ID verification, analytics can help agencies fight fraud

Fraudsters have been running schemes on government programs essentially since those programs were first created. In a sense, if there is money to be had, criminals will try to get their hands on it -- many of them going to great lengths to do so.

That said, it should come as no surprise that COVID-19 created an environment especially ripe for fraudulent activity. When the pandemic hit in early 2020, government unemployment offices were flooded with both legitimate requests as well as hits from scammers looking to take advantage of the system and the chaos caused by the flood of claims.

In fact, in December 2020, nearly 68% of the country’s entire workforce had submitted applications for unemployment insurance; however, the actual number of unemployed individuals was just under 10%, according to the Bureau of Labor Statistics. At least five states had more UI claims than the entire pool of civilian workers residing in that state in early 2020, and the U.S. Department of Labor reported that at least 10% of the $872 billion in pandemic unemployment benefits (as of Sept. 30, 2021) was paid improperly, likely the result of fraud.

Access to new technology like bots and artificial intelligence has given criminals, both those acting individually and larger organized crime syndicates, the power to submit fraudulent benefit applications on a tremendous scale.

First, fraudsters either buy stolen IDs, many of which are purchased from the dark web or create synthetic (or “Frankenstein”) IDs by combining various bits of identity data from different sources. Then, they employ bots to completely inundate government systems and slip in fraudulent applications, which often go unnoticed among the flood of legitimate ones.

Many organized crime rings use underpaid workers from abroad to input data into application portals for UI and other related programs. They solve captchas or find email addresses and match ZIP codes to keep the scheme’s wheels moving forward.

As government attempts to limit criminal activity, many agencies are working to deploy technology solutions that allow them to capture anomalies and detect fraud in programs like UI, Medicare/Medicaid and even the Supplemental Nutrition Assistance Program.

Automated identity verification

With nearly 30% of the fraudulent UI claims in larger states based on stolen Social Security numbers, it’s much more difficult for government agencies to catch anomalies. Implementing an automated identity verification (AIV) system can be a lifesaver for agency IT teams that are understaffed and overworked for several reasons:

• Improved processing time - By automating ID verification, government agencies can quickly process more applications. Using real-time credit data can help eliminate fraudulent claims before they get into the system. Faster processing also contributes to a higher user satisfaction rate among legitimate applicants who experience more efficient turnaround.
• Reduced human error - AIV eliminates the potential for human error common when staff are feeling the stress of doing more with less. Even with well-trained and experienced employees in place, errors, omissions and misunderstandings can let fraudulent claims pass.
• Less expensive than deploying new workers - The growing demand for qualified IT professionals makes these positions very competitive and often cost-prohibitive for agencies on a set annual budget.
• Scalability – Even government IT shops that can find, hire and train qualified new employees must still deal with seasonal (end of quarter, end of year) or event-based (disaster, pandemic) scaling challenges that test their normal day-to-day workload. AIV can provide flexibility during times of peak demand. 

When automated ID verification is implemented correctly, government entities can gain insight and clarity into which claims require investigation, saving precious time and resources often spent looking into perfectly legitimate claims.

Data analytics for fraud detection and prevention

Banks, hospitals, educational institutions and manufacturing firms have been using data analytics, artificial intelligence and machine learning to aid in fraud detection in for several years now. Both internal IT and outside contractors have found it to be a valuable analytical tool for detecting fraud, monitoring transactions and ensuring compliance for both employees and clients. 

A 2020 report commissioned by researchers at the Administrative Conference of the United States found that federal agencies were closing the gap and that 45% of the 142 agencies surveyed were also using AI and/or machine learning to assist in fraud analysis in two key areas:

1. Using data analytics to detect and diagnose fraud after the fact.

Data analytics can help supplement IT and financial auditing teams and improve the overall efficiency and effectiveness of their post-mortem audits. Analytics make it possible to quickly and efficiently compare the data from disparate systems, more confidently identifying anomalies between them. 

These data mining and data matching techniques can produce reports that detect potential false, inflate or duplicate invoices or payments and can identify fraud or improper payments that have already been awarded. This allows agencies to identify bad actors, recover funds more quickly and identify and eliminate fraud, waste and abuse moving forward.

2. Implementing behavioral analytics to prevent fraud.

As important as fraud detection, prosecution and recovery are, using behavioral analytics to help prevent fraudulent activity by verifying identity before a claim is ever paid out is the real opportunity. The U.S. Department of Labor reported that in 2020, one individual used a single Social Security number to file for unemployment in 40 different states -- 29 of which paid up. Deploying preventative behavioral analytic technology can detect these types of anomalies and discrepancies before fraudsters get their money.

Cross-matching or AI-based cluster and pattern recognition algorithms can compare well-established facts and behavior patterns from every aspect of a typical transaction and then compare them to what is being actually being submitted. For example, if it typically takes users about six to eight minutes to fill out an online UI application, then applications that come in under a minute should be flagged as suspicious and investigated as they are likely being produced by a bot.

Manual verification tends to focus on the validity/veracity of the data being submitted and may overlook other anomalies. Because analytics monitoring software can be run 24/7, it can help detect when an account has been compromised. Identifying an unusual set of login patterns -- several times a day, from a strange IP, or at times that wouldn’t normally correspond to the user’s time-zone -- can send up red flags. 

As far back as 2018, the Government Accountability Office and industry groups such as the Association of Certified Fraud Examiners issued a report that risk-assessment programs using proactive data analytics had reduced fraud losses by more than 50%.

The time to upgrade fraud systems is now

As we head into 2022, governments must implement modern technologies that fight fraud. Agencies are just coming back to full strength and capacity after adjusting to the demands of the pandemic, and expectations have never been higher. Combining an automated verification system and data analytics for prevention and detection help create a powerful tool for today’s government IT professionals who are experiencing a sea-change in volume and manner in which fraud is being perpetrated.

Criminals will always attempt to steal from well-funded government programs, but this is a fight that we can’t afford to lose. Implementing these technologies can make deeper and longer-lasting changes in the fraud landscape overall.  Armed with hard data to illustrate potential risks to impacted benefits programs, IT-savvy government leaders can provide documentation to request additional resources, funding and even policy changes.