News & Politics » Miscellany

Land of Dreams

It's a cherished American truism that humble beginnings don't limit a person's ability to succeed. But a local economist has good evidence that, in fact, they do and our society is nowhere near as mobile as we thought.

by

comment

America is supposed to be a place where anybody who wants to can move up the economic ladder through his or her own efforts. People generally believe that having poor parents isn't much of a handicap. But is this assumption true?

The few economists who asked the question in the 1970s and '80s thought it was. They found only a 16 to 20 percent correlation between the earnings of parents, rich or poor, and the earnings of their children. That's about what you'd expect if your economic life is pretty much what you make of it and not what your parents make of it for you. So when University of Chicago economist Gary Becker gave his presidential address to the American Economic Association in 1987, he was expressing both a professional consensus and an American piety when he said, "Low earnings as well as high earnings are not strongly transmitted from fathers to sons."

But ever since then, with little fanfare, that consensus has been unraveling, as economists pile up evidence that earnings success in the United States is more inherited than achieved. Unfortunately, they've had less success in figuring out how it's inherited.

In 1992 Gary Solon of the University of Michigan and David Zimmerman of Williams College began the demolition job when they published separate studies in the American Economic Review. They both found that the true correlation between parents' and children's earnings wasn't 16 to 20 percent but 40 percent or higher. The country hadn't changed; in Solon's words, the earlier research had simply been based on "error-ridden data, unrepresentative samples, or both." In 1997, in his book Parental Priorities and Economic Inequality, U. of C.'s Casey Mulligan upped the ante again: "More than 50 percent of earnings differences among parents are passed on to children."

This summer, in an article that should have made news but didn't, Bhashkar Mazumder of the Federal Reserve Bank of Chicago examined the biggest and best set of data yet, and concluded that the U.S. is a "relatively immobile society" in which parents pass along a strong 60 percent of their earnings advantage (or disadvantage) to their children. His research was published in the September issue of the bank's Chicago Fed Letter and has been submitted to a peer-reviewed economics journal.

If Mazumder is right, it's much harder to climb the economic ladder--or fall down it--than most people believe. And that means existing inequalities in American society won't go away quickly. Even in America, the most important decision you'll ever make is choosing the right parents.

The economists who've kept coming up with increasingly dire numbers haven't been following some new academic fashion. They've just been examining larger and more representative groups over longer periods of time. All the studies reached their conclusions using a similar process. They recorded the earnings of a parent and compared them to the earnings of a grown child of that parent. For historical reasons, these parent-child pairs usually consist of fathers and sons, but Mazumder says father-daughter comparisons look about the same.

The more parent-child pairs the economists can get information on, and the more representative these pairs are of the general population, the more reliable the results will be. Even more important, says Mazumder, is having information on what the pairs earned for a number of years. The earliest studies sometimes used only one year of earnings, which isn't a reliable indicator of lifetime earnings for anyone. Mazumder's data set, drawn from confidential social security data files, enabled him to average together many more years than any of his predecessors.

Mazumder and his fellow economists then combined this parent-child information with a century-old insight from Victorian social scientist Francis Galton. He observed that children of tall parents tend to be taller than average, but not as tall as their parents. Similarly, children of short parents tend to be shorter than average, but not as short as their parents. Galton called this phenomenon "regression to the mean"--that is, extremes even out over time. Experience has shown that the same tendency applies to extremes of wealth and poverty. Children of rich parents are better off than average, but usually not quite as well-off as their parents; children of the poor are worse off than average, but usually not as poor as their parents.

The first question Becker, Solon, Zimmerman, Mulligan, Mazumder, and others want to answer is, how fast does this evening-out process work? In theory it could happen in a single generation--or it might not happen at all.

At one extreme, in a society with a zero percent correlation, your earnings would be entirely unaffected by your parents' earnings, and as a result society-wide inequalities in one generation would be randomized as the next generation grew up. At the opposite end of the spectrum, in a society with a 100 percent correlation, each child would earn exactly what his father did relative to others in society. The result would be that any society-wide inequalities would be preserved intact over time. (If an economy worked the way Marx thought capitalism worked, with inequality steadily increasing generation after generation, the percentage would already exceed 100. It doesn't.)

Real societies fall at various points between these extremes. For years after Solon's and Zimmerman's 1992 articles the consensus among economists was that the correlation in this country hovered around 40 percent. Mazumder's work puts us at 60 percent--which is especially disquieting since the best study available for Canada (using good data) suggests that the correlation there is only 20 percent.

Mazumder brings these numbers closer to real life by translating the percentages into time. In a 20 percent society, it takes only one generation--25 years or so--for an inequality to substantially dissipate. In a 60 percent society, by contrast, it takes three generations--under ideal circumstances. He strongly suspects that it takes even longer for those on the bottom; preliminary research suggests that earnings transmission for low-earning parents and children may be significantly higher than for those with high earnings.

Enough diagnosis. This isn't the America we thought we lived in. How can we fix it?

The economists don't know. Parents give their children eye color, hair color, stature, native intelligence, race, attitudes toward life, familiarity (or lack thereof) with jobs, and education. But they don't give their children a head start or a delayed start in lifetime earnings in any direct or obvious way. (They might give them stocks or bonds, but that would be income or wealth, not earnings.) What mix or combination of factors explains that 60 percent transmission of earnings?

Economists Samuel Bowles and Herbert Gintis summarize the research on this question in a July 14 draft paper (www.umass.edu/preferen/gintis/intergen.pdf). They found that the correlation between parents' and children's earnings does have something to do with wealth, race, attitudes (fatalistic or not), IQ, and education. But all these factors combined don't predict even half of the correlation. Even when the quality of schooling is factored in, "more than half of the intergenerational transmission coefficient is unaccounted for."

Mazumder nevertheless suspects that schooling and its quality may be a more important factor for the poor, because of what he calls "borrowing constraints." Poor parents are often unable to raise the money to move to a neighborhood with better schools or to borrow money to send their kids to a good college. This is hardly news; the question is whether either of these constraints is crucial to keeping the children's earnings down. With a new set of data large enough that he can zero in on low-income parent-child pairs, Mazumder hopes to be able to confirm or deny his plausible hypothesis.

Mazumder is cautious when asked to speculate on social remedies that would reduce the 60 percent correlation. "I see this number as a descriptive statistic," he says. "It's not clear that there's any optimal number, but it is a signal of where we are as a society. We should look more closely at what it may imply," and then look for "appropriate policies to foster greater income mobility."

Art accompanying story in printed newspaper (not available in this archive): photo/Yvette Marie Dostani.

Add a comment