California’s Unemployment Backlog Grew to Over 1 Million Applications. Will Automation Speed Up Payments?

In this May 7, 2020 file photo, a person looks inside the closed doors of the Pasadena Community Job Center in Pasadena, Calif., during the coronavirus outbreak.

In this May 7, 2020 file photo, a person looks inside the closed doors of the Pasadena Community Job Center in Pasadena, Calif., during the coronavirus outbreak. AP Photo


Connecting state and local government leaders

State unemployment agencies have started automating identity verification in the hopes of more quickly processing record numbers of new applications. But experts say more fixes to flawed systems are also necessary.

From May to June, the unemployment applications of 1.3 million out-of-work people in California were set aside for state workers to review, an arduous process that delayed benefits for weeks or even longer. 

Now, California is among a growing number of states using software to automate more of the unemployment insurance application process, with the goal of more quickly getting payments to the people who need them. 

The automated identity verification tool is just one step the state has taken in recent weeks to try and get the problem under control.

A “strike team” appointed by Gov. Gavin Newsom last month released a 103-page report with recommendations for the Employment Development Department. Most drastically, the agency followed advice to shut down new applications for two weeks in September, allowing employees  to catch up on the growing backlog of cases that had been waiting for at least 21 days for resolution. 

But even with that reset, EDD has a long way to go. Agency staff reported that during the two-week break they were able to go through 246,000 cases out of a backlog that climbed to 1.6 million by early fall, although a recent report shows they continue to make progress. The agency earlier this month told frustrated state lawmakers that it will need until January to fully clear out the rest.

California isn’t alone in dealing with massive unemployment system problems that have resulted in people eligible for benefits just not getting them. As nationwide unemployment hit a high of nearly 15% in April and Congress created a new program to allow self-employed workers to receive assistance, agencies struggled to process applications and get checks to desperate families. 

In March, as the pandemic was just starting to affect unemployment numbers, nearly 95% of unemployment applicants nationwide received payments within two weeks of applying—the U.S. Labor Department’s benchmark for timeliness for states like California that require a waiting week before applications receive benefits—but by June, the rate was less than 45%. In September, the nationwide average was 53%. 

In California, the strike team report, written by Yolanda Richardson, the state’s government operations secretary, and Jennifer Pahlka, a former U.S. Deputy Chief Technology Officer under President Obama, found that manual reviews to confirm applicants’ identities created a bottleneck that slowed payments, without meaningfully preventing fraud. While about 60% of claims are approved automatically, the remaining claims that require manual reviews are the largest factor slowing payments well beyond federal standards for timeliness.

Of the applications that require manual review, 78% were due to a problem identifying the claimant’s identity, according to the report. Part of the issue in California and other states was just the volume of claims as the pandemic and associated shutdowns started to ripple through the economy. Nationally, initial claims for unemployment benefits rocketed from about 225,000 per week in March 2019 to 6.9 million in one week in March 2020. 

“What states need is automation to deal with their backlogs,” said Waldo Jaquith, who leads the unemployment insurance team at U.S. Digital Response, a group that provides free technology assistance to governments across the country. “States have received more applications than they possibly have the staff to deal with. They can’t even staff up to the appropriate levels.”

The California EDD contracted the vendor to automate its system this month. When applicants file for benefits, the system will allow them to upload identification documents, as well as a photo they take through the system. The idea is to both make for a quicker process and curtail fraud, the agency has said

After the company’s first day operating in the state on Oct. 6, the agency processed 64% of applications automatically. The September report found the previous average to be about 60% and said a vendor could help that number reach 91%. 

At a California Assembly hearing on Oct. 7, Democratic Assemblymember David Chiu said he remained concerned there is still a long way to go. But by Thursday, EDD reported that automatic verification was now processing 85% of claims. did not return a message seeking comment. On its website, the company advertises recent work for Arizona, Georgia, Nevada and Pennsylvania in addition to California. Jaquith said is one of two or three companies providing a similar service. 

Fraud Failures

Automation is not a panacea for all unemployment insurance problems.

Jennifer Kwart, a spokeswoman for Chiu, said in an interview that although Chiu has confidence in the authors of the report, the California EDD still needs to improve its outreach and accessibility. Non-native English speakers, disabled users and people who use only mobile devices for internet access have difficulty signing up for benefits, she said.  

The strike team report also identified a significant EDD staffing problem. While the state went on a hiring binge, these new employees lack the training and expertise to tackle the stacked-up cases. On top of this, the team found that experienced staffers were being tasked with training the new employees, which meant the agency had actually become less productive. 

Going forward, EDD has pledged to redirect experienced staff from training duties or answering phones, saying they would instead be focused on resolving past cases. 

California’s automation efforts with also won’t itself be enough to stop the kind of systemic fraud that has plagued unemployment systems across the country during the pandemic, the report says. 

In just one example, Arizona has estimated it paid out $500 million in fraudulent claims. Federal authorities also believe a criminal ring based out of Nigeria used stolen Social Security numbers and other pieces of information to submit fraudulent claims across the country. Washington state alone believes it made $576 million in payments to these scammers and have so far recovered $346 million, the New York Times reported. 

While automated systems seem to be at least as good as employees at catching such fraud, states are viewing them as tools to get through their backlogs. Improving fraud protection hasn’t “even been part of the calculus,” Jaquith said. 

Along with creating headaches for agency staff, fraud is a major problem for the people who depend on unemployment insurance and other state benefits, said Jessica Bartholow, a policy advocate at the Sacramento-based legal advocacy group Western Center on Law and Poverty.

When fraudsters gain access to a claimant’s identity to secure unemployment payments, it robs the person who is entitled to that aid and can even cut off other benefits like food stamps, Bartholow said. It’s also more likely to hurt disabled people, homeless people and others who are in the greatest need.

In California, EDD this month froze 350,000 debit cards containing benefits because of fraud concerns, including multiple applications at one address. For people with legitimate benefits caught up in the freeze, they will have to work out the problem with the bank or EDD, the state agency said

“I would qualify that as a serious problem,” Bartholow said. “This is not pretend. This is real fraud.”

But existing systems just don’t have much of a response to these types of schemes. State unemployment systems generally have strong methods of catching individuals who lie about their own situation to boost their benefits, but are less prepared to stop the wide scale fraud efforts that are much more costly to states. 

“The ‘gotcha’ approach to a lot of state UI systems is optimized to the mythical welfare queen and not to who is actually doing the stealing, which is organized crime rings stealing vast amounts of money,” Jaquith said.

Jacob Fischler is a state policy reporter based in Portland, Oregon.

NEXT STORY: NOAA Partners with Google to Boost Weather Forecasts with AI