November 14, 2002
The Apple Falls Close to the Tree
seems increasingly apparent that the secret to success is to have a successful parent. Consider some prominent examples: George H. W. Bush and George W. Bush; Bobby Bonds and Barry Bonds; Henry Fonda and Jane Fonda; Estée Lauder and Ronald Lauder; Julio Iglesias and Enrique Iglesias; Sam Walton and Jim, John, S. Robson and Alice Walton.
As more recent and better data have become available, economists have marked up their estimate of the impact of parents' socioeconomic status on their children's likelihood of economic success.
It turns out that the famous line attributed to Andrew Carnegie — "from shirt-sleeves to shirt-sleeves in three generations" — is an understatement. Five or six generations are probably required, on average, to erase the advantages or disadvantages of one's economic origins.
This represents a marked departure from past thinking. In the 1980's, when Gary S. Becker of the University of Chicago pioneered the economic theory of intergenerational transmission of economic status, it was believed that the correlation between a father's and son's income was only around 0.15 — less than half the correlation between fathers' and sons' heights.
The early studies suggested that if a father's income was twice the average, his son's expected income would be 15 percent above average, and his grandson's just 2 percent above average. This is fast "regression to the mean," a concept Sir Francis Galton used to describe the progression of offspring toward the average height.
Landmark studies published by Gary Solon of the University of Michigan and David J. Zimmerman of Williams College in The American Economic Review a decade ago, however, led economists to revise substantially upward the estimate of the similarity of fathers' and sons' incomes. They noted that income fluctuated for idiosyncratic reasons from year to year — an employee could lose a job, for example — so estimates that depended on a single year were based on "noisy" data. Also, the samples previously analyzed represented only a narrow slice of the population at different points in individual careers. These factors caused the correlation in annual incomes to understate the correlation in "lifetime" incomes.
Averaging earnings over five years produced a correlation of around 0.40 for fathers' and sons' earnings — the same as the correlation between their heights. If people's incomes were represented by their heights, the similarity in income between generations would resemble the similarity observed in the heights of fathers and sons.
New studies by Bhashkar Mazumder of the Federal Reserve Bank of Chicago suggest that the similarity in income is even greater. Using Social Security records, he averaged fathers' earnings over 16 years (1970 through 1985) and sons' earnings over four years (1995 through 1998), and found that around 65 percent of the earnings advantage of fathers was transmitted to sons. The wider window provides a better reflection of lifetime earnings.
The relationship between fathers' and daughters' earnings was just as strong.
So that grandson (or granddaughter) mentioned previously could expect to earn 42 percent more than average. After five generations, the earnings advantage would still be 12 percent.
Furthermore, the degree of persistence across generations is strong for both rich and poor. Thomas Hertz of American University finds that a child born in the bottom 10 percent of families ranked by income has a 31 percent chance of ending up there as an adult and a 51 percent chance of ending up in the bottom 20 percent, while one born in the top 10 percent has a 30 percent chance of staying there and a 43 percent chance of being in the top 20 percent.
In another study, David I. Levine of Berkeley and Dr. Mazumder found that the impact of parental income on adult sons' income increased from 1980 to the early 1990's.
Why is there such a strong connection between parents' socioeconomic status and their children's? A large part of the answer involves intergenerational transmission of cognitive ability and educational level.
But these factors can "explain at most three-fifths of the intergenerational transmission of economic status," Samuel Bowles and Herbert Gintis of the University of Massachusetts wrote in the latest issue of The Journal of Economic Perspectives. They suggest that the intergenerational transmission of race, geographical location, height, beauty, health status and personality also plays a significant role.
Arthur S. Goldberger of the University of Wisconsin has long questioned whether knowledge of the "heritability" of income is of much use. Even if the father-son correlation is high because traits that affect earning power are inherited, well-designed interventions could still be cost effective and improve the lot of the disadvantaged.
To take an extreme example, the correlation in incomes between fathers and sons was high in South Africa under apartheid because race is an inherited trait. The abolition of apartheid reduced the correlation. The organization of society matters.
Perhaps the only legitimate use of the intergenerational correlation in income is to characterize economic mobility. The data challenge the notion that the United States is an exceptionally mobile society. If the United States stands out in comparison with other countries, it is in having a more static distribution of income across generations with fewer opportunities for advancement.
Anders Björklund of Stockholm University and Markus Jäntti of the University of Tampere in Finland, for example, find more economic mobility in Sweden than in the United States. Only South Africa and Britain have as little mobility across generations as the United States.
Luke Skywalker and Darth Vader are an unusual father-son pair; in most families, the apple does not fall so far from the tree.