The “tax gap”—the difference between taxes owed to the government and taxes actually collected—often is equated with evasion. But the tax gap is more than the willful disregard of the law. Rather, it appears in many shades of gray. In a new report, we, together with University of Chicago law professor Daniel Hemel, describe the implications of these nuances for tax policy and administration.
The IRS’s measure of the tax gap includes unpaid federal income, payroll, excise, and estate taxes. In its most recent study of noncompliance, the agency estimated the gross annual tax gap was $441 billion for tax years 2011-2013. Late payments and enforcement revenue reduced the gap by $60 billion each year.
But the IRS estimate does not distinguish between taxpayer mistakes and intentional evasion. For example, it includes the honest errors made by taxpayers trying to comply with ever-changing and often-opaque laws. The IRS also includes the liabilities of well-intentioned taxpayers who acknowledge what they owe but are unable to pay.
Moreover, the IRS may be overstating or understating the true tax gap. The agency uses a statistical method called “detection controlled estimation” (DCE) to account for noncompliance missed in audits of individual income tax returns. But that approach is being reconsidered or already has been rejected by other countries out of concern for its accuracy. Some researchers have found DCE effectively triples the estimate of the individual income tax underreporting gap.
At the same time, the IRS omits unambiguous evasion from its tax gap estimates. For example, it excludes unreported income from illegal sources, such as drug dealing. It also may exclude underreporting by passthrough businesses such as partnerships and S corporations because the IRS does not routinely audit these entities in its compliance studies.
Then there are those shades of gray where taxpayers, IRS examiners and appeals officers, the courts, and Congress may disagree over what is legal. For example, the IRS bases its tax gap estimates on the post-audit recommendations of examiners, even if they are reversed on administrative appeal or by court challenge. As a result, the IRS may include taxpayer behavior that failed to pass muster with auditors but ultimately was found by IRS appeals officers or the courts to be legal avoidance that should not be included in the tax gap.
Our paper describes three implications of properly framing the tax gap for tax policy and administration:
1. The Tax Gap Should Not Be A Performance Target.
IRS research reveals important sources of noncompliance and sheds light on the amount of unpaid taxes that could be collected under current law. But aiming to reduce the tax gap by a particular percentage or amount could result in potentially perverse incentives. For example, the IRS could meet the target simply by encouraging examiners to accept aggressive positions adopted by sophisticated taxpayers.
2. The IRS Needs Skilled Examiners And Up-To-Date Technology To Better Address Noncompliance.
Detecting noncompliance is challenging when sophisticated taxpayers and their advisers aggressively exploit ambiguities in the tax code. Addressing those avoidance techniques is especially resource-intensive and would require a substantial increase in IRS funding to hire and train skilled examiners.
3. Tackling Noncompliance Requires Substantive Legal Reforms.
Even the most skilled and experienced examiners will struggle to enforce tax laws that are ambiguous at their core. Audits are no substitute for legal clarity. For example, current law practically invites owners of passthrough businesses to adopt aggressive reporting positions. Some are challenged by examiners, others are accepted as legal avoidance, and many never are observed because the IRS rarely audits these entities. If Congress wrote better tax rules for passthroughs, the IRS would have a much easier job and its enforcement would be more equitable.
The tax gap is a significant problem. But the first step towards addressing it is understanding its shades of gray.
