The Agenda

Henry Harpending on How Class Might Become Caste

One of Matt Yglesias’s running jokes is that Greg Mankiw believes in the rule of a hereditary aristocracy:

Many proponents of low taxes on high-income individuals are “supply-siders” who claim that such a tax policy will maximize overall welfare. But other proponents of low taxes on high-income individuals such as Greg Mankiw deny that this is the relevant consideration, and simply say that progressive taxation is immoral

Mankiw has replied to this characterization by quoting from “Spreading the Wealth Around,” the essay Matt links to above:

Public goods and Pigovian subsidies lead naturally to a tax system in which higher income individuals pay more in taxes. Surely, those with higher income and greater property benefit more from a governmental system that protects property rights. Moreover, the monetary value attached to other public goods (such as parks and playgrounds) and to positive-externality activities (such as basic research) very likely rises with income as well. Indeed, if the income elasticity of demand for these services exceeds one, as is plausible, a progressive tax system is perfectly consistent with the Just Deserts Theory.

What about transfer payments to the poor? These can be justified along similar lines. As long as people care about others to some degree, antipoverty programs are a type of public good. [Thurow 1971] That is, under this view, the government provides for the poor not simply because their marginal utility is high but because we have interdependent utility functions. Put differently, we would all like to alleviate poverty. But because we would prefer to have someone else pick up the tab, private charity can’t do the job. Government-run antipoverty programs solve the free-rider problem among the altruistic well-to-do.

Mankiw asks whether this sounds like someone who objects to progressive taxation, and I think we can agree that the answer is pretty clearly no. 

Matt has used much stronger language to characterize Mankiw’s views in the past, e.g.:

Conversely, Republicans say they oppose increasing taxes on the rich. They offer the pragmatic argument that high levels of taxation on very financially successful people slow economic growth over the long run. Others, like Greg Mankiw, have offered the argument that even if higher taxation of the rich would increase overall human welfare we should ignore this fact (PDF) because it would deprive America’s genetic elite of the money to which they’re entitled.

Once again, Mankiw has a strong case that Matt’s characterization isn’t a very good or accurate one. But let’s take this “genetic elite” idea seriously for a moment. Henry Harpending, an anthropologist and population geneticist who co-authored one of my favorite books, The 10,000-Year Explosion, has written a fascinating post on caste and meritocracy. He begins as follows:

An article by Sabrina Tavernise appeared in the New York Times a few days ago describing increasing perceptions of class conflict in America, and there is a lot of recent commentary in the press about this report from the Pew Charitable Trust that claims there is less class mobility here than in several other northern countries.   It is not very clear to me what the complaints really are or what alternatives exist. If there is any substantial heritability of merit, where merit is whatever leads to class mobility, then mobility ought to turn classes into hereditary castes surprisingly rapidly.

It is, of course, very difficult to determine whether or not there is “any substantial heritability of merit,” as Harpending acknowledges in his next paragraph:

A start at looking into genetic consequences of meritocracy is to create the simplest possible model and follow its implications. Consider free meritocracy in a two class system, meaning that each generation anyone in the lower class who has greater merit than someone in the upper class immediately swaps class with them. Mating then occurs at random with class. There are no fitness differences, no selection at all, everything is neutral. It is convenient to discuss what happens in terms of IQ because IQ is familiar and, in northern industrial nations, closely related to merit. On the other hand we have to keep in mind that our focus on IQ is like that of the drunk searching for his keys under the lamppost: we can measure IQ but have no good way to measure honesty nor time preference nor conscientiousness and so on. 

Harpending models a scenario in which “the additive heritability of merit is set to 0.6 so there are substantial random environmental effects.” So what happens?


Class mobility after the first generation is 30% while after four generations it has declined to 10% and continues to decline after that.  The average merit in the two classes is about -1SD in the lower and +1SD in the upper on the original scale, corresponding to IQs of 85 and 115.

Recall that there are no fitness differences in this model. Still, after four generations, about 70% of the variance is between classes, which can be compared to  about 35% of the variance among continental human groups for random genetic markers, i.e. colloquially class differences are twice neutral race differences. (The familiar among-population figure of 15% made famous by Lewontin refers to gene differences while here we are comparing genotype differences of diploids, hence the difference between 15 and 35.)

A surprise to me from this model was the rapidity with which classes turn into castes: most of the action is in the first generation or so. In retrospect this seems so obvious that it is hardly worth saying but it wasn’t so obvious to me when I started toying with it.

Even though everything here is selectively neutral, I wonder about the extent to which this free meritocracy mimics selection. Any mutant that boosts merit in carriers will be concentrated in the upper class and vice versa. Greg and I discuss in our book how environmental change initially selects for dinged genes that are “quick fixes” in carriers but detrimental in homozygotes, citing sickle cell in humans, broken myostatin in beef cattle, and numerous others. Does this social system mimic selection?

A correlate of IQ in humans is myopia, one idea being that IQ boosters relax early developmental constraints on CNS growth resulting in eyeballs too big for the socket, leading to myopia. I have read somewhere that myopia is positively related to income in the US. Time to try to find that literature.

If Harpending’s stylized scenario bears any resemblance to reality, it doesn’t have any obvious normative implications. For example, one might conclude that “substantial heritability of merit” provides a solid normative foundation for a strongly redistributive tax-and-transfer regime. Indeed, I actually think that this is the gut instinct that informs many so-called “luck egalitarians.”

It might also lead us to look upon “faculty taxes,” i.e., taxes tied to underlying human capital rather than income as such, somewhat more favorably. (Mike Konczal offers a critique of “faculty taxes” from the left, and he wonders if elites secretly like the idea.) Or it could lead us to conclude that fretting about social mobility is ultimately futile, the view that is often attributed to conservatives, reactionaries, etc. That is, the “lower orders” are simply undeserving, and thus should be left to their fate.

Suffice it to say, I think that last conclusion is wrongheaded. Absolute upward mobility seems like an important thing to strive for, and relative social mobility is clearly impaired by mass incarceration, the rise in the number of disabled mentally ill individuals, the low-quality of America’s K-12 schools and the broader set of institutions devoted to providing social uplift to the currently poor, and other problems that would be worth addressing even if social mobility weren’t a concern as such. For me, being concerned about social mobility in at least some sense is massively overdetermined.

But what if we actually addressed mass incarceration, etc., and found that members of some families and groups tended to cluster at the top of the income and wealth spectrums while others did not? As an example, one could think of the success of members of various South Asian caste and ethnoreligious communities that, in the words of a friend, were “pre-adapted” to modernity. Yuri Slezkine brilliant book The Jewish Century makes a related argument about Ashkenazi communities in central and eastern Europe and how they prefigured the “service nomadism” that defines advanced economies. I tend to think that the fact that different groups are distributed unevenly across the spectrum, and that different groups tend to have different preferences regarding the relative importance of market and household production, etc., is not intrinsically problematic. 

Harpending’s scenario lead us to an interesting set of questions about nepotism and how we feel about the intergenerational transmission of human and social capital. As we’ve discussed in the context of “pregnancy envy,” I suspect that our feelings about the intergenerational transmission of wealth, and more broadly of advantage, are rooted in the inchoate sense that kinship networks should take precedence over the state and over wider social obligations. Somehow “undoing” this deeply ingrained sense is, as I’ve argued elsewhere, fraught and potentially counterproductive.  


The Latest