Policy Research Working Paper 10810 Women’s Labor Force Participation in Nepal An Exploration of The Role of Social Norms Jumana Alaref Aishwarya Patil Tasmia Rahman Ana Maria Munoz Boudet Jasmine Rajbhandary Social Protection and Jobs Global Practice June 2024 Policy Research Working Paper 10810 Abstract Whether and the extent to which social norms matter for behaviors. Overall, the study finds that personal beliefs and women’s labor force participation has been shown to vary social expectations are generally not very restrictive among by context. This paper presents rigorous evidence on how respondents, and that there are limited linkages between these relationships hold in the case of Nepal, where female social norms and women’s work outcomes. However, the labor force participation remains among the lowest in the study also shows that norms matter for selected subgroups world. Using a representative survey covering four provinces and under certain circumstances that are related to the in Nepal, data were collected from 2,000 married Nepali woman’s role as a mother and in the household as well as women and men on their own beliefs about norms-related to her job characteristics. The findings indicate that relaxing behaviors, their expectations of how common it is for norms in those specific circumstances can help to promote others in their social group to engage in those behaviors, women’s labor force participation in Nepal. and the expected social consequences surrounding those This paper is a product of the Social Protection and Jobs Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at jalaref@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Women’s Labor Force Participation in Nepal: An Exploration of The Role of Social Norms Jumana Alaref, Aishwarya Patil, Tasmia Rahman, Ana Maria Munoz Boudet, Jasmine Rajbhandary1 JEL Codes: C93, D91, J16, J21 Keywords: social norms, female labor force participation, South Asia, Nepal 1 Social Protection & Jobs Global Practice, Mind, Behavior, and Development (eMBeD) team, and the Gender Group, World Bank. Acknowledgments: We thank Solutions Consultant (P.) Ltd team in Nepal, led by Suraj Pradhan and Dipa Shrestha, for their diligent efforts in implementing the survey that underlies this work. From World Bank management, the team is grateful to Nicole Klingen, Faris Hadad-Zervos, Cem Mete, Stefano Paternostro, Lada Strelkova, and Ayesha Vawda for their guidance and support. Iman Sen, Maurizio Bussolo, Nele Warrinnier, Sarah Elizabeth Haddock, and the American Institute Research (AIR) team headed by Thomas de Hoop provided advice. Abhilasha Sahay, Ioana Botea, and the Social Sustainability and Inclusion South Asia Team provided valuable feedback at the peer review stage. We appreciate the feedback received from participants from academia and civil society in Nepal during a virtual consultation held at the study design stage on April 6, 2023. The study benefited from the generous funding of the World Bank’s Rapid Social Response Program (RSR), which is gratefully acknowledged. All errors remain our own. The views and findings expressed in this paper are those of the authors alone and do not necessarily represent those of the World Bank or its Executive Directors. INTRODUCTION Female labor force participation (FLFP) in South Asia has been a focus of policy concerns for the past decade. The average participation levels in the region remain among the lowest in the world and have stayed at around 30 percent since the early 2000s.2 Nepal is no exception. Labor force participation for Nepali women aged 15- 64 was only 26.3 percent in 2017-2018, compared to 53.8 percent among working-age Nepali men.3 The drivers behind these low levels of economic participation have included the stage of economic growth, following Goldin’s (1995) U-shaped relationship between female employment and the transition out of agriculture to services; the structure of a country’s economy (Klasen, 2019); women’s human capital and returns to education (Peet et al., 2015; Chicoine, 2021); access to services such as childcare (Halim, Perova and Reynolds, 2023a); and safe transport (Field and Vyborny, 2022; Buchmann et al., 2023). In the case of South Asia, social norms are also frequently discussed as a factor preventing women’s participation in the labor market (Jayachandran 2021; Bussolo et al., 2024), whether by themselves or in combination with the more traditional supply and demand-side barriers discussed in the literature (for example, Deshpande and Kabeer, 2024; Deshpande and Singh, 2021). In Nepal, a high share of women outside the labor force are engaged in subsistence farming for household consumption as their main economic activity, owing to Nepal being a highly rural economy. And while the economy is shifting away from agriculture, and jobs created over the last decade have been a mix of low- and high-productivity in industry and service sectors, men have benefitted disproportionately. Of the 3.8 million new wage jobs added to the economy by 2018, men took up over two-thirds of those jobs (Ruppert Bulmer, 2020), including jobs in manufacturing, finance, and business services sectors. These outcomes cannot be explained by Nepali women’s education levels – progress in educational attainment of girls is sizable and women enjoy high returns to education (Ruppert Bulmer et al., 2020).4 Young women, particularly married ones, have low attachment to the labor market. Nearly half (47 percent) of them are not in employment, education, or training (NEET), vs. 22 percent of young men.5 Over a third of young married women aged 20 and 24 report to not have done any work other than housework in the 12 months before the 2022 Nepal Demographics and Health Survey (DHS), as compared to three percent of young married men within the same age group. Social norms related to the unequal gender roles in families may compel women to shoulder a disproportionate burden of household work, thereby reducing their availability for wage employment and restricting them to jobs with lower employment potential than men. Women are more likely than men to be in unpaid farm work (6 million women compared to 2.8 million men). They also work, on average, 9 hours per week less than men do, which partly explains a gender gap in mean monthly earnings of almost 6,000 Nepali rupees (NPR) in favor of men (Ruppert Bulmer et al., 2020). 2 World Development Indicators (WDI), latest year available. Modeled ILO LFP. The modeled ILO comparable number for FLFP in Nepal is 29 percent in 2016, an increase from 22 percent in 2003. 3 Source: Nepal Labor Force Survey 2017/2018. The labor force participation rate is a measure of the proportion of a country's working-age population (individuals 15 years and older) that engages actively in the labor market, either by working or looking for work. Following the 19th International Conference of Labour Statisticians in 2013 – the new standards that Nepal adopted for the Labor Force Survey in 2017/18 – labor force participation includes work performed for pay, profit, or family income, and excludes subsistence work (production for own final use), whether in agriculture or in non-agriculture family or household enterprises. In past publications (e.g., Ruppert Bulmer et al., 2020), women’s labor force participation was estimated at 75 percent using the 2017/2018 Labor Force Survey as subsistence activities were counted in the definition of work. 4 Over the last three decades, women achieved strong education gains. Average years of schooling rose from 2 years for females born in 1960 to over 9 years for those born after 1990. A woman with a college degree, on average, earns double what a woman with incomplete primary earns, controlling for sector of work and other factors (Ruppert Bulmer et al. 2020). 5 Nepal Labor Force Survey 2017/2018. 2 This paper contributes to the evidence on female labor force participation in Nepal by exploring the role of social norms in explaining women’s decision to work for pay outside the home in the non-subsistence economy, where their share remains low and where normative influences also tend to be more prominent barriers for women, as compared to the subsistence economy. Social norms are defined as informal rules about acceptable or appropriate behaviors in a given context (Muñoz Boudet et al., 2023). When social norms reinforce expectations that men and women will occupy distinct and unequal social roles, they reproduce gender inequality in access to resources and opportunities. While there has been research on how to improve women’s economic participation, the overall literature on binding and women-specific constraints in Nepal, especially in relation to the role of social norms vis-à-vis structural and other factors is limited.6 This study adds to the literature aiming to empirically identify prevalent social norms related to FLFP. Existing studies (Gauri et al., 2019; Sen et al., 2022; Zeitoun et al., 2023) measure social norms by gauging not only respondents’ own beliefs, but also their expectations of how common it is for others in their social group to engage in related behaviors and the expected social approval or disapproval, including social consequences surrounding those behaviors. By rigorously measuring these components of social norms, this methodology allows identification of the strength of social norms and how closely associated they are with the outcomes of interest. This paper is one of the first applications of these measurements to the topic of FLFP in South Asia, presenting rigorous evidence on the extent to which social norms impact Nepali women’s economic participation. We add to the existing literature on the topic by highlighting nuances associated with social norms under various circumstances faced by married young women in Nepal. Given the prevalence of outmigration in Nepal, the paper also explores the flexibility of norms among households with migrant spouses. Male migration not only changes intrahousehold power allocations and dynamics, as well as upholding of norms (Ahmed, 2020; Ram Mohan et al., 2023), but also local labor markets. However, the impacts of male out-migration on labor supply of non-migrating members of the household, such as wives, remain ambiguous. On the one hand, receiving remittances could increase members’ consumption of leisure and decrease their labor supply (an income effect). On the other hand, a large outflow of migration could lead to a reduction in aggregate labor supply, which would in turn increase aggregate wages in the local labor market, thereby increasing the opportunity cost of not participating in the labor market (substitution effect). Empirical evidence from Nepal shows support for both hypotheses. Shrestha (2017), using data on large-scale male migration between 2001 and 2011, finds that a rise of 1 percentage point (p.p.) in the share of migrants in the population at the village level increases wages by 2.3 percent and increases LFP by 0.4 p.p. in non-migrant households, driven largely by increased female participation and wages in nonagricultural employment. The LFP impact for households with a migrant is lower, at 0.2 p.p., suggesting some degree of an income effect. Similarly, Lokshin and Glinskaya (2009) and Raju et al. (2018) find that FLFP falls for households with a migrant due to an income effect from receiving remittances. Phadera (2016) finds that male migration either increases women's workload, potentially overburdening them, or forces them to realign their priorities and fill home production and family farm roles vacated by men. This is consistent with findings from Ghimire et al. (2021), who find that wives of migrant workers increase their participation in farming, but also daily activities outside the home, leisure activities, and media use, suggesting a mixed impact on their workload and autonomy. 6 This study does not focus on norms that may impact a range of outcomes that start earlier in life, with implications on economic participation at later stages of life. These can include differential investments in girls’ and boys’ education, child marriages, as well as other harmful practices related to sexual and reproductive health, such as the practice of menstrual isolation (or “chhaupadi”) (Holland et al., 2023). 3 By shedding light on underlying beliefs and expectations that uphold norms, including for specific sub- groups such as migrant households, findings from this paper can help inform the design of interventions to improve FLFP, as well as reduce the disparity in job quality between men and women. Better information on the type, strength, and specific circumstances under which norms bind women’s economic participation can inform policies and services to increase their access to the labor market. The paper is structured as follows. Following a summary of the country context, Section 1 describes the methodology used for measuring social norms. Section 2 discusses the sampling approach for data collection and its limitations. Section 3 highlights the demographic and socio-economic characteristics of the sample. Section 4 details the empirical strategy employed to assess the role of social norms in determining women’s paid work outcomes outside the home. Section 5 presents the findings, starting with descriptive statistics on labor market outcomes among sampled respondents, behaviors, personal beliefs, and social expectations towards thematic areas associated with women’s work, followed by a systematic exploration of whether and to what extent norms and personal beliefs do predict women’s paid work outcomes. This section also discusses the role of social norms when considering ‘rational’ factors, observables that may explain liberality (or conservativeness) of beliefs and social norms, and whether the influence of social norms on women’s paid work outcomes varies for certain groups or under specific circumstances. Section 6 concludes. NEPAL CONTEXT Existing data on personal attitudes and beliefs around gender roles and women’s economic participation in Nepal shows a contrast between views and practices when it comes to women’s employment. On the surface, most Nepalis have a favorable outlook towards women’s jobs. Per a Survey of the Nepali People – a series of public opinion surveys of a nationally representative sample of Nepalis 18 years and older from all 7 provinces – over 90 percent of respondents disagree that women should not be encouraged to work outside their homes and a similar proportion also disagree that higher education is more important for a boy than for a girl child.7 However, when asked about attitudes and practices that create an enabling environment for a woman to go out to work, the results are not so positive. Almost a quarter of ever- married women report that their husband is jealous or angry if they talk to other men. Similarly, a wife neglecting the children is reported by men and women as the most common circumstance justifying wife beating, followed by the wife going out without telling her husband.8 Even though over 73 percent of the Nepalis believe that men have equal responsibility in raising children as their wives do,9 girls and women disproportionately bear the burden of unpaid work in the household. Almost half of the girls (aged 5-17 years) are engaged in household chores as opposed to less than a fifth of the boys and almost all women are involved in caring for the family, irrespective of their labor market status.10 Not surprisingly, “family and household responsibilities” is the most common reason stated by economically inactive Nepali women for not trying to find a job or to start a business, quitting the last job, or for working less than 40 hours a week (Table A.1, Annex A). While in the labor market, women are more likely to be unpaid family workers and less likely to be wage-employed, both formally and informally (Table A.2, Annex A). Moreover, women may face perceived societal biases that could contribute to their occupational 7 A Survey of the Nepali People, 2020. A series of this public opinion survey is led by School of Arts, Kathmandu University (KUSOA), in collaboration with Interdisciplinary Analysts (IDA) and The Asia Foundation. 8 DHS, 2022. 9 A Survey of the Nepali People, 2018. 10 Nepal Labor Force Survey, 2017/18. More than 90 more percent of females (but less than half among males) are involved in at least one activity of producing services for own final use (activities include preparing meals, doing the dishes, cleaning the house, caring for elderly and children). 4 segregation, and to preventing them from realizing their potential as entrepreneurs (Ruppert Bulmer et al., 2020; USAID, 2019; Xheneti and Karki, 2016; Bushell, 2008). Evidence shows that some working women endure hardships that could also discourage them and, potentially, other women from working. For example, women who contribute financially report that their spouses subject them to physical and emotional abuse. While women who are employed for cash are as likely as those not employed for cash or not employed at all to have ever experienced spousal physical, sexual, or emotional violence, they are twice as likely to be threatened with divorce.11 While international migration for better wage job opportunities abroad is an option for most men as an alternative to limited labor demand in the domestic market, it is not an option for most women. Just over 5 percent of external migrants over the last decade have been women, out of over 4 million labor approvals since 2008-09.12 Once again, constraints such as safety concerns including gender-based violence or exploitation of female migrants, social norms around female care-giving activities, and norms around occupations considered “acceptable” for women in sending and receiving countries limit women’s labor mobility (Ruppert Bulmer et al., 2020). SECTION 1: METHODOLOGY We collected data from married men and women across four provinces in the country. This survey is combined with qualitative data collection with similar individuals across these provinces to gain further insights on the determinants of labor market participation, decision-making, and how social norms inform (or, not) such decisions. The qualitative data collection preceded the survey in order to inform the questionnaire’s social norms module. We collected social norms’ data following an adaptation of the framework proposed by Bicchieri et al. (2014) and developed by Bussolo et al. (2024, Forthcoming). The survey tool specifically seeks to collect information on the main components of social norms, including: (1) individual behaviors (IB) – what people do; (2) personal beliefs or personal normative beliefs (PNB) – what people think they ought to do; (3) social empirical expectations (SEE) – what people think others in their social group typically do; and (4) social normative expectations (SNE) – what people think others in their social group think are appropriate behaviors within the group. Additionally, we collected information on (5) reference groups – people who are reported to have influence on one’s perception of social norms or to enforce the norms, (6) sanctions – the positive or negative consequences of conforming with or digressing from the perceived social norm, and (7) conditionality of norms – specific circumstances or characteristics that might make a behavior more or less acceptable or that might trigger a stronger normative response.13 Figure 1 summarizes the measurement framework used in this study to explore whether and to what extent social norms influence behavior around women working across the two broad themes identified, namely gender roles within the home and women in the public space. The specific focus on these domains and on married women is informed by the formative qualitative data collection conducted prior to the survey.14 The model specifies that for a behavior to be driven by social norms, it must be conditional on both the social empirical and social normative expectations, with some level of social sanctions or rewards expected for 11 Nepal Demographic and Health Survey, 2016 and 2022. 12 Ministry of Labor, Employment, and Social Security, Nepal, 2020. 13 For a deeper discussion of social norms theory and concepts, refer to Bussolo et al. (Forthcoming) 14 Annex D includes details about the qualitative survey. 5 conforming with or deviating from the norm. If individuals engage in a behavior irrespective of what they think others are doing or what others believe, then the behavior in unconditional and, therefore, not driven by social norms. While we expect behaviors to also be influenced by individuals’ normative beliefs about appropriate behavior, without a corresponding social expectation, we do not assume these to be dependent on a social norm. Table A.3, Annex A includes examples of questions in the quantitative tool as an illustration of one of the themes explored in the study, and Box C.1, Annex C provides a detailed illustration of how different components associated with social norms may work together to influence behavior in a context such as Nepal. Figure 1. Social norms measurement framework Source: Authors’ compilation based on literature and the quantitative tool In addition, a complementary methodology – vignettes – was used to understand how support for women’s employment varies under different circumstances and to assess the causal association between women’s work status, and respondents’ beliefs and social expectations. Vignettes place a hypothetical character in a situation that is typical for an average person in the context of study and ask respondents how they think the character would behave in that situation (Hainmuller et al., 2015). By prompting respondents to think from the perspective of a third person (the hypothetical character), vignettes can mitigate social desirability bias, which is a major challenge when measuring social norms (Samman, 2019). Two types of vignettes were used in this study. The first set of vignettes asked respondents whether a hypothetical woman, Sarita, would accept a job offer available to her given varied levels of social expectations. This methodology sheds light on whether certain components of social norms drive behavior more than others. We present a base scenario wherein both social normative and social empirical expectations of the hypothetical character are conservative, and then change each element to observe if the respondent’s approval of the hypothetical character working outside the home for pay changes with each manipulation. Analysis of the differences in responses to each change tells us if 6 social norms are driven by empirical expectations or normative expectations or both, thus helping arrive at causality (Bicchieri et al., 2014). Table 1 summarizes the vignette scenarios used in the study. Table 1: Vignette to test if the respondent’s approval of Sarita taking up the job changes based on changing social expectations. Conservative Normative Expectations Liberal Normative Expectations Conservative Few women in Sarita’s community work Few women in Sarita’s community work Empirical outside their homes for pay and few people in outside their homes for pay, but most people Expectation her community believe that it is okay for in her community believe that it is okay for women to work outside their homes. women to work outside their homes. Liberal Most women in Sarita’s community work Most women in Sarita’s community work Empirical outside their homes for pay, but few people in outside their homes for pay and most people Expectation her community believe that it is okay for in her community believe that it is okay for women to work outside their homes. women to work outside their homes. Notes: (1) The situation presented is, “Sarita is a woman whom you don’t know. She lives in a village with her husband who works and with her in-laws. She has two children who go to school. She recently came across a good work opportunity. The workplace is not too far from her home.” (2) The question asked is, “Do you think that it is okay for Sarita to take up the job? (Yes/ No).” The strength of social norms may also vary based on the woman’s circumstances, including her own characteristics, type of job, and spouse’s employment status. The second set of vignettes explores a few of these circumstances to examine the context dependency of norms for women’s LFP and approval for women’s participation. Varying the personal, job, and spouse’s circumstances, the vignette asks respondents whether they approve of and if they think others approve of the woman’s decision to work. This context dependency is important to understand from a policy perspective, as relaxing norms in those specific circumstances can help to promote women’s labor force participation. Figure B.1, Annex B summarizes the circumstances that were explored in the study. Lastly, the study identifies relevant reference groups for respondents across two major aspects of their lives: Decisions about important events such as education, marriage, etc. and decisions regarding the main outcome of interest – paid work outside the home (Box C.2, Annex C includes examples of these questions). SECTION 2: DATA COLLECTION The quantitative survey instrument was administered to 2,000 Nepali men and women between May 29 and June 24, 2023. While individual respondents were selected using an in-field random sampling method, a set of criteria and targets guided the selection of provinces and households to ensure sample diversity in terms views about social norms and women’s work. Earlier studies have shown regional and provincial differences in key labor market indicators as well as differences in views about norms based on ethnic groups and migration status of men (Kathmandu University and others, 2019; Paudel, 2019; Phadera, 2016). To capture this diversity, we selected 4 out of 7 provinces in the country as opposed to a nationally representative sample to ensure that each selected province has a relatively substantial sample size:15 15While Karnali Province was not considered primarily owing to its remoteness, making it less accessible to enumerators, its economic activity is similar to that of rural areas in Sudurpashchim Province. Both Madhesh Province and Koshi Province have the highest migration rate as of 2020, however, since Madhesh Province has the highest female unemployment rate, it suited the sample better. Lumbini Province was not selected since it is closer to Madhesh Province and the national average in terms of female LFP, and closer to Sudurpashchim Province in terms of female 7 1. Bagmati Province, which has the highest employment-population ratio (EPR) and female LFP in the country, 2. Sudurpashchim Province, which performs poorly on EPR and LFP indicators and has the lowest share of migrant population, 3. Madhesh Province, which has one of the highest rates of migration in the country and has the highest female unemployment rate, 4. Gandaki Province. While the above provinces comprise major ethnic groups such as Madheshi, Muslims, Newars, and Dalits, vulnerable ethnic groups such as the Janajati population comprise a higher share in Gandaki Province. Given the low female LFP in Nepal, sampling targets for working and non-working women were set to ensure sufficient representation of each group. Additionally, to understand how social norms may impact women with migrant spouses in a country where male out-migration is highly prevalent, targets were also set to include sufficient respondents from migrant households. As such, the study defined 3 categories of households, with a pre-identified target number of respondents within each: (1) Households with working female respondent (425 households), (2) households with non-working female respondent (425 households), and (3) households with a husband who migrated for work (domestically or abroad) at the time of the survey (hereinafter referred to as ‘migrant households’) (300 households). For category 1 and 2 households, a married couple was sampled and interviewed,16 and for category 3, only the female spouse of the migrant husband was interviewed. To ensure representativeness at the province level, households were randomly selected within the four selected provinces. Each province was divided into clusters, or wards, the smallest administrative unit, which served as the primary sampling unit (PSU) for this survey. The sampling frame consisted of all wards within each selected province. The number of clusters per province was selected based on population size of the province and divided across the three categories of households. Ten households were targeted within each cluster. Wards for categories 1 and 2 were selected using the probability proportionate to size (PPS) method, giving an independent chance of selection to each ward as per its population size. Out of these wards, a subset was selected for category 3 households. Finally, in-field random sampling of households within each selected ward was done using the spin-the-bottle technique, employing a skip interval to survey every third household in rural and every fifth household in urban areas – continuing this pattern until the determined number of households were interviewed for each category. Interviewers used a screening questionnaire to ascertain eligibility of respondents.17 The final sample includes: 630 respondents from Madhesh Province, 760 from Bagmati Province, 360 from Gandaki Province, and 250 from Sudurpashchim Province. Further details on sampling are included in Annex E. Study limitations The study has two main sampling limitations. First, findings cannot be generalizable for all women and men in Nepal as the sample is not nationally representative, including not being generalizable to unmarried women and men, given that the study focused on married couples only. Second, setting targets for households with working women resulted in a higher-than-average employment-to-population ratio (EPR) among women in the sample, as compared to the population average (EPR among female respondents in the sample using the reference period of 7 days, as Section 5 notes, is 44 percent, whereas female EPR in the 4 provinces is 23.8 unemployment rate. 16 Married couples were interviewed separately; not together. Measures were taken by enumerators to ensure that each individual respondent had privacy, to the extent possible, when responding to the survey. 17 The screening questionnaire defined working as “participating in any activities for a wage, salary, commission, profit, or payment of any kind; this includes working as an employee, or in your own business, or out of own initiative, or paid apprenticeship, paid domestic work, or paid farm work.” 8 percent).18 This could generate concerns that the sample is skewed towards liberal views if one is to believe that working women are more likely to be liberal. However, it is important to note that the main outcome variable for this study is, "I work outside the house for pay" (at the time of the survey), and there, the share of women working is 29.39 percent, which is comparable to the average population statistic for the four provinces. SECTION 3: SAMPLE CHARACTERISTICS Household composition and socio-economic outcomes of members Out of the 850 households in the sample, 26 percent are in rural areas. Most households are concentrated in Nepal’s two ecological regions: hills (47 percent of households) and Terai (48 percent of households). Only 5 percent of households are from the mountains, given the remoteness of those areas. Table A.4, Annex A provides further description of the sampled households. Most households are male-headed and have an average of 4.25 members. An average household has 1-2 children under the age of 18 years and around 2 working age adults (18-59 years). About a third of the sampled households co-reside with a mother and/or father-in-law. On average, 68 percent of working-age adults are in the labor force (Table A.5, Annex A). The share is higher among men - 87 percent vs. 58 percent for women. While most men are salaried workers (59 percent compared to 46 percent among women), women are more likely to work as self-employed (39 percent compared to 26 percent among men). The proportion currently engaged in unpaid employment is low, among both males and females. The average monthly income of a household in the sample is reported at 38,738 NPR.19 Most households (80 percent) report having earned income from employment (Table A.6, Annex A). Other reported sources of income in the month preceding the survey include international remittances (26 percent), loans (13 percent), and saving or sale of assets (10 percent). A small share of the sample (6 percent) reports receiving government transfers from safety net programs. The average age of a respondent is 36 years (with an average gap of 5 years between men and women). Most respondents are Hindus (86 percent), followed by Buddhist (9 percent), but there is more diverse representation of ethnicities – 33 percent identify as Adibasi and Janajati, 29 percent as Khas and Aryan, 27 percent as Dalit, and 10 percent as Madheshi. Migrant Households Migrant households in the sample mostly comprise husbands who migrated internationally (93.3 percent) to countries such as Saudi Arabia, Qatar, Malaysia, United Arab Emirates (UAE), and India. According to female spouses, work is the main reason for their husbands’ migration, with construction worker s, security guards, and drivers being the most common occupations. Almost half of migrants have not completed basic education. Migrant households in our sample differ on some important characteristics when compared to non- migrant households. Given the absence of the male spouse, migrant households are less likely to be headed by a male. They tend to be smaller and have more children and fewer household members of working age and that are in Authors’ calculation using the Nepal Labor Force Survey 2017/2018. 18 This likely places the sample in the middle of the income distribution. According to the 2014/2015 Nepal Household 19 Budget Survey, the average monthly household income for the third consumption quintile group is estimated at 24,516 NPR and for the fourth consumption quintile group (second richest group), it is at 32,042 NPR. 9 the labor force. They are less likely to report earned income from employment and more likely to report international and domestic remittances. The average amount for international remittances received monthly is reported at 41,166.67 NPR, which is much higher than the average monthly income from employment among non-migrant households (24,904 NPR). Overall, migrant households report higher total monthly income than non-migrant households (49,021 NPR compared to 35,109 NPR monthly, respectively).20 Female spouses of migrant husbands also tend to be younger and less likely to work both 30 and 7 days prior to the survey, compared to women in non-migrant households. Among those who have ever worked, women from non-migrant households are more likely to work/have worked longer hours per week and to earn more. They are also more likely to work/have worked as own-account worker, and less likely to work/have worked as an employer or wage employee. SECTION 4: EMPIRICAL STRATEGY In order to assess the role of social norms in women’s decision to work for pay outside the home (i.e., the main outcome variable used throughout the analysis as “Paid Work”), we proceed in two stages. First, we explore the correlations between norms and our behavioral outcomes of interest. Examining associations tells us whether one’s beliefs and social expectations seem to have any influence on behaviors, and whether they are in the direction we hypothesize, i.e., conservative beliefs being positively correlated with conservative behaviors and vice-versa. For each norm included in the study, we explore whether conservative beliefs and social expectations are associated with corresponding behaviors. Equation (1) presents a bivariate regression of a behavior on PNB, SNE, SEE, and sanctions, regressed for each norm. Equation (2) regresses the main outcome variable – women working for pay outside the home at the time of the survey – on the same set of explanatory variables. Equation (1) = 0 + 1 + Equation (2) = 0 + 1 + is the individual behavior reported by female respondents or the spouse’s behavior reported by male respondents, is the corresponding personal normative beliefs or social normative expectations or social empirical expectations or expected sanctions for norm for respondent . There are 8 norms examined in each specification of this equation: female caregiver (household work mainly done/should be done by women), male breadwinner (financially providing for family solely done/should be done by men), working women (women do not/should not work outside the home for pay), mobility freedom (women do not/should not go outside the home alone/unaccompanied), honor (women working is dishonorable to family), safety (women working are subject to harassment), motherhood (women do not /should not leave children with someone else when going out to work), and mixed-gender workplace (women do not/should not work in mixed-gender places) norms. In equation (2), is the work status for individual (which is a binary variable that takes the value of 1 if the female respondent or, in the case of male respondents, their spouse is engaged in paid work outside the house at the time of the survey). Next, we consider the three core elements of social norms together (PNB, SNE, and SEE) to further learn which element best predicts women’s work status. To do so, we explore the relationship between these elements and the outcome of interest using multivariate regressions. Aggregating responses to beliefs and 20This finding is consistent with existing literature that finds that Nepali migrant households are more well-off than the average household in Nepal (IMF, 2020). 10 social expectations pertaining to multiple norms by theme, we create indices across the two themes (gender roles within the home and women in the public space) using Bussolo et al.’s (2024, Forthcoming) methodology.21 Our use of personal beliefs and normative expectations to explore the relationship between social norms and related outcomes is consistent with the approach taken by other literature as well (Bernhardt et al., 2018; Bursztyn et al., 2022; Goldstein et al., 2024).22 Additionally, a single index for sanctions is used in the analysis. The index for household roles consists of 2 items – female caregiver and male breadwinner norms, while the index for women in public space comprises 5 items – working women, mobility freedom, honor, safety, motherhood, and mixed-gender workplace norms. In these indices, each item ranges from 0 to 1, with 1 being the most conservative and 0 being the most liberal.23 Beliefs and social expectations for men and women across most indices follow similar distributions. As such, we do not report findings from household-level regressions. Equation (3) shows the regression specification that explores association between women’s work status and indices for each of the two core elements of social norms. Across the different specifications, household-level and individual controls are included. Equation (3) = 0 + 1_ + 2_ + + is the work status for individual (which is a binary variable that takes the value of 1 if the female respondent or, in the case of male respondents, the spouse is engaged in paid work outside the house at the time of the survey), _ and _are indices of relevant personal normative beliefs, and social normative expectations for theme for respondent . is a set of control variables that include respondent characteristics such as age squared and years of education, as well as other household control variables including having child(ren) younger than 5 years old, living in an urban area, ethnicity, province, log of total household income, presence of at least one of the parents-in-law in the household, and whether it is a migrant household. Next, we run an OLS regression that incorporates, in addition to the indices, “rational” factors that may play an important role in determining women’s work decisions. Equation (4) = 0 + 1_ + 2_ + + The outcome variable and indices are the same as in equation (3). includes educational attainment, household income, the presence of other working-age women who are in the labor force as well as adolescent girls in the household, and awareness of the respondent regarding job search methods. 21 Using the data collected for this survey, the authors developed two indices using rigorous measures of reliability and internal consistency (Cronbach’s alpha and factor analysis) of all items within each social norms’ element, as well as testing for their predictive validity when it comes to predicting women’s work status. The two indices derived from this analysis were mostly consistent with the classification proposed earlier in this paper, with the exception of the motherhood norm that was not deemed suitable for either index. 22 Internal consistency of the SEE variables, however, was low and the factor loadings were inconsistent and incomparable with the SNE and PNB indices. As per Bussolo et al.’s conclusion, SEE variables and indices are thus excl uded from the regression analyses. 23 The distribution of the aggregated belief and social norm indices across these elements is provided in Figures B.2, Annex B, broken down by gender. 11 To explore if and which observable characteristics about an individual or their household are associated with the prevailing liberality (or conservativeness) of beliefs and social expectations, we run an OLS regression of each thematic index on household and individual-level characteristics of respondents, using the specification presented below. Equation (5) = 0 + 1 + 2 + 3 + 4ℎ + 5 + 6ℎℎ_ + 7 + 8ℎℎ + 9 + 10 − + is the PNB, SNE, or Sanctions index score and 1 to 10 are various individual and household characteristics of individual . We conduct heterogeneity analysis by estimating equation 5 for each index across the two components (PNB, SNE) and separately for each of the two themes. Equation (6) = 0 + + 1_ + 2_ + 1_ ∗ + 2_ ∗ + + The outcome variable, indices, and the set of controls being the same as in equation (3). captures the dimension of heterogeneity: migration status of the male spouse in the household (migrant vs. non- migrant households) and ethnicity (Madheshi vs. Non-Madheshi). captures the estimate on the association between each index and outcome when = 0, and + is the estimate when = 1. SECTION 5: FINDINGS Labor market outcomes and constraints faced by Nepali men and women. Among the sample interviewed for this study, we find low levels of human capital investments which are comparable to the national average.24 Education levels are low, with a third of respondents having completed some basic education (started a grade between one and eight but did not complete it), and 26 percent reporting not being educated at all. Only 19 percent of respondents have completed grade eight or some secondary education. On average, women have less education than men, with 34 percent having no education (compared with 15 percent among men), and only 7 percent of women having completed basic education (compared to 13 percent among men). Table 2 presents sample statistics pertaining to key labor market outcomes of Nepali men and women (further details provided in Table A.7, Annex A). Most male respondents report having worked in the past 7 and 30 days prior to the survey. On the other hand, over half of all women indicate they have not worked in the past 30 days or in the week prior to the survey (57 percent). Almost 40 percent of women are not working nor looking for work. The main reasons for not working in the past 30 days among women include household chores, lack of trustworthy childcare options, and lack of suitable jobs, with the latter being the most common reason for not working among men. Among non-working respondents, while 59 percent of men report having worked at some point in the past (most of them in the construction sector), only 17 percent of women report the same (most of those worked in the agriculture, forestry, and fishing sector). Among non- working women 24A third of all working age individuals (15 years and above) in Nepal are illiterate, and 27 percent have up to basic education (Nepal Labor Force Survey, 2017/2018). 12 who have worked in the past, the main reasons for stopping include the burden of juggling household chores and work (43 percent), having children (24 percent) and getting married (13 percent). Men, on the other hand, face other distinct barriers, as a majority cite illness, injury, or disability (24 percent), lack of desire to work (21 percent), or lack of suitable jobs (19 percent) as the main reasons for stopping work. Additional barriers reported by both women and men include lack of resources for job search. Most respondents (99 percent among women, and almost all men) report that access to networks followed by having own/family business (37 percent among women, and 36 percent among men) are the two channels to find jobs. A negligible share of both men and women has heard about employment service centers (ESCs) run by the Ministry of Labor and Social Security (MoLESS) and only a few mention job agencies as a resource. Over 95 percent of men and women found their job using networks or having a family/own business. Over 82 percent of non-working women and men expressed willingness to take a job if an opportunity becomes available in their city/village, but it drops to 40 percent among women if the opportunity becomes available outside their city/village. Distance appears to be a factor for women. Among women who currently work or have worked in the past outside the home, most walked to their job, and very few take public transportation or own a car or a bike in contrast to working men. On average, it takes women 21 minutes to commute to work if working outside the house, compared to 26 minutes for men. Almost all respondents who worked in the past 30 days are engaged in paid work. Women are twice more likely to work inside the house (41 vs. 18 percent among men), work as own-account workers (45 vs. 26 percent among men), and work in the agricultural sector (25 vs. 14 percent among men). Similarly, women are also slightly more likely to work as paid helpers in a family business and/or farm, while men are more likely to work as salaried/wage workers (46 vs. 31 percent among women) or casual laborers (20 vs. 15 percent among women). Overall, women are more willing to work for lower wages than men –15,030 NPR a month, compared to 22,060 NPR a month for men. The average monthly wages received by women and men show a clear raw gender wage gap of 34 percent, with men receiving on average 20,071 NPR monthly and women 13,235 NPR monthly in gross income from paid work in cash.25 However, there are no gender differences in jobs offering social security contributions and a written contract, with only 8 and 15 percent of respondents in the sample receiving those benefits, respectively. Working women in the sample are still confronted with constraints that other non-working women face. For example, working women rely on family for childcare and very few rely on non-family options, such as workplace-provided childcare and a babysitter. Non-family options are largely unavailable or not trustworthy (as reported by 48 and 53 percent, respectively, of women who report working in the past 30 days or ever worked). Moreover, among women with a desire to start a business, 32 percent report that their childcare, eldercare, or household responsibilities are major barriers to starting a business (compared to these responsibilities reported as a barrier by just 7 percent of men). 25The raw gender wage gap is the difference between average gross monthly income of men and women as a percentage of men’s gross income. 13 Table 2. Key labor market outcomes for Nepali men and women in the sample Indicator Women (%) Men (%) Education No education 34.09 15.29 Some basic 29.39 32.12 Basic complete (complete grade 8) 7.22 12.94 Some secondary (start grade 9 or 10) or 14.17 22.71 complete (pass S.E.E/S.L.C.) Some higher secondary (start grade 11 or 12) 12.60 13.06 or complete (pass H.S.E.B final) Diploma/bachelor’s complete or higher 2.53 3.88 Worked in 30 days preceding the survey 44.78 88.35 Worked in 7 days preceding the survey 43.48 86.47 Work status of all respondents Self-employed 23.56 24.71 Salaried/wage employee 19.04 60.35 Unpaid worker in the family business 2.17 3.29 Does not work and not looking for work 39.92 5.30 Does not work but is looking for work 15.30 6.35 Reasons for not working, among non-working respondents (multiple responses) Illness, injury, or disability 2.20 14.14 Childcare, no other options I trust 36.85 0.00 Eldercare 7.72 1.01 Household chores, no one to take over 57.95 16.16 No support from family 9.61 2.02 No suitable jobs available 31.18 55.56 Other 16.69 29.29 Notes: (1) Self-employed includes those who are own-account workers with no employees and those who are employers with at least 1 employee, the share of the latter being less than 1 percent. (2) Salaried/wage employees includes paid workers in family businesses, army. (3) Does not work and not looking for work includes not looking because currently studying, training, or upgrading – the shares for which are small (0.35 and 0.24 percent for females and males resp.) (4) The category “Other” under reasons for not working is an aggregate of the following: no desire to work, retired or too old, currently studying/upgrading qualifications, want to migrate abroad, no financial need. Among females, the largest share is no desire to work at 9.45 percent. Among males, it is retired/too old at 13.13 percent, followed by no desire to work. Deciding to work, whose opinion matter in driving behavior? Identifying the reference group is crucial for understanding how social norms affect women’s LFP. Spouses and other household members are the most influential voices for Nepali women’s (Figure 2) decisions around working outside the house for pay. Other family members who do not reside in the same household, and community members are important, but to a lesser extent, with 28 percent of respondents indicating their opinions matter to some or to a large extent. 14 These answers are consistent with decision-making influencers when it comes to life events such as education, migration, children’s education, and their marriage (Figures B.3, Annex B). Women are slightly more likely than men to report that their spouses’ and non-household family members’ opinions matter to them when taking decisions about life events and about paid work outside the home. This may be the case as typically, married women in Nepal do not live with their natal families, and thus may consult with their parents and/or siblings when making these decisions. Among female respondents, those who are engaged in paid employment are more likely to report that their spouse’s opinion matters when taking decisions about major life events and less likely to report that their non- household family’s opinion matters to them. When it comes to decisions about working outside the home for pay, there are no significant differences in the importance of a spouse’s opinion between working and non- working women. On the other hand, non-working women are more likely than working women to value community member’s opinions about working outside for pay. Men with working wives consider their spouse’s opinions about decisions regarding both life events and paid work outside to a greater extent than do men with non-working wives. This may point to higher agency for working women; when a woman is working, her spouse is more likely to value her opinions about important decisions. However, it could also indicate that women who have more influence on their spouses’ decisions to begin with are able to negotiate better at the intra-household level when making employment decisions. Women with migrant husbands are less likely to report that their spouse’s opinions about life events matter to them, perhaps given the physical absence of the husband, women are the ones responsible for major decisions around the household. Their natal families appear to be more influential about such decisions and about paid work outside the home, and their community’s opinions seem to matter more than it does for women from non-migrant households. Figure 2. Reference groups for working outside the home for pay Notes: Family include parents or parents-in law, siblings or siblings-in-law, children, etc. that live within the same household as the respondent. Non-HH (household) family includes relatives who do not live in the same household as the respondent. Community includes neighbors, friends, religious leaders, groups that the respondent is part of such as women’s groups, cooperatives, agricultural groups, religious groups, etc. 15 Behaviors, personal beliefs, and social expectations regarding women in the public space Beliefs and social expectations are explored across two themes as noted in above: women in public spaces, and traditional gender roles in the household. Figure 3 presents behavior, personal beliefs, and social expectations across five questions that relate to women in the public space and their mobility. There is widespread support for women working outside the home for pay. On a scale of 0 to 100 (100 being the most conservative), the average support for women not working outside for pay is 16, indicating widespread liberal opinions among both male and female respondents. Respondents’ own beliefs are in line with what they believe others think as well.26 On average, respondents expect only 23 percent of others in their reference group to find it not acceptable for women to work outside for pay, with only 8 percent expecting negative societal sanctions for women who work outside for pay. We also elicited respondents’ personal opinions (PNB) about aspirations they have for their daughters and daughters-in-law to learn whether beliefs are in line with respondents’ expectations of their children (not shown in Figure 3; refer to Table A.8, Annex A). Most respondents aspire for the next generation of female members to work outside the home and earn a living (with 87 percent of respondents wanting their daughters to work and 82 percent wanting their daughters-in-law to work). Similarly, when it comes to views that relate to women’s mobility and being in the public space – working in mixed-gender environments, going out unaccompanied, not being exposed to harassment or not bringing dishonor to family because of working – few respondents hold conservative personal beliefs and expect others in their reference group to hold them as well. Consistent with social normative expectations, very few expect negative sanctions for women acting liberally on those dimensions. A low share of female respondents (or in the case of male respondents, their spouses) in the sample behaves in a conservative manner on some of those dimensions. For example, only 30 percent of working women report not working in gender-mixed environments or unwillingness to do so and only 8 percent report not going outside the house to the market unaccompanied by a family member. In fact, female and male respondents appear to overestimate how conservative women in their reference group are in terms of how they behave on those two dimensions. They estimate that 42 percent of those that work do not work or are unwilling to work in gender-mixed environments, and that 26 percent do not go outside unaccompanied. The only exception within the public space, where female respondents’ behavior aligns with more conservative or traditional gender norms is the low share of women that actually works for pay outside the home. On average, 70 percent of women in the sample report not working outside for pay at the time of the survey.27 In contrast, respondents appear to think that in in their reference group less than half of the women do not work outside (46 percent). 26For ease of interpretation, the scales for behaviors, beliefs, social expectations, and sanctions have been adjusted so they are comparable. The original scale of 0-10 for social expectations has been adjusted to go from 0- 100 for this chart. 27Note that the recall period for the statement was “I (my wife) work(/s) for pay outside the home” was “at the time of the survey”, and hence the share of women not working (70 percent) differs from the earlier reported statistics that used different recall periods (last 7 days and 30 days). Responses to this question, "I work for p ay outside the home” (at the time of the survey) is consistently used in the analysis as the outcome variable. 16 Figure 3. The behavior, beliefs, and social expectations of Nepalis regarding social norms around women in the public space Notes: Individual behavior is a binary variable that presents the share of female respondents (or as it pertains to the behavior of the spouse if asked to the male respondent) behaving in accordance with the conservative statement. Values for statements pertaining to personal beliefs show the average support among male and female respondents for a statement that takes the value on a scale of 0 to 100, where 0 is the most liberal and 100 is the most conservative. Values for statements pertaining to social normative expectations and social empirical expectations show the share of male and female respondents’ reference group members whom they expect to hold more conservative beliefs or act in a conservative manner, respectively. Sanctions is a binary variable that presents the share of male and female respondents who expect a negative sanction for not adhering to the norm in question. Questions on sanctions and individual behavior were not asked on the two dimensions related to harassment and dishonor. Table 3. Bivariate regressions of behaviors on beliefs – Women in the public space Beliefs and Individual behavior (IB): IB: Women do not work in IB: Women do not go social Women do not work outside mix-gender environment outside the house expectations the house for pay unaccompanied PNB 0.15** 0.20* 0.52*** (0.06) (0.11) (0.08) SNE 0.16** 0.12 0.52*** (0.07) (0.09) (0.07) SEE 0.21*** 0.06 0.26*** (0.07) (0.06) (0.06) Sanctions 0.05 0.15*** 0.27*** (0.04) (0.04) (0.04) Notes: (1) A positive correlation implies that more conservative beliefs are associated with conservative behaviors. (2) All items on beliefs and social expectations are scaled from 0 to 1, with 0 being the most liberal and 1 being the most conservative. Sanctions is a binary variable that presents the share of respondents who expect a negative sanction for not adhering to each statement. Outcome variable is coded as 1 if the individual respondent answered “yes” to conforming to the conservative behavior. If the behavior pertains to the other gender, the respondent was asked about his/her spouse’s behavior and the variable was coded accordingly. (3) No controls and regional fixed effects are included. (4) SEs clustered at the ward level. (5) Robust standard errors in parentheses. (6) *** p<0.01, ** p<0.05, * p<0.1. 17 Overall, behaviors seem to be generally correlated with beliefs. Table 3 shows that conservative behaviors, especially around working outside for pay and going out unaccompanied, are correlated with conservative beliefs and social expectations. When it comes to working in mixed-gender environments, however, while beliefs are correlated with behavior, we find no association between behaviors and social expectations. The association with sanctions, however, is mixed. There is no correlation between respondents’ expected sanctions and behavior when it comes to women working outside for pay. On the other hand, women not working in gender-mixed environments and not going outside the house unaccompanied are strongly associated with respondents’ expectations about high sanctions, for those that expect them. Behaviors, personal beliefs, and social expectations regarding traditional gender roles within the household When it comes to fulfilling gender roles within the household, Nepali men and women in the sample are not as liberal in their beliefs and expectations (Figure 4). For example, for statements that ask about personal beliefs around women leaving their young children with someone else when they go out to work for pay (i.e., motherhood norm), as well as around spousal roles in the household, such as household work being done by women (i.e. female caregiver norm) and financially providing for family is the man’s sole responsibility (i.e. male breadwinner norm), respondents report support of around 50, 40, and 37, respectively, on a scale of 0 to 100 (with 100 being the most conservative). This indicates an overall higher adherence to traditional gender roles as compared to norms around women in the public space. Figure 4. The behavior, beliefs, and expectations of Nepalis regarding social norms around gender roles within the home Respondents expect others in their reference group to strongly adhere to traditional gender roles, especially around the motherhood norm. For example, they estimate that two-thirds of the people in their reference group will not approve of women leaving their young children with someone else if they go out to work and will face sanctions for breaking the norm. About half of the women in the sample (53 percent) report not leaving or not being willing to leave their young children with someone else to go to work and expect a higher share (67 percent) of people in their reference group to behave accordingly. These personal beliefs and social 18 expectations around motherhood translate into respondents’ views about other females in the household – we observe a 17-23 p.p. decrease in respondents’ aspirations for their daughters and daughters-in-law to work outside the house for pay if they have school-going children.28 Respondents estimate that about half of those in their reference group agree or behave in a way that allocates women the caregiver role, in line with their own behaviors. However, the average support for this norm among male and female respondents is 40 and less than half of respondents (40 percent) would expect sanctions for households that do not adhere to this norm. When it comes to norms around men’s breadwinner role, personal beliefs and behavior are less conservative compared to social expectations, but for this norm, overall individual views and expectations are less conservative compared to women’s roles as mothers and caregivers. Only 30 percent of respondents reported that financially providing for their households is solely the responsibility of the man, despite estimating that 47 percent of their community live in households with such arrangements. Only 20 percent of the respondents expect negative sanctions for households where men are not the sole providers. Conservative behaviors correlate very strongly with conservative beliefs and social expectations across all three norms around gender roles, as shown in Table 4. There seems to be no notable differences in the correlation coefficient between the three components (PNB, SNE, SEE) for the two norms on women’s and men’s household roles. However, on the motherhood norm, SEE coefficient, while still statistically significant, is much lower than PNB and SNE. Table 4. Bivariate regressions of behaviors on beliefs – Gender roles within the home Beliefs and social Individual behavior expectations It is a woman’s primary A man is the sole Women do not leave young responsibility to take care of financial provider for his children with someone else the household (cooking and family when they go out to work for caring for children) pay PNB 0.72*** 0.49*** 0.72*** (0.06) (0.05) (0.03) SNE 0.70*** 0.49*** 0.81*** (0.06) (0.05) (0.05) SEE 0.75*** 0.56*** 0.58*** (0.06) (0.04) (0.08) Sanctions 0.24*** 0.15*** 0.36*** (0.05) (0.03) (0.04) Notes: (1) A positive correlation implies that more conservative beliefs are associated with conservative behaviors. (2) All items on beliefs, and social expectations are scaled from 0 to 1, with 0 being the most liberal and 1 being the most conservative. Sanctions is a binary variable that presents the share of respondents who do not expect a negative sanction for not adhering to each statement. Outcome variable is coded as 1 if the individual respondent answered yes to conforming to the behavior. If the behavior pertains to the other gender, the respondent was asked about his/her spouse behavior and the variable was coded accordingly. (3) No controls and regional fixed effects are included. (4) SEs clustered at the ward level. (5) Robust standard errors in parentheses. (6) *** p<0.01, ** p<0.05, * p<0.1. 28The decrease in aspirations for daughters is from 87 to 64 percent, whereas that for daughters-in-law, it is from 82 to 65 percent. 19 Association between social norms and women’s employment This sub-section examines whether and the extent to which decisions about women’s work for pay outside the home may be influenced by personal beliefs, social empirical and normative expectations, and perceptions of sanctions for not conforming to norms across the two themes explored in this study. Bivariate regressions reported in Table A.9, Annex A show that the association between personal beliefs and social expectations and the outcome of interest (women’s work for pay) are in the direction we hypothesized, i.e., more conservative norms are associated with lower participation of females in paid work outside the home For example, on a scale of 0-1, with 1 being the most conservative, going from most liberal to most conservative in one’s personal belief that a woman should not work for pay outside the home is associated with an average 15 percentage points lower probability of the woman working for pay outside the home. Similarly, going from most liberal to most conservative in one’s social expectations about the proportion of reference group members who do not approve of women working for pay and the proportion of other women who do not work for pay reduces the probability of a woman working by 16 percentage points and 21 percentage points, respectively. The only exception is the motherhood norm, where beliefs and expectations do not correlate with women’s work outcomes. However, important differences emerge across the two themes as well as between male and female respondents. First, beliefs that relate to traditional gender roles for men and women in the household tend to correlate with women’s work status more so than the beliefs around women in public spaces. Social norms – both empirical and normative expectations – appear to correlate with women’s work status when work challenges gender traditional roles. This is also true for social expectations specifically around women working outside the home for pay, but it is not the case with most other dimensions of women in public places. Second, social sanctions, which are relatively low to begin with, are not correlated with women’s work status, with the exception of when households breach the male provider norm. For both men and women, higher expected severity of social sanctions on the male breadwinner role is associated with a lower probability of the woman working. Overall, women's personal beliefs and social expectations are more significantly correlated with their work status as compared to their spouses’. For men, social expectations rarely correlate with their spouses’ work status, with the exception of empirical expectations about other women working for pay outside the home and working in mixed-gender environments. Their personal beliefs also correlate less with women’s work status and do not appear to matter when it comes to beliefs about appropriateness of women working outside the home and about men’s provider role. When we analyze the individual level predictors of women’s work outcomes (Table 5), we find that for norms around household roles, personal beliefs, rather than social expectations tend to be more strongly correlated with women’s work status. Going from most liberal to most conservative in the PNB index for household roles is associated with an average 8 percentage points lower probability of women working for pay outside the home. In contrast, when it comes to norms around women in the public space (column 2) associations with personal beliefs and social normative expectations are not statistically significant. These results also hold when variables across both themes are considered in the same regression specification. For male respondents, however, even their own beliefs do not appear to influence employment outcomes of their spouses.29 Overall, these findings suggest that mainly women’s own beliefs around gender roles influence women’s employment outcomes, and that there is limited social disapproval expected towards women who participate 29 The same regressions were run without any controls, the results for which are reported in Table A.11, Annex A. 20 in the labor force. This is further confirmed by the low expected sanctions and lack of association between women’s work and expected sanctions, as shown in Table A.10, Annex A. These findings are consistent with the descriptive results that show unanimous high acceptance among respondents for women’s mobility and being in the public space, and less so for norms that challenge the status quo related to traditional gender roles. However, these results also show that when considering all elements of social norms together, the relative strength of the relationship between these elements and women’s work outcomes can vary. We further contextualize findings in Table 5 by unpacking the two key items within the gender roles within the household index (Table 6). For the female caregiver norm, women’s conservative beliefs are negatively associated with their own employment outcomes, but not men’s beliefs on this issue. On the other hand, for the male breadwinner norm, men’s conservative beliefs negatively correlate with women’s work outcomes. In addition, women’s beliefs that their reference group holds conservative views on this issue also negatively correlate with their work outcomes. Tables A.12, Annex A include these results for each norm within the public space index. When looking at the full sample, we find no association of beliefs and social expectations about each norm that forms the public space index with the outcome variable – women’s paid work outside, which in line with results of Table 5. Table 5. Individual-level regressions of social norms indices predicting women’s work outcomes Full sample Females only Males only (1) (2) (3) (4) (5) (6) (7) (8) (9) Paid Paid Paid Paid Paid Paid Paid Paid Paid work work work work work work work work work PNB Index – -0.08** -0.08** -0.09** -0.08** -0.08 -0.08 HH Roles (0.04) (0.03) (0.04) (0.03) (0.06) (0.06) SNE Index – -0.04 -0.04 -0.08 -0.08 0.01 0.01 HH Roles (0.05) (0.05) (0.06) (0.06) (0.07) (0.06) PNB Index – -0.03 -0.01 -0.03 -0.02 -0.02 -0.00 Public Space (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) SNE Index – -0.00 -0.00 -0.00 0.00 0.00 -0.01 Public Space (0.03) (0.03) (0.03) (0.03) (0.04) (0.04) -0.04 -0.12 -0.03 -0.00 -0.12 0.01 -0.09 -0.14 -0.08 Constant (0.26) (0.24) (0.26) (0.26) (0.24) (0.27) (0.26) (0.25) (0.26) Obs. 2,000 2,000 2,000 1,150 1,150 1,150 850 850 850 Adjusted R2 0.03 0.02 0.03 0.04 0.02 0.03 0.01 0.00 0.01 Controls YES YES YES YES YES YES YES YES YES Regional FE YES YES YES YES YES YES YES YES YES Notes: (1) Negative correlation means that more conservative beliefs and social expectations are associated with lower probability of women working for pay outside the house. (2) For all indices, higher the value, more conservative the beliefs and expectations. Outcome variable “Paid work” is binary with 1 if the female reported a “yes” to “I work outside the house for pay.” And if a male reported “yes” to “My wife works outside the house for pay.” (3) Standard errors clustered at the ward level. (4) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 21 Table 6. Individual-level regressions of norm items within the HH roles index predicting women’s work outcomes Full Sample Females Only Males Only (1) (2) (3) (4) (5) (6) Paid work Paid work Paid work Paid work Paid work Paid work PNB – Female caregiver -0.13* -0.17** -0.07 (0.06) (0.06) (0.10) SNE – Female caregiver -0.07 -0.09 -0.05 (0.10) (0.12) (0.12) PNB –Male breadwinner -0.18** -0.15 -0.23** (0.08) (0.09) (0.09) SNE – Male breadwinner -0.07 -0.18* 0.08 (0.10) (0.10) (0.12) Constant -0.06 -0.03 -0.04 -0.00 -0.11 -0.09 (0.26) (0.25) (0.27) (0.25) (0.26) (0.26) Obs. 2,000 2,000 1,150 1,150 850 850 Adjusted R2 0.02 0.03 0.03 0.04 0.01 0.01 Controls YES YES YES YES YES YES Regional FE YES YES YES YES YES YES Notes: (1) Negative correlation means that more conservative beliefs and social expectations are associated with lower probability of women working for pay outside the house. (2) For all items, the higher the value, the more conservative the beliefs and expectations are. Outcome variable “Paid work” is binary with 1 if the female reported a “yes” to “I work outside the house for pay.” And if a male reported “yes” to “My wife works outside the house for pay.” (3 ) Standard errors clustered at the ward level. (4) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.10. The role of social norms when considering “rational” factors determining women’s work outcomes Findings thus far suggest that social norms may not be very restrictive in Nepal, as evidenced by limited social disapproval expected towards women who participate in the labor force. Specifically, social normative expectations are not strong predictors of women’s LFP in Nepal, though personal beliefs (i.e., preferences), especially around household roles, do appear to matter. However, we also know that practical, “rational” factors - such as educational attainment, household income, the presence of other working women within the household, the presence of adolescent girls (10-19) in the household, and awareness regarding job search methods, among others- play an important role in decisions related to women’s employment.30 We run an OLS regression that incorporates these factors, in addition to personal beliefs and social normative expectations and find that when rational factors are considered, only personal beliefs about household roles matter (Table 7), whereas normative expectations overall and personal beliefs about women in the public space are not predictors of women’s work outcomes.31 Not surprisingly, financial need shows up as a strong predictor of whether women take up paid jobs outside the home: An increase in family income (without factoring in the income of working females within the household) reduces the likelihood that women will work for pay. The presence of other working age women within the same household who are in the labor force is also associated with a reduced probability of the female respondent working for pay outside the home, 30 It is important to note that this is not meant to be an exhaustive list of rational factors that can influence female LFP. 31 These factors reflect only those variables that were asked to all respondents in the survey. Robustness check using the postestimation variance inflation factor test reveals that multicollinearity is not a concern; VIF values for all variables are less than 10 with a mean VIF of 2.54. 22 perhaps due to the need for at least one woman in the household to be available to assume responsibilities such as cooking and cleaning and looking after children. Relatedly, a woman is more likely to work outside the home for pay if there is presence of adolescent girl(s) in her household, indicating that young girls could act as substitutes for working women’s domestic responsibilities.32 Table 7. Explanatory power of “rational” factors on women’s work outcomes vs. that of social norms PAID WORK PNB Index - HH Roles -0.08** (0.03) SNE Index - HH Roles -0.05 (0.04) PNB Index - Public Space -0.01 (0.03) SNE Index- Public Space -0.01 (0.03) Years of education 0.00** (0.00) Log of household's total income minus working females' -0.03*** earned income (NPR) (0.01) Presence of other working-age women in HH in labor force -0.12*** (0.03) Indicator for adolescent girl(s) in respondent's HH 0.07* (0.04) Respondent is aware of non-network methods of job search 0.01 (0.03) Constant 0.64*** (0.07) Observations 2,000 Adjusted R-squared 0.05 Regional FE YES Notes: (1) Apart from the “rational” factors included and shown in the table, no additional variables were included as controls. (2) SEs clustered at the ward level. (3) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. What explains liberal beliefs around social norms? What predicts liberal personal beliefs and social normative expectations among survey respondents? We explore this question by looking at the association between beliefs and norms and various observable household and individual-level characteristics of the respondents. We find that a higher level of education is negatively and significantly correlated with conservative beliefs and social expectations, but only when it comes to norms around women in the public space (Table 8). This suggests that more educated respondents are not more likely than those less educated to hold more liberal beliefs and expectations around men and women’s household roles. However, they do tend to expect lower sanctions for norm transgression overall, compared to their less educated counterparts.33 32 In our sample, 35.05 percent of respondents have adolescent girl(s) in the household, and 26.65 percent of respondents report the presence of other working-age women in the household who are in labor force (excluding those studying). 33 We repeat the same OLS regression separately for women and men (Tables A.13, Annex A). 23 Table 8. Explaining Social Norms (1) (2) (3) (4) (5) PNB Index - PNB Index - SNE Index - SNE Index - Sanctions HH Roles Public Space HH Roles Public Space Index Respondent is female -0.00 -0.03 -0.01 -0.07* 0.02 (0.03) (0.04) (0.02) (0.04) (0.11) Respondent’s age in years, 0.00 -0.00 -0.00 -0.00** -0.00 squared (0.00) (0.00) (0.00) (0.00) (0.00) Years of education, derived -0.01 -0.02* -0.00 -0.02*** -0.08*** from education level (0.01) (0.01) (0.01) (0.01) (0.02) Indicator for respondent -0.01 0.05 -0.04 -0.00 0.33 having child younger than 5 (0.02) (0.04) (0.03) (0.05) (0.23) years Respondent lives in urban area 0.00 0.02 -0.08 -0.08 -0.93* (0.05) (0.13) (0.05) (0.13) (0.48) Log of household's total 0.01* 0.00 0.00 -0.01 0.11 income minus working (0.01) (0.01) (0.01) (0.01) (0.07) female’s earned income (NPR) Madhesh Province 1.09*** 0.68*** 0.97*** 0.82*** -0.28 (0.08) (0.11) (0.06) (0.10) (0.54) Gandaki Province -0.16** -0.31*** -0.23*** -0.28*** -1.65*** (0.06) (0.08) (0.06) (0.09) (0.46) Sudurpashchim Province 0.46*** 0.50*** 0.50*** 0.30** -1.76** (0.08) (0.13) (0.14) (0.14) (0.66) Respondent belongs to 0.13* 0.00 0.20*** 0.09 0.72** Madheshi (non-Dalit) ethnic (0.06) (0.21) (0.05) (0.14) (0.34) group Migrant HH -0.00 -0.03 -0.01 -0.00 -0.52* (0.03) (0.05) (0.03) (0.04) (0.30) At least one among MIL and -0.01 -0.04 -0.00 -0.04 0.02 FIL live in the HH (0.02) (0.05) (0.03) (0.04) (0.19) Constant 0.32** 0.60** 0.63*** 1.21*** 17.16*** (0.14) (0.25) (0.15) (0.24) (0.97) Observations 2,000 2,000 2,000 2,000 2,000 Adjusted R-squared 0.66 0.27 0.63 0.34 0.08 Notes: (1) Higher the outcome variables, higher the conservativeness of beliefs and social expectations and higher the expected sanctions. (2) Reference group for province is Bagmati Province. (3) SEs clustered at the ward level. (4 Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. SEs clustered at the ward level. Contrary to intuition, we find no evidence of respondents in rural areas being more likely to hold conservative beliefs either about norms around women’s household roles, or about women being in the public space as compared to those in urban areas. However, those in rural areas are more likely to expect negative sanctions for breaching a norm. These variables, together, explain between 27 and 66 percent of the variation in the social norms indices, and 8 percent of the variation in the sanctions index. Geographically, respondents from Madhesh and Sudurpashchim Provinces are more likely to hold conservative beliefs and social expectations, as compared to those from Bagmati Province. On the other hand, respondents from Gandaki Province are more likely to be liberal. Relatedly, those who belong to the Madheshi ethnic group tend to be more conservative. These differences are supported by earlier literature showing that Nepal’s socio-cultural practices differ along the lines of castes and ethnicities. For example, Chakravarty et al. (2019) find that in Southern Nepal (Terai and Madhesh), women are more frequently confined to the 24 household and are unable to travel outside of their immediate community for any work. In addition, in conservative parts of the Terai districts, cultural norms prohibit women from interacting with men other than their family members. Within the Hindu-based caste structure, Dalit women experience worse discrimination due to oppressive structures rooted in caste and class, as well as higher incidences of gender-based violence (Whetstone & Luna, 2023). Heterogeneity in association between social norms and women’s work outcomes While there is limited linkage between social norms and women’s work outcome for the average respondent, we explore whether there is heterogeneity in this association for some groups. We explore variations by migration status of the male in the household, ethnicity, locality (urban vs. rural), presence of one of the parents-in-law in the household, and respondent having children under 5 years of age. We do not find evidence of any difference in the effect of change in a person's personal beliefs or social normative expectations on women's participation in paid work based on whether one lives in an urban or rural area, whether at least one among mother-in-law or father-in-law reside in the household, and whether the respondent has young children. We find some variation with ethnicity and migration experience.34 When examining the theme of gender roles within the household, there appears to be no statistically significant differences in the association between women’s work status and PNB and SNE indices across the Madheshi and non-Madheshi ethnic groups. However, for norms around women in the public space theme, conservative personal beliefs may be associated more strongly with women’s work outcomes for non-Madhesi than for Madhesi respondents, though this result does not hold when examining both themes together. On the other hand, conservative social expectations about one’s reference group around norms related to women in the public space do appear to be more strongly associated with Madheshi women's participation in paid work outside the home as compared to non-Madheshi women (Table A.14, Annex A). Similarly, for migrant households (based on migration status of the spouse), no statistically significant differences are observed in the association between women’s work status and PNB and SNE indices for household roles (Table A.15, Annex A). Association between conservative social expectations and negative work outcomes for women do appear to be stronger for migrant households, but this disappears when examining both themes together. When do norms matter? We find that social normative expectations are not strong predictors of women’s work outcomes among respondents of this study. Given the generally low level of social sanctions and liberal social expectations around various domains of gender roles and women in public places, this is not altogether surprising. However, this is not to say that social norms are not important at all in the context of women’s employment in Nepal. In fact, the study does find that varying the circumstances under which employment decisions are made can have a bearing on the level of approval or support for women’s work. To capture the circumstantial variation of social norms on women’s employment decisions, we use two types of vignettes that asked about decisions of a hypothetical woman on whether or not to accept a job offer and personal and social approval of her decisions under specific conditions or circumstances. While regression analysis examines the association of personal beliefs and social expectations with women’s work status, these types of vignettes are tools that allow us to examine which element of social expectations (normative or empirical) appear to be driving women’s employment decisions. In our sample, 15 percent of respondents live in migrant households. Almost 10 percent of respondents identify as 34 Madheshi. 25 The first set of vignettes asked respondents whether Sarita would accept a job offer available to her given varied levels of social expectations (Section 1 presents details). Analysis of these vignettes shows a statistically significant difference between respondents’ approval of Sarita taking up the job opportunity when both SNE and SEE change, indicating that respondents’ approval is conditional upon whether other women in the community are working and whether others in the community approve of women working. Respondents also approve of Sarita working if she expects most (instead of a few) members of her community to believe that it is okay for women to do so, given her empirical expectations remain conservative. However, changing empirical expectations alone does not appear to be sufficient to drive a change in approval (Table A.16, Annex A).35 Similarly, there is evidence that the generally high approval for women’s work outside the home drops once specific conditions are imposed that relate to women’s characteristics and roles and their job’s characteristics. The second set of vignettes explore personal approval and social approval for a woman’s decision to work under varied conditions of the woman and her household, the job characteristics, and circumstances around her husband’s employment status. Figure 5. Do beliefs and expectations of Nepalis around women working outside the home for pay change under certain situations? Figure 5 shows that personal and expected community support drop for women who would have to leave young children with non-family members or when they are unable to keep up with household chores. Support also drops when the job requires travel outside the city, it is physically demanding, and if it is in the hospitality sector, with the latter suggesting that norms around jobs considered “safe” or “suitable to women” are contributing to occupational segregation. On the other hand, personal and social support remains high under 35Each of the 3 scenarios are compared with the base scenario of conservative SEE and SNE using t-tests. The p- values associated with the t-tests are small (p<0.05) when comparing conservative SEE & SNE with (i) liberal SEE & SNE, and (ii) conservative SEE & liberal SNE. The p-value associated with the t-test is not small when comparing conservative SEE & SNE with liberal SEE & conservative SNE. Results when changing both SEE & SNE hold even when taking males only and females only samples. 26 circumstances that challenge the husband’s traditional roles within the household such as inability to provide financially for his family. This is consistent with findings noted earlier that suggest low negative sanctions for households that challenge men’s breadwinner norm, as compared to women’s caregiver norm. It is worth noting that across all circumstances, respondents’ own views are less conservative that than their reference group’s views, although the extent of the divergence is not consistently large across all situations. For example, while 51 percent of respondents approve of paid worker in the hotel/restaurant sectors, they estimate that only 28 percent of others in their reference group would approve. Table 9. Bivariate regression coefficients – Beliefs about whether women should work under different contexts predicting women’s work for pay outside the home Paid work Core elements Full Female Male Circumstances/ measuring Theme sample sample sample contexts social norms (self) (spouse) A woman’s Children <5 looked Respondent 0.08*** 0.14*** 0.01 characteristics after by non-family approval (0.03) (0.03) (0.03) and roles Community 0.12** 0.14*** 0.09 approval (0.04) (0.04) (0.05) Unable to keep up Respondent 0.09*** 0.11*** 0.07** with household tasks approval (0.02) (0.02) (0.03) Community 0.07*** 0.11*** 0.02 approval (0.02) (0.02) (0.04) A woman’s job Occasionally travel to Respondent 0.10*** 0.09*** 0.10*** characteristics work outside the approval (0.02) (0.02) (0.03) village/city Community 0.05* 0.08*** 0.01 approval (0.03) (0.02) (0.04) Job physically Respondent 0.07*** 0.07*** 0.08*** demanding approval (0.02) (0.02) (0.03) Community 0.06* 0.06** 0.05 approval (0.03) (0.03) (0.04) Works in a Respondent 0.09*** 0.10*** 0.08** restaurant/hotel approval (0.02) (0.03) (0.03) Community 0.05 0.08** 0.02 approval (0.04) (0.03) (0.05) Notes: (1) Positive correlation means that approvals are associated with higher likelihood of women working for pay outside the house. (2) Respondent and community approval variables are binary variables coded as 1 if the respondent approves of the behavior. Outcome variable is coded as 1 if the female respondent answered yes to working outside the house for pay. In the case of male respondent, he was asked whether his wife works outside for pay and the variable was coded accordingly. (3) No controls and no regional fixed effects are included. (4) SEs clustered at the ward level. (5) Robust standard errors in parentheses. (6) *** p<0.01, ** p<0.05, * p<0.1. 27 Table 9 shows that women who are more likely to approve or to expect community approval of work under any of the listed potential circumstances are also more likely to work than women who do not approve nor expect community approval under the listed scenarios.36 For example, women who approve of leaving children under five years of age to be looked after by non-family members while working are 14 percentage points more likely to be working than women who do not approve of working under the same circumstance. When it comes to job characteristics, women’s own personal approval seems to be more closely correlated to their work outcomes compared to their perception of community approval, in line with earlier findings. For men, it is only their personal approval of a woman working under the listed set of circumstances that is associated with women’s work outcomes. The exception is leaving children younger than five with non-family members. Approval of women’s work under specific circumstances also varies between several groups (reported in Tables A.18, Annex A). Findings reveal that spouses of migrant men are less likely to approve of a woman working outside the home for pay if she works further away from home or travels for work. The absence of husbands from home may contribute to making the wives feel more responsible toward being nearer to their homes. SECTION 6: DISCUSSION AND CONCLUSION Whether and the extent to which social norms matter for women’s labor force participation has been shown to vary by context. This paper is the first to provide rigorous evidence on these relationships in Nepal. Table 10 summarizes the paper’s findings. Our findings show that, overall, Nepali men and women hold relatively supportive views and social expectations around women working outside the home for pay, despite actual FLFP being low in Nepal and among the survey respondents. This is especially true for norms related to women’s mobility and presence in public spaces, where beliefs and social expectations are generally very liberal, with low expected sanctions, and not associated with women’s employment outcomes. In comparison, we find more restrictive views and expectations of others when it comes to traditional gender roles within the household, in particular around the motherhood norm. We find that views on spousal roles in the households (male breadwinner norm and female caregiver norm) are associated with lower employment for women, suggesting that these norms are relevant for women’s labor market decisions. However, beliefs and expectations around motherhood, which are the most restrictive, are not associated with employment outcomes. Indeed, we find that for the most part, personal beliefs are statistically significant predictors of women’s employment outcomes, but normative expectations are not. This suggests that factors other than norms are likely to be stronger determinants of Nepali women’s decisions to work for pay outside the home. Further analysis shows that when rational factors related to employment – such as educational attainment, household income, other working women within the household, presence of adolescent girls in the household, and awareness regarding job search methods – are considered, only personal beliefs about household roles are associated with women’s work outcomes, confirming further that normative expectations are not binding constraints. 36 Table A.17, Annex A includes results for all contexts surveyed. 28 Table 10. A snapshot of the key findings Association with women's decision to work outside the home for pay (Full sample) Beliefs and social expectations ↓* Personal beliefs Female caregiver norm (-) Social normative expectations ↓ if woman unable to keep up with household tasks ↓** Personal beliefs Male breadwinner norm (-) Social normative expectations Motherhood norm (-) ↓ if childcare by non-family Working women norm ↓ for job characteristics: travel outside village/city, physically demanding, hospitality sector Norms on: Mobility freedom, Honor, Safety, (-) Mixed-gender workplace. Rational determinants ↓*** if other working-age women are present in HH ↓** with increase in HH income (less female income) ↑* with presence of adolescent girls ↑** with increase in education (-) awareness of job-search methods Impact by sub-groups Increases for Madhesi women and for women with migrant spouses (-) Presence of children <5, presence of in-laws, living in an urban area Notes: (-) not significant; ↑ significant positive association; ↓ significant negative association (*** p<0.01, ** p<0.05, * p<0.1). Text in italics refers to personal beliefs and social expectations for norms under specific circumstances/contexts. Associations reported in this table indicate overall results, refer to Table 6 above and Tables A.12, Annex A for details and results separately for male and female samples. Each variable and its association with the outcome variable should be interpreted independently and in conjunction with messages highlighted in this section. Given that personal beliefs and social expectations are generally not very restrictive among the study respondents, it is likely that a norm tipping point has been reached when it comes to FLFP in Nepal, where a critical share of the community has eased up on enforcing a traditional norm- as suggested by low sanctions- or adopted new, more supportive norms towards FLFP. Bicchieri (2017) suggests that norm change gains momentum when about a third of the population abandons an old norm, encouraging moderately norm- sensitive people to reconsider their behavior, with the tipping point occurring once more than half the group deviates. In the case of Nepal, further research is required to unpack when and how norms have shifted over time, and what the enabling factors might have been. World Bank (2022) finds that in contrast to Nepal, progressive attitudes towards gender equality have either progressed slowly or even regressed over time for most South Asian countries, although women and higher educated people tend to have less restrictive attitudes. While there is limited linkage between social norms and women’s work outcome for the average respondent, there appears to be some heterogeneity in this association, particularly by ethnicity (Madheshi vs. non- Madheshi) and migration status of the spouse. Heterogeneity in normative expectations across groups and geographies is important to note as social norms may be more influential for some, depending on the extent of restrictive views expected in their respective reference groups. This is also validated by the vignette analysis, which shows that in extreme hypothetical cases (where most people are conservative in their beliefs or behaviors), social norms can be binding constraints to women’s FLFP decisions. This, however, seems unlikely in the context of Nepal given our findings. 29 Support for women’s work also tends to vary depending on certain circumstances, in particular, when it comes to delegating childcare to non-family members or being unable to keep up with household chores. Support also drops depending on women’s job’s characteristics, especially if the job requires travel outside the city or village of residence, if it is physically demanding, and if it is in the hospitality sector. There is some evidence of pluralistic ignorance with regard to these beliefs and expectations, however, as normative expectations are generally more restrictive than personal beliefs. Pluralistic ignorance is a common conformity trap around the tipping point, wherein behavior may be driven by misperceived social expectations that are more conservative than one’s own beliefs. When this is the case, disclosing strong but hidden social support could be a direct intervention strategy to correct these norm perceptions and encourage behavioral change (Bursztyn et al., 2020). However, when both beliefs and social expectations are restrictive, as is the case with leaving children under 5 years with non-family members, norm-transformative interventions may be useful to consider. For example, a new norm can be created by reframing childcare around the value of investing in early childhood development for better prospects in the future. It is also important to consider policy interventions that indirectly address normative barriers around the motherhood norm and appropriate jobs for women. For example, interventions that address supply-side challenges to childcare, such as lack of affordable, accessible, and quality options are important given that care options beyond immediate family appear to be a barrier to women. Demand-side constraints as evidenced by low personal and community support for leaving children with non-family members also matter, which may necessitate information and awareness campaigns about the benefits of early childhood development centers. Similarly, interventions to increase workplace safety, improve access to affordable and safe transport, and promote female role models may support women’s work outside female-dominated sectors and in cases where the job requires commuting. Lastly, longer-term interventions that can help adjust beliefs and practices related to the husband’s own involvement in household work would be beneficial to improve young mothers’ work- life balance and allow them to participate in productive employment. Overall, the combination of relatively high support for and low normative restrictions around women working outside the home in Nepal creates conditions that are ready for a huge push in FLFP. Policy interventions should consider addressing both structural barriers and conformity traps around motherhood norms and appropriate jobs for women. On the former, in particular, a deeper dive into types of services, preferences, and social expectations around childcare is necessary to gain more insights on adequate policies and framing of childcare services that can allow mothers to avail such services without triggering social sanctions, thereby creating an enabling environment for them to pursue employment. Lastly, more research is needed to understand how social norms interventions can be tackled in combination with other non-norm barriers that address limited labor demand and improved employability and agency among women. 30 REFERENCES Ahmed, Sarah. "Women left behind: Migration, agency, and the Pakistani woman." Gender & Society 34, no. 4 (2020): 597- 619. Bicchieri, Cristina, Jan W. Lindemans, and Ting Jiang. "A structured approach to a diagnostic of collective practices." Frontiers in psychology 5 (2014): 1418. Bicchieri, Christina. “Norms in the wild: How to diagnose, measure, and change social norms.” Oxford University Press (2017). Bernhardt, Arielle, Erica Field, Rohini Pande, Natalia Rigol, Simone Schaner, and Charity Troyer-Moore, “Male Social Status and Women’s Work.” AEA Papers and Proceedings, (2018), 108, 363–67 Bursztyn, Leonardo, Alessandra L. González, and David Yanagizawa-Drott. "Misperceived social norms: Women working outside the home in Saudi Arabia." American economic review 110, no. 10 (2020): 2997-3029. Bushell, Brenda. "Women entrepreneurs in Nepal: what prevents them from leading the sector?" Gender & Development 16, no. 3 (2008): 549-564. Buchmann, Nina, Carl Meyer, and Colin D. Sullivan. Paternalistic discrimination. Working Paper, 2023. Bussolo, Maurizio, Jessy Amarachi Ezebuihe, Ana Maria Muñoz Boudet, Stavros Poupakis, Tasmia Rahman, and Nayantara Sarma. "Social Norms and Gender Disparities with a Focus on Female Labor Force Participation in South Asia." The World Bank Research Observer 39, no. 1 (2024): 124-158. Bussolo, Maurizio, Ana Maria Munoz Boudet, Chinmaya Holla, Thomas de Hoop, Tasmia Rahman, Garima Siwach, Nele Warrinnier. “Measuring Gender Norms in the Field”. The World Bank (2024, Forthcoming). Central Bureau of Statistics (CBS). "Report on the Nepal labour force survey 2017/18." (2019). Chakravarty, Shubha, Mattias Lundberg, Plamen Nikolov, and Juliane Zenker. "Vocational training programs and youth labor market outcomes: Evidence from Nepal." Journal of Development Economics 136 (2019): 71-110. Chicoine, Luke. "Free primary education, fertility, and women’s access to the labor market: Evidence from Ethiopia." The World Bank Economic Review 35, no. 2 (2021): 480-498. Crowne, Douglas P., and David Marlowe. "A new scale of social desirability independent of psychopathology." Journal of consulting psychology 24, no. 4 (1960): 349. Cislaghi, Ben, and Lori Heise. " Measuring Gender-related Social Norms: Report of a Meeting, Baltimore Maryland.” Learning Group on Social Norms and Gender-based Violence of the London School of Hygiene and Tropical Medicine" (2016): 44. Deshpande, Ashwini, and Naila Kabeer. "Norms that matter: Exploring the distribution of women’s work between income generation, expenditure-saving and unpaid domestic responsibilities in India." World Development 174 (2024): 106435. Deshpande, Ashwini, and Jitendra Singh. “Dropping out, being pushed out or can’t get in? Decoding declining labour force participation of Indian women.” No. 14639. IZA Discussion Papers, 2021. Dhar, Diva, Tarun Jain, and Seema Jayachandran. "Reshaping adolescents' gender attitudes: Evidence from a school-based experiment in India." American economic review 112, no. 3 (2022): 899-927. Field, Erica, and Kate Vyborny. "Women's mobility and labor supply: experimental evidence from Pakistan." Asian Development Bank Economics Working Paper Series 655 (2022). Gauri, Varun, Tasmia Rahman, and Iman Kalyan Sen. "Measuring social norms about female labor force participation in Jordan." World Bank Policy Research Working Paper 8916 (2019). Ghimire, Dirgha, Yang Zhang, and Nathalie Williams. "Husbands’ migration: increased burden on or more autonomy for wives left behind?" Journal of Ethnic and Migration Studies 47, no. 1 (2021): 227- 248. Goldin Claudia. “The U-Shaped Female Labor Force Function in Economic Development and Economic History .” Schultz TP Investment in Women’s Human Capital and Economic Development. University of Chicago Press, 61-90 (1995). Goldstein, Marcus, Paula Gonzalez, Talip Kilic, Sreelakshmi Papineni, and Philip Randolph Wollburg. "Breadwinners and 31 Caregivers: Examining the Global Relationship between Gender Norms and Economic Behavior." World Bank Policy Research Working Paper 10692 (2024). Hainmueller, Jens, Dominik Hangartner, and Teppei Yamamoto. "Validating vignette and conjoint survey experiments against real-world behavior." Proceedings of the National Academy of Sciences 112.8 pp.2395-2400 (2015). Halim, D, E Perova, and S Reynolds. “Childcare and mothers’ labor market outcomes in lower-and middle- income countries,” The World Bank Research Observer, 38: 73–114. (2023a). Halim, Daniel, Michael B. O’Sullivan, and Abhilasha Sahay. "Increasing Female Labor Force Participation." World Bank Group Gender Thematic Policy Notes Series; Evidence and Practice Note (2023b). Harper, Caroline; Nicola Jones, Anita Ghimire, Rachel Marcus, and Grace Kyomuhendo Bantebya (Eds.) “Empowering Adolescent Girls in Developing Countries: Gender Justice and Norm Change (1st ed.).” Routledge (2018). Holland, Jeremy; Rishal Poonam. “Measuring Social Norm Change Through Storytelling. Baseline Report.” UN Women (2023). International Monetary Fund. 2020. Nepal Select Issues: April. IMF Country Report No. 20/97. Jayachandran, Seema. "Social norms as a barrier to women’s employment in developing countries." IMF Economic Review 69, no. 3 (2021): 576 595. Klasen, Stephan. "What explains uneven female labor force participation levels and trends in developing countries?" The World Bank Research Observer 34, no. 2 (2019): 161-197. Lokshin, Michael, and Elena Glinskaya. "The effect of male migration on employment patterns of women in Nepal." The World Bank Economic Review 23, no. 3 (2009): 481-507. Ministry of Health and Population [Nepal], New ERA, and ICF. “Nepal Demographic and Health Survey 2016.” (2017). Ministry of Health and Population [Nepal], New ERA, and ICF. “Nepal Demographic and Health Survey 2022.” (2023). Ministry of Labor, Employment, and Social Security [Nepal]. “Nepal Labor Migration Report 2020.” (2020). Muñoz Boudet, Ana María, Tasmia Rahman, Nour Nasr, and Abigail Dalton. "Addressing Social and Gender Norms to Promote Gender Equality." World Bank Group Gender Thematic Policy Notes Series: Evidence and Practice Note (2023). Nepal Rashtra Bank. “Fifth Household Budget Survey Nepal 2014/15.” (2016). Paudel, Anil. "Girls' Transitions to Work through Higher-Quality TVET Programs in Nepal. Echidna Global Scholars Program, Policy Paper." Center for Universal Education at The Brookings Institution (2019). Peet, Evan D., Günther Fink, and Wafaie Fawzi. "Returns to education in developing countries: Evidence from the living standards and measurement study surveys." Economics of education review 49 (2015): 69-90. Phadera, Lokendra. “International migration and its effect on labor supply of the left-behind household members: Evidence from Nepal.” No. 333-2016-14201. Annual Meeting, July 31-August 2, Boston, Massachusetts, Agricultural and Applied Economics Association (2016). Raju, Dhushyanth, and Jasmine Rajbhandary, eds. “Youth employment in Nepal.” The World Bank, (2018). Ram Mohan, Rohini, Ranjitha Puskur, D. Chandrasekar, and Harold Glenn A. Valera. "Do gender dynamics in intra- household decision making shift with male migration? Evidence from rice-farming households in Eastern India." Gender, Technology and Development 27, no. 2 (2023): 157-183. Ruppert Bulmer, Elizabeth, Ami Shrestha, and Michelle Marshalian. "Nepal Jobs Diagnostic." World Bank Group (2020). Ruppert Bulmer, Elizabeth. “Job Quality in Nepal is Improving, but Women Lag Behind.” World Bank Jobs and Development Blog (2020). https://blogs.worldbank.org/jobs/job-quality-nepal-improving- women-lag-behind Samman, E. “Quantitative measurement of gendered social norms.” London: ALiGN (2019). Kathmandu University, School of Arts; Interdisciplinary Analysts, and The Asia Foundation. “A Survey of the Nepali People in 2018.” (2019). School of Arts, Kathmandu University, Interdisciplinary Analysts, and The Asia Foundation. “A Survey of the Nepali People in 2020.” (2020). 32 Sen, I. Zeina Afif, Varun Gauri, and Gohdar Mohamed. "Women’s Labor Force Participation in the Kurdistan Region of Iraq." (2022). Shrestha, Maheshwor. "The impact of large-scale migration on poverty, expenditures, and labor market outcomes in Nepal." World Bank Policy Research Working Paper 8232 (2017). USAID. “Advocacy for the promotion of women’s entrepreneurship in Nepal. Study Brief.” (2019). World Bank. “Reshaping Norms: A New Way Forward”. (2022). South Asia Economic Focus (April), World Bank, Washington, DC. 33 Annex A: Tables Table A.1. Reported reasons among inactive Nepali women for not trying to find a job or start a business in 30 days preceding the survey Reason Share of females (%) Waiting for results 0.9 Waiting for season to work 2.9 Waiting to be recalled to past job 0.1 Tired of looking for work 4.8 No jobs matching skills, lacks experience 21.4 Considered too young/old by employers 0.7 Look for oversees employment 0.6 In studies/ training 10.1 Family/ household responsibilities 34.3 Family member made obstacle 2.3 Engaged in a subsidiary agriculture 16.2 Have other sources of income 0.6 Disability, injury, illness 2.1 Other 3.2 Source: Authors’ compilation using Nepal Labor Force Survey, 2017-2018 Table A.2. Employment type for Nepali men and women Employment Type Share of males (%) Share of females (%) Wage Formal 16.2 14.8 Wage Informal 45.2 26.4 Unpaid Family Worker 6.7 27.6 Self, Agriculture 7.1 10.8 Self, Non-Agriculture 24.8 20.4 Source: Authors’ compilation using Nepal Labor Force Survey, 2017-2018 34 Table A.3. Questions included in the quantitative tool under “gender roles within the home” theme Core element Definition Example question Individual Behavior Behavior - What I do In my household, financially providing for the family is solely my spouse's responsibility. Personal Normative Attitude - What I believe one should do To what extent to you agree with the following Beliefs statement: Financially providing for the family should solely be the man’s responsibility. Social Normative What I believe others in my reference Think about the people whose opinions matter Expectations group approve or disapprove of me to you or your family: How many out of 10 such doing; based on second-order beliefs of people believe that financially providing for the what others in my reference group family should solely be the believe. man’s responsibility? Social Empirical What I believe others in my reference Think about the people whose opinions matter Expectations group do; based on my observations of to you or your family: How many out of 10 such how others in my reference group behave people live in households where financially providing for the family is the man’s responsibility only? Consequences of digressing from or Think about the people whose opinions matter Sanctions/Rewards benefits of adhering to prevailing social to you or your family: What would people say norms about households where the men are not solely responsible for financially providing for the family? 35 Table A.4. Household Demographic Characteristics Variable Full Sample Mean/(SE) N 1,150 Total number of HH members 4.25 (1.60) Total number of minors (under 18 years) 1.64 (1.18) Total number seniors (above 60 years) 0.27 (0.57) Total number of disabled HH members 0.05 (0.22) Number of respondent's children under 5 years 0.36 (0.58) Number of respondent's children under 18 years 1.55 (1.15) Number of working-age (>=18 & <=59) HH members 2.30 (0.99) Number of female-working-age (>=18 & <=59) HH members 1.27 (0.51) Number of male-working-age (>=18 & <=59) HH members 1.03 (0.72) Indicator for a male member as head of HH 0.71 (0.46) Mother-in-law lives in the HH 0.28 (0.45) Father-in-law lives in the HH 0.23 (0.42) At least one among MIL and FIL live in the HH 0.32 (0.47) Hindu 0.86 (0.35) Buddhist 0.09 (0.28) Muslim (religion) 0.02 (0.13) Christian 0.03 (0.18) Khas & Aryan 0.29 (0.46) 36 Adibasi & Janajati 0.33 (0.47) Madhesi (Non dalit) 0.10 (0.29) Dalit 0.27 (0.44) Muslim (ethnicity) 0.02 (0.13) Rural 0.26 (0.44) Mountain 0.05 (0.22) Hill 0.47 (0.50) Terai 0.48 (0.50) Table A.5 Labor Market Outcomes of Household Working Age Members Variable (in percent) Full Sample Male Female Mean Mean Mean Share of HH members who are of working age 57.78 52.97 67.18 Share of HH members who are of working age in labor force 67.78 86.66 57.83 Share of HH members who are of working age out of labor force 32.22 13.34 42.17 Out of those in labor force, share of those who are: Self-employed/own-account worker (no employees) 29.72 25.23 38.44 Business owner/employer with at least 1 employee 0.45 0.53 0.27 Salaried/wage employee 45.63 58.82 26.40 Army 0.10 0.24 0.00 Student who also works 0.28 0.29 0.20 Unpaid worker in the family business 3.63 3.67 4.47 Paid worker in the family business 3.13 2.81 3.93 Does not work but is looking for work 17.06 8.41 26.29 Note: Working age is defined as 18-59 years. 37 Table A.6. Household Sources of Income Variable N Full Sample Mean/(SE) Earned income, reported by female 1,150 0.80 (0.40) International remittances, reported by female 1,150 0.26 (0.44) Domestic remittances, reported by female 1,150 0.03 (0.16) Other transfers from family or friends, reported by female 1,150 0.02 (0.12) Safety nets, reported by female 1,150 0.06 (0.23) Savings or selling assets, reported by female 1,150 0.10 (0.30) Loan, reported by female 1,150 0.13 (0.34) Earned income amount, reported by female 923 23,454.93 (12,060.22) International remittances amount, reported by female 299 39,802.68 (21,085.27) Domestic remittances amount, reported by female 31 16,500.00 (8,630.37) Other transfers from family or friends amount, reported by female 18 23,166.67 (17,161.00) Safety nets amount, reported by female 65 8,652.31 (7,021.51) Savings or selling assets amount, reported by female 118 11,248.31 (13,535.7) Loan amount, reported by female 149 54,910.07 (118,200.22) Total income of HH reported by female 1,150 38,738.87 (49,166.48) Note: An observation from international remittances was deemed as an outlier and excluded from the calculations above, as the sum reported by that respondent was a lump-sum as opposed to the monthly amount reported for all. 38 Table A.7. Respondents’ Demographic and Labor Market Outcomes Variable Male Female Full Sample Mean/(SE) Mean/(SE) Mean/(SE) N 850 1,150 2,000 Age (in years) 38.35 33.67 35.65 (0.31) (0.25) (8.86) Work last 30 days 0.88 0.45 0.63 (0.01) (0.02) (0.48) Work last 7 days 0.87 0.44 0.62 (0.01) (0.02) (0.49) Ever migrated (self - asked only to males) 0.14 NA NA (0.35) NA NA Spouse ever migrated (asked only to females) NA 0.37 NA NA (0.48) NA N (only if worked in last 30 days) 751 515 1,266 Paid work last 30 days 0.99 0.98 0.98 (0.00) (0.01) (0.12) N (only if worked in last 7 days) 735 500 1,235 Paid work last 7 days 0.98 0.98 0.98 (0.01) (0.01) (0.14) N (only if not worked in last 30 days) 99 635 734 Ever worked 0.59 0.17 0.23 (0.05) (0.02) (0.42) N (only if paid in cash; not kind) 789 605 1,394 Gross monthly income from paid work 20,070.85 13,235.04 17,104.09 (350.54) (362.43) (10,039.28) N (only if worked in past 30 days or ever worked) 809 623 1,432 Weekly work hours 53.53 47.68 50.99 (0.53) (0.72) (16.56) Work inside the house 0.18 0.41 0.28 (0.01) (0.02) (0.45) Has a written contact 0.13 0.17 0.15 (0.01) (0.02) (0.36) Employer paid social security contributions 0.08 0.08 0.08 (0.01) (0.02) (0.27) Own-account worker, without employees 0.26 0.45 0.34 (0.02) (0.02) (0.48) Employer, with at least 1 employee 0.01 0.01 0.01 (0.00) (0.00) (0.08) 39 Wage employee 0.46 0.31 0.39 (0.02) (0.02) (0.49) Army 0.01 0.00 0.00 (0.00) (0.00) (0.06) Unpaid helper in family business and/or farms 0.03 0.04 0.04 (0.01) (0.01) (0.19) Paid helper in family business and/or farms 0.03 0.05 0.04 (0.01) (0.01) (0.19) Casual laborer (Y/N) 0.20 0.15 0.18 (0.01) (0.01) (0.39) Work in agriculture 0.14 0.25 0.19 (0.01) (0.02) (0.39) Table A.8. Aspirations of respondents about their daughters’ and daughters-in-laws’ work outcomes Aspiration statement Share of respondents who agree (Mean/SE) I would want my daughter to work outside the house and earn a living. 0.87 (0.34) I would want my daughter to work outside the house and earn a living, 0.77 even when she is married. (0.42) I would want my daughter to work outside the house and earn a 0.64 living, even when she has school-going children. (0.48) I would want my daughter in law to work outside the house and earn a 0.82 living. (0.38) I would want my daughter in law to work outside the house and earn a 0.65 living, even when she has school-going children. (0.48) 40 Table A.9. Bivariate regression coefficients – Beliefs about gender roles within the home and women in public spaces predicting women’s work for pay outside the home Theme Norm Core Paid work Paid work Paid work elements (Full sample) (self) (Female- (Female measuring only sample) spouse) social norms (Male-only sample) A woman’s primary PNB -0.10** -0.14*** -0.04 responsibility is taking care (0.04) (0.04) (0.03) of HH SNE -0.10* -0.14** -0.03 (0.05) (0.06) (0.05) SEE -0.09* -0.14** -0.03 (0.05) (0.05) (0.05) Sanc. -0.02 -0.04 0.00 (0.03) (0.03) (0.03) Financially providing is solely a PNB -0.13*** -0.17*** -0.07* Gender roles in man's responsibility (0.04) (0.04) (0.04) the household SNE -0.12** -0.18*** -0.03 (0.05) (0.05) (0.05) SEE -0.10* -0.15** -0.05 (0.06) (0.05) (0.07) Sanc. -0.08** -0.07** -0.09* (0.03) (0.03) (0.04) Women should not / do not PNB -0.05 -0.05 -0.05 leave children in the care of (0.04) (0.05) (0.04) someone else when they go SNE -0.05 -0.07 -0.04 to work or to look for work (0.05) (0.06) (0.05) SEE 0.03 0.03 0.02 (0.06) (0.06) (0.06) Sanc. 0.01 0.00 0.01 (0.03) (0.03) (0.03) Women should not / do not PNB -0.15** -0.19** -0.12* work for pay outside the (0.06) (0.07) (0.06) house SNE -0.16** -0.24*** -0.06 (0.07) (0.06) (0.08) SEE -0.21*** -0.20*** -0.22** (0.07) (0.07) (0.08) Sanc. -0.05 -0.03 -0.07 (0.04) (0.05) (0.05) Women should not / do not PNB -0.08 -0.10 -0.06 leave the house go to places (0.08) (0.10) (0.08) Women in the such as the market SNE -0.04 -0.11 0.05 public space unaccompanied by a family (0.08) (0.10) (0.07) member SEE -0.05 -0.07 -0.02 (0.05) (0.04) (0.06) Sanc. -0.06 -0.12*** 0.04 (0.04) (0.04) (0.05) Women going outside brings PNB -0.15* -0.20** -0.08 (0.08) (0.09) (0.08) 41 dishonor to families SNE -0.11 -0.15 -0.06 (0.09) (0.10) (0.09) SEE -0.04 -0.06 -0.06 (0.10) (0.09) (0.13) Women going outside PNB -0.13* -0.22*** -0.03 exposes them to harassment (0.07) (0.07) (0.08) SNE -0.12 -0.12 -0.12 (0.10) (0.10) (0.10) SEE -0.08 -0.06 -0.09 (0.09) (0.09) (0.11) Women should not / do not PNB -0.09 -0.11 -0.06 work in gender-mixed (0.07) (0.07) (0.10) environment SNE -0.11 -0.14* -0.07 (0.07) (0.07) (0.08) SEE -0.18*** -0.22*** -0.14* (0.05) (0.04) (0.07) Sanc. -0.02 -0.0370 0.01 (0.03) (0.03) (0.03) Notes: (1) Negative correlation means that more conservative beliefs are associated with lower probability of women working for pay outside the house. (2) All items on attitudes/beliefs are scaled from 0 to 1, with 0 being the most liberal and 1 being the most conservative. Outcome variable is binary with 1 if the female reported a "yes" to "I work outside the house for pay." and if a male reported "yes" to "My wife works outside the house for pay." (3) No controls and no regional fixed effects are included. (4) SEs clustered at the ward level. (5) Robust standard errors in parentheses. (6) *** p<0.01, ** p<0.05, * p<0.1. Table A.10. Individual-level regressions of sanctions index predicting women’s work outcomes Full sample Females only Males only (1) (2) (3) (4) (5) (6) Paid work Paid work Paid work Paid work Paid work Paid work Sanc. Index -0.01 -0.01* -0.01* -0.01** -0.01 -0.01 (0.01) (0.01) (0.00) (0.00) (0.01) (0.01) Constant -0.00 0.47*** (0.09) 0.02 0.50*** -0.06 0.44*** (0.30) (0.30) (0.08) (0.30) (0.12) Obs. 2,000 2,000 1,150 1,150 850 850 Adjusted R2 0.02 0.01 0.03 0.01 0.01 0.00 Controls YES NO YES NO YES NO Regional FE YES NO YES NO YES NO Notes: (1) Negative correlation means that more conservative beliefs and social expectations are associated with lower probability of women working for pay outside the house. (2) Higher the sanctions index, higher the perceived social sanctions. Outcome variable “Paid work” is binary with 1 if the female reported a “yes” to “I work outside the house for pay.” And if a male reported “yes” to “My wife works outside the house for pay.” (3) Standard errors clustered at the ward level. (4) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 42 Table A.11. Individual-level regressions of social norms indices predicting women’s work outcomes, without controls Full sample Females only Males only (1) (2) (3) (4) (5) (6) (7) (8) (9) Paid Paid Paid Paid Paid Paid Paid Paid Paid work work work work work work work work work PNB Index – HH -0.05* -0.05* -0.05 -0.04 -0.05 -0.05 Roles (0.03) (0.03) (0.03) (0.03) (0.04) (0.04) SNE Index – HH -0.01 -0.01 -0.04 -0.04 0.03 0.03 Roles (0.05) (0.04) (0.05) (0.05) (0.06) (0.06) PNB Index – -0.02 -0.00 -0.02 -0.01 -0.02 -0.01 Public Space (0.03) (0.03) (0.03) (0.03) (0.03) (0.04) SNE Index – -0.01 -0.00 -0.03 0.00 0.00 -0.00 Public Space (0.03) (0.04) (0.03) (0.04) (0.04) (0.04) Constant 0.35*** 0.33*** 0.35*** 0.37*** 0.34*** 0.37*** 0.33*** 0.33*** 0.33*** (0.03) (0.02) (0.03) (0.03) (0.02) (0.03) (0.03) (0.02) (0.03) Obs. 2,000 2,000 2,000 1,150 1,150 1,150 850 850 850 Adjusted R2 0.01 0.00 0.01 0.01 0.00 0.01 -0.00 -0.00 -0.00 Controls NO NO NO NO NO NO NO NO NO Regional FE NO NO NO NO NO NO NO NO NO Notes: (1) Negative correlation means that more conservative beliefs and social expectations are associated with lower probability of women working for pay outside the house. (2) For all indices, higher the value, more conservative the beliefs and expectations. Outcome variable “Paid work” is binary with 1 if the female reported a “yes” to “I work outside the house for pay.” And if a male reported “yes” to “My wife works outside the house for pay.” (3) Standard errors clustered at the ward level. (4) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 43 Tables A.12 Individual-level regressions of norms within each index predicting women’s work outcomes Full Sample (1) (2) (3) (4) (5) (6) Paid work Paid work Paid work Paid work Paid work Paid work PNB – Working woman -0.11 (0.09) SNE – Working woman -0.11 (0.11) PNB – Mobility freedom -0.10 (0.11) SNE – Mobility freedom 0.07 (0.10) PNB – Honor -0.13 (0.11) SNE – Honor 0.01 (0.11) PNB – Safety -0.09 (0.09) SNE – Safety -0.01 (0.10) PNB – Motherhood -0.03 (0.04) SNE – Motherhood -0.01 (0.05) PNB – Mixed-gender -0.02 workplace (0.08) SNE – Mixed-gender -0.05 workplace (0.09) Constant -0.09 -0.14 -0.13 -0.13 -0.11 -0.12 (0.24) (0.24) (0.25) (0.24) (0.25) (0.23) Obs. 2,000 2,000 2,000 2,000 2,000 2,000 Adjusted R2 0.02 0.02 0.02 0.02 0.02 0.02 Controls YES YES YES YES YES YES Regional FE YES YES YES YES YES YES Notes: (1) Negative correlation means that more conservative beliefs and social expectations are associated with lower probability of women working for pay outside the house. (2) For all items, the higher the value, the more conservative the beliefs and expectations are. Outcome variable “Paid work” is binary with 1 if the female reported a “yes” to “I work outside the house for pay.” And if a male reported “yes” to “My wife works outside the house for pay.” (3) Standard errors clustered at the ward level. (4) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.10. (5) The results for female caregiver and male breadwinner norms are reported in the main body of the paper (Table 6). 44 Females only (1) (2) (3) (4) (5) (6) Paid work Paid work Paid work Paid work Paid work Paid work PNB – Working woman -0.07 (0.11) SNE – Working woman -0.20* (0.10) PNB – Mobility freedom -0.04 (0.13) SNE – Mobility freedom -0.00 (0.12) PNB – Honor -0.18 (0.12) SNE – Honor 0.04 (0.11) PNB – Safety -0.22** (0.08) SNE – Safety 0.08 (0.11) PNB – Motherhood -0.02 (0.05) SNE – Motherhood -0.04 (0.06) PNB – Mixed-gender -0.01 workplace (0.09) SNE – Mixed-gender -0.05 workplace (0.11) Constant -0.09 -0.14 -0.13 -0.14 -0.10 -0.13 (0.25) (0.25) (0.25) (0.24) (0.25) (0.24) Obs. 1,150 1,150 1,150 1,150 1,150 1,150 Adjusted R2 0.03 0.02 0.02 0.02 0.02 0.02 Controls YES YES YES YES YES YES Regional FE YES YES YES YES YES YES 45 Males only (1) (2) (3) (4) (5) (6) Paid work Paid work Paid work Paid work Paid work Paid work PNB – Working woman -0.17 (0.11) SNE – Working woman 0.02 (0.15) PNB – Mobility freedom -0.18 (0.12) SNE – Mobility freedom 0.18 (0.13) PNB – Honor -0.05 (0.12) SNE – Honor -0.04 (0.14) PNB – Safety 0.10 (0.12) SNE – Safety -0.17 (0.13) PNB – Motherhood -0.04 (0.06) SNE – Motherhood 0.02 (0.06) PNB – Mixed-gender -0.02 workplace (0.16) SNE – Mixed-gender -0.06 workplace (0.13) Constant -0.12 -0.16 -0.14 -0.12 -0.15 -0.13 (0.25) (0.25) (0.25) (0.25) (0.26) (0.24) Obs. 850 850 850 850 850 850 Adjusted R2 0.01 0.01 0.00 0.01 0.00 0.00 Controls YES YES YES YES YES YES Regional FE YES YES YES YES YES YES 46 Tables A.13. Explaining conservative social norms, by gender Table 13.1. Explaining Social Norms – Females only sample (1) (2) (3) (4) (5) PNB Index - PNB Index - SNE Index - SNE Index - Sanctions HH Roles Public Space HH Roles Public Space Index Respondent’s age in years, 0.00 -0.00 -0.00 -0.00*** (0.00) -0.00 squared (0.00) (0.00) (0.00) (0.00) Years of education, derived from -0.01** -0.02*** -0.01 -0.02*** (0.01) -0.11*** education level (0.01) (0.01) (0.01) (0.03) Indicator for respondent having -0.02 0.00 -0.05 -0.02 0.23 child younger than 5 years (0.02) (0.03) (0.03) (0.04) (0.17) Respondent lives in urban area 0.01 0.04 -0.07 -0.11 -0.92* (0.05) (0.15) (0.05) (0.14) (0.48) Log of household's total income 0.02*** 0.01 0.01 -0.00 0.12* minus working female’s earned (0.00) (0.01) (0.01) (0.01) (0.07) income (NPR) Madhesh Province 1.07*** 0.64*** 0.96*** 0.78*** (0.10) -0.30 (0.08) (0.11) (0.07) (0.62) Gandaki Province -0.14** -0.29*** -0.22*** -0.25** (0.09) -1.42*** (0.07) (0.08) (0.06) (0.49) Sudurpashchim Province 0.43*** 0.54*** 0.50*** 0.34** -1.54** (0.08) (0.12) (0.13) (0.15) (0.55) Respondent belongs to Madheshi 0.16*** -0.04 0.22*** 0.09 0.96** (non-Dalit) ethnic (0.05) (0.2) (0.06) (0.15) (0.40) group Migrant HH -0.01 -0.03 -0.02 -0.00 -0.57* (0.03) (0.05) (0.03) (0.04) (0.31) At least one among MIL and FIL -0.00 -0.02 0.00 -0.02 0.04 live in the HH (0.02) (0.06) (0.03) (0.04) (0.19) Constant 0.30** 0.61*** 0.65*** 1.18*** 17.33*** (0.11) (0.19) (0.12) (0.20) (0.81) Observations 1,150 1,150 1,150 1,150 1,150 Adjusted R-squared 0.67 0.28 0.65 0.34 0.08 Notes: (1) Reference group for province is Bagmati Province. (2) SEs clustered at the ward level. (3) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. SEs clustered at the ward level. 47 Table 13.2 Explaining Social Norms – Males only sample (1) (2) (3) (4) (5) PNB Index - PNB Index - SNE Index - SNE Index - Sanctions HH Roles Public Space HH Roles Public Space Index Respondent’s age in years, 0.00 0.00 0.00 -0.00* -0.00 squared (0.00) (0.00) (0.00) (0.00) (0.00) Years of education, derived from -0.00 -0.01 0.00 -0.02* (0.01) -0.05 education level (0.01) (0.01) (0.01) (0.03) Indicator for respondent having -0.01 0.10* (0.06) -0.04 0.01 0.40 child younger than 5 years (0.02) (0.04) (0.06) (0.33) Respondent lives in urban area -0.01 -0.01 -0.09 -0.06 -0.96* (0.05) (0.12) (0.05) (0.14) (0.49) Log of household's total income -0.06** -0.06 -0.04 -0.05 -0.07 minus working female’s earned (0.02) (0.04) (0.03) (0.05) (0.21) income (NPR) Madhesh Province 1.11*** 0.71*** 0.98*** 0.85*** (0.12) -0.24 (0.08) (0.12) (0.05) (0.49) Gandaki Province -0.19*** -0.36*** -0.25*** -0.35*** (0.09) -1.96*** (0.07) (0.08) (0.06) (0.4) Sudurpashchim Province 0.45*** 0.44*** 0.49*** 0.25 -2.19** (0.09) (0.14) (0.15) (0.15) (0.79) Respondent belongs to Madheshi 0.06 0.07 0.15*** 0.11 0.16 (non-Dalit) ethnic (0.09) (0.17) (0.05) (0.15) (0.30) group At least one among MIL and FIL -0.02 -0.07 0.00 -0.05 0.00 live in the HH (0.02) (0.05) (0.02) (0.05) (0.22) Indicator for ever migrated 0.09 -0.03 0.07 0.01 0.80** (0.06) (0.12) (0.04) (0.11) (0.32) Constant 0.96*** 1.14** 0.97** 1.53** (0.62) 18.56*** (0.31) (0.54) (0.36) (2.46) Observations 850 850 850 850 850 Adjusted R-squared 0.64 0.26 0.60 0.33 0.09 Notes: (1) Reference group for province is Bagmati Province. (2) SEs clustered at the ward level. (3) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. SEs clustered at the ward level. 48 Table A.14. Differences by ethnicity in association between social norms indices and women’s paid work outcomes HH Roles Public Space Both themes (1) (2) (3) Paid work Paid work Paid work PNB Index_HH Roles # 0.04 -0.04 Madheshi () (0.07) (0.08) PNB Index_HH Roles () -0.09** (0.04) -0.07* (0.04) SNE Index_HH Roles # -0.09 0.01 Madheshi () (0.12) (0.12) SNE Index_HH Roles () -0.03 -0.04 (0.06) (0.05) PNB Index_Public Space # 0.10* (0.05) 0.09 Madheshi () (0.06) PNB Index_ Public Space () -0.04 -0.03 (0.03) (0.03) SNE Index_ Public Space # -0.12*** -0.13*** (0.04) Madheshi () (0.04) SNE Index_ Public Space () 0.01 0.02 (0.03) (0.04) Madheshi -0.01 -0.04 0.05 (0.13) (0.07) (0.13) Obs. 2,000 2,000 2,000 Adjusted R2 0.03 0.02 0.03 Controls YES YES YES Regional FE YES YES YES Notes: (1) Controls include a binary variable for belonging to the Madheshi ethnic group, respondents’ age, years of education, having children younger than 5 years, log of household’s total income, at least one among mother -in-law or father-in-law living in the household, being a migrant household, living in an urban area, and provinces. (2) Interaction term () can be interpreted as the difference between the two subgroups (in this case, Madheshi vs Madheshi respondents) in the association between conservative beliefs and women’s paid work status. If the interaction term is negative and statistically significant, it means that the association is stronger for Madheshi compared to non-Madheshi respondents. If the interaction term is positive and statistically significant, the association is stronger for non-Madheshi compared to Madheshi respondents. 49 Table A.15. Differences by household migration status in association between social norms indices and women’s paid work outcomes HH Roles Public Space Both themes (1) (2) (3) Paid work Paid work Paid work PNB Index_HH Roles # migrant -0.08 -0.07 HH () (0.08) (0.08) PNB Index_HH Roles () -0.07* (0.04) -0.06 (0.04) SNE Index_HH Roles # migrant -0.04 -0.00 HH () (0.09) (0.09) SNE Index_HH Roles () -0.03 -0.04 (0.06) (0.06) PNB Index_Public Space # 0.03 0.03 migrant HH () (0.05) (0.05) PNB Index_ Public Space () -0.03 -0.02 (0.03) (0.03) SNE Index_ Public Space # -0.13** (0.06) -0.09 migrant HH () (0.06) SNE Index_ Public Space () 0.02 0.01 (0.04) (0.04) Migrant HH 0.01 0.03 0.05 (0.06) (0.05) (0.06) Obs. 2,000 2,000 2,000 Adjusted R2 0.03 0.02 0.03 Controls YES YES YES Regional FE YES YES YES Notes: Controls include a binary variable for belonging to a household with a current migrant male spouse, respondents’ age, years o f education, having children younger than 5 years, log of household’s total income, at least one among mother -in-law or father-in-law living in the household, living in an urban area, belonging to the Madheshi ethnic group, and provinces. Table A.16. Analysis of vignettes Comparison scenarios p-values associated with t-test Full sample Females only sample Males only sample 1. Change SNE only: low-low vs high- 0.62 0.48 1.00 low (conservative SEE & SNE vs. liberal SEE & conservative SNE) 2. Change both SEE and SNE: 0.00 0.00 0.00 low-low vs high-high (conservative SEE & SNE vs. liberal SEE & SNE) 3. Change SNE only: 0.05 0.08 0.32 low-low vs low-high (conservative SEE & SNE vs. conservative SEE & liberal SNE) 50 Table A.17. Bivariate regression coefficients – Beliefs about whether women should work under different contexts predicting women’s work for pay outside the home Theme Circumstances/contexts Core elements Woman Woman Female spouse measuring social works works outside works outside for norms outside for for pay pay pay (Female-only (Male-only (Full sample) sample) sample) A woman’s Married Respondent 0.23*** 0.18** 0.32*** characteristics approval (0.05) (0.08) (0.01) and roles Community 0.18*** 0.20*** 0.15** approval (0.04) (0.03) (0.06) Children younger than Respondent 0.05 0.08* 0.02 five looked after by approval (0.04) (0.04) (0.05) family Community 0.07*** 0.08*** 0.06* approval (0.02) (0.02) (0.04) Children younger than Respondent 0.08*** 0.14*** 0.01 five looked after by approval (0.03) (0.03) (0.03) non-family Community 0.12** 0.14*** 0.09 approval (0.04) (0.04) (0.05) Unable to keep up with Respondent 0.09*** 0.11*** 0.07** household tasks approval (0.02) (0.02) (0.03) Community 0.07*** 0.11*** 0.02 approval (0.02) (0.02) (0.04) Educated Respondent 0.01 0.02 0.00 approval (0.10) (0.09) (0.14) Community 0.04 0.09 -0.04 approval (0.09) (0.08) (0.14) A woman’s job Returns home after 5 Respondent 0.17*** 0.20*** 0.11** characteristics approval (0.03) (0.04) (0.05) Community 0.14*** 0.14*** 0.15*** approval (0.03) (0.03) (0.03) Workplace not close to Respondent 0.09*** 0.08*** 0.11** home approval (0.02) (0.02) (0.04) Community 0.09*** 0.12*** 0.05 approval (0.02) (0.02) (0.03) Occasionally travel to Respondent 0.10*** 0.09*** 0.10*** work outside the approval (0.02) (0.02) (0.03) village/city Community 0.05* 0.08*** 0.01 approval (0.03) (0.02) (0.04) Job physically Respondent 0.07*** 0.07*** 0.08*** demanding approval (0.02) (0.02) (0.03) Community 0.06* 0.06** 0.05 (0.03) (0.03) (0.04) 51 approval Works in a government Respondent 0.00 0.01 -0.01 job approval (0.04) (0.04) (0.05) Community 0.01 0.00 0.03 approval (0.04) (0.04) (0.05) Works in a Respondent 0.09*** 0.10*** 0.08** restaurant/hotel approval (0.02) (0.03) (0.03) Community 0.05 0.08** 0.02 approval (0.04) (0.03) (0.05) Job pays very well Respondent 0.10 0.12 0.07 approval (0.08) (0.08) (0.13) Community 0.01 -0.02 0.06 approval (0.09) (0.08) (0.14) Husband’s Not working and no Respondent 0.05 0.03 0.08 employment income approval (0.05) (0.07) (0.08) status Community 0.05 0.00 0.11 approval (0.05) (0.05) (0.07) Previously unemployed Respondent 0.13*** 0.15*** 0.11** but now employed approval (0.04) (0.04) (0.04) Community 0.08*** 0.08** 0.09** approval (0.02) (0.03) (0.04) Migrant worker who Respondent 0.08* 0.06 0.11** does not send approval (0.04) (0.04) (0.04) remittances Community 0.09** 0.07* 0.11** approval (0.03) (0.04) (0.04) Notes: (1) Positive correlation means that approvals are associated with higher likelihood of women working for pay outside the house. (2) Respondent and community approval variables are binary variables coded as 1 if the respondent approves of the behavior. Outcome variable is coded as 1 if the female respondent answered yes to working outside the house for pay. In the case of male respondent, he was asked whether his wife works outside for pay and the variable was coded accordingly. (3) No controls and no regional fixed effects are included. (4) SEs clustered at the ward level. (5) Robust standard errors in parentheses. (6) *** p<0.01, ** p<0.05, * p<0.1. 52 Tables A.18. Approval of women’s work under different contexts, by sub-groups Table 18.1 Approval of women’s work under different contexts, non-working vs. working Working women Non-working women Pairwise t-test Mean Mean/(SE) Mean/(SE) difference Individual Perceived Individual Perceived Individual Perceived approval community approval community approval community (PNB) approval (PNB) approval (PNB) approval (SNE) (SNE) (SNE) A woman's characteristics and roles She is married 0.99 0.95 0.98 0.94 -0.01 -0.01 (0.01) (0.01) (0.01) (0.01) She has small 0.90 0.80 0.84 0.72 -0.06*** -0.09*** children under 5 (0.01) (0.02) (0.02) (0.02) who can be looked after by family members She has small 0.32 0.20 0.21 0.12 -0.11*** -0.08*** children under age (0.02) (0.02) (0.02) (0.01) 5, who she will need to drop with non-family members while she is at work She is unable to 0.52 0.39 0.41 0.30 -0.11*** -0.10*** keep up with all (0.02) (0.02) (0.02) (0.02) household tasks without help She is well 0.98 0.98 0.98 0.97 -0.00 -0.01 educated (0.01) (0.01) (0.01) (0.01) A woman's husband's employment status Her husband is 0.97 0.94 0.94 0.94 -0.02* -0.00 not working and has (0.01) (0.01) (0.01) (0.01) no income Her husband was 0.95 0.84 0.89 0.78 -0.06*** -0.06*** previously (0.01) (0.02) (0.01) (0.02) unemployed [e.g., no income] but has now found a job Her husband is a 0.94 0.91 0.91 0.866 -0.030* -0.04** migrant worker (0.01) (0.01) (0.01) (0.014) and is not sending her money A woman's jobs' characteristics 53 She returns home 0.92 0.78 0.84 0.69 -0.08*** -0.08*** from work after 5 (0.01) (0.02) (0.02) (0.02) pm She works in a 0.92 0.91 0.90 0.90 -0.02 -0.01 government job (0.01) (0.01) (0.01) (0.01) Her workplace is 0.82 0.65 0.76 0.56 -0.06*** -0.09*** not close to her (0.02) (0.02) (0.02) (0.02) home The job she is in 0.99 0.97 0.97 0.96 -0.02* -0.01 pays very well (0.01) (0.01) (0.01) (0.01) She has to 0.74 0.47 0.63 0.41 -0.12*** -0.06** occasionally travel (0.02) (0.02) (0.02) (0.02) for work outside the village/city Her job is 0.67 0.50 0.63 0.47 -0.04 -0.03 physically (0.02) (0.02) (0.02) (0.02) demanding She works in a 0.56 0.31 0.45 0.25 -0.11*** -0.06** hotel/restaurant (0.02) (0.02) (0.02) (0.02) Observations 515 635 1,150 Table 18.2 Approval of women’s work under different contexts, non-migrant vs. migrant households Females w/o migrant spouse Females with migrant spouse Pairwise t-test Mean Mean/(SE) Mean/(SE) difference Individual Perceived Individual Perceived Individual Perceived approval community approval community approval community (PNB) approval (PNB) approval (PNB) approval (SNE) (SNE) (SNE) A woman's characteristics and roles She is married 0.98 0.95 0.99 0.94 -0.01 0.01 (0.00) (0.01) (0.01) (0.01) She has small 0.86 0.75 0.89 0.77 -0.03 -0.02 children under 5 (0.01) (0.02) (0.02) (0.02) who can be looked after by family members She has small 0.25 0.16 0.30 0.17 -0.05* -0.02 children under age (0.02) (0.01) (0.03) (0.02) 5, who she will need to drop with non-family members while she is at work 54 She is unable to keep 0.48 0.36 0.41 0.28 0.07** 0.08** up with all household (0.02) (0.02) (0.03) (0.03) tasks without help She is well 0.98 0.98 0.99 0.96 -0.01 0.03** educated (0.01) (0.01) (0.01) (0.01) A woman's husband's employment status Her husband is 0.96 0.94 0.94 0.93 0.01 0.01 not working and (0.01) (0.01) (0.01) (0.02) has no income Her husband was 0.91 0.81 0.92 0.79 -0.01 0.02 previously (0.01) (0.01) (0.02) (0.02) unemployed [e.g., no income] but has now found a job Her husband is a 0.93 0.91 0.90 0.81 0.03 0.10*** migrant worker (0.01) (0.01) (0.02) (0.02) and is not sending her money A woman's jobs' characteristics She returns home 0.88 0.75 0.85 0.67 0.03 0.09*** from work after 5 (0.01) (0.02) (0.02) (0.03) pm She works in a 0.93 0.93 0.86 0.86 0.06*** 0.07*** government job (0.01) (0.01) (0.02) (0.02) Her workplace is 0.80 0.63 0.74 0.52 0.06** 0.11*** not close to her (0.01) (0.02) (0.03) (0.03) home The job she is in 0.98 0.96 0.98 0.97 -0.01 -0.00 pays very well (0.01) (0.01) (0.01) (0.01) She has to 0.67 0.46 0.63 0.38 0.07** 0.08** occasionally travel (0.02) (0.02) (0.03) (0.03) for work outside the village/city Her job is 0.70 0.50 0.65 0.43 -0.01 0.07** physically (0.02) (0.02) (0.03) (0.03) demanding She works in a 0.51 0.29 0.48 0.24 0.03 0.05 hotel/restaurant (0.02) (0.02) (0.03) (0.03) Observations 850 300 1,150 55 Table 18.3 Approval of women’s work under different contexts, males vs. females Females Males Mean/(SE) Pairwise t-test Mean Mean/(SE) difference Individual Perceived Individual Perceived Individual Perceived approval community approval community approval community (PNB) approval (PNB) approval (PNB) approval (SNE) (SNE) (SNE) A woman's characteristics and roles She is unable to 0.46 0.34 0.52 0.41 0.05** 0.07*** keep up with all (0.02) (0.01) (0.02) (0.02) household tasks without help A woman's jobs' characteristics She returns home 0.87 0.73 0.88 0.77 0.01 0.04** from work after 5 (0.01) (0.01) (0.01) (0.01) pm Her workplace is 0.79 0.60 0.83 0.64 0.04** 0.04* not close to her (0.01) (0.01) (0.01) (0.02) home Observations 1,150 850 2,000 Note: Only select contexts that have a significant difference are reported, other contexts explored in the survey not reported for brevity. 56 Annex B: Figures Figure B.1. Social norms measurement using context-dependent scenarios Source: Authors’ compilation based on literature and the quantitative tool. Figure B.2. Distribution of social norms indices, by gender Gender roles within the household - Personal Beliefs Index Note: The higher the values on each scale, the more conservative beliefs become. 57 Gender roles within the household - Social Normative Expectations Index Women in the public space - Personal Beliefs Index 58 Women in the public space - Social Normative Expectations Index Sanctions Index Note: The higher the value, the higher the likelihood that respondents expect households in their community to face negative sanctions for breaking a norm. 59 Figure B.3. Reference Groups Figures B.3(a) Reference groups for major life events 100 All 91 Sahre of respondents (in percent) 80 60 59 51 40 39 30 20 28 28 23 18 0 8 6 8 5 5 NOT AT ALL 0 1 LITTLE EXTENT SOME EXTENT LARGE EXTENT Spouse Family Non-HH family Community 100 Females 94 Sahre of respondents (in percent) 80 60 59 51 40 40 29 20 28 27 25 17 7 0 6 6 5 5 NOT AT ALL 0 1 LITTLE EXTENT SOME EXTENT LARGE EXTENT Spouse Family Non-HH family Community 60 100 Males 86 Sahre of respondents (in percent) 80 60 59 51 40 38 31 20 27 29 19 20 10 12 0 5 5 6 0 NOT AT ALL 1 LITTLE EXTENT SOME EXTENT LARGE EXTENT Spouse Family Non-HH family Community Notes: (1) “Life events” include important events in one’s life related to one’s education, responsibilities, marriage, migration, and the child’s education, their marriage etc. Figures B.3(b) Reference groups for working outside the home for pay 100 Females 93 Sahre of respondents (in percent) 80 60 53 49 40 38 35 32 20 23 19 19 13 0 8 7 6 4 1 NOT AT ALL 1 LITTLE EXTENT SOME EXTENT LARGE EXTENT Spouse Family Non-HH family Community 61 100 Males 86 Sahre of respondents (in percent) 80 60 49 52 40 34 36 35 20 25 21 14 16 12 0 6 7 4 0 NOT AT ALL 2 LITTLE EXTENT SOME EXTENT LARGE EXTENT Spouse Family Non-HH family Community 62 Annex C: Illustrations Box C.1. Illustration of How Social Norms Influence Behavior Let us take the example of a girl named Seema, who lives in a conservative village in Nepal’s Terai region. Her understanding of the social norms can develop by observing the behaviors and beliefs of people around her (her “reference group”), which could include, among others, her peers in the neighborhood, extended family members, and other females in her household. Let us say she observes that it is common for females in her village quit their jobs after getting married – just like her sister-in-law did. This informs her belief that women, once married, prioritize taking care of their families over their own careers (her “social empirical expectation”). As she gets trained as a hairdresser, she sees and hears about other women, who seek employment in a barber shop, being shunned or criticized (experiencing “sanctions") by religious leaders and community members. Based on these reactions from her community, she forms a second-order belief that people in her community do not approve of women cutting men’s hair or taking up similar jobs that require physical interaction with men (“social normative expectation”). Although Seema might aspire to pursue her hairdressing career and believe that she should (“personal normative belief” or “personal belief”), what she thinks other women do and others in her reference group think is the right thing to do, coupled with expected negative consequences if she does not follow what others do and think, can collectively influence her decision to not pursue this career (“individual behavior”). This is how social norms, in practice, can influence behavior. Now let us say Seema moves to Kathmandu alone, where she knows no one and sees women around her pursuing careers in all kinds of fields. The social norm that guided her decision in the village applied within her community – which was her reference group then. With the change in a way people around her behave and think about women’s careers, she may finally take up the job at a barber shop in line with what she believes. This implies that social norms apply within a reference group, as highlighted by Bicchieri at al. (2014). Box C.2. Examples of Reference Groups’ Questions For 5 types of pre-defined reference groups (spouse, other household members, non-household family/relatives, community, other (specify), respondents were asked the following set of questions. 1. Would [person/group] have an opinion about whether or not you work outside the house? 2. To what extent do [person/group's] opinions matter to you? 3. To what extent do [person/group's] opinions matter to your household members? The same set of questions was asked about a second theme: important issues or events in life. 63 Annex D: Qualitative Data Collection The quantitative survey tool was developed using insights from qualitative interviews, the analysis of which also provides essential context to interpreting the results of this quantitative study. The fieldwork was conducted in 3 provinces during November 2022: Madhesh Province, Bagmati Province, and Sudurpashchim Province. In-depth semi-structured interviews were conducted with both men and women, divided equally between those working for pay and those not working for pay within each gender group. A detailed analysis of the qualitative data is presented in World Bank (2024, Forthcoming). Qualitative interviews revealed that social norms, along with beliefs, culture, traditions, intra-household relationships, and family circumstances shape women's decisions to work and the nature of their jobs. A key finding was the strong adherence to traditional gender roles within the home, with women primarily seen as caretakers in the household for which men were not considered good substitutes. Women’s presence outside the home came with concerns about the family’s reputation being harmed if the woman interacted with other men or the possibility of her being seen as unfaithful, as well as disapproval of women having a long commute for work. These findings guided the focus of social norms explored in the quantitative survey across two broad themes: (i) gender roles at the household level, and (ii) women’s mobility in the public space. Interviews also revealed that women’s participation in paid work outside the home is context dependent, influenced by factors such as having young children, financial status of the family, and job characteristics such as work location. As such, specific situations are explored in the quantitative study using a set of context dependent behaviors and characteristics. Additionally, marital status and reference groups seem to be crucial in shaping women's work decisions. Married women faced more significant social restrictions, and their decision to work was observed to be primarily influenced by household dynamics, particularly their husbands' beliefs and expectations. Though women's social circles also include other community members, such as teachers and social workers, men, overall, have much wider social circles. These findings highlighted the value of focusing on social norms governing married women’s decisions and reference groups of individuals within the household to better understand how norms are enforced. 64 Annex E: Sampling The processes employed to select the wards, households, and respondents was as follows: Ward (cluster) selection: We used the list of wards and its population figures from the 2021 Nepal Census. The sampling frame consisted of the list of all wards within each selected province. Each province comprises of districts and within each district are municipalities (urban and rural municipalities) which are further broken down into wards – the smallest administrative units. Districts were stratified by urban and rural areas to ensure greater statistical power for detecting differences between the 2 localities. The stratification by urban-rural was done to be proportionate to the population share of each group within a province, resulting in a self-weighted sample, allowing for analysis of data at province level and further at locality level within each province. To select the wards, a random start point was generated to negate any bias in the list and to provide an independent chance of selection from the list. The probability proportionate to size (PPS), gives an independent chance of selection to each wards as per its population size, i.e., a higher chance of selection to wards with higher population size.37 As a first step of random selection of wards, the cumulative frequency (CF) of population of households in a ward was calculated. Since the unit of analysis for our study purpose was households having certain criteria and we expected the main outcome variables (social norms) to vary at household levels (as opposed to at an individual level), the household population figures served as the basis for sampling purpose (as opposed to population size of individuals for a ward). Applying PPS, in the first step, the required number of wards were selected for Categories 1 and 2 households (households with working and non-working females respectively). Following this, the clusters allocated for Category 3 (households with migrant population) households were taken as a subset of the wards selected for Categories 1 and 2. The survey was carried out in both rural and urban locations in a total of 4 provinces. Selection of the random starting point within each ward during in-field random sampling of households: For every ward, a predefined landmark for starting point was chosen. The predefined landmark consisted of i) school, ii) health post, iii) central marketplace or iv) ward office. The chosen landmark for every cluster was rotated to account for randomization and to avoid interviewer bias. Once the landmark was chosen, each enumerator used the spin-the-bottle method to randomize direction in which the survey took place. After starting with a household, enumerators used a skip interval to survey every third household in rural areas and every fifth household in urban areas. Once the household was chosen, the interviewer used the screener to ascertain the eligibility as per the category quota set aside for them.38 Respondent selection: The respondents were selected based on a screener instrument that surveyed the following factors: 1. Gender: Both males and females within a household were interviewed. However, for households with migrant men, only the women were interviewed. 2. Age group: Only women within the economically active age range, i.e., between the ages of 18-59 years 37 38 With a target of 10 households per cluster, Categories 1 and 2 households were selected from 85 clusters and those for Category 3 from 30 clusters. The 85 clusters allocated for Categories 1 and 2 were distributed across the provinces based on proportion of household population, while the 30 clusters allocated for the Category 3 were distributed as per the proportion of migrant population within the 4 provinces. 38 Note that selection of wards during the pilot (for which we have selected Bagmati Province) did not use the PPS method. Since Bagmati is a Province shortlisted for the final survey as well, the Solutions team first finalized wards sampled for the final survey, and for the pilot, selected wards not included in the final list and that are in proximity to Kathmandu, ensuring both rural and urban wards were included. In-field sampling within each ward was random and enumerators had to use the same process that they will follow during the final survey. 65 were interviewed. For spouses of female respondents, the restriction was 18 or older. 3. Ethnicity: The major 8-10 ethnic groups in Nepal are captured in the sample. A screener ethnic composition was applied to ensure that marginalized ethnic groups such as Dalits were sufficiently represented in the survey. 4. Marital Status: Only married men and women were interviewed. 5. General demographic factors include: • Perceived economic situation: Low to middle low income • It was ensured that both the respondents (male and female for Categories 1 and 2) and female respondent for Category 3 belonged to the second generation of the selected household (for example, not the in-laws residing in a HH but their son and his wife. 66