A new beta incorporating analysts' forecasts

T he standard beta risk measure, based on historical data, is unable to account for the superior returns on small firms or for the returns on securities neglected by analysts. The superior risk-adjusted returns on small firms were first rigorously documented by Banz 1981 and Keim 1982, while the returns on neglected firms were reported by Arbel and Strebel 1982. When we compute the standard beta, the historical return distributions are employed without modification as the best estimate of investors’ future expectations. The standard measurement of beta makes no allowance for the information contained in analysts’ forecasts, or for estimation risk in general. Thus, it gives no attention to the impact of analysts’ forecasts on investors’ expectations concerning the future volatility of returns, nor does it take account of instability in the historical return time series. These factors are not critical in the case of large, highly researched firms with relatively stable return data, where beta provides a reasonable estimate of future risk. For small neglected firms, however, the historical measurement of anticipated risk may be grossly inaccurate. We present a new beta derived from a respecified capital asset pricing model (CAPM), incorporating the information contained in analysts’ forecasts by allowing for uncertainty in the mean of the expected return distribution. The conceptual basis of the new beta is described in the next section. We then use a sample of 660 securities to test the new beta relative to the old beta in terms of its ability to 81