Piketty's U.S. Wealth Figures Match Recent Sources

There is a piece in FT critical of Piketty's data. It's long and involves multiple countries. I will only focus on the US here as that is what I know. The good data we have on wealth inequality in the US is broadly consistent with Piketty's story.

Piketty's US wealth data sources are a combination of estimates based on old estate tax data and then more recent cross-sectional surveys conducted by the Federal Reserve (sometimes referred to variously as Wolff or Kennickel, who are using the same microdata with slightly different definitions of wealth). The old estate tax data does not appear to be that great, though it was all Piketty had to work with at the time the book was published. When the cross-sectional data comes on line late into the 20th century, we see that pretty clearly because the estate tax data way understates wealth concentration at the top relative to the direct survey data.

Since Piketty published the book, we have a new wealth series from Gabriel Zucman and Emmanuel Saez that is not mentioned in the FT article. This is the best series so far. It takes US tax records from 1913 (so a single administrative data source, not mixed data sources) and then backs out wealth concentrations by applying observed rates of return to the capital income reported in the records.

This is what those concentrations look like for the top 10%, top 1%, and top 0.1%:

What you are looking for here is the U shape. Wealth concentration falls during the Great Depression and the World Wars and is crawling back.

The Saez-Zucman data is also entirely consistent with the direct cross-sectional survey data from the Federal Reserve at the top 10% level for the short period that data has been collected:

Comparing the SCF 1% and 0.1% data to the Saez-Zucman 1% and 0.1% data is not particularly helpful because 1) SCF data does not survey enough households to make the 1% of its survey figures super precise, and 2) SCF almost certainly does not get very much participation from the actual 1% as they are not generally game for answering such surveys.

So the point is: the basic wealth story of Piketty is entirely backed up by more recent (and much better) data that is from one tax record source, as well as direct survey data.

Comments