Stephen Gordon reports on a 'plausible' reason for the apparent structural break in the relationship between the US dollar-Canadian dollar exchange rate and oil prices around 1993:
Worthwhile Canadian Initiative: Oil prices and the Canadian dollar: A mystery solved: In an earlier post, I noted that the relationship between the CAD-USD exchange rate and oil prices hasn't always strong as it has apparently been over the past few years. Since I hadn't been able to sort this problem out on my own, I went into last weekend's session on this topic organised by the Bank of Canada at the meetings of the Canadian Economics Association with a certain amount of the hope. Happily, I wasn't disappointed.
The story starts with Bob Amano and Simon van Norden's paper written in the early 1990's, in which they found that an increases in the price of oil prices were generally associated with depreciations in the value Canadian dollar. Unsurprisingly, the Bank of Canada has found this to be a less-than-spectacularly-successful forecasting model over the past few years, so they've been taking a second look.
Robert Lafrance's (et al) paper suggests a structural break in the relationship between oil prices and the exchange rate sometime around 1993. Sure enough, when you go to the data, the correlation coefficient between the WTI oil price (divided by the US GDP deflator) and the CAD is -0.69 before 1993, and 0.71 afterwards. Why?
The third paper of the session, given by the IMF's Tamim Bayoumi, suggests a plausible explanation. Instead of using oil prices as an independent variable, they use the value of net oil exports, and their model tracks the last few years quite well. Here's a graph of the real value of net oil exports and oil prices:
Click on figure to enlarge
Ordinarily, we'd expect that a rise in oil prices would increase the value of net oil exports, thus leading to an appreciation of of the CAD. And if you look at the data since 1993, that story seems to be consistent with what we observed.
But the story from the 1970s and 1980s is very different. The rise in oil prices during the 1970s was associated with a reduction in the value of net oil exports, so it's not surprising that an econometric model that used data from this period would pick up that negative relationship.
After this was pointed out, the general consensus of those who were at the session seemed to be that the experience of the 1970s and 80s was explained by the dirigiste oil policies of the era; the reaction of the governments of the day to the increase in oil prices was to reduce exports in order the ensure 'secure' oil supplies for the domestic market. So even though oil prices were increasing, net exports of oil went negative, thus putting downward pressure on the CAD.
Once oil production and trade had been liberalised (and especially after NAFTA), the correlation between the CAD and oil prices went back to being positive.