Empirical Tests of the ZCAPM

This chapter presents evidence in support of the empirical ZCAPM. Based on U.S. common stock returns in the period January 1965 to December 2018, across a variety of test asset portfolios, out-of-sample cross-sectional tests of the traditional approach to specifying return dispersion (RD) in an asset pricing model yield insignificant or marginally significant estimated prices of risk in most cases. The failure of this traditional model justifies using the ZCAPM to take into account the symmetric effects of RD on stock returns. Graphical analyses of the empirical ZCAPM estimated via the EM algorithm show a close relation between zeta risk and one-month-ahead (out-of-sample) cross-sectional excess returns for a variety of test asset portfolios. Interestingly, zeta risk exhibits strong goodness-of-fit with the one-month-ahead returns of widely used size and book-to-market equity portfolios by Fama and French (1992, 1993). Further graphical results show that the empirical ZCAPM outperforms popular Fama and French (1992, 1993, 1995, 1996, 2015, 2018) three-, five-, and six-factor models in terms of predicting one-month-ahead stock returns, especially among industry portfolios. The latter poor cross-sectional performances of these popular models across different industry portfolios is a major shortfall that appears to be attributable to their inability to add much explanatory power in time-series regressions to the CAPM market model. In sum, graphical evidence based on out-of-sample returns lends strong support for the ZCAPM.