Semiparametric approaches to welfare evaluations in binary response models

Applied welfare analysis is commonly based on mean or median benefit estimates. However, the whole conditional distribution of benefits is often of interest for policy makers. This article compares the distributional information recovered by competing semiparametric methods. Results obtained using a valuation study for the improvement of water resources in Brazil suggest that the semiparametric approaches are in line with each other, capturing a rich heterogeneity structure that is ignored by the logit approach. Failure to take this structure into account may lead to the undue rejection of the project according to the Kaldor-Hicks criterion.