Prediction Model Bias
Brad Plumer provides a very practical reason why he believes environmental cost predictions tend to be over-estimates:
[M]arkets always tend to be smarter than these forecasters, and adjust in ways that no one expected.
Let’s assume arguendo that this is correct (and I think it is, as a general statement about markets vs. planners – I’m about as Hayekian as they come). This leads to two questions, one far more important than the other.
1. The narrow point is that, per prior discussion on this, if one could actually observe a reasonably consistent over-estimate in a provably-relevant reference class of prior predictions, this would lead a competent forecaster to make a transparent “topside” adjustment to get a best-available forecast.
2. The more important point is that this same logic ought to apply to the damage estimates that similar bodies forecast for the costs of climate change. That is, markets should be smarter than forecasters think when it comes to adapting to climate change as well (e.g., crop selection, transitions to other economic sectors, clever infrastructure developments and so on). To only focus on this kind of error on one side of the cost / benefit calculation is loading the dice.
This estimate doesn’t right off strike me as correct. It would take a lot of hubris to think that you were going to actually estimate the energy portfolio and the individual source costs to get a total cost estimate. Not least because you are creating conditions which will foster technological change.
More sensibly you would try to estimate a long run demand curve for the dirty good that inside of it included cross-price elasticities of substitution. That is, don’t try to figure out what the market will come up with just look at how the market has responded to past prices and try to build a demand schedule.
More likely is that because there is a lot of guesswork involved in this process, modelers are erring on the side of caution. For example, what is the long, long run price elasticity of carbon emitted by coal. Hard to know but its got to be greater than long long run price elasticity of coal. Lets just use that as an upper bound and call that “high price scenario.”
The analysis is probably more sophisticated than that but the idea is that you often have to bound your estimate by things that you are pretty darn sure are greater or less than the target.
— Karl Smith · Jun 23, 05:54 PM · #
How does the market or a market predict the cost of “environmental costs?”
— cw · Jun 23, 06:08 PM · #
That was a very clumsy sentence. I just meant to say “predict environmental costs.”
— cw · Jun 23, 06:09 PM · #
Karl:
At a very high level of abstraction, your outline describes how, for example, the inetgrated environment-economics models that are used to generate long-run predictions of climate change impacts work. They ususally go sector-by-sector, but use historical market-based reactions to changes to estimate response functions. A classic simple case is the so-called spontaneous technology improvement function. I think the debate would be “Yes, but couldn’t this response function itself change in this new environemnt?”.
— Jim Manzi · Jun 23, 06:17 PM · #
I don’t see that the situation is as symmetrical as this post posits. Consider the following analogy. Suppose that we compare the costs of stronger building codes in earthquake-prone areas versus the costs of damages due to earthquakes.
It’s not clear to me that markets can affect the latter to the same extent that they can the former. It wouldn’t surprise me if the estimates of building code costs were too large, partly for the reason that Brad says, but for other reasons as well. But the cost of earthquake damage seems much more closely tied to the physical properties of the buildings, the frequency of earthquakes, etc. Once a building is built, it will either collapse in an earthquake or it won’t, no matter what the market wants.
I suppose one could argue that the market will itself predict the proper engineering for buildings, but if that were true, we wouldn’t ever have market bubbles.
— Chris · Jun 23, 08:06 PM · #
Never mind. I see that I have misread this post.
— cw · Jun 23, 10:52 PM · #