A few weeks back, Gernot made reference to Martin Weitzman's work on analyzing climate change as a low probability, high-stakes event. After a few weeks, I've had a little time to give some blog-level reflection to what Weitzman has to say and quite frankly, I'm not sure I like where the logic leads (I'm not sure I don't like where the logic doesn't lead either, but that's for another day).
First, a little technical stuff. Here is the abstract from Weitzman's paper:
With climate change as prototype example, this paper analyzes the implications of structural uncertainty for the economics of low-probability high-impact catastrophes. Even when updated by Bayesian learning, uncertain structural parameters induce a critical tail fattening of posterior-predictive distributions. Such fattened tails have strong implications for situations, like climate change, where a catastrophe is theoretically possible because prior knowledge cannot place sufficiently narrow bounds on overall damages. This paper shows that the economic consequences of fat-tailed structural uncertainty (along with unsureness about high-temperature damages) can readily outweigh the effects of discounting in climate-change policy analysis.
I know, blah-blah-blah. So here's the blog version--If the likelihood of some really, really bad outcome is unknown, but unlikely and not impossible, the consequences of not avoiding the really, really bad outcome are really, really bad. Action to avoid the really really bad (but unlikely) outcome is economically justified.
Sounds like a reasonable statement of the precautionary principle and Weitzman's paper goes a long way toward giving a mathematical formalization to the principle. I don't want to argue with Weitzman's conclusions, but rather point out the slippery slope of using Weitzman's logic for policy analysis for other low probability high stakes events.
First let's look at an example. Let's say two families of 5 are on vacation at a wooded resort in say, oh I don't know, Kentucky. And let's say that upon arrival at said resort, the security personnel warn visitors to be cautious in secluded areas because the area is know for (poisonous) copperhead snakes, but it is unlikely (but not impossible) that a snake could wander--do snakes wander?--near the resort. And for the sake of this example, let's suppose that the parties involved are city dwellers and don't know the difference between a copperhead and a harmless rat snake. Now, suppose the six kids--all under the age of 12--are playing in the side yard when a six foot snake wanders onto the patio. What is the correct policy decision here?
According to Weitzmanian logic, the correct policy decision is to take action to prevent the unlikely, but not impossible, catastrophic outcome. Why? Because the expected losses from not acting are so high--that is, the losses if the unlikely catostrophic event happen are close to infinity--that the immediate costs of acting are warranted.
In the snake example, the probability of a catastrophic event--death of a child--is low, but uncertain. In such a case, it is economically justifiable to remove the threat by beheading the beast. Wow, that actually makes me feel a little better.
Now consider another example. Suppose that it is unlikely, but not completely impossible that a country with a totalitarian government possesses weapons capable of killing thousands, and that this country has shown a willingness in the past to use such weapons...
Like I said, a slippery slope.
