On May 15, 2019, the IRS’s Large Business and International Division (LB&I) announced a new program, the Large Corporate Compliance (LCC) program, which will rely on the use of data analytics to determine which large and complex corporate taxpayers it will audit. IRS News Release, IR-2019-95 (2019). The LCC program replaces the Coordinated Industry Case (CIC) program. The IRS’s latest announcement signals another step towards more sophisticated means of audit selection.
The persistent refrain from the IRS and tax professionals over the last decade has been that the IRS’s budget and resources are dwindling, resulting in fewer audits. The use of advanced data analytics will allow the agency to perform its mission with a smaller work force and a tighter budget, the IRS says, because newer technologies will allow its people to work smarter and more efficiently. Tax practitioners will appreciate that data-driven selection may result in a fairer process of selecting returns for audit, but the use of automation and data analytics may also raise concerns about selection bias and transparency.
Pointing LB&I Towards Returns with the Highest Compliance Risk
Like the CIC program it replaces, the LCC program will use pointing criteria to determine which cases should be in the program. Pointing criteria are a list of factors, each of which is assigned a points value. Under the LCC program, seven pointing factors will be used to determine whether a case belongs in the LCC program: gross assets, gross receipts, operating entities, multiple industry status, total foreign assets, total related transactions, and foreign tax. Internal Revenue Manual (I.R.M.) 18.104.22.168 (Apr. 22, 2019); I.R.M. Exhibit 4.50.2-1. If after adding together the point criteria for each factor the total equals 15 or more points, the case will qualify as a LCC case. I.R.M. 22.214.171.124 (Apr. 22, 2019). Unlike the CIC program, which applied pointing criteria on a manual, localized basis, the LCC will apply the pointing criteria using computer automation.
After automated pointing determines which cases will be in the LCC program population, data analytics will be used to identify which returns pose the highest compliance risk. LB&I indicated that the LCC program allows for “continuous improvement using an agile model principle to continually monitor and improve based on feedback from stakeholders including field teams, practice networks, and data scientists.” IRS News Release, IR-2019-95 (2019). LB&I implemented “agile model” in 2014, which it explained was driven by “testing and experimentation to explore and validate compliance risks and associated issues.” See IRS LB&I Concept of Operations (Dec. 31, 2014).
The IRS may use the large universe of available data to select taxpayers’ returns for audit. Some potential data sources are public or commercial, such as credit reports, social media (such as Twitter, Facebook, Instagram), public filings with the Securities and Exchange Commission (such as Forms 10-K), news sources, city and state records (like deeds and registries), credit and debit card processors, eBay, PayPal, and marketing data. See, e.g., Department of Treasury, Internal Revenue Service, Request for Information Regarding Social Media Research Request (Dec. 18, 2018, modified Jan. 30, 2019, and Feb. 19, 2019). The IRS also has access to non-public information, such as prior tax returns and information returns, data collected from third parties (such as offshore John Does summonses), and data collected through information sharing with other countries. This data comprises a taxpayer’s digital footprint and may factor into a decision about whether to audit a return.
An Increased Focus on Data Analytics
The IRS’s use of data analytics in audit decisions is not new, but the processes and methods have become more sophisticated. The IRS previously used the Discriminate Inventory Function (DIF) system, which dates back to the 1960s. The DIF system assigned numeric scores to individual and some corporate returns. If a taxpayer received a higher DIF score, the taxpayer’s return was more likely to be audited. The new automated and data-driven approaches used by the IRS are like DIF scores on steroids.
The IRS has increasingly focused on data analytics over the past several years. In November of 2016, the IRS formed the Research, Applied Analytics and Statistics Division (RAAS) out of the previous Office of Compliance Analytics and the Research, Analysis and Statistics offices. Within the IRS organization, RAAS falls under the Deputy Commissioner for Operations Support. The IRS says that RAAS’s mission is to “lead a data-driven culture through innovative and strategic research, analytics, statistics, and technology services in partnership with internal and external stakeholders.” See I.R.M. 126.96.36.199(1) (Sept. 28, 2018). RAAS aims to achieve its mission by “combin[ing] advanced analytics, dynamic testing, reporting, and prototyping with appropriate scientific rigor and deep IRS domain expertise to deliver valid and actionable insights using diverse sources of data.” See I.R.M. 188.8.131.52(3) (Sept. 28, 2018). RAAS provides its services to the various divisions and programs within the IRS, including the LB&I division.
The IRS also has invested heavily in data analytics technology. In September 2018, the IRS signed a deal with Palantir Technologies for $99 million over seven years. See IRS, Contract Proposal, Performance Work Statement, Jan. 11, 2017. Palantir Gotham is a project under the Palantir Technologies umbrella with other U.S. government clients, including the CIA, Department of Defense, and Department of Homeland Security. The IRS uses the Palantir Gotham platform to run its Lead Case Analytics service, which allows the IRS to find, analyze, and visualize connections between disparate sets of data. Special agents and investigative analysts in IRS Criminal Investigations use Lead Case Analytics to “generate leads, identify schemes, uncover tax fraud, and conduct money laundering and forfeiture investigative activities.” See IRS, Privacy Impact Assessment: Lead Case Analytics, LCA (Jul. 28, 2015).
With the announcement of the LCC Program, the IRS continues in this data-driven direction. If LB&I’s strategic goals are any indication, this is likely just the beginning. LB&I has emphasized the need to “use data to drive compliance decisions.” See IRS Publication 5319, FY2019 LB&I Strategic Goals (Rev. Feb. 2019). The IRS is also seeking input from the tech community on ways it can use artificial intelligence, machine learning, cognitive computing, and data analytics techniques. See Department of Treasury, Internal Revenue Service, Request for Information Regarding Advanced Analytics, Artificial Intelligence and Machine Learning Capabilities for the Cybersecurity Cloud Solution Program (Jun. 27, 2018).
Effects of the Large Corporate Compliance Program on Taxpayers
The IRS’s data-driven mindset could be beneficial for compliant taxpayers. Using data may result in the IRS choosing only those returns that have the greatest risk of compliance issues. Where the IRS selects a taxpayer for audit, it may limit the scope of the audit to select issues with a higher risk of noncompliance.
Although the use of data analytics in audit selection has a great deal of potential upside, there are still concerns. One issue with using data analytics for audit selection is that the results are only as good as the underlying algorithm that the computer employs in selecting returns for audit. The IRS will have to carefully evaluate whether any unconscious biases are built into its algorithms that could disproportionately target certain taxpayers or groups of taxpayers.
Another concern is transparency. The IRS so far has engaged with the public regarding its use of new data analytics programs and techniques. The IRS however is unlikely to disclose details about the way that it selects returns for audit. For example, past attempts by taxpayers seeking information related to how DIF scores were calculated were unsuccessful due to the protection of tax code Section 6103(b)(2), which states that the IRS does not need to disclose to the taxpayer the “standards used or to be used for the selection or returns for examination, or data used or to be used for determining such standards.” In Gillin v. IRS, the court held that the “IRS closely guards information concerning its DIF score methodology because knowledge of the technique would enable an unscrupulous taxpayer to manipulate his return to obtain a lower DIF score and reduce the probability of an audit.”
Tax practitioners should be cognizant of the new technologies that the IRS is employing and adapt accordingly. Tax practitioners should try to learn as much as they can about their client’s digital footprints and potential data about their client that already may have been collected by the IRS. This will aid tax practitioners and their clients in responding to Information Document Requests (IDRs), preparing for interviews, and other interactions with the IRS in an environment where the IRS will have amassed and analyzed large amounts of data even before selecting a taxpayer’s return for audit.
This column does not necessarily reflect the opinion of The Bureau of National Affairs, Inc. or its owners.
Carina Federico is an associate in Crowell & Moring’s Tax Group in the firm’s Washington, D.C. office. Carina focuses her practice on federal tax controversy and tax litigation matters before the IRS and in trial and appellate courts across the U.S. Her experience includes serving as first chair at trial, taking and defending depositions, briefing a wide range of tax issues, negotiating settlements, and representing clients in IRS Appeals conferences.
David B. Blair is a partner and chair of Crowell & Moring’s Tax Group in the firm’s Washington, D.C. office. Mr. Blair’s practice is in the area of federal tax litigation and controversy. With over 20 years of tax litigation and trial experience, he has handled large tax litigations in the areas of transfer pricing, foreign tax credits, partnerships, tax-exempt bonds, consolidated returns, excise taxes, employment taxes, and tax accounting issues.