November 24, 2007

Greg Mankiw on Social Mobility

Filed under: Economics, Politics, Science — duncan @ 8:12 pm

Greg Mankiw is one of the most widely read economics bloggers. He teaches at Harvard, and was chairman of Bush’s Council of Economic Advisers from 2003-2005. I’ve recently been looking at some of his old posts.

On April 27 2006 Mankiw discussed “intergenerational transmission of inequality.” In a talk to the Centre for American Progress, the economist Tom Hertz said: “the chances of [an American] getting rich are about 20 times higher if you are born rich than if you are born in a low-income family.” Mankiw expresses annoyance with Hertz’s liberal “spin”.

Mankiw: “One might ask why being born into a high-income family means you will likely have higher income. Is it the good genes that you inherited from your successful parents or the nice neighbourhood and expensive private schools that their high income could purchase for you? Is it nature or nurture?

The evidence suggests that nature trumps nurture.”

Mankiw refers to a study about adoption by the economist Bruce Sacerdote. [For some reason I can’t get the link to work – but you can download the study from Mankiw’s site.] Sacerdote studied 1117 families who adopted children through Holt International Children’s Services from 1970 to 1980. (The data was collected in 2003). He attempted to calculate “the transmission of income, education and health characteristics from adoptive parents to adoptees. I then compare these coefficients of transmission to the analogous coefficients for biological children in the same families, and to children raised by their biological parents in other data sets.”

Here’s the passage Mankiw quotes:

“Having a college educated mother increases an adoptee’s probability of graduating from college by 7 percentage points, but raises a biological child’s probability of graduating from college by 26 percentage points. In contrast, transmission of drinking and smoking behaviour from parents to children is as strong for adoptees as for non-adoptees. For height, obesity, and income, transmission coefficients are significantly higher for non-adoptees than for adoptees.”

As Sacerdote puts it, parents’ education has an “economically meaningful” effect on adoptees’ education. “Each additional year of mother’s educational attainment raises the adoptee’s educational attainment by .07 years. But the effects for adoptees are modest when compared with the corresponding effects for non-adoptees… [F]or educational outcomes, the level effects of parental education are quite important, but only about one quarter of the story.”

So ‘nature’ and ‘nurture’ both play a role in educational attainment – and ‘nature’ seems to play a larger role than one might expect. But Mankiw’s post focuses on income, and here Sacerdote’s conclusions are counter-intuitive. “The adoptee’s income appears to have almost no relationship to parental income.”

Mankiw: “Sacerdote suggests that income is like height. Having a tall father means you are likely to be tall, but it is because he has given you the tall gene, not because he has created an environment that fosters height. The same appears to be true of income.”


I hardly know where to start. Perhaps its worth mentioning that human growth is massively influenced by environmental factors. As Wikipedia puts it: “Genetics is a major factor in determining the height of individuals, though it is far less influential in regard to populations. Average height is increasingly used as a measure of the health and wellness (standard of living and quality of life) of populations. Attributed as a significant reason for the trend of increasing height in parts of Europe is the egalitarian populations where proper medical care and adequate nutrition are relatively equally distributed… Genetic potential plus nutrition minus stressors is a basic formula.”

So income is indeed a lot like height, in the sense that there’s a causal relationship between them. Living in “an environment that fosters height” – i.e. having a high standard of living – is going to influence how tall you are. Still, Mankiw and Sacerdote are only talking about U.S. families, so Mankiw’s shorthand is, perhaps, acceptable.

Far more dubious is Mankiw’s interpretation of Sacerdote’s paper. In the first place, Sacerdote does not discuss the range or distribution of the adopting parents’ incomes. Although Hertz makes strong claims about transmission of income levels from parents to children (“our parents’ income is highly predictive of our incomes as adults”), he also argues that this relationship is far stronger at the extremes. “Children born to the middle quintile of parental family income ($42,000 to $54,000) had about the same chance of ending up in a lower quintile than their parents (39.5 percent) as they did of moving to a higher quintile (36.5 percent).” The problem, put simply, is that the poor stay poor, while the rich stay rich. Within the broad range of middle-income families, the relationship between parent and child income is not strong.

We don’t know details of the incomes of the families in Hertz’s study. But we do know that Holt’s background check requires adopting families to have a “minimum income.” Sacerdote doesn’t say what this income threshold is – but it must surely exclude the low-income extreme at which, according to Hertz, the intergenerational transmission of income is strongest. In a word, adoption agencies have a responsibility to place children with families who can support them. Sacerdote’s study therefore cannot adequately incorporate many of the most important factors by which income inequality is perpetuated. (Not just factors directly related to income – Holt also runs a criminal record check, for instance, and requires parents to have been married for at least three years.) This is not “a data set in which adopted children were, literally, assigned randomly” (as Mankiw writes), because a number of strict criteria must be met before the random allocation of children to families can begin.

But even when this is taken into account, one would expect to find a relationship between the incomes of parent and adopted child. The apparent lack of such a relationship is particularly striking given the reported link between different generations’ educational attainment. If there really is no intergenerational transmission of income to adopted children, one would need an explanation for why the relationship you’d expect between income and education has here broken down.

As it happens, Sacerdote offers two such explanations. This result “could be driven by the restriction of range among Holt families, or by higher measurement error in my survey.” Higher measurement error, that is, than in similar surveys which do find a positive link between the incomes of parent and adopted child. In his Table 3a, Sacerdote compares “Transmission in Holt Sample Versus Transmission in Other Samples.”

Holt Adoptees:

Transmission of Years of Education: 0.069
Transmission of Income: -0.087

Swedish Adoptees:

Transmission of Years of Education: 0.144
Transmission of Income: 0.154

NLSY Adoptees:

Transmission of Years of Education: .277
Transmission of Income: .112

There are, of course, many possible reasons for the disparities between these figures. But it’s striking that Mankiw picks an apparently anomalous result as the basis for his post.

Sacerdote himself seems to favour “measurement error” over “restriction of range” as the explanation for his counter-intuitive income transmission result. “The lower income transmission in my sample is quite possibly driven by higher measurement error in my income survey question.” “Loss of parental income is not statistically significant in predicting child’s years of education, which may be a statement about the measurement error in my parental income variable.” But Mankiw doesn’t mention this possibility. For Mankiw, “Sacerdote suggests that income is like height.”

In other words, Mankiw is misrepresenting the paper he quotes. He is picking a result that conflicts with previous studies of intergenerational social mobility, and with common sense, in order to advance his political views. These views are, of course, that the poor are to blame for their poverty, while the rich deserve their wealth. Don’t talk about social justice: entrenched inequality has nothing to do with society.

Generally the right talks about weakness of character here. (See, for instance, Mankiw’s New York Times piece on American health-care. One reason for the failures of the American health care system is “the sexual mores of American youth”.) In this post Mankiw has an alternative explanation. It’s all in the genes.

[NB:  It turns out that Sacerdote’s original study prompted an absolute deluge of internet comment, which I’ve  just belatedly discovered.  If I get time I’ll try to see if any of it blows apart what I say above.]  [I never did get round to it.  If you spot any grievous flaws in the above, please leave a comment.]


  1. Intelligence, like other behavioural traits is significantly heritable.

    “Data from more than 8000 parent-offspring pairs, 25,000 sibling pairs, 10,000 twin pairs and adoption studies provide evidence that genetic factors play a substantial role in the variation of general intelligence, with heritability estimates ranging from 40 to 80%” —Burdick et al, Cognitive variation in DTNBP1 influence general cognitive ability. Human Molecular Genetics, 2006, Vol 15, No. 10.

    “Heritability estimtes for intelligence quotient (IQ) range from 0.50 to 0.80. This makes IQ a suitable target for attempts to identify the specific genes involved.” Chorney et al, Role of the cholinergic muscarinic 2 receptor (CHRM@) gene in cognition. Molecular Psychiatry (2003) 8. 10-13.

    “A substantial body of literature from twin, family and adoption studies documents significant genetic effects on human intelligence. Heritability estimates range from 40 to 80% and meta-analyses suggest an overall heritability of around 50%” Dick et al,

    (2006) “Association of CHRM2 with IQ: Converging Evidence for Genes Influencing Intelligence.” Behavioral Genetics.

    “Multivariate genetic analyses indicate that general intelligence is highly heritable, and that the overlap in the cognitive processes is twice as great as the overall phenotypic overlap, with genetic correlations averaging around .80.”

    Plomin et al (2004) “A functional polymorphism in the succinate-semialdehyde dehydrogenase genes is associated with cognitive ability,” Molecular Psychology 9, 582-586.

    Also, the brains of more intelligent people are different in terms of cortical thickness and myelination (affects processing speed – Einstein had a larger number of glial cells which produces this). This is significantly heredible:

    “The UCLA researchers took the study a step further by comparing the white matter architecture of identical twins, who share almost all their DNA, and fraternal twins, who share only half. Results showed that the quality of the white matter is highly genetically determined, although the influence of genetics varies by brain area. According to the findings, about 85 percent of the variation in white matter in the parietal lobe, which is involved in mathematics, logic, and visual-spatial skills, can be attributed to genetics. But only about 45 percent of the variation in the temporal lobe, which plays a central role in learning and memory, appears to be inherited.”

    (see New Scientist 9 March 2009 ‘High Speed Brains in the Genes’ and MIT Technology Review, March 24, 2009)

    Comment by M Stein — August 30, 2009 @ 6:25 am

  2. M – I think you have failed to understand the point of my post. The post does not contest, or indeed discuss, the question of whether biological and behavioural traits describable as intelligence are genetically inheritable. IMO, they very likely are. [For what it’s worth, I also think that ‘intelligence’ is far too general, vague, and culturally determined a category to pick out an object of scientific study with any useful precision. I also think that there are substantial flaws in the methods, such as IQ testing, generally used to assess intelligence. But my post is not discussing this question.]

    Rather, the post is discussing the relation between the genetic transmission of biological characteristics, and the intergenerational transmission of social class or income-level, with regard to social mobility in the US. The post points out that 1) Mankiw’s understanding of the nature of genetic inheritance is flawed. (Mankiw believes that height is entirely genetically determined (an incoherent concept), when in fact height is also, massively, determined by living standards – which are in turn often determined by income.) 2) Mankiw egregiously misrepresents the paper that he cites to support his claims.

    The latter, in particular, is worth pointing out, since most of Mankiw’s readers won’t also read the paper Mankiw cites, and discover that he is distorting its contents for political purposes.

    Comment by duncan — August 30, 2009 @ 11:55 pm

  3. NB: I notice that Alex Tabarrok is currently pulling exactly the same stunt here.

    Comment by duncan — August 31, 2009 @ 9:51 pm

  4. Great Article , I thought it was remarkable

    I look forward to more similar postings like this one. Does This Site have a RSS I can subscribe to for new posts?

    Comment by greheannycrar — November 23, 2009 @ 11:46 pm

  5. It’s clear that there is a genetic component to our rudimentary measures of intellect. Adoption studies have shown genetic links to obesity, mortality, antisocial behavior and criminality. I haven’t seen such a study probably because it’s hard to quantify, but my gut tells me that the somewhat interrelated ideas of motivation and “work ethic” (somewhat like metabolism) are also likely heritable. In sum, smart hardworking people who get along well with others have children predisposed to be like them, the antisocial “lazy” and unintelligent do likewise. It’d be no surprise that adoptees would move towards their biological parents social class. One of the popular studies, possibly the Swedish one, shows that the biological component becomes stronger as the subject ages (they revert further towards their genetic norm).

    This is why the Marxist fantasy can never be reality without a strong and perpetual police state enforcing this mythical equality. As soon as you take your jackboot off the throat of a population they will revert to their genetic predispositions very quickly, magnified by the effects of their new divergent environment (as you would well recognize).

    In this case right has it right. It may not seem noble, or ideal, or fair, but nature and reality have no such mandate.

    Comment by Ryan — December 10, 2009 @ 4:18 am

  6. I’m a little uncertain if this is a wind up – probably, depressingly, not. In any case, you’ll notice that the post above discusses… actual research. You, on the other hand, seem mostly to be relying on that organ of gnostic knowledge, “my gut”. Perhaps your gut is geneteically superior to mine, which is why it can gather and analyse statistical information about the intergenerational transmission of inequality.

    Comment by duncan — December 10, 2009 @ 7:17 am

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.

Blog at

%d bloggers like this: