The Price Elasticity of the Demand for Oil

Kevin Drum, Megan McArdle, Jim Manzi and Stuart Staniford are all worried by an IMF report that has very low price elasticities of oil such that “a 10 percent permanent increase in oil prices reduces oil demand by about 0.7 percent after 20 years.” Three quick notes.

First, do note that the IMF estimates are below others in the literature which estimate an elasticity of 0.2 to 0.3, meaning that a 10% increase in price would reduce demand by 2 to 3 percent, still small but three times the IMF estimates. Moreover, the US estimates tend to be higher still in the range of 0.4-0.5. All of the estimates are certainly low so we are not going to solve the climate change problem overnight with a tax on oil.  I’m not sure where the surprise is, however. Oil is necessary for civilization–given today’s technology–so people aren’t going to give it up easily.

Second, as Ryan Avent notes, a smaller elasticity makes a better case for taxing oil and reducing labor taxes. Moreover why the focus on oil?  What matters for climate change is total carbon and there are likely to be many carbon users with high elasticities relative to oil, which has special properties.

Third, this all depends on substitute technologies. In the past, there were few good substitutes for oil. If there are more good substitutes in the future then the elasticities will get larger. We do know that as the price of oil increases so does energy innovation (as measured by the number of energy patents).


If the demand is inelastic, then wouldn't a very small decrease in demand mean a large decrease in the price of oil? I'm imagining an almost vertical line in the demand/price slope.

You could have a situation where an alternative technology is cheaper than oil at first, but then isn't, after demand for oil declines only a little bit.

See late 2008.

It would depend on the elasticity of supply. A shift of the demand curve and a movement along the supply curve.

Yes, but a carbon tax and cap-and-trade were never expected to reduce oil consumption significantly. The carbon reductions are projected* to come largely from the electricity sector, by reducing coal consumption in favor of natural gas and renewables (and maybe nuclear).

On the other hand, there might be reasons to reduce oil consumption other than climate change. For example, national security issues, reduced macroeconomic vulnerability/volatility, etc.

*Models from places such as DOE, think tanks (e.g. RFF), universities (e.g. MIT, Stanford's EMF), etc.

Right, elasticity implies a linear relationship. My guess is that the demand curve is highly concave, because there are very high fixed costs to switching over infrastructure away from oil. For small changes in the price of oil we're not going to switch to hydrogen or synthetic gasoline. If oil's at $250 barrel these might look pretty attractive...

I thought the economics of synthetic gasoline (Fischer Tropsch) makes it competitive above $60 / bbl crude?

You might be remembering that value from much earlier when oil was also cheaper. The key number to check when energy prices are rapidly changing is Energy Return on Energy Invested.
That number is low for the synthetic oil processes.

Well, point elasticities don't necessarily "imply" any particular relationship. But generally people assume constant elasticity, which implies a convex, not linear, relationship.

This actually strengthens your case. If you're right about the concavity, then the low point elasticities could underestimate the potential for large shifts away from oil and certain (higher) price points, whereas the convexity assumption implies quite the opposite.

People and companies already have an incentive to economize on the use of fossil fuels- it isn't like these things were free without the taxes.

There are five important issues related to high oil prices due to increases in demand. First, what may happen to the fundamental forces driving the increase in demand. Second, how the price elasticity of demand changes with time and with the price of oil. Third, how the price elasticity of supply changes with time and with the price of oil. Fourth, how the prospect of high oil prices affects econ growth. Fifth, how the prospect of high oil prices and its effect on econ growth conditions the proposed taxation of consumption.
Tyler, hope you can draw tentative answers to those questions from the papers you have reviewed. My impression is that the experience of 1973-1986 is not relevant. Thanks.

I think you've misread the elasticity estimates in the James Hamilton paper you link to. The 0.2 to 0.3 elasticity value is for the short-term. Long-term values are around 0.6 to 0.9, 10 times more than the value the IMF got. (See table 3, pg 35) It looks like the IMF has done something wrong with their estimate.

I hate to criticize Avent on this blog, but just because oil has low price elasticity does not mean that we should tax it heavily. Sure, it means this is less distortionary than taxing a higher elasticity good, but we should also worry about the tax incidence: it's difficult to make a progressive tax on oil.
More importantly, if you're talking about replacing another tax, shouldn't you phase out something *more* elastic? Labor supply elasticity is usually estimated somewhere between 0.1 and 0.6, depending on whether extensive margin is properly accounted.

My prior for all price elasticities is -1. The long run estimates Talosaga cites must therefore be correct.

More seriously, the urban poor don't drive. They will be least affected. The urban rich can pay, if they wish. Welcome to the demanding, politically troublesome middle class.

Go nuclear!


Let me see if I get this right: Economists justify taxes on oil cause of externalities that would be corrected by the right tax. But if elasticity is low, those externalities due to mispricing are low so a tax won't help much which means an externality tax is not really justified (or won't matter much which equals low changeable externalities). So economists justify the taxes anyway cause it leads to more revenue?

This sounds like a giant rationalization for more gas taxes without regard to the rationale or nature of impact.

I'm with georgi on this one. I don't really understand the economic argument implied here. As an economics grad student studying environmental economics, I know that the idea behind a carbon tax is to internalize externalities caused by damages from CO2. While I realize the damage function isn't well-defined (given uncertainty), I don't see how one can say it "doesn't lower gas consumption enough" if your goal is to internalize damage costs. So long as the tax is set at an appropriate level given the expected damages, it is lowering gas consumption "enough." If you're ultimate goal is to raise revenue or pursue an environmental ideal, then that's another story.

How dumb can you be? The people over at HuffingtonPost (author and leftwing comenters) believe you can earn a risk-free return borrowing for 4-weeks from the Fed at 0.25%, buying 10-year Treasuries yielding 3.64% and selling the 10-year Treasuries four weeks later to repay the fed. How financially illiterate can you be?

The headline says: Taxpayer Arbitrage - Report Details Big Banks' Low Rate Money Machine

If Megan McArdle is worried about something, it is not a problem, and vice versa. You don't really need to do much more work than that.

People and companies already have an incentive to economize on the use of fossil fuels- it isn’t like these things were free without the taxes.

I I tried to make up an imaginary (yet realistic) fuel that was better than oil. Short of a cold fusion in a AA battery, it's tough.

Comments for this post are closed