Response to Conor Clarke, Part I

Last week Conor Clarke at The Atlantic blog , apparently as part of a running argument with Jim Manzi, raised four substantive issues with my study, “What to Do About Climate Change,” that Cato published last year. Mr. Clarke deserves a response, and I apologize for not getting to this sooner. Today, I’ll address the first part of his first comment. I’ll address the rest of his comments over the next few days.

Conor Clarke: 

(1) Goklany’s analysis does not extend beyond the 21st century. This is a problem for two reasons. First, climate change has no plans to close shop in 2100. Even if you believe GDP will be higher in 2100 with unfettered global warming than without, it’s not obvious that GDP would be higher in the year 2200 or 2300 or 3758. (This depends crucially on the rate of technological progress, and as Goklany’s paper acknowledges, that’s difficult to model.) Second, the possibility of “catastrophic” climate change events – those with low probability but extremely high cost – becomes real after 2100.

Response:  First, I wouldn’t put too much stock in analyses purporting to extend out to the end of the 21st century, let alone beyond that, for numerous reasons, some of which are laid out on pp. 2-3 of the Cato study. As noted there, according to a paper commissioned for the Stern Review, “changes in socioeconomic systems cannot be projected semi-realistically for more than 5–10 years at a time.”

Second, regarding Mr. Clarke’s statement that, “Even if you believe GDP will be higher in 2100 with unfettered global warming than without, it’s not obvious that GDP would be higher in the year 2200 or 2300 or 3758,” I should note that the conclusion that net welfare for 2100 (measured by net GDP per capita) is not based on a belief.  It follows inexorably from Stern’s own analysis.

Third, despite my skepticism of long term estimates, I have, for the sake of argument, extended the calculation to 2200. See here. Once again, I used the Stern Review’s estimates, not because I think they are particularly credible (see below), but for the sake of argument. Specifically, I assumed that losses in welfare due to climate change under the IPCC’s warmest scenario would, per the Stern Review’s 95th percentile estimate, be equivalent to 35.2 percent of GDP in 2200. [Recall that Stern’s estimates account for losses due to market impacts, non-market (i.e., environmental and public health) impacts and the risk of catastrophe, so one can’t argue that only market impacts were considered.]

The results, summarized in the following figure, indicate that even if one uses the Stern Review’s inflated impact estimates under the warmest IPCC scenario, net GDP in 2200 ought to be higher in the warmest world than in cooler worlds for both developing and industrialized countries.


Source: Indur M. Goklany, “Discounting the Future,” Regulation 32: 36-40 (Spring 2009).

The costs of climate change used to develop the above figure are, most likely, overestimated because they do not properly account for increases in future adaptive capacity consistent with the level of net economic development resulting from Stern’s own estimates (as shown in the above figure).  This figure shows that even after accounting for losses in GDP per capita due to climate change – and inflating these losses – net GDP per capita in 2200 would be between 16 and 85 times higher in 2200 that it was in the baseline year (1990).  No less important, Stern’s estimate of the costs of climate change neglect secular technological change that ought to occur during the 210-year period extending from the base year (1990) to 2200. In fact, as shown here, empirical data show that for most environmental indicators that have a critical effect on human well-being, technology has, over decades-long time frames reduced impacts by one or more orders of magnitude.

As a gedanken experiment, compare technology (and civilization’s adaptive capacity) in 1799 versus 2009. How credible would a projection for 2009 have been if it didn’t account for technological change from 1799 to 2009?

I should note that some people tend to dismiss the above estimates of GDP on the grounds that it is unlikely that economic development, particularly in today’s developing countries, will be as high as indicated in the figure.  My response to this is that they are based on the very assumptions that drive the IPCC and the Stern Review’s emissions and climate change scenarios. So if one disbelieves the above GDP estimates, then one should also disbelieve the IPCC and the Stern Review’s projection for the future.

Fourth, even if analysis that appropriately accounted for increases in adaptive capacity had shown that in 2200 people would be worse off in the richest-but-warmest world than in cooler worlds, I wouldn’t get too excited just yet. Even assuming a 100-year lag time between the initiation of emission reductions and a reduction in global temperature because of a combination of the inertia of the climate system and the turnover time for the energy infrastructure, we don’t need to do anything drastic till after 2100 (=2200 minus 100 years), unless monitoring shows before then that matters are actually becoming worse (as opposing merely changing), in which case we should certainly mobilize our responses. [Note that change doesn’t necessarily equate to worsening. One has to show that a change would be for the worse.  Unfortunately, much of the climate change literature skips this crucial step.]

In fact, waiting-and-preparing-while-we-watch (AKA watch-and-wait) makes most sense, just as it does for many problems (e.g., some cancers) where the cost of action is currently high relative to its benefit, benefits are uncertain, and technological change could relatively rapidly improve the cost-benefit ratio of controls. Within the next few decades, we should have a much better understanding of climate change and its impacts, and the cost of controls ought to decline in the future, particularly if we invest in research and development for mitigation.  In the meantime we should spend our resources on solving today’s first order problems – and climate change simply doesn’t make that list, as shown by the only exercises that have ever bothered to compare the importance of climate change relative to other global problems.  See here and here.  As is shown in the Cato paper (and elsewhere), this also ought to reduce vulnerability and increase resiliency to climate change.

In the next installment, I’ll address the second point in Mr. Clarke’s first point, namely, the fear that “the possibility of ‘catastrophic’ climate change events – those with low probability but extremely high cost – becomes real after 2100.”