This paper discusses extensions of standard Markov switching models that allow estimated probabilities to reflect parameter breaks at or close to the end of the sample, too close for standard maximum likelihood techniques to produce precise parameter estimates. The basic technique is a...