Friday, January 8, 2010

Breaking the US Dollar/Commodities Link?

Yesterday's WSJ ran a report on how commodities prices appear to be shifting such that they are no longer as closely linked to the US dollar. Historically, as most all commodities are priced in US Dollars (by virtue of their being listed on US Exchanges), there has been an inverse correlation between the strength of the dollar and the prices of commodities; as the value of the dollar goes up, commodities' prices go down. This makes sense, as a weaker dollar means you'd have to use more of those weak dollars to purchase the same amount of some object.
If the WSJ report is accurate, this could be a startling trend. Let's take a look. Here's the AMEX Dollar Index over the last year:
And here's three major commodities often used to gauge the strength of the dollar, and the economy in general: Gold, Oil, and Copper over the last year (respectively):
Huh. I don't seen any sort of positive correlation there, it looks like textbook inverse correlation. What the hell are they... oh, wait (from the article):
"But since the end of November, both the dollar and commodity prices have been gaining ground."
Since the end of November? You're talking about a sample size of one month, which, with all the holidays and weekends and other gaps in trading, results in about 22 data points. And if I look at those graphs on a one month horizon (the charts are dynamic, just change the time horizon to 1m) I see oil and copper following a general upward trend, while gold is looking like the bottom of a parabola, and the dollar seems to have peaked right around December 22 and has been gradually working it's way down since then. Actually, gold and the dollar look to be in close inverse synchronicity (what one would call the "usual tie" that this article purports they are breaking) while oil and copper appear to have very little correlation with the dollar. In fact, if I plot the correlation, between the price of oil and the USD/CHF exchange rate for the month of December (as that was the data I had available), we get this, oh so meaningless scatter plot:
And a full regression analysis yields an R_Squared value of .04717 (adjusted R_squared of -.0004742). So no, Wall Street Journal, you are incorrect, dollar and commodities prices are not gaining ground together, though they DO seem to be casting off the shackles of their typical inverse correlation.
The headline should have read: "Dollar's correlation with some commodities appears to have weakened over an incredibly specific and short period of time."
Color me underwhelmed. If this trend keeps up for all of 2010, okay, then you've got something.
Full Disclosure: Short March 2010 Mini Crude Oil (QMH10) as of writing.

No comments: