Approaches to underdevelopment based on misallocation of resources have two premises. First, that there is huge heterogeneity in terms of underlying productivity among potential and actual entrepreneurs. Second, that the mechanisms that guide resource allocation do not necessarily result in the resources going to the most productive entrepreneurs. Using the Townsend Thai data and the Million Baht program studied by Kaboski and Townsend(2012), we show evidence for both these premises. First, using the fact that the Townsend Thai data include a long time series of pre-intervention information, we estimate TFP household by household. We then show that the effect of the Million Baht program, which was a source of additional short-term credit in the village, varies dramatically by pre-program TFP. There is no discernible effect in terms of income or business profits among low pre-program TFP households but the high TFP households show a large increase in profits (more than 1.5 THB increase in profits for 1 THB in loans). This effect doubles when we restrict to high TFP households that had a non-agricultural business before the intervention. On the other hand, program credit is not allocated based on baseline TFP. However market credit partly mitigates the disparity.