Wednesday 16 November 2011


Martin Weitzman is presenting here at Canterbury, tomorrow afternoon, on the economics of climate change.

I'm really looking forward to his talk; I hope to see a lot of local Offsetting readers there too. RSVP now if you haven't already!

Weitzman argues that a fat downside tail in the distribution of potential outcomes make the case for strong action to combat global warming - it's insurance that's well worth the premium.

It's a pretty reasonable argument. And, I broadly accept it: it's worth spending a bit on insurance. My general problem with it is that there are all kinds of potential fat tailed risks with deep structural uncertainty. So I have the same problem with Weitzman's argument as I have with Posner's "Catastrophe": there are a non-trivial number of potential states of the world in which all human life is extinguished. It's not immediately clear how we ought to expend resources in insuring against all of the different risks.

At the margin, I'd strongly favour transferring money from general spending to these kinds of insurance expenditures. But how much can or should we spend in total? And how to allocate across risks? I wouldn't be averse to transferring some money from global warming abatement to asteroid abatement: if NASA or somebody else discovers an asteroid that's on track to hit us in a decade, it's remarkably unclear whether there's anything we can do about it. I don't think the decade-long end-game would be particularly fun. There's also the potential emergence of all kinds of different superbugs. The risk of bad global warming outcomes in the medium term is perhaps on par with the risk of emergent superbugs, and much higher than the risk of asteroid (although the costs of asteroid approach infinity). From there we move to less and less likely events - all the way down to Shoe Event Horizons and Grand Collapsing Hrung Disasters. We need some mechanism to help us allocate resources correctly between global warming and other plausible (and, on an insurance argument, even implausible) potential disasters. And if Cost-Benefit analysis fails us when facing issues of deep structural uncertainty, on what basis ought we make decisions?

I hope to get the chance to ask Weitzman at tomorrow's lecture. And I hope to see you there.


  1. In Australia we will soon see how well this 'insurance' works. Impose a huge carbon tax burden on your economy for a less than detectable effect on temperature.

    Oh what value.
    Ask this clown if he likes paying for insurance that won't actually pay out.

  2. And what was his answer? @V no detectable effect on T but perhaps a detectable effect on exportable innovation

  3. @Cam: That there are a small number of such risks that are plausibly worth insuring against, and this one is both reasonably likely and high cost if it eventuates.