One reason so many Americans tolerate inequality is their belief that it’s not a permanent condition. Yes, you might start out life without a lot of money. But if you work hard and play by the rules, then you’ll get ahead. You might never become a millionaire, but you’ll still find your way into the middle class. And then your kids will have a shot to do even better. Experts call this income mobility. The rest of us call it "the American dream."
But for too many of us, it really is a dream—and nothing more. The comparison to other developed nations is striking. In Europe and in Asia, the countries most similar to the U.S. have both more equality and more mobility. In other words, they have fewer rungs on the income ladder to climb, and they have an easier time making each step.
As my former colleague Tim Noah wrote last year, “A nation that prides itself on its lack of class rigidity has, in short, become significantly more economically rigid than many other developed countries.” This has (or should have) significant political implications, particularly at a time when economic growth seems both too slow and too uneven. “While liberals often complain that the United States has unusually large income gaps, many conservatives have argued that the system is fair because mobility is especially high, too: everyone can climb the ladder,” Jason DeParle observed last year in The New York Times. “Now the evidence suggests that America is not only less equal, but also less mobile.”
Scholars have been exploring these trends for a while, trying to understand exactly why it’s gotten so hard to get ahead in the U.S. And that’s produced some fascinating pieces of research. The latest comes from “The Equality of Opportunity Project,” an ongoing effort led by a team of scholars at Harvard and the University of California-Berkeley—Raj Chetty, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez. You may have heard or read about their new study, which was the subject of a front-page story by David Leonhardt in the Times last week. The study’s key insight was that income mobility is not uniform across the U.S. It varies enormously by region. And the regional pattern has some familiar characteristics, with mobility generally low across the poorest parts of the South.
Does the data really tell such a clear story? What policy implications can we draw from that? And what’s next for this team of researchers? To answer those questions, I conducted an email interview with Professor Hendren. Our conversation appears below. It’s lightly edited—very, very lightly. Hendren got pretty wonky with some of the numbers, but I thought more specialized readers might actually appreciate that.
Jonathan Cohn: Let's start with a very general question. Income mobility is obviously a subject that's received a lot of attention from scholars over the years. This study—how does it add to our understanding? Or, to put it a little more bluntly, what does it tell us that we didn't know before?
Nathaniel Hendren: You're absolutely right that there have been many previous studies focused on intergenerational income mobility, especially in the U.S. I think the main innovation in this study is to look at the geographic variation in this mobility across the U.S. Here, we were surprised to find substantial variation. For example, on average a kid born in Charlotte, NC whose parent earns less than 25K per year has a 4.3% chance of earning more than $100K at age 45. In contrast, a kid born in Salt Lake City, UT to parents earning less than 25K per year has an 11% chance of earning more than $100K. So, I think we knew before that kids from richer families, on average, earn more money than kids from poor families. But, I don't think we previously knew the extent to which parental income determined kid income across the U.S.
JC: The geographic pattern is one of the most interesting parts of your research. But before we get to that, let's drill a bit deeper on what might be driving the changes in mobility. In the research you guys basically tested a bunch of different factors, everything from tax breaks for working families to the prevalence of two-parent families. Which ones correlated most strongly with mobility, and which ones didn't? I realize correlation isn't the same as causality.
NH: Yes, that's right. The initial motivation for the study was to analyze the impact of tax policies, such as the earned income tax credit, on intergenerational mobility. What we found were modest correlations between state and local tax policies and mobility, but I think it's clear that there is a lot of remaining variation to be explained.
To this aim, we explored many additional factors. The factors that were most highly correlated fell into four broad categories:
1. The quality of the K-12 education system (things like average test scores, dropout rates, etc.)
2. Economic inequality and segregation (the size of the middle class in a local area, income and racial segregation)
3. Measures of the strength of social ties and social capital. There are many proxies here that we borrowed from previous work done by sociologists, such as census response rates and voter turnout rates. These were positively related to mobility.
4. Measures of the strength of the family. The fraction of kids born to a single mother in an area and the divorce rate of an area were all very highly negatively correlated with mobility.
Of all four factors, the family structure variables—in particular the fraction of kids born to single mothers—were the most strongly correlated with mobility. But, it's important to note this is not the impact of having a single parent on one's kid. In fact, we found lower rates of mobility for kids of married parents in areas with a higher fraction of single mothers. Therefore, the correlation appears to be picking up more of a property of the location than the family per se.
JC: It sounds like sorting out the causality of this is difficult. We don't know, for example, whether single-parent families are a cause of lower mobility or a consequence of it. Of course, both things could be true, right?
NH: Yes, that's right. What we can say is that it's not entirely an effect operating within single-parent households that drives the spatial differences across the U.S.
JC: Let's talk about the geographic pattern. For the most part, we're seeing lower mobility where we see the most poor people—i.e., in the Deep South. Or am I wrong about that? We're also seeing higher mobility in areas with more active government—higher taxes, more generous safety net programs—although there's one part of the country that bucks this pattern. In parts of the Plains and Rocky Mountain states, with pretty minimal government, we're also seeing a lot of mobility.
NH: Your first statement is, perhaps surprisingly, not really what's going on. It's true the South has generally lower rates of upward mobility. But if you correlate the measures of upward mobility with mean parent income of an area the correlations are generally small. Recall, we're thinking of upward mobility as the outcomes of kids whose parents earn a fixed amount of money. So, there's not really any reason to think that upward mobility is higher or lower in a place with greater income. Even if the maximum income in a place is, say $100K, a kid could leave town and go get a job in NY earning $1M. We're measuring upward mobility using the place where the kid grew up regardless of where they live in adulthood. This is particularly relevant if you look at the rural/urban split. In general, mobility is lower in urban areas. But, we know from other research that the highest paying jobs are in cities.
JC: A reader to Andrew Sullivan's blog The Dish noted that mobility seems to track pretty closely to the concentration of African American communities. On the other hand, didn't you control for race? I'm curious what, if anything, you make of the apparent correlation.
NH: Thanks for the opportunity to clear up the race question a bit. It is absolutely true (and obvious from the link you send) that the fraction of African Americans is strongly negatively correlated with upward mobility. So, why don't we focus on race?
Three reasons … First, although we do not observe race in our data, we analyze upward mobility using a subset of the population who live in zip codes that are at least 80 percent white. The resulting map is very highly correlated (>0.7). Put differently, poor whites in the south have lower rates of upward mobility than poor whites in San Francisco. One might be concerned that the 80 percent threshold is not enough, but here I refer any interested reader to Figure A6 in the paper, which replicates this correlation for varying thresholds between zero and 95 percent. The pattern is clear: Although upward mobility is lower in areas with higher percentages of blacks, it's higher for both poor whites and blacks.
Second, although the correlation between upward mobility and race is strong, the correlation goes away if you control for the fraction of people born to single parents in an area. (See Table 7, columns 3 and 4 in the paper). In contrast, the four sets of factors I mentioned previously all remain correlated with upward mobility in a multivariate regression. In this sense, race is not a robust correlate of the amount of upward mobility in a place.
Third, we have an ongoing analysis looking at the experiences of people who move across cities in the U.S (see Figures A7-A8). Here, the results suggest if poor parents move to a less (more) upwardly mobile place, their kids do worse (better). Obviously, ones race doesn't change when they move. So, the map really seems to be about differences in places that cannot simply be explained by differences in races or racial composition.
It's place not race.
JC: I'm just curious—what kind of policy lessons do you draw from these results? Or if the data is too inconclusive to draw firm lessons, what kind of hunches do you have—say, about what kind of investments make sense?
NH: I'm very cautious to draw strong policy lessons from our results. What we're really hoping to do is to highlight the dramatic differences in mobility across areas in the U.S., with the hopes that we can learn something from these differences. At this point, I don't have a recommendation to give to a city like Atlanta on how to obtain mobility rates like Seattle. But, if you wanted me to point to one factor that I think could improve mobility given what we know now, I'd say schooling. We know not only from our study, but also from much previous work that schooling is fundamental for improving economic outcomes. Cliché as it sounds, ensuring equality of access to quality education is at the top of my list.
JC: Is this an ongoing project—i.e., should we expect more research from you guys, perhaps from the same data set? If so, can you tell me what you're exploring?
NH: Yes, we'll be busy. Our immediate next step is understanding the experiences of kids who move from one place to another. Beyond that, we hope to explore more specific types of policies we can undertake to promote mobility. For example, we know that segregation has a negative correlation with upward mobility. A natural follow-up question is whether transportation policies that decrease the functional barriers imposed by segregation can help foster greater upward mobility. It's these types of questions we hope to explore. But we hope we're not the only ones in this endeavor. By making the mobility data available online, we hope other researchers will be motivated to use the data to study their own hypotheses about what can be done to improve social mobility in the U.S.
Lede image via Shutterstock.