Larvatus Prodeo

There, gentle reader, I have avoided using the term ‘catastrophe’ so that strong messages are appropriately wrapped. Martin Weitzman himself has been more forthright by titling his paper On Modeling and interpreting the Economics of Catastrophic Climate Change. But then he has written 45 pages of mostly econospeak, which carries a different kind of fright. Not existential, but bad enough. Much of it is impenetrable to ordinary mortals. With the help of a couple of interpreters I’ll try to render his message understandable. Anyway this is what I took from Weitzman.

1. Doubling atmospheric CO2 carries a small but unacceptable risk of catastrophic climate change.

2. By catastrophic he means an outcome that threatens civilisation as we know it. We simply can’t ignore the risk. It must be attended to.

3. In the face of truly catastrophic risk the normal practice adopted by economists of applying standard cost-benefit analysis is “wrong, unhelpful, and a dead-end” (from
Kevin Ummel, one of my interpreters). We should do whatever it takes.

4. The uncertainties that give rise to this risk are not the result of inadequate science. Rather they are inherent in the climate system itself.

First, let’s look at the nature of the risk.

You will recall that in the ‘Perdition’ post we looked at a figure from the Stern Review (download from here the full Executive Summary - Figure 2):

Figure 1: Stabilisation temperature for various levels of atmospheric GHGs

Weitzman is concerned with the middle line. His 560ppm (doubling pre-industrial CO2) is close enough to 550ppm. Stern’s graph shows that at the 95th percentile according to the Hadley ensemble study (continuous horizontal line) gives a temperature increase of 4.7C. The dotted line, which represents the average of 11 studies produces a 95th percentile result of over 6C. Weitzman tells us that it is in fact 7C.

Weitzman points out that at the extremity there is a one in a hundred chance that the temperature outcome will be 10C. You can see the problem of what he calls “thick” or “fat” tails, or again “an extreme tail that is heavy with probability” in this graph:

Figure 2: Estimates of equilibrium climate sensitivity (not including slow feedback processes).
From !PCC [AR4], Figure 9.20, page 720.

Weitzman says that the chance of the temperature stabilising below 1.5C is very unlikely but in terms of an upper bound, effectively there is none.

But on top of that Weitzman is cognisant of what Hansen has said about doubled climate sensitivity when long-term feedbacks, which are cutting in now, are taken into account, citing his paper Climate change and trace gases (pdf). With long-term feedbacks Weitzman points out that we have a one in a hundred chance of a 20C warming.

Spratt and Sutton remind us in Climate Code Red that we regard it as unacceptable if an aeroplane has a better than one in a million chance of falling out of the sky when we board it. Clearly we should demand even more stringent odds for the end of life on the planet or the end of civilisation as we know it - phraseology that Weitzman uses.

As to what 20C warming would be like at the rate we are pushing the climate, Weitzman says it hasn’t happened before, not in a billion years (page 3).

There would be no summer ice on the planet, perhaps none at all in winter either. This would result in a sea level rise of some 80 meters. The unknown to me is thermal expansion as the huge body of water, now on average 4km deep, heats up.

One problem here is that more than 2C warming brings danger of tipping points, including the big one of the release of methane clathrates from under the sea. At 6C it is a likely outcome, at 10C you could put your house on it.

Warming of 2-4C becomes increasingly difficult and brings eventual rising sea levels at about the rate of 15 meters for every degree. 5-6C could change the face of the planet in a way that is essentially unimaginable..

For 10C the best analogy in terms of impact on the biosphere is probably the End Permian or Permian–Triassic extinction event of 251.4 million years ago. But each of these events has it’s own characteristics. One of the most important of ours is the speed of the carbon pulse.

Weitzman didn’t have access to the paper by Sheehan, Jones et al (pdf) survey who found in the real world that:

an atmospheric CO2 concentration level of over 900ppm CO2e and warming of 2.2C to 4.7C are projected by 2100, even if aggressive emission reductions after 2030 are achieved. (Emphasis added)

So if you look back to Figure 1, imagine what a line representing 900ppm would look like, adjust it in terms of Hansen’s doubled climate sensitivity for long-term feedbacks, and you can see that there is only a minute chance of avoiding 2C warming, given the path we are on.

It’s true that Weitzman did not say, as in my point 3 above, that conventional cost-benefit analysis was a “dead end” (Ummel’s words). What he said was:

policy advice coming out of conventional thin-tailed CBAs of climate change must be treated with (possibly severe) scepticism until this fat-tailed high-input aspect is addressed seriously and included empirically in a true fat-tailed CBA.

In the light of everything Weitzman says, I think Ummel’s words are justified. At the very least Weitzman is saying “Look at the fat-tailed possibilities first, resolve them if you can, and then go on from there.”

Concerning point 4 the relationship between climate feedbacks and temperature outcomes is represented by the following diagram:

Figure 3: How a symmetric PDF (probability density function) for the feedback factor will lead to an asymmetric PDF for temperature.

If you take the shaded area on the y-axis, flip it to the mirror image and place it over the x-axis in Figure 2 above and you’ll get the idea.

This diagram, shown by both my guides Kevin Ummel and Peter Wood (pdf), comes from am article by Roe and Baker cited by Weitzman Why Is Climate Sensitivity So Unpredictable? (behind the paywall). There is more about it at Slate, Global Dashboard and at RealClimate.

To paraphrase Ummel further, while scientific work might trim the feedback possibilities a little (although this would be really hard going), the possibility of “really scary stuff” in terms of temperature outcomes will remain.

Peter Wood (pdf) provides a more than useful guide to the thicket of Weitzman’s prose. You might like to read him first, or tackle Weitzman’s paper (pdf) directly. Weitzman uses concepts such as VSL (the value of statistical life - virtually infinite) and the Dismal Theorem. I can’t find a common language explanation of the Dismal Theorem from Weitzman’s text, but I gather from James Annan that when considering the impacts of fat tails “the loss in utility arising from such impacts increases more rapidly than their probability decreases” and from Yohe and Tol (pdf) that we can’t experientially learn anything about such impacts before we are called on to do something about them.

We shouldn’t pretend that the parameters for a normal cost-benefit analysis are certain when they are not.

Now try this:

The Dismal Theorem says that in the limit as the VSL-like parameter becomes very large, agents are willing to pay a very high price to eliminate or reduce deep uncertainty - a vexing idea that is difficult to wrap one’s mind around.

Weitzman worries about taking the precautionary principle too far.

Should we have foregone the industrial revolution because of the GHGs it generated?

A pointless question, I think. We didn’t. Now we should use the technology of the industrial revolution and beyond to help us go on from where we actually are.

He also seems to assume that fixing the situation so as to eliminate the feedbacks will cost a very high price. Spratt and Sutton show that the price compared to the war effort in WW2 for example is modest. We just have to get cracking now.

Weitzman would agree about starting now. He points out that that leaving it for our progeny who will be richer than us to fix is not sensible, because by then it might not be fixable. Also if things really go pear-shaped they may not be richer than us.

But Weitzman, like Hansen, might be just another voice crying in the wilderness. Peter Wood, he who comments hereabouts and who rendered me some assistance (thanks Peter) tells me that he asked Garnaut after his H.W. Arndt Lecture recently whether he would be making any implicit or explicit assumptions about the value of Weitzman’s “VSL-like parameter”. The answer was “No” - the shortest answer he gave that night!

To do so would shift the paradigm of how we approach the economics of global warming mitigation. We can’t have that, now, can we?!

To me the direction of policy is clear. We need a concerted planet-wide effort to reduce atmospheric levels of GHGs to 350ppm CO2e in the first place, because it’s lower than where we are now, and because Hansen et al tell us that 350ppm is the lower bound for an ice-free planet. Personally, I think we should try to find our way back to 265-285ppm which prevailed in the Holocene for nearly 10,000 years. It suited our species very well. It would also give the animals a go not to mention the rest of the biosphere. It is our ethical responsibility and would take the Dismal Theorem out of play.

As a co-operative project involving the whole world, that would be a big ask, but it’s worth a try.