Abstract:
We study the role of OPEC meeting dates and production announcements for predicting jumps in the oil market. The analysis covers the daily period from 2nd December 1997 to 26th May 2017, with the start and end dates corresponding to our availability of intraday oil-price data. We start our analysis by applying the standard linear Granger non-causality test to analyze whether the OPEC-based predictors cause jumps. This test fails to detect predictability from OPEC-based predictors to oil-market jumps. Yet, given the strong evidence of nonlinearity between jumps and OPEC production announcements and meeting dates, we then use a nonparametric causality-in-quantiles test. Upon employing this data-driven robust approach, we find strong evidence that the OPEC-based predictors do predict oil-market jumps, ranging from the lower end of the conditional distribution of jumps to around the median.
