Debunking the Jevons Paradox: Nobody goes there anymore, its too crowded

The “Jevons paradox,” asserts that increasing “the efficiency with which a resource is used tends to increase (rather than decrease) the rate of consumption of that resource.” It is mostly if not entirely bunk, as the scientific literature and leading experts have demonstrated many times (see “Efficiency lives “” the rebound effect, not so much”).But it lingers on in part because it is one of those quirky, ill-defined contrarian notions that the media can’t get enough of and in part because those who oppose clean energy, often for bizarre ideological reasons, keep pushing it.So I’m reposting two debunkings written by Real Climate Economics expert Dr. Jim Barrett. As noted in the second post (whose Yogi Berra quote I repeated for my headline), “Though he discovered it nearly 100 years after Stanley Jevons, I believe [Berra’s] exploration of the Jevons effect is more complete and accurate than Jevons’ own, as well as being vastly shorter. The notion that we could get so efficient at using energy that we’d end up using more is about as valid as the idea that a restaurant could get so crowded that it was empty.

“Barrett “has 13 years of experience working in the nexus of climate change, energy efficiency and economics and has written extensively on the role of efficiency in achieving environmental and economic goals.” He was a senior economist on the Congressional Joint Economic Committee and is now Chief Economist at the Clean Economy Development Center.This post by Dr. Barrett originally appeared on the Great Energy Challenge blog, in partnership with National Geographic and Planet Forward. It is reposted from Real Climate Economics.

Rebounds Gone Wild

Energy efficiency has become very popular in recent years. So much so that it’s becoming cool for the truly hip to hold it in disdain.

Case in point: David Owen’s piece in this week’s New Yorker: The Efficiency Dilemma” (subscription required).

It reads like he’s being contrary just for the sake of being contrary. I don’t want to make a habit of highlighting this type of work, and to do a thorough job of dismantling the piece would take more time and space than I have. But it generated some genuinely interesting conversations in my email this week and I have a hard time letting such poor and frankly lazy reasoning pass without comment.


As a compromise, I’ll try to focus more on the serious issues in the article and less on the serious issues I have with the article itself. Wish me luck.

The focus of the article is something called the Jevons paradox (named after economist William Jevons), or the more common and more broadly defined “rebound effect.” In essence the rebound effect is the fact that as energy efficiency goes up, using energy consuming products becomes less expensive, which in turn leads us to consume more energy.

Jevons’ claim was that this rebound effect would be so large that increasing energy efficiency would not decrease energy use. The rebound effect would eat up all (or more than all) of the energy savings.

To be clear, the rebound effect is real. The theory behind it is sound: Lower the cost of anything and people will use more of it, including the cost of running energy consuming equipment. But as with many economic ideas that are sound theory (like the idea that you can raise government revenues by cutting tax rates), the trick is in knowing how far to take them in reality. (Cutting tax rates from 100% to 50% would certainly raise revenues. Cutting them from 50% to 0% would just as surely lower them.)

The problem with knowing how far to take things like this is that unlike real scientists who can run experiments in a controlled laboratory environment, economists usually have to rely on what we can observe in the real world. Unfortunately, the real world is complicated and trying to disentangle everything that’s going on is very difficult.

Owen cleverly avoids this problem by not trying to disentangle anything.

One supposed example of the Jevons paradox that he points to in the article is air conditioning. Citing a conversation with Stan Cox, author of Losing Our Cool, Owen notes that between 1993 and 2005, air conditioners in the U.S. increased in efficiency by 28%, but by 2005, homes with air conditioning increased their consumption of energy for their air conditioners by 37%.

Owens presents this as clear and obvious proof of a Jevons effect. Case closed.

Here is where Owen gets lazy: A few key facts disprove the point. Facts that are not hard to track down. I write for this blog in my spare time (for free), and I managed to find it without breaking a sweat. I’m not sure why a paid writer for a magazine like The New Yorker couldn’t do the same.

Consider the following:

Real (inflation adjusted) per capita income increased by just over 30% over that time period. All else being equal, when people have more money, they buy more stuff, including cool air.


The average size of new homes increased from 2,095 to 2,438 square feet, over 16%. More square feet means more area to cool and more energy needed to cool it.

In 1993, of homes that had A.C., 38% only had room units while 62% had central air. By 2005, 75% of air conditioned homes had central units. Bigger units covering more rooms means more cool air and, you guessed it, more energy.

(Real electricity prices were mostly flat over this time period, falling by just over 1%, contributing little, if anything, to the increase.)

Finally, even though air conditioners were 28% more efficient in 2005 than in 1993, air conditioners last between 15 and 25 years. Using the mid-range lifespan of 20 years, and assuming that efficiency increased gradually from 1993 to 2005, and accounting for the introduction of new AC units associated with new home construction (about 1.5% of the housing stock in any given year), I calculated the efficiency of the average central air unit in service in 2005 to be about 11.5% more efficient than the average unit in 2009.

Accounting only for the increased income over the timeframe and fixing Owen’s mistake of assuming that every air conditioner in service is new, a few rough calculations point to an increase in energy use for air conditioning of about 30% from 1993 to 2005, despite the gains in efficiency. Taking into account the larger size of new homes and the shift from room to central air units could easily account for the rest.

All of the increase in energy consumption for air conditioning is easily explained by factors completely unrelated to increases in energy efficiency. All of these things would have happened anyway. Without the increases in efficiency, energy consumption would have been much higher.


Worse, and even more transparently wrong, Owen points to the increasing use of air conditioning in the developing world, especially India and China, as evidence of a globally expanding Jevons effect. Never mind the fact that income in China is growing something like three times as fast as in the U.S. and that the cost of air conditioning as a share of average incomes are falling at an even greater rate.

It’s one of the well-established frustrations of the energy efficiency world that people pay too much attention to the up-front cost of goods and not enough to the cost of energy needed use them. More expensive highly-efficient products have a hard time competing. Suddenly, however, Owen wants us to believe that falling up-front prices and rising incomes couldn’t possibly explain the accelerated market penetration of air conditioners in China and that rising efficiency is the reason.

It’s easy to be sucked in by stories like the ones Owen tells. The rebound effect is real and it makes sense. Owen’s anecdotes reinforce that common sense. But it’s not enough to observe that energy use has gone up despite efficiency gains and conclude that the rebound effect makes efficiency efforts a waste of time, as Owen implies. As our per capita income increases, we’ll end up buying more of lots of things, maybe even energy. The question is how much higher would it have been otherwise.

The even more interesting question is whether efficiency growth can ever overpower the effect of income growth and start reducing energy consumption in absolute terms.

— James Barrett

Rebounds and Jevons: Nobody Goes There Anymore. It’s Too Crowded

This is the second post in a series on the rebound effect and energy efficiency by Real Climate Economics blogger James Barrett. It is reposted from Real Climate Economics.

My last post on David Owen’s piece in the New Yorker and on the Jevons effect stirred up some interesting questions and discussion that I want to follow up on here. My last one purposely avoided some of the more technical parts of the issue to keep it readable and under my word limit. I think I’m about to undo that.

But first we should pay thanks to the great 20th Century philosopher, Yogi Berra, from whom I shamelessly stole the title of this post. Though he discovered it nearly 100 years after Stanley Jevons, I believe his exploration of the Jevons effect is more complete and accurate than Jevons’ own, as well as being vastly shorter. The notion that we could get so efficient at using energy that we’d end up using more is about as valid as the idea that a restaurant could get so crowded that it was empty.

Dictating Terms

Though I hate having arguments about how we should argue, there are a few things we need to get straight:

First, as originally observed and defined by Jevons, the Jevons effect is a decidedly micro issue. He observed that increased energy efficiency in coal fired steam engines resulted in increased use of coal to fire steam engines as they were used in more applications and more intensively in existing ones.

Further, the central point of Jevons’ theory was that advances in energy efficiency forced increases in energy consumption. Not merely that consumption increased despite efficiency, but that efficiency caused the increase.

So, if you believe that energy consumption would have been higher without advances in energy efficiency then, by definition, you do not also believe in the Jevons effect.

Much of the debate around this, including the comments to my last post, seem to center around a weaker form of the Jevons effect, i.e. that energy efficiency can’t keep up with demand growth. This is a very different argument than the strict form of the Jevons effect.

The distinction is important, because more than once I have found myself arguing that the Jevons effect doesn’t widely exist, and demonstrating that with examples, only to have people say that Jevons doesn’t mean we shouldn’t invest in efficiency and that I’m missing the macro question, neither of which are consistent with the definition of the Jevons effect, at least in the strict form.

I believe it is this second argument, over the weaker form of Jevons that most people are really arguing about, which is really about the size of the rebound effect.

With this in mind, let’s take a look at the different potential sources of rebound.

Home Economics 101

Looking first at household energy use (which is very different from using energy as a factor of production), I believe there is very little evidence for any Jevons effect.

Increased energy efficiency can often be treated as a decrease in energy prices, which can be broken down into two separate effects. The substitution effect and the income effect. When the price of any good falls, it frees up some amount of money we would have spent on it, essentially increasing disposable income (thus the name “income effect”). If consumption patterns don’t change, we would expect people to buy more of everything more or less proportionately.

Of course, falling energy prices do impact peoples’ consumption patterns. When energy prices fall, we tend to buy more of it because it is more affordable. This is the substitution effect. Because energy has become cheaper relative to other goods, we buy more energy and less of some other things, substituting one for the other.

For households, both these effects tend to be small. The substitution effect is small because energy is already so cheap that it tends not to influence our decisions to watch TV, use our computers, or put lights on. Most people have no idea how much it costs to run a TV (about 5 cents an hour), so owning a more energy efficient TV will have little impact on how much TV they watch.

The income effect also tends to be fairly small because (as you can see in this nifty graphic) the average household spends less than 12% of its income on gasoline and utilities (which here includes water, garbage etc as well as electricity). For the average household, if everything suddenly became 10% more efficient, energy would fall to 10.8% of their income, and their disposable income would go up by 1.2%. Energy consumption would rebound to about 10.93% (10.8% x 101.2%), producing an 11% rebound, far below the 100% needed for Jevons.

The places where you should expect the combined substitution and income effects to be large are where energy consumption is more of a luxury than a necessity. Using an air conditioner or heating our homes above a certain minimum level are good examples because when energy is expensive or the budget is tight, this is where people will economize. But, as I showed about air conditioning in my previous post, and Matt Kahn showed about driving in his, even for non-necessities, the Jevons effect is hard to find.

And admit it, while high gasoline prices might make you cancel the family road trip to the Grand Canyon this summer, nothing, not even free gasoline, could make you do it twice.

Industrious Efficiency

For industry, the income effect is the fact that with the price of energy (more appropriately “energy services”) falling, they become more profitable on a per unit basis. This increase in profitability should lead to an increase in production levels, though how large is hard to know. If we make the conservatively high assumption that businesses plow all of the increased profits back into making product, we can do the same type of calculation as we did for households. In 2006 (the most recent data available), energy consumption made up just over 3% of total input costs in the manufacturing sector (labor was about 21%). So, by the same calculation as above, the income effect would create a very small rebound effect: a 10% increase in operating efficiency in the manufacturing sector would result in a 9.7% reduction in energy use, a 3% rebound.

The substitution effect is even smaller in industry. Though we can often think of increases in efficiency as reductions in energy prices, that’s really just a form of intellectual shorthand. Increases in industrial end-use efficiency typically result from things like investments in more efficient equipment or higher expenditures on labor for operations and maintenance. These represent substitutions of physical capital and labor for energy in the industry’s production process. Because increasing expenditures on labor and machinery are what cause the cost savings in energy, firms can’t take advantage of this efficiency by shifting expenses back to energy. That would undo the savings they created in the first place.

And all of this is supported, if not proved, by actual data. Again, I can’t let Owen (remeber him? He started all this) off the hook for laziness. All of this is easily accessible from the DOE’s Manufacturing Energy Consumption Survey and BEA’s Input Output tables. I took a look at some of the heavier and more intensive industries, ones that would be most sensitive to energy prices and most likely to show large rebounds or even a Jevons effect. Between 1998 and 2006 (the periodicity of some of the data is a little odd), here is what I found:

This shouldn’t be surprising to anyone: Energy efficiency leads to reduced energy use, with some rebound. The pattern is consistent across the manufacturing sector. Despite a 26% increase in GDP and a 7% increase in manufacturing output over that time period, both energy intensity and energy use fell for the sector as a whole and for almost all of the sub-sectors that I looked at. Try it for yourself.

So you’ll have to pardon my incredulity when I hear people like Owen claim that Jevons effects are everywhere, because everywhere I look, I can’t find them.

Some of the Whole and All of Its Parts

A final word about the macro question. There is nothing particularly magical about the macroeconomy, it is merely the sum of all the micro parts. If we can’t find a Jevons effect in all the individual places we look, it will continue to be absent if we sum them all up.

The one exception to this is the question of productivity. If energy efficiency increases overall productivity (it does), which in turn accelerates economic growth (it should), then there is a third source of rebound effect that won’t show up in the income or substitution effects I describe. A self-fulfilling income effect, perhaps.

This is tough to address in a few words (or even in many), because it links back to very technical questions of multifactor productivity and the like. But looking at past increases in gasoline prices, history shows that for every 1% increase in gasoline prices, GDP tends to decline about 0.05%, so that a doubling of gas prices might knock a half a point off of GDP. If we’re willing to assume that this is at least the right neighborhood for all energy types and that the response to a reduction in prices would be about as large as it is for an increase in prices (and there are good reasons to doubt this), then this productivity effect would still be pretty small. A 1% across the board increase in energy efficiency might produce an increase in GDP of 0.2% (not insignificant given historic annual growth rates in the 3% neighborhood). This, in turn, would lead to an increase in energy consumption of slightly less than the same relative magnitude, a rebound effect of less than 20%.

Putting all of these things together, the most you can reasonably expect is a total rebound effect of 30% under some pretty generous assumptions. Combine this with income growth caused by other things, population growth, and other factors, and the gains from efficiency can get buried in the weeds. All that’s really clear is that for significant periods, energy efficiency has not increased fast enough to cause energy use to go down.

But, and this is the key point, this is not the same as saying, and it does nothing to justify saying that efficiency can’t grow fast enough to reduce overall energy use. That is exactly as valid as saying that nuclear power can never reduce our use of fossil fuels, because even when we made large investments in nukes, fossil use still increased. I’ll have plenty more to say about nukes later, but nuclear advocates would rightly respond that this is proof only of the fact that our investments in nuclear were not large enough to offset increases in electricity demand. The same is true for efficiency. If efficiency investments have never forced energy use to decline, that’s proof only that we, as a nation, have never really given it a try.

So while there may be many good reasons to go to a crowded restaurant, being alone isn’t one of them.

— James Barrett

Related Post: