The Myth of Microfinance? Why MFIs Shouldn’t Be Freaking Out (Yet)

This is a joint post with Michael Clemens.

The headline in the Boston Globe on September 20, 2009 was catchy: "Billions of dollars and a Nobel Prize later, it looks like 'microlending' doesn’t actually do much to fight poverty." The article referred to the findings of two recent impact evaluations in microfinance in India and the Philippines conducted by researchers at MIT and Yale, respectively. The studies, which were randomized controlled trials (RCTs) of microfinance interventions, found “weak and in some cases nonexistent effects” of microfinance on profits, expenditures and well-being. Privately and publicly, donors, MFIs and practitioners are expressing concern about the impact of these studies on the future of microfinance. Are they right to be worried?

Not yet. Let’s look at why we can’t (or shouldn’t) reject credit and savings programs for the poor based solely upon these studies.

1. India and the Philippines are not Mali and Guatemala. It’s a simple concept, right? We know that India is quite distinct from the Philippines, which in turn is distinct from Africa and Latin America. The economists know this too. While RCTs have a high degree of internal validity – meaning that we can “trust” the findings – they have low degrees of external validity. In other words, the findings are valid for that particular country, area, program and context, but cannot be easily extrapolated. So unless your MFI is operating in the exact same way, using the same methodology and targeting the same client group as Spandana and First Macro Bank, we can’t use these studies (alone) to predict results in other countries. Does this mean we have to do a separate microfinance impact evaluation in every country? Certainly not. But it does mean that we need to look at a body of evidence across countries for similar products before we can start drawing conclusions. Our colleague, David Roodman, has discussed these studies and their strengths (and weaknesses) on his blog.
2. Not all microfinance products are created equal. This seems self-evident, but it’s also often forgotten. One of us (Jenny Aker), is working on impact evaluations of cowpea and rice production in West Africa. Both of the projects facilitate farmers’ access to seeds, training and inputs. If we find that the projects don’t improve cowpea and rice yields, would we therefore conclude that all agriculture projects don’t work? Hopefully not. It’s the same with microfinance. Providing credit could technically be a “microfinance” program. But the devil is in the details. How much is the loan size? Does it target men or women? Are they individual or group liability loans? What can loans be used for? What are the interest rates? How long are the repayment periods? You get the idea. The two programs being evaluated in India and the Philippines are pretty different MF products. So while these studies provide some important insights into the impact of these specific products, they aren’t (and shouldn’t) be a general condemnation of all microfinance (yet).
3. The potential benefits of credit take time. The effects of any credit program are dynamic in nature and quite diverse. Let’s skip the economic models and imagine a straightforward scenario: A MF client takes the loan and invests it in a microenterprise. The business does well, the client gets higher profits, and he/she spends part of that money repaying the loan and the other part on household expenses. Seems simple, right? But this pathway hides a lot of assumptions, and we can see the myriad of places where things might not go as planned. A client might, for example, use the loan to better manage her cash flow, whereas another might invest in durables. Even if we assume that everything follows our simplistic model, when would we expect to see benefits for the individual and the household? 6 months? 12 months? 2 years? Both of these studies rely upon a period of about 18 months – and in the case of Spandana, a single, 50-week loan. This time frame becomes increasingly complicated when we consider that many microfinance products start out with small loans, which might not allow clients to make huge changes to their businesses during the first cycle. So what happens to these clients 1-2 years from now?
4. Size counts – but it’s relative. The Spandana evaluation found that microfinance did not have significant effects on health, education and women’s decision-making, but it did increase durable expenditures and profits for those households who already had a business. This might seem disappointing, but the results are theoretical unless we put it into context. In addition to asking, “Is there a positive impact?”, our evaluations should ask, “How do these benefits compare to the costs? And how does that cost-benefit ratio compare with other potential interventions for the poor?” If, for example, Spandana doesn’t provide a very large loan subsidy, then maybe (small) positive effects are sufficient to justify the costs. And if we compare the cost-benefit ratio for microfinance with other development interventions (agriculture, health and education), then maybe microfinance doesn’t look so bad (Of course, it could look worse). The point is, we need to compare the magnitude of the treatment effect with its costs, and to the costs and benefits of other poverty-reducing interventions.
5. We’re on the same side. While we should applaud more recent rigorous impact evaluations of microfinance programs, it’s also important to recognize that some MFI practitioners were asking tough questions about microfinance long before these studies even started. In fact, some NGOs have shifted the emphasis away from pure microcredit towards microsavings, partly in response to these concerns. (Of course, whether microsavings becomes the “transformational panacea that lifts people out of poverty” is a separate issue, although there is a recent evaluation of microsavings in Kenya). Since both MFI practitioners and academics have raised questions about the “miracle” of microfinance, there is a unique opportunity to work together, review our current knowledge of the industry and discuss what this means for future interventions. But this also implies that any future microfinance or microsavings interventions should have built-in, iterative and custom-made evaluation mechanisms to guide the program as it evolves -- using RCTs where feasible, and other rigorous evaluation methods when not (as Chris Blattman, Dani Rodrik and many others have argued).