By Graham Peterson
An old complaint from economics about verbal social theory is that, changes in behavior are predicated on changes in preferences. And as the by now pretentious cliche goes De Gustibus non est Disputandum. That which explains everything explains nothing. Taste based explanations offer researchers, as George Stigler put it, “endless degrees of freedom.” Quite literally: a regression equation for which each of k variables on the right hand side constitutes a taste that is specific and idiosyncratic to each individual of n individuals, n = k and the equation predicts perfectly every data point.
Tautologies are useless.
This is a point that should harmonize with qualitative, comparative researchers as well. The comparative researcher looks for similar elements within categories, and thence searches for differences among those categories in order to suggest how weakly or strongly the common attribute actually correlates to other attributes of those categories. Or to approach matters deductively, he claims X should appear in all of Y categories, which can therefore be disproved by a single differing case. Either way, one cannot merely tack a new taste onto the original hypothesis every time one encounters differences.
Tautologies are useless.
So nobody wants ad hoc theory. The question is: when does updating or amending an original research hypothesis or mechanism become ad hoc?
I think the answer to that question is decidedly not “when we start talking about tastes,” as Gary Becker (God rest his soul) and George Stigler said. Theory begins to become ad hoc when, regardless whether one is using an English alphabet or Bourbaki symbols, one begins to add to one’s theory in order to explain additional variation without:
- using a consistent and transparently stated criterion of the empirical threshold after which an original hypothesis is to be accepted or rejected, and then
- adding as little as possible to the amended theory in order to reduce the number of dimensions along which new empirics must be taken in order to deal with the additional variation
The key here is to be conservative, I think. You can always add more salt, but you can’t take it back out. You can always cut more hair off, but you can’t glue it back on. And you can always add more variation to your theory, but you can’t disentangle a rats nest once you’ve got one.
This would seem to be how we get a handle on our criterion for belief, remain clear and communicable, and enjoy parsimony. Science only requires us to hold constant or get some threshold of control on variation and isolate causal pathways. And that means that we can, and indeed should often, hold changes in relative prices and capital and income endowments constant, varying tastes — the reverse of the usual economic approach.
Becker and Stigler claimed in their paper that apparent changes in tastes can be accounted for by investment in skill and item specific capital. You get addicted to music you like or heroin because you invest in capital that makes the knowledge and techniques you would need to enjoy other things more expensive in relative terms. This moves the problem into the budget constraint. You get “addicted” to certain collections of behaviors we call “traditions and customs” because you have environment-specific capital that makes responding to changes in the environment expensive, and your substitution to investment in new “customs capital” relatively expensive.
But I argue in the reverse, that we can start from taste based analyses, and get back to stable preferences in the economic sense, so taste based analyses are not a threat to price theory, nor is price theory a threat to taste based explanations. For a variety of reasons it is clear to structuralists that social structures are quite stable in the long run. If one combines this with the classic idea in economic sociology that tastes are embedded, and drawn from these stable social structures, we can see the stability and relative homogeneity of individual level preferences that have always made economics tick, as a function of social structure.
It is not ad hoc, thus, in either the economic or sociological analysis, to carefully update and amend the basic framework, because the basic frameworks are both indeed rather parsimonious and appear to imply one another’s basic assumptions quite well, and both theories imply the same approach to validating them empirically.