Thursday 4 April 2013

Culture and agents

Frances Woolley very nicely points out the problem with hand-wavy references to 'culture' as explanation: by explaining everything, it explains nothing. She also points out that it violates economists' methodological individualism:
"Culture" as an explanation also violates what Eric Crampton calls "the first rule of the microeconomists club": methodological individualism. A good economic explanation starts with the choices of individual rational actors. It begins with the premise that people are, fundamentally, all alike - we want social status, reproductive success (or sex, anyways), economic security. What looks like differences in tastes are, in fact, differences in ways of achieving those fundamental goals that we all share. [emphasis added]
Frances's point here is important. I'm not sure that I'm 100% committed to the idea that we all do in fact share fundamental goals; we at least put pretty different weights on the different goals. I'm not even sure that I'm committed to the idea that individual preferences are stable or even knowable outside of the context of choice. But people having different constraints and production functions for similar base goals is a much better starting point than looking at the outcomes of others' optimisations and deeming them to be doing a bad job of fulfilling their own utility functions.

Rumplestatskin at Macrobusiness read Frances as saying culture can't matter, then defended the study of culture in economics. He concluded:
It is genuinely unfortunate that to be taken seriously by the economics ‘in crowd’ you have to start with a theory that ignores time, has a unique equilibrium, and aggregates all agents into one.  There are so many avenues of research being pursued at the fringe of the profession, and within other disciplines, such as sociology anthropology, and psychology, that have so much to offer in terms of explaining and predicting social and economic phenomena by happily embracing cultural explanations.
I'm not sure I follow Rumplestatskin's path from "methodological individualism" to "ignores time, has a unique equilibrium, and aggregates all agents into one." There are plenty of models of that sort, but they're hardly required; they're just more tractable. Hayek was one of the biggest proponents of methodological individualism and was hardly this kind of theorist. And I tend to think representative agent models when taken too seriously can do harm: it can be too tempting to assume that deviations from the modeled path stem from a policy-relevant irrationality or stupidity rather than from differences in individual production technologies or utility functions that we'd assumed away for tractability.

Further, Frances was simply saying that 'culture' can't be the sum of one's explanation. As she'd written:
The Becker-Stigler view does not imply that culture or preferences are unimportant. Rather, it is rallying cry for economic imperialists, a call for economists to take culture, preferences, and all of the other phenomena ignored by previous generations, and explain them using the tools of rational choice theory.
According to this approach, "culture" is the beginning of an explanation, not the end. Why is there son preference is some places and not in others? What are the advantages and disadvantages, the costs and benefits, of sons? How can son preference persist? 

3 comments:

  1. Nice post - not sure how I missed the initial Woolley post. I was writing about something similar, I'll be sure to lean on these posts ;)

    ReplyDelete
  2. Eric, thanks for this. I left a comment on Rumplestatskin's blog too. The title of my post - "Why culture is a lousy explanation" - probably overstates my position. I agree with you that there may be times when it is inappropriate to say everyone has basically the same preferences - when analyzing gender, for example (there's a whole school of thought called "difference feminism" that takes as its starting position that men and women are different.)

    ReplyDelete
  3. Let me know how it goes.

    ReplyDelete