January 19, 2011
Between 2008 and 2009, unemployment increased from 5.8 percent to 9.3 percent, the largest one-year increase on record (which goes back to 1948). Over the same period, the number of Americans without health insurance coverage rose by more than four million—from 46.3 million in 2008 to 50.7 million in 2009—and low-income people lost insurance at a greater rate than Americans overall. Thus, it isn’t surprising that the Census Bureau’s official poverty estimates show that the number of people who were impoverished in 2009 increased by 3.74 million, and the poverty rate increased from 13.2 percent in 2008 to 14.3 percent in 2009.
More surprising is an “alternative” poverty estimate Census quietly released earlier this month. This estimate, highlighted today in a New York Times editorial, shows no increase in poverty between 2008 and 2009. Given the record increase in unemployment and huge decline in health insurance coverage, especially among low-income people, could this alternative estimate showing no increase in poverty really be correct?
The short answer is no, but it takes some unbundling of the numbers to understand what’s going on. Certain benefits that were expanded in the Recovery Act, particularly food stamps and refundable tax credits, are not currently counted in the official poverty measure. The alternative poverty measure cited by the NYT counts these benefits. Thus, one explanation of the apparent disconnection between unemployment and poverty is that the food stamp and refundable tax credit provisions in the Recovery Act were large enough (and targeted in a way) that prevented any actual increase in poverty.
I’m certainly a fan of the Recovery Act—although as economic stimulus, it was much too small, as Eileen Appelbaum explained in an earlier post here—and I think that the Act’s expansions of food stamps and refundable tax credits kept poverty from increasing as much as it would have in their absence. But I don’t believe that these expansions kept poverty from increasing at all. As a practical matter, what families lost last year as a result of the massive increase in unemployment and uninsurance was just too large to be completely offset by the Act’s food stamp and tax credit expansions.
What the NYT doesn’t note in its editorial is that the alternative poverty threshold used by Census to produce the no-increase result appears to be about $233 lower—about 1 percent lower—in 2009 than in 2008. For example, the alternative poverty threshold for a family of four was $24,755 in 2008; the same threshold in 2009 was $24,522. The official threshold also declined, but only by about .36 percent. Thus, at least part of the explanation for the no-increase result using the alternative measure is that it set the poverty “bar” lower in 2009 than it did in 2008.
Why did the value of the alternative poverty line decline at more twice the rate of the official one? The alternative poverty line is tied to the amount that moderate-income households (roughly those at the 33rd percentile) spend on housing, utilities, and food. Thus, during say a housing bubble, when housing prices are outpacing inflation, the alternative poverty line will tend to rise at a faster rate than a line adjusted only for inflation, while during a slump, like the one we’re having now, the threshold will tend to fall behind inflation. Whether or not we think this is a conceptually sound way to measure poverty—my own preference is to tie the poverty threshold to a percentage of median income, rather than a subset of consumer spending, and not allow it to decline in nominal dollars—it’s worth noting as a factor when explaining otherwise counter-intutitive results like poverty not increasing in a year when unemployment rose at record rate.
A second issue has to do with the alternative measure’s treatment of health insurance and medical costs. Unlike housing and food, medical necessities are not included in the threshold of the alternative measure cited by the NYT. Instead, the alternative measure only accounts for medical expenditures that were paid out-of-pocket in 2009.
Thus, if you were one of the 4.4 million people to have lost health insurance in 2009, you may or may not look worse off that year according to the alternative measure. If you were able to afford paying for medical care out of pocket, you’ll look worse off, but if couldn’t afford necessary care and didn’t make any additional out-of-pocket payments, you’ll look no worse off. And, if, in fact, you made fewer out-of-pocket payments precisely because you lost health insurance (for example, fewer co-pays for previously covered services), you will actually look better off, even through the value of health services you lost may far exceed those you gained. A related problem is that the people who are able to afford to pay for health care costs out-of-pocket may be more economically secure—because of substantial assets or low housing costs—than the ones who can’t afford to pay, even holding income equal.
Interestingly, Census also published some experimental poverty measures that include medical expenses in the poverty thresholds, but these have gotten little attention in the media. Unlike the alternative thresholds that exclude medical needs, the value of these thresholds actually increased between 2008 and 2009. These thresholds show (more accurately, in my view) poverty increasing by between 1.1 and 1.8 million people between 2008 and 2009.
A final issue has to do with how the alternative measure counts refundable tax credits. The Recovery Act increased both the Child Tax Credit and EITC for certain categories of low- and middle-income families for tax years 2009 and 2010. For nearly all of the families benefitting from this increase, the benefits were received as a lump sum in 2010 (after they filed their 2009 tax returns) rather than in 2009. However, the alternative measure used by Census counts these benefits as having been received in 2009. I imagine Census has practical reasons for doing this, but it provides an inaccurate picture of real poverty in 2009. In general, it makes sense to take account of taxes paid and taxes benefits received when measuring poverty, but in this case, the official poverty line’s trend between 2008 and 2009 arguably provides the more accurate picture of poverty because it doesn’t count an increase in tax credits received in 2010 as having reduced poverty in the previous year.
Finally, I would be remiss if I didn’t note two important things the NYT got right. First, they correctly note that the recent tax cut deal includes a one-year payroll tax cut that “is less helpful to the lowest-income workers than a now-expired tax break in the stimulus”, a fact I’ve noted in previous posts, but one that hasn’t been widely reported by the media. Second, they correctly argue that “reducing federal help now”—the course urged by conservatives—will ensure more poverty later. But these are points that can be made without making the incorrect assertion that there was no increase in poverty in 2009.
