This paper combines income tax returns with macroeconomic household balance sheets to estimate the distribution of wealth in the United States since 1913. We estimate wealth by capitalizing the incomes reported by individual taxpayers, accounting for assets that do not generate taxable income. We successfully test our capitalization method in three micro datasets where we can observe both income and wealth: the Survey of Consumer Finance, linked estate and income tax returns, and foundations’ tax records. We find that wealth concentration was high in the beginning of the twentieth century, fell from 1929 to 1978, and has continuously increased since then. The top 0.1% wealth share has risen from 7% in 1978 to 22% in 2012, a level almost as high as in 1929. Top wealth-holders are younger today than in the 1960s and earn a higher fraction of the economy’s labor income. The bottom 90% wealth share first increased up to the mid-1980s and then steadily declined. The increase in wealth inequality in recent decades is due to the upsurge of top incomes combined with an increase in saving rate inequality. We explain how our findings can be reconciled with Survey of Consumer Finances and estate tax data.

This is the abstract of “Wealth Inequality in the United States since 1913: Evidence from Capitalized Income Tax Data” by Emmanuel Saez and Gabriel Zucman, forthcoming as an article in the Quarterly Journal of Economics.

Here’s Alan Reynolds’s criticism of a PowerPoint version of this article, and an excerpt of that criticism:

Zucman-Saez concludes that there was a “large increase in the top 0.1% wealth share” since the 1986 Tax Reform, but “no increase below the top 0.1%.” In other words, all of the increase in the wealth share of the top 1% is attributed to the top one-tenth of 1%–those with estimated wealth above $20 million. This is quite different from the graph in Mr. Piketty’s book, which showed the wealth share of the top 1% (which begins at about $8 million, according to the Federal Reserve’s Survey of Consumer Finances) in the U.S. falling from 31.4% in 1960 to 28.2% in 1970, then rising to about 33% since 1990.

In any event, the Zucman-Saez data are so misleading as to be worthless. They attempt to estimate top U.S. wealth shares on the basis of that portion of capital income reported on individual income tax returns–interest, dividends, rent and capital gains.

This won’t work because federal tax laws in 1981, 1986, 1997 and 2003 momentously changed (1) the rules about which sorts of capital income have to be reported, (2) the tax incentives to report business income on individual rather than corporate tax forms, and (3) the tax incentives for high-income taxpayers to respond to lower tax rates on capital gains and dividends by realizing more capital gains and holding more dividend-paying stocks.

Reynolds then goes on to consider each of those 3 factors.

It turns out, as Reynolds pointed out in an email, that Milton Friedman criticized the Saez-Zucman approach in–are you ready?–1939.

Here’s an excerpt from Friedman’s critique:

The difficulties with this method are of two types. There are, first, the difficulties arising from the character and reliability of the data: the difficulty of accurately estimating the capitalization factor; the empirical necessity of using the same capitalization factor for all income classes; the fewness of the returns in the very high, and the absence of any returns in the very low, wealth classes and the consequent necessity of extrapolation; the decidedly different age distribution of the individuals covered by the estate data and those covered by the income tax data; the use of figures based on unaudited returns; the biased nature of the sample of individuals filing income tax returns; the absence of a wealth total thar might be employed to correct at least partly for this bias; the conceptual difficulties with the income total used to classify individuals by income classes; and so on. Second, there are the difficulties inherent in the method that could not be removed by any conceivable improvement in the data employed.

HT2 Alan Reynolds and Phil Magness.