Faias, JoséOnyshchenko, Pavlo2015-03-062015-03-062014-10-212014http://hdl.handle.net/10400.14/16835Common predictors of U.S. equity market risk premium fail out-of-sample. We provide a new cross-sectional measure of stock market tail risk. This performs better than the historical risk premium and other commonly used predictors for short- and long-term horizons. The predictive power of cross-sectional tail risk is especially remarkable for one-month horizon forecast and during contractions. We show that under a mean-variance setting, there is an economic increase in the expected return by more than 100% in the short-term and more than 50% for longer horizons.engmaster thesis201181738