Most Americans have been wedded to the idea that the United States is a uniquely open country, where anyone can make it if they try. So strong is this belief that we have a phrase to encapsulate it: the American Dream. But until recently, there was little data available to confirm or disprove the truth of our self-perception. Today, the conventional wisdom among commentators, following an academic consensus, is that the American Dream is better pursued in other countries. Levels of economic mobility in the United States are comparatively low, not unusually high.
Recent research, however, complicates the picture dramatically. The new evidence does not suggest that the U.S. has especially high economic mobility, but it does indicate that America is not the international laggard that has been portrayed by earlier studies. In this multi-part series, I will lay out the case for this surprising conclusion. In this installment, I review how the old consensus developed and discuss the methodological details necessary for understanding why the early mobility research gave the wrong impression. The background information provided here should clear up some longstanding conceptual confusion pervasive even among mobility researchers today, who have tended to blur, without realizing it, the distinction between mobility and income inequality.
In 2006, two influential academic papers independently determined the U.S. to have low mobility relative to its peers. The first, by Canadian economist Miles Corak, concluded on the basis of an extensive review of the literature that the United States—and to a lesser extent, the United Kingdom—had much less mobility than France and Germany, which had less mobility than the Scandinavian countries and Canada. Along the same lines, Finnish economist Markus Jantti and his colleagues compared the United States with the Scandinavian countries and the United Kingdom and found lower levels of mobility in the U.S.
By 2008, the Pew Charitable Trusts could accurately describe the state of academic knowledge as showing that “the economic mobility of families across generations is lower in the United States than in many other countries.” Jason DeParle summarized the evidence for this position in a front-page New York Times story at the start of 2012. Regarding the new consensus that America is uniquely immobile, Isabel Sawhill of the Brookings Institution remarked (accurately), “I don’t think you’ll find too many people who will argue with that.”
There was reason to be skeptical though, as I had pointed out in a cover essay for National Review in late 2011. First, the cross-national literature on occupational and educational mobility is complicated and does not point to obviously lower mobility in the U.S.
Second, there was practically no cross-national evidence on the extent to which children grow up to have higher incomes than their parents. It is entirely possible that children in the U.S. are less able to move from “the bottom” to “the middle” than those in other countries, but because the income levels that constitute “the bottom” and “the middle” differ over time and across countries, American children might still end up better off than their parents in absolute terms, and to a greater extent than children in other countries. That would occur, for instance, if movement between “the bottom” and “the middle” were similar across all countries but stronger economic growth in the U.S. made everyone richer over time to a greater extent than in other countries. Perhaps when Americans think about the American Dream, it is “absolute mobility” we have in mind. At any rate, there is still no cross-national evidence on absolute mobility.
Instead of absolute mobility, we might be concerned that the U.S. lags other countries in terms of “relative mobility.” Relative mobility is about whether children can move from “the bottom” to “the middle” or “the top,” whether children starting at “the middle” can move to “the top” or tend to fall to “the bottom,” and whether children raised at “the top” fall “downward.” This sort of movement ignores the question of whether people in “the bottom” (regardless of where they came from or where they are going) are richer over time in absolute terms or whether people at “the top” become richer at a faster rate than people in “the bottom.” A child experiences (downward) relative mobility if she starts in, say, the middle fifth of family income but as an adult ends up in the second-lowest fifth, regardless of whether she is richer or poorer than her parents were. In theory, everyone could become richer than their parents, but if a person’s income rises relatively less than it does for other people, she might find herself in the second-lowest fifth of family income as an adult, as others pass her by. She experiences (downward) absolute mobility if she ends up poorer than her parents, regardless of whether she falls out of the middle fifth.
This absolute/relative distinction points to a third, more important, problem with the cross-national literature circa 2012. Namely, there was significant confusion about what exactly the mobility indicators that were most commonly used were measuring. Most research comparing mobility across countries has used a measure called the intergenerational elasticity, or IGE. Simply, the IGE indicates the extent to which having more parental income than someone else in childhood translates into having more income in adulthood than that person. When the IGE is high, it indicates that childhood income is more strongly associated with income in adulthood—that family origins matter more for adult income levels.
At the time, researchers routinely described the IGE as measuring relative mobility, a convention that continues today. But this was and is inaccurate. Contrasting the IGE with the “rank-rank slope” used by Harvard economist Raj Chetty and his colleagues in their groundbreaking work on mobility reveals why. Chetty begins by converting childhood and adulthood incomes into percentiles. If someone’s income makes them richer than ten percent of children but poorer than 90 percent of them, they are at the tenth percentile of the income distribution. After converting all childhood incomes into percentile ranks and all adulthood incomes into percentile ranks, Chetty and his colleagues determine how important childhood income rank is for adulthood income rank.
Since the incomes are all expressed in terms of percentiles, this rank-rank slope summarizes the relative mobility experienced by each child. It tells us the extent to which the degree and direction of individual relative mobility—movement in terms of percentiles—produces smaller percentile gaps between children in adulthood than existed in childhood. A rank-rank slope of 40 indicates that in adulthood, the difference between the poorest child (with a percentile rank near 0) and the richest child (with a percentile rank near 100) will tend to shrink from 100 percentiles to 40 percentiles.
Note that whether or not income levels or income inequality grows between generations is immaterial for the rank-rank slope. Differences across countries in relative mobility must also be independent of which countries are richer or poorer than others and which have more or less inequality.
Now consider the IGE. It summarizes all of the absolute mobility experienced by each child over a generation. Some children move up in absolute terms and others move down; some move up or down by much more than others do. The IGE tells us the extent to which the degree and direction of absolute mobility experienced by individual children produces more or less income inequality between them in adulthood than existed in childhood.
The absolute mobility summarized in the IGE is expressed in terms of percentage changes in income. This expression does not mean that the IGE is measuring relative mobility. Comparing poor children who tend to experience a 30 percent increase in income to middle-class children tending to see a 20 percent increase in income is not the same thing as comparing poor children who tend to move up 30 percentiles to middle-class children tending to move up 20 percentiles.
In practice, the pattern of absolute mobility in all countries and in all periods for which data is available is such that the initial income inequality between children falls in adulthood—the richer children tend to experience absolute percentage gains that are smaller than the percentage gains of poorer children (and income losses that are larger in percentage terms than the losses of poorer children). This is the well-known phenomenon of regression to the mean, and it is the reason why IGE values are smaller than one. (At the same time, rich children still tend to do better than poor children in adulthood, which is why IGEs are larger than zero.) The IGE, then, summarizes the extent to which absolute mobility across children narrows or widens the initial inequality between them.
Because the absolute mobility summarized in the IGE is expressed in percentage terms, differences in how rich or poor countries are do not affect IGE comparisons. However, the IGE can be higher in one country than in another depending on how much income inequality grows between generations in each country—even if their relative mobility rates are the same. When the IGE is higher in the U.S. than in Canada, indicating less “mobility,” that need not imply that it is more unusual for American children of fast-food workers to become top executives than it is for similarly situated Canadian children. It may simply mean that top executives make much more relative to fast-food workers in America than they do in Canada. As I wrote in my National Review essay, “In this scenario, it’s not that opportunities to obtain the best slots in the United States are less fairly distributed than in other countries, it’s simply much more lucrative to occupy those slots here.”
This is a subtle but crucial distinction that cuts to the heart of what we mean by “mobility” and “opportunity.” When most people speak of “equal opportunity,” they have in mind the idea that family background does not limit what a child realistically can aspire to be in adulthood. Poor children—no less than rich children—can make it to “the top.” This is fundamentally a value centered on relative mobility. If we had equal opportunity to fill any role in the economy, we might value equality of income to a greater or lesser extent. But that is a value distinct from equality of opportunity, and the two may be reflected in and addressed by a nation’s public policies independently of each other.
Imagine two scenarios. In Scenario A, we manage to reduce the income gap between top executives and fast-food workers, say, by raising the minimum wage or raising taxes at the top. But that is all we do, and it is no easier for the children of fast-food workers to become top executives than it ever was. In Scenario B, we manage to help the children of fast-food workers so that they are better able to become top executives in adulthood, say, by increasing their school readiness prior to kindergarten. But at the same time, the pay gap between fast-food workers and top executives widens.
Scenario A features declining income inequality but stagnant relative mobility. In contrast, Scenario B involves rising income inequality alongside increasing relative mobility. Relative mobility is lower in Scenario A than in Scenario B, but the IGE could very well indicate that “mobility” is higher in Scenario A. In the same way, comparing IGEs across countries can indicate that a country with slower growth in income inequality but lower relative mobility has more “mobility” than a country with greater income inequality growth but greater relative mobility.
The IGE blurs these distinctions. It may be useful for some purposes, but it cannot be considered a measure of “mobility.” Chetty and his coauthors argue that when inequality rises faster between generations while relative mobility is stable, so that the IGE is higher (worse), that indicates relatively lower opportunity. It does so because “the consequences of the ‘birth lottery’—the parents to whom a child is born—are larger.” That is, it may be no more difficult to reach the top, but if the gap between the top and the bottom grows, then people who can’t get to the top are that much worse off relative to the top. While this is true, it begs the question of why we should be concerned about inequality as distinct from relative mobility.
To return to my National Review article, I thus argued it was possible not only that the U.S. has higher absolute mobility than other countries but higher relative mobility too. Evidence from the Jantti study, however, indicated that the U.S. had worse relative mobility than other countries, relying on true measures of relative mobility. Corak’s subsequent work comparing the U.S. and Canada suggested the same.
Even here, however, there was reason to wonder about these cross-national comparisons. The Jantti paper found only small relative mobility differences between the U.S., Scandinavia, and the U.K.—with one exception. Upward mobility from the bottom among American men was lower than in the other nations. But upward and downward mobility from the middle and downward mobility from the top were similar across the countries, and upward mobility from the bottom among women differed little. As it turns out, even the upward mobility problems of American men in this study were overstated, as we shall see.
Such was the state of cross-national mobility research at the start of 2012.
In the next installment, I will show how the consensus solidified by Corak's Great Gatsby Curve has fallen apart as recent research by Corak and others has focused on measuring relative mobility better and more carefully. I will also reveal that the Jantti paper involved an apples-to-oranges comparison of Scandinavia to the U.S. that obscured the similarity in relative mobility between the two.
Read Part II of this series here.
Scott Winship is the Walter B. Wriston Fellow at the Manhattan Institute for Policy Research. You can follow him on Twitter here.
Interested in real economic insights? Want to stay ahead of the competition? Each weekday morning, e21 delivers a short email that includes e21 exclusive commentaries and the latest market news and updates from Washington. Sign up for the e21 Morning eBrief.