Nordhaus, Tol, and Climate-Change Economics: Turning Around the Conventional Wisdom
“The scientific modeling of climate change, and its possible impacts on human welfare, are very technical…. When experts try to summarize the fields for the layperson, they sometimes present matters in misleading ways, however inadvertent. William Nordhaus’s treatment of the economics literature, and RealClimate’s discussion of the accuracy of climate models’ temperature predictions, are good examples.”
At the Institute for Energy Research (IER) blog, I rebutted Yale economist William Nordhaus’s New York Review of Books criticism of a Wall Street Journal editorial by 16 “global warming skeptic” scientists, including MIT’s Richard Lindzen.
To understand the full point-counterpoint, the interested reader should consult the above links chronologically. Elsewhere I have challenged the entire case for a carbon tax. However, in the present post I want to focus on just two issues in the overall debate, that were raised in the wake of my initial IER post:
(1) the timing and amount of net damages from climate change, according to the best economic models, and
(2) what the climate scientists mean when they talk of a “confidence interval” in temperature projections.
The casual reader in these areas may be very surprised at what I have to report.
Richard Tol: Climate-Change Damages
There were several things that Nordhaus had written in his piece that bothered me. One, in particular, was misleading and a disservice to the debates. When talking about the economic approach to classifying greenhouse gas emissions, Nordhaus wrote (bold added):
In economics, a pollutant is a form of negative externality—that is, a byproduct of economic activity that causes damages to innocent bystanders. The question here is whether emissions of CO2 and other greenhouse gases will cause net damages, now and in the future. This question has been studied extensively. The most recent thorough survey by the leading scholar in this field, Richard Tol, finds a wide range of damages, particularly if warming is greater than 2 degrees Centigrade. Major areas of concern are sea-level rise, more intense hurricanes, losses of species and ecosystems, acidification of the oceans, as well as threats to the natural and cultural heritage of the planet.
I hope the reader will agree with me that Nordhaus is certainly inviting the reader to infer that now and in the future, the best available studies (as summarized by leading scholar Richard Tol) show that emissions of GHG will cause net damages.
The above quotation is not taken out of context; this is all the information Nordhaus provides on the topic. The only ambiguity the reader can imagine concerns just how much damage GHG emissions will cause, now and in the future.
Yet if we consult Tol’s paper—the very one cited by Nordhaus in support of the above quotation—we find that most economic studies find global warming will confer net benefits on humanity at least through the years 2050 – 2060. Only after we get at least another 2 degrees Celsius of warming (and that is compared to a recent baseline, not a preindustrial benchmark), do most studies in this literature say that the damages to certain parts of the world begin to overwhelm the benefits to other parts of the world.
As I mentioned in my IER post, I hope the average reader will agree with me that Nordhaus’s summary of Tol’s findings was extremely misleading (perhaps unintentionally). I daresay the average person, relying on mainstream media treatment of the issue, has been led to believe that “the consensus” of experts believes climate change is right now causing incredible damage and will only get worse as time passes.
And yet, the very person Nordhaus singled out as the leading scholar in the field, shows that the majority of the best available studies show global warming leading to net benefits at least for another four decades.
Tol Clarification—Global Warming’s “Sunk Benefits”
Now when I made the above point in my IER post, I got some pushback from various critics, saying I was misrepresenting Tol’s work. Indeed, Richard Tol himself made the following clarification in the comments of my post:
It’s easy to misinterpret Figure 1 from [my paper] Tol (2009).
Initial warming is indeed likely to be beneficial: CO2 fertilization of crops, reduced spending on heating homes, and fewer cold-related deaths are the main factors.
However, totals do not matter. The incremental impact turns negative around 1.2K. If we were able to control climate, we would warm the planet by 1.2K and stop there. However, the momentum of the climate system and the energy system is such that, if you accept the mainstream view of the workings of the climate, we cannot avoid 1.2K warming, or 2.0K warming for that matter.
The initial benefit is thus a sunk benefit: We will enjoy it regardless of what we do.
This is a very interesting and subtle point. I rush to mention that I made this very point myself in my original IER post footnote:
I note that just because global warming might arguably confer net benefits on the world for the next sixty years, doesn’t by itself mean that no mitigation efforts should be undertaken beforehand. If the standard climate models are correct, then the trajectory of global temperatures will only respond sluggishly, even with drastic changes in emissions down the road.
Thus, Tol’s clarification in the comments is perfectly consistent with my take in the original post. I wasn’t trying to draw conclusions about climate policy from that one graph from Tol’s 2009 paper. Rather, I was showing that Nordhaus had misleadingly summarized what the literature says on the impacts of global warming.
Now I am not here trying to be coy. We all recognize that the political battles over climate change can become downright nasty, and it is understandable that Tol was concerned his work might be used by “deniers” to say that no government intervention is needed. In that light, I probably should have amplified the caveat (which I had only put in a footnote) to make sure Tol’s own position were clearly stated.
A Very Crucial Point
With this straight, let me emphasize why I think this is so critical: The general public has no idea that the “consensus” (if we wish to use such terminology) of economic studies shows net benefits from anthropogenic climate change for decades.
If the general public did know that, it is undeniable that public opposition to a massive new carbon tax or cap-and-trade scheme would be that much greater.
Restated: It is one thing to tell the average citizen, “We are currently doing damage to everyone on Earth, particularly people who live in tropical and coastal regions. Emitting CO2 is exactly analogous to dumping poisonous chemicals into a river and polluting drinking water downstream, which is why the government ought to penalize the activity.”
It is a far different thing to tell people, “Unrestricted GHG emissions are currently hurting some people, but benefiting others. In fact, considering humanity as a whole, the gains to the winners from a warmer planet outweigh the losses to the losers. We estimate this situation will continue for the next 50 years or so, if governments do nothing and let businesses and individuals emit GHG as they please. However, according to our best computer simulations, about the year 2060 our grandchildren—who will be far richer than we are now—will, on net, be made poorer, and will wish we had restricted our own economic growth in order to keep the planet cooler for them.”
We have very little idea what the relevant tradeoffs and technologies will be 50 years from now, and it’s also not nearly as obvious how to properly evaluate the “social utility” of our grandchildren against our own. The case for climate activism does indeed become much more dubious when people learn the true state of the economics literature.
Yes, it is true that many of the scholars working in this field still think there is a case for (initially modest) government mitigation efforts such as a global carbon tax, but the confidence we should have in their recommendations is much lower once the time element is highlighted.
To repeat my main point, and to underscore why I was so upset with his original article: The innocent reader of William Nordhaus’s treatment in the New York Review of Books would have absolutely no idea that the second description above, is the more accurate picture of what the economics literature has to say about the impacts on human welfare from anthropogenic climate change.
Climate Trends: Models vs. Reality
Now we have seen just how important the accuracy of the computer simulations are. After all, Nordhaus is asking us to go along with a massive new tax to eliminate a “negative externality” that won’t manifest itself for at least half a century.
In assessing the predictive power of the standard suite of models, in my IER post I linked to climate scientist Chip Knappenberger’s MasterResource discussion of the following chart from Santer et. al’s 2011 survey article:
Fig. 2. A comparison between modeled and observed trends in the average temperature of the lower atmosphere, for periods ranging from 10 to 32 years (during the period 1979 through 2010). The yellow is the 5-95 percentile range of individual model projections, the green is the model average, the red and blue are the average of the observations, as compiled by Remote Sensing Systems and University of Alabama in Huntsville respectively (adapted from Santer et al., 2011).
Looking at the above chart, Knappenberger concluded that the standard suite of climate models was “on the verge of failing.” (Note that Santer et al. did not come to this conclusion in their paper.)
Knappenberger’s reasoning was straightforward enough: In the chart above, the green line is the average model prediction of global temperatures. The red and blue lines are actual observations (from three different sources) of the global temperature trend from 1989 through 2010, and they have been consistently below the model’s projected trend. In fact, unless observed temperatures begin rising fairly quickly in the next few years, the observed trend will breach the lower boundary of the yellow envelope. If the suite of models were accurately capturing the true climate system, we would only expect natural variability to generate such a breach (on the low side) 5% of the time.
A Different View: Are the Climate Models Doing Just Fine?
Regarding the above, a critic pointed me to this February 2012 RealClimate blog post, which produces a graph and analysis that seems to give an entirely different accounting of the success of the climate models:
Indeed, this graph appears to give a much different picture from the Santer et al. graphic that Knappenberger discussed. The climate models look to be right on the money, predicting a bit too much warming in recent years but well within the uncertainty bounds.
Counterintuitive Confidence Intervals With Climate Models
The resolution of this conundrum relies on two things. First, the confidence interval is tighter in the first graphic from Santer et al. In other words, the yellow envelope in the first chart shows the range in which the actual observations need to remain, if we only want to have a “false negative” 10 percent of the time, whereas the second chart’s gray region is more generous and will only allow researchers to incorrectly reject a correct model 5 percent of the time.
The second crucial difference is that the Santer et al. graph (with the yellow band) looks at temperature trends, whereas the RealClimate graph (with the gray band) looks at single-year observations. Yes, it is true that “natural variability” and other noise can keep the suite of climate models from being “falsified” in any given year. Yet if we look at a 30-year trend, the noise should largely cancel out, with the underlying predicted trend being quite dominant. On this score, the climate models have indeed overpredicted warming, such that they are on the verge of being falsified.
While contemplating “confidence intervals” in the context of climate models, something just didn’t sit right in my mind. In my personal discussions with Chip Knappenberger, it was hard to convey why this procedure seemed so counterintuitive to me. Let me elaborate on the point, because this may be an area where economists and climate modelers come with different preconceptions.
In a standard economic regression analysis, we typically approach things the way one is taught in high school when learning basic statistics. Namely, you set up a null hypothesis that is the opposite of the causal relationship you (the researcher) actually think exists. Then, if there is an apparent relationship in the data (such that you get a positive value on the coefficient for a certain term in a least-squares regression, say) you can see if the result holds up at a 90 percent, 95 percent, or 99 percent confidence interval.
In this normal context, the higher the confidence interval, it means the more confident you are that the apparent relationship between two measured variables isn’t spurious. You are in effect saying, “If there really weren’t any relationship between variable X and variable Y, then I wouldn’t be getting this type of result 99 percent of the time. Therefore, I reject the null hypothesis—which says there is no relationship—and think that there really is a relationship.”
Yet in charts of climate model projections, the “confidence interval” works the other way around. Here, the higher the number, the less confident we can be that an apparent match between the model and nature is due to the underlying accuracy of the model. To put it in other words, here the null hypothesis is that “this suite of climate models is accurately simulating global temperature.” Thus if we make it harder to reject the null (by ramping up the confidence level), then it gives more wiggle room for the models.
Specifically, the “95% range” in the second graph above comes from looking at all of the observed “runs” of the suite of climate models, and then plotting the gray boundary that captures the realizations of 95% of the runs centered around the average. Ironically then, the less agreement there is between the individual climate models, then the wider the gray zone would be, and the harder it would be for Nature to “falsify” the suite of climate models.
Now to be fair, there are rational vetting procedures that the IPCC uses, to determine which climate models are included in the set, the various runs of which are used when constructing these graphics. But to make sure the reader understands the crucial point I am making: Suppose for the sake of argument that one particular model accounted for 3% of the total simulated runs, and it predicted global temperature anomalies of 20 degrees Celsius from the year 2010 forward, while another particular model accounted for a different 3% of the total simulated runs, and it predicted global temperatures of minus 20 degrees C from 2010 forward.
In this (absurd) situation, the RealClimate post would show a massive gray zone covering the “95% confidence interval,” and (barring an asteroid collision or a massive change in the sun), it would be inconceivable that temperature observations would fall outside of this range. Yet that would hardly shower confidence on the suite of models.
To repeat, I am not claiming that there are models included in the IPCC suite, which are palpably absurd and generate the high/low ranges yielding the gray confidence interval. I am simply trying to ensure that outsiders understand exactly what climate researchers mean when they talk about a “95% confidence interval.”
Their interpretation is almost the exact opposite of the statistical tests performed in other disciplines, such as economics. This difference isn’t because the climate researchers are charlatans; it flows from the nature of their task. Even so, it’s crucial for everyone to be aware of this difference.
The scientific modeling of climate change, and its possible impacts on human welfare, are very technical areas requiring years of study to master. When experts try to summarize the fields for the layperson, they sometimes present matters in misleading ways, however inadvertent. William Nordhaus’s treatment of the economics literature, and RealClimate’s discussion of the accuracy of climate models’ temperature predictions, are good examples.