New Study of Credit Impacts in Mexico

On CGAP's main blog, Jake Kendall has summarized a new World Bank study of the impacts of access to credit on poor people in Mexico:

The analysis by World Bank economists Miriam Bruhn and Inessa Love, examined the economic impacts of Banco Azteca, launched seven years ago by Grupo Elektra, an electronics and household goods retailer.Grupo Elektra obtained a banking license and opened Banzo Azteca branches in its 815 stores across Mexico in October 2002, combining its physical presence with a database of four million current and former customers and 3,000 motorcycle loan agents to reach a low-income segment of the population previously considered "unbankable". Banco Azteca also took over the issuance and management of installment loans, averaging $250, which had previously been issued by Grupo Elektra’s finance arm.

I would emphasize the last point in Jake's post: can we trust the implications about causality in this non-randomized study? Imagine if a retailer in the United States had stores only in California and New Jersey, and in, say, 2007, started banking through all of these stores. If we learned that people near the stores, or clients of them, experienced different changes in employment and income over the next two years than people in Wyoming, would we confidently conclude that the retailer's new venture in banking was the cause? Big economic changes occurred then that affected different parts of the country differently. (Or if you are not American, devise your own Nationally Appropriate Metaphor for Econometrics (NAME).)The concern, in other words, is that Grupo Elektra placed its stores in a non-random way across Mexico and that areas where it operated experienced systematically different economic changes from those where it did not, quite aside from the expansion of credit. In the data, that would obscure (or masquerade for) the impacts of credit. No doubt the authors of the study appreciate this concern as well as anyone, but chose to "thoughtfully make the most of their unrandomized situation," as I wrote about another new World Bank study.