A Lesson?
It is painfully evident that the financial engineering profession–of which I am a member, at least of the academic branch–created models that were painfully inadequate for the purpose of valuing, and appraising the risks of, highly credit sensitive mortgage derivatives. The real products were too complicated, too non-linear, too dependent on unknown and arguably unknowable parameters to value accurately.
Thinking about that sparked thoughts of a topic that I blogged about extensively some time back–climate science and climate models–but which I have largely ignored since. Financial engineering and climate modeling have a lot in common. Both attempt to characterize sometimes highly non-linear things by solving numerically partial differential equations. The results generated by both types of models are highly dependent on the values of certain parameters that are not directly observable, and often times very difficult to estimate even approximately given the limited data available; hence, parameter inputs for both types of models are often guesstimates, at best.
The pronounced deficiencies of financial engineering models, especially for credit sensitive mortgage products may hold a lesson for climate science, and for policymakers who base (directly or indirectly) decisions affecting literally trillions of dollars of wealth on the predictions of these models. That lesson is a simple one: be very skeptical about the predictions of these models. The one-eyed man is king in the land of the blind, and having a model is arguably better than not having one, in the sense that one’s predictions are more precise with a model than without one–but that does not mean that the precision is high in any absolute sense. Indeed, the very elegance and deceptive precision of the models–they produce, after all, numbers to machine precision–can make them extremely alluring. The models are so much more elegant, neater, ordered, and logical than messy reality. Thus, modelers too often tend to fall in love with what they have created. The financial crisis makes manifest the dangers inherent in this attraction.
If anything, climate and climate models are more complicated than CDOs and the models to value them (though that proposition may be arguable, given than climate lacks the strategic, and potentially psychological, aspect that is present in any situation involving human interaction, such as a financial market.) Thus, the stunning demonstration of model error currently playing out in the financial markets should give pause to anyone who makes sweeping policy recommendations based on the predictions of climate models. Not that I expect that it will.
Not a lesson at all.
Green techs adopted, little effects of global warming – you lose a few hundreds of billions of dollars and reduce pollution
Green techs not adopted, little effects of global warming – you save a bit of money (those which aren’t spent invading countries for more oil and mitigating the effects of diminishing hydrocarbons resources)
Green techs adopted, large effects of global warming – you lose a few hundreds of billions of dollars and save trillions of dollars and hundreds of millions of lives
Green techs not adopted, large effects of global warming – you save some money, then lose trillions in adjustment to climate changes in rich countries (Florida going under the sea, the South-West and Great Plains turning into a desert, invading Canada for Lebensraum, etc) and hundreds of millions of lives in poor countries which don’t have the resources to mitigate the effects of GW.
Considering the scale of the problem and the huge amount of evidence for climate change (not only models – also historical records!), I know which option (hopefully) most sane, ethical people will prefer.
Actually, you of all people should support action of climate change because Russia will be one of the very few countries to actually benefit from GW! 🙂
Comment by Da Russophile — October 18, 2008 @ 11:03 pm
Add this to our metastasizing agree-to-disagree file. Re “huge amount of evidence for climate change”: I agree that climate changes all the time, and that there is massive evidence for that. The question is what is the anthropogenic contribution to that change. There the evidence is more ambiguous, to say the least, and the common anthrocentric assumption is that if the climate changes, it must be due to us. Color me skeptical.
Re the models, (a) I think it is indisputable that they are highly parameterized, that the predictions are highly sensitive to these parameters, and modelers have no real clue as to the correct values of many of the most crucial parameters (e.g., clouds), and (b) the predictions of the state of the art models are at odds with the data (esp., tropical temperatures.)
I think nitpicking the models is overdoing it. The way I see things is:
1. CO2 levels are the highest they’ve been in the past half a million years (according to very reliable ice core measurements)
2. CO2 levels have shot up so fast in the past few decades that on a geologic timescale they look like something close to a singularity
3. These decades have featured massive and increasing CO2 anthropogenic CO2 emissions. What are the chances that this happens to be a natural thing, just in the exact time period that humans have dotted the world with factories, cars and industrial farms?
4. Historic data indicates an extremely close correlation between CO2 levels and mean global temperatures.
5. Historic data indicates that even with just the current CO2 levels the world should be 2-3C warmer. Delays in the climate system and the “global dimming” phenomenon indicate that the close to 1C warming we have seen in the past century is far from over.
6. It also neglects the fact that even a little warming could destabilize CO2 sinks and convert them into CO2 sources in uncontrolled feedback mechanisms, and ignores that most projections see CO2 emissions expanding rapidly well into the next few decades.
7. Ironically, to an extent I agree with you that there is perhaps ultimately not that much point in spending money trying to curb CO2 emissions, but not because of concerns about whether the models are right but because I think we’re already pretty much screwed, and that those resources would be better used to develop planetary climate engineering projects that are the sole hope for future climate stability.
Comment by Da Russophile — October 19, 2008 @ 10:46 pm