Beware Of Models

One of the healthier outcomes from the financial crisis was that the scales fell from our eyes when it came to models. Think back to the really good articles and blog posts that exposed the data faults and logical faults of the mathematical models that underpinned the failed strategies of the financial system. We had learned, or so it seemed, to view these falsely airtight theorems with an appropriate amount of skepticism.

And then along came Deepwater Horizon and with it no end of scientific models purporting to demonstrate how the oil spill would doom the Gulf of Mexico and other bodies of water to potentially permanent dormancy. In and of itself, that isn’t surprising, but the fact that those same media outlets that had so justly pointed out the limits of science when applied to finance somehow lost their new found sobriety.

For instance, the WSJ this week had a remarkably good opinion piece on the oil spill and its potential effects. Titled ” How Far Will The Gulf Gusher Spread” it postulated that we would feel the effects of the spill for years and years and that it would spread to Europe and even the Arctic Ocean. The author makes these projections despite starting the article with this observation:

Now the Loop Current is in the news once again. Oil from the Deepwater Horizon gusher—please don’t call it a “spill”—has begun trickling into the current, prompting anxious speculation as to how much will be swept up and where it will be borne. Only a small quantity of surface oil has been seen entering the current, but much more swirls below. Given the complex natures of both petroleum and marine waters, these underwater plumes will be extremely difficult to measure and track.

So, the author says that underwater plumes are “… extremely difficult to measure…” yet he assumes their existence as one of the central points of his argument. Recall that most financial models assumed that housing prices would not fall.

Before you attack, let me say two things. One, I consider the oil spill to be an unmitigated disaster and, as I said, I found this particular article to actually be quite interesting. At the same time, I am not inclined to accept the predictions of scientific models with the same openness as I once had, particularly when the data supporting the model is admittedly sketchy.

Don’t lose that healthy skepticism that you developed during the financial crisis. As we learned, things are infinitely more complicated than most models assume and, while they are worthwhile tools, they need not be treated as scripture. They are frequently wrong.

You can leave a response, or trackback from your own site.

Leave a Reply