Weitzman\’s argument

This might be one for John Quiggin here. JQ I know is quite taken with Weitzman\’s arguments about the desirability and costs of climate change abatement. Bjorn Lomborg a great deal less so.

So the question is, is this a fair representation of Weitzman\’s views and is it indeed an argument against them?

But it is interesting to assess Weitzman\’s argument (My arguments are partly indebted to Professor Nordhaus (pdf)). Tickell (and many other campaigners) fancies Weitzman, because his economic argument seems to support draconian climate policies. While very technical, it relies on a fairly straightforward gist. All risks you can think of – even catastrophic ones – have non-zero risk. Thus, it is possible (if not very likely) that global warming will not only increase the planet\’s temperature by 4C, but 10C. Heck, it might even increase beyond 20C – which Weitzman with armchair climatology, suggests might have a probability of 1%. Since evidence for or against such extremes is scarce, accumulating evidence can only slowly close us in on their true probability. Yet, for any given amount of evidence, there will always be sufficiently outrageous risks (think 30C) that are sufficiently unbound by evidence and sufficiently close to negative infinite utility that the total net utility is negative infinity. Thus, we should be willing to spend all our money to avoid it.

Now, in principle all economists would agree that non-trivial risks should be included in the model, and for example, Nordhaus has done that analytically in cost-benefit models (they still show that large emission cuts are not warranted). However, the Weitzman result curiously means that the more speculative and fuzzy the extreme event, the more it counts in the total utility.

This is an argument driven by a technicality – essentially a claim that we are willing to pay an infinite amount to avoid even an infinitesimal risk of annihilation. Yet we demonstratively aren\’t – and shouldn\’t be. Civilization-ending asteroids hit the earth once every 100m years, but at present we only spend $4m per year to track them. Maybe we should pay $1bn. But we shouldn\’t spend everything.

This underscores the fatal flaw in the Weitzman argument. When we allow all scary, fuzzy concerns onto centre stage, there is no end to where we should spend all our money. Every conceivable policy measure has a non-zero risk of catastrophe and so should be avoided at any cost. Biotechnology, http://www.wisegeek.com/what-is-a-strangelet.htm strangelets, runaway computer systems, nuclear proliferation, rogue weeds and bugs, pandemics, and asteroids are just a small sample of the areas each of which we should spend all our money on.

Professor Quiggin responds thusly:

I think this gives you a pretty good case against spending infinite amounts to protect against remote threats. But you don\’t need that to say that, where one action is associated with a long tail of risk and the alternative is not, you\’d better think carefully about the tail, rather than focusing on median or mean damage estimates. 
For example, if you buy Weitzman\’s numbers, you should certainly be willing to spend more than 1 per cent of income to protect against catastrophic global warming. How much more is tricky, I agree.

The striking thing about global warming is that, starting with a mean warming of 3 degrees under BAU, a symmetrical estimate suggests 0 and 6 are equally possible. In the event that 0 is correct, money spent in mitigation has no effect at all, so we are out (say) 3 per cent of income. On the other hand, 6 while perhaps not wiping out life on earth would be catastrophic. So if you think the standard deviation of your projection is 1.5 (that is, 2.5 per cent probability of being outside the [0,6] range you don\’t need to go to tiny probabilities to reach a point where the tails play a big role in any estimate of expected damage.

Hmmm…some thinking to do here for me.


3 thoughts on “Weitzman\’s argument”

  1. Pingback: Recent Links Tagged With "argument" - JabberTags

Leave a Reply

Your email address will not be published. Required fields are marked *