It seems that no matter what financial series you look at, there’s a similar pattern of ups and downs over the last few years. I was curious to get a quick quantitative impression of how much of a contribution aggregate factors have been making to recent movements in the price of oil.
I commented on Sunday on recent big swings in stock prices, which rose in the fall of 2010, declined sharply over sovereign debt worries last summer and fall, rebounded early this year, and over the last month have moved back down. The same basic patterns are seen in the dollar price of oil.
How much of the recent moves in oil prices can be explained by changing perceptions of global economic activity? One way I thought to get an impression of this was to look at the extent to which recent oil price movements are mirrored in other commodities. I was able to assemble a quick data set on spot prices for copper, corn, palladium, platinum, soybeans, and wheat, and calculated the principal component of the weekly percentage changes in these 6 commodity prices over January 2005 to September 2011. The value of this principal component for any particular week in fact turns out to be pretty close to the average change across the 6 commodity prices for that week. I’m a big believer in using parameters that are a priori plausible values in preference to over-fitting a given sample, and the principal components analysis suggests that a simple average would be an excellent summary statistic to use for these purposes.
I then regressed the weekly percentage change in the price of crude oil on the average change for the other six commodities over that same week, and came up with a coefficient of 1.15. Again the principle of parsimony suggests that a value of 1 is a pretty good a priori reasonable value to use to represent that regression. That gives me then an extremely simple-minded way of answering the question, that in fact is pretty close to what an exact fit to the historical data would lead you to choose, namely, the predicted percentage change in oil prices this week is just the average percentage change observed for other commodities over this week.
The black line in the graph below shows what happened to the actual price of oil since September 2010. The green line is the price you would have predicted since 2010 if you knew nothing other than what happened to other commodity prices since 2010 and assumed that oil would do exactly the same thing as other commodities. An increase in the price of oil from $80 to $100 in the last quarter of 2010, followed by a decline back down to $80 in the fall of 2011, would have been exactly in line with what happened to other commodities. One need appeal to nothing special happening to the oil market over this period, other than an increase in demand for oil along with other commodities as prospects for global economic growth first picked up and then declined.
What we’ve seen since last fall, however, is a little different. If oil had risen no more than these other 6 commodities, it wouldn’t have gone above $93 at its peak in March. More than half of both the move up in the price of oil at the start of this year, as well as of the sharp drop in the price of oil over the last month, may reflect influences specific to the oil market.
Among those factors, efforts by refiners to find alternative suppliers for boycotted Iranian production played a role, though I’m wondering what others see that I don’t suggesting that tensions in Iran are now settling down. There are also some moderately encouraging developments in recent production numbers.
But I think the basic impression from the first two figures plotted above is correct– much of the drop in oil prices over the last month is a response to a worsening outlook for global economic activity.
This post originally appeared at Econbrowser and is posted with permission.