clock menu more-arrow no yes mobile

Filed under:

Trump wants to change how poverty is calculated — to make fewer people eligible for benefits

It’s a subtle but important change.

Bloomberg Asks Fed Gov’t For Permission To Ban Food Stamp Purchases Of Sugary Drinks
It’s all about the prices that people like food stamps recipients pay.
Spencer Platt/Getty Images
Dylan Matthews is a senior correspondent and head writer for Vox's Future Perfect section and has worked at Vox since 2014. He is particularly interested in global health and pandemic prevention, anti-poverty efforts, economic policy and theory, and conflicts about the right way to do philanthropy.

The Trump administration has been incredibly consistent, from day one, about its desire to slash benefits for poor Americans, including Medicaid and food stamps (aka Supplemental Nutrition Assistance Program or SNAP). But with the threat of Democratic filibusters and now a Democratic majority in the House, Trump’s efforts to make those cuts a reality have been repeatedly stymied.

That hasn’t stopped the administration from coming up with new attempts, and the latest is subtle but profound: changing the inflation rate used to update the poverty line.

That might sound like a merely technical change. It’s not.

The poverty line is used to determine eligibility for a wide array of government programs. The change the administration is proposing would, over the course of many years, shrink the size of Medicaid, food stamps, free school breakfasts, Head Start, and many, many other programs. To be clear, the administration could certainly change the inflation measure in a way that makes more people eligible — but given its calls for spending cuts, it appears likely it’ll pick a measure that reduces eligibility.

It would not be a big change at first. In its first few years, the policy wouldn’t save much money or affect many people. But make no mistake: seen in the context of the Republican Party’s priorities, this is effectively a cut to government programs. If you think we don’t do enough to provide health care and food assistance to the poor, it’s a bad deal.

The difference between CPI and chained CPI

Here’s what’s happening: Trump’s budget agency has put out a “request for comment” on different ways to adjust the Official Poverty Measure (OPM) for inflation. In other words, it’s essentially laying the groundwork to change how we measure poverty, with an eye toward shrinking the pool of who’s eligible for benefits.

The official poverty measure was developed by the Social Security Administration’s Mollie Orshansky in 1963 and defined as three times the “subsistence food budget” for a family of a given size. As former acting Commerce Secretary Rebecca Blank (then a Brookings Institution fellow, now chancellor of the University of Wisconsin Madison) explained in 2008 congressional testimony:

The subsistence food budget for a family of four was based on the Economy Food Plan developed within the USDA in 1961 using data from the 1955 Household Consumption Survey. It was described as the amount needed for “temporary or emergency use when funds are low.” ... If the average family spent one-third of its income on food, then three times the subsistence food budget provided an estimated poverty threshold. This calculation was done for a family of 4, and so-called ‘equivalence scales’ were used to estimate how much was needed by smaller or larger families.

The current poverty line is this number, calculated in 1963 and based on 1955 data, updated by the Consumer Price Index in each year since.

It’s worth dwelling on this for a second. The way we measure poverty is based on a 58-year-old analysis of 64-year-old data on food consumption, with no changes other than adjusting the poverty line for inflation.

That makes the inflation adjustment incredibly important from the standpoint of figuring out who counts as poor.

Currently, the inflation measure the government uses to adjust the poverty line is the Consumer Price Index (CPI). But many economists believe that CPI overestimates inflation — which would imply that the poverty measure has risen too rapidly over time.

What would cause an overestimate? There are several sources, but the most important one in this context is substitution bias. Basically, the way inflation measures work is that they try to estimate the cost of a representative basket of goods at different points in time. The basket’s composition is meant to reflect the goods that consumers are buying — a little bit of housing, a little bit of food, a little bit of electronics, and so on.

But which goods consumers buy change over time, due to their relative prices and capabilities. Take MP3 players. Those were a significant part of US consumer spending in, say, 2004. But as storage on cell phones got bigger, and streaming got easier, fewer people were buying MP3 players, and consequently they make up a miniscule share of consumer spending in 2019. A similar story happened with laptops and tablets. Because tablets are a lot cheaper than laptops, and do most of what many home users want out of a laptop, a number of people have moved toward tablets and away from laptops over time.

But conventional inflation measures like CPI don’t incorporate that kind of substitution. While the details are complex, they usually keep the basket of goods whose price they’re measuring pretty constant over time, updating it only every few years.

In 2002, the Bureau of Labor Statistics introduced “chained” CPI (or C-CPI), which updates the basket on a monthly basis. Chained CPI typically grows more slowly than typical CPI, because it incorporates ways that consumers save money by switching to cheaper goods.

There are other alternative inflation measurement available, most notably the Personal Consumption Expenditures Price Index (PCEPI), which unlike either the CPI or C-CPI is measured on the business side, not the consumer side, and includes expenses like employer-sponsored health care that the CPI measures exclude. This is the metric upon which the Federal Reserve relies most often in deciding how to control inflation. There’s also CPI-W, which is used for Social Security cost-of-living increases and measures retail prices specifically as it affects hourly workers in cities, and CPI-E, which specifically measures spending by the elderly and tends to rise faster than the regular CPI.

The Trump administration request asks for comments on the usage of all of these measures for adjusting the poverty line. It’s typically neutral on which the administration might pick. But its interest is most likely in switching to chained CPI or PCEPI. Because they rise more slowly, using those metrics would lead to lower enrollment in programs pegged to the poverty line over time, whereas using CPI-E would lead to higher enrollment (and also be a strange choice for programs not targeting the elderly).

Indeed, chained CPI is the likeliest option for the administration to pick if for no other reason than because it was incorporated in the 2017 Republican tax bill. The thresholds for different tax brackets are indexed to inflation, and so this change means that over time more people will be pushed into higher tax brackets, raising revenue. The Obama administration for years pushed for a deal where chained CPI would be applied to both Social Security and taxes, in conjunction with increases in Social Security benefits for longtime recipients (who’d be hardest hit by the change, due to years of compounding).

It’s not clear whether, legally, the Trump administration can make this change on its own. As attorney and liberal welfare policy expert Shawn Fremstad notes, there are several statutes instructing the Secretary of Health and Human Services to update the poverty line in accordance with the CPI — not the chained CPI.

But the use of a request for comment suggests the administration might try to do this through regulation anyway, and fight it out in court later.

Georgetown law professor David Super noted in an email, “To the best of my knowledge, previous discussions have almost always assumed that legislation is required.” He thinks it would most likely be required to make this change too.

How to evaluate the proposal

It’s hard to get a measure of the impact of this change, but in the near term switching to chained CPI would not do a whole lot.

In 2013, the CBO estimated what changing to chained CPI across the government would do to spending. This would be a broader shift than simply changing adjustment of the Official Poverty Measure. The biggest budgetary savings would come from applying chained CPI to Social Security and the tax code, neither of which are included in the current Trump proposal. The CBO concluded that there’d be a $28.5 billion reduction in health spending over 10 years — but that includes changes to provider reimbursement that wouldn’t be affected by a poverty line change, and even if all of it came from a reduction in Medicaid eligibility, it’d only represent a cut of 0.6 percent of the $4.4 trillion in Medicaid spending the CBO projected over those same 10 years.

All that said, the cut would be very concentrated. It would cut some people off from Medicaid and food stamps entirely, while leaving most people on the programs unaffected. That means a small average effect, but an acute effect on the small number of people getting kicked off.

Defenders of the idea argue that what Trump’s proposing, beyond saving money, would simply make the poverty line more accurate.

I’m sympathetic to that argument; as an intellectual matter, a chained price index strikes me as clearly superior as it more closely tracks how the prices of goods people actually buy are changing — though as the Center on Budget and Policy Priorities’ Sharon Parrott notes, inflation might be higher for low-income families, in which case moving to chained CPI would arguably be a loss for accuracy.

But this is fundamentally a political decision, not a technical one; we’re not debating “what is the best way to measure the cost of living” but “are too many people getting Medicaid and food stamps, or too few”?

One could answer “too few” and still support using a more accurate price index, if it’s paired with an increase in the overall level of benefits. But that’s not what the Trump administration is doing. It’s seeking out ways to cut benefits over the long run, not attempting a recalibration of government metrics for greater accuracy.

In 2016, the conservative American Enterprise Institute conducted a big poll on poverty asking Americans, among other things, how much money a family of four — two adults, two kids — would have to earn to qualify as not poor. The median response was $30,000. By contrast, in 2016 the poverty line for a family of four was $24,300. The American people appeared to want a threshold that was 23 percent higher than the existing threshold.

It’s also important to think about this in the context of broader political goals. Do you support Medicare for all, or otherwise providing health insurance to all Americans? Then narrowing Medicaid is a move in the wrong direction. Do you support a basic income or other policies to expand cash benefits or cash-like benefits to the poor? Then shrinking food stamps, which are quite similar to cash, is a move in the wrong direction.

In any case, you can’t come to a conclusion on the merits of this idea just by thinking about the technical details of inflation measurement. It always comes down to how big or small you want the welfare state to be.

Sign up for the Future Perfect newsletter. Twice a week, you’ll get a roundup of ideas and solutions for tackling our biggest challenges: improving public health, decreasing human and animal suffering, easing catastrophic risks, and — to put it simply — getting better at doing good.