Credit Program Participation and Decline in Violence: Does Self-Selection Matter?

Studies on the effects of gendered policy design in microcredit on spousal violence generally suffer from self-selection and under-reporting bias. I overcome such shortcomings by constructing a new measure of violence. The new measure indicates whether there was a decline in violence by subtracting current-experience of violence from ever-experience. Such differencing eliminates time-invariant sources of bias. Further, sensitivity of the effect of female membership to self-selection is examined by gauging the size of self-selection based on selection on the observables. Once self-selection is addressed, the initial positive effect of membership on violence reduction disappears.