Wednesday, September 21, 2005

The "paradox of fuel efficiency" and the misuse of statistics

Jeff Jocoby, one of the Boston Globe's conservative columnists wrote "The paradox of fuel efficiency" for today's op-ed page. He argues, on the basis of a book entitled "The Bottomless Well", that increased fuel efficiency increases dependency on fuel. If there's anything that economists (and physicists) love presenting, it's counterintuitive results. I haven't read the book, but, if Jacoby is presenting the argument accurately, I doubt that it will convince me.

Here are some excerpts from the column, and why he's got everything totally wrong:

He analogizes cars to computers:

Consider how much more use you get from your computer today than you did from the far less efficient PC you owned 15 years ago. As the efficiency of computers has climbed, so has the demand for them. Today it costs less than ever to process a byte of data by computer -- but more resources are devoted to computing than ever before. Driving is no different. If American cars averaged 45 miles per gallon, it would take less fuel than it does now to move a car from point X to point Y -- but the total amount of driving would rise, and so would the amount of gasoline consumed.

The analogy is far from accurate. It is true the software has become increasingly wasteful of computing resources (memory, hard disk space). More is demanded from today's computers because today's computers can do more than PC's fifteen years ago. In 1990, a Pentium IV would have been the equivalent of a supercomputer. Then, most Americans had never even heard of the Internet. In the personal computer's history, its uses have increased from those of a specialized office device to a game playing console to an essential communication device. The same cannot be said of cars, whose basic tasks are still the same as they were when the automobile was first invented over a hundred years ago: to get people from one place to another.

From that claim, he continues:

This ''paradox of efficiency" is as true of cars and computers as of light bulbs, jet turbines, and air conditioners, Huber and Mills write. ''The more efficient they grew, the more of them we built, and the more we used them -- and the more energy they consumed overall."

This argument is true, but only up to a point. I don't know very much about jet turbines, so let's take light bulbs and air conditioners as examples. I recently replaced some of my 60W and 100W incandescent light bulbs with 13W and 24W fluorescent bulbs, respectively. The newer, more efficient bulbs produce equivalent light levels to the old ones and last longer, but use less energy. But, I only require a limited amount of light in each room, and I only require the lights to be on when it's dark outside. So, in fact, I use less electrical energy on lighting than I did before. A similar argument can be made with regard to air conditioners. If an air conditioner is highly inefficient, making it very expensive to run, a consumer might limit the amount of time he runs it or keep it on a higher setting. But, once air conditioners have reached an efficiency level where they can be run at a comfortable setting, consumers will set them to that setting no matter how energy-efficient the internal mechanisms are. It is hightly unlikely that someone who buys a highly efficient air conditioner will make his house freezing cold and leave it on during the winter just because it's cheaper to run. My argument holds as long as the asymptotic limit on usefulness has been reached. For these examples, that limit is reached as long as each house already has enough light or is at a comfortable temperature. Technology, no matter its purpose, tends to become less expensive over time, as more units are produced and sold. This is true for both inefficient technology and efficient technology. If the asymptotic limit on usefulness has not been reached, there will be increased consumption whether or not efficiency is mandated. Increased energy efficiency tends to make the initial costs of producing and buying a consumer good higher, and, in a simplistic model, mandated efficiency would tend to reduce purchases of new consumer goods, not increase it.

And, finally, to the misuse of statistics:

In ''The Bottomless Well," a myth-busting new book on energy and how we use it, Peter Huber and Mark Mills acknowledge that this paradox -- ''the more efficient our technology, the more energy we consume" -- strikes many people as heretical. But the numbers bear it out. Thirty years ago, the energy cost of transportation was nine gallons per 100 vehicle miles. Today it is six gallons -- a 33 percent drop. Yet over the same period, the total amount of fuel consumed rose 56 percent -- from 115 billion gallons a year to more than 180 billion gallons.

The comparison is like the proverbial apples to oranges. The former is a normalized value: Gallons of gasoline per 100 vehicle-miles. The latter is an absolute value: Total amount of fuel consumed. In order to make sense of the numbers, I will assuming that the fuel value quoted is only for automobile fuel. According to the U.S. Bureau of Transportation Statistics' National Transportation Statistics for 2005, there were 138 million automobiles on American roads in 1975. In 2005, thirty years later, there were 236 million. During the period when fuel efficiency increased 33%, and comsumption increased 58%, there was a 71% increase in the number of automobiles. Taking Jacoby's numbers at face value, in 1975, each car used an average of 833 gallons of gasoline, and in 2005, that number dropped to 7361. And, that decrease in yearly gas consumption doesn't even take into account two more factors that would lead to an increase fuel consumption. Firstly, urban sprawl has been increasing since the 1970's, as populations from cities move to suburbs and exurbs. Because of the separation of housing and commercial zones, suburbanites tend to be much more dependent on cars for simple errands than city dwellers. Also, cities are still the centers of business. The number of daily commuters between cities and suburbs has increased, and so has the average daily commute time. More time spent waiting in traffic leads to more wasted fuel. The numbers show that increased fuel efficiency has not kept up with increased demand for automobiles and increased reliance on automobiles. But, they also show that, on average, even though cars are more efficient, individuals are not being increasingly wasteful to compensate.

The policy implication of Jacoby's (and Huber and Mills') argument is that we should stop worrying about fuel efficiency and let the free market run its course, because, ultimately, any measures to oppose it are counterproductive. Even if the argument held any water, there is still the problem of peak-oil. Without conservation and development of alternative fuels, the world supply of petroleum will eventually run out2. The prospect threatens our entire way of life. Petroleum is burnt as fuel, and is used as a raw material in plastics, and even agricultural fertilizer.

The market isn't driven by energy efficiency. It's driven by prices. As the past few months have begun to show, high oil prices tend to reduce consumption. If people do tend to use more energy when given a more efficient device, then, an alternative policy would be to encourage the use of efficient devices in addition to artificially raising the price of oil through taxation whenever the market price drops below a certain threshold.

Technorati tags: , ,

1 This analysis is incredibly simplistic, and undoubtedly hides a good number of inaccuracies.

2 The problem of peak-oil goes well beyond the petty question of whether there should be drilling in the Arctic National Wildlife Refuge. Even if the most optimistic scenarios play out, the ANWR will delay the onset of peak-oil by less than fifty years. As far as I know, there is no agreement as to when peak-oil will be reached, but most estimates place it within the 21st century.

Although I agree with your analysis, it feels like you are carefully sidestepping his point rather than refuting it. Perhaps he is saying that it is precisely because cars are more efficient today than in 1975 that we can afford sprawl and more cars, thus creating more consumption overall. That notwithstanding, the policy implication of his argument remains pretty strange to me.
I doubt it. Urban sprawl is primarily driven by the housing market, not the auto market.
Post a Comment

<< Home

Links to this post:

Create a Link