A response to Scott Alexander on Jewish achievement
Why I still don't think it's that interesting, all things considered.
In an extremely popular post over at Astral Codex Ten (the successor blog to Slate Star Codex), Scott Alexander argues that Jews’ disproportionate achievements are significant, and that we should care about them. His post is arguing against my own post from 2013, in which I argued that once we account for things like selective immigration, urbanization, and temporary effects, Jewish achievement looks less stunning than if we simply look at it without context. Scott argues that some of the factors I list don’t change the basic statistical story much, while others might be true but don’t really bear on the overall discussion.
First, a quick word about how I think about this issue. Scott writes:
Noah admits that his goal is to make the hypothesis of Jewish specialness sound "less interesting". I'm against this. I would like it to remain interesting and something that people pay attention to.
In fact, that is not my goal. It was merely my conclusion, after hearing about Jewish achievement for pretty much my entire life and finally sitting down to think about it carefully. It’s true that some pundits do start with their conclusions and cast around for evidence to support those conclusions — that’s certainly an easy way to pump out high volumes of content — but I try to always take the more challenging road of letting the evidence persuade me. Hence the pun “Noahpinion”. ;-)
Anyway, I’ll go through Scott’s points one by one, but first, I’d like to talk about why this question matters at all.
Why would we care about Jewish achievement?
There are, as far as I can see, at least four reasons why people might care about Jewish achievement. Let’s go through these.
1) Arguing against disparate impact/systemic racism
Disparate impact is the legal doctrine that says that if a policy or practice affects different groups of people differently, it must be discriminatory in some way. More generally, it’s related to the idea that socioeconomic disparities between racial groups in the U.S. are due to systemic racism. Scott, like Andrew Sullivan and many others, thinks that Jewish achievement shows that this theory of racial disparities is wrong. He writes:
I would like [Jewish achievement] to remain interesting and something that people pay attention to.
Why? The Standard Model of American Ethnicity says that there are whites and non-whites, whites are rich, non-whites are poor, and this is because of structural racism where whites are oppressing everyone else. Reality gets beaten and twisted until it can be shoehorned into this model - gifted programs that are 80% Asian "perpetuate whiteness", etc. The reality is that every ethnic group is different from every other ethnic group, including in socioeconomic status, with white people usually somewhere around the middle…
If you dismiss every group that does better than whites, then you can tell a story where all inequality is caused by white people controlling everything and creating covert structures/institutions that favor whites. If you don't dismiss those groups, the story becomes harder. Anti-Semites had their own story about problems caused by Jews controlling everything and creating covert structures/institutions that favored Jews. Nowadays we rightly reject that story. But in order to continue rejecting it, we have to come up with strained explanations to make Jewish achievement less interesting, because we've already committed to using the structural racism explanation for every group difference that seems relevant to us.
OK. But if you want to argue against disparate impact — or against the broader invocation of structural racism as the cause of all racial disparities — you definitely don’t need to use Jewish achievement as a counterexample. In fact, Jewish achievement isn’t even close to being your best counterexample!
Zach Goldberg has a recent post with lots of data comparing the socioeconomic performance of various groups in the U.S. Here’s a graph showing a measure of occupational status by national ancestry:
Indians come out on top, with Iranians in second place. This is unsurprising to anyone who knows that Indian Americans have the highest median household income of any ancestry group (in terms of median personal income adjusted for age and sex, Belarusian and Chinese Americans pull a bit ahead). Breaking it down by religion, Hindu Americans have an even higher percentage of adults making over $100,000 a year than Jewish Americans:
And when Goldberg looks at the medical profession, the difference is especially stark:
Now, we know exactly why this is the case: Selective immigration. There is no real argument about this. Our immigration system selects for highly skilled individuals, and the larger the ratio of the source country’s population to the number of people who come over here, the more selected the immigrant group tends to be. This is a fancy way of saying that the Indian people who come to America tend to be the educational and economic cream of the crop of 1.4 billion people — engineers, business executives, doctors, entrepreneurs and so on. That kind of selection very predictably creates an elite minority.
And unlike in the case of Jews, selective immigration very clearly demonstrates a shortcoming of disparate impact doctrine — and of the idea that systemic racism is the sole cause of racial disparities. The fact that selective immigration regularly and quite predictably creates minority groups with different aptitude for business and education — whether you believe that aptitude comes from culture, genetics, social capital, or whatever — it demonstrates a way that group outcome disparities come from sources other than structural racism. And it gives a clear reason for us to temper the application of disparate impact.
But note that in his post, Scott spends quite a bit of time arguing against the idea that selective immigration is responsible for Jewish achievement. Why? If his point here is really just to show that disparities arise for non-racist reasons, he should focus on selective immigration instead of pooh-poohing it. Indian Americans (or Filipino Americans, or Iranian Americans, etc.) provide a much stronger argument against overapplication of disparate impact/systemic racism than Jews do, precisely because we know why these minorities are elite, whereas with Jews it’s kind of mystery meat.
2) Group pride
It’s fun to sit around reading about the historic empires that were ruled over by your own racial or religious or ethnic group, and to think “Wow, I am the heir to greatness!” This is very silly (and I plan to write more about why it’s silly in a subsequent post). But it’s a natural human tendency, I think.
But Jewish guys like Scott and myself can’t really do that, because our ancestors never ruled any empire of note. So in order to feel like we’re the Heir to Greatness, Jews are forced to look at other types of achievement — Nobel prizes won, companies founded, great theories discovered, vaccines developed, and so on. Jewish achievement confers a sense of (undeserved) group pride — instead of puffing ourselves up by looking at maps of the old Spanish Empire or the old Tang Dynasty or whatever, we can look at pictures of Albert Einstein and Richard Feynman, as if their brilliance somehow magically confers prestige upon us as well.
But this is hardly a reason why society as a whole should care.
3) Cultural lessons
If Jewish achievement is due to culture (as Scott allows might be the case), then perhaps we might be able to learn and reproduce some secrets of Jewish success. In fact, there is a whole literature of books in China that attempts to do just this, with titles like “Jewish People and Business: The Bible of How to Live Their Lives” and “The Eight Most Valuable Business Secrets of the Jewish”.
And like…OK, if you want to try that, that seems absolutely fine with me. As Scott notes in his post, some of my caveats to Jewish achievement still allow for the possibility that Jewish culture motivates Jewish people to succeed in certain ways (or at least, did so in the past). So I don’t see any reason people shouldn’t study Jewish culture to glean possible secrets of success, if they think that will help them succeed.
But I doubt this is a reason for society to care. It seems implausible (though perhaps not impossible?) that the American government could implement the secrets of Jewish culture in our educational system.
4) Genetic superpowers
Some people believe that Jewish achievement is due to genetic superpowers. The most common theory I’ve heard is that brain diseases like Tay-Sachs, which are common in Jewish populations, are merely a rare, excessively strong form of some sort of brain modification that gives lots of Jewish people super smarts. But there are other theories too.
If you’re a person who’s deeply invested in hereditarian explanations for group outcomes, then I guess it makes sense to embrace theories like this — or at least to try to rule out alternative theories and let people arrive at ideas like this by default. Or if you’re one of the scientists who’s searching for the “smart genes”, then maybe you’ll look at Jewish achievement and decide to study Jews (though in practice I think this is probably not going to work, since Jews are such a small population).
But I am not at all invested in hereditarian explanations for group differences, and I don’t see any good reason why our society should be invested in that, either. If scientists can find “smart genes” by looking at Jews, more power to them, I guess — I’m happy to give them my cheek swab, but I think it would just add noise to their sample. ;-)
So overall, I don’t see a lot of reason for society to care about Jewish achievement. It’s not necessary — or even that helpful — if you’re trying to argue against disparate impact. And none of the other reasons seem particularly important.
But in any case, let’s go through Scott’s substantive points.
In my original 2013 post, I listed five factors that each make me think Jewish achievement is less impressive than many people believe. These are:
The possibility of selective immigration
Jewish people’s tendency to live in big cities
Selective assigning of “Jewish” status to more successful people of mixed ancestry when assessing achievement
Temporary cultural effects that fade over time
The temporary scientific and industrial outperformance of the United States and Europe, where most Jewish people historically have lived
You can think of these five things as controls in a regression equation, good old y = Xβ + ε. “y” here is Jewish achievement, the X’s are various factors that explain that achievement, and the error term ε is the “mystery meat” — the unexplained portion of Jewish achievement that might represent cultural factors, genetic superpowers, pure luck, or any number of other things no one has thought of yet.
My post was about suggesting some X’s. I argued that introducing factors like selective immigration should shrink our mental estimate of the size of the unexplained residual. That means we shouldn’t expect any one of these factors to be, by itself, a sufficient monocausal explanation for Jewish achievement. But added up, they leave less to be explained.
1) Selective immigration
Despite what you may have heard, U.S. immigration is highly selective. Shortly before he died, I saw the great economist Ed Lazear give a talk about a paper of his that decisively shows just how selective we are. The abstract:
Success, measured by earnings or education, of immigrants in the US varies dramatically by country of origin. For example, average educational attainment among immigrants ranges from 9 to 16 years, depending on source country. Perhaps surprisingly, immigrants from Algeria have higher educational attainment than those from Israel or Japan. Also true is that there is a strong inverse relation of attainment to number of immigrants from that country. These patterns result because in the US, immigrant slots are rationed. Selection from the top of the source country’s ability distribution is assumed and modeled. The main implications are that average immigrant attainment is inversely related to the number admitted from a source country and positively related to the population of that source country. The results are unequivocally supported by results from the American Community Survey. Additionally, a structural model that is more explicit in the assumptions and predictions fits the data well.
Note the part where Algerians do better than Israelis!
Immigrants’ upward mobility in the U.S. is astonishing. Here’s a graph of data from Abramitzky et al. (2019), showing that nearly all immigrant groups are more upwardly mobile than the native-born:
All of this is why I usually use selective immigration as a good baseline assumption. The U.S. immigration system might or might not have been a bit less selective a century ago, but Abramitzky et al. show that the immigrant mobility advantage existed back then as well.
Now, in the case of Jews, there are a few caveats that mean we shouldn’t chalk Jewish achievement entirely up to selective immigration. First, as Lazear’s abstract notes, the larger the source population relative to the immigrant population, the more selective the immigration tends to be. The entire world Jewish population has never been above 20 million; about 2.5 million Jews immigrated to America. That’s a sizeable fraction! We should expect that selection effect to be much weaker than the effects for China, India, etc.
Second, as Scott points out, selective immigration can’t explain Jewish achievement in Europe — or at least, the achievement of subgroups like Ashkenazi Jews who only really existed in Europe. He marshals a couple of scattered pieces of evidence to show Jewish achievement in European countries.
This shouldn’t be discounted, but I think we should use more systematic measures of achievement in Europe. There were a bunch of Jewish physicists at the Fifth Solvay Conference, and Jews were disproportionately likely to be scientists in the USSR, but I feel like this is not a great measure of group achievement. Maybe scientist was one of a circumscribed set of jobs available to Jews, maybe Jews went into physics because of role model effects, and so on. A systematic study is needed, and this is not it. But I’ll keep an open mind!
In any case, selective immigration is so prevalent in America that I feel like it has to be included in our mental regression when explaining Jewish achievement.
I think Scott underrates the importance of urbanization here. In my original post I claimed that a Jewish preference (or in the case of Europe, a forced preference) for living in big cities meant that Jewish people had more opportunities to do professional and scientific jobs, earn higher incomes, network, and so on. Scott retorts that Jews aren’t particularly special in this regard:
82% of Americans live in cities, and a bunch of ethnic groups are almost entirely urban (are there a lot of Haitian-American farmers?) I don't think this really makes Jews particularly special.
But that 82% “urbanization” includes the suburbs, so it’s not be as meaningful as you might think. (Also, Haitian immigrants presumably suffer from the same anti-Black racism that other Black Americans suffer from, yet make a lot more money, so maybe the urbanization helps!)
And the world wasn’t always so urbanized. When most Jews came to America in the late 1800s and early 1900s, the urbanization rate was 50% or less, and was even lower in Europe:
Jews were thus less likely to move to rural immigrant enclaves where economic success was less likely. For groups like Swedish immigrants, economists have shown that these rural ethnic enclaves held back immigrants for several generations.
Furthermore, in 1800s and early 1900s Europe, urbanization was low, and many farmers were trapped in rural poverty with few opportunities. But Jews, not being able to own land and being afraid of pogroms, tended to cluster in cities. So the urban-rural thing is probably more important in the past, and more important in terms of explaining historic Jewish achievement in Europe.
But what about the present day? Scott argues that city life doesn’t give much of an advantage these days:
In 2000, the median household income of US Jews was $72,000. The median household income of people who lived in New York City was $40,000; in San Francisco, $55,000. So I don't think you can explain Jewish success by saying they are more likely to live in places like New York City and San Francisco.
(Slight mistake here: Scott uses numbers for New York state instead of NYC. But I don’t think it makes much of a quantitative difference.)
Why did Scott choose 2000, instead of either 2019 (the most recent available year) or some earlier time when Jews had recently immigrated? I don’t know. But in 2000, U.S. cities were just starting to come out of a period of disinvestment and decay, when high-earning people fled urban cores, and “inner city” became synonymous with poverty. In 2019, after two decades of urban revival, the picture looks a bit different — Jewish median household income was up to around $111,000, which was actually lower than the median household income in San Francisco or San Jose. The average American Jewish household still have a sizeable income advantage over the average American household in most cities, but “slightly poorer than the average household in San Jose” is a little less impressive than the numbers Scott lists!
In any case, a more thorough study of the impact of urbanization — forced or chosen — on Jewish incomes is beyond the scope of this blog. I hope someone does that. But the real point here is that urbanization is just one of a number of controls in our mental regression — one factor that makes Jewish achievement a little bit less mysterious.
3) Selective attribution of Judaism
In my original post, I wrote:
In Jewish tradition, you're a Jew if your mom was Jewish, or if you convert to the faith. But in the modern world, that definition is often...stretched, to include people with Jewish-sounding last names, people with a Jewish father and a non-Jewish mother, etc. That can distort the statistics in a variety of ways.
First, the "one-drop rule" may be applied selectively more to successful people than to unsuccessful people; see the footnotes in this list of "Jewish" Nobel Medicine Prize winners, for example. Or see this song by Adam Sandler ("Harrison Ford's a quarter Jewish; not too shabby!").
Second, we may engage in confirmation bias and attribution error; upon seeing a successful person with one Jewish parent, we may attribute her success to the Jewish parent, but upon seeing an unsuccessful person with one Jewish parent, we may ignore her Jewish ancestry completely.
Scott acknowledges doing exactly this in counting the Jews at the Solvay Conference, but asserts that other measures of Jewish achievement are not subject to this bias:
But this is why we have objective data. The income, net worth, and education numbers all come from self-report, which shouldn't be vulnerable to this problem. Or you can look at Nobel prizes won by Israel vs. other countries, using Israeli residence as a non-one-drop-rule-biased proxy for Judaism. Or at the Russian numbers, which were presumably based on the Russian census.
The problem is that this “objective” data isn’t so objective. For example, Israel made the following amendments to its Law of Return in 1970:
The Law of Return was amended in 1970 to extend the right of return to some non-Jews. Amendment number 2, 4a, states:
The rights of a Jew under this Law and the rights of an oleh under the Nationality Law, 5712-1952, as well as the rights of an oleh under any other enactment, are also vested in a child and a grandchild of a Jew, the spouse of a Jew, the spouse of a child of a Jew and the spouse of a grandchild of a Jew, except for a person who has been a Jew and has voluntarily changed his/her religion.
The law since 1970 applies to the following groups:
Those born Jews according to the orthodox interpretation; having a Jewish mother or maternal grandmother.
Those with Jewish ancestry – having a Jewish father or grandfather.
So the one-drop rule is definitely in effect here. Regarding self-reported data, there is definitely some amount of one-drop rule there too; self-identification with an ancestry group tends to be stronger for higher-income people. I’m not sure about the Russian census.
In any case, it seems like this is an issue no matter how you slice it.
4) Temporary group and country effects
In my original post, I argued that Jewish achievement shows signs of fading relative to the past (there is some evidence of this). And that’s just within America; as the U.S. and Europe become less significant in the global economy and in global research efforts, Jewish achievement is likely to decline proportionally.
Scott says that this doesn’t really matter for the question of Jewish achievement:
This is a kind of weird argument - we don't have to think about or explain something, because maybe it will stop happening in the future. Do we accept this for any other social question? For all we know, maybe in a few decades black people will earn exactly as much money as white people - does that mean it's not worth talking about racial inequality today?
But it’s not a weird argument. If Black-White income and wealth gaps weren’t so persistent, and showed signs of mean reversion, people wouldn’t be so upset about them.
If a group’s outperformance is temporary, it changes the whole question of why they outperform. First of all, it makes it less relevant, since phenomena that are no longer happening are of a more academic interest relative to phenomena that are ongoing. People are typically not as interested in explaining long-finished episodes of minority outperformance, for instance. Second, temporary outperformance should change our assessment of the likelihood of various hypotheses for what caused that outperformance; it could make the “residual” of our mental regression less likely to be either genetic superpowers or enduring cultural practices, and more likely to be a “right place, right time” sort of thing.
If you think that these factors (and others, like the tendency of religious minorities to outperform in many countries) still leave a substantial residual amount of Jewish achievement that’s interesting to investigate, then go for it! Send anthropologists to study Jewish communities and networks, do genetic tests on Jews, and so on. Personally, I’m less interested in the question given the factors I named, especially because — as I explained above — I don’t think it really bears on the social issue of disparate impact and systemic racism. But if you think it’s a big deal, don’t let me stop you from investigating further!
If it was still the nineteenth century, you could compare European Jews in big urban centers to those who lived in shtetls, who would be more rural but might be more (culturally/genetically/etc) Jewish.
It's interesting that people in China are selling self-help books about the secret to Jewish success. Have they invented a euphemism equivalent to "tiger mom?" In the west it seems like most people who talk about Jewish achievement spend very little time exploring possible cultural differences and are most interested in providing support for hereditarian theories of intelligence.
Occurs to me that disparate impact and other explanations (cultural or genetic) aren’t exclusionary. Everyone always looks for the “REASON” as if the one fact will Single handedly explain all things.
Literally everything could be true at the same time (or not true).