The "general laws" that some claim economics has discovered are often either obvious, or wrong. Sometimes both. In Brian's economic science post there's the following example:
"The banality of economic "predictions" is especially vivid in Posner's claim that "an increase in the severity of punishment will (ceteris paribus) reduce the amount of crime." But surely we don't need economics--or any putative social science--to draw that conclusion!"
The conclusion is not only obvious but will sometimes be wrong -- punishments that are so severe as to violate someone's sense of justice may make them more likely to engage in the offending behavior. (If you give a severe beating to your teenager for bringing home a "B" on his report card, he may respond by running away and dropping out of school altogether).
I think you go wrong when you look to economics for general laws, even when you sprinkle in a liberal helping of ceteris paribus clauses. But I still find economic theory immensely valuable. Mathematical economic models have a lot of strengths as a way to organize a causal story about social processes. Econ is rigorous about modelling the interactions of large numbers of opportunistic individuals, and deriving predictions about the unintended emergent properties of those interactions. These kinds of emergent properties are pretty non-intuitive (unless you are Adam Smith). So econ is great training in interpreting and getting a feel for social systems, especially those involving markets (the collective system that economists understand by far the best).
What people call the "general laws" of economics -- demand curves slope down, supply curves slope up, etc. -- are generally instead the modelling assumptions that are being made to impose mathematical form on social interaction, and thus to derive particular predictions. Those predictions themselves are limited in usefulness -- they are general, qualitative and directional rather than specific and quantitative, and the level of abstraction economics is conducted at means they are wrong in specific cases on a fairly frequent basis. They also tend to be extremely difficult to genuinely test empirically. The fact that there is rarely a quantitative magnitude in the prediction means that a failed empirical test looks identical to a correctly predicted effect that just turned out to be small.
But I would argue that going through the modelling process still creates a fairly substantial improvement on untutored common sense. Even when the predictions are wrong they usually have some validity, in that they alert you to genuine forces operating in the system, give you a feel for what to look for. (Social systems rarely reach the "frictionless" equilibriums hypothesized by economists, but the pressures toward equilibrium still exist).
I do think that the standard market model in economics produces a wide variety of non-obvious predictions that have turned out to have at least some empirical force -- predictions about tax incidence (e.g. given reasonable labor supply assumptions taxes on employers are paid by workers), commodification and falling profit rates as industries mature, greater efficiency of market-organized production and distribution of goods (countries that have abandoned markets entirely have not had good economic outcomes). They are don't always come true (e.g. successful brands can prevent products from being commodified as firms age) but they point you toward pressures that are really operative.
Of course modelling in economics has moved far beyond the familiar standard market model over the last few decades. One could argue that this movement has almost gone in the opposite direction from "useful prediction". As modellers took apart the standard model and varied its assumptions, they demonstrated that a wide range of predictions can be rationalized by varying theoretical assumptions in defensible ways. Just by making some quite reasonable assumptions about information, transaction costs, market frictions, or returns to scale, you can get results that look very "unreasonable" according to standard market models. Identically skilled workers can be paid different wages in a competitive market equilibrium. Some workers might not be paid a compensatory benefit for taking a dirty or dangerous job. Free trade might not be mutually beneficial. That doesn't mean market logic doesn't continue to have validity -- market forces are still very operative in these models -- but it does mean that the range of possible real world outcomes compatible with theory is greatly expanded. (Economists are perhaps overly quick to downplay these results "in public", or just jump to the conclusion they are not applicable, in order to maintain disciplinary clout behind the predictions they feel most comfortable with).
In some sense these new models may have expanded the space of possible theoretical outcomes beyond our empirical capacity to test them. But I don't think this is necessarily an unhealthy development, since I think social processes are too complex to allow for a simple model of definitive theory testing as we see in certain hard sciences. These models can be valuable tools for describing and thinking about social processes if combined with close measurement and the proper types of auxiliary hypotheses. But this might happen in a way that brings economics closer to being a highly mathematical, internally rigorous, measurement-oriented version of more "narrative" social sciences, rather than a hard science along the lines of a social physics.
Marcus Stanley
Recent Comments