Evaluating the distributive effects of a micro-credit intervention

Abstract

Most analyses of randomized controlled trials of development interventions estimate an average treatment effect on the outcome of interest. However, the aggregate impact on welfare also depends on distributional effects. We propose a simple method to evaluate efficiency–equity trade-offs in the utilitarian tradition of Atkinson (1970). This involves an estimation of the average treatment effect on a monotone concave function of the outcome variable, whose curvature captures the degree of inequality aversion in the welfare function. We argue this is preferable to the current practice of examining distributional impacts through sub-group analysis or quantile treatment effects. We illustrate the approach using data from a credit delivery experiment we implemented in West Bengal, India.