How America made a mess of measuring poverty
Jeff Spross

Illustrated | Three Lions/Getty Images, Yevhenii Dubinko/iStock
February 13, 2020

Since 1963, the United States has been measuring what percentage of its population is in poverty. But it hasn't changed the way it measures poverty since 1963, either. Outside of annual adjustments for inflation, the Official Poverty Measure (OPM) works the same way now that it did six decades ago.

Rep. Alexandria Ocasio-Cortez (D-N.Y.) aims to fix that. She introduced her Recognizing Real Poverty Act in September of last year, part of a larger suite of measures to aid low-income Americans. According to her office, the bill requires federal agencies to "adjust the federal poverty line to account for geographic cost variation, costs related to health insurance, work expenses for the family, child care needs, and new necessities, like internet access."

Granted, this is a call for a reform, and a study of what that reform should be. It doesn't lay out precise end goals. But Ocasio-Cortez is right: The way America defines poverty right now is a conceptual trainwreck.

Let's run through the problems.

First off, the OPM focuses solely on food: Using 1955 data, it defined a basket of foodstuffs any household would need for basic living, adjusted for family size, and determined that a household was in poverty if that basket cost more than a third of its overall budget. Basically, the Johnson administration needed some way to set eligibility for its new welfare programs, so it grabbed some work being done by government statisticians and retrofit it as an overall poverty measure. Other than the inflation adjustments, that's still how we define the poverty line. And it's still what we use to decide eligibility for a whole bunch of programs, from Medicaid to food stamps, school lunch programs, Obamacare's subsidies, and other grants and forms of assistance.

But focusing on food leaves out things that existed at the time that any reasonable person would also consider necessary for basic dignified living standards (health care, child care). And, of course, it leaves out other needs that have arisen since (like internet access or mobile phone service). Yet another problem is what sources of income are and aren't counted towards the total family budget, which the food budget is then measured against.

The answer to this last question is basically a random hodgepodge: Wage income is counted, along with dividends and interest payments; but capital gains from selling assets are not. Granted, capital gains aren't exactly relevant to the average low-income American, but government aid certainly is: The OPM counts unemployment insurance, Social Security, workers' compensation, and other benefits that are straight cash aid. It does not count assistance that isn't simple cash — say, health coverage in the form of Medicaid, or benefits linked to certain needs, like food stamps or housing assistance. Nor does the OPM count government aid that is distributed via the tax code, like the Earned Income Tax Credit or the Child Tax Credit.

This creates a perverse situation: Lawmakers, citizens and journalists all cite the OPM in discussions and arguments over how good a job America does fighting poverty, or what it should be doing differently. Yet a huge swath of the programs that aim to alleviate poverty have no effect on the OPM! Indeed, outside of a dramatic fall from 1960 to 1970, the official poverty rate has bounced between 11 percent and 15 percent ever since.

Now, American policymakers are hardly ignorant of this situation. Even in 1963, they knew they weren't creating the best measure — they just needed a measure. And over the last decade the government has developed an alternative metric, the Supplemental Poverty Measure (SPM), which addresses some of the OPM's problems.

The SPM cleans up a lot of the contributions to income: It includes non-cash benefits like food stamps and housing, and it includes transfers within the tax code, like the aforementioned tax credits. It also treats as necessities other expenditures beyond food, by subtracting spending on things like child care and out-of-pocket medical expenses out of its final measurement of a family’s income. The result is the SPM arguably comes closer to doing what we demand of the OPM: it actually tells us what effect our poverty-fighting efforts have had over time. When researchers took the SPM methodology and extended it back through time, they found a pretty consistent fall in America's poverty rate, from 26 percent in 1967 to 16 percent in 2014.

But the Supplemental Poverty Measure hardly ends the debate, either. It does not include government programs like Medicaid, for example — there remains a big argument over how to reduce the value of something like health care coverage to a simple dollar figure. The SPM also adjusts for different costs of living in different geographies, meaning the poverty rate it might find in a low-income area of Mississippi will actually be lower than the rate the OPM would find there. Finally, while the SPM treats expenditures like child care and medical bills as necessities — i.e. you're defined as in poverty if you can't afford them — it does not do the same for education costs or the ability to save. Those are effectively treated as luxuries, and whether you can afford them or not doesn't factor into the SPM's assessment.

Perhaps the most interesting change from the SPM to the OPM is how they set their thresholds. For the OPM, if that basket of foodstuffs, defined way back when, costs more than a third of your family budget, then you're in poverty. But the SPM does something more complex. It includes a bundle of needs — like food, clothing, shelter, and more — and sets its value at 33 percent of median income. This may sound really technical, but think of it this way: If every American, from the richest to the poorest, suddenly got $10,000 more a year, the OPM rate would drop to zero or something close to it. But the SPM rate would remain largely unaffected.

In other words, the OPM measures how many people fall below a standard of living that remains fixed over time. The SPM measures how many households fall past a certain distance from the country's median living standard. Both measures of the poverty rate can theoretically be reduced to zero. But while the OPM could be reduced to zero without affecting inequality at all, inequality would have to shrink — at least for the bottom portion of earners — for the SPM to fall to zero. This is generally understood as a debate between "absolute" measures of poverty, like the OPM, and "relative" measures, like the SPM. (Though technically, both measures are relative comparisons: The SPM to how other people are doing now, the OPM to how other people did in the past.) Which approach is better is a hot topic of debate among people who pay attention to this stuff.

And this really gets us to the core problem with defining poverty: There's no right way to do it. We're sufficiently used to the official poverty rate at this point that we treat it almost as an objective, scientific measure. But it’s nothing of the sort. How to define poverty is an inescapably political question, and an inescapably moral one.

Should education count as a necessity in the bundle of goods we must be able to afford to not count as impoverished? What about savings, or child care, or health care, et cetera? What about leisure time? (Some European measures of poverty do indeed include the ability to go on vacation as a life necessity.) Should our poverty measures factor in inequality or not? Answers will depend on a person's values and worldview and ideology. (For what it's worth, none other than Adam Smith came down in favor of the relative approach, in a famous parable about a linen shirt.)

One last interesting factoid: Gallup has actually been asking Americans for decades how much a family of four would need to make "to get along in your local community." In 1963, Americans said it was $5,304, which was relatively close to the OPM's threshold of $3,128. By 2007, the gap had expanded dramatically, with the "get along" response reaching $52,087, while the OPM threshold lagged behind at $21,500. Turns out if you just adjusted the 1963 answer for inflation, you'd get $37,500, which is still way higher than the updated OPM. The conservative American Enterprise Institute (AEI) surveyed Americans in 2016 to find out what they thought it took for a family of four to not be poor, and the average answer was $32,293. The OPM threshold for a family of four in 2016 was $24,339.


And get this: The most common international poverty metric is a relative measure, set at 50 percent of median income. Both the OPM threshold of $3,128 in 1963 and the AEI survey response of $32,293 in 2016 are pretty close to 50 percent of median income at the time. Decades ago, America's official poverty line used to match the 50 percent of median income threshold, then fell behind. But Americans' general opinion of what counts has poor has kept up with or even exceeded that measure.

Rep. Ocasio-Cortez may be opening up a can of worms by asking the federal government to rethink its Official Poverty Measure. But she can take heart that her fellow citizens also think there's something way off in how we decide who is and isn't poor.