Whatever Happened to the Bank of Canada’s Exchange Rate “Model”?

Think back to last October-November.  The Canadian dollar went psycho for a few days — rising as high as $1.10.  Then it came back down to mere parity with the greenback (leaving us a mere 25% over fair value, measured by PPP).  Some time after that, then-Governor Dodge indicated in Senate testimony (on December 6) that he was “relieved” by the decline from $1.10 to more normal, justifiable levels.  He argued that the Bank of Canada’s own exchange rate “model” (which is actually just a rolling regression of the Canada-U.S. exchange rate on several key determinants) predicted that an exchange rate in the 95-98 cents rangs was quite legitimate.

His exact words:  “Now 98 cents sounds very specific. I don’t intend it to be nearly that specific but something in the mid- to upper 90s seemed to be pretty consistent with that.”

I wasn’t the only person shocked that the central bank governor would be so explicit about a specific trading range for the dollar that he was comfortable with.  Yes, I agree wholeheartedly that 98 cents was a lot better than $1.10.  But 98 cents is neither natural nor tolerable.  It is almost 25 percent above the fair value of the dollar (measured by PPP levels – which the OECD pegs at around 81 cents), and still far too high for any non-resource export industry (including tourism and other tradeable services, not just manufacturing) to stay in business.

Making matters worse, later on, Finance Minister Jim Flaherty seized on Dodge’s words and ran amok with them.  He said, on January 29:  “The Bank of Canada has targeted a range of 95 to 98 cents or so for the dollar, and that remains the target of the Bank of Canada, which we support.”  He thus converted Mr. Dodge’s “comfort level” with 95-98 into an actual target range.  The Minister obviously misspoke, and must have been quickly corrected by the Bank of Canada and his own officials.  (Even mis-speaking, that’s a stunning conceptual error for a Finance Minister – one for which he has never been held to account.)  The Bank, officially, only has one target for anything, and that’s core inflation.  But clearly, Mr. Flaherty’s and Mr. Dodge’s words sent a signal – intended or otherwise – to financial markets that Canada’s top policy-makers were quite happy with a dollar near parity.

I and others engaged with the Bank of Canada about exactly how their so-called “model” indicated that a 95-98 cent dollar was somehow “normal” or expected (and, implicitly anyway, actually desirable).  The model is a running regression of the exchange rate on its key determinants: largely world commodity prices (both oil and non-oil), and the Canada-U.S. interest rate differential (reflecting cyclical and monetary policy differences).  The problem with all these regressions (and the Bank’s is no exception) is that their coefficients tend to be unstable over time.  In fact, the Bank’s model is so unstable that even the sign of the oil price coefficient (that is, whether a higher oil price translates into a higher loonie or a lower loonie) depends on the time frame (sample period) covered by the regression.  (If you use only more recent data, the sign is positive; if you use a long-term sample, the sign is negative.)  It describes at best a statistical correlation (and an unstable one, at that), not an actual structural model of exchange rate determination.  And in no way should it be interpreted as reflecting any theory about the “fundamental” value of our exchange rate.

Fast forward to April 2008.  Lo and behold, the loonie has stayed pretty closely in that near-parity range mapped out by Dodge and Flaherty for the last half-year.  Yet the underlying determinants of the exchange rate (according to the Bank’s “model”) have changed dramatically.  In particular:

  • energy commodity prices are 50% higher, and even non-energy commodity prices (such as agricultural prices) are up substantially
  • The Canada-U.S. interest rate differential has widened by a full 100 basis points since last November (the Fed has cut by 2.5 points since then, versus 1.5 points for the Bank of Canada).

Both factors should have caused a significant increase in the Canada-U.S. dollar – one which (thankfully) has not occurred.

This leaves me wondering about the status of the Bank’s exchange rate “model,” about which I was highly skeptical all along.  The coefficients on their running regression may have changed again.  Their most recent monetary policy report assumes that the dollar stays around 98 cents (despite the pro-appreciation changes in its determinants).  I do not think that assumption is broadly compatible any more with the coefficients of their former model, and I’d be interested to hear how they explain that.

It also leaves me wondering about the power of “jawboning.”  Here we’ve had Canada’s top policy-makers map out, in surprisingly strong language, a highly unofficial “target” for the dollar — one that, surprise surprise, has actually come true (despite changes in the crucial variables underlying that exchange rate).  Are currency traders actually organizing their behaviour around the Dodge-Flaherty target zone???

Whether or not that’s the case (and only time will tell), it is abundantly clear that Canada cannot tolerate an exchange rate near par without continuing massive damage to all non-resource tradeable industries, a continuing enormous switch of production into low-productivity non-tradeables, and the continuing disintegration of our current account position (which is already in deficit).  On that basis, I humbly suggest that Flaherty and Mr. Carney (Dodge’s replacement) think about ways of reducing that unofficial “target” considerably – to something more tolerable and sustainable.

One comment

  • What is the NAIRU at you ask?

    The actual rate of unemployment.

    What model did you use to determine this you ask?

    The one with the coefficients that gave us the existing rate of unemployment!

    What is the natural rate of exchange (NROE *tm) between the CDN and US dollar you ask?

    The actual exchange rate!

    What model did you use to determine this natural rate you ask?

    The one with the coefficients that gave us the existing exchange rate!

    I expect their model has changed.

Leave a Reply

Your email address will not be published.