about me
art
biz
Chess
corrections
economics
EconoSchool
Finance
friends
fun
game theory
games
geo
mathstat
misc
NatScience
... more
Profil
Logout
Subscribe Weblog

 
heckman2Over on Cafe Hayek they note an interview with economist and Nobel Laureate James Heckman, who chides economists because "formal economic models view the development process solely in terms of raising IQs. Or else they assume that IQ is purely heritable." I found this outrageous on several levels.

First, when Hernstein and Murray came out with The Bell Curve, the economics profession circled the wagons, proclaimed their PC bona-fides, and shunned it. There has not been much research on IQ or aptitude and labor outcomes, before or since (in my opinion because it is too inegalitarian). The Journal of Economic Literature gave primo space and an expedited review to a facile hatchet job on the Bell Curve, just so everyone would know what right-minded economists should think. Unlike all other books on economic subjects (books don't have economist referees), the JEL noted The Bell Curve committed the following key error:
[Hernstein and Murray] and their publishers have done a disservice by circumventing peer review. The Bell Curve was sprung full blown without external scientific scrutiny
Around that time Heckman himself criticized Hernstein and Murray because it "ignored" environmental factors, as if all the cross-tabbing of IQ and socioeconomic variables, twin studies, adoption studies, all don''t count--though ignoring IQ is perfectly acceptable when running a regression that measures the effects of discrimination. The idea that unless a factor can be isolated and uncorrelated with all other relevant factors, you can't say anything about it, is just an obvious pretext because clearly this critique would invalidate all econometrics, and no econometrician is that much of a nihilist.

Secondly, his statement that "it's not nature versus nurture, but rather nature with nurture." is really banal. No modern scientist in this field ever argues that IQ is deterministic, 100% heritable. It's a typical straw man argument you see again and again, like when someone responds to a criticism of excess lawsuits by saying "but if there were no lawyers you would be at the whim of various government authorities!" True, but irrelevant. Repetition doesn't make it a better argument, it merely shows it's the best one they have.

So what is the result of his "unconventional" approach: "We find that both social and emotional skills are essential in producing successful people. These findings change the way economists think about the human capital formation process." The idea that "social and emotion skills are essential" is an incredibly obvious remark, to insiders and outsiders. There isn't a reluctance to measure these things and test their imacts, there's a difficulty in measurement. To imply these important soft variables are being ignored merely because they are not appreciated, and meanwhile to assert there has been too much emphasis on IQ in the literature, by such a smart and educated guy like Heckman, is in my opinion a calculated disingenuousness.
AleksJ meinte am 22. Jun, 09:25:
If someone wants to actually look at the Herrnstein&Murray data, it's here: http://pacioli.bus.indiana.edu/erasmuse/bellcurve/bellcurve.htm A good re-analysis is at http://www.srv.net/~msdata/bell.html 
HedgeFundGuy antwortete am 22. Jun, 19:43:
Neat reference, but I found the re-analysis rather unconvincing. Classification tables are problematic measures of fit, because that is a test of the level at which one turns a dependent variable to 0 or 1, which is independent from the issue of how the independent variables are correlated with the dependent variable. Given the researcher finds all poverty cases were predicted inaccurately, that's so bad it means the test is meaningless, like when you get a p-value of 0.999.

A big book like The Bell Curve (ie, lots of date presented in different ways) is easy to pick off peicemeal, but like A Monetary History of the United States, the bottom line is the thesis: when you control for SES variables, does a g-loaded variable explain various outcomes? I think the answer is still yes. 
AleksJ antwortete am 22. Jun, 21:41:
I found the re-analysis quite supportive of the H&M work.

But SES/g correlation is a very small bit of the big picture that's been turning around over the past years: it's the inadequancy of the standard social science model and of "political correctness" (better term: politico-demagogical self-interest) often taking over the (sometimes bitter) truth.

Heckman agrees with the big picture, but softens the edge a little. My own take at this is also that monetary units, IQs and g-factors are overrated. There is a forest out there. 
ps (anonymous) meinte am 25. Apr, 20:27:
why Heckman's work is important
The essay says that the idea behind importance of social and emotional skills is "incredibly obvious", but I would argue that when you are up to a certain level every good idea sounds obvious. Many economic ideas may sound obvious even for non-specialists (unlike that in, say, mathematics) and this is not just expressing this idea what makes the work of Heckman extremely important. Heckman and his co-authors created innovative econometric techniques to measure the importance of non-cognitive skills and derived important policy implications.

Heckman’s results allow calculating the efficiency of investments in early child education. The returns from such programs for children from disadvantaged families can exceed costs in about 10 times and this is not due to IQ change, but due to influencing non-cognitive skills
(see http://www.tompaine.com/articles/2006/01/11/catch_em_young.php )
Note that a standard way to evaluate a program to children is just to measure IQ (so, what is obvious for the author of the essay was not at all obvious for American educators and policy makers!). If you stick to this measure you would conclude that the programs for children are useless, but they are quite the opposite! These findings can change approaches to government social programs by shifting attention from the unemployed and older children to very small children which, according to Heckman, should greatly reduce inequality.