HETEROGENEOUS EXPECTATIONS AND MARKET DYNAMICS: A NON-LINEAR VERSION

The first task of this paper is to determine whether heterogeneity in expectation formation is detected from actual economic data. The second task is to measure the extent to which the evolutionary dynamics between different expectation schemes has affected the fluctuations of the data. The dynamic model presented in this paper assumes two types of economic agents. One (rational agent) has rational expectations and the other (boundedly rational agent) has cobweb expectations. The fraction of boundedly rational agents is assumed to be determined by a non-linear function of the performance measures of each expectation. In particular, the function is formed so that the fraction of boundedly rational agents increases as its relative performance improves. If the fraction is fixed to be zero, the model in this paper collapses to a conventional rational expectations model with a linear quadratic objective function. The two models, one with an evolutionary expectation formation function and the other with zero fraction of bounded rational economic agents, are tested and compared. In preliminary GMM estimations using the actual U.S. hog data, the two models have produced similar estimated values for the deep parameters. However, specification test results support the model with heterogeneous expectations against the model assuming homogeneous rational expectations. Simulation experiments based on the heterogeneous expectations model and the estimated parameter values will be performed to measure the extent to which the heterogeneous expectations have affected the fluctuations of the market. Specifically, we will measure how much the variances of the quantity and price data will be reduced if all economic agents hold homogeneous rational expectations.