Millions of the newly jobless are going without benefits as the unemployment system buckles under the weight of new claims, according to our new national survey, conducted in mid-April.

For every 10 people who said they successfully filed for unemployment benefits during the previous four weeks:

• Three to four additional people tried to apply but could not get through the system to make a claim.
• Two additional people did not try to apply because it was too difficult to do so.

These findings imply the official count of unemployment insurance claims likely drastically understates the extent of employment reductions and the need for economic relief during the coronavirus crisis. To quantify the undercount, we look at the 21.5 million workers who filed for unemployment benefits from March 22 to April 18. Our results suggest:

• An additional 7.8 to 12.2 million people could have filed for benefits had the process been easier.
• After accounting for these workers—who applied but could not get through or did not try because of the difficult process—about half of potential UI applicants are actually receiving benefits.

When we extrapolate our survey findings to the full five weeks of UI claims since March 15, we estimate that an additional 8.9–13.9 million people could have filed for benefits had the process been easier.

These findings on the millions of frustrated filers and the UI system’s low payment rate highlight the need for policies to improve rather than hinder the UI application process. At a minimum, states should presume everyone is eligible and immediately pay benefits, only verifying eligibility and reviewing claims after the unprecedented wave of claims slows down.

Quantifying reports of frustrated filers for unemployment

An estimated 24.4 million workers filed for unemployment benefits from March 15 through April 18. But while the CARES Act significantly expanded workers’ UI eligibility and benefits during the coronavirus crisis, widespread reports indicate that long-neglected state UI systems are unable to handle the volume of applications, preventing laid-off or furloughed workers from receiving necessary unemployment benefits. To gauge how well the UI system is handling the new caseloads, we used Google Surveys to ask 25,000 people, “Did you apply for unemployment benefits in the last 4 weeks?” (Google Surveys collects responses from visitors to news and other internet sites; see Google Survey’s description for more details and see the paper by Brynjolfsson et al. for an application of Google Survey data to measure the prevalence of remote work during the pandemic.) From April 14 to April 24, 24,607 respondents (98%) answered the survey.

To our knowledge, ours is the one of the first national surveys that asks about the social safety net and its performance in the COVID-19 era. In the process of writing this draft, we became aware of another survey, conducted by Navigator and Global Strategy Group April 17–22, that asked 1,022 registered voters about their experience and difficulty with accessing unemployment benefits. In addition, a survey conducted by Bick and Blanden April 22–23 found that many respondents have yet to apply for UI benefits.

Figure A shows the shares of people who responded each of the six answers to the question, “Did you apply for unemployment benefits in the last 4 weeks?” (respondents could select one of the six).

The 9.4% of survey respondents who said they “applied successfully” for benefits is reassuringly similar to what one would expect based on official UI claims data over a similar time frame. According to Google Surveys, the survey population is balanced to represent the age 18 and older internet-using population of the U.S., or about 254.7 million people according to the latest 2019 Census population estimates. The survey results therefore imply there were about 23.8 million UI claims applications. For comparison, total initial UI claims reported to the U.S. Department of Labor from March 22 to April 18 were only somewhat lower, at about 21.5 million. It makes sense that our survey-based estimate is slightly higher because of the backlog and delays of UI claim determination. Moreover, to the extent that we are overstating the rate of successful applications, we are necessarily understating the rate of unsuccessful attempts to apply for benefits.

The 3.4% share of respondents who tried to apply but could not get through the system represents a clear lower bound on those frustrated by the process and prevented from applying. (If states improve the processing of UI claims, some of these frustrated filers may eventually be able to apply and have their applications accepted.)

To put this number in perspective, if we conservatively equate the 9.4% of the respondents who applied successfully with the total 21.5 million initial UI claims filed over the four weeks ending April 18, then 7.8 million tried to apply and couldn’t get through. Similarly, if we add to this 7.8 million the 1.9% of respondents who did not apply because it was too difficult, a total of 12.2 million people would be UI claimants but were not.

All told, additional 7.8 to 12.2 million people could have applied for benefits had the process been easier. This estimate is conservative, because some of the 1.6% of respondents (3.7 million people) who tried to apply but were rejected may have had their applications rejected because states were unable to process applicants with expanded eligibility under the CARES Act, such as newly covered self-employed workers. When we extrapolate our survey findings to the full five weeks of UI claims since March 15, we estimate that 8.9–13.9 million people could have filed for benefits had the process been easier.

From other research on official UI claims data, we know that only 71% of UI applicants are receiving benefits, in that their initial claims have become continuing insured claims. But this calculation excludes those prevented from applying in the first place. Accounting for the 7.8 to 12.2 million people who could have applied, we estimate that the UI system’s payment rate could be as low as 45% to 52%.

In order to put the “frustrated filers” into perspective, it is helpful to compare them to the number of people who said they applied successfully. Figure B shows the results. For every 10 successful UI applicants, three to four more people tried but couldn’t get through, and two people did not apply because it was too difficult.

Technical details about the survey

We conducted a single-question survey targeted at 25,000 respondents through Google Surveys, beginning April 14. The survey was about 98% complete and had 24,607 respondents as of April 24; the results in this blog post use that data. Google Surveys attempts to make the population representative of the age 18 and older internet-using population, balancing demographics to match the national population sample from the U.S. Census Current Population Survey Computer and Internet Use Supplement; see here for more details.

Our single question survey asked “Did you apply for unemployment benefits in the last 4 weeks?” Respondents could select only one of six answers, which were presented in this order:

• I did not apply because I did not lose a job
• I did not apply because I am not eligible
• I did not apply because it was too difficult
• I tried but my application was rejected
• I tried but I could not get through
• I applied successfully

Using inferred demographics from respondents’ search histories, Google Surveys assigns sample weights to about three-quarters of the sample. In order to use the entire sample of respondents, we presented unweighted results. Table 1 shows the survey results are very similar with or without sample weights.

We thank Josh Bivens, Virginia Rutter, and John Schmitt for help in designing the survey instrument.