Time Series Variation in the Factor Zoo

Should we be surprised at the number of “animals” in the “Factor Zoo"? The ability of the CAPM as well as workhorse three- to six-factor models to explain the cross-section of returns varies substantially over time, providing scope for a broad set of factors. We study 205 previously-identified factors, documenting time variation in their significance, prior to, during, and after the periods studied by the original authors. The number of statistically significant factors, as well as the number of principal components obtained from them, varies with the cross-sectional dispersion in individual stock CAPM alphas. The number of significant factors is strongly related to the number of publicly-listed firms, and is also related to institutional ownership, recession indicators, interest rates, and measures of diversity in firm characteristics. On balance, our results suggest that the large number of factors with significant explanatory power reflects the complexity of the economic environment, including changes in investor composition, the types of firms listed, and competitive conditions