Heterogeneity in Stock Pricing: A STAR Model with Multivariate Transition Functions

Stock prices often diverge from measures of fundamental value, which simple present value models fail to explain. This paper tries to find causes for these long-run price movements and their persistence by estimating a STAR model for the price-earnings ratio of the S&P500 index for 1961Q1 - 2009Q3, with a transition function that depends on a wider set of exogenous or predetermined transition variables. Several economic, monetary and financial variables, as well as linear combinations of these, are found to have nonlinear effects on stock prices. A two-step estimation procedure is proposed to select the transition variables and estimate their weights. This STAR model can be interpreted as a heterogeneous agent asset pricing model that makes a distinction between chartists and fundamentalists, where the set of transition variables is included in the agents’ information set.