Campaign News

If there’s one thing everybody knows about carbon pricing, it’s that there’s not much effect on consumption of transportation fuels. In econ-speak, the explanation is that demand is very inelastic: a price increase of (say) 10 percent reduces consumption by much less than 10 percent. In plain English, the explanation is that driving is something that people “have to do”, so price changes don’t have much impact on how much driving people do. 

But it’s looking more and more likely that what everybody knows is wrong. Carbon pricing—and especially carbon taxes—are likely to be much more effective than we thought.

A look at the models

Look at just about any economic model out there and you can see clear evidence of what everybody knows. (The graphs below tell the story, so feel free to just scroll quickly through them if you’re getting weighed down by the econ-speak in the text.)

For a national example, consider the NEMS model used by the Energy Information Administration. The figure below shows the impact on motor gasoline consumption of a carbon tax that starts at $25 per ton CO2 in 2015 and rises at 5 percent per year through 2040. Baseline emissions (in blue) are expected to decline by a fairly impressive 25 percent between 2015 and 2040, presumably because of some combination of less driving and more efficient vehicles (driven in part by tough new fuel economy standards). But the extra impact from adding in a carbon tax (in purple) is essentially zero.

Source: EIA Annual Energy Outlook 2014. Their carbon tax starts at $25 per ton CO₂ in 2015 and rises at 5 percent annually.

 

Look in particular at the early years—before there’s much turnover in the vehicle fleet—to see how the model treats driving as something that people “have to do”. In 2015, for example, a $25 carbon tax is projected to reduce consumption from 8.14565 million barrels per day to 8.117913, a reduction of only 0.34 percent. (For those of you keeping score at home, that means the implied short-run elasticity here is in the ballpark of -0.034: a $25 carbon tax is about 25 cents a gallon, which is about 10 percent of the retail price of gasoline, so if a 10 percent price increase reduces consumption by 0.34 percent then a 1 percent price increase—which is the hypothetical specified in the definition of elasticity—should reduce consumption by something like 0.034 percent.)

The most popular model for Washington State—called CTAM, a model that’s popular because it’s tractable, and tractable because it’s simple —shows a similar result because it’s based on similar assumptions about inelastic demand. In fact, the treatment of motor gasoline in CTAM involves just one number: an estimate for the long-run price elasticity of demand for gasoline. Following what seems like solid reasoning, CTAM gets that number (-0.62, in case you’re curious) from a weighted average of long-run price elasticity estimates from the economics literature; and, since these are long-run estimates, CTAM phases in the elasticity over ten years: -0.062 in year 1, -0.124 in year 2, etc.

Because CTAM is so tractable, I was able to modify it to model the estimated impacts of the BC carbon tax. (Note that CTAM is under revision; the version I used was I think this one.) The blue line below shows actual per-capita petroleum consumption for Canada as a whole between 2007 and 2012 (the former being the year before the BC carbon tax started, the latter being the latest year for which data are available). I then used this as a baseline for BC—in other words, I assumed that per-capita consumption in BC would have tracked Canada as a whole if BC had not implemented its carbon tax—and then I used the CTAM elasticities to model the impacts of the carbon tax. The result is the purple line below. (It diverges slowly from the blue line because the BC carbon tax increased from $10 in 2008 to $30 in 2012 and because CTAM phases in the elasticity over ten years.)

Source: CANSIM 134-0004, 051-0001, calculations based on CTAM.

A look at reality

As I wrote last year, here’s what actually happened to petroleum consumption in BC. (See below for an update.)

Source: CANSIM 134-0004, 051-0001, calculations based on CTAM.

 

The economic analysis comes from two Canadian academics, Nicholas Rivers and Brandon Schaufele, in a 2013 working paper:

Our main result is that the BC carbon tax generated demand response that is 7.1 times larger than is attributable to an equivalent change in the carbon tax-exclusive price… We find that the BC policy reduced carbon dioxide emissions from gasoline consumption by more than 3.5 million tonnes. Of this total, 85.6%, or 3.1 million tonnes, is due to the additional saliency of the carbon tax –- i.e., it is an environmental bonus that would not have been achieved if individuals responded to carbon taxes in the same way as to identical changes in gasoline prices caused by other factors. [Italics added.]

The idea of salience is a new one (at least to me!), and it comes from the intersection of environmental economics and behavioral economics. As Rivers and Schaufele put it, “Tax saliency refers to the hypothesis that tax-induced price changes generate distinct demand responses when compared with equivalent market-determined price movements.” In other words: Carbon taxes are better than you think.

I tend to be a pretty mainstream neoclassical economist, so this salience idea blew my mind when I first encountered it. But now I get it, partly because of an explanation that involves media amplification: gas prices go up or down by ten or twenty cents on a regular basis and so is no big deal, but when the BC carbon tax created a similar price change it became the central issue in the 2009 provincial election campaign. (An explanation that runs parallel to salience is persistence: increases in market price are reversed all the time, but increases in gas taxes tend to persist.)

It’s important to note that Rivers and Schaufele aren’t alone:

Our results are in line with Li, Linn and Muehlegger (2012) who find that consumers are more responsive to changes in gasoline excise taxes than to tax-exclusive prices. In particular, Li, Linn and Muehlegger estimate a tax saliency ratio (i.e., the mean consumer response to an increase in gasoline taxes divided by an equivalent increase in market prices) equal to 8.1, a value that is within the range of our estimates. Although in a different context, our results also complement Finkelstein (2009) and Chetty, Looney and Kroft (2009) who suggest that consumers exhibit more elastic demand response if prices increase due to a highly visible tax than if prices increase for some other reason.

Let me wonk out for a minute: “Finkelstein” is Amy Finkelstein, and “Chetty” is Raj Chetty, and they are (respectively) the winners of the 2012 and 2013 John Bates Clark medal, an annual award to the best American economist under the age of 40 that is highly correlated with future winners of the Nobel prize in economics. These are very high-powered economists.

Li, Linn, and Muehlegger are less well known, but their paper (published in November in American Economic Journal: Economic Policy) is amazing. They study five decades of changes in gasoline prices and taxes across the fifty US states, and here’s their conclusion:

We find that rising gasoline taxes are associated with much larger reductions in gasoline consumption than comparable increases in gasoline prices… suggest[ing] that a $0.05 increase in the gasoline tax would reduce gasoline consumption by 0.86 percent, three times as large as the effect one would find from an empirical framework that does not separate gasoline prices into tax and tax-exclusive components.

[O]ur work indicates that fuel taxes may be a more effective measure for reducing gasoline consumption or inducing consumers to adopt more fuel efficient vehicles than previously thought.

In other words: Carbon taxes are better than you think.

A look ahead

It’s important not to get too carried away with all this.

For one thing, transportation fuels are really not all that important in the big picture. That’s because of global warming’s global context, which I described elsewhere:

Transportation accounts for about 1/2 of Washington State greenhouse gas emissions, about 1/4 of national greenhouse gas emissions and about 1/8 of global greenhouse gas emissions. The global climate challenge is mostly about coal and electricity generation, not about petroleum or transportation fuels.

For another thing, this is a new line of economic research, so there may be surprises ahead. This is especially true with respect to BC’s carbon tax, which has only been in effect since 2008. There are many questions about, for example, what we might call the persistence of salience: will the carbon tax continue to have a disproportionate effect in the future? (One could even ask the flip side of this question, which is whether just having a public debate about carbon pricing will influence demand even if carbon pricing doesn’t actually happen. Ah, dissertation topics galore!) For a sense of the importance of these issues, consider this update of the slide above that includes 2013, the latest year for which data is available:

Source: CANSIM 134-0004, 051-0001, calculations based on CTAM.

Finally, it’s worth remembering that transportation fuels are still going to be a tough nut to crack. Demand for motor gasoline may not be as inelastic as everybody thinks, but it’s still no Slinky. Perhaps a BC-style $30 carbon tax will reduce transportation emissions by 10 percent rather than 1 percent. That’s still pretty modest, and the major emissions reductions from carbon pricing are still likely to come from other sectors, notably electricity generation. Happily, there’s good news in those sectors, too. I’ll address that in my next post about how carbon taxes are better than you think.

One last point: The salience and persistence arguments may depend on the presence of open and honest public debate about and acknowledgement of the fact that carbon pricing is going to raise gasoline prices. This is potentially relevant in the debate between carbon taxes and cap-and-trade systems: the former approach tends to be open and transparent while the latter approach tends to be more opaque. (At a Seattle Chamber luncheon panel I went to last year, a speaker from California refused to admit that cap-and-trade would raise gas prices. The same dynamic is in evident in this KUOW story from earlier this month, which quotes state Rep Joe Fitzgibbon saying that “I don’t think it’s at all a foregone conclusion that the energy companies will choose to increase their prices.”) There are more questions than answers here, but it seems plausible that honesty really is the best policy.

Update May 8 2015: See also this Yahoo news article: “A new preliminary study by Matthew Lewis, a professor of Economics at Clemson University, and three other researchers suggests a 12 percent swing in pump prices would raise or lower gasoline demand by 1 percent, roughly five times higher than the EIA formula.”