Indiana began a big crackdown on identity crooks this year and the results are startling: The state has saved $85 million so far by not paying out bogus tax refunds.
The savings came from using research firm LexisNexis' giant database to verify the identities on state income tax returns. By spotting false or stolen identities, the state can determine which returns are concocted and block the refunds.
The Indiana Department of Revenue's identity-matching effort is indicative of the types of data-driven programs most states have undertaken to combat an exploding number of sham tax refund filings, false Medicaid and unemployment claims and public assistance fraud that can cost governments billions of dollars.
Indiana's results are proof that using data has a payoff. And they provide a tantalizing glimpse of the cost-savings states could get from applying a government-wide "big data" approach to combatting the fraudulent claims states face in an Internet age when identity theft is rampant.
Indiana spotted 74,782 returns filed with stolen or manufactured identities as of the end of last month with its new identity-matching effort. Without it, the Department of Revenue caught just 1,500 cases of identity theft out of more than 3 million returns filed in 2013.
Other Indiana agencies are mining data, matching data sets and performing some data analysis to detect fraud, said Matthew Donahue, the Revenue Department's leader of strategic transformation initiatives. But, he said, there's no state government-wide data sharing, combination of databases or use of data analysis across them to attack fraud in a big-data way.
That's true in nearly all the states, said Doug Robinson, executive director of the National Association of State Chief Information Officers.
Despite the vast databases of personal, business and other information that states possess, Robinson said, only a few states are starting to leverage them in a strategic, big-data way to detect and prevent fraud, or to apply them to other pressing management, economic or social problems states face.
Instead, he said, states largely house their databases separately in tax, health, revenue, welfare, motor vehicles, voter registration and professional licensing departments and offices. Other agencies cannot easily access the data, which is often not in the same format. Thus, it cannot be quickly managed, matched, mined or otherwise analyzed in an enterprising fashion to catch and prevent fraud or attack other problems.
"States are data engines at their core," Robinson said. But, he said, "just because states have a lot of data doesn't mean they've met the definition of big data. ... Very few states have taken an enterprise approach to it."
Like Indiana, many other states increasingly have taken action against scams and other improper payments by using identity verification and other data tools, especially in claims for unemployment insurance benefits.
By matching unemployment claims against the identities of people jailed in country prisons, Pennsylvania discovered and stopped almost 19,000 payments made in 2013 to inmates who can't lawfully collect the benefits. The state estimates the program will save about $100 million a year.
New Mexico cut unemployment insurance fraud by about $10 million, or 60 percent, last year after integrating its unemployment tax and claims systems. The new system lets the state quickly match documentation between employers and claimants to reduce fraud, and to detect and recover improper payments.
New Jersey has blocked nearly $450 million in improper unemployment payments the past four years by identifying 300,000 people who tried to wrongfully collect benefits through identity theft or other schemes or mistakes.
Florida has taken an aggressive approach to rooting out fraudulent or improper claims for food stamps, cash assistance for needy families and Medicaid by verifying claimants are who they say they are with the LexisNexis identity database before issuing payments.
The state kept $32.8 million in inappropriate payments "from going out the door" in its last fiscal year that ended June 30, said Andrew McClenahan, director of public benefits integrity for Florida's Department of Children and Families. It also detected $34.4 million in overpayments and recovered more than $19.4 million of that money.
New York's Department of Taxation and Finance has expanded its program to prevent and track down tax cheats in recent years. It now uses data screening and analytics to review the 10 million personal income tax returns it receives each year. Last year, it spotted more than 255,000 inappropriate refunds and kept $413 million from being paid.
Some states are taking the experience they've gained in one agency and applying it to others. After launching a program to recover improper unemployment insurance payments, Michigan began using data analytics software to look for anomalies or patterns that suggest fraud in other areas, such as its Food Assistance Program.
To integrate its data and leverage the information across all of state government, North Carolina has consolidated its data analytics efforts, including its data-based fraud detection program, under its Office of Information Technology Services. The goal is to apply big data-style analytics across all agencies to improve business decisions.
In trying to take a big-data approach to problems, states face hurdles beyond getting disparate data out of agency silos and making it compatible. Some data, such as voter registrations, reside in separate state constitutional offices, which poses a jurisdictional problem. And there are privacy concerns.
Most states use third-party software or services offered by private companies to analyze and apply their data to a task, just as Indiana's Department of Revenue uses LexisNexis to authenticate identities.
Indiana doesn't send LexisNexis tax returns. Instead, it sends portions of information from returns that LexisNexis compares to its identity database, which is pulled from public and commercial records and is one of the largest in the world.
The state also sends questions to taxpayers that only real tax filers can answer.