Partial Identification of Treatment Effects in Binary Response Models with Applications in Health Economics

The broad objectives of this project are to study the issues of partial identification in the context of models involving binary endogenous treatment variables and binary outcomes, and to investigate the implications for econometric estimation of policy effects in empirical economics. Identified sets for treatment effects for several Australian health economic applications will be estimated and compared with conventional point identified estimates. Performance of alternative bound estimators will be examined and particular attention given to the issue of the weakness of the instruments and the size of the bounds. The new theoretical developments in this literature have significant implications for empirical economics.