One-size-fits-all taxation fails to accommodate diverse taxpayer circumstances. This Article proposes allowing taxpayers to contract into alternative tax regimes administered by private intermediaries. Participating taxpayers would make payments to the intermediaries pursuant to contract, and the intermediaries would be required to pay to the government at least as much as these taxpayers would have paid the government otherwise. That amount is determined based on the actual tax receipts of a control group, taxpayers who wish to contract with an intermediary but instead are chosen at random to continue under the status quo. These alternative tax regimes might better accommodate taxpayers' preferences, leaving the taxpayers with greater utility without reducing government revenue. An intermediary could offer different substantive law, different procedural rules, or both. Taxpayers, for example, might receive lower tax rates in exchange for forgoing deductions that cause the taxpayer to engage in socially wasteful behavior. Advances in artificial intelligence make contractual tax reform feasible.
INTRODUCTION I. APPLICATIONS A. Changing the Tax Base 1. Expanding the Base 2. Contracting the Base B. Accommodating Other Sources of Heterogeneity 1. Earnings Potential 2. Work Versus Leisure Preferences C. Reducing Waste 1. Saving on Compliance Costs 2. Closing the "Tax Gap" D. Opening the Overton Window II. IMPLEMENTATION A. Basic Mechanics: Treatment Group and Control Group B. The Role of Intermediaries 1. Division of Responsibility 2. Organizational Form 3. Duration 4. Solvency Assurance C. Regulation of Contractual Tax Reform 1. Limitations 2. Filing Status 3. Innovation Incentives III. POTENTIAL OBJECTIONS A. Taxpayer Exploitation 1. Deceptive Practices 2. Privacy Violations 3. Unconstitutional Conditions B. Equality and Equity 1. Worsening Inequality 2. Violating Horizontal Equity CONCLUSION INTRODUCTION
The same body of tax law applies to all taxpayers. (1) This Article proposes upending this bedrock principle, allowing private intermediaries to offer alternative tax regimes. The government would insist that it receive at least as much tax revenue as it would have received under existing tax law, so the intermediaries would have powerful incentives to find packages of tax changes that would benefit individuals without lowering tax collections. For example, in exchange for lower rates, businesspersons might forsake the deduction for business travel (2) or might link their cash register directly into the tax reporting system. (3) Some taxpayers might agree to procedural rules that favor the government, in exchange for smoother tax refund processing. (4) Allowing some parents a deduction for childcare expenses might make them more likely to work outside the home and thus actually increase tax collections. (5) Alternative tax-rate structures with higher inframarginal and lower marginal rates might encourage some taxpayers to work harder, pay more taxes, and yet be happier as a result. (6)
Contractual tax reform requires data about taxpayers that would help predict how much they would pay in an alternative tax regime. Much of the tax literature presumes that taxpayers have private information about themselves. (7) Sometimes, the tax system can harness that information by allowing private parties to choose among different regimes. (8) But for the exchanges suggested above, one cannot allow all taxpayers to opt in, because those most willing to give up benefits would be those least likely to use them. A businessperson who does not travel for business would be happy to give up the deduction for business travel in exchange for a lower rate. (9) Similarly, the goal might be to find a married taxpayer who would reenter the workforce after having children only in the absence of tax distortions. (10) The challenge is to identify groups of taxpayers who can be expected, on average, to pay at least as much tax in the alternative tax regime.
Artificial intelligence (AI) may now make it possible in many cases to identify such taxpayers. Three revolutions--in computing power, in the availability of data, and in the computer algorithms used to analyze the data (11)--mean that computers can increasingly predict human behavior with remarkable accuracy. (12) As early as 2011, the retailer Target Corporation famously sent coupons for baby clothes and cribs to a teenager, whom its data scientists had predicted was pregnant. (13) The teenager's father came to a Target store furious about the mailing, but later apologized when he found out that his daughter was in fact pregnant. (14) Since 2011, AI has grown ever more capable, and today, such anecdotes seem unsurprising. (15) Legal scholars take for granted that algorithms can make reasonably accurate predictions, focusing instead on questions of when and how the legal system should be able to rely on them. (16)
Even the best analysis will not provide foolproof predictions about how different taxpayers will respond to alternative tax regimes. But tax law already relies on predictive analytics. In the United States, the Internal Revenue Service (IRS) has long used data-driven computer models to determine the most promising audit targets ex post. (17) But tax law has never used AI or other data-driven models to optimize or target tax rules ex ante, either to improve efficiency or to maximize tax revenues. (18) Tax scholars have ignored the possibility of optimizing tax law using data science. (19)
Perhaps the reason for this gap in the literature is the justifiable fear that a government empowered to use AI to change individual tax regimes might make serious errors or, more nefariously, favor some taxpayers and discriminate against others. But this Article proposes alternative regimes entered into through voluntary private contractual arrangements. Such an approach not only reduces the danger of governmental abuse but also ensures that private parties have robust incentives to identify areas in which available data allows sufficiently confident predictions.
Contractual tax reform would require careful implementation. (20) Private intermediaries would design alternative tax regimes and decide which taxpayers to invite. These private intermediaries must have proper incentives to identify regimes that improve taxpayer utility while producing at least as much tax revenues for the government. Our proposal provides these incentives by randomly assigning some taxpayers who would like to be subject to an alternative regime to a control group subject to generally applicable tax law. This group's tax receipts would determine how much the intermediary must pay to the government. The design ensures that the arrangement will not harm the government, and the requirement that taxpayers affirmatively opt in ensures that taxpayers expect it to benefit them. It might seem that the only losers are the taxpayers stuck in the status quo by random chance, yet this Article will demonstrate how these taxpayers can benefit too. (21)
Suppose, for example, that a private intermediary called "Taxes, Inc." hires tax experts and AI experts to collaborate. The firm identifies one million candidate taxpayers to invite to opt into an alternative tax regime. Taxes, Inc. would send these taxpayers an invitation to opt in, along with disclosures about the upsides and downsides of the alternative regime. (22) Suppose that 100,000 of the invitees agree to participate and are deemed suitable candidates after further voluntary disclosures to Taxes, Inc. A randomly selected subset of these opting-in taxpayers (say, 10 percent, meaning 10,000 taxpayers) would be assigned at random to the control group. But the other 90,000 would be bound by the alternative tax regime; the alternative tax regime would be a contract between them and Taxes, Inc. If the 90,000 taxpayers--the "treatment group"--paid more than 9.0 times the taxes paid by the control group, then Taxes, Inc. would receive the excess (or some fraction thereof) as profits. But if the treatment group paid less than 9.0 times the taxes paid by the control group, Taxes, Inc. would have to reimburse the government the difference (or the same fraction thereof). (23)
Private intermediaries could offer alternative tax regimes to individuals or to business entities such as corporations. (24) The alternative tax regimes could be purely substantive, purely procedural, or a combination of both. Some limitations on alternative tax regimes are desirable. Many tax benefits aim to achieve nontax policy goals. (25) For example, the research and development (R&D) tax credit encourages scientific and engineering expenditures. (26) The underlying theory is that businesses do not capture all the benefits of their R&D expenses, (27) and so society benefits from favorable tax treatment of R&D. (28) Similarly, many other tax expenditures aim to encourage taxpayers to create positive externalities or to reduce negative externalities. Examples include various tax benefits for higher education, (29) clean energy, (30) and homeownership. (31) The simplest solution is for Congress to bar alternative regimes that remove specified tax benefits or, particularly in early implementations, to limit the scope of contractual tax reform to specific provisions. (32) Blocking some tax benefits will, of course, be less of an issue in countries (or states) that make less use of tax benefits to further nontax policy goals. (33) Though most of our examples will focus on the U.S. federal tax system because of its familiarity, contractual tax reform might be as or more desirable in other jurisdictions.
There are strong theoretical reasons to believe that welfare-improving alternative tax regimes could exist even in the U.S. federal tax system. The U.S. system is extraordinarily expensive to comply with and to administer, (34) yet it leaves hundreds of billions of dollars owed to the...