A Markov Switching Cookbook

This paper examines maximum likelihood estimation via hill climbing and the expectations maximization (EM) algorithm in the context of Hamilton's Markov switching framework. The techniques are explained in detail and are followed by a discussion of both analytic and computational issues. Both algorithms tend to be computer intensive, but an approximation technique is shown to significantly reduce the computational demands of the EM algorithm relative to a Davidon Fletcher Powell hill-climbing routine.