Return Predictability: Learning from the Cross-Section

Mardi | 2015-05-12
Sully 5, 16h-17h20


This paper develops an estimation framework in which the true parameters of international return processes share a common distribution. The model (i) makes e fficient use of the cross-sectional correlation in the residuals, (ii) incorporates cross-sectional information in the estimation process, and (iii) introduces economic constraints on equity premium forecasts. The eff ect on estimation precision is remarkably strong and manifests itself both in- and out-of-sample. Once cross-sectional information is accounted for, the international evidence of return predictability appears much less heterogeneous than previously reported. The United States stands out as the exception rather than the rule in having both an unusually large long term equity premium and an unusually strong return predictability.