Policy Research Working Paper 8840 Improving the Measurement of Rural Women’s Employment Global Momentum and Survey Research Priorities Gayatri Koolwal Development Economics Development Data Group May 2019 Policy Research Working Paper 8840 Abstract Rural economies are in transition around the world; in require improved counting of individuals’ employment, many countries, improved technology and linkages across accounting for the complexity of measuring rural women’s sectors have expanded access to markets and accelerated labor force participation, as well as data on social, economic, production for some farmers. At the same time, rural and institutional constraints that women face in seeking areas globally are facing a growing base of landless and better economic opportunities. Using recent rounds of the smallholder farmers, out-migration to urban areas, and per- Ethiopia, Malawi, Nigeria, and Uganda Living Standards sistence of low-skilled, informal, and seasonal jobs where and Measurement Study-Integrated Surveys on Agriculture, women are often heavily concentrated. Recent global initia- as well as findings from recent country pilots conducted by tives are examining programs that can effectively raise rural the International Labour Organization, this paper discusses incomes, including how addressing shortfalls in women’s best practices and issues to consider when examining rural hours worked and earnings can raise rural productivity and women’s employment in socioeconomic surveys, as well as growth. But well-designed policies to address these issues a survey research agenda to improve measurement. This paper is a product of the Development Data Group, Development Economics. 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 author may be contacted at koolwalg@gmail.com. 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 Improving the Measurement of Rural Women’s Employment: Global Momentum and Survey Research Priorities Gayatri Koolwal* JEL Codes: C8, J16, J21, J22, J24, J46. Keywords: Gender, Agriculture, Employment, Unpaid Work, Labor Underutilization, Household Surveys. *Consultant, Living Standards Measurement Study (LSMS), Survey Unit, Development Data Group, The World Bank, Rome. email: koolwalg@gmail.com . This work was supported by the International Fund for Agricultural Development. I am very grateful to Kathleen Beegle, Valentina Costa, Sam Desiere, Cheryl Doss, Talip Kilic, Paula Nagler, Joann Vanek, Kieran Walsh and Alberto Zezza for their detailed comments on the paper, as well as feedback from participants in the ICID/IFAD/C4D2 workshop on “Improving the Availability and Quality of Individual-Level Data on Women and Youth in Living Standards Measurement Study (LSMS) Surveys” , November 2018, Rome. 1 I. The policy case for improved data on rural women’s employment Across developing countries, rural women are heavily engaged in a range of employment activities — including smallholder farming, contributing work in the family farm or business, as well as in off-farm jobs that are typically seasonal and informal (ILO, 2018a, 2018b). Employment surveys typically have difficulties in collecting data on these types of jobs, as well as on social and economic barriers that women face in achieving better-quality employment outcomes. These include time burdens in unpaid household and care work, transport and safety issues, and other local norms and institutional barriers around marriage, work and having children (Jayachandran, 2015). Rural women often find themselves in particularly vulnerable employment situations, given limited access to resources, infrastructure and markets, as well as diversified sources of income. A better understanding of rural women’s work, paid and unpaid, 1 and the constraints they face is critical for policy design as rural economies evolve with economic growth. Depending on country context, growth (particularly in the nonfarm sector) has led to rapid out-migration to urban areas, as well as increased diversification of rural incomes, although the quality and expected earnings of off-farm opportunities are often low (Davis et al 2017, Mellor, 2017). A growing urban middle class in many developing countries has led to increased demand for food and other resources, renewing policy efforts to improve incomes and productivity in agriculture, including investments in inputs and new technology. At the same time, however, rural land ownings are increasingly fragmented across countries, leading to a rising global base of less-productive, smallholder farmers, as well as growing casual labor on farms among the landless (Lowder et al, 2017). Since rural women are often concentrated in these activities, they are increasingly targeted by country programs aiming to raise rural productivity and economic growth (Kochhar et al, 2017). But these programs need to be informed by data that more accurately count women who are employed, and constraints to working. These concerns have also helped motivate recent global initiatives in policy and statistics to regularly track outcomes of small-scale rural work across developing countries, and understand what policies and programs work for raising rural incomes and welfare. International momentum on measuring paid and unpaid work for rural women At the international level, Target 2.3 of the Sustainable Development Goals and the recently-launched 50 x 2030 Initiative emphasize the role of data in policies to raise productivity and earnings among smallholder farmers. 2 Resolution I of the 19th Conference of International Statisticians (19th ICLS, 2013) also narrowed the international definition of employment to work for pay or profit, requiring more careful data collection on (a) economic activities that rural women are concentrated in, including those that are heavily subsistence-based and/or unpaid for significant parts of the year, as well as contributing family work; (b) complementary data collection on individuals’ total work burdens across paid and unpaid activities; as well as (c) additional data on labor underutilization, to provide a comprehensive view of demands and constraints among individuals who either want to engage in more paid work, or are out of the labor force for various reasons. The Women’s Work and Employment Partnership, a multi-institutional collaboration 1 In this paper, paid work entails time spent on an income earning activity (even if explicit wages or salary are not given to the person); this also includes work in non-subsistence farming, nonfarm household enterprises, and as an employee. 2 Specifically, SDG Target 2.3 looks at doubling agricultural productivity and incomes of small-scale farmers, with a particular focus on women and poorer/marginalized groups. The 50 x 2030 initiative is a global partnership to conduct regular surveys of small-scale farmers across 50 countries, and make these data available to countries for policy making. 2 between FAO, ILO and World Bank, is also conducting methodological work in countries to operationalize the 19th ICLS definitions in employment surveys. Rural women are also a crucial demographic as countries look to develop strategies for targeting and collecting data on informal employment, as part of the ILO’s Recommendation No. 204 on transitioning from the informal to formal economy. The recently-revised International Classification of Status in Employment (ICSE-18) during the 20th ICLS also includes, relevant for poorer and rural women, the creation of a new employment status category on dependent contractors, including seasonal/casual and short-term workers, as well as recommending a survey question on place of work that will also improve data on types of employment that are typical for women such as home-based work and street vending. What additional data could inform rural program design? Many countries have recently introduced employment-related initiatives where rural women are beneficiaries (Box 1). Many of these programs have faced difficulties in targeting women, however, owing to lack of information about who is employed or available/wanting to work, as well as constraints women face in participating. This includes the need for regular data collection on earnings and hours worked among men and women, as well as on constraints to participation — as in the case of India’s National Rural Employment Guarantee Scheme, which has had success in keeping poor rural women in the labor force (Sarkar et al, 2018), but suffers from lack of available jobs in some areas, lower female labor force participation (LFP) in poorer states, and women working longer hours to receive the same daily wage rate as men (Dutta et al, 2014; de la O Campos, 2015). Rural skills/technology initiatives could benefit from similar data. Ethiopia’s Rural Capacity Building Project (RCBP), for example, has faced difficulties in targeting agricultural extension services to women farmers, due to household-level outreach where the reported decision makers/points of contact are often men, as well as gender differences in human capital and networks. Targeting of RCBP and other rural entrepreneurship skills training programs could be improved by devising different delivery approaches based on a more accurate counting of self-employed women, and constraints women face to participating (Buehren et al, 2017). And programs to expand productivity-enhancing finance to women farmers/small enterprises, while expanding rapidly across countries, are also often hampered by a clear understanding of who manages the farm or business, and whether men are ultimately making decisions over investments (Holloway et al, 2017). Box 1. Examples of large-scale country initiatives that target rural women and employment outcomes Workfare programs • India’s National Rural Employment Guarantee Scheme (NREGS), launched in 2005 and which reached 76 million workers in 2016-17, targets the rural poor and guarantees 100 days of wage work per year, with a quota of at least 30 percent women participants. In 2016, 52 percent of participants were women (Fletcher et al, 2018). Under the program, men and women earn equal wage rates, and childcare is also provided for groups of women with children under 6 years. • Ethiopia’s Productive Safety Net Program (PSNP) was introduced in 2005 to help the rural poor withstand shocks and address food insecurity. Men and women participate in public works projects, and women are also entitled to childcare services. Projects also include supporting agricultural work on private land owned by female-headed households (Jones et al, 2010). Skills development, and access to markets and technology • The Self-Employed Women’s Association (SEWA) in India, the largest trade union in the country, supports self-employed urban and rural women (about 900,000 women in rural areas) (de Luca et al, 2013). SEWA helps women entrepreneurs through training/skills development, accessing supplier networks and organizing producer groups, and in turn easing access to markets, finance, and technology. 3 • Ethiopia’s Rural Capacity Building Project (RCBP) was implemented in 10 regions of the country beginning in 2007. RCBP strengthens agricultural extension services through raising awareness of improved and environmentally sustainable techniques, as well as better targeting farmers’ needs, including a focus on women farmers — who constitute (depending on the data source) between 29 percent and 45.5 percent of agricultural labor (FAO, 2011; Palacios-Lopez et al, 2017). • Malawi’s national Farm Input Subsidy Program for smallholder farmers, part of the country’s Agriculture Sector-Wide Approach (ASWAp), includes female-headed households as targets for access to inputs, extension and on-farm technologies, as well as supporting access to input and output markets (Deijl et al, 2017). Access to finance for self-employment • Mexico’s Expanding Rural Finance project, launched in 2015, is targeted towards small farmers and micro, small and medium enterprises (MSMEs) in rural areas. In 2017, women agricultural producers and MSMEs accounted for more than 70 percent of beneficiaries. • Afghanistan’s Women Economic Empowerment - Rural Development Program (WEE-RDP), launched in 2018, will cover 34 provinces. A main focus will be on improving access to finance for women-owned MSMEs. Understanding how rural employment responds to market factors Better data on factors enabling women’s economic mobility, and opportunity costs for employment, can also shed light on how rural women respond to changes in broader economic shifts. Using recent cross- country data from developing-country labor force surveys, Figure 1 shows that LFP among younger rural women follows a U-shaped pattern (discussed in Goldin, 1995; Mammen and Paxson, 2000) — where women’s LFP is higher in poorer countries, declines initially with growth as they pursue more education and leisure, and then rises again with increases in income/expected earnings opportunities. However, employment for older age groups is less responsive to economic growth (right panel of Figure 1). This is an important policy question which countries like India have recently been grappling with (Fletcher et al, 2018). Younger women, who are less likely to have children and may have greater mobility (Including opportunities for education), could be more responsive to changes in income and market conditions — but better data are needed on economic opportunities and constraints faced by rural women overall. Figure 1. Rural women’s labor force participation against log GDP per capita (2016-17), by age group Younger women (aged 15-24) Women aged 25+ Source: LFP data obtained from ILOSTAT (Key Indicators of the Labour Market (KILM) data), 2016 and 2017. GDP data obtained from WDI. 4 Overview of the paper The paper examines methodological approaches to improving socioeconomic survey data on rural women’s employment amid rural economic transitions, presenting stylized facts from recent World Bank Living Standards and Measurement Study-Integrated Surveys on Agriculture (LSMS-ISA) conducted in Ethiopia, Malawi, Nigeria and Uganda, as well as recent country pilot studies conducted by the ILO to test survey question design under the 19th ICLS recommendations. These surveys are of particular interest in the context of rural employment, given their detailed modules on agriculture along with other sectors, as well as a broad range of individual and household socioeconomic characteristics that can be used to better understand employment trends. The paper covers five main priority areas (from both research and policy perspectives) in improving the measurement of rural women’s employment in surveys: (a) Accurate counting of rural women’s employment, since their labor contributions are often underreported, and particularly given the revised definitions of work and employment under the 19th ICLS; (b) Survey questions to better understand labor underutilization; (c) Measuring unpaid work burdens and the role of time use surveys; (d) Survey respondent choice in intra-household decision-making questions on earnings and farm/business management roles; and (e) Capturing additional features of women’s work amid rural transition — including migration, skills development, access to productivity-enhancing technologies — along with barriers to economic mobility (or being able to seek better economic opportunities). The aim of the discussion is not to recommend that all issues be tackled together, particularly given cost and other considerations, including country context — but to highlight important measurement issues for further research and policy design, the implementation of which could be determined based on local priorities. Variations also exist across surveys in how employment questions are asked and modules are structured, including within the LSMS. This paper compares different survey approaches to understand best practices in presenting a more comprehensive view of rural women’s employment that is also relevant for policy making. II. Ensuring rural women’s employment is not undercounted Improving the counting of rural women’s employment in surveys can strengthen the targeting and design of rural policies. Aspects of rural women’s work, such as informality and contributing family roles, can lead to underreporting of their employment, including if they perceive themselves mainly in “domestic work” when their earnings are low and/or seasonal (see, for example, Benes and Walsh, 2018a; Koolwal, 2018; Comblon and Robilliard, 2017; and Fox and Pimhidzai 2013). Boundary questions for small-scale farming Within farming, women have multiple roles, including planting and harvesting crops, raising livestock, resource collection, and processing and marketing output. Women farmers tend to manage smaller plots/parcels, however — often with only a small share of output sold seasonally, with the remainder for household use (FAO, 2011). Under the narrowed definition of employment under the 19th ICLS, “boundary” 5 survey questions for agriculture are therefore needed to distinguish respondents’ intended production for sale versus their own use. Box 2 presents boundary questions that have been used in recent rounds of the LSMS-ISA in Malawi, Nigeria, and Ethiopia (currently in the field); these were informed by a recent labor force survey (LFS) pilot study across 10 countries in different global regions to better understand how the 19th ICLS definitions should be implemented across different country contexts (Benes and Walsh, 2018a). Box 2. Comparing survey questions across LSMS-ISA on main intended destination of agricultural products Malawi (2016-17) (only for crops): For each of the five most important crops you worked on in the last 7 days (question asked separately for each crop): - Was production from this crop (a) all intended for sale, (b) some are intended for sale and rest for consumption by family, or (c) all intended for family use/consumption? - If some intended to be sold, was it (a) less than ¼, (b) ¼, (c) ½, (d) ¾, or (e) more than ¾ ? - In the past, were these products mainly sold or mainly kept for consumption? Nigeria (2015-16 post-planting and post-harvest rounds, and current round in the field); Ethiopia (current round in the field) (one question covering all agricultural activities): Over the last 7 days, are products obtained from [NAME’s] household agricultural production/family farming (one question pertaining to all crop/livestock/fishing activity): - (a) only for sale/barter; (b) mainly for sale/barter but some for own/family use; (c) mainly for own/family use but some for sale/barter; or (d) only for own/family use? There are variations in question design — the Malawi survey asks boundary questions for each of the five most important crops; the Nigeria, Ethiopia, and planned future LSMS-ISA surveys ask one question across all agricultural activities. The data across these surveys reflect similar trends, however. Table 1 shows that in both Malawi and Nigeria, own-use production activity in agriculture is high, and significantly higher for women than men (for Nigeria, in the post-harvest round). Under the 19th ICLS standards, these individuals would no longer be considered self-employed in agriculture — Desiere and Costa (2018) also find that reported employment in agriculture in the Nigeria survey falls considerably under the new standards because of high shares of work in own-use production. Including separate questions in surveys on own- use production can ensure that the significant time women spend in this work, particularly in poorer agricultural contexts, is not missed in employment surveys. Table 1. Malawi and Nigeria LSMS-ISA: intended destination of own-farm products in last 7 days, men and women aged 15+ (own-reporting sample) Poor households(b) Non-poor households(b) Men Women Gender gap Men Women Gender gap (A) (B) (A-B)(c) (A) (B) (A-B)(c) Malawi 2016-17: Share of individuals in crop agriculture: Any crop being mainly intended for sale(a) 0.40 0.34 0.06*** 0.49 0.39 0.10*** [0.49] [0.47] [0.50] [0.49] All crops for household use/ consumption 0.51 0.60 -0.09*** 0.41 0.51 -0.10*** [0.49] [0.49] [0.59] [0.50] Number of individuals in crop agriculture 1366 2191 1337 1737 6 Share of total own-reporting sample 0.48 0.45 0.36 0.32 Nigeria 2015-16: Share of individuals in any own-farm work: Post-planting round 0.27 0.29 -0.02 0.35 0.28 0.06*** Only/mainly intended for sale(a) [0.44] [0.45] [0.48] [0.45] Mainly intended for HH 0.54 0.51 0.03 0.47 0.47 -0.004 use/consumption, some for sale [0.49] [0.50] [0.50] [0.50] 0.19 0.20 -0.01 0.18 0.24 -0.06*** Only for HH use/consumption [0.39] [0.40] [0.39] [0.43] Number of individuals in own-farm work 3163 1706 1800 1452 Share of total own-reporting sample 0.47 0.26 0.29 0.22 Post-harvest round Only/mainly intended for sale(a) 0.41 0.34 0.07** 0.48 0.39 0.09*** [0.49] [0.47] [0.50] [0.49] Mainly intended for HH 0.40 0.44 -0.04** 0.36 0.40 -0.04* use/consumption, some for sale [0.49] [0.49] [0.48] [0.48] Only for HH use/consumption 0.19 0.22 -0.03 0.17 0.21 -0.04** [0.40] [0.42] [0.37] [0.41] Number of individuals in own-farm work 1157 474 796 692 Share of total own-reporting sample 0.53 0.28 0.34 0.27 Notes: Standard deviations in brackets. T-tests of equality of means were conducted across men and women within each set of columns; *** p<0.01, ** p<0.05, * p<0.1. Sampling weights used. (a) Mainly intended for sale = either all, or at least half of output intended to be sold (b) In Malawi, households were compared above and below the poverty line. In the Nigeria data, the poverty line variable was not available, so households in the bottom two quintiles of consumption expenditure were compared against other households. A few gender-related measurement issues need to be addressed with boundary questions. One is proxy response — in separate results from the Malawi survey, women who were proxies for other women/men reported significantly lower market activity on average, compared to individuals reporting for themselves. 3 Women proxies might have less information about farm market activity (similar patterns arise with estimates of hours worked, as discussed in Section III), but this requires further investigation across countries. Addressing seasonality is also important. Table 1 shows that in Nigeria, reported production that is only/mainly for sale is higher in the post-harvest as opposed to post-planting season, which would affect employment estimates; significant gender differences also emerge post-harvest among poor respondents. Asking about intended use of production over the last 12 months in addition to the last 7 days may be one way to address seasonality (similar to other wage/self-employment questions in the LSMS- ISA, and other household surveys). However, further testing may be needed to address recall issues, particularly for proxy respondents (Gaddis et al, 2019). Integrating boundary questions for each plot/agricultural activity within the survey’s agricultural module could also improve reporting precision, and provide a more detailed view of women’s farming roles along with agricultural module questions on production and decision-making/management (GSARS, 2017). Given the additional burden placed on 3 Results available upon request. 7 respondents, this approach may be better suited for more focused agricultural surveys conducted on a rotating basis, that could then be compared with standard employment survey data. Counting women’s off-farm employment: The role of informality Aside from own-farming, rural women are also heavily involved in informal jobs that are more difficult to pick up in surveys. Casual wage labor on farms, for example, is increasingly common amid a growing share of landless households (Mueller and Chan, 2015). Figure 2, for example, shows that poor individuals in the Malawi 2016-17 LSMS-ISA, and particularly women, reported nearly all wage work in ganyu (or short-term, temporary labor). As this work is highly seasonal and typically with little pay, and has traditional nomenclature, it can often go unreported if not explicitly asked in surveys (Beegle et al, 2017). Valenzuela (2003) also discusses other contexts where day labor can be underreported for similar reasons — In Japan, for example, “laborer” is closely associated with the word romusha, which implies low-skilled, menial labor, and many respondents in this work may not self-identify in these roles. The second panel of Figure 2 also shows that among wage earners, the gender discrepancy in ganyu labor widens with age; older women wage laborers are as likely as younger women to continue working in ganyu, while the share of men in this work declines rapidly with age. Close attention is needed in surveys to local labor arrangements that the poor engage in — and which might get missed with standard questions on wage and self-employment, particularly for women who are often highly engaged in these seasonal forms of work, and even among more vulnerable/older groups. Figure 2. Malawi LSMS-ISA, 2016-17: share of men and women in ganyu (short-term, temporary) labor Share of men and women (total sample) that have worked in ganyu and Among poor men and women wage earners, share in ganyu other wage labor, by poor/non-poor households and reference period in the last 12 months, by age Within self-employment, another issue is identifying women who run their own business — and separately those in contributing family work. Many rural women who have a business do not necessarily count themselves as employed/entrepreneurs, because their operations are typically small in scale, and this work is often conducted alongside other family responsibilities (ILO, 2018b). Cognitive testing that informed the 8 ILO LFS pilot study also found that in some contexts words like “business” were interpreted mostly as formal enterprises, and so respondents (often women) would be less likely to report their work as a business if they were self-employed in informal settings and/or had unpaid supporting roles in the family farm or business. Women in contributing family roles are also considered employed, since it contributes to household profit, but are also often less likely to be identified (or self-identify) in this form of employment. The ILO LFS pilot study found that women in contributing family work often tend to identify themselves first as engaged only in domestic work, because of social and cultural norms that diminish the economic value of contributing family work, and since this work is also often conducted with domestic activity. As a result, follow-up or “recovery” questions were needed to hone in on this as well as other economic activity (also see Benes and Walsh, 2018a; Data2X, 2017). Women respondents in the cognitive testing stage, for example, often said they helped with stocking, storing, and other activities on the farm, but they had to be prompted specifically to report these activities. Figure 3 also shows, based on recent analysis of the ILO pilot data (Koolwal, 2018), that recovery questions on contributing family work re-classifies substantial shares of men and women — with the greatest effects for rural women — back into employment. Figure 3. ILO LFS pilot countries: share of men and women re-classified as employed, through recovery questions on contributing family work Rural Urban Notes: Similar patterns observed for recovery questions on paid work in agriculture, and to a lesser extent on small/secondary jobs. Source: Koolwal (2018), based on ILO LFS pilot study data accessed at ILO. Careful attention to wording choice, adapted to local context, is therefore important in the framing of employment questions. A recent study by Comblon and Robilliard (2017) using data from more than 40 surveys and censuses conducted in Cameroon, Mali and Senegal between 1976 and 2012 also discusses how women’s reporting of employment is more sensitive to questionnaire wording and design, owing to the effects of social norms/interpretation of wording, and the specific types of work women in low-income contexts tend to be involved in. Most socioeconomic surveys also do not distinguish between running and supporting a family business in their labor modules — within the LSMS, the 2016-17 Malawi and 2013-14 Uganda surveys do ask separate questions on whether respondents “run a business of any size for 9 themselves or the household” or “helped without being paid in any kind of business run by the household”, but in other surveys, “running or helping” in a business is asked in one question. The ILO LFS pilot study findings also recommend maintaining separate questions for running versus contributing family work in a business, to better identify household members’ roles (Benes and Walsh, 2018a), although similar care needs to be taken in the guidance and framing of questions to ensure that respondents do not default to traditional gender norms where women are always considered as “helping” in the business and men are “running” the business, even if their roles are similar. Summary and ways forward Rural women’s employment can be underreported in employment surveys, owing to multiple factors including their concentration in informal and seasonal work, social norms about how they view their labor contributions, and other domestic responsibilities they are engaged in. Box 3 summarizes key points highlighted in this section. Box 3. Summary: Ways forward on counting rural women’s work • Boundary questions that distinguish between market- and own-use production in agriculture are crucial under the 19th ICLS. Further exploration is needed on adapting these questions to account for seasonality, as well as proxy reporting. • Within off-farm wage work, amid shrinking land ownings and rising seasonal work on farms, asking about traditional forms of casual labor on others’ farms that may be missed by standard questions on wage employment is also important (as in the case of ganyu in Malawi, where rural women wage workers across most age groups are almost entirely concentrated). • Greater attention is needed to how questions are framed around self-employment, to identify individuals running a farm/enterprise separately from contributing family work. Owing to social norms and other time burdens, rural women are more likely to underreport this work. Recovery questions have also been shown to be particularly effective in re-classifying women as contributing family workers, when they initially report not being employed. III. Labor underutilization and time use Labor underutilization Given the narrower definition of employment under 19th ICLS, and the role of seasonality in rural jobs, the 19th ICLS recommends collecting data on two measures of labor underutilization — time-related underemployment, and the potential labor force — to better understand outcomes for individuals who would like to work more, or are out of the labor force. Time-related underemployment refers to whether individuals in employment, over a short reference period (a) wanted to work additional hours, (b) whose working time in all jobs was less than a specified hours threshold, and (c) who were available to work additional hours given an opportunity for more work (ILO, 2013; para 4). 4 Time-related underemployment has significant bearing on women as well; often when women get paid jobs, for example, their household responsibilities do not decline, and they are more likely to work part-time even if they wanted to work more (Folbre, 2018). 4 The 19th ICLS also provides guidance on how to construct different thresholds based on working hours (ILO, 2013). 10 The measurement of time-related underemployment relies on thresholds against which to compare working time/hours worked, as well as questions on desire and availability to work. Desire and availability to work are still not included in many socioeconomic surveys, although some LSMS-ISA surveys (including the 2013-14 Uganda survey, presented in Table 2 and Figure 4 below) include variants of these questions. Accurate reporting of hours worked can be challenging when unpaid activities are conducted simultaneously with paid work — as in the case of contributing family workers, whom the ILO LFS pilot studies found to report, on average, the least number of hours compared to other employed groups (Benes and Walsh, 2018c). Better counting of employment, as discussed in Section II, can help address some of these issues. Proxy reporting (and in particular, women proxies) can also lead to lower estimates of hours worked, as discussed further in this section. Table 2, which presents data from the 2013-14 Uganda LSMS- ISA, also shows that women proxies from poor households (who are mostly rural) are significantly less likely to report desire to work than male proxies, even though the difference between own-reporting men and women is not significant (for the non-poor sample, gender gaps emerge in both own and proxy reporting). As with the issues raised in Section II around proxy reporting of market activity in agriculture, further investigation is needed across countries on how women proxies, particularly in poorer contexts, might affect reporting. Table 2. Uganda National Panel Survey, 2013-14: share of women and men who would like to work more paid hours, by own- versus proxy reporting, and poverty status Women (W) Men (M) Gender gap Gender gap (W-M, own (W-M, proxy reporting) reporting) Own Proxy Diff. (own- Own Proxy Diff. (own- reporting proxy) reporting proxy) Below poverty line (sample) 545 475 327 584 Would like to work more (additional 0.62 0.47 0.15*** 0.65 0.54 0.11*** 0.03 -0.07** paid hours in same/different job) [0.49] [0.50] [0.48] [0.50] 0.36 0.42 -0.06** 0.34 0.36 -0.03 0.02 0.06** Would not like to work more [0.48] [0.49] [0.47] [0.48] 0.02 0.10 -0.08*** 0.01 0.10 -0.09*** 0.01 0.002 Don’t know [0.13] [0.30] [0.12] [0.30] Above poverty line (sample) 1786 1496 1173 1992 Would like to work more (additional 0.49 0.34 0.15*** 0.59 0.42 0.17*** -0.10*** -0.08*** paid hours in same/different job) [0.50] [0.47] [0.49] [0.49] 0.48 0.57 -0.09*** 0.39 0.48 -0.08*** 0.09*** 0.09*** Would not like to work more [0.50] [0.49] [0.49] [0.50] 0.02 0.10 -0.08*** 0.03 0.09 -0.06*** -0.01 0.01 Don’t know [0.14] [0.30] [0.17] [0.29] Notes: (1) Source: UNPS 2013-14 round. The question on desire to work additional hours was asked, for those employed, as follows: “In the last 7 days, would [NAME] have liked to work more hours than [NAME] actually worked, provided the extra hours had been paid?” (1=Yes, in the current job 2=Yes, in taking an additional job 3=Yes, in a different job with more hours 4=No, and 9=Don’t know). On availability to work, recent LSMS surveys ask respondents if they would “want to work if a job or business opportunity becomes available.” In the ILO LFS study, respondents were asked two separate questions on (a) whether they would want to work more paid hours per week than usual (desire to work), and (b) whether they could start working those hours in the next two weeks (availability to work). (2) T-tests of equality of means were conducted between proxy and own-reporting, as well as across men and women within each of these categories. *** p<0.01, ** p<0.05, * p<0.1. 11 Figure 4 also underscores the importance of examining these relationships by income and age — among poor households, younger women and men are equally likely to want to work additional paid hours. This diverges substantially, however, among groups aged 35 and older, likely due to women’s greater responsibilities within the home, including care of children. Also, some women may want to work more but are time constrained, so may consider themselves as neither available nor willing to work given their circumstances. As discussed further in Section V, asking additional contextual questions on reasons for wanting/not wanting to work more, including factors related to the quality of available economic opportunities as well as time constraints, can help in better understanding the trends in Figure 4. Figure 4. Uganda National Panel Survey, 2013-14: relationship between age and desire to work more (own-reporting sample) Households below the poverty line Households above the poverty line Finally, to complement statistics on unemployment, the 19th ICLS also introduced the concept of the potential labor force to provide additional data on those out of the labor force, but who could be responsive to employment policy. This includes (a) unavailable jobseekers, who are seeking employment but not available; (b) available potential jobseekers, who are available but not seeking; and (c) willing potential jobseekers, who are neither seeking nor available for employment, but who want employment. A question on desire to work, in particular, is critical for capturing (c) – and with important implications for women, who may be out of the labor force and not actively seeking work 5 because of unpaid work burdens, lack of suitable economic opportunities, not being allowed to work/other barriers to seeking work, or other social/cultural norms affecting their time allocation (see Section V as well). The ILO LFS study found, for example, that a much higher percentage of women cited family responsibilities as a reason for not searching for employment (Walsh, 2018). Wording can also matter — the ILO LFS study found that more women tended to report job search when the question on search was split into two (seeking a job, and seeking to start their own business) as opposed to a single question on seeking work for pay or profit (Benes and Walsh, 2018c). 5 19th ICLS defines the concept of unemployment as “active job search by persons not in employment who are available to work.” 12 Unpaid work burdens and the value of time use data As seen above, desire and availability for paid work, and hours worked, are not independent of unpaid responsibilities. Recent evidence from available time use surveys across countries shows that women’s total (paid and unpaid) work burdens are substantially greater than men’s (United Nations, 2015). Figure 5, using the 2013-14 Uganda LSMS-ISA, shows that time spent across paid and unpaid work is also higher for rural women, stemming from their engagement across paid and contributing family employment, own use production, and domestic activity. Figure 5 also indirectly highlights the role that unpaid care work plays in rural women’s work burdens; women’s time spent in unpaid work rises steeply in their childbearing years, whereas for men it falls, and women’s unpaid work burdens also fall much more slowly than time spent in paid work as men and women age. Figure 5. Uganda National Panel Survey, 2013-14: hours spent across paid and unpaid non-market activities in the last week Rural (own-reports) Urban (own-reports) Few socioeconomic surveys have detailed data on time spent across different areas of unpaid work (among the LSMS surveys, for example, the Uganda survey has the most detailed breakdown of time spent across multiple unpaid activities; see Figure 6). Well-designed time use surveys are critical in highlighting total work burdens, but are not without significant costs and measurement issues (Buvinic and King, 2018), and may be best as standalone surveys. Among socioeconomic surveys, narrowing the set of activity categories to those that are most contextually and policy-relevant is a priority measurement initiative under the EDGE- 2 project with UNSD and ILO (ILO, 2018d). Figure 6 shows, for example, that despite asking about time spent over multiple non-market labor activities, reported hours in the Uganda survey were concentrated in only a few areas (catch-all categories for agriculture and domestic work, in addition to water and fuel collection). Rather than a broad range of activities, therefore, surveys could break out specific activities within agriculture that are relevant for policy targeting. One such activity is time spent marketing/selling output — Koolwal (2018) shows that across the ILO LFS pilot countries, for example, around a third of women who were involved entirely in own use production in agriculture (and thus classified as not employed) nevertheless reported regularly selling a share of their agricultural production. Within domestic work, as well, breaking out time spent in childcare 13 is crucial. Collecting data on time spent in childcare in standard surveys can be complex, as it is often a secondary activity conducted with other domestic or paid work, and as a result underreported. The new Women’s Empowerment in Agriculture Index (WEAI) survey module includes more detail in the interviewer-guided introduction about time spent with children to lower the risk of being masked by other activities (Seymour, 2018); other approaches such as activity tracking devices or phone-based apps (Kilic, 2018; Daum, 2018) could help in recording which activities are conducted together/simultaneously, with less recall error. Figure 6. Uganda National Panel Survey, 2013-14: hours spent across paid and unpaid non-market activities in the last week Rural Urban Summary and ways forward By narrowing the definition of employment as well as highlighting the need to collect data on individuals’ total work burdens, the 19th ICLS motivates the need for additional data on labor underutilization, as well as ways that data on unpaid work could be better incorporated in socioeconomic surveys. This has important implications for rural women. Box 4 summarizes areas for future work. Box 4. Summary and ways forward: Labor underutilization and time use • Labor underutilization: Including questions on desire to work, along with reasons for wanting/not wanting to work, is important, to accompany questions on working hours, availability, and search. • Unpaid work burdens: in developing a refined set of unpaid work categories for socioeconomic surveys, key areas to cover for rural women include market activity in agriculture (significant even among own-use producers), as well as care work. Exploring newer, more cost-effective and direct approaches (including using mobile and other technology) to collect these data could possibly shed greater insight on their working day and secondary activities as well. • Proxy reporting also has a significant downward effect on reporting of hours worked in paid and unpaid work, particularly for women proxies. Reasons for this pattern need further investigation, and have implications for respondent choice going forward. 14 IV. Earnings and decision-making roles Complementing missing data on earnings In surveys, earnings are measured through (a) individual wage/salary earnings; (b) profit from the farm (measured at the household or plot level); and (c) profit from the nonfarm enterprise (measured at the enterprise level, which if own-account could be assigned to an individual household member). The quality of rural earnings data can vary substantially across country surveys, often resulting in missing and outlier values for occupations where rural women also tend to be overrepresented. In wage employment, for example, this includes informal and seasonal jobs that can be paid out irregularly and/or in-kind; and in self-employment, in family farms/businesses where the value of household members’ labor is not recorded (if, as is common in rural areas, multiple family members are involved). For individual wage/salary earnings, carefully collected community-level data on average wages or earnings for men and women, for formal/casual work across different sectors, could be useful as complementary data to understand expected labor market returns — particularly in contexts when many women are out of the labor force. For self-employment, within farming, several recent studies have discussed the use of individual-level data on productivity that is associated with farm income (see, for example, studies from Sub-Saharan Africa including Doss, 2018; Kilic et al, 2015). And as discussed in more detail below, for cases where income is pooled (across individual earnings, and farm/non-farm enterprises) the LSMS and other household surveys are exploring alternative questions about decision making over earnings to better understand individuals’ control over income. A joint pilot study in Sri Lanka by the World Bank and ILO under the WWEP, to coordinate survey questions across the LFS and LSMS, will also test questions on decision making in agriculture (Gaddis and Palacios-Lopez, 2018, Walsh, 2018). A closer look at decision making over earnings Questions on decision-making over wage earnings, as well as income from self-employed farm/off-farm operations are receiving increasing attention in surveys, as well as recent rounds of the LSMS-ISA. A main issue with decision-making questions, however, is the choice of respondent, which as seen earlier with Ethiopia’s RCBP can lead to gender inequalities in program targeting. A growing number of within and cross- country studies are also showing that men and women within the same household can respond very differently to the same sets of questions on the main decision maker (see Donald et al, 2017 for a cross- country study from Sub-Saharan Africa; Anderson et al, 2017 from rural Tanzania). One approach to address this issue (albeit more costly) is to ask both the most knowledgeable adult man and adult woman in the household. The Demographic and Health Surveys ask both husbands and wives, for example, in decision making questions over their spouse’s earnings and large household purchases. Within farming, the WEAI survey module also asks both men and women about decision making roles in production and income (Malapit and Quisumbing, 2015). The Uganda GSARS pilot also had two respondents for the module on decision making – (a) the plot holder, and (b) the holder’s spouse/partner if they lived in the household and were also engaged in agricultural work on the holding (GSARS, 2017). Similarly with enterprise income, cross-checking ownership and management of the enterprise across male/female respondents would be important, to understand whether gender norms affect reporting of these outcomes — including whether women entrepreneurs themselves underreport their decision- making roles in the business, as was found in an entrepreneurship training program in rural Pakistan (Giné and Mansuri, 2014). Using the 2016-17 Malawi survey, for example, Figure 7 compares men’s and women’s responses on whether they ran a business in the household labor module, with responses in the non-farm 15 enterprise module on who owns/manages each business (in the enterprise module, the respondent for each enterprise may not be the same as the business owner/manager; in the data, there was also nearly total overlap between ownership versus management). Where men were the respondents in the non-farm enterprise module, they agreed with women who reported in the labor module that they ran a business in 68 percent of those cases. Women who were respondents in the enterprise module, however, agreed with men who reported operating a business in the labor module in 86 percent of cases. Gendered views of business ownership are likely a major factor for this discrepancy. In this case, the value-added of decision making questions over ownership/management of the enterprise is questionable without ensuring consistency of responses across men and women in the same household. For example, a similar analysis of the 2013-14 Uganda LSMS-ISA revealed a much lower degree of inconsistency, as there was closer matching of respondents across the labor and non-farm enterprise modules. Figure 7. Malawi LSMS-ISA, 2016-17: share of respondents in NFE module, that agree with men and women who report in labor module that they run a business Total sample (proxy+own-reporting in labor module) Notes: Sample sizes of men/women respondents in NFE module presented on each bar. Within self-employment, an additional issue is that contributing family workers, often women, may also have significant roles in managing the business. Asking contributing family workers whether they were involved in the day-to-day administration of the business and earnings, as well as the frequency of involvement (on a regular basis; only during the current period/season; from time to time; or rarely), can help clarify economic roles for men and women beyond legal ownership of the business, and in doing so inform policy making (ILO, 2018c). Such questions were also recommended by the 20th ICLS in the revision of ICSE-18. Summary and ways forward Since a large share of rural women’s employment is concentrated in contributing work in family farms/businesses, as well as informal wage employment that can be paid irregularly, strictly relying on individual earnings data to make comparisons on the value of women’s and men’s labor may not be very informative. Box 5 presents some recommendations from this section. 16 Box 5. Summary and ways forward: Earnings and decision making • Particularly in contexts where seasonal/irregular wage employment is common, using community data on average wages for men and women, across more disaggregated categories/sectors of employment, can help complement individual-level data on wage/salary earnings — and enhance an understanding of expected labor market returns. • Within household farms/businesses, given difficulties in valuing individual family members’ labor contributions to household income, decision-making questions over earnings and management of operations can be helpful. Reported decision-making roles, however, may not be consistent across different household members, and trying to ensure the same respondents across modules is important (for example, across labor and non-farm enterprise modules). Asking about contributing family workers’ decision-making roles is also important for policies on rural women. V. Rural transition and barriers to economic mobility Rural economies are in transition around the world, although country experiences vary in the pace of growth. In some countries, rural women have benefited from improved access to improved inputs and technology, as well as greater access to labor and output markets and accelerated growth in production stemming from agricultural commercialization (see IFC, 2016 for case studies from Bangladesh and Indonesia). However, rural women have been left out in many other cases. This section discusses areas of data collection needed to better understand, for policy making, how (a) rural women’s employment is affected by structural changes occurring in rural economies and (b) factors affecting their ability to take advantage of new economic opportunities. Additional features of women’s employment amid rural transitions Migration Rapid out-migration to urban areas has significant implications for rural women. Rural women often face greater barriers to migrating for work, but nonetheless find their work and time allocation are affected when other male household members do so. Mu and van de Walle (2011) and Slavchevska et al (2016), for example, discuss how rapid economic development and rural-urban migration led to lower engagement in paid work and greater on-farm work for women who remained in rural areas, but not for men. And when rural women do migrate, they are much more likely to be engaged in vulnerable, informal forms of work (ILO, 2018a). Additional individual-level questions on migrants’ employment outcomes, including place and date of migration, would help shed light on how economic development is associated with work-related migration and outcomes of remaining household members, including women. Among the LSMS-ISA, currently only the 2015-16 Ethiopia survey asks individual-level questions on migrants spanning these topics. The 2013- 14 Uganda LSMS-ISA asks on the second visit to the household whether any member from the roster has left, reasons why (economic/education/marriage/conflict), and where (region of the country, or abroad) they went. The question on locality of migration, if within the country, is asked at the district level.
 Having a similar, consistent set of questions across surveys and survey rounds is valuable in cross-country comparisons, as well as the longer-term effects of migration in these areas. 17 Data on remittances received from migrants (whether men or women, and if they had migrated for work) is also useful in understanding changes in rural men’s and women’s opportunities and income from rural- urban migration. Among the LSMS-ISA, the 2016-17 Malawi survey is the only one that asks about money sent back to the household from children aged 15 and older who have migrated (including the amount, frequency, mode of payment, and recipient within the household). Extending these questions to include parents who have migrated, across different country surveys going forward, would also be helpful. Skills development Despite the narrowing of rural gender gaps in school enrollment over time, rural women face substantial inequalities in skills development and the ability to seek improved economic opportunities. Broad questions remain on (a) understanding how to target interventions to narrow these gaps and (b) how skills and economic mobility translate into better employment outcomes. Several randomized controlled trial (RCT) program-level studies have introduced formal business training for women entrepreneurs but often with little effect, due perhaps to skills not improving enough or the right skills not being taught (see McKenzie and Woodruff, 2014). 6 Underreporting of rural women’s employment also makes targeting training programs and extension services difficult. At present, most nationally-representative household surveys typically do not ask skills-related questions beyond literacy and having formal schooling, which as seen in Van den Broeck and Kilic (2018) may not be relevant for designing policy in contexts where only low-skilled opportunities are available. The World Bank’s STEP skills measurement program has implemented household and employer-based surveys between 2012 and 2017 in 17 countries, spanning a range of cognitive and non-cognitive skills through objective and self-reported measures, but only covers urban adults aged 15-64. 7 To inform areas that surveys should cover on collecting data on different types of skills, and in turn how to evaluate outcomes from different types of employment-related programs, guidelines were developed during the 20th ICLS on the measurement of qualifications and skills mismatch — where skills span (a) job- specific/technical skills, (b) basic skills such as literacy, numeracy and ICT skills, and (c) skills that can be transferred from one job to another, including cognitive as well as non-cognitive skills such as personal initiative and managerial acumen) (ILO, 2018e). In low-income, wage labor contexts, relevant job-specific and basic skills would typically be focused on physical/labor-intensive capacities, but would range more widely for those running their own business/farm operations where numeracy-related skills (including financial acumen, for example) are relevant. Better data on non-cognitive factors associated with employment are also highly relevant in low-income, rural contexts, where gender norms on employment opportunities tend to be highly entrenched — approaches used in psychology have also been introduced in the WEAI survey module to understand the extent to which respondents’ economic actions are driven by self-motivation as opposed to others’ preferences (Malapit et al, 2017). Similarly, the importance of personal initiative (as opposed to acting out of necessity) for women entrepreneurs has been raised by Campos et al (2017) and Buvinic and Furst-Nichols (2014). 6 Few studies are focused in rural areas, however (see Calderon et al, 2013 for one exception from rural Mexico). 7 See http://microdata.worldbank.org/index.php/catalog/step/about for more details. 18 Productivity-enhancing technology in rural areas On economic mobility, another important area for future work is developing individual-level questions on use of technologies, including frequency and purpose of use, that can help expand economic opportunities and access to markets. Mobile phones, for example, can reduce costs of business and farm-related transactions, provide information on market prices and job opportunities, broaden personal and business networks, and ease access to financial services, including mobile money accounts that allow women better control over their income (Demirgüç-Kunt et al, 2017). Among the LSMS, the 2016-17 Malawi and 2013-14 Uganda LSMS-ISA ask about individual ownership of a mobile phone (in the household roster for Malawi and the module on asset ownership for Uganda), but not questions about how they use their phones that can improve policy design; only the 2015-16 Ethiopia LSMS-ISA asks about individual mobile money accounts in the module on savings. The link between mobile phone use and agriculture is also missing. The 2017 Global Findex provides guidance on use-related questions that can be policy-relevant (for example, how individuals have used their mobile phones in financial transactions, whether these transactions are for agriculture, and with what frequency). Another important issue within rural areas is the use of improved agricultural technology and inputs. Recent LSMS-ISA agriculture modules ask about the use of improved seeds and other inputs at the plot level, that could be compared across men and women plot owners/managers (along with jointly managed/owned plots), and linked to indicators of production and income. However, because women also make up a much smaller share of independent farmers, and are often not equally targeted by agricultural extension and technology initiatives, simply looking at use of improved technologies without understanding who is left out — and why — can skew an understanding of demand for technology. Direct questions on reasons for/against adoption that have been documented in more focused country-level studies are lacking in broader socioeconomic surveys, including skills/knowledge (Udry, 2018); information networks (Munshi, 2004; Moser and Barrett, 2006); other economic activities farmers may be involved in (Ibrahim, 2013); credit and other liquidity constraints (de Janvry et al, 2016); and other factors such as individuals’ desire to minimize volatility in expected profits (Zhu, 2019). Constraints to economic opportunity Along with data on migration, access to skills and technology, socioeconomic surveys need to better capture constraints that rural individuals often face in seeking these economic opportunities. Some surveys have introduced questions on whether women need permission to leave the house for work (as in the DHS), as well as safety and commuting time to work; these are important for understanding mobility constraints for wage workers, but are less relevant for home-based workers or farmers. Asking about a wider range of constraints in surveys is also in line with Kabeer (2012), who presents a research agenda on specific barriers and alleviating factors to women’s economic mobility. Survey questions in this area would essentially be qualitative/subjective in nature, but if asked carefully (and honed through cognitive testing to account for language and local context) could be helpful in understanding where policy might be more effectively targeted. Respondents could choose from multiple options affecting their economic mobility, including (a) time/distance to work, as well as markets and other institutions (financial, administrative, etc.); (b) social and work-related norms affecting the ability to work, work longer hours, and/or advance in one’s position — including discrimination in hiring and advancement, and the risk of harassment; (c) unpaid work burdens at home affecting available time, including childcare; and (d) lack of skills in that field (specifying which skills are needed). For self-employed individuals, additional response categories could include (a) access to inputs/technology (across both farm and off- 19 farm activities), and (b) market-related factors including demand for one’s products, prices, and costs, as well as access to local markets. Individuals’ motivations and confidence are also important, as discussed earlier. More contextual questions were also included in the ILO LFS pilot study on reasons for respondents wanting to change their employment situation, that could help in a better understanding of women’s labor supply decisions across different countries (for example, to better match skills, improve other working conditions, earn more income, etc.). In household surveys like the LSMS, questions on constraints to economic mobility are gradually being introduced, although not necessarily in a systematic way. The non-farm enterprise modules of the 2015- 16 Ethiopia and Nigeria LSMS-ISA, for example, ask questions about constraints faced in enterprise operations and growth, as well as opening a new business (see Box 6). To allow for comparisons across different household members who own enterprises, this question needs to be asked at the enterprise level (as is done in the Ethiopia survey; in the Nigeria data, this question appears to be aggregated at the household level.) Adding a similar question to the individual labor module is also important, to include household members who may not necessarily be running an enterprise but have aspirations to work in other areas. As questions in this area evolve, response categories also need to be relevant for both women and men. Looking at Box 6, for example, potential constraints to nonfarm enterprise growth span infrastructure, financing, market access, technology, and institutional issues. However, a key response category on unpaid work, and in particular child care, is missing. Given the importance of estimating women’s time burdens, including a set of options around this issue is imperative. Skills development, as well as social norms around areas that women are expected to be involved in (often lower-earning sectors) also need to be addressed in these questions (both in enterprise and labor modules). Box 6. 2015-16 Ethiopia and Nigeria LSMS-ISA: Questions on constraints to non-farm enterprise growth Question from nonfarm enterprise module: List three most important constraints to (a) non-farm business operations and growth (for each enterprise), and (b) preventing household members from opening a non-farm enterprise: Response categories: 20 Using the Ethiopia LSMS-ISA, Figure 7a also shows that response categories can be refined/condensed. Compared to other reported constraints, most enterprise owners overwhelmingly report “market access” as the biggest factor affecting enterprise growth — and within this group, distance and cost are the most common issues. Developing a more detailed set of response options around these market access issues is therefore useful, and conducting testing prior to survey implementation to tailor options to specific country contexts. Differences between men and women across all three categories of market-related constraints were also statistically significant, and Figure 7b also shows that reporting of these constraints followed markedly different trends for men and women by age (reporting of market-related constraints increases with men’s age, and falls — albeit with a flattening-out in between — with women’s age). Figure 7a. Ethiopia 2015-16 LSMS-ISA: Reported constraints to non-farm enterprise growth (own-reporting sample) Across all reported constraints Within those reporting market-related constraints Notes: There were 818 male and 811 female entrepreneurs in the own-reporting sample (and within those reporting market-related constraints, 489 and 511 men and women, respectively). Figure 7b. Ethiopia 2015-16 LSMS-ISA: Reported constraints to non-farm enterprise growth (own-reporting sample) 21 Overall, the data reveal important target areas from a gender perspective (for example, younger women and the prominence of market-related constraints) that policy could be targeted towards — but more refined data in this category of constraints would be useful. Specifically, looking more carefully within market access issues related to distance and cost — for example, commuting times, how often individuals can reach markets, infrastructure quality and modes of transport for products — would better inform how policy might be directed, particularly since distance/cost affect marketing of products as well. In surveys with a community module like the LSMS, better data on access to child care and caregiving arrangements can also help in understanding women’s labor underutilization amid rural economic transitions. This could include, for example, community-level data on care centers along with existing data collection on local schools — including how long schools and centers are open — as well as other questions on informal caregiving arrangements among family members or other individuals within the community, which could be more relevant for poorer, rural contexts. Summary and ways forward There are large survey data gaps in understanding how rural economic transition is associated with changes in individual employment outcomes. Data on constraints faced by rural women in seeking better economic opportunities or improving productivity are also very important for policy targeting (for example, in technology adoption programs). Box 7 summarizes these issues and priorities for survey data collection going forward. Box 7. Features of work amid rural transition, and barriers to economic mobility • Consistent individual-level data on work-related migration across surveys is needed (including data on individuals from the household that have migrated for work, place of migration, sector of work, and remittances) to understand how rural-urban migration has affected rural women left behind. • Survey questions on skills acquired (or needed), as well as questions on economic mobility (including, relevant to women, information on respondents looking to move to paid work in the formal sector, and those transitioning from home and subsistence production into market work), are still rare and need further exploration in socioeconomic surveys. • In addition to asking about desire to work, to better inform programs to improve economic mobility in rural areas, questions on constraints to seeking economic opportunities — questions relevant to both wage and self-employment, across safety, market access, and time-related constraints —should be included regularly in surveys. These outcomes are also tied to non-cognitive factors affecting employment, an important area to examine as well. As with constraints to seeking economic opportunities, understanding constraints to using technology are also important, given inequalities faced by women in areas such as mobile phone ownership and access to improved agricultural inputs and extension services. VI. Conclusions Policies can be better targeted with more accurate counting of rural women’s employment, as well as other important factors that affect their decisions to work and seek better economic opportunities. Shifts in international standards for data collection and measurement have also brought many work- and employment-related issues faced by rural women to the forefront — including underreporting of their 22 market activity, unpaid work burdens, informality, constraints to accessing markets, and inequalities in access to skills and productivity-enhancing technologies and inputs. Along with an overview of the policy and research landscape, this paper presents data from recent LSMS-ISA surveys (Ethiopia, Malawi, Nigeria, and Uganda), as well as recent country pilots conducted by the ILO, to highlight priority areas on measuring rural women’s employment. The recommendations in this paper outlined at the end of each section can broadly be categorized into two areas: first, what we already know and should currently be incorporated in surveys; and second, where survey methods still need to be explored through research and testing. Under the first set of recommendations, ensuring the accurate counting of rural women’s employment is important through boundary questions in agriculture, asking about different forms of casual labor, and framing/recovery questions to ensure key areas of self-employment are not underreported. Adding questions on desire to work can also be explored in the short term. On decision making, ensuring the same respondents are reporting across labor/non-farm enterprise modules is important (or that both men and women are asked). A question on constraints to seeking better economic opportunities is also needed, across safety, market access, and time-related constraints. The second category of recommendations requires further research, including how unpaid work burdens can be better elicited in surveys, why proxy reporting tends to under-report time in paid and unpaid work, and how community data could complement individual data on hard-to-measure topics such as wages and access to child care. It also includes developing a survey research agenda around measuring work amid rural economic transition, including individual-level data on work-related migration across surveys, skills and access to technology. Overall, this paper provides an initial overview of specific issues that should be highlighted in surveys to ensure more accurate counting of rural women’s employment, and in turn improving links to the design of rural employment policies. When collected over time, additional data in these areas can also help provide insights on the dynamics of women’s employment during economic cycles and rural economic shifts, and help in broader economic planning as well. References Anderson, C. Leigh, Travis W. Reynolds and Mary Kay Gugerty. 2017. “Husband and Wife Perspectives on Farm Household Decision-making Authority and Evidence on Intra-household Accord in Rural Tanzania.” World Development 90: 169–183. Beegle, Kathleen, Emanuela Galasso and Jessica Goldberg. 2017. "Direct and Indirect Effects of Malawi’s Public Works Program on Food Security.” Journal of Development Economics 128: 1-23. Benes, Elisa, and Kieran Walsh. 2018a. “Measuring Employment In Labour Force Surveys: Main findings from the ILO LFS pilot studies.” International Labour Organization. Benes, Elisa, and Kieran Walsh. 2018b. “ILO LFS Pilot Studies Cognitive Interviewing Tests: Methodology, Process and Outcomes.” International Labour Organization. 23 Benes, Elisa, and Kieran Walsh. 2018c. “Measuring Working time and Time-related Underemployment in Labour Force Surveys: Main findings from the ILO LFS pilot studies.” International Labour Organization. Buehren, Niklas, Markus Goldstein, Ezequiel Molina and Julia Vaillant. 2017. “The Impact of Strengthening Agricultural Extension Services: Evidence from Ethiopia.” World Bank Policy Research Working Paper 8169. Buvinic, Mayra, and Elizabeth M. King. 2018. “Invisible No More? A Methodology and Policy Review of How Time Use Surveys Measure Unpaid Work.” Data2X Report. Buvinic, Mayra, and Rebecca Furst-Nichols. 2014. “Promoting Women's Economic Empowerment: What Works?” World Bank Research Observer 31(1). Calderon, Gabriela, Jesse M. Cunha, and Giacomo De Giorgi. 2013. “Business literacy and development: Evidence from a randomized controlled trial in rural Mexico.” Banco de Mexico Working Paper, No. 2013- 24. Campos, Francisco, Michael Frese, Markus Goldstein, Leonardo Iacovone, and Hillary C. Johnson. 2017. “Teaching personal initiative beats traditional training in boosting small business in West Africa.” Science 357: 1287-1290. Comblon, Virginie, and Anne-Sophie Robilliard. 2017. “Are female employment statistics more sensitive than male ones to survey design? Evidence from three West African countries.” DIAL Working Paper, DT/2015-22. Data2X. 2017. Women’s Work & Employment Partnership: Phase One Activities & Outcomes. Technical Summary, October 2017. Davis, Benjamin, Stefania Di Giuseppe, and Alberto Zezza. 2017. “Are African households (not) leaving agriculture? Patterns of households’ income sources in rural Sub-Saharan Africa.” Food Policy 67: 153– 174. Daum, Thomas. 2018. “Using a Smartphone App to Collect Time Use Data in Zambia.” Presentation made at the Workshop on Improving Data Collection on Women and Youth Employment, World Bank and IFAD, November 2018, Rome. Deijl, Claudia, Agnes Andersson Djurfeldt, and Magnus Jirström. 2017. “Agricultural policy in Sub-Saharan Africa and its Relevance for Smallholder Farmers, Women and Youth.” Agriculture for Food Security 2030 Report 2017-1. de Janvry, Alain, Elisabeth Sadoulet, Manzoor Dar, and Kyle Emerick. 2016. “The Agricultural Technology Adoption Puzzle: What Can We Learn From Field Experiments?” Working Papers P178, FERDI. de la O Campos, Ana Paula. 2015. Empowering Rural Women Through Social Protection. Rural Transformations – Technical Papers Series #2, Social Protection Division, Food and Agricultural Organization. 24 de Luca, Loretta, Hélène Sahy, Saba Joshi, and Mayra Cortés. 2013. Learning from Catalysts of Rural Transformation. International Labour Organization, Geneva. Demirgüç-Kunt, Asli, Leora Klapper, Dorothe Singer, Saniya Ansar and Jake Hess. 2017. The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution. World Bank. Desiere, Sam, and Valentina Costa. 2019. “Employment data in household surveys: taking stock, looking ahead.” Working Paper, Development Data Group, World Bank. Donald, Aletheia; Gayatri Koolwal, Jeannie Annan, Kathryn Falb, and Markus Goldstein. 2017. “Measuring Women's Agency.” World Bank Policy Research Working Paper 8148. Doss, Cheryl. 2018. “Women and agricultural productivity: Reframing the Issues.” Development Policy Review 36:35-50. Dutta, Puja, Rinku Murgai, Martin Ravallion, and Dominique van de Walle. 2014. Right to Work? Assessing India's Employment Guarantee Scheme in Bihar. Equity and Development; Washington, DC: World Bank. Fletcher, Erin K., Rohini Pande and Charity Troyer Moore. 2018. “Women and Work in India: Descriptive Evidence and a Review of Potential Policies.” Faculty Research Working Paper, RWP18-004, Harvard Kennedy School. Folbre, Nancy. 2018. Developing care: Recent research on the care economy and economic development. International Development Research Centre. FAO. 2011. The State of Food in Agriculture 2010-11: Women in Agriculture: Closing the Gender Gap for Development. Rome. Fox, Louise, and Obert Pimhidzai. 2013. “Different Dreams, Same Bed: Collecting, Using, and Interpreting Employment Statistics in Sub-Saharan Africa: the Case of Uganda.” World Bank Policy Research Working Paper 6436. Gaddis, Isis, Gbemisola Oseni, Amparo Palacios-Lopez, and Janneke Pieters. 2019. “Measuring Farm Labor: Survey Experimental Evidence from Ghana.” World Bank Policy Research working paper no. WPS 8717. Gaddis, Isis, and Amparo Palacios-Lopez. 2018. “Counting Women’s Work and Employment: Recent Methodological Advances and Tracing the Links Between Data and Policy.” Presentations made at the 20th ICLS, October 2018, Geneva; and at the workshop on Improving Data Collection on Women and Youth Employment, World Bank and IFAD, November 2018, Rome. Giné, Xavier and Ghazala Mansuri. 2016. “Money or Management? A Field Experiment on Constraints to Entrepreneurship in Rural Pakistan”, Mimeo. World Bank. Global Strategy to improve Agricultural and Rural Statistics (GSARS). 2017. “Intra-household decisionmaking in agriculture: The value added of asking who makes the decisions around specific agricultural activities.” Working Paper No. 18. GSARS Working Paper: Rome. 25 Goldin, Claudia. 1995. “The U-Shaped Female Labor Force Function in Economic Development and Economic History.” In: Schultz, TP, Investment in Women’s Human Capital and Economic Development. University of Chicago Press, pp. 61-90. Holloway, Kyle, Zahra Niazi and Rebecca Rouse. 2017. Women’s Economic Empowerment Through Financial Inclusion: A Review of Existing Evidence and Remaining Knowledge Gaps. Innovations for Poverty Action, Financial Inclusion Program. Ibrahim, Kasirye, 2013. "Constraints to Agricultural Technology Adoption in Uganda: Evidence from the 2005/06-2009/10 Uganda National Panel Survey," Research Series 151979, Economic Policy Research Centre (EPRC). International Finance Corporation (IFC). 2016. Investing in Women along Agribusiness Value Chains. World Bank. International Labour Organization (ILO). 2018a. Women and Men in the informal economy: A statistical picture (3rd edition). Geneva. ILO. 2018b. World Employment Social Outlook: Trends 2018. Geneva. ILO. 2018c. Data collection guidelines for ICSE-18. 20th International Conference of Labour Statisticians, Room Document No. 4. ILO. 2018d. “Exploring Light Time-Use approaches for measuring productive activities.” What Counts: Statistics on Work, Brief No. 1, August. ILO. 2018e. “Guidelines concerning measurement of qualifications and skills mismatches of persons in employment.” ICLS/20/2018/Guidelines. ILO. 2013. Resolution I concerning statistics of work, employment and labour underutilization. 19th International Conference of Labour Statisticians, Geneva. Jayachandran, Seema. 2015. “The Roots of Gender Inequality in Developing Countries.” Annual Review of Economics 7: 63-88. Johnston, Deborah, Sara Stevanob, Hazel J. Malapit, Elizabeth Hulla, and Suneetha Kadiyala. 2018. “Time Use as an Explanation for the Agri-Nutrition Disconnect: Evidence from Rural Areas in Low and Middle- Income Countries.” Food Policy 76: 8-18. Jones, Nicola, Yisak Tafere and Tassew Woldehanna. 2010. “Gendered risks, poverty and vulnerability in Ethiopia: to what extent is the Productive Safety Net Programme (PSNP) making a difference?” Working paper, Overseas Development Institute. Kabeer, Naila. 2012. “Women’s Economic Empowerment and Inclusive Growth: Labour Markets and Enterprise Development.” IDRC Research Paper. 26 Kilic, Talip. 2018. “Focus on Women’s Employment: The Value of Time.” Presentation made at the Workshop on Improving Data Collection on Women and Youth Employment, World Bank and IFAD, November 2018, Rome. Kilic, Talip, Amparo Palacios-Lopez and Markus Goldstein. 2015. “Caught in a Productivity Trap: A Distributional Perspective on Gender Differences in Malawian Agriculture.” World Development 70: 416– 463. Kochhar, Kalpana, Sonali Jain-Chandra, and Monique Newiak (eds). 2017. Women, Work and Economic Growth: Leveling the Playing Field. International Monetary Fund, Washington DC. Koolwal, Gayatri. 2018. “Counting Rural Women’s Work Under 19th ICLS: New Insights from ILO’s Country Pilots.” Working Paper, Data2X. McKenzie, David, and Christopher Woodruff. 2014. “What are we learning from business training evaluations around the developing world?” World Bank Research Observer 29(1): 48-82. Mammen, Kristin, and Christina Paxson. 2000. “Women’s Work and Economic Development.” Journal of Economic Perspectives 14(4): 141-164. Malapit, Hazel J., Kathryn Sproule, and Chiara Kovarik. 2017. “Using cognitive interviewing to improve the Women’s Empowerment in Agriculture Index survey instruments: Evidence from Bangladesh and Uganda.” Journal of Gender, Agriculture and Food Security 2(2): 1-22. Malapit, Hazel J. L., and Agnes R. Quisumbing. 2015. “What dimensions of women’s empowerment in agriculture matter for nutrition in Ghana?” Food Policy 52: 54-63. Mellor, John W. 2017. Agricultural Development and Economic Transformation: Promoting Growth with Poverty Reduction. Palgrave Studies in Agricultural Economics and Food Policy (Christopher Barrett, ed.), Palgrave Macmillan. Moser, Christine M., Barrett, Chris B. 2006. “The complex dynamics of smallholder technology adoption: The case of SRI in Madagascar.” Agricultural Economics 35 (3): 373–388. Mu, Ren, and Dominique van de Walle. 2011. “Left behind to farm? Women's labor re-allocation in rural China.” Labour Economics 18: issue S1, S83-S97 Mueller, Bernd and Man-Kwun Chan. 2015. Wage labor, agriculture-based economies, and pathways out of poverty: taking stock of the evidence. LEO Report No. 15, USAID. Munshi, Kaivan. 2004. “Social learning in a heterogeneous population: Technology diffusion in the Indian Green Revolution.” Journal of Development Economics 73(1): 185–213. Palacios-Lopez, Amparo, Luc Christiaensen, and Talip Kilic. 2017. “How much of the labor in African agriculture is provided by women?” Food Policy 67:52–63. Sarkar, Sudipa, Soham Sahoo, and Stephan Klasen. 2018. “Employment Transitions of Women in India: A Panel Analysis.” Working Paper, IZA Institute of Labor Economics. 27 Seymour, Greg. 2018. Discussant comments, Workshop on Improving Data Collection on Women and Youth Employment, World Bank and IFAD, November 2018, Rome. Seymour, Greg; Hazel J. Malapit, and Agnes Quisumbing. 2017. “Measuring Time Use in Development Settings.” World Bank Policy Research Working Paper No. 8147. Slavchevska, Vanya, Susan Kaaria, and Sanna- Taivalmaa, Sanna-Liisa. 2016. “Feminization of Agriculture in the Context of Rural Transformations: What is the Evidence?” World Bank, Washington, DC. Udry, Chris. 2018. “Information, Market Access and Risk: Addressing Constraints to Agricultural Transformation in Northern Ghana.” Presentation made at Center for Development Economics and Policy workshop, Columbia University School of International and Public Affairs, November. United Nations. 2015. The World’s Women 2015: Trends and statistics. Department of Economics and Social Affairs, Statistics Division. Valenzuela, Abel. 2003. “Day Labor Work.” Annual Review of Sociology 29: 307-333. Van den Broeck, Goedele, and Talip Kilic. 2018. “Dynamics of Off-Farm Employment in Sub-Saharan Africa: A Gender Perspective.” World Bank Policy Research Working Paper 8540. Walsh, Kieran. 2018. “Measuring Women’s and Men’s Work: the 19th ICLS, Purpose and Progress.” Presentations made at the 20th ICLS, October 2018, Geneva; and at the workshop on Improving Data Collection on Women and Youth Employment, World Bank and IFAD, November 2018, Rome. Zhu, Jessica. 2019. “Heterogeneous Farmers’ Technology Adoption Decisions: Good on Average Is Not Good Enough.” Working Paper, Department of Agricultural and Applied Economics, University of Wisconsin-Madison. 28