Unemployment Insurance Utilization Rates Differ Substantially Across States


Weekly releases of unemployment insurance (UI) claims data by the US Department of Labor (DOL) have indicated a very high utilization of state and federal UI programs during the pandemic. If these UI claims reports are accurate, UI benefit recipients comprised more than 150% of unemployed workers and 16% of the labor force between late March and the beginning of October. The unprecedented high ratio of UI benefit recipients to unemployed workers raises questions about the accuracy of the UI claims data.1  

The high ratio of UI recipients to unemployed workers may be because:

(i) UI recipients in the DOL data may be poorly measured and potentially double counted, 

(ii) the official measure of unemployed workers may substantially understate the number of jobless or underemployed workers eligible for UI benefits, or 

(iii) there may be considerable fraud in the UI system. 

Our analysis also finds that reliance on state and federal UI programs varies widely across states. For example, since late March UI recipients in Pennsylvania have comprised about 250% of the state’s unemployed workers and about 30% of the state’s labor force. In contrast, since late March UI recipients in South Dakota have comprised less than 58% of the state’s unemployed workers and about 4% of the state’s labor force.


Three Unemployment Insurance Programs Account for Most of the Benefit Recipients

The three primary unemployment insurance programs for which jobless Americans are eligible are outlined below: 
  1. Regular State Unemployment Insurance: This is the standard unemployment program in the United States administered by each state, where eligible workers who become unemployed through a job loss can collect unemployment benefits.2
  2. Pandemic Unemployment Assistance (PUA): This new emergency unemployment program was enacted to mitigate the impact of widespread unemployment caused by COVID-19 lockdowns. This program has expanded the types of individuals eligible for unemployment benefits to include business owners, self-employed workers, and independent contractors.3 
  3. Pandemic Emergency Unemployment Compensation (PEUC): This new emergency unemployment program has extended the time period for eligibility for unemployment benefits. Typically, state UI benefits are limited to no more than 26 weeks of benefits, but this program provides the opportunity for eligible individuals to collect an additional 13 weeks of benefits.


Weekly Unemployment Insurance Continued Claims in a Given Week Likely Include Benefits Paid in Prior Weeks

Between the weeks of March 28th and October 10th there has been substantial variation in the use of each of these UI programs across weeks and between states. Each week the DOL reports the number of new claims and continued claims for each UI program. In principle, UI continued claims in a week should equal the number of UI recipients during the week. However, because of the unprecedented surge in claims for the new PUA and PEUC programs, weekly continued claims figures in some states likely include backlogs of weekly recipients for these programs. In other words, continued claims may include the number of benefit weeks paid to jobless workers in prior weeks as well as the current week.

For example, the DOL reported that there were 6.98 million continued PUA claims during the week of August 22 in California. This reliance on the PUA program in a single week would be remarkable because the Bureau of Labor Statistics (BLS) reported that at the time there were only 2.17 million unemployed workers in California’s labor force of 18.7 million workers. It has been reported that the 6.98 continued claims include benefit recipients for prior weeks, in addition to recipients in the current week, due to backlogs in the system. In the October 1st, 2020 weekly release of unemployment insurance claim, the DOL included a technical note to address California’s backlog issue. The note stated that California will be instituting a two week pause on the processing of initial claims to reduce its backlog and implement fraud prevention measures.5 The inclusion of recipient backlogs in weekly continued claims figures creates a challenge for statisticians to accurately measure weekly variation in the utilization of the PUA and PEUC programs.6

Measuring Unemployment Insurance Utilization
If continued UI claims include benefits paid for previous weeks due to backlogs in the system, but are not double counted across adjacent weeks, it is still possible to calculate the average UI utilization rate during the pandemic. We use two measures of utilization: UI benefit recipients relative to the number of unemployed workers and relative to the overall labor force. We sum all UI continuing claims between late March and the beginning of October to measure the total number of person-weeks of benefits paid since the CARES Act was passed and the PUA and PEUC programs were established. We sum all person-weeks of unemployment and person-weeks of labor force participation since late March as benchmarks for the UI benefit recipient totals. We calculate these pandemic-wide utilization rates for each state and for the U.S. overall. These UI utilization rates are effectively weighted averages of weekly UI utilization rates. We do not estimate UI utilization rates by week as they would be imprecise due to the possibility that “weekly” continued claims include benefits paid for prior weeks due to system backlogs.
Figure 1 presents UI utilization rates measure relative to the number of unemployed workers. Nationally the three primary UI programs have paid out about 1.5 times as many person-weeks of benefits than person-weeks of unemployment since late March. Figure 1 also shows substantial variation across states and indicates that states such as Michigan, Montana and Pennsylvania have paid more than 2.1 times as many person-weeks of UI benefits than person-weeks of unemployment since late March. In contrast states such as Florida, South Dakota, Utah and Wyoming have paid UI benefits equivalent to less than 73% of the number of person-weeks of unemployment since late March.


Figure 1

Figure 2 presents utilization rates of the UI program relative to the size of the labor force. Nationally the three primary UI programs have paid out the person-weeks of benefits equivalent to 16% of the person-weeks of labor force participation since late March. Figure 2 also shows substantial variation across states and indicates that states such as Michigan and Pennsylvania have paid benefit weeks equivalent to over 28% of each state’s person-weeks of labor force participation. In contrast states such as Idaho, South Dakota, Utah and Wyoming have paid UI benefits equivalent to only about 5% of the person-weeks of each state’s person-weeks labor force participation since late March.

Figure 2

 Reasons for Wide Swings in Utilization Across Weeks and Between States
Differences in utilization rates may be the result of more than just variation in employment conditions by state. Recently, there have been investigations into fraud and public statements regarding backlogs in processing UI claims, offering skepticism of the accuracy of the figures presented and how to accurately interpret them.

Unemployment insurance backlogs depend on how long it takes for states to process claims. While some states were easily able to process claims and payout benefits immediately, other states have taken more time to pay out claims. A state that processes claims efficiently and pays out all benefits for the current week will report continued claims that accurately reflect the actual number of people requesting benefits each week. In contrast, a state that slowly processes claims will report continued increases in claims for several weeks in a row. This is because continued claims include claims for the current week as well as for the weeks those benefits remained unpaid on backlog.

As noted above, it was suspected that some of California’s recent spike in PUA claims was capturing this backlog.7 One week after the number of continued claims for California peaked at almost 7 million, continued claims dropped from about 6.4 million to about 3.4 million (almost 50%!) in a single week. Large decreases in PUA claims in a single week is consistent with the idea that weekly counts include numerous instances in which individuals are collecting unemployment benefits for multiple prior weeks of unpaid benefits in a single week. These additional headlines offer a glimpse into how backlogs might be contributing to low utilization rates:

– Kentucky: “Kentucky has made little progress on unemployment backlog, figures show”8

– Nevada: “Scope of Nevada’s jobless claims backlog still unclear”9

These weekly spikes are distinct from a second key potential cause of the high utilization rates reported here: fraud. There has been substantial evidence that millions, if not billions of dollars of benefits have been claimed fraudulently. A surge in fraudulent claims would dramatically increase the number of continued claims and provide an additional explanation for why cases have spiked in certain states. Workers who previously weren’t eligible for unemployment benefits (such as gig workers, self-employed workers, etc.) can apply for PUA and may have an easier time committing fraud because it is more difficult for a prior company to verify their unemployment status. The headlines below offer a glimpse into why fraud might be affecting the utilization rates between states:
– Pennsylvania: “New PUA payments paused in Pa. due to uptick in suspicious claims”10
– Hawaii: “PUA fraud in Hawaii up to $92 mil in claims, 6,000 victims so far”11
– Michigan: “400K unemployment claims now flagged in Michigan fraud investigation” 12
– California: “Fraud likely driving suspicious spike in unemployment claims”13
In addition to fraud and backlogs, there have been reporting discrepancies for the PUA claims. For example, Florida has listed zero continued claims for its PUA program in all data releases. It is unclear why this is but is further demonstration of the confusing figures we are seeing regarding unemployment claims.


We find that during the pandemic UI benefit person-weeks paid have amounted to more than 150% of the person-weeks of unemployment over the same time period. In some states, UI benefit person-weeks have been more than twice as high as the number of person-weeks of unemployment since late March. While this high utilization of UI programs may be due, in part, to an underestimate of the number of jobless and underemployed workers eligible for UI benefits, the most likely explanations are that: (i) continued claims in the major UI benefit programs may be reported with considerable error and may be double or triple counting benefits across weeks, and (ii) there is considerable fraud in UI benefit programs, especially in certain states with exceptionally high apparent utilization.


  1. In a recent Brookings Institution publication Gary Burtless noted that while the share of unemployed workers collecting UI benefits rises during a recession, even during the Great Recession the share rarely exceeded 70% and in earlier recessions rarely exceeded 60%. In contrast, during the pandemic we see implausibly high shares of unemployed workers collecting UI benefits. (See https://www.brookings.edu/wp-content/uploads/2020/03/Burtless_UI_updated_20200402.pdf).
  2. https://www.dol.gov/general/topic/unemployment-insurance 
  3. https://edd.ca.gov/About_EDD/coronavirus-2019/cares-act.htm 
  4. https://edd.ca.gov/About_EDD/coronavirus-2019/cares-act.htm 
  5. https://www.dol.gov/ui/data.pdf. See also, https://www.edd.ca.gov/About_EDD/pdf/news-20-49.pdf
  6. Backlogs have been reported in other states including Pennsylvania, Oregon, Wisconsin, and Nevada. https://www.inquirer.com/business/pennsylvania-coronavirus-unemployment-claims-backlog-90000-20200723.html; https://www.koin.com/news/health/coronavirus/oregon-employment-department-completes-pua-backlog/.
  7. https://www.capolicylab.org/publications/september-15th-analysis-of-unemployment-insurance-claims-in-california-during-the-covid-19-pandemic/ 
  8. https://www.wdrb.com/in-depth/kentucky-has-made-little-progress-on-unemployment-backlog-figures-show/article_b1d553a4-e964-11ea-abf1-1f3f6e897005.html 
  9. https://www.reviewjournal.com/business/scope-of-nevadas-jobless-claims-backlog-still-unclear-2121966/ 
  10. https://www.phillytrib.com/news/local_news/new-pua-payments-paused-in-pa-due-to-uptick-in-suspicious-claims/article_39620efc-98ad-5378-a339-1dba11194746.html 
  11. https://www.khon2.com/always-investigating/pua-fraud-in-hawaii-up-to-92-mil-in-claims-6000-victims-so-far/ 
  12. https://www.bridgemi.com/business-watch/400k-unemployment-claims-now-flagged-michigan-fraud-investigation
  13. https://www.politico.com/states/california/story/2020/09/10/california-fraud-likely-driving-suspicious-spike-in-unemployment-claims-1316454


Stephen G. Bronars, PhD


Related Capabilities

Share on facebook
Share on linkedin
Share on twitter


Stephen G. Bronars, PhD

Subscribe to our Newsletter

This website uses cookies to improve functionality and performance. By continuing to use this website, you agree to the use of cookies in accordance with our Privacy Policy.