A number of studies have presented evidence rejecting the validity of
the Sharpe-Lintner capital asset pricing model (CAPM). Possible altern
atives include risk-based models, such as multifactor asset pricing mo
dels, or nonrisk-based models which address biases in empirical method
ology, the existence of market frictions, or the presence of irrationa
l investors. Distinguishing between the alternatives is important for
applications such as cost of capital estimation. This paper develops a
framework which shows that, ex ante, CAPM deviations due to missing r
isk factors will be very difficult to detect empirically, whereas devi
ations resulting from nonrisk-based sources are easily detectable. The
results suggest that multifactor pricing models alone do not entirely
resolve CAPM deviations.