PREDICTING EXCESS RETURNS IN FINANCIAL-MARKETS

This paper attempts to reproduce the time series properties of nominal excess returns in a variety of financial markets using a representative agent cash-in-advance model, modified to allow for time variation in the conditional variances of the exogenous processes. The exogenous fundamental processes of the model are estimated from the data and the remaining free parameters are estimated with a simulated method of moments technique. Simulations demonstrate that the model can replicate some of the predictability features of observed excess returns for the period 1978-1991, but that it fails to account for the serial correlation and for the joint properties of one and three months excess returns.