American income inequality is commonly told as a story of divergence: since the 1970s, the share of income going to the top of income distribution has skyrocketed, while the share of income going to the bottom of the income distribution has seen large declines. Even in absolute terms, there is evidence that incomes for low-wage workers have eroded and incomes for workers at the middle of the income distribution have only modestly increased.
But is this the full story?
According to a thought-provoking essay by Cornell economist Richard Burkhauser, the well-documented trends in income inequality often focus on the pre-tax and pre-transfer market wages of income earners, but that does not answer the question of how much financial resources are available to families at different points in the income distribution. To get at the latter question, he and his coauthors have revisited time series data from 1979 to 2007 from the Current Population Survey (CPS) adding in transfer income, accounting for taxes, and adjusting for household size.
Methodological note: prior analyses have attempted to account for these factors using the CPS but have not adequately addressed the problem of top-coding (the data for very high income earners are capped in order to ensure confidentiality). The other leading account of income inequality, that of Picketty and Saez, addresses the top-coding problem by using tax return data, but have not been able to make all of the adjustments for transfers and household size. As Burkhauser notes, Picketty and Saez were concerned with the question solely of market income shares for the wealthiest individuals, not the set of financial resources available to families throughout the income distribution.
Below are two tables from Burkhauser’s paper in JPAM. Table 1 displays the median income displays the median income growth during different business cycles under different adjustments and Table 2 shows the income growth for different quintiles making the same adjustments. Both charts can be read the same way: the first panel shows the trend without taking into account taxes, transfers, and household status, the second panel adjusts for household, the third for household and transfers, the third adjust for household size, the fourth for taxes and transfers and household size, and the sixth for the insurance value of health care coverage.
As it turns out, making these adjustments paints a dramatically different picture. As Table 1 illustrates, real income growth between 1979 and 2007 at the median is very modest without taking into account taxes and transfers, but is more than ten times larger after including various taxes and benefits (3.2% versus 36.7%). Looking across the income distribution, the picture is even more dramatic – the bottom quintile without adjustment lost 33.0% of its real income between 1979 and 2007, but the picture is entirely reversed (even before taking into account health insurance). Notably, income growth is also quite a bit higher for the top of the income distribution after making these adjustments (partially a reflection of the declining tax rate on high income earners since the 1980s).
As Burkhauser states in summary:
“It is the case that the market income of low-income tax units has declined substantially since 1979, but it is not the case that the post-tax, post-transfer house- hold size–adjusted resources available to the bottom quintile of the population have declined. They are up almost 15 percent over that period and 26 percent when the insurance value of health insurance is included. This value would be even higher if other in-kind transfers targeted on low-income people were included in our measure (e.g., food stamps, WIC).”
What should we make of these results?
The first point is methodological. Burkhauser rightly points out that the question of resources available to families can only be answered after making these kinds of adjustments. Poverty researchers should welcome analyses that use corrected income series and make the adjustments for taxes and transfers – and this is precisely what analysts of the American welfare state have been calling for in recent years (as I touched upon in an earlier discussion). Some of the more dramatic results that Burkhauser obtains depend upon certain assumptions about how to count the value of health insurance, and as I have also said, there’s not a simple method of accounting for the value of health insurance to households. If the question is how much does having health insurance change the quality of life of families, then it probably is an overstatement to look at changes in insurance value (we need a better measure of insurance value that accounts for dramatic price inflation over time). Still, the results hold up without any inclusion of insurance.
The second point is political. There are many critics of the American welfare state on the right that have been trying to make the basic point that “poor” Americans are a lot richer than they claim, and they point to evidence that shows that many of these households can actually purchase cars, televisions, and computers. A less than responsible policy entrepreneur might seize upon these results as further evidence that things are really not as bad for low-income families as is often portrayed by sympathetic poverty researchers. It is therefore important that poverty researchers not dogmatically retreat behind the position that the standard of living has only improved for wealthy Americans, but rather to more carefully assess how much additional poverty is now prevented by transfers (such as cash welfare), and also to be realistic about how much could be further prevented through more protective and redistributive policies.