After nearly 30 years of the microcredit movement, we've finally started seeing rigorous impact evaluations in the last few years. Randomized control trials of some variant of microcredit have been conducted in India, Morocco, Mongolia and the Philippines. Each of these trials adds to the evidence, but each is in a specific context, with differences in contracts, eligibility, loan size and structure, and most importantly among the borrowers. That’s why it’s still exciting to see new trials which provide evidence in a different context.
“Microfinance at the Margin: Experimental Evidence from Bosnia and Herzegovina,”a new working paper presented at the 2012 Innovations for Poverty Action (IPA) Conference, gives us yet another different context to examine how households use microcredit and its impact on their lives. The authors of the study – Britta Augsburg, Ralph De Haas, Heike Harmgart and Costas Meghir – look at a group of randomly-selected loan applicants who normally would have been rejected during the loan screening process, in many cases because they lacked the necessary collateral to secure a loan (an approach similar to one used by Dean Karlan and Jonathan Zinman in South Africa and the Philippines). By relaxing the collateral requirements and other screening criteria, individuals who had previously been unable to access credit became eligible.
The pool of “marginal applicants” in this study was 60 percent male, had an average age of 37 years and an average income equivalent to USD 11,123. Approximately two thirds of the clients were married and slightly more than half of them were employed at the baseline survey. Compared to a nationally-representative survey, the population of marginal applicants was younger, less educated and more likely to be male and married than the population as a whole. In terms of the lending criteria, the marginal applicants failed to meet an average of 2.6 of the six main lending requirements for loan eligibility. Seventy-seven percent of the applicants did not possess sufficient collateral or did not meet one or more of the other requirements, which included an assessment of the borrower’s character. About one third of the applicants were judged to have a weak business proposal and loan officers were concerned about the repayment capabilities of about a quarter of the applicants.
By instructing loan officers to lend to these higher-risk marginal clients who they normally would have rejected, the authors were able to study the effects of microlending on individuals who previously had limited access to credit. The researchers found that taking a loan led to increased levels of business activity and higher self-employment among borrowers, but that these developments did not necessarily translate into increased profits or higher household income. Although these results do not align with traditional views of the microfinance sector that claim that microcredit leads to increased business profits and higher household income, they are consistent with the findings of other studies, including Karlan and Zinman’s South Africa study.
The authors did, however, find that the loans had a significant impact on household’s spending and consumption patterns. Families with little savings reduced their consumption, while families with a reserve of assets drew down their savings to complement the loan. These results are also consistent with previous evidence that microloans are lumpy and generally too small to support an investment that can change the scale of a business on their own. In many cases, a household must supplement the loan with their own resources in the form of reduced consumption or savings. Keep in mind that this isn’t necessarily a bad thing—the loan does appear to get the household close enough to the necessary sums for investments that had been out of reach. In fact, standard risk management protocol from banks is to loan less than a business needs as this can help control moral hazard for borrowers. As a lender, you want borrowers to have some significant skin the game, and thus incentive to work hard to make an investment pay off.
Another important finding of the study is that the loans led to a decline in school attendance rates and an increase in the number of young adults working. On average, children of loan recipients with a business at baseline worked 20 hours more than children of the same age in the control group. These increased work hours were at the expense of school attendance: teenage children ages 16-19 were 9 percent less likely to attend school after their families received a loan. The authors explain that these findings may be interpreted in two ways – either that families are denying their kids a valuable education by pulling them out of school, or that the returns to education are so low that employing children in a new family business is a better allocation of time and resources. The fact that this effect only appears among teenagers suggests the latter interpretation is more likely. Families aren’t trading off all schooling, just the final few years of high school.
Augsburg, De Haas, Harmgart and Meghir’s study is the first microcredit RCT conducted during an economic downturn: it occurred at the height of the 2008-09 global financial crisis amidst widespread fears of over-indebtedness. By investigating the important questions of impact during a global economic downturn, this working paper is a valuable contribution to our understanding of microcredit.