Middle Class Earnings Are Stagnant! (Because Retirees Have No Earnings)
The latest National Review features my full critique of current It-Book Capital in the Twenty-First Century, by Celebrity Economist Thomas Piketty. The review combines the discussion from my two earlier columns of his income concentration estimates with an assessment of his argument that rising inequality is a threat to social stability. I view the book as a landmark work of data collection and presentation, but one marred by its sketchy theorizing about the fate of capitalism.
That's not to say that the data he presents are without problems. As important as they are, as I argued in my inaugural Forbes column, the income concentration estimates compiled by Piketty with his colleague Emmanuel Saez cannot be used to assess how the middle class has done over time. Incomes have risen strongly after taxes and federal benefits are taken into account and other improvements to the Piketty numbers are factored in. But I also showed that middle-class incomes have risen robustly among the working-age population before taking higher federal benefits and lower taxes into account.
Astonishingly, the latter result has been systematically ignored by a number of writers and researchers responding to my column. Most prominently, Jared Bernstein (in a response posted at the New York Times Upshot website) and Piketty himself (in an interview with The New Republic) claimed that middle class incomes have risen only because of government benefits and lower taxes. The implication is that the American labor market is fundamentally broken as a vehicle for prosperity.
This is a claim that, I showed in the column, many people (including Saez) had made earlier using Piketty and Saez's data. But now, rather than addressing the evidence I presented (or rather, presented again), Bernstein, Piketty, and others are simply ignoring it in their responses to my critique. By citing estimates that do not separate out the working and non-working populations, they obscure the strong gains in earnings that have accrued to working-age households.
You don't have to look far to see this—you can scroll a few tabs over in the spreadsheet Bernstein uses. Bernstein turns not to Piketty's data but to related estimates—I would say better ones for recent decades—from the Congressional Budget Office. CBO statistically matches people in the Department of Commerce's Current Population Survey to tax returns in the IRS data used by Piketty, thereby attempting to leverage the strengths of both data sources. Bernstein notes that post-tax and -transfer household income for the middle fifth of Americans rose 36 percent from 1979 to 2010 before reassuring his audience that 90 percent of that increase came from transfers and lower taxes. Household wage and salary income declined by seven percent.
But the figures he is citing, like those of Piketty, combine retirees and members of working-age households. Retirees have lower pre-tax and -transfer incomes than everyone else, concentrated in—you guessed it—retirement benefits. More importantly, with the aging of the baby boomers, retirees have been a growing fraction of the middle class. Since Bernstein includes a rising share of non-working Social-Security-and-Medicare-receiving people in his trend analyses, it should come as no surprise to find that earnings growth was less important than the growth of transfers over the period.
But don't believe me, believe CBO. In the spreadsheet table used by Bernstein, people are ranked by the pre-tax, post-transfer income of their household (adjusted for household size). The middle fifth of people are then identified, and CBO's Table 6 displays their average household income for different types of income. The easiest way to see that Bernstein's results are being driven by retirees is to click over to Tables 14, 15, and 16, which show the same results for people in childless nonelderly households, in elderly (age 65+) households, and in households with children (respectively).
Here are the figures Bernstein presents (in the first row), along with the same ones broken down by whether or not the head is elderly, and among the nonelderly whether or not children are present:
Wages and salaries only detract from middle-class income growth (that is, they only decline) for elderly households. Rather than constituting 91 percent of income growth, taxes and transfers only account for 54 percent of income growth among nonelderly households. The growth of elderly households is the entire reason that wages and salaries detracted from income growth and that taxes and transfers accounted for nearly all of the growth. People in such households grew from 15 percent of all people in the middle fifth in 1979 to 26 percent of them in 2010. It is not just that the retiree population has grown—thanks to Social Security and Medicare, the number of retirees in the middle fifth specifically has grown much faster than their rate of growth in the general population.
Wages and salaries are the single biggest factor explaining income growth for those middle class families living in households with children. Wages and salaries are nearly as important for childless nonelderly households. The fact that when the two nonelderly groups are combined wage and salary growth looks less important than it does in either group individually is an instance of Simpson's Paradox. The paradox is explained by the fact that among the nonelderly middle class, people in childless households became a much bigger group relative to those in households with children.
It is not entirely clear how to interpret the paradox. One possibility is that households with children that are also in the middle class are a more advantaged group today than in the past, in which case looking at their changes overstates the importance of earnings growth to overall income growth. The greater dependence on wages and salaries and lower dependence on transfers and taxes would reflect the fact that only higher-earning parents make it to the middle class today, which now includes more retirees and other childless households.
However, among the nonelderly, childless households have lower incomes than households with children, in part because they have fewer earners. Therefore, pooling nonelderly households may understate the importance of earnings growth because the group with fewer earners has grown so much bigger over time. The ideal way to deal with this is to adjust incomes for the number of household members before computing averages and looking at changes in those averages. However, while CBO determines who is in the middle fifth of households on the basis of size-adjusted household income, it then reports income averages that are not adjusted for household size.
A second issue is that older “nonelderly” households may be headed by a fully or semi-retired head. Workers become eligible for partial Social Security benefits at age 62, and private retirement savings can generally be tapped without tax penalties before age 60. The reliance on Social Security among retirees may be behind the greater importance of transfers among childless nonelderly households than among households with children. In addition, CBO puts households that have children but that also have an elderly head in this “nonelderly childless” group, further expanding the presence of retirees in the group.
I will return to these issues in my next column, but before concluding this one, let's consider what the increases in wage and salary income look like, along with gains in a broader measure of labor income that includes worker contributions to 401k-type retirement plans, employer provided health insurance, and the employer's share of payroll taxes (which is viewed by CBO and most economists as income received by workers that is then taxed away to pay for social insurance benefits). I also show the change in pre-tax and -transfer income, which includes other private sources such as business, capital, and retirement income.
Middle class households with children had earnings $7,000 to $13,000 higher in 2010 than in 1979 (after accounting for the rise in the cost of living), a gain of 14 to 23 percent. If one adds other forms of pre-tax and -transfer income, the increase was over $15,000, or 25 percent. Childless nonelderly households also saw significant gains in pre-tax and -transfer income ($8,000, or 20 percent). The gains for nonelderly households as a whole are smaller than for either group individually—a clear sign that grouping childless households and those with children understates improvement if incomes are not adjusted for household size. Only elderly households in the middle class saw declines in earnings and pre-tax and -transfer income. Don't cry for them, though—their after-tax and -transfer income (not shown in the table) rose by 45 percent.
Here is the same table but comparing 1979 to 2007, like 1979 a business cycle peak. That avoids misleading the reader to think that the 1979 to 2010 changes reflect long-term patterns rather than being affected by the Great Recession.
Yes, that's a $21,000 increase in household income among middle-class families with children before taking transfers or taxes into account. It's a $13,000 increase for childless nonelderly households. Seniors experienced declines. Good thing Social Security and Medicare were enough to raise their after-tax and -transfer incomes by 48 percent (not shown).
Among both nonelderly groups, household income grew by about one-third—without any consideration of federal benefits or taxes. In fact, labor income also grew by about one-third among the nonelderly groups individually. Even if you are inclined to prefer the nonelderly figures when the two groups are pooled, we are talking about a $10,000 increase in earnings broadly defined. But in my next column, I will show using another data source that adjusting for household size also indicates an earnings increase of one-third among working-age households.
This piece originally appeared in Forbes
This piece originally appeared in Forbes