No one knows. There is approximately an infinity of exchange rate models out there, and according to the one(s) used at the Bank of Canada, the CAD is over-valued. I don't think anyone will strenuously object to the notion that the CAD is in some sense too high - that is, higher than what most models would predict. But is this gap significantly greater than the normal range of fluctuations that we've seen in the past? And is this gap important enough to induce a change in how the Bank conducts policy?
We'd have a better understanding of this problem if people routinely reported error bands with their estimates (a pet peeve of mine), but they don't. So instead, I'm just going to plot some graphs and do some eyeball econometrics.
Everyone knows that movements in the CAD are closely related to movements in oil prices, so that's where I'll start. Here is a graph of daily data for oil prices and the CAD-USD exchange rate since 2000; the data from July 2009 are plotted in red.
According to this graph, there's nothing particularly alarming about recent movements in the CAD. Although the most recent data are at the upper end of the observed range of variation, they are by no means outliers.
Of course, oil prices aren't the only thing to consider. As we know from this post, movements in the Bank of Canada's commodity price index - of which oil is a component - are closely correlated with movements in our terms of trade. So here is a plot of the commodity index against the exchange rate. The data are weekly, and observations from July 2009 are plotted in red.
Here, we see that we've moved to the edge of the familiar range of variation, but the CAD has not boldly gone where it has not gone before.
There are doubtless more factors that should be taken into account here, but I don't think that adding them would alter the conclusion that the gap between the CAD's current value and that predicted by its 'fundamentals' is not large enough to justify abandoning the Bank's current exchange rate policy.
At least, not yet.
For what it's worth, my toy model (MERT) suggests the Canadian Dollar should be at 92.314399. It is currently at 92.440, suggesting the Canadian Dollar is overvalued by 1/8th of a cent.
Posted by: Mike Moffatt | August 07, 2009 at 07:57 PM
So, yes, in fact I *do* object!
Posted by: Mike Moffatt | August 07, 2009 at 07:57 PM
I think BoC has a time bomb in its hands. The CAD will crash hard as soon as real estate starts dropping where it is supported by the fundementals and the truth about CMHC gets out. Fanny and Freddy will look like extremmly well run agencies compaired to CMHC. Also it is just a matter of time before the price of oil comes down to its historical trend of 40usd, but in the process it will undershoot. I do not exclude oil hitting 25usd will all the supply out there. I expect CAD in the coming tow yeras to go below 70cents where it should be on a purchasing power basis.
Posted by: Casanova | August 07, 2009 at 09:56 PM
Casanova: CMHC isn't remotely similar to Fannie or Freddy. Since it provides insurance, for CMHC to get into trouble I think you actually need to have defaults, not just a drop in prices. Canadian mortgages are recourse loans, and we have an approximately humane social safety net so I have a hard time imagining a massive wave of defaults on Canadian mortgages, even if prices fall substantially.
BTW, the data is hinting that prices are stabilizing.
Posted by: Patrick | August 07, 2009 at 11:46 PM
I would like to believe you, but three months dont make a trend. Have a look at this very well researched article, you will understand that CMHC is in dire straits.
http://americacanada.blogspot.com/
also this guy has some very valid points:
http://marketdepth.typepad.com/marketdepth/2009/06/sell-the-banks.html
I think that as usual Canada follows US with approx 2 years lag. Home prices peaked in US in 2006 and they did in Canada in 2008. If anything, we overdid things here compared to the US and all statistics coming out show that things are considerably getting worse here than in the US.
But I guess, only time will tell. I have difficulty seeing a bright future here short term when 78% of our exports go to the US. I am placing my bets that teh CAD will go below 70cents and this will bw the best stimulus Canada will get which will pull it out of the recession. This is 2 years down the road in my opinion. As to the value of CAD it is overshooting right now, but will come down as the deterioration of canadian economy start picking up speed.
Posted by: Casanova | August 08, 2009 at 12:45 AM
I have to confess those graphs again surprised me Stephen: both on the closeness of fit; and how close we currently are to the fit.
OK, oil is a biggish part of the Canadian economy, but not THAT big. How come it correlates so closely with the exchange rate? Maybe it's that the price of oil has such a big variance relative to the other stuff that it swamps all the other stuff. Maybe it's that the price of oil correlates with the other stuff, so acts as a proxy.
Posted by: Nick Rowe | August 08, 2009 at 07:14 AM
"Maybe it's that the price of oil correlates with the other stuff, so acts as a proxy."
I suspect this is the case. I wish I knew exactly what this 'other stuff' is and why it correlates so closely to oil.
Since the late 1990s, if you do a simple y = a + bx where y is the CDN dollar and x is oil prices, you get an R-squared in the range of .8-.85. If you include interest rate differentials, that climbs to around .9.
Results of the regression here:
http://economics.about.com/od/exchangeratesbycountry/a/mert_predictor.htm
The funny thing is the regression came from an assignment I gave for a 2nd year undergrad econometrics course I was teaching. When my students came back with such tight fights I was convinced they had done something wrong. :)
Posted by: Mike Moffatt | August 08, 2009 at 07:33 AM
Oil prices (when denominated in US Dollars) may be, in part, a proxy for the strength of the U.S. dollar itself. That is, when the price of oil rises against the U.S. dollar, it is due to a fall in the price of the U.S. dollar not a rise in the price of oil.
I guess you could test this by running a similar regression using oil prices to predict the value of, say, the Japanese yen.
Posted by: Mike Moffatt | August 08, 2009 at 07:43 AM
Casanova, there is a big hole in the "Sell the Banks" comparison of US to Canada.
Because of a different income distribution and health care costs the median Canadian is better off than the median American. Also 35 year amortizations move a lot of money over time into lenders pockets, but they don't have carrying cost explosions like ARM mortgages do. It's individually annoying for someone to own a house that is worth less than the mortgage, but as long as they can carry it it's not a systemic problem.
The key question is how many people who recently bought houses are losing their jobs. If that is a problem you would expect to see a spike in mortgage defaults. It's not visible now.
As long as we keep the right wingers trying to bring back 1933 at bay we should be OK.
Posted by: Jim Rootham | August 08, 2009 at 03:28 PM
My guess is that the BoC is not so much worried by the current level of the exchange rate than by its rate of change. Although it has been more or less stable for the last 2 months, an exchange rate that appreciates too rapidly is terrible for exporters even if they have hedges in place. As the market moves, their hedges become deep in the money and their credit lines get maxed out. Due to this, they are unable to add to their existing hedges and become fully exposed to currency fluctuations on their ongoing business. Sharp swings in the exchange rate can really decimate small businesses that do not have enough capital to withstand them.
Posted by: jg | August 08, 2009 at 08:21 PM
For what it's worth, my model (essentially a slightly more sophisticted version of Mike's that I use in a quarterly projection model) says that the loonie will average 90 cents in Q3 based on the EIA's forecast of oil prices.
For Stephen - the confidence interval is 86 to 94 cents.
Posted by: brendon | August 08, 2009 at 10:12 PM
Ah. Do you recall offhand what the oil price forecast was?
And that confidence interval sounds right; it looks to be about the same as the width of those data clusters.
Posted by: Stephen Gordon | August 09, 2009 at 08:01 AM
The Oil forecast was high 60's for the rest of 2009 and about $70 on average for 2010.
Posted by: brendon | August 09, 2009 at 12:07 PM
Q3 average specifically was $65 - a little lower than the current price of around $70.
Posted by: brendon | August 09, 2009 at 12:10 PM
I'd love to see your model sometime, brendon, if it isn't proprietary!
$65 barrel oil = 88.6 cent CDN$ according to MERT, so it sounds like in agreement. Confidence intervals roughly similar as well. I'd bet the added complexity helps your model more in 'outlier' situations ($20 or $120 barrels of oil). I wouldn't trust MERT too much if oil prices got really out of whack.
Posted by: Mike Moffatt | August 10, 2009 at 08:10 AM
Thanks to Stephen and contributors. Valuable.
Nick: I suspect that oil prices have been roughly correlated with other commodity prices and massive inflows of foreign investment targeting resource sectors.
Shameless requests:
Would love to see graphs for the 1970s decade as well as the 1980s and 1990s even if the data is noisy.
What are the current Purchasing Power of Parity (PPP) estimates for the value of the Canadian dollar? I would guess it to be in the low US$0.80 range.
Has anybody attempted to estimate how much non-resource activity was crowded out by the higher Canadian dollar in recent years?
Posted by: westslope | August 19, 2009 at 05:02 AM