Thomas Piketty and Emmanuel Saez Respond to Alan Reynolds

Here's the response of Thomas Piketty and Emmanuel Saez to Alan Reynold's op-ed in the WSJ that has been under discussion here. As they note, Reynold's critiques "do not invalidate our findings and contain serious misunderstandings on our academic work":

Response by Thomas Piketty and Emmanuel Saez to: The Top 1% . . . of What? by ALAN REYNOLDS: In his December 14 article, “The Top 1% … of What?”, Alan Reynolds casts doubts on the interpretation of our results showing that the share of income going to the top 1% families has doubled from 8% in 1980 to 16% in 2004. In this response, we want to outline why his critiques do not invalidate our findings and contain serious misunderstandings on our academic work.

First and most important, Alan Reynolds points out that, in contrast to our results, the official Census Bureau figures show only a modest increase in the top 5% income share. The reason for the discrepancy is that the Census Bureau estimates are based on survey data which are not suitable to study high incomes because of small sample size and top coding of very high incomes. In contrast, tax return data provide a very accurate picture of reported incomes at the top. Our key contribution was precisely to use those tax data to construct better inequality estimates. We found that only families within the top 1% experienced very large gains relative to the average since 1980 and that upper middle class families (the next 4% below the top 1%) experienced only modest gains (similar to the modest increases found in the Census Bureau figures for the top 5%). This shows that the Census Bureau figures, based on data which cannot measure top 1% incomes, misses the extraordinary gains going to the top 1%, which is perhaps the most striking change in the US income distribution in recent decades.

Second, Alan Reynolds asserts that our estimates are upward biased because our total income measure is smaller than personal income from National Income and Products Accounts. Our measure of income is cash market income defined as gross income reported on tax returns less government transfers such as Social Security or Unemployment Insurance. Personal income is a broader measure of income which also includes non-cash market income such as fringe benefits from employers, imputed rent for homeowners, under-reported income (due to tax evasion) but also government transfers such as Medicare, Social Security. Conceptually, it makes more sense to focus either on market income (before deducting taxes and including transfers) or on disposable income (market income net of taxes and including transfers). We chose to estimate inequality based on (cash) market income but it would certainly be interesting to estimate inequality based on disposable income as well to assess the effects of government taxes and transfers on inequality. The official concept of personal income is not appropriate for either computation because it mixes market income with transfers but does not subtract taxes. Alan Reynolds points out that transfers have increased since 1980 but taxes on high incomes have decreased substantially. Actually, we have estimated that the average Federal tax burden on top 1% families has decreased from 44.4% in 1980 to 30.4% in 2004. The decrease in taxes at the top outweighs the increase in transfers at the bottom. Therefore, the top 1% disposable income share has most likely more than doubled since 1980.

Third, Alan Reynolds points out that reported incomes may not reflect true incomes because of tax evasion or tax avoidance. This is a legitimate concern and we, along with a number of colleagues, have actually spent substantial time investigating this issue. Alan Reynolds has picked some of the facts in order to provide a very skewed view. Most of the scenarios described by Alan Reynolds, such as a shift from corporate income to individual income or from qualified stock-options to non-qualified stock options, would imply that high incomes used to receive capital gains instead of ordinary income. For example, a closely held C-corporation which does not distribute its profits increases in value and those accumulated profits would appear as realized capital gains on the owner individual tax return when the business is sold. Yet, our top 1% income share series including realized capital gains has also doubled from 10.0% in 1980 to 19.8% in 2004.

In contrast to what Alan Reynolds suggests, there is a debate among economists on whether reported incomes respond to tax rates. The emerging consensus is that there can be substantial responses in the short-run due to retiming of income such as realizing capital gains before a tax rate increase, but that the long-term response is small. For example, as shown by Austan Goolsbee at the University of Chicago, the Clinton 1993 top rate increase from 31% to 39.6% did induce executives to exercise their stock-options in 1992 instead of 1993 but executive pay resumed its dramatic surge after 1994. Indeed, our series focusing exclusively on W2 wage and salary income (therefore excluding both business income and capital income), show that the top 1% wage income share has increased from 6.4% in 1980 to 11.6% in 2004, no doubt a very large increase as well. There is a very large literature on executive compensation and one point of agreement is that executive pay has surged relative to average pay over the last two decades, whether one counts stock-options when exercised or when granted, or even if one excludes stock-options entirely.

Even the small point on 401(k)s is conceptually mistaken: pension income is reported on tax returns when withdrawn during retirement and hence returns on pension funds are implicitly included in our income measure. Furthermore, before 401(k)s where introduced in the 1980s, workers had traditional Defined Benefits pensions which also generated capital income which were not reported on tax returns before retirement.

In sum, our work has shown the top 1% income share has increased dramatically in recent decades and has reached levels which had not been seen since before World War II and even since before the Great Depression when including capital gains. The reduction in taxes at the top since 2001 has mechanically exacerbated the discrepancy in disposable income between the rich and the rest of us. Thus, it is obvious that the progressive income tax should be the central element of the debate when thinking about what to do about the increase in inequality. Even conservatives like Alan Reynolds would agree and that is why they prefer to dismiss the facts about growing income inequality rather than face the debate on income tax progressivity at a time of growing economic disparity.

Thomas Piketty is Professor of Economics at the Paris School of Economics Emmanuel Saez is Professor of Economics at the University of California at Berkeley December 20, 2006.

Here is Alan Reynold's "stunned silence" from a comment to another post. This isn't much of  rebuttal, but he acknowledges that with a promise of more to come:

...Please take a look at the reply that Emmanuel Saez posted on his site. I will respond to it in due course. Piketty and Saez do not seem to address the data break after 1986, but they do suggest that personal income is too broad a measure. Simply adding transfer payments alone would cut the top 1 percent's share by 3 percentage points, but they say that requires also subtracting taxes. The CBO and Census both do that, and that is some of the data I referred to but did not show.

The short op ed could not and did not explain why I suggested there is no clear evidence of a signficant and sustained increase in any reasonably broad measure of inequality since 1988 (or in some cases since 1985-86).

The top 1%, after all, leaves distribution within the other 99% a total mystery, even if one really believes that gains at the top are somehow subtracted from someone else (if Smith gets a raise, must Jones get a pay cut?). I'm talking about Gini coefficients for disposable income and consumption, 90/10 ratios for salaries of full-time employees, comparative growth of SCF real median income by decile and quintile -- that sort of thing. If this seems baffling, I know of a textbook I can recommend. However, a small sampler of such evidence will be online at cato.org next week, and comments are always welcome.

Piketty and Saez assert that because of sampling error and top-coding Census is missing gigantic amounts of income in the top 1% -- that is, above $265,000. But no plausible guess at the difficulties of internal censorship (which Burkhauser taught us all) begins to account for the gap between Census and Piketty-Saez estimates for the top 5%. Even if we left out all income above $5 million in Piketty and Saez -- the top one-hudreth of one percent -- that would only narrow the gap by less than one percentage point.

There is a lot of explaining left to do, so stay tuned.

Yes Alan, there is. Lots. Starting with your promise that:

If anyone can demonstrate I misquoted such scholars, or am mistaken about any facts in last month's Wall Street Journal piece, I will gladly correct the record just as readily as I have now corrected the record about my 1992 piece.

We'll be waiting for the correction.

Update: See Brad DeLong's follow-up here and here.