The trend toward inequality has slowed, but so has income growth

Every couple of years the Congressional Budget Office publishes estimates of the trend in federal tax burdens and the distribution of U.S. incomes. CBO’s newest estimates, published last week, show that income inequality continues to rise, but at a sharply slower pace compared with the three decades ending in 2007. Unfortunately, the upward trend in real income has also slowed sharply.

Before the Great Recession began in 2008, median household income climbed about 1.4 percent a year when income is measured on an after-tax basis. In the seven years between 2007 and 2014—which include the first five years of the current economic expansion—median real income edged up just 0.2 percent a year. Median pretax income did not improve at all. In fact, it remained lower than it was just before the Great Recession. The new CBO estimates suggest we are seeing a slowdown in the trend toward more inequality. But we are also experiencing very slow income improvement. Contrary to a popular impression, the painfully slow gains can be seen up and down the income distribution.

As gloomy as these statistics are, many readers may be surprised that they are not worse. Some of our best-known statistics about income growth suggest middle-class income gains have been slower than the gains just reported by CBO. The average growth rate of after-tax income before the Great Recession means that the median U.S. household saw its annual income climb 50 percent between 1979 and 2007 (see Chart 1). This is a much bigger gain than the one reported by the Census Bureau and usually cited by reporters. The Census numbers show the median household enjoyed an income gain of just 15 percent between 1979 and 2007 (Chart 1). This estimate reflects the trend in before-tax, rather than after-tax, income, and it covers a narrower range of income items received by most families. In addition, it does not account for the shrinking average size of a typical American household. Finally, the Census estimate is based on income amounts reported by survey respondents, whereas the CBO estimate reflects income amounts reported on tax returns as well as in Census surveys.

In sum, the quality of the data available to the CBO is much better than that of the survey reports used by the Census Bureau. The measure of income used by the CBO is also more comprehensive. Two income items are reasonably well reported on Census surveys, wage income and Social Security benefits. Other income items, including self-employment earnings, pensions, interest and dividend payments, and most kinds of means-tested benefits, are subject to serious misreporting, and the reporting problems have grown worse over time. Some types of income, such as the monetary value of Medicare and Medicaid insurance policies and employers’ health plans, are unknown to most survey respondents. While the CBO makes reasonable estimates of the value of the subsidies provided by these health plans, no such estimates are included the Census Bureau’s most widely cited measure of household income.

Changes in the federal tax system, shifts in household income, and growing misreporting on household surveys have adversely affected the quality of Census income reports. Last week’s CBO report shows how these issues can affect standard measures of income growth and inequality. The federal tax system has become more progressive over the full period covered by the report (1979 through 2014). Though federal tax burdens have declined up and down the income distribution, they have fallen much faster in the case of households in the bottom half of the income distribution. Americans in all ranks of the distribution have seen sizeable increases in the value of subsidies they receive under employer-provided and government health plans, but the gains represent a proportionately bigger fraction of the incomes received by households at the very bottom of the income distribution. When we exclude health benefits from our definition of family income, we are ignoring a growing fraction of our total income, except in the case of families at the very top of the distribution.

Both the Census survey data and the CBO estimates show that inequality rose substantially over the 1979-2014 period. CBO latest numbers show plainly that the increase was driven by surging inequality in Americans’ market incomes—wages, self-employment earnings, business income, interest and dividends, and income from rent.  A commonly used indicator of inequality is the Gini coefficient, which ranges between zero (which indicates perfect equality) and one (indicating all income is received by a single household). Using this measure, CBO estimates that inequality of market income increased from 0.47 in 1979 to 0.60 in 2014, an increase of 28 percent (Chart 2). Rising social insurance, increasing health benefits targeted on low income families, and a somewhat more progressive federal tax system reduced the growth of after-tax, after-government-benefit inequality slightly. The Gini coefficient of after-tax income increased only 26 percent.  Of course, inequality is considerably smaller when it is measured with after-tax income, inclusive of government benefits. (In 2014, the Gini coefficient using this measure was 0.44, or 27 percent less than when inequality is measured using market income.)

The good news in the latest CBO report is that inequality has increased much more slowly since 2007 than it did in the previous 28 years. In fact, when inequality is measured using after-tax incomes, inclusive of government benefits, inequality actually fell slightly between 2007 and 2014. The recent stabilization of inequality, even at a high level, would be much better news if Americans’ real incomes were climbing robustly. Unhappily, CBO’s newest numbers suggest that income gains in the recovery through 2014 were anemic.