As Congress reviews the latest proposals to assist the unemployed and help the economy recover, the most recent round of unemployment benefit claims data continues to offer policymakers confusing signals, to put it mildly.
As I reviewed previously, the “Pandemic Unemployment Assistance (PUA) gap” suggests the Department of Labor (DOL) count of “persons” claiming unemployment benefits in all programs has been significantly overstated because it includes what the DOL report calls “backdated claims.” In the week ending July 4, the PUA gap appeared to shrink to about 2.6 million, or apparently half the level it was just two weeks ago. But as I describe below, current data appear to be missing over 2.6 million “continuing claims” in Arizona, which, if counted, would keep the PUA gap near its recent levels.
Digging deeper into PUA data for individual states reveals even greater cause for confusion. Owing to the same “backdated claims” issue, several state reports suggest implausible levels of PUA continuing claims. For example, the DOL report released yesterdaysuggests that in the week ending July 4, Pennsylvania had 3.4 million continuing claims for PUA — or more than four times the 815,000 continuing claims under the state’s UI program. If that 3.4 million figure actually reflected current PUA recipients, it would mean that over half of the 6.4 million Pennsylvanians in the civilian labor force in June 2020 were currently collecting benefits from the PUA program. Clearly that’s not the case. That massive figure also contributes greatly to the PUA gap nationwide, as Pennsylvanians appear to constitute 26 percent of the almost 13.2 million “persons” currently reported as being on PUA nationwide. For comparison, the 815,000 Pennsylvanians on the state UI program constitute less than 5 percent of the national UI caseload.
Then there’s Arizona. For the week ending June 27, Arizona reported nearly 2.3 million continuing claims for PUA. If each were really a “person,” that would mean nearly two-thirds of the 3.5 million people in Arizona’s civilian labor force were collecting PUA benefits then. Again, not so. But then this week’s DOL report suggests that in the week ending July 4 — the very next week — exactly zero people were collecting PUA benefits in Arizona. That 2.3 million swing in Arizona more than explains the apparent 1.1 million decline in the national PUA continuing claims “caseload” last week. Meanwhile, data publicly available on the website of the Arizona Department of Employment Security reports there were actually more than 2.6 million “continuing claims” in Arizona that week, noting that such claims “represent weekly filings and are not unique individuals.” The absence of this data means national “continuing claims” didn’t fall last week by 1.1 million as DOL reported, but likely rose by at least that amount if Arizona’s true figures had been included.
Factors other than data definitions and reporting issues are also contributing to both elevated PUA claims and large swings from one week to the next. For example, last week Maryland Governor Larry Hogan reported the state “uncovered a massive criminal enterprise involving over 47,500 fraudulent unemployment insurance claims totaling more than $501 million.” The attack was conducted by a “massive and sophisticated criminal enterprise” targeting the PUA program. Maryland’s Labor Secretary noted that the design of the PUA program leaves it open for fraud: “The PUA Program, in particular, allows individuals to self-certify that they are unemployed due to the coronavirus, eliminating the regular check-and-balance that exists under the regular state UA program, increasing the potential for fraud.” In recent weeks the number of PUA initial claims in Maryland has been cut in half, driven by sharp declines in claims from individuals outside the state.
The bottom line is PUA data — and thus national unemployment benefit data that include PUA figures — need to be handled with great care. First, PUA data simply do not reflect the “caseload” that “continuing claims” normally mean. And second, the data are subject to large and — in the case of Arizona last week — completely illogical swings from one week to the next, rendering national figures even more suspect. Rising numbers of fraudulent claims — which inflate figures until uncovered, and then cause sharp declines — add still more confusion. That complicates policymakers’ efforts to understand both the number of people on benefits as well as the direction in which caseloads are trending, both key factors in determining prudent next steps to continue to assist large numbers in need.