Sunday, March 08, 2015

Andrew Gelman smacks me down on social science rivalries

Somehow I missed it when it first came out, but back in December Andrew Gelman wrote a great smackdown of a Bloomberg piece by yours truly. I had written that the reason economists get paid much more than sociologists was that they have more technical skills, especially statistics. Gelman retorts:
Smith writes a bunch of things I disagree with — but then, as a political scientist, I guess it’s no surprise to see me disagreeing with an economist!... 
Freudian psychiatrists got a lot of money and a lot of prestige back in the 1950s, and they didn’t know any statistics at all. But they had one useful skill, and that was the ability to convince rich people that their services were personally valuable. In the national discourse, psychiatrists were taken seriously. They were mocked, but they were a central part of the culture and their pronouncements were respected. And they managed to do so without following Smith’s advice to “stop whining and tech up.” 
So perhaps a better piece of advice, if sociologists (and psychologists and political scientists?) want to join economics at the big boys’ table and up their salary and their influence, would be: Stop whining and make your work more appealing to rich people.
Gelman is right. My post left out some big, important things.

First of all, yes, teching up is not necessary for an academic field to get paid big bucks in the private sector (and hence, bigger bucks in the academic sector). Freudian psychology shows that sometimes you can pull off the feat without technical skills.

Second of all, as others have also pointed out since I wrote my post, teching up is not sufficient to get money in the private sector. Your techniques also have to have applications. Gelman terms this "making your work appealing to rich people", but I think usually it's just making your techniques appealing to business (Freudian psychology, with its direct marketing to rich individuals, is the exception rather than the rule). This is why operations research profs get paid more than stats profs. Pretty similar skills, but stats profs would have to pay a higher transaction cost to jump to the private sector - they'd have to learn how to do private-sector stuff. Operations research people already know that stuff, and could switch really quickly. 

Usually, when you have profs with business-related skills, they are in business schools or engineering schools. The less applied disciplines housed in the "literature, science, and arts" schools - physics, math, econ, etc. - are usually paid a lot less. Econ is the biggest exception, and it is not only because of economists' technical skills. 

It's also because economists have figured out ways to make the private sector purchase their techniques directly, without the need for too much retooling. These include, but are not limited to:

1. Legal consulting

2. Macroeconomic forecasting

3. Various assorted other kinds of business consulting

4. Market design (this is new!) 

These, of course, are in addition to at least one hugely important area where economists can apply their skills with retooling: Management consulting.

For sociologists to emulate this financial success completely, it would require that they find ways to get paid for doing sociology. This seems unlikely to happen (though by all means, go for it!). But if they improve their stats techniques so that with a bit of retooling the average sociologist could go to work at a management consultancy or similar firm, they will raise the value of their outside option. At the same time, if they put more stats and applied math in the undergrad sociology curriculum, they will make sociology graduates more appealing to the business world, thus raising demand for the sociology major and probably making it more inelastic at the same time. This was another big thing my Bloomberg post left out.

Whether sociologists want to do this, of course, is another question.

But it's interesting to note that Andrew Gelman's field, political science, has teched itself up a great deal in recent decades, including with game theory (a technique that first found purchase in economics), and with statistics (as Andrew Gelman's own career demonstrates). So even though my prescription for raising sociology salaries is incomplete, it does describe the path that poli sci has chosen to take, and that Gelman himself has chosen to take within poli sci! So that's something.

Gelman's point about Freudian psychology also raises another interesting question. He seems to think that Freudian psychology is (and I am using my own words, not his) basically a bunch of bullshit that rich people were tricked into paying for - a belief that seems very widespread nowadays. But if economists are getting paid to dish out bullshit, what's the bullshit that they're getting paid to dish out?

Gelman claims that Gary Becker's "imperialist" econ stuff is obviously wrong:
Gary Becker’s writing doesn’t scare me, but a lot of it strikes me as wrong, indeed so obviously wrong that it causes me to question how it gets so much respect within the field of economics. I’ve talked with some economists whom I know and respect, and they in turn respect much of Becker’s work, so the story here is far from simple. But let me say this again, my concern about work such as Becker’s — and, I believe, the concern of many other social scientists — is not fear of imperialism, it’s disquiet at such extreme ideas being treated as mainstream.
That's harsh! But even if it were true, I don't think it would be that relevant, since economists don't really get paid to do Beckerian stuff in the private sector.

If economists are getting paid to bullshit, a la Freudian psychologists, who is paying them? Management consultancies might make economists use a teensy tiny little bit of econ theory, but mostly they're just running regressions and doing other stats stuff. As for market design, that ain't bullshit.

That leaves 1. Law and economics, 2. Macroeconomic forecasting, and 3. Various other forms of business consulting.

Law and econ seems like it will inevitably contain some big element of bullshit, because all legal stuff contains large amounts of bullshit. Lawyers are legally obligated to fling whatever bullshit they can fling. But economic considerations are important in legal matters. There's no way around that, really. You could say that the economics being done in legal cases is not the right economics, or not the best, but it's very hard to make an argument that economics in general should just be kicked out of the courtroom.

As for macroeconomics forecasting, that is indeed mostly bullshit, but it's bullshit for which the source of demand - the desire to know the future of the economy - is deep and fundamental. There's no getting rid of this one, I don't think.

That leaves the nebulous 3. Various other forms of business consulting. This is way too nebulous and diffuse to characterize. Some - I have no idea how much - is probably very valuable and useful. But then you have stuff like the corporate shilling in Inside Job. Anyway, if you are pissed at the econ profession, and you think that economists are getting overpaid to do economics for the private sector, it is probably somewhere within this category that you want to look. Good luck with that!


In the comments, Ryan Decker writes:
Probably one of the bigger components of the last category is transfer pricing. I think it would be hard to argue that economists bring nothing of value to that process (though it involves lots of collaboration with accountants and lawyers).
I didn't even realize that economists did much of this, actually! All the people I know who work in transfer pricing are, indeed, accountants and lawyers. But it makes sense that a lot of economists would be hired for this. And yes, I think this is a pretty clear case of creating real value ("real" as opposed to a fleeting fad, an artifact of imperfect institutions, etc. - yes, I know there's no good definition for "real value").

Another chunk of (3) might be strategy consulting by industrial organization economists. It seems like there should be a lot of that, but I have no idea how much there actually is, or how well it actually works.


  1. Probably one of the bigger components of the last category is transfer pricing. I think it would be hard to argue that economists bring nothing of value to that process (though it involves lots of collaboration with accountants and lawyers).

    1. Excellent point. Bumped.

    2. The money for economists in transfer pricing is advising multinationals how to minimize tax liability. The social value is dubious.

    3. Some of it, yeah. But transfer pricing needs to be done *somehow*, otherwise we can't collect tax on multinationals.

    4. Ryan P7:33 PM

      You could go formulary I suppose. Or just take it one level more primitive and do it with capital, labor, and consumption taxes. (Still need some enforcement, but I strongly suspect those are inherently easier. At the very least you can basically do those with accountants & lawyers, whereas TP without economists isn't viable.)

    5. Formulary is the way to go. In tax rules you want to avoid areas that get so complex the well-funded will be able to outgun the IRS.

  2. Anonymous4:03 PM

    As a Sociologist, i find amazing how we basically call out economist for being "sell-outs" and then complain about how little we're getting paid. If we want to be paid more, Academia is really not the way to go. And i ask why be paid more? I don't think any one of us had getting rich in mind when we decided to follow this career. What bothers me more is the lack of demand for our work in ares which we're actually good at, like when an economist is called to analyze an election instead of a Political Scientist, but this is (slowly) changing.

    1. Alvy Singer: [addressing the camera] There's an old joke - um... two elderly women are at a Catskill mountain resort, and one of 'em says, "Boy, the food at this place is really terrible." The other one says, "Yeah, I know; and such small portions."

      Annie Hall

  3. The kind of BS that Gelman's comment suggests to me is not in your list: i.e., afflicting the afflicted and comforting the comfortable. I am thinking of the kind of work put out by CATO, AEI and E(mployment)PI, "justifying" policies that favor the 1%, typically under the banner of economic efficiency.

    1. Yep there is that but it doesn't pay well for most of the people doing it. A few do make bank.

    2. Nathanael1:45 PM

      It pays pretty well for unskilled labor! The interesting thing is that the people paying for it have figured out that proper credentials aren't really necessary for this kind of hackwork, so they're paying the uncredentialed rate.

      They still have to pay quite a bit because the group of people who are willing to be completely unprincipled turns out to be fairly small. THANK GOODNESS.

  4. Noah, about management consulting and sociology, have you considered organizational behavior? These folks do quite a bit of work with firms, and they are sociologists or psychologists. Anthropologists, communications people and others with org-type skills are also in demand. Their fields as a whole have not prospered, however. In part, what needs to be explained is the absence of spillover effects.

    Also, and this relates to psychoanalysis, there is little doubt that economics currently serves an ideological function. In saying this, I’m not denying all the useful, or purportedly useful, stuff that also goes on, but ideology is a part of it. A lot of trade theory is about why certain types of trade arrangements are socially beneficial, a lot of industrial org is about why we do or don’t need particular systems of regulation, similarly for finance, and so forth. Some of these “optimal x” specialists are hired on to help design systems, but most of them simply promote one institutional/policy framework or another. Their function is ideological, and they are effective because economic judgments are held in high esteem. This in turn is a consequence of and contributor to the “superiority of economists”, which pumps up the whole profession. I think other social scientists get jealous or resentful of this.

    1. Yep, OB is an important case to show how marketable skills raise academic salaries!

  5. "For sociologists to emulate this financial success completely, it would require that they find ways to get paid for doing sociology. This seems unlikely to happen (though by all means, go for it!)."

    In a world of Twitter and Facebook (i.e., increasingly digitized social interactions), this seems increasingly _likely_ to happen. The applications for advertising, market research, and product development alone appear vast.

    1. I know psych people get paid for that, bit do soc people also?

    2. The question is not _do they_ but _will they_.

      Duncan Watts is clearly ahead of his time, but I suspect in 20 years most socialogists will be similarly "teched up".

      Another good opportunity are online games. MMOs already hire economists to design their internal markets, as they get more elaborate and engrossing I anticipate they will hire sociologists to design their internal community structure.

    3. Iiiiinteresting.

    4. Yep. Jure Leskovec is also impressive.

      You could say, hey, Jure is in a computer science department ... but computers are eating everything, and I think a substantial fraction of the future sociology department and curriculum looks like Jure.

  6. "But it's interesting to note that Andrew Gelman's field, political science,"

    Well, Gelman does statistical work applied to political science, but he got his PhD in statistics and at Harvard, and with B.A.s in math and physics, and he has a joint appointment at Columbia in statistics and political science, and his first academic position was statistics.

    He's a super high tech statistician first, not a typical political science professor at all. In fact, I remember him from my PhD days as a co-author of a high tech Bayesian book that I liked, "Bayesian Data Analysis".

    But he is right the pay has to do with more than tech. Accounting is a good example, from what I saw a lot less high tech than econ, but much higher pay. Tech, though, is a factor. It does lower the supply of people willing to become econ profs.

  7. Also, in Gelman's post, he criticizes a graph:

    "— What’s with the y-axis labels? Every 20,000 dollars on the odd numbers? That’s just weird."

    Why do you think he doesn't like that to a seemingly large degree? I guess people prefer even numbers, but $50,000 seems like a good starting point here.

  8. Anonymous10:52 AM

    I generally believe that "appealing to rich people" is identically equal to "appealing to the business world" for all intents and purposes.

  9. In the olden days, people majored in business because they weren't bright enough for a serious subject. That may be less true these days; but outside of technical specialists, the business folks I've dealt with are not a sophisticated lot, especially when it comes to mathematics or statistics. I used to write marketing copy for optimization software and can report for that for the average manager, linear programming might as well be string theory. Under the circumstances, waving equations at these guys made sense even if the equations didn't do anything. Psychoanalysts aren't the only witch doctors that ever made a living impressing the natives with mumbo jumbo.

  10. Nathanael1:41 PM

    "But if economists are getting paid to dish out bullshit, what's the bullshit that they're getting paid to dish out?"

    Arguments for why rich people should pay lower taxes, and be allowed to do whatever they want without regulation. Duuuuuh.

    1. Which involves disguising or denying the role of power, especially that which underlies economic rent and rent-seeking, based on "experts" and "tech." Just follow the trail of economics departments that rich people fund. The Koch Bros. have created whole econ departments single-handedly, for instance. Oh, and did I mention the present push to influence the high school curriculum as well. Get 'em young.

  11. I just stumbled upon your blog and wanted to say that I have really enjoyed reading your blog posts.
    Educational Company in India, Global Schools in India

  12. The biggest bullshit I encountered during my economics studies was econometric forecasting. The example there was about inflation and the Bank of England. Since the Bank of England had a 2% inflation target, the forecast said that inflation will move from its current level to the 2% target. The only "skill" was the speed of adjustment. I cannot imagine how this exercise was of value to anyone, because it was basically the Bank of England telling inflation os whatever we want it to be. And you don't need econometrics to do that. In fact, the only gain in knowledge from this was a demonstration on how wrong the forecast was.
    Similar examples concerned the regression to the mean or to the trend, which mathematically is absolutely correct - and the logic is irresistible - but it's also a pretty big cloud of nothing. Politicians like to have a prediction of the unusual events that will happen, not a prediction of what the world will be if nothing happens. The "perfect" world in which nothing unusual ever happens is the neoclassical general equilibrium. But history is a chain of events that deviate from the general equilibrium. In this way, economics sets itself up for failure.
    Finally, Freudian psychology and economics have one thing in common, they have their idea of how people think and they run with it. This is the source of the bullshit.

  13. Anonymous9:29 AM

    Yeah, economic forecasting is hard. Part of the reason for this is because the forecasts themselves can affect behavior and change things. At least with weather forecasts you don't have to worry about forecasts affecting what really happens.

  14. Anonymous10:09 AM

    Sociologist who become administrator in the late stage of their career have a better understanding of society than economist. And In Canada, these sociologists have better wages than business school graduate at the same stage of their career.

  15. Look, when I was a political sciencer, quantitative methods was a mandatory class. praps wind the econ cock in?

  16. Anonymous9:51 PM

    Andrew Gelman is half right. Both economists (Noah especially) and sociologists (Gelman obviously) tend to be like Freudian psychologists. Mostly full of shit.

  17. Anonymous8:49 AM

    Generally you don't need a lot of math - certainly not high-tech math - for consulting. To get the job, you need to be good - very good - at doing what is essentially high-school math quickly & reliably while under pressure. To keep the job, you also need to be good at working with large excel data sets. Having a PhD in math will not get you a job if you can't do that - and that math PhD is no guarantee that you will be able to. As a recruiter from one of the top consultancies recently observed to me, "you'd be amazed at how bad many math PhDs are at basic arithmetic." (This is before we get into the fact that there are numerous other skills necessary for a consulting career besides good high-school math.)