International Stock Return Predictability and Asset Pricing Models

We propose a new methodology for predicting international stock returns and evaluating international asset pricing models. Our Bayesian framework performs probabilistic selection of predictors and factors that can shift at multiple unknown structural break dates. The approach generates significantly more accurate forecasts of international stock returns than a range of popular models that are economically meaningful for a risk-averse mean-variance investor. Allowing for regime-specific variable selection reduces considerably the international diversification of an unhedged U.S. investor's portfolio. Our framework improves the ability of international asset pricing models to explain the cross-section of expected returns