The endogenous grid method without analytical inverse marginal utility

The computational time required to solve and estimate dynamic economic models is one of the main constraints in empirical research. The Endogenous Grid Method (EGM) proposed by Carroll (2006) is known to offer impressive speed gains over more traditional stochastic dynamic programming methods, such as Value Function Iterations (VFI). However, existing EGM implementations implicitly require an analytical expression for the inverse marginal utility, which is not known in many interesting cases. We propose a simple and fast approach, which we refer to as the interpolated EGM (iEGM), that can be applied even when the inverse marginal utility is not known analytically. We show through two applications that the iEGM inherits the speed and accuracy of the EGM and that our approach is an order of magnitude faster than traditional approaches.