Beyond Purposivism in Tax Law


Tax shelters are a classic problem in tax law. From the loss-generating mink farms of the 1950s1 to today’s complex, multi-part corporate schemes,2 tax theorists have struggled to block abusive tax shelters without squelching legitimate tax planning. Tax shelters have cost the U.S. Treasury billions of dollars,3 significantly color public perceptions of tax fairness,4 and are a cornerstone of tax classes in law schools.5

But what is a tax shelter? One popular definition is that a tax shelter complies with the text of the statute, but not the statute’s underlying purpose.6 And the obvious remedy is purposivism, which elevates purposes of statutes above mere text.7 While tax law contains a number of anti-abuse doctrines specifically designed to combat tax shelters, scholars have criticized these doctrines as inconsistent and ineffective, proposing instead that purposivism alone would be preferable.8

This Article argues that purposivism is not enough. The version of purposivism most common among tax scholars considers the “specific purpose” of the statute or “general principles” of tax law when no specific purpose seems applicable.9 It typically finds these general principles in the structure of the tax code itself, like the realization requirement that gain or loss should only be recognized upon a sale or similar event.10 But analysis of specific purposes and structural principles would also invalidate any number of tax structures widely considered legitimate.11 Thus the traditional approach to purposivism in tax law fails at the crucial task of separating valid tax strategies from abusive ones. Moreover, any attempt to rescue purposivism by finding structural principles of tax law12 encounters several problems. Structural principles are practically difficult to ascertain and highly subjective; they are descriptively odd, given that the tax code was enacted piecemeal over time as a series of atomic political compromises; and they often attempt to cement contested or normatively questionable principles, like the realization principle itself.13

Much of purposivism’s appeal in tax law is its ambition of fidelity to the legislature, which elevates the values reached through democratic deliberation and avoids criticisms of judicial legislation. But the definition of “tax shelters” that focuses on statutory purpose is ultimately both overinclusive and underinclusive.14 This Article argues that a better and more direct test of whether a transaction is a tax shelter is simply whether it violates the normative preferences of tax experts. While this definition does not achieve universal consensus, it better describes the underlying problem of tax shelters and resituates the conversation over abusive transactions as an explicit discussion of policy rather than faithful agency.

Armed with this normative approach, how should we go beyond purposivism as it presently stands in tax scholarship?15 This Article considers two possible alternatives: pragmatism and doctrinalism.16 Pragmatic purposivism frankly acknowledges situations where statutes lack specific purposes, filling the gaps with pragmatic judgments about the superior policy.17 Doctrinalist purposivism applies interpretive doctrines (like the economic substance doctrine) to resolve statutory ambiguities, leaving less room for case-by-case discretion but providing greater certainty to taxpayers.18 Policy considerations ultimately motivate both pragmatism and doctrinalism, but pragmatism moves more quickly by allowing interpreters to make normative judgments on specific legal questions, while doctrinalism moves more slowly (like the common law in general) based on judicial consensus regarding broad doctrines.

This Article advocates a hybrid of pragmatism and doctrinalism: a focus on pragmatism in ex ante agency rulemaking, and on doctrinalism in ex post adjudication. This split plays to the particular strengths of pragmatism and doctrinalism. Pragmatism allows the law to produce better social outcomes. Doctrinalism provides certainty and therefore makes tax planning easier and more predictable. But doctrinalism is unnecessary when agencies are issuing forward-looking guidance, which provides clear rules for taxpayers to follow. Pragmatic purposivism is thus best suited to this forward-looking Treasury guidance, while doctrinalist purposivism is best suited to adjudication of already-completed transactions.

This hybrid approach clarifies existing doctrine on the role of agencies and courts. Because agency guidance receives judicial deference under Chevron19 and Skidmore,20 agencies are permitted (and expected) to make rules on pragmatic grounds.21 Courts, in contrast, are bound by precedent and thus by doctrines of interpretation. Under the status quo, agencies and courts alike employ both pragmatic and doctrinalist reasoning in their decisions. This Article argues that pragmatism and doctrinalism should be separated and explicitly analyzed. The ex ante/ex post split is both normatively preferable and consistent with current law.

Explicit discussion of theories beyond simple purposivism is especially crucial today. Absent such discussion, textualist judges and commentators have attacked our present ad hoc blend of pragmatic and doctrinalist purposivism as free-wheeling judicial policymaking, offending ideals of notice and procedural fairness.22 Focusing pragmatic purposivism on ex ante guidance gives it additional rigor and balances the predictability of doctrinalism with the effectiveness of pragmatism.

Part II of this Article describes the conventional argument for purposivism in the interpretation of tax statutes. It shows how that argument fails to distinguish between tax shelters and valid tax structures. Part III proposes two new variants of purposivism, pragmatism and doctrinalism. It then proposes a framework for the Treasury and the courts to go beyond purposivism, namely that they should emphasize pragmatism ex ante when issuing prospective guidance and doctrinalism ex post when considering novel fact patterns brought by taxpayers. Part III also connects this framework to existing practice by the Treasury and the courts, describing how it is consistent with judicial deference and stare decisis.

Purposivism in Tax Law

A huge quantity of tax scholarship considers methods of statutory interpretation, mostly concluding that tax statutes ought to be interpreted in a purposivist fashion.23 The conventional view in this literature is that textualism leaves the tax code more vulnerable to tax shelters, which are designed to comply with the literal text of tax statutes.24 On this view, tax shelters can be rejected on purposivist grounds because they are inconsistent with the underlying purposes of tax statutes. But this easy solution overestimates the role of purposivism as it is usually understood by tax scholars. This Part surveys theories of purposivism in tax scholarship and discusses their limitations. Like the rest of this Article, this Part focuses on purposivism and is addressed primarily to proponents of purposivism, rather than advocating for the abandonment of purposivism in favor of another methodology, like textualism.

Purposivism in Tax Law

Purposivism is an expansive concept, and one prone to misuse. A judge could defend any ruling by generically declaring it consistent with the purpose of the statute, without providing any evidence of that purpose. Or, a tax scholar unfamiliar with statutory interpretation might believe that any application of the business purpose doctrine is an instance of purposivist reasoning. But neither of these approaches reflects “purposivism” as generally understood in statutory interpretation, either in theory or in practice.

This Article focuses specifically on the most comprehensive statement of purposivism in tax scholarship to date, as described by Shannon McCormack. McCormack proposed a system of purposivist interpretation that borrows from the legal process theory of Henry Hart and Albert Sacks. McCormack argues that a tax strategy should be permitted or denied based on the “specific purpose” of the statute in question, as well as “general principles of tax law.”25

According to McCormack, the specific purpose of a statute is “what [it] is trying to achieve.”26 For example, “[t]he specific purpose of the casualty loss deduction is clearly to compensate taxpayers for the diminution in the value of personal property caused by certain events, such as an automobile accident.”27 Specific purposes are similar to what Hart and Sacks called the “immediate purpose of a statute”28 and presumably can be investigated using similar evidence: formally enacted statements of purpose,29 the nature of the problem the enactment attempted to address,30 contextual aids from the time of enactment (like legislative history),31 and contextual aids after enactment.32

General principles are trickier.33 According to McCormack, these are “overarching theoretical constructs throughout the entire [Internal Revenue] Code [(the ‘Code’)] that are intended to be captured by its individual provisions.”34 Examples include “the concept that one’s basis in an asset should reflect one’s economic investment in that asset” and “the . . . realization requirement, which generally holds that gain or loss should not be recognized until a qualifying event occurs (such as a sale).”35

Although McCormack does not elaborate on the source of these general principles, they largely resemble a kind of “structural” purposivism, which finds general principles in the overall structure of the tax code.36 Through close reading of the Code as a whole and experience with how its various provisions interact, structural purposivists purport to extract principles underlying all of tax law, allowing them to decide cases where specific purposes are silent.37

Structural purposivists maintain that structural purposes in tax law are both objectively determinate and practically useful in blocking tax shelters. Articles in the structural purposivist genre cite historic tax shelters in order to demonstrate that these tax shelters violate the fundamental structure of the Code. These principles can be very broad: Deborah Geier, for instance, lists as principles that “the same dollars should not be taxed to the same person more than once or deducted by the same person more than once”38 (a principle that McCormack endorses as well)39 and “that what we are trying to reach under an income tax is, essentially, consumption and net increases in wealth.”40

Structural purposivists assert that these principles are not merely normatively desirable, but objectively verifiable. In theory, by careful study, any interpreter should come to the same conclusion about the structure of the tax code. This idea is not new. Midcentury tax scholar Stanley Surrey attributed most of his work to the discovery of a “rational framework” in tax law, divined from a close reading of the tax code at the age of 23.41 Structural purposivism therefore mixes the objective and the esoteric: It claims that structural principles are found, not invented, but these structural principles are generally only available to those experienced enough to internalize the norms of “tax logic.”42

By emphasizing the internal coherence of tax law, structural purposivism embraces some strands of Hart and Sacks’s thinking and departs from others.43 Hart and Sacks argued that an ambiguous statute should be read “so as to harmonize it with more general principles and policies.”44 Other thinkers have developed these ideas further, most prominently Ronald Dworkin. Dworkin proposed what he called “constructive interpretation,” in which interpreters “impos[e] purpose on an object or practice in order to make of it the best possible example of the form or genre to which it is taken to belong.”45 This philosophy parallels the structuralism of Geier and others by attempting to extract general purposes from a potentially contradictory body of law. However, it represents just one part of the jurisprudence of Hart and Sacks. At other times, Hart and Sacks embraced the kind of pragmatic broad normative inquiry46 and the doctrinalist canons of construction that this Article ultimately endorses.47 Each of these approaches is consistent with Hart and Sacks’s theory of “more general and thus more nearly ultimate purposes of the law.”48

Because it focuses on specific purposes and general principles that theoretically exist independently of the case at hand, purposivism in tax law has largely followed the “faithful agent” model of purposivism, where interpreters subordinate their judgments to the will of the legislature.49 For faithful agents, statutory purpose is not made or invented by the interpreter; it is found, either in legislative history and other materials surrounding the original enactment, or through an objective process of reasoned elaboration using the tax code as a whole. Faithful agency is a popular defense against the accusations of interpretive activism frequently lobbed by modern textualists.50 But as we will see, the objectivity of faithful agency comes at a cost.

Conventional Tax Purposivism Cannot Distinguish Tax Shelters

Specific Purposes Cannot Distinguish Tax Shelters

In many ordinary tax cases, the specific purpose of the relevant statute might be sufficient to make its meaning clear.51 But in many other cases, a court might be left scratching its head, even after careful review of the statute, legislative history, and context of enactment.52 This criticism is not new; William Eskridge has argued that purposivism can be indeterminate in important cases,53 and other scholars of statutory interpretation have made similar arguments for more than a hundred years.54 But while it is well-known in the statutory interpretation literature, most tax law scholars who advocate purposivism have not addressed this important shortcoming.55

Because modern tax law is so complicated, tax planning frequently combines sections of the tax code in a manner that falls outside the specific purpose of any particular statute. Advocates of purposivism in tax law often claim that tax shelters can be repudiated whenever they exceed the specific purposes of tax statutes.56 But this is true both of legitimate tax planning and illegitimate tax shelters, and one cannot be rejected without rejecting the other.

First consider what may be the most famous modern tax shelter: the contingent liability shelter litigated in Black & Decker.57 In that transaction, Black & Decker contributed $561 million of cash and roughly $560 million of contingent liabilities to a newly formed corporation, so that the new corporation’s net value was approximately $1 million. Black & Decker then sold the corporation to a third party for $1 million. Black & Decker structured the transaction to qualify under sections 357(c)(3) and 358(d)(2) of the Code, so that it would have basis in the new corporation’s stock equal to the cash contributed.58 Given its roughly $561 million basis in the new corporation at the time of the sale for $1 million, Black & Decker claimed a tax loss of around $560 million, which allowed it to pay no taxes on its income for the year and even claim a $57 million refund.59 But the economic justification for this substantial tax loss was thin; Black & Decker had merely reshuffled its assets and liabilities and sold some of them off at fair market value.60

At the time, most commentators agreed that Black & Decker’s transaction was a tax shelter. Many argued that the shelter should be disallowed on purposivist grounds.61 The argument focused on specific purposes: The loss should be disallowed because “Congress never contemplated that section 357(c)(3) would apply” in the manner that Black & Decker proposed.62 The government took up this argument in its briefs before the Fourth Circuit, emphasizing the specific mischief that Congress sought to remedy with the statute: “to protect a parent corporation from a tax double whammy when transferring both assets and associated liabilities to a subsidiary in exchange for stock.”63 McCormack, applying her purposivist test to Black & Decker, similarly concluded that the transfer of liabilities separate from the underlying business was “not the situation the deductible liability exception sought to address.”64 Because Black & Decker had transferred only liabilities and cash, without associated assets, it departed from the prototypical, “routine” transaction that Congress had in mind.65 Thus the loss should be disallowed as falling outside this prototypical transaction.66

This narrow view of specific purposes would indeed bar strategies like the Black & Decker shelter. But as a broader survey of tax law will show, this argument proves too much. Just as Congress had not considered the Black & Decker shelter, Congress cannot feasibly anticipate every possible application of a draft tax statute, either in the statutory text or in relevant legislative history. The mere fact that Congress has not contemplated a particular structure cannot imply that structure is invalid. This is true of all sorts of tax provisions—there are various generally accepted strategies that taxpayers can use to reduce their tax liability, even though the strategies cannot be supported by specific statutory purpose or Congressional intent.

One example of a permissible non-abusive tax strategy that falls outside a statute’s specific purpose is the corporate double-dummy merger, which combines two preexisting corporations under a newly formed holding corporation. In a double-dummy merger, shareholders of each preexisting corporation contribute their shares to the new holding company in exchange for holding company stock. Thus, before the merger, the shareholders separately own stock of two independent corporations; after the merger, they all own stock in the joint holding company.67

A properly executed double-dummy merger is tax-free.68 However, the section of the tax code dealing with typical reorganizations does not exempt double-dummy mergers.69 Instead, these mergers rely on section 351, the part of the Code prototypically addressing the formation of new corporations.70 Because the shareholders in a double-dummy merger contribute their old stock in exchange for control of a new corporation, they satisfy the literal requirements of § 351 in the same way that a taxpayer would upon contributing property to a new corporation.71

But does the specific purpose of section 351 support a double-dummy merger? The prototypical section 351 transaction involves a single business—either the incorporation of a new business or the transfer of new property into corporate form.72 A double-dummy merger, on the other hand, involves the reorganization of multiple existing businesses already in corporate form. While this does not contradict the specific purpose of section 351, it is also not a natural corollary of that purpose.

One could argue that since the specific purpose of section 351 is to leave mere changes in form tax-free, transactions like double-dummy mergers that leave assets within corporate solution are consistent with that purpose.73 On the other hand, Congress probably did not intend for this purpose to be achieved at all costs and in all situations; and the existence of a separate section of the Code (section 368) specifically dedicated to mergers arguably suggests that section 351 should not be read to facilitate mergers.74 The tension between the potential competing interpretations underscores the difficulty in reading ancient legislative history—the earliest predecessor to section 351 was added to the Code in 1918,75 long before the first double-dummy merger. Again, specific purpose simply does not give a clear answer. Yet despite this ambiguity, double-dummy mergers are widely used and have been accepted by the Treasury and courts.76

A second example is the prepayment of expenses by cash-method taxpayers. Cash-method taxpayers generally recognize income only when cash is received and take deductions only when cash is paid.77 They can exploit this system by prepaying expenses; for example, a taxpayer might prepay rent for January 2022 in December 2021, even though the rent is not due until January. By doing so, the taxpayer can deduct the rental expense a year earlier, in their tax return for 2021 rather than 2022.

How does this strategy align with the specific purposes of the tax code? The cash method itself is as old as the modern income tax; it was the only method available in the Revenue Act of 1913.78 From the start, the statute has consistently allowed a deduction for business expenses like rent. In 1913, this included “the necessary expenses actually paid in carrying on any business,”79 with similar language surviving in the modern tax code.80 The Revenue Act of 1918 imposed a new requirement that the taxpayer’s method of accounting “clearly reflect the income,”81 which again has survived in similar form in the modern tax code.82

What specific purposes can we infer from the text and legislative history of the Act? On one hand, prepayment might be a necessary concession to taxpayers of a piece with the cash method’s general departure from economic reality. We could argue, perhaps, that the purpose of the cash method is administrative simplicity, and that opportunities for gamesmanship are an accepted cost of that simplicity. On the other hand, we could also argue that the purpose of the cash method is specifically to prevent this kind of gamesmanship—to avoid abuse by tying deductions to actual outlays of cash. From this perspective, to allow a different kind of gamesmanship would be a violation of the statute’s specific purpose.

Similarly, what should we make of the requirement that the method of accounting “clearly reflect the income”? Arguably, the purpose of this provision is to prevent taxpayers exploiting accounting methods to artificially defer or eliminate tax liability. But, also arguably, it was surely apparent that the cash method might sometimes allow deferral of tax liability relative to the more economically grounded accrual method, and Congress in 1918 gave no indication that the clear-reflection requirement prohibited the use of the cash method in general.83

In reality, there is little evidence for any of these perspectives in the historical record. The legislative histories of the 1913 Act and the 1918 Act include few technical accounting details. The President of the American Association of Public Accountants in 1913 criticized the codification of the cash method on the grounds that business accounts were generally kept (then as now) using the accrual method;84 his suggestions were entered into the Congressional Record, but neither discussed nor acted upon. Indeed, members of Congress at the time did not always fully grasp the difference between the cash method and the accrual method at all.85 Further, the requirement that method of accounting clearly reflect income is notoriously ambiguous—it was never clearly defined and has been called “one of the most perplexing of all tax accounting requirements.”86

Thus, specific purposes again give no clear answer. It is impossible to say how Congress expected prepayments of rent to be treated; the historical record is so thin that it is even difficult to imaginatively reconstruct how Congress would have wanted the prepayments to be treated if they were hypothetically confronted with the issue. But despite the lack of support from specific purposes, it is clear based on decades of case law that this prepayment strategy is allowed.87

A third example is the check-the-box election. Unlike the prior two examples, the check-the-box election is the product of Treasury regulations rather than clever planning by taxpayers. However, because courts only give Treasury regulations Chevron deference when the underlying statutes are unclear,88 the election is another example of tax law operating within statutory ambiguities, and it is a good example of the kind of pragmatic purposivism discussed in Part III.A.

Under current law, many business entities—including most limited liability companies (“LLCs”) and limited liability partnerships (“LLPs”)—may elect to be taxed either as corporations or partnerships for federal income tax purposes.89 Check-the-box elections were introduced in 1996 to simplify entity classification, which had previously relied on a complex multi-factor test interpreting the term “corporation” in the tax code.90 By 1996, this multi-factor test had become ineffective and needlessly costly: As the Joint Committee on Taxation (“JCT”) noted, “the entity classification regulations in effect prior to the check-the-box regulations were manipulable and were effectively elective for well-advised taxpayers.”91 The old rules therefore required taxpayer acrobatics and advantaged those with legal resources, without substantially altering results compared to an explicitly elective regime. As a solution to this problem, the check-the-box rule was widely praised by commentators when first introduced.92 Although they have caused some consternation among international tax practitioners for facilitating “hybrid” entities (which have different tax classifications under U.S. and foreign law) that can facilitate tax avoidance,93 there seems little doubt today that the check-the-box regime is valid law.94

However, the check-the-box election does not follow from statutory purpose. The key statute is section 7701(a)(3) of the Internal Revenue Code (the Code), which defines “corporation” to include “associations.”95 The word “association” is famously ambiguous; one early case described its interpretation as “seemingly in a hopeless state of confusion.”96 In the absence of a compelling plain meaning, most purposivists would turn to other tools, like legislative history and statutory context, to determine whether the check-the-box regulations appropriately implement the statute. The question becomes: What was the purpose of the corporate income tax? What were the objectives of that statute, what mischief did it seek to remedy, and what sorts of entities was it meant to target?

One conventional answer, proposed by President Taft in the 1909 letter to Congress that ultimately led to the modern corporate income tax,97 is that the corporate income tax imposes a charge for “the privilege of doing business as an artificial entity and of freedom from a general partnership liability enjoyed by those who own the stock.”98 This view suggests that, rather than freely electing entity type, LLCs and LLPs should be taxed as corporations, since they are artificial entities immune from general partnership liability. Another answer, suggested by Steven Bank, is “that the corporate income tax was originally adopted as a substitute or ‘proxy’ for taxing corporate shareholders directly.”99 On this view, the distinction between corporations and partnerships was largely one of administrative convenience, that “corporate dividends were visible and easy to track” while “[p]artnerships divided their earnings on an irregular basis and without formal notice.”100 This answer might support corporate treatment of LLCs and partnership treatment of LLPs, but again it would not support an elective regime.101

The best answer is simply that statutory purpose provides little guidance on modern-day entity classification. The first state law authorizing LLCs was enacted in Wyoming in 1977.102 Contemporary double-taxation of corporations (one tax on corporate profits, another on shareholder dividends) did not exist until the shareholder dividend exemption was repealed in 1936.103 The primary distinction between partnerships and corporations today, and thus the central issue in entity classification, is the double tax.104 Because the tax statutes dictating entity classification rules far predate the policy facts most relevant to those rules today, it makes little sense to investigate specific purpose through legislative history or an inquiry into the mischief the statute sought to remedy.

In this example as elsewhere, static specific purposes fall flat; they do not provide an adequate guide to the scope of modern corporate tax law. The purpose of the original statute has only a limited relationship to its current use, so it is hardly surprising that today’s elective regime is not supported by the original purpose of a statute enacted more than a century ago. A court reviewing the check-the-box regulations agreed, ruling that the regulations were a “reasonable response to the changes in the state law industry of business formation,” despite their apparent disconnect from the statute’s specific purpose.105

All these examples interpret the tax code in ways that arguably fall outside the specific purposes of the relevant statute; none are regarded as abusive.106 What separates them from the tax shelters that have been the subject of so much scholarly criticism? Not their relationship to congressional intent; not their relationship to specific statutory purpose. Because specific statutory purposes are essentially static and originalist, they cannot provide guidance on new questions in a constantly changing interpretive landscape.107

Moreover, the problem is becoming worse over time. We have increasingly entered an era of “unorthodox lawmaking,”108 where lawmakers eschew the careful deliberation portrayed in Schoolhouse Rock! in favor of emergency bills, last-minute revisions, and provisions meant to win key stakeholders (like the crucial 50th senator who can break the filibuster in a reconciliation bill).109 This new regime permits significantly less public debate prior to the passage of statutes. That means less legislative history for interpreters to use in teasing out statutory purpose; it also renders statutory purpose less coherent and less clear in general. Tax laws have especially suffered under this new regime because tax laws are often passed through omnibus reconciliation bills meant to address many different policy goals. Several recent tax reforms were the product of truncated congressional debate, including the Affordable Care Act110 and the Tax Cuts and Jobs Act of 2017,111 which contained drafting oversights that caused significant headaches for interpreters after they were signed into law.112

The overall picture for specific purposes is therefore bleak. Congress has always struggled to anticipate just how tax statutes will apply in practice, and unorthodox lawmaking has made the struggle more difficult than ever. The specific purposes of old statutes shed little light on modern interpretive problems, and new statutes present fewer cognizable specific purposes because of the modern Congress’s truncated deliberative process. Specific purpose alone therefore provides little help in separating legitimate and illegitimate tax structuring.

Structural Purposes Are Elusive

If specific purposes are inadequate against tax shelters, can general principles come to the rescue?

The search for a general principle in Black & Decker illustrates just how tricky and controversial that search can be. One potential general principle might be that a contributor’s basis in a new corporation (“Newco”) following a section 351 contribution should not exceed Newco’s economic value at the time of the contribution.113 But this proposed general principle does not apply in all cases—for one, a contribution of loss property would result in basis exceeding Newco’s fair market value.114 If this is acknowledged as an exception to the general principle, the principle begins to seem more like a situation-specific statement of one’s own preferences.

Moreover, structural purposivists must contend with a range of structural principles that occasionally may contradict each other. Black & Decker’s strategy seems like a straightforward case of selective realization in line with the realization principle. It is uncontroversial that an investor may sell stocks with built-in losses and retain stocks with built-in gains; why should Black & Decker not be able to do the same with respect to liabilities that it has paid a third party to assume in an economically legitimate transaction?

Even if we were to agree that some determinate general principle weighs against Black & Decker, specific statutory purpose trumps it. Consider the contribution of assets and contingent liabilities of a going concern, which critics of Black & Decker acknowledged as the prototypical, non-abusive case where sections 357(c)(3) and 358(d)(2) should apply to give the taxpayer higher basis.115 If Taxpayer Terry were to contribute her entire business, including $100 of assets and $50 of related liabilities, to Newco, the economic value of Newco would be $50, but it is clear that Terry would have a basis of $100 in Newco under Sections 357(c)(3) and 358(d)(2).116 Thus the general principle that basis cannot exceed economic value probably does not exist, but especially does not apply to this case, where it conflicts with the specific purpose of the statutes in question.

But perhaps we need to think bigger. What if the right general principle is something like Geier’s statement “that what we are trying to reach under an income tax is, essentially, consumption and net increases in wealth[?]”117 This principle might bar the Black & Decker shelter because the shelter generated paper losses without any real economic losses.

But where do these principles come from? The answer seems to be the policy preferences of the interpreter rather than any actual structural features of tax law.118 Geier’s principle that income should equal consumption plus increases in wealth is stated nowhere in the tax code—it is, in fact, a restatement of the Haig-Simons definition of income.119 The tax code notoriously contains so many “tax expenditures,” incentive programs that give tax benefits unmoored from the measurement of income,120 that it cannot be said to have been drafted with any fidelity to Haig-Simons income.

Moreover, these principles are debatable even normatively. David Weisbach famously argued that lines around what is or is not income should be drawn based on policy grounds (like efficiency) rather than an abstract platonic ideal of income,121 an idea that has since become widely accepted among tax scholars. Should the Haig-Simons principle therefore be abandoned? Should it be replaced with a principle that income equals whatever is most efficient?

It is easy to say that some policy preference is a structural principle when using that principle to rebut a tax shelter one dislikes. But it is harder to argue that these principles are comprehensively reflected in the tax code. They are more often asserted than proven, and it is not clear how they could be disproven. And even if they are useful for tax scholars, they are difficult for judges to wield, especially generalist judges who lack the necessary familiarity with the structure of tax law.122

Moreover, these very broad general principles seem to reflect an unrealistic view of the legislative process. Borrowing from the political science and public choice literatures, scholars like William Eskridge and Philip Frickey argued in the 1980s and 1990s that statutes did not actually have a single idealized purpose,123 but rather reflected “backroom deals” between legislators and interest groups with diffuse, contradictory, and often selfish goals.124

If single statutes arguably lack determinate purposes, the problem is vastly worse for the tax code as a whole. Some statutes set a baseline level of tax; some statutes are intended to incentivize prosocial behavior; some statutes are bald giveaways to favored interest groups, perhaps as the price of buying those prosocial statutes. The tax code was enacted piecemeal, by many different Congresses, over many different periods. As Eskridge notes with respect to statutory interpretation in general: “[T]here are few if any principles or policies that we accept at any and all cost, some principles and policies may be in tension with one another, and given this complexity it is hard for an interpreter to weigh these competing concerns determinately.”125 Given the multiplicity of different purposes and policies, how can a coherence-based theory extract a single overarching structural purpose in any particular case?

One answer might be that even if structural purposivism is not descriptively true, it is a useful legal fiction that rationalizes the tax code.126 But this response just shines harsher light on the arbitrariness involved in discerning these structural principles. If they rely, as they must, on the policy preferences of the interpreter, it would be better for those policy preferences to be stated explicitly. Otherwise, structuralism simply serves as a confusing fiction, and one that is strictly inferior to pragmatism; structural principles that nicely resolve one case might lead to the wrong result in another, so that an interpreter’s commitment to those principles comes at the cost of flexibility.127

Even if we can agree on what general principles exist, we may not agree on whether these principles are normatively desirable. Consider again the realization principle. As a descriptive matter, it is subject to several important exceptions—for example, broker-dealers are required to account for the value of their securities on a mark-to-market basis, meaning that broker-dealers will recognize any gains or losses regardless of realization.128 But the realization principle encounters normative problems as well. Many scholars have argued that mark-to-market taxation should become the general rule;129 in a recent survey, a slight majority of tax professors agreed.130 In practice, as the Black & Decker case shows, application of the realization principle to complement specific purposes would tend to exacerbate the problem of tax shelters rather than relieving it. Thus, even if we assume arguendo that the realization principle is a valid one, treating it as a general principle may ossify it in a way that does more harm than good.

A final problem with general principles is that they are often uninformative. General principles are general—they do not allow us to draw specific lines in difficult cases,131 which are the cases that are actually litigated and which are sufficiently ambiguous that general principles are necessary. Take the example of Cottage Savings, in which the taxpayer exchanged a portfolio of mortgages for a technically different but economically similar portfolio of other mortgages.132 Was this trade a “disposition of property” under section 1001(a) of the Code? The specific purpose of the statute was ambiguous; and the realization principle that “gain or loss should not be recognized until a qualifying event occurs (such as a sale),”133 provides no guidance as to what that “qualifying event” should be. Thus, in the absence of purposive guidance, the Court instead decided Cottage Savings based on the language of relevant regulations and judicial deference to the Treasury.134

The great ambition of structuralism, of course, is fidelity to the law as written.135 The hope is that judges can avoid allegations of loose construction or judicial activism by inferring structural principles from the law itself, rather than bringing their own policy preferences to bear. But in reality, any judge claiming to apply structural principles inevitably finds principles consistent with their own policy preferences. In contrast, pragmatism and doctrinalism more overtly reflect normative goals that may not be found within the statutes themselves. They are ultimately more ambitious, but also more transparent.

The Case for Explicit Normative Analysis

Because both specific and general purposes can remain ambiguous, purposivism, honestly applied, may not catch out structures generally regarded as abusive. Consider the transaction in Compaq Computer Corp.,136 where Compaq bought and immediately resold stock of Royal Dutch Petroleum.137 Compaq bought the stock “cum dividend” (i.e., with the right to receive a forthcoming dividend) for $887.6 million and resold the stock “ex-dividend” (i.e., without the right to receive the dividend) for $868.4 million, buying and selling from the same counterparty. Compaq paid $1.5 million in transaction costs in the transaction and also became entitled to a $22.5 million dividend from Royal Dutch, on which it paid $3.4 million in withholding taxes to the Netherlands, leaving it with a net dividend of $19.2 million.138 Crucially, though, Compaq was entitled to a foreign tax credit in its U.S. tax return equal to the $3.4 million that it paid to the Dutch government. After the dust settled, and including the benefit of the foreign tax credit, Compaq made $1.25 million on the purchase and resale.139

How was this possible? The key was that the spread between the value of the stock cum and ex-dividend was only the net value of the dividend, not the gross value. This was because most shareholders could not take advantage of the foreign tax credit from Dutch withholding tax. Thus, Compaq engaged in a form of tax arbitrage—“the transactions were essentially transfers of foreign tax credits from owners who could not use them to taxpayers who could.”140

McCormack considers the Compaq transaction in a detailed analysis of statutory purposes. After finding that “th[e Compaq] transaction does not flout any general principles,” she focuses on evidence of specific statutory purposes.141 She observes that, according to the legislative history, “[t]he purpose of the foreign tax credit generally is to eliminate the possibility of double taxation (once by the foreign jurisdiction and again by the United States) on the foreign source income of a U.S. person.”142 And she reasons that, since “it is quite difficult to determine who actually bears the economic incidence of a tax,”143 and given Congress’s apparent decision to choose a potentially over-inclusive, easy-to-administer bright line rule, it is within the statutory purpose to allow the nominal payor of withholding tax (Compaq) to take the foreign tax credit.144 McCormack concludes that the Compaq transaction is not an abusive tax shelter at all, and that “it is inappropriate to deny Compaq the claimed credits.”145

McCormack’s statutory analysis is deep and careful and is probably the best answer for a coherence theorist using tools rooted solely in statutory purposes. But in doing so, it illustrates the potential shortcomings with that approach. While McCormack does her best to extract statutory purposes from the materials available to her, a simpler answer may be that Congress could not have anticipated that the foreign tax credit would be used in this way and left insufficient clues in the legislative history for us to reconstruct statutory purposes that cast light on this particular case. In situations like these, as I have argued, we must look elsewhere to resolve statutory ambiguities.

More importantly, McCormack’s response talks past the essentially normative concerns of other critics who saw the Compaq transaction as wasteful tax planning.146 To these critics, the problem was never simply that Compaq was reaping a statutory benefit Congress had not intended; the fundamental issue was that Compaq had engaged in purely wasteful tax arbitrage. Compaq had incurred $1.5 million in transaction costs merely by shuffling paper around, with the costs ultimately borne by the U.S. government and by other taxpayers. And it had done so in a purely artificial transaction totally detached from its ordinary course of business. If this was not a tax shelter, then nothing is! And a theory that fails to categorize a transaction like this one as abusive seems to misunderstand what we mean when we talk about tax shelters.

So what really separates tax shelters from legitimate transactions? Simply that tax shelters are too good to be true.147 Tax shelters violate principles of fairness and equity. They generate huge tax losses without any underlying economic activity;148 they allow savvy taxpayers to avoid taxation entirely.149 In short, tax shelters are tax shelters not because they contradict statutory purpose, but because they contradict the normative preferences of tax experts.

Rather than debating what is or is not a tax shelter, we should therefore take a more explicitly normative view of tax law.150 Tax experts will never reach complete agreement about what qualifies as a tax shelter, since this assessment will depend on personal policy preferences. Instead, the interpretation of the tax code should be analyzed with normative goals in mind: efficiency, fairness, process, fidelity to Congress, the institutional capacities of the courts and the Treasury, and more. The idea that tax law should target tax shelters, however defined, is only a proxy for what should ultimately be a normative inquiry.151

In the current judicial climate of strict construction and fidelity to text, scholars might be afraid of advocating any interpretive method that incorporates normative preferences. While these are reasonable concerns, they are unavoidable. If double-dummy mergers, prepayments, and check-the-box elections are to be separated from tax shelters, it cannot be based on legislative history and other conventional purposivist tools alone. Tax interpretation must go beyond purposivism in this everyday sense.

To be clear, the problem is not just that purposivism gives the wrong result, but rather that it often gives no result. Conventional structural purposivist tools are indeterminate in a wide variety of important cases, including those involving most tax shelters.152 Promoters of tax shelters attempt to claim benefits that Congress did not contemplate, which are not entailed by the specific purpose of the underlying tax statute, and which no general principle seems to legitimize. But the same is true of structures like double-dummy mergers, prepayments, and check-the-box elections; we cannot conclude from the mere absence of specific purpose or a general principle that a tax benefit should not be allowed.

One solution might be simply to adopt a rule of strict construction when purposively interpreting tax benefits. This rule would be distinct from a version of strict constructionism that urges judges to stick to the text of the statute, which usually means more generous distribution of tax benefits.153 It would instead place statutory purpose first and foremost, while prohibiting any tax benefit not entailed by immediate statutory purpose. This is the approach that many tax purposivists seem to already apply under the status quo.154

Strict construction of tax benefits imposes a clear-cut rule when ambiguity remains about statutory purposes. Importantly, it does not distinguish between the benign and abusive tax shelters discussed above—so it would prohibit double-dummy mergers, prepayments, and check-the-box elections along with contingent liability tax shelters. This might be acceptable in practice since Congress could enact a statutory remedy if it felt that any tax benefit truly ought to have been granted. The asymmetric rule against tax benefits might suit our modern era of tax shelters, where the primary concern of many tax scholars is the protection of revenue.

In addition to placing a greater onus on Congress to reenact non-abusive tax structures, however, strict constructionism requires a disregard for statutory text that many modern interpreters might find uncomfortable. Tax shelters are typically constructed to comply with the literal language of tax statutes. Thus, in order to be effective against tax shelters, strict constructionism must prohibit tax benefits even when they are consistent with the text of the statute and not inconsistent with the specific purpose of the statute.155 Few modern textualists or purposivists (outside of tax law) would reject statutory text when statutory purpose is ambiguous. Moreover, given current realities of legislative deadlock, Congress may be too slow in enacting legislative remedies where tax benefits are mistakenly denied. Strict construction of tax benefits suffers from all these impediments, muting its appeal.

Beyond Purposivism

If specific purpose is insufficient and general structural principles are too vague, what are the alternatives? This Part considers two: pragmatism and doctrinalism. It argues that when specific purpose runs out, tax interpreters should refer to a hybrid of pragmatic and doctrinal tools to decide whether a tax structure is acceptable. While these methods differ from those generally proposed by tax scholars in the past, they are still consistent with a purposivist scheme—one could argue that they are just a different approach at finding the general principles endorsed by McCormack, Hart, and Sacks.156

Pragmatic Purposivism

In this Article, I use the term “pragmatism” to refer specifically to an approach to statutory interpretation that aims to achieve the best results, however those results are measured, rather than prioritizing some other doctrinal goal, like faithful agency to the will of the legislature or stare decisis. Pragmatic157 purposivism acknowledges that sometimes ambiguity remains after applying conventional tools of statutory interpretation. In those cases, a pragmatic interpreter picks the best interpretation on normative grounds, based on their judgment of which interpretation would lead to the best policy outcomes.158

One of the best exemplars for modern pragmatic purposivism is Justice Breyer. Breyer’s jurisprudence closely follows the purposivism of Hart and Sacks, which Breyer studied in law school.159 Much like Hart and Sacks, Breyer tries to reconstruct the intent of the “reasonable member of Congress.”160 Breyer also recognizes that specific purposes sometimes run out.161 In those cases, he concludes that the reasonable legislator would have decided the case on consequentialist, pragmatic policy grounds162—an approach which a strict constructionist would be quick to point out allows the judge to rely on their own policy preferences. The fusion of pragmatism and purposivism is characteristic of Breyer’s judicial philosophy in general, as well as the emphatically pragmatic academic writing he produced before joining the Supreme Court.163

Pragmatic purposivism successfully distinguishes between abusive and non-abusive tax planning. To illustrate, first consider the example of the check-the-box election through a pragmatic purposivist lens. As noted above, entity classification was functionally elective even prior to the check-the-box regime, because the factors in the entity classification test could be easily manipulated. The old rule distorted economic activity and required taxpayers to incur additional costs, while advantaging the rich and well-advised. So, faced with a statutory ambiguity, the Treasury took the pragmatic purposivist route of allowing check-the-box elections on policy grounds.164

Now consider the Black & Decker tax shelter. What would the practical consequences be if this shelter were respected for tax purposes? More fees for lawyers; more time wasted in inefficient corporate restructuring; less money for the federal government. There is no policy reason to permit this tax shelter, and many reasons to prohibit it. So the pragmatic purposivist could easily conclude that this shelter should be prohibited. Thus, pragmatism succeeds where mere purposivism fails: It distinguishes the (permissible) check-the-box regime from the (impermissible) contingent liability tax shelter.

Although pragmatic purposivism has achieved some popularity among judges outside of tax law, its most enthusiastic practitioner within tax law is undoubtedly the Treasury. The Treasury has remained largely “purposivist[,] despite the rise of the . . . new textualism” in courts.165 Moreover, over the past hundred years, the Treasury has substantially shifted in its published decision-making away from the consideration of statutes and toward the consideration of policy objectives.166 Much of this shift can be attributed to the advent of judicial deference regimes like Chevron167—as administrative law scholars like Peter Strauss have argued, agencies have latitude within a “Chevron space” of statutory ambiguity to make rules based on whatever grounds they see fit, which typically means pragmatic grounds.168

The idea of a Chevron space closely parallels pragmatic purposivism. The Treasury first applies conventional purposivist criteria in order to determine whether a statute is ambiguous. If the statute is ambiguous, Chevron deference169 permits the Treasury to write rules on pragmatic grounds. Under this framework, the Treasury applies pragmatism where specific purposes run out. Thus, pragmatic purposivism explains the Treasury’s reasoning in examples like the check-the-box regulations much more cleanly than purposivism alone.

Doctrinalist Purposivism

Although tax scholars are often private pragmatists, they are just as often public doctrinalists. Tax experts have a certain cultural affinity for rules—a love of the cut and dry that motivated many to become tax specialists in the first place. This tendency manifests in the anti-abuse doctrines that populate tax law. These include doctrines like the economic substance doctrine, which requires that each transaction have a substantial non-tax business purpose and a substantial non-tax economic effect in order to be respected for tax purposes.170

These anti-abuse doctrines should be considered substantive canons—interpretive tools that courts use to reach the best reading of the tax code.171 Specifically, they should be considered presumptions rebuttable by statutory purpose—so that, for example, if a tax structure lacked a non-tax business purpose or a non-tax economic effect, it would be presumptively invalid and disallowed unless consistent with the specific purpose of the underlying statute.172

Because these doctrines can be rebutted by specific purposes, they start exactly where specific purposes end. Although some modern purposivists are skeptical of substantive canons—which have come to be identified with textualism in academic literature173—they formed an important part of Hart and Sacks’s legal process. Hart and Sacks described presumptions as “[t]he court’s last resort, when doubt about the immediate purpose of a statute remains.”174

How do doctrinal presumptions like these operate in practice?175 Consider again the contingent liability tax shelter, an easy target for the economic substance doctrine. Black & Decker stipulated that it had entered the transaction only for tax reasons, thus failing the business purpose prong.176 So an interpreter applying the economic substance doctrine could likewise have rejected this shelter. And, indeed, doctrines like the economic substance doctrine have historically been the courts’ main weapons against tax shelters. In contrast, prepayments and double-dummy mergers pass the economic substance doctrine because they involve real transactions with substantial business purposes and substantial economic effects. Prepayments involve real payments of cash, and double-dummy mergers involve the actual combination of pre-existing corporations; while structured with taxes in mind, both transactions have a legitimate non-tax business purpose. Thus, doctrinalist purposivism succeeds in separating tax shelters from legitimate transactions.177

If pragmatic purposivism is the hallmark of the Treasury, doctrinalist purposivism is the hallmark of the courts. Courts are bound by precedent, and precedent generates doctrines that are passed down and refined from case to case.178 This trend has been challenged by proponents of the new textualism, which of course rejects purposivism entirely and has in some cases rejected the substantive tax canons as well.179 But to the extent that purposivism survives among the courts in tax cases, it often does so in doctrinal garb.180

Doctrinalism fits the modern judiciary better than pragmatism. Today more than ever, judges disagree on their normative goals and their perceptions of the common good. Some judges are purposivists, some are textualists, some are pragmatists, and some are dynamic theorists. Some judges in tax cases are pro-government, and some are pro-taxpayer. To allow judges to rely on personal normative preferences in individual cases risks making the judiciary itself unpredictable and arbitrary. Unpredictability, in turn, makes tax planning more difficult, opens up the possibility of forum shopping by litigants in tax cases, and undermines the rule of law.

Moreover, judges’ insulation from the political process means that they are less democratically accountable and therefore theoretically less democratically responsive than agencies. Treasury regulations go through notice and comment before they are finalized, and they are widely vetted both inside and outside the Treasury. Not so with judicial decisions. Thus, it makes sense that judges would couch policy preferences within canons of construction. These canons move more slowly over time, reflecting the wisdom of many successive judges and accruing authority with those judges’ agreement.

What legitimizes some canons and disqualifies others? There are several potential answers. First, doctrinalist canons might be background norms sufficiently widespread that legislative drafters are presumptively aware of them, so that these canons form a necessary general background to the faithful interpretation of statutes.181 Second, substantive canons might also be justified in positivist terms by their widespread acceptance among judges.182 Third, regardless of which canons are presently widely accepted (either among legislators or judges), a different set of canons might be preferable on normative grounds.

My view, mirroring Hart and Sacks’s, is that normative considerations ultimately justify substantive canons but, because the canons reflect consensus among judges, they represent a higher, more abstract level of normative judgment. We could imagine normative preferences playing out at several possible levels of specificity. Most granularly, a judge could apply those preferences ad hoc with respect to each case, without even aiming for consistency between cases or respecting stare decisis. At a higher level, a pragmatic purposivist might try to find the best rule in this case and similar future cases, ultimately choosing the outcome that produces the best results in general within the constraints of statutory purpose. At an even higher level comes doctrinalist purposivism: normative concerns underlie the doctrines (protecting government revenue, fighting tax shelters), but the doctrines must apply across a wide variety of cases and statutes. Thus, doctrinalist purposivism takes pragmatic considerations seriously, but it incorporates them in a bounded fashion based on common law reasoning.

Although doctrinalism has many advantages, it is not a perfect solution and encounters its own share of hard cases, which create uncertainty in the doctrine’s application. Most prominently, David Hariton has incisively argued that cases involving substantive canons turn on how the transaction is framed.183 Because tax-motivated transactions can often be bundled with legitimate transactions entered into purely for economic (non-tax) reasons, the narrowly tax-motivated transaction might appear to violate a substantive canon if considered alone, while still surviving scrutiny if paired with a legitimate transaction. For example, a loss-generating transaction that lacks a substantial business purpose or substantial economic effect standing alone could be paired with a legitimate sale of assets to a third party.

This is a valid and important critique, but it is not a unique critique of doctrinalism. Hariton’s arguments equally apply even if we solely consider congressional intent, since Congress might not intend to facilitate the narrow transaction even if it intended to facilitate the broad transaction.184 Indeed, the level at which a transaction is framed is important even in pragmatic analysis. Pure tax planning is widely regarded as inefficient and distortive; thus, any specific piece of tax planning, viewed narrowly, might be prohibited on pragmatic grounds. Even so, there may be a pragmatic justification for allowing tax planning in the context of a broader legitimate transaction. Thus, the appropriate framing is as important to pragmatic analysis as it is to doctrinalist, intentionalist, or purposivist analysis. Hariton himself does not argue that the appropriate frame is impossible to determine or that it makes objective analysis of tax cases impossible, although he does point out that it is an underappreciated source of subjectivity in difficult tax questions.185 He instead argues that the Treasury should take a more consistent line when it comes to framing transactions as shelters or not.186

But even doctrinalism rooted in the substantive canons proves inadequate in some cases. Some transactions are problematic for policy reasons but do not violate any substantive canon of taxation. One recent example is the boom in corporate inversions that occurred prior to tax reform in 2017.187 In a corporate inversion, a foreign corporation domiciled in a low-tax jurisdiction would acquire a U.S. corporation in order to permit tax planning that ultimately reduces U.S. taxes—for example, by using intra-company debt to shift profits from the United States to a lower-tax jurisdiction. Inversions were a problem because they eroded the U.S. tax base and incentivized wasteful tax structuring that only benefited corporate tax havens, like the Cayman Islands and Ireland.188

Although corporate inversions were problematic for tax policy, they clearly were not barred by the substantive canons. The acquisition of the U.S. corporation in an inversion is a bona fide transaction; it has a substantial economic effect and, in most cases, taxpayers can legitimately claim that it has a substantial business purpose in the form of corporate synergies. Thus, the courts lacked the ability to recast inversion transactions on doctrinal grounds.

The Treasury, on the other hand, was not so constrained, and issued a series of notices and regulations that significantly curbed the benefits of inversion transactions.189 This guidance was well within the Treasury’s regulatory and subregulatory authority under Chevron and Skidmore deference and stood as an example of agency action that would not have been available to courts.

Similarly, sometimes the courts lack authority to alter statutory interpretations that have outlived their usefulness. Consider again the example of check-the-box elections. Check-the-box elections would fail under many anti-abuse doctrines, including the substance-over-form doctrine: An election solely for tax purposes has no substantive non-tax purpose and should theoretically be disallowed. Doctrinalist purposivism alone might therefore misfire by prohibiting this tax strategy. This again suggests that doctrinalist purposivism is not a complete solution—the best method may ultimately be a hybrid approach.

A Hybrid Approach

One core benefit of doctrinalism is predictability—it allows taxpayers to plan their transactions with the confidence that they will receive their hoped-for tax treatment, so long as they do not violate any anti-abuse doctrine. One core benefit of pragmatism is tailoring of results to circumstances—it casts a tighter net around tax shelters, allowing more granular distinctions between acceptable and unacceptable tax structuring.

To the doctrinalist, predictability is especially important because the vast majority of tax planning is legitimate and merely incidental to ordinary business transactions. These ordinary transactions are often planned by tax lawyers who view the whole phenomenon of tax shelters with distaste.190 Unlike an agenda of broad-based pragmatism, which always carries some risk that an interpreter will mistakenly condemn a non-abusive transaction, doctrinalism promises consistency and thereby avoids chilling ordinary business deals.

To the pragmatist, doctrinalism fails to capture many modern tax shelters, which are often designed not only to comply with the text of the tax code, but also to satisfy canons like the economic substance doctrine. Some of the substantive canons are sufficiently flexible that they still arguably apply to these transactions—but their application is not clear cut, and judges would undoubtedly have an easier time rejecting these transactions with the flexibility that pragmatism provides.

The situation is complicated by the fact that the Treasury and courts seem to follow conflicting approaches. Does the doctrinalist predictability of the courts counter the pragmatic unpredictability of the Treasury? Would one of these interpreters ideally just adopt the methods of the other?

Not quite, because the Treasury and the courts operate in very different contexts. The Treasury receives judicial deference only in published decision-making—Chevron deference for regulations191 and Skidmore deference for subregulatory guidance.192 It is only in these circumstances that past work has found a pragmatic shift in IRS decision-making. In contrast, the Treasury does not receive Chevron or Skidmore deference in cases not addressed by existing published guidance. Novel factual situations are also the bread and butter of a court’s docket—taxpayers are generally risk-averse and rarely challenge an IRS publication that is directly on point.

This more nuanced account of administrative and judicial activity suggests that both pragmatism and doctrinalism may be useful in different contexts. Pragmatism is especially appropriate for published guidance, because published guidance is necessarily predictable: It tells taxpayers exactly how the Treasury will treat a particular fact pattern. Taxpayers might dislike when a regulation or revenue ruling forecloses a potential tax strategy, but they cannot complain of unpredictability or unfair ex post decision-making. Likewise, doctrinalism is especially appropriate when a fact pattern falls within an ambiguous area of the law not addressed by published guidance, where predictability is at a premium.

Viewed this way, pragmatic and doctrinalist purposivism do not necessarily divide different branches of government, but rather different types of cases.193 Ex ante guidance should rely primarily on pragmatic purposivism; ex post adjudication should rely primarily on doctrinalist purposivism. While ex ante guidance is the exclusive province of the Treasury and the IRS, the IRS also frequently must engage in ex post judgments about how to treat taxpayers with novel fact patterns.194 This distinction matches the difference between “rules” subject to agency rulemaking procedures and “orders” subject to agency adjudication procedures under the Administrative Procedure Act.195

This hybrid approach also suits the relative institutional competencies of courts and agencies.196 Courts are diffuse and unpredictable; each circuit develops its own precedents in federal tax law, and no federal court has particular expertise in tax law other than the Tax Court, whose own decisions may still be appealed to circuit courts. In contrast, the Treasury is a single department staffed by experts and with significant institutional capacity to consider which policies will result in the best outcomes across the entire country. Its expertise means that it will generally make better pragmatic judgments, and its unity means that it will generally produce guidance that is more coherent and easily understood than courts’. In addition, the Treasury is better situated than courts to make the threshold decision of whether statutory purpose is ambiguous in the first place, because it is intimately involved in the drafting of statutes and will have a better institutional understanding of statutory purpose.

The hybrid proposal is not quite how the courts and Treasury currently operate.197 Reality is much messier—rather than cabining pragmatic and doctrinal approaches to ex ante and ex post interpretation, courts and the Treasury generally believe themselves to be bound by substantive canons of tax law,198 while also demonstrating themselves to be motivated by pragmatic considerations on the margins.199 Because judicial anti-abuse doctrines in tax law have not been considered substantive canons until recently, scholars, judges, and the Treasury itself have not considered the possibility that they may be more appropriate in some interpretive contexts than others.

To be clear, I do not claim that pragmatic purposivism is solely appropriate ex ante, or that doctrinalist purposivism is solely appropriate ex post. I argue instead that interpreters should place greater weight on pragmatic considerations when reliance interests are not at stake, and more weight on doctrines when those interests are at stake. In particular, policy issues may sway judges in extreme cases even though they are making ex post determinations, even when no anti-abuse doctrine applies. In this sense, I am not advocating a strictly formalist division of interpretive method, but rather something reminiscent of the “practical reason” approach promoted by professors William Eskridge and Philip Frickey.200

This framework aims to strike a balance between efficiency and fairness. In addition to its normative desirability, it follows current administrative law. Ex ante IRS guidance receives either Chevron or Skidmore deference, which (particularly for Chevron deference) gives the Treasury freedom to set rules on pragmatic, normative grounds.201 In contrast, Treasury interpretations of statutes during audits do not receive any special judicial deference,202 and therefore the Treasury must conform its reasoning to the anticipated reasoning of reviewing courts. Those courts, which always operate ex post, are in turn bound by doctrinalist canons of interpretation that derive from common law.203

Some caveats apply. The ex ante/ex post framework is a first-best solution, which relies on coordination between judges and widespread acceptance of ex post doctrinalism. If coordination cannot be achieved—if, for example, a substantial contingent of pragmatic purposivist judges insists on applying individualized policy judgments rather than doctrinalist canons—then we may well be back to the status quo, such that the second-best solution (the solution accounting for the constraint of noncompliance from pragmatic purposivist judges) would be thoroughgoing pragmatic purposivism. Consequently, the choice between pragmatism and doctrinalism is essentially a coordination problem. Even if we agree that an ex ante/ex post split would be ideal, its success (and its desirability as optimal policy) turns on whether judges and IRS administrators actually follow it.204

Ex ante pragmatism and ex post doctrinalism may also be more promising within tax law than in other areas of law, because of the predominance of the IRS and the Tax Court in tax administration. As noted above, because relatively few tax audits ever go to trial, the IRS is the sole tax authority that most taxpayers will ever face and the interpreter whose views are taken most seriously in practice.205 But whether interpretive policy can be coordinated even within the IRS, especially in a manner that acknowledges the subtle distinction between ex ante and ex post interpretation, remains to be seen.


Past scholarship has argued that statutory purpose can serve as a dividing line between tax shelters and ordinary, non-abusive transactions. But that line is unworkably fuzzy; both legitimate and illegitimate tax structures sometimes exceed the specific purposes of tax statutes, and structural principles of tax law are too vague and subjective to fill the gap. Conventional appeals to statutory purpose therefore fail to distinguish abusive tax planning.

This Article invites explicit consideration of the interpretive possibilities beyond faithful-agent purposivism. It proposes ex ante pragmatism and ex post doctrinalism as the ideal method to discourage tax shelters and encourage legitimate tax planning. But the Article’s core contribution is broader: Any given scholar might favor a different mix of pragmatism, doctrinalism, or structuralism depending on their own normative preferences. This Article provides a theoretical framework to make that evaluation, by discussing the strengths and weaknesses of competing interpretive approaches.

  1. [1]. E.g., United States v. Cook, 270 F.2d 725, 726 (8th Cir. 1959); see also generally Cedarburg Fox Farms, Inc. v. United States, 283 F.2d 711 (7th Cir. 1960) (fox farm); Greer v. Comm’r, 17 T.C. 965 (1951) (chinchilla farm).

  2. [2]. E.g., Summa Holdings v. Comm’r, 848 F.3d 779, 781–84 (6th Cir. 2017); Black & Decker Corp. v. United States, 436 F.3d 431, 433–35 (4th Cir. 2006) (considering a contingent liability tax shelter).

  3. [3]. Tanina Rostain & Milton C. Regan, Jr., Confidence Games: Lawyers, Accountants, and the Tax Shelter Industry 25 (2014) (“Abusive tax shelter activity at the turn of the twenty-first century cost the U.S. Treasury billions of dollars ... .”).

  4. [4]. James B. Lewis, The Treasury’s Latest Attack on Tax Shelters, 11 Tax Notes 723, 723 (1980) (arguing that tax shelters result in “impairment to the fairness of the income tax, the perception of unfairness by the rest of the taxpaying public, and the feared adverse impact on the level and temper of voluntary compliance”).

  5. [5]. E.g., Stephen Schwarz & Daniel J. Lathrope, Fundamentals of Corporate Taxation: Cases and Materials 619–35 (10th ed. 2019) (discussing corporate tax shelters, as well as judicial and legislative responses to the proliferation of tax shelters).

  6. [6]. E.g., Steven A. Dean & Lawrence M. Solan, Tax Shelters and the Code: Navigating Between Text and Intent, 26 Va. Tax Rev. 879, 882 (2007) (“[W]e adopt the position, taken by others, that tax shelters are generally characterized as transactions that appear to comply in a literal manner with the Code, but which are designed to reach a tax result that Congress would not have intended.” (footnote omitted)); Shannon Weeks McCormack, Tax Shelters and Statutory Interpretation: A Much Needed Purposive Approach, 2009 U. Ill. L. Rev. 697, 703 (“It is commonly agreed that tax shelters refer to transactions that are carefully designed to fit within the letter of the tax law to derive benefits that tax planners and taxpayers know are (or likely are) outside the purposes of the provisions on which they rely.”).

    Some theorists distinguish between intentionalism, focused on congressional intent, and purposivism, which focuses on statutory purposes more broadly construed. Theories focused on actual congressional intent have been out of favor for almost a hundred years; Max Radin famously argued that intentionalism of this sort was untenable due to individual congresspeople’s lack of attention to specific factual situations, the difficulty of reconstructing specific intent from the historical record, and the impossibility of aggregating diffuse specific intents. Max Radin,

  7. [7]. See John F. Manning, The New Purposivism, 2011 Sup. Ct. Rev. 113, 113 (“Congress enacts statutes to achieve certain purposes, and ... judges should construe statutory language to fulfill those purposes.”).

  8. [8]. See generally Leandra Lederman, W(h)ither Economic Substance?, 95 Iowa L. Rev. 389 (2010) (advocating for replacing the economic substance doctrine with an inquiry into congressional intent). See McCormack, supra note 6, at 720–31. Sandra O’Neill has argued that these anti-abuse doctrines are simply purposivism by another name. Sandra Favelukes O’Neill, Let’s Try Again: Reformulating the Economic Substance Doctrine, 121 Tax Notes 1053, 1053 (2008) (describing the economic substance doctrine as “no more than a doctrine of purposeful statutory construction ... .”).

  9. [9]. See McCormack, supra note 6, at 721–27; Lederman, supra note 8, at 398, 443.

  10. [10]. See infra notes 33–36 and accompanying text.

  11. [11]. The structures are double-dummy mergers, see infra notes 67–76 and accompanying text, as well as check-the-box elections, see infra notes 88–103 and accompanying text.

  12. [12]. Cf. Deborah A. Geier, Interpreting Tax Legislation: The Role of Purpose, 2 Fla. Tax Rev. 492, 497 (1995) (“Code provisions ought to be construed so as not to damage [their] fundamental structure, even if doing so requires that a statutory term be construed in a nonliteral (nontextual) fashion.”); McCormack, supra note 6, at 731–42 (arguing for and applying a new framework to derive purposes of tax codes and regulations).

  13. [13]. See infra Section II.B.2.

  14. [14]. See infra Section II.B.1.

  15. [15]. This Article goes “beyond” purposivism in the sense that it takes an existing purposivist framework, elaborates it, and extends it. Cf. Hubert L. Dreyfus & Paul Rainbow, Michel Foucault: Beyond Structuralism and Hermeneutics (1982) (describing a theoretical evolution that goes “beyond” structuralism without necessarily contradicting structuralism). Note that while this Article goes beyond purposivism as currently understood among tax scholars, it does so in a way that is arguably consistent with the legal process purposivism espoused by Hart and Sacks. See infra notes 46–48, 156 and accompanying text.

  16. [16]. See infra Part III.

  17. [17]. A similar but distinct school of thought is the “practical reason” school of interpretation, proposed by William Eskridge and Philip Frickey, and applied to tax law by Michael Livingston. Practical reason rejects “foundationalism” that focuses on any one theory of interpretation, instead proposing that each court consider a “broad range of textual, historical, and evolutive evidence when it interprets statutes.” William N. Eskridge, Jr. & Philip P. Frickey, Statutory Interpretation as Practical Reasoning, 42 Stan. L. Rev. 321, 322 (1990); see Michael Livingston, Practical Reason, “Purposivism,” and the Interpretation of Tax Statutes, 51 Tax L. Rev. 677, 679 (1996) [hereinafter Livingston, Practical Reason]. This Article speaks instead to those (like myself) who embrace purposivism, which is one of the “foundationalist” techniques criticized by Eskridge and Frickey, but emphasizes next steps when specific purpose runs out. In other words, this Article assumes that the specific purposes of statutes still control so long as those purposes can be determined, contra Eskridge, Frickey, and Livingston.

  18. [18]. I discussed one doctrinalist approach in a recent article, although I did not label it “doctrinal” at the time. Jonathan H. Choi, The Substantive Canons of Tax Law, 72 Stan. L. Rev. 195, 201 (2020) [hereinafter Choi, Substantive Canons].

  19. [19]. Chevron U.S.A., Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 844 (1984).

  20. [20]. Skidmore v. Swift & Co., 323 U.S. 134, 139–40 (1944).

  21. [21]. See generally Chevron, 467 U.S. (holding that an agency’s interpretation of an ambiguous statute warrants deference so long as it represents a “reasonable policy choice”). Cf. Covad Commc’ns Co. v. F.C.C., 450 F.3d 528, 537 (D.C. Cir. 2006) (requiring the agency to “articulate a satisfactory explanation for its action including a rational connection between the facts found and the choice made” (quoting Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983))); Jonathan H. Choi, An Empirical Study of Statutory Interpretation in Tax Law, 95 N.Y.U. L. Rev. 363, 365–66 (2020) [hereinafter Choi, Empirical Study] (“[M]any have suggested that judicial deference regimes, like Chevron deference, empower agencies to make rules based on normative policy concerns, rather than merely seeking the ‘best reading’ of a statute (using purposivism, textualism, or any other methodology).” (footnotes omitted)); Richard J. Pierce, Jr., How Agencies Should Give Meaning to the Statutes They Administer: A Response to Mashaw and Strauss, 59 Admin. L. Rev. 197, 200 (2007) (arguing that, under Chevron, agencies can choose among permissible interpretations of a statute “only by engaging in a policymaking process”). But see Aaron Saiger, Agencies’ Obligation to Interpret the Statute, 69 Vand. L. Rev. 1231, 1232 (2016) (“In circumstances where a reviewing court is expected to defer to agency interpretation, the agency bears a legal and ethical duty to select the best interpretation of its governing statute.”).

  22. [22]. Summa Holdings, Inc. v. Comm’r, 848 F.3d 779, 782, 786–89 (6th Cir. 2017); Benenson v. Comm’r, 910 F.3d 690, 699 (2d Cir. 2018); see Gitlitz v. Comm’r, 531 U.S. 206, 219­–20 (2001).

  23. [23]. For some examples of scholarship arguing in favor of a purposivist approach to tax law interpretation, see generally Noël B. Cunningham & James R. Repetti, Textualism and Tax Shelters, 24 Va. Tax Rev. 1 (2004); Geier, supra note 12; Mary L. Heen, Plain Meaning, the Tax Code, and Doctrinal Incoherence, 48 Hastings L.J. 771 (1997); Lederman, supra note 8; Richard Lavoie, Analyzing the Schizoid Agency: Achieving the Proper Balance in Enforcing the Internal Revenue Code, 23 Akron Tax J. 1 (2008); Michael Livingston, Congress, the Courts, and the Code: Legislative History and the Interpretation of Tax Statutes, 69 Tex. L. Rev. 819 (1991) [hereinafter Livingston, Congress]; McCormack, supra note 6; Lawrence Zelenak, Thinking About Nonliteral Interpretations of the Internal Revenue Code, 64 N.C. L. Rev. 623 (1986); and Livingston, Practical Reason, supra note 17; see also generally Andre L. Smith, The Deliberative Stylings of Leading Tax Law Scholars, 61 Tax Law. 1, 19 (2007) (“Textualism’s capacity for producing undesirable or absurd consequences is well explored by Zelenak, Geier, Heen, Cunningham & Repetti, and Lavoie.”). All of the authors cited by Smith favor purposivism, except for Cunningham and Repetti, who sometimes favor intentionalism. In contrast, tax scholars that favor textualism are relatively rare. See generally John F. Coverdale, Text as Limit: A Plea for a Decent Respect for the Tax Code, 71 Tul. L. Rev. 1501 (1997) (defending a textualist approach to tax code interpretation); Edward A. Zelinsky, Text, Purpose, Capacity and Albertson’s: A Response to Professor Geier, 2 Fla. Tax Rev. 717 (1996) (same).

  24. [24]. For a discussion of how textualism leaves the tax code vulnerable to tax shelters, see generally Cunningham & Repetti, supra note 23; Lederman, supra note 8; and McCormack, supra note 6.

  25. [25]. McCormack, supra note 6, at 721–27.

  26. [26]. Id. at 722.

  27. [27]. Id.

  28. [28]. Hart & Sacks, supra note 6, at 1380.

  29. [29]. Id. at 1377.

  30. [30]. Id. at 1378 (“call[ing] for a close look at the ‘mischief’ thought to inhere in the old law and at ‘the true reason of the remedy’ provided by the statute for it.”).

  31. [31]. Id. at 1379.

  32. [32]. Id. at 1380.

  33. [33]. Hart and Sacks also believed that sometimes the immediate purpose of the statute would fail to decide the case at hand, noting the importance of “more general and thus more nearly ultimate purposes of the law.” Id. at 148. But their account of how to determine general purposes “[wa]s notably incomplete.” Vincent A. Wellman, Dworkin and the Legal Process Tradition: The Legacy of Hart & Sacks, 29 Ariz. L. Rev. 413, 432–33 (1987). The Legal Process tradition is arguably compatible with structuralism, doctrinalism, and pragmatism, which I describe in further detail below. See infra notes 43–48, 156 and accompanying text.

  34. [34]. McCormack, supra note 6, at 723.

  35. [35]. Id.

  36. [36]. The term “structural purposivism” comes from a recent article of mine discussing scholars who “argue that the structure of the Code itself should be read to imply certain principles that overcome statutory text.” Choi, Substantive Canons, supra note 18, at 241.

  37. [37]. See id. at 241–42 & nn. 235–38.

  38. [38]. Geier, supra note 12, at 497.

  39. [39]. McCormack, supra note 6, at 723.

  40. [40]. Geier, supra note 12, at 497; see also generally Joseph M. Dodge, The Logic of Tax (1989) (describing the federal tax code using broad conceptual themes similar to Geier’s).

  41. [41]. Stanley S. Surrey, A Half-Century with the Internal Revenue Code: The Memoirs of Stanley S. Surrey 7 (Lawrence Zelenak & Ajay K. Mehrotra eds., 2022) (“I surveyed almost the entire structure of the income tax ... . A rational framework could be devised for the structure ... . I saw the income tax not as a random body of rules and edicts but as an internally consistent framework. All of my later work has been dominated by that approach.”). Of course, even if this approach was feasible in Surrey’s time, the modern tax code is vastly different: more complex and written without the spirit of consensus that tended to characterize midcentury politics.

  42. [42]. Livingston, Congress, supra note 23, at 829; Livingston, Practical Reason, supra note 17, at 683–84.

  43. [43]. McCormack acknowledges that her proposal diverges in some respects from legal process purposivism—in particular, she believes that her use of general principles may require more radical departures from statutory text than Hart and Sacks would accept. McCormack, supra note 6, at 723 n.138 (“Hart and Sacks oppose this concept entirely ... .”).

  44. [44]. Hart & Sacks, supra note 6, at 148. This appeal to coherence is part of Hart and Sacks’s theory of “reasoned elaboration” by which statutes ought to be interpreted. Theodore W. Jones, Textualism and Legal Process Theory: Alternative Approaches to Statutory Interpretation, 26 J. Legis. 45, 53 (2000) (providing a structuralist perspective based on Hart and Sacks’s theory of “reasoned elaboration”); Hart & Sacks, supra note 6, at 143–58 (applying reasoned elaboration in statutory interpretation).

  45. [45]. Ronald Dworkin, Law’s Empire 52 (1986).

  46. [46]. Hart & Sacks, supra note 6, at 102 (describing the ultimate purpose of all statutory interpretation as “establishing, maintaining and perfecting the conditions necessary for community life to perform its role in the complete development of man”). To be sure, Hart and Sacks were far from thoroughgoing pragmatists, and their intent was to craft a legal process whose neutral principles would legitimate interpretation above mere appeals to interpreters’ personal normative preferences. But by narrowly endorsing pragmatic purposivism in the context of agency interpretation, I also attempt to constrain normative judgments in a process-driven way.

  47. [47]. See John F. Manning, What Divides Textualists from Purposivists?, 106 Colum. L. Rev. 70, 89 n.67 (2006) (“Among other things, the Legal Process approach gave unflinching effect to the sort of substantive canons that judges have developed over time ‘to promote objectives of the legal system which transcend the wishes of any particular session of the legislature.’” (quoting Hart & Sacks, supra note 6, at 1376)); Hart & Sacks, supra note 6, at 1376–77, 1380 (instructing interpreters to apply clear statement rules and presumptions).

  48. [48]. Hart & Sacks, supra note 6, at 148.

  49. [49]. E.g., Michael Herz, Purposivism and Institutional Competence in Statutory Interpretation, 2009 Mich. St. L. Rev. 89, 92 (“[P]urposivism is grounded on, respects, and seeks to advance legislative preferences; it is a form of faithful agency.”); Peter J. Smith, Textualism and Jurisdiction, 108 Colum. L. Rev. 1883, 1899 (2008) (“Intentionalism and purposivism proceeded from the premise of legislative supremacy: If, in a constitutional democracy, judges must be faithful agents of Congress, then judges must attempt to decipher as accurately as possible Congress’s statutory instructions.”). One prominent alternative to the faithful-agent model was proposed by Justice Harlan Stone, who argued against the “illusion that in interpreting [statutes] our only task is to discover the legislative will,” suggesting that judges “treat a statute much more as we treat a judicial precedent, as both a declaration and a source of law, and as a premise for legal reasoning.” Harlan F. Stone, The Common Law in the United States, 50 Harv. L. Rev. 4, 13, 15 (1936). However, Stone’s proposal is “not the prevalent view today.” John F. Manning & Matthew C. Stephenson, Legislation and Regulation: Cases and Materials 303 (4th ed. 2021).

  50. [50]. Manning & Stephenson, supra note 49, at cxxv–cxxxiv. The new textualism is an umbrella term for the revival of textualist thought over the past four decades, spearheaded by figures like Justice Scalia. The new textualism was initially characterized by its particular skepticism of legislative history. See William N. Eskridge, Jr., The New Textualism, 37 UCLA L. Rev. 621, 623 (1990) (“The new textualism posits that once the Court has ascertained a statute’s plain meaning, consideration of legislative history becomes irrelevant.”). But see John F. Manning, Textualism as a Nondelegation Doctrine, 97 Colum. L. Rev. 673, 731–37 (1997) (defending certain textualist uses of legislative history). Of course, as textualism has garnered more adherents, many branches of textualism have emerged with more complex commitments than mere skepticism regarding legislative history.

  51. [51]. E.g., Laue v. Comm’r, No. 3842-18S, 2020 WL 1929271, at *2 (T.C. Apr. 20, 2020) (applying an example from legislative history speaking directly to the facts of the case).

  52. [52]. For example, section 7502 of the Code provides that the date of a tax filing’s “postmark ... shall be deemed to be the date of the delivery.” I.R.C. §7502(a)(1). But what should the date of delivery be if the post office fails to apply a postmark? Was the statute’s reference to a postmark a mere oversight, or an attempt to force taxpayers to send important filings by certified mail (where a postmark is guaranteed)? The legislative history was not particularly clear, and the Tax Court ultimately addressed this uncertainty through its own estimate of when the filing was most likely mailed. Seely v. Comm’r, 119 T.C.M. (CCH) 1031, 2020 WL 201751, at *2; see also Bryan Camp, Lesson from the Tax Court: The Common Law Mailbox Rule Lives!, TaxProf Blog (Feb. 3, 2020),
    mon-law-mailbox-rule-lives.html [](describing the common law rule as “a backstop to fill in gaps in the statutory and regulatory scheme”).

  53. [53]. William N. Eskridge, Jr., Dynamic Statutory Interpretation 29-30 (1994); William N. Eskridge, Jr., Philip P. Frickey & Elizabeth Garrett, Legislation and Statutory Interpretation 230 (2d ed. 2006) (“Purposivism does not yield determinate answers when there is no neutral way to arbitrate among different purposes. Even if there were agreement as to which purpose should be attributed to a statute, the analysis in the hard cases might still be indeterminate. Often an attributed policy purpose is too general and malleable to yield interpretive closure in specific cases, because its application will depend heavily upon context and the interpreter’s perspective.”).

  54. [54]. Marcus P. Knowlton, Legislation and Judicial Decision: In Their Relations to Each Other and to the Law, 11 Yale L.J. 95, 100 (1901).

  55. [55]. But see Livingston, Practical Reason, supra note 17, at 679 (arguing against conventional purposivism in favor of the “practical reason” approach advanced by Eskridge). As discussed above in note 17, this Article differs from Livingston’s and Eskridge’s prescriptions in that it suggests that courts remain faithful to statutory purposes when they can be discerned.

  56. [56]. See, e.g., infra notes 61–66 and accompanying text.

  57. [57]. Black & Decker Corp. v. United States, 436 F.3d 431, 432 (4th Cir. 2006); see also I.R.S. Notice 2001-17, 2001-9 C.B. 730 (describing the contingent liability tax shelter).

  58. [58]. If sections 357(c)(3) and 358(d)(2) had not applied, then Black & Decker’s basis in the new corporation would have been reduced by the amount of the liabilities assumed, meaning that it would have had near-zero basis in the new corporation and would not have subsequently recognized a loss upon selling the new corporation.

  59. [59]. Black & Decker, 436 F.3d at 434.

  60. [60]. The statute has since been amended to explicitly block contingent liability tax shelters like the Black & Decker transaction. I.R.C. § 358(h).

  61. [61]. See infra notes 62–66.

  62. [62]. Karen C. Burke, Deconstructing Black & Decker’s Contingent Liability Shelter: A Statutory Analysis, 108 Tax Notes 211, 212 (2005).

  63. [63]. Black & Decker, 436 F.3d at 436–37 (emphasis omitted).

  64. [64]. McCormack, supra note 6, at 749.

  65. [65]. Burke, supra note 62, at 214.

  66. [66]. Arguing from a more textualist perspective, the IRS also cited section 357(b) of the Code, which treats the assumption of a liability as “‘money received’ by the taxpayer on the exchange,” Black & Decker 436 F.3d at 438, “for purposes of section 351 or 361,” I.R.C. § 357(b)(1)(B), so long as the “principal purpose of the taxpayer ... was ... to avoid Federal income tax ... or ... was not a bona fide purpose,” id. The IRS first argued that a bad purpose under section 357(b) would require Black & Decker to decrease its basis in the new corporation by the amount of the assumption of the liability, under section 358(a)(1). Black & Decker, 436 F.3d at 438. The Court rejected this argument, emphasizing that a provision that applied “for purposes of section 351 or 361,” I.R.C. § 357(b)(1)(B), could not apply to section 358(a)(1) without more explicit statutory language. Black & Decker, 426 F.3d at 438–39. The IRS also attempted to argue that a transaction that failed the anti-abuse language in section 357(b) could not qualify under section 357(c)(3) (and therefore would not qualify for the exception from basis reduction in section 358(d)(2)), because section 357(c)(2) barred transactions that failed section 357(b) from qualifying for section 357(c)(1). Id. The Court rejected this argument too, observing that 357(c)(1) was distinct from 357(c)(3), and pointing out that if Congress had wanted a similar carve-out from section 357(c)(3) it should have added one. Id. at 439–40.

  67. [67]. The double-dummy merger differs from some more conventional merger methods, in which one company merges into the other or into a subsidiary of the other. A straightforward merger of one corporation into another is a straightforward “A reorganization,” named after the section of the tax code granting this sort of merger tax-free treatment. I.R.C. § 368(a) (2020). Another popular style of merger is the “reverse triangular merger” or “(a)(2)(E)” merger, again named after the relevant section of the tax code. I.R.C. § 368(a)(2)(E).

  68. [68]. More specifically, it will be a nonrecognition event for federal income tax purposes.

  69. [69]. See I.R.C. §§ 354, 368.

  70. [70]. See id. § 351.

  71. [71]. In particular, because the shareholders as a group have “control” of the holding corporation after the merger is complete, they satisfy the requirements of § 351(a) of the Code. Id. § 351(a).

  72. [72]. Marvin A. Chirelstein, Tax Pooling and Tax Postponement—The Capital Exchange Funds, 75 Yale L.J. 183, 190 (1965) (“What is undoubtedly conceived of as typical of section 351 is a relatively small-volume transaction involving either the incorporation of an existing business by its owners or the establishment of a new business by a limited number of individuals desiring to combine their capital and skills.”).

  73. [73]. The Senate and House reports for the 1921 bill that introduced the predecessor to section 351 emphasized the prior law under which taxpayers may be taxed on gain even if they “actually realize[] no cash profit.” S. Rep. No. 67-275, at 11–12 (1921); and H.R. Rep. No. 67-350, at 10 (1921). They noted that nonrecognition “w[ould], by removing a source of grave uncertainty[,] ... not only permit business to go forward with the readjustments required by existing conditions but also w[ould] considerably increase the revenue by preventing taxpayers from taking colorable losses in wash sales and other fictitious exchanges.” S. Rep. No. 67-275, at 11–12; H.R. Rep. No. 67-350, at 10.

    One in-depth comparison of the legislative histories behind the treatment of corporate formations and reorganizations concludes:

    [T]hat the predominant congressional concern in enacting the nonrecognition provisions for both reorganizations and certain transfers of property to corporations was the need to stimulate the economy after World War I by both removing any tax disincenties and providing tax incentives for a variety of corporate activities, including corporate combinations and the incorporation of ongoing or new businesses.

    Kathryn L. Powers, “Decontrol” of Section 351 of the Internal Revenue Code: Facilitating Capital Formation by Small Corporations, 31 Case W. Rsrv. L. Rev. 814, 830–31 (1981). But this more realistic interpretation of the broader purposes behind nonrecognition of formations and reorganizations is unworkably broad as the basis for a purposivist analysis; it would fail to prohibit any corporate tax shelter, since tax shelters do generally incentivize corporate activities.

  74. [74]. As additional evidence against the argument that section 351 may appropriately address any mere changes in the form of an investment, section 368(a)(1)(F) classifies as a reorganization any “mere change in identity, form, or place of organization of one corporation.” I.R.C. § 368(a)(1)(F). The categorization of some mere changes in form as reorganizations could again be taken as indirect evidence that transactions not listed in the statute should not be considered reorganizations and therefore should not be accorded nonrecognition treatment, under the principle that expressio unius est exclusio alterius (i.e., the express mention of one thing excludes others that are not mentioned). William N. Eskridge, Jr., Philip P. Frickey, Elizabeth Garrett & James J. Brudney, Cases and Materials on Legislation and Regulation: Statutes and the Creation of Public Policy 668 (5th ed. 2014).

    Moreover, sections 351 (on corporate formations) and 368 (on corporate reorganizations) were enacted by different Congresses with subtly different rationales for enactment. The Senate Bill that ultimately led to the Revenue Act of 1918 originally provided for both tax-free reorganizations and tax-free exchanges of property for stock, the latter being what we would now recognize as a section 351 formation. The Senate Report argued that the purpose of both these provisions was “to negative the assertion of tax in the case of certain purely paper transactions.” S. Rep. No. 65-617, at 5 (1918). However, nonrecognition for contributions to new corporations (the antecedent to section 351) was later deleted, leaving only nonrecognition for corporate reorganizations (the antecedent to section 368). Revenue Act of 1918, Pub. L. No. 65-254, ch. 18, 40 Stat. 1057 (1919). Three years later, nonrecognition for formations was added in the Revenue Act of 1921.

  75. [75]. Revenue Act of 1918 § 202(b).

  76. [76]. See Rev. Rul. 84-71, 1984-1 C.B. 106 (reversing Rev. Ruls. 80-284, 1980-2 C.B. 117; and 80-285, 1980-2 C.B. 119 to permit acquisitive section 351 transactions).

  77. [77]. See I.R.C. § 446(c)(1) (permitting use of “the cash receipts and disbursements method”).

  78. [78]. Alan Gunn, Matching of Costs and Revenues as a Goal of Tax Accounting, 4 Va. Tax Rev. 1, 4 (1984) (“The Revenue Act of 1913 required the use of cash-method accounting by all taxpayers.”). Even earlier, the Excise Tax Act of 1909, which applied solely to corporations, also imposed an income tax solely on a cash basis. Corporation Excise Tax Act of 1909, Pub. L. No. 61-5, ch. 6, § 38(2)(d), 36 Stat. 11, 112–17 (1909).

  79. [79]. Higgins v. Comm’r, 312 U.S. 212, 215 (1941) (citing Revenue Act of 1913, Pub. L. No. 63-16, § 2(b), 38 Stat. 114, 167 (1913)).

  80. [80]. I.R.C. § 162(a) (“There shall be allowed as a deduction all the ordinary and necessary expenses paid or incurred during the taxable year in carrying on any trade or business ... .”).

  81. [81]. Revenue Act of 1918, H.R. 12863, 65th Cong. § 212(b) (1918) (“[I]f the method employed does not clearly reflect the income, the computation shall be made upon such basis and in such manner as in the opinion of the Commissioner does clearly reflect the income.”).

  82. [82]. I.R.C. § 446(b) (“[I]f the method used does not clearly reflect income, the computation of taxable income shall be made under such method as, in the opinion of the Secretary, does clearly reflect income.”).

  83. [83]. Stephen F. Gertzman, Federal Tax Accounting ¶ 3.02 (2021) (“[T]here was nothing in the 1918 Act that suggested that Congress believed that the cash method did not clearly reflect income or that its use should be discouraged.”).

  84. [84]. Specifically, Robert Montgomery argued that the phrase “actually paid” in the act, which implied that deductions could only be taken on a cash basis, should be substituted with “losses actually ascertained” or “interest actually accrued,” in order to authorize the use of what we would now call accrual-basis accounting. Tariff Hearings: Hearing Before the H. Comm. on Ways & Means, 62d Cong. 6285 (1913) (statement of Robert H. Montgomery, President, American Association of Public Accountants). These changes were not made. However, regulations issued shortly after permitted taxpayers to file their taxes using the same method with which they keep their books. Gunn, supra note 78, at 5 n.19.

  85. [85]. H.R. Rep. No. 922, 64th Cong., 1st Sess. 4 (1916) (describing the Revenue Act of 1913 as a tax “on the accrued basis”). Alan Gunn notes that this “seems to have been a mistake, perhaps inspired by the regulations and forms, which did require a sort of accrual accounting.” Gunn, supra note 78, at 5 n.18.

  86. [86]. Gertzman, supra note 83, ¶ 2.02[2].

  87. [87]. Historically, the IRS primarily challenged prepayments on the grounds that they are capital expenditures (and thus give rise to assets with basis, rather than an immediate deduction) if they have benefits relating to future taxable years. See I.R.C. § 263(a) (denying deductions for “[a]ny amount paid out for new buildings or for permanent improvements or betterments made to increase the value of any property or estate”). However, a long line of cases has held that prepayments are immediately deductible, even if they relate to expenses attributable to a future taxable year, so long as the expense is incurred within one year. This is known as the “one-year rule.” E.g., Hotel Kingkade v. Comm’r, 180 F.2d 310, 312–13 (10th Cir. 1950); United States v. Akin, 248 F.2d 742, 744 (10th Cir. 1957); Zaninovich v. Comm’r, 616 F.2d 429, 432 (9th Cir. 1980); Agro-Jal Farming Enters., Inc. v. Comm’r, 145 T.C. 145, 151–52 (2015). The Treasury finalized regulations endorsing the one-year rule in 2004. Treas. Reg. § 1.263(a)-4(f) (2020).

    The question of whether a prepayment is a capital expenditure is separate from the more fundamental question of whether the purpose of the cash method is consistent with prepayment of expenses by taxpayers, which is implicitly accepted in much of this case law. Because the capital expenditure issue is a straightforward interpretive question regarding the meaning of section 263(a), I do not consider it in the body of this Article.

  88. [88]. See Chevron U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 844–45 (1984).

  89. [89]. Treas. Reg. § 301.7701-3 (2020).

  90. [90]. Treas. Reg. §§ 301.7701-1–11 (1996); T.D. 6503, 1960-2 C.B. 409. These regulations were known as the “Kintner regulations,” after a Ninth Circuit ruling that classified a professional organization as a corporation for tax purposes. Kintner in turn relied on the Supreme Court’s ruling in Morrissey v. Commissioner, which addressed whether a golf club should be considered a corporation for federal income tax purposes. See United States v. Kintner, 216 F.2d 418, 421–24 (9th Cir. 1954); Morrissey v. Comm’r, 296 U.S. 344, 350–56 (1935).

  91. [91]. Staff of the J. Comm. on Tax’n, 105th Cong., Review of Selected Entity Classification and Partnership Tax Issues 17 (1997).

  92. [92]. E.g., William S. McKee, William F. Nelson & Robert L. Whitmire, Federal Taxation of Partnerships and Partners¶ 3.06[1] (2021) (“The simplicity and flexibility of the check-a-box Regulations have evoked virtually universal approbation and support from private practitioners and taxpayers.”); Am. Bar Ass’n Section of Tax’n, Comments on Notice 95-14, 1995-14 I.R.B. 7, Proposed Revisions to the Entity Classification Rules, Tax Notes Today (1995); N.Y. State Bar Ass’n Tax Section, Report on the “Check the Box” Entity Classification System Proposed in Notice 95-14, at 19–37 (1995); Victor E. Fleischer, Note, “If It Looks Like a Duck”: Corporate Resemblance and Check-the-Box Elective Tax Classification, 96 Colum. L. Rev. 518, 529–32 (1996).

  93. [93]. See William J. Bricker, Jr., Karen B. Brown, Alan W. Granwell & Paul R. McDaniel, Use of Hybrids in International Tax Planning: Past, Present and Future, 13 St. John’s J. Legal Comment. 79, 82 (1998).

  94. [94]. Gregg Polsky has criticized the check-the-box regulations as inappropriate in light of Supreme Court precedent establishing criteria for what constitutes an “association,” as set out in the 1935 case of Morrissey v. Commissioner, 296 U.S. 344, 356–58 (1935). See generally Gregg D. Polsky, Can Treasury Overrule the Supreme Court?, 84 B.U. L. Rev. 185 (2004) (arguing that check-the-box regulations are invalid). However, current Supreme Court doctrine gives agencies deference when they override a prior court ruling, so long as the ruling did not establish that the statute was “unambiguous.” Nat’l Cable & Telecomms. Ass’n. v. Brand X Internet Servs., 545 U.S. 967, 982 (2005) (“A court’s prior judicial construction of a statute trumps an agency construction otherwise entitled to Chevron deference only if the prior court decision holds that its construction follows from the unambiguous terms of the statute and thus leaves no room for agency discretion.”). This is important to the interaction between Morrissey and the check-the-box regulations because the Morrissey Court acknowledged the ambiguity in the mere term “association.” Morrissey, 296 U.S. at 356–57. What’s more, the Court even suggested (almost fifty years before Chevron) that in light of the statutory ambiguity, the Treasury was authorized to fill in the gaps with regulations, which provides the modern reader a further hint that Chevron step one (statutory ambiguity) was satisfied and that Brand X therefore would warrant deference to a subsequent agency determination, even if contrary to the Morrissey ruling. See id. at 354–55 (“As the statute merely provided that the term ‘corporation’ should include ‘associations,’ without further definition, the Treasury Department was authorized to supply rules for the enforcement of the Act within the permissible bounds of administrative construction.”).

  95. [95]. I.R.C. § 7701(a)(3) (2018).

  96. [96]. Coleman-Gilbert Assocs. v. Comm’r, 76 F.2d 191, 193 (1st Cir. 1935).

  97. [97]. The Payne-Aldrich Tariff Act of 1909, Pub. L. No. 61-5, 36 Stat. 11, 112 (1909), imposed an excise tax on corporate income that was functionally equivalent to an income tax but was phrased as an excise tax in order to avoid the constitutional challenge that had defeated the Wilson-Gorman Tariff Act. The Wilson-Gorman Tariff Act had imposed income taxes (including a corporate income tax) in 1894 but was struck down by the Supreme Court in Pollock. See generally Pollock v. Farmers’ Loan & Tr. Co., 157 U.S. 429 (1895) (invalidating the Wilson-Gorman Tariff Act). The excise tax on corporate income was later folded into the federal income tax after the passage of the Sixteenth Amendment, which authorized a federal income tax. See U.S. Const. amend. XVI.

  98. [98]. Charles W. Pierson, Is the Federal Corporation Tax Constitutional?, 39 Nat’l Corp. Rep. 544, 545 (1910).

  99. [99]. Steven A. Bank, Entity Theory as Myth in the Origins of the Corporate Income Tax, 43 Wm. & Mary L. Rev. 447, 452 (2001). Bank relies in part on evidence from the legislative history of the Wilson-Gorman Tariff Act, which was overturned in Pollock. The Act imposed an income tax on “all other corporations, companies, or associations doing business for profit in the United States, no matter how created and organized, but not including partnerships.” Act of Aug. 27, 1894, Pub. L. No. 53-349, § 32, 28 Stat. 509, 556 (1894). This language matches the Senate’s version of the Act; however, an earlier House draft had imposed the tax on all “corporations or associations organized for profit by virtue of the laws of the United States or of any State or Territory, by means of which the liability of the individual stockholders is anywise limited.” Bank argues that:

    The House version reflects the grant or concession theory’s emphasis on the corporation’s special privilege of limited liability, while the Senate version taxes all corporations regardless of their individual attributes. The enactment of the latter version thus appears to lend credence to a natural entity rather than an artificial entity or grant/concession theory explanation.

    Bank, supra, at 497 (footnote omitted).

  100. [100]. Bank, supra note 99, at 524.

  101. [101]. Another view, proposed by Marjorie Kornhauser, is that the corporate income tax was intended as a “corporate regulatory measure.” Marjorie E. Kornhauser, Corporate Regulation and the Origins of the Corporate Income Tax, 66 Ind. L.J. 53, 53 (1990). Kornhauser argues that the federal government essentially used the corporate income tax as a means to collect information on corporations: “[T]he tax enabled the government to acquire information to help it legislate more knowledgeably ... .” Id. at 54. This approach gives little guidance on the interpretation of the term “association” today, however, because both partnerships and corporations file returns that provide similar amounts of information to the federal government.

  102. [102]. Susan Pace Hamill, The Limited Liability Company: A Catalyst Exposing the Corporate Integration Question, 95 Mich. L. Rev. 393, 399–400 (1996).

  103. [103]. Revenue Act of 1936, Pub. L. No. 74-740, 49 Stat. 1648.

  104. [104]. Schwarz & Lathrope, supra note 5, at 6 (“The concept at the heart of Subchapter C is the double taxation of corporate income.”).

  105. [105]. Littriello v. United States, No. 04CV-143-H, 2005 WL 1173277, at *3 (W.D. Ky. 2005), aff’d, 484 F.3d 372 (6th Cir. 2007), cert. denied, 552 U.S. 1186 (2008). As Part III will discuss, the ambiguity of specific purpose gives the Treasury latitude to interpret the statute on policy grounds rather than focusing solely on fidelity to originalist purposes.

  106. [106]. Other examples of transactions that appear to fail this purposivist test but are nonetheless allowed under current law are the Cottage Savings transaction, infra notes 132–34 and accompanying text, and umbrella partnership real estate investment trusts (“UPREITs”), which combine a real estate investment trust (“REIT”) with a partnership in order to generate tax benefits for investors with legacy real estate assets. UPREITs are highly technical workarounds to the gain recognition rule in Section 351(e), but they have been expressly blessed by Treasury regulations. Treas. Reg. § 1.701-2, ex. 4 (1994).

    One response might be that each of these transactions is in fact illegitimate, but that they have survived because no litigant has had standing to challenge them. This is part of a generally observed problem in tax law, that taxpayers may sue for overtaxation, but that no citizen has standing to sue in cases of undertaxation, even though in theory the undertaxation of one taxpayer necessitates increased taxes on all other taxpayers. Polsky, supra note 94, at 238–43; Lawrence Zelenak, Custom and the Rule of Law in the Administration of the Income Tax, 62 Duke L.J. 829, 847–53 (2012). Proponents of this line of thought might happily discard check-the-box elections and double-dummy mergers, or at least might have been happy to discard these structures at their birth, prior to the development of reliance interests. See infra Section III.B. (describing strict constructionist approaches that would reject these structures alongside conventional tax shelters).

  107. [107]. Many others have made the same critique of static intentionalism and purposivism. See, e.g., Eskridge, supra note 53, at 14–34; T. Alexander Aleinikoff, Updating Statutory Interpretation, 87 Mich. L. Rev. 20, 33–37 (1988).

  108. [108]. See generally Barbara Sinclair, Unorthodox Lawmaking: New Legislative Processes in the U.S. Congress (5th ed. 2017) (coining the term “unorthodox lawmaking”).

  109. [109]. See Abbe R. Gluck, Comment, Imperfect Statutes, Imperfect Courts: Understanding Congress’s Plan in the Era of Unorthodox Lawmaking, 129 Harv. L. Rev. 62, 99 (2015) (discussing the implications of Congress’s truncated deliberative process in the context of King v. Burwell).

  110. [110]. Patient Protection and Affordable Care Act, Pub. L. No. 111-148, 124 Stat. 119 (2010).

  111. [111]. Tax Cuts and Jobs Act, Pub. L. No. 115-97, 131 Stat. 2054 (2017).

  112. [112]. Justice Roberts’s majority opinion in King v. Burwell chided the Affordable Care Act for “more than a few examples of inartful drafting,” including the apparent typo that the Court overrode in that case. King v. Burwell, 576 U.S. 473, 491 (2015). The Tax Cuts and Jobs Act was arguably even worse, and congresspeople have been debating “technical corrections” to correct drafting errors ever since the original act was passed. Jane G. Gravelle, Cong. Rsch. Serv., R46754, “Technical Corrections” and Other Revisions to the 2017 Tax Revision (P.L. 115-97), at 1 (2021).

  113. [113]. See Marvin A. Chirelstein & Lawrence A. Zelenak, Tax Shelters and the Search for a Silver Bullet, 105 Colum. L. Rev. 1939, 1964 (2005) (“As a matter of tax logic, the basis of stock received in a § 351 incorporation cannot exceed the value of that stock—$1 million in Black & Decker
    —unless the value of the property transferred had declined in the hands of the transferor prior to the transfer”); McCormack, supra note 6, at 751.

  114. [114]. Chirelstein and Zelenak acknowledge this exception in their discussion of “tax logic.” Chirelstein & Zelenak, supra note 113, at 1964. By default, the corporation’s “inside” basis with respect to the loss assets would be reduced to the fair market value of the assets at the time of the contribution, so the loss would not be duplicated. I.R.C. § 362(e) (2018). If Black & Decker were thought to give rise to a double tax benefit, a general principle against double tax benefits might be cited. However, based on case law at the time, it was unclear whether Black & Decker would be able to deduct the contingent expenses as they came due. See Ethan Yale, Reexamining Black & Decker’s Contingent Liability Tax Shelter, 108 Tax Notes 223, 223 (2005); McCormack, supra note 6, at 752–53. The rule against double tax benefits is sufficiently widely known that I have called it a canon of construction elsewhere, so that this rule is included in doctrinalist purposivism, as discussed below.

  115. [115]. Burke, supra note 62, at 214, 216.

  116. [116]. McCormack, supra note 6, at 751–52. The example of Taxpayer Terry is based on a thoughtful example that McCormack provides in her article.

  117. [117]. Geier, supra note 12, at 497.

  118. [118]. See generally Choi, Substantive Canons, supra note 18 (discussing differences in judicial and scholar interpretation of tax statutes); Zelinsky, supra note 23 (offering a rebuttal to Geier’s preference for courts acting proactively in tax controversies); Livingston, Practical Reason, supra note 17 (discussing judicial interpretation of tax statutes).

  119. [119]. Joel Slemrod & Jon Bakija, Taxing Ourselves: A Citizen’s Guide to the Debate over Taxes 28 (4th ed. 2008) (“Economists’ standard definition [of income]—called Haig-Simons income after the two people who first developed it—is ‘the increase in an individual’s to consume during a given period of time.’ In other words, your annual income is the value of the goods and services you consume during a year, plus the net change in your wealth (saving) that occurs in that year.”).

  120. [120]. E.g., Kristin E. Hickman, Administering the Tax System We Have, 63 Duke L.J. 1717, 1721
    –22 (2014) (“The I.R.C. now contains hundreds of tax expenditure items representing more than $1 trillion of indirect government spending each year.”).

  121. [121]. See David A. Weisbach, Line Drawing, Doctrine, and Efficiency in the Tax Law, 84 Cornell L. Rev. 1627, 1627–28 (1999) (“[L]ine drawing in the tax law can and should be based on the efficiency of competing rules rather than on doctrinal concerns or traditional tax policy”).

  122. [122]. As a cultural aside, it may be unsurprising that tax professors are the most enthusiastic proponents of structuralism since their work requires them to distill complicated tax statutes into simple organizing principles for the benefit of students. Even if structuralism is descriptively incorrect, it may still be a useful pedagogical tool for students, as evidenced by Geier’s excellent casebook. See generally Deborah A. Geier, U.S. Federal Income Taxation of Individuals 2022 (2022) (reviewing the U.S. federal income tax code through a structural lens).

  123. [123]. Eskridge and Frickey argued that statutory purpose was frequently “indeterminate”: that statutes reflect “complex compromises” between “a range of interest groups, each of which will have their own reasons for supporting, or at least not opposing, the statute.” Eskridge & Frickey, supra note 17, at 335.

  124. [124]. See Livingston, Practical Reason, supra note 17, at 681–82. Michael Livingston applied this general argument specifically to tax law, arguing “statutes have multiple purposes, and ... purposive analysis is likely to be indeterminate in hard cases.” Id. at 706.

  125. [125]. Eskridge, supra note 53, at 148 (1994). Eskridge discussed how “public values,” as reflected in canons of construction, provide an alternative to coherence-based theories of interpretation. Id. at 148–51.

  126. [126]. This parallels similar arguments made by proponents of legal-process purposivism against the classic public choice critique. See generally Hart & Sacks, supra note 6 (describing the “benevolent presumption . . . that the legislature is made up of reasonable men pursuing reasonable purposes reasonably”).

  127. [127]. Of course, this argument against structural purposivism in tax law is not a general argument against Dworkinian constructive interpretation. Cf. Dworkin, supra note 45, at 52 (discussing constructive interpretation). An advocate of constructive interpretation in all areas of law might argue that its prescriptions carry into tax law as well. This Article makes a more modest argument against sui generis structural purposivism in tax law.

  128. [128]. I.R.C. § 475(a) (2020).

  129. [129]. Ari Glogower, Taxing Capital Appreciation, 70 Tax L. Rev. 111, 128–33 (2016) (describing past proposals for mark-to-market taxation).

  130. [130]. Jonathan H. Choi, A Survey of Law Professors on Tax Reform, Yale J. Reg.: Notice & Comment (2021), []. Note that a majority of professors who expressed an opinion agreed; counting those who indicated they were neutral or had no opinion, slightly less than 50 percent of professors agreed.

  131. [131]. See Weisbach, supra note 121, at 1633–37 (arguing that the platonic concept of realization is uninformative in determining which actual transactions are or are not realization events).

  132. [132]. Cottage Sav. Ass’n v. Comm’r, 499 U.S. 554, 556–57 (1991).

  133. [133]. McCormack, supra note 6, at 723.

  134. [134]. Cottage Sav., 499 U.S. at 560–62. Cottage Savings is another example of a transaction with shelter-like qualities that might have been deemed illegitimate by tax purposivists but was ruled legitimate by the Supreme Court, which found in favor of the taxpayer.

  135. [135]. Anthony J. Sebok, Reading The Legal Process, 94 Mich. L. Rev. 1571, 1574 (1996) (Book Review ) (“Hart and Sacks’s theory of statutory interpretation ... depended critically on the presumption that procedures existed that could identify the purposes selected by the legislature without actually substantively evaluating those purposes.”); Cottage Sav., 499 U.S. at 568.

  136. [136]. Compaq Computer Corp. v. Comm’r, 277 F.3d 778, 779 (5th Cir. 2001), rev’g 113 T.C. 214 (T.C. 1999).

  137. [137]. Id. at 779–80. Technically, Compaq bought and sold American Depository Receipts (“ADRs”), which represent ownership of foreign corporation stock. “Foreign stocks are customarily traded on U.S. stock exchanges using ADRs.” Id. at 779.

  138. [138]. Id. at 780.

  139. [139]. The post-tax profit was not as simple as canceling the loss from the sale against the net dividend and then subtracting the $1.5 million in transaction costs from the $3.4 million foreign tax credit, because Compaq was obligated to pay U.S. tax on the gross dividend, rather than the net dividend. McCormack, supra note 6, at 763 & n.390.

  140. [140]. George K. Yin, The Problem of Corporate Tax Shelters: Uncertain Dimensions, Unwise Approaches, 55 Tax L. Rev. 405, 407 (2002).

  141. [141]. McCormack, supra note 6, at 765.

  142. [142]. Id. (internal quotation marks omitted) (quoting S. Rep. No. 106-120, at 135 (1999)).

  143. [143]. Id. at 765–66.

  144. [144]. Id. at 764–68. McCormack also argues that the statutory purpose of the foreign tax credit does not prohibit Compaq from recouping the cost of the foreign tax elsewhere. Id. at 769–71.

  145. [145]. Id. at 770.

  146. [146]. See Daniel N. Shaviro & David A. Weisbach, The Fifth Circuit Gets it Wrong in Compaq v. Commissioner, 94 Tax Notes 511, 513–14 (2002).

  147. [147]. Cf. Michael J. Graetz, 100 Million Unnecessary Returns: A Simple, Fair, and Competitive Tax Plan for the United States 116 (2010) (“[A] tax shelter is a deal done by very smart people that, absent tax considerations, would be very stupid.”).

  148. [148]. See, e.g., Black & Decker Corp. v. United States, 436 F.3d 431, 432 (4th Cir. 2006).

  149. [149]. For an example of such tax avoidance, see the factual recitation in Gitlitz v. Comm’r, 531 U.S. 206, 210 (2001).

  150. [150]. Other scholars have also criticized analysis of whether a structure is a “tax shelter” as a trivial semantic debate. E.g., Lederman, supra note 8, at 402 (“[T]he real question is not whether a transaction is a ‘tax shelter’ but rather whether the claimed tax results are consistent with the intent of Congress.”).

  151. [151]. This argument parallels David Weisbach’s influential thesis that lines should be drawn between different activities (for example, whether an instrument is debt or equity) on grounds of efficiency rather than on conventional doctrinalist grounds (for example, whether the instrument meets the general legal definition of debt). See Weisbach, supra note 121, at 1627–31.

  152. [152]. Some purposivists might take subsequent legislative inaction as tacit approval of an interpretation of the tax code. Thus, although check-the-box elections and double-dummy mergers may not have been legitimate at the outset, they may have been implicitly legitimized over time because they were not legislatively overturned.

  153. [153]. See Zelenak, supra note 23, at 666–70 (discussing a sample of cases where the government mostly argued in favor of “nonliteral” interpretations, while the taxpayer mostly relied on the plain text of the statute to try to claim benefits). Examples from case law include the recent Summa Holdings and Benenson cases, in which several circuit courts blessed a seemingly abusive tax shelter on textualist grounds. Summa Holdings, Inc. v. Comm’r, 848 F.3d 779, 789–90 (6th Cir. 2017); Benenson v. Comm’r, 887 F.3d 511, 523 (1st Cir. 2018); Benenson v. Comm’r, 910 F.3d 690, 698–700 (2d Cir. 2018).

  154. [154]. See supra Section II.B.1.

  155. [155]. In this sense, strict constructionism differs from the existing widely recognized rule against implied tax exemptions. Choi, Substantive Canons, supra note 18, at 251–54 (describing the rule). That rule operates as a mere tiebreaker when none of the conventional tools of statutory interpretation can select between two competing interpretations. Strict constructionism is a much stronger thumb on the scale against tax benefits.

  156. [156]. The structuralist approach is similar to a kind of reasoned elaboration, attempting to create the most coherent image of the tax code as a whole, which Hart and Sacks endorse in their legal process materials. See Jones, supra note 44, at 53 (providing a structuralist perspective based on Hart and Sacks’s theory of “reasoned elaboration”); Hart & Sacks, supra note 6, at 143–58 (applying reasoned elaboration in statutory interpretation). At other times, Hart and Sacks appear to endorse a kind of broad, pragmatic normative inquiry. Hart & Sacks, supra note 6, at 102 (describing the ultimate purpose of all statutory interpretation as “establishing, maintaining and perfecting the conditions necessary for community life to perform its role in the complete development of man”). At still other times, Hart and Sacks endorsed the sort of canons of interpretation that this Article describes as doctrinal. See Manning, supra note 47, at 89 n.67 (“Among other things, the Legal Process approach gave unflinching effect to the sort of substantive canons t[h]at judges have developed over time ‘to promote objectives of the legal system which transcend the wishes of any particular session of the legislature.’”); Hart & Sacks, supra note 6, at 1376–77, 1380 (instructing interpreters to apply clear statement rules and presumptions). Each of these approaches is consistent with Hart and Sacks’s theory of “more general and thus more nearly ultimate purposes of the law.” Hart & Sacks, supra note 6, at 148.

  157. [157]. The term “pragmatism” has a rich history in law and philosophy, and pragmatic purposivism should be distinguished from some other schools of thought also described as pragmatic. In particular, it is distinct from philosophical pragmatism as originally expounded by philosophers including Charles Sanders Peirce and William James. See, e.g., Catherine Legg, Pragmatism, Stan. Encyclopedia of Phil. (Apr. 6, 2021),
    agmatism [] (describing pragmatism). It is also distinct from the “practical reason” school of statutory interpretation advanced by William Eskridge and Philip Frickey and applied to tax law by Michael Livingston. For a discussion of this school of statutory interpretation, see generally Eskridge & Frickey, supra note 17; and Livingston, Practical Reason, supra note 17.

    Pragmatism as a general judicial philosophy is most often associated with Judge Richard Posner. But Posner was a radical pragmatist rather than a purposivist searching for a methodology to complement immediate statutory purpose. Posner argued that purposivism “runs the risk of attributing to legislation not the purposes reasonably inferable from the legislation itself, but the judge’s own conceptions of the public interest.” Richard A. Posner, Statutory Interpretation—In the Classroom and in the Courtroom, 50 U. Chi. L. Rev. 800, 819 (1983) [hereinafter Posner, Statutory Interpretation]. This suggests that he believed pragmatism was not merely preferable, but inevitable. See id. at 820 (“[M]ethods of imputing congressional intent are artificial; and as I argued earlier, it is not healthy for the judge to conceal from himself that he is being creative when he is, as sometimes he has to be even when applying statutes.”). Especially in his later work, he urged judges “not to worry initially about doctrine, precedent, and the other conventional materials of legal analysis, but instead to try to figure out the sensible solution to the problem[.]” Richard A. Posner, The Federal Judiciary: Strengths and Weaknesses 80 (2017).

    Posner’s earlier work was relatively closer to what I call pragmatic purposivism. His theory of “imaginative reconstruction,” which requires the judge to “try to think his way as best he can into the minds of the enacting legislators and imagine how they would have wanted the statute applied to the case at bar,” is reminiscent of Breyer’s “reasonable legislator.” Posner, supra at 817; Breyer, supra note 6, at 98–101. Yet Posner was critical of the famous Hart and Sacks line that a court “should assume, unless the contrary unmistakably appears, that the legislature was made up of reasonable persons pursuing reasonable purposes reasonably.” Posner, Statutory Interpretation, supra at 819 (internal quotation marks omitted) (quoting Hart & Sacks, supra note 6, at 1415). He argued, instead, that often “lines of compromise are not clear,” and in those cases “[i]t is inevitable, and therefore legitimate, for the judge in such a case to be moved by considerations that cannot be referred back to legislative purpose.” Id. at 820. These considerations could include “judicial administrability” or “the public interest.” Id.

  158. [158]. See Robert J. Pushaw, Jr., Talking Textualism, Practicing Pragmatism: Rethinking the Supreme Court’s Approach to Statutory Interpretation, 51 Ga. L. Rev. 121, 123–24 (2016). Although this Article describes pragmatism in consequentialist terms, and in practice most of its proponents seem to be consequentialist, in theory an alternative normative framework could be used, like deontology or virtue ethics.

  159. [159]. Ken I. Kersch, Justice Breyer’s Mandarin Liberty, 73 U. Chi. L. Rev. 759, 767–68 (2006) (reviewing Stephen Breyer, Active Liberty: Interpreting Our Democratic Constitution (2005)).

  160. [160]. Breyer, supra note 6, at 87–88 (internal quotation marks omitted).

  161. [161]. Id. at 96 (“It is unlikely that anyone in Congress thought about this question, for it is highly technical.”).

  162. [162]. Id. at 95–101 (discussing the case [“Case Three”] of an ambiguous statute and applying pragmatic criteria in the absence of statutory guidance).

  163. [163]. Cass R. Sunstein, Justice Breyer’s Democratic Pragmatism, 115 Yale L.J. 1719, 1720 (2006) (reviewing Stephen Breyer, Active Liberty: Interpreting Our Democratic Constitution (2005)) (describing the “pragmatic dimension” of Breyer’s jurisprudence); Kersch, supra note 159, at 763 (describing Breyer’s emphasis on “purposes and consequences”).

  164. [164]. In the notice that first introduced the check-the-box regime, the IRS stated that “the purpose of this approach is to simplify the rules in order to reduce the burdens on both taxpayers and the Service.” I.R.S. Notice 95-14, 1995-1 C.B. 297. However, this is not the only outcome that could potentially be justified on normative grounds. For example, if the IRS were to take a general anti-business stance, it might have continued the old formalistic entity classification regime in spite of the administrative costs. One problem with pragmatism is that different observers will vary in what they consider pragmatic or socially desirable.

  165. [165]. Choi, Empirical Study, supra note 21, at 397–98. That article considered the sources that the Treasury tends to cite in justifying its published guidance, concluding that the Treasury frequently cites legislative history and other tools generally associated with purposivism. Id. However, the degree of the Treasury’s true fidelity to statutory purpose cannot be discerned from citations to legislative history alone. It could be, for example, that the Treasury cites legislative history only when useful to support conclusions that it prefers for pragmatic policy reasons.

  166. [166]. Id. at 392–95.

  167. [167]. Jonathan H. Choi, Legal Analysis, Policy Analysis, and the Price of Deference: An Empirical Study of Mayo and Chevron, 38 Yale J. on Reg. 818, 823–24 (2021) [hereinafter Choi, Legal Analysis] (finding that a shift toward Chevron deference increased the frequency of normative terms used by the Treasury in regulatory preambles).

  168. [168]. Peter L. Strauss, Essay, “Deference” Is Too Confusing—Let’s Call them “Chevron Space” and “Skidmore Weight, 112 Colum. L. Rev. 1143, 1163–64 (2012); see also Yehonatan Givati, Strategic Statutory Interpretation by Administrative Agencies, 12 Am. L. & Econ. Rev. 95, 96 (2010) (“In the model, the agency, which maximizes some objective function, adopts a rule that interprets a statute . . . .”); Matthew C. Stephenson, The Strategic Substitution Effect: Textual Plausibility, Procedural Formality, and Judicial Review of Agency Statutory Interpretations, 120 Harv. L. Rev. 528, 536, 544 (2006) (assuming that agencies “are interpretive instrumentalists, attaching no intrinsic importance to textual fidelity or analogous concerns” but instead attempting to “secure whatever interpretation would best advance [their] substantive policy agenda”); John R. Wright, Ambiguous Statutes and Judicial Deference to Federal Agencies, 22 J. Theoretical Pols. 217, 226–29 (2010) (modelling agency action as a function of policy goals).

    The Supreme Court has recently trended toward skepticism of unlimited deference to agencies, even within the Chevron space. See Michigan v. EPA, 576 U.S. 743, 750 (2015) (“‘Not only must an agency’s decreed result be within the scope of its lawful authority, but the process by which it reaches that result must be logical and rational.’ It follows that agency action is lawful only if it rests ‘on a consideration of the relevant factors.’” (citation omitted) (quoting Motor Vehicle Mfrs. Ass’n. of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983))); Encino Motorcars, L.L.C. v. Navarro, 579 U.S. 211, 221 (2016) (ruling that a regulation can be procedurally defective if the agency does not “give adequate reasons for its decisions”). It remains to be seen how much the Supreme Court will actually curb agency discretion, perhaps by engaging in a “hard look” review in Chevron step two.

  169. [169]. Or, formerly, National Muffler deference, a form of deference specific to tax law which defers to regulations so long as they “implement[ed] the congressional mandate in some reasonable manner.” Nat’l Muffler Dealers Ass’n v. United States, 440 U.S. 472, 476 (1979) (internal quotation marks omitted) (quoting United States v. Cartwright,411 U.S. 546, 550 (1973)). Despite its superficial similarity to Chevron deference, National Muffler deference was generally thought less deferential than Chevron deference. Choi, Legal Analysis, supra note 167, at 838 (“Because National Muffler’s factors are so specific, it was thought to be less deferential than Chevron, which set out a relatively vague standard for reasonableness.”). National Muffler deference was abolished by the Supreme Court’s decision in Mayo. Mayo Found. for Med. Educ. & Rsch. v. United States, 562 U.S. 44, 55 (2011).

  170. [170]. I.R.C. § 7701(o) (2018).

  171. [171]. See generally Choi, Substantive Canons, supra note 18 (arguing that anti-abuse doctrines in tax law should be considered canons of construction, paralleling canons used in other areas of law).

  172. [172]. See id. at 199–201. In that article, I did not refer explicitly to the “immediate” purpose of a statute, since the framework was written more broadly to accommodate textualists as well as purposivists. However, this is the interpretation of the framework in that article most appropriate to modern purposivism.

  173. [173]. Anita S. Krishnakumar, Reconsidering Substantive Canons, 84 U. Chi. L. Rev. 825, 826–27 (2017) (“There is a popular belief among statutory interpretation scholars that substantive canons of statutory construction ... act as an ‘escape valve’ that helps textualist judges eschew, or ‘mitigate,’ the rigors of textualism.”). Of course, some substantive canons appeal more to textualists (like federalism or avoidance canons) than others. Choi, Substantive Canons, supra note 18, at 231.

  174. [174]. Hart & Sacks, supra note 6, at 1380; see also id. at 1210 (“Can the body of statutory law ever attain any semblance of rationality and consistency unless the courts continue unremittingly the effort to discern and articulate principles such as these?”).

    Presumptions are just one type of substantive canon. A form of substantive canon that plays an even more important role in Hart and Sacks’s typology is the clear statement rule, which can only be rebutted by a “clear statement” in the text of the statute. According to Hart and Sacks, “these policies of clear statement may on occasion operate to defeat the actual, consciously held intention of particular legislators, or of the members of the legislature generally.” Id. at 1376; see also William N. Eskridge, Jr. & Philip P. Frickey, Quasi-Constitutional Law: Clear Statement Rules as Constitutional Lawmaking, 45 Vand. L. Rev. 593, 597 (1992) (“‘[C]lear statement rules’ . . . can only be rebutted by clear statutory text.”). When Hart and Sacks create a “[c]oncise [s]tatement” of the task of statutory interpretation, they devote one out of the four clauses to the importance of clear statement rules. See Hart & Sacks, supra note 6, at 1374.

  175. [175]. Substantive canons vary in how rule-like they are, and, in some respects, they resemble standards more than rules. An exceptionally broad and therefore less formal canon is the substance-over-form doctrine, which broadly mandates that transactions be taxed in accordance with their substance rather than their form. See, e.g., Gregory v. Helvering, 293 U.S. 465, 469 (1935); United States v. Phellis, 257 U.S. 156, 168 (1921); Estate of Weinert v. Comm’r, 294 F.2d 750, 755 (5th Cir. 1961). Because it is relatively ambiguous and the scope of its application is relatively less clear, the substance-over-form doctrine has arguably been used as a fig leaf for decisions that are ultimately pragmatic rather than doctrinal, to satisfy the urges of judges to root their decisions in precedent while actually reaching a sensible policy result.

  176. [176]. Black & Decker Corp. v. United States, 436 F.3d 431, 441 (4th Cir. 2006) (“Taxpayer conceded for purposes of deciding the motion that ‘tax avoidance was the sole motivation underlying Black & Decker’s decision to outsource its healthcare management function to BDHMI.’”). Because Black & Decker was decided before the economic substance doctrine was codified, the Court applied a variant on the doctrine known as the “sham transaction doctrine.” Id. at 440. The sham transaction doctrine invalidates a transaction if both of its prongs are violated, as opposed to the modern economic substance doctrine, which is failed if either of its prongs are violated. See id. at 441 (describing the test to reject a transaction as a sham as conjunctive, meaning that both prongs would need to be failed for the transaction to be rejected). The first prong of the sham transaction doctrine requires a good business purpose, which Black & Decker seemed to lack. Id. The second is violated if “no reasonable possibility of a profit exists,” which also seems to have been the case for Black & Decker. See id. (citing Rice’s Toyota World, Inc. v. Comm’r, 752 F.2d 89, 91 (4th Cir. 1985)).

    Note that Black & Decker’s admission that it lacked a good business purpose for the transaction was primarily a consequence of the procedural posture of its case before the Fourth Circuit. The trial court had granted Black & Decker summary judgment in its favor, and for purposes of that summary judgment, Black & Decker had stipulated that it lacked a good business purpose. See id. It is possible that after the Fourth Circuit remanded the case for a factual determination of whether the sham transaction ought to apply, Black & Decker could have successfully argued that it did have a good business purpose. Moreover, it is possible that, even if Black & Decker failed in this particular case, other taxpayers could be more careful to provide plausible business purposes in light of the lessons from Black & Decker, making the doctrinalist case against tax shelters more difficult.

    Note that Black & Decker’s admission that it lacked a good business purpose for the transaction was primarily a consequence of the procedural posture of its case before the Fourth Circuit. The trial court had granted Black & Decker summary judgment in its favor, and for purposes of that summary judgment, Black & Decker had stipulated that it lacked a good business purpose.

  177. [177]. The prominence of substantive canons in tax law is a notable case of strong methodological stare decisis. Many scholars believe that courts do not generally give precedential weight to doctrines of statutory interpretation. See, e.g., Abbe R. Gluck, The States as Laboratories of Statutory Interpretation: Methodological Consensus and the New Modified Textualism, 119 Yale L.J. 1750, 1754 (2010) (“Methodological stare decisis—the practice of giving precedential effect to judicial statements about methodology—is generally absent from the jurisprudence of mainstream federal statutory interpretation ... .”); Stephen M. Rich, A Matter of Perspective: Textualism, Stare Decisis, and Federal Employment Discrimination Law, 87 S. Cal. L. Rev. 1197, 1197 (2014) (“When the Supreme Court rules on matters of statutory interpretation, it does not establish ‘methodological precedents.’ The Court is not bound to follow interpretive practices employed in a prior case even if successive cases concern the same statute.” (footnote omitted)). One recent article by Aaron-Andrew Bruhl concludes that “evidence of [methodological precedent] is pervasive in the lower courts.” Aaron-Andrew P. Bruhl, Eager to Follow: Methodological Precedent in Statutory Interpretation, 99 N.C. L. Rev. 101, 106 (2020). Bruhl contrasts methodological stare decisis at lower courts with the Supreme Court’s relative willingness to overturn methodological precedent. Tax law doctrine is primarily determined by lower courts, since Supreme Court tax cases remain rare. Id. This may explain the persistence of methodological precedents like the anti-abuse doctrines in tax law. The substantive canons are notable within this conversation because they are not precedents specific to any particular statute but are read across the tax code as a whole, consistent with Bruhl’s thesis. See id. at 106–07.

  178. [178]. Each of the substantive tax canons has substantially evolved over time, generally becoming more specific and rule-like over the decades. The economic substance doctrine, for example, has developed from a broad requirement that transactions must have economic substance, to a specific two-prong test. See Choi, Substantive Canons, supra note 18, at 205–06.

  179. [179]. E.g., Summa Holdings, Inc. v. Comm’r, 848 F.3d 779, 782 (6th Cir. 2017) (“Each word of the ‘substance-over-form doctrine,’ at least as the Commissioner has used it here, should give pause... . ‘Form’ is ‘substance’ when it comes to law. The words of law (its form) determine content (its substance).”).

  180. [180]. In particular, courts often attempt to reject tax shelters under a doctrine like the substance-over-form doctrine. E.g., Summa Holdings, Inc. v. Comm’r, T.C.M. 2015-119, 2015 WL 3943219, at *7 (T.C. June 29, 2015), rev’d, 848 F.3d 779, and rev’d sub nom. Benenson v. Comm’r, 887 F.3d 511 (1st Cir. 2018), and rev’d sub nom. Benenson v. Comm’r, 910 F.3d 690 (2d Cir. 2018). The Fourth Circuit, in Black & Decker, initially declined to reject the taxpayer’s tax shelter on the basis of statutory purpose alone. 436 F.3d at 437 (“The legislative history argument does not persuade us... . [W]e find no ambiguity in the statute that requires us to parse the congressional record ... .”). However, the court subsequently indicated that the shelter might fail under the sham transaction doctrine. See id. at 440–43.

  181. [181]. Choi, Substantive Canons, supra note 18, at 235 (“[A]lthough this Article focuses on substantive canons whose legitimacy comes from their status as widely accepted background norms, there are other theories that would endorse a wider swath of canons.”).

  182. [182]. William Baude & Stephen E. Sachs, The Law of Interpretation, 130 Harv. L. Rev. 1079, 1123 (2017) (“[L]egal canons operate if—indeed, especially if—the drafters are unaware of them.”).

  183. [183]. David P. Hariton, The Frame Game: How Defining the “Transaction” Decides the Case, 63 Tax Law. 1, 8–13 (2009).

  184. [184]. E.g., id. at 3–4 (“[T]he taxpayers used the Son of Boss [tax shelter] not to generate a taxable loss in the absence of any economic loss, but rather to step-up the basis of property they were about to sell and thereby avoid a taxable gain.”).

  185. [185]. Id. at 27.

  186. [186]. Id. at 35–42.

  187. [187]. See Eric L. Talley, Corporate Inversions and the Unbundling of Regulatory Competition, 101 Va. L. Rev. 1649, 1650–51 (2015) (describing the “wave” of multinational corporations inverting abroad).

  188. [188]. The first generation of inversion transactions involved jurisdictions like the Cayman Islands; as these jurisdictions became recognized as tax havens, the most recent generation of inversions tended to favor countries like Ireland, which is importantly a member of the European Union in which corporations could make a stronger claim to conducting actual business activity. See id. at 1669–72.

  189. [189]. Inversions and Related Transactions, 81 Fed. Reg. 20,858, 20,873–81 (Apr. 8, 2016) (to be codified at 26 C.F.R. pt. 1); Inversions and Related Transactions, 83 Fed. Reg. 32,524, 32,530 (July 12, 2018) (to be codified at 26 C.F.R. pt. 1).

  190. [190]. Peter Canellos distinguishes between the “tax bar” and the “tax shelter bar.” Peter C. Canellos, A Tax Practitioner’s Perspective on Substance, Form and Business Purpose in Structuring Business Transactions and in Tax Shelters, 54 SMU L. Rev. 47, 55–57 (2001); see also Joseph Bankman, The Business Purpose Doctrine and the Sociology of Tax, 54 SMU L. Rev. 149, 150 (2001) (“Most members and leaders of the New York State Bar Association regard the shelter phenomenon as deplorable ... . This side of the tax bar would readily push the (alas, nonexistent) button that would eliminate the ‘modern’ shelter, even if so doing would deprive them of this slight part of their current practice.”).

  191. [191]. Mayo Found. for Med. Educ. & Rsch. v. United States, 562 U.S. 44, 56 (2011) (“We see no reason why our review of tax regulations should not be guided by agency expertise pursuant to Chevron to the same extent as our review of other regulations.”). Some scholars have criticized the Treasury’s current practice of making “temporary” regulations effective prior to notice and comments. See Leandra Lederman, The Fight Over “Fighting Regs” and Judicial Deference in Tax Litigation, 92 B.U. L. Rev. 643, 662–63 (2012); Kristin E. Hickman, Coloring Outside the Lines: Examining Treasury’s (Lack of) Compliance with Administrative Procedure Act Rulemaking Requirements, 82 Notre Dame L. Rev. 1727, 1760 (2007) (“Treasury is not the only agency that promulgates binding regulations in advance of seeking and considering public comments. Nevertheless, the courts generally consider regulations issued through such a process procedurally invalid unless one of the four exceptions listed in APA section 553 applies. Many if not most Treasury regulations do not fall within the scope of those exceptions.”) (footnotes omitted).

  192. [192]. Michael I. Saltzman & Leslie Book, IRS Practice and Procedure ¶ 3.03 (2021) (“Prior to the Supreme Court’s decision in Mead, some courts applied Chevron deference to revenue rulings while others gave no deference whatsoever. After Mead, the general consensus is that Skidmore is the more appropriate standard . . . . The Supreme Court itself, however, has not expressly ruled on the question . . . .”) (footnotes omitted); Choi, Empirical Study, supra note 21, at 373 n.39 (“[I]t is widely believed that Skidmore deference applies to IRS subregulatory guidance . . . .”).

  193. [193]. This Article emphasizes normative justifications to prefer one methodology over another, especially predictability, fairness, and the protection of the public fisc. I believe that it also conforms to constitutional constraints on agency and court action, but that discussion falls outside the scope of this Article.

  194. [194]. Most prominent tax shelters will fall into the ex-post camp because they tend to involve novel tax shelters planned by creative tax lawyers. The fact that tax shelters might be addressed using doctrinalist doctrines ex post, and then subsequently prohibited by the application of pragmatic doctrines ex ante, raises interesting questions about the prospectivity of new interpretations of law that would be an interesting subject for future research.

  195. [195]. Manning & Stephenson, supra note 49, at 830 (“The key feature of a rule, at least under the APA definition, is that a rule is future-oriented, whereas adjudication concerns events that happened in the past.”).

  196. [196]. Many scholars have emphasized the institutional competence of agencies in using pragmatic decision-making to interpret statutes. William N. Eskridge, Jr., Expanding Chevron’s Domain: A Comparative Institutional Analysis of the Relative Competence of Courts and Agencies to Interpret Statutes, 2013 Wis. L. Rev. 411, 420–23 (arguing that Chevron should be expanded to cover a wider range of agency interpretations in light of agencies’ superior interpretive competence); Cass R. Sunstein & Adrian Vermeule, The Unbearable Rightness of Auer, 84 U. Chi. L. Rev. 297, 305 (2017) (“When a statute is unclear, and especially when a complex modern regulatory statute is unclear, resolution of the ambiguity will inevitably require policy-making competence—which courts lack and which agencies have.”); Cass R. Sunstein & Adrian Vermeule, Interpretation and Institutions, 101 Mich. L. Rev. 885, 928–30 (2003) (describing, with approval, an EPA regulation formulated on pragmatic public policy rather than purely statutory grounds). These scholars build on the “comparative institutional” approach first developed by Neal Komesar. See generally Neil Komesar, In Search of a General Approach to Legal Analysis: A Comparative Institutional Alternative, 79 Mich. L. Rev. 1350 (1981) (articulating and justifying the “comparative institutional” method of analysis).

  197. [197]. One possible explanation for the status quo is that it reflects an intuition that there exists a single best interpretation for any particular statute. The best-known articulation of this view may be Ronald Dworkin’s “right answer” thesis. Ronald Dworkin, A Matter of Principle 119–21 (1985). In contrast, the hybrid approach that I sketch could lead to different interpretations of the law depending on context. In practice, ex ante and ex post contexts are sufficiently dissimilar that the distinction should not lead to confusion, but to some, the essential concept may seem unintuitive, not “law-like,” and contrary to some schools of thought, like Dworkin’s.

  198. [198]. Some of these doctrines persist as common law rules, like the step transaction doctrine. True v. United States, 190 F.3d 1165, 1174–77 (10th Cir. 1999) (applying the step transaction doctrine); Associated Wholesale Grocers, Inc. v. United States, 927 F.2d 1517, 1521–26 (10th Cir. 1991) (same). Some have been codified, like the economic substance doctrine. See I.R.C.
    § 7701(o) (2018).

  199. [199]. Preambles to tax regulations routinely identify policy rationales supporting the position reflected in the regulation. See, e.g., Guidance Under Section 1061, 86 Fed. Reg. 5452, 5463, 5465 (Jan. 19, 2021) (to be codified at 26 C.F.R. pt. 1) (discussing a variety of policy concerns, including the “complexity” of the regulations and “the potential for abuse”).

  200. [200]. Eskridge & Frickey, supra note 17, at 322 n.3. However, my approach still differs from Eskridge and Frickey’s in the sources of authority it emphasizes. In the case of doctrinal formalism, it recommends greater emphasis on canons of construction than do Eskridge and Frickey. Moreover, I suggest that courts and the Treasury should still hew to statutory purposes when they are “clear,” which may be reminiscent of the Eskridge/Frickey “funnel of abstraction” but places much greater weight on the more concrete interpretive elements. See discussion supra notes 17, 55 (discussing the differences between this Article’s proposals and Eskridge and Frickey’s practical reason).

  201. [201]. See supra notes 20–21, 165–69 and accompanying text.

  202. [202]. The Treasury’s interpretations of its own regulations, as opposed to statutes, could theoretically receive Auer deference (also known as Seminole Rock deference). However, the Supreme Court has developed a number of limitations on Auer deference that limit the degree of deference for ex post interpretations of ambiguous regulations. See, e.g., Christopher v. SmithKline Beecham Corp., 567 U.S. 142, 156 (2012) (indicating that an agency should not receive Auer deference if its interpretation is an “unfair surprise” or the agency has failed to provide “fair warning of the conduct [a regulation] prohibits or requires” (alteration in original) (internal quotation marks omitted) (quoting Gates & Fox Co. v. Occupational Safety & Health Rev. Comm’n, 790 F.2d 154, 156 (D.C. Cir. 1986))); Udall v. Tallman, 380 U.S. 1, 17 (1965) (granting Auer deference and emphasizing that the agency’s “interpretation had, long prior to respondents’ applications, been a matter of public record and discussion”). The Tax Court has similarly limited Auer deference to situations where the Treasury’s interpretation “is a matter of public record and is an interpretation upon which the public is entitled to rely when planning their affairs.” CSI Hydrostatic Testers, Inc. v. Comm’r, 103 T.C. 398, 409 (1994), aff’d 62 F.3d 136 (5th Cir. 1995); see also S. Pac. Transp. Co. v. Comm’r, 75 T.C. 497, 541 (1980) (“Respondent is relying on the general rule that an administrative agency's interpretation of its own regulation or other directive is controlling unless plainly erroneous or inconsistent with the directive.”). Moreover, the Department of the Treasury and the IRS have recently stated that “[i]n litigation before the U.S. Tax Court, as a matter of policy, the IRS will not seek judicial deference under Auer . . . or Chevron . . . to interpretations set forth only in subregulatory guidance.” Dep’t of Treas., Policy Statement on the Tax Regulatory Process 2 (2019) (citations omitted), https://home.treasu [https://perma.
    cc/XQL7-2YNN]. In general, scholars and judges have begun to criticize Auer deference more vociferously in recent years. See generally Christopher J. Walker, Attacking Auer and Chevron Deference: A Literature Review, 16 Geo. J.L. & Pub. Pol’y 103 (2018) (outlining arguments against the two deference doctrines).

  203. [203]. An interesting twist is the interaction between judicial interpretations and agency guidance. If a court has interpreted a statute one way, does an agency have the subsequent authority to overturn that interpretation in rulemaking? The Supreme Court answered this question in Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967 (2005), which held that if the underlying statute is ambiguous (the first step of Chevron analysis), then an agency may overrule a court’s interpretation of that ambiguous statute. Id. at 991–92. Whether or not this ruling is constitutionally valid exceeds the scope of this Article, but as a normative matter Brand X reflects the compromise embodied in this Article’s ex ante/ex post framework. Within the bounds of statutory purpose, an agency should be able to issue prospective guidance on pragmatic grounds even if that guidance would not have been appropriate in an ex post judgment.

  204. [204]. A parallel concern might be that the Treasury is incapable of applying pragmatism effectively, so that even at the Treasury doctrinalism is preferable. This would only be the case if we take a very dim, in my opinion an unrealistically dim, view of Treasury competence.

  205. [205]. Choi, Substantive Canons, supra note 18, at 215 (“The IRS ... is the sole federal authority that the vast majority of taxpayers will ever deal with ... .”).


Associate Professor, University of Minnesota Law School.

Thanks to Reuven Avi-Yonah, Allison Christians, Kristin Hickman, Leandra Lederman, Shannon Weeks McCormack, Amy Monahan, Patrick Murray, Gregg Polsky, Jim Repetti, Steve Shay, Sloan Speck, Richard Stanger, Jacob Victor, Clint Wallace, Larry Zelenak, and Ed Zelinsky, as well as the participants in the New York University Lawyering Scholarship Colloquium, the Law and Society Association Conference, the USD-Davis-Hastings Tax Law Speaker Series, the University of Florida Tax Policy and Public Finance Colloquium, the Junior Tax Scholars Workshop, the Break into Tax series, the Indiana Maurer School of Law Tax Policy Colloquium, the University of Wisconsin Law School Faculty Workshop, the University of California Irvine Tax Policy Colloquium, the Boston College Tax Policy Workshop, and the Duke Law School Tax Policy Seminar, for their helpful comments. Last but not least, thanks to the editors of the Iowa Law Review for their excellent editorial work.