30 August 2008

Neoclassical Deathwatch

In past posts, I have pointed out the important difference between risk and uncertainty. (Risk is something that you can quantify in terms of probabilities, while uncertainty is something that you cannot.)

For those of a statistical inclination (i.e., neoclassical economists using expected utility), risk is the way to go and uncertainty is ignored. That's a fine idea in the casino but a miserable one when you are dealing with a mad dog.

Climate change is one such dog, and this essay [PDF] takes apart the conventional wisdom of using risk and cost-benefit analysis (CBA) to examine climate change (as the Stern Review did):*
It is now acknowledged that the economics of climate change is now more appropriately concerned with uncertainty rather than return, a feature of the problem that has been evident from the early 1990s, when the scientific assessments began in earnest. It can also reasonably be argued that CBA is useless for the climate problem because of the uncertainty and risks of catastrophe. The discounting of costs and benefits in which risks are converted into certainty equivalents and discounted at market rates has been shown to be misleading and biased. This in turn implies that the economic problem is one of achieving political targets, based on scientific evidence, at lowest costs compatible with equity and effectiveness, rather than with the economics of choosing the targets themselves.
Bottom Line: Economic analysis must use methods appropriate to the problem, and climate change requires that we consider uncertainty. Tough.

hattip to DS

* and this paper [pdf] critiques Stern as too neoclassical.


TokyoTom said...

David, it seems to me that the chief critic of classical CBA as applied to climate change is Harvard's Marty Weitzman.

Here is more on Weitzman and climate CBA generally.

Eric said...

I don't understand the difference between 'risk' and 'uncertainty.' To me all the cases can be quantified. Some are harder and messier than others. For instance, I am trying to quantify various large systemic risks and even climate collapse in net present value terms and in political incentive terms. Quantification seems to be hard but not impossible or unknowable. Quantification also seems to be critical if I want anyone else to pay attention to me.

albionwood said...

If you can't understand the difference between risk and uncertainty, then I hope nobody pays attention to you.

Eric said...


OK, I will take your challenge and challenge back.

Please define in scientific terms, preferably statistical, preferably using Bayesian statistics, a quantitative distinction between the two. I do not think that you can do it.


Eric said...

@ albionwood

Risk can be defined as “the threat or probability that an action or event will adversely or beneficially affect an organisation's ability to achieve its objectives”[1]. In simple terms risk is ‘Uncertainty of Outcome’, either from pursuing a future positive opportunity, or an existing negative threat in trying to achieve a current objective.


Risk and uncertainty seem to be different words for the same thing.

Your ad hominem attack struck me as foolish, by the way.

David Zetland said...

@Eric and Albionwood -- FYI, I am distinguishing between risk that can be quantified via a probability distribution and uncertainty that cannot. For example, risk of death from heart disease in the population vs uncertainty of a *particular" girl calling you for a date. Read Robert Frank (1924?) and Taleb's Black Swan for original and recent discussions of the differences...