SYSTEMATIC REVIEW ON WOMEN’S ECONOMIC EMPOWERMENT MARCH 2023 WHAT DO WE KNOW ABOUT INTERVENTIONS TO INCREASE WOMEN’S ECONOMIC PARTICIPATION AND EMPOWERMENT IN SOUTH ASIA? FINANCIAL PRODUCTS Amna JavedŦ, Najaf ZahraŦ, and Ana Maria Munoz BoudetŦ BACKGROUND evidence for policies and financial inclusion programs im- plemented in any South Asian country and which directly The World Bank’s South Asia Region Gender Innovation Lab aimed to change women’s economic outcomes or have (SAR GIL) is conducting a systematic review and meta-analysis indirectly done so. The study summarizes the effects of pro- of interventions with direct or indirect effects on measures viding access to grants or loans, as well as varying features of women’s economic empowerment. The review focuses of credit contracts, such as repayment timelines. In addition, on changes in labor market outcomes, income, and other the brief distinguishes between programs targeting current empowerment indicators. The goal is to document what has business owners as well as programs that seek to encourage and has not worked for women in the region (covering all entrepreneurship and discusses why program impacts vary countries: Afghanistan, Bangladesh, Bhutan, India, Maldives, by gender within the South Asia region. The brief concludes Nepal, Pakistan, and Sri Lanka), understand the types of by highlighting gaps in the literature and recommending interventions implemented, and identify gaps in knowledge areas for further research. and action. Interventions are organized into five categories: The review includes English-language studies published Skills, Assets, Financial Products, Care, and Empowerment.1 between January 1990 and March 2020 across white and This brief summarizes the main findings from the Financial gray literature (peer reviewed journals, working papers, Products category. program or agency reports, and academic theses, among others) identified via an extensive search of multiple da- tabases.2 Intervention inclusion was not limited by time, WHAT IS INCLUDED? duration, frequency, or method of exposure. Figure 1 summarizes the three-stage identification process. The Existing systematic reviews have evaluated the provision of first stage filtered select papers relevant to the region and financial products within multi-regional contexts but with- programs that were specifically for women or included out distinguishing the effects for subgroups of women alone. female beneficiaries. The second stage filtered for inter- This review includes experimental and quasi-experimental vention type and the third stage for methodology.3 Two Ŧ World Bank. 1 Previously, the interventions were divided into six categories. However, the Credit and Entrepreneurship categories have been pooled into the Financial Products theme. This was done because the Entrepreneurship theme primarily included business training or business training plus credit programs. The Skills theme includes the business training interventions, while the Credit theme now includes the business training plus credit interventions. The Labor Markets theme has been replaced with a Care category, which covers any child or elder care interventions that may affect women's empowerment or labor market outcomes in the region. This replacement was made due to a lack of studies uniquely associated with the Labor Markets theme. 2 The search included the following databases: Econlit, Web of Science, Science Direct, National Bureau of Economic Research (NBER), Google Scholar, World Bank e-Library, UNWider, Abdul Latif Jameel Poverty Action Lab (J-PAL), Institute of Labor Economics (IZA), Center for Global Development (CGD), International Growth Center (IGC), American Economic Association (AEA), AEA Registry, International Initiative for Impact Evaluation (3iE), Research Papers in Economics (RePEc), IDEAS database, and JSTOR. 3 Second stage search terms included entrepreneur, entrepreneurship, business, start-up or startup, vendor, training, microcredit, microfinance, loan, loans, microloans, cash transfer, credit transfer, credit, borrow, borrower, lottery, saving or savings, lending, group lending, finance, digital finance, transaction, interest rate, interest, banking, and digital banking. Third stage search terms included comparison group, counterfactual, counter-factual, evaluation, assessment, impact, rct, randomized control trial, impact evaluation, quasi experiment, quasi-experiment, propensity score matching, psm, regression discontinuity design, rdd, and discontinuous design. Figure 1: Search Methodology Identi cation A preliminary list of Backward and forward Resources were rechecked Stage 2: Snowballing databases was searched snowballing was conducted using the World Bank Stage 1: Base Search library lists and the Stage 3: Recheck connected paper’s website Studies were stored in a The process was repeated Final studies were added snowballing repository un�l no new studies were to a thema�c database Screening iden�fied Eligibility decisions were made a�er reading the �tle and abstract of each resource. Addi�onal scoping was done to iden�fy outcomes and methodology for some ar�cles. Each study outcome was assigned to a category, either Eligibility employment, empowerment, or income. Poten�al papers were also checked for their iden�fica�on strategy. Key informa�on about programs and par�cipants was extracted for each study including type of interven�on, sample popula�on, econometric methodology, and impact details. All studies added were given unique IDs based on interven�on type, popula�on, econometric methodology/specifica�on, year, and outcomes. If a study reported impact es�mates using more than one specifica�on, it was coded using different codes for methodology/specifica�on. Key Information reviewers independently searched and extracted data worked), income or profit, and empowerment (including, from the list of finalized articles, including impact effects, among others, agency, well-being, happiness, mobility, design, and intervention components. Additional out- financial or political empowerment). come-specific data, such as units of reporting, coefficient The selection criteria required the inclusion of those stud- significance, and standard errors were also extracted. If ies that evaluated programs with a rigorous methodology a study reported impact estimates using more than one and included outcomes for women.5 Of the 29,332,410 specification, all were recorded, but only the researchers’ papers identified in the first stage of the search process, preferred specification is used in this brief. approximately 6 percent (1,537,066) remained after Eligible studies were those that: filtration using the second stage search terms. Further • Evaluated a financial product, specifically grants or credit refinement in the third stage, removal of repetitions and (such as microfinance, individual or group-based loans).4 refinement by title and abstract led to 95 studies on rele- • Employed experimental or quasi-experimental evaluation vant financial products being compiled in a preliminary list. methods. These papers were read for methodology and relevance, • Reported outcomes for women, either as the direct then snowballed backward and forward, resulting in a final target population or a subpopulation of interest. list of 19 articles meeting the predefined inclusion criteria. • Reported required outcomes, including labor market The final sample includes studies from Bangladesh (2), outcomes (such as self-employment, participation, days India (9), Pakistan (5) and Sri Lanka (4).6 4 Asset-based or savings programs are included in previously published Asset and Self-Help Group systematic reviews and are not included here (Javed et al., 2022; Zahra et al., 2022). 5 The brief focuses on studies employing methods to rigorously separate treatment effects from pre-trends, biases, or any confounding effects. 6 Bernhardt et al. (2019) include samples from both India and Sri Lanka. 2  |  Systematic Review on Women’s Economic Empowerment PROGRAM DESIGN small home-based manufacturing (handicrafts, bag or mat manufacturing, tailoring). For re- The studies in the sample evaluate various financial spondents with existing businesses, the age of products including grants, loans, and microfinance. the business varies between 6 and 10 years on Interventions may have multiple arms in addition to the average, where reported. financial product, such as the inclusion of business training or flexibility in loan contracts. Table A.1. in the Appendix PROGRAM COST provides program and sample descriptions in detail, while Three studies in the sample provided program select intervention characteristics are summarized below. cost estimates. Giné and Mansuri (2021) offered business training and a loan lottery TYPE OF INTERVENTION to a sample of rural microfinance clients In the sample, five articles evaluate micro- and showed that the recurring costs (travel finance interventions alone or jointly with allowances, actual training session, salaries of training, four evaluate cash or capital grants staff) of training clients was $20.44, while the alone or jointly with business training, four total cost inclusive of fixed expenses (lodging, study credit alone or jointly with training salaries for trainers, development of material) provision, and six articles assess the impact was $126.32. The average loan amounted to of varying features of loan contracts.7 Loan approximately $764 (45,095 Pakistani rupees). sizes vary from 4,000 to 40,000 rupees, with In comparison, Field et al. (2016) offered busi- the median loan between 10,000 and 15,000 ness counselling and provided a subsample rupees. Interest rates for loans vary between with the option of inviting a friend to join the 12 and 24 percent. Standard loan contracts training. The authors calculated an average are characterized by repayments every week, cost of 157 Indian rupees per person for their every two weeks, or once a month depending two-day business counseling intervention, on the loan size, the frequency of meetings, including instructor fees, snacks, and trans- and the credit provider. portation reimbursement. In Sri Lanka, women were offered a nine-day training while half of PROGRAMS TARGET URBAN AREAS the sample were provided an additional cash Most interventions in the sample target grant of $129 (or 15,000 Sri Lankan rupees). participants residing in urban or peri-urban Apart from the cost of the grant, the authors areas. There are two exceptions. Montgom- calculated that the training provision amount- ery and Weiss (2011) study a sample with ed to $126-131 per person for current business only 25 percent of participants residing in owners, and $133-140 for potential business urban areas. Similarly, Giné and Mansuri owners (de Mel et al., 2012). (2021) study respondents in the context of rural Pakistan. IMPACT TIMELINE Studies measure impacts after two years on PROGRAM SAMPLE average, with four months and six years being Programs usually sample poor and low-income the shortest and longest interval between households. Respondents tend to be married the intervention and follow-up, respectively. and between 18 and 50 years old with an Most studies use randomized control trials to average age of approximately 35 years. Female evaluate the intervention, while three studies business owners usually operate in retail sales use quasi-experimental methods to determine (street vendors, small grocery stores) and program impacts. 7 In this brief, we adopt the terminology used by the authors to describe the type of intervention as either a credit or microfinance program. Microfinance interventions are usually offered by microfinance institutions which can require members to participate in other activities such as savings, and which target women from low-income households with initially small credit lines. However, we note that in our sample, loan sizes and household income levels are similar for other credit interventions. march 2023  |  3 PROGRAM IMPACTS a business in the next year, but this effect dissipated by the second year. In Sri Lanka, the authors found that both For this review, program impacts are separated into busi- the provision of training alone and the grant plus training ness ownership (measured as a binary variable, or as the increased the likelihood of potential business owners op- number of businesses owned), profit and other business erating a business four and eight months later (de Mel et outcomes (such as the number of employees or sales), al., 2014). However, these effects disappeared 16 months broader labor market, and empowerment outcomes. For after the intervention. Those who received the grant plus ease of interpretation, we present estimates in percent or training were 10 percentage points more likely to have percentage point changes wherever possible. In all cases, opened a business since baseline but 9 percentage points the results presented are estimated impacts for women, more likely to have opened and then closed their business, even where interventions have separate treatment arms leading to a net effect of zero on business ownership (de for men. Mel et al., 2014). The provision of training alone led to a smaller increase in business ownership after four months, A. BUSINESS OWNERSHIP but this effect also lost significance in subsequent survey rounds. Thus, it appears that in these contexts, providing A question of particular interest to policy makers is wheth- training plus a grant or loan speeds up entry into a busi- er the provision of financial products such as credit or ness but has no long-term impact on ownership rates for grants is an effective mechanism in encouraging women to take up entrepreneurship. aspiring business owners. Two studies in our sample provided financial products plus The impact of microfinance is similar. For example, training to female participants who did not own an existing households in India which received access to microfinance business at baseline but who were likely to enter self-em- were not more likely to have a business and did not have ployment, as measured by having developed a business a greater number of businesses on average, compared plan or knowing the type of business to start. In particular, to households without access to microfinance (Banerjee Said et al. (2017) provided a 12-month loan of 10,000- et al., 2015). However, the provision of credit alongside 30,000 Pakistani rupees in addition to a two-hour training repayment flexibility can have a positive effect on business class on marketing, networking and capacity building to ownership. Field et al. (2013) showed that groups who aspiring female entrepreneurs in urban and peri-urban received a two-month grace period before their first loan areas in Pakistan.8 Similarly, aspiring female entrepreneurs repayment were three times more likely to start a new in urban Sri Lanka were divided into two groups, with the business within six months of the loan receipt, presumably first receiving a nine-day business training program, and because the grace period allowed them to make riskier the second group receiving the same training and a grant investments. of $130 (15,000 Sri Lankan rupees) which they could use Not all women were equally likely to be attracted to en- for any purpose (de Mel et al., 2014).9 In both studies, the trepreneurship or have their business survive. In Pakistan, sample of women were on average 34 to 37 years of age younger women and women who saw their mothers and between 84 to 87 percent were married. However, the manage a business were more likely to set up a business study in Pakistan sampled women with 2.5 years of school- themselves; this result was not symmetrical for other ing, on average, while the aspiring entrepreneurs sampled relatives (Said et al., 2017). In Sri Lanka, training was more in Sri Lanka had approximately 10 years of education. likely to induce women with lower measured analytical Interestingly, both studies found their interventions in- skills to run a business, while training plus a grant led to duced women to start their own businesses in the short lower analytically skilled and poorer women to become run, but these effects were not sustained over time. Figure entrepreneurs (de Mel et al., 2014). In India, almost all the 2 visualizes the results for select outcomes. Specifically, businesses that were created after access to microfinance Said et al. (2017) found that women who received the were female owned, though these marginal businesses loan and training were five percent more likely to set up tended to be smaller and less profitable compared to 8 The sample of aspiring business owners in this study includes skilled women from low-income households who have a practical business plan and require a line of credit. 9 Aspiring business owners in this study were women who planned to enter self-employment in the next year. The authors further restrict the sample to those women who were able to identify which type of business they planned to start, and those who had childcare available for children younger than five years of age. 4  |  Systematic Review on Women’s Economic Empowerment Figure 2: Business Ownership, Select Outcomes Note: The figure shows select business ownership outcomes. The labels indicate the outcome measurement and time since program implementation. The intervention and authorship details are provided in the legend. Coefficients have been converted to percentage point changes. 95 percent confidence intervals displayed. the average business in the area (Banerjee et al., 2015). stock by an amount similar to the grant, though profits In Pakistan, Said et al. (2022) found that women who ran only increased temporarily for the first eight months (de a business from home and obtained advice from their Mel et al., 2014). The provision of business training alone husband were more likely to see their business survive. did not improve profits, sales, capital stock, or the number In addition, women who lived in households with other of hours that owners spent in their business four to 25 members who ran a business were less likely to have shut months after the program. These average effects often down their business. mask informative heterogeneity in impact. For example, Banerjee et al. (2019) found that six years after access to B. PROFIT AND BUSINESS OUTCOMES microfinance, business outcomes improved specifically for The impact of financial products on profit or business-re- individuals who owned a business prior to receiving micro- lated outcomes (such as capital stock or the number of finance. Those who had a pre-existing business spent 83 employees) is mixed and any positive effects are not found percent more on business inputs, produced 104 percent consistently across program, intervention type, subgroup, more in business revenue, owned 3,000 Indian rupees or time. more in business assets, and earned approximately 28 percent higher profits. Figure 3 displays select profit outcomes. In India, access to microfinance led to greater investments in existing One reason for the variation in business profit impacts businesses and profits increased for those enterprises that is that investment decisions can vary by empowerment were already profitable (Banerjee et al. 2015). Field et al. levels and the size of the loan. In Sri Lanka, men who (2016) found that while access to microfinance led to an received a treatment of 10,000 Sri Lankan rupees invested increase in income generated by women by 9 percent, this 138 percent of it (by including complementary resources effect faded in the long run. In Sri Lanka, business training or reinvesting business profits), but those who received a combined with a cash grant of $129 improved capital 50 percent larger treatment of 20,000 Sri Lankan rupees march 2023  |  5 Figure 3: Profit, Select Outcomes Note: The figure shows select profit outcomes. The labels indicate the outcome measurement and the time since program implementation. The intervention and authorship details are provided in the legend. Coefficients have been converted to percentage point changes. 95 percent confidence intervals displayed. only invested 63 percent. In line with these investment data from India and Sri Lanka, Bernhardt et al. (2019) found decisions, profit increases were greater for those who that while profits for female-owned enterprises did not invested more. In contrast, women did not invest the increase with the provision of loans or grants, households smaller treatment in their enterprise but invested about of treated women earned significantly more in aggregate 85 percent of the larger amount (de Mel et al., 2009). income. This is because women often invested in other However, real profits did not increase for women in household members’ enterprises. In India, women in either case. The authors suggest these differences are single-enterprise households who received credit in- not explained by entrepreneurial ability, attitudes toward creased their weekly profits by 81 percent, relative to the risk, or profit reporting behavior. Instead, they suggest control group. In contrast, women who received credit in that the types of investments female entrepreneurs make multi-enterprise households saw no impact on their profits, are constrained by their empowerment levels. Women although other household members’ enterprises had a 44 who are more empowered invest in inventories while less percent increase in weekly profits compared to the control empowered females invest in equipment that is difficult group. In Sri Lanka, grants had no impact on business profits to decapitalize (de Mel et al., 2009). Suboptimal decisions for women in multi-enterprise households, though monthly for profit improvement ensue when business investment profits of female entrepreneurs in single enterprise house- decisions are constrained by empowerment levels. holds increased by 22 percent (Bernhardt et al., 2019). Another factor contributing to the varying profit results is The impact on profit of flexible repayment interventions, whether the individual receiving credit resides in a single compared to standard credit contracts, is largely positive. For or multi-enterprise household. This is important because example, Aragón et al. (2020) provided a sample of female women may not always invest in their own business. Using street vendors in Maharashtra, India, with a flexible credit 6  |  Systematic Review on Women’s Economic Empowerment line which allowed them to obtain credit when they wanted, D. EMPOWERMENT for the amount they needed, and to be repaid at their choos- The evidence showed that financial product interventions ing. Six months after the intervention, women who were of- did not have a significant effect on empowerment outcomes. fered the flexible credit line had seven percent higher profits For example, in Pakistan, women who set up their own compared to clients receiving a standard credit line. Along business did not see an improvement in their household similar lines, women in Kolkata, India, who were offered a decision-making power or asset holdings within a one-year two-month grace period before their first individual-liability period (Said et al., 2017). Montgomery and Weiss (2011) also loan repayment, reported 57 percent higher weekly profits found no improved empowerment for female microfinance after three years, compared to respondents who did not borrowers, though wives of male borrowers saw small im- receive a grace period and made their first repayment two provements in certain measures of family decision making, weeks after receiving the loan (Field et al., 2013).10 such as whether to have another child. Similarly, Banerjee Flexible credit lines can increase profits by allowing et al. (2015) did not find any significant effect on an index of business owners to better match their cash flow to their 16 social empowerment outcomes a year or two years after borrowing and repayments and therefore make more receiving access to microfinance.11 In India, Field et al. (2016) profitable investments (Aragón et al., 2020). Similarly, showed that access to microfinance does not change em- grace periods in loan repayments allow clients to invest in powerment, as measured by managing financial resources illiquid but high return investments, particularly benefiting or with respect to household expenditure decisions, though women who are relatively risk averse and have more busi- it did positively affect the probability of a respondent having ness experience and skills. In fact, profits increased by 92 separate savings from her husband and the likelihood that percent for a risk averse subgroup of clients who received she had the major say with respect to investment in the a two-month grace period before their first loan repay- household business. In Bangladesh, respondents with access ment (Field et al., 2013). In contrast, explicit commitment to microfinance were more likely to be prevented from work- features, such as text reminders regarding repayments, ing by their husbands or other members of the household. do not significantly impact profit or business ownership, The women were also likely to have more traditional beliefs possibly because the demand for commitment is already toward gender and social norms after access to microfinance; met by regular repayment structures (Afzal et al., 2019). they were 20 percentage points more likely to believe that women should not earn more than their husbands, and 14 C. LABOR MARKET percentage points more likely to believe that women should not take up employment outside the home if they have The impact on labor market outcomes is mixed, with young children (Asadullah et al., 2021). some evidence for increased participation in household businesses. For example, Banerjee et al. (2015) found that the household head and spouse in households with access to microfinance increased their labor supply entirely in the CONCLUSION household business by 3.2 hours. In India, Field et al. (2016) found significant increases in labor force participation after In conclusion, this review highlights several patterns from access to microfinance, driven by greater participation in the literature. businesses that were actively operated and managed by First, the provision of financial products can induce wom- female respondents. In Bangladesh, access to microfinance en to enter entrepreneurship, but these effects are not was associated with a 2-percentage point increase in female sustained over time. labor force participation in traditional employment, defined as homestead farming and raising livestock, though there Second, evidence for improvements in profit, labor was no change in self-employment (Asadullah et al., 2021). force participation, and empowerment is mixed. One 10 The literature also underlines the positive impacts of increased group meeting frequency. In West Bengal, India, women who met in weekly meetings with weekly repayments were 5 percent less likely to default on their loans and socialized 37 percent more often with their group members outside of the meetings, compared to the control group which met and repaid in monthly intervals (Feigenberg et al., 2013). The lower default rates are driven by increased social interactions rather than differences in repayment frequencies. In a later study, the authors show that the gains are greatest for women who have less developed social networks at baseline and social empowerment, as measured by having to ask their spouse for permission to go to a neighbor’s house (Feigenberg et al., 2014). 11 The authors include 16 social outcomes in their index: indicators for women making decisions on each of food, clothing, health, home purchase and repair, education, durable goods, gold and silver, investment; levels of spending on school tuition, fees, and other education expenses; medical expenditure; teenage girls’ and teenage boys’ school enrollment; and counts of female children under one year and one- to two-years-old. march 2023  |  7 explanation for these results is that women who are not suggest that the binding constraint for women may not suited for business ownership may be incentivized to opt always be capital and skills, but a lack of empowerment. for entrepreneurship when there is a lack of other suitable Research designed to ease financial constraints jointly employment opportunities. In addition, female entrepre- with empowerment interventions can improve our neurs face multiple constraints beyond the liquidity and understanding of the interaction of a woman’s baseline financial limitations also faced by their male counterparts. empowerment with her entrepreneurial success. The Not only are their investment decisions bound by their inclusion of treatment arms targeting male household members or partners to remove social barriers regarding empowerment and the threat of expropriation faced from female entrepreneurship can further our understanding within the household, but their business knowledge and of the importance of considering household skills may not translate into profitable enterprises if they constraints when implementing interventions to boost lack control and decision-making power in their business. entrepreneurship. Moreover, mobility restrictions, child and elder care, • Will one-on-one training make a difference? There is a lack household responsibilities, and a lack of empowerment of research in South Asia on the effectiveness of providing can all contribute to women choosing smaller scale or credit jointly with more intensive one-on-one training home-based businesses in industries considered accept- for female entrepreneurs. Evidence from Peru suggests able for female entrepreneurs, as opposed to pursuing the that general training cannot transfer the necessary most profitable opportunities. In one example, women in management skills that are required to improve business Pakistan were willing to give up 3,000 Pakistani rupees in success, but more intensive training can have a positive profits, amounting to 60 percent of the median profits in effect on business revenue (Valdivia 2015). the sample, just to keep the business in the home (Said et • What is the long run impact of interventions? There al., 2022). is a need to study the impact of financial product interventions over a longer period to trace the trajectory Third, the provision of flexibility in repayment schedules of profits, business survival, and investments beyond a has a positive and large impact on women’s business three-year horizon. profits. Traditional microfinance and credit interventions • What impact will scaled-up programs have? More have the potential to remove credit constraints. However, research is needed to understand the effects of loan terms such as immediate repayments, inflexible providing financial product interventions at scale. terms, and high interest rates can make borrowing for Specifically, can scaled-up programs increase business entrepreneurship risky and unattractive. Providing flexi- ownership and local competition and thereby bility in repayments by delaying the first reimbursement incentivize women to choose non-traditional or non- or allowing respondents to postpone repayment in in- home-based enterprises? come-constrained months can encourage women to make • Can financial products encourage women to cross high-return investments even if they are riskier. over into male-dominated industries? There remains a need to evaluate the effectiveness of financial product The current evidence highlights many new questions of interventions that ease capital constraints while interest and further areas for research: simultaneously encouraging female entrepreneurs to • Are the right women being targeted? The lack of enter male-dominated industries that offer greater consistent improvement in business outcomes and profit profitability. This encouragement may include providing in the medium- to long-run suggests that perhaps the women with skills-based training, industry-specific “correct” group of women has not yet been introduced knowledge, and even networking or apprenticeship to entrepreneurship with credit and microfinance opportunities for industries typically dominated by men. interventions. More research is required to understand • Will digital marketplace training benefit women business both the combination of traits possessed by successful owners? Certain internet-based marketplaces, such female entrepreneurs and the environment that enables as Etsy, Instagram, or Facebook Marketplace, allow them to thrive. Further, can interventions replicated to a businesses to target a much wider consumer base while sample of female non-business owners with these traits permitting women to remain home-based. Programs that provide sustained and positive results? provide women with capital and the necessary training to • What role do social and household constraints play? The exploit and navigate these digital marketplaces should be evaluations of combined training and credit interventions tested for their success. 8  |  Systematic Review on Women’s Economic Empowerment REFERENCES Feigenberg, Benjamin, Erica Field, Rohini Pande, Natalia Rigol, and Shayak Sarkar. 2014. “Do Group Dynamics Influence Social Capital Gains Among Afzal, Uzma, Giovanna D’Adda, Marcel Fafchamps, Simon R. Quinn, and Microfinance Clients? Evidence from a Randomized Experiment in Urban Farah Said. 2019. “Implicit and Explicit Commitment in Credit and Saving India,” Journal of Policy Analysis and Management, 33 (4): 932-949. DOI: Contracts: a Field Experiment.” NBER Working Paper 25802. https://www. 10.1002/pam.21790. nber.org/system/files/working_papers/w25802/w25802.pdf Field, Erica, Jose Martinez, and Rohini Pande, 2016. “Does women’s bank- Aragón, Fernando M., Alexander Karaivanov, and Karuna Krishnaswamy. ing matter for women? Evidence from urban India,” IGC Working Paper. 2020. “Credit lines in microcredit: Short-term evidence from a random- https://www.theigc.org/wp-content/uploads/2016/11/Field-et-al-2016- ized controlled trial in India.” Journal of Development Economics, 146. Working-Paper.pdf DOI:10.1016/j.jdeveco.2020.102497. Field, Erica, Rohini Pande, John Papp, and Natalia Rigol. 2013. “Does the Asadullah, M Niaz, Nudrat Faria Shreya, and Zaki Wahhaj. 2021. “Access Classic Microfinance Model Discourage Entrepreneurship among the to Microfinance and Female Labor Force Participation.” WIDER Working Poor? Experimental Evidence from India.” American Economic Review, Paper 2021/30. https://www.wider.unu.edu/sites/default/files/Publica- 103 (6): 2196-2226. DOI: 10.1257/aer.103.6.2196 tions/Working-paper/PDF/wp2021-30-access-microfinance-female-la- bour-force-participation.pdf Field, Erica, Seema Jayachandran, Rohini Pande, and Natalia Rigol. 2016. “Friendship at Work: Can Peer Effects Catalyze Female Entrepreneurship?” Banerjee, Abhijit, Emily Breza, Esther Duflo, and Cynthia Kinnan. 2019. American Economic Journal: Economic Policy, 8 (2): 125-53. DOI: 10.1257/ “Can Microfinance Unlock a Poverty Trap for Some Entrepreneurs?” NBER pol.20140215 Working Paper 26346. http://www.nber.org/papers/w26346. Giné, Xavier and Ghazala Mansuri. 2021. “Money or Management? A Field Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Kin- Experiment on Constraints to Entrepreneurship in Rural Pakistan.” Eco- nan. 2015. “The Miracle of Microfinance? Evidence from a Randomized nomic Development and Cultural Change, 70 (1). DOI: 10.1086/707502. Evaluation.” American Economic Journal: Applied Economics, 7 (1): 22-53. DOI: 10.1257/app.20130533. Javed, Amna, Najaf Zahra, and Ana Maria Munoz Boudet. 2022. "What do We Know About Interventions to Increase Women's Economic Participa- Battaglia, Marianna, Selim Gulesci, and Andreas Madestam. 2021. “Repay- tion and Empowerment in South Asia? Self-help Group Programs." South ment Flexibility and Risk Taking: Experimental Evidence from Credit Con- Asia Gender Innovation Lab, World Bank Group. tracts.” Working Paper. Montgomery Heather, John Weiss. 2011. “Can Commercially-oriented Bernhardt, Arielle, Erica Field, Rohini Pande, and Natalia Rigol. 2019. Microfinance Help Meet the Millennium Development Goals? Evidence “Household Matters: Revisiting the Returns to Capital among Female Mi- from Pakistan.” World Development, 39 (1): 87-109. DOI: 10.1016/j.world- croentrepreneurs.” American Economic Review: Insights, 1 (2): 141-60. dev.2010.09.001. DOI: 10.1257/aeri.20180444 Said Farah, Mahreen Mahmud, and Azam Chaudhry. 2017. “Assessment de Mel, Suresh, David McKenzie, and Christopher Woodruff. 2008. “Re- of loans to female start-ups: a randomized control trial.” International turns to Capital in Microenterprises: Evidence from a Field Experiment.” Growth Center Working Paper F-37300-PAK-1. The Quarterly Journal of Economics, 123 (4): 1329-1372. Said Farah, Mahreen Mahmud, Giovanna d’Adda, and Azam Chaudhry. 2009. “Are Women More Credit Constrained? Experimental Evidence 2022. “Home-Based Enterprises: Experimental Evidence on Female Pref- on Gender and Microenterprise Returns.” American Economic Journal: Ap- erences from Pakistan.” Economic Development and Cultural Change, 71 plied Economics, 1 (3): 1-32. DOI: 10.1257/app.1.3.1. (1). DOI: 10.1086/714739. 2014. “Business training and female enterprise start-up, growth, and Valdivia, Martin. 2015. “Business training plus for female entrepreneurship? dynamics: Experimental evidence from Sri Lanka.” Journal of Development Short and medium-term experimental evidence from Peru.” Journal of De- Economics, 106: 199-210. DOI: 10.1016/j.jdeveco.2013.09.005. velopment Economics, 113: 33-51. DOI: 10.1016/j.jdeveco.2014.10.005. Feigenberg, Benjamin, Erica Field, and Rohini Pande. 2013. “The Eco- Zahra, Najaf, Amna Javed, and Ana Maria Munoz Boudet. 2022. "What nomic Returns to Social Interaction: Experimental Evidence from Mi- do We Know About Interventions to Increase Women's Economic Partic- crofinance.” Review of Economic Studies, 80: 1459-1483. DOI: 10.1093/ ipation and Empowerment in South Asia? Asset Transfers and Property restud/rdt016. Laws." South Asia Gender Innovation Lab, World Bank Group. STAY CONNECTED We gratefully acknowledge funding from the South Asia Trade Facilitation Program (SARTFP) and the Umbrella Facility for Gender Equality (UFGE). SARTFP is a trust fund administered by the World Bank with financial contribution from the Government of Australia’s Department of Foreign Affairs and Trade. UFGE is a multi-donor trust fund administered by the World SARGENDERLAB@WORLDBANK.ORG Bank to advance gender equality and women's empowerment through experimentation and WORLDBANK.ORG/SARGENDERLAB knowledge creation to help governments and the private sector focus policy and programs on scalable solutions with sustainable outcomes. The UFGE is supported with generous contributions from Australia, Canada, Denmark, Finland, Germany, Iceland, Ireland, Latvia, the Netherlands, Norway, Spain, Sweden, Switzerland, United Kingdom, United States, the Bill and Melinda Gates Foundation, and the Wellspring Philanthropic Fund. APPENDIX Table A.1: Included Studies and Program Details ID Title and author(s) Intervention Region Sample Method and Program description design 1 Money or Business Pakistan: Rural microfinance clients who are Randomized Clients of a microfinance institution management? A training and Bahawalpur, part of a community organization. Control Trial, were randomly assigned to receive field experiment loan lottery. Hyderabad, While women are primarily engaged Ordinary Least an eight-day training course focused on constraints to Attock in small home-based manufacturing, Squares on business planning, marketing, and In collaboration entrepreneurship in men are involved primarily in the financial management. Clients from both with the rural Pakistan agribusiness sector. Average sales training and non-training groups were Pakistan Poverty for male businesses are 13,001 also randomly assigned to a loan lottery if Giné, Mansuri (2021) Alleviation Fund, Pakistani rupees ($217) but only they had repaid one loan on time and had the National 4,104 Pakistani rupees ($68) among no overdue loans. Average approved loan Rural Support businesses run by female members. size was 45,096 Pakistani rupees ($764) Program, and Women are 37 years on average, 94 among lottery winners. the World Bank. percent are married, and 60 percent have a business at baseline. 2 Access to microfinance Microfinance Bangladesh Average age of 29 years, average Regression Estimate impact of access to microfinance and female labor force monthly income of 12,328 Bangla- Discontinuity using the branch locations of major micro- participation deshi takas ($159), 5 years of school- Design finance institutions in the country. Treated ing on average, 88 percent Muslim. households are within 4-5 kilometers of a Asadullah, Shreya, 41 percent in paid employment, 10 branch. Control households are outside of Wahhaj (2021) percent self-employed. the 4-5 kilometer radius. 3 Repayment flexibil- Credit with flex- Bangladesh The Dabi loan sample is restricted ANCOVA Dabi loans are offered to microenterpris- ity and risk taking: ibility for two to female BRAC members with es, typically with no employees except Experimental evidence repayments. good credit histories. The average for family workers. Average Dabi loan from credit contracts respondent is 38-39 years old, size is $275 with a 22 percent annual In collaboration has 4.5 years of schooling, has a interest rate, a 12-month repayment Marianna Battaglia, with BRAC. household labor income of $7,000 cycle, and monthly repayments of equal Selim Gulesci, Andreas per year, and 45 percent own a installments. Madestam (2021) business at baseline. The flexible contract allows borrowers to delay up to two monthly repayments at any point during the loan cycle using repayment vouchers. 4 Home-bias among Microloan and Pakistan: Sample includes 630 respondents. Randomized Microenterprise loans are provided to female entrepreneurs: training. Bahawalpur, Women are 37 years of age; 90 control trial, women wanting to set up a new business. experimental evidence Gujrat, Sialkot percent are married, and 50 percent ANCOVA Treatment loans range from 10,000 Pakistani In collaboration on preferences from are literate. Average household rupees to 40,000 Pakistani rupees ($67-267), with Kashf Pakistan expenditure of 14,000 Pakistani to be repaid over a year, with repayments Foundation. rupees (low income). 20 percent of starting one month after disbursement. Said, Mahmud, d’Adda, respondents have a business and 20 Loans were complemented by three-hour Chaudhury (2022) percent had a business in the past. training sessions on the importance of marketing, networking, and capacity building. Clients must provide business plans and use the loans as intended. 5 Credit lines in Credit line with India: Sample consists of 360 female street Randomized The treatment is the introduction of a credit microcredit: short-term flexible loan Maharashtra vendors (entrepreneurs), with small control trial line which allows borrowers to withdraw evidence from a amount and to mid-sized business and daily sales or repay at any time a flexible amount, up randomized control repayments. around 3,000 Indian rupees (or to a preapproved limit. Required minimum trial in India $50) and daily profits of 600 Indian payment is the 24 percent interest accrued In collaboration rupees. 1/3 sold perishables (fruits, on the debt balance. Joint liability, weekly AragÓn, Karaivanov, with Mann vegetables). 90 percent are married, meetings, lending to female borrowers. Krishnaswamy (2020) Deshi Mahila in a household with average 4.7 Drawing limits set to 10,000 Indian rupees Bank. members and 40 percent are or 20,000 Indian rupees initially, based on literate. Businesses are 10 years old characteristics of borrowers. Compare treat- on average. ed clients to control group which is offered a standard term loan with weekly repayments. Bank employee collects payments on fixed collection day. Loan period set to three years for credit line clients, and 1, 1.5, or 2 years for term loan clients. Two hours of training provided on basic financial education and prevention of over-indebtedness. 10  |  Systematic Review on Women’s Economic Empowerment Table A.1: Included Studies and Program Details ID Title and author(s) Intervention Region Sample Method and Program description design 6 Can microfinance un- Microfinance India: The average household in the Randomized 52 of 104 poor neighborhoods were lock a poverty trap for Hyderabad sample has 6.8 members and Control Trial randomly selected for the opening of In collaboration some entrepreneurs? 31 percent have a business. The a Spandana branch. Groups of 6 to 10 with Spandana. average business generates 25,240 women are formed and women are jointly Banerjee, Breza, Duflo, Indian rupees in sales and 33.5 responsible for the loans in their group. Kinnan (2019) percent have more than one worker. Loans start at 10,000 Indian rupees ($200 at market exchange rate), and second loans are between 10,000 Indian rupees to 12,000 Indian rupees . The interest rate is 12 percent and reimbursements are made over 50 weeks. Clients must be female, between 18 to 59, have resided in the area for a year at least, have valid identification, and a minimum of 80 percent of women in a group must own their home. 7 Implicit and explicit Financial Pakistan: In first group, restrict to female ANCOVA Individual financial product. The experiment commitment in credit commitment Bhakkar, clients of NRSP (past and current) varied the number of weeks, size of and savings contracts: products. Chakwal, whose household owns at least one installments, the week of payment received. a field experiment Jhelum, business. 60 percent of this sample For first group of clients, set contract to 6 In partnership Rawalpindi, is self-employed. weeks, lumpsum can be 5,000/4,500/5,500 Afzal, D’Adda, with the Khushab, Pakistani rupees with installment amount Fafchamps, Quinn, Said National The second group includes clients Mandi of 1,000 Pakistani rupees. For second group (2019) Rural Support whose households may or may not Bahauddin of clients, set to 8-week contract, lumpsum Programme own a business. can be 3,500/3,200/3,800 Pakistani rupees (NRSP). and installment amount is set to 800 Pakistani rupees. Four other treatments: 1. reduce flexibility (cancellation fee of 500 Pakistani rupees for defaulting on contract), 2. increase flexibility (participant can choose to delay one installment by a week only), 3. remind respondents to pay one day before a payment is due, text message sent directly to participant, 4. reminder sent to a family member of participant. 8 Household matters: India sample India: Kolkata. India: 474 households, participants Ordinary Least India: individual liability loans ranging revisiting the returns to provided with are 18 to 55 years old, reside Squares from 4,000 Indian rupees to 10,000 Indian Sri Lanka: capital among female loans. in household with at least one rupees ($90 to $225). A random subset Colombo, microentrepreneurs entrepreneur. of participants made repayments every Sri Lanka sample Greater Kandy two weeks, beginning two weeks after Bernhardt, Field, provided with Sri Lanka: 202 firms run by males, the loan disbursal. The second group was Pande, Rigol (2019) cash and capital 187 by females. Participants owned given a two-month grace period before grants. USD1,000 or less in business capital, the first repayment. had no paid employees and were between 20 to 65 years of age. Sri Lanka: a random prize drawing which consists of one of four grants: 10,000 Sri Lankan rupees ($100) in materials for business, 20,000 Sri Lankan rupees in materials, 10,000 Sri Lankan rupees in cash, or 20,000 Sri Lankan rupees in cash. Cash can be used for any purpose. 9 Assessment of loans to Loan plus Pakistan: 899 clients in urban and peri-urban Randomized Clients of Kashf Foundation were provided female run start-ups training. Bahawalpur, areas, 440 in treatment group. Control Trial, loans between 10,000-30,000 Pakistani Gujrat, On average, female participants Ordinary Least rupees for 12 months, allowing monthly Said, Mahmud, In collaboration Sialkot, were 37 years old with 2.5 years Squares installments. Service charges are 22% Chaudhry (2017) with Kashf Foun- Multan of education and 2.8 children. 18 per annum of the loan balance. Female dation (KF) and percent are self-employed. beneficiaries receive two-hour training Pakistan Poverty on marketing, networking and capacity Alleviation Fund building. (PPAF). march 2023  |  11 Table A.1: Included Studies and Program Details ID Title and author(s) Intervention Region Sample Method and Program description design 10 Does women’s banking Access to micro- India: Women with savings accounts The study exploits the expansion of matter for women? finance through Ahmedabad in SEWA Bank, who had taken doorstep banking executed by SEWA Evidence from urban doorstep a loan from SEWA and lived in Bank from 1999 onwards, which involved India banking. a region where SEWA Bank had an increase in loan officers who are expanded. Sample comprises 3,676 literate, know basic math, have been with Field, Martinez, Pande In partnership respondents. SEWA for at least 5 years, and receive a (2016) with SEWA. commission for enrolling new clients. 11 Friendship at Credit. India: The sample size is 6,863 households. Randomized The first loan is 10,000 Indian rupees work: can peer Hyderabad Approximately 32 percent of Control Trial, ($200), at an interest rate of 12 percent. effects catalyze female households have an existing Ordinary Least Women are jointly responsible for loans entrepreneurship? In collaboration business Squares of their group (groups of 6 to 10 women). with the Individual loans provided to females aged Field, Jayachandran, Centre for Micro between 18 to 59, who have resided in Pande, Rigol (2016) Finance and the same area for at least one year and Spandana. have valid identification. Loan usage is not restricted. 12 The miracle of mi- Microfinance India: The average household in the Randomized 52 of 104 poor neighborhoods were crofinance? Evidence Hyderabad sample has 5 members, 0.32 Control Trial randomly selected for the opening of In collaboration from a randomized self-employment activities, and a Spandana branch. Groups of 6 to 10 with Spandana evaluation 0.145 self-employment activities women are formed and women are jointly and the managed by women. The average responsible for the loans in their group. Banerjee, Duflo, Centre for Micro business generates 15,991 Indian Loans start at 10,000 Indian rupees ($200 Glennerster, Kinnan Finance (CMF). rupees in revenue and has 0.169 at market exchange rate), and second (2015) employees. loans are between 10,000 Indian rupees to 12,000 Indian rupees. The interest rate is 12 percent and reimbursements are made over 50 weeks. Clients must be female, between 18 to 59, have resided in the area for a year at least, have valid identification, and a minimum of 80 percent of women in a group must own their home. 13 Do group dynamics Meeting and India: Kolkata Sample of 739 clients. Women are Randomized The intervention randomized 148 influence social repayment 35 years old on average, and house- Control Trial five-member client groups into two repay- capital gains among frequency for holds have some income generating ment schedules, one with weekly group microfinance clients? credit. activity in the form of a business. meetings and repayments and the other Evidence from a 29 percent are first time borrowers, with monthly meetings and repayments. In partnership randomized experi- 27.7 percent had previously taken Loans range from 4,000 Indian rupees with ment in urban India out one VFS loan, 43.3 percent had to 12,000 Indian rupees ($66 to $197), Village Financial taken out two VFS loans. Average interest rates are 22 percent, and loan Feigenberg, Field, Services Private, have completed primary education, tenures are 10 months. Pande, Rigol, Sarkar Ltd. (VFS). are predominantly Hindu, and 90 (2014) percent are married. 26 percent are housewives, others business owners or in casual labor. 14 The economic returns Meeting and India: West 961 clients surveyed at baseline Randomized Clients were offered individual liability to social interactions: repayment Bengal and endline. Respondents tend to Control Trial loans of 4,000 Indian rupees and were experimental evidence frequency for be 33 years old on average. 86.5 assigned to either monthly group from microfinance credit. percent are literate and 72 percent meetings and repayment (control), of households own a business. weekly group meetings and repayment Feigenberg, Field, In partnership (treatment 1), or weekly meetings with Pande (2013) with monthly repayments (treatment 2). Village Financial Services. 12  |  Systematic Review on Women’s Economic Empowerment Table A.1: Included Studies and Program Details ID Title and author(s) Intervention Region Sample Method and Program description design 15 Does the classic Loan repayment India: Kolkata Sample of 845 clients. Low-income Randomized Each client received an individual liability microfinance model timing. women from urban areas. Control Trial, loan of 4,000 Indian rupees to 10,000 discourage entrepre- Ordinary Least Indian rupees ($90 to $225). Clients In collabo- neurship among the Squares formed groups of five. 85 groups assigned ration with poor? Experimental to regular loan repayments beginning Village Financial evidence from India two weeks are disbursal. 84 groups began Services. repayment after two-month grace period. Field, Pande, Papp, Both groups faced similar interest charges. Rigol (2013) 16 Business training and Training and Sri Lanka: Two separate samples considered. Randomized 400 women from each sample were female enterprise training plus a Colombo, The current business owner sample Control Trial offered business training and half of start-up, growth, and grant of $130. Greater Kandy includes 628 women who worked this group were preselected to receive a dynamics: experimen- more than 20 hours per week in grant of 15,000 Sri Lankan rupees ($129) tal evidence from Sri self-employment and had monthly conditional on completing training. Funds Lanka profits of Rs. 5,000 or less. Women could be used for any purpose. in this group were 36 years, married, De Mel, McKenzie, The ILO Start and Improve Your Business had run their business for 6 years, Woodruff (2012) (SIYB) training program was used; the and had 10 years of schooling, on potential and current owners were average. offered 9 and 7 days of training with 7 The potential business owner hours of instruction every day, respective- sample includes 628 women who ly. Training cost between $126 to $131 were out of the labor force but per current business owner and $133 to were likely to start working within $140 per potential owner (this includes the next year. Women who had transportation and food allowance). identified the type of business they planned to start and had child care were included. 82 percent of women in this sample had worked before and were 34 years old on average. 17 Can commercially-ori- Microfinance. Pakistan: 11 Sample of 2,881 households, 25 Difference in Compare households who are members ented microfinance districts. percent in urban areas. 40 percent Difference in old villages and have access to loans In collaboration help meet the of the sample is female, 64 percent already, with households who are with Khushhali millennium develop- of females were literate, 70 percent members in new villages and are eligible Bank. ment goals? Evidence of households live below the for loans but do not yet have access. from Pakistan poverty line. Loan terms are between 3 and 12 months Montgomery, Weiss at 20 percent interest, loan size varies (2011) between 3,000 Pakistani rupees to 30,000 Pakistani rupees, to be used for business or agriculture but not for consumption. Loans are made individually but if an individual defaults the whole group (3 to 25 members) is ineligible for loans. 18 Are women more Cash grant and Sri Lanka: 405 firms included in sample, 197 Randomized Conducted a random prize drawing which credit constrained? capital stock Kalutara, are run by males, 190 by females, 18 Control Trial consists of one of four grants: 10,000 Sri Experimental grant. Galle, Matara are run jointly. Most firms are in the Lankan rupees ($100) in materials for evidence on gender retail sector (small grocery stores) or business, 20,000 Sri Lankan rupees in and microenterprise manufacturing sector (sewing, lace materials, 10,000 Sri Lankan rupees in returns or bamboo products etc.) cash, or 20,000 Sri Lankan rupees in cash. Cash can be used for any purpose. de Mel, McKenzie, Females are about 42 years of Woodruff (2009) age, with approximately 9 years of schooling. 19 Returns to capital in Cash grant and Sri Lanka: 405 firms included in sample, 197 Randomized Conducted a random prize drawing which microenterprises: capital stock Kalutara, are run by males, 190 by females, 18 Control Trial consists of one of four grants: 10,000 evidence from a field grant. Galle, Matara are run jointly. Most firms are in the Sri Lankan rupees ($100) in materials experiment de Mel, retail sector (small grocery stores) or for business, 20,000 Sri Lankan rupees McKenzie, Woodruff manufacturing sector (sewing, lace in materials, 10,000 Sri Lankan rupee in (2008) or bamboo products etc.) cash, or 20,000 Sri Lankan rupees in cash. Cash can be used for any purpose. march 2023  |  13