Monthly Forecasts of Systematic Risk: An Evaluation

Abstract

Recent advances in covariance and variance estimators coupled with improvements in the quality of intra-day data have made possible more precise measurement of beta (systematic risk). In this paper we examine the forecastability of monthly betas for Dow Jones stocks. The out-of-sample forecasting exercise conducted in our study results in a dramatic reduction of forecast error of beta on average by over 80%, relative to the industry standard of the constant model. This finding has vast implications for all aspects of finance as precise forecasting of the beta parameter is of crucial importance.

Presented at the 2007 North Americian Summer Meetings of the Econometric Society at Duke University.

Keywords: Beta, Forecasting, Systematic Risk