Unemployment insurance fraud during the pandemic cost states $135B—and counting

COVID-era unemployment insurance fraud is estimated to have cost state governments up to $135 billion, the Government Accountability Office reported last year.

While some states have moved to prevent future waste, fraud and abuse by modernizing their unemployment systems, often taking advantage of federal grants along the way, many state systems still lag behind and remain vulnerable. And although the pandemic and the government aid that accompanied it has ended, multiplying threats from bad actors make benefits fraud an ongoing issue for everyone.

“The pandemic lowered the guardrails,” said James Cotter, director of the Integrity Data Hub at the National Association of State Workforce Agencies, or NASWA. “What we've seen since then is an emergence in fictitious employer schemes, where somebody sets up a fake company, [gets laid off] and thereby collects against multiple [unemployment]  claims.”

To combat the latest schemes, Cotter said continued efforts to modernize state unemployment systems and enhanced information sharing are the two best ways to get the problem under control.

NASWA has tried to bolster information sharing between states, which often have disparate systems. Cotter said the association’s Integrity Data Hub, which allows states to submit unemployment claims to a database and then crossmatches them against other state data, has helped prevent around $4 billion in improper payments, and has conducted more than 64 million identity verifications.

Cotter, who made these remarks at an Amazon Web Services conference in Austin, Texas, this week, said it is “challenging and interesting” given how different each state’s workforce agencies are, even down to where they are housed within the government. 

The hub also offers states a support team that, among other things, tweaks agency websites and makes other fixes to encourage people to “do the right thing,” Cotter said.

States can send their claims to NASWA either daily or weekly. In addition to crossmatching claims, the hub also verifies IP addresses, home addresses, Social Security numbers and bank accounts to see if, for example, someone has tried filing in multiple states—a key indicator of fraudulent activity.

When states submit unemployment claims to the data hub, NASWA notifies them if cases are potentially fraudulent, but leaves it up to individual states to investigate further and prosecute, if necessary. Cotter estimated that 55% of claims submitted to the hub are returned to the states to investigate. The onslaught of referrals turned out to be “overwhelming” for some governments, partly due to a lack of staff to handle them. That prompted NASWA to add tools to help states prioritize cases by sorting and filtering them, making it easier for states to know where to “drill down.”

Meanwhile, progress modernizing unemployment systems varies by state. Some have embraced cloud technology and artificial intelligence. Others still rely on decades-old mainframe technology driven by outdated coding languages.

The North Carolina Division of Employment Security moved away from mainframe technology in 2018 for its benefits administration and then moved its tax system onto the cloud last year. Raju Gadiraju, the agency’s chief information officer, said the strategy has to be one of “continuous evolution” to keep up with advances in technology and fraud.

“It's not a one-time deal,” he said. “It's just not enough to say you’ve modernized it. In order to keep evolving, you can't just have hardware and add on to it, you’ve got to refresh yourself.”

North Carolina struggled during COVID with UI fraud, even with all this technology. The U.S. Department of Labor estimated that the state overpaid more than $200 million in unemployment insurance at a fraud rate of just over 4% in 2020, which was better than many states at that time but still higher than officials would have liked.

The year presented enormous challenges for a state employment agency that had until COVID-19 presided over an unemployment rate of just over 3%. That figure ballooned to over 14% once the pandemic hit, Gadiraju said, while the number of claims received jumped from a typical average of 3,000 per week to 300,000 in just the last three weeks of March 2020. One day, claims hit 54,000, a dramatic and unprecedented uptick.

The human cost was high, too. The agency usually has 600 full-time employees, but the high demand meant it needed to scale up to 2,500, with many borrowed from five other agencies. Those employees had to cope with 150,000 calls a day. Having a modernized system allowed the agency to scale up quickly and embrace strategies like self-service so people could check the status of their claims autonomously.

North Carolina also made use of its centralized Government Data Analytics Center to crossmatch claims and try to fight fraud, as well as tapping into NASWA’s national hub. “Why reinvent the wheel?” Gadiraju said, noting that the Division of Employment Security had considered procuring various data analytics pieces for itself before seeing that NASWA offered them to states for free.

Gadiraju said the state has since added features to improve security like multifactor authentication, and created various “fraud rules” that help analysts identify patterns in the data like home addresses and IP addresses not matching. The agency already has 102 fraud rules, and they are always “evolving,” he said, to keep up with how threats evolve.

“The system we built was a good start,” Gadiraju said.