Is it legitimate to use the same estimator of beta both to predict portfolio behavior and to evaluate historical performance? The answer depends on whether one can reasonably expect the value of beta not to change. Examples of cases in which this expectation is unwarranted are (1) a firm that has recently acquired another, less risky firm and (2) a firm that uses the proceeds of a new equity offering to retire debt, reducing its leverage.

If the firm’s beta has been consistently rising or falling over the period for which its historical beta is computed, its average value will lag behind the current true value, and still further behind a projection for the future that allows for continuation of this trend. Or, if a firm is exposed to a macroeconomic event whose uncertainty is known to be increasing, its historical beta will underestimate its future beta.

When there is a discrepancy between past and future beta, it is necessary to incorporate into predictions of future beta knowledge about fundamentals of the firm. Thus, for example, if there is a demonstrated tendency for a company’s beta to revert toward the norm for the industry, then any estimate of future beta should allow for this tendency toward reversion.

The more information brought to bear in predicting beta, the smaller will be the average prediction error. If the information is properly used, it will always pay to incorporate fundamental as well as purely technical, or price historical, information into prediction of beta.