Dynamics of a market with market participants switching their expectation formation functions: an empirical application to the U.S. hog market

This paper explores the dynamics of the U.S. hog market with three different dynamic models that are distinguished only by their assumptions with regard to market participants' expectations of future prices. The first model assumes that all the producers in the market have rational expectations. The second model assumes that a constant fraction of the producers have static expectations. The third model, the main focus of this paper, assumes producers choose either rational expectations or static expectations every year based on the past performance of the two different types of expectations. Empirical tests such as log-likelihood ratio type specification tests of GMM estimations and one-step ahead forecasts indicate the third model best captures the movements of the price and quantity data in the hog market, even though the value added by the third model over the second is not great. Simulation experiments illustrate that the market will reach a steady state in the framework of the first and the second model if external shocks are zero. In contrast, in the framework of the third model the market never reaches a steady state but generates cyclical movements of economic variables. Artificially increasing the intensity of choice higher than that suggested by empirical tests, makes the dynamics of the market described by the third model become chaotic.