In 1923 and 1924, thanks to a short-lived law allowing the publication of tax data, the New York Times reported the tax payments of more than 30,000 Americans.
Some of the names of wealthier taxpayers on that list were rare, meaning that their descendants can be identified by name today — names such as Vanderbilt, Roosevelt and Winthrop.
Generations later, in 2000, those same last names still signaled high status — they were overrepresented among physicians by a factor of nearly three.
That is one of many such findings discovered by Gregory Clark, a University of California, Davis, economist who undertook a wide-ranging study of surnames throughout history. His conclusion, which has enormous implications for politics, policy and social science, is that mobility is far lower than is commonly thought.
Clark's research on the persistence of surnames' status, published as a book titled The Son Also Rises, suggests that nothing -- not even the introduction of the welfare state in Europe or the the Cultural Revolution in China -- does much to change social mobility.
For example, the names of the Norman conquerers of England who were recorded as property owners in the Domesday Book of 1086 -- names such as Sinclair, De Arcis (modernized as D'Arcy) and De Mon Gomerie (Montgomery) -- were still about 25 percent overrepresented in the names of students enrolled at Cambridge and Oxford in 2012.
That means the Normans’ social advantage hadn’t totally washed out nearly a millennium later.
The same stickiness holds true for low-status people. Surnames associated with the poorest Hindu groups of India’s colonial era are “extremely rare” among physicians — a representative high-status group. Today, names widely adopted by emancipated U.S. slaves lacking last names after the Civil War, such as Washington and Merriweather, appeared among physicians at just one-third of their frequency in the general population in 2000.
The bottom line is Clark estimates that the persistence rate of social status is in the range of 0.75, meaning that about half a given person’s outcome is predicted by his ancestors, in every time and place.
That estimate is far higher than mainstream estimates of income or wealth mobility, which place the intergenerational elasticity of income or wealth in the United States at between 0.4 and 0.6., which works out to about a quarter or one-third of outcomes explained by inheritance. It is significantly lower in some Scandinavian countries.
But Clark explains that his estimates differ for a few reasons. One is that by analyzing surnames he’s not limited to comparing two generations, a constraint imposed on most researchers because of the lack of data.
Another is that he measures social status, as opposed to income. While a philosopher and a stockbroker might both be high-status, he explained, “their income is going to be different.”
And regression to the mean plays a large role. If an individual with high underlying social competence has a bad outcome, his descendants would nevertheless tend, slowly, back toward higher status.
Clark predicted in his book that his conclusions would prove controversial.
“He’s right that that’s going to be controversial,” said Erin Currier, the director of the Economic Mobility project at Pew Charitable Trusts, noting that Clark’s estimates of mobility fall don’t “jibe” with the mainstream (she hadn’t reviewed Clark’s research).
"The point I would really disagree with is the idea that mobility is predetermined,” said Currier, pointing to recent research led by Harvard's Raj Chetty that found significant differences in mobility among regions of the United States with differing policies.
Clark responded that the true underlying level of mobility likely doesn't differ across the United States, as Chetty found, given that Clark's research suggests it doesn’t even vary between the United States and Sweden.
Instead, Chetty's findings might be explained by differing elements of the social structure, such as race, that cause differences in income mobility — not status.
Furthermore, Clark noted, so far no one has proposed an alternative interpretation of his data.