Conditional Moments of the Distribution of Abnormal Returns, the Cross-Section of Stock Returns, and Monetary Policy

Jensen, Johnson, and Mercer (1997) demonstrate that ME and BE/ME do not significantly explain the cross-section of stock returns during periods of contractionary monetary policy. In this study we use a dataset of stocks screened in the manner of a Fama and French 1992. We use a simple AR(1) model estimated via Quantile regression and OLS to forecast the first four moments of the distribution of abnormal returns. Where the abnormal returns come from the Fama-French four-factor time-series model applied to a pre-sample of returns for each of the stocks. In the overall sample the t-statistics from Fama-MacBeth regressions indicate the significance of the conditional mean, even in the presence of ln(ME), BE/ME, and cumulative lagged returns (to control for momentum). When the sample is bifurcated and the coefficients from periods of expansive monetary policy are examined separately from those for periods of contractionary monetary policy, none of the conditional moments are significant during periods of loosening monetary policy, while both the conditional mean and conditional kurtosis are significant during periods of monetary tightening. That is to say, that the conditional mean and the conditional kurtosis of the abnormal returns are significant in explaining the cross-sectional variation in stock returns in those periods where size and book-value are not