Exchange Rates and Gasoline Prices?

Well the price of gasoline just spiked upwards across Canada and the usual media analysis has begun.  Five key reasons were summarized as follows by Shawn McCarthy in the Globe and Mail: 1) Approach of the summer driving season leading to a switch to summer gas formulations which leads to a reduction in supply. 2) Higher crude prices. 3) A lower Canadian dollar. 4) Lack of retail competition in Canada and 5) U.S. exports of gasoline and diesel out of North America creating tighter demand and supply conditions. I thought explanation number three was rather intriguing.

According to explanation number three, since the dollar is off about 10 percent against the U.S. dollar and since crude prices are set in U.S. dollars, it means that Canadian refiners are paying more for oil and importers have to pay more for their gasoline.  This explanation has been floated around quite a bit over the last few days.  Well, that is an easy one to quickly get a bit of data on.  In Figure 1, the monthly value of the Canadian dollar ($CAD per $US, noon spot rate, average) for the period 1990 to 2014 is plotted against the average regular unleaded gasoline price for 18 Canadian cities.*  

 

Well, from 1990 to about 1999, the average gasoline price hovers around 60 cents a litre while the Canadian dollar depreciates – from 1.17 to about 1.49 CAD for a US dollar.  It would appear during this period the depreciating Canadian dollar did not seem to do much to gasoline prices.  The Canadian dollar depreciates further after 1999 and gas prices rise from about 60 cents a litre to 72 cents.  There may be some exchange rate effect here.  The Canadian dollar begins to appreciate after 2002 and hits parity for a while in 2007 but there is no drop in gasoline prices.  The dollar stays strong relative to the US dollar after 2007 and gas prices continue to rise.  Actually, during this period the average price of gasoline rises quite dramatically.  If anything, I would have to conclude that on average a stronger Canadian dollar is associated with higher gasoline prices while a weaker dollar is associated with lower gas prices.  Of course, no other confounding factors over time are being controlled for here but it was still an interesting picture. 

 

Figure 2 takes the first difference of both series and plots them against each other and adds a linear trend.  Increases in the amount of Canadian dollars needed to buy a US dollar (depreciation) seem to be inversely related to gasoline price increases.  Again, a depreciating currency seems to be weakly correlated (a very low r-squared) with lower gasoline prices.  Maybe there is a lag?  I also plotted the first difference of gasoline prices against the exchange rate lagged three months and still got the same weak negative relationship. 

So, I don’t find much support in a depreciating currency as part of the explanation for the rise in prices. As for the other factors – decreasing retail competition over time may be a better explanation for some of the rise in prices.  After all, in the late 1980s there were over 20,000 gas retail outlets in Canada whereas by the mid 1990s they had shrunk to 16,500.  Between the mid-1990s and 2006, they fell further to 13,772.  Today, there are apparently 12,285 retail gasoline outlets.  A full explanation would need to control for this change in the market and any changes in the concentration of ownership as well as the price of crude and the onset of more fuel-efficient cars.  Still, cutting the number of retail outlets in half while the gasoline price more than doubles over twenty years suggests that this must be part of the explanation.

*They are: St. John’s, Winnipeg, Regina, Saskatoon, Edmonton, Calgary, Vancouver, Victoria, Whitehorse, Yellowknife, Charlottetown, Halifax, Saint John, Quebec, Montreal, Ottawa, Toronto, Thunder Bay. Source: Table 3260009 - Average retail prices for gasoline and fuel oil, by urban centre, monthly (Cents per litre). Exchange rate: v37426 Canada; United States dollar, noon spot rate, average.

 

APRIL 26 UPDATE

The point of the post was to simply examine the relationship between the exchange rate and the price of gasoline given the argument made in the media that a depeciating dollar was a factor driving the spike in gas prices.  It was obviously too simple an examination even though I thought I had qualified matters by acknowledging other factors (including the price of crude) in the last paragraph.  I have had a number of blog comments (as well as e-mails) telling me I cannot look at gasoline prices without taking the price of crude oil into account.  Most of the comments have been polite and not questioned my credentials as an economist.  The points are that crude oil prices are a major driver of gasoline prices (with which I agree even if the post did not explicitly state this point) and that the Canadian dollar is also driven by the price of crude oil (Well perhaps. I always thought it was commodity prices in general of which crude oil is of course an important component).  Okay. I went out and got the monthly price of West Texas Intermediate.  I plot two more figures below - one with the exchange rate and  the price of crude over time, the other with the average price of gasoline variable I used in the post and the price of crude over time. I think the figures speak for themselves. The one plotting the exchange rate over time with the price of crude suggests that there are periods where rising crude prices move with an appreciating dollar (certainly from 2001 to 2008 the relationship looks strong).    However, from 1990 to about 2001, the dollar depreciates while crude oil prices stay pretty flat.  As for gasoline prices and crude? Well, the price of gasoline and the price of crude seem to move together very nicely.  Let me state at this point that a full explanation would require a more methodical and sophisticated empirical analysis perhaps using some type of simulataneous regression framework to account for potential simulataneity in the relationship between crude oil prices, the exchange rate and the  price of gasoline as well as other variables such as a measure of retail market concentration.  As well, one might want to try a variety of different crude oil prices. I'm sure that would be a fun and interesting project.

 

Related articles

Gasoline Prices

Comments

[jt]

Usually, integrated oil's financial reports say "they price (wholesale) gasoline as a unified North American market" (i.e. gasoline futures in USD, ... http://www.cmegroup.com/trading/energy/refined-products/rbob-gasoline.html).
The current price being $C 0.89.
PetroCanada displays their wholesale prices (http://www.petro-canada.ca/en/wholesalefuel/5531.aspx), and the lowest prices (i.e. at a refinery) are very close to this price.
But, as you noted, the wholesale price of gasoline is only about 2/3's of the current retail price, so other factors can be important at the retail level.

[Phil]

Hmmmm ... I wonder even if there were an effect, whether you'd see it in that kind of chart. If gas goes up (say) 80%, and the US dollar down 20% ... well, it could just be that the price of gas would have gone up 125% if the US dollar had stayed flat. But how can you tell if 80% or 125% is the "right" number?

I think you'd need to go more granular. I'd be willing to bet that on a day-to-day basis, the price of gas is more likely to rise on a day the US/CAN exchange rate also rises. Even on a weekly basis. That assumes that rising crude prices don't push the Canadian dollar up. If they do, my guess is you'd have to control for that, too.

Hmmm ... OK, how's this prediction? If you ran a regression to predict weekly gas price changes from two independent variables -- change in crude price, and change in US/CAN exchange rate -- both would be significant and positive. It might work for monthly, too -- the regression in the post doesn't include the crude price, right?

[Livio Di Matteo]

No monthly crude price in the regression Phil. Interesting points. Ideally, I would want to put together a data set with a larger number of variables to control for other confounding factors.

[Shan]

Just plot a chart showing Brent crude divided by gas price, and the exchange rate. No more sophisticated model for other things that influence gas prices is required to see what looks to me like a pretty obvious relationship between them.

If you're looking for other influential variables, I would think that regional and global refinery capacity utilization might be a more promising place to start than retail gasoline outlets. Going by anecdotal evidence from the gas stations that have closed around my area in recent years, vigorous competition keeping profit margins on fuel sales very small to negative was part of the story.

[John C]

To the extent Canada is petroleum exporter, would n't the value of the C$ movement to move with price of crude? This might explain the positive correlation to some degree.

[PelinoC]

There are three markets behind the consumer's gasoline price: market for crude oil (which is nominally global, but because of a glut in North America somewhat regional - see the difference in prices between Brent Crude, West Texas Intermediate, and Alberta heavy oil); wholesale market for refined products (definitely regional, because of transportation costs); retail market based on distribution chains and retail competition.

The various indicators for crude prices are available, so crude price movements could be extracted from the data. Intermediate prices for refined products are also available (because there are major distribution centres where refined products are sold to independents, and they post prices online - though this data only goes back so far and is harder to get). There have been consumer price spikes in the past that have been largely caused by refinery market issues, such as supply constraints caused by refineries going offline during busy summer seasons, for example. Again, these types of movements could be extracted from the retail price data you presented (note that the intermediate price is different in different cities across the country, because of transportation costs from refineries to retail distribution centres).

"Seeing" the impact of the Canadian dollar on all of these markets is tricky, because while crude prices are fully affected by changes in the dollar (because as the newspaper notes, crude prices are denominated in US dollars - even Alberta prices are just backed out from Texas prices), refined product prices are only partially affected by the dollar (some grades of gasoline are "Canada only" and are not traded, and some refineries produce only for the Canadian market, while others sell into both Canada and the US interchangeably).

It's messy, and not surprising that there is not a clear signal in the retail price.

[Andrew_leach]

Livio,

You write that, "from 1990 to about 1999, the average gasoline price hovers around 60 cents a litre while the Canadian dollar depreciates – from 1.17 to about 1.49 CAD for a US dollar." That's true, so far as it goes. However, over that time period, the average annual growth rate for world oil prices denominated in Canadian dollar terms was less than 1% per year (0.8% per year on average). So, as the commenter above points out, given that the Canadian dollar input costs didn't change over that time period, there'd be very little reason to expect an increase in gasoline prices. Further, if you track traded gasoline, NY HArbour gasoline in $CAD terms actually depreciated over that time period, although only by an annual rate of -0.4%, so again, there's no reason you'd have expected Canadian cash costs of gasoline not to reflect global commodity prices in $CAD terms.

Contrast that with the most recent data we have, which I wrote about at Macleans yesterday. Over the past 12 months, world oil prices have increased (in $US terms) about 6% and in Canadian dollar terms, about 14%. These have been matched by an 8% increase in Toronto gas prices, which is actually a little smaller than you might expect. In the March-March comparison, Toronto gas prices follow world crude perfectly, but April has some weirdness in the data. If you look at the more relevant comparable, world gasoline prices, these have jumped over the past year in Canadian dollar terms by much more than in $US terms, and Canadian gas prices generally track the Canadian dollar value of US tidewater gasoline very well.

I think your analysis over time needs to account for their being a relationship between crude and the Canadian dollar - over the 1990s time period you cite, the real dollar decrease in crude prices globally (they increased by 1.1% per year in $US terms) was exacerbated in Canada by a decrease in the dollar, which kept crude prices even flatter in Canadian dollar terms than they were in the world market over that period.

All else equal, it's hard to argue that a depreciation of the Canadian dollar would not increase the domestic cash costs of a traded commodity. Of course, all is not equal when it comes to the Canadian dollar and oil prices, so my analysis is as subject to that critique as yours.

[Livio Di Matteo]

Andrew: You are certainly correct that a fuller explanation requires taking the price of crude into account- I think I mention it along with some other factors at the tail end of the piece. I guess the point of the post was to deal with the argument being made in the popular media that the falling dollar was driving the current spike in prices. I looked at the data and could not find a lot of support for that point. However, you probably would want to do a fuller analysis whereby some simultaneity between crude prices and the exchange rate is taken into account when trying to isolate the effect on gasoline prices. Cheers. Livio.

[Livio Di Matteo]

I have updated the post. Livio.

[Andrew_leach]

Agreed, although I think in the same way that people made comments that "if the dollar went down, exports would be worth more", they can make comments now saying "if the dollar went up, gas would be cheaper". In both cases the simultaneity issue is important. While we might want a lower dollar in and of itself, we likely don't want the circumstances most likely to cause a low dollar, and vice-versa.

[genauer]

a comment from Germany:

Our gasoline price is, usually within 1%, World market Brent price / 200 liter * Euroexchange rate + taxes (about 60% of the total) + all others (refining, shipping, sales, etc, ca 10%, and adjusted for 2.0% inflation) Actually as sheer logic would expect it.

We had a similar discussion here before.

It was somewhat higher, after so many refineries closed, that we apparently had to import larger quantities of finished good. Refinery open again, 4 weeks later the premium disappeared.

In summer when large provinces start school holidays, premiums of 3 or 4 cents also occur on occasion.

Works perfectly in my hands since 1990, but because of frequent tax changes in the past, different currency deutschmark and Euro, and the occasional need to know the german calendar, the excel sheet would look like a mess for you : - )

Until about 3 years ago, the difference between WTI and Brent was so small, that you could forget it, now East Coast / Canadian RBOB prices should be a mix of about 80% Brent and 20 % WTI

Back in 2003 the formula also worked in upstate New York

[Nathanael]

Asking whether exchange rate effects or other effects are dominant?
I'd start with
(1) the ratio of the Canadian gasoline price to the US gasoline price,
and plot that against
(2) the ratio of the Canadian dollar to the US dollar

This eliminates crude oil price effects, which are clearly the first-order effect.

If the two ratios track one-for-one over time, then the remaining effect is due to the exchange rate. If they don't, then something interesting is happening, such as less competition or whatever.