Peak Gasoline?
Literally, I could spend all day debunking crap like this off of the web. Econbrowser highlights that
Calculated Risk caught today’s big story from the Energy Information Administration, which was that the drop in U.S. gasoline demand (discussed by Big Picture and William Polley last week) has turned into a freefall.
Generally speaking, anything that’s picked up by both Calculated Risk and Big Picture is sure to be viewed, and presented, through “Bear-Tinted Glasses,” and there’s a good chance the content is pure crap. Econbrowser proceeds to show a chart with 2004 and 2005 summer demand. Kedrosky even chimes in, with a chart (lifted from some other site) showing two years of gasoline demand.
Here’s the debunking, and it’s easy. Download the weekly gasoline demand data from the DOE and make a chart for yourself from 1991 to the present, the longest timeframe possible with that file. Add a 14-week (one quarter) moving average and a 52-week moving average. Never mind, I’ve done that below.
Notice how the quarterly moving average of demand, shown in blue, oscillates around the longer-term annual moving average of demand, shown in red. Wow, looks like there’s a good chance this is just another low point in the oscillation around trend. Notice also how the red line has turned briefly negative in the past, such as in 1996, with no real impact on long-term trend.
Here’s another feature of the chart, easily seen but less easily quantified. Notice how the gray area seems to be getting tighter over time, in relation to the long-term moving average? That means demand for gasoline is becoming more consistent throughout the year, i.e. winters going up and summers going down, relative to each other. You could quantify this by taking the 52-count standard deviation of each week’s ratio to its trailing 52-week average (done, but not shown in the chart, feel free to learn your own Excel skilz), and noting that it has fallen from 4.5-5.5% in the early 1990s to under 3% now, with a pretty clear demarcation around 2002. Fascinating what looking at more than 2-3 years worth of data can tell you.
Freefall? Falling off a cliff? Peak gasoline? What a load of crap. One could have said the same thing at many other points in the last 17 years, and they would have been just as wrong then, as they are now.
It’s up to the reader to decide if the “peak gasoline” commentators actually believe what they wrote, or if they were just being scary or sensationalistic for other reasons; I’m pretty sure there’s some sensationalism for the sake of expanding readership mixed in there with the “believing their own BS” aspect, but your mileage may vary. Regardless, a 1-3 year look at something so obviously erratic about its trend as gasoline consumption is worth demolishing with a longer-term view. Go back to the chart and see how easy it would be, to take any one date where the consumption was below average, and cherry-pick a time the year before it, two years before it, or three years before it, to show how consumption had “fallen off a cliff” since then. Hogwash!
The demand for (and production of) energy per capita is on the upswing, and barring some humanity-threatening catastrophe, it will continue to be on the upswing. Ditto with world population, and unless (until?) the U.S. experiences either a Balkanization or massive economic crash, double-ditto with the U.S. population. Demand is going UP over the long term, and will keep doing so. Don’t be fooled by the shorter-term oscillations around the uptrend line.
The Stone Age didn’t end for a lack of stones; neither will the Petroleum Age end for a lack of petroleum. Bank on it. The only thing that changes paradigms and “ages” are true technological breakthroughs, and of the many game-changing tech breakthroughs over the centuries, would you care to determine the proportion that were funded by taxes funneled into big corporate R&D departments?
For those so inclined, the next question: is there a contrarian opportunity to be found here? Or do these assorted “peak gasoline” pundits’ dislocated and distorted views on the gasoline consumption trend merely represent a fringe element to be ignored (after being debunked, of course)?


February 25th, 2008 at 7:29 pm
Bill,
That was a great post.
Jeff
February 25th, 2008 at 8:51 pm
The quote of mine that you use was made September 14, 2005, and referred specifically to the consequences of the hurricanes. I believe that episode shows up rather clearly in your gray line for 2005. I suppose someone reading your remarks might instead construe these to be comments I am alleged to have made about the current situation.
I for one had never heard this phrase “peak gasoline” until seeing your entry. I wonder who these “peak gasoline pundits” might be?
February 25th, 2008 at 9:16 pm
The Kedrosky link got me rolling, and I can’t find the linkage or search I used to get to your post. I had thought it current when I made the link to you. That’s what I get for not assiduously checking dates of posts.
Let’s examine this quote, James.
“Do demand curves slope down? I’ve been telling my students they do for years. And I imagine I’ll be telling them the same thing for a few more years to come.”
Would you care to discuss what happened to both gasoline prices and gasoline demand since then? It appears that you, Barry, and Calc.Risk were all mentioning a demand curve that isn’t very downward sloping at all, and mentioning it at almost an exact “bottom” for demand. Chart. Both prices and demand are higher today. Did that turn out how you expected it to?
I expect that demand will continue to oscillate about that upward trend line, with seasonality, regardless of price action.
February 26th, 2008 at 7:00 pm
I think your graph sucks. You’re eyeballing a 52wk moving average to discern growth rate rather than plotting it directly.
http://www.imagebam.com/image/2ab50c3177163
Displays YoY change in weekly Gasoline supply and it’s regression.
I think the data supports two conclusions:
1) The data has become less volatile (it does not speak to the source as a decrease in sampling error or a decrease in demand volatility - as you imply)
2) The second derivative of supply (the slope of the regression line) is clearly negative and has fairly stationary residuals (suggesting a decent fit). This shows that the increase in demand has been slowing… .
February 26th, 2008 at 7:17 pm
[…] was some previous discussion along these lines in Peak Gasoline. This entry was written by Bill and posted on February 26, 2008 at 7:16 pm and filed under […]