Impact of alternative length estimation and prediction periods on the stability of security and portfolio betas

Abstract

This study examines how the forecast errors of beta predictions are influenced by the following: 1) the length of the estimation period, 2) the length of the prediction period, 3) the size of the portfolio, and 4) the risk class of the security or portfolio. The mean-square error is utilized as the forecast error measure, and the components of the mean-square error (bias, inefficiency, and random error) are analyzed to determine the source of the forecast error.