Women, immigrants, and microcredit in Europe: a Bayesian approach

We use structural modeling to address the allocation process of a microcredit provider granting loans to a heterogeneous pool of applicants. Our theoretical model accounts for technology, risk preferences, and information asymmetry. We test the model with a hand-collected database that includes detailed information on the applicants of a microcredit institution funding European micro-enterprises. Non-parametric Bayesian methodology is used to unpack between-group differences in approval probabilities associated with gender and country of origin and identify (demand-side differences), while differences in unexplained approval probabilities would suggest supply-side biases. The empirical analysis shows that applicants coming from outside of the European Union tend to be more productive than EU-born citizens. They also enjoy a higher approval probability, except for applicants from Latin America, which appear to be riskier borrowers. This result suggests that the microcredit provider treats immigrants fairly. By contrast, the higher productivity and the lower risk of female entrepreneurial projects is only partially compensated by easier access to credit.