Wednesday, April 14, 2010

Carbon Prices are a Very Blunt Tool

I have been discussing the following assertion by Paul Krugman (speaking for a larger community of environmental economists):
Restricting emissions would slow economic growth — but not by much. The Congressional Budget Office, relying on a survey of models, has concluded that Waxman-Markey “would reduce the projected average annual rate of growth of gross domestic product between 2010 and 2050 by 0.03 to 0.09 percentage points.”
I pointed out on Monday that if you assume trend economic growth of 3%/year (since we are being told that the effect on growth is less than 0.1%, and thus probably less than the measurement error), and noting that Waxman-Markey has the goal of an 83% reduction in the absolute level of carbon emissions by 2050, this implies that the carbon efficiency of the economy has to improve by 7%/year between now and 2050 on average.

This morning, I made the graph above which shows US historical experience with the price of carbon and the rate at which carbon efficiency improves.

What I did was to take data from the BP Statistical Review for 1965-2008, which gave me US consumption of coal, oil, and natural gas. I (approximately) corrected to give the carbon content of each fuel*.  I then took prices of each fuel over time, and produced a weighted average price per tonne of carbon (where the weight of each fuel price was based on the amount of carbon input into the economy via that particular fuel type).  This allows us to get an approximate price that the rest of the economy paid to use fossil fuel carbon, and we can look at the trends of this over time, along with the levels of carbon emissions and GDP (here real chained GDP from the Bureau of Economic Analysis).

This allows to try to draw conclusions from the natural experiments that fluctuations in energy prices have created in recent decades.  (One could argue that this is a strange combination, since the price of carbon in the different fuels varies enormously, but it mimics the spirit of trying to reduce carbon emissions by a single carbon price).

To give some feeling for the data, here is the price of carbon (RHS), together with the quantity of carbon emissions and the level of GDP (with both the latter two on the LHS expressed as a ratio to the 1965 value).

As you can see, a massive -- roughly five-fold -- increase in the price of carbon in the 1970s did cause the two lines to diverge (there's considerable debate about the extent of the split between actual efficiency improvements and offshoring of the most carbon intensive parts of the production process).  A slightly smaller run-up in prices since 1998 has (so far) failed to have as much impact.

To get a better feel for this, we can look at the ratio of GDP produced per tonne of carbon emitted.  The year on year change in that is below, along with the price of carbon again on the right side:

This is the quantity that would have to improve at 7% a year for 40 years for Krugman's rough consensus to be correct.  I have shown that level as the red line - the actual data are in green. (Note, the red line is to be interpreted relative to the LHS efficiency gain scale, not the price scale). As you can see, we have never gotten particularly close to 7%, even when carbon prices increase by factors of four or five.  Also, once prices go back down, the levels of efficiency gain drop off again.

In particular, stare at the efficiency trend from 1998-20008. Efficiency gains are pretty much about flat at about 2 1/2% per year despite the huge price increase during that interval.

My conclusion is to get to 7% and stay there, energy prices would need to increase by a large but imprecisely known factor - probably at least another factor of 2 or 3 from 2008 levels, and then stay there instead of falling back down.

This is the rough equivalent of hitting the economy over the head with a hammer in the hope of bringing it to its senses.

I predict that, if carbon prices are used as the main tool to address carbon emissions, one of two things will happen.  Either
  • Carbon prices will go extremely high in order to bring the required efficiency gains, and this will bring about one or a series of recessions (and thus a massive political backlash)
  • So many fudges will be built into the scheme that carbon prices will not go high enough, and then the gains will fall far short of desired.

*Gruesome methodological detail: I took oil to be 85% carbon by weight, used the coal figures expressed in MTOE and corrected those by 85% also, and took 1 bcm of natural gas to be 725 tonnes of NG, then multiplied by 12/16 for carbon content.  My prices came from BP for oil (expressed in $2008 by them), from the EIA for coal (expressed in $2000 by them, but corrected to $2008 using BEA GDP deflators), and from the EIA for natural gas (corrected with GDP deflators for inflation to $2008) .  All of this is a bit rough and ready and absolute values could be off by O(10%), but the methodology is consistent over time so trend analysis should be fine.


Datamunger said...

Glad you mentioned the off-shoring of carbon emissions.

And we have to keep in mind that in addition to "fudges built into" the emissions regulation system, large-scale cheating become more likely as the price of carbon is artificially raised. Krugman should write a companion volumn for us, The Return of Prohibition Economics.

And additionally, as time goes on, one could expect the gains would be harder to achieve -- low hanging fruit & such.

Datamunger said...

All of which makes me root for peak oil and greedy Arabs.

kjmclark said...

I think you're missing three key points. First, the increase in energy prices was unexpected, second they weren't based on the fuels' carbon content, and third the increases were exogenous shocks, not a revenue-neutral tax policy.

For the first, people react very differently when faced with an unexpected price increase as compared to an expected one. When people don't expect change, they often wait to see what will happen next. When people can anticipate the change, they are more likely to take steps to prepare for it.

Next, my understanding is that the oil shocks of the 70s resulted in moving electricity production away from oil to coal, and small improvements in transportation efficiency. I haven't heard much about outsourcing energy-intensive production to other countries, probably since those other countries were also experiencing an oil price shock. Further, this EIA document ( suggests that coal prices peaked in 1978, fell quickly after that, and have fallen pretty much consistently since then. The BP price data doesn't go back before 1987. So the price data for the most carbon-intense fuel is fairly different from that of oil.

Finally, when oil prices increase, for the most part that looks like a foreign tax on oil consumers. The increase in oil revenue goes almost directly out of the country to producing nations. That's very different from what happens with a revenue-neutral tax. In that case, any revenue generated would be used for tax credits for energy efficiency or for other measures to reduce our GHG emissions. (Like funding a better electrical grid to distribute electricity from wind and solar generation.) So the money all stays in our economy instead of mostly heading overseas, as in the price history you're looking at.

Burk said...

Put me down for option 1. Carbon pricing may be blunt, but it works. The other options don't work. Thus, blunt or not, carbon pricing wins.

There is also the example of Europe and Japan, which have higher fuel prices than we do, partly from taxation, and still have high usage, though less waste than we have. How much price pressure would it take for us to get to European levels of carbon efficiency and how far would that get us to the overall goal? That might be a good future topic to discuss.

Thanks for your work!

Gary said...

I'm agreeing with kjmclark. More to the point might be to look at price/demand for coal vs. natural gas for electricity production, to get some idea how elastic the carbon trade-off could be. The low hanging fruit in the carbon game is coal - not liquid transportation fuels.

Stuart Staniford said...


Responding to your points in turn:

1) Expected vs Unexpected. Who knows, right? We have no historical experience of planned price increases producing large efficiency gains, and indeed we have no experience of efficiency gains this large from any cause. Personally, I would speculate that given the extent of the average citizens information level, it's still going to come as a surprise to them if congress passes a law.

2) Agreed the coal/oil price trends are quite different. The curve in coal usage shows very little response to the oil price shocks (the different fuels actually show surprisingly little substitutability in the data) - I'll try to post it later (I made those graphs but didn't want to weigh the piece down with them). I used EIA price data to get back to 1965.

3) Your point about the shift in revenues overseas is a good one, and I agree that would tend to mitigate the effect on the economy (though some green investment would likely leak overseas too, absent trade barriers - China could well become the main supplier of wind/solar equipment for example).

Datamunger said...

I'm not sure I'd be so quick to endorse kmj's point #3 as it could run squarely into Jevons' Paradox.

"revenue generated would be used for tax credits for energy efficiency or for other measures to reduce our GHG emissions. "

Note first that efficiency doesn't translate into lower emissions. eg. A building boom of carbon-neutral palaces (juiced by tax credits) could actually raise emissions, construction being very carbon intensive.

In fact efficiency is a total red herring unless it's kept in mind that it's pointless emissions-wise unless there is a cap (lowered annually) on total emissions including those off-shored. Tax money would have to go not just to grid upgrades but mostly to pulling coal plants off-line and paying miners to close their operations.

There is room for endless delusion on this issue.

kjmclark said...

Um, Datamunger, if the construction were carbon-intensive, then it would run smack into the wall of more expensive carbon, wouldn't it? And if we weren't meeting our target with a given level of carbon tax, if we had any sense whatsoever, the first change would indeed be to raise the tax level. Notice that if we choose the cap and trade route, all of that is neatly handled if we set the caps to achieve our goals.

Jevon's Paradox has always struck me as a bit of a misunderstanding in the peak oil world. It assumes that the supply can increase, so that if price increases induce efficiency improvements, we can now do more with the same amount and we'll be encouraged to use more. But if the supply isn't increasing, or if the supply in this case is the carbon on which we've set limits, then the supply is actually decreasing, and Jevon's Paradox doesn't apply. The costs are constantly rising in that case, so next year, when you would be taking advantage of your new efficiency, you can't afford as much, so you're back to square one.

Remember that Jevons was writing about coal in Britain in the 1860s. There were no real constraints to supply except price. If we're artificially constraining supply, then there are certainly other constraints, and prices will be continually rising due to those other constraints.

Datamunger said...


if the construction were carbon-intensive, then it would run smack into the wall of more expensive carbon, wouldn't it?

Agreed, in a strict cap and trade scheme. But no, if there are generous tax credits being thrown around funded by hefty carbon taxes. This is because a system of carbon taxes and credits merely favours certain 'efficient' and desirable types of economic activity over others but doesn't put a bound on the quantity of the favoured activity or even on the supply of carbon used by it.

Tax the SUV and its fuel heavily and feed the money into a rebate on subcompacts and you may find total emissions from the little cars and the industrial complex building them them doesn't keep falling and may even rise despite the price of fuel, especially if we consider the world as a whole.

Datamunger said...

you may find total emissions from the little cars and the industrial complex building them them doesn't keep falling

That should have read, "total emissions from autos and..."

sofistek said...

Just a thought ... how accurate can any analysis be if the data that it is built upon is not accurate?

I ask because I've read a lot about how GDP measurements have changed over the years, including changes that could be though of as subjective, which may distort the true picture of carbon efficiency. If the real GDP figure is much lower than that which governments release, then carbon intensity is much greater (assuming the level of carbon emissions is known with some accuracy).

If we don't have the correct data, we can't tell the real story.

Mike Aucott said...

Most of the comments above make good points, but in my view the case presented by Stuart remains convincing - a carbon price much higher than Krugman indicates will be necessary to reduce emissions. What to do?
At some point, a steadily rising carbon tax - with the money collected rebated to the public - would have to work, no? And, there's a case to be made for good old-fashioned command and control. Make it illegal to produce electricity by burning coal, for example. In my view, rules are the only thing that worked for automobiles; mpg increased dramatically as the CAFE standards were implemented in the late 70s, and mpg will increase again with the new rules. Perhaps once America and the rest of the industrialized world works through the current spate of climate contrarian nonsense, it will be ready to impose some serious carbon emission rules.