Getting the Question Right in Debates about Changes in Income and Income Inequality

Richard Burkhauser, Jeff Larrimore, Kosali Simon,Cornell University, Joint Committee on Taxation, Indiana University
September 20, 2011

In their influential work tracing the market income of Americans from 1913-1998, Piketty and Saez (2003) ask a fundamentally different question than has generally been posed in the past literature on U.S. income inequality. Their question is: How has the share of market income controlled by the top 1, 5 or 10 percent of tax units changed over the 20th century? This is an important question but it suffers from several drawbacks. It focuses on market income alone, uses the tax unit as its sharing unit, ignores the fact that tax units contain different numbers of people, and most importantly, by excluding the role of government in the measure of resources available to people, implicitly assumes that the level and trends in the distribution of market income they report over time would have been the same in the absence of government.

About This Exchange
Burkhauser, Larrimore and Simon examine research by Piketty and Saez (2003) and its implications for the inequality debate. Read a response to this essay from Tim Smeeding and Irv Garfinkel, of University of Wisconsin and Columbia University respectively, here.

What is the Income Change Question that Policy Researchers Should Answer?

Piketty and Saez focus only on top income shares, but this is not the only income measure that is impacted by their focus on market income. In the first column of Table 1, which come from Burkhauser, Larrimore, and Simon (2010) we ask the Piketty and Saez type question: How much did the market income (pre tax, pre transfer income) of the median tax unit in the United States change across the last three business cycles and overall between 1979 and 2007?

The answer is quite startling. Real median income measured in this way shows little change over this almost 30 year period. The gains experienced in the first two business cycles are almost completely erased by the losses over the last business cycle. If we updated these numbers to 2009, and hence include the current recession, the overall change is negative. However, this measure of income growth does not recognize that tax units are a subset of households and that income sharing can occur across tax units within households. When a household is used and the in-cash government transfers going to that household is also included, the overall increase in the pre-tax, post-government in-cash transfers of the median household (column 2) is 15.2 percent, almost five times the growth found in column 1.

Related RSF Resources

When we acknowledge that households are of different sizes and adjust our measure of income in column 3 using a standard adjustment for household size, the increase of the median person’s household size-adjusted pre-tax, post-government in-cash income increased by 23.6 percent, more than seven times the growth in column 1. This definition of income is most often used in the United States poverty, income and income inequality literature.

In column 4 we use the NBER tax simulation package to subtract income and payroll taxes and add the income from the Earned Income Tax Credit to estimate the increase in disposable income (household size-adjusted post-tax, post-transfer income) of the median person. We find that the disposable income of the median person increased by 29.3 percent or nearly 10 times the growth found in column 1.

Finally, column 5 recognizes that workers are compensated by employers both in cash wages and in fringe benefits. We include the ex ante or insurance value of the most important of these fringe benefits, employer provided health insurance, as well as the insurance value of government provided Medicare and Medicaid going to these households. Doing so, we find that the overall increase of the median person’s household size-adjusted pre-tax, post-government in-cash plus the value of health insurance income increased by 36.7 percent, almost 12 times the value reported in column 1.

Piketty and Saez offer a perfectly reasonable measure of how much the owners of the land, labor and capital used in the production of goods and services report on their personal income tax statements (pre-tax, pre-transfer income) and how it has changed over time. But this is a far cry from accounting for the post-tax, post-transfer resources available to the median American and how that has changed over time.

Table 1 suggests that the market income of a tax unit is a poor indicator of the household size-adjusted resources available to people. Those using answers to questions about the levels or distribution of the market income of tax units to answer questions about the levels or distribution of the total resources available to people and how they change over time will surely get it wrong.

RICHARD BURKHAUSER is the Sarah Gibson Blanding Professor of Policy Analysis Department of Policy Analysis and Management at Cornell University. He contributed to the RSF volume Assets for the Poor and is part of RSF's U.S. 2010 project. JEFF LARRIMORE is an economist with the Joint Committee on Taxation. KOSALI SIMON is a professor at Indiana University.

Burkhauser, R. V., & Larrimore, J., & Simon, K. (2010). A second opinion on the economic health of the American middle class and why it matters in gauging the impact of government policy. Cornell Working Paper.

Piketty, T., & Saez, E. (2003). "Income inequality in the United States, 1913–1998." Quarterly Journal of Economics. 118:1, 1–39.


RSF: The Russell Sage Foundation Journal of the Social Sciences is a peer-reviewed, open-access journal of original empirical research articles by both established and emerging scholars.


The Russell Sage Foundation offers grants and positions in our Visiting Scholars program for research.


Join our mailing list for email updates.