The bias in the long run relation between the prices of Brent and WTI crude oils
This paper is about the magnitude of association between two crude oils, the UK Brent and the Texan WTI. Practice presumes and theory predicts a unitary and a proportionate association, especially for the log-log specifications. A battery of cointegration tests are conducted to test whether the slopes, or the cointegrating vectors, are statistically significant and are statistically higher than +1. All cointegrating vectors are found to be statistically higher than +1 whatever the sample frequency selected, monthly, weekly, or daily, and whatever the functional forms or the econometric procedures adopted. Subsidiary results are that the samples chosen do not contain calendar structural breaks, and that regionalization of the oil market is strongly denied. These results reject the underlying intuition and theory, and set the stage for a possible financial anomaly.
Keywords: Brent and WTI crude oil spot prices, cointegration, error-correction models, GARCH methods, bias in the association, three data frequencies, oil market integration.
JEL Classifications: C22, C12, G15, Q41