People tend to like big, sweeping theories of economic history. When Karl Marx foretold a supposedly inevitable series of class conflicts and revolutions that would end in a communist utopia, the idea was so powerful that it inspired revolutions, alternative economic systems and wars. Other thinkers depicted economic history as the triumph of a particular culture, or the inevitable ascendance of free-market capitalism.
French economist Thomas Piketty, the latest to win popular acclaim for a grand theory of economic history, was a bit different. Unlike some of those who came before, Piketty brought a wealth of data. Painstakingly exploring historical records, he constructed measures of income and wealth going back centuries and covering a number of different countries.
His thesis was also unusually simple—instead of a complex theory of social class or the vagaries of human culture, he merely predicted that the rich get richer until a big war or revolution resets things. His reams of historical data purported to bear this pattern out—the wealth and income shares of the rich tend to rise and rise, unless a calamity like World War II temporarily levels the playing field.
Whenever a thesis like this gains popularity—Piketty’s much lauded book was a best-seller—smart people are going to line up to take a shot at it. Broad theories of history, after all, are always too simple to be true in their purest form. History is a complex process, for one thing, and overarching theories are gross simplifications. Also, because history only happens once, it’s impossible to verify a historical theory with anywhere near the degree of empirical confidence of a scientific theory.
Early Piketty critiques often focused on his calculation of capital’s share of income. Many noted that Piketty didn’t take capital depreciation into account. Matt Rognlie, now a professor at Northwestern University, estimated that most of the rise in capital income was due to the increased value of land, not to corporations eating the world. These are problems for Piketty’s theory of why the rich get richer, but they don’t really contradict his historical argument that the rich usually do get richer. A few people raised doubts about Piketty’s data, but overall the criticisms did little to blunt the consensus that the French economist’s work was magisterial in that regard.
Now, however some economists are going back and checking Piketty’s history more thoroughly. Richard Sutch of the University of California—Riverside has a new paper arguing that Piketty’s measurement of the share of wealth owned by the richest Americans is deeply flawed and unreliable.
Sutch points out that Piketty’s wealth data, which goes back to 1810, patches together two very different data sources—the early data comes from estate tax records and covers individuals, while the later data comes from government surveys and covers households. In order to compare wealth patterns in the 20th century with those in the 19th, Piketty contrasts household wealth in the former against individual wealth in the latter, when the two can in fact behave very differently.
Sutch argues that Piketty’s procedure for harmonizing the two data sources is simplistic and arbitrary. He also identifies some places where Piketty seems to have introduced fudge factors into his analysis, drew a line across a period in the 1970s where the household data had a gap, or apparently made calculation errors.
For the 20th century, Sutch concedes that his concerns are relatively minor, and that the broad pattern Piketty describes isn’t in very much doubt. But the 19th century is a different story. Sutch contends that much of Piketty’s analysis of that century is based on a single data point from 1870, collected by a potentially unreliable analysis of census records, and massaged by yet more fudge factors and guesses. Piketty also has a data point for 1810, but Sutch argues that Piketty essentially made up this data point based on older data from 1774.
The scant quantity and unreliability of the 19th century data, and the heroic assumptions Piketty makes in order to fit that data, drive Sutch to exasperation:
“(Piketty’s) heavily manipulated data, the lack of clarity about the procedures used to harmonize and average the data, the insufficient documentation, and the spreadsheet errors are more than annoying. Very little of value can be salvaged from Piketty’s treatment of data from the nineteenth century.”
Others are sure to check Sutch’s criticisms carefully. But if his analysis holds up, it will be the most spectacular refutation the economics profession has seen since the Reinhart-Rogoff affair back in 2013.
The broad strokes of Piketty’s analysis of 20th century inequality are almost certain to survive intact. There’s too much high-quality data there. That inequality in the US and other developed nations rose in the early 20th century, fell in the mid-century, and then rose again in the latter decades isn’t in doubt.
But this one-century pattern is too short to prove anything about grand cycles of history. Without reliable information on the 19th century, it will be quite a stretch to conclude that the rich usually tend to get richer.
That doesn’t mean Piketty is wrong. His theory could still be valid. And there’s no question that the rise in inequality since 1980 is concerning, or that the parallels with the Gilded Age a century earlier are disturbing. But it’s always important to remember the difficulties and dangers in crafting all-embracing narratives of economic history. Bloomberg View
Noah Smith is a Bloomberg View columnist.
Comments are welcome at [email protected]