Nonlinear Dynamics and Structural Change in the U.S. Hog-Corn Cycle : A Time-Varying STAR Approach

The linearity of the U.S. hog-corn cycle has been questioned by Chavas and Holt (1991). Even so, attempts have not been made to model the potential nonlinear dynamics in the hog-corn cycle by using regime-switching models. One popular alternative is Terasvirta's smooth transition autoregressive (STAR) model, which assumes regime switching is endogenous and potentially smooth. In this article, we examine monthly data for the U.S. hog-corn cycle, 1910-2004. A member of the STAR family, the time-varying STAR, is fitted to the data and its properties examined. We find evidence of nonlinearity, regime-dependent behavior, and time-varying parameter change