Citing a new paper by Bruce Meyer and James Sullivan, a Bloomberg editorial opines that “poverty isn’t as high as the U.S. government says it is.” The reason according to Bloomberg is that the current official measure: 1) doesn’t count the EITC and certain other means-tested benefits, and 2) sets the poverty line too high by “ignoring discount prices at big-box outlets such as Wal-Mart Stores Inc., where many low-income families shop.”
The current measure is deeply flawed. It should use a more comprehensive after-tax measure of income, so Bloomberg critique number 1 is mostly right.
But critique number 2 is completely off base: the problem with the current poverty threshold—about $22,000 for a family of four in 2010—is that it is much too low to serve as a measure of a minimally decent living standard in today’s economy.
If we fixed both of these problems, the poverty rate would be higher than currently reported, not lower. And, in fact, this is confirmed by the Meyer-Sullivan paper, which finds that the poverty rate is around 17 percent if we used a modern poverty threhsold.
The current threshold is too low because it was set in the early 1960s using late-1950s data, and has only been adjusted for inflation since then, and not for the general increase in average living standards over the last half-century. A “base-1959” poverty measure made sense in the early 1960s. It’s absurd and indefensible half a century later.
If you're not sure you agree with me on this, ask yourself if you should only receive “base-1959” Social Security benefits, i.e., the 1959 benefit level adjusted for inflation. The Social Security benefits we receive when we retire are initially adjusted for the increase in the national average wage index, which ensures, as SSA explains, that “a worker's future benefits reflect the general rise in the standard of living that occurred during his or her working lifetime.” I’m guessing most Americans think this is a sound approach.
Meyer and Sullivan exacerbate the base-1959 problem with the FPL by adjusting the poverty threshold down further using the CPI-U-RS, a research version of the CPI. There are times to use the CPI-U-RS to show trends, but this is not one of them. The main effect of using the CPI-U-RS here is to lower an already too-low poverty threshold by even more—so, for example, the threshold for a family of four, is cut by about $2,500 or 11 percent. One of the results: the Meyer-Sullivan-adjusted poverty threshold falls from 55 percent of median income for a family of four in 1959 to only one-third of it in 2009. In essence, deprivation is defined even further down that it is using the current base-1959 FPL.
That said, the Meyer-Sullivan paper does include some useful information and analyses. In particular, they also calculate income poverty rates since 1960 using a threshold set at 50 percent of median income, roughly the level at which our current FPL was set at initially. This measure tells a much different story: the poverty rate ends up being about 17 percent in 2010, rather than 11.7 percent (both figures here use a comprehensive measure of income that includes the EITC, SNAP, etc). In other words, when poverty is measured in a completely reasonable fashion—using the same measure used in most other wealthy Western nations—it is higher than the official measure suggests.
Finally, Meyer-Sullivan think that poverty should be measured using consumption data rather than income data, and that doing so would, again, show that poverty rates are much lower than official data show. Consumption vs. income is a complex issue that merits a post of its own when I get the chance. I think consumption poverty measures might play a useful supplementary role some day, but they're not ready for prime time. Here it is worth noting that the Meyer-Sullivan consumption poverty estimates show a steep and implausible decline in poverty in the 2000s during the Bush Administration, one that diverges from the trend we see using income data (and other hardship data, such as food insecurity). In this area, I find more plausible this new working paper by Jonathan Fisher, David Johnson, and Tim Smeeding, which finds that that both consumption inequality and income inequality trends track each other fairly closely between 1985 and 2005.