The Limitations of Economics and Models

In an interesting bit of synchronicity, an Economist piece today on the advances and limitations of economics as a discipline contrasts with a news item on the pseudo science of certain types of hucksterism, eh, modeling.

The Economist’s Free Exchange section discusses the limitations of the dismal science in being able to assess the possible costs of global warming:

GLOBAL warming sceptics often urged the world to wait until the science got better. This cry has abated somewhat as better science has confirmed the scope of the problem. But after spending the weekend at the American Economics Association, with side conversations into the Stern Report, I wonder if the same might be said to economists trying to put a cost on future global warming.

I came away from AEA with a renewed sense of wonder about the rapid improvements in economic science. Behavioural economics and neuroeconomics are revolutionising micro; experimental economics is pushing frontiers out in both micro and macro. Better data is finally allowing economists to test a wide range of common theoretical assumptions; I had a fascinating discussion yesterday with a fellow who is challenging the traditional view of switching costs and pricing.

But the richness of the change is making it clear how much there is still to learn. At a panel on housing prices, one of the discussants pointed out that housing economics is a field that should be at least as big as finance, given how much it matters to the economy, and yet they don’t even have basic models for things like land. No one even knows how to build such a model–what do you do, superimpose it on a map? Housing economics is still awaiting its Modigliani-Miller, its Black-Scholes, and so forth.

The questions that the Stern Report attempts to answer about future utility and so forth are essentially insoluble; given the number of variables and intangibles, it’s hard to see how they could be much better than random. That’s why the chocie of discount rate has to do most of the heavy lifting. But perhaps this just gives us a sense of false precision. Perhaps we should simply go with the intuitive reaction to global warming, which is that it’s bad and we should try hard to stop it, maybe up to 10% of income worth. As the science gets better, it can fill in. After all, we manage to have housing markets even without a good model of land prices.

As an aside, there are people who know how to model land prices, and they aren’t economists. A physicist I know realized one of his algorithms could be used to forecast urban development and used it to decide where and how much to bet in Las Vegas. He made $100 million (personally, mind you, his partners made even more). With profit potential like that, why should they share their findings?

But back to the real issue. In global warming, we have a problem that has so many variables that the outcome is hostage to the assumptions. This has dire consequences: it means the debate can be muddied by people who have a lot to lose, say, from the imposition of a carbon tax. The author of the piece argues it is better to get on and tackle the problem first, because any attempt to get it mathematically right (at least at this juncture) will be spurious.

The contrasting story du jour shows how easily people can be fooled by faux science. One film industry hedge fund (a dubious proposition to begin with, since movies as a whole deliver T-bill like returns with considerably more risk), Stark Investments, marketed its deals claiming it could estimate how much to invest in specific projects. Here is Bloomberg (via ) on the fund’s partners Benjamin Waisbren and Jack Kavanaugh:

Most of Waisbren’s plan, particularly the part about being smarter about borrowing money to reduce the cost of capital sounds pretty good. But the description of Waisbren’s partner Jack Kavanaugh using investment banking style business modeling techniques is pretty entertaining:

Like Waisbren, Kavanaugh has devised a plan to make Hollywood pay for investors. Taking advantage of the relationships he developed investing on behalf of Hollywood heavies, Kavanaugh compiled as much proprietary historical data as he could about movie returns.

He attaches a numerical value to each participant in a film — the talent, the director, the screenwriter, the producer and so on — based on their historical returns, and then he runs the numbers through Monte Carlo simulations to figure out how much the movie will sell for in foreign territories and, ultimately, the most prudent amount to invest in the budget.

The methodology sounds pretty dubious (too many dependent variables for any regression to be valid) and the proof is in the pudding. The model led the fund to invest heavily in such box-office disappointments as the Poseidon remake and “V for Vendetta.”

But it gets back to the same issue implied in the Economist piece: people hold math in high regard, even though the mathematical representation of any situation is an abstraction, and it’s hard to know ex ante how accurately it will capture the elements you care most about. But investors and policy makers seem to prefer the risk of being precisely wrong to that of being directionally correct but not able to pin a number to it.

Print Friendly, PDF & Email