What Matters in a Characteristic

We investigate how different components in firm characteristics affect expected returns and comovements in international stock markets. We decompose characteristics into country, industry, and adjusted components. Then, we use these components to capture time-series and cross-sectional variations in stock-level alphas and factor exposures. We show that decomposing characteristics is crucial to model jointly expected returns and comovements: (i) country (adjusted) components capture systematic risk exposures (alphas), (ii) component-based models outperform benchmark models, and (iii) alphas in international markets are significant, contrary to the U.S. market. However, trading on predicted alphas does not generate significant out-of-sample net performances, indicating that they are related to limits to arbitrage.