Tax season is upon us. While individuals might see tidy personal returns, identifying tax fraud can be a multimillion-dollar saving endeavor for governments.
In 2013 alone, the IRS reported paying out more than $5 billion in tax returns that ended up being fraudulent, StateTech reports. But the IRS is turning things around with technology, introducing new data elements to system filters in its 2016 season that caught $4.1 billion of identity-related tax fraud, according to a September report by the Treasury Inspector General for Tax Administration.
States are also seeing success with new technologies and practices. For example, the Utah Tax Commission pinpointed an attempted $11 million in fraudulent income tax returns last year, but reported that the state lost less than $20,000 as a result of modified practices and new software that analyzes returns to spot fraud, the Salt Lake Tribune reported.
Better data analytics can help beef up fraudulent findings for local governments and help out otherwise overwhelmed staff.
Charleston County, S.C., for example, was able to identify and prevent over 600 cases of erroneous approvals, preventing $2.1 million in losses, by instituting LexisNexis technologies that made use of identity analytics. The technology cross-checks customer records against more than 20,000 public records and commercial data sources.
“The revenue lost from erroneous or fraudulent legal residence filings hits counties where it really hurts — eliminating funds for key programs, such as education and public safety,” said Haywood Talcove, CEO of government and LexisNexis Risk Solutions in a press release. “By using identity analytics to detect anomalies beyond its jurisdiction, Charleston County is truly serving as a steward of taxpayer dollars and ensuring everyone pays their fair share.”
Maryland is also calling on data analytics to fight fraud and is seeing staggering results. Andrew Schaufele, director of Maryland’s Bureau of Revenue Estimates, revealed to StateScoop that the state introduced a new set of algorithms that proved successful in identifying 65 to 70 percent of fraudulent returns last year — a big increase from the 55 percent success rate in 2015.
The success was largely due to shifting from an algorithm that proved too far-reaching and overwhelmed fraud analysts to a more narrow and refined model that was able to better zero in on instances of fraud.
“As we began to learn about the fraud, we found consistencies amongst the fraud and developed a regression,” Schaufele told StateScoop. “We were scoring returns, finding a probability that it was fraud, and we set a threshold in our system, and started flagging returns.”
In California, the Franchise Tax Board is moving from analog to digital, embracing online filing and turning to digital storage methods that allow the state to sort through its millions of returns more efficiently, Government Technology reports.
While the FTB still deals with 3 million paper returns a year, the organization is able to scan and connect each return with a digital record and clean up its data stores. It can also overlay analytics on the tax records, creating a more streamlined auditing and fraud-spotting process.
“If you have 18 million tax returns, and one of the things we do is select people for audits, having good models to pick the best ones, that’s all about having the right data and using data really smart," CIO Cathy Cleek told Government Technology. "EDR [Enterprise Data to Revenue] gave us better models so we could pick better cases to audit, find more fraud, find better collection cases."