Whither The Bottom 90 Percent, Thomas Piketty?
While not quite inducing Beatlemania, French economist Thomas Piketty's visit this week to America has inspired the Washington analog of teenage frenzy. On Tuesday, the inequality expert spoke here in D.C. (top one percent share: 28%) at multiple events. Wednesday he was in New York (Manhattan top share: 54%), and he will head from there to Boston (36%) and San Francisco (37%). Among those of us interested in inequality, the pump for this visit was primed months ago. Word had trickled (gushed, really) across the pond about Piketty's book, Capital in Twenty-First Century, released in France last August but the translated version arriving in the U.S. only last month. To a certain crowd that is convinced that inequality is a dire economic problem, it's kind of like “I Want to Hold Your Hand.”
Piketty's book lays his cards on the table from the start. He titles it to evoke Marx and begins with an epigraph quoting the Declaration of the Rights of Man and the Citizen to the effect that all inequality should be viewed as suspect. He poses the question in which he is interested as whether capitalism is fundamentally self-correcting in a way that prevents inequality from getting out of control or whether it will produce ever-rising inequality. While he allows that his answer is “imperfect and incomplete,” his modesty goes out the door before that paragraph ends. Piketty's thesis, in his own words:
““When the rate of return on capital exceeds the rate of growth of output and income, as it did in the nineteenth century and seems quite likely to do again in the twenty-first, capitalism automatically generates arbitrary and unsustainable inequalities that radically undermine the meritocratic values on which democratic societies are based.”
In other words, Piketty is suggesting that we may have entered a period in which concentrated wealth will produce a sort of inequality death spiral. With economic growth sluggish, and the returns to wealth high, owners of “capital”—land, housing, buildings, businesses, and other income-producing property—will receive a rising share of income as they re-invest their returns. Eventually, this dynamic, combined with inheritance, will produce a leisure class of capitalists who do not have to work to grow ever richer, while the bottom half is left behind in intolerable conditions. An earlier quote Piketty gave Daniel Altman of the New York Times on the effect of the Bush tax cuts is particularly revealing. The cuts, he predicted would
““eventually contribute to rebuild a class of rentiers in the U.S., whereby a small group of wealthy but untalented children controls vast segments of the U.S. economy and penniless, talented children simply can't compete….there is a decent probability that the U.S. will look like Old Europe prior to 1914 in a couple of generations.”
There is a lot to unpack here. Broadly speaking, Piketty's argument has four parts: (1) his account of trends in inequality and growth, (2) his explanation for these trends, (3) his interpretation of the implications these trends have for the future, and (4) why we should care about these past and future trends. I have thoughts on all four of those components that I'll share in the coming weeks, but let me focus on his income trend data here and in my next post. While I slowly collect my thoughts on the other parts of Piketty's argument, you should definitely check out American Enterprise Institute scholar Kevin Hassett's brilliant critique of Piketty from one of Tuesday's events—particularly relevant to Piketty's interpretation of what the future holds.
While Piketty's efforts to improve our understanding of income concentration have been invaluable, the tax-return-based estimates that he and others have compiled are not without problems for certain applications. At Tuesday's event at the Tax Policy Center, Piketty rightly discounted evidence presented by Hassett that the consumption share of the top fifth of Americans has not risen. Most household surveys are unlikely to adequately capture trends in income and consumption at the top. They do not capture enough people who are at the very top of the top, who drive the trends in concentration. A sample as large as 120,000 people (such as in the survey Hassett used) would be expected to have no more than twelve households from the top one percent of the top one percent, and year-to-year variations in that small sub-sample can make a big difference in trends. To the extent that surveys do include some of the richest households, sponsors of the surveys blur the information collected on them out of privacy concerns, including capping the amount of their income or consumption reported in the data.
At the same time, it is no less true that tax return data cannot be used to assess trends in income below the top—at least in the U.S., and I suspect elsewhere. The Piketty and Saez data for the U.S. indicate that between 1979 and 2012, the bottom 90 percent's income dropped by over $3,000. However, the official Census Bureau estimates indicate that the bottom 80 percent of households saw an increase of nearly $3,500. Median income—the income of the household in the middle of the distribution—rose by $2,500. If you are underwhelmed by these initial differences, stick around.
The Census Bureau figures are superior to the Piketty and Saez estimates when looking below the top ten percent in two ways. First, the measure of income derived from tax returns excludes a significant amount of income, and people below the top are disproportionately recipients of that income. Most importantly, in the United States, most public transfer income is omitted from tax returns. That includes not just means-tested programs for poor families and unemployment benefits, but Social Security. Many retirees in the Piketty-Saez data have tiny incomes because their main source of sustenance is rendered invisible in the data. The Census Bureau figures include some transfers, though even they omit non-cash transfers like food stamps, school lunches, public housing, Medicare, and Medicaid.
You might think that that means the Piketty-Saez data still does a good job capturing “market income”—what people make before the government steps in to redistribute. But their data also excludes non-taxable capital gains (such as those accruing to middle-class households when they sell a home), employer benefits (like health insurance), and other sources of non-taxable income. More subtly, it is impossible to get an accurate read on trends in market income concentration when retirees (with little to no market income) are included in the data (as they always are). The share of retirees has been growing for some time, and that puts downward pressure on the market income trend.
One can use the Census Bureau data to estimate trends in market income for households with a head under age sixty (and so unlikely to be retired). Among those with any market income, I find an increase of $3,400 in the median (using the same cost-of-living adjustment as the Census Bureau and Piketty and Saez). This estimate does not include the value of employer-provided health insurance or other fringe benefits and does not include capital gains either.
The second reason that tax return data are inferior to Census Bureau estimates for incomes below the top is that tax returns—or “tax units,” which essentially means potential tax returns if everyone filed—are different from households. The Piketty and Saez data include as tax units all returns filed by dependent teenagers with summer jobs and undergraduates with work-study positions. They count roommates and unmarried partners as separate tax units rather than as one household, ignoring all of the shared living expenses that make living with someone cheaper than living alone. As a consequence, incomes are much lower among tax units than among households.
It's also worth reiterating that there are ways of improving on income measurement that none of the above figures incorporate. Income trend estimates should account for declines in household size—fewer mouths to feed for a given income—and they should use a better cost-of-living adjustment. When I raised these issues with Saez recently on a conference panel in which we both participated, he agreed that in principle, incomes should be size-adjusted, though he favored adjusting them according to the number of adults rather than the convention of adjusting by the number of adults and children. He also agreed that the inflation measure favored by the Congressional Budget Office and targeted by the Federal Reserve Board (the “Personal Consumption Expenditures deflator”) is more appropriate than the adjustment used by the Census Bureau and by himself and Piketty.
Another improvement to the above income estimates would incorporate non-cash public benefits and employer-provided health insurance. Finally, particularly if we are concerned about inequality, income measures should account for the redistribution that occurs through progressive taxation (as Piketty and Saez have done in less-cited work).
When I incorporate these improvements using the Census Bureau data, I find that median post-tax and -transfer income rose by nearly $26,000 for a household of four ($13,000 for a household of one) between 1979 and 2012. If you don't like the household-size adjustment, the non-adjusted increase was over $20,000 at the median. If you think that valuing health care as income is problematic, that figure drops to $10,400 under the implausible assumption that third-party health care benefits have no value to households. The income of the bottom 90 percent rose nearly $12,000 under that assumption instead of dropping by $3,000 as in the Piketty and Saez data, and it rose by nearly $21,000 if health benefits are included. For a household of four, median market income for non-elderly households (not counting employer-provided health care as income) rose $9,400.
The distinction between income as measured by Piketty and Saez and more comprehensive definitions of household income measured more accurately matters for how we should think about rising income concentration. The Piketty and Saez numbers for the bottom 90 percent are routinely cited by other researchers and journalists. In response to a remark on Twitter by Timothy Noah, author of The Great Divergence (the last big inequality book), that, “Nobody thinks market income has failed to grow since '79,” I easily turned up a number of mentions that essentially make that very claim.
The Center on Budget and Policy Priorities has cited the Piketty-Saez data to conclude that “since the late 1970s, the incomes of the bottom 90 percent of households have essentially stagnated.” Economist John Schmitt of the Center for Economic and Policy Research noted that “the average income of the bottom 90 percent barely budged.” The Center for American Progress's ThinkProgress, citing journalist David Cay Johnston's computation from the Piketty-Saez data, reported that “average income rose just $59 from 1966 to 2011 for the bottom 90 percent”. The Economic Policy Institute's president, Larry Mishel, faithfully reported the supposed five percent increase between 1979 and 2007 (which comes to less than $1,700) that the Piketty-Saez data indicate preceded a bigger drop after 2007. My post-tax and -transfer estimates including health benefits show a rise between these years of $18,000.
In short, Piketty seems to draw too strong a conclusion (“terrifying,” in his words) about what continued rising inequality would entail for the bottom 90 percent (at least in the U.S.). Rising income concentration has not been accompanied by stagnation below the top, and there is no reason to think that it will be in the future. (In fact, there are reasons to think that income concentration might level off in the future and incomes lower down might rise more robustly, a point to which I will return in a future post. Those of you who heard my question at Piketty's Tax Policy Center event already can anticipate it.) The slowdown in median income growth began in the 1970s, years before income concentration began to rise. Indeed, income growth for the top one percent was no better than for the middle fifth between 1969 and 1979.
For that matter, sociologist Lane Kenworthy has found that income concentration and median incomes are only modestly associated looking at changes in both across developed countries—and that correlation disappeared when he allowed for the possibility that income concentration might affect economic growth and redistribution. The correlation across counties in the U.S. between the share of income received by the top one percent of parents and median parent's income is 0.14, where 0.0 indicates no relationship and 1.0 indicates the strongest relationship possible. The correlation between the top one percent's share and the 25th percentile of parent income (the income level three-quarters of the way down the income ladder) is 0.07. New York City's 54 percent top share has not dissuaded people from around the U.S. and the world from putting down roots there. What does Piketty know that they don't?
If rising inequality is compatible with higher incomes further down, then while income grows more concentrated at the top, everyone else could still end up richer, as they have in recent decades. That would be a very different problem—if a problem at all—than if incomes fall within the bottom 90 percent, which Piketty seems to regard as inevitable.
Next up, I'll take on the Piketty-Saez trend estimates for the top one percent. While their tax-return-based figures do a much better job conveying the levels of income concentration in the U.S. than survey estimates, they may overstate the rise in income concentration pretty badly.
This piece originally appeared in Forbes
This piece originally appeared in Forbes