Thomas Piketty’s Questionable Data
Salim Furth /
Thomas Piketty made some questionable choices in adjusting and presenting the data that underlies his bestselling economics tome, “Capital in the Twenty-First Century.” Chris Giles, economics editor of the Financial Times newspaper, published a detailed list of apparent fudges in Piketty’s data.
Giles’ most explosive accusation is that Piketty chose data sources that were friendliest to his own preconceived ideas. For example, both the United States and the United Kingdom have two potential data sources for wealth: estate tax records and surveys of living households. In the U.S., Piketty uses the household survey, which showed rising wealth concentration. But in the U.K., he chose to use the inferior-quality estate tax data, which also showed rising wealth concentration. If he’d flipped both choices, he would have found falling inequality in the U.K. and steady inequality in the U.S. Giles is correct when he says, “Choices matter.” Giles’ estimates of U.K. wealth inequality in recent ecades are much lower than Piketty’s, and Piketty will need to defend his choices if we are to believe that U.K. wealth inequality has been rising.
Piketty presents data showing that wealth inequality rose slightly in Sweden from 2000 to 2010. But his “2000” data point actually is 2004 data, and his “2010” data point actually is an average of 2005 and 2006. When Giles used the data from 2000, he found that inequality actually fell slightly from 2000 to 2006 (the last year available). Perhaps Piketty had a good reason to use the years he did, but he has not offered an explanation.
These questionable choices have been reported as “errors” or “mistakes,” but the questions about Piketty’s data pertain to the choices he made, not the minor goofs. Historian Phillip Magness presents Piketty’s summary data on U.S. wealth inequality alongside its pre-1970 source. The graphs tell very different stories. Perhaps Piketty’s adjustments were valuable and moved the data in the right direction. But it is incumbent on Piketty to explain those adjustments, and it is incumbent on the reader to understand that the data was uncertain and incomplete to begin with and then was adjusted as the author believed necessary.
Even the best data on wealth distributions is uncertain. One of Piketty’s central ideas is that the amount and concentration of wealth has been rising steadily since 1980. He contends that the same economic forces are at work now and he projects the recent changes into the future. But if there is substantial uncertainty about each estimate and disagreement among data sources, then “trends” are highly subjective. As Yogi Berra may have said, “Predictions are hard to make, especially about the future.”
So how should we read Piketty? As others have noted, Capital can be divided into three components: history, prediction and prescription. One can believe the history without agreeing with Piketty’s predictions about the future. And if Piketty’s predictions are correct, he’s still wrong to prescribe brutal, confiscatory taxation, because that would increase poverty and lower wages, especially in poor countries.
What is at stake in Giles’ critique is Piketty’s account of history. Piketty’s story makes broad claims about global trends in the 19th and 20th centuries. If the trends turn out to depend on making specific choices, interpolations and adjustments in his collection of data, then we might have to conclude that predictions are hard to make, even about the past.