The Economy Type II Error Built

Our obsession with worthiness

You likely think that the US social welfare system—loosely defined as the benefits, subsidies, and tax credits the federal government transfers back to Americans through everything from housing subsidies to the Earned Income Tax Credit—is meant to help those in need. You might even think that it’s designed to fight poverty.

Consider for a moment: there is not a single public program for which you can receive benefits in the United States based on need alone. There is always some kind of “poor and…” modifier for the target population. For example:

• Supplemental Security Income (SSI) - Poor and

  • old or disabled,

  • with less than $2,000 in assets.

If you are poor but not disabled, or poor, disabled, and have $20,000 in a 401k, you cannot get SSI. But SSI eligiblity is at least understandable, take a look at SNAP:

• Supplemental Nutrition Assistance (SNAP aka food stamps) - Poor and

  • i) not on strike, ii) not undocumented, iii) not in college at least half-time, iv) not certain type of legal immigrants, v) not a drug felon

  • with less than $2,750 in liquid assets,

  • compliant with general work requirements, applied to everyone age 16-59, unless exempt (there’s a lot of exemptions), meaning participant must:

    • register for work

    • accept suitable work

    • be in a training program

    • not quit or reduce work below 30 hours a week,

  • compliant for ABAWD work requirements, applied to able bodied adults without dependents who are age 18-56 (the debt ceiling deal raised this to 56 from 49), meaning participant must;

    • work 80 hours a month

    • participate in a work program for 80 hours a month, such as the SNAP Employment and Training program

    • participate in a combination of work, work program, or workfare for 80 hours a month.

It’s not like we’re exactly giving away the Taj Mahal here. This complex, overlapping compliance system is to get $200-$700 monthly coupons for food, coupons that have very strict limits on what can and cannot be purchased.

If food stamps are meant to help the needy, why do we make them jump through so many hoops?

And the number of hoops that poor people have to jump through to get benefits, not just food stamps but numerous other programs, has spawned its own area of research called administrative burden. It’s best laid out in a book of the same name by two Georgetown faculty (with companion syllabus), who explain how bureaucratic difficulty is a policy choice. They give the example of voter ID laws—a paperwork hurdle that is neither necessary nor accidental, but a means of curtailing voting rights through administration of those rights.

The phrase that tells you all you need to know is the title of lecture on the administrative burdens of immigration law called, “A Thousand Petty Fortresses.”

So why the hoops petty fortresses?

The answer comes from statistics and the push and pull of Type 1 and Type 2 error, the proverbial yin and yang of accuracy. A Type 1 error means a test found something that wasn’t there, Type 2 error means a test didn’t find something that was there. This is intuitive, even if you can’t name it in statistical terms. If you need a covid test, you know that the rapids can miss a positive case at first—they have a high Type 2 error. But if you take a PCR, they can find a case that’s completely asymptomatic—which yes is technically accurate but feels like Type 1.

Type 1 and Type 2 error are commonly described as the false positive and false negative. In life, and statistics, these two errors are tradeoffs; you cannot minimize one without increasing the risk of the other.

Translated to the US social policy system, the accuracy here is about deserving. Benefits should only to go worthy, deserving people. Type 1 error would mean that someone benefited who wasn’t worthy. Type 2 error would mean that someone worthy didn’t benefit.

Our system is built on avoiding Type 1 error. I’d even go so far as saying it’s the North Star of policy design, because it trumps every other consideration.

Take the temporary expansion to the Child Tax Credit it 2022 that 1) made the credit fully refundable and 2) paid out the credit in advance in monthly installments. In effect, it created a universal monthly income for all children in America. An additional 19 million children received the benefit through the expansion. The maximum credit was $300 per month per kid.

This credit was small but mighty. It lifted around 3 million children out of poverty the second the money went out the door, the largest reduction in history. In the long run, reducing child poverty has returns on investment in the ten-to-one range, because poverty in childhood has lasting deleterious effects on education, income, and health well into adulthood. Poverty isn’t an isolated event for a kid. They don’t just shake it off.

And yet, the expansion expired without being renewed or made permanent. Key senators claimed parents would use the money to buy drugs. Conservative economists claimed that over a million parents might (operative word there is MIGHT) stop or reduce working if they had cash for their children with no strings attached, while others said it’d be closer to a half million who, again, might.

In sum, a benefit that immediately improves the lives of poor children and makes a generational investment in fighting poverty risks giving money to parents who are derided as druggy or lazy.

Type 1 error: Give the money anyway, knowing some parents would be unworthy of it or misuse it.

Type 2 error: Take the money away, knowing that many children and worthy parent benefited greatly from it.

We erred on the side of Type 2. Bye bye historic decreases in child poverty.

Do this enough times and in enough programs, and we build a Type 2 world, where we’d rather reduce or cut benefits for people we know are worthy rather than potentially help someone who is not.

It might have been lost in the swirl of negotiations around the debt ceiling but around 80% of SNAP households have at least one worker already. But this is America, where the minimum wage has been $7.25 for 14 years; our labor market produces many workers who are still in poverty. That creates an identification challenge for worthiness. People who can work should work—that’s the credo averred by every president since Kennedy, and the worthiness test for public programs. But, they also can be working and still be income eligible for benefits because their earnings are so low. So how do you find who is worthy?

Work requirements, or rather, paperwork requirements is the answer policymakers have settled on. Make people prove that they are working or looking for work. The snag is work requirements prove to be more of a hurdle than a boost. They are very good at getting people kicked off of programs for not filing paperwork correctly, but they do not increase work or earnings to any significant degree, and by kicking so many off of support, may increase poverty.

Type 1 error: Provide food stamps with no work strings attached, knowing that some may not work or work as much as a result.

Type 2 error: Put onerous paperwork requirements on a food program, knowing that many who meet those requirements will be kicked off the program for administrative reasons.

We erred on the side of Type 2, again.

This is as good of time as ever to point out that there is no federal law requiring employers to provide paystubs to employees. Most states do, but not the federal government.

The result of our obsession with avoiding Type 1 error—of having an unworthy person receive public funds—has made a Type 2 economy, where worthy people in need of help are often punished, forgotten, or overlooked.

I go back, again and again, to this figure from the Report on the Economic Well-Being of U.S. Households, produced annually by the Federal Reserve. It shows the share of adults who could meet an unexpected $400 expense in cash or cash equivalent (like a credit card that they could pay off that month). In a year when the economy is weak, like in 2013 when we were still crawling out of the Great Recession, only half of Americans could. But even in a year when the labor market is red hot, a third still can’t.

Two notes about this: First, for parents the number is always a notch lower. In 2022, it was 57 percent who could meet an expense (as opposed to 63 percent above). Second, this statistic is often misinterpreted as “one car repair away from financial ruin.” That’s not quite right, as most people who said they couldn’t meet the expense said they would put in on a credit card and pay it off later, others said they would borrow from a family member, sell something, or get a loan. Only a third of the third, or about 13 percent of all Americans, could straight up not pay it. More accurate is to say, “one car repair from financial hardship that could be severe.”

But going back to Type 2 error, a measure like this, that gets at precariousness, should upend our notion of who is deserving or why we judge people in need that way. Our economy fails to provide widespread economic security. And that’s the error we never try hard enough to correct.