Abstract:
This paper provides a novel perspective to the predictive ability of rare disaster risks for West Texas Intermediate (WTI) oil market returns and volatility using a nonparametric quantile-based methodology over the monthly period of 1918:01–2013:12. We show that a nonlinear relationship and structural breaks exists between oil returns and various rare disaster risks; hence, linear Granger causality tests are misspecified and the linear model results of non-predictability are unreliable. However, the quantile-causality test shows that rare disaster-risks strongly affect both WTI returns and volatility, with stronger evidence of predictability observed at lower quantiles of the respective conditional distributions. Our results are robust to alternative specification of volatility (based on a GARCH model), and measure of rare disaster risks (based on the number of crises).