Financial Analysts Journal, May/June (1991): 49-56
Wayne E. Ferson
University of Chicago
Campbell R. Harvey
Duke University
Abstract
Despite the random walk hypothesis, which asserts that asset returns should be completely unpredictable, it is well known that returns are predictable to some extent. Does this predictability reflect inefficiencies such as market fads, or is it more systematic?
A multi-beta asset pricing model using risk factors related to the stock market, unexpected inflation, consumer expenditures and interest rates indicates that most of the predictable variation in asset returns can be explained by shifts in the assets' risk exposures (betas) and by shifts in the market's compensations for holding these exposures (risk premiums). Little variation remains to be explained by market inefficiency.
Both betas and risk premiums change predictably over time, although changes in risk premiums are far more important than changes in betas, at least at the portfolio level. The evidence suggests that investors rationally update their assessments of expected return. It also suggests that the relative contributions of different risk factors to predictable return will differ across assets.