Is Economics a Science? (Of course it is...)
What follows is an unpublished letter to the Economist, responding to their negative conclusion to the above question...:-)
In "News from the lab" (May 8, 1999 ) it is flattering to read that Kagel and Roth ( Handbook of Experimental Economics) is "the indispensable reference" on experimental economics. But it is distressing to read that because "...unlike physics, economics yields no natural laws or universal constants" it follows that "...with or without experiments, economics is not and never can be a proper science." By this definition, astronomy, geology, biology, perhaps parts of physics itself, and certainly psychology must not be proper sciences either.
Rather than quibbling about definitions, it may help to consider how laboratory experiments complement other kinds of investigation in economics, as they do in those other sciences. Let me give an example.
One strategy for looking at field data (as opposed to laboratory data) is to search out "natural experiments," namely comparable sets of observations that differ in only one critical factor. The benefit of using field data is that we are directly studying markets and behavior we are interested in, but the disadvantage is that in natural markets we can seldom find comparisons that permit sharp tests of economic theory.
In a 1990 paper (in the informatively named journal, Science) I studied such a natural experiment, involving the markets for new physicians in different regions of the U.K. in the 1970's. The markets in Edinburgh and Cardiff succeeded while those in Newcastle and Birmingham failed, in ways that can be explained by how these markets were organized. But as will be apparent to readers of the Economist, there are other differences than market organization between Edinburgh and Cardiff on the one hand and Newcastle and Birmingham on the other. So, how are we to know that the difference in market organization, and not those other differences, accounts for the success and failure of the markets?
One way to approach this question is with a laboratory experiment. In a paper in the Quarterly Journal of Economics, John Kagel and I report such an experiment, in which we study small, artificial markets that differ only in whether they are organized as in Edinburgh and Cardiff or as in Newcastle and Birmingham. Unlike in those naturally occurring markets, the market organization is the only difference between our laboratory markets. And our laboratory results reproduce, on a smaller scale and despite far smaller incentives, the results we see in the natural markets. So the experiments show that the differences in market organization by themselves can have the predicted consequences.
Does this "prove" to a mathematical certainty that the different market organizations are the cause of the differences observed in the natural markets? Of course not. Does it provide powerful additional evidence in favor of that hypothesis? Of course it does.
· Roth, A.E. "New Physicians: A Natural Experiment in Market Organization," Science, 250, 1990, 1524-1528.
· Kagel, John H. and A.E. Roth, "The dynamics of reorganization in matching markets: A laboratory experiment motivated by a natural experiment," Quarterly Journal of Economics, February, 2000, 201-235.
· The Handbook of Experimental Economics, John H. Kagel and Alvin E. Roth, editors, Princeton University Press, 1995. Paperback edition, Fall 1997.
Charlie Plott, at Cal Tech, also wrote an (also unpublished) letter in response to the same piece: at his suggestion it is posted here.