orgtheory.net

what economists should learn from sociology

Fabio

I was giving a talk about my research on the politics of the anti-war movement at the GMU econ department’s public choice seminar. At lunch, my good friend Bryan Caplan and his colleague Ilea Rainer asked me: what are five things that economists should learn from sociology? We didn’t get through all five, but here are three that came up:

  1. Sociologists have a deep appreciation of imitation and conformity as a basic feature of human behavior. Economists rarely model this explicitly. If it is so important, as sociologists have shown, then economists are really missing the boat.
  2. Social networks/social structure matters. Simple idea but few economists sit around and model the effects of social structure. Usually the best you get is a recognition that “norms” matter. My argument is that economists should care because social structures produce opportunities (e.g., you hear about jobs through networks). Most models assume that information/resources are out there but few wonder or model how networks/institutions create them.
  3. Collect some real data. Most economists are “data downloaders.” They are afraid to get out there and survey or observe people, which limits your questions to what major data sets have asked. Experimental work is an improvement, but it is still a highly artificial construction. Go out and observe some markets with your eyes and get your hands dirty. Live a little!

What would you add?

Advertisements

Written by fabiorojas

March 7, 2007 at 2:28 am

20 Responses

Subscribe to comments with RSS.

  1. […] from the other Fabio at orgtheory tells what economists should learn from sociologists, while Mark Thoma informs that neuroscientists are borrowing from […]

    Like

  2. In 2005 the Journal of Economic Perspectives had an issue about the potential contributions of economic sociology to economics. Mark Granovetter, Mike Hannan, Barbara Reskin, and Denise Bielby all contributed to the article. Robert Gibbons provides a nice introduction titled “What is Economic Sociology and Should Any Economists Care?” (although he considered titling the paper “Why Robert Merton is my Hero”).

    The most surprising paper (“Identity and the Economics of Organizations”) in the issue comes from economists George Akerlof and Rachel Kranton, who argue that identity is a mechanism missing from most economic theories. Organizations, they say, “place workers into jobs with which they identify and the creation of such identities are central to what makes organizations work” (pg. 11). Identity can explain why commitments form in organizations that provide weak monetary incentives, like the military, or may contribute to middle-level productivity as witnessed in machine shops, like those described by Roy and Burawoy. The Akerlof and Kranton paper just scratches the surface though. Identity, as conceived by sociologists and organizational scholars, is multifaceted and operates at multiple levels (e.g. group identity versus organizational identity).

    Like

    brayden

    March 7, 2007 at 3:50 pm

  3. Yes, identity is a good one. Though I’ve been critical of how Akerlof and Kranton apply the idea in some cases, I think they’ve made important steps in bridging sociology and economics. You also get this in the economics of religion. You can re-interpret arguments about club goods as arguments about identities.

    Like

    Fabio Rojas

    March 7, 2007 at 3:56 pm

  4. Fabio, your point about data collection is a fair one, but I think for economists the critical distinction is not between data you collect yourself and data you download, but between “objective” and “subjective” data. Economists are perfectly happy to collect data by hand (assuming, of course, that the marginal benefits of doing so exceed the marginal costs), but they are very wary of survey data that collect respondents’ subjective perceptions of things. This comes from the economist’s belief in revealed preference or demonstrated preference: It’s not what you say you’ll do that matters, but what you actually do. Hence the reliance on data on goods and services actually delivered, prices actually paid, contracts actually signed, organizational forms actually chosen, and so on. Likert-scale data on what people say they like, or say they do, isn’t as good.

    In the management literature you see lots of papers on the performance of particular contractual arrangements, strategies, and so on. To get the dependent variable, the researcher simply asks. “Rate your satisfaction with this transaction on a 1-to-7 scale.” This strikes most economists as bizarre. The only way to find out if the transaction “performed” is to see if it was repeated, or if profits went up, etc. How well the manager thinks it performed is beside the point. (Of course, I’m sure this strikes the sociologists as equally bizarre.)

    Like

    Peter Klein

    March 8, 2007 at 6:54 pm

  5. Hi, Peter. I agree it is important to appreciate revealed preference. But my comment was more about taking numerical data from surveys at face value, not so much about the differences between attitudes/opinions vs. observed behavior. [Aside – smart field researchers know the difference and the write up is often about the disconnect between statements and actions.]

    Here’s a simple example: A little while back in sociology, a number of authors wrote on the pricing of legal services based on published price data (e.g., how much your charge / hour). What was interesting was that this data is highly ambiguous. If you actually hang out at law firms, you learn that the same recorded price is often attached to very different behaviors. For example, say client A requires 2 hours of research @ $100 per hour. Then B comes along and asks for the same service. Since you’ve already done the work, it only takes you 1/2 hour. So do you charge $200 or $50? If you hang out with lawyers, you learn it depends a lot on the client and your relationship.

    My point is simple: any researcher who takes published price data on its face value runs the risk of missing very important behavior. In some cases, it probably doesn’t matter. But in the case above, it might matter quite a bit. And the only way you would find out is actually getting off your butt and into the field!

    Like

    Fabio Rojas

    March 8, 2007 at 8:28 pm

  6. That is a good point. You sound almost like a Coasian!

    Like

    Peter Klein

    March 8, 2007 at 9:20 pm

  7. Exactly. The over reliance on secondary data fails to take into account many of the societal changes, which will come about only by a ‘qualitative research’ supplemented by a quantitative one rather than it being only reduced to a quantitative one.

    Like

    Alex M Thomas

    March 13, 2007 at 1:04 pm

  8. #3 probably fits well with this quote I saw yesterday:

    “”The government are very keen on amassing statistics. They collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But you must never forget that every one of these figures comes in the first instance from the village watchman, who just puts down what he damn pleases.” (quoting an anonymous English judge.)”

    Like

    scott cunningham

    March 13, 2007 at 3:31 pm

  9. […] Comments scott cunningham on what economists should learn from sociologyAlex M Thomas on what economists should learn from sociologyPeter Klein on organizations as […]

    Like

  10. My #4 is slightly different from #2, ie, economists need (!) to spend more time on the dynamics of interaction. Formal models can’t handle it; experiments look at outcomes. There is a LOT going on when people negotiate, communicate or even walk down a busy street. The blinders of a model (or calculus!) keep us from exploring, appreciating and using those complex interactions in theories with testable hypotheses.

    Like

    David Zetland

    March 14, 2007 at 3:06 am

  11. Isn’t there quite a body of economic literature on why people like things like celebrities, and on bubbles, both of which are effects of social networks, and crowd behaviour?

    Like

    Marcin

    March 15, 2007 at 12:19 pm

  12. On #1, this appears to provide support precisely for economists’ methodological individualistic approach: “representative agents” and the Becker-Stigler rejection of idiosyncratic tastes to explain outcomes.

    Like

    Jack P

    March 15, 2007 at 2:30 pm

  13. Marcin: You are correct that there is a literature on bubbles and more generaly information cascades, but you see very little modelling of social networks. For example, in sociology, it is a basic procedure to ask who is connected to whom in a group and then to answer questions about social structure with that data. This is almost never done in economics. Usually, economsits assume the network is out there and fail to model it.

    Like

    Fabio Rojas

    March 15, 2007 at 3:05 pm

  14. What about Matthew O. Jackson and co-authors? I realize his papers on the topic are

    Like

    Jack P

    March 15, 2007 at 3:14 pm

  15. […] wonderful weblog orgtheory.net has two posts with “quick, highly-stylized” on what economists should learn from sociologists, and what sociologists should learn from economic reasoning (via Marginal Revolution). From the […]

    Like

  16. Dr. Rojas, your conviction about field research and qualitative data collection cannot be emphasized enough. However, I do agree with Dr. Klein’s skepticism about the utility of survey research.

    I believe that the combination of field research and “official data” will bring social science closest to understanding human action.

    Like

    Brian Pitt

    March 18, 2007 at 9:07 pm

  17. Hi Brian – I agree with the general thrust of your point. Field research + “hard” numbers can be enormously useful, but I wouldn’t underestimate surveys either. First, surveys, if nothing else, measure what people want you to believe about them. That’s pretty important in my book. If you believe that voter opinions drive policy, knowing how people see the world seems important.

    Second, surveys can be easily manipulated to create experiments, which are hard to do in field research or with official numbers.

    Third, you can do quality control with surveys, while official numbers can be biased in really bizarre ways that you can’t fix.

    Surveys have their uses!

    Like

    Fabio Rojas

    March 18, 2007 at 9:17 pm

  18. Excellent points!

    But I must ask, does your optimism about survey data extend to multinomial logit regressions that seek to predict increases in Likert-style dependent variables (e.g., regressing race on how much one likes G.H. Bush)?

    I think I should have applied to Indiana!

    Like

    Brian Pitt

    March 19, 2007 at 1:26 am

  19. Hi, Brian. My optimism about surveys depends mainly on the quality of the survey. I’ve recently completed a working paper on attitudes from a survey I conducted. I don’t use the attitudes by themselves, but I use in them with data about actual behavior.

    In my case, I was asking about changing boundaries within an academic profession. So I did do the sort of regression of what predicts attitudes, but I also did regressions to see if the same factors predict classroom behavior (i.e., if they assigned books that promote the ideas they claim to believe).

    And, yes, you should have applied to Indiana!

    Like

    Fabio Rojas

    March 19, 2007 at 3:06 pm

  20. […] while back, I wrote a post called “what economists should learn from sociology.” Consider this a follow up post – what sociologists can learn from economists. Let’s […]

    Like


Comments are closed.

%d bloggers like this: