To my disappointment, some advocates of less-skilled immigration are drawing on my column on ethnic attrition to suggest that concerns about less-skilled immigration are exaggerated. The theory, as I understand it, is that because ethnic attrition means that economic outcomes for the entire universe of individuals with at least (say) one Mexican-born grandparent are to some degree understated, we shouldn’t be concerned about allowing a continued large-scale influx of less-skilled immigrants. Let’s think through why this is wrongheaded:
1. We are conflating ethnicity and educational attainment, and this creates a misleading impression. When we are forced to rely on an ethnic category as a proxy for less-skilled immigration, as often happens in this discussion due to data limitations, we miss the fact that some number of Hispanic immigrants to the U.S. are skilled and some number of Asian and European and African immigrants to the U.S. are less-skilled. It is true that there is data indicating that economic progress between second- and third-generation Latinos seems to stall. Yet this data is far less valuable, in my view, than data concerning the second- and third-generation descendants of college-educated immigrants. Unfortunately, we don’t appear to have good data on the latter. Conceptually, we’d be better off placing all skilled immigrants — whether they’re from Latin America, Asia, Europe, or Africa — in one bucket and all less-skilled immigrants in another. The continued emphasis on ethnicity is, frankly, a major advantage for those who aim to increase the non-college-educated share of the future U.S. workforce, as it allows them to characterize those who believe that U.S. immigration policy should contribute to raising the average skill level of the U.S. workforce as motivated by fear or dislike of immigrants from a particular region.
2. One of the core implications of ethnic attrition is that it appears that the second- and third-generation Americans who “defect” from identifying as Hispanic have higher levels of educational attainment than those who continue to identify as such. This could be a random coincidence. Or it could be that there are differences between Hispanics who intermarry as opposed to those who do not. For example, it could be that Hispanics with higher levels of educational attainment are more likely to intermarry, as they are more likely to have the same levels of educational attainment as native-born individuals. (Intermarriage rates might also decrease if educated members of a given ethnic group don’t have to marry outside of the group to find a partner with a comparable level of educational attainment.)
Consider the following from Matthjis Kalmijn of the University of Amsterdam:
A common claim in the literature is that higher-educated persons are more likely to marry outside their ethnic/racial group than lower-educated persons. We re-examine this “educational gradient” with a multilevel analysis of 46 immigrant groups in the Current Population Survey. We find that there are positive effects not only of individual education on intermarriage but also of the educational level of a group. Moreover, the educational gradient declines when the aggregate level of education of an immigrant group is higher. The aggregate effect of education points to cultural explanations of the gradient that emphasize the role of interethnic attitudes. The interaction effect points to a structural explanation that explains the gradient in terms of opportunities of finding similarly educated spouses within the group.
We also have findings on people of Asian origin in the U.S. and Canada from Sharon M. Lee and Monica Boyd:
In both countries, those with less than a high school education are most likely to be in-married, a pattern that is similar for both men and women. As education increases, intermarriage also increases but the increase is not linear. The percentage inmarried increases for the two highest education categories in both the U.S. and Canada.
And then we have a report from the Pew Research Center released in February of last year:
However, these overall similarities mask sharp differences that emerge when the analysis looks in more detail at pairings by race and ethnicity. Some of these differences appear to reflect the overall characteristics of different groups in society at large, and some may be a result of a selection process. For example, white/Asian newlyweds of 2008 through 2010 have significantly higher median combined annual earnings ($70,952) than do any other pairing, including both white/white ($60,000) and Asian/Asian ($62,000). When it comes to educational characteristics, more than half of white newlyweds who marry Asians have a college degree, compared with roughly a third of white newlyweds who married whites. Among Hispanics and blacks, newlyweds who married whites tend to have higher educational attainment than do those who married within their own racial or ethnic group. [Emphasis added]
Later on, the report provides more detail on Hispanic intermarriage:
About one-in-five intermarried white/Hispanic couples who married in 2000 or later are college-educated. That is lower than white couples (26%) but higher than Hispanic couples (6%). A similar pattern exists among those couples who married in earlier years, but the shares are somewhat lower, especially for couples who married prior to 1980.
On the one hand, a 20 percent share isn’t extremely high. On the other hand, it is much, much higher than a 6 percent share. And one assumes that there are unobserved variables at work as well.
Taken together, we have at least suggestive evidence that intermarriage is more common among the more highly educated. If we believe that social networks are an important part of upward mobility, the fact that skilled immigrants are better-positioned to join more privileged social networks suggests that skilled immigrants and their descendants are more likely to achieve upward mobility. This in turn implies that they are less likely to rely heavily on means-tested benefits and more likely to build wealth. So the fact that attrition biases the results for third-generation Americans who identify as Hispanic is not actually a coincidence: rather, it applies that the difficulties facing immigrants who arrive with limited skills and limited English language proficiency have reverberations beyond the first generation.
3. The arguments I’ve been making about immigration are discomfiting for many. The reason, I would suggest, is that my arguments draw on recent social science findings — ably summarized by Scott Winship in National Review in 2011 — which reveal that upward mobility from the bottom of the income ladder is actually much lower than Americans have commonly assumed:
The EMP/Brookings analyses break the parent and child generations into fifths on the basis of each generation’s income distribution. If being raised in the bottom fifth were not a disadvantage and socioeconomic outcomes were random, we would expect to see 20 percent of Americans who started in the bottom fifth remain there as adults, while 20 percent would end up in each of the other fifths. Instead, about 40 percent are unable to escape the bottom fifth.[6] This trend holds true for other measures of mobility: About 40 percent of men will end up in low-skill work if their fathers had similar jobs, and about 40 percent will end up in the bottom fifth of family wealth (as opposed to income) if that’s where their parents were.
Is 40 percent a good or a bad number? On first reflection, it may seem impressive that 60 percent of those starting out in the bottom make it out. But most of them do not make it far out. Only a third make it to the top three fifths. Whether this is a level of upward mobility with which we should be satisfied is a question usefully approached by way of the following thought experiment: If you’re reading this essay, chances are pretty good that your household income puts you in one of the top two fifths, or that you can expect to be there at age 40. (We’re talking about roughly $90,000 for an entire household.) How would you feel about your child’s having only a 17 percent chance of achieving the equivalent status as an adult? That’s how many kids with parents in the bottom fifth around 1970 made it to the top two-fifths by the early 2000s.[8] In fact, if the last generation is any guide, your child growing up in the top two-fifths today will have a 60 percent chance of being in the top two fifths as an adult. That’s the impact of picking the right parents — increasing the chances of ending up middle- to upper-middle class by a factor of three or four.
Advocates of less-skilled immigration, meanwhile, often fall back on the Horatio Alger myth, and the notion that today’s less-skilled immigrants are in all relevant respects identical to the less-skilled immigrants of a century ago, the ancestors of many of today’s native-born Americans. This ignores the fact that the native-born skill level is much higher today than it had been a century ago, and that assimilation and economic upward mobility are more attainable goals for immigrants who match or exceed the native-born skill level. To put this differently, opponents of an increase in less-skilled immigration rely on a realistic appraisal of how good U.S. institutions are at mitigating hardcore multigenerational poverty — not very — rather than the romantic notion that all it takes to make your way in America today is a little bit of pluck and a little bit of determination. I would argue that we ought to craft an immigration policy that identifies individuals who are likely to flourish in the United States as it really is, not as we’d like it to be.