US Tail Risk and International Stock Return Predictability

In this paper, we aim to investigate whether lagged U.S. tail risk can predict non-U.S. returns and whether lagged non-U.S. tail risk can predict U,S. returns. We measure the U.S. tail risk by constructing a portfolio that long the CBOE Put Protection Index (PPUT) and short the S&P 500 index. We find that lagged U.S. tail risk displays strong predictive ability for non-U.S. returns by means of predictive regression model, pairwise Granger causality test, adaptive elastic-net estimation, variable importance and out-of-sample forecast gains. Whereas, non-U.S. tail risk exhibits limited ability in predicting U.S. returns. Keywords: tail risk; Granger causality; LASSO; variable importance.