Should people with limited work history qualify for unemployment insurance?

When you lose a job in the U.S., you can turn to unemployment insurance to bridge the gap.

Except for when you can’t.

Unemployment insurance (UI) reaches only a fraction of people out of work. The unemployment rate typically exceeds the portion of workers claiming UI by a factor of three or more (Figure 1). With 10 out of 100 American workers unemployed following the Great Recession, only three were receiving UI support.

There is no UI for self-employed people, independent contractors, or “gig economy” workers, like rideshare and delivery drivers. In most cases, you can’t claim UI if you quit or if you’re fired for misconduct.

Those examples likely come as no surprise. But what about when a layoff strikes one of the following workers?

• A new graduate three months into a first job
• A retiree who takes occasional shifts at a gift shop to supplement Social Security
• A manual laborer who works seasonally and sporadically
• A stay-at-home parent who picks up weekend hours at the hardware store to supplement the family income

Actual eligibility depends on the details, of course, and a plethora of state rules, no two of which are the same.¹ Every state, however, has some requirement for sufficient work history, which is where folks like these might fall short. Even if you have an eligible job, you must have earned a certain amount of money within a recent period of time.

Excluding “a large portion of society”

There has been little academic research to date on workers excluded because of insufficient work duration or earnings. That’s a significant research need, given how many of these people there appear to be. A new working paper by Amanda Michaud, senior research economist with the Opportunity & Inclusive Growth Institute, estimates such ineligible workers make up 26 percent of the labor force—that would be more than 43 million Americans as of early 2023.

“This is a large portion of society, and they’re facing rules I have not seen the economics literature try to understand,” said Michaud. In “Expanding Unemployment Insurance Coverage,” she develops a theoretical framework to study how these ineligible workers react to conditions and incentives. She then enlists machine learning (a form of artificial intelligence, or AI) for the challenging task of identifying them in U.S. census data.

This allows Michaud to explore how these ineligible workers responded when the doors to UI were suddenly opened wide to them in 2020, before shutting again 18 months later. The results could inform proposals for expanding UI to cover more workers, either on a permanent basis or as a temporary, automatic measure during recessions.

A matter of moral hazard

Why exclude these low-earning or low-work-duration workers from UI in the first place?

Under federal law, Department of Labor regulations require states to interpret and apply a standard known as “able and available for work.” Although UI functions as a safety net, the objective is not social support but getting workers back on the job. Thus, regulations require states to determine that a worker seeking UI benefits will be credibly available for reemployment. This includes that states may consider “the previous work history of the individual (including salary and fringe benefits), and how long the individual has been unemployed.” An insufficient work history, per the regulation, could be evidence that a person has “limit[ed] his or her availability in such a way that the individual has withdrawn from the labor market.”

Another way to put it: “One rationale for these rules around UI is to stop people from cheating,” said Michaud. The logic is that someone who does not work much (as evidenced by recent history) has a stronger motivation to collect benefits for as long as possible, rather than look for a job and leave UI sooner. Providing unemployment benefits creates what economists call “moral hazard”—for all workers, to some degree, but particularly (or so the logic goes) for someone less attached to the labor market to begin with.

The Labor Department does not specify a minimum work-history requirement for states to comply with the “able and available” standard. In practice, however, all states have established mathematical formulas that, per Michaud’s research, appear to exclude many working people.

No such thing as a typical ineligible worker

Michaud’s analysis highlights one challenge with many states’ work-history formulas: They carve out workers with a wide range of motivations and incentives.

One reasonable assumption might be that people with limited work or earnings histories live in poor households. Indeed, Michaud finds that many of them do. Compared with eligible workers, they are more likely to be in poverty or suffer from food insecurity (Figure 2).

However, “you don’t want to study individuals in isolation,” said Michaud. “You want to understand the families that they live with and the other financial resources they have.” Alongside low-income people, she said, ineligible workers include “secondary-earners, retirees, young people who are going to school who have great financial support behind them and can move back in with their parents.”

For example, Michaud’s data show that 1 in 5 ineligible workers has an annual household income over $100,000. Nearly 60 percent have a college degree (Figure 3). Many households might fit the profile of “wealthy hand-to-mouth”—higher income, but similarly high debt payments and other expenses.

Whether these different types of workers deserve UI benefits is a reasonable subject for ethical debate. As a matter of economics—and per the stated legislative purpose of the UI program—the central question is: How motivated are they to go back to work?

The Great Recession: Without UI, ineligible workers went back to work faster

Michaud uses AI to identify workers in U.S. census data who were highly likely to be eligible or ineligible for UI based upon insufficient work or earnings histories. She then applies her data and model to the Great Recession (when UI was not extended to ineligible workers) and the COVID-19 pandemic recession (when it was).

During the Great Recession (2007–9) and its recovery, Michaud finds that unemployed workers ineligible for UI went back to work at a 25 percent faster rate, on average, than those who were eligible. This does not mean that the lack of UI was the only reason these workers were more quickly back on the job—the two populations could differ in other ways. Notably, however, the job-finding rates of eligible and ineligible workers became largely similar as eligible workers ran out of their benefits, suggesting UI (or the lack of it) was playing a motivating role.

Michaud’s theoretical model predicts an outcome similar to these actual data from the Great Recession. That lets her confidently use the model to test how things might have been different if the ineligible workers had received UI during this period. She finds that the job-finding rate of ineligible workers would have been 53 percent lower if they had been able to rely on UI during the Great Recession and its aftermath. In other words, the “moral hazard” introduced by UI for these workers would have been substantial.

The COVID recession: Suddenly, UI for everyone

With jobs of many types upended by the pandemic, in March 2020 Congress provided funds for states to extend UI to nearly all workers by establishing Pandemic Unemployment Assistance (PUA). At its peak, PUA was paying weekly benefits to more than 15.3 million unemployed people who would not otherwise have qualified for UI—more than 9 percent of the U.S. labor force. For most of the pandemic, PUA served more workers than regular UI (Figure 4).

PUA rolled out alongside an extended benefits period and increased benefit levels for all unemployed workers. States had different application processes, start dates, and end dates, making post-pandemic UI complicated terrain for researchers like Michaud to navigate.

But after accommodating these many moving parts into her analysis, Michaud finds that PUA recipients—workers who would have been ineligible in normal times—stayed on UI 1.7 times longer than other unemployed workers. She finds that if PUA claimants had been exiting UI at the same rate as regular recipients, total claims would have been up to 18 percent lower.

This “is not alone a smoking gun,” Michaud writes, since other factors common to PUA claimants’ jobs could have made it more difficult to return to work. However, the outcome is the opposite of normal economic times or the Great Recession, when these ineligible workers returned to work faster than others.

Taken together, the evidence from the Great Recession and the pandemic seems to support the original logic of excluding these workers: With UI benefits, they take longer and are less likely to go back to work. But Michaud’s model and AI-driven data let her look deeper, and the story gets more nuanced. Although PUA recipients did collect unemployment longer than other workers, they also went back to work much faster than expected.

A pandemic puzzle

Boosted pandemic UI payments were worth much more, in many cases, than these workers could expect to make in the workplace. Michaud’s theoretical model—accurate for predicting outcomes during the Great Recession—predicts PUA recipients would have expended minimal effort on job searching until their benefits were about to expire.

And yet in the real world of 2020–21, these workers were busy searching for and landing new jobs. They went back to work and ceased their UI benefits about one-third faster than the model would predict. No adjustments Michaud made to the parameters of the model could account for the PUA exit rate of actual workers during the pandemic. “Through the lens of the model,” she writes, “a surprising puzzle emerges of ‘why were PUA claims so short?’”

It could be that unique pandemic conditions were a factor, with workers behaving differently than they would in a typical recession. It could also be that these workers—even those who work few hours or earn relatively little—place some additional, nonmonetary value on having a job that Michaud’s economic model does not fully capture. Even with UI during the pandemic, they wanted to work—at least, more than we would have thought.

Different (ineligible) workers, different incentives

For policymakers, Michaud’s research highlights an essential distinction between two types of workers ineligible for UI because of work history: those who earn too little and those who lack sufficient duration on the job. Many state formulas exclude them both, but state or federal expansions of UI could treat them differently.

“It matters which lever you move, the duration or the earnings,” Michaud said. In the data, workers without enough duration show a stronger drive to get reemployed than even workers who do quality for regular UI. Think of that new graduate barely into their first job. “They are trying to get experience and climb the career ladder,” Michaud said. “If we want to expand unemployment insurance, we should take away the duration requirement first.”

Workers ineligible because of low earnings are a diverse group, such as the stay-at-home parents or retirees earning some extra household income on the side. “Some of them don’t need as much insurance, because a lot of them have other means,” Michaud said. Others earn so little that a steady (if smaller) UI check would simply outweigh the expected value of working.

Expanding UI to this low-earnings group seems to have a stronger moral hazard effect, keeping them on the sidelines. “Maybe unemployment insurance is not the right tool to help these groups get through bad times,” Michaud said.

Informing calls to expand UI

Before the pandemic, The Hamilton Project at the Brookings Institution called for extending UI to part-time workers. Gabriel Chodorow-Reich of Harvard and John Coglianese of the Federal Reserve Board write that “these reforms would better align the eligibility criteria to a modern labor market in which many individuals have short spells out of the labor force or prefer part-time employment due to family or other considerations.” They acknowledge the “microeconomic” work-disincentive effects of UI, but argue that including more workers would, on net, “make UI a better macroeconomic stabilizer.”

The pandemic experience has brought fresh suggestions to expand UI, such as a 2022 report from the JPMorgan Chase Institute that looks closely at PUA recipients from Ohio and New Jersey (via their Chase bank accounts). The report found PUA recipients exited UI at a rate “generally comparable” to other workers, especially after the supplemental benefit fell from $600 a week to $300. They also found that PUA recipients had 25 percent lower pre-pandemic income than regular UI recipients.

“PUA mitigated labor income risk for workers more marginally attached to [the] labor force, with no clear evidence of increased work disincentive effects,” the authors write. “This potentially warrants UI reform to broaden eligibility more permanently or to create a second tier level of income support for unattached workers.”

By building a theoretical framework and enlisting AI to identify and better understand the millions of Americans whose limited work histories exclude them from UI, Michaud brings new data to the debate—and an economic model to help interpret what these other researchers are reporting.

“Where did these rules come from? Should they be different? Should we expand UI over the business cycle?” Michaud asks. “It affects so many people. I just want people to get interested in these questions.”

Endnote

1 The UI system was established by Congress in the Social Security Act of 1935 and programs must follow basic federal guidelines. However, UI is primarily state-funded (via taxes on employers) and states have wide latitude to set and administer program rules.