Has vehicle efficiency really curbed U.S. oil demand?

Improving vehicle efficiency is often cited as a major reason for declining U.S. oil consumption since 2005. But is there any evidence for that claim?

Falling domestic oil consumption is a key factor in the new narrative about the U.S. achieving “energy independence” in the next eight years through a combination of improved vehicle efficiency and increased domestic drilling. An April research note by financial services company Raymond James depicted its forecast in this chart:

The much-ballyhooed May, 2012 forecast by Edward Morse of Citigroup also cited efficiency as a key pathway to energy independence by 2020, asserting, “US liquid fuels demand is in structural, secular decline due to demographics, fuel efficiency, transport technology shifts.”

Last December, Daniel Yergin, chairman of the oil consultancy IHS CERA, explained the decline in US oil consumption thusly in an editorial for the Wall Street Journal: “What’s happening? Part of the answer is demand. U.S. oil consumption reached what might be called ‘peak demand’ in 2005 and has since declined. The country has become more efficient in its use of petroleum, and that will continue as vehicle fuel economy goes up. The economic slump has also muffled demand.”

In April, I critiqued the supply side of these new energy independence forecasts, but the demand side remained opaque. I wondered: How much of our oil demand decline since 2005 owes to efficiency, and how much to the recession?

The efficiency argument seemed dubious. Before global oil supply stopped growing in 2005 and kicked off the new era of much higher prices, the fuel efficiency of new vehicles sold in America was going nowhere fast. Passenger car efficiency improved by just 2 mpg in the 15 years up through 2005, while light trucks gained only 1.3 mpg. But then efficiency began improving more quickly, growing from 30.1 mpg in 2006 to 33.8 in 2011 for cars, and from 22.5 mpg to 24.5 mpg for trucks, according to BTS.

I decided to see what the data had to say. And soon regretted it.

Light vehicle efficiency gains

The decline in U.S. oil demand is easy enough to find. According to the Energy Information Administration, U.S. oil demand fell from 20.8 million barrels per day (mbpd) in 2005 to 18.9 mbpd in 2011, a 1.85 mbpd decline.

Although the question seemed simple enough, calculating the effects of improved efficiency in new light vehicles (cars and light trucks, including pickups, cross-overs, minivans, and SUVs) proved to be remarkably complicated.

After many hours of digging up, rationalizing, and analyzing data from the U.S. Bureau of Transportation Statistics (BTS), Autodata Corp., and WardsAuto, then filling in holes in those data series from other miscellaneous sources and trying to ensure that I was using the most recent revisions of these frequently-revised data sets, along with a good deal of head-scratching, I decided to approach the question this way: For each model year 2006 - 2011, I took the number of vehicles sold, the average fuel economy, and the average vehicle miles traveled (VMT), for cars and light trucks separately. Then I assumed that the vehicles sold in each year replaced vehicles that were 10 years older, and calculated the fuel savings for the same VMT in each year. Then I translated the gallons of fuel saved per year into barrels per day by dividing that number by 42 (the number of gallons in a barrel) and 365 (the number of days in a year).

The result: A little over 1 mbpd in oil consumption was saved through 2006 - 2011 new light vehicle sales, or about 56 percent of the 1.85 mbpd decline in U.S. oil demand. The assertions of the energy independence squad were looking good.

But that’s just the fuel saved per model-year of sales. It does not include how much additional savings resulted from, for example, vehicles purchased in 2005 being driven in the subsequent years. To simply the analysis, I left it out at first, knowing that the actual savings would be more than 56 percent.

Other complicating factors also surely played a role. For example, the average life of a U.S. light vehicle was around 10 years in 2006, but that has grown to 10.8 years in 2012, implying somewhat lower fuel savings. On the other hand, new vehicles are typically driven more than older ones, implying greater savings than the average miles driven per vehicle each year would indicate. And there are other real-world considerations, like how much driving is in the city or on the highway. Again, I put those questions aside

Despite its difficulty, this approach was still no more than a gross approximation.

Vehicle miles traveled adjustment

For one thing, people have been driving less during the recession, so we must be careful not to count that as efficiency gains. Per-vehicle VMT for new cars declined by 15 percent from 2005 to 2010, from 12,490 miles per year to 10,625. (BTS does not have separate VMT data for cars and light trucks for 2011, but total U.S. VMT only fell by 1.2 percent in 2011 over 2010, according to a different data series from the Federal Highway Administration, so I simply used 2010 per-vehicle VMT data for 2011.) At the same time, curiously, per-vehicle VMT for light trucks grew a whopping 41 percent, from 10,963 per year to 15,472. I found this hard to explain, and I suspect it will be revised lower in typical fashion in the future, but for my current purposes I used the given data. After accounting for the decline and gain in vehicle miles traveled, the fuel savings fell to 967 thousand barrels per day, or about 52 percent of the total.

As a rule of thumb, according to EIA data, the 12-month moving average of total U.S. VMT has bounced around 3 trillion miles per year since 2003, plus or minus about 40 billion miles. In December 2011, it was 35.8 billion miles under the 3 trillion mark.

Source: EIA

Refining adjustment

Next, the fuel savings should really be adjusted for refining cracks. Only about 19 gallons of gasoline, or 42 percent, can be refined out of a 42-gallon barrel of oil. Assuming that the vast majority of the light vehicles sold in this period run on gasoline and not diesel, and adjusting the VMT-adjusted number further, the total savings fell to 465 thousand barrels per day, or about 25 percent of the total demand loss.

Sales-weighted adjustment

Finally, the fuel savings needed to be adjusted for actual vehicle sales. With so many different models of cars and trucks sold, a detailed calculation of their improved efficiency would be hugely complex, so I simply tried to get within the right ballpark by using the average efficiency of new cars and trucks sold in each model year. However, the efficiency of various models with each category can vary enormously.

For example, the average efficiency shown in the BTS data for a 2011 model year car is now 33.8 mpg, while the average for a light truck is 24.5. But according to fresh September data from Motorintelligence, the top-selling vehicle in America, as always, is the Ford F-series pickup, which gets just 17 mpg in the city and 23 on the highway. Next is the Toyota Camry, with 22 mpg in the city and 32 on the highway, for a combined 26 mpg. The third best-selling vehicle is the Chevy Silverado pickup, with an unimpressive 15 mpg in the city and 22 mpg on the highway.

According to the latest data from WardsAuto, Americans buy more light-duty trucks than cars, at 6.95 million vs. 6.09 million respectively in 2011. So we must take into account the fact that some of the most popular vehicles in each category have appallingly low fuel economy.

Recent research by the University of Michigan tried to weight fuel economy by sales, and offered a monthly data series from October 2007 to August 2012. That wasn’t long enough for my purposes, but it did illustrate the point nicely, showing that the sales-weighted efficiency of new cars sold in 2012 is still only 23.8 mpg — a far cry from the 2011 model year averages. But how to take this into account without complete data for 2005 through 2007?

After playing with the data awhile, I found that the U of M fuel economy data was fairly consistently about 25 percent lower than the equivalent BTS data for each year. This wouldn’t give me an exact result, but it would be in the right ballpark. Adjusting the BTS fuel economy data downward by 25 percent, then re-running the same calculations, gave a sales-weighted, VMT-adjusted, and crack-weighted final result of just 86 thousand barrels per day from more efficient cars and trucks with model years 2006 through 2011.

The bottom line

At 86 thousand barrels per day, new vehicle efficiency can only account for about 4.6 percent of the total decline in U.S. oil consumption since the end of 2005.

Now let’s revisit my assumptions for some of the factors I left out. What if all 81 million new cars and light trucks sold from 2006 through 2011 were driven twice as far as the averages – around 21,000 miles per year for cars, and 31,000 miles for trucks? Compared against the same number of miles driven by the 10-year-old vehicles they replaced, the savings is, rather intuitively, doubled to 9.3 percent.

What if we count the cumulative gains each model year – six times the fuel savings for 2006 vehicles, five times for 2007 vehicles, and so on? That would bring the total savings to 21 percent of the total oil demand decline. Double it, assuming that all new vehicles are driven twice as far, and it’s 776 thousand barrels per day — still just 42 percent. That number would be near the most optimistic end of the spectrum.

For a final point of reference on vehicle efficiency: According to the EPA, the adjusted average efficiency of the entire existing fleet of 240 million cars and light trucks in the U.S. was just 22.5 mpg in 2010. And we’re replacing those vehicles slowly, at the rate of around 13 million a year, with new ones that currently average 23.8 mpg on a sales-weighted basis. Looking at the data this way, it’s hard to imagine how the rosy demand-side forecasts of the “energy independence” squad can come true.

Now, other analysts might — with far better research budgets and data — come up with different estimates, after accounting for additional variables in this admittedly tortuous analysis. But as it stands, my conclusion is inescapable: The majority of the U.S. oil demand decline since 2005 was due to the recession, not more efficient vehicles.

Although I won’t get into it here in detail, there is additional evidence to support this notion. According to EIA data, the decline from 2006 through 2011 in aviation fuel, “residual fuel oil consumed by the transportation sector” (which I interpret to mean bunker fuel used for waterborne shipping), and petroleum products consumed by the electric power generation sector adds up to 681 thousand barrels per day, or 38 percent of the total decline in oil consumption. On top of that we would have to add the fuel demand lost by the trucking sector.

In short: It’s not fuel efficiency. It’s the recession, stupid.

Now let’s revisit the Raymond James forecast at the top. None of the analysts who are bullish on vehicle efficiency have explained how U.S. demand can drop 3 - 4 mbpd over the next eight years in a healthy economy. If more than half of the decline in U.S. oil demand since 2005 owes to the recession, then suggesting that continuing the current trends is a positive thing is a bit like saying, “Hey, since Johnny got cancer two years ago and lost his appetite, he’s eating less than half of what he used to. If he keeps going like this, in another two years he might be eating another 25 percent less! Super!”

And if Mr. Yergin had been candid in his answer, he would have said, “What’s happening? Part of the answer is demand. U.S. oil consumption has declined since 2005 because of the economic slump. Improved vehicle efficiency has also muffled demand.”

Photo: Buckminster Fuller’s 1933 Dymaxion concept car, which got 30 mpg, seated 11 people, and had a top speed of 120 mph. (supermac/Flickr)