How many people in your state escape state taxes on RVs by registering them in Montana? How many jailed criminals receive unemployment benefits? How many people claim unemployment benefits while working? How many recipients trade low-income food credit for cash to buy liquor, household items and cigarettes? How many Medicaid recipients drive high-end BMWs?
If you are a state, city or county jurisdiction, you probably can't answer those questions with any degree of accuracy. First, your budget is tight -- and answers require lots of investigative legwork. Second, many government services are now provided online, so verifying identity and confirming self-reported information can be difficult.
But what if all levels of government had a "big data view" of of their citizens -- a view that would inform the unemployment agency in one state that a resident purchased a vehicle in another state that doesn't charge sales tax, or let officials know when a Medicaid recipient purchases a Bentley, BMW or Mercedes?
Believe it or not, such a view exists. Governments and other entities can now get a better view and understanding of identity thieves, criminal networks and fraudsters who are taking them for billions -- something that used to be a lot more difficult.
"Imagine if you lost your job 20 years ago," said Andy Bucholz, director of Government Fraud Solutions for LexisNexis. "You'd have to go to the unemployment office, stand in line and present information. So if I stole [someone's] identity, I'd have to go to a lot of effort."
Someone there would verify who Bucholz is, after which that person would give him a check that he would have to take to the bank -- where a bank teller would also verify that Bucholz is who he says he is. "I'd have to go back and stand in line every week to get this money," he said, adding that while identity theft was possible with face-to-face transactions, it was much more difficult.
"Now, if you lose your job, you register online, and we'll pay you with a debit card, because you may not have a bank," Bucholz said. "That's great for fraudsters; they can sign up online and get paid with a debit card."
Virtual transactions are only a part of the problem, Bucholz said. "Your view of the citizens in your jurisdiction is actually quite narrow. So the question is, 'Have you authenticated who this person is that you're giving this [money] to?' Are we giving unemployment benefits to prisoners? Well, yes."
A check of inmates in neighboring counties or states against one's unemployment recipients can spot suspicious data, which can then be followed up with potential savings to the jurisdiction.
A resident of Massachusetts, for instance, can buy an expensive RV online and register it in Montana, which has no sales or use tax, and no vehicle inspection requirement. This saves the owner potentially tens of thousands of dollars, but it violates the law in the person's state of actual residence and deprives that state of legal revenues. And because the state of residence most likely lacks a "big data view" that includes Montana, it has no idea of the lost revenue.
Small jurisdictions have an even smaller view of residents. One small county in Indiana, for example, has a homestead tax exemption for a primary residence. LexisNexis looked at the county's data and found millions of dollars of homestead exemption fraud, which for a small jurisdiction could give a considerable boost to the budget.
But not all fraud is a good fit for big data -- governments must be motivated to find the fraud in the first place. Take abuse of food stamps, for instance. In the Supplemental Nutrition Assistance Program (SNAP), as it is now called, the federal government is giving out the money, Bucholz said, and the states aren't incentivized to handle enforcement, so there's not much traction.
When it comes to tax revenue, however, it's a different story. "Governors want that money," he said, adding that ownership of those funds makes it easy to determine whether such a solution will be successful, such as with housing, Medicare and Medicaid.
In that vein, LexisNexis Risk Solutions did a test of the large Northeastern states' Medicaid, said company data scientist Jo Prichard, in which the names of Medicaid recipients were checked against the types of vehicles owned.
"There were a whole bunch of Medicaid recipients who were claiming benefits and driving everything from Bentleys to Aston-Martins to high-end Mercedes-Benz vehicles," he said, adding that while it's easy to use big data to find people who are gaming the system, the really interesting part is analyzing collusion among groups of individuals committing fraud. "It's about building a lie detector, so if they're being less than truthful about one thing, what are the chances that they're being untruthful about other things?"
While the company has some 4 petabytes of data in its big-data arsenal, subscribing jurisdictions only need to set up an interface and provide specific data points related to the problem at hand.
Once agencies or jurisdictions look at what they want to solve, Bucholz said, it opens up the door.
"For instance, there's a lot of counties that have debts to be collected," he said, noting that locate-and-contact tools are useful to find people who moved and left no forwarding address, or are deceased.
"People may be underreporting their taxes or not filing, or all kinds of things," Bucholz said. "Could our problem be solved if we had a much more holistic view? That's when they should reach out, because the answer is most likely yes."
Image courtesy of iStockPhoto
NEW ON THE PODCAST