Structural Estimation of Continuous Choice Models: Evaluating EGM and MPEC

In this paper, I evaluate the performance of two recently proposed approaches to solving and estimating structural models: The Endogenous Grid Method (EGM) and Mathematical Programming with Equilibrium Constraints (MPEC). Monte Carlo simulations confirm that both EGM and MPEC have advantages relative to standard methods. EGM proved particularly robust, fast and straight forward to implement. Approaches trying to avoid solving the model numerically, therefore, seem to be dominated by these approaches.