Pindyck vs. Weitzman: How much should we spend to curb global warming?

I spent last weekend at an excellent NBER Conference, Climate Change: Past and Present. There was lots of interesting stuff I’d love to blog about, and hopefully will soon. But a highlight was a fantastic exchange between two proverbial giants on the Big Climate Change Question.

I'm paraphrasing arguments from memory here, so hopefully I don't misrepresent either of these guys. And apologies in advance if this seems too technical..

Robert Pindyck went first. He presented a more-or-less standard representative agent macro model of the world economy and built in a lot of assumptions about various kinds of uncertainty surrounding the effect of warming on output. His model had welfare as a function of output and output growth as a function of temperature change. Importantly, it seems, he assumed temperature change could not affect utility in any manner other than output—a seemingly strong assumption in my book (we all study state-dependent utility, no?). He emphasized many assumptions that were generous toward those most worried about cataclysmic consequences of global warming. And in anticipation of Weitzman, he used probability distribution functions of uncertainties with “fat tails.” Pindyck concluded that society’s willingness to pay to prevent or severely limit global warming could be no more than about 2.5 percent of world income, and likely much, much less.

While Pindyck did use some relatively “green” assumptions that might favor a large number, it was also clear the deck was stacked. One key assumption: it makes no difference whether or not the world ends with certainty in 400 years. I had a hard time with that one. Stephen Salant, one of the discussants, had a hard time with it too. He noted that changing this assumption alone could increase willingness to pay to 99 percent of income.

I noted Pindyck’s graphs omitted from consideration very large levels risk aversion (relative risk aversion was less than or equal about 2). For relatively small amounts of risk aversion and uncertainty, greater risk aversion reduces willingness to pay to stop warming. But for very large levels of risk aversion, this reverses, and willingness to pay explodes with risk aversion (in my view, Pindyck tried to ignore this possibility and bushed it off too casually when questioned). This reversal, I think, is also tied to the amount of growth in economy. It bothers me because empirical finance says risk aversion is crazy large and way off the charts of what Pindyck was willing to consider—like a CRRA of 20. This assumption alone would seem to push everything to infinity or zero depending on the other assumptions.

Weitzman’s view was the polar opposite of Pindyck’s. Weitzman finds uncertainty everywhere, and finds thoughtful account of this uncertainty can easily lead to infinite valuations for preventing global warming. Weitzman acknowledged that plausible models give very low willingness to pay, but argued that equally plausible models could give infinite valuations. No one questioned his math. I, at least, found his assumptions and the effort he took to justify them as reasonable. Weitzman has clearly dug in deep in terms of the scientific climate change literature and thinking seriously about the long run.


A few interesting highlights:

Weitzman suggested that it’s hard to take any of the climate models and economic models very seriously. He said there is little reliable temperature data because inferences from ice and sediment core samples and tree rings were noisy. In his view, the only good data are CO2 concentrations. These data go back hundreds of thousands of years because the air was trapped in the ice and so accurate CO2 concentrations can be extracted. Those data show that we now have higher CO2 concentrations than at any point in many hundreds of thousands of years. And concentrations are growing at rate that almost surely exceed anything in millions and perhaps hundreds of millions of years. It is also clear that, broadly speaking, CO2 and temperatures moved together in history.

We are in unchartered territory, and this is the principal source of uncertainty. Even a small probability that our activities will extinguish the planet give good cause to stop emissions, even at high cost, per Weitzman’s assumptions.

In the end, Weitzman said economic benefit-cost analysis is no help in situations like this one. He argued that climate change was rather unique in this way. Decisions must be made by some other means.

Pindyck argued that giving up on economic benefit-cost analysis would cede the debate to radical environmentalists. (He didn’t say why not to radical climate-change denialists.) He argued it was critical for economists to wrestle with the issue and find a reasonable and balanced valuations and targets. He emphasized learning (have we really much more to learn before it’s too late?). Pindyck argued climate change was not unique. Other potential problems were equally ominous, including: running out of water, infectious diseases, nuclear war and nuclear terrorist attacks. So if willingness to pay for preventing climate change was infinite, it left no room for these other problems. I think Weitzman concurred that climate change risks needed to be balanced against other concerns, but where to spend money for the other concerns seemed less clear. And it wasn’t clear how economists might help to quantify, realistically, these competing ominous concerns.

The inescapable conclusion to me was that we should worry a lot about what the future might hold. Technology and growth hold great promise and potential. But we also have a remarkably dangerous ability to influence our environment and destroy ourselves. How do we work collectively to make prudent choices and while preserving individual liberties? I’m pretty sure the answer is not to pretend these problems don’t exist.

There was a lot of technical debate about the separability of temperature effects from output effects on utility. I found the debate rather strange because all of it was in terms of a single, globally representative individual. From where I sit, looking at how climate change will affect food supply, I can easily imagine a situation where the richer two-thirds of the world goes on growing happily as ever toward utopia while the other third starves to death. I have a hard time reconciling that view with a non-separable utility function.