Manzi argues that Krugman and Weitzman get it wrong when they argue that the prospect of a fat-tailed probability distributions motivate urgent action to deal with climate change. He argues that the tails are not fat and that spending unbounded amounts to address arbitrarily small prospects of catastrophe is analogous to the fallacy involved in Pascal’s Wager. Thought provoking argument that I will think about.
Update: The essence of the Manzi argument is his claim: “I will argue that his (i.e. Krugman’s) consideration of the role of uncertainty in determining our course of action is also incorrect, because he fails to provide any principle by which we could establish any limit to much we should spend to reduce a risk which can never be eliminated entirely”. Recall that Pascal’s Wager claims that even though belief in God cannot be proven one should believe in God because this belief costs nothing but offers unbounded rewards (e.g. ever-lasting life) if correct. We can modify this a bit and recognise that belief in God might cost a little (listening to boring clergymen might be one cost) but that this cost must be assessed against the non-negligible probability of gaining an immense reward. In a different setting this argument might be that it is worth spending $100 trillion to (a) avoid climate change, (b) avoid an asteroid, (c) protect a natural species, (d) …etc… if not doing so might involve a non-negligible probability of a disaster costing $1000 trillion. The argument is pretty much the same argument as Pascal’s for advocating belief in religion.
Manzi’s claims are first that the probability of something like a climate catastrophe is much lower than suggested by Krugman or Weitzman and is, in fact, a very remote event. Second, he argues that there are many catastrophic risks in the world (many more than just possibilities (a), (b), (c)) - perhaps thousands – and we could not address them all. Moreover these risks can never be entirely eliminated. Hence we need to limit the number of potential catastrophes we address and limit the amounts we spend on individual potential catastrophes. This is analogous to a resolution of Pascal’s Wager which suggests that, on Pascal’s logic we should believe in many religions i9f we were to avoid the high opportunity cost of not achieving possibly immense rewards.
Weitzman claims that the probabilities of climate catastrophe are not remote. Specifically:
Weitzman argues that if only gradually ramped up remedies are applied that in two centuries the probability distributions of temperature outcomes have decidedly ‘fat’- tails:
- With probability 0.05 the increase in mean global surface temperatures will be greater than 10oC.
- With probability 0.01 the increase in mean global surface temperatures will be greater than 20oC.
These non-negligible probabilities of disaster are backed up by a recent MIT study. The other examples I cite (asteroid strike, biodiversity disaster) seem much less likely and hence would seem to be much less worthy of focus than that of a climate catastrophe.
If probabilities of a disaster are very small then in everyday life we trend to ignore them. I don’t avoid driving my car because of the probability of a car accident. That is rational both because the probability of a disaster is low – it is not negligible – and because the costs of avoiding this activity and many other low risk (but catastrophic) activities with cumulatively be high. In this case the Manzi logic works.
I read the article on the article, not the actual article, but if this is the argument:
“In very short form, Weitzman’s central claim is that the probability distribution of potential losses from global warming is “fat-tailed”, or includes high enough odds of very large amounts of warming (20°C or more) to justify taking expensive action now to avoid these low probability / high severity risks.
The big problem with his argument, of course, is that the IPCC has already developed probability distributions for potential warming that include no measurable probability for warming anywhere near this level for any considered scenario.”
It’s very silly, because even increases in the range of 4 degrees are going to be exceptionallty bad news, and they are well within the realm of possibility.
I understand but I am sure Weitzman would defend his estimates. Others confirm the non-negligible possibility of catastrophe. If this is correct then Manzi’s argument dies.
Well the redistributive sector can argue the toss over probabilities all they like, but the real probability is that nothing will piss the productive sector off more than the actions of those who would lecture them about such probabilities but fail to walk the talk- http://www.bostonherald.com/news/regional/view/20100425home_field_advantage_bradys_pigskin_palace_slammed/srvc=home&position=0
It’s like banging on about all the ice melting and the seas rising and then in the next breath jumping on the bandwagon and bidding up coastal RE. What’s the probability of a few carbon offsets clinching that skinny taled argument we no longer need to ask?
I read the article on the article, not the actual article, but if this is the argument:
“In very short form, Weitzman’s central claim is that the probability distribution of potential losses from global warming is “fat-tailed”, or includes high enough odds of very large amounts of warming (20°C or more) to justify taking expensive action now to avoid these low probability / high severity risks.
The big problem with his argument, of course, is that the IPCC has already developed probability distributions for potential warming that include no measurable probability for warming anywhere near this level for any considered scenario.”
It’s very silly, because even increases in the range of 4 degrees are going to be exceptionallty bad news, and they are well within the realm of possibility.
Thanks for the very thoughtful post on my post.
A couple of quick things:
1. Re: “It’s very silly, because even increases in the range of 4 degrees are going to be exceptionallty bad news, and they are well within the realm of possibility.” According to the IPCC (AR4, WG2, SPM) the economic costs of a 4C increase in global temperature are estimated to be 1% – 5% of global economic output.
2. This is why weitzman has to argue for a 1% chance of a 20C increase. There are specific studies that show higher and lower climate sensitivity, and have been for decades, yet the consensus estimate for climate sensitivity has remained extremely stable at about 3C since the 1979 Chaerney COmmission report.
Best,
Jim Manzi
Well the redistributive sector can argue the toss over probabilities all they like, but the real probability is that nothing will piss the productive sector off more than the actions of those who would lecture them about such probabilities but fail to walk the talk- http://www.bostonherald.com/news/regional/view/20100425home_field_advantage_bradys_pigskin_palace_slammed/srvc=home&position=0
It’s like banging on about all the ice melting and the seas rising and then in the next breath jumping on the bandwagon and bidding up coastal RE. What’s the probability of a few carbon offsets clinching that skinny taled argument we no longer need to ask?