The Big Lebowski Defense
This article originally appeared in Forbes.
The past two weeks have seen a conversation between liberals and conservatives around the decline in marriage and its relationship to economics and cultural change. My last column argued that while marriage has declined, men’s earnings have deteriorated little if at all, making it difficult to link the two. One reaction was a sort of theoretical contortionism where declines in male earnings reduce marriage while, asymmetrically, improvements in earnings fail to raise marriage rates. Appeals to “volatility” were marshalled, despite the fact that economic volatility hasn’t risen by much. In the chart below, based on data from two surveys I analyzed for my National Affairs essay, Bogeyman Economics, the share of working-age adults experiencing a 25 percent drop in income rises over time from 7 percent between 1968 and 1969 to 9 percent between 2006 and 2007.
The other interesting response was a kind of Big Lebowski Defense regarding which index to use for inflation adjustment. It turns out that price measurement is imperfect! And the theory of consumer utility lacks the neatness of natural laws! Who knew?? And who cares if the Bureau of Labor Statistics, Census Bureau, and Congressional Budget Office all abandoned this one index 20 to 25 years ago, or that the unambiguous improvements that the Bureau of Labor Statistics has made to it only go back to 1999, and only go back to 1978 using the “research series” version of it? Who cares that the latest version created by BLS more closely tracks my preferred index? Who cares that experts in price measurement know that we’re not adequately accounting for all the free stuff we get from the internet today and how it makes us better off?
“Yeah, well, that’s just, like, your opinion, man.”
The bottom line is that at comparable points in the business cycle, there hasn’t been nearly as much change in male earnings on the low end of the earnings distribution as people seem to think. That’s without taking into account the rising share of compensation accounted for by health insurance or the expansion of the Earned Income Tax Credit. People can argue that income and price measurement are complicated—they surely are—but until someone gives me specific reasons why in the face of these complications their choices are better than mine, rather than simply throwing up hands and saying who knows?, I’m going to declare victory here.
This same sort of you-say-potato logic is the basis for Tom Edsall’s New York Times column today. Edsall discusses the conclusion of my mentor and dissertation advisor, Christopher Jencks, that the poverty rate fell from 19 percent in 1964 to 5 percent in 2013. Jencks makes no claim that only 5 percent of the population is actually in economic distress, his argument is clearly and obviously about the magnitude of the decline in poverty. More than 5 percent are in some kind of distress today, but more than 19 percent were in 1964 too.
Edsall emphasizes that Jencks has “unassailable liberal credentials,” which means that the reader and he are safe to treat this as non-crazy-conservative-talk. (Actually, Jencks’s liberal credentials have been assailed many, many times, and at any rate, after knowing him for twenty years, I’m still not sure he thinks of himself as liberal. That’s a good thing, by the way.) Edsall gets feedback from researchers representing the whole ideological spectrum, from A to C, and concludes of Jencks that “his conclusion…understates the scope of hardship in this country.”
Edsall and the liberal researchers he interviewed (many of whom I like a lot, for the record) all want us to focus on relative poverty rather than absolute poverty—to view people as no less poor even if everyone’s income doubles over time. They are welcome to prefer looking at that kind of a measure, which blurs the distinction between material hardship and income inequality. But no one provides any reason to privilege it over looking at material hardship. Look closely and you’ll find that no one even says anything about whether relative poverty has increased. The reliable Jacob Hacker notes that our relative poverty rates are on par with Chile’s and Turkey’s, which obviously indicates we should be indifferent between living in the U.S. or those countries.
But the real highlight is when Edsall goes into “that’s just your opinion” mode, getting into, once again, the issue of inflation adjustment. He quotes attorney Sean Fremstad as saying that using Jencks’s (and my) preferred inflation adjustment, the “PCE deflator,” “an elderly couple was not poor in 2013 if they had an income, including housing and food benefits, equal to $11,300.” Under the CPI-U, Fremstad says that they were poor so long as they had an income under $14,095.
What Fremstad seems to be saying is that the poverty line for an elderly couple in 2013 would have been $11,300 if the line had been updated using the PCE starting in 1970. Why 1970? It’s the year in which the math works out to $11,300, but it’s an arbitrary year. If the poverty line had been updated using the PCE starting in 1980, it would be $12,200 today.
Let’s back up, in 1970, the poverty line for Fremtad’s two-person elderly family was $2,349. Question: if the poverty line had risen at the true cost of living, what would it have been in 2013? We don’t know that because we don’t know how much the true cost of living increased. If we believe the CPI-U is a close approximation, then the answer is that the poverty line would have been $14,095 in 2013. If we believe the PCE is a close approximation, then the answer is that the poverty line would have been $11,300 in 2013.
Whether a two-person elderly family needs $11,300 or $14,095 to get by in 2013 isn’t really the issue. If the PCE tracked the true change in the cost of living, we might still think that $11,300 is too low for a poverty line today, but that would mean that $2,349 was too low in 1970. If the CPI-U tracked the true change in the cost of living, we might think $14,095 is close to adequate for a poverty line, which would mean that $2,349 was close to adequate in 1970. But the true change in absolute poverty is (1) the percent of people in 2013 below a $2,348 poverty line after raising the line to account for the *true* change in the cost of living since 1970 minus (2) the percent of people in 1970 below a $2,348 poverty line.
Edsall and Fremstad are getting too hung up on levels of poverty rather than trends, which was (again) the real subject of Jencks’s piece. Bump up the 1970 and 2013 poverty lines by 10 percent each and the PCE would still show the same decline in poverty, and it would still be a bigger decline than the CPI-U shows. And because the change in the PCE is unambiguously closer to the “true” increase in the cost of living than the change in the CPI-U, poverty declined by more than the CPI-U shows and declined by a lot. Full stop. (Do I need to repeat again how all the federal statistical agencies ditched the CPI-U 20 to 25 years ago? I will.)
Edsall goes on to cite the work of Kathryn Edin and Luke Shafer, which finds an increase in the number of Americans living on $2 a day—and increase that did not happen because no one in America lives on $2 a day if that standard is to mean anything. I made that point the last time Edsall cited them.
The Economic Policy Institute’s Larry Mishel provides wage growth figures for the bottom fifth (and those don’t even use the PCE deflator). But two paragraphs later Edsall is citing “Mishel’s case that falling private sector wages are a major factor” for the poor (emphasis mine). Wait, did Mishel say wage growth? He did! Off message!
Finally, Edsall shows a chart indicating how much higher poverty would be if not for the safety net. You can’t tell from the chart, but the entire post-tax & -transfer poverty decline he shows between 1967 and 1990 is due to a drop in poverty among the elderly; safety net programs didn’t reduce poverty among children until the 1990s, when we started the welfare reforms and work supports that make Edsall nervous.
I am tired of writing about this stuff.