In June of 2018, a lawsuit filed against Harvard University led to the disclosure of an internal report that the school had commissioned. The report revealed the extensive use of race as a factor in undergraduate admissions. Specifically, if Harvard were to select only for academic ability, Hispanics would be just two percent of the freshman class, and blacks would be one percent. After allowing for non-academic factors—legacy, athletic, extracurricular, and a cryptic “personal” factor—Hispanics would still be just four percent, and blacks two percent, of the incoming class. Yet the actual class ended up being nine percent Hispanic and ten percent black.1
To conform to the federal law first articulated in the Supreme Court’s convoluted Bakke case, universities often portray race as merely one factor among many, with no accompanying quotas or balancing goals. But Harvard’s internal numbers, in conjunction with similar disclosures at other schools facing lawsuits over the years, suggest it is quite a stretch to portray race as a “a small plus factor” that “breaks ties.” In fact, according to the plaintiffs’ analysis in the Harvard case, a black applicant would have a 95 percent chance of admission if he had the same resume as an Asian applicant with a 25 percent acceptance rate.2
Why did the Supreme Court adopt such a wishy-washy stance on racial preferences in the first place? Perhaps for the same reason that schools had always been loath to admit they used informal quotas. Racial preferences strike many people as wrong, but there is a certain level of discomfort with a demographically unrepresentative student body as well. Since we must have one or the other, schools have adopted a muddled, dishonest policy on race that leaves no one satisfied.
The impasse is to some degree an inevitability born of our history, as the legacy of slavery and Jim Crow affects white-black relations to this day. The school admissions mess has been exacerbated, however, by our country’s more recent choice to accept large scale, low-skilled immigration. As with all immigration, the low-skilled variety comes with both costs and benefits to the United States. What is underappreciated, however, is the persistence of some of those costs across generations. When the children and grandchildren of low-skilled immigrants do not rise to the same socioeconomic level as natives, they add to the number of underrepresented groups. The newer groups lodge familiar requests for representation and generate familiar quandaries for elite schools. Put bluntly, affirmative action in college admissions is no longer just about black Americans, and low-skilled immigration is the main reason for it.
Oddly, the role of low-skilled immigration in generating a long term skills deficit is rarely stated explicitly. Academics who study inequality often treat the immigration issue as if it does not exist. Others acknowledge the role of immigration but assume that full economic assimilation will occur in one or two generations. The data say otherwise.
Incomplete Assimilation: The Evidence
In studying the assimilation of low-skilled immigrants, the most natural group to consider is Mexican Americans, who make up about 59 percent of Hispanic immigrants and 28 percent of the foreign-born overall. Mexican Americans are not the only immigrant group with an education level below the U.S. average, but they are the largest and easiest to track over time. For purposes of this discussion, “first generation” immigrant means a foreign-born person; second generation means U.S.-born with at least one foreign-born parent; and third-plus generation means U.S-born with two U.S.-born parents. How much do Mexican Americans improve from the first generation to the third-plus? To answer that question with survey data, we must know the birthplaces of respondents and their parents. Fortunately, the Census Bureau’s Current Population Survey (CPS) has been providing that information dating back to 1994. Since that time, the assimilation pattern has been clear. The children of Mexican immigrants make substantial strides in income and education, but then progress appears to stall. The third-plus generation does little better than the second, and Mexican Americans remain substantially behind other third-plus generation Americans, particularly non-Hispanic whites.
For example, just seven percent of working aged Mexican immigrants have a college degree, compared to 39 percent of white Americans. As expected, second generation Mexican Americans improve markedly, increasing their college graduation rate to 19 percent. Unfortunately, no further progress is observed. College completion for the third-plus generation stands at 18 percent, less than half the white rate.
Income patterns are only slightly more encouraging. Among working aged employed men, Mexican immigrants earn, on average, 53 percent of white income. The figure rises to 68 percent in the second generation, but then tops out at 74 percent in the third-plus generation. As for poverty, about 22 percent of adult Mexican immigrants are below the poverty line, compared to 9 percent of whites. Poverty falls to 14 percent in the second generation, but progress seems to stop there, as 16 percent are poor in the third-plus generation.3
As discouraging as these statistics are, they are vulnerable to a criticism—namely, that they are based on cross-sectional data. A cross-section is a single snapshot in time, so the “second generation” Mexican Americans in this analysis are not the children of the first generation. They are actually the children of Mexican immigrants who arrived before the first generation we observe today. In the same way, our third-plus generation is not directly descended from the second generation in the data. An alternative method would be to start observing Mexican immigrants at some point in the past, then track the progress of their children and grandchildren over time.
When RAND Corporation economist James P. Smith constructed such a study using historical CPS and Census data, the results were mixed. Third-plus generation Mexicans growing up in the mid-twentieth century do seem to have had a smaller education gap with whites compared to their parents’ generation. However, they made no progress on the income gap, which suggests the education gains were part of a secular trend of Americans staying in school longer—not actual skill gains relative to whites.4
Optimists have raised one other data issue in hopes of finding assimilation, and that is the phenomenon of “ethnic attrition.” In defining the first and second generations with the CPS, we use only the birthplace of respondents and their parents—not self-identification. In other words, a Mexican immigrant is someone who was born in Mexico, regardless of whether he or she identifies as having Mexican ancestry. For the third-plus generation, however, respondents and their parents are all born in the U.S., so we must rely on respondents’ self-identification as Mexican American (in the CPS ethnicity question) in order to define the group. However, not everyone with Mexican ancestry identifies as such, and the rate of identification declines with assimilation. It could be that the most successful third generation Mexican Americans “attrit”—that is, they cease identifying as Mexican. If so, our observed third-plus generation could be a less assimilated subset of the much broader third generation.
Before discussing studies that account for ethnic attrition, we should clarify the overall research question. If we want to know whether Mexican immigration has had a long term negative effect on U.S. per capita income and education levels, then clearly we do need complete third generation data, including people who no longer identify as Mexican. If, however, we want to know whether Mexican immigration has helped foster a large group of less skilled Americans who share a common ethnic identity, then the ethnic attrition issue is irrelevant, and we already know the answer to our question. In 2017, over 11 million people self-identified as third-plus generation Mexican Americans, and, as we have seen above, their income and education lag white Americans by a significant margin.
Adding underrepresented minority groups who organize on the basis of their identity has profound consequences for social policy, most notably affirmative action. As noted in the introduction, Harvard accepts Hispanics at more than four times the rate we would expect based on academic achievement alone. This happens only because there are enough self-identified Hispanics who appear to need affirmative action. To Harvard, it is irrelevant that there exist some successful people with Latin American roots who do not identify as Hispanic.
In any case, ethnic attrition does not appear to have a large impact on the results. In 2008, sociologists Edward E. Telles and Vilma Ortiz revisited a 1960s-era cross-sectional survey of Mexican Americans living in the Southwest. By tracking down the children of the original respondents, they created a longitudinal dataset free of ethnic attrition. Results were disappointing. The authors found improvement in high school graduation rates from the second to the third generation, but much smaller gains in college graduation and income.5
The Telles and Ortiz study focused only on Los Angeles and San Antonio. By contrast, a 2017 paper from University of Colorado economist Brian Duncan and his coauthors uses a nationally representative—albeit small—subsample of the National Longitudinal Survey of Youth (NLSY). Taking advantage of the NLSY’s unique data on grandparents’ places of birth, the authors identify 81 true third generation Mexican Americans, including those who do not self-identify as Mexican or Hispanic. Although this group does show improvement over the second generation, its assimilation is still far from complete. The third generation college graduation rate of 24 percent, for example, is well below the white rate of 39 percent.6
In sum, the economics literature tells a depressing story about low-skilled immigrant assimilation. Although the children of Mexican immigrants move upward, progress is halting in subsequent generations. Depending on the dataset and the analytic technique, the third generation either improves mildly or stays at about the same level as the second generation. We cannot rule out Mexican assimilation by the fourth or fifth generation, but the prospects seem unlikely. For all practical purposes, the lagging economic status of Mexican Americans persists indefinitely. As the largest group of low-skilled immigrants, their experience should also raise concerns about low-skilled groups for which we currently have less data, such as those from Central America and the Caribbean.
The Policy Option That Must Not Be Named
Immigrant assimilation is not a verboten topic. If it was, then the studies discussed above would not exist. What is remarkable, however, is how often academics and policymakers fail to acknowledge that immigration exacerbates the problems they study. Immigration restriction seems to be their Lord Voldemort—the policy option that must not be named! For example, in 2009 the Pew Charitable Trusts put together a long list of recommendations to tackle poverty in the U.S. In dramatic language, the authors emphasized that the American Dream itself hangs in the balance if social mobility does not improve. “We are calling for nothing less than a fundamental shift toward government policies that are mobility-enhancing,” the authors wrote. And so they recommended healthcare subsidies, public preschool, incentives for college attendance, disability reform, teen pregnancy reduction, and on and on. Immigration, of course, was never mentioned.7
Robert Putnam, the famous political scientist, is likewise interested in every means of reducing inequality except immigration restriction. His recent book, Our Kids, presents a series of case studies that motivate some quite specific policy recommendations, including making high school sports equipment free to team members. At one point in the book, Putnam laments that today’s technology-driven economy offers little opportunity for low-skilled workers to climb the ladder. The logical implication would be that low-skilled immigration is unwise economic policy, especially given the persistence of skill deficits across generations. Predictably, however, Putnam leaves immigration unaddressed.8
Those who travel in Washington, D.C. policy circles know that suggestions to limit immigration are considered practically uncouth. If one hears or reads a proposal to restrict immigration, it is almost always from someone who has built a longstanding reputation as a restrictionist—not from a run-of-the-mill policy wonk casually opining.
When academics, wonks, and media personalities do encounter skepticism about mass immigration, they often respond with superficial talking points. Some version of “immigrants are assimilating as fast as previous ones” is quite common. In fact, that was the title of a 2015 New York Times article, which highlighted the positive findings from a National Academies study. “The force of integration is strong,” the reporter quotes the study’s lead author saying. Not until the fourteenth paragraph do readers discover that children of Mexican and Central American immigrants “do not reach the levels of their American peers.”9
The Times article is part of a pattern of uncritical acceptance afforded any seemingly positive data point about assimilation. Here is another one: “Fully 89 percent of U.S.-born Latinos spoke English proficiently in 2013,” according to a Pew Research Center report.10 That sounds moderately impressive, but the Pew data come from a Census question that asks, simply, “How well do you speak English?” Pew accepts self-assessments of “very well” as evidence of proficiency. What Pew did not mention was an objective test of English literacy administered by the Program for the International Assessment of Adult Competencies (PIAAC). U.S.-born Latinos score at just the 35th percentile on the PIAAC test, and 23 percent score at a level that is sometimes described as functional illiteracy.11 Furthermore, English adoption does not mean that Spanish disappears. Over 21 million native-born Americans now speak Spanish at home, up from 15 million in 2000. Pew itself has found that 30 percent of third generation Hispanics are at least as proficient in Spanish as they are in English.12
As the language example indicates, bromides about immigrant assimilation tell an airbrushed story, but policymakers seem uninterested in the harsher reality. After all, a deep dive into the data may tempt someone to suggest the policy option that must not be named.
Back on Campus
We began with the messy state of racial preferences in university admissions. Although rarely acknowledged, low-skilled immigration helped transform affirmative action from a program for black Americans into a more complicated ethnic balancing act. In particular, “Hispanic” is now a large and discrete student classification, requiring admissions preferences that rival the size and scope of affirmative action for black students. Because low-skilled immigrants do not assimilate within the several generations for which we have data, affirmative action for multiple groups now appears to be a more or less permanent feature of university admissions.
The problems exacerbated by low-skilled immigration go beyond admissions, however. University campuses have become microcosms of interethnic disputes in American society. The same kind of battles over fair representation and social justice occur within universities as they do in the wider world. The difference, however, is that the wealthy and protective campus environment enables such disputes to be more strident. Earlier this year, a student group at the Harvard Kennedy School distributed the “Faculty Pledge on Equity, Diversity, Inclusion & Belonging.” Explaining that the Kennedy School is “one of the worst environments for people from marginalized communities,” the pledge required professors to “receive specialized professional development detailing the ways in which their discipline [sic] have been shaped by race and racism and receive diversity or anti-racism training.” Professors would also need to document their own personal diversity efforts and submit them for “regular and systematic” review.
The pledge caused consternation on campus. Some professors saw it as a proposal for re-education camps and diversity tribunals. It appeared to be an attack on their academic freedom, and an attempt at intimidation to boot. Eventually, the dean’s office issued a statement promising no official pressure to sign, but it was little comfort. The us-against-them mindset of student activists now permeates the campus culture. Assuaging them has led to enormous amounts of time and money spent on campus diversity programs, including six-figure salaries for the administrators of various multicultural offices. Faculty hiring is now even more suffused with politics, as the candidates’ gender or skin color or sexual orientation competes with their scholarship as determinative factors. And as the faculty changes, so does the curriculum, with whole disciplines—e.g., “whiteness studies”—developed around the oppression mindset. Even the liberal arts, once thought to convey universal truths, now must be made “relevant” with course offerings that cater directly to various identity groups.
These conflicts have deep roots, but one aggravating factor is low-skilled immigration, which has added to the number of underrepresented groups that organize on their own behalf, both on campus and in the nation at large.