In this study, we investigate the existence of long-run common trends shared by West Texas Intermediate (WTI) oil prices and rig count in the US in the presence of structural breaks. Co-integration between variables would allow for co-movements in the long-run and regressing one variable on the other would not yield spurious results. The policy implication of co-integration is that each variable could assist in predicting and explaining the long-run behavior of the other one.
For robustness check, we use three different co-integration tests, the Johansen procedure (based on the maximum likelihood estimation) and two residual-based tests, the Eagle-Granger 2-step procedure and the Gregory-Hansen procedure (which allows for structural breaks in the data and endogenously estimates the break date). As Gregory and Hansen (1996) argue, conventional co-integration tests fail to detect co-integration in most cases where time series exhibit structural breaks. This justifies the use of the Gregory-Hansen co-integration test in the present study.
Additionally, in this study we examine the existence of any Granger-causality between the two variables. The implication of finding either a uni-directional or bi-directional causality is that lagged values of each variable add predictive power in the estimation of the other variable, a property that can be exploited in forecasting.
The two variables of interest are oil prices and the rig count. Brent and WTI are the two most important international crude oil types which serve as benchmarks to international oil prices. Brent crude, extracted from the North Sea, trades typically at prices slightly higher than the WTI crude which pertains to most of the crude oil extracted from North America. In this study, we use data on monthly WTI oil prices expressed in dollars per barrel obtained from the Federal Reserve Economic Data. In order to transform it into constant dollars, we use the producer price index data from the same source. The US monthly rig count data is obtained from Baker-Hughes. It represents the total number of oil rigs in operation in the US during a given month. The time span covered runs from July 1987 to December 2017.