Aspects of the Identification and Inference for Distributional Treatment Effects in Partially Identified Models using Panel Data

This thesis develops new econometric methodologies for estimating the causal impacts of policy interventions using real-world observational data when interventions may not be randomly allocated to individuals. Rather than only estimating the average policy effects, the methods focus on estimating the whole distribution of the heterogeneous policy effects across individuals using longitudinal data, including those who gain and those who lose, providing more nuanced information for policy evaluation. The models were used to evaluate the impacts of exercises on individuals’ body weights.