SIEPR researchers examine exposure to racial diversity
Federal safety net programs commonly face concerns about who is getting financial assistance from the government. Fears of cheaters are often part of the debate.
Now with the coronavirus effect, such programs could become more impacted.
In his latest research on the efficiency of federal disability insurance programs, which preceded the COVID-19 outbreak, Stanford economist Luigi Pistaferri examined the other side of the coin — those who are wrongfully not receiving aid. And he discovered a striking disparity: Women facing permanent or severe work-limiting impairments are 20 percent more likely than men to have their claims wrongly rejected.
“The consequences of bias are potentially severe: Those rejected for disability insurance despite not being able to work have often very few alternative avenues of support, and this is a long-term problem,” Pistaferri and his co-author Hamish Low of Oxford, stated in their working paper posted by the National Bureau of Economic Research.
“The finding of substantial gender-based errors raises important policy issues,” they concluded.
While the pandemic doesn’t change the research findings, Pistaferri, a senior fellow at the Stanford Institute for Economic Policy Research (SIEPR), says it could affect the current landscape of disability insurance programs in multiple ways.
More people may end up applying for disability benefits amid increased job losses and exacerbated mental and physical health problems, he said. And with shelter-in-place restrictions, the processing of claims could take longer than usual. Reviews of existing disability beneficiaries have also been suspended.
Low added that the diversion of resources to treating COVID-19 patients means the treatment of other illnesses will be delayed, allowing them to become more debilitating and therefore increasing the need for disability insurance.
“In these times, making sure the people that need the insurance receive the insurance will be more important than ever,” he said.
Here, Pistaferri discusses the research and its implications.
Why is disability insurance an important regulatory issue to study?
The Social Security Disability Insurance program and the Supplemental Security Income program, which are both administered by the Social Security Administration, pay benefits to about 6 percent of the working-age population, and spending on benefits has almost doubled since 1985. That’s because 1984 was the last time the disability programs were reformed in a significant way. It became easier for people with mental health and musculoskeletal conditions to get into them.
What did you set out to examine in this study?
We were interested in how the screening process works, whether errors are made at the point of screening, and how big those errors are. When you apply for disability insurance, a big problem is that an evaluator doesn't know whether you are truly disabled under the government criteria set, so they could make two types of errors.
The first type is where they err on the side of caution, and they turn you down even if you are truly disabled and are unable to work. Or, they interpret the evidence you provide in a benign way, so you can actually get into the program even though you're not technically disabled and can still do some work.
My study focused on the first type of error and wanted to see if these errors differed by characteristics of individuals — race, gender or the type of condition you have.
What’s at stake when such errors are made?
A lot of people actually die or their health gets significantly worse while waiting for disability insurance that they’re likely entitled to. So our study is about cases like that and how these errors are really costly for people who are not on disability insurance.
What did you find?
We actually did not start with the gender focus. We started by thinking, ‘Is it the case that errors are worse with conditions that are harder to detect or verify?’
Back pain is a classic example. A lot of people have back pain from anatomic issues but they can’t provide medical evidence showing exactly what’s wrong.
There is really no test that can show how much pain you are in. There is no “pain-o-meter” for back pain.
In contrast, some people might have a physical problem, like a herniated disk, but they don't actually have pain, so it is actually easier for them to exaggerate their disability. And mental health problems — there’s an entire spectrum for that.
While we were studying these issues, we discovered one characteristic that really seemed to matter a lot — gender.
It jumped out at you?
Yes, it was too high to neglect. It was so evident there were big differences by gender. We threw everything at this to actually make sure the effect was there. And we found there was no statistical difference between blacks and whites; there doesn’t seem to be a racial bias.
Tell me about these gender differences.
Imagine two very similar people applying for disability insurance. They have the same level of education, labor market experience, self-reported health characteristics and clinical indicators, like back pain or cardiovascular disease. They both say, ‘I have a disability, and because of that condition, I cannot do certain kinds of work.’ And there's only one thing that differs — one is a man, and one is a woman.
If you're a woman, you're 20 percent more likely to be turned down than a man with identical characteristics. That’s the key result of the study.
What’s the potential explanation of that? Could it be an inherent bias, or is it built into the assessment somehow?
This was an important question we asked ourselves, and in the end, we were left with evidence supporting that there seems to be a higher threshold set by the Social Security Administration for admitting women into the program. The Social Security Administration concludes that women are more likely to work than men with the same self-reported disability.
That's why there are more errors on women than there are on men.
Previous studies by you, SIEPR Director Mark Duggan, and others have revealed problems with the disability insurance program. Can you connect how this all fits into the larger employment picture?
We think that the big problem with the Social Security Disability Insurance program is that it has been used as a sort of substitute for a long-term employment insurance program that doesn't exist in the U.S. A lot of people were employed by industries or occupations that were killed by automation, international trade or technological change, so something led them to lose their jobs or to have permanently low wages.
In normal circumstances, these people would work, or prefer to work. But when they don't have a lot of economic opportunity — whether there are no jobs for them, or maybe they're stuck in low wage jobs — they might actually prefer to move into a program like disability insurance as a substitute. This means that applicants for disability insurance are a mix of healthy individuals who cannot find work, and truly work-limited individuals, creating a difficult problem for the Social Security Administration in separating these types of individuals.
Our focus is on one side of this problem: How well are the work-limited identified and insured? However, the underlying issue is how much insurance and retraining should be available to those unable to work, whether due to poor health or because their industries or occupations have declined.
Given your latest findings here, how can the application process be improved?
You can potentially make all applications gender-blind, but that’s not super feasible because there are disabilities that are gender-specific, like breast cancer or testicular cancer. In cases like musculoskeletal or mental health problems, there should be no need for a Social Security evaluator to see your gender, so you can decide to make those applications gender-blind.
There’s another thing. The nature of the application process already puts women at a bit of a disadvantage. If you get to the last step of the screening process, Social Security sends another form where you have to list daily living activities. They ask: Do you take care of pets? Do you take care of a parent? Do you take care of the shopping for the household? And we think that because of the role that society has assigned to women, women are more likely than a man to say, ‘I'm not sitting on the couch all day, and despite the disability, I still do the shopping and take care of the kids and my mom, etc.’
So we should reform perhaps the way the information gets collected. We’re not the first to say the screening process is imperfect, but we’ve just revealed a striking element of imperfection in the process.