The issue of tax refund fraud by identity theft provides a unique opportunity to analyze significant data over a period of years in an effort to learn the right lessons about the phenomenon. The sequence in which changes were made also makes it possible to assess the impact of different approaches to combating this crime. The ongoing case study described in earlier Blog posts outlines a form that analysis might take. http://www.fgs.org/rpac/2015/07/05/id360-closing-death-records-silver-bullet-or-detour-a-case-study/
Examining this essentially comparable data from Tax Years 2010, 2011, and 2012 seems to validate my assertion that this approach helps to give us visibility over the effectiveness of the IRS efforts to combat tax refund fraud. Recognizing that it may take over a year for a final determination to be made as to the validity of a tax return, we should anxiously pursue and evaluate the equivalent TY 2013 data at the earliest possible moment. Meaningful information may be gleaned from subsequent years as well.
I am reminded that it would have been TY 2010 transactions generating CY 2011 headlines and, thus, TY 2010 data provides our best baseline against which to assess the impact of the evolving IRS use of filters to flag potentially fraudulent tax returns claiming inappropriate refunds. Little, if any, fraud detection filtering was in place as TY 2010 returns were being processed with the primary emphasis being placed upon quick refund processing and payment. In fact, refund checks were being written within days of receipt early in the filing season even before the IRS would have received information returns that would eventually be used to validate returns. The Government Accountability Office has described this IRS business model as a “Pay and Chase” enforcement approach. http://www.gao.gov/products/GAO-15-482T
Addendum 12 Jan 2016: During this baseline period, and until the closure mandated by the Bipartisan Budget Act of 2013 took effect in March of 2014, DMF data was widely available from the official site at NTIS and elsewhere. It should be noted, however, that in December of 2011 all of the major genealogical sites responded to the possibility that thieves might have gotten information used to file fraudulent tax returns from their resources by masking the display of deceased persons SSNs for a minimum of three years after their death.
If, as reported, the IRS began the use of fraud detection filters (likely including the Death Master File) in December 2011, we should not be surprised that TY 2011 data reflects a significant reduction in the number of undetected fraudulent returns involving the SSNs of deceased individuals. In TY 2010, the 104,950 deceased cases represented slightly over 7% of the total 1,492,215 potentially fraudulent tax refunds paid that year. In TY 2011, the 19,102 deceased cases represented approximately 2% of the total 1,086,998 cases. The pleasant surprise occurs when one considers the extent to which the IRS development of a series of filters, scoring algorithms and clustering methodologies over the last four years has improved the detection of identity theft returns involving the living, a far more difficult task.
TY 2012 improved further with 12,338 deceased cases representing slightly over 1.5% of the total 787,343 cases determined to be potentially fraudulent. The dollar amounts of the TY 2012 questionable refunds involving the SSNs of deceased individuals ($22,239,751) represented barely 1% of the $2,137,397,982 total.
These results are clearly consistent with the assertion that the IRS, once they began to develop filters to flag potentially fraudulent tax returns in December 2011, continued to expand and improve the events which would prompt additional scrutiny of a return before a refund check would be processed. That initial set of filters used in processing Tax Year 2011 returns using 11 filters was increased to more than 80 for Processing Year 2013 generating the further improvements seen in the Tax Year 2012 results above.
A subsequent blog entry will provide further analysis of this information and explore the implications of these findings.