The Dismal (Climate) Science: On Marty Weitzman, Fat Tails, And How Economists Could Better Help Us to Overcome Global Warming

The World Bank recently invited Harvard economist Marty Weitzman to present his latest work on catastrophic climate change, with federal judge Richard Posner, millionaire math whiz John Seo, and Nobel Laureate Tom Schelling providing comment. Since Weitzman's paper is an important insight into the uncertainty of economic analysis of global warming, it deserves some investigation -- as does the appropriate role and necessary modesty of economists with respect to climate change more generally.First, kudos to the Bank for holding the event at all. It was a welcome sign that climate change as a development issue is now fully accepted. Anybody who has been paying any attention at all now knows that global warming will hurt poor people in developing countries first and worst (See for example, Bill Cline's book on Global Warming and Agriculture). Moreover, there is a rapidly growing recognition (Another Inconvenient Truth) that rich and developing countries must both reduce greenhouse gas emissions rapidly to avoid planetary disaster.It's therefore unfortunate that the World Bank event repeated an all-too-common Washington mistake: Offering a supposedly learned discussion of the vagaries of climate change without an actual climate scientist on the panel. That kind of unsupervised play can lead to erroneous comments -- like the puzzling suggestion by Tom Schelling that the atmosphere may become saturated with carbon dioxide, thereby limiting warming -- going entirely unchallenged.Climate scientists know (and have for 50 years) that the amount of CO2 in the atmosphere could be increased to 10,000 times the present level, and there'd still be room for more of the heat-trapping gas and resulting warming.¹The same goes for the event's considerable discussion of geoengineering,² and the support voiced for the potential, purposeful injection of chemicals (sulfate precursors) into the upper atmosphere to create a "sulfur sun shade" to reduce incoming radiation and cool the planet. Discussing sulfur-spraying is perfectly fine -- reputable geophysicists study and publish on the subject -- but if it's going to be treated in a public setting, let's try to have a better collective handle on the science and unknowns. For example, "more acid rain" is an exceedingly simple list of the downsides. Most critically, we have little idea how such action might affect atmospheric circulation, cloud formation, evaporation, or patterns of rainfall -- major components of the climate system that could lead to serious unintended (and unimagined) consequences.But critiques of the broader discussion aside, Weitzman's specific argument -- detailed in On Modeling and Interpreting the Economics of Catastrophic Climate Change -- deserves serious attention. I can personally vouch that the paper is a laborious, technical read (as noted by more than one panelist), but the primary, take-away point is this: When economists attempt to weigh the costs and benefits of climate change to determine the "efficient" scale and speed of humanity's response, the analyses largely ignore the implications of low-probability catastrophes. The costs of truly catastrophic warming (i.e. leading to rapid sea-level rise, mass extinctions, collapse of agriculture, disruption of ocean currents and carbon sinks…you get the idea) are so high that even when they are relatively unlikely, their inclusion in the analysis is enough to swamp any standard cost-benefit considerations. It doesn't take an economist to understand that in the face of such risks we should be willing to bear just about any cost to avoid even the small chance of planetary catastrophe. Indeed, this is the conclusion that Weitzman’s calculations come to.Weitzman's contribution is the wedding of expected utility theory (the language of cost-benefit analysis) with what was already a well-known source of uncertainty from the perspective of climatology. The key variable to any climate model, or economic assessment based on their results, is what's known as "climate sensitivity" -- the estimated, equilibrium increase in average surface temperate given a doubling of the atmospheric concentration of CO2. Because humans have never witnessed a doubling of atmospheric CO2 and its effects, scientists derive their estimates from the study of similar events in the distant geologic past.Weitzman notes that estimates of climate sensitivity follow a probability distribution that is wide and "thick-tailed" (i.e. there's a small but not insignificant probability of really dramatic warming). Below is a graph of recently-published climate sensitivity estimates from an excellent paper by Malte Meinshausen (in Avoiding Dangerous Climate Change ), who has been assessing the probabilities of warming scenarios for years. You can see that the area of highest probability is around the IPCC's "best estimate" of 3°C, but the tail extends far to the right (technically off the graph) and retains a fair amount of probability (hence the term "thick tail"). Weitzman ballparks that there is a 5% probability of a climate sensitivity of 7°C or more.The additional and essential point that needs to be made, especially when thinking about catastrophic climate change, is why the probability distribution of climate sensitivity looks the way it does. Its shape is not an accident but due to the fundamental nature of the climate system. The eventual change in temperature is driven not primarily by the assumed doubling of CO2 but by the system's resulting "feedback" processes (i.e. a warmer world has more water vapor and less ice, which leads to more warming, and so on). In the absence of feedbacks, a doubling of CO2 would only increase temperature by about 1.3°C. But with them, we get potential values of 8°, 9°, or even 10°C or more. To see how, consider the illustrative graph below from a recent Science article by Gerard Roe and Marcia Baker. The y-axis shows the distribution of climate sensitivity (flip the shaded area around and it will look familiar). The curve running between the shaded figures shows the relationship between the strength of the climate system's feedbacks ("Feedback factor" on the x-axis) and the total change in temperature. ³Or you can think of the x-axis as an interest rate and the y-axis as your savings account balance. Imagine you were going to put $1,000 away for 50 years, and I told you that the interest rate would be determined by flipping a coin: 5% for heads and 6% for tails. Due to the power of compounding, the seemingly small difference between 5% and 6% translates into really big differences over time -- $8000 in fact.⁴ The climate system has the same principals, except that we're not flipping a coin to choose between 5% and 6%. Instead, we face a wide range of potential feedback factors, some of which would translate into very large temperature increases.As the figure shows, the climate sensitivity estimate will always take its characteristic, fat-tailed shape as long as we have a more-or-less symmetric distribution (i.e. uncertainty) for the feedback factor. But the important point is this: Although significantly reducing the uncertainty of feedback effects can decrease the probability of more than about 8°C warming for a doubling of CO2, it doesn’t do much of anything to reduce the probability of warming events in the 5°-8°C range (about 9°-14°F).⁵ In other words, the basic uncertainty that drives Weitzman's conclusion is a permanent feature of the climate. No amount of research is going to change that fact that we are spinning a roulette wheel and the little white ball could well land on the temperature equivalent of "really scary stuff."So what are economists do to? Massaging the statistical techniques isn't going to help. The real prospect of catastrophic outcomes means the question of whether or not to take serious and decisive action is a forgone conclusion. One could argue that the probability of German scientists acquiring a nuclear weapon during WWII was "very low" (like the probability of a climate sensitivity of 10°C), and judging by the state of the program at the war's end, this seems reasonable. Yet the United States alone spent the equivalent of $27 billion in today's dollars to address this small probability, while spending more than $4 trillion more to counter the "heavier" part of the distribution -- the loss of Europe and the Pacific to the Axis. Nobody now suggests that the U.S. was wrong to spend the $27 billion to address the relatively small risk that the Nazis might create a nuclear bomb.What this tells us -- and what Weitzman's paper implies when the impressive economic jargon is stripped away -- is that applying standard cost-benefit analysis in the face of truly catastrophic risk is wrong, unhelpful, and a dead-end. If economics wants to remain relevant, I believe it needs to ditch the integrated assessment models, which have often presented their dubious results as reason for caution and delay, and start doing what economists did in the 1940's: they showed how to turn the American economy into a fantastic engine of production to equip its allies and counter the common threat. The equivalent today is to show how -- and at what cost -- the world can quickly and convincingly confront global warming and eliminate emissions.The climatologists and engineers have done their part in documenting the problem and preparing the technical solutions, respectively. It's time for us to stop fussing about discount rates and the like and get down to the business of finding the least-cost applications of solutions and making the case to the private and public sectors. I think the fundamentals are there to make a very convincing one. We can also help determine the reasonable sharing of those costs among countries and communities. And perhaps most importantly, we can try to keep up on the science as we go. Doing so will remind us that we don't have the time to remain mired in the arcane.--------------------------------------------------------¹The saturation argument is sometimes voiced by skeptics claiming that additional CO2 cannot possibly absorb further radiation and produce greater warming. Unfortunately, the science points the other direction. Consider that the Earth’s atmosphere is about 0.04% CO2 by volume. On Venus the figure is 96.5%, and some infrared radiation still makes it way through the greenhouse and into space.²A class of interventions that propose to alleviate global warming not by reducing greenhouse gas emissions but by engineering a geophysical solution.³The formula for the curve should look familiar to economists. It’s the basic way to express the multiplier effect given an initial rise in spending.⁴You would have $12,000 if the coin came up heads (5% interest) and $20,000 if it came up tails.⁵See the Roe and Baker paper for more on this point.