Friday, November 26, 2004

Is economics a science?

Over at Leiter Reports, guest bloggers Marcus and Jason Stanley are having an interesting discussion about whether or not economics is a science. Much of this depends on semantics, how is a science defined? If you are like Tyler Cowen then a science is defined by "[producing] empirical knowledge which is subject to [a] process of testing, [broad interpretation], and feedback" as well as maybe having controlled experiments. Certainly there are a number of methodological similarities between economics and other disciplines that are considered sciences, like physics and chemistry. To name a few,
  • Economics makes falsifiable claims, which are then
  • tested empirically
  • and are refined using mathematical models.
But for Brian Leiter, "this isn't, however, what 'science is all about'
on any plausible account." So what is science all about?
It is important to remember that a lot of the credibility of economics depends precisely on its claim to be a science, in the precise sense of generating successful predictions. (Predictive power may be neither a necessary nor sufficient condition for science, but economists generally view it as what makes their discipline "scientific.")
There are two points to make here. First, I don't think that most economists view the ability of economics to generate successful predictions is the valuable part of economics. At least to me, the important part is understanding human behavior in a clear and consistent manner. Some fields are good at making predictions (macro, financial, etc.) But many are not.

The second point here is that Leiter has now shifted the debate from whether economics is a science to whether economics can make good predictions. Despite the fact that he states that "Predictive power may be neither a necessary nor sufficient condition for science." He proceeds to use the remainder of his lengthy post to make three basic points. First, that economics is based on assumptions that are not realistic and thus cannot make predictions about reality. Second, that economics cannot make quantitative predictions, only qualitative. I'm prepared to concede the second, but not without giving a litany of reasons why this is true in economics and not in the natural sciences:

  • On most economic phenonmena, economists can't experiment. Instead economists rely on nature and society to set up experiments. This is a haphazard way of conducting empirical research.
  • Humanity is significantly more complex than, say, a chemical reaction. Everyone wants something different out of society and this makes studying it very difficult.
  • Part of why it is difficult is that so often economists are forced to deal with things that are hard to quantify, like utility and preferences.
  • Furthermore, society will not sit still. Everything is constantly changing, which makes setting up experiments even harder.
  • Very realistic models of economies would not be very pretty. This creates disincentive effects, since most people would like to study elegant and clear models.
  • Physics is lucky enough that very nice models are also very realistic. Economics also has very nice models, but while every science has to deal with complexity, economic complexity is located closer to the heart of economics. Whereas a physicist can model a missile flight very easily, since the complexity only arises when one approaches the extremes of physical behavior, an economist modeling a market must deal with complexity in the very heart of the market.
  • In economics, there are many more important variables to consider. Economists must deal with demand schedules instead of nice vector measurements of velocity and acceleration.
  • Lastly, and most interestingly, what we write affects how people act. The very object of study is affected by how economists study it. Indeed this is the driving force behind rational expectations, and it is a valid one. If economists offer some new insight into the workings of the economy, this changes the perceptions and strategies of economic actors, which in turn changes the way in which the economy functions. No other science has this problem.
On the first point, that economics makes unrealistic assumptions and thus cannot even make qualitative predictions, I refer you to the great Paul Krugman:
But of course there is no such model. Worse yet, not only don't I know of any such model, I don't even know how one might begin to create one. My perplexity is not, I believe, a reflection of my ignorance of political science, or sociology. Economists may make lots of bad predictions, but they do have a method - a systematic way of thinking about the world that is more true than not, that gives them genuine if imperfect expertise. (That is also, of course, why lay commentators and other social scientists tend to hate them). Other social sciences haven't yet found anything remotely equivalent. Oh, there are bits and pieces, and some of them are very exciting; try taking a look at Robert Axelrod's stuff. But basically when it comes to most of the questions that I am really interested in, one man's view is as good as another.

When I was young and naive, I used to imagine that my career as an economist could eventually branch out into one as a general social scientist. And I still love to read and think about the broader questions - a book like Jared Diamond's Guns, Germs, and Steel can keep me happy as a clam for weeks. But the clarity and power of economic analysis can spoil you: once you have a taste of what it means to have a really insightful model, you tend to be inhibited about looser speculations.

Overall, I was not impressed by Leiter's criticisms, they seem to me to be rather banal, which is coincidentally what he thinks about economic theory. After all, his chosen field of analytical philosophy has its own problems with being useful and worthwhile. Is this a case of the pot calling the kettle black?


Post a Comment

<< Home