Microcredit Impact Evaluation: 2 Where Credit Is Due (2015)

bulletin policy bulletin [ february 2015 ] where credit is due Seven randomized evaluations from around the world show that microcredit does not have a transformative impact on poverty, but it can give low-income households more freedom in optimizing the ways they make money, consume, and invest....

bulletin policy bulletin [ february 2015 ] where credit is due Seven randomized evaluations from around the world show that microcredit does not have a transformative impact on poverty, but it can give low-income households more freedom in optimizing the ways they make money, consume, and invest. art phaneuf photography | shutterstock.com Key Results: Demand for many of the microcredit products was modest. In Ethiopia, India, Mexico, and Morocco, when MFIs offered loans to eligible borrowers, take-up ranged from 13 to 31 percent, which was much lower than partner MFIs originally forecasted. Expanded credit access did lead some entrepreneurs to invest more in their businesses. In Bosnia and Herzegovina and Mongolia, access to microcredit increased business ownership. All but one study showed some evidence of expanded business activity, but these investments rarely resulted in profit increases. Microcredit access did not lead to substantial increases in income. Despite some evidence of business expansion, none of the seven studies found a significant impact on average household income for borrowers. Expanded access to credit did afford households more freedom in optimizing how they earned and spent money. Six studies suggest that microcredit played an important role in increasing borrowers’ freedom of choice in the ways they made money, consumed, invested, and managed risk. There is little evidence that microcredit access had substantial effects on women’s empowerment or investment in children’s schooling, but it did not have widespread harmful effects either. Microcredit did not lead to increases in children’s schooling in the six studies in which it was measured, and only one of the four studies that measured women’s empowerment found a positive effect. Across all seven studies, researchers did not find that microcredit had widespread harmful effects, even with individual-liability lending or a high interest rate. www.povertyactionlab.org www.poverty-action.org From its beginnings as a lending experiment in Bangladeshi villages in the 1970s, microcredit—providing small loans to underserved entrepreneurs—expanded rapidly in the 1990s and 2000s, and now serves over 200 million clients worldwide. Traditionally, financial institutions excluded the poor, finding it too costly to make small loans to borrowers without credit histories or collateral. Yet through the expansion of group-liability lending, community-based banks, and new repayment models, microfinance institutions (MFIs) and banks have brought credit and other financial products to the poor on an unprecedented scale. Throughout its history, microcredit has been both celebrated and vilified as a development tool. It was initially embraced by policymakers, donors, and funders as an important financial product to help small-scale entrepreneurs invest more in their businesses, increase profits, earn additional income, and potentially lift themselves out of poverty. Yet as microcredit gained widespread support, some questioned the validity of these claims. Early critics noted that reports of microcredit’s success were often based on anecdotes or simple before-and- after comparisons. Some suggested that expanding credit access could even be harmful. Business expansion is risky, and if entrepreneurs’ investments are not profitable, increased debt could potentially pull them deeper into a poverty trap. Starting in the early 2000s, researchers began to conduct randomized evaluations to contribute rigorous evidence to this debate. Seven randomized evaluations have assessed some of the most pressing and important questions about microcredit: What is the impact of access to microcredit on financial behavior, business activity, and household welfare? Do borrowers’ investments translate into increased income? Does access to microcredit help empower women or increase household investments in education or health? john wollwerth | shutterstock.com www.povertyactionlab.org www.poverty-action.org evaluations This bulletin reviews seven randomized evaluations that can inform policy debates about the impact of microcredit on low-income borrowers. These studies, conducted between 2003 and 2012,1 cover products tested in seven countries spanning four continents and a wide range of contexts and borrower types. Taken together, they are fairly representative of the global microcredit industry. Researchers partnered with eight relatively large MFIs to conduct randomized evaluations of one or more of their products.2 Four were for-profit lenders (in India, Mexico, Mongolia, and the Philippines), three were non-profit (two heatheronhertravels.com | flickr.com in Ethiopia and one in Morocco), and one chose to remain anonymous (in Bosnia and Herzegovina). In all seven studies, In Morocco ➏, Al Amana opened new branches in randomly the MFI extended microcredit to randomly assigned individuals assigned rural areas with low credit access and offered or communities who had not borrowed from it before. Four of group-liability loans to both men and women. 4 Finally, in the the five studies that randomly offered microcredit across an Philippines ➐, First Macro Bank randomly offered individual- entire community (in Ethiopia, India, Mexico, and Morocco) liability loans to applicants with credit scores slightly below the incorporate potential spillover or displacement effects on eligibility line. nearby businesses and measure the impact of microcredit expansion on the community as a whole. Table 1 also outlines the main features of the seven products. All of the lenders except Spandana in India ➌ and XacBank On the next page, Table 1 summarizes the evaluations in Mongolia ➎ explicitly targeted entrepreneurs, but none numbered ➊ through ➐ in the text for this bulletin. In Bosnia restricted or monitored how loans were spent. The loans’ and Herzegovina ➊, an MFI offered individual-liability loans nominal annual interest rates ranged from 12 to 60 percent, with to a randomly assigned group of marginally creditworthy loan the exception of Mexico ➍, which had a rate of about 110 percent. applicants. In Ethiopia ➋, peasant associations called kebeles Every loan’s interest rate (except that in India ➌) was below the were randomly assigned to receive access to group-liability loans median market microloan interest rate for the country. from the Amhara Credit and Savings Institute or the Oromiya Credit and Savings and Share Company. In India ➌, Spandana3 opened new branches in randomly-assigned neighborhoods in Hyderabad and offered group-liability loans to women. In Mexico ➍, researchers partnered with Compartamos Banco to randomly assign some areas in the state of Sonora to receive access to group-liability loans for women. In randomly assigned villages in Mongolia, poor women who expressed interest in a loan received offers for either individual- ➎a or group-liability loans ➎b from XacBank, allowing researchers to test the relative effectiveness of each model. 1 Researchers are currently conducting seven- and eight-year follow-up 3 Spandana Sphoorty Financial Limited. surveys in India and Morocco to measure the longer-run impacts of microcredit. 4 While households took out almost only group-liability loans, individual loans 2 Five of the MFI partners had at least US$190 million in outstanding microloans were also introduced during the study period. as of 2012. www.povertyactionlab.org www.poverty-action.org 3  product and evaluation features table 1 Approximate loan size in PPP USD Annual Offer Time between Evaluation Gender Repayment Repayment Country Researchers Partner MFI Targeted to entrepreneurs? Eligibility Liability model percentage randomization microcredit offer number of borrowers (% of average household frequency performance rate (APR) level and endline survey annual income) Augsburg, De Haas, Yes Marginally creditworthy Monthly payments Bosnia and 46% ever late, ➊ Herzegovina Harmgart, Anonymous 40% female (borrower planned to invest loan applicants Individual $1,815 (9%) over an average of 22% 26% written off Individual 14 months in a new or existing business) with collateral 14 months Meghir Amhara Credit Tarozzi, and Savings Institute 13% female Yes Poor households with a “Regular payments” Peasant ➋ Ethiopia Desai, Oromiya Credit household head (borrower had a business plan) business plan and collateral “Small groups” $500 (118%) over 12 months 12% 5% default* associations 36 months Johnson and Savings and Share Company Banerjee, Women from non-migrating 15–18 months Spandana (first endline) Duflo, No restriction households identified as likely Weekly payments ➌ India Glennerster, Sphoorty Female to entrepreneurs borrowers; at least 80% Groups of 6 to 10 $600 (22%) over 12 months 24% 49% ever late Neighborhoods and 39–42 months Financial Limited Kinnan must be homeowners (second endline) Angelucci, Yes Weekly payments 10% ever late, Neighborhoods 16 months ➍ Mexico Karlan, Compartamos Banco Female (borrower owned or wanted Women Groups of 10 to 50 $450 (6%) over 4 months 110% 1% default or villages (average exposure) Zinman to start a business) Attanasio, Augsburg, No Monthly payments 27% 19 months ➎a Mongolia De Haas, XacBank Female (borrower interested Poor women Individual $470 (29%) over an average of 5% ever late Villages Fitzsimons, in receiving a loan) 8 months Harmgart Attanasio, Augsburg, No Monthly payments ➎b Mongolia De Haas, XacBank Female (borrower interested Poor women Groups of 7 to 15 $700 (43%) over an average of 27% 9% ever late Villages 19 months Fitzsimons, in receiving a loan) 6 months Harmgart Crépon, Households with Weekly, biweekly, Devoto, Yes 6% female a business other or monthly payments ➏ Morocco Duflo, Al Amana household head (borrower owned a business other than non-livestock Groups of 3 to 4 $1,080 (21%) over an average 15% no data Villages 24 months than non-livestock agriculture) Parienté agriculture of 16 months Marginally creditworthy Karlan, First Macro Bank loan applicants with businesses Weekly payments 33% ever late, ➐ Philippines 85% female** Yes Individual $220 (3%) 60% Individual 13 months Zinman (FMB) who are homeowners or over 3 months 7.4% default long-term renters * In Ethiopia, the repayment rate is based on MFI-reported historical general repayment rates. ** In the Philippines, while FMB targeted marginally creditworthy female microcredit applicants, 15 percent of borrowers were male. 4 www.povertyactionlab.org www.poverty-action.org abdul latif jameel poverty action lab innovations for poverty action 5 evaluations Follow the Money: Microcredit’s Theory of Change Figure 1 illustrates one theory of change for microcredit. Increased access to credit may benefit a borrower in many ways. It can directly finance household needs, help borrowers shift away from risky borrowing practices, or provide a cushion from unexpected economic shocks such as job loss or home damage. But for microcredit to have a transformative effect on the poor, it must enable them to sustainably expand their earnings potential. sergio hayashi | shutterstock.com First, MFIs must create a microcredit product that appeals to credit-constrained entrepreneurs. When entrepreneurs take increases in educational attainment, health outcomes, and up loans, they get an additional source of capital to invest in greater life satisfaction. Financial services targeted to women their enterprises through a variety of different channels, from or otherwise marginalized populations could also potentially restructuring or making capital investments in a current improve borrowers’ decision-making power or social standing enterprise, to expanding inventory, hiring staff, or starting a by encouraging community interactions, allowing them new business altogether. These investments may in turn lead more control over the use of borrowed funds and potentially to increased sales and potentially increased profits. improving their earnings power. If entrepreneurs’ returns on their investments are higher than Microcredit’s theory of change has a clear path to improved the cost of the loans, credit may lead to increases in household well-being, but it is necessary to consider how a borrower income. With additional income, borrowers and their families may veer from this path. The studies in this bulletin evaluate can increase spending on food, health care, housing, education, microcredit from input to output, outcome, and ultimate their businesses, leisure, or any number of other goods impact, and provide important insights into how microcredit and services. In turn, these spending increases may lead to programs both succeed and fail to follow this theory of change along each step of the way. figure 1 a theory of change for microcredit inputs outputs outcomes impacts Microcredit Product Business Investment Business Activity Increased Increased Household Physiological Household Saving/Spending On Health Start new business Increased sales Income Improve current business Increased profit Health Intellectual Education Education Other Assets Psychological Life satisfaction Social investment improved standard of living Empowerment slawek jurczyk, fábio testa, richard schumann, juan pablo bravo, improved taras pastushchuk, tina abi hachem, sergey zenzin | nounproject.com well-being 6 www.povertyactionlab.org www.poverty-action.org  results 1. Demand for many of the microcredit In Ethiopia ➋, 31 percent of households offered microcredit had products was modest. outstanding loans at the time of follow-up. In four studies where MFIs offered microloans to a In Bosnia and Herzegovina ➊, Mongolia ➎, and the general population of eligible borrowers, take-up ranged Philippines ➐, microcredit was offered exclusively to those from 13 to 31 percent, which was much lower than partner who had already applied for or expressed direct interest in a MFIs originally forecasted. The first important question is: loan. Thus, it is not possible to determine what microcredit when MFIs offer people access to credit, do they take it up? take-up rates among a more general population would have For microcredit, demand can be an important reality check. been in these contexts. In Bosnia and Herzegovina ➊ and the When microcredit was marketed to potential borrowers in Philippines ➐, extending loans to marginally creditworthy Ethiopia ➋, India ➌, Mexico ➍, and Morocco ➏, relatively loan applicants resulted in 100 percent and 40 percent take-up, few took it up (Figure 2). In rural areas of Morocco ➏ with no respectively. In Mongolia ➎, where XacBank marketed loans to previous access to microcredit, only 13 percent of villagers in women who had previously indicated interest at a community the treatment group decided to create small groups to take a meeting, take-up was 50 percent for individual-liability loans loan. In India ➌ and Mexico ➍, 18 and 19 percent of eligible and 57 percent for group-liability loans. Taken together, these borrowers in each respective treatment group borrowed from results suggest that microcredit may be valued as a useful the partner MFI within 18 months of gaining access to credit. financial tool by some, but not all, borrowers. figure 2 microcredit take-up was modest when mfis offered it to a general population of eligible borrowers 100% Comparison Group Treatment Group Statistically significant difference from comparison group 57.0% 50.0% Take-up rate (%) 39.5% 31.2% 32.4% 33.2% 17.8% 17.3% 13.2% 6.0% 5.1% 5.8% 6.2% 6.2% 0.0% Ethiopia India Mexico Morocco Bosnia and Mongolia Mongolia The Herzegovina (individual) (group) Philippines Representative population of eligible borrowers People in the sample expressed interest in or applied for microcredit Note: Statistical significance is noted at the 90 percent confidence level or higher and error bars represent 90 percent confidence intervals; In Ethiopia, India, Mexico, Mongolia, and Morocco, take-up is measured as having any loans from the partner MFI at the time of the endline survey; In India, the results displayed are from the first endline survey (1.5 years), and there is also a statistically significant difference after 3.5 years; In Bosnia and Herzegovina, comparison group take-up is measured as having any outstanding loan from any MFI and treatment group take-up is a direct measurement of those who took up the partner MFI’s microcredit offer (76.3 percent of borrowers in the treatment group reported having any loans from any MFI at the time of the endline survey); In the Philippines, take-up is measured as having any loan from any financial institution in the month preceding the endline survey. www.povertyactionlab.org www.poverty-action.org 7 results getty images (us), inc. Borrowers reported using microloans for many purposes, Microcredit did not cause borrowers to take on additional including both business development and consumption. MFIs debt from other sources. While some theories suggest that generally target microcredit products towards entrepreneurial microcredit may spur demand for additional credit elsewhere as efforts, but borrowers’ financial priorities may differ. Four entrepreneurs attempt to adequately expand business activities, studies summarized borrowers’ reports on their loan use. four studies find evidence of the opposite. In India ➌, These self-reported data must be taken with a grain of salt, as informal borrowing from families or communities fell in self-reports have been shown to vary significantly depending on neighborhoods where Spandana offered microcredit. Women when, how, and by whom borrowers are surveyed.5 Around 83 who gained access to group loans in Mexico ➍ were no more percent of borrowers in Ethiopia ➋ reported using microcredit likely to borrow from other formal sources, but were 1.1 for business purposes, while 9 percent spent their loans on percentage points more likely to hold an informal loan relative to schooling, ceremonies, or general consumption. In India ➌, 5.1 percent in the comparison group. In Mongolia ➎, researchers 30 percent of borrowers reported using at least part of their found that those offered microcredit were nearly 30 percent loans to start new businesses and 22 percent to buy stock for less likely to hold a loan from other formal credit sources. In existing businesses. Additionally, 30 percent reported using a the Philippines ➐, microcredit offers did not affect loans from portion of their loans to repay existing loans, 15 percent to buy friends, family, or moneylenders. While there is some evidence durable household goods, and 15 percent to smooth household of substitution among different types of credit, researchers did consumption. In Bosnia and Herzegovina ➊, 8.5 percent of not find evidence supporting the claim that expanded access clients reported spending the majority of their microloans to from one MFI led borrowers to take on additional debt from purchase goods. Around two-thirds of Moroccan ➏ microcredit other sources. applicants declared to be planning to use loans for animal husbandry projects, with just over one-quarter intending to invest in trade-related businesses and 6 percent in other nonagricultural businesses (e.g. services and handicrafts). 5 Karlan, Dean, Adam Osman, and Jonathan Zinman. 2013. “Follow the Money: Methods for Identifying Consumption and Investment Responses to a Liquidity Shock.” 8 www.povertyactionlab.org www.poverty-action.org  results 2. Expanded credit access did lead some later. They were also 5 percentage points more likely to hold inventory. In Mongolia ➎, microcredit’s effect on business entrepreneurs to invest more in their businesses. ownership varied by loan type. Individual-liability loans did not increase business ownership. However, women Increased entrepreneurial activity is a vital step in microcredit’s who were offered group-liability loans were 9 percentage theory of transformative change. If microcredit does not points more likely to own a business relative to 39 percent increase business ownership, size, or profits, it is unlikely in the comparison group, and less educated women were 31 that it will deliver increased income by relaxing credit percentage points more likely to own a business. Researchers constraints that inhibit business growth. All studies except hypothesize that joint liability may have dissuaded borrowers the one conducted in the Philippines ➐ showed evidence from using loans for non-investment purposes in this context. of expanded business activity, but these investments rarely resulted in significant increases in profits. The evaluations in Ethiopia ➋, India ➌, Mexico ➍, and Morocco ➎ found no effect on business ownership. In the In Bosnia and Herzegovina and Mongolia, access to Philippines ➐, some borrowers closed their businesses; on microcredit expanded business ownership. In Bosnia and average, treated clients operated 0.1 fewer enterprises. Herzegovina ➊, where half of all comparison group households owned a business, those offered a loan were 6 percentage points more likely to report owning a business 14 months figure 3 microcredit access increased business ownership in two of the seven studies Comparison Group Treatment Group Statistically significant difference from comparison group 99.1% 99.4% 83.2% 81.7% 66.2% 56.5% 58.5% Business owner (%) 50.7% 35.7% 34.9% 24.8% 25.4% 24.3% 23.9% Ethiopia India Mexico Morocco Bosnia and Mongolia The Herzegovina (group) Philippines Note: Statistical significance is noted at the 90 percent confidence level or higher and error bars represent 90 percent confidence intervals; In Ethiopia, ownership is measured for non-farm businesses; In India, displayed results are from the first endline survey (1.5 years), and there is also no statistically significant difference after 3.5 years; In Bosnia and Herzegovina, differences in business ownership are not significant for multiple hypotheses testing; In Mongolia, displayed results are for household businesses. There was also a positive statistically significant difference for respondent businesses. www.povertyactionlab.org www.poverty-action.org 9 results Generally, microcredit access increased business activities, 3. Microcredit access did not lead to substantial but rarely resulted in profit increases. Evaluations in six countries increases in income. provide some evidence of increased business activity, but to varying degrees. In Mexico ➍, expanded credit access increased the size of While microcredit helped some entrepreneurs invest, none existing businesses. In Morocco ➏, for borrowers who previously of the seven studies found that it had a significant impact on farmed and owned livestock, access to microloans increased income for the average borrower. In Morocco ➏, borrowers’ sales from these businesses by around 20 percent. Profits also business sales and profits increased, yet they cut back on wage increased for the larger businesses, but decreased in smaller ones. labor and reduced household asset sales to near-zero levels, Researchers observed similar trends for larger businesses in leaving their total income unchanged. In Mexico ➍, business Ethiopia ➋, India ➌, and Mexico ➍. In India ➌, Mexico ➍, and sales expanded among those offered microcredit, but both Morocco ➏, overall business investments or expenses increased for profit and household income remained at previous levels. borrowers offered microcredit. In Bosnia and Herzegovina ➊, Ethiopia ➋, India ➌, and Mexico ➍, researchers did not find An analysis of consumption (often used as a proxy for any overall effect on borrowers’ profits. Contrary to Morocco ➏, material well-being) reveals similarly mixed impacts. In however, excluding the top 1 percent of business profits from India ➌, total consumption among treatment households analysis in Bosnia and Herzegovina ➊ does reveal evidence of a was no different from that of households in comparison significant profit increase for the remainder of the sample. In neighborhoods. In Mongolia ➎, food and total consumption India ➌, for businesses that existed before their owners gained increased for households in group-liability villages, but not in access to microcredit, researchers found that average profits individual-liability villages. In Morocco ➏, microcredit offers more than doubled in a year and a half. However, these effects had no effect on total household consumption. In Ethiopia ➋, were driven exclusively by businesses that were among the most researchers found evidence that microcredit offers may have profitable prior to gaining access to credit. In the Philippines ➐, actually resulted in increased food insecurity. On average, applicants offered microcredit had 0.3 fewer employees on average. Ethiopian ➋ households in microcredit communities reported Contrary to the other studies, this suggests that expanded access to an additional half-month of food insecurity on top of the 1.3 credit shrunk business size in the Philippines ➐. months reported by households in comparison communities. In another randomized evaluation in South Africa, researchers collaborated with a large for-profit micro-lender The Problem with Profit to randomly relax its scoring criteria for some microcredit applicants (similar to the Bosnia and Herzegovina ➊ and Why was microcredit’s impact on business profit so limited in Philippines ➐ studies). Even though these loans targeted these contexts? There are many potential reasons. Borrowers may choose to reduce their wage labor when they can earn more employed individuals, not entrepreneurs, and charged relatively high interest rates (about 200 percent APR), researchers found from their businesses, as was the case in Morocco ➏. Many that access to this credit product did increase income. In this small-scale entrepreneurs may not be good at growing their case, loans helped employed individuals absorb economic businesses without additional training or support. In Mexico ➍, shocks (e.g. family health emergencies), which allowed them women’s business revenues increased with access to credit, but to keep their jobs. Additionally, there was no evidence of any so did their expenses. As several studies found, some borrowers negative impacts on average.6 chose to use loans for consumption rather than investment. There are numerous other potential explanations that the existing studies do not examine. Often low-income households own businesses that are relatively undifferentiated in a saturated market, such as owning a small shop in an outdoor market. Given that many other entrepreneurs are similarly employed, a large increase in profits for any one entrepreneur is unlikely. Furthermore, some entrepreneurs may be entrepreneurs out of necessity rather than choice, given limited job prospects in their economies, and others may simply not be motivated, or have the skills or time to substantially grow their businesses. 6 Karlan, Dean and Jonathan Zinman. 2010. “Expanding Credit Access: Using Randomized Supply Decisions to Estimate the Impacts.” 10 www.povertyactionlab.org www.poverty-action.org results table 2 summary of microcredit's impact on various outcomes Outcome Bosnia and Herzegovina Ethiopia India Mexico Mongolia Morocco Philippines Business ownership — — — + — — Business revenue — — — — — Business inventory/assets + no data + no data + + — Business investment/costs — — no data Business profit — — — — — + — Household income — — — — — — — Household spending/consumption — - — — — Social well-being — — — — — Note: Green (red) arrows represent statistically significant positive (negative) differences in outcomes between the treatment and comparison groups at the 90 percent confidence level or higher, dashes represent no statistically significant difference; Ethiopia: While none of the individual business outcomes showed a positive impact, a combined business outcomes index did; a decline in household spending/consumption is measured as an increase in food insecurity; India: The increase in assets occurred only after 3.5 years, while the increase in inventories occured only after 1.5 years; Mexico: Household spending is measured as the value of assets purchased in the past two years; social well-being is measured as a combination of women’s empowerment outcomes and trust in people; Mongolia (group): Business assets measured as an index of listed assets increased, while assets measured as monetary stock did not; Morocco: There was an increase in combined business sales and home consumption, an increase in business costs, and no change in investment; The Philippines: There was a decrease in the number of businesses and number of paid employees; household spending/consumption was measured as changes in food costs and quality; a combined social well-being index showed a negative effect. 4. Expanded access to credit did afford households more freedom in optimizing how they earned and spent money. Despite mixed results on income and consumption, evidence from six studies suggests that microcredit can play an important role in expanding the ways in which borrowers make employment decisions, consume, and invest. In Morocco ➏, borrowers invested more in their businesses, increasing both sales and profits, but decided to concurrently cut back on their casual wage labor, potentially due to its less desirable and less stable nature as a source of income. Similarly, in Bosnia and Herzegovina ➊, microcredit access allowed borrowers to increase their self-employment. In Mexico ➍, microcredit helped women avoid selling assets to pay off debts. In India ➌ and Mexico ➍, households with access to microcredit decreased spending on “temptation goods”—such as alcohol, cigarettes, and gambling—to invest more in their businesses. In Mongolia ➎, about half of all microcredit was getty images (us), inc. used for household consumption; with group-liability loans ➎b, households bought more and healthier foods. In the Philippines ➐, microcredit access helped borrowers cope with risk, strengthen community ties, and expand access to informal credit. Collectively, these results suggest that although microcredit may not be transformative in lifting people out of poverty, it can afford people more freedom in their choices (e.g. of occupation, or financing assets) and the possibility of being more self-reliant. www.povertyactionlab.org www.poverty-action.org 11 results Product Design Recent evidence suggests that relatively simple tweaks to microcredit products may change their impact on poverty and financial institutions’ bottom line. MFIs and banks must make many decisions about the terms of their credit products. They must determine the interest rate, whether to offer group- or individual-liability loans, whether a loan should include an initial grace period, and how often borrowers must repay, among many other choices. Several randomized evaluations by IPA and J-PAL affiliates can help inform these product design questions. For example, a study in West Bengal, India found that offering a two-month grace period before borrowers had to begin making payments led to a 20 percent increase in monthly income after three years relative to borrowers who had to begin repayment two weeks after receiving a loan. However, those same grace period clients were more than three times as likely to default in the short run, and reported riskier business practices in the long run. This suggests that the traditional immediate repayment model simultaneously limited default and income growth.7 getty images (us), inc. Immediate repayment may not be the only way to reduce delinquency. In Malawi, introducing a cost-effective fingerprint identification technology substantially increased repayment rates among high risk borrowers.8 In Kenya, ongoing research suggests that using water tanks as collateralized loans may increase take-up rates with no effect on late payments.9 In an evaluation in West Bengal, India, researchers found that switching from weekly to monthly repayment meetings resulted in the same high repayment rates, but dramatically reduced collection costs for the partner MFI, as well as client stress.10 Two evaluations tested the relative merits of group versus individual liability. In Mongolia ➎, both group- and individual-liability loans resulted in similar repayment rates, providing evidence against the hypothesis that repayment rates are high only because of groups’ screening and perceived repayment enforcement. In another evaluation with a bank in the Philippines, researchers randomly selected existing group-lending centers to convert to an individual-liability model. They found no difference in repayment rates between the two models. Additionally, although individual entrepreneurs opted for relatively smaller loans, an overall increase in the number of borrowers allowed MFIs to maintain comparable profit levels.11 Researchers are evaluating more product designs worldwide and their findings can help inform future microcredit offerings. Donors and NGOs may have an important role to play in innovating, piloting, and testing more client-centered products in cases where they may not be immediately commercially viable. 7 Field, Erica, Rohini Pande, John Papp, and Natalia Rigol. 2013. “Does the Classic 10 Field, Erica and Rohini Pande. 2008. “Repayment Frequency and Default in Microfinance Model Discourage Entrepreneurship Among the Poor? Experimental Micro-Finance: Evidence from India.” Evidence from India.” 11 Giné, Xavier and Dean Karlan. 2014. “Group versus individual liability: 8 Giné, Xavier, Jessica Goldberg, and Dean Yang. 2012. “Credit Market Short and long term evidence from Philippine microcredit lending groups.” Consequences of Improved Personal Identification: Field Experimental Evidence from Malawi.” 9 Jack, William, Michael Kremer, Joost de Laat, and Tavneet Suri. 2012. “Expanding Access to Micro Credit: The Role of Asset-Collateralized Loans.” 12 www.povertyactionlab.org www.poverty-action.org results 5. There is little evidence that microcredit access had substantial effects on women’s empowerment or investment in children’s schooling, but it did not have widespread harmful effects either. Researchers did not find that access to microcredit led to substantial increases in women’s empowerment or investment in schooling. In the six studies in which it was measured, microcredit access did not increase children’s schooling. Examining education in India ➌, over 90 percent of boys and girls aged 5 to 15 were in school, regardless of whether their household had been offered microcredit 1.5 or 3.5 years beforehand. Results from Morocco ➏ were similarly insignificant, where around 45 percent of all children aged 6 getty images (us), inc. to 15 were in school, regardless of whether or not their villages received microcredit offers. In Bosnia and Herzegovina ➊, Microcredit access did not have harmful effects on the there was a 9 percentage point decline in school attendance average borrower. Some critics believe that expanding access among 16 to 19 year olds. In households with relatively to credit may crowd out existing businesses, cause households lower education levels, these adolescents also worked 1.1 to overextend themselves and fall into debt traps, or increase hours more per week on average compared with similar stress or depression levels among borrowers. There is little households not offered a loan. In contrast, researchers found evidence to support these claims. In Mexico ➍, microcredit was no change in teenagers’ average workload in India ➌, and offered at a 110 annual percentage rate (APR), meaning that in teenage girls worked about two hours less per week in just four months microloan recipients had to pay 28 percent in treatment neighborhoods. interest for every peso borrowed, as well as repay the principal In three of the four studies that evaluated women’s loan amount. Researchers found no evidence of negative empowerment, microcredit access had no effect. Many MFIs impacts on business revenues, profits, or household decision- offer microcredit exclusively to women, creating a potential making power for the poorest clients offered microcredit. opportunity to increase female empowerment. Four studies Additionally, there were no differences in life satisfaction or examined the role of microcredit and women’s empowerment, job stress between those who did and not receive a microcredit and three found no effect on female decision-making power or offer. In the Philippines ➐, male borrowers reported higher independence. In Mexico ➍, where Compartamos emphasized stress levels, but overall stress levels among borrowers were empowerment as part of its product, women did enjoy a small no different from the comparison group. In Ethiopia ➋, where but significant increase in decision-making power. Microloan average loan sizes equated to 118 percent of annual income, recipients in Mexico ➍ also reported less depression and there were no detrimental effects on businesses, income, youth more trust in others. In the Philippines ➐, households offered labor, or youth education. In Bosnia and Herzegovina ➊ microcredit reported more trust in their neighborhoods, and Mongolia ➎a, where borrowers had individual-liability potentially as a secondary result of increased opportunities for loans, there was no evidence suggesting that microcredit community risk-sharing. trapped or otherwise harmed borrowers. www.povertyactionlab.org abdul latif jameel poverty action lab innovations for poverty action 13 www.poverty-action.org open questions danm12 | shutterstock.com Open Questions How does access to credit affect specific populations? The seven studies featured in this bulletin provide a strong There is growing concern among policymakers, advocates, foundation of evidence on the impact of microcredit, but and funders that certain types of people may do themselves there are still many important questions for researchers, more harm than good by borrowing (e.g. falling into debt policymakers, and financial service providers to consider: traps). While there is currently little evidence to support this, understanding microcredit’s effects on different types of How can simple tweaks to microcredit products improve their borrowers is an important consideration for targeting clients accessibility and effectiveness? and informing consumer protection policy. From improving take-up rates to targeting specific social Should policymakers and banks work to expand microcredit outcomes, increasingly nuanced products may strengthen to more borrowers in communities that already have access to microcredit’s impact in some contexts. Researchers and microcredit services? practitioners have made some progress, but more research is needed to deepen decision-makers’ understanding of effective There is a limited body of evidence on the effects of expanding credit product design. credit access in communities that already have it. A randomized evaluation with Compartamos Banco in Mexico suggests that What are appropriate consumer protection policies to support further increasing access was a financially sustainable decision expanded credit access while enabling people to make informed for the bank and could have ultimately benefited borrowers. decisions about borrowing? When Compartamos reduced interest rates, it increased take- up by new customers without reducing its profits. Yet in Mali, Historically, political decisions have both enabled and disabled researchers found that offering an agricultural loan induced microcredit’s rapid growth. Further exploration of the effects effective self-selection initially. The farmers who benefited of consumer protection policies, the establishment of credit most from an influx of capital were those who chose to borrow bureaus, and general regulatory oversight must be examined at the first opportunity. alongside product design innovations. 14 www.povertyactionlab.org www.poverty-action.org further reading Field, Erica and Rohini Pande. 2008. “Repayment Frequency ➍ Angelucci, Manuela, Dean Karlan, and Jonathan Zinman. 2015. “Microcredit Impacts: Evidence from a Randomized Microcredit and Default in Micro-Finance: Evidence from India.” Journal Program Placement Experiment by Compartamos Banco.” of the European Economic Association 6(2-3): 501-509. American Economic Journal: Applied Economics 7(1): 151-82. Giné, Xavier, Jessica Goldberg, and Dean Yang. 2012. “Credit Market Consequences of Improved Personal Identification: Field Experimental ➎ Attanasio, Orazio, Britta Augsburg, Ralph De Haas, Emla Fitzsimons, and Heike Harmgart. 2015. “The Impacts of Evidence from Malawi.” American Economic Review 102(6): 2923-2954. Microfinance: Evidence from Joint-Liability Lending in Mongolia.” Giné, Xavier and Dean Karlan. 2014. “Group versus individual American Economic Journal: Applied Economics 7(1): 90-122. liability: Short and long term evidence from Philippine microcredit lending groups.” Journal of Development Economics 107: 65-83. ➊ Augsburg, Britta, Ralph De Haas, Heike Harmgart, and Costas Meghir. 2015. “The Impacts of Microcredit: Evidence from Jack, William, Michael Kremer, Joost de Laat, and Tavneet Bosnia and Herzegovina.” American Economic Journal: Suri. 2012. “Expanding Access to Micro Credit: The Role Applied Economics 7(1): 183-203. of Asset-Collateralized Loans.” Working Paper. ➌ Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Karlan, Dean and Jonathan Zinman. 2010. “Expanding Kinnan. 2015. “The Miracle of Microfinance? Evidence Credit Access: Using Randomized Supply Decisions to Estimate from a Randomized Evaluation.” American Economic Journal: the Impacts.” Review of Financial Studies 23(1): 433-464. Applied Economics 7(1): 22-53. ➐ Karlan, Dean and Jonathan Zinman. 2011. “Microcredit Banerjee, Abhijit, Dean Karlan, and Jonathan Zinman. 2015. “Six in Theory and Practice: Using Randomized Credit Randomized Evaluations of Microcredit: Introduction and Further Scoring for Impact Evaluation.” Science 332(6035): 1278-1284. Steps.” American Economic Journal: Applied Economics 7(1): 1-21. Karlan, Dean, Adam Osman, and Jonathan Zinman. 2013. Beaman, Lori, Dean Karlan, Bram Thuysbaert, and Christopher “Follow the Money: Methods for Identifying Consumption and Udry. 2014. “Self-Selection into Credit Markets: Evidence Investment Responses to a Liquidity Shock.” Working paper. from Agriculture in Mali.” NBER Working Paper No. 20387. Reed, Larry, Jesse Marsden, Amanda Ortega, Camille Rivera, ➏ Crépon, Bruno, Florencia Devoto, Esther Duflo, and William and Sabina Rogers. 2014. The State of the Microcredit Summit Parienté. 2015. “Estimating the Impact of Microcredit on Those Campaign Report. Washington, DC: Microcredit Summit Who Take It Up: Evidence from a Randomized Experiment in Campaign. Accessed December 16, 2014. Morocco.” American Economic Journal: Applied Economics 7(1): 123-50. http://stateofthecampaign.org/2014-report-executive-summary/. Field, Erica, Rohini Pande, John Papp, and Natalia Rigol. 2013. ➋ Tarozzi, Alessandro, Jaikishan Desai, and Kristin Johnson. “Does the Classic Microfinance Model Discourage Entrepreneurship 2015. “The Impacts of Microcredit: Evidence from Ethiopia.” Among the Poor? Experimental Evidence from India.” American Economic American Economic Journal: Applied Economics 7(1): 54-89. Review 103(6): 2196-2226. Lead Author: Justin Loiseau | Co-author: Claire Walsh | Design: Amanda Kohn Suggested Citation: J-PAL and IPA Policy Bulletin. 2015. “Where Credit is Due.” Cambridge, MA: Abdul Latif Jameel Poverty Action Lab and Innovations for Poverty Action. We thank implementing partners and funders for supporting this research. By country we acknowledge: Bosnia and Herzegovina: Cowles Foundation, European Bank for Reconstruction and Development, European Research Council, and the Institution for Social and Policy Studies at Yale University; Ethiopia: Amhara Credit and Savings Institute, the David and Lucile Packard Foundation, Family Health International, and Oromiya Credit and Savings and Share Company; India: ICICI Bank, J-PAL, Spandana Sphoorty Financial Limited, and The Vanguard Charitable Endowment Program; Mexico: The Bill & Melinda Gates Foundation, Compartamos Banco, and the National Science Foundation; Mongolia: European Bank for Reconstruction and Development, European Research Council, the Institute for Fiscal Studies, and XacBank; Morocco: Al Amana, Agence Française de Développement, International Growth Centre, and J-PAL; The Philippines: The Bill & Melinda Gates Foundation and the National Science Foundation. www.povertyactionlab.org www.poverty-action.org 15  policy lessons If their main goal is poverty reduction, donors should microcredit has managed to expand financial freedom in all these facilitate, but not finance, standard microcredit lending. ways. The lack of transformative impacts on poverty should not While the seven studies summarized in this bulletin show that obscure these more modest, but potentially important effects. The traditional microcredit can be a useful financial tool, there is rapid growth and unprecedented reach of the global microcredit little evidence to suggest that it leads to substantial improvements industry points to the fact that low-income families around the in income or social well-being. As the microcredit industry has world value credit as a tool that can help them better manage their grown and raised private capital, many international development complex financial lives. donors have reduced or stopped subsidizing microcredit institutions. Given that microcredit has not led to transformative Piloting and testing more innovative credit products may impacts on poverty, donors and private investors interested in lead to greater impacts on poverty. Despite their success in increasing the social impact of credit could use philanthropic numbers, microcredit institutions could innovate more. money to encourage the design, piloting, and testing of more Piloting lending models that more closely match the cash flexible credit products (see third policy lesson). Donors and flow needs of borrowers may prove more transformative. governments should also focus their efforts on creating supportive Traditionally, microcredit has emphasized the importance of regulatory frameworks for financially viable and responsible small loans, short repayment periods, and immediate and product offerings for the under- and unbanked. frequent repayment. These terms may limit the types of investments the loans can finance. Recent evidence shows that Microcredit may not be the best instrument to improve small tweaks to loan features can make a big difference (see business profitability, but it can be an important tool Product Design on page 12). In India, adjusting repayment for increasing people’s freedom of choice in deciding schedules allowed poor businesswomen to start more businesses, their occupations and managing their finances. In Bosnia invest more in them, earn more profit, and increase household and Herzegovina ➊, India ➌, Mexico ➍, Mongolia ➎, and income, though the product would not initially be commercially Morocco ➏, access to a microloan enabled some entrepreneurs viable.12 And in Mali, rearranging microcredit cash flows to match to start, expand, or invest in new assets for their businesses. agricultural seasons increased farmers’ investments in their land.13 Yet those investments did not translate into significantly higher These results point to opportunities for more nuanced investment profits or income. These results do not imply that microcredit by providers and donors in piloting, testing, and scaling new institutions and banks should stop offering credit, or that credit credit models. As these new products may not be immediately is not an important financial service for the poor; they instead commercially viable, the donor and nonprofit community could suggest that conventional microcredit alone should be neither the play an important role in supporting products that could improve primary nor only instrument used to promote entrepreneurship. the impact of credit on the lives of the poor. Finally, more evaluation is needed to better understand how credit products Rather than increasing profitability or household income, the affect different types of borrowers to help inform consumer seven studies suggest that microcredit may be most effective protection regulation and MFIs’ and banks’ targeting strategies. as a mechanism to improve people’s freedom of choice in how they earn and spend money. Financial freedom has many facets, including the ability for individuals to decide their occupation, 12 Field, Erica, Rohini Pande, John Papp, and Natalia Rigol. 2013. “Does the Classic Microfinance Model Discourage Entrepreneurship Among the Poor? Experimental make household spending decisions, finance larger purchases Evidence from India.” for their homes or businesses, refrain from selling off assets in 13 Beaman, Lori, Dean Karlan, Bram Thuysbaert, and Christopher Udry. 2014. times of crisis, and better manage risk. Across various contexts, “Self-Selection into Credit Markets: Evidence from Agriculture in Mali.” ipa & j-pal offices around the world Innovations for Poverty Action (IPA) is an international research organization that discovers and promotes effective solutions to global poverty problems. www.poverty-action.org The Abdul Latif Jameel Poverty Action Lab (J-PAL) is a network of affiliated professors around the world who are united by their use of randomized evaluations to answer questions critical to poverty alleviation. J-PAL’s mission is to reduce poverty by ensuring that policy is informed by scientific evidence. www.povertyactionlab.org IPA Country Programs J-PAL Regional Offices