Some Tests of APT Mispricing Using Mimicking Portfolios.

Cross-sectional and time-series tests using mimicking portfolios are used to assess the exactness of the APT with(out) a residual market factor. The first factor seems to be sufficient to span the efficient set, whether the model is estimated using (un)conditional variance-co-variance matrices that are (un)adjusted for nonsynchronous trading. Although the conditional standard deviations of the mimicking portfolios significantly explain the time-variability of security volatilities, the residuals of the mean equation still exhibit heteroskedasticity. Similar results are obtained for portfolios of CAPM-beta-ranked securities, and for randomly selected individual securities. Copyright 1994 by MIT Press.