Brent update

James Hamilton has an interesting posting over at Econobrowser.    While Cushing OK WTI is considered the US benchmark grade of crude oil and is the NYMEX deliverable grade, gasoline in the US is currently priced off of North Sea Brent crude.  This is due to pipeline constraints which have prevented midwest oil producers from getting their product to market - so Brent serves as the marginal barrel of oil for most US refiners.  (A discussion of that point can be found here.)   Hamilton shows that the price of US gasoline and Brent crude are cointegrated with an intercept approximately equal to the sum of taxes and refining margin on a gallon of gasoline, and with the coefficient on the price of Brent being the conversion factor from barrels to gallons (1/42).

For those who have never heard of cointegration here is a brief explanation.   If you you have two time series, each of which in isolation follows a random walk (or worse a random walk with drift), and you attempt to regress one of the variables on the other (both in levels) then you can end up with test statistics which imply that the two series are highly related when in fact they are not.  The problem is that when there is a random walk in the data the test statistics do not converge at the expected speed.  You can generally see the problem by looking at the regression residuals.  If your test statistics are significant but the regression residuals appear to follow a random walk then you likely have a "spurious regression" problem.

However there are cases in which you have two variables, each of which follow a random walk in isolation, but they maintain a stable relationship between themselves - like say gasoline and Brent crude prices!  In this case if you regress one of the variables on the other then the regression residuals will appear to be white noise.  In this case we say that the two variables are cointegrated of order 1 and there are meaningful test statistics for the coefficients in the regression.  Hamilton's textbook is the definitive source on this topic.

Now back to Brent:  it should not be at all surprising that gasoline and Brent prices are cointegrated.  66% of the cost of a gallon of gasoline is due to the crude oil input price.  Gasoline demand is notoriously price inelastic ie the demand for gasoline responds slowly to the price of gasoline.  Hence producer cost shocks will be passed through almost completely to the price of gasoline.  Since taxes and refining margins are fairly stable (except for that brief Katrina / Rita period) we should expect a fairly stable relationship between gasoline and the underlying oil price.

So Hamilton's good news is that with Brent prices falling from 125 USD / bbl in March 2012 to 95 USD / bbl today we should expect to see gasoline prices follow.  The bad news is the reason that oil prices fell was that oil demand has been slowly falling as well as the world economy has struggled.   Then on Friday Brent was up over 6 USD / bbl - the 4th largest move on record.  Still you have to be happy that crude oil prices are responding to the slowdown.  It will provide a bit of extra stimulus.