Using interest rate differentials to predict exchange rate movements is a good lesson in hubris. In 1997, the Bank of Canada’s Board of Directors had one of their regional meetings in Quebec City, and as part of this exercise, local notables were invited to an off-the-record dinner with then-Governor Gordon Theissen and other Bank officials. Two places were set aside for Laval economics profs, and since I hadn’t yet had the pleasure, I and another junior prof went.

I set myself the task of coming up with a Penetrating, Insightful Question as a way of paying for a very nice supper (plus drinks). I took a look at some recent numbers, and noted that the US-Canada real interest rate differential had increased by quite a bit over the previous months. So that meant that the markets were expecting the CAD-USD rate to increase (or at least, stop falling), right? So here was my question, paraphrased since a) it was a long time ago, and b) I said it in French:

The Bank of Canada has made inflation targeting its central goal for the past few years. During that time, the CAD-USD exchange rate has declined, and our export sector has benefited greatly. But over the past few months, Canadian interest rates have fallen with respect to US rates, so it would appear that the exchange rate will start rising soon. Will the Bank allow the exchange rate to rise, even if it starts to hurt manufacturing exports?

I don’t think I’d be violating any rules of confidentiality if I reported that Mr Theissen’s response was ‘Yes’. I thought it was a good question, but as time went on, I was more and more pleased that the dinner was off the record. Instead of increasing, the CAD-USD exchange rate continued its downward trend for another four years.

The standard relationship between interest rates and exchange rates is the following:

real US rate = real Canadian rate + expected appreciation of CAD-USD exchange rate

As Menzie Chinn notes in this excellent post at Econbrowser, the horizon at which this relationship works is somewhat problematic. If you define the real rate as the nominal long-run bond yield less y/y CPI inflation, this relationship works best for a horizon of seven years. And even there, the correlation coefficient between actual and predicted appreciation is only 0.55.

Here is a graph of the real, long-run interest rate differentials and the average rate of growth of the CAD-USD exchange rate over the following seven years. In retrospect, my question wasn’t quite so embarrassing after all – thanks to the sharp rise that started in January 2002, the average growth rate of the exchange rate over the seven years following that supper was around 4% a year.

But here’s the weird thing: since 1998, there’s been no substantial movements in the interest rate differentials. During this time, the CAD-USD exchange rate has gone from 85% of PPP to roughly parity. Does this mean that the market expects the CAD to keep rising beyond its PPP level?

Stay tuned.