the new economic imperialism: methods not models

Two economic graduate students affiliated with Duke’s Center for the History of Political Economy have just released a new working paper on the history of quasi-experimental methods in economics. Panhans and Singleton document the dramatic takeoff of the use of techniques like regression discontinuity, difference-in-difference, and instrumental variables in the top economics journals, and most of its subfields. The paper’s a nice introduction to this history, and for readers unfamiliar with the older approaches, sets up a nice quick contrast between a 1970s “structural” approach to the returns to education vs. a 1990s “quasi-experimental” approach. What really struck me, though, was an ending reflection on the nature of economic imperialism.

Panhans and Singleton return to a 1984 article by Stigler on the nature of economic imperialism. Stigler argued that economists had entered four new fields, including history, law, politics, and sociology, and they had brought with them the Chicago doctrine of rationality, e.g. “Marriage is viewed as a rational act by which the partners seek to maximize their incomes relative to remaining single or choosing different mates” (Stigler 1984: 308, cited on p. 20). Panhans and Singleton note that the contemporary discipline of economics has continued its tackling of diverse subjects previously found in other fields, but that the narrow emphasis on maximizing behavior characteristic of the old economic imperialism is gone. They write:

From a superficial point of view, Stigler’s predictions appear prophetic. Not only has work by economists in the applied areas of crime, education, and politics expanded greatly, in large part marking “the applied turn,” the fields of health and development economics are today at the forefront of the profession in prestige and the public view. Further, the disciplinary interactions and applied orientation have been institutionalized in schools of public policy, applied economics, public health, and others that train graduate students for academia, industry, and government and recruit researchers from multiple disciplines. However, there appears to be a key distinction with how this imperialism proceeds: rather than the application of a behavioral model of purposive goal-seeking, “economic” analysis is increasingly the empirical investigation of causal effects for which the quasi-experimental toolkit is essential. To return to Stigler’s analogy, the missionary’s Bible is less Mas-Colell et al. (1995)* and more Mostly Harmless Econometrics (Angrist and Pischke 2009).

What do you think? And if Panhans and Singleton are right, what does that mean for debates about the relationship of sociology (or political science, etc.) to economics? Do we have much to fear if most of what economists bring to the table are a fancy, but easily borrowable, set of quasi-experimental tools for analyzing the questions we were already asking? What are the consequences of this new empirical imperialism?

* The dominant textbook for graduate microeconomics.

Author: Dan Hirschman

I am a sociologist interested in the use of numbers in organizations, markets, and policy. For more info, see here.

5 thoughts on “the new economic imperialism: methods not models”

  1. Economists don’t have an exclusive jurisdiction over statistics. This “new economic imperialism” is weak when compared to the older one, the scary one.

    Like

  2. Well, that was a fun paper. It seems like the “threat” in this case is a different one — not one that really challenges sociologists’ view of the world, as Stigler’s was, but one in which our relative lack of quantitative skills (on average, of course) puts us at a disadvantage. Of course, these methods are adoptable. But the undergrad background of most soc grad students means they have to do a lot more work to reach an econ level of quant skills. So maybe this speaks to Noah Smith’s advice that we should “tech up.”

    OTOH, as someone who does, and values, nonquantitative work, I would be sad if you couldn’t enter a soc program without a math or physics degree. (I can’t find it online anymore, but at one point last year I looked at the MIT econ PhD job market candidates and IIRC most had one of the two as undergrad degrees.)

    To what extent has strong identification become a must-have in sociology at present? I don’t follow the relevant subfields closely enough to know. My impression is that while it’s favored and spreading it’s not an absolute requirement. But there’s a lot of room to use this stuff to improve our understanding of things economists might be less interested in. E.g. Hassan El Menyawi’s paper “The Great Reversal,” which looks at change in attitudes toward same-sex practices over three centuries in the Muslim world (http://www.fondationtrudeau.ca/sites/default/files/great_reversal_el_menyawi_2014_asa_0.docx). It uses distance from Mecca to identify changes in the relationship between being a Muslim nation and legality of same-sex practices, and won a whole slew of student paper awards.

    Liked by 2 people

  3. Thanks for sharing this Dan. Its nice to know someone is writing up this bit of intellectual history.

    There is a risk, when we frame these issues as being about “economic imperialism,” that we lose the ability to objectively appraise the substantive issues. I think that when one discipline is influenced by another discipline it is usually “learning” and therefore, all to the good. I think that is definitely true in the specific case of quasi-experimental empirical research techniques.

    Liked by 2 people

  4. Nice paper, but they probably can’t make the point about prevalence without looking at the proportion of papers that have gone that direction. The graphs are all measured in absolute terms.

    I hear a lot more about reduced form modeling than identification when I’m around economists (identification might a taken-for-granted norm and more prevalent, though). As I understand it, reduced form modeling requires one to assume a Cobb-Douglas, or some other form of utility or production function, then estimate its parameters. Still rat choice.

    Interestingly, Pete Leeson just gave a talk at the last Adam Smith Fellows meeting urging a new kind of economic imperialism that roughly equates to the things quants in sociology say to quals — you guys are good for finding interesting things to study (cataloguing descriptions), but you’ll have to leave the causation, mechanisms, and explanations to us.

    I think identifying disciplines as collections of ontological commitments (e.g. to rational choice or all the -isms and -ists and -ians, turns us tribal. It’s probably better to see departments as empirical-context-specific sets of questions for which some methods and theories are better and worse at answering.

    That way we can at least pretend that the data are guiding us rather than our quasi-religious identities and communities.

    Like

  5. ^should have been clearer that I think the idea that some social sciences and methods are just journalistic fodder that’s only good for providing ammunition to the big guns is a stupid argument.

    Like

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s