In May 2013, the Treasury Inspector General for Tax Administration (“TIGTA”) reported audit results showing that between May 2010 and May 2012, the Internal Revenue Service (“IRS”) used “inappropriate criteria” to identify which organizations’ applications for tax-exempt status it would give heightened scrutiny.  TIGTA found that IRS selections had been based on groups’ names or policy positions rather than objective indicia that groups might act outside the requirements and limitations for tax-exempt status under 26 U.S.C. § 501(c).  As restated by TIGTA in its latest Review of Selected Criteria Used to Identify Tax-Exempt Applications for Review, published this month, “using names and/or policy positions instead of developing criteria based on tax-exempt laws and Treasury Regulations is inappropriate.”

The IRS asserts it has been reforming its review-selection process ever since in an effort to develop “data-driven” criteria and use “data analytics to inform decision-making.”  For example, the latest work plan for the Tax Exempt and Government Entities 2018 fiscal year says the IRS will “continue to improve Form 990, 990-EZ, and 990-PF compliance models” as part of a new “data-driven approach” to checking compliance and selecting returns for examination.  The phrases “data-driven selection criteria” and “data analytics” can conjure algorithms and black-box calculators that are supposed to mirror impartial decisions and whose lack of bias is notoriously difficult to understand without expert background and sophisticated mathematics.

The IRS’s description of its approach, however, proffers something much simpler that does not involve sophisticated mathematics or algorithmic analysis.  Instead, its new “data-driven” approach involves analyzing informational returns for indicia that the group is operating outside the restrictions of the tax-exempt statutes or not complying with reporting obligations.  So, under the 2018 Plan, examinations will target organizations whose returns show the “highest risk of Employment Tax non-compliance” (such as 1099 information showing “high distributions” or numerous employees but “zero or minimal Medicare and/or Social Security wages paid”).  And examinations will focus on entities that do not file schedules required by their Form 990 information.  More generally, when an entity’s Form 990 suggests it has taxable business income unrelated to its charitable purposes, an examination should ask whether the entity filed Form 990-T and if not, why not.

If this “data-driven” approach is new, a few questions arise:  How was the IRS using data from Form 990 series returns before now?  Why weren’t these compliance criteria used previously?  Shouldn’t the IRS have been flagging these potential problems all along?

In any event, to the extent “data-driven” analysis does not rely on names or policy positions but instead focuses on objective indicia of compliance with the law and Treasury regulations, the new approach will be better than the IRS’s past inappropriate practices.  But if that’s all that’s being changed, and the data being relied upon is still from Forms 1023 and 990, then perhaps “machine-driven” better captures the new examination-selection criteria than “data-driven.”

Mike Geske is Counsel at Cause of Action Institute.