ABSTRACT
Establishing comprehensive properties of linear, Gaussian time series models is fairly straight-forward. As noted in Chapter 3, however, even slight departures from the linear model, such as a second-order Volterra series (Section 3.2), can be extremely difficult to work with. Quite often, properties of a nonlinear process, such as existence of a stationary solution, must be done on a case-by-case basis. In this chapter, we discuss stochastic difference equations. also referred to as random coefficient autoregression. This approach gives us a general framework within which to establish properties for a variety of nonlinear models.
