Few Unemployed Service Sector Workers Received UI and Many Experienced Hardships
The effect of the coronavirus outbreak on the U.S. labor market has been profound. In the early weeks of the outbreak, the unemployment rate skyrocketed from 4% in February to almost 15% in April of 2020. Although the economy has partially recovered since April, as of August 2020, the unemployment rate stood at over 8%, more than twice as high as it had been just 6 months prior. The economic toll of the coronavirus outbreak has been particularly severe for service sector workers. As state-wide orders to close businesses went into effect, many retail, food service, and hospitality workers experienced layoffs and furloughs. Jobs in these sectors are currently at half to two-thirds of their previous levels.
In response to the sudden spike in unemployment, the Federal government took some emergency measures, notably passing the Coronavirus Aid, Relief, and Economic Stimulus (CARES) Act. The CARES Act temporarily expanded both benefit amounts and also eligibility. The expanded benefit amounts expired in July and were replaced, but only in some states, with more modest increases in benefit levels. These emergency responses reinforced the Unemployment Insurance (UI) safety net but also added additional complexity to a UI system that was already complex and widely variable across states.
In conjunction with the rapid rise in unemployment, UI claims have also spiked. Initial UI claims have been over a million every week since the end of March and continued UI claims have been above 25-30 million since the end of April. At the end of June, nearly one in five workers filed for UI claims.
However, and importantly, not all of these claims will materialize into actual benefits received. In the years prior to the pandemic, workers faced a number of hurdles in the process of applying for UI. Workers needed to document their job searches weekly, experienced long response times, and technical glitches on state websites made it difficult for workers to receive and stay qualified for UI benefits. These existing inefficiencies have been exacerbated by the unprecedented levels of initial UI claims filed in the wake of COVID19. While the Pandemic Unemployment Assistance (PUA) program under the CARES Act broadened UI eligibility criteria, allowing some with low or irregular earnings to newly qualify, many workers did not know of the program or had trouble applying. PUA rollout timing also varied widely between states and some states required applicants to first go through the regular application process and be rejected before applying for the PUA program.
These barriers to accessing UI could have dire consequences for unemployed workers. American households, and especially front-line service sector workers, were already financially fragile before the pandemic struck. Many service sector workers were already on the edge of a financial cliff, with little financial buffer against an economic shock. UI has the potential to make the difference between getting by and experiencing severe material hardships.
When unemployed workers access UI, it can serve as an effective safety net. The federal aid provided by the CARES Act has kept consumption high and may keep poverty levels near their pre-pandemic levels. Prior research found that UI recipients worried less about meeting their basic needs, and food insecurity declined. UI benefits and the additional benefit amounts available through the CARES Act have kept workers from facing worse economic hardship and poverty levels. The rollback in these supplemental UI benefit amounts could lead to more food insecurity and higher eviction rates.
In this brief, we draw on survey data from The Shift Project collected in the Spring of 2020 from around 2,500 workers who had lost their service sector jobs. We address key questions about the UI response to the economic crisis triggered by the pandemic:
- To what extent are economically vulnerable workers able to access UI benefits? When workers are not accessing UI, what are the hurdles that get in their way?
- How does access to UI benefits vary across U.S. states?
- What are the consequences of having or lacking access to UI benefits in terms of experiencing material hardships?
We find that only about 1 out of 4 unemployed workers in April and May received a UI payment, suggesting major barriers to access. We also find a huge gap across states in access to UI with 77% of unemployed workers in Minnesota accessing benefits compared with a scant 8% in Florida. Finally, we show that UI benefits make a big difference in helping workers to avert material hardships. When UI benefits are accessible, they are an effective safety net, but a large majority of workers are not accessing this important benefit when they need it.
Hurdles in the UI Process
Recent debates over the appropriate amount of Unemployment Insurance benefits often assume that unemployed workers will actually receive these benefits. In Spring 2020, we surveyed 2,561 unemployed people who had been laid-off or furloughed from their jobs at one of 110 of the largest firms in the retail, food service, hospitality, grocery, pharmacy, fulfillment, or hardware sectors. We found that most of those who experienced job loss before June were in fact not receiving Unemployment Insurance. Specifically, for every 100 unemployed former “essential workers,” we find that just 27 had actually received unemployment insurance benefits. Unemployed people are cut out from receiving benefits at several points along the way.
To illustrate the several ways that many unemployed workers are diverted before receiving UI, Figure 1 depicts the percentage of workers who pass through each stage of the UI application process. Figure 1 shows how the Unemployment Insurance system functions as a funnel for these former workers.
Of all workers who are unemployed, 24% did not even try to apply for UI. A large share of these non-applicants report that they did not apply because they did not believe they were eligible (52% of non-applicants). Smaller, but significant, shares report that they did not apply because they did not know how (11% of non-applicants) or because they had not yet had time to apply (7% of non-applicants). Another 5% of unemployed people tried to apply, but either found the application too complicated or experienced technical problems that prevented them from completing the application process. In all, just 71% of unemployed workers report that they were able to complete an application for UI.
Seven percent of unemployed workers who completed a UI application reported that their application was denied. Survey respondents’ understandings of why they were denied were revealing: In spite of expanded eligibility criteria, more than one-third were denied because of insufficient prior earnings or hours. Another 17% reported they were denied because of technical difficulties. (The remaining share were denied for other reasons or did not specify.)
Twenty-five percent of the unemployed had completed a UI application, had not yet been denied, but also had not yet been approved. They were still awaiting a response to their application.
Of the remaining 39% of unemployed workers whose UI application had been approved, about one-third had not yet received a benefit payment. Just 27% of all of those who were unemployed said that they had applied, had been approved, and had received benefits.
Although the share of unemployed workers receiving UI benefits may increase over time, our data suggest that workers experience a sizable period of time without benefits. Even among those who had been unemployed for two months or more, only 25% had received a UI payment.
It Doesn’t Have to Be This Way
The COVID19 pandemic has posed a uniquely difficult set of circumstances, with an unprecedented number of unemployed former workers coming up against an all-to-easily overwhelmed UI system. The “funnel” shown in Figure 1 shows the scale of the consequences when we look across the country – only a small minority of unemployed workers successfully received UI. But, we can also break this down by state.
We find significant variation across states in the share of UI applicants who actually received UI, after adjusting for differences in applicants’ age, race/ethnicity, parental status, gender, marital status, educational attainment, school enrollment, length of unemployment spell, former sub-sector, and reason for unemployment.
As shown in Figure 2, unemployed former workers fared the best in Minnesota, where we estimate that 77% of those who applied received their UI benefits. The unemployed also reported high rates of receipt in Massachusetts (65% of applicants) and Virginia (64% of applicants). About half of applicants in California, Missouri, North Carolina, New York, and Tennessee had received their UI benefits. Rates of receipt were much lower in Colorado (25%), Illinois (24%), Indiana (27%), and Ohio (24%). But, Florida stands alone at the bottom of the list, with just 8% of applicants having received their UI benefits.
Failures in the UI System Lead to Household Economic Insecurity
Delays and dysfunction in the UI application process have consequences. Essential workers, even when employed, live close to the economic margin – struggling to make ends-meet on low-wages and in the face of unpredictable schedules. With little economic buffer, when unemployment strikes, hunger and housing insecurity can quickly follow. Unemployment Insurance, if received quickly, can provide a crucial buffer against economic calamity for these workers.
Unemployed workers who had applied to UI but either had not yet heard back or not yet received their approved benefits experienced deep deprivation and very high levels of economic insecurity. The top panel of Figure 3 shows that, over the prior month, 26% of these former workers had gone hungry because they couldn’t afford enough to eat and 13% experienced housing insecurity – having to double-up or staying in a shelter or other place not meant for housing. Nearly one in five (18%) unemployed people who hadn’t received UI also reported that someone in their household didn’t get medical care they needed because of the cost. As shown in the bottom panel of Figure 3, even larger shares had trouble paying essential utility bills (35%) or struggled to cover their expenses and pay bills (43%). These unemployed people also appear to have quickly depleted private safety nets – by the time we surveyed them, 57% reported that would be unable to cope with just a $400 expense shock.
Actually receiving UI benefits made a tremendous difference. Unemployed workers who weren’t stuck waiting for UI experienced significantly less hardship and economic insecurity. These former workers were just half as likely to experience hunger and only a third as likely to experience housing insecurity as those who were unemployed but hadn’t received UI, and they were also significantly less likely to have to forgo medical care, not be able to pay their bills, or be unable to cope with a $400 shock.
In fact, unemployed people who actually received Unemployment Insurance generally fared just about as well as those who were still working. Receiving unemployment insurance meant that unemployed former workers and those who were still working had the same level of hunger hardship, of housing insecurity, and of forgoing medical care. Unemployed workers who received UI were just slightly more likely to have trouble paying bills or not being confident in their ability to cope with an expense shock as those who were still employed.
But it is vitally important to recognize that those levels of deprivation and economic insecurity were still very high. Indeed, 14% of these currently employed “essential workers” reported going hungry in the past month because they couldn’t afford enough to eat, and 5% experienced housing insecurity. That the receipt of unemployment insurance brought unemployed people down to that same level of deprivation as employed people is evidence of the vital role that UI can play, but it does not diminish the fundamental precarity and poverty with which “essential workers” contend.
Essential workers in the service sector have played a key role in combating the COVID19 pandemic, allowing Americans to effectively socially distance and quarantine by staffing the front lines of grocery stores, restaurants, and pharmacies, and doing the work of fulfillment and delivery. But workers in the service sector have also borne some of the worst economic shocks of the COVID19 recession. We report on the experiences of thousands of essential workers in the service sector during the late spring of 2020.
We find that workers who were laid off or furloughed struggled to get timely Unemployment Insurance payments, with just a quarter of workers who became unemployed getting benefits by the time of survey.
This delay was not inevitable. We find radically different experiences across US states, with three-quarters of unemployed workers in Minnesota who applied for UI receiving benefits, but just 8% of workers in Florida. These state differences cannot be put down to differences in worker demographics, industries, or the timing of their applications.
These delays had huge consequences for workers and their families. Service sector workers struggle to get by even in relatively good economic times. With little economic margin for error, unemployment can have immediate catastrophic consequences. We find that 26% of unemployed workers who had not received UI benefits went hungry because they couldn’t afford enough to eat, 13% experienced housing insecurity, and nearly one in five said someone in their household had to forgo needed medical care. Actually receiving UI benefits made a huge difference, taking the level of hardship down to the same amount experienced by service sector workers who kept their jobs.
As the pandemic continues over the next several months or longer, the economic toll on unemployed workers is likely to worsen over time. Our research findings indicate that the UI system has been an effective tool for averting serious hardships for those who can access benefits. Unfortunately, the majority of unemployed and furloughed workers fell through holes in this safety net in the early months of the pandemic. The wide state differences demonstrate that it doesn’t have to be this way.
The Shift Project has collected survey data from hourly service-sector workers employed at large retail and food establishments since the fall of 2016. This brief focused on a subsample of 8,350 respondents interviewed in April and May of 2020. The survey data collection was national in scope and the survey sample includes respondents from all 50 U.S. states and Washington, D.C.
The Shift Project recruits survey respondents using online Facebook/Instagram advertisements, targeted to workers employed at large retail and food-service employers. In the Spring of 2020, we leveraged Facebook’s advertising targeting system to reach both workers who were currently employed at (5,789) and workers who were recently laid-off or furloughed from (2,561) the same set of 126 large service sector employers. We were able to identify laid-off and furloughed workers because Facebook’s advertising targeting platform continued to identify them with their former employer.
Those who responded to the Shift survey invitation were automatically routed to a survey landing page where they were asked to consent to participate in the study, then began the online self-administered survey using the Qualtrics platform. As an incentive, those who completed the survey and provided contact information were entered into a lottery for a $500 gift card. The survey included modules on job characteristics, work schedules, demographics, economic stability, health, parenting, and child outcomes. For furloughed and laid-off workers, we asked a series of questions on UI. To screen out invalid survey responses, we used an attention filter (a question that instructed respondents to select a particular response category to verify the accuracy of their responses).
The survey recruitment approach yields a non-probability sample of workers, which may differ from the broader population of service-sector workers. To mitigate this potential bias, we have applied weights that adjust our sample to reflect the universe of service-sector workers in the United States.
We construct survey weights to adjust the demographic characteristics of the Shift survey sample to match the demographic characteristics of service-sector workers in the American Community Survey (ACS) for the years 2008-2017. We align the ACS sample with the Shift sample by selecting workers in the ACS who are employed in the same occupations and industries as the Shift sample.
In our analysis of workers’ progression through the UI application process, we estimate weighted unadjusted descriptive statistics. These descriptive statistics are reported for the 2,561 unemployed or furloughed workers who completed this section of the survey. These analyses describe the share of respondents who reached various benchmarks of the UI process by the time of survey. On average, when we surveyed them, these respondents had been unemployed for 1.2 months, with 34% laid-off of furloughed less than 1 month prior, 38% 1 month prior, 13% 2 months prior, and 14% 3 months or more prior. We find that there were not significant differences between states in these durations or between those who had received a response from UI or not.
For our analysis of state differences in workers’ progression through the UI application process, we first limit our sample to respondents in states that had at least 50 unemployed respondents in the survey data. We define a dichotomous dependent variable equal to one if respondents reported receiving their UI benefits and equal to 0 if the respondent reported applying, but had not received their benefits. We then estimate an OLS regression model predicting UI receipt as a function of a set of indictors for state, along with controls for respondent age, race/ethnicity, gender, marital status, parental status, educational attainment, and school enrollment. We further adjust for length of unemployment, whether the respondent was laid-off or furloughed, and the sub-sector in which they were formerly employed. We limit the sample to respondents who began the UI application process. The model is weighted using the ACS weights described above. After estimation, we take predicted values of UI receipt by state after adjusting for these covariates. We then map the predicted values.
For our analysis of the protective effect of UI receipt on household economic poverty and insecurity, we estimate a series of OLS regression models. We specify six outcome variables.
- Hunger Hardship. Respondents were first asked if, “In the past 12 months, were you ever hungry, but didn’t eat because you couldn’t afford enough food?” and then, if they answered in the affirmative, were asked, “In the past month, were you ever hungry, but didn’t eat because you couldn’t afford enough food?” We code a dichotomous variable equal to 1 if respondents reported that they had been hungry because they couldn’t afford enough food in the last month, and zero if they did not experience hunger due to not being able to afford food or only experienced it in months 2-12.
- Housing Hardship. We construct this variable using the same logic as above, but here respondents are coded as “1” if they either “moved in with other people because of financial problems” or “stayed in a shelter, in an abandoned building, an automobile, or any other place not meant for regular housing, even for one night” in the prior month and are coded as “0” otherwise.
- Medical Hardship. We construct this variable using the same logic as above, with respondents coded as “1” if there was “anyone in their household who needed to see a doctor or go to the hospital but couldn’t go because of the cost” in the prior month and are coded as “0” otherwise.
- Utility Hardship. We construct this variable using the same logic as above, with respondents coded as “1” if they did “not pay the full amount of gas, oil, or electricity bill because you didn’t have enough money” in the prior month and are coded as “0” otherwise.
- Difficulty with Bills/Expenses. We construct a dichotomous variable equal to “1” if respondents report that it was either “very difficult” or “somewhat difficult” for them to “cover your expenses and pay all your bills” and “0” otherwise.
- Coping with Expense Shock. We construct a dichotomous variable equal to “1” if respondents report that they “could probably not” or “certainly not” come up with $400 if an unexpected need arose within the next month,” and “0” if they report that they “could probably” or “certainly” do so.
First, we estimate a model that compares currently employed and currently unemployed workers, with controls for age, gender, race/ethnicity, parental status, marital status, educational attainment, school enrollment, and industry. We generate predicted values on each outcome for employed respondents. Second, we estimate a similar model that is restricted to those respondents who are unemployed and compares those who received UI benefits with those who either (a) had not heard back on their UI application or (b) reported that they had been granted benefits, but that these benefits were yet to arrive. We control for age, gender, race/ethnicity, parental status, marital status, educational attainment, school enrollment, and former industry, duration of employment, and whether laid off or furloughed. These models omit respondents who did not apply for UI or who were denied UI based on the logic that such respondents are more likely to differ from respondents who received UI on unobservable and potentially confounding characteristics. We generate predicted values on each outcome for respondents who received UI and those who had not. Both models employ the survey weights. We then plot the predicted values on each outcome for employed respondents, unemployed respondents who received UI, and unemployed respondents who had not yet received UI.
For a detailed discussion of The Shift Project data collection, methodology, and data validation, see:
Schneider, D. and K. Harknett. 2019. “What’s to Like? Facebook as a Tool for Survey Data Collection.” Sociological Methods & Research. https://doi.org/10.1177/0049124119882477.
The authors gratefully acknowledge the work of Rebecca Wolfe, Connor Williams, and Megan Collins in the preparation of this report.
The Shift Project gratefully acknowledges funding from the Bill and Melinda Gates Foundation, the Robert Wood Johnson Foundation, and the W.T. Grant Foundation. The views expressed here do not necessarily reflect the official views of the sponsors.