Fighting Against Unemployment Insurance Fraud
Apr 09, 2020
Unemployment insurance (UI) has been a key tool for government responses to economic downturns, since it was established during the Great Depression. However, despite being one of the country’s largest benefit programs, state unemployment offices struggle with fraud and waste. Billions of dollars are lost through erroneous payments, both to malicious fraud and unintentional individual error. With most unemployment insurance claims being handled and processed by states, many have turned to analytics to catch fraud and reduce incorrect payments.
There are a multitude of reasons people submit incorrect unemployment claims. Some are fraudsters looking to take advantage of how long it can take state unemployment offices to validate a claim. Between underfunded offices, outdated technology, and lag time for employer submitted information to compare claims against, fraudsters can operate without detection for a while. A fraud ring based out of Cook Country, Illinois submitted over 900 fraudulent unemployment claims, stealing more than $8.7 million from UI departments in Illinois, Indiana, and Minnesota before they were caught. Fraudsters like these have been known to use stolen identities, some of them from dead people or prisoners, to apply for unemployment insurance. Often, the fraud isn’t caught until the person whose identity has been stolen tries to apply for unemployment and finds they are ineligible. Even when the fraud is caught by the state, it can be months after the fact, by which point the money is usually unrecoverable. However, not all incorrect claims are due to fraud. Another major source of improper payments comes from individuals misreporting their employment information on their claims. Whether knowingly or by accident, such misreporting can lead to individuals getting unemployment payments they are not entitled to. Advanced analytics can help unemployment insurance offices catch intentional fraud and reduce overpayments due to incorrectly filled out forms.
Unemployment insurance offices have a lot of data at their disposal, and analytics can help them harness it. Predictive modeling can alert offices to red flags on claims that may need closer scrutiny. Utah created a model to analyze claims and look for patterns that could be indicators of fraud, such as those originating from suspicious locations or groups of claims that have very similar information in certain fields. Utah’s experience allows them to identify and stop fraud. Analytics can also nudge people to fill out their forms accurately. New Mexico found that most of their overpayments were due to individuals with incorrectly filed out forms. New Mexico built a predictive model that could identify individuals that were at risk of receiving improper payments. They combined that approach with reminders and pop-ups to nudge individuals to accurately report their data, doing research into which messages most reduced fraud. They found that targeted pop-up messages reduced individual fraud by 40 percent. However, implementing analytics needs to be done with care. A poorly implemented model can have unforeseen repercussions. In Michigan, an unemployment model’s high false positive rate, which was as high as 93%, lead to many innocent people being accused of fraud and lawsuits against the state.
Analytics is a valuable tool for any unemployment insurance office. If implemented thoughtfully and thoroughly tested, as in Utah and New Mexico, predicative models can keep states from paying out millions of dollars to incorrect claims and catch fraud in its tracks.