One engages the economic arguments of a Nobel Prize winning economist with some caution, but in this instance, I think it is safe to say that Paul Krugman’s assessment of the costs of cap and trade is overly certain, and probably wrong. Krugman argues correctly (unlike many cap and trade supporters) that cap and trade imposes costs on consumers, but he believes that these costs will be small based on several economic analyses:

A cap-and-trade system would raise the price of anything that, directly or indirectly, leads to the burning of fossil fuels. Electricity, in particular, would become more expensive, since so much generation takes place in coal-fired plants.

Electric utilities could reduce their need to purchase permits by limiting their emissions of carbon dioxide — and the whole point of cap-and-trade is, of course, to give them an incentive to do just that. But the steps they would take to limit emissions, such as shifting to other energy sources or capturing and sequestering much of the carbon dioxide they emit, would without question raise their costs.

If emission permits were auctioned off — as they should be — the revenue thus raised could be used to give consumers rebates or reduce other taxes, partially offsetting the higher prices. But the offset wouldn’t be complete. Consumers would end up poorer than they would have been without a climate-change policy.

But how much poorer? Not much, say careful researchers, like those at the Environmental Protection Agency or the Emissions Prediction and Policy Analysis Group at the Massachusetts Institute of Technology. Even with stringent limits, says the M.I.T. group, Americans would consume only 2 percent less in 2050 than they would have in the absence of emission limits. That would still leave room for a large rise in the standard of living, shaving only one-twentieth of a percentage point off the average annual growth rate.

You can take a look at the most recent MIT study here (in PDF). What I’d like to focus on is the implicit assumption of spontaneous decarbonization in the report, which can be inferred from the Appendix provided by the authors (here in XLS).

The report assumes for its reference scenario — which is what “business as usual” is called, i.e., in the absence of any climate policies — that GDP growth will average 2.64% from 2010 to 2050. At the same time it assumes that carbon dioxide emissions will increase by only 1.185% per year 2010 to 2050. The difference of about 1.5% per year must be made up by improvements in energy efficiency and expansion of carbon-free energy supplies, which occurs in the absence of cap and trade policies. It would be fair to ask how this background rate would be achieved. (How, I ask?)

Without the assumption of spontaneous decarbonization the challenge of meeting emissions reductions targets would be much larger, about 100% larger in terms of emissions reductions from the reference scenario baseline (in 2050). What would a different assumption about spontaneous decarbonization do to the cost estimates? If the background rate of spontaneous decarbonization is not met, then it will necessarily lead to increased costs.

One simple way to get a ballpark estimate of possible increased costs is to multiply the additional cumulative emissions reductions needed times a marginal rate of those reduction. For instance, if the marginal costs of emissions reductions are, say $200 dollars per tonne of carbon dioxide, then this would add as much as $24 trillion dollars to the cost of emissions reductions. If the marginal cost of reductions are instead $500 dollars per tonne of carbon dioxide, then this would add as much as $59 trillion to the cost of emissions reductions presented in the MIT study. You can do your own math by knowing that assumptions of spontaneous decarbonization in the MIT study eliminate 118 gigatonnes of carbon dioxide emissions 2010 to 2050 in the reference scenario. And if you really want to explore scenario space you can vary assumptions of future GDP growth. For instance, if GDP growth 2010 to 2050 averages 2.8% per year rather than 2.64% then this adds another 16 gigatonnes of carbon dioxide emissions on top of the 118 mentioned above, and so on.

If you are getting the impression that cost estimates of cap and trade are really sensitive to assumptions, then you are correct. The costs could be presented as being small with one set of assumptions or large with another. Krugman should know this.

And we haven’t even gotten into issues of technological scalability and substitutability, much less the politics.

For further reading on spontaneous decarbonization:

Pielke, Jr., R. A., Wigley, T., and Green, C., 2008. Dangerous assumptions. Nature, Vol. 452, No. 3, pp. 531-532 (PDF)