The IRS claimed success in its fight against refund fraud based on a November statistic showing a decline in confirmed identity theft returns, but only after touting an increase in the same statistic in June -- a discrepancy Commissioner John Koskinen attempted to clarify November 15.
Koskinen acknowledged the potential appearance that the IRS tried to have it both ways, but emphasized that a new trend came into focus after the June estimate. "You have to look at the data, and it's a good -- perfectly good question," he said after addressing attendees of the American Institute of CPAs' Fall Tax Division meeting in Washington.
"At the front end of this year, it looked like we were stopping more" refund fraud, Koskinen said, but "as a result of the filters and the information and the ability to shut things down, suddenly those returns weren't coming through anymore at the same level."
On November 3 the IRS said that it had stopped fewer "confirmed identity theft returns" from January to September 2016 as compared with the same period in 2015 due to better pre-processing filters. In a release, the IRS reported a nearly 50 percent drop in fraudulent returns that made it into the IRS tax processing systems -- which showed, the agency said, that its anti-fraud security summit coalition's efforts "are working upfront in the tax process."
The IRS said that in the first nine months of 2016, it "stopped 787,000 confirmed identity theft returns, totaling more than $4 billion," compared with 1.2 million returns worth about $7.2 billion over the same period in 2015. That equated to 34.4 percent fewer returns worth some 44.4 percent less money.
However, in a June 28 release, the IRS reported that from January through April 2016, it caught $1.1 billion of fraudulent refunds claimed on more than 171,000 returns, compared with $754 million claimed on 141,000 returns in the same period in 2015 -- equating to 21.3 percent more returns worth 45.9 percent more money.
An IRS spokesperson confirmed with Tax Analysts November 10 that those estimates -- an increase in June, decrease in November -- applied to the same measure of refund fraud.
"The June statement was as of April, a snapshot of what we knew then," the spokesperson said in an email. "As we continued processing returns and did analysis, we were able to see that the IRS is preventing more fraud [and identity theft] and detecting fewer fraudulent refund claims. This broader analysis is reflected in the Nov. figures."
Asked to clarify why the IRS put a positive spin on divergent values of the same statistic, the spokesperson reiterated November 14 that the June estimate was based on what information the IRS had at that point in time. "The November announcement was based on a review of more complete data; we had a fuller picture," the spokesperson said.
Koskinen explained that "there's a lag time in the data," adding, "So when we are looking at the front end of the year, we're two or three months behind the data."
"One of the reasons we're hesitant to put out data too early is because it keeps coming in," Koskinen said. And when the reversal emerged, he said he questioned the data's consistency: "My concern was, 'Is this a flip?'" Some returns may have been missed because they were sidelined for further review, then later found to be legitimate, he noted.
The commissioner admitted that new data and differing trends in returns filed later in the year both probably played some part in the reversal of the metric at hand from positive to negative, but he expressed optimism that the IRS and its partners are detecting or blocking more fraudulent returns upfront, which is why the value of refunds found to be false declined.
"All of a sudden it's gotten much harder just to file, if you're a criminal," Koskinen said.
Follow Luca Gattoni-Celli (@TheGattoniCelli) on Twitter for real-time updates.