Raj Chetty’s Family Problem

Why are some poor kids able to escape their meager circumstances—and contra today’s widespread cynicism about the American dream, a significant number still do—while others stay tangled in poverty’s net? No researcher has dug into this question more deeply or more fastidiously than Harvard economist Raj Chetty, chief author of two new, much celebrated, papers (see here and here). In many respects, the hype is justified, but one critical evasion has gone unnoticed by an overeager media.  

Let’s begin, though, with the papers’ primary finding. The authors conclude that cross-class relationships, or what they call “economic connectedness” (EC), are the strongest predictor of low-income children’s success. More colloquially, poor children who get to mingle with higher-income people (meaning higher than median income, not “rich” or “wealthy” as many journalists have mistakenly written) earn more as adults than poor children who only interact with other poor people. A “dosage effect” supports the theory: the younger children are when they move to an area with strong EC, the stronger their chance of moving up the income ladder as adults.  

The thesis echoes a long-standing sociological insight about “concentrated poverty.” More than 30 years ago, William Julius Wilson perceived that as civil rights advances of the 1960’s opened up more residential opportunities, middle-class and stably-employed, working-class blacks would leave inner city ghettoes for more affluent communities. When neighborhoods were emptied of their middle-class residents like doctors, teachers, and social workers, those areas devolved into scenes of crime and disorder. Perhaps the best-known examples of this dynamic are urban housing projects like Cabrini Green in Chicago, whose dysfunction became so bad officials offered residents housing vouchers with the hope they would use them to move to more economically integrated neighborhoods. They then promptly destroyed buildings that four decades earlier were believed to be a godsend for the poor.   

Chetty’s new research enhances our understanding of social capital and thereby tweaks our notion of concentrated poverty. Where earlier studies might have compared, say, high school graduation or teen pregnancy rates in different places, access to massive IRS and Facebook data sets gave the scholars the chance to drill down to the level of zip codes and census tracks. Add to that a decades-worth of both policy trial and error and research on social capital, and the authors are now able to tease out patterns and draw distinctions invisible to previous scholars. They discovered that though it tends to be highly correlated with concentrated poverty—almost none of the lowest-income zip codes in the United States exhibit high levels of EC—connectedness varies not just across counties but also across neighborhoods within counties. For reasons that are still unclear, some poor districts manage to overcome their economic isolation. Low-income areas of Minneapolis, for example, are more connected across SES lines than those in Indianapolis, and, in fact, low-income children who grow up in the former have higher incomes than those who grow up in the latter. (Note to aspiring poverty researchers: get thee to Minneapolis, ASAP.)   

It’s no slight against Chetty and his co-authors to point out that “economic connectedness” is an idea well suited to Facebook data: 70 million ties, or “friends,” are an invitation to study exactly that. But the authors are careful to test EC’s clout by evaluating how much two other kinds of social capital affect mobility. They examine zip codes for levels of Robert Putnam-style civic engagement using rates of volunteerism as a proxy. They also look at social cohesion or clustering, “the rate at which two friends of a given person are in turn friends with each other.” Most people might reasonably assume that both these two kinds of social capital would be a good thing for any community. They could well be, but their benefits do not include enabling children’s upward mobility. Only EC has the magic touch. In fact, in Akron, Ohio—an example mentioned in the paper—social cohesion and upward mobility are actually negatively correlated. Close ties may sound nice, but parents have always worried about their children hanging around “the wrong sort of people” for good reason. It’s one reason urban charter schools have such long waiting lists. 

Economic connectedness is not an easy problem to address. Though they didn’t use the term, ancient philosophers were well aware that homophily—the attraction of like to like—is a powerful force in human affairs. In modern times, sociologists and social network theorists have shown how people are drawn to others of the same age, place of birth, race, religion, and hobbies; social and economic status—SES—is no different. Chetty’s group finds “an almost linear relationship between the SES of an individual and the average SES of their friends,” an example of what they call “friending bias.” In large, mixed-income high schools, for instance, kids tend to sort themselves into groups of friends from similar backgrounds. (Smaller high schools simply have fewer people in which students can find other kids of “their kind.”) Notably, the only institution that successfully promotes economic connectedness is churches; evidently religious homophily is more powerful than SES homophily.  

These are all impressive results. The problem is that Chetty and his team leave the impression that economic connectedness is (if not the sole) by far the most powerful association with a neighborhood’s upward mobility. But look carefully at the study, and you’ll find that’s not quite true. Several charts accompanying the text of the main paper compare economic connectedness with other neighborhood characteristics often assumed to make a difference in kids’ life outcomes. A lot of them turn out to be pretty weak when you control for EC.

Segregation? A negative for upward mobility, but only a small one. Neighborhoods with high employment and job growth? A small positive. Inequality? A modest negative. The share of black residents? Now, we’re looking at a meaningful (negative) effect on upward mobility, although it’s worth noting the researchers also find that in largely black communities, economic connectedness can offset some of the disadvantage. William Julius Wilson, one of the first to describe the problem of isolated poverty, would not be surprised to learn that EC has as much positive impact on upward mobility in black as in predominantly white communities. 

There is one neighborhood characteristic that actually rivals economic connectedness in explaining upward mobility, however: family structure. In the first paper’s univariate analysis, the proportion of single-parent households has a more powerful (negative) correlation with upward mobility in their analysis of zip code data than EC has a positive. (The multivariate analysis of zip code data suggests that family structure is about as predictive as economic connectedness). You can find this information in several charts included in the first paper and in supplementary figures associated with it, yet Chetty and his team mention family structure only in passing in the discussion. Media coverage followed his lead. 

What makes this oversight—if that’s what it is—especially striking is that family structure, as a powerful variable in strengthening neighborhoods and children’s upward mobility, has popped up in a previous Chetty paper, where it was once again largely avoided. His 2018 study, “Race and Economic Opportunity in the United States,” found that adult employment rates are higher and school suspension rates are lower when black boys grow up in “areas with higher black father presence,” even if their own father is not living with them. An executive summary of the paper concludes that “what matters is not parental marital status itself, but rather community-level factors associated with the presence of fathers, such as role-model effects or changes in social norms.” But in the large majority of cases where a father lives with the child—i.e., “father presence”—the parents are married. As Brad Wilcox wrote in a detailed take-down in these pages, the fact that most married-couple households have two workers distorts the paper’s findings on adult household income. In any case, if the share of two-parent households has a significant positive effect on neighborhoods’ social environment and upward mobility in 2018, you’d think it would be worth at least a nod, and hopefully more than that, when it shows up in 2022. It might even be that neighborhood family make-up has some positive correlation with economic connectedness.

We know something about why children tend to do better growing up in two-parent families, but why does EC also advance upward mobility? The most obvious possibility is that higher-income people bring news to less plugged-in acquaintances about job and education opportunities. A 2019 paper from the Philadelphia Federal Reserve found that children living in a gentrifying area—that is, a lower income area that has had an influx of college-educated professionals—were more likely to go to college than those living in ungentrified communities. In its coverage of the Chetty research, The New York Times interviewed a young woman from a troubled, working-class family who didn’t know anything about the SAT until she learned about the test from high school friends from more well-to-do families. The parents of her friends looked over her college essays, and after she became the first in her family to get a post graduate degree, friends of those friends connected her to her current job as a criminal defense attorney.  

But that’s an easy case. Middle-class people also influence poorer friends and acquaintances in less concrete ways. Years ago, I interviewed a number of poor teen mothers. When I asked them if they had wanted to get married before having children, they were nonplussed. It gradually emerged that with the exception of some characters on television sitcoms, none of them knew any married couples. Two-parent families were as foreign an idea to these young mothers as the SAT had been for the criminal defense attorney profiled by the Times. Inter-class contact doesn’t just give the poor “role models,” it expands people’s understandings of possible life scripts and gives them a tangible vision of a better future, including a stable family life.

The prevalence of single-parent families is an uncomfortable topic for many people these days; all the more reason for scholars studying upward mobility to deal with it frankly.

Kay S. Hymowitz is the William E. Simon Fellow at the Manhattan Institute and a contributing editor of City Journal. She writes extensively on childhood, family issues, poverty, and cultural change in America.