As we approach the filing season, I thought it would be a good idea to discuss an issue that affects many taxpayer returns, namely the IRS processes for identifying and stopping refund fraud. Attempted refund fraud has become a significant problem in our tax system. According to the most recent figures available, in calendar year (CY) 2016, identity theft (IDT), refund fraud alone, cost the government roughly $1.7 billion. I fully support the IRS’s efforts to reduce refund fraud and protect revenue. However, I have expressed concern over several years that the refund fraud false positive rate (FPR) is too high and that the IRS takes far too long to process legitimate taxpayers’ returns once it has determined that they have been inaccurately selected. For some taxpayers who rely on their tax refund to pay for necessary living expenses, their anxiety increases every day that their refund is delayed.
One of the main drivers behind these issues is the timing between when the IRS selects returns to be analyzed for possible refund fraud, and when it receives payor information that would either verify or disprove this possibility. But before we get into specific concerns surrounding the IRS’s fraud detection program, here is a little background on how the systems that select possible fraudulent returns work.
The IRS’s Return Integrity Operations office (RIO), which is housed in the Wage and Investment Division (W&I), is tasked with detecting and preventing both IDT and non-IDT refund fraud. RIO oversees both the Taxpayer Protection Program (TPP) and the pre-refund wage verification program. TPP is the program responsible for detecting and preventing IDT fraud, and the pre-refund wage verification program is responsible for detecting and preventing non-IDT refund fraud, such as creating fake or altering W-2s with inflated wages or withholding to receive larger credits or refunds.
When RIO selects a taxpayer’s return into the TPP, the return will be suspended until the taxpayer contacts the IRS and authenticates his or her identity. When a taxpayer’s return is selected into the pre-refund wage verification program, the income and withheld tax reported on the return will be matched against third-party data provided by employers and other payors (e.g., Forms W-2 and 1099-MISC-Nonemployee Compensation). Form W-2 information must be submitted to the Social Security Administration (SSA) by January 31, and the SSA then forwards the information on to the IRS. 1099-MISC-Nonemployee Compensation forms must be submitted directly to the IRS by January 31.
The IRS uses two systems for the detection of refund fraud: the Dependent Database (DDb), which is used to detect only IDT, and the Return Review Program (RRP), which is used to detect both IDT and non-IDT refund fraud. The DDb contains filters which are comprised of rules and are binary in nature (i.e., if the rule is broken, the return will be selected for further analysis; if the rule is not broken, the return will continue through normal processing). RRP’s filters are comprised of both rules and models. In the most recent filing season (2018), significantly more taxpayers were selected into the IRS’s pre-refund wage verification program as a result of the IRS adding two filters to the RRP system for non-IDT refund fraud. Specifically, non-IDT refund fraud filters selected in excess of one million returns from January 1 through October 17, 2018, a roughly 400 percent increase when compared to the same time period last year.
The addition of these filters is partially responsible for the high false positive rate (FPR) for non-IDT refund fraud in particular. For January 2018 through October 17, 2018, non-IDT refund fraud had an FPR of about 81 percent, while IDT refund fraud had an FPR of about 63 percent.
Just as concerning is the delay that taxpayers who filed legitimate returns experience in receiving their refunds. The IRS determined that sixty-three percent of returns selected into the non-IDT refund fraud program in 2018 were legitimate even though more than two weeks elapsed from the time of selection until the IRS released the refund – this is in addition to a two-week screening time prior to selection. The IRS refers to this 63 percent figure as the “operational performance rate” (OPR). Unfortunately, a similar measure for IDT refund fraud is currently not tracked. This is because the IRS relies on taxpayers taking action to authenticate their identity (i.e., calling the IRS and authenticating over the phone or authenticating in person at a Taxpayer Assistance Center [TAC]), whereas the non-IDT refund fraud program is entirely reliant on the IRS taking action in order for the refund to be released (i.e., the IRS must verify the income reported by the taxpayer with the income reported by the employer/payor).
While the OPR provides data on the number of returns delayed no more than four weeks, it still lacks detail. Many of the taxpayers selected into the non-IDT refund fraud program had their returns delayed far beyond four weeks. In my 2017 Annual Report to Congress, I reported that about 37 percent of returns selected into the pre-refund wage verification program during the 2017 filing season took 11 weeks or more to be processed.
Dealing with a refund delayed beyond normal processing times may cause great anxiety for taxpayers, especially low income taxpayers who often rely on their refunds for emergencies such as paying medical expenses, or to pay other day-to-day bills. For those taxpayers selected into the non-IDT refund fraud program, this anxiety is magnified by the fact that IRS customer service representatives (CSRs) do not have access to the non-IDT refund fraud case management system, leaving them in the dark as to what is delaying their return, and when they can expect their refund.
After a review of the 2018 filing season, TAS identified several issues that contributed to the delays in the non-IDT refund fraud program. The first issue was timing. For instance, the SSA must receive employer’s information by January 31. The SSA then forwards the information to the IRS who uses it to verify the information on a taxpayer’s return. If the IRS does not receive the third-party information by the time a taxpayer files his or her return, a fraud detection filter may select the return because there is no third-party information available to verify the return. However, the IRS may receive the employer information within a day or two of selecting the return, but the IRS would not be aware that it received that third-party information for at least a week because, in 2018, it checked for third-party information weekly, not daily.
The fraud detection systems also select returns because there is a discrepancy between income shown on the return and income shown on third-party documentation. However, we identified that the system selected returns where changing the income amount on the return to match that of the third-party documentation either increased the refund amount or did not change it.
Finally, a major contributor to the delay of the release of selected non-IDT refund fraud returns which were verified as legitimate is the use of the outdated Electronic Fraud Detection System (EFDS), which is the case management system for non-IDT refund fraud. One of the new filters used during the 2018 filing season selected about 303,000 Earned Income Tax Credit (EITC) and Additional Child Tax Credit (ACTC) returns as potentially fraudulent because no third-party wage information had been posted as of February 15, 2018, about two weeks after the January 31 deadline established by law. Once these accounts were selected as potentially fraudulent, the IRS anticipated that EFDS would be able to release refunds in bulk when income on the returns could be verified with third-party information that come in later. However, because EFDS does not interact with the IRS system that maintains third-party wage information, IRS employees had to manually enter the third-party data into EFDS one document at a time to release the refunds.
NTA appreciates comments and suggestions.
Subscribe to TaxConnections Blog
Enter your email address to subscribe to this blog and receive notifications of new posts by email.