Ideas to Action:

Independent research for global prosperity


David Roodman's Microfinance Open Book Blog


A revised chapter 6 (.docx and .pdf) is up.

For me, the most interesting thing to come out of my editing was a greater appreciation of the implications of Dean Karlan and Jonathan Zinman's study of "cash lending" in South Africa. In that study, they pioneered the methodology they would later use in their 2009 study of the impacts of microcredit in Manila, in which some applicants for credit are randomly "unrejected."

Unlike the Manila study and the other randomized microcredit impact study of 2009, the South Africa study found that credit reduced poverty. Here are the two things I came to understand, or understand better:

  1. The effective interest rate on the loans in question was an astonishing 586%/year, compounded. The loans were repaid over four months, with monthly interest equal to 11.75% of the opening balance. Since the average balance was about half the opening balance, the monthly rate was more like 2 × 11.75% = 23.5%. Multiply that by 12 to get 282%/year.

    ...or, more precisely, the loan balance was 100% of the original amount for the first month, 75% for the second, 50% for the third, and 25% for the fourth. The average of those four numbers is 62.5%, so the average balance over the four months was 62.5% of the opening balance, not 50%, and the effective rate was 11.75%/62.5% = 18.8%/month or 18.8% × 12 = 225.6%/year.

    If you do the math properly (use Excel's IRR() function on a disbursement of +1000 and four payments of --367.5), the rate with compounding is 586%. (On the interpretation of such monstrous numbers see Does Compartamos Charge 195% Interest? and Reflections on Transparency.

    What is remarkable is that people came out ahead on average despite paying such a high rate.

  2. The key benefit of the loans lay in helping a subgroup of people get or retain jobs, not start microbusinesses. As I write in the text, citing an e-mail from Dean Karlan, "They might use the loans to buy required uniforms, or sample kits for sales work, or fix or buy a vehicle to get to work."

Something I already appreciated, which is worth keeping in mind as you cast your mind over this study, is that subjects had a daily per-capita income of $6.50, using a purchasing power parity exchange rate of 3.87 rand/dollar. So they were better off than the people usually imagined as targets for microcredit.

Related Topics:


CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.