The use of heuristic optimization algorithms to facilitate maximum simulated likelihood estimation of random parameter logit models

Abstract

Applications of random-parameter logit models can be found in various disciplines. These models have non-concave simulated likelihood functions and the choice of starting values is therefore crucial to avoid convergence at an inferior optimum. Little guidance exists, however, on how to obtain good starting values. We apply an estimation strategy which makes joint use of heuristic global search routines and gradient-based algorithms. The central idea is to use heuristic routines to locate a starting point which is likely to be close to the global maximum, and then to use gradient-based algorithms to refine this point further. Using four empirical data sets, as well as simulated data, we find that the strategy proposed locates higher maxima than more conventional estimation strategies.