Regression models we have seen so far involve an equation describing a relationship between endogenous (dependent) and exogenous (independent) variables with coefficients or weights that remain constant. A natural extension of the model involves cases where the structural form of relationship between endogenous and exogenous variables remains the same, but coefficients and error term variance change depending upon the state of some variable. The variable that determines the regime could be latent, i.e., not observed explicitly, or an observable variable. Dynamic regime switching models compute the probability of a particular regime manifesting itself when an observation is recorded. For simplicity, it is customary to use the Markov model for the regime switching process. This assumption implies that the probability of occurrence of a regime is dependent only upon the prior period’s regime and other observable variables. For this reason, models analyzed in this chapter will belong to the class of Markov dynamic regime switching models.