THE TAX REDISTRIBUTION GAP Eric Baudry* Abstract The tax revenue gap—the difference between how much the IRS collects in tax revenue and how much it should collect based on the text of the Internal Revenue Code— is both well-defined and well-studied. But raising revenue is just one purpose of taxation; the tax code also operates to redistribute wealth. Drawing from the tax revenue gap and redistribution literatures, this article coins a parallel concept, the tax redistribution gap, to map the extent to which the tax system falls short of its redistributive goals. Introducing a tax redistribution gap measure challenges background assumptions in current tax discourse: first, it would call out a reliance on pre-market income as a distributive baseline, which serves to overstate the redistributive impact of the tax code; second, it would increase the profile of redistribution among policymakers and the public— understandably, measures like the tax revenue gap and tax expenditure budgets focus dialogue on tax evasion and over-spending by the government. Ultimately, the tax redistribution gap would provide a single measure that displays how we are falling short of a key task of the state (redistribution). And by understanding and comparing the component ways our current tax systems falls short of it intended outcomes, we can better tailor redistributive policy solutions. * Faculty Fellow, University of Michigan Law School. I thank Anne Alstott, Reuven Avi-Yonah, William Eskridge, Ari Glogower, Jacob Goldin, Richard Kaplan, Ariel Jurow Kleiman, Robert Lawless, the faculties of the University of Illinois and University of Michigan Law Schools, and the attendees of the 2024 Michigan Law Junior Faculty Forum for their helpful comments and insightful feedback. Thanks also to the staff of the Columbia Journal of Tax Law for their excellent editorial work. 2025] THE TAX REDISTRIBUTION GAP 87 I. INTRODUCTION ................................................................................................ 88 II. IRS TAX MEASURES ........................................................................................ 89 A. The Tax Revenue Gap ............................................................................... 91 1. Tax Revenue Gap Form ....................................................................... 92 2. Measuring the Tax Revenue Gap ......................................................... 93 3. Tax Revenue Gap Functions ................................................................ 94 B. The Absence of a Tax Redistribution Gap ................................................ 96 1. Shortcomings of Distributive Analysis ................................................ 97 2. Understanding Redistributive Avoidance ............................................ 98 3. Finding Space for a Tax Redistribution Gap Measure ...................... 100 II. BUILDING THE TAX REDISTRIBUTION GAP MEASURE .................................... 104 A. What is Redistribution? ........................................................................... 104 1. Redistributive Subjects ...................................................................... 105 2. Redistributive Mechanisms and Goods ............................................. 105 3. Redistributive Baselines .................................................................... 107 B. Defining Redistributive Taxation ............................................................ 109 1. Direct Transfers ................................................................................. 109 2. Differential Contributions ................................................................. 111 III. IMPLEMENTING THE TAX REDISTRIBUTION GAP MEASURE ........................... 114 A. The EITC ................................................................................................. 115 B. The Implementation Gap ......................................................................... 117 1. Nonparticipation ................................................................................ 117 2. Underclaiming ................................................................................... 119 3. Underreceiving .................................................................................. 120 IV. CLOSING THE TAX REDISTRIBUTION GAP ...................................................... 124 A. Nonparticipation Gap .............................................................................. 125 B. Underclaiming Gap ................................................................................. 127 C. Underreceiving Gap ................................................................................ 127 VI. CONCLUSION .................................................................................................. 129 88 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 I. INTRODUCTION Why do we tax? The Constitution lays out Congress’ power “to lay and collect taxes . . . to pay the debts and provide for the common defence and general welfare of the United States,”1 and the 16th Amendment specifically establishes Congress’ ability to collect taxes “on incomes, from whatever source derived[.]”2 This language squarely frames income taxation’s revenue raising purpose; unsurprisingly, the IRS measures the taxes it receives each year.3 But modern tax scholars recognize that taxation’s functions go beyond raising funds: taxes also operate as both a redistributive tool for “reducing the unequal distribution of income and wealth that results from the normal operation of a market-based economy” and a regulatory lever by which the government can “steer private sector activity in . . . [desired] directions.”4 As with total revenue raised, several government agencies measure the distributive impact of the tax code.5 These baseline measures seem to put revenue and redistribution on equal footing. But government attention to revenues goes beyond gross receipts. In 1964, the IRS began to study not just how much it raised in taxes, but also how much it should have received under the tax code as written—we call the difference between these figures the tax gap.6 As one former Commissioner of the IRS wrote in 1979, this secondary measure reflects both that “[k]nowledge about overall compliance is important for the efficient administration of the tax laws” and that “there has also been general public interest in this problem.”7 While government agencies do measure the distributive impact of the tax code, what is missing from this analysis is a scorecard for taxation’s redistributive function—specifically, is the tax code creating as much redistribution as is built into its design? 1 U.S. CONST. art. I, § 8, cl. 1. 2 U.S. CONST. amend. XVI. 3 For Fiscal Year 2023, the IRS processed 271.5 million tax returns and collected almost $4.7 trillion in gross tax revenue. Internal Revenue Serv., SOI Tax States – IRS Data Book, https://www.irs.gov/statistics/soi-tax-stats-irs-data-book (Aug. 22, 2024) [https://perma.cc/4NP2- ED6V]. 4 Reuven Avi-Yonah, The Three Goals of Taxation, 60 TAX L. REV. 1, 3 (2006). Avi-Yonah distinguishes between taxation’s purposes in another way: he has argued that only tax provisions that primarily seek to raise revenue should be subject to the Taxing Clause. Reuven Avi-Yonah & Yoseph Edry, Constitutional Review of Federal Tax Legislation, 1 ILL. L. REV. 1, 3-4 (2023). By contrast, tax provisions effectuating redistribution “are inherently political, and thus, their distributive function of reducing inequality should not be subject to judicial review.” Id. 5 See Revenue Estimating, JOINT COMM. ON TAX’N, https://www.jct.gov/operations/revenue- estimating/ [https://perma.cc/FF8V-P7MA]; Trends in the Distribution of Household Income from 1979 to 2020, CONG. BUDGET OFF. (Nov. 14, 2023), https://www.cbo.gov/publication/59510 [https://perma.cc/97YC-6YXL]; Office of Tax Analysis, U.S. DEP’T OF THE TREASURY, https://home.treasury.gov/policy-issues/tax-policy/office-of-tax-analysis [https://perma.cc/C2V3- K5SU]. 6 Daniel Hemel et al., The Tax Gap’s Many Shades of Gray 6 (U. Chic. Coase-Sandor Inst. for L. & Econ., Working Paper No. 938, 2021). This article will refer to the concept as the tax revenue gap. 7 INTERNAL REVENUE SERV., DEP’T OF THE TREASURY, PUBL’N 1104, ESTIMATES OF INCOME UNREPORTED ON INDIVIDUAL INCOME TAX RETURNS ii (1979). https://www.irs.gov/statistics/soi-tax-stats-irs-data-book https://www.jct.gov/operations/revenue-estimating/ https://www.jct.gov/operations/revenue-estimating/ https://www.cbo.gov/publication/59510 https://home.treasury.gov/policy-issues/tax-policy/office-of-tax-analysis 2025] THE TAX REDISTRIBUTION GAP 89 In this article, we will examine tax measurements: what the IRS measures when it comes to taxes, what it does not, and why it matters. Part I highlights the disparity between respective measures for two of taxation’s core purposes: revenue- raising and redistribution. It studies the form and impact of the tax revenue gap and concludes by exploring why redistribution does not receive the same attention. Part II draws from writings on redistribution and the tax revenue gap to coin the latter’s conceptual mirror image, the tax redistribution gap. It first defines redistribution and redistributive taxation to establish a clear object of study; next, it builds a framework for a tax redistribution gap consisting of three parts— nonparticipation, underclaiming, and underreceiving gaps. Together, these components offer a new measure by which the IRS can assess the extent to which the tax system falls short of its redistributive goals. Part III applies the framework developed in Part II to sketch how we might calculate the tax redistribution gap. The nonparticipation gap occurs when eligible taxpayers fail to file for the refundable credits; the underclaiming gap results from taxpayers who claim credits but do not claim the full amount to which they are entitled; and the underreceiving gap is caused by legal processes that result in otherwise eligible taxpayers losing their credit refunds. Finally, Part IV establishes the conceptual utility of a tax redistribution gap measure. Introducing a tax redistribution gap can challenge normative assumptions and priorities embedded in the tax code; it can also help shape social and political discourse by raising the profile of redistribution within government and society. But perhaps most importantly, a tax redistribution gap would offer an easy-to- understand measure that reflects how we are falling short with respect to a key task of the state. II. IRS TAX MEASURES Measurement matters, or so the idiom goes.8 Contemporary culture ripples with calls to measure: business blogs counsel adopting measurable targets under acronymic frameworks like SMART and OKR;9 the popular book and subsequent movie Moneyball tells the story of how Oakland Athletics general manager Billy Beane revolutionized baseball strategy through the use of “sabermetrics”—or sports analytics—to identify diamond players in the rough;10 a decade and a half 8 Some attribute the sentiment to 16th century mathematician Rheticus, who professed that if we can measure something, we can exercise control over it. See Matthew Cornell, What’s your feed reading speed? (July 30, 2007), http://www.matthewcornell.org/blog/2007/7/30/whats-your-feed- reading-speed.html#1 [https://perma.cc/9HTS-4Z36] (citing GEOGRAPHY MATTERS! (Doreen Massey & John Allen eds., 2010)). Others point to 19th century scientist Lord Kelvin as having said that “when you can measure what you are speaking about, and express it in numbers, you know something about it.” Id. (citing HAN T.M. VAN DER ZEE, MEASURING THE VALUE OF INFORMATION TECHNOLOGY (2001)). 9 See Leon Ho, 5 Reasons Why It Is Important for Goals to Be Measurable, LIFEHACK (Aug. 10, 2023), https://www.lifehack.org/884896/why-is-it-important-that-goals-be-measurable [https://perma.cc/6G7H-5R64]; Execution is Everything, WHAT MATTERS, www.whatmatters.com [https://perma.cc/T25E-322P]. 10 See MICHAEL LEWIS, MONEYBALL: THE ART OF WINNING AN UNFAIR GAME (2003). http://www.matthewcornell.org/blog/2007/7/30/whats-your-feed-reading-speed.html#1 http://www.matthewcornell.org/blog/2007/7/30/whats-your-feed-reading-speed.html#1 https://www.lifehack.org/884896/why-is-it-important-that-goals-be-measurable http://www.whatmatters.com/ 90 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 after Fitbit revolutionized the pedometer, wrist adornments that measure our steps have largely replaced traditional watches.11 The impulse to measure and track is not without support. Business professor Bill Tayler contends that “[h]umans are hardwired to respond to what is being measured because if it’s being measured, it feels like it matters.”12 The rise of Fitbit offers one data point: the act of wearing a pedometer, even if the user did not look at it, increased daily step counts.13 Hybrid car owners experienced a similar phenomenon when their dashboards began displaying fuel efficiency, triggering subtle incentives to adjust their driving in order to keep the number up.14 Psychology calls this the Hawthorne Effect: the phenomenon that awareness of a measurement alone causes an increase in performance with respect to the measure.15 The power of measurement to motivate change also extends to government. Take poverty: in 1963, food economist Mollie Orshansky developed a new measure of poverty levels for the Social Security Administration based on family size; in 1965, the Office of Economic Opportunity, the government lead for President Johnson’s newly declared War on Poverty, adopted Orshansky’s poverty thresholds; and in 1969, Orshanky’s calculations became the government’s first official statistical definition—what we now call the Official Poverty Measure.16 While modern debates rage on the long-term success of anti-poverty efforts,17 national poverty rates fell significantly during the first years of the War on Poverty, from 15.6% in 1965 to 11.9% in 1972.18 But there is a flipside to fixating too much on what we measure, which is that we can remain unaware or lose sight of that which we do not. Business leaders 11 See Camille Gonzalez, How Fitbit Made the Pedometer Cool (Again?), MEDIUM (Aug. 21, 2018), https://medium.com/@camillegonzalez/how-fitbit-made-the-pedometer-cool-again-f8da4c03a6a4 [https://perma.cc/8TXX-HMWE]. 12 Christie Allen, Wearing a pedometer—even if you don’t look at it—may boost step counts, BYU MARRIOTT SCH. OF BUS. FAC. RSCH. (Oct. 28, 2022) https://marriott.byu.edu/stories/faculty- research/wearing-a-pedometer-even-if-you-dont-look-at-it-may-boost-step-counts [https://perma.cc/SG3C-MAHX]. 13 William Tayler et al., The Effect of Wearable Activity Monitor Presence on Step Counts, 46(4) AMER. J. HEALTH BEHAV. 347 (2022). 14 See Michael Rosenwald, For Hybrid Drivers, Every Trip Is a Race for Fuel Efficiency, WASH. POST (May 26, 2008). Energy scholar Sarah Darby explained that “[o]nce you start making fuel consumption more visible, you have something that comes to the forefront of people's minds instead of lurking in the background.” Id. 15 See Jim McCambridge et al., Systematic review of the Hawthorne effect: New concepts are needed to study research participation effects, 67 J. CLINICAL EPIDEMIOLOGY 267, 267 (2014). 16 Gordon Fisher, Remembering Mollie Orshansky—The Developer of the Poverty Thresholds, 68(3) SOC. SEC. BULL. (Dec., 2008), https://www.ssa.gov/policy/docs/ssb/v68n3/v68n3p79.html [perma.cc/28DY-5FC7]. 17 Compare Matthew Desmond, Why Poverty Persists in America, N.Y. TIMES MAG. (Apr. 3, 2023), https://www.nytimes.com/2023/03/09/magazine/poverty-by-america-matthew-desmond.html with Dylan Matthews, Why even brilliant scholars misunderstand poverty in America, VOX, (Mar. 10, 2023) https://www.vox.com/future-perfect/2023/3/10/23632910/poverty-official-supplemental- relative-absolute-measure-desmond [https://perma.cc/C6CD-7XY4]. 18 See Robert Haveman et al., The War on Poverty: Measurement, Trends, and Policy, 34(3) J. POL’Y ANALYSIS & MGMT. 593, 600 (2015). https://medium.com/@camillegonzalez/how-fitbit-made-the-pedometer-cool-again-f8da4c03a6a4 https://marriott.byu.edu/stories/faculty-research/wearing-a-pedometer-even-if-you-dont-look-at-it-may-boost-step-counts https://marriott.byu.edu/stories/faculty-research/wearing-a-pedometer-even-if-you-dont-look-at-it-may-boost-step-counts https://www.ssa.gov/policy/docs/ssb/v68n3/v68n3p79.html https://www.nytimes.com/2023/03/09/magazine/poverty-by-america-matthew-desmond.html https://www.vox.com/future-perfect/2023/3/10/23632910/poverty-official-supplemental-relative-absolute-measure-desmond https://www.vox.com/future-perfect/2023/3/10/23632910/poverty-official-supplemental-relative-absolute-measure-desmond 2025] THE TAX REDISTRIBUTION GAP 91 call this “measurement myopia.”19 This nearsightedness can also show up in government programs. Since the launch of the oft-maligned No Child Left Behind program, the emphasis on test scores in American education has resulted in a decline in other aspects of teaching and learning. Public schools across the country have slashed music and arts programs to leave more time for reading, writing, and math;20 research suggests that schools that improve standardized test scores do not simultaneously see student improvement in cognitive skills like processing speed, working memory capacity, and fluid reasoning.21 Work by the New Economics Foundation sums up this tension by restating our original idiom: “Measurement matters. It matters because it both reflects and reproduces the priorities of government. Ultimately, it determines where public resources are allocated and therefore what goals will be pursued.”22 Given that measurements can both spur progress and result in tunnel vision, politicians, policymakers, advocates, and citizens should want to know what the government is measuring. The rest of this part looks at IRS measures with respect to revenue and redistribution; specifically, it examines the form and function of the tax revenue gap to shine a lot on the corresponding absence of a tax redistribution gap. A. The Tax Revenue Gap The tax revenue gap, simply defined, is “the difference between tax revenues collected and those that would be theoretically expected to be collected in the absence of any evasion or late payment.”23 Our tax code as written stands to raise a particular amount in revenue—for tax year 2021, taxpayers owed an estimated $4.57 trillion dollars.24 But when the IRS totaled its tax receipts, taxpayers had only paid approximately $3.88 trillion.25 The difference—about $688 billion, or 15% of expected tax revenues—makes up the “gross tax [revenue] gap,” i.e. “the difference between the aggregate tax liability imposed by law for a given tax year and the amount of tax that taxpayers pay voluntarily and timely for that year.”26 Of course, some taxpayers voluntarily pay their liabilities late, and IRS enforcement efforts recoup an additional sum. These amounts reduce the gross tax revenue gap to result in the “net tax [revenue] gap.”27 For 2021, the IRS estimated 19 Paul Zak, Measurement Myopia, THE DRUCKER INST. (Jul. 4, 2013), https://drucker.institute/thedx/measurement-myopia/ [https://perma.cc/H6X5-TPHL]. 20 Mark Aleynick et al., Fitting the Arts into Today’s Public Education System, RUTGERS UNIV. COMM. REPOSITORY, 2 (2012). 21 See Amy Finn et al., Cognitive Skills, Student Achievement Tests, and Schools, 25(3) PSYCH. SCI. 736, 742 (2014). [changed Phsychol. To Physch. Per T6] 22 EILÍS LAWLOR ET AL., NEW ECONOMICS FOUND., SEVEN PRINCIPLES FOR MEASURING WHAT MATTERS: A GUIDE TO EFFECTIVE PUBLIC POLICY-MAKING 2 (2009). 23 Norman Gemmell & John Hasseldine, The Tax Gap: A Methodological Review, 20 ADVANCES IN TAX’N 203, 204 (2013). 24 MELANIE KRAUSE ET AL., IRS RESEARCH, APPLIED ANALYTICS & STATISTICS, FEDERAL TAX COMPLIANCE RESEARCH: TAX GAP PROJECTIONS FOR TAX YEARS 2020 & 2021 8 (OCT. 2023). 25 Id. 26 JAMES M. BICKLEY, CONG. RSCH. SERV., RL33882, TAX GAP AND TAX ENFORCEMENT 1 (FEB. 16, 2007). 27 Id. https://drucker.institute/thedx/measurement-myopia/ 92 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 that an extra $63 billion per year would come in through late payments and IRS enforcement, making the net tax revenue gap $625 billion.28 Over the two and a half decades following the introduction of the tax revenue gap measure in 1964, the IRS issued three additional studies—in 1979, 1983, and 1988—estimating its magnitude.29 Policy recommendations to address the tax revenue gap quickly followed: in a March 1988 report to the Senate Budget Committee, General Government Division Associate Director Jennie Stathis noted that “IRS’ estimates show the tax [revenue] gap remains substantial” and “[o]ur recommended improvements to several IRS enforcement programs should enable IRS to better identify unreported income and collect more taxes.”30 Similarly, law review articles on the tax revenue gap appeared as early as 1989.31 The American Bar Association devoted resources to addressing the tax revenue gap with a July 1987 report.32 Today, the IRS dedicates a webpage and numerous publications to information about the tax revenue gap, including annual projections and backward- looking estimates.33 The tax revenue gap is also subject to contemporary political attention—for example, President Biden aimed to reduce the gross tax gap by 10% over the next decade as part of his 2021 American Families Plan Tax Compliance Agenda.34 Similarly, the idea regularly makes headlines in major newspapers.35 1. Tax Revenue Gap Form What makes up the tax revenue gap? The IRS slices it in two ways, by type of tax and type of noncompliance. The total tax liability comes from the five 28 KRAUSE ET AL., supra note 24. These numbers mean that any reduction in the tax revenue gap would have a large impact: in 2019, the Government Accountability Office noted that “a 1 percent reduction of the 2008-2010 average annual net tax [revenue] gap . . . could fund about 82 percent of IRS’s enforcement budget.” See JAMES R. MCTIGUE, JR., U.S. GOV’T ACCOUNTABILITY OFF., GAO- 19-558T, TAX GAP: MULTIPLE STRATEGIES ARE NEEDED TO REDUCE NONCOMPLIANCE 2 (May 9, 2019). 29 Gross Income Tax Gap Estimates: Hearing before the S. Budget Comm. to Discuss Estimates by the Internal Revenue Service of the Size, Scope, and Composition of the Income Tax Gap, 100th Cong. 88-22 (1988) (statement of Jennie S. Stathis, Associate Director, General Government Division). 30 Id. at 2. 31 Tom Weiksnar & Todd Van Valkenburg, Attacking the Tax Gap, 6 AKRON TAX J. 165, 166 (1989) (“The tax [revenue] gap should be more than a passing statement. It should be a major issue to the American people.”). 32 Id. 33 IRS: The tax gap, INTERNAL REVENUE SERV., https://www.irs.gov/statistics/irs-the-tax-gap (last updated Oct. 10, 2024) [https://perma.cc/C5HH-65FF]. 34 U.S. DEP’T OF THE TREASURY, THE AMERICAN FAMILIES PLAN TAX COMPLIANCE AGENDA 13 (2021). 35 See Ashlea Ebeling, Americans Failed to Pay a Record $688 Billion in Taxes. The IRS Says That Will Change, WALL ST. J. (Oct. 12, 2023), https://www.wsj.com/personal-finance/taxes/americans- failed-to-pay-a-record-688-billion-in-taxes-the-irs-says-that-will-change-631ce518; see also Alan Rappeport, Tax cheats cost the U.S. $1 trillion per year, I.R.S. chief says., N.Y. TIMES (Apr. 13, 2021), https://www.nytimes.com/2021/04/13/business/irs-tax-gap.html; Lawrence Summers & Natasha Sarin, The IRS is leaving billions on the table. Here’s how it can collect that money., WASH. POST (June 22, 2020), https://www.washingtonpost.com/opinions/2020/06/22/how-irs-could-fix- tax-gap/. https://www.irs.gov/statistics/irs-the-tax-gap https://www.wsj.com/personal-finance/taxes/americans-failed-to-pay-a-record-688-billion-in-taxes-the-irs-says-that-will-change-631ce518 https://www.wsj.com/personal-finance/taxes/americans-failed-to-pay-a-record-688-billion-in-taxes-the-irs-says-that-will-change-631ce518 https://www.washingtonpost.com/opinions/2020/06/22/how-irs-could-fix-tax-gap/ https://www.washingtonpost.com/opinions/2020/06/22/how-irs-could-fix-tax-gap/ 2025] THE TAX REDISTRIBUTION GAP 93 component taxes enforced by the IRS: the individual income tax, corporate income tax, employment tax, estate tax, and excise taxes.36 In turn, the IRS breaks down noncompliance for each tax into three categories: nonfiling (cases in which a taxpayer does not file a return at all), underreporting (when a taxpayer files a return but understates their tax liability), and underpayment (representing cases where a taxpayer files an accurate return but does not pay their entire liability).37 (The tax revenue gap does not include missing tax revenues from unlawful acts, whether because the government would rather eliminate than tax illegal economic activities or because estimating such lost revenue would be an impossible task.)38 Of these pieces, underreporting comprises the largest portion of the overall gap—80%—with individual income tax underreporting making up the majority of that slice (56%).39 (For tax year 2021, $396 billion of the total $688 billion gross tax revenue gap resulted from individual underreporting.40) Within individual income tax underreporting, sole proprietor underreporting—that is, understatement of tax liability by individuals who own an unincorporated business—presents the most significant problem: GAO data estimates that 57% of sole proprietor income is misreported, compared to only 1% of wages and 18% of capital gains.41 Broken down by filers, the GAO found that 30% of sole proprietors misreported gross business receipts, and just over three-quarters (76%) misreported business expenses.42 The IRS attributes $80 billion of the gross tax revenue gap to sole proprietor income.43 2. Measuring the Tax Revenue Gap The IRS derived early tax revenue gap estimates using the Taxpayer Compliance Measurement Program, through which IRS agents performed “line-by- line examinations of several different types of tax returns” in order to generate estimated underpayments across a variety of tax categories.44 In the 1990s, Congress objected to the IRS method due to “the high cost to the IRS and the compliance burden placed on taxpayers who were selected.”45 Rather than abandon the enterprise of measuring the tax revenue gap, the IRS introduced the National Research Program. By only directly auditing items that could not be independently verified, the IRS sought to minimize the burden to taxpayers.46 36 KRAUSE ET AL., supra note 24, at 13. 37 Id. at 9. 38 Mark Mazur & Alan Plumley, Understanding the Tax Gap, 60 NAT. TAX J. 569 (2007). 39 U.S. GOV’T ACCOUNTABILITY OFF., GAO 23-106448, TAX GAP: MODEST REDUCTIONS IN THE GAP COULD YIELD LARGE FISCAL BENEFITS 1 (2023). 40 KRAUSE et al., supra note 24, at 10. 41 U.S. GOV’T ACCOUNTABILITY OFF., GAO 24-105281, SOLE PROPRIETOR COMPLIANCE: TREASURY AND IRS HAVE OPPORTUNITIES TO REDUCE THE TAX GAP 21 (2023). 42 Id. at 22. 43 Id. at 1. 44 BICKLEY, supra note 26, at 1. 45 Id. 46 Id. at 2. 94 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 Daniel Hemel et al. explain the tax revenue gap measurement method in more detail.47 The National Research Program relies on individual IRS examiners detecting underreporting and underpayments in individual returns. Examiners auditing similar tax issues with the same propensity for underreporting identify varying levels of tax noncompliance; the IRS assumes that the most “successful” examiners are also the most accurate, and it scales underreporting identified by less successful examiners to that level through a process called “Detection Controlled Estimation.”48 One study of the NRP and DCE estimates that DCE adjustments triple the amount of suspected underreporting.49 Using this new method, the IRS most recently generated tax revenue gap estimates in 2023 for tax years 2020 and 2021.50 As noted above, the gross tax revenue gap for tax year 2021 is $688 billion, a $87 billion increase over tax year 2020’s $601 billion gap.51 Indeed, the tax revenue gap is growing in absolute value over time—in 2016, the IRS estimated a gross tax revenue gap of $458 billion per year for tax years 2008 to 2010, while a 2022 study estimated an annual gap of $496 billion for 2014 to 2016.52 But the tax revenue gap remains stable in relative terms, at approximately 15% of the total estimated tax burden.53 3. Tax Revenue Gap Functions Why does the tax revenue gap matter? Its relative components can tell us about breakdowns in the tax process. For example, the prevalence of sole proprietor underreporting demonstrates a primary root of the tax revenue gap: information asymmetry between taxpayers and the IRS. As Leandra Lederman explains, “the taxpayer knows the facts regarding the relevant transactions he or she engaged in during the tax year—or at least has ready access to that information. The government is forced to obtain that information after the fact, either from the taxpayer or from third parties.”54 When the government does not have a way to access that information—as is the case with sole proprietorships—tax compliance becomes a question of taxpayer honesty and understanding tempered by the risk of an audit. Unsurprisingly, compliance rates are proportionate to the level of information the government receives: taxpayers report 98.8% of income subject to 47 Hemel et al., supra note 6, at 3-4. 48 Id. at 7. 49 Id. at 8. 50 KRAUSE ET AL., supra note 24, at 4. 51 Id. at 7. 52 MCTIGUE, supra note 28, at 2; U.S. GOV’T ACCOUNTABILITY OFF., GAO 24-105281, SOLE PROPRIETOR COMPLIANCE: TREASURY AND IRS HAVE OPPORTUNITIES TO REDUCE THE TAX GAP 1 (2023). 53 Demian Brady & Andrew Wilford, Minding the Gap: Recommendations for Assessing, Addressing, and Ameliorating the Tax Gap, NATIONAL TAXPAYERS UNION FOUNDATION 4 (May 17, 2024), https://www.ntu.org/foundation/detail/minding-the-gap-recommendations-for- assessing-addressing-and-ameliorating-the-tax-gap [https://perma.cc/V37J-BJR3]. See also MCTIGUE, supra note 28, at 16 (explaining that despite IRS efforts to “improv[e] services to taxpayers, and enhanc[e] enforcement of the tax laws. . . the percentage at which taxpayers pay their taxes voluntarily and on time has remained relatively constant over the past three decades.”). 54 Leandra Lederman, Reducing Information Gaps to Reduce the Tax Gap: When is Information Reporting Warranted?, 78 FORDHAM L. REV. 1733, 1735 (2010). https://www.ntu.org/foundation/detail/minding-the-gap-recommendations-for-assessing-addressing-and-ameliorating-the-tax-gap https://www.ntu.org/foundation/detail/minding-the-gap-recommendations-for-assessing-addressing-and-ameliorating-the-tax-gap 2025] THE TAX REDISTRIBUTION GAP 95 withholding (such as wages and salaries), 95.5% of income subject to third-party information reporting but not withholding (like interest and dividends), 91.4% of income subject to partial reporting (capital gains), but only 46.1% of income subject to neither reporting nor withholding.55 But information asymmetry is not the sole cause of the tax revenue gap. The GAO reports that reduced audit rates, low levels of taxpayer service, the complexity of the tax code, and opportunities to create tax shelters also contribute to missing revenue.56 The relationship between the magnitude of these factors and the pieces of the tax revenue gap can offer policymakers nuanced information about the overall efficiency of the tax system. Our understanding of these intersecting causes has been consistent since initial tax revenue gap studies. Summarizing the 1987 American Bar Association Report, Weiksnar and Van Valkenberg noted the primacy of information reporting: recognizing that “[t]he opportunity to underreport exists when the IRS asks a taxpayer to supply information or record imputed transactions, like depreciation,” they concluded that “[i]ncreased third-party reporting is the most promising method available to increase compliance.”57 They also recognized other contributing factors, including negative public opinion towards taxes, frequent tax reform leading to public misunderstanding, and overall complexity of the tax code.58 Importantly, achieving a particular tax revenue gap is an imperfect policy goal because tax revenue gap empirics will always be imprecise—they are based on noncompliance, which is both difficult to observe and subject to ambiguity given the complexity of the tax code.59 Moreover, the lag time between a given year and IRS analysis of that year’s tax revenue gap means that “the tax [revenue] gap will reflect tax policy and administration from several years in the past.”60 But the value of tax revenue gap analysis is not in a perfectly calculated number—achieving one would be impossible, leading some commentators to argue that “[t]he tax [revenue] gap should not be a performance target.”61 Instead, understanding the tax revenue gap’s scale is enough understand its harms.62 Overall, the tax revenue gap harms equity, trust, and efficiency. It harms equity because tax noncompliance affects compliant taxpayers: “noncompliance can create an unfair competitive advantage among businesses because those that do not pay tax debts are avoiding costs that tax-compliant businesses are incurring.”63 This creates real unfairness that undermines both horizontal and vertical equity 55 Id. at 1738. 56 U.S. GOV’T ACCOUNTABILITY OFF., supra note 39, at 2-3. 57 Weiksnar & Van Valkenburg, supra note 31, at 173. 58 Id. at 174. 59 Hemel et al., supra note 6, at 5. 60 Id. at 25. 61 Id. at 4. 62 Neil Warren, Estimating Tax Gap is Everything to an informed response to the Digital Era, UNSW BUSINESS SCHOOL TAXATION AND BUSINESS LAW, 10 (Apr. 5, 2018), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3200838 [https://perma.cc/H7SL-JEMJ] (writing that the tax gap “is not just about establishing a single number or range of estimates of tax gap—[i]t is about understanding the nature, drivers and incidence of non-compliance behavior as reflected in tax gap estimates, which can help guide the best responses to improve compliance.”). 63 MCTIGUE, supra note 28, at 8. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3200838 96 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 (taxpayers with equal incomes will pay different amounts in tax, while some taxpayers with more income will pay relatively less).64 By consequence, the tax revenue gap erodes public trust in the tax system.65 And, as Logue and Vettori explain, it also creates economic inefficiency: “people and resources will tend to move into the areas of the economy and forms of organization with respect to which noncompliance is easiest and thus taxes are lowest.”66 These harms implicate a broad range of stakeholders: from the Treasury (focused on tax policy design); to the IRS (focused on tax enforcement); to taxpayers and voters (concerned with tax fairness); to politicians (concerned with good government); to statisticians (concerned with data reliability).67 The harms also dictate possible solutions: responding to the asymmetry of information between sole proprietors and the government, Logue and Vettori propose implementing a presumptive tax, under which small business owners would be required to report gross receipts but instead of allowing business expense deductions, the IRS would impute expenses via “presumed profit percentages based on historical average profit ratios within the taxpayer’s line of business.”68 Addressing the underreporting that results from the use of either unskilled or corrupt tax preparers, the GAO counsels Congress to provide the “IRS with the authority to regulate paid tax return preparers.”69 Given the attention the tax revenue gap receives, it is worth exploring why we do not study or discuss the parallel slippages in implementation of our redistributive provisions. B. The Absence of a Tax Redistribution Gap As noted above, taxation’s redistributive function receives some degree of government study. As part of its broader study of household income distributions, the Congressional Budget Office estimates the distributive impact of the tax code as a whole, with calculations going back to 1979.70 Similarly, the Treasury Department’s Office of Tax Analysis calculates the distributive impact of both federal taxes and proposed changes to tax law.71 And upon request from members of Congress, the Joint Committee on Taxation will analyze the distributional effects of proposed tax amendments.72 Private entities also monitor distribution: like the Joint Committee on Taxation, the Tax Policy Center issues distributional analyses 64 Kyle D. Logue & Gustavo G. Vettori, Narrowing the Tax Gap through Presumptive Taxation, 2 COLUM. J. TAX L. 100, 113 (2011). 65 MCTIGUE, supra note 28, at 8. 66 Logue & Vettori, supra note 64, at 114. 67 Warren, supra note 62, at 27. 68 Logue & Vettori, supra note 64, at 129. 69 MCTIGUE, supra note 28, at 18. 70 In 1975, Michael Graetz bemoaned that scholarly advances in the study of distributive impact of taxation had “not been combined to evaluate distributional consequences of changes in the personal income tax.” Michael J. Graetz, Assessing the Distributional Effects of Income Tax Revision: Some Lessons from Incidence Analysis, 4 J. LEGAL STUD. 351, 351 (1975). 71 Office of Tax Analysis, U.S. DEP’T OF THE TREASURY, https://home.treasury.gov/policy- issues/tax-policy/office-of-tax-analysis [https://perma.cc/5L6J-U8YM]. 72 Revenue Estimating, JOINT COMM. ON TAX’N, https://www.jct.gov/operations/revenue- estimating/ [https://perma.cc/FF8V-P7MA]. https://home.treasury.gov/policy-issues/tax-policy/office-of-tax-analysis https://home.treasury.gov/policy-issues/tax-policy/office-of-tax-analysis https://www.jct.gov/operations/revenue-estimating/ https://www.jct.gov/operations/revenue-estimating/ 2025] THE TAX REDISTRIBUTION GAP 97 of tax reforms, and in 1995, the American Enterprise Institute released a 300-page report on distributional analysis of tax policy.73 1. Shortcomings of Distributive Analysis With respect to redistribution, each of these distributive measures suffers from both a definitional and a completeness problem. As Ari Glogower explains, each metric uses “market” income as a baseline against which it measures distribution by comparing changes to income caused by government taxes and spending.74 (For example, according to the Congressional Budget Office, in 2009 the federal tax system lowered societal inequality compared to pre-tax levels by about 8.5%.75) Murphy and Nagel explain that the idea of a “pre-tax” distribution is oxymoronic: “Private property is a legal convention, defined in part by the tax system; therefore, the tax system cannot be evaluated by looking at its impact on private property, conceived as something that has independent existence and validity.”76 Without the societal systems and structures created by government because of tax revenues, none of us would have the capacity to earn an income. (Hobbes wrote that our lives in such circumstances would be “solitary, poore, nasty, brutish, and short.”77) Simply comparing pre- and post-tax market income erases the portion of an individual’s tax burden that equals the suite of public goods they consume, or, in other words, the indirect benefits they receive from government spending. Consider a two-person society consisting of A and B, in which person A has a market income of 10,000 a year and person B 2,000. A pays 3,000 in taxation, while B receives 200 in tax credits. Distributional analysis would conclude that the tax system has a net distributional impact of 3,200—A lost 3,000 vis-à-vis their market income and B gained 200. But this analysis ignores that the government uses its net 2,800 in tax revenue to fund public goods that both A and B consume. Assume that A benefits the equivalent of 2,400 from public goods and B 600. Then, A has only lost 600 from the entire tax and spending scheme, while B has gained 800, for a net redistributive swing of 1,400.78 This redistributive picture looks very different from the distributive effects of the same system: measuring redistribution 73 Measuring the Distribution of Tax Changes, TAX POL’Y CTR., https://www.taxpolicycenter.org/resources/measuring-distribution-tax-changes [https://perma.cc/QM2A-WMWE]; DAVID E. BRADFORD, AM. ENTER. INST., DISTRIBUTIONAL ANALYSIS OF TAX POLICY, (1995). 74 Ari Glogower, A Basic Needs Baseline for Distributional Analysis, 48 BYU L. REV. 1697, 1703- 04 (2023). 75 David Kamin, Reducing Poverty, Not Inequality: What Changes in the Tax System Can Achieve, 66 TAX L. REV. 593, 595 (2013). 76 LIAM MURPHY & THOMAS NAGEL, THE MYTH OF OWNERSHIP 8 (1st ed. 2002). Glogower describes the critique another way: a distributional study seeking to evaluate the net distributional effects of government policies necessarily begins with a baseline of antecedent market income that already reflects the effects of government policies. Glogower, supra note 74, at 1739-40. 77 THOMAS HOBBES, LEVIATHAN, Ch. XIII (St. Paul’s Churchyard, Andrew Crooke 1651), https://www.gutenberg.org/files/3207/3207-h/3207-h.htm [https://perma.cc/B8UZ-VVVE]. 78 Law and economics scholars disagree over the proper proxy for aggregate benefits received from public goods. See infra Part III.A.3. https://www.taxpolicycenter.org/resources/measuring-distribution-tax-changes 98 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 without taking into account the benefits provided by public goods would be to significantly overstate the redistribution provided by the wealthy. Even correcting for this definitional error, none of the above studies combine prospective and retrospective analysis to tell us whether a particular reform or the system as a whole moved the needle as much as it should have. When we measure the distributive impact of the tax system as a whole, we should also measure the potential impact of the tax system as written to learn the ways in which our implementation falls short of our actual policy. When we project the impact of a proposed tax reform, we should follow up with studies of its real-world impact to assess whether the reform achieved its projected goals. Without both pieces, we are only getting half the story. 2. Understanding Redistributive Avoidance Despite redistribution being a core function of taxation, it may be that there is no collective appetite for redistribution discourse beyond distributive analysis. Distributive analysis simply measures the impact of the tax system as it exists; explicitly measuring redistribution would forefront the government’s role in tackling inequality through taxation. We explore two possible explanations for why the government might shy away from doing so: either the public does not want redistribution and government follows suit in policy and priorities, or the government deprioritizes redistribution and public opinion follows. The standard intuition, under what is known as “the workhorse political economy model,” is that public demand for redistribution increases as inequality increases in society. But real-world surveys show the opposite: as inequality has increased in the United States, public support for the idea that the government should use taxation to reduce income differences has remained flat or declined slightly.79 Jonathan Hopkin offers one explanation rooted in the political system of the United States.80 Unlike many other OECD countries which allow for proportional representation across multiple parties, the U.S. two-party system forces voters to make a binary choice. A median voter who might benefit from redistribution from the wealthy may still vote against a leftist party for fear that progressive policies would focus redistributive gains amongst the poor, leaving middle-class taxpayers with a higher tax burden.81 Alesina et al. describe another based on the psychology of the United States population. According to their research, American taxpayers over-believe in the potential for class mobility relative to its actual likelihood (and their optimism is notably high in states where actual mobility is relatively low).82 This disparity could lead taxpayers to reject government redistribution due to an inflated sense of their own economic potential. 79 Vivekian Ashok et al., Support for Redistribution in an Age of Rising Inequality: New Stylized Facts and Some Tentative Explanations, 2015 BROOKINGS PAPERS ON ECON. ACTIVITY 367, 367. 80 Jonathan Hopkin, The Politics of Tax Justice in Democracies: Redistribution Beyond the Median Voter Theorem, 2 LSE PUB. POL'Y REV. 4 (2022). 81 Id. at 5. 82 Alberto Alesina, et al., Intergenerational Mobility and Preferences for Redistribution, 108 AM. ECON. REV. 521, 522-23 (2018). 2025] THE TAX REDISTRIBUTION GAP 99 Stefanie Stantcheva sees less of a tension between growing inequality and American views towards redistribution. Her survey work shows that “[t]rust in government (or lack thereof) . . . seems to be a critical element in driving support for redistribution.”83 Although 70% of her survey respondents reported believing that “money and wealth in the U.S. should be more evenly distributed,” almost 90% agreed with the statement that “politicians in Washington work to enrich themselves and their largest campaign contributors, instead of working for the benefit of the majority of citizens.”84 Public opposition may be less against redistribution and more against government as its chief implementer. Under any of these explanations, voters might reject strong redistribution by the government, and politicians and policymakers might respond by deprioritizing it. Hopkin describes this as the “dominant view” in political science theory, under which “voter preferences” are the key “drivers of change.”85 On the other hand, politicians and policymakers may be driving redistributive inertia. Political science research shows that public opinion can follow policy changes—known as the “leadership effect”—as opposed to vice- versa.86 Politicians may have reasons for deprioritizing redistribution apart from voter preference, with the result that nothing catalyzes change in popular support for redistributive policies. First, redistribution is exceptionally polarized. The 70% of survey respondents who believed that wealth should be more evenly distributed in the United States are split across partisan lines, with 92% of Democrats and 42% of Republicans agreeing.87 (When asked if inequality was a serious or very serious issue, 69% of Democrats but only 25% of Republicans agreed.)88 Politicians seeking to ride the middle may be hesitant to center a divisive issue. Still, a majority of Stantcheva’s respondents agreed that inequality is too high; a second explanation for the leadership effect may be that popular opinion is underrepresented in policymaking. Hopkin explains that as politics has become dominated by party elites, the influence of the wealthy over party platforms has surged.89 The sway of the wealthy could operate as a dampening effect on redistributive policy and discourse. A third reason for a lack of government attention to tax-based redistribution separate from public opinion is that the combination of taxation and progressive social policy can feel like a misnomer. German political scientist Florian Fastenrath explains that left wing parties tend to focus on spending policy to carry out their 83 Stefanie Stantcheva, Why People Vote Against Redistributive Policies That Would Benefit Them, MIT PRESS READER (Nov. 20, 2021), https://thereader.mitpress.mit.edu/why-do-we-not-support- redistribution/ [https://perma.cc/69D4-2CTC]. 84 Stefanie Stantcheva, Understanding Tax Policy: How Do People Reason? 136 Q. J. OF ECON. 2309, 2345 (2021). 85 Hopkin, supra note 80, at 10. 86 Kathleen M. McGraw et al., “What They Say or What They Do?” The Impact of Elite Explanation and Policy Outcomes on Public Opinion, 39 AMER. J. POL. SCI. 53, 68 (1995). See also Florian Fastenrath et al., Why is it so Difficult to Tax the Rich? Evidence from German Policy-Makers, 29(5) J. EUR. PUB. POL’Y 767, 769 (2022). 87 Stantcheva, supra note 84, at 2346. 88 Id. 89 Hopkin, supra note 80, at 10-11. https://thereader.mitpress.mit.edu/why-do-we-not-support-redistribution/ https://thereader.mitpress.mit.edu/why-do-we-not-support-redistribution/ 100 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 policy visions, leaving the nuts and bolts of tax policy to deemed experts.90 This can further inflate the wealth effect on politics, as lobbyist groups for corporations and the wealthy can take narrative control of tax policy.91 3. Finding Space for a Tax Redistribution Gap Measure Despite potential public or political reticence to frontline redistribution, the lack of a tax redistribution gap measure is odd for at least two reasons. First, primary theories of distributive justice agree on the importance of redistribution in a well-functioning society. 92 Utilitarians, who believe “that the morally right action is the action that produces the most good,”93 arrive at a redistributive mandate from the concept of diminishing marginal utility, which concludes that “the poorest households, those in abject poverty, will derive much greater utility from an additional $1 of income compared to better-off households, even those that are only slightly better off.”94 Egalitarianism places special concern on the most marginalized in society, and in particular on the difference in resources afforded to those people vis-a-vis their better-off peers; the core intuition that “an unequal distribution of resources is inherently suspect” brings the theory to a redistributive mandate even more straightforwardly than utilitarianism.95 And while libertarianism on its face appears to accept the justness of market distributions, 96 libertarian theorists often call for a minimal social safety net in order to reduce the risk of “unjustified encroachment” towards private property.97 (Indeed, this reasoning infiltrated conservative politics in the 1960’s, with economist Milton Friedman advocating for a negative income tax and President Nixon incorporating it into his Family Assistance Plan.98) If redistribution is part of any distributive justice mandate, it should follow that a government deploying redistribution should want to know whether its policies are operating efficiently and maximally. 90 Fastenrath et al., supra note 86, at 768, 772. 91 Id. at 774. 92 Broadly, contemporary theories of distributive justice fall into three primary categories: utilitarianism, egalitarianism, and libertarianism. See, e.g. Joseph Bankman & Thomas Griffith, Social Welfare and the Rate Structure: A New Look at Progressive Taxation, 75 CAL. L. REV. 1905, 1915-16 (1987) (describing the difference between entitlement and welfarist theories of justice). 93 Julia Driver, The History of Utilitarianism, STANFORD ENCYCLOPEDIA OF PHILOSOPHY (Edward Zalta & Uri Nodelman eds., 2014), https://plato.stanford.edu/entries/utilitarianism-history/ [https://perma.cc/P9K9-HG8M]. 94 Ariel Jurow Kleiman, Low-End Regressivity, 72 TAX L. REV. 101, 131 (2018). See, e.g., Sarah B. Lawsky, On the Edge: Declining Marginal Utility and Tax Policy, 95 MINN. L. REV. 904 (2011). 95 Kleiman, Id. at 110. 96 Julian Lamont and Christi Favor, Distributive Justice, STANFORD ENCYCLOPEDIA OF PHILOSOPHY (Edward Zalta ed., 2017), at https://plato.stanford.edu/entries/justice-distributive/. 97 Daniel Hemel & Miranda Perry Fleischer, Atlas Nods: The Libertarian Case for a Basic Income, 2017 WIS. L. REV 1189, 1210 (2017) (citing Eric Mack, Non-Absolute Rights and Libertarian Taxation, 23 SOC. PHIL. & POL’Y 109, 125, 140 (2006)). Hemel and Miranda Fleischer push libertarian redistribution further by arguing that it is necessary to justify market outcomes in the first instance. Id. at 1219. 98 Id. at 1198. https://plato.stanford.edu/entries/utilitarianism-history/ https://plato.stanford.edu/entries/justice-distributive/ 2025] THE TAX REDISTRIBUTION GAP 101 Second, and unsurprisingly given the first, the tax code already deploys significant amounts of redistribution, both through direct cash transfers and through its progressive curve.99 As Blum and Kalven, Jr. point out, “[t]here are not so many methods of accomplishing peaceful redistribution.”100 For many economists and legal scholars asking how a government interested in pursuing rich-to-poor distribution might utilize its power to do so, the answer is simple: taxation. Welfare economics theory places redistribution squarely in the realm of tax, with other laws organized around principles of efficiency.101 This is because taxation’s inherent relationship to distribution gives it an edge over other potential methods for redistribution. Louis Kaplow and Steven Shavell provide the primary analysis: “[U]sing legal rules to redistribute income distorts work incentives fully as much as the income tax system—because the distortion is caused by the redistribution itself—and also creates inefficiencies in the activities regulated by the legal rules.”102 Based on this concept of “double-distortion,” they suggest that every non-tax redistributive is dominated by a parallel tax scheme: “[E]ven though the income tax distorts work incentives, any regime with an inefficient legal rule can be replaced by a regime with an efficient legal rule and a modified income tax system designed so that every person is made better off.”103 Kaplow and Shavell also note an additional reason why tax is superior to other legal regimes for effectuating redistribution: “[T]he income tax system (including transfer programs) can redistribute from all the rich to all the poor, whereas legal rules have substantially less redistributive potential.”104 This is for two reasons: first, non-tax legal rules seeking to redistribute will result in contracted 99 See infra Part II.B. 100 Walter Blum & Harry Kalven, Jr., The Uneasy Case for Progressive Taxation, 19 U. CHI. L. REV. 415, 487 (1952). 101 Edward McCaffery & Jonathan Baron, The Political Psychology of Redistribution, 52 UCLA L. REV. 1745 (2005) (primarily citing work by Kaplow and Shavell). See also Jeremy Bearer-Friend et al., Taxation and Law and Political Economy, 83 OHIO ST. L.J. 472, 479 (2022); Kyle Logue & Ronen Avraham, Redistribution Optimally: Of Tax Rules, Legal Rules, and Insurance, 56 TAX L. REV. 157, 158 (2003) (making “the safe bet that a majority of legal economists hold the following view: Whatever amount of redistribution is deemed appropriate or desirable, the exclusive policy tool for redistributing to reduce income or wealth inequality should always be the tax-and-transfer system.”). 102 Louis Kaplow & Steven Shavell, Why the Legal System Is Less Efficient than the Income Tax in Redistributing Income, 23 J. LEGAL STUD. 667, 667-68 (1994). 103 Id. at 669 (emphasis omitted). Kaplow and Shavell’s double-distortion argument has not been immune to scholarly critique. Matthew Dimick traces a number of responses to their analysis, including those of Christine Jolls and Chris Sanchirico. Matthew Dimick, The Law and Economics of Redistribution, 15 ANN. REV. L. & SOC. SCI. 559, 568-73 (2019). Christine Jolls applies behavioral economics insights—namely, that humans respond differently to known tax burdens than to uncertain events through which legal rules might redistribute—to complicate the comparison of distortion from taxes and legal rules. Id. at 568-69. (citing Christine Jolls, Behavioral Economics Analysis of Redistributive Legal Rules, 51 VAND. L. REV. 1653, 1658-63 (1998)). Chris Sanchirico rejected Kaplow and Shavell’s simplifying assumption that humans are homogenous in their precaution-taking strategies, concluding that legal rules that tie damages to income can theoretically work to improve overall welfare. Id. at 569-70 (citing Chris William Sanchirico, Taxes Versus Legal Rules as Instruments for Equity: A More Equitable View, 29 J. LEGAL STUD., 797 (2000)). But as Dimick concludes, “[t]hese responses appear to have had only limited success in challenging the argument’s sturdy reputation.” Dimick, Id. at 560. 104 Kaplow & Shavell, supra note 102, at 674. 102 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 parties adjusting the price of transactions to reflect cost of legal rules, thereby capturing or circumventing the redistributive aims of the law; and second, redistribution through legal rules is limited to those who become parties to lawsuits.105 Zach Liscow further explains that taxation brings an efficiency advantage over redistributive alternatives because cash redistribution promotes the autonomy of redistributive recipients and minimizes the burden on the rich from whom redistribution is extracted.106 The primacy of redistributive taxation makes sense given tax’s function and history. As Murphy and Nagel explain, taxation is the primary vehicle through which the government determines “how the social product is shared out among different individuals, both in the form of private property and in the form of publicly provided benefits.”107 That is, the government sets both the tax rates—which determines “how much of a society’s resources will come under the control of government”—and how to spend the resulting revenue “in accordance with some collective decision procedure.”108 These two pieces primarily dictate levels of societal redistribution. Indeed, Blum and Kalven, Jr., in their canonical analysis of progressive taxation, connect this role to taxation’s inherent redistributive potential.109 Despite questioning many of the common justifications for progressive tax schemes, they note that taxation has the undisputed ability to redress inequality: “However uncertain other aspects of progression may be, there is one thing about it that is certain. A progressive tax on income necessarily operates to lessen the inequalities in the distribution of that income.”110 Still, there is nuance to be had. Despite the strong arguments in favor of tax- based redistribution, Logue and Avraham leave space for redistribution to occur outside of taxation under specific conditions. While they conclude that “[i]ncome redistribution may be best accomplished primarily through the tax-and-transfer system (with some possible room for a supplementary lump sum tort tax),” they argue that “there are nonincome measures of inequality with respect to which the legal system should play an essential redistributive role.”111 Liscow agrees: he writes that “features other than income are often desirable bases for redistributing income, and legal rules may be institutionally better equipped than taxes—or the only option—to redistribute based on such non-income features.”112 But the recognition that taxation need not be the exclusive site for governmental redistribution does not lessen the claim that it serves as its primary site. Avi-Yonah adds historical context to the above structural point by locating redistributive intent in the earliest U.S. income tax schemes. Tracing the post-Civil- 105 Id. at 674-75. 106 Zachary Liscow, Redistribution for Realists, 107 IOWA L. REV. 495, 506 (2022). 107 MURPHY & NAGEL, supra note 76, at 76. 108 Id. 109 Blum & Kalven, supra note 100, at 486-88. 110 Id. at 486. 111 Logue & Avraham, supra note 101, at 169. They point to insurance as one such alternative redistributive site, noting that a “nondiscrimination rule” in health insurance coverage might be the most efficient way to redistribute with respect to inequality of lifetime health costs flowing from differences in genetic makeup. Id. at 226. 112 Zachary Liscow, Reducing Inequality on the Cheap: When Legal Rule Design Should Incorporate Equity as Well as Efficiency, 123 YALE L. J. 2478, 2482 (2014). 2025] THE TAX REDISTRIBUTION GAP 103 War shifts from a mostly agrarian society to “one dominated by large industrial corporations,” he explains that “lawmakers of both parties” found existing tax regimes—consisting of federal consumption taxes and state property taxes— insufficient to combat “a sharp rise in inequality.”113 The result was a “focused and sustained effort to enact a federal income tax on both individuals and corporations, as well as an estate tax,” ultimately resulting in the 16th Amendment and the foundation for the income tax scheme most familiar to Americans today.114 Although the early income tax applied to only a fraction of the population (the National Archives reports that less than one percent of the population had income tax liability in 1913115), Avi-Yonah emphasizes that “a primary goal of the income tax historically was seen as redistributing wealth from the rich to everyone else.”116 Anne Alstott and Ben Novick find another data point of redistributive tax policy beginning in 1924, when the government kept taxation high after World War I in part to provide benefits to veterans. They write, “The financial cost of the war itself and the nation’s lingering obligations to veterans kept the federal government’s revenue needs higher than otherwise and helped build taxing capacity that would be critical to Thirties and Forties efforts to build the welfare state and to finance the second World War.”117 Particularly interesting is that these moves towards redistribution come despite the historical tendency, noted by Blum and Kalven, Jr., “of separating progression from equalitarianism and soft-peddling its redistributive effects, [which] probably served an important function in keeping progression a respectable idea in our society.”118 It seems intuitive that a government would want to know whether its redistributive tools are having the intended effect. Indeed, the tax revenue gap framework invites the parallel concept of a tax redistribution gap. A 2011 statement by the GAO Director of Strategic Tax Issues invokes redistribution in passing: [I]n 2001 taxpayers underreported $6.3 billion in net income due to misreported Individual Retirement Arrangement (IRA distributions). But taxpayers also may underclaim benefits to which they are entitled. . . . [O]f tax filers who appeared to be eligible for a higher-education tax credit or tuition deduction in tax year 2005, about 19 percent . . . failed to claim any of them.119 Australian tax scholar Neil Warren proposes tax redistribution gap analysis more directly: “there is no reason why tax [revenue] gap studies cannot be expanded beyond taxes by a revenue administration if it was also administering negative taxes 113 Avi-Yonah, supra note 4, at 11. 114 Id. 115 16th Amendment to the U.S. Constitution: Federal Income Tax (1913), NATIONAL ARCHIVES: MILESTONE DOCUMENTS, https://www.archives.gov/milestone-documents/16th-amendment [https://perma.cc/U8R9-UG62]. 116 Avi-Yonah, supra note 4, at 12. 117 Anne Alstott and Ben Novick, War, Taxes, and Income Redistribution in the Twenties, 59 TAX L. REV. 373, 438 (2006). 118 Blum & Kalven, supra note 100, at 486-87. 119 U.S. GOV’T ACCOUNTABILITY OFF., GAO-11-747T TAX GAP: COMPLEXITY AND TAXPAYER COMPLIANCE 1 (2011). https://www.archives.gov/milestone-documents/16th-amendment 104 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 (such as tax credits and subsidies) or social (income-contingent) transfer programs.”120 With a tax redistribution gap measure, the IRS would not be introducing new redistribution—it would instead be evaluating current policy. That alone could exempt introduction of the measure from public opinion or political concerns around redistribution. Nor is the IRS a stranger to new tax measures. Before the 1960’s, the IRS did not measure the tax revenue gap; that figure is now an enshrined piece of tax policy analysis. Similarly, Stanley Surrey coined the concept of tax expenditures—a tax provisions that carve out from the tax base economic activity that would otherwise fall under accepted concepts of net income— in the 1960’s121; by 1974, Congress required the President’s annual budget to include an account of all tax expenditures in the code.122 If the IRS or another federal agency were to introduce a tax redistribution gap measure, how would it do so? The next two parts of this article seek to answer that question. II. BUILDING THE TAX REDISTRIBUTION GAP MEASURE In this second section, we start to build a framework under which the IRS might begin to measure the tax redistribution gap measure. We begin by defining redistribution. A. What is Redistribution? To measure something, we must first define it, and so we start with theory: what does it mean to redistribute? Colloquially, we think of Robin Hood, stealing from the Sherwood Forest rich to provide for the less fortunate.123 But Robin Hood’s antics are a kind of private redistribution, by which the people take the allocation of resources into their own hands to correct perceived injustices flowing from government action. Here, we are concerned with redistribution that results from, and not in response to, governmental decisions. While redistribution by government—that is, state-sanctioned, legal reapportionment of resources— operates differently than renegade takings, both forms get at the issues at the heart of this article: “what constitutes a just society, what citizens owe to each other, and what limitations on liberty are acceptable.”124 Philosopher Christian Barry defines four parameters indicative of governmental redistribution: 120 Warren, supra note 62, at 5. 121 STANLEY S. SURREY, THE UNITED STATES INCOME TAX SYSTEM—THE NEED FOR A FULL ACCOUNTING, in TAX POLICY AND TAX REFORM: 1961–1969, 573 (William F. Hellmuth & Oliver Oldman eds., 1973). 122 STANLEY S. SURREY & PAUL R. MCDANIEL, TAX EXPENDITURES 31 (1985). 123 HOWARD PYLE, THE MERRY ADVENTURES OF ROBIN HOOD (1883). 124 Alex Raskolnikov, Accepting the Limits of Tax Law and Economics, 98 CORNELL L. REV. 523, 562 (2013). 2025] THE TAX REDISTRIBUTION GAP 105 (1) The subjects, such as individual persons or rigidly and non- rigidly defined groups whose holdings of goods are modified through the redistribution; (2) The baseline, the initial distribution of goods to which some other distribution is seen as a redistributive modification; (3) The social mechanism, such as a change in tax laws, monetary policies, or tort law, that engenders the redistribution of goods among these subjects; and (4) the goods, such as income and property (or perhaps opportunities and liberties), that are redistributed through this mechanism.125 1. Redistributive Subjects In terms of redistributive subjects, Barry’s definition is multidirectional and vast. That is, it encompasses redistribution from both rich to poor subjects and poor (or middle-class) to rich, in addition to redistribution along other lines (such as identity markers like race or gender and social groups like parents or veterans). Indeed, “[v]irtually any legislation has some redistributive effect to the extent that it imposes on an individual taxpayer a personal cost in excess of the value that the taxpayer receives from the expenditure” 126 or if “a subgroup of residents receives benefits that are substantially disproportionate to the related . . . costs that the same subgroup bears.”127 In these terms, Clayton Gillette equates “municipal funding of a homeless shelter” to “municipal funding from general tax revenues of a public golf course frequented by a small percentage of the local population” as redistributive enterprises.128 This comparison makes clear that (nearly) everything can be redistribution. Gillette argues that “[a]ttempts to provide a more precise definition of redistribution are inherently problematic,”129 but too broad of a definition threatens to swallow the concept’s usefulness. Accepting the broad umbrella of redistribution, we focus in this article on just the redistribution centered in distributive justice theory: redistribution down the income spectrum.130 2. Redistributive Mechanisms and Goods This is an article about IRS measures, and so unsurprisingly we focus on taxation as our redistributive mechanism.131 This is no accident given the general 125 Christian Barry, Redistribution, THE STANFORD ENCYCLOPEDIA OF PHILOSOPHY (Edward N. Zalta ed., 2018), https://plato.stanford.edu/archives/spr2018/entries/redistribution [https://perma.cc/R78Z-V6C6]. 126 CLAYTON P. GILLETTE, LOCAL REDISTRIBUTION AND LOCAL DEMOCRACY 53-54 (2011). 127 Id. at 57. 128 Id. at 54. 129 Id. 130 Indeed, redistribution in tax-policy terms takes on the colloquial meaning—that from rich to poor. See, e.g., Kaplow & Shavell, supra note 102, at 667. 131 While tax-based redistribution is coined “direct” redistribution, governments can also impact the distribution of welfare across society through “indirect” redistribution such as regulating the provision of goods and services (for example, minimum wage laws or anti-discrimination provisions). Emiliano Huet-Vaughn, Minimum Wages and Voting: Assessing the https://plato.stanford.edu/archives/spr2018/entries/redistribution 106 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 consensus that taxation is the government’s most efficient redistributive tool. Accordingly, the salient “good” is money. But there is one additional gloss to add before we move on: why limit our analysis to federal tax redistribution? The answer is relatively uncontroversial. Citing a litany of economists, Bahl et al. write that central governments should bear responsibility for redistribution because 1. the benefits of redistribution are not limited to the geographic boundaries of states and localities and 2. redistribution at a lower level of government can be compromised by in- and out-migration of people seeking to benefit from or escape redistributive taxation.132 In 2018, Stephen Calabrese et al. substantiated this intuition by modeling the relationship between taxation and majority rule at the local and federal level.133 As preface, they quote 20th century American economist George Stigler, who argued that “redistribution is intrinsically a national policy” because “if 99 communities tax the rich to aid the poor, the rich may congregate in the hundredth community, so this uncooperative community sets the tune.”134 Richard Musgrave reiterated the point: “policies to adjust the distribution of income among individuals must be conducted on a nationwide basis. . . . regional differentiation leads to severe locational inefficiencies. Moreover, regional measures are self-defeating, as the rich will leave and the poor will move to the more egalitarian-minded jurisdictions.”135 Calabrese et al.’s model confirmed Stigler and Musgrave’s intuition but on slightly different grounds.136 They concluded that local taxes for redistribution do not arise in part because the difference between the median and mean income of local communities tends to be small, such that it is difficult for voting blocks of poorer constituents to move for significant redistribution.137 Out-migration to escape redistribution is unnecessary because redistribution is unlikely to arise in the first instance.138 Consistent with this analysis, the overwhelming effect of state taxation is regressive. The Institute on Taxation and Economic Policy recently released its 7th edition of a 50-state survey of state tax schemes.139 Despite a post-COVID trend of Political Returns to Redistribution Outside the Tax System 1 (IZA – Inst. Lab. Econ., Discussion Paper No. 16416, 2023). 132 ROY BAHL ET AL., FISCAL RESEARCH CENTER REPORT NO. 49: STATE AND LOCAL GOVERNMENT CHOICES IN FISCAL REDISTRIBUTION 1 (Ga. St. Univ. Andrew Young Sch. Pol’y Stud. 2000). 133 Stephen Calabrese et al., Majority Choice of Taxation and Redistribution in a Federation, 217 J. PUB. ECON. 104782 (2023). 134 Id. at 2 n.7. 135 Id. 136 Id. at 2. 137 Id. 138 Id. 139 INST. ON TAX’N AND ECON. POL’Y, WHO PAYS? A DISTRIBUTIONAL ANALYSIS OF THE TAX SYSTEMS IN ALL 50 STATES (7th ed. 2024). 2025] THE TAX REDISTRIBUTION GAP 107 states strengthening refundable tax credits,140 the report concludes that forty-four states’ tax systems exacerbate income inequality.141 Focusing on federal taxation as the government’s primary tool to affect the redistribution of resources across the population, the final definitional question becomes against what baseline, other than pre-market income, ought we measure redistribution? 3. Redistributive Baselines Applying his definitional framing to the tax-and-transfer system, Barry writes that to know whether redistributive taxation has occurred, we must identify: (1) the extent of the benefits enjoyed by different people within a social system (or the costs that they have imposed on others); (2) the costs of providing these benefits or averting imposed costs; and (3) the contribution of each person to the provision of social benefits and compensation for costs imposed. Redistributive tax-and-transfer occurs whenever people have paid taxes that are above and beyond what is required to cover the costs of the public benefits that they have received and the costs they have imposed on others.142 Baked into this definition is the concept of a redistributional baseline as the point at which someone’s tax liability exactly equals the net value they receive from the suite of public goods funded by general tax revenues.143 Jim Hines, addressing redistribution from an economic rather than philosophical lens, arrives at the same baseline. He writes that when taxes are set to equal the value individuals receive from public goods, “[a]ny deviations [from this baseline] . . . can then be interpreted as net redistributions due to the fiscal system inclusive of both taxes and expenditures.”144 While using relative benefits to determine the shape of a tax scheme has fallen out of favor,145 it can still provide a benchmark for measuring redistribution. But as Blum and Kalven, Jr. wrote in the early 1950’s, “the difficulties of isolating and measuring particular benefits are for the most part insurmountable.”146 140 Samantha Waxman, Record Number of States Create or Improve EITCs to Respond to COVID- 19, CTR. ON BUDGET & POL’Y PRIORITIES BLOG (Aug. 11, 2021, 4:35 pm), https://www.cbpp.org/blog/record-number-of-states-create-or-improve-eitcs-to-respond-to-covid- 19. 141 INST. ON TAX’N AND ECON. POL’Y, supra note 139, at 10. 142 Barry, supra note 125, at 8. 143 Of note: Barry describes redistribution from the rich (whose taxes might exceed the value of public benefits received); similarly, redistributive taxation to the poor occurs when people either pay less in taxes than the benefits they receive or have a negative tax liability outright. But either calculation avoids the problems, noted earlier, of using pre-tax market income as a baseline. 144 James R. Hines Jr., What is Benefit Taxation?, 75 J. PUB. ECON. 483, 484 (2000). 145 See Ira K. Lindsay, Benefits Theories of Tax Fairness, 9 STUD. IN TAX L. HIST. 93, 98-101 (2019). 146 Blum & Kalven, supra note 100, at 455. 108 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 Indeed, as Robert Scherf and Matthew Weinzierl note, “an individual’s benefit from-and thus willingness to pay for-public goods is private information.”147 Like much of law and economics analysis, deploying a redistributive baseline requires identifying an appropriate proxy. Economics supports the use of either a proportional or regressive proxy. Scherf and Weinzierl survey four competing models to measure relative benefits from public goods: Erik Lindahl’s 1919 model, which ties benefits from public goods to an individual’s marginal willingness to pay; Geoffrey Brennan’s from 1976, which challenges Lindahl’s approach by arguing for uniform pricing of public goods; Hervé Moulin’s from 1987, which extends Lindahl’s marginal valuation approach to calculate inframarginal valuations—that is, the sum of benefits an individual receives from all state activity; and Jim Hines’ at the turn of the millennium, which draws from Brennan’s approach to identify the point at which individuals would be equally well off if they were to purchase public goods at a common price from the private market.148 Ultimately, the models differ in how to conceptualize “benefits”—as Scherf and Weinzierl explain, “Moulin’s approach measures benefit relative to a setting in which the public goods in question are not provided; Hines’s approach measures benefit relative to a setting in which the individual’s preferred level of public goods is provided.”149 These differing definitions of benefits result in different solutions to the proxy problem: Lindahl’s and Moulin’s models result in a proportional tax of about 20%, while Brennan’s and Hines’ result in regressive taxes. We need not definitively resolve this debate here.150 For our purposes, it is enough to assume that for wealthy taxpayers, a redistributive baseline (the benefits 147 Robert Scherf & Matthew Weinzierl, Understanding Different Approaches to Benefit-Based Taxation 2 (Nat’l Bureau of Econ. Rsch., Working Paper No. 26276, 2019). 148 Id. at 3-7. 149 Id. at 7. 150 Indeed, theoretical arguments support both potential proxies. Hayek writes what Barbara Fried calls the “classic argument” for arriving at a proportionate tax scheme from benefits analysis: “‘[A] person who commands more of the resources of society will also gain proportionately more from what the government has contributed’ to the provision of those resources, and taxation ought to be levied in proportion to the benefits so provided.” Barbara Fried, The Puzzling Case for Proportionate Taxation, 2 CHAP. L. REV. 157, 160 (1999). Fried, however, dismantles the idea of proportional taxation as an appropriate proxy for benefits from public goods. To her, Hayek’s claims do not stand up to serious scrutiny because individuals do not consume proportionately more public goods as incomes increase (she cites “clean air, defense, and broadcast spectra” as collectively shared public goods), nor can we establish that the wealthy derive additional utility from public goods as opposed to “social opportunities, talents, and hard work.” Id. at 166. Fried concludes that tying taxation to benefits is likely to result in a regressive scheme. Id. Still, Avi-Yonah, citing Graetz, articulates another theoretical justification for measuring benefits proportionately to income: “[A]ll income-generating activities depend on . . . a variety of services provided by the government,” such that “all income-generating activities can be conceived as a partnership between individuals and the government, and taxation can be justified as the government receiving its share of partnership income”—in other words, higher income should be subject to higher taxation. Reuven S. Avi-Yonah, Why Tax the Rich? - Efficiency, Equity, and Progressive Taxation, 111 YALE L.J. 1391, 1404 (2002). Graetz’s framing echoes Adam Smith’s arguments from centuries earlier: much like Nagel and Murphy today, Smith recognized that individuals have the capacity to earn income because they exist ‘“under the protection of the state,”’ and as such a proportional tax scheme is most appropriate. Matthew Weinzierl, Revisiting the Classical View of Benefit-Based Taxation 2 2025] THE TAX REDISTRIBUTION GAP 109 they receive from public goods) will be lower than either our tax system as written or as implemented. Whether a proportional or regressive scheme best approximates benefits from public goods, the baseline will cancel out when we calculate the tax redistribution gap: that is, if total potential redistribution equals the space between the total tax burden under our tax system as written and a total benefits from public goods according to a redistributive baseline (PR = TW – RB), and actual redistribution equals the space between the actual tax burden and a redistributive baseline (AR= TI – RB), then the tax redistribution gap—the space between total potential redistribution and actual redistribution—equals the space between our tax system as written and our tax system as implemented (RG = PR – AR = (TW – RB) – (TI – RB) = TW – TI). B. Defining Redistributive Taxation According to our working definition, redistribution occurs when an individual’s tax burden varies from the redistributive baseline of benefits they receive from public goods. Despite the multiplicity of tax provisions, only two redistributive mechanisms exist: first, provisions that directly transfer money to the poor via taxation (“direct transfers”151), and second, provisions that shift the relative burden of funding public goods away from a redistributive baseline (“differential contributions”152). 1. Direct Transfers Direct transfers are the simpler of the two in that they are easier to observe and measure. They occur when a taxpayer ends up with a negative effective tax rate—that is, when an individual taxpayer receives more money from the government than they pay in taxes such that the tax transaction between the taxpayer and the government flows in the opposite direction than usual. Also referred to as “pure resource distribution,” direct transfers can “be implemented by a substantial personal tax exemption, by a negative income tax (or earned income tax credit), by wage subsidies, by family allowances, or by a sizable demogrant.”153 While much (though not most) of the federal tax code has an anti-poverty focus, most anti-poverty tax provisions do not replicate the function of a direct transfer.154 The only mechanism through which the current tax code achieves an actual cash transfer is a refundable credit—defined in lay terms by the IRS as “a credit you can get as a refund even if you don't owe any tax.”155 (Nat’l Bureau of Econ. Rsch., Working Paper No. 20735, 2014). I plan to revisit the question of redistributive baselines in future work. 151 MURPHY & NAGEL, supra note 76, at 89. 152 Id. 153 Id. at 92. 154 In 2014, Susannah Camic Tahk catalogued the extensive number of tax provisions that have been passed since the 1990s with the goal of alleviating poverty. Susannah Camic Tahk, The Tax War on Poverty, 56 ARIZ. L. REV. 791 (2014). 155 Refundable tax credits, I.R.S. (Jan. 17, 2025), https://www.irs.gov/credits- deductions/individuals/refundable-tax-credits [https://perma.cc/Y4Y6-HMHC]. 110 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 Our tax code deploys four refundable tax credits that work as cash transfers to different degrees: the Earned Income Tax Credit, the Child Tax Credit, the Premium Tax Credit, and the American Opportunity Tax Credit. As Tahk explains, “[t]he EITC provides a cash subsidy to low-income families in proportion to their earned income up to a certain limit, above which the credit phases out.”156 Today, the EITC represents the seventh-largest tax expenditure in the tax code, distributing an estimated $67 billion for fiscal year 2024 according to the Department of the Treasury.157 The Child Tax Credit (CTC) similarly aims to support taxpayers with children. Like the EITC, the CTC requires a taxpayer to have earned income and varies with number of children.158 However, the CTC is only partially refundable.159 In fiscal year 2024, CTC distributions are estimated to total $45 billion.160 Passed as part of the Affordable Care Act, the Premium Assistance Tax credit operates to help defray the cost of marketplace health insurance.161 The credit operates to fix health insurance premiums at a percentage of a taxpayer’s income (ranging from 2% to 9.5%, with higher incomes resulting in higher percentages).162 In fiscal year 2024, distributions are estimated to reach over $66 billion.163 Finally, the American Opportunity Tax Credit (AOTC) provides up to $2,500 against qualified higher education expenses.164 Like the CTC, the AOTC is only partially refundable (up to 40% of the credit, or $1,000).165 For fiscal year 2024, the AOTC is estimated to distribute $2.6 billion.166 Still, tax direct transfers do not capture the full picture of direct transfers in the United States. In addition to tax credits, both Temporary Assistance for Needy Families (TANF) and Social Security Income provide cash support (TANF for low- income families, and Social Security for low-income individuals who are aged, blind, or disabled).167 The federal government also administers several in-kind direct transfers, such as the Supplemental Nutrition Assistance Program (food 156 Tahk, supra note 154, at 798. 157 OFFICE OF MGMT. & BUDGET, ANALYTICAL PERSPECTIVES: BUDGET OF THE UNITED STATES GOVERNMENT FISCAL YEAR 2024 tbl.20–4 (2023), https://www.whitehouse.gov/wp- content/uploads/2024/03/ap_20_tables_fy2025.xlsx [https://web.archive.org/web/20240320140215/https://www.whitehouse.gov/wp- content/uploads/2024/03/ap_20_tables_fy2025.xlsx] [hereinafter OFFICE OF MGMT. & BUDGET, ANALYTICAL PERSPECTIVES]. 158 See I.R.C. § 24. 159 Id. § 24(d). 160 OFFICE OF MGMT. & BUDGET, ANALYTICAL PERSPECTIVES, supra note 157, tbl.20–4. 161 See I.R.C. § 36B; The Premium Tax Credit – The Basics, I.R.S (Sept. 13, 2024), https://www.irs.gov/affordable-care-act/individuals-and-families/the-premium-tax-credit-the- basics [https://perma.cc/7ZNG-CWJY]. 162 I.R.C. § 36B. 163 OFFICE OF MGMT. & BUDGET, ANALYTICAL PERSPECTIVES, supra note 157, tbl.20–4. 164 See I.R.C. § 25A(b). 165 I.R.C. § 25A(i). 166 OFFICE OF MGMT. & BUDGET, ANALYTICAL PERSPECTIVES, supra note 157, tbl.20–4. 167 Ariel Jurow Kleiman, Inequality of Deservingness, 23 J. CONTEMP. LEGAL ISSUES 235, 241-42 (2022). 2025] THE TAX REDISTRIBUTION GAP 111 stamps), the Special Supplemental Nutrition Program for Women, Infants & Children (WIC), and Medicaid.168 These programs vary in size,169 but they all do important redistributive work and could easily be included in a broader analysis of federal direct transfers. As Ganz and Brill explain, “If a federal program that primarily benefits low-income households is included in the tax-and-transfer system, it will make the system appear more progressive. If it is excluded, the system will appear less progressive.”170 But for our analysis, non-tax direct transfers are more appropriately considered as adjusting the redistributive baseline of low-income taxpayers. When a poor individual receives a non-tax direct transfer, the total benefits they receive from public goods increases, raising their redistributive baseline.171 As noted above, the redistributive baselines cancel out when calculating the tax redistribution gap. The size and efficiency of these non-tax programs is of course a critical piece of understanding national redistribution as a whole, but they remain conceptually distinct from our present object of study—redistribution via the tax code. 2. Differential Contributions Differential contributions result from the fact that deviations from a benefits tax scheme result in individuals paying more or less in taxation than a proportional or regressive baseline. The shape of the tax curve and associated deductions dictate the separate price of aggregate public goods for each taxpayer: more progressive curves and fewer top-end deductions will result in high earners paying disproportionately more for public goods and, all other things being equal, lead to more redistribution.172 Individuals who pay a higher average tax rate than a particular redistributive baseline can be said to have contributed to redistribution, while individuals paying 168 Id. at 242. 169 In fiscal year 2022, federal and state expenditures for TANF totaled $31.3 billion. See Office of Family Assistance, U.S. Dep’t of Health & Hum. Servs., TANF and MOE Spending and Transfers by Activity, FY 2022 (Jan. 22, 2024), https://www.acf.hhs.gov/ofa/data/tanf-and-moe-spending-and- transfers-activity-fy-2022 [https://perma.cc/UFR5-BW67]. Federal costs for WIC totaled $6.6 billion in fiscal year 2023, while SNAP spending reached $112.8 billion. JORDAN W. JONES & SAIED TOOSSI, ECON. RSCH. SERV., U.S. DEP’T OF AGRIC., THE FOOD AND NUTRITION ASSISTANCE LANDSCAPE: FISCAL YEAR 2023 ANNUAL REPORT 5 tbl.1 (2024). Social Security and Medicaid eclipse all other programs. For fiscal year 2023, Social Security payments totaled $1.237 trillion. Office of Retirement and Disability Policy, Soc. Sec. Admin., Summary: Actuarial Status of the Social Security Trust Funds (May 2024), https://www.ssa.gov/policy/trust-funds-summary.pdf [https://perma.cc/7BKM-YRYR]. Medicaid spending reached $1,029.8 billion in 2023. NHE Fact Sheet, Centers for Medicare and Medicaid Services (last modified Dec. 18, 2024), https://www.cms.gov/data-research/statistics-trends-and-reports/national-health-expenditure- data/nhe-fact-sheet [https://perma.cc/9BMK-9NK4]. 170 Alex Brill & Scott Ganz, Progressivity, Redistribution, and Inequality, NAT’L AFFS., Summer 2020, at 74, 76. 171 Tax direct transfers could also be conceptualized under the umbrella of differential contributions in that they push an individual’s tax liability further below the redistributive tax baseline. We separate them out because of the significant role refundable credits play in our tax scheme. 172 Of course, all other things are not equal. As discussed in the next section, progressive curves come with efficiency tradeoffs given the elasticity of labor and income. 112 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 lower rates (or receiving transfers) have received redistribution. This is not to say that progressivity always equates with redistribution: Splinter explains that the two concepts are distinct because while “progressivity is independent of the tax level, redistribution changes with the tax level.”173 Splinter is right conceptually, but at any fixed level of tax revenue, higher progressivity will by definition result in more redistribution. Unlike the number of direct transfers, the list of tax provisions affecting differential contributions is extensive—in large part due to the extreme complexity of the U.S. tax code. Of course, a baseline level of differential contributions is set by our tax rates and their deviation from a regressive or proportional redistributive baseline. The United States income tax curve is—and has always been— progressive. Still, “[d]espite the ambiguity of economic theory, public policy over the last three decades has steadily moved toward lower marginal tax rates on high earners.”174 As former Secretary of Labor Robert Reich noted on X, top marginal tax rates peaked at 91% in 1960 and dwindled to 35% by 2010.175 But tax rates alone do not define differential contributions; tax provisions that adjust the rate structure further adjust relative tax liabilities. In 2014, Tahk catalogued the provisions that “directly subsidize low-income individuals.”176 Moreover, direct transfer provisions also impact differential contributions by reducing average tax rates for low-income taxpayers even if they do not lead to a cash refund.177 As Kleiman notes, citing Congressional Budget Office analysis, “the EITC has brought average tax rates down for taxpayers in the first income quintile, and the child tax credit has done the same for those in the second, third, 173 David Splinter, U.S. Tax Progressivity and Redistribution, 73 NAT. TAX J. 1005, 1007 (2020). 174 N. Gregory Mankiw et al., Optimal Taxation in Theory and Practice, 23 J. ECON. PERSPS. 147, 154 (2009). 175 Robert Reich (@RBReich), X (Apr. 29, 2023, 8:15 PM), https://x.com/RBReich/status/1652466548202766336 [https://perma.cc/MJ86-PH6C]. Most recently, the top rate increased to 39.6% in 2012, but it dropped to 37% in 2018. History of Federal Income Tax Rates: 1913 – 2024, Bradford Tax Institute (last visited June 28, 2024), https://bradfordtaxinstitute.com/Free_Resources/Federal-Income-Tax-Rates.aspx [https://perma.cc/3EQN-3C7H]. 176 Tahk, supra note 154, at 793. Tax provisions such as personal exemptions (I.R.C. §§ 151(b); 151(d)(3)(A)), dependency exemptions (I.R.C. § 151(c)), and the standard deduction (I.R.C. § 63(c)(1)) reduce a taxpayer’s liability; if a provision pushes liability below the redistributive baseline, it creates redistribution towards that taxpayer. 177 The breakdown between direct transfers and differential contributions from refundable credits depends on the fraction of eligible claimants who do not have tax liabilities. For fiscal year 2024, EITC outlays are estimated at $67 billion, whereas revenue lost (that is, total EITC credits reducing tax liability but not resulting in a refund for the taxpayer) is estimated at only $3 billion. OFFICE OF TAX ANALYSIS, U.S. DEP’T OF THE TREASURY, TAX EXPENDITURES FY2025 25 tbl.1, 37 tbl.5 (2024), https://home.treasury.gov/system/files/131/Tax-Expenditures-FY2025.pdf [https://perma.cc/E4Z5-BUS4]. The EITC therefore primarily redistributes through direct transfers as opposed to differential contributions. By contrast, the CTC resulted in foregone tax revenue of almost $64 billion and outlays of $45 billion, making it a provision that predominately results in a differential contribution. Id. The refundable Premium Tax Credit is more similar to the EITC—it is estimated to distribute almost $67 billion but will lead to revenue losses of only $15 billion. Id. The AOTC is grouped with the Lifetime Learning Credit, a nonrefundable credit. Like the CTC, these credits result primarily in differential contributions; together, they are estimated to result in $2.5 billion in outlays and $14 billion in foregone tax revenue. Id. 2025] THE TAX REDISTRIBUTION GAP 113 and fourth income quintiles.”178 Each of these policies push differential contributions towards redistribution. Conversely, the many income tax expenditures targeted towards the higher ends of the income spectrum, such as deductions for charitable contributions and mortgage interest expenditures, steer in the opposite direction. The Treasury Department estimates the deductibility of mortgage interest on owner-occupied homes, I.R.C. § 25, to be the sixth most costly tax expenditure over the next decade.179 Scholars regularly criticize the mortgage interest deduction as regressive.180 Focusing on the individual income tax alone provides an incomplete picture of differential contributions. We have a multi-faceted tax system, with separate tax curves for capital gains, corporate income, and payroll taxes. The impact of these different taxes on redistribution is defined in part by their incidence: the standard intuition is that the corporate income tax is imposed on the corporate entity itself; the incidence of capital gains taxes is mixed, with the Congressional Budget Office and Joint Committee on Taxation assuming “that capital bears three-fourths of the economic burden and workers bear one-fourth;” and the economic consensus is that payroll taxes fall entirely on workers, “regardless of where the statutory burden is imposed.”181 The preferential tax rate for capital gains in particular lowers differential contributions from the wealthy; a 2019 survey by the Tax Policy Center concluded that the top 20% of income earners in the United States held 92% of realized capital gains.182 Ultimately, differential contributions result from the composite of all tax rates and the many tax expenditures that adjust them; the composite of direct transfers and differential contributions sets the overall level of redistribution that results from the tax system.183 Indeed, scholarly literature disagrees as to whether progressive rates alone correlate with redistribution. One camp contends that 178 Kleiman, supra note 94, at 112 (citing CONG. BUDGET OFF., PUB. NO. 53597, THE DISTRIBUTION OF HOUSEHOLD INCOME, 2014 25 (2018); CONG. BUDGET OFF., PUB. NO. 51361, THE DISTRIBUTION OF HOUSEHOLD INCOME AND FEDERAL TAXES, 2013 20 (2016)). 179 OFFICE OF TAX ANALYSIS, U.S. DEP’T OF THE TREASURY, TAX EXPENDITURES FY2025 30-32, tbl.2b (2024), https://home.treasury.gov/system/files/131/Tax-Expenditures-FY2025.pdf [https://perma.cc/J927-ECE6]. 180 See, e.g., Andrew Hanson, Ike Brannon & Zackary Hawley, Rethinking Tax Benefits for Homeowners, NAT’L AFFS., Spring 2014, at 40, 41. 181 Brill & Ganz, supra note 170, at 75. 182 CHUCK MARR, CTR. ON BUDGET & POL’Y PRIORITIES, PROPUBLICA SHOWS HOW LITTLE THE WEALTHIEST PAY IN TAXES: POLICYMAKERS SHOULD RESPOND ACCORDINGLY 9 (2021), https://www.cbpp.org/sites/default/files/7-15-21tax.pdf [https://perma.cc/L5W2-5ML3]. 183 Tahk notes a third category of tax provisions with redistributive potential: “‘indirect’ policies that create incentives for third parties to fulfill certain needs of the poor.” Tahk, supra note 154, at 793. See, e.g., I.R.C. § 42, the Low-Income Housing Tax Credit, which rewards developers for building low-income housing; I.R.C. § 45D, the New Market Tax Credit, which encourages investment in low-income communities; and I.R.C. §§ 51 and 1396, the Work Opportunity and Empowerment Zone Employment Credits, which reward employers for hiring marginalized workers. But the net redistributive impact of these credits is hazy: they indirectly increase the share of goods (housing, economic investment, and jobs) available to low-income individuals, but they directly create a financial boon for developers, investors, and employers. I save redistributive analysis of these provisions for future work. 114 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 “countries that redistribute more have less progressive tax structures.”184 Another concludes “there is a negative relationship between inequality and tax progressivity, and a positive relationship between tax progressivity and redistribution.”185 Despite this disagreement, there is data showing that the U.S. tax system is progressive overall. In 2020, David Splinter, a member of the Congressional Joint Committee on Taxation, compiled six measures of federal tax rates, each of which showed considerable progressivity.186 Splinter further found the tax system, as measured by the Kakwani index,187 had become more progressive overtime: “between 1979 and 2016, . . . the Kakwani index of tax progressivity increased by 46%. Since 1986, tax progressivity has more than doubled.”188 The tax system is also redistributive. Splinter analyzed Congressional Budget Office data to conclude that in 2016, taxes and transfers “more than quadruple[d] the market income of the bottom quintile,” whereas “[h]igher-income groups receive[d] fewer transfers relative to market income.”189 While Splinter uses market income as a redistributive baseline (and therefore likely overstates redistribution relative to a benefits baseline), he still shows an increase in redistribution over time: “Between 1979 and 2016, . . . the [Reynolds-Smolensky] index of tax-and-transfer redistribution increased 59 percent. . . . Since 1986, it increased 66 percent.”190 Splinter succinctly explains that overall progressivity and redistribution do not follow declining top marginal tax rates because only “a small share of tax returns has been subject to top rates.”191 With these pieces in place, we can now illustrate the function of a tax redistribution gap by the identifying key slippages in the implementation of redistributive taxation. III. IMPLEMENTING THE TAX REDISTRIBUTION GAP MEASURE To know whether a system is working as intended, we need to know both the system’s current results and its intended outcomes. Estimates of the tax revenue gap compare two figures: total taxes paid and total taxes owed. Similarly, the tax redistribution gap can be calculated by comparing the total redistribution created under the current system to the maximum redistribution contemplated by statute. To do so, we break total redistribution created into its component parts. 184 Pablo Beramendi et al., Progressive Taxation and Redistribution 4 (Nat’l Rsch. Found. Of Korea, 2020) (working paper), https://daniel- stegmueller.com/papers/BeramendiDimickStegmueller2020.pdf, [https://perma.cc/A3XW-YST9] (emphasis omitted). 185 Id. at 23. 186 Splinter, supra note 173, at 1008 (mapping progressivity according to six different estimates of average federal tax rates). 187 The Kakwani index is the difference between the distribution of income across the population, demarcated by the Lorenz curve, and the distribution of the tax burden across the population, demarcated by a concentration curve. 188 Splinter, supra note 173, at 1005. 189 Id. at 1016 (emphasis omitted). 190 Id. at 1018. 191 Id. at 1020. 2025] THE TAX REDISTRIBUTION GAP 115 Measuring total direct transfers from taxation is as simple as aggregating total IRS outlays from refundable credits. While measuring differential contributions according to our definition is more difficult than totaling direct transfers, we have the tools to do so. As noted above, the Treasury Department releases annual distribution tables showing the relative impact on the federal tax system by income decile. Economists could similarly calculate the proportional or regressive tax rate (according to Moulin’s or Hines’ work, respectively) needed to generate the same revenue levels as the current system, model the resulting tax system, and compare the results to the Treasury reports.192 The more challenging task is to ascertain how much the tax system as written could redistribute. The existence of the tax revenue gap makes clear that some redistribution is left on the table come tax time. According to analysis by the National Bureau of Economic Research, tax evasion is concentrated at the top of the income scale: “under-reported income as a fraction of true income (i.e., income that should legally be reported on tax returns) rises from about 10% in the bottom 90% of the income distribution to 16% in the top 1%.”193 This suggests that the tax revenue gap significantly decreases redistributive differential contributions by lowering average tax rates paid by the wealthy. Closing the tax revenue gap would increase the progressivity of the tax system and increase redistribution; distributional analysis akin to that issued by the Department of Treasury could quantify roughly by how much. While that calculation could help approximate missing differential contributions, tax revenue gap estimates provide a roadmap for measuring missing direct transfers. To construct the tax revenue gap, the IRS estimates total tax liability via the National Research Program by aggregating missing revenue from a sample of tax returns. Warren describes this method as a “bottom-up” approach, by which a revenue administration uses “operational data and available 3rd party data at the taxpayer level which allows summing across the population of taxpayers to obtain national aggregates.”194 The same approach can be used to model the tax redistribution gap from direct transfers. Using the EITC as an example, the rest of this section will explore the resulting contours of the tax redistribution gap for direct transfers. A. The EITC After President Nixon’s Family Assistance Plan failed to garner enough political support, the Earned Income Tax Credit (EITC) emerged as a proposal to minimize the impact of payroll taxes on low-income workers.195 The challenge for its designers was to offset Social Security payments and support working families while still promoting work. As Democratic Senator Russell Long, the credit’s original champion, explained on the Senate floor, the EITC came out of a search 192 Such a model might or might not incorporate shifts in income-earning behavior caused by the difference in rates between a baseline and the progressive reality. 193 John Guyton et al., Tax Evasion at the Top of the Income Distribution: Theory and Evidence 4, (Nat’l Bureau of Econ. Rsch., Working Paper No. 28542, 2023). 194 Warren, supra note 62, at 13. 195 See 118 CONG. REC. 33010 (1972). 116 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 “[t]o find a dignified way to provide help to a low-income working person, whereby the more he works the more he gets, and at the same time to phase it out in such a way as not to decrease the incentive to work.”196 Other scholars aptly trace the credit’s subsequent political history and steady growth197—this article will not retread old ground. The Earned Income Tax Credit, codified in I.R.C. § 32, does exactly what its name describes: it provides a tax benefit to workers in proportion to their earned income.198 The credit phases as a percentage of earned income until it reaches a statutory maximum, which fluctuates with the number of qualifying children the taxpayer has.199 (For tax year 2025, households with three or more children can earn a maximum credit of $8,046; two children correlate with a maximum credit of $7,152; one with $4,328; and childless households can receive a maximum of $649.200) As Goldin explains, up until this point, “the EITC induces negative marginal tax rates, with tax liability declining with each additional dollar earned.”201 Once a taxpayer’s income reaches a set phaseout amount, the amount of credit declines at the statutory phaseout percentage.202 (Married households do not receive greater credits; instead, they have higher phaseout amounts.203) The recipients of EITC redistribution today are the same as at its inception: low-income workers (predominantly those with children). For tax year 2020, approximately 16% of taxpayers filing individual income tax returns claimed the EITC.204 Commentators recognize the EITC as “the federal government’s largest anti-poverty program.”205 Bird-Pollan describes it as “a powerful tool in the U.S. government’s redistributive program” and “one of the largest wealth transfer programs currently operating in the United States.”206 Perhaps unsurprisingly given the scope of the credit, there is much political and scholarly focus on EITC overpayments207—that is to say, instances in which a taxpayer or household claims and receives redistributive credit to which they are not entitled. That focus is not without foundation: the Tax Foundation notes that EITC overpayments have increased in recent years (in parallel with overpayments 196 Id. at 33011. 197 See, e.g., Jennifer Bird-Pollan, Who's Afraid of Redistribution - An Analysis of the Earned Income Tax Credit, 74 MO. L. REV. 251 (2009); Dennis J. Ventry, Jr., The Collision of Tax and Welfare Politics: The Political History of the Earned Income Tax Credit, 1969-99, 53 NAT’L TAX J. 983 (2000). 198 See I.R.C. § 32. 199 MARGOT CRANDALL-HOLLICK ET AL., CONG. RSCH. SERV., R43805, THE EARNED INCOME TAX CREDIT (EITC): HOW IT WORKS AND WHO RECEIVES IT (2023). 200 Rev. Proc. 2024-40, 2024-45 I.R.B. 1100. 201 Jacob Goldin, Tax Benefit Complexity and Take-up: Lessons from the Earned Income Tax Credit, 72 TAX L. REV. 59, 64 (2018). 202 CRANDALL-HOLLICK ET AL., supra note 199, at 5. 203 Id. at 5-6. 204 Id. at 24. 205 Tahk, supra note 154, at 794. 206 Bird-Pollan, supra note 197, at 254, 283. 207 See 2 TAXPAYER ADVOCATE SERVICE, 2018 ANNUAL REPORT TO CONGRESS 93 (2018). 2025] THE TAX REDISTRIBUTION GAP 117 from other refundable credits).208 In fiscal year 2022, almost one third of EITC dollars were overpayments.209 The Department of the Treasury reports quarterly on EITC overpayments as part of a “Payment Integrity Scorecard.”210 But the credit’s breadth and writing on overpayments make it easy to ignore a subtler problem—people who under the terms of the statute qualify for its redistribution but do not, for one reason or another, receive it. As the Taxpayer Advocate Service writes, “the focus on [improper payments] masks both the successes and challenges in improving EITC compliance . . . The improper payment rate does not take into account that for every dollar of EITC improper payments, 40 cents of EITC went unclaimed by taxpayers who appear to be eligible.”211 The existence of unclaimed credits does not take away from or negate the policy problem of EITC overpayments. Instead, it highlights a separate policy issue: the redistribution implementation gap. B. The Implementation Gap Each component of the tax revenue gap—nonfiling, underreporting, and underpaying—maps onto a parallel portion of the tax redistribution gap— nonparticipation, underclaiming, and underreceiving. Using the EITC as an example, we explore the contours of each. 1. Nonparticipation To claim the Earned Income Tax Credit, eligible taxpayers must complete two steps: First, file a tax return, and second, claim the credit. Consequently, eligible taxpayers who do not file a tax return (say, because they have no tax liability for a particular tax year) or file but do not claim the credit will not receive redistribution to which they are otherwise entitled. These taxpayers make up the nonparticipation gap. The IRS tracks EITC participation by state and nationally.212 To calculate rates, the IRS compares Census Bureau data to individual tax records. More specifically, the IRS models the number of eligible EITC claimants using demographic census information and matches it against the number of actual EITC claims filed.213 For tax year 2021 (the most recent year for which the IRS has published its data), 80.8% of eligible taxpayers claimed the EITC.214 The number 208 Scott Hodge, Why Congress Is More to Blame than IRS for $26 Billion in Refundable Tax Credit Overpayments, TAX FOUND. (June 6, 2023), https://taxfoundation.org/blog/irs-tax-credits- overpayments/ [https://perma.cc/SW35-ULKA]. 209 Id. 210 High-Priority Programs, PaymentAccuracy, Off. of Mgmt. & Budget, https://www.paymentaccuracy.gov/payment-accuracy-high-priority-programs/ [https://perma.cc/R5QA-9YTW] (last visited Jan. 22, 2025). 211 TAXPAYER ADVOCATE SERVICE, supra note 207, at 91-92. 212 See EITC participation rate by states: Tax years 2013 through 2020, I.R.S. https://www.eitc.irs.gov/eitc-central/participation-rate-by-state/eitc-participation-rate-by-states [https://perma.cc/55WU-NFS2] (last updated Jan. 17, 2025). 213 Id. 214 Id. 118 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 remained quite stable over the preceding decade, never dropping below the 76.3% and never surpassing this 80.8% figure.215 Detailed polling from tax year 2005 provides a more granular breakdown of eligible taxpayers who did not claim or receive the credit. Taxpayers eligible for an EITC of $1–$99 (according to IRS modeling) had a participation rate of just 42%, while 90% of taxpayers eligible for a credit of more than $4,000 claimed it.216 Unpublished IRS data on file with Goldin suggests a similar nonparticipation gap with respect to total available dollars. According to his analysis, taxpayers claimed 86% of available EITC dollars in 2013.217 Likewise, a report generated by the Treasury Inspector General for Tax Administration (TIGTA) based on 2014 Census data concluded that eligible taxpayers claimed $40.3 billion in EITC credits—85% of the calculated maximum credit amount— leaving $7.3 billion in refunds unclaimed.218 Compare this to the nonfiling tax revenue gap: For tax year 2021, individual taxpayers contributed $67 billion to the tax revenue gap through nonfiling. This sum represents just 2.5% of 2021’s $2.7 trillion individual income tax liability.219 By contrast, the EITC nonparticipation gap of 15% is six times as large a portion of total potential EITC dollars. (Of course, underreporting is a much larger portion of the tax revenue gap, whereas the underclaiming redistribution gap is likely quite small in scale.220) Still, there is a strong argument that the nonparticipation gap is insignificant relative to the scale of the program. No social welfare provision can expect 100% take-up, and other federal welfare programs, such as the Supplemental Nutrition Assistance Program (SNAP), the Supplemental Nutrition Program for Women, Infants, and Children (WIC), and Temporary Assistance for Needy Families (TANF), as well as other federal redistributive credits like the Child Tax Credit, have nonparticipation gaps.221 What contributes to this nonparticipation gap? The Tax Policy Center suggests that factors such as absence of tax liability, lack of knowledge of eligibility, and tax complexity combine to reduce total claims.222 The IRS has identified particular categories of workers who disproportionately underclaim 215 Id. 216 Dean Plueger, Earned Income Tax Credit Participation Rate for Tax Year 2005, in RECENT RESEARCH ON TAX ADMINISTRATION AND COMPLIANCE: SELECTED PAPERS GIVEN AT THE 2009 IRS RESEARCH CONFERENCE 151, 182 (2009). For the most part, participation rates increased consistently between these two poles, with the exception of a disproportionate percentage of taxpayers claiming $500–$599 credits—but that group was underrepresented in the study, which may have created “an overstated participation estimate due to sampling variability. Id. 217 Goldin, supra note 201, at 70. 218 OFF. OF INSPECTIONS & EVALUATIONS, TREASURY INSPECTOR GEN. FOR TAX ADMIN., THE INTERNAL REVENUE SERVICE SHOULD CONSIDER MODIFYING THE FORM 1040 TO INCREASE EARNED INCOME TAX CREDIT PARTICIPATION BY ELIGIBLE TAX FILERS 1-2 (2018). 219 KRAUSE ET AL., supra note 24, at 8. 220 See infra Section III.B.2. 221 Goldin, supra note 201, at 66-67. 222 Briefing Book: Do all people eligible for the EITC participate?, TAX POL’Y CTR., https://taxpolicycenter.org/briefing-book/do-all-people-eligible-eitc-participate [https://perma.cc/A69J-RGD2] (last updated Jan. 2024). 2025] THE TAX REDISTRIBUTION GAP 119 refundable credits: people living in rural areas, self-employed workers, individuals with disability pensions (or caring for children with disability pensions), workers without a qualifying child, non-English speakers, grandparents raising grandchildren, and workers with recent changes to their marital, parental, or employment status.223 One study explains why rural taxpayers may have lower claim rates: “Professional tax preparation services are typically concentrated in poor urban neighborhoods,” and “in rural settings, many low-income families may lack the types of regular social contacts that would lead to awareness of the EITC.”224 Logistical burdens are also higher; rural families face greater time and transportation costs involved with filing returns or receiving preparation assistance.225 Over the two-year study of 314 rural mothers, 89–94% of respondents were eligible to claim the EITC, but only 41–58% claimed it.226 The study also affirmed that non-English speaking taxpayers are less likely to claim the EITC: in the first year of the study, participants interviewed in a language other than English were 97% less likely to claim the EITC.227 The study’s conclusions parallel the 2005 IRS data showing participation increasing in proportion to credit size. Focusing on income, the study found that “for every thousand dollars in earned income, the odds of EITC participation were 6% greater.”228 But even filing a tax return is not a guarantee that an eligible taxpayer will receive the EITC. According to Goldin’s analysis of unpublished IRS data, 8.5% of eligible filers do not attempt to claim the EITC on their returns.229 2. Underclaiming Where the nonparticipation gap includes taxpayers who file but do not claim the EITC, the underclaiming gap consists of taxpayers who claim the EITC but do not claim the full amount to which they are entitled. In this way, the underclaiming gap is the conceptual parallel to the underreporting tax revenue gap. Goldin categorizes EITC complexity into two elements—informational complexity, representing the difficulty of obtaining the necessary information to determine a taxpayer’s eligibility and benefit amount, and computational complexity, the difficulty the taxpayer has in applying the information to actually determine credit eligibility.230 By 2015, only 2% of EITC claimants filed taxes without using some form of assisted preparation method (tax software or a free or 223 Earned Income Tax Credit & Other Refundable Credits, I.R.S., https://www.eitc.irs.gov/ [https://perma.cc/56FE-B2G9] (last visited June 10, 2024). 224 Clinton G. Gudmunson et al., EITC Participation and Association With Financial Distress Among Rural Low-Income Families, 59 INTERDISC. J. OF APPLIED FAM. SCIS. 369, 370 (2010). 225 Id. (citing Donald P. Hirasuna & Thomas F. Stinson, Urban and Rural Differences in Use of Earned Income Credits: A Study of Minnesota’s Working Family Credit, 30 INT’L REG’L SCI. REV. 408, 413 (2007)). 226 Id. at 372, 374. 227 Id. at 375. 228 Id. 229 Goldin, supra note 201 at 71. 230 Id. at 60. 120 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 paid tax preparer).231 As Goldin explains, filing assistance “dramatically reduce[s] the computational complexity associated with claiming a tax benefit,” effectively eliminating the need for taxpayers to perform their own credit calculations and therefore negating the likelihood of credit underclaiming as a result of computational complexity.232 But taxpayers must still determine and provide the relevant information in order for the tax preparation process to properly calculate credit eligibility and amount. Therefore, EITC informational complexity can contribute to the underclaiming gap. Goldin and Kleiman argue that the EITC’s complicated child- claiming systems likely result in some number of eligible children going unclaimed, which would lower or eliminate the credit received by their caretakers.233 This has particularly significant consequences for redistribution, as “[t]he most vulnerable households are hit hardest by the exclusions and complexity that the current child- claiming rules create.”234 3. Underreceiving But what about taxpayers who file a return, claim the full amount to which they are eligible, and do not receive some or all of the credit? The final third of the tax redistribution gap is the underrecieving gap—the conceptual mirror to the tax underpaying gap. This article discusses two contributors to the underreceiving gap: IRS audits and bankruptcy proceedings. As the EITC grew, political concerns over noncompliance and fraud led to a dramatic increase in audits of tax returns claiming the credit.235 Soon, the IRS audited low-income EITC returns more frequently than higher-dollar returns: by 2004, taxpayers claiming the EITC were 1.76 times more likely to be audited than households earning more than $100,000; in 2005, 43% of individual audits targeted taxpayers who claimed the credit.236 High audit rates persisted into recent years: in the decade prior to 2022, EITC claimants have made up 20% of individual taxpayers but were the targets of 40% of individual audits.237 As noted above, these concerns were not motivated by thin air.238 But while EITC audits likely recapture some amount of overpayments, they create a hefty procedural burden and almost certainly freeze or deny refunds to eligible taxpayers. There is also a racial justice element to this IRS practice. In 2023, new analysis of IRS audit rates revealed that Black taxpayers were audited at 2.9 to 4.7 times the 231 Id. at 91. 232 Id. at 92. 233 Jacob Goldin & Ariel Jurow Kleiman, Whose Child Is This? Improving Child-Claiming Rules in Safety-Net Programs, 131 YALE L. J. 1719, 1724 (2022). 234 Id. at 1726. 235 See Jonathan P. Schneller et al., The Earned Income Tax Credit, Low-Income Workers, and the Legal Aid Community, 3 COLUM. J. TAX L. 176, 187 (2012). 236 Id. at 186. 237 MARGOT L. CRANDALL-HOLLICK, CONG. RSCH. SERV., IN11952, AUDITS OF EITC RETURNS: BY THE NUMBERS 3 (2022). 238 See, e.g., Leslie Book, Refund Anticipation Loans and the Tax Gap, 20 STAN. L. & POL'Y REV. 85, 93-94 (2009) (noting that EITC error rates between 25–35% place it above error in other parts of the tax system and well above error in other benefits programs). 2025] THE TAX REDISTRIBUTION GAP 121 rate of non-Black taxpayers and that 78% of the disparity was driven by disproportionate audit rates of Black EITC claimants.239 EITC audits overwhelmingly occur via correspondence, a process by which the IRS requests and receives information by mail.240 Between fiscal years 2008 to 2016, the IRS annually conducted 400,000 to 500,000 correspondence audits of returns claiming the EITC, compared to between 30,000 and 50,000 field audits.241 By fiscal year 2021, the proportion of correspondence audits increased even further: 99.8% of EITC audits closed were correspondence audits, compared to 74% of non- EITC audits.242 The correspondence audit process is difficult for EITC claimants to navigate. As Schneller et al. note, EITC claimants are disproportionately transient or homeless, making communication by mail challenging if not impossible.243 Further, EITC claimants have lower literacy rates and more fear of the government, distrust in or lack of access to financial institutions required to prove credit eligibility, and language barriers.244 These hurdles mean that even when a taxpayer successfully receives a correspondence audit, they often cannot effectively engage with it. A 2007 Taxpayer Advocate Service study found that more than one in four auditees did not understand from the notice that they were under audit, and approximately half did not understand what they were required to do in response.245 Even taxpayers who engaged with the process faced challenges: another National Taxpayer Advocate study found that “more than half of audited EITC claimants reported difficulties in obtaining the requested documents, and nearly half of the same group did not understand why the documents were requested in the first place.”246 Unsurprisingly, logistical and comprehension barriers translate into low success rates. Less than one-third of EITC audits in fiscal year 2018 reached completion.247 43% of audited taxpayers did not respond at all to the initial audit notice, and another 26% responded but did not complete the audit process.248 These low rates are not new: in 2010, only 30% of EITC claimants who received a correspondence audit responded.249 239 HADI ELZAYN ET AL., STANFORD INSTITUTE FOR ECONOMIC POLICY RESEARCH, MEASURING AND MITIGATING RACIAL DISPARITIES IN TAX AUDITS 3-4 (2023). 240 Schneller et al., supra note 235 at 188. 241 KATHLEEN BRYANT ET AL., TAX L. CTR. AT NYU L. & CTR. FOR TAXPAYER RTS., EXCLUSIONARY EFFECTS OF THE IRS CORRESPONDENCE AUDIT PROCESS WARRANT FURTHER STUDY 1 (2022). 242 CRANDALL-HOLLICK, supra note 237, at 4. 243 Schneller et al., supra note 235, at 188. 244 Leslie Book, The IRS’s EITC Compliance Regime: Taxpayers Caught in the Net, 81 OR. L. REV. 351, 393-405 (2002). 245 BRYANT ET AL., supra note 241, at 1 (quoting 1 TAXPAYER ADVOCATE SERVICE, 2018 ANNUAL REPORT TO CONGRESS 137 (2018)). 246 Schneller et al., supra note 235, at 189 (citing 2 TAXPAYER ADVOCATE SERVICE, 2007 ANNUAL REPORT TO CONGRESS 95 (2007)). 247 CRANDALL-HOLLICK, supra note 237, at 5. 248 Id. 249 Taxpayer Advocate Service, IRS Correspondence Examinations: Are they really as effective as the IRS thinks?, NTABLOG (last updated Feb. 8, 2024) at 122 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 Low success rates in turn reduce or remove access to the EITC, even for eligible creditors. If a taxpayer cannot complete the audit process, the IRS will disallow the claimed credit “due to undelivered mail, nonresponse, or insufficient response.”250 But disallowance does not mean the taxpayer was ineligible. The 2004 National Taxpayer Advocate Annual Report recognized that many disallowed tax benefits “are actually correctly claimed.”251 In 2007, 43% of EITC claimants who requested reconsideration of an IRS denial ultimately received the credit at an average of 96% of the initial claimed amount.252 As Schneller et al. write, “[t]hese statistics . . . suggest that the IRS's initial auditing process is only slightly more accurate than a coin toss.”253 To calculate the total amount of the underreceiving gap attributable to EITC audits, the IRS could deploy a process similar to the NRP but with respect to redistribution: that is, it could measure the average amount by which IRS agents improperly disallow EITC claims in the audit process and use those results to model national aggregates. But reducing EITC audits cannot on its own fully close the underreceiving gap because other legal mechanisms work to strip EITC benefits from eligible taxpayers. Specifically, taxpayers entering bankruptcy can see their EITC credits applied to debts as part of the bankruptcy process. When a taxpayer files for bankruptcy, bankruptcy statutes create an estate that consists of “all legal or equitable interests of the debtor in property as of the commencement of the case” for the purpose of paying bankruptcy creditors.254 Recognizing that certain federal benefits exist to support economically insecure debtors, the statute exempts from the bankruptcy estate the “debtor's right to receive . . . a social security benefit, unemployment compensation, or a local public assistance benefit.”255 This general exemption does not apply to tax refunds, nor does it name refundable credits like the EITC. Consequently, it is up to the states, which can either adopt federal exemptions or enumerate their own, to provide protection for EITC amounts.256 Today, most states exempt EITC refunds from the bankruptcy estate because of the EITC’s welfare-adjacent status, but not all do.257 Two recent bankruptcy cases highlight the consequences for EITC recipients facing bankruptcy. https://www.taxpayeradvocate.irs.gov/news/nta-blog/ntablog-irs-correspondence-examinations- are-they-really-as-effective-as-the-irs-thinks/2012/03/ [https://perma.cc/DMU8-2KLU]. 250 BRYANT ET AL., supra note 241, at 3 (quoting Guyton et al., supra note 193, at 3). 251 Statement of Janet Spragens before the IRS Oversight Board (Feb. 1, 2005), at https://www.taxnotes.com/research/federal/other-documents/testimony-other-than-irs-and- treasury/professor-calls-for-focus-on-low-income-taxpayers-before-irs/yj4c [https://perma.cc/Y6UG-S8P4]. 252 Schneller et al., supra note 235, at 188 (citing 1 TAXPAYER ADVOCATE SERVICE, 2009 ANNUAL REPORT TO CONGRESS 160 (2009)). 253 Id. 254 11 U.S.C. § 541(a)(1). 255 11 U.S.C. § 522(d)(10)(A). 256 Timothy M. Todd, Tax Credits in Bankruptcy, THE TAX ADVISER, Jan. 2019, at 75, 76. 257 Keith Fogg, Exempting the Earned Income Tax Credit from the Bankruptcy Estate, TAX NOTES (Jan. 11, 2023), https://www.taxnotes.com/procedurally-taxing/exempting-earned-income-tax- credit-bankruptcy-estate/2023/01/11/7h696 [https://perma.cc/X9HP-PKB5]. 2025] THE TAX REDISTRIBUTION GAP 123 In In re Moreno, the Bankruptcy Court for the Western District of Washington reviewed a claim by a bankruptcy debtor that EITC benefits were exempt from the bankruptcy estate.258 Washington law as interpreted by the Washington Supreme Court exempted public “assistance” from the bankruptcy estate.259 The bankruptcy court turned to Washington law defining “public assistance” as “public aid to persons in need thereof for any cause,” including “federal aid assistance.”260 In turn, Washington law defined “federal aid assistance” as including payments from “a federally administered needs-based program.”261 The question before the court was therefore whether the EITC is a needs-based program.262 The court began its analysis by reviewing treatment of the EITC by other bankruptcy courts. Reviewing caselaw from Idaho, Minnesota, New York, and South Dakota, the court noted that “courts have generally regarded the EITC to be a form of needs-based public assistance,” and that “[i]n cases where the EITC has been found nonexempt under state law, it is generally because the particular statutory scheme explicitly or implicitly gives reason to exclude the EITC.”263 The court declared that “the EITC is specifically aimed at providing monetary assistance to lower-income individuals” and “Washington’s statute offers no reason to exclude the EITC or other federal tax credits.”264 The court then held that the debtor’s EITC was “protected from bankruptcy administration.”265 A Ninth Circuit Bankruptcy Appellate Panel upheld the court’s holding that the EITC is a needs-based program and therefore exempt from the bankruptcy estate.266 In its decision, the court cited Supreme Court precedent recognizing that “[t]he earned-income credit was enacted . . . [in part] to provide relief for low- income families hurt by rising food and energy prices.”267 In a 2022 case in which the debtor also asserted that her EITC refunds were exempt from the bankruptcy estate, the Bankruptcy Court for the District of New Mexico reached the opposite holding.268 Like the Washington court, the court in In re Medina recognized the “anti-poverty” purpose of the EITC.269 But the court emphasized that while many states expressly exempt the EITC from the bankruptcy estate, New Mexico has not, nor does it have a general public assistance exemption.270 Without a statutory hook, the court held that the EITC was not exempted from creditors’ claims.271 258 In re Moreno, 629 B.R. 923 (W.D. Wash. Bankr. 2021). 259 Wash. Rev. Code § 74.04.280 (1959); Anthis v. Copland, 270 P.3d 574 (2012) (referring to assistance protection statutes as “exemption statutes”). 260 629 B.R. at 929 (citing Wash. Rev. Code § 74.04.005(11)). 261 Id. at 930 (citing Wash. Rev. Code § 74.04.005(8)). 262 Id. 263 Id. at 933. 264 Id. at 934. 265 Id. 266 In re Moreno, No. WW-21-1124-LBS, 2021 WL 6140115 (B.A.P. 9th Cir. Dec. 23, 2021). 267 Id. at *4 (emphasis added) (citing Sorenson v. Secretary of Treasury, 475 U.S. 851, 864 (1986)). 268 In re Medina, No. 22-10233-j7, 2022 WL 17742527 (Bankr. D.N.M. Dec. 16, 2022). 269 Id. at *3. 270 Id. at *4. 271 Id. at *6. 124 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 In states like New Mexico that do not exempt the EITC from the bankruptcy estate, low-income taxpayer bankruptcies contribute to the underreceiving gap by allowing redistribution to flow from the government to creditors. Trustees in these states have financial incentive to pursue the EITC: 11 U.S.C. § 326(a) authorizes trustees to earn 25% of assets they administer under $5,000;272 an EITC is easy money in the sense that a trustee does not have to sell it, as they might other bankruptcy assets. This phenomenon is not just hypothetical. As one bankruptcy scholar puts it, “[e]arned-income tax credits . . . are a bit of a thing in consumer bankruptcy at the moment.”273 But without empirical study, it is impossible to compare the underreceiving gap from bankruptcy proceedings to that of IRS audits or to other portions of the tax redistribution gap. These three components—the nonparticipation gap, the underclaiming gap, and the underreceiving gap—comprise the overall redistribution implementation gap with respect to the EITC. Calculating the magnitude of each of these components would allow for an estimate of the overall implementation gap; adding that implementation gap to EITC payments would allow us to measure the total potential redistribution of the EITC as written and approximate the tax redistribution gap. In the final section, we further build out the tax redistribution gap’s conceptual value by mapping proposed tax reforms to their respective tax redistribution gap components. IV. CLOSING THE TAX REDISTRIBUTION GAP Having used the EITC as a focal point, we now have a preliminary framework of the tax redistribution gap. For whom should it matter? Warren lists the many stakeholders of tax revenue gap analysis (the Treasury, the IRS, taxpayers and voters, politicians, and statisticians)274—each of these stakeholders should have an interest in understanding the tax redistribution gap. A clear measure of the tax redistribution gap has both procedural and substantive value: it can highlight whether the tax system is operating as intended, and to the extent it is not, how far short it falls of its redistributive goals. To begin, we address one critique of the redistribution implementation gap: it measures an inevitability. Just as it is not reasonable to expect 100% tax compliance, 100% redistribution implementation is not a feasible goal.275 Still, there is value in understanding how far short of 100% compliance the current system falls and what slippages contribute to its underperformance. Assuming that the tax code as written represents society’s preferred level of redistribution, slippages in implementation demonstrate overall system inefficiency, the degree of which may be more than society deems acceptable. 272 11 U.S.C. § 326(a). 273 E-mail from Robert Lawless, Max L. Rowe Professor of L., Univ. of Ill. Coll. of L., to Eric Baudry, Mich. Fac. Fellow, Univ. of Mich. L. Sch. (May 31, 2024, 12:40 P.M. EST) (on file with author). 274 Warren, supra note 62 at 27. 275 Id. at 20. 2025] THE TAX REDISTRIBUTION GAP 125 And while some amount of tax revenue gap is also inevitable, solutions abound. A 2023 Government Accountability Office report recognized 17 proposed solutions to narrow the tax revenue gap focused on increasing education and outreach to taxpayers, adjusting the IRS’ enforcement approach, and leveraging improved technology to increase government efficiency and efficacy.276 While discrete proposals to address portions of the tax redistribution gap exist, this article aggregates some of the suggested reforms to demonstrate how the concept of the tax redistribution gap allows for a more bird’s-eye view of the status quo. A. Nonparticipation Gap To the extent that much of the nonparticipation gap flows from a lack of taxpayer knowledge and understanding, our primary intuition might be to close the gap by increasing taxpayer education and outreach. Indeed, the IRS, state governments, and nonprofit organizations run extensive awareness campaigns aiming to increase EITC participation rates.277 But as noted below, evidence regarding the efficacy of taxpayer outreach is mixed at best and counsels for alternative solutions to closing the nonparticipation gap. Each year, the IRS sends notice letters to a group of taxpayers identified by computer filters as likely eligible for the EITC reminding them of the credits; the notices also create an alternative pathway to claiming it—if taxpayers complete and return the notice (which contains an eligibility questionnaire), IRS employees can issue a refund.278 But this strategy only marginally affects overall EITC participation: in 2014, these notices prompted 175,000 taxpayers to claim the credit, increasing overall estimated taxpayer and dollar participation rates by less than one percentage point.279 A TIGTA report concluded that these notices “have a limited impact because they are issued to a small percentage of potentially eligible tax filers, and half do not respond.”280 One California study paints a starker picture. The study compared outreach methods to millions of likely eligible EITC recipients. It found that although “[m]ore personalized outreach led to higher levels of engagement with online resources,” none of the outreach methods increased credit take-up—“[r]egardless of the information provided, the framing of the message, or who the messenger was.”281 The study’s authors determined that it was likely no intervention method had an impact on tax filing or credit claiming “larger than a half percentage point.”282 The data pointed to one conclusion: “Outreach alone is likely not enough to improve take-up among non-filing populations.”283 Together, these studies imply that successful efforts to close the nonparticipation gap likely require more than digital or third-party communication 276 U.S. GOV'T ACCOUNTABILITY OFF., supra note 41 at 17. 277 See Goldin, supra note 201 at 72-73. 278 OFF. OF INSPECTIONS & EVALUATIONS, supra note 218, at 4-5, 4 n.11. 279 Id. at 6. 280 Id. 281 ELIZABETH LINOS ET AL., CAL. POL’Y LAB, INCREASING TAKE-UP OF THE EARNED INCOME TAX CREDIT 16, 23 (2020). 282 Id. at 15. 283 Id. at 23. 126 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 with taxpayers. Data suggests that in-person contact is much more impactful. When rural mothers participated in 90–120-minute interviews about tax filing (of which the EITC was only a limited part), they were 21% more likely to claim the EITC.284 Pushing taxpayers to engage with free tax preparation services could also help. A study of an IRS notice pointing taxpayers to such services “yielded only a small (absolute) increase in EITC and CTC participation,” but it confirmed that there is a strong correlation between filing a tax return and claiming the EITC.285 In this vein, Goldin argues for continued expansion of in-person assistance programs like Volunteer Income Tax Assistance (VITA) and Tax Counseling for the Elderly (TCE).286 An alternative would be to reduce or replace the taxpayer’s role in the filing process entirely. Just as much of the tax underreporting gap results from information asymmetry between taxpayers and the government, with taxpayers having income data that the IRS cannot access without third-party reporting, some of the nonparticipation gap flows from low-income taxpayers lacking knowledge and resources to claim the EITC despite the government having most if not all of the information it needs to determine eligibility. The TIGTA study finding that EITC notices were ineffective contended that modifying the basic income tax return forms to include EITC eligibility questions could “improve eligibility determinations and increase participation.”287 In 2020, two United States Senators picked up the call for the IRS to update Form 1040 “to include sufficient information to provide an eligible filer with an automatic refund.”288 Developments in artificial intelligence also enable new possibilities for closing the nonparticipation gap that would not require additional IRS personnel time or changes to paper forms. While programs like TurboTax of tax filing past led users down a pre-defined decision tree based on their inputs, AI tools can dynamically respond to user information in real time. The Taxpayer Advocate Service notes that many tax preparation companies are already deploying AI functionality to improve the filing process, though the Service warns that “AI assistants may encounter difficulties interpreting complex tax laws correctly or considering unique circumstances that could impact a taxpayer’s return.”289 284 Gudmunson et al., supra note 224, at 379. 285 Jacob Goldin et al., Tax Filing and Take-Up: Experimental Evidence on Tax Preparation Outreach and EITC Participation 13 (Nat’l Bureau of Econ. Rsch., Working Paper No. 28398, 2021). 286 See Goldin, supra note 201, at 100. 287 OFF. OF INSPECTIONS & EVALUATIONS, supra note 218, at 9-10 (noting that a revised form could collect residency and age information for a taxpayer’s children). 288 Press Release, Sherrod Brown, Senators Sherrod Brown and Catherine Cortez Masto, Brown, Cortez Masto Mark EITC Awareness Day, as They Press IRS to Ensure All Americans Eligible for EITC Receive It (Jan. 31, 2020), https://www.brown.senate.gov/newsroom/press/release/brown- cortez-masto-mark-eitc-awareness-day-as-they-press-irs-to-ensure-all-americans-eligible-for-eitc- receive-it- [https://web.archive.org/web/20240627114556/https://www.brown.senate.gov/newsroom/press/rel ease/brown-cortez-masto-mark-eitc-awareness-day-as-they-press-irs-to-ensure-all-americans- eligible-for-eitc-receive-it-]. 289 Is AI Generated Tax Advice Making the Grade?, TAXPAYER ADVOC. SERV. (June 11, 2024), https://www.taxpayeradvocate.irs.gov/news/tax-tips/is-ai-generated-tax-advice-making-the- grade/2024/06/ [https://perma.cc/X3KF-Q3SJ]. 2025] THE TAX REDISTRIBUTION GAP 127 The IRS is similarly deploying artificial intelligence models into the NRP process in an attempt to close the tax revenue gap.290 However, the Government Accountability Office reports that the IRS has yet to document either its decision- making process to determine what AI model to run or the technical specifications of its ultimate choice.291 The IRS could engage in a parallel process and utilize AI to better detect eligible taxpayers who failed to claim refundable credits. Agency use of AI is not without its dangers. Danielle Citron and Ryan Calo caution that when agencies replace technical tasks with automation, “guarantees of transparency, accountability, and due process [could] fall away.”292 Other tech scholars flag that “the accountability mechanisms and legal standards that govern decision processes have not kept pace with technology.”293 The IRS’ failure to document the specifics of its AI usage invokes these concerns. Still, governmental use of AI will continue to grow—a 2019 executive order directed every federal agency to explore the potential for AI to increase government efficiency294—and with it comes the potential for innovation in redistribution. B. Underclaiming Gap To the extent that an underclaiming gap flows from informational complexity about direct transfers, its solutions likely mirror those for the nonparticipation gap: connecting taxpayers with in-person support, reducing the taxpayer role in the filing process, or reducing taxpayers’ information costs. But policymakers could also address the underclaiming gap from the policy side, reworking child-claiming rules to reduce complexity. C. Underreceiving Gap Proposed solutions to the underreceiving gap must account for the difficulty taxpayers face understanding and engaging with legal processes. Key to these solutions will be ensuring support throughout the process, not just before it. In 2005 testimony to the IRS Oversight Board, Low-Income Taxpayer Clinic Director Janet Spragens reiterated the importance of sustained communication to positive outcomes for taxpayers: “low income taxpayers need multiple mailings, phone calls and/or other communications or contacts to obtain information, documents and other proof supporting their deductions, credits and other tax return positions.”295 Indeed, a study of IRS EITC audits found that only 38% of taxpayers who received 290 Tax Gap: IRS Should Take Steps to Ensure Continued Improvement in Estimates, U.S. GOV’T ACCOUNTABILITY OFF. (May 6, 2024), https://www.gao.gov/products/gao-24-106449 [https://perma.cc/UK4G-D6QQ]. 291 Id. 292 Ryan Calo & Danielle Keats Citron, The Automated Administrative State: A Crisis of Legitimacy, 70 EMORY L.J. 797, 802 (2021). 293 Joshua A. Kroll et al., Accountable Algorithms, 165 U. PA. L. REV. 633, 636 (2017). 294 Exec. Order No. 13,859, 84 Fed. Reg. 3967 (Feb. 11, 2019). 295 Spragens, supra note 251 (citing 2 NATIONAL TAXPAYER ADVOCATE, I.R.S., EARNED INCOME TAX CREDIT (EITC) AUDIT RECONSIDERATION STUDY (2004)). 128 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 no calls from the IRS were found eligible for the EITC; when taxpayers received three or more phone calls, eligibility rates increased to 67%.296 One method to provide support throughout an EITC audit would be to increase access to low-income taxpayer representation. As I have noted in other work, legal representation for low-income tax issues has historically been a misnomer.297 But as the tax code became a more common vehicle for federal benefits, the need for legal assistance increased. Even so, tax representation today is frequently left out of general legal services.298 The supply of legal assistance for low-income tax issues comes almost entirely from I.R.C. § 7528, which authorized the IRS to create and sustain a grant program for low-income taxpayer clinics. Still, the IRS spends many times more on EITC audits than Congress allocates to Low Income Taxpayer Clinics, leading some to call for increased resources dedicated to EITC representation.299 Access to representation leads to materially better results for EITC claimants. A 2007 Taxpayer Advocate Service (TAS) study found that “low- income filers with representation were twice as likely as their non-represented counterparts to emerge from an IRS audit with no change in their claimed EITC, at rates of 41.5% and 23.1%, respectively.”300 But a 2019 study revealed that 97% of low-income taxpayers do not have professional representation during a correspondence audit.301 Empirical analysis of Tax Court outcomes suggests that the benefits of representation persist for the duration of a tax controversy. Analyzing a random sample of Tax Court cases, Lederman and Hrung found that “taxpayer representation has a significant effect on financial outcome in cases that go to trial” and “the magnitude of the effect increases with the experience of the attorney and remains highly statistically significant.”302 On average, represented taxpayers in tried cases received financial outcomes that were 18% more favorable.303 These statistics suggest that lack of counsel for audited taxpayers leads directly to the underreceiving gap. Indeed, a 2007 TAS study estimated that a lack of representation resulted in pro se taxpayers being denied $300 million in EITC credits for which they were otherwise eligible.304 Apart from ensuring additional support for taxpayers facing EITC audits, the IRS could also close the underreceiving gap by pulling back on its audit 296 Id. (citing 2 NATIONAL TAXPAYER ADVOCATE, I.R.S., EARNED INCOME TAX CREDIT (EITC) AUDIT RECONSIDERATION STUDY 10 (2004)). 297 See Eric Baudry, Pedagogy, Service, Justice, Tax: Reconciling the Tension Between Academic Low-Income Taxpayer Clinics and an Access Rights Theory of Change (Apr. 2, 2024) (unpublished manuscript) (on file with author) (largely citing Keith Fogg, Taxation with Representation: The Creation and Development of Low-Income Taxpayer Clinics, 67 TAX LAW. 3 (2013)). 298 Schneller et al., supra note 235, at 191. 299 E.g., id. at 203-04. 300 BRYANT ET AL., supra note 241, at 5 (quoting Schneller et al., supra note 235, at 192). 301 Id. at 5-6 (citing TAXPAYER ADVOCATE SERVICE, 2021 ANNUAL REPORT TO CONGRESS 294 (2021). 302 Leandra Lederman & Warren B. Hrung, Do Attorneys Do Their Clients Justice? An Empirical Study of Lawyers’ Effects on Tax Court Litigation Outcomes, 41 WAKE FOREST L. REV. 1235, 1281 (2006) (finding that presence of an attorney did not impact outcomes of settled cases). 303 See id. at 1255. 304 Schneller et al., supra note 235, at 208. 2025] THE TAX REDISTRIBUTION GAP 129 practices. Fewer EITC audits would result in fewer eligible taxpayers being denied credits because of information or access barriers (while admittedly reducing one strategy to address EITC overpayments). In September 2023, the IRS announced that it would attempt exactly that by shifting its audit focus in future tax years away from EITC recipients.305 While the change has yet to take full effect, the move is promising in that it addresses the heart of the underreceiving gap. Whereas attorneys may help reduce the underreceiving gap from audits, they are unlikely to help with the gap that results from bankruptcy procedures. Courts interpreting state law regarding bankruptcy exemptions have found little grey area: state law either does or does not exempt the EITC and other public assistance from the bankruptcy estate. In states without an exemption, the routing of EITC refunds to creditors is not a result “for which the court should be blamed.”306 The patchwork legal landscape calls for a straightforward legal reform to close this portion of the underreceiving gap: as Fogg argues directly, “this is an area where Congress should step in. We should not require a state by state determination of whether to exempt federal anti-poverty payments.”307 VI. CONCLUSION Measurements matter. When it comes to taxation, we currently measure both total tax revenue received and the amount missing due to noncompliance or evasion. But when it comes to redistribution, we limit our measures to the overall distributive impact of the tax code; we do not attempt to aggregate the amount of redistribution that goes unclaimed or undelivered. This article coined the tax revenue gap’s mirror image, the tax redistribution gap, and offered an initial framework of how we might begin to measure it. Progressive rates and refundable credits like the Earned Income Tax Credit result in both differential contributions by rich and poor taxpayers and direct transfers to low-income taxpayers. In turn, the tax revenue gap decreases differential contributions from the rich while implementation slippages—in particular, a nonparticipation gap, underclaiming gap, and underreceiving gap—reduce direct transfers received by the poor. Introducing a tax redistribution gap measure challenges background assumptions in current tax discourse: first, it would call out a reliance on pre-market income as a distributive baseline, which serves to overstate the redistributive impact of the tax code; second, it would increase the profile of redistribution among policymakers and the public—understandably, measures like the tax revenue gap and tax expenditure budgets focus dialogue on tax evasion and over-spending by the government. Ultimately, the tax redistribution gap would provide a single measure that displays how we are falling short of a key task of the state (redistribution). And by 305 IRS Announces Sweeping Effort to Restore Fairness to Tax System with Inflation Reduction Act Funding, I.R.S., IR-2023-166 (Sept. 8, 2023), https://www.irs.gov/newsroom/irs-announces- sweeping-effort-to-restore-fairness-to-tax-system-with-inflation-reduction-act-funding-new- compliance-efforts [https://perma.cc/5DTQ-G7HD]. 306 Fogg, supra note 257. 307 Id. 130 COLUMBIA JOURNAL OF TAX LAW [Vol. 16:2 understanding and comparing the component ways our current tax system falls short of its intended outcomes, we can better tailor redistributive policy solutions.