It has been about 6 months since we last looked at MERT - my toy for examining the relationship between oil prices, interest rates and the value of the Canadian dollar (relative to the greenback). It turns out that in 2010 the toy has done a very good job at 'predicting' the value of the Canadian dollar, as on average it has only been off by 4/10ths of a cent.
MERT value = 48.26 + (0.6211*OIL) + (1.8306*INTGAP)
where:
- MERT = The predicted value of the Canadian dollar
- OIL = WTI Cushing Spot price in dollars
- INTGAP = Target for the Overnight Rate (Canada) - Target for the Federal Funds rate (U.S.).
Over the last 10 1/2 months, the value of the Canadian dollar and the MERT prediction for the Canadian dollar have followed each other closely:
The MERT series is a little more volatile, as in 2010 oil prices have taken more dramatic swings than the Canadian dollar. However, the swings have been in the same direction, as shown by the simultaneous drops in the Canadian dollar and oil prices in February and May.
If we took MERT literally (which we shouldn't, since it's a toy), we could use it to see if the Canadian dollar is overvalued or undervalued relative to the price of oil:
Given the current price of oil, the Canadian dollar should be just under parity, rather than the 96.7 cents it is at. Or alternatively, given the Canadian dollar, the price of oil should be priced a few dollars lower than it currently is. On average, the MERT predicted value of the Canadian dollar has been 0.4 cents higher than the actual value of the Canadian dollar, which is rather remarkable for a toy model calibrated with data from the 2002-2006 period.
Thanks for sharing this.
I wonder how successful someone could be if they used your "toy" to make money by taking advantage of changes in the CAD/USD exchange rate? It appears it could be used as a Fx valuation tool much like "Graham Price" (amongst others) for individual stocks.
Posted by: Adam M | October 27, 2010 at 01:16 PM
Hi Adam,
In theory you could look at the residuals (that is the gap between the actual vs. forecasted) and see if that was correlated with future rises/falls in the CAD/USD (or future rises/falls in the price of oil). There's probably something there, but I wouldn't be confident enough in it to start making large bets on forex/commodity markets - if I was, I would have kept MERT to myself and would be doing so now.
But maybe I should examine the issues of residuals. I'll try that in a future post.
Posted by: Mike Moffatt | October 27, 2010 at 01:30 PM
One thing I have noticed is that the relationship starts to break down once oil goes over $82 or so a barrel and MERT starts predicting a dollar above parity. It could be that parity is acting as a psychological barrier that keeps the dollar down at this price point and once the dollar passed the barrier, the relationship would go back to normal. Unfortunately (or fortunately) there's not enough periods with a $90+ oil to test this.
Posted by: Mike Moffatt | October 27, 2010 at 01:50 PM
I don't know if it's a psychological barrier, but *something* happens when the dollar hits parity. Whenever I do a scatter plot of oil and commodity prices against the exchange rate, the relationship looks stable up to parity, and then there's a sort of a kink in what had been stable linear relationship.
Posted by: Stephen Gordon | October 27, 2010 at 02:06 PM
Yeah, the same thing happens in MERT. The only thing that makes me thing it's a psychological barrier is that it happens right at parity and not at a non-focal point such as 87 or 112 cents.
Posted by: Mike Moffatt | October 27, 2010 at 02:40 PM
So, introduce the glass ceiling correction factor when it approaches parity.
Posted by: Just visiting from Macleans | October 27, 2010 at 04:56 PM
"So, introduce the glass ceiling correction factor when it approaches parity."
That's the basic idea, yeah. Problem is, there's not enough data points to really determine what that factor looks like. In particular, do things go back to 'normal' once the psychological barrier is crossed, or does the Canadian dollar rise at a significantly slower rate than 0.62 per $1 increase in oil? I suspect it's the latter, but I'm not sure what the rate should be.
Posted by: Mike Moffatt | October 27, 2010 at 05:14 PM
Any merit in trying to correlate to the TSX index for energy? The psychology might be factored in there.
Posted by: Just visiting from Macleans | October 27, 2010 at 05:27 PM
I've noticed, observationally only, a recent (last few quarters?) strong correlation to risk assets in general, specifically changes in the S&P 500 , at least for day-to-day moves. Given the size of cross-border investment flows that smells right- when there's a flight to safety in the US you would expect USD appreciation against all other currencies, & commodities. So perhaps this is only another face of the oil price change, but I wonder.
And there are definite shorter-term range limits and resistance/support points, which isn't all that surprising given some relatively large corporate hedges out there, which often encourage these sorts of patterns. But they aren't long term, as the hedges are steadily replaced at new levels, so maybe adding in a function related to the gap to a rolling average, or perhaps a gap to back-checked max/min?
And yes these all reflect real-market behaviour, in a previous incarnation in FX sales the underlying drivers routinely came up in discussions with corporate, investor & speculative clients all.
Posted by: Anders Kruus | October 27, 2010 at 07:18 PM