Other titles for this post: “Assortative Mating Much?” and “Which Petard?”
Tim Noah has written a column on equivalence-adjusted inequality numbers. The column does an excellent job of explaining how the Census measures inequality, but Noah’s gloss on the numbers seems a bit confused.
The Census Bureau’s standard measure of income inequality is based on household incomes. For sticklers, that’s problematic because a household can be a sprawling family of 12, or it can be a single elderly retiree. Different groupings require varying quantities of cash to get by. That’s often why those who deny or minimize the income inequality claim that the official numbers are illegitimate. When you factor in differences in household size and circumstance, they say, income inequality is not as bad. They’re right.
I can’t speak for all sticklers, but I worry that Noah is missing the point of what the sticklers are actually worried about. Here’s what he writes on that subject:
What’s interesting about the alternative measure Yen used is that it (or something like it) happens to be favored by conservatives (and also some nonideological contrarians like Georgetown economist Steven J. Rose, author of Rebound, and Reuters blogger Gregg Easterbrook, author of Sonic Boom) who argue that all this talk about income inequality is overblown. To me, the news in the AP story isn’t “Income Inequality on Rise.” Rather, it’s “Inequality Minimizers and Deniers Hoist by Own Petard.”
Speaking only for myself, I’d say Noah has the wrong petard in mind. The point of using equivalence-adjustment is not to “deny” the rise in income inequality. Rather, it is to give a better, real-world picture of overall household consumption and income levels. If we’re comparing apples to apples, of course we’ll see that inequality has increased. But we sticklers are sticklers: we want a better, more accurate portrait of how American households are faring.
To answer this criticism [re: household incomes], the Census published, alongside its standard calculation, a separate calculation that was equivalence-adjusted. The Census’s equivalence-adjusted measure of income inequality factored in the number of people occupying a household and also took into account three key variations: children consume less than adults; a single parent’s first child costs more than a couple’s first child; and as families grow, they realize certain efficiencies of scale. By this unofficial measure, which is what the AP’s Yen chose to use, income inequality is not as great as it is by the Census’ standard measure. The Gini coefficient isn’t 0.468; it’s 0.458.
Looking only at equivalence-adjusted income, the fourth and fifth quintiles are slightly but perceptibly less well-off, partly because those quintiles are already dominated by multi-person households — and, as we’ve discussed previously, these two quintiles contain almost half of the total population. The second and middle quintiles are slightly better off, and these quintiles tend to be dominated by single-person households. The bottom quintile, also heavily dominated by single-person households, doesn’t budge.
It’s not clear that Noah has taken in contrarian Stephen Rose’s actual point. Equivalence-adjusted income is a useful tool. But if we want to understand how the world works and not just score points it helps to take into account life cycle effects as well as the earnings of persons and not households.
As Rose wrote in Rebound, “if we reported the family incomes of each person separately, the median income in 2006 would be $5,000 higher than the census’s estimate.” Why is it useful to think about persons? Because it gives us a window into actual social problems, like enduring segregation and mass incarceration, and how assortative mating exacerbates social isolation.
But that’s not the end of the story. When you calculate using equivalence-adjusted data, income inequality is shown to have grown at a much faster rate over the past three decades than it is when you calculate using the Census’ standard measure.
If I may, this strikes me as one of the least surprising findings of all time. A couple of weeks ago, I wrote a short post on D.C.’s Democratic primary and I highlighted a striking statistic:
The median income of a white man in the District is $71,823.38. For a white woman, it is $51,085.39. In stark contrast, the median incomes for black men and women are $29,822.39 and $31,691.27 respectively. I prefer looking at median earnings for individuals to median household income because doing so helps clarify the role of assortative mating and household formation.
If the median white man and woman marry to form a household, the resulting household income would be $122,908.77. In an expensive metropolitan area like the District, this couple could feel cash-strapped, particularly if the couple has school-aged children. Tuition for private education is a burdensome expense, and the prospect of better schools is powerfully attractive, as it has a big impact on disposable income.
Among African Americans, marriage rates are lower than for whites. Note that black women earn more than black men in the District, and a disproportionately large share work in the public sector, including municipal government. African American families with children, many of them female-headed households, are undoubtedly concerned with the quality of public schools, yet there are also cross-pressures. And for a married African American couple, with a household income of $61,513.66, the prospect of cutbacks in public sector compensation or wage freezes or the end of tenure pose an immediate, potent threat, as Courtland Milloy suggests in his Washington Post column.
The implication is fairly straightforward. Anne Murphy Paul wrote a helpful introduction to the implications of this form of “educationally homogamous” marriage in 2008 for the New York Times.
In a world defined by a decent-sized college wage premium, the fact that college-educated women marry college-educated men while women with no more than a high school diploma marry men with no more than a high school diploma means that inequality is going to increase really, really fast. To find this remotely surprising is, to me at least, unbelievably strange.
Another small point: one of the things that happens during a recession is that the rate of household formation slows down, as Edward Glaeser has explained. So a single-person household that in a better economic climate might have reported a modest income is now part of a multi-person household with three adults sustained by, say, one wager-earner. Quite rightly, the equivalence-adjustment reflects the reality that this household’s economic status has deteriorated. Is this an inequality problem or is an economic stagnation problem? I’d suggest it is the latter.
I don’t pretend that this revelation about equivalence-adjusted income will silence the small band of people dedicated to arguing that income inequality is a minor problem at best. But it ought to give them pause when they consider that their preferred measurement shows income inequality to be accelerating more quickly than we (or they) previously knew.
Why do I think that income inequality is a minor problem? It’s because I think that the facts that
(1) our public schools are failing children from disrupted families,
(2) that a large number of African Americans and Latinos find themselves on the margins of our society,
(3) that we incarcerate a vast number of young men from a relative handful of neighborhoods defined by concentrated poverty,
(4) that rates of violent crime and physical abuse remain scandalously high,
and (5) that we expend more resources on middle-income social service providers than on wage subsidies, demogrants, and other measures that would directly alleviate poverty
would be pretty big problems even if the top 0.1% moved to Bermuda and thus “improved” our income inequality numbers.
There are many, many policies we could and should pursue to tackle the problems I’ve outlined. Improving organizational discipline in the public sector is probably the highest priority. Were the worst-performing, least cost-effective public institutions to perform as well as the highest-performing, most cost-effective public institutions, the most vulnerable members of our society would be vastly better off at negligible cost to taxpayers. It is possible that we should raise more tax revenue to pay for better public services. (Over time, I’ve grown increasingly convinced that the inefficiencies that plague the public sector are not an inescapable fact of life, and that Baumol’s cost disease can be mitigated if not cured.) One would think that we’d want to do this in the least economically damaging way, i.e., through relatively flat consumption taxes. Note that none of these steps would curb the creation of enormous fortunes at the top of the income distribution.
So why the fixation on inequality rather than addressing the public sector failures and cultural pathologies that reproduce poverty and isolation in our society? I could advance all kinds of psychological theories. But that would make me no better than the people who’ve been using Todd Henderson — a person I’ve never met — as their personal rhetorical piñata.