Larry Reed of the Microcredit Summit Campaign has a letter to the editor in the Washington Post about my piece last weekend.
He makes the valid point that in writing, "Microcredit rarely transforms lives," I shouldn't be as certain as I seem since there are many variants of microfinance for which I don't have impact evidence that I trust.
The impact of financial services depends a lot on how those services are provided and the other types of support the clients receive. We should study and learn from those microfinance providers who can show that their clients are moving out of poverty.
But as a practical matter, one must judge based on the evidence one has (even as one supports the gathering of more good evidence). And that is what I have done. It is certainly possible that variants of microfinance not yet tested produce much better results. But the assertion that the manner of delivery matters a lot is also not grounded in evidence.
Meanwhile, we are getting broadly consistent results from more contexts and more kinds of microfinance: individual loans in urban Philippines, group loans in urban India, group and individual loans in rural Morocco and Mongolia.
I think Larry would reply that the microfinance institutions so far tested--Spandana, Al Amana, XacBank, First Macro Bank---are hardly representative of the 3,600 counted by the Microcredit Summit Campaign. But I suspect that they represent a broad swath of the microfinance delivered. In microfinance, there is a 20/80 rule...actually more like a 3/80 rule: some 3% of the institutions provide 80% of the loans. Those 3% are businesses or business-like institutions, like the ones tested, that have for the most part grown large by focusing purely on financial services, and delivering them in ways that cover costs. Key exceptions are BRAC and Pro Mujer and perhaps some Freedom from Hunger affiliates such as CRECER, which bundle financial and non-financial services for thousands or millions of people. Those are ones I'd love to put to the randomized test if a way can be found. Maybe Dean Karlan and IPA are doing it. Their test of health education for microcredit clients in Peru found no health benefits.
If the strongest response to my article is that there are other forms of microfinance not yet tested that may work better, that supports my contention that the evidence base is weak.
I think the stronger argument against relying on the existing body of evidence is that it doesn't look at effects beyond 18 months or so. Several of the randomized studies have found that microcredit raises investment. One can imagine that the full benefits of that investment would take longer to manifest.