Sunday, May 22, 2011

Diagrams in Economics: They Actually Aren't All That

Via Mark Thoma, Daniel Little has an interesting post on the history of diagrams in economics.

One thing that I find interesting is the extent to which mathematical diagrams have been such a dominant part of the toolbox of many economists, especially in modern times. That classical economists did not have access to these modern tools; in many ways, the classical economists had an advantage as a result.

It hardly needs to be said that mathematically precise diagrams are not scientific. They are not empirical, but merely describe a situation assuming very rigid premises are always true. They are most useful for engaging in thought experiments and, perhaps, helping to clarify thinking. (Although there is the risk of the opposite, since not everything that is important is necessarily represented in the model!)

It shouldn't need to be said that doing nuanced mathematical proofs about nuanced details contained in such diagrams are going to tend to be useless to the extent that the truth of such nuanced details depends on false assumptions. But economists waste a lot of time trying to prove such little details about diagrams... This time would probably be better spent contemplating and understanding the real world. In this way, modern economics can be somewhat inefficient; but you would think that a field that claims to obsess over optimality and efficiency, the way economics does, would be more efficient.

Milton Friedman defended models with untrue premises by suggesting that it does not matter how unrealistic the assumptions of a model are, as long as those models predict accurately. An analogy might be Newton's laws of motion which depend on false assumptions about how the world works compared to quantum mechanics. Nonetheless Newton's laws of motion are practically useful for some applications, especially since it is much easier to intuitively understand and use than quantum mechanics. What matters, or so argues Milton Friedman, is the truth of the predictions produced by the model more than whether the premises built into the model are true.

Well, there is some merit here in Milton Friedman's argument. But it is also excessively apologetic and sweeps far too many problems under the rug. First, this is no different than any other area of life. We ultimately have no choice but to reason from premises that are not perfectly true, because absolute truth is not available to us. For example, in law, we infer intent from actions, because we do not have direct access to someone's intent, which resides in their minds. That does not mean that false premises do not matter, however. When our inferences in law are wrong, we end up giving someone the death penalty when perhaps the appropriate punishment would have been merely some prison time or even probation. In economics, you have economists who risk giving policymakers policy advice that is wrong and therefore harmful.

Indeed, the absence of perfection regarding the truth of premises is not an excuse for ignoring false premises. Economists still should do the best they can to get their premises right. They should obsess at least as much over arriving at empirically correct premises as they obsess over rigid mathematical features of their models that do not really matter much in the real world anyway, to the extent that that house of cards is built on a foundation of false assumptions in any case. What is the point of being greatly detailed about a model when many of those details are based on premises that are not only false, but known to be false? If your goal is to make empirical predictions, the more nuanced mathematical details often are not going to matter anyway. A more efficient use of time would be to try to do better in terms of the truth of the premises that are used in the models in the first place.

Also, regarding Milton Friedman's argument, some economists can be awfully sloppy empiricists, as demonstrated by Noah Smith and Dean Baker here and here. While, it is really not true that one can fully make up for untrue premises by using empirical research to verify predictions, to the extent that economists are going to try to do just that, they must ensure that they enforce much higher standards of empirical rigor.

Finally, the entire Milton Friedman defense assumes that the same approach that is applicable to the much more narrow subject of physics is applicable to the much more complex subject of economics. The law of diminishing marginal returns should have convinced any rational knowledge maximizing economist to diversify their methodological approach long ago. The fact that they haven't just proves that economists, like everyone else, are far from rational. One need not go so far as to say that models are completely useless to make the obvious point that they are not all that or the be all and end all.

It is a large missed opportunity that economics has gone so long without working more on getting basic premises right. Economists ignore so many factors that are obviously important in economic transactions (like culture or social meaning) that they can be said to be rather ignorant of the actual functioning of the economy. Many economists don't study the actual economy; they study a fantasy world which consists of economic models built on demonstrably false premises.

Now, this is not true of all economists. You have the rise of behavioral economics which, finally, tries to work on getting premises about how humans actually behave right instead of proceeding based on known falsehoods concerning rationality. It is, however, still true of too many economists. While I am happy about the rise of behavioral economics, rather than be excessively gushy about this development, a more appropriate response is: "It is about time! How long were you economists going to waste playing in your sandbox before emerging to deal with the complexities of the real world anyway???"

What is most remarkable about economics is how little economists actually know, not how much they know. This is not said to bash on economists. The thing economists are trying to study is incredibly complex. But it is a reason to bash on arrogant economists who proclaim to know more than they actually do. So, I think that Greg Mankiw's advice to those listening to economists, both on the left and the right, is well taken: "If you find an economist who says he knows the answers, listen carefully, but be skeptical of everything you hear."

No comments:

Post a Comment