassment out- that if women work, they most likely work from home (as side the home (figure 7). These findings are in line with Women, on average, regardless of their working status, spend little time outside home and are usually accompanied when they do.  TABLE 6. TIME SPENT OUTSIDE THE HOME AND REASONS FOR LEAVING THE HOUSE QUHS item Employed women Nonemployed women All working-age women In a typical week before the COVID-19 pandemic, how many days 3.0 2.6 2.7 would you go outside your home? (average number of days per week) Reasons for leaving the home before the COVID-19 pandemic (% positive responses for each item) To visit family, friends, or neighbors 84 82 82 To go to shops to buy groceries/clothes 49 44 45 To visit a clinic or health worker 42 38 38 To go to social events 30 26 26 To walk/for leisure 11 16 15 To attend school/literacy classes 10 13 12 To take children to school 4 4 4 To go to work 12 0 3 For Quran classes, dars, or other gathering 2 3 3 TABLE 7. OBSERVANCE OF PURDAH AND BEING ACCOMPANIED WHEN LEAVING THE HOME Nonemployed women All working-age QUHS item Employed women (%) (%) (15–64) women (%) Do you observe purdah? Yes 97.3 98.5 98.3 Who usually accompanies you? (if any) Child 18.7 14.5 15.2 Husband 29.9 28.9 29.0 Male relative 7.9 7.0 7.1 Female relative or nonrelative 41.8 48.1 47.1 Other 1.8 1.6 1.6 Total 100.0 100.0 100.0 Note: In the case of purdah, yes indicates any of the following responses to the question, “When you go outside for work or schooling or market, do you…?”: cover head only, cover body but not face, or cover whole body. No indicates not observing any purdah. 14 Safety concerns can further limit women’s mobility.  FIGURE 7. EXPERIENCE OF SEXUAL HARASSMENT OUTSIDE THE HOME Share of women respondents (%) 30 27.2 25 20 19.2 15.6 15 11.3 10 8.1 5 0 Any type of harassment Inappropriate Inappropriate use of Gestures/actions of Other harassment staring/comments phone/email sexual nature BY TYPOLOGY At least a quarter of working-age women (regardless of working status) express safety concerns when walking alone in public spaces.  TABLE 8. WOMEN’S FEELINGS OF SAFETY WALKING ALONE IN THEIR NEIGHBORHOOD Nonemployed All working-age QUHS item Employed women (%) women (%) women (%) Do women feel safe walking alone outside in their neighborhood? Yes, anytime 45.7 45.9 45.8 During daytime only 26.1 29.0 28.6 No 28.2 25.2 25.6 results for urban Peshawar, where 30.8 percent of women mean trip duration is higher for walking and personal au- reported an episode of sexual harassment. In both cities, tomobile trips but considerably lower for travel by bus, inappropriate comments/staring stands as the most bicycle, or other means of travel. Potential safety issues common episode of harassment but with very differ- and interaction with unwanted men seem to affect wom- ent shares (19 percent in Quetta, 28 percent in Pesha- en’s trips and choice of mode the most (Adeel, Yeh, and war). The second-most common form of harassment in Zhang 2013). Quetta is inappropriate use of phone/email (16 percent), whereas in Peshawar it is gestures/actions of a sexual Furthermore, these findings on women’s experience nature (11 percent of women). Table 8 shows that most of harassment, safety concerns, and preferred mode of women (74.4 percent) consider walking alone in their transportation suggest that the prevailing social norms own neighborhood to be safe; however, 29 percent would that restrict women’s mobility are also consistent with feel safe only during the day, and over a quarter would feel high risks to their personal safety and dignity. unsafe walking alone outside the neighborhood. As table 8 shows, safety perceptions and concerns are very simi- Access to Information lar between the employed and the nonemployed, which shows that safety is a concern for all women. Lack of information about labor market opportu- nities significantly hampers FLFP. Similar to findings The chosen mode of transport when women go outside from the LFS, official unemployment is very low in Quetta. is public taxi (47.6 percent), followed by own/household According to QUHS 2021, the unemployment rate is 0.5 car (25.3 percent). Less preferred methods are walking percent for women and 2.4 percent for men. In the case of (14 percent) and public bus (9.3 percent).13 This is consis- women, this implies that nonworking women are mainly tent with prior analysis from the Time Use Survey 2007 out of the labor force rather than unemployed (women showing that women in Pakistan rely on personal (rather out of the labor force represent 84 percent of working-age than public) modes of motorized transport. For instance, women). Interestingly, about 8 percent of women (and 13 The question on mode of transport refers to “during times of coronavirus”; therefore choices might also be influenced by perceived risk of contagion or lockdowns. 15 10 percent of men) who are out of the labor force report Household Welfare willingness to work even though they are not currently looking for a job (this refers to the economically inactive, Women living in poorer households tend to have who answered that “at present” they want to work, cor- higher participation rates, and FLFP decreases as responding to 275 observations for women and 132 for household welfare increases, suggesting that women men). Among these women, the most reported reason take up employment due to necessity and to increase for not searching for a job (figure 8) is lack of knowledge household consumption. QUHS data allow for estima- about labor market functioning (37 percent), followed tion of an asset index at the household level based on the by care responsibilities (13.7 percent) and cultural and household’s ownership of selected assets. The index fol- family prohibition (11.5 percent). Men who are out of the lows a methodology similar to that of the wealth index labor force represent 28 percent of working-age men, and from the DHSs. Using the asset score as a proxy for house- the share of these who would like to work is 10 percent. hold welfare, the analysis shows that female employment The distribution of reasons among men is quite different and FLFP are higher at the lower quintiles of the score and less biased than for women; still, 17 percent of men and decrease as household welfare increases. For in- do not know how to look for employment either. stance, while 27.7 percent of women from households in the first wealth quintile are in the labor force, the share This lack of knowledge is mostly related to the job search goes down to 9.7 percent for women living in the wealth- process: 36 percent of women who are willing to work re- iest households.14 These findings suggest that women are port not looking for a job because they do not know how. often required to take up employment (typically informal An additional 1 percent do not know what types of jobs jobs) to increase household consumption. In fact, among they can do for pay, possibly proxying for lack of educa- women, economic necessity is the fourth-most cited tion and/or specific skills. Furthermore, if these women condition that makes women working acceptable, after who lack knowledge about job opportunities were to join home-based or close-to-home work or working while the labor market, FLFP in urban Quetta could increase observing purdah. from 16 to 19 percent. Characteristics and Quality of Women lack knowledge about labor market functioning.  Women’s Jobs FIGURE 8. MAIN REASONS THAT PEOPLE WHO ARE OUT OF THE LABOR FORCE BUT WANT TO WORK DO NOT TRY TO Employment status in urban Quetta differs sub- FIND A JOB OR START A BUSINESS stantially by gender: working women are mainly own-account workers, particularly those with low Number of respondents education, while men are mostly paid employees. 6 Labor market outcomes in Quetta tend to be strongly 7 Women 7 segmented by gender, with more than half of employed 25 women (58.5 percent) working on their own account 37 and 60 percent of men working as paid employees (ei- 16 ther as casual or by piece rate) (figure 9). However, when 27 Men 16 looking at employment status by educational attainment, 4 the pattern among working women reverses.15 While 17 own-account is the most popular status for women with 0 10 20 30 40 low education (69.9 percent), the majority of highly edu- Waiting for potential Family/household cated female workers (only a quarter of working women) job/job offer constraints are paid employees (65.6 percent). By contrast, employ- Coronavirus related Don't know how to ment status among men does not vary with educational (business closed, fear) look for work attainment, though highly educated men have a higher Discouraged workers probability of working as paid employees than those with Note: Single-person households, single parents with children, and couples less education. (See box 2 for definitions of employment without children may have other relatives at home. statuses discussed in this note.) 14 While the same pattern is observed among men, the difference in LFP between the lowest and highest quintiles is smaller. 15 In this report, low education refers to less than lower secondary education (below matric/grade 10). High education refers to having completed at least lower secondary education. 16 Employment status in urban Quetta differs substantially by gender, as women are mostly own-account workers. FIGURE 9. EMPLOYMENT STATUS BY GENDER AND EDUCATIONAL ATTAINMENT Share of respondents (%) 100% 4.9 5.2 2.5 5.4 90% 7.3 4.7 80% 24.5 34.3 Own account 70% 54.1 Employee 60% 60.0 65.7 65.6 50% Contributing family worker 40% Employer 69.4 30% 58.5 Other 20% 39.3 31.3 10% 23.3 23.5 0% Men Women Men Women Men Women ALL EMPLOYED LOW EDUCATION HIGH EDUCATION Note: Own account includes own-account workers (nonagriculture) and owner cultivators (not in subsistence agriculture). Employees include regular and casual paid employees, paid workers by piece rate, and paid nonfamily apprentices. Contributing family workers are those who work in a business owned by a family member or help a family member who works for someone else. “Other” includes paid nonfamily apprentices, sharecroppers, contract cultivators, members of a producer’s cooperative, and others. High education means completing at least lower secondary education. Low education means less than lower secondary education (below matric/grade 10). BOX 2. DEFINITIONS OF WORKING STATUS PRESENTED IN THIS REPORT Employees are workers who hold jobs defined as paid employment jobs, where incumbents hold explicit (written or oral) or implicit employment contracts that give them a basic remuneration that does not directly depend upon the revenue of the unit for which they work. In this report, based on QUHS 2021, this category includes the following workers: regular paid employees with fixed wages, workers receiving a fixed salary from a family business, casual paid employees, and paid workers by piece rate or work performed. Employers are workers who, working on their own account or with one or a few partners, hold jobs defined as self-employment jobs (jobs where the remuneration directly depends on the profits derived from the goods and services produced) and, in this capacity, engage on a continuous basis one or more persons to work for them as employees. Own-account workers are workers who, working on their own account or with one or more partners, hold jobs defined as self-employment jobs but have not engaged on a continuous basis any employees to work for them. In this report, this category includes contributing workers in nonagricultural activities (representing 31 percent of employed men and 56 percent of employed women in urban Quetta), workers in agriculture (representing 0.5 percent of employed men and 2 percent of employed women), and owner cultivators (0.2 percent of em- ployed men and 0.1 percent of employed women). Contributing family workers are those who work in a market-oriented business owned and operated by a family member or those who help a family member who works for someone else. In this report, this category corresponds to contributing family workers, mainly in nonagriculture. Other workers include the following categories per QUHS 2021: paid nonfamily apprentices, sharecroppers, contract cultivators, members of producers’ cooperatives, and other workers not classifiable by status. Home-based workers are defined as (a) own-account workers and contributing family workers involved in production of goods and services in their homes for the market and (b) workers carrying out work in their homes for remuneration, resulting in a product or service as specified by their employer(s) irrespective of who provides the equipment, materials, or other inputs used, and contributing family workers helping such 17 workers.16 HBWs work from their home or a family friend’s home and include employees (as defined above), employers, own-account workers not in agriculture, and contributing family workers not in agriculture. Al- though the International Labour Organization (ILO) and WIEGO do not count employers as HBWs, given the nature of jobs women perform in urban Quetta, they are counted as such in this report (employers represent only 1 percent of employed women in urban Quetta). ILO (2021) recognizes homeworkers as a subgroup of HBWs. In addition to working from home, homeworkers are defined statistically as employees or dependent contractors. According to QUHS 2021, regular paid employ- ees and casual paid employees (who are more likely to be homeworkers) represent only 1 percent of women HBWs (see table 11). Women HBWs who are paid by the piece (20 percent) are mainly garment and handicraft workers. This suggests that the share of women HBWs who are homeworkers in urban Quetta is very small.17 Working women are highly segregated by industry Occupational sex segregation in urban Quetta is in and occupation, often performing low-value-added line with similar findings for urban Peshawar and ur- activities aligned with the skills gap and norms on ban Pakistan. Data for urban Peshawar from the Pesha- socially accepted jobs. Figure 10 shows that working war Urban Household Survey (Mancini 2021) show that women in urban Quetta are employed mainly in two almost 80 percent of all employed women are concen- sectors: manufacturing, which employs the majority of trated in the 10 most common occupations for women women (61.7 percent), mostly in the textile sector pro- workers, as opposed to 60 percent of employed men in ducing garments, followed by education, which employs a smaller share of women (9.9 percent).18 Segregation Working women are highly segregated by industry and of women by industry is apparent when observing the occupation.  two most prevalent working statuses among women (own-account and employee), though women working FIGURE 10. DISTRIBUTION OF WOMEN ACROSS SECTOR OF as wage employees are overrepresented in the education ACTIVITY BY EMPLOYMENT STATUS (ACCORDING TO ISIC sector relative to all working women (figure 10). Along CLASSIFICATION) with limited sectoral diversity, female employment is Share of women workers (%) also concentrated in socially accepted occupations such C - Manufacturing 79.1 36.4 as garment workers, handicraft workers, and teachers or 61.7 teachers’ aides, and there is limited representation of ur- H - Transportation and storage ban women in services and retail. The top 10 occupations M - Professional, scientific, among women account for 92 percent of female employ- and technical activities ment, whereas in the case of men, the top 10 occupations 2.4 account for only 62 percent of employment (table 9). Fur- P - Education 23.6 9.9 thermore, segregation by occupation increases among Q - Human health and social workers with low education. Women’s most frequent work activities occupations differ greatly by education level, more so S - Other service activities than among men. Women with low education tend to be manufacturers, refuse workers, cleaners, or shop sales- T - Activities of households as employers; people, whereas highly skilled women tend to be teachers, undifferentiated goods childcare workers, or nursing professionals. Overall, the 0 20 40 60 80 100 occupational profile of working women reflects stereotyp- Own-account worker Employee All employed ically female roles and aligns with preferences expressed by men regarding the conditions under which it is accept- Note: Graph shows International Standard Industrial Classification (ISIC) able for women to work for pay (table 10). sectors that employ at least 1.5 percent of all working women. 16 For more information on HBWs, see the definition by Women in Informal Employment: Globalizing and Organizing (WIEGO) at https://www.wiego.org/ definition-home-based-workers. 17 For more on the ILO’s definitions, see https://ilostat.ilo.org/resources/concepts-and-definitions/description-employment-by-status. 18 According to the latest wave of the LFS (2020–21), for women ages 15–64 in Quetta District, the distribution of employed women (excluding those in agriculture) follows the same pattern across industries seen in QUHS 2021, but the share of women in each industry is different. For instance, the share of women working in manufacturing is smaller (29 percent) in the LFS than in the QUHS, whereas the shares of those in education and in human health are larger (34 and 24 percent, respectively). In addition, the LFS yields a greater share of women classified as service and sales workers (30 percent) and a smaller share of craft workers (16 percent). While the QUHS yields a higher FLFP rate, it also yields a slightly different composition of female employment, suggesting that the gap in LFP is not random. 18 The top 10 occupations among women account for 92 percent of female employment.  TABLE 9. TOP 10 MOST COMMON OCCUPATIONS BY GENDER AND EDUCATION LEVEL (ACCORDING TO ISCO 3-DIGIT CLASSIFICATION) Women (%) Cumul. (%) Men (%) Cumul. (%) 1 Garment and related trades workers 63.6 63.6 Shop salespersons 15.0 15.0 2 Handicraft workers 11.8 75.4 Street and market salespersons 14.5 29.5 3 Secondary education teachers 4.6 80.0 Building frame and related trades workers 6.4 35.9 4 Primary school and early childhood teachers 4.5 84.5 Car, van, and motorcycle drivers 6.3 42.3 5 Refuse workers 2.7 87.2 Garment and related trades workers 4.3 46.5 6 Hairdressers, beauticians, and related 1.4 88.6 Machinery mechanics and repairers 4.0 50.5 7 Other teaching professionals 1.1 89.7 Business services agents 3.8 54.3 8 Childcare workers and teachers’ aides 1.0 90.7 Numerical clerks 3.2 57.5 9 Other health professionals 0.9 91.6 Regulatory government associate professionals 2.5 60.0 10 Shop salespersons 0.9 92.4 Domestic, hotel, and office cleaners and helpers 2.5 62.5   Women with low education Men with low education 1 Garment and related trades workers 74.8 74.8 Street and market salespersons 17.6 17.6 2 Handicraft workers 13.9 88.7 Shop salespersons 14.6 32.2 3 Refuse workers 3.6 92.4 Building frame and related trades workers 11.5 43.7 4 Shop salespersons 1.2 93.5 Car, van, and motorcycle drivers 8.8 52.5 5 Domestic, hotel, and office cleaners and helpers 1.0 94.5 Garment and related trades workers 6.3 58.8 6 Personal care workers in health services 0.9 95.4 Machinery mechanics and repairers 6.1 64.9 7 Hairdressers, beauticians, and related 0.8 96.1 Food processing and related trades workers 3.3 68.2 8 Painters and building structure cleaners 0.5 96.7 Manufacturing laborers 3.3 71.5 9 Primary school and early childhood teachers 0.4 97.1 Domestic, hotel, and office cleaners 2.7 74.1 10 Building frame and related trades workers 0.3 97.4 Building finishers and related trades workers 2.4 76.5   Women with high education Men with high education 1 Garment and related trades workers 28.6 28.6 Shop salespersons 15.4 15.4 2 Secondary education teachers 18.6 47.2 Street and market salespersons 11.6 26.9 3 Primary school and early childhood teachers 17.0 64.1 Numerical clerks 5.8 32.7 4 Handicraft workers 5.3 69.4 Business services agents 5.6 38.4 5 Other teaching professionals 4.6 74.1 Regulatory government associate professionals 4.4 42.8 6 Childcare workers and teachers’ aides 4.0 78.0 Car, van, and motorcycle drivers 4.0 46.8 7 Other health professionals 3.8 81.8 Protective services workers 2.6 49.5 8 Client information workers 3.4 85.2 Primary school and early childhood teachers 2.5 51.9 9 Hairdressers, beauticians, and related 3.3 88.5 Garment and related trades workers 2.4 54.3 10 Nursing and midwifery professionals 2.3 90.8 Secondary education teachers 2.3 56.6 Note: Men and women with high education are those who have completed at least lower secondary education, while men and women with low education have not. the 10 most common occupations for working men. In it is easier to create jobs aligned with socially acceptable urban Pakistan, the top five occupations make up two- occupations. For instance, according to the Pakistan LFS thirds or more of the share of employment for women, 2020–21, the female employment rate (ages 15–64) in whereas there is more diversity for men. Urban men are urban Quetta District was 2.2 percent, whereas in rural more likely to be engaged in trades such as construction Quetta District it was 9.7 percent. Similarly, for Baloch- and services (shopkeepers), whereas urban women are istan Province, the female employment rate was 3.8 per- more likely to be engaged as domestic help or in apparel cent in urban areas and 17.2 percent in rural. A previous and textiles (Amir et al. 2018). This partially explains why round of the LFS (2017–18) was also in line with these FLFP rates are higher in rural areas of Pakistan: on a farm, results: the employment rate among women ages 15–64 19 in urban Balochistan was 5.8 percent, whereas the rate FIGURE 11. CHARACTERISTICS OF AN IDEAL JOB increased to 9 percent in rural areas. FOR WOMEN, ACCORDING TO WOMEN Share of women respondents (%) A closer look at the two most prevalent working sta- 0.5 tuses of women—own-account workers and paid employees—confirms the pattern of segregation by 4.4 industry but also reveals that industry/occupation choice varies by status. Women working as own-account Work from home workers are more likely to be found in manufacturing (79 Work outside the percent), while female paid employees have greater rep- home: government job 41.2 54.0 Work outside the resentation in education (23.6 percent) compared to the home: private/NGO job overall average (figure 10). Similarly, in terms of occupa- Others tion, own-account women work mainly as craft (and re- lated) workers, whereas paid employees are more evenly split between craft workers (46 percent) and profession- als (32 percent). This further reflects lower representa- tion of women in high-skilled professions and indicates Note: Women were asked what their ideal job was before the COVID-19 that own-account jobs held by women tend to be low pandemic. skilled (small-scale and home-based). of women reported a preference for it being home-based There is a strong emphasis on employment that is (figure 11). However, it is noteworthy that for 45 percent acceptable to men and ideal to women, reflecting of women, work outside the home is the ideal job, par- the actual job profile of women. Working from home ticularly a government job. Such preferences are clearly is the most important condition that makes female em- reflected in the actual profile of jobs held by women. For ployment acceptable to men and women and is consid- instance, less than 1 percent of men work inside the home ered the ideal form of employment for women. However, versus 83 percent of women (table 11). As a reference, there are important gender differences in the conditions work from home is more prevalent among women in ur- that make paid work acceptable for women (table 10). ban Quetta than among those in urban Peshawar. In the According to women, the two most important conditions case of Peshawar, 65 percent of women work inside the besides home-based work are that women can observe home, compared to just 5 percent of men. purdah while working and, if the job is outside the home, the workplace must be close by. In contrast, men consider In addition, as table 11 shows, women with low educa- having no interaction with non-mahram men and work- tion are more likely to work from home, whereas women ing as a teacher or nurse make it acceptable for women with high education are much more likely to work in a to work (only 0.3 percent of women cite the latter). When shop or office (43 percent). This is consistent with the asked about the characteristics of an ideal job, 54 percent vicious cycle described in the previous section whereby Strong emphasis on forms of employment “acceptable” to men and “ideal” to women   TABLE 10. CONDITIONS THAT MEN AND WOMEN VIEW AS ACCEPTABLE FOR WOMEN TO WORK FOR PAY Condition Men (%) Women (%) Home-based work 54.1 47.4 Work as a teacher or nurse 10.9 0.3 No interaction with non-mahram men 8.8 5.4 No overnight travel or travel outside the city 7.1 0.5 Ability to work while observing purdah 6.9 20.7 If work is outside the home, the workplace should be close by. 6.2 12.8 If work is outside the home, proper coronavirus safeguards are in place. 2.0 0.0 A good salary 1.7 0.0 If work is outside the home, the workplace is sex segregated. 1.4 0.4 Economic necessity 1.0 6.5 Note: Responses shown for categories with a response of 1 percent or more. 20 Women engage in jobs that do not require physical interaction, and most work from home, particularly those with low education.  TABLE 11. LOCATION OF MAIN EMPLOYMENT BY EDUCATION Low education (%) High education (%) All employed (%) Women At home 92.7 45.5 81.3 At other’s home (family friend or employer) 3.2 0.5 2.5 On the street or outside 1.7 6.2 2.8 In a shop, office, or factory 1.8 42.9 11.8 Other 0.6 4.9 1.6 Men At home 0.5 0.5 0.5 At other’s home (family friend or employer) 2.7 2.3 2.6 On the street or outside 33.1 13.1 22.9 In a shop, office, or factory 61.9 81.9 71.9 Other 1.8 2.3 2.2 Note: Men and women with high education are those who have completed at least lower secondary the social norms that determine women’s educational at- half of men working from home indicated they did so be- tainment (or the decision to go to school) later determine cause it was cost-effective, and 30 percent stated it was a path toward employment outside the household, which “easier to work from home.”19 These results show that in often generates better earning opportunities. the case of women, working from home is not much of a choice but rather an alternative to other forms of work HBWs make up the majority of employed women in that are less aligned with existing social norms. urban Quetta, in line with national figures. Under a statistical definition of HBWs in line with ILO/WIEGO— Most women HBWs in urban Quetta are own-account which includes workers who carry out remunerative garment workers who have low chances of upward work in their homes (work resulting in a product or ser- mobility. The second-most prevalent employment sta- vice), whether as own-account workers, paid workers, tus among women HBWs is paid employment by piece or contributing family workers—the share of employed (20 percent) (table 12). Women HBWs are also largely women in urban Quetta who are HBWs is 78.6 percent. employed in manufacturing as garment workers, though This is in line with national figures showing that home- about 10 percent are teachers (likely telecommuting due based work has grown in recent years in Pakistan due to pandemic-related school closures or tutoring neigh- to an increase in female workers and a decrease in male borhood children from home). The share of HBWs is workers. In fact, nonagricultural home-based work has higher among women with low education (88 percent), become a primary source of employment for women in compared to women with high education (40 percent). Pakistan. According to LFS data, between 2013/14 and Moreover, women HBWs with high education tend to 2017/18, the share of female HBWs in nonagricultural have completed lower secondary education only and employment increased from 20.5 to 46.2 percent (Akhtar work in manufacturing (garments, handicrafts) or teach- 2020). See box 2 for detailed definitions of HBWs and ing. This suggests that while these workers have flexibil- homeworkers. ity in terms of their hours, working from home limits the quality and type of jobs women can take and, thus, nega- QUHS 2021 included a question about the reasons why tively affects their upward mobility and income (Amir et the respondent works from their own dwelling or home. al. 2018). Additionally, home-based work affords women In the case of women, 51 percent answered they were fewer opportunities for networking, knowledge sharing, “not allowed to leave,” followed by 21 percent who an- or learning from other peers in the same occupation/ swered it is “easier to work from home” and 17 percent industry, as well as fewer opportunities for improving who reported they “can’t leave home because [they need] the quality of their employment and their productivity to attend family.” Men’s reasons were very different. About growth, further depressing their earnings. 19 The survey item gave “regular workplace closed because of COVID-19” as an answer choice, but it did not register any observations. 21 Most women HBWs are own-account workers, followed by paid workers by piece. TABLE 12. EMPLOYMENT STATUS OF WOMEN HBWs Employment status Percentage (%) Regular paid employee with fixed wage 0.5 Casual paid employee 0.5 Paid worker by piece or work performed 20.2 Paid nonfamily apprentice 0.0 Employer 1.2 Own-account worker (nonagriculture) 71.5 Contributing family worker (nonagriculture) 6.3 Other 0.0 Total 100.0 Childcare responsibilities play an important role in children (who are a minority, as more than 90 percent of shaping women’s employment profile. The vast ma- households have children). Most of these women (60 per- jority of households in urban Quetta (85 percent) have cent) work as paid employees. The presence of children at least one child in the 0–14 age group, a sharp contrast increases the likelihood of women working as HBWs, to the share of households with an adult 65 or older (21 which could signal that the decision to have children and percent). The presence of children (ages 0–14) in the the expectation to work from home are driven by the household increases the likelihood that women work same underlying factors. These findings further reflect as own-account workers. Among working women living the primary role of women in caregiving and household in a household with children, the share of own-account (unpaid) work and help to explain why women spend on workers is 61.2 percent, whereas a quarter are paid em- average three hours a day less performing market work ployees. The trend reverses among women living with no than men (6.4 and 9.6 hours per day, respectively). BOX 3. LABOR PROFILE OF AFGHAN REFUGEE WOMEN IN QUETTA At present, an estimated 1.4 million registered Afghan refugees live in Pakistan, mainly in the KP and Baloch- istan Provinces and in urban/semiurban centers. According to QUHS 2021, the Afghan refugee population represents approximately 13.5 percent of the Quetta population and a lower share (11.8 percent) of Quetta’s working-age population. This is in line with what was previously observed for Afghan refugees in Peshawar. The QUHS shows a higher number of young dependents in Afghan refugee households, with an average of 4.7 children below age 15, compared with 3.9 in Pakistani households. This is not surprising since the Afghan refugee population in Quetta is somewhat younger than the Pakistani population (for both men and women), by an average of two years (21.5 versus 23.6 years). Table A.1 in appendix A shows that Afghan refugee women are more likely to be active in the labor force (and employed) than Pakistani women, particularly in the younger cohorts. While Afghan refugee women represent 12 percent of the female working-age population (for men, the share is 11.3 percent), their share among those in the labor force is almost double (21 percent) (for men, the share is 13.2 percent). In addition, the overall rate of FLFP is 28 percent for Afghan refugee women and 15 percent for Pakistani women (table B3.1).20 This result is in line with the lower level of welfare observed among Afghan refugee households in Quetta and, therefore, their relatively greater need to participate in the labor market. Analysis of educational attainment of working-age women reveals a significant education gap between Afghan refugees and hosts, with the former being most likely to be illiterate and having substantially lower educa- tional attainment. About 82 percent of Afghan refugee women have less than primary education, whereas the corresponding figure among Pakistani women is 49 percent. The share of Afghan refugee women with less than 20 The overall rate of male labor force participation is 84 percent for Afghan refugees and 71 percent for Pakistanis. 22 TABLE B3.1. LABOR MARKET INDICATORS FOR WOMEN 15–64 IN URBAN QUETTA, BY NATIONALITY Indicator Pakistani (%) Afghan refugee (%) All nationalities (%) LFP rate 14.5 27.6 16.1 Employment rate 14.0 27.2 15.6 Unemployment rate 0.5 0.4 0.5 LFP extended, including work for own consumption 16.1 28.2 17.6 primary education increases when considering the population of working-age women (86 percent, compared to 55 percent among the Pakistani female labor force). The lower human capital of the Afghan refugee female labor force in Quetta is reflected in its sectoral and occupational structure. Compared to Pakistani women, Afghan refugee female workers are almost exclusively employed in manufacturing (76.5 percent work in this sector, compared to 58.1 percent of their Pakistani peers). Female Afghan refugee workers are mostly employed as craft and trade-related workers. Furthermore, most of them are HBWs (93.2 versus 74.5 percent of Pakistani women), own-account workers (68.6 versus 52.8 percent of Pakistani women), or garment workers in the textile sector. The share of Afghan refugee women working as professionals is barely 1.5 percent, whereas among Pakistani women this figure is 16.7 percent. Businesses operated by Afghan refugee workers (regardless of their sex) tend to be on average smaller than those owned by Pakistani workers. Sources: Figures from Operational Data Portal, UNHCR, Geneva, Switzerland (accessed May 31, 2022), https://data.unhcr.org/ en/country/pak. Other data in this section are from Redaelli (2022) and the Quetta and Peshawar Urban Household Surveys. The earnings differential between men and women is fully explained by observables, highlighting prevailing social norms.  TABLE 13. OAXACA-BLINDER DECOMPOSITION OF GENDER GAP IN HOURLY EARNINGS (WITH HECKMAN CORRECTION FOR SELECTION) Indicator Coefficient Difference (%) Dependent variable: Log (hourly earnings)  Difference men-women 1.034*** (0.131)   Explained 0.894*** (0.149) 86 Unexplained 0.140 (0.153) 14 Observations 1,978   Note: Robust standard errors in parentheses. Hourly earnings are reported monthly earnings normalized by days per week and by hours per day usually worked at main job. A coefficient of positive sign indicates an increase in the wage gap. Explanatory variables include age, age-squared, education level, employment status, workplace, and occupation. Inclusion of sector (alone or with occupation) returns nonsignificant coefficients. Variables used for Heckman correction for selection include age, age-squared, education level, marital status, presence of children 0–14 in household and dummy for Afghan refugee populations. *p < 0.1  **p < 0.05  ***p < 0.01 Overall, women have a lower-quality job profile than icraft workers; only a minority perform more skilled jobs, men, which mostly explains the gender earnings gap. such as teachers or health professionals. These low-quality Working women in urban Quetta are mostly employed in jobs do not create incentives to increase their participation low-value-added activities that display a higher prevalence in market work. Not surprisingly, the decomposition of the of own-account, informal, home-based work. They work in earnings differential between men and women (table 14) the manufacturing industry mainly as garment and hand- is largely explained by observable demographic and job 23 characteristics—namely, age, educational attainment, em- ing hours, and their pay. Women’s predominantly domes- ployment status, work location, and occupation).21 tic roles and responsibilities toward their families appear to be a strong influence, even when they are employed. The results presented in this section indicate that the The need for reconciliation of house care and work for polarization of men’s and women’s working lives ex- pay is reflected by the characteristics of their jobs. While tends far beyond the decision to join the labor force. this is universally true, educated women are far more They also highlight that education makes a difference in likely to work outside the home and have careers that are working women’s experiences. Employed women are not more similar to their male counterparts. Women with low just a minority but a segregated one in terms of their oc- education, by contrast, are mostly engaged in informal, cupations, the restricted location of their jobs, their work- own-account home-based work, often in manufacturing. BOX 4. IMPACT OF THE COVID-19 PANDEMIC ON WOMEN’S PARTICIPATION IN THE WORKFORCE On March 21, 2020, as COVID-19 cases rose, Pakistan authorities imposed a lockdown. The first lockdown lasted until May 9, 2020. Thereafter, sporadic temporary lockdowns ensued. Baseline data from different COVID-19 surveys in Pakistan22 show that the pandemic led to severe household economic and food insecurity; job losses due to the economic lockdown, particularly in urban areas; and slowdowns/closures in business ac- tivity. Nonwage workers (own-account workers not in agriculture), daily/weekly wage workers, and youth and less educated workers were affected the most. Fortunately, as of August 2020, a V-shaped recovery process seemed to have begun, with the employment rate reaching close to prepandemic levels (Pakistan Bureau of Statistics 2020; World Bank 2021). In a recent study focusing on the gender effects of COVID-19 in Pakistan,23 Taş et al. (2021) concluded that the sectors where women are most likely to be employed, such as education and health, were the most severely affected. Further, the postpandemic recovery has been faster for males.24 As in many countries, the pandemic has led to a disproportionate increase in women’s unpaid care work in Pakistan, as well as increases in their reported rates of stress, anxiety, and exposure to violence. The QUHS includes a module on labor before the pandemic (February 2020) and during the lockdown period (March–July 2020), along with the labor situation at the time of the survey (collected between November 2020 and March 2021). In line with findings at the national level, the data show that in urban Quetta, 14.8 percent of women employed before the pandemic had lost their job by the time of the survey (compared to 1.8 percent of men). At the same time, 1.6 percent of nonworking women (versus 7.5 percent of nonworking men) took on a job, mostly in low-skilled occupations, such as craft or elementary workers, probably in response to income loss in the household. It is noteworthy that for both men and women, the change in LFP rates before and after the start of the COVID-19 pandemic is not statistically significant at any level (table B4.1). Possible explanations might be that social distancing policies, which had a profound impact in other countries, were implemented as micro lockdowns in Pakistan (as opposed to citywide lockdowns). Also, the type of jobs that women in Quetta have, which are mostly home-based and rarely contact-intensive, exposed them less to the adverse effects of lockdowns. More- over, female employment has been extremely low since the COVID-19 pandemic began. The data suggest that the only significant change in urban Quetta’s labor market after the pandemic began is an increase in male employment, which is consistent with a faster postpandemic recovery among men than among women. These findings should also be interpreted with caution because the number of observations with a valid (nonmissing) LFP status is significantly lower in the prepandemic labor module relative to labor at the time of the survey.25 21 The QUHS includes a module on earnings from an individual’s main job. Data show high nonreporting of earnings. Among respondents who indicated being employed in a paid job and reported a valid (nonmissing) sector and occupation, 35.8 percent are missing data on earnings. Analysis of the probability of nonreporting shows that nonreporting does not seem to be systematic. For instance, 33 percent of employed women and 36 percent of employed men are missing earnings; most missing earnings are found in manufacturing (where most women work) and in construction (which is a male-dominated sector). Only three respondents (women) indicated having zero earnings, which makes it difficult to distinguish between paid and unpaid workers. 22 The Pakistan Bureau of Statistics COVID-19 special survey was collected October–November 2020. The World Bank COVID-19 phone survey was conducted November 2020–April 2021. 23 The study used the administrative database of Pakistan’s largest online job platform and an online COVID-19 survey. 24 According to the authors, male-dominated sectors such as hotels, restaurants, food service, and transportation were also hit hard in terms of job losses. 25 While there were 210 missing observations on labor force status at the time of the survey (in a sample of 10,056 working-age individuals), the number of missing observations increases to 1,853 when the question refers to labor market status in February 2020. 24 TABLE B4.1. DISTRIBUTION OF EMPLOYED YOUTH (15–29) ACROSS ECONOMIC ACTIVITY, BY SEX Women Men Pre–COVID-19 Pre–COVID-19 pandemic At time of pandemic At time Indicator Difference Difference (February survey (February of survey 2020) 2020) LFP (%) 16.8 16.1 0.7 71.4 72.2 –0.8 Employment (%) 14.2 15.6 –1.4 66.3 69.8 3.6*** Unemployment (%) 2.5 0.5 2.1*** 5.1 2.4 2.7*** Out of labor force (%) 83.2 83.9 –0.7 28.6 27.8 0.8 Observations 4,417 4,733 3,786 5,113 Note: Observations refer to number of respondents with a valid status in the labor force (nonmissing observations). *p < 0.1  **p < 0.05  ***p < 0.01 A closer look at respondents who were employed before the pandemic but experienced job loss shows that female-dominated sectors and occupations were severely affected by the containment measures—namely, lockdowns and school closures. QUHS 2021 shows that women in professional activities, education, and hu- man health—which are also high-skilled sectors—were more likely to lose their jobs (table B4.2). These sectors concentrate female employment in Quetta, along with manufacturing. In the case of men, there is less varia- tion by sector in the share of men experiencing job loss. The highest share is found in the construction sector, where 3.8 percent of men lost their jobs. TABLE B4.2. JOB LOSSES AMONG WOMEN AFTER THE COVID-19 PANDEMIC Share of employed women in Share of women employed in the sector who experienced ISIC sector of activity the sector pre-COVID (%) job loss after the COVID-19 pandemic began (%) C - Manufacturing 70.8 3.0 E - Water supply; sewerage, waste management, and 0.9 0.0 remediation activities F - Construction 0.7 0.0 G - Wholesale and retail trade; repair of motor vehicles 0.6 0.0 and motorcycles H - Transportation and storage 3.9 3.2 M - Professional, scientific, and technical activities 5.4 26.9 P - Education 9.5 13.1 Q - Human health and social work activities 2.4 19.6 R - Arts, entertainment, and recreation 0.9 0.0 T - Activities of households as employers; undifferen- 1.3 2.8 tiated goods Analysis using the latest waves of the LFS for urban Balochistan (2017–18 and 2020–21; the 2017–18 round is representative at the province level only), gives a more representative picture of the situation before and after the pandemic. The comparison between the distribution of employed women (ages 15–64) across industries (excluding agriculture) before and after the pandemic supports the findings from the QUHS. According to the LFS rounds, there was an increase in the share of women working in manufacturing (from 17 percent in 2017 to 44 percent in 2020), suggesting that women who took on jobs did so mostly in manufacturing. At the same time, the shares of female employment in education and human health showed the greatest decline (5 and 12 percentage point decreases, respectively), suggesting that these two sectors were hardest hit by the pandemic. 25 Conclusions and Policy Options fields and help women gain market information (Field et al. 2016). Only 16 percent of working-age women in urban Quetta participate in the labor market, compared to Investing in girls and young women’s education 72 percent of men. Efforts aimed at increasing FLFP and skills-based training (with a gender focus) is will likely contribute to higher economic growth and an important precondition to increasing FLFP and poverty reduction. Along with experiencing low levels breaking the vicious cycle of low education and low of labor market participation, women employed in ur- employment. As demonstrated by a large body of evi- ban Quetta mostly work in low-value-added activities, dence and the Peshawar and Quetta household surveys, with a high prevalence of own-account, informal, and women are more likely to be involved in the labor force home-based work. At the same time, they perform jobs if they are more educated. Addressing both demand and in line with socially accepted occupations, likely a func- supply constraints that limit girls’ education remains a tion of how easily these jobs can be done from home. key priority. Similarly, a lack of marketable skills can Eberhard-Ruiz and Gutierrez (forthcoming) estimate discourage women from seeking jobs. Skills-based in- the potential job and GDP gains from closing the coun- terventions can improve income, empowerment, and try’s employment gap between men and women relative labor market outcomes for women through increased to peer countries with a similar level of development. business knowledge (such as financial planning, mar- According to the study, 7–19 million new jobs could be keting, and other business-related skills), improved life created, and the estimates for GDP gains range from 5 to skills (such as outlook on life, motivation, self-esteem, 23 percent, depending on the benchmark scenario.26 The and career aspirations), and greater decision-making remainder of this section discusses policy options aimed inside and outside the home (Chinen et al. 2017). In a at promoting greater FLFP in urban Pakistan.27 recent systematic review of skills-based interventions in South Asia, researchers concluded that interventions Working women are mainly employed in the manu- sensitized to the prevailing social and logistical barriers facturing industry as garment and handicraft work- for women—household work, family obligations, child- ers. Only a minority (the highly educated) perform care, and gendered norms against travel—had larger more skilled jobs such as teachers or health profes- impacts.28 Examples of these interventions are programs sionals. Representation in other nontraditional sec- that provide monetary incentives, childcare services, tors is very low. Women’s career fields are perceived mentoring for life skills, organized training sessions in to align with traditional gender roles. Employment in villages and close to women’s homes, training delivered other fields, especially nontraditional sectors, can be through local providers, and advertising campaigns especially challenging for women. Rigorous evidence that employ social mobilizers (Zahra, Javed, and Munoz on what works in assisting women to cross over into Boudet 2021). male-­dominated nontraditional sectors is scarce but promising. For instance, informational nudges—par- Social norms seem to be the most powerful factor ticularly those that emphasize the differential earnings in determining women’s interactions with the pub- between female- and male-dominated occupations—can lic sphere and workforce. Household attitudes and encourage women to enroll in training programs to en- behavior and social norms play an important role ter male-dominated trades (Hicks et al. 2011). Providing in determining whether, when, and how women can information on sector-­ specific profitability could also work for pay. In this context, steady long-term policy change beliefs about profitability. Schools could offer efforts are needed to influence social norms toward en- information through career guidance, informational ses- couraging women’s empowerment. Research indicates sions accompanying skills training programs, or edutain- that possible interventions to influence norms include ment (Bjorvatn et al. 2020). Similarly, early exposure to strategic use of positive messaging about strong female male role models has been shown to improve the likeli- role models. Furthermore, global evidence suggests that hood of women crossing over into male-dominated sec- engaging men is crucial in changing norms surrounding tors and occupations (Alibhai et al. 2017; Campos et al. women’s economic activities. For example, men can act as 2015). Exposure to a successful role model may pro- “gatekeepers” for women by providing access to capital, vide information about the returns in male-dominated information, and networks. 26 Eberhard-Ruiz and Gutierrez (forthcoming) chose Bangladesh as a benchmark country because it has managed to substantially increase women’s employment in recent years while sharing similar cultural and labor market characteristics with Pakistan. 27 For a comprehensive review, see World Bank (2021). 28 Women targeted by these training programs, either as intended beneficiaries or as a subsample, generally have low levels of education, are poor or from marginalized backgrounds, work in low-skilled occupations, or are active in the informal sector. 26 The COVID-19 pandemic could worsen already un- Information and communication technology (ICT) favorable prospects for women's labor participation has the potential to provide women with increased and employment, so the gendered effects of the pan- access to better markets while allowing them to demic should be considered in recovery efforts. As circumvent obstacles related to mobility and social businesses close temporarily or permanently, jobs have norms. The QUHS shows that 55 percent of working-age disappeared for both men and women. However, the women in urban Quetta do not have internet access and sectors where women are more likely to be employed, 13 percent do not know about the internet (versus 32 such as education and health, have been most severely percent and 2 percent, respectively, among men). These affected. Likewise, the COVID-19 pandemic has led to a shares are much higher among women with low educa- disproportionate increase in women’s unpaid care work, tion. Emerging ICT jobs could provide new opportunities which, if prolonged, will make women more likely to quit for women, especially women living in urban areas. In the labor market altogether. In addition, the fall in house- Pakistan, freelancers in ICT generally work 34 hours per hold income and the rise in unpaid work are likely fac- week, with flexible hours, and a gender gap in earnings tors creating higher stress and anxiety among women, as does not seem to exist. Some records even suggest that well as increased exposure to domestic violence. Recov- female freelancers in Pakistan earn more than their male ery policies must incorporate elements aimed at restor- counterparts (Cho and Majoka 2020). ICT can help con- ing household dynamics and incentives that encourage nect women and men with the labor market in different women to work, such as childcare support services and ways—for example, by expanding their skills, expanding targeted social safety nets for informal and home-based their options for the job search, providing access to on- female workers who do not benefit from social protection line and remote learning trainings, and providing access coverage. Women who have lost their jobs in the hard- to e-commerce platforms. ICT could also boost female est-hit sectors can serve as frontline workers to roll out employment by enabling women to telework from their public COVID-19 response programs for contact tracing, home in more productive jobs compatible with prefer- testing, vaccination, and remote learning. ences around home-based work and care responsibili- ties—especially as these preferences have become more Most employed women (78.6 percent) in urban pronounced and unavoidable due to the COVID-19 pan- Quetta are HBWs who are largely employed in in- demic. Measures to promote women’s ICT jobs include formal jobs of low upward mobility. For women, creating workspaces for women with internet connec- working from home is an alternative used to work tions, networks, and mentors. around existing social norms. Effective implemen- tation of recent legislation to recognize the status Additional key areas for action to support FLFP of HBWs can improve women’s economic participa- include investments in infrastructure to (a) facili- tion in the province.29 In April 2022, the Balochistan tate transportation and safety of public spaces and assembly passed the Home-Based Workers Bill, aimed (b) adapt workplace environments to the needs of at protecting the rights of women and other workers women. Security concerns and gender norms that in- involved in home-based work in the province. The law hibit mobility of women directly stunt women’s LFP. In recognizes informal HBWs (the majority of whom are this context, affordable, safe public transport systems women) as formal workers and entitles them access to responding to the specific needs of women and sup- social security benefits and a minimum wage. This rec- porting their participation in the workforce are critical ognition will enhance HBWs’ access to decent wages (ADB 2016). Relevant measures include (but are not and social security benefits. Furthermore, this will limited to) pedestrian walkways that are adequately also improve measurement of the overall FLFP rate in lit and easy access to reporting incidents of harass- Balochistan, as research shows the number of HBWs ment and swift resolution of these issues with support is underestimated in Pakistan and other parts of the from law enforcement. Equally important is investing world due to definitional issues in standard LFSs and in workplace environments where there is access to use of proxy respondents for women.30 Effective imple- facilities such as childcare, dedicated transport, and mentation of provincial and federal HBW laws will aid separate rest areas for women, as all of these are con- economic participation of women who are constrained ducive to women’s work. There is compelling evidence to work from home due to social norms and other care of the positive impact of childcare availability on wom- work responsibilities. en’s employment, including in low- and middle-income 29 The World Bank–financed development policy credit program SHIFT 1&2 recently supported the government in passing 11 laws in Sindh, KP, and Balochistan. HBW laws were passed in KP and Balochistan to recognize informal HBWs as formal workers and support HBWs’ access to decent wages and social security benefits. 30 The World Bank is undertaking mixed-methods research on HBWs in KP and Balochistan to support the provincial governments in implementing HBW laws. This will also entail supporting the government in developing rules and aiding HBW registration. 27 countries (see the review of evidence in Devercelli The analysis presented here suggests that existing and Beaton-Day 2020). In the context of Pakistan, surveys typically underestimate women’s work, so state-subsidized childcare programs, public-private different approaches are needed to better measure partnerships for day care facilities, and communal all of women’s economic contributions inside and childcare facilities are all feasible options to support outside the home. As demonstrated by the Peshawar working women and ease the burden of household and and Quetta surveys, techniques to measure FLFP should market work. Provisions for separate toilets and rest/ be modified to capture more robust data on productive prayer areas for women are also crucial to establish activities by expanding estimates of economic contribu- comfortable workplaces for women. tions. Future labor surveys could also be adapted to the approach of collecting data from all household members, going beyond proxy respondents for women. For additional information, please contact Uzma Quresh, uquresh@worldbank.org, and Maria Beatriz Orlando, morlando@worldbank.org, co–task team leaders of the Pakistan Gender and Social Inclusion Platform, and Moritz Meyer, mmeyer3@worldbank.org, task team leader of the Pakistan Poverty and Equity Program. 28 APPENDIX A: QUHS SAMPLING METHODOLOGY AND DESCRIPTIVE STATISTICS The Quetta Urban Household Survey (QUHS) was planned with the objective of delivering a representative sample of the city of Quetta (Metropolitan Corporation) as a whole and of Afghan refugees living in the city. To this end, sampling was conducted in two stages. First Stage: Selecting Primary Sampling Units for Listing The Pakistan Bureau of Statistics has demarcated Quetta Metropolitan Corporation into 508 urban enumeration areas, or primary sampling units (PSUs). Using sampling proportional to size, 220 PSUs were selected for listing. Second Stage: Selection of Households for Interview In the second stage, 2,020 households were selected—11 in each of the 220 PSUs selected in the first stage. The sample frame for the second stage was a full list of all structures (both dwellings and nonresidential units) and households in the 220 PSUs. The household listing operation identified 35,913 Pakistani and 3,745 Afghan refugee households in total. The latter were very concentrated in certain PSUs. The number of Afghan refugees and Pakistani households to visit in each PSU was defined as follows: If the PSU had less than 10 Afghan refugee households, all were visited; otherwise nine were selected. Then, taking into account the number of Afghan refugee households, as many Pakistani households as needed to visit 11 households in total were selected. If there were not enough Pakistani households in the PSU, additional Afghan refugee households were selected. The sample of households, or the target sample, was selected by systematic equal-probability sampling from the list of all households in the PSU, sorted by structure number,1 independently for each nationality. The households of each nationality not selected in the target sample were assigned a serial mobilization number to indicate the order in which they could be used to substitute nonrespondent households of the same nationality from the target sample. Households with mobilization numbers 1 or 2 are referred to as the reserve sample. Selection Probabilities and Sampling Weights The probability of interviewing a household of nationality (Pakistani or Afghan refugee household, determined at time of listing) in block is the product of (a) selecting the PSU for listing in the first stage and (b) selecting the household from the listing data in the third stage. These probabilities are given by equations 1 and 2: , (1) , (2) where  is the total number of blocks selected for listing;  is the number of households in the block, per the 2017 Census;  is the total number of households in Quetta Metropolitan Corporation, as per the 2017 Census;  is the fraction of households in the block for which a nationality (as defined at listing) was reported;  is the number of households of the nationality (as defined at listing) interviewed in the block; and  is the total number of households of the nationality (as defined at listing) listed in the block. To obtain unbiased estimates from the survey, the data reported from a household must be affected by a sampling weight , equal to the inverse of its selection probability ( ). 1 Households were identified using a structure number written by fieldworkers near the door of the dwelling and recorded in the listing data sets. A substructure number was assigned to avoid ambiguity in the few cases where the same structure number was mistakenly reported for more than one household. 29 TABLE A.1. DESCRIPTIVE STATISTICS AT THE HOUSEHOLD LEVEL Statistic All households Non–Afghan refugee households Household size 7.7 7.6 Number of children (ages 0–14) living at home 3.0 2.9 Dependency ratio (expressed as % of adults 15–64) Child 90.5 86.4 Older adult 8.2 8.2 Total 98.7 94.6 Nuclear households (%) 56.5 56.0 Female-headed households (%) 3.4 3.7 Household composition (%) Single, no children 3.8 3.9 Single with children 6.7 6.9 Couple without children 4.7 4.5 Couple with 1–3 children, no others at home 29.4 30.2 Couple with 4+ children, no others at home 38.2 37.4 Couple with children and other family members at home 17.2 17.1 Note: Child dependency ratio refers to number of children ages 0–14 per adult ages 15–64. Older adult dependency ratio refers to number of seniors over 65 per adult ages 15–64. Total dependency ratio refers to children ages 0–14 and older adults over 65 per adult ages 15–64. Nuclear households refer to couples and children only. Households that consist of a single adult, a single adult with children, or a couple without children may have other relatives as well. 30 APPENDIX B: REGRESSION RESULTS TABLE B.1. AVERAGE MARGINAL EFFECTS FROM PROBIT PARTICIPATION EQUATIONS   (1) (2) (3) (4) Variable dy/dx SE dy/dx SE dy/dx SE dy/dx SE Dep. var: LFP = 1 at time of survey Age 0.024*** (0.005) 0.023*** (0.005) 0.020*** (0.006) 0.015*** (0.005) Age-squared –0.000*** (0.000) –0.000*** (0.000) –0.000*** (0.000) –0.000*** (0.000) Afghan refugee = 1 0.120*** (0.026) 0.105*** (0.028) 0.008 (0.029) 0.007 (0.028) Married = 1 –0.060*** (0.022) –0.057** (0.022) –0.020 (0.025) –0.006 (0.022) Own education (completed grades) Reference: below primary Primary –0.062** (0.025) –0.059** (0.027) –0.047* (0.025) Secondary –0.055** (0.026) –0.044 (0.031) –0.043 (0.027) Tertiary or more     0.016 (0.028) 0.021 (0.039) 0.026 (0.035) Education of household head (completed grades)  Reference: below primary Primary –0.064** (0.030) –0.078** (0.030) Secondary –0.088*** (0.029) –0.094*** (0.029) Tertiary or more –0.013 (0.046) –0.033 (0.038) Food adequacy = 1 –0.021 (0.025) –0.025 (0.023) Asset index –0.027*** (0.006) –0.025*** (0.006) Access to cell phone = 1         0.020 (0.023) 0.024 (0.019) Household composition (number of members in age/ sex group)  0–5 0.010 (0.008) 6–14 –0.008 (0.006) 15–24 0.007 (0.006) females 25–44 0.029 (0.022) males 25–44 –0.039*** (0.014) females 45–64 –0.008 (0.023) males 45–64 –0.040* (0.022) 65+             0.034 (0.023) Feels safe outside own neighborhood             –0.003 (0.022) Purdah = 1                 Own belief: in favor of female work                  Own decision: work inside                 Own decision: work outside Own decision: community activity Own decision: political activity Own decision: shopping Own decision: education Own decision: marriage Own decision: health Pseudo-R2 0.0289   0.0351   0.0652   0.0804   F statistic 11.50   8.533   6.307   4.420   Observations 4,711   4,643   3,730   3,730   Note: Standard errors in parentheses. The sample refers to working-age adults (15–64) able to work (not in school or ill/disabled). Survey weight applied. Food adequacy takes value of 1 if the male primary respondent considers the household’s food consumption adequate or better. Asset index estimates follow a methodology similar to that of DHS. The minimum value is –4.78, and the maximum is 6.91. Access to cell phone includes both owning a phone and accessing one through a spouse, brother, or friend. The own-decision dummy takes the value of 1 if a woman is included in the decision-making, whether she makes the decision alone or with partner. *p < 0.1  **p < 0.05  ***p < 0.01 31 TABLE B.1. AVERAGE MARGINAL EFFECTS FROM PROBIT PARTICIPATION EQUATIONS (CONTINUED) (5) (6) (7) (8) Variable dy/dx SE dy/dx SE dy/dx SE dy/dx SE Dep. var: LFP = 1 at time of survey Age 0.015*** (0.005) 0.015*** (0.005) 0.016*** (0.005) 0.015*** (0.005) Age-squared –0.000*** (0.000) –0.000*** (0.000) –0.000*** (0.000) –0.000*** (0.000) Afghan refugee = 1 0.002 (0.028) 0.007 (0.027) 0.016 (0.028) 0.009 (0.028) Married = 1 –0.012 (0.022) –0.014 (0.022) –0.012 (0.023) –0.011 (0.024) Own education (completed grades) Reference: below primary Primary –0.046* (0.025) –0.047* (0.025) –0.058** (0.026) –0.052** (0.026) Secondary –0.045* (0.026) –0.046* (0.026) –0.055** (0.027) –0.064** (0.028) Tertiary or more 0.026 (0.035) 0.025 (0.035) 0.014 (0.034) 0.017 (0.034) Education of household head (completed grades)  Reference: below primary Primary –0.076** (0.030) –0.072** (0.030) –0.067** (0.030) –0.063** (0.031) Secondary –0.096*** (0.030) –0.094*** (0.030) –0.092*** (0.030) –0.100*** (0.031) Tertiary or more –0.030 (0.038) –0.029 (0.038) –0.029 (0.037) –0.030 (0.038) Food adequacy = 1 –0.030 (0.023) –0.030 (0.022) –0.023 (0.023) –0.029 (0.023) Asset index –0.026*** (0.006) –0.025*** (0.006) –0.024*** (0.006) –0.024*** (0.007) Access to cell phone = 1 0.021 (0.019) 0.023 (0.020) 0.021 (0.020) 0.014 (0.021) Household composition (number of members in age/ sex group)  0–5 0.010 (0.008) 0.011 (0.008) 0.012 (0.008) 0.012 (0.008) 6–14 –0.007 (0.006) –0.008 (0.006) –0.007 (0.006) –0.007 (0.006) 15–24 0.007 (0.006) 0.007 (0.006) 0.008 (0.006) 0.010 (0.006) females 25–44 0.029 (0.023) 0.028 (0.023) 0.025 (0.022) 0.027 (0.022) males 25–44 –0.039*** (0.015) –0.039*** (0.015) –0.040*** (0.015) –0.038** (0.015) females 45–64 –0.007 (0.023) –0.007 (0.023) –0.006 (0.023) –0.001 (0.024) males 45–64 –0.043* (0.023) –0.041* (0.022) –0.044** (0.022) –0.039* (0.022) 65+ 0.033 (0.023) 0.034 (0.023) 0.031 (0.023) 0.027 (0.023) Feels safe outside own neighborhood –0.006 (0.022) –0.006 (0.022) –0.012 (0.022) Purdah = 1     –0.113** (0.056) –0.104* (0.055) –0.100* (0.056) Own belief: in favor of female work      0.157*** (0.043) 0.151*** (0.046) Own decision: work inside             0.023 (0.058) Own decision: work outside 0.015 (0.070) Own decision: community activity 0.051 (0.072) Own decision: political activity –0.021 (0.052) Own decision: shopping –0.014 (0.023) Own decision: education 0.007 (0.028) Own decision: marriage –0.036 (0.043) Own decision: health –0.007 (0.028) Pseudo-R2 0.0814   0.0834   0.0964   0.0971   F statistic 4.163   3.986   4.996   3.589   Observations 3,676   3,665   3,636   3,499   Note: Standard errors in parentheses. The sample refers to working-age adults (15–64) able to work (not in school or ill/disabled). Survey weight applied. Food adequacy takes value of 1 if the male primary respondent considers the household’s food consumption adequate or better. Asset index estimates follow a methodology similar to that of DHS. The minimum value is –4.78, and the maximum is 6.91. Access to cell phone includes both owning a phone and accessing one through a spouse, brother, or friend. The own-decision dummy takes the value of 1 if a woman is included in the decision-making, whether she makes the decision alone or with partner. *p < 0.1  **p < 0.05  ***p < 0.01 32 REFERENCES Bardasi, E., K. Beegle, A. Dillon, and P. Serneels. 2011. “Do Labor Statistics Depend on How and to Whom ADB (Asian Development Bank). 2016. “Policy Brief on the Questions Are Asked? Results from a Survey Female Labor Force Participation in Pakistan.” Experiment in Tanzania.” World Bank Economic Review 25 (3): 418–47. Adeel, M., A. G. O. Yeh, and F. Zhang. 2013. “Gender, Mobility and Travel Behavior in Pakistan: Benes, E., and K. Walsh. 2018. Measuring Employment in Analysis of 2007 Time Use Survey.” MPRA Labour Force Surveys: Main Findings from the Paper 55474, University Library of Munich, ILO LFS Pilot Studies. Statistical Methodology Germany. https://mpra.ub.uni-muenchen. Series 4. Geneva: ILO. de/55474/1/MPRA_paper_55474.pdf. Bjorvatn, K., A. W. Cappelen, L. Helgesson Sekei, E. Adeel, M., and A. G. O. Yeh. 2018. “Gendered Immobility: Sørensen, and B. Tungodden. 2020. “Teaching Influence of Social Roles and Local Context on through Television: Experimental Evidence Mobility Decisions in Pakistan.” Transportation on Entrepreneurship Education in Tanzania.” Planning and Technology 41 (6): 660–78. Management Science 66 (6): 2308–25. http://eprints.lse.ac.uk/88340/1/Adeel_ Gendered%20Immobility_Accepted.pdf. Campos, F., M. P. Goldstein, L. McGorman, A. M. Muñoz- Boudet, and O. Pimhidzai. 2015. “Breaking Akhtar, S. 2020. “Home-Based Workers in Pakistan: the Metal Ceiling: Female Entrepreneurs A Statistical Profile.” Statistical Brief No. 26, Who Succeed in Male-Dominated Sectors.” WIEGO, Manchester, UK. https://www.wiego. Policy Research Working Paper 7503, org/sites/default/files/publications/file/ World Bank, Washington, DC. https:// WIEGO_Statistical_ documents1.worldbank.org/curated/ Brief_N26_Pakistan_final.pdf. en/753711467997247654/pdf/WPS7503.pdf. Alibhai, S., N. Buehren, S. Papineni, and R. S. Pierotti. Chinen, M., T. de Hoop, L. Alcázar, M. Balarin, and J. 2017. “Crossovers, Female Entrepreneurs Who Sennett. 2017. “Vocational and Business Enter Male Sectors: Evidence from Ethiopia.” Training to Improve Women’s Labour Market Policy Research Working Paper 8065, World Outcomes in Low- and Middle-Income Bank, Washington, DC. Countries: A Systematic Review.” Campbell Systematic Review 13 (1): 1–195. doi:10.4073/ Ambler, K., S. Herskowitz, and M. Maredia. 2021. “Are csr.2017.16. We Done Yet? Response Fatigue and Rural Livelihoods.” Journal of Development Economics Cho, Y., and Z. Majoka. 2020. Pakistan Jobs Diagnostic: 153 (November): 102736. doi:/10.1016/j. Promoting Access to Quality Jobs for All. Jobs jdeveco.2021.102736. Series No. 20. Washington, DC: World Bank. https://openknowledge.worldbank.org/ Amir, S., A. Kotikula, R. Pande, L. L. Y. Bossavie, and U. handle/10986/33317. Khadka. 2018. Female Labor Force Participation in Pakistan: What Do We Know? Washington, Desiere, S., and V. Costa. 2019. “Employment Data DC: World Bank. https://openknowledge. in Household Surveys: Taking Stock, worldbank.org/handle/10986/30197. Looking Ahead.” Policy Research Working Paper 8882, World Bank, Washington, DC. Anjum, G., T. Kessler, and M. Aziz. 2019. “Cross-Cultural https://openknowledge.worldbank.org/ Exploration of Honor: Perception of Honor handle/10986/31872. in Germany, Pakistan, and South Korea.” Psychological Studies 64 (2): 147–60. https:// Devercelli, A. E., and F. Beaton-Day. 2020. Better Jobs ir.iba.edu.pk/faculty-research-articles/190. and Brighter Futures: Investing in Childcare to Build Human Capital. Washington, DC: World Asadullah, M. N., and Z. Wahhaj. 2016. “Missing from the Bank. https://openknowledge.worldbank.org/ Market: Purdah Norms and Women’s Paid Work handle/10986/35062. Participation in Bangladesh.” IZA Discussion Paper No. 10463, Institute of Labor Economics, Dillon, A., E. Bardasi, K. Beegle, and P. Serneels. 2012. Bonn, Germany. https://ftp.iza.org/dp10463.pdf. “Explaining Variation in Child Labor Statistics.” 33 Journal of Development Economics 98 (1): Experiences-of-School-Closures-Insights-from- 136–47. doi:10.1016/j.jdeveco.2011.06.002. Interviews-with-Girls-and-Mothers-in-Punjab- Pakistan. Eberhard-Ruiz, A., and V. Gutierrez. Forthcoming. Assessing the Economic Gains from Closing Mancini, G. 2021. Women in the Workforce in Pakistan’s Female Employment Gap: Peshawar. Washington, DC: World Bank. Background Note Prepared for Pakistan Country http://documents.worldbank.org/ Economic Memorandum. Washington, DC: curated/en/099055002222210536/ World Bank. P1748030cd0f5b00f08d6d01018b077bdd7. Field, E., and K. Vyborny. 2015. “Female Labor Force National Institute of Population Studies and ICF. 2019. Participation in Asia: Pakistan Country Study.” Pakistan Demographic and Health Survey Unpublished manuscript. 2017–18. Islamabad, Pakistan: NIPS; Rockville, Maryland: ICF. https://dhsprogram.com/pubs/ Field, E., S. Jayachandran, R. Pande, and N. Rigol. 2016. pdf/FR354/FR354.pdf. “Friendship at Work: Can Peer Effects Catalyze Female Entrepreneurship?” American Economic Pakistan Bureau of Statistics. 2020. “Special Survey for Journal: Economic Policy 8 (2): 125–53. Evaluating Socio-economic Impact of COVID-19 doi:10.1257/pol.20140215. on Wellbeing of People.” Hicks, J. H., M. R. Kremer, I. Mbiti, and E. Miguel. 2011. Paterson, R. 2008. “Women’s Empowerment in “Vocational Education Voucher Delivery Challenging Environments: A Case Study and Labor Market Returns: A Randomized from Balochistan.” Development in Practice Evaluation among Kenyan Youth.” Working 18 (3): 333–44. http://www.jstor.org/ Paper, Abdul Latif Jameel Poverty Action stable/27751928. Lab, Massachusetts Institute of Technology, Cambridge, MA. Redaelli, S. 2022. Profile of Afghan Nationals in Pakistan Urban Centers. Washington, DC: World Bank. ILO (International Labour Organization). 2014. Engaging Men in Women’s Economic Rutkowski, M. 2020. “How Social Protection Can Help Empowerment and Entrepreneurship Countries Cope with COVID-19.” Voices (blog), Development Intervention. Geneva: Women’s April 15, 2020. https://blogs.worldbank. Entrepreneurship Development Programme. org/voices/how-social-protection-can-help- countries-cope-covid-19. ILO (International Labour Organization). 2021. Working from Home: From Invisibility to Decent Work. Sanauddin, N. 2015. “Proverbs and Patriarchy: Analysis Geneva: ILO. https://www.ilo.org/wcmsp5/ of Linguistic Sexism and Gender Relations groups/public/---ed_protect/---protrav/--- among the Pashtuns of Pakistan.” PhD thesis, travail/documents/publication/wcms_765806. University of Glasgow. http://theses.gla.ac.uk/ pdf. id/eprint/6243. International Conference of Labour Statisticians. 2013. Tanaka, S., and M. Muzones. 2016. “Female Labor Force “Resolution Concerning Statistics of Work, Participation in Asia: Key Trends, Constraints, Employment, and Labor Underutilization.” and Opportunities.” ADB Briefs, October 2016. https://www.ilo.org/wcmsp5/groups/ https://www.adb.org/sites/default/files/ public/---dgreports/---stat/documents/ publication/209666/female-labor-force- normativeinstrument/wcms_230304.pdf. participation-asia.pdf. Malik, R., N. Zahra, A. Tahir, and K. M. Geven. 2022. Taş, E. O., T. Ahmed, N. Matsuda, and S. Nomura. 2021. “Girls’ Lived Experiences of School Closures: “Impacts of COVID-19 on Labor Markets and Insights from Interviews with Girls and Household Well-Being in Pakistan: Evidence Mothers in Punjab, Pakistan.” South Asia Gender from an Online Job Platform.” South Asia Gender Innovation Lab Policy Brief, January 2022. Innovation Lab Policy Brief, February 2021. http://documents.worldbank.org/curated/ http://documents.worldbank.org/curated/ en/920331642085666052/Girls-Lived- en/366361617082088695/Impacts-of-COVID- 34 19-on-Labor-Markets-and-Household-Well- Aspirations-Experiences-and-Challenges- Being-in-Pakistan-Evidence-From-an-Online- for-Urban-Educated-Pakistani-Women- Job-Platform. Discussions-in-Four-Metro-Cities.pdf. UN Women. 2019. Families in a Changing World: World Bank. 2021. “Pakistan COVID-19 Phone Survey.” Progress of the World’s Women 2019–2020. New York: UN Women. http://www.unwomen.org/ World Bank. 2022. Reshaping Norms: A New Way en/digital-library/progress-of-the-worlds- Forward. Washington, DC: World Bank. women. https://openknowledge.worldbank.org/ handle/10986/37121. USAID (United States Agency for International Development). 2012. Women’s Economic Zahra, N., A. Javed, and A. M. Munoz Boudet. 2021. Empowerment: Balochistan. Washington, DC: “What Do We Know About Interventions to USAID. https://europa.eu/capacity4dev/ Increase Women’s Economic Participation file/79001/download?token=-5zf_14k. and Empowerment in South Asia?” Systematic Review on Women’s Economic World Bank. 2019. Labor Force Aspirations, Experiences Empowerment, June 2021. http:// and Challenges for Urban, Educated Pakistani documents.worldbank.org/curated/ Women: Discussions in Four Metro Cities. en/235351625080690161/What-Do-We- Washington, DC: World Bank. https:// Know-About-Interventions-to-Increase- documents1.worldbank.org/curated/ Women-s-Economic-Participation-and- en/190071611611113613/pdf/Labor-Force- Empowerment-in-South-Asia. 35