Beta Forecasting: A Two-Decade Evaluation

Abstract:     
The purpose of this paper is to compare a series of models to forecast beta, a fundamental parameter in finance. Realized measures of asset return covariance and variance are computed and applied to forecast beta following Andersen, Bollerslev, Diebold and Wu (2005a and 2005b). This approach is compared with the traditional constant beta model and a variant, the random walk model. It is shown that an autoregressive model with two or three lags that is estimated on the previous 80 quarters of realized beta produces the lowest or close to the lowest error for quarterly stock beta forecasts. In general, the AR(3) model has a mean-squared-forecast-error half that of the constant beta model.

Number of Pages in PDF File: 30

Keywords: Portfolio Management, Realized Beta, Autoregressive Model

JEL Classification: G0

working papers series