« The Correlation Between Oil Prices and the Canadian Dollar: 1995-2010 | Main | The outlook for long-term unemployment »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Thanks for the data. I included as well the US Dollar Trade Weighted Major Currencies index in my calculations. Correlation between that series and oil is 0.72.

The 0.90 correlation between the USD/cad and USD/barreloil is less interesting when you see that not just the c-dollar, but all non-US currencies (ie. USD/majorcurrencies) are correlated to USD/barreloil.

I have a feeling that much of what we are seeing in these correlations is simply a general weakening of the US dollar since 2000. This shows up so strongly in the correlations because, after all, USD is the numerator in both sides of the equation. In other words, it can't be simplified to a story about the c-dollar and barrels of oil.

Agreed JP, though it does bring up an interesting question: "How much of the rise of the price of oil is simply due to a devaluation of the U.S. dollar?" I guess one way to test that would be to examine the price of oil to other currencies. But since the USD hasn't devalued 4x RE: other currencies since 2000 (but the price of oil has increased 4x), you'd think it can't *all* be simply USD depreciation.

I've tried to be very, very careful to avoid describing any kind of causal relationship since there are so many possible explanations for the relationships we see in the data.

JP and Mike

I agree with the observation that it appears that US dollar weakness is an important factor in both the strengthening of CAD/USD and WTI, and may expain the majority of the correlation between the latter two. When commenting on previous posts, I had pointed out that CAD/EUR, which should gauge Canadian dollar strength independent of the US dollar, does not appear to be particularly correlated with WTI. But after a bit more thought, I believe an correlation between WTI and CAD/USD independent of the USD can be discerned.

I haven't run any regressions to this effect, but instead have a new chart located here http://special----k.blogspot.com/2010/04/yet-more-on-cadusd-and-wti.html

In the top panel I have plotted WTI on a log scale with 1/US Dollar Index on a linear scale superimposed. If you look at the chart, you'll note that from 2003 on, 1/US Dollr index is strongly correlated with log(WTI). I can't explain why log(WTI) is correlated to linear 1/USD, but that's what the chart says. However, using log(WTI)does account for Mike's observation that WTI has gone up four-fold versus USD while the currencies have "only" gone up 50% (using log(WTI) the appreciation is similar).

Now using the basic model that log(WTI) ~ 1/USD, one observes periods where log(WTI) outpeforms 1/USD. During these periods, CAD/USD outperforms EUR/USD (and CAD/EUR goes up). One also observes periods where log(WTI) underperforms 1/USD, and during these periods CAD/USD underperforms EUR/USD (and CAD/EUR goes down). These observations suggest to me that WTI and CAD/USD are correlated independently of USD, but this correlation only appears when the price of oil diverges from what one would expect from a model based solely on USD.

Adding to the above, it seems that CAD/USD ~ 1/USD + WTI (or log(WTI) since at least 2003). The problem with this regression is that 1/USD and WTI are cointegrated. My graphical approach is an attempt to isolate WTI's contribution to CAD/USD independent of 1/USD. For a regression analysis, using cointegration techniques (which I am not familiar with) might bear fruit.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad