WASHINGTON—The U.S. Internal Revenue Service has launched several data analytics projects that are improving compliance, combating fraud, identifying abusive tax shelters and saving money, according to several representatives of the federal tax collection agency. The officials spoke at the Predictive Analytics World Government Conference held here on Sept. 17 and 18.
Dean Silverman, senior advisor to the commissioner at the IRS and director of the IRS Office of Compliance Analytics (OCA), delivered a keynote speech on the first day of the conference. Silverman, who worked for 25 years at various consulting firms, including Bain & Co. and Mercer Management Consulting, joined the IRS last year.
OCA’s mission is to improve compliance with tax laws and to make data analytics a key part of IRS strategic decision-making, said Silverman. Reporting directly to the IRS commissioner, the office is charged with “building a robust culture of analytic problem-solving,” he said. Part of that means change from the top down, he said, noting that his job involves helping executives “understand what good analytics support looks like and then to expect and demand it.”
Another part of that means change from the bottom up, getting staff to pay the right amount of attention to the right details. “You’d be surprised how many discussions are about what data was selected for what analysis and how that data” was prepared for analysis by decision-makers, Silverman said.
Silverman described three ways the IRS is using analytics today:
Pattern recognition to identify mistakes in filed tax returns. During the 2012 tax filing season, the IRS piloted and tested a program that analyzes tax returns as they are being processed to identify patterns of mistakes or “issues” on returns and correlating that with particular professional tax preparers. The IRS then contacted those preparers immediately to discuss and correct the problems. “We didn’t want to wait [until] after the filing season, because most refund claims come in very early in the filing season,” said Silverman.
Those early interventions prevented the IRS from erroneously refunding several hundred million dollars, he said. The also helped improve preparer compliance and avoid errors in subsequent filings. The agency plans to expand the program in 2013.
Fraud detection. Tax refund fraud “is growing very significantly, especially refund fraud that’s the result of identity theft,” said Silverman. OCA and several IRS business units are developing data analytics models to improve the detection of such fraud. They tested and improved the models during the 2012 tax filing season and now have “various ID theft screening filters in place to improve our ability to stop false returns before they are processed and before a refund is issued,” said Silverman. “We’re also creating a taxonomy of identity theft fraud, [which is] helping to shape a comprehensive strategy.”
Audit trails from credit and debit card transactions. Starting this year, credit/debit card processors and payment aggregators like PayPal are required to report to the IRS the value of all card transactions. “This new requirement will give the IRS third-party reporting data on business receipts for the very first time,” helping the department to identify underreporting of business income. OCA is working with other parts of the IRS on data analytics models that will help them “to test and learn where enforcement attention is most required, and where education and outreach can help us improve voluntary compliance.”
Analytics Tools for an Aging Agency Workforce
In a separate presentation, IRS staffers described two other data analytics projects.
One involves the key issue of workforce management. In the next five years, more than 35 percent of the IRS workforce and half of its leadership will become retirement eligible, said Jonathan Edelson, supervisory management and program analyst in the IRS Office of Research, Analysis and Statistics. To help plan and manage that change and as part of the IRS Commissioner’s Workforce of Tomorrow Initiative, the IRS is developing a Strategic Workforce Analysis Tool (SWAT).
Emily Shammas, analyst in the IRS Office of Research, Analysis, and Statistics described SWAT as a set of three models designed to project trends in the IRS workforce and workloads. Today, the agency is using the tool to predict workforce attrition rates, said Shammas, and is in the process of validating and verifying other capabilities of the tool.
A second example involves analyzing the structure of certain transactions. Rahul Tikekar, computer scientist in the Intelligent Business Solutions group the IRS Office of Research, explained a tool being developed to identify possible “abusive tax-avoidance transactions.” An example of such a transaction might be a taxpayer setting up an S corporation to receive income from a partnership, thus avoiding the self-employment tax. The tool would enable an IRS agent to search for transactions in filings by entering a simple diagram of the structure of the transaction. Tikehar said the tool would be rolled out and tested with a small number of IRS users over the next year.
Tam Harbert is a freelance writer based in Washington, D.C. She can be contacted through her website.