This paper discusses the problems that are encountered when dynamic, discrete choice models are specified with continuous, serially correlated state variables. A variety of approximation methods that can deal with these problems is examined, and an empirical example that allows continuous variables to be serially correlated is presented.
