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 representati ve agent cash-in-advance model, modified to allow for time variation i n 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 mo ments 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 correla tion and for the joint properties of one and three months excess retur ns.