Constituencies and Control in Statutory Drafting: Interviews with Government Tax Counsels

I.     Introduction

The end of 2017 saw Congress pass the most sweeping tax legislation in over thirty years.1 The media, academics, and policymakers have derided this new tax bill based on both its substantive policies and the process of its creation. Commentators have protested the speed with which the bill passed through Congress,2 its shoddy drafting,3 its complexity,4 and its multitude of loopholes (which were often attributed, at least in part, to the speed of enactment).5 Relatedly, Democratic Senators have complained that they did not have time to read the Senate bill before it passed, and that the bill was passed in a form containing hastily scribbled handwritten notes.6 At all stages of the bill’s journey through Congress, commentators criticized the fact that the process happened too fast and that votes were taken without sufficient opportunity to consider the actual legislative language.7 In short, substantive policy aside, the passage of the 2017 tax bill brought significant attention to the tax legislative process and the way tax statutes are actually written and passed as part of that process.

Surprisingly, while there is widespread familiarity with the basics of the tax legislative process,8 there has been little in-depth research about how substantive tax policy is turned into actual statutory language.9 As a result, we have little means to understand why the tax statutes in the Internal Revenue Code (“I.R.C.” or “Code”) are drafted the way they are. For example, why does the Code sometimes state a broad general rule followed by exceptions that swallow the rule, rather than simply stating a narrower general rule? Why do exceptions sometimes appear in the same Code provision as the general rule, but are other times buried far away? How do drafters determine when to create a new defined term, as opposed to borrowing a pre-existing definition or leaving the term undefined? Who makes such choices and how are they made?

While these questions may seem dry and technical, they are anything but. The way our tax statutes are drafted can determine how accessible they are and to whom, can impact how taxpayers understand and interact with the tax system, can shape societal attitudes towards tax law, and can even affect how legislative power is allocated.

In this Article, we undertake the first extensive empirical examination of how those responsible for creating tax legislation make drafting and articulation choices, and what factors they consider when they do so. We report the findings of 26 in-depth interviews we conducted with government counsels who have participated in the tax legislative process over the last four decades, to explore how and why certain drafting and formulation decisions are made and the implications of these decisions.

We found that a majority of those interviewed did not pay much attention to how a statute was formulated, and did not think it mattered, as long as the intended rule was accurately articulated. The reason for this was that many interviewees viewed the Code as written for experts, the U.S. Department of the Treasury (“Treasury”), and software companies, rather than for ordinary taxpayers. We further found that interviewees perceived tax statutes to be sticky and prioritized preservation of existing formulations: Once a particular provision is placed in the Code, congressional dynamics and a desire to not upset settled expectations make changing prior statutory text difficult. Interviewees also reported that drafters at the House and Senate Office of Legislative Counsel took primary responsibility and control over turning policies into statutory language, and that other participants in the process almost completely deferred to Legislative Counsel on formulation issues, even though these other participants would participate in higher level substantive policy discussions and decisions.10 

However, interviewees also described the ways the tax legislative process (and within it, the drafting process) has shifted over time and varies depending on context. For example, some suggested that the process has moved from an expertise-driven model to one with more amorphous control, with more lobbyist and industry group participation. Others noted how drafting of major legislation differs from drafting in the ordinary course.

These findings have important implications. First, they affirm the findings of an important body of legal literature regarding the implications of legislative drafting practices for judicial theories of interpretation, particularly textualism.11 Notably, scholars have previously found that members of Congress (“Members”) focus on legislation at a policy level, not at the level of actual drafting of the legislative text.12 Our findings support some of the conclusions drawn in that literature. However, by revealing the sometimes idiosyncratic reasons why formulation decisions are actually made in drafting tax legislation, and exposing the strong inertial tendencies inherent in statutory formulations, our findings also call textualism and other interpretive canons even more deeply into question.

Second, our study has broad implications for the tax system, not just in the small number of cases in which statutory language is litigated, but also in the much broader circumstances where tax law is applied without judicial intervention. Our findings show how a variety of drafting dynamics—including many counsels’ views that formulation choices do not matter, the difficulty of making technical corrections or cleaning up the Code, the desire not to upset existing users’ understandings of the law, and other constraints on drafters—lead to strong inertial tendencies, whereby the law accrues in a layer-cake fashion over time and becomes ever more convoluted. These dynamics are a significant but underappreciated contributor to legal complexity, and they point to distributional tradeoffs in the law’s formulation. By focusing on the benefits of keeping the law’s formulation the same for existing users, drafters may inadvertently privilege entrenched interests relative to newcomers and laypersons who may struggle with the law’s formulation. Unearthing these dynamics allows tradeoffs between groups to be more consciously evaluated as such.

Our finding that ordinary taxpayers are not the intended audience of tax statutes also reveals a tension, whereby potentially disproportionate responsibilities are placed on taxpayers who are not expected to understand the law. Even though drafters do not expect ordinary taxpayers to understand or even read tax statutes, principal responsibility—backed up by attestation clauses and penalties—is nonetheless placed on taxpayers to fill out accurate and compliant tax returns. This is done even if the taxpayer uses a preparer or tax preparation software. Resolving this tension through rethinking of penalty systems or fundamental reforms is difficult. This serves as a caution about the costs of drafting expert-centric law that its principal subjects are not expected to understand.

Finally, our findings provide fresh insights regarding the tax legislative process, revealing a contrast between public expectations and narratives about the legislative process and how the process actually works. This raises critical questions about the legitimacy of the legislative process, and how power is allocated and exercised as part of that process.

We proceed as follows. In Part II, we summarize the process for enacting tax law, show that there are different potential approaches the drafter can choose in articulating given substantive content, and then explore the potential effects of these choices.13 In Part III, we describe our research methodology and summarize our empirical findings from the in-depth interviews we conducted with government counsels regarding tax legislative drafting.14 In Part IV, we explore the implications of our findings for judicial theories of statutory interpretation, for the tax system, and for the legislative process.15 In Part V, we conclude.16

II.     The Tax Legislative Process and the Drafting of Tax Statutes

Tax academics have long-focused on substantive tax policy choices.17 They have paid much less attention to drafting choices, or how substantive choices are formulated into actual legal content. In this Part, we discuss the types of drafting choices that drafters must make in turning substantive tax policy into statutory form, and explore why these drafting choices deserve attention.18 In order to understand drafting choices, it is important to appreciate the context in which they are made. Drafting is a part of the political process of making laws. Therefore, we first provide in Section II.A a brief overview of the tax legislative process, which is the process by which tax legislation is conceptualized, written, and (sometimes) passed by Congress.19

A.     Understanding the Tax Legislative Process

This Section provides what is often presented as the textbook tax legislative process. As scholars have increasingly recognized, there has been a rise in “unorthodox legislation,” or legislation passed through practices and procedures that do not follow the textbook process.20 In reporting our findings in Section III.B, we note situations in which the descriptions that our interviewees provided deviate from the textbook process.21

1.     Origination in the House

The U.S. Constitution provides that “[a]ll Bills for raising Revenue shall originate in the House of Representatives; but the Senate may propose or concur with Amendments as on other Bills.”22 The effect of this “origination clause” is that proposed legislation is initiated in the House. Both the House and Senate consider tax legislation via tax-writing committees. The House Ways and Means Committee is the chief tax-writing committee of the House and begins the “formal” legislative process by considering the proposed tax legislation.23 Conceptually, however, ideas for proposed legislation may come from other sources, including the Treasury Department, the Administration, the IRS, or outside interests (e.g., industry and trade associations) and their lobbyists.24

The House Committee on Ways and Means membership consists of majority and minority members (generally based on House representation).25 The process of considering legislation traditionally begins with hearings where witnesses answer Member questions.26 After hearings, the proposed legislation is considered at a Ways and Means Committee “markup” session, in which the committee “reach[es] tentative decisions on specific issues.”27 Markups are open to the public unless committee members vote to close the session.28 For example, the House markup session for the 2017 tax reform was publicly broadcast and is available online.29 Specialists have traditionally played an important role in the markup process. Staff of the Joint Committee on Taxation (“JCT”) (a nonpartisan congressional committee closely involved in the tax legislative process) assist in the development of the proposed legislation,30 as do Ways and Means Committee Staff and Member staff.31 Treasury has also traditionally attended markup sessions, supplying revenue estimates, statistical data, and other advice.32

Once the Ways and Means Committee reports the bill, the full House considers and votes on the bill, pursuant to procedures issued by the Rules Committee of the House.33 If the vote is favorable, the bill becomes an Act of the House and is then sent along to the Senate.

2.     Senate Consideration

As is the case with the House, the Senate’s consideration of tax legislation is done via a tax-writing committee, the Senate Finance Committee. Like Ways and Means, the Senate Finance Committee traditionally conducts hearings and holds a markup of its proposal.34 It may adopt the bill the House passed, or it may amend it. Senate changes may be small or large—large changes may constitute a whole new Senate bill.35 The bill (and accompanying committee report) is debated on the Senate floor, pursuant to Senate procedures.36 If the Senate version of the bill passes, and assuming that the Senate and House versions differ, the bill is returned to the House and a Conference Committee—a committee appointed by the House and Senate—is convened to reconcile the differences between the House and Senate bills. The reconciled Conference Committee version of the bill is then reported back to House and Senate, together with a Conference Committee report that explains the Conference Committee decisions.37 If both Houses approve the Conference Committee version, the enrolled bill goes to the President for signing. Absent a presidential veto, the bill becomes law.

3.     The Joint Committee on Taxation 

As noted, specialists have traditionally played an important role in the tax legislative process. Such specialists include the staff of the nonpartisan JCT.38 Established in 1926, the JCT’s role is to assist majority and minority Members of both Houses with tax legislation.39 Its staff includes tax attorneys, accountants, and economists.

By statute and practice, the JCT’s role includes investigating the operation and effects of taxes and tax administration; investigating methods of tax simplification; making reports and recommendations to the House Ways and Means Committee and Senate Finance Committee (or House and Senate) on the results of such investigations and studies; making reports on the general state of the tax system to the Ways and Means Committee and Senate Finance Committee at least once every Congress (subject to the necessary appropriations); and assisting directly in the development and formation of statutory language.40

The JCT is primarily responsible for writing the report that traditionally accompanies legislation in the House, including a detailed explanation of the bill, which forms part of the bill’s legislative history.41 In addition, the JCT is charged with “scoring” or providing official revenue estimates for tax legislation considered by either the House or Senate.42

B.     Drafting Choices

The broad-strokes description set forth above and the background literature on how tax bills are passed43do not capture how drafters make particular choices to translate substantive tax policy prescriptions into statutory language. Often, there is more than one way to articulate the same substantive tax rule, so the statutory drafter must make choices. In this Section II.B, we highlight some of the different choices that drafters face. 

1.     Rules vs. Exceptions

It is well recognized that any exception can also be formulated as a general rule.44 Frederick Schauer has famously argued that, for this reason, exceptions and general rules are indistinguishable, and exceptions therefore carry no independent jurisprudential weight.45Schauer uses the example of fornication to illustrate the point.46A rule that sex outside of marriage is illegal may be formulated in a number of ways. It can be stated as a rule:

Rule Formulation: Fornication is illegal.

However, it can also be stated as an exception:

Exception Formulation: Sex is legal, except sex outside marriage is illegal.47

The same principle applies for tax statutes.48 For instance, I.R.C. § 121 excludes from income gain on the sale of a principal residence up to $250,000.49 The exclusion is only available if the taxpayer has owned and used the property as a principal residence for at least 2 of the 5 years preceding the sale.50 The I.R.C. § 121 exclusion could be structured as a rule or an exception.

Rule Formulation: If taxpayer owns and uses a property as a principal residence for 2 of last 5 years, then she may exclude gain on sale up to $250,000.

Exception Formulation: A taxpayer may exclude up to $250,000 gain on the sale or exchange of her property, except that such exclusion is not available unless the taxpayer has owned and used the property as a principal residence for 2 out of the last 5 years.51

While the exception framing and the rule framing are equivalent, exception-based framing is pervasive in the Code. Even I.R.C. § 61, arguably the foundational section of the federal income tax, begins with an exception:

(a) General Definition

     Except as otherwise provided in this subtitle, gross income means all income from whatever source derived, including (but not limited to) the following items:

[list of included items] . . . .52

Many of the exceptions to gross income are contained elsewhere in the Code and themselves contain exceptions. For example, I.R.C. § 101, which creates an exception from gross income for certain life insurance proceeds, states that it is subject to a number of exceptions.53 Likewise, I.R.C. § 102(a) states that gross income does not include property acquired by gift or bequest, but creates an exception for employee gifts, which are not excluded from gross income by § 102(a).54 I.R.C. § 103 excludes interest from state and local bonds from gross income, but creates an exception for certain private activity bonds, arbitrage bonds, and bonds not in registered form.55 This pervasive use of exceptions may suggest that the practice is inevitable. However, as we explain next, all of these exception-drafted statutes could, as a matter of logic, instead be articulated as more narrowly tailored rules.

2.     Narrow Rules with Defined Terms vs. Broad General Rules with Large Exceptions

An extension of the idea that any exception can also be formulated as a general rule is the idea that legal distinctions can be drafted using narrow rules with defined terms, or using broad general rules followed by large exceptions that may threaten to swallow the general rule. 

Exception-swallows-the-rule drafting occurs frequently in the Code: One example is I.R.C. § 163, which addresses the deductibility of interest. That statute states a broad general rule: “[t]here shall be allowed as a deduction all interest paid or accrued within the taxable year on indebtedness.”56 But, much farther down in the statute, it creates a large exception: § 163(h)(1) states that “personal interest” is not deductible, thereby excepting personal interest from the ambit of the general rule.57 Making matters more complicated, § 163(h)(2) then creates an exception to that exception: It defines “personal interest” to exclude certain items including, significantly, “qualified residence interest.”58 This is the basis for the home mortgage interest deduction. 

The entire disallowance of personal interest in §163(h) threatens to swallow the broad general rule in § 163(a) that all interest is deductible, and in fact substantially changes the principle stated in the general rule to the point where one wonders why the broad general rule was necessary. Moreover, the exception to the exception in § 163(h)(2) buries the intricate and important rules allowing home mortgage interest deductions deep within the statute.59

The statute could, alternatively, be written using a much narrower rule that is circumscribed using a defined term, while achieving the same substantive outcome. For example:

Narrow Rule: Approved Business Interest and Approved Personal Interest are deductible. 

Defined Term 1: Approved Business Interest includes ___.

Defined Term 2: Approved Personal Interest includes ___.

In short, rather than stating a broad general rule that all interest is deductible followed by large exceptions, the statute could have articulated a narrower rule stating that only certain “approved interest” is deductible, and then could have defined that term.

Defined terms are thus a way for a statute, particularly a complex statute, to state a narrow rule without having to introduce an explicit exception to do the narrowing. Defined terms are especially important in a technical area like tax law. They allow the drafter to articulate narrowly circumscribed rules even if there is no broader social meaning to the terms used (as exists in the case of “fornication”).60 For example, the term “controlled foreign corporation” (“CFC”) or “applicable high yield discount obligation” are Code-made terms that would have little meaning to the outside world aside from the definitions in § 957 and § 163, respectively.61 But having been defined into existence, they allow the Code to draw distinctions (for example, between CFCs and non CFCs) without introducing an explicit exception.

3.     Free-Standing Provisions v. Intra-Section Drafting

Statutes also vary in terms of where exceptions and defined terms are placed in relation to the main rules to which they relate, and in terms of how explicitly or implicitly they are cross-referenced. The Code sometimes places exceptions and defined terms in the same Code section as the main rule, but sometimes places them in different Code sections, and may do so with or without an explicit cross reference.

i.     Placement of Exceptions

The Code’s varying approaches to placement of exceptions can be seen in the rules governing contributions of property to corporations and partnerships upon entity formation. These rules generally provide “[n]onrecognition” (i.e., tax free) treatment for such contributions.62 There are numerous exceptions to nonrecognition—that is, situations in which tax will in fact be imposed. It turns out that drafters have taken contrasting approaches to placement and cross-referencing of these exceptions in the corporate and partnership contexts.

In the case of corporate contributions, exceptions to the general nonrecognition rule are contained in the same Code section as the general rule.63 The general rule providing that transfers of property to a controlled corporation are not taxable is in I.R.C. § 351(a).64 The exception to nonrecognition is contained in I.R.C. § 351, right after the § 351(a) general rule.65 I.R.C. § 351(b) describes at length circumstances in which property transfers to the corporation will be taxed, including in very specific circumstances (such as the receipt of “nonqualified preferred stock”).66 I.R.C. § 351 is therefore an example of an exception explicitly written into the same Code section as the general rule.

In contrast, the partnership contribution rules take a different approach. There is also a general nonrecognition rule for transfers of property to partnerships, I.R.C. § 721.67 But that section is short.68 Its general rule provides that “[n]o gain or loss shall be recognized to a partnership or to any of its partners in the case of a contribution of property to the partnership in exchange for an interest in the partnership.”69 While I.R.C. § 721 offers a few within-section exceptions to this general rule, most of the significant exceptions to nonrecognition treatment (including what happens when the contributor receives cash or prohibited property in exchange for the property contributed) are left to other Code sections. Examples of such “free-standing” exceptions include the so-called “disguised sales” rules of I.R.C. § 707(a)(2)(B) and the rules regarding what happens when there is liability relief in excess of outside basis, which can yield gain recognition.70 

Not only are these free-standing exceptions to nonrecognition treatment not offered within the general rule of I.R.C. § 721, they also do not even refer to § 721, and § 721 does not refer to them. Instead, these other rules are presented in the Code as “independent rules.” In the case of liability relief in excess of outside basis (which produces a functional exception to § 721), the exception comes about through linkages between multiple, other rules, which themselves do not explicitly refer to each other.71

In short, the partnership exceptions to nonrecognition treatment are far less apparent upon a casual glance than the exceptions in the corporate contribution context.

ii.     Placement of Defined Terms

The same issue can arise for defined terms. Sometimes a Code section will offer its own definition for a term and may limit the definition to only that Code section. An example of such a section is I.R.C. § 108, which provides a tax exclusion for the income resulting from forgiveness of a taxpayer’s debt under certain conditions. I.R.C. § 108(d)(1) provides the definition of indebtedness of the taxpayer and limits its application to only § 108:

For purposes of this section, the term “indebtedness of the taxpayer” means any indebtedness—

(A) for which the taxpayer is liable, or

(B) subject to which the taxpayer holds property.72

Other Code provisions instead offer cross-referenced definitions, referring the taxpayer to another section of the Code. This occurs, for instance, in the corporate tax rules, which explicitly cross reference the definition of “control” of the corporation from one section (I.R.C. § 368(c)) to define control elsewhere, for example, in I.R.C. § 351, which addresses transfers to corporations controlled by the transferors.73 Among such Code provisions offering definitions by cross reference, some might borrow the definition from elsewhere unchanged, but others may modify or impose substituted language when employing the cross-referenced definition. For example, I.R.C. § 304 (pertaining to redemptions through related corporations) employs a cross reference to the attribution rules of I.R.C. § 318 but substitutes “5%” for the 50% thresholds appearing in I.R.C. § 318.74 I.R.C. § 199A cross-references I.R.C. § 1202(e)(3)(A) in defining the term “specified service trade or business” but notes that that section should be “applied without regard to the words ‘engineering, architecture.’”75

Still other Code provisions use terms that are defined in a “catch-all” Code section that contains defined terms—I.R.C. § 7701. Section 7701 states:

When used in this title, where not otherwise distinctly expressed or manifestly incompatible with the intent thereof—

[list of defined terms].76

“Title” refers to all of Title 26, the entire Internal Revenue Code. Thus, a drafter may decide to insert a new defined term into I.R.C. § 7701 that applies broadly to all of Title 26 unless “otherwise distinctly expressed or manifestly incompatible with the intent.”77 A drafter may also rely on the existing I.R.C. § 7701 definitions in drafting new statutes.

Note that some provisions of the Code use terms that could be defined terms, but they are either not defined at all, or a stated definition elsewhere in the Code does not apply to a given use of the term. An example of a term that is never defined is the term “trade or business,” which occurs in several places throughout the Code without a definition.78 An example of a term that has a stated definition that is not applicable to all uses of the term in the Code is the I.R.C. § 61 definition of gross income. I.R.C. § 61 lists “[i]ncome from discharge of indebtedness” as one of the items that is explicitly included in income.79 While “income from discharge of indebtedness” also occurs in and is defined in I.R.C. § 108(d)(1), the § 108 definition expressly only applies “[f]or purposes of this section [i.e., section 108]” itself.80 And § 61 provides no alternative definition, either in § 61 itself or by cross reference. Therefore, the term as used in § 61 remains undefined.

4.     Drafting Features in Combination

Finally, it is important to note that often all of these features interact and occur in combination. Take for example I.R.C. § 82, pertaining to reimbursement of moving expenses, which states:

Except as provided in section 132(a)(6) [pertaining to “qualified moving expense reimbursement[s]”], there shall be included in gross income (as compensation for services) any amount received or accrued, directly or indirectly, by an individual as a payment for or reimbursement of expenses of moving from one residence to another residence which is attributable to employment or self-employment.81

This is an example of an explicit exception from gross income inclusion that is created by cross-reference to a defined term. The defined term contained in § 132(a)(6) (“qualified moving expense reimbursement”) is itself part of another substantive Code exception to gross income: If the reimbursement is a “qualified moving expense reimbursement” then it is excluded from gross income as a § 132 fringe benefit.82 Here, the choices we have discussed work in combination, and certain of such choices made in combination may make the statute particularly difficult to parse.

C.     Do Drafting Choices Matter?

Understanding the array of drafting choices that confront the statutory drafter begs the question: Do these choices matter? Should we care whether drafters write broad general rules paired with exceptions as opposed to narrow rules using defined terms? Should we care where they place such exceptions or defined terms in the Code, or whether they use explicit as opposed to implicit cross references? Or, are such drafting choices irrelevant as long as the statutes accurately convey the intended law?

One possible answer to this question is that these choices do not matter because the substantive law is the same and will have the same impact regardless of how the statute is formulated. However, it is also possible that even if the substantive content of the statute is unchanged, drafting choices may have some important effects. The potential importance of drafting choices can be illustrated via metaphorical comparison to the decision regarding placement of bathrooms in a building. Clearly, the number of bathrooms in a building (a decision about content) is important—the number of bathrooms in a concert hall, for example, will affect how many people can go to the bathroom and how long they have to wait. But the decisions about where and how to place and design bathrooms are also important. Whether bathrooms are placed next to a ramp on the first floor or at the top of a steep set of stairs tucked in a back corner, whether bathrooms are well lit, and whether they have appropriate signage, can all affect how people actually find, use, and experience the bathrooms. Bathroom placement also affects perceptions of whether the concert hall is handicapped accessible and is generally an appealing place to see a concert. Likewise, how tax statutes are formulated can affect accessibility and how various parties (including lawyers, the government, courts, and policy makers) interact with the law, thereby influencing the law’s impact.

We now outline some potential impacts of drafting choices. This discussion does not necessarily prove that drafting choices do matter or quantify the extent to which any particular drafting decision matters. However, examining the ways drafting choices may matter does help sharpen our inquiry into what counsels are thinking about when they draft. Specifically, if counsels are operating under the hypothesis that drafting choices are irrelevant, it would be important to understand why they think this, in light of the possible effects that we outline below.83 Alternatively, if counsels do consider drafting choices to be significant, it would be important to understand which consequences they regard as most important, which ones they had not considered, and why.

1.     Effects on Taxpayers, Representatives, and Government

Certain drafting choices can make the statute difficult to interpret accurately. For example, a spiral of exceptions to exceptions to exceptions can be quite difficult to read. Placement of an exception to a general rule in a distant Code provision may cause readers to miss the exception and may invite interpretive errors.84 In contrast, narrowly tailored general rules or a general rule followed by exceptions within the same section may be more intuitive for the reader. 

While perhaps such differences may matter less for sophisticated tax experts who are deeply familiar with the Code,85 certain choices make the law significantly more costly to some users. For example, less sophisticated users of the Code (including new lawyers and lawyers in smaller generalist law practices that provide some tax services) may find convoluted statutory language or hidden exceptions difficult to comprehend, interpret, or detect.86 This can cause errors, create barriers to entry, and make tax advice more costly for taxpayers. Anecdotal evidence from the tax bar suggests that it takes longer for new tax lawyers to get up to speed on tax law, due in part to the increasing length and complexity of the tax law. This can result in a small pool of experts.87 And one can expect costs to taxpayers to be high if the pool of available experts is small.

A system in which taxpayers must rely on experts to read the Code may also create additional costs for the government, because the government must rely on lawyers and advisors to help taxpayers navigate the tax law. Additional costs may arise, for example, if lawyers and tax advisors take aggressive pro-taxpayer interpretations, which may decrease revenues collected and shift the burden of detecting these positions to the government.88 Such costs may arise if companies that write tax software misunderstand, misinterpret, or mis-implement the law. Such errors in software algorithms and programming may be difficult for tax policymakers to detect and correct.

Moreover, as the Code gets harder to read, government agencies may have to play a bigger role in translating the Code for taxpayers.89 Depending on how a statute is worded, the gulf between the statutory language and IRS guidance language (such as FAQs, Forms, and Instructions on the website) may be small or large. The larger the gulf, the more work the IRS will have to do, and the more discretion it will have to interpret, translate, or simplify the law.90

2.     Effects on Judicial Interpretations

How a statute is formulated may have consequences for judicial interpretation. When courts engage in statutory interpretation, they are, of course, engaging in interpretation of the statutory text, even if this is only an entry point into a more holistic or purposive interpretation.91 The fact that courts focus on the text on the page and apply various interpretive canons in construing statutory text means that the formulation choice for identical substantive legal content can matter in judicial outcomes, even in cases where the statutory content is the same as a matter of logic.

For instance, at least some courts discount examples as incapable of offering dispositive law, even though examples may in effect offer the same substantive content that can be formulated as a rule.92 Thus, the choice to formulate substantive content in the form of an example, rather than a seemingly identical rule, may impact a court’s treatment of the same legal content. 

To take another example of the impact of drafting choices, the interpretive canon expressio unius provides that inclusion of one thing implies the exclusion of another.93 So, for instance, a list of “any tree, bush, or shrub” may be interpreted to exclude potted plants. This canon could mean that using a general rule followed by a list of exceptions may imply an exclusion of other exceptions, in a way that other formulations may not. 

3.     Consequences for Tax Law’s Development

The way the statute is formulated may also create intended or unintended presumptions and effects as law develops. For example, a statute drafted as a broad general rule followed by exceptions creates a presumption that items not explicitly excepted are included in the general rule. This may yield more widespread inclusion of items in the general rule over time than a “narrower rule” formulation, for example if new situations and categories arise over time that the listed exceptions did not contemplate.

Take, for instance, the definition of “capital asset” in the Code. The Code provision that defines a capital asset provides that “the term ‘capital asset’ means property held by the taxpayer (whether or not connected with his trade or business), but does not include [a series of listed exceptions].”94 This definition of a capital asset as “all property” followed by delineated exceptions means that, as new types of assets arise that the original drafters may not have contemplated, such assets are presumptively capital assets (because they do not fall within the delineated exceptions). In contrast, the new type of asset would not presumptively qualify as a capital asset if the rule were drafted as a narrower general rule that affirmatively provided the specific assets that qualify as capital assets.

On the flip side, it is also possible that a statute drafted as a general rule followed by a list of exceptions—or really any statute that delineates lists of items that fall within a certain category—may make it easier from a political process perspective for additional exceptions to be added to the list over time.95 It is possible that tax policymakers and drafters may find it easier to insert new exceptions than to unwind a narrowly tailored specific rule.

4.     Framing Effects on Policy Debates

The way the tax law is articulated may also exert a framing influence on tax policy myths and debates. As Frederick Schauer has noted, the choice to formulate something as an exception may be purposefully done in order to characterize one position as socially dominant and conversely to cast the exception as the “outsider.”96 This insight is applicable in the context of tax statutes: For example, a broadly articulated rule that “all income is taxable” reads like an income tax and creates the impression that income is the current and normatively desirable tax base, even if it is then riddled with exceptions. Such a rule can obscure the consumption-like features of existing law and may affect debates over the choice of tax base.97

In addition, an ugly, unwieldy, or verbose Code may provide fodder for those who might complain about the Code’s complexity and who might suggest that the Code is “broken.”98 Complexity, of course, has multiple dimensions, and complexity of the drafted word is only one dimension.99 However, at least some charges of complexity stem from how the rules are formulated and the length of the Code, both of which result in part from drafting choices. 

In short, substantive content aside, these are a variety of ways in which the way the content is formulated may potentially matter. Again, the above discussion is not meant to definitively prove that formulation choices matter or why they matter. Rather, the point is to identify various potential effects, and to examine the extent to which government counsels consider these effects (and others) in creating and drafting tax legislation.

III.     Interviews with Government Counsels

We turn in this Part to the findings of interviews we conducted with current and former government counsels who participated in the tax legislation drafting process over the past four decades. Our goal was to understand (1) how those involved in the creation of tax legislation actually made drafting choices in formulating statutory language, and (2) to what extent they viewed these choices as significant. We outline our methodology and then report our findings. 

A.     Methodology

In identifying the pool of interviewees, we started with current and former counsels in the tax field we were able to identify and then employed a purposeful sampling technique called “snowball sampling” or “referral sampling.”100 That is, we first identified interview subjects who obviously met the criteria for our study (e.g., former counsels who had been involved in the tax legislation drafting process) and interviewed them. We located these individuals via internet and other research, which yielded names of counsels who had occupied key positions in important tax writing committees and other offices over time. We then requested referrals from the subjects we interviewed to identify additional subjects.

Because the interviewee population is former or current government employees and officials, and the activity being studied (i.e., legislative drafting) was not embarrassing or ethically questionable, we considered it acceptable to ask for suggestions regarding additional potential subjects directly (i.e., by asking for names). In fact, the subjects we interviewed would often spontaneously volunteer names of other subjects they thought we should speak with without us having to ask. A minority of interview subjects chose to contact other potential subjects to request their consent before passing their contact information on to us. In a few cases, subjects expressed strong opinions regarding other potential interviewees who we should absolutely talk to about drafting. In many cases, we were able to locate and contact additional subjects using publicly available email addresses. In some cases, we asked those we interviewed to provide us contact information of the individuals they had referred to us. There were some suggested subjects for whom we were unable to locate contact information.

1.     Interview Subjects

We sent solicitation emails to 45 current and former counsels and staff who have had involvement in the tax legislation drafting process. We received 31 responses, with five declining to be interviewed.

The counsels and staff we spoke to included individuals who had worked on the Joint Committee on Taxation, the Senate Finance Committee, the House Ways and Means Committee, the Department of Treasury, the IRS, Senate Legislative Counsel, and counsels to individual Congress Members.101 Interview subjects were almost evenly split between those who had been involved with the legislative drafting process during or before the enactment of the 1986 Code (the last major tax reform before the 2017 changes), and those who only became involved in subsequent years. The length of involvement among interview subjects varied widely. Because of the relatively small pool of individuals interviewed, we are constrained in our ability to provide specific numerical breakdowns by committees, years of service, political affiliation, or other criteria due to confidentiality considerations.

2.     Size of Pool and Representativeness

The pool of tax counsels we spoke to is small, but the range of individuals we spoke to was broad enough to provide a well-textured sense of drafting choices and practices.102 Our goal was to understand how drafting choices were made across time to produce the Code. In light of our research question, we attempted to speak with counsels who had served over different periods of time and in various capacities. By using qualitative interviews, we were able to have extensive discussions with our interviewees about various aspects of drafting.

3.     Interview Approach

We conducted semi-structured phone interviews with those individuals who responded to our email and expressed willingness to be interviewed. This meant that we approached the interviews with a prepared list of questions, but we added additional questions and also skipped questions as appropriate. Our initial list of questions is attached in the Appendix. Interviews lasted between approximately 40 minutes to an hour and a half.

The semi-structured approach allowed us to tailor our interview questions to take into account that individual’s experience and familiarity with the tax legislation drafting process. For example, if an interview subject mentioned early in an interview that the individual had high-level managerial experience with drafting, rather than day-to-day experience with particular provisions, we might focus on questions about the general drafting process and responsibilities.

Interviewees sometimes spoke at length about various aspects of the legislative drafting process. In such instances, we allowed the conversation to flow naturally, rather than forcing the conversation back to the interview script. Because of this semi-structured approach, we were able to obtain insights into the drafting process that were unexpected, did not directly reflect the questions we asked, or illuminated aspects of drafting that we had not previously considered.

4.     Limitations

Our research methodology yielded a wealth of information regarding how counsels approached the process of creating tax legislation. However, our methodology was also subject to important limitations. First, the snowball or referral sampling approach may yield a sample in which subjects refer the researchers to others who share their opinions or whom they know well.103 This means there is no guarantee of representativeness, and the sample may be biased. In addition, we had no way of preventing our interview subjects from discussing the interviews with each other. If this happened, it had the potential to affect the answers we received. On the flip side, the pool of tax counsels involved with the legislation drafting process is relatively small and largely known to each other, so snowball sampling was an effective way to reach this relatively contained target population.

Second, because the number of interview subjects was small, we are constrained in how we report our results due to research ethics and the interests of maximizing confidentiality. For example, we do not identify the years during which each interviewee was involved with legislative drafting or the committees they served on because doing so risks identifying the research subjects. Similarly, we do not report quotations that run a significant risk of identifying the interviewee.

Third, because many current counsels and staffers were busy working on tax reform in the fall of 2017, this may have decreased the response rate to our solicitation email. It may also mean that we have a disproportionate number of declines or non-responses from Republican counsels and other counsels who were working on the 2017 legislative reform proposals.

Fourth, House Legislative Counsel declined our interview requests, citing confidentiality restrictions. Thus, we were not able to speak with Legislative Counsel in the House Office of Legislative Counsel.

Fifth, in many cases there was some time lag between the interviewee’s participation in the legislative process and the time of the interview. This was particularly so for counsels active in or prior to the 1980s. This means that interview responses may be limited by human memory. It is possible that the passage of time may influence how certain interviewees recall and describe their experiences and approaches.

 We must note a final methodological point. In our findings, we report interviewee responses regarding the significance (or lack thereof) of certain drafting choices. However, these responses cannot definitively prove or disprove the impacts of drafting choices on taxpayer behavior, tax policy debates, the law’s trajectory, or judicial decision-making. A different type of research would be required to prove the actual effects of drafting choices on various actors and constituencies.

B.     Findings

Our interviews yielded significant information regarding how counsels and others involved in the tax legislative process made choices regarding drafting, and how they regarded the significance of such choices. The interviews also shed light on other aspects of tax drafting, such as how statutory language accretes over time, who controls the drafting process, and how that process has changed.

1.     Audience

i.     Majority View: Drafting Choices Do Not Matter; Ordinary Taxpayers Are Not the Audience

When asked whether drafting choices matter, a majority of interviewees responded that they did not pay much attention to drafting choices and that they did not really matter.104 This may be our most surprising finding. While we set out to understand what motivated drafters to make certain choices (such as the use of a general rule followed by an exception, rather than a more narrowly tailored general rule), it was somewhat difficult to get our interviewees to discuss what might motivate such a choice. By and large, they simply had not paid much attention to these types of drafting choices and were not particularly interested in talking about them (though they did explain that choices tended to be motivated by a desire to preserve existing language and follow existing patterns in the Code, discussed in more detail in Section III.B.2.i below). Putting aside any path-dependent reasons for drafting choices, our interviewees generally had few thoughts on what might motivate one choice over another, or the implications of such choices.

Part of what seemed to underpin this view was drafters’ perceived audience for tax statutes. One of the questions we specifically asked interviewees was who the intended audience was when drafting tax statutes. We received a variety of responses to this question. A view commonly expressed was that tax statutes are not drafted for the ordinary taxpayer; therefore, it did not matter whether statutes were convoluted because ordinary taxpayers would not be reading the statute. In fact, almost no interviewees indicated that they were drafting for taxpayers themselves (other than very sophisticated taxpayers and their tax counsel). Taxpayers, we were told, would not be reading the tax law from tax statutes. Some interviewees specifically indicated that it did not matter if the statute itself was complex because ordinary taxpayers who used TurboTax or an accountant would not experience that statutory language complexity.105

Many interviewees emphasized the fact that the underlying goal in drafting tax legislation was to effectuate the intent of the Members of Congress. As discussed below, however, the overwhelming view among interviewees was that Members did not pay much attention to reading the statutory language. Having said that, however, interviewees had a variety of other constituencies in mind when drafting statutes.

Some interviewees said they were drafting with the Treasury Department in mind, crafting statutes so that it would be possible for Treasury to effectively write regulations implementing the statute. Some thought that the IRS was an important audience—they were drafting statutes in a way that allowed the IRS to administer the law, write forms and instructions, etc. Others opined that sophisticated tax counsel, who had expertise in reading and interpreting the rules and advising clients, is the primary audience. A couple of interviewees indicated that the primary goal was to articulate a statute so that software companies such as TurboTax could effectively write a program that implemented the rule that a large number of taxpayers could use. One interviewee suggested that an important audience in the drafting process was their peer group—similarly positioned tax professionals, including both other drafters and tax experts in private industry. Such individuals would be evaluating a particular drafter’s expertise, acumen, and, perhaps, future employability, so one wanted to appear smart and competent to these tax professionals. Yet another stated that drafting was “inward focused,” staffers “are writing for themselves” and that “[t]he audience is that little group of people who actually read the stuff and then explain it to the Members.”106

But, as noted, the answer that we almost did not hear at all was that those involved in the drafting process were drafting for taxpayers themselves. Indeed, a few interviewees felt very strongly that the tax law was not meant to be understood by the average taxpayer and that charges that the face of the statute was too complex or convoluted were unjustified for that reason.

Likewise, although interviewees occasionally mentioned that they had courts in mind during the drafting process, this was not a dominant focus.107 To the extent judicial interpretation was mentioned, the focus was on how prior judicial interpretations served as an incentive for drafters to retain existing statutory language because taxpayers had relied on court rulings interpreting it. On the other hand, potential future interpretations of statutory language and judicial doctrines appeared not to be at the forefront of interviewees’ minds. One interviewee stated, in response to our questions about audience:

Not courts. Forget the courts. Forget the courts don’t matter. Remember maybe ninety percent of tax law, the courts don’t matter. The IRS doesn’t matter. What matters is millions of people are relying on what’s in that Code. Like as it has been interpreted by millions of interpreters, whether they were accountants or whatever, and a usage evolves. If you disturbed that, you are disturbing a whole lot of things that may never get to court. It’s just tax, the administration and the operation of the tax system is absolutely an iceberg. We spent all of our time worrying about the tip that is above the water. So in order of priority, the most important parts of the tax system once a bill is a bill, it starts with the IRS. This is what they put in their publications. Single most important thing and number one is what is in the form. Number two is what is in the publication that is explaining the form. At that point, the public is relying on that. Everything after that in terms of what affects economic behavior is vastly less important. It’s the most important thing in a tax system from administration of course is how do people file their return? The penalties only affect a few people. Court cases affect even fewer people. Everything else is what are you doing that affects what goes on the return . . . . I am never worried about what the court will do as a practice. I am overstating it, but I just don’t think that’s important.108

In other words, what was important was how taxpayers interfaced with the tax system through the forms and publications put out by the IRS and through other interpreters. It did not really matter how courts would interpret the tax law because challenges rarely reach the courts. Another interviewee noted that he felt that, in retrospect, he should have paid more attention to potential judicial interpretation than he actually did.109

Notably, despite being able to point to various audiences for tax statutes other than taxpayers (e.g., Treasury, tax experts, TurboTax, IRS), most interviewees did not seem to think that drafting choices mattered a great deal for these constituencies. For example, most interviewees seemed to assume that experts would be able to understand, analyze, and advise taxpayers accurately despite drafting choices in the Code that might make interpreting the law non-obvious (such as free-standing exceptions or numerous cross references). Some of the interviewees who indicated that Treasury and IRS were a primary audience for statutory drafters did indicate that tax legislation containing explicit statements of statutory purpose or explicit delegations of authority would be helpful for regulation writers. In general, however, interviewees did not consider the types of drafting choices we mentioned during our interviews to be very important for the target audiences they identified. As discussed further below, several interviewees attributed drafting choices—for example, the choice to use exceptions paired with general rules or to use defined terms—to the particular unique “style” of drafting that drafters in the House and Senate Offices of Legislative Counsel used.110

ii.     Minority View: Drafting Choices Matter

A minority of interviewees indicated that drafting choices do matter. One suggested that formulation decisions might matter for purposes of how the legislation is scored. That interviewee indicated that perhaps articulating a broadly inclusive general rule followed by an exception might enable the legislation to get a better score from the Joint Committee on Taxation.111 This was a political process effect that we had not thought about.

Other interviewees noted that drafting choices may matter for reasons that we had discussed in Section II.C. For instance, one interviewee suggested that the way that the law was formulated had significant impacts on the public’s ability to read and interact with the tax law. The interviewee suggested that there are potential problems with Legislative Counsel’s formulation decisions being left somewhat unchecked and unmonitored. Another interviewee thought that placing exceptions in a free-standing Code provision separate from the generally applicable rule was an obvious error. That interviewee also opined that the way a statutory rule is articulated might affect how a court might interpret that language. Another raised the possibility that poor drafting might give credence to arguments that the Code is too long and complex and that the Code is broken and needs to be overhauled and simplified, noting:

[A]nything that makes people think [the Code] is crazy [matters and] . . . every time you [do] something screwy in the Code, and people can show, ‘Look how convoluted this is. This . . . over there really is associated with this thing over here,’ yes, it matters. It’s just another nail in the coffin of, ‘We have to do reform, but things are crazy, there’s too much of it, and we need to simplify it.’ Yeah, it does matter—it matters less than the policy that is implemented matter. But is it yet another [factor].112

Another interviewee thought that drafting style might matter for consistency and effects on judicial interpretation. Yet another noted that starting with a general rule followed by exceptions provided clarity and discipline to the Code. But these perspectives were in the minority among the interviewees we surveyed.

2.     Change Over Time; Time Pressures

Interviewees also pointed to several dynamics of legislative drafting as responsible for the way that the tax law is actually formulated. Two dynamics often mentioned were (1) the accretion of statutory language over time, and (2) limitations drafters faced in how much they can change existing statutory language. In addition, several interviewees noted that legislation drafters face significant time pressure while drafting tax legislation that could affect the statutory language.

i.     Accretion Over Time and Stickiness

When asked about why tax statutes are sometimes drafted such that exceptions swallow the general rule, or why the Code contains seemingly independent exceptions contained in free-standing Code provisions with no obvious cross reference, many interviewees suggested that the way the tax law appears now is an artifact of how the law came to be over time. The original drafters of the law may have crafted a general rule while not appreciating the complexity or full extent of the problem. Alternatively, subsequent legislators may have wanted to change the statutory rule. Either situation might necessitate amendments in the future. Any such amendments would tend to be addressed with additional provisions that are layered onto the original statute, rather than a wholesale rewrite of the original rule.

Interviewees specifically noted that drafters tend not to reformulate the original rule due to an interest in making the minimum number of changes necessary. The result would be a complicated and difficult to read statutory edifice. Interviewees gave various reasons for why drafters tend to make the minimum possible changes to an old provision. One reason was that the existing Code provision would have created various, settled expectations, reliance, and interpretations. For example, court cases would already have interpreted existing statutory language; regulations may have been written interpreting or implementing a statutory provision; tax lawyers would have invested hours in learning, interpreting, and researching statutory language; and new Code sections may have sprung up that cross-referenced other Code sections or relied on definitions elsewhere in the Code.113One interviewee suggested that if a certain Code section uses a defined term, that Code section is much more likely to be used later on by other sections than a Code section that did not contain a defined term. Drafters would make small tweaks to the statutory language, rather than engaging in wholesale statutory overhaul or reformulation, to try to avoid unsettling these existing interpretations, cross-references, accumulated bar knowledge, and regulatory interpretations.

Relatedly, other interviewees suggested that the nature of the tax legislative process means that agreement can often be reached to make small tweaks, but that a broader reformulation (even if the purpose was just clarification) would require more extensive legislative buy-in, which is costly and more difficult to obtain. One of the questions we asked interviewees was whether it would be possible to simply clean up parts of the Code that had accreted over time (without making substantive changes). In response, interviewees routinely suggested that cleaning up the Code just to make it clearer was difficult or impossible because of underlying political dynamics, and that there was little appetite for or constituency that prioritizes such a cleanup. One stated:

It’s not the end of the world [to have a convoluted Code]. It’s inefficient and so on and so forth. But it doesn’t really rise to a level that you’re going to go fight a war about. And it takes enormous amount of work to get that done . . . . But anyway, nobody cared. Nobody pushed it because Members don’t care because the affected tax writers don’t care. Go out and find some taxpayer who actually cares about . . . rules and how they should be drafted.114 

In this way, many interviewees suggested the original drafting decisions could prove to be sticky. If a broad but poorly tailored general rule had previously been written, for instance, that rule might have staying power due to stickiness, even if numerous exceptions had to be added later. The process that yields this layer-cake statutory structure can be described as statutory inertia, whereby not only the content of the law,115 but also its formulation, is difficult to change. As one interviewee explained, “[t]here’s a lot of inertia . . . in the tax world and . . . in Congress in general. . . . [I]t takes a lot to overcome that. And it usually ends up not happening.”116

ii.     Low Likelihood of Technical Corrections

In response to our questions about the possibility of cleaning up the Code, several interviewees alluded to technical corrections legislation.117 Several interviewees pointed to the difficulty in getting even technical corrections legislation (which fixes drafting errors) passed, let alone changes merely geared towards readability and clarity. A concern repeated by several interviewees was that Members would hijack a technical corrections bill to do much more than make technical corrections. One interviewee noted:

If you introduce a technical corrections bill in the Senate what you have done is you have given every member of the Senate an opportunity to put on his favorite piece of crap. And so it never happens. Technical corrections bills generally ride on the heels of some other piece of tax legislation because nobody – in the House you can control the process, but you can’t control the process in the Senate.118

Some interviewees noted that certain Members would have a very hard time being convinced that only technical corrections were being made and might block legislation for that reason. Interviewees noted that Members might even oppose corrective follow-up legislation that is clearly only technical in nature if they did not like the original legislation. Leaving flawed legislation as is, it was suggested, might make future reform more likely.

One interviewee did suggest that it is possible to get technical corrections done for legislation that the same Congress had just passed because the Members might want to avoid looking bad for what they, themselves, had passed. But, overall, interviewees agreed that it is very difficult to use technical corrections to clean up the law that a different Congress passed.119

iii.     Limited Time and Resources

Another feature of the legislative drafting process that several interviewees mentioned was time pressure and limited resources. These were cited as a reason why tax legislation may have been drafted in a less than comprehensible fashion in the first instance. Some interviewees suggested that ideally drafters would identify how a particular provision would affect and interface with other Code provisions. However, interviewees pointed out that drafters were often operating under immense time pressure. As a result, drafters may not do as good of a job thinking about the interrelationships between provisions as might otherwise be desirable. 

Time pressure was also cited as a reason why there is little appetite for revising the Code to make it more readable. As one interviewee put it:

[W]e’ve got really scarce time and would you rather make it readable where the only person who would be reading . . . is us. Everybody else is going to like look it up and read it online. Would you rather get us to substantively improve the law some way? I think what’s hard to comprehend is how much time it would take to get past the technical correction that literally just re-wrote the section to have the absolute identical meaning but be more readable. I would not be surprised if that took hundreds of hours to get that passed.120

In short, scarce time meant that cleaning up the Code takes a back seat to substantive policy activities and considerations.

3.     Who Controls Drafting

i.     Legislative Counsel Control over Language

All interviewees indicated that the staff at the House and Senate Offices of Legislative Counsel (“Legislative Counsel”) was primarily responsible for making drafting choices.121 This was not surprising given what is previously known in the academic literature about legislative drafting.122 As discussed below, however, a couple of interviewees suggested that drafting became less disciplined and more collaborative when dealing with major legislation, and we also detected a shift towards more amorphous control over drafting and language over time.123

The two Legislative Counsel offices hold primary drafting responsibility for the tax law, as they do with other areas of the law. The stated roles of House and Senate Legislative Counsel, respectively, are “to work with committees and Members to understand their policy preferences in order to implement those preferences through clear, concise, and legally effective legislative language”124 and “turn . . . request[s] into clear, concise, and legally effective legislative language.”125

Consistent with these stated roles, our interviewees routinely reported that staff of the Legislative Counsel offices took the policies that Congress wished to enact and turned them into legislative language. This involved choices as to how to articulate particular rules (whether as a general rule followed by exceptions, or as a more narrowly defined rule, or whether to use a cross-referenced definition or a definition within a section, or some other option). Our interviewees also described how Legislative Counsel more generally had jurisdiction to consider—and did consider—the interaction between a particular, proposed change to the tax law and existing portions of the tax law, and how, if at all, the interface between the various provisions should be formulated in the Code.  

In making these choices, several interviewees noted that Legislative Counsel consults formal guides to drafting that exist for both House and Senate Legislative Counsel. These guides are the product of Legislative Counsel efforts over time. For example, the 1995 Edition of the House Legislative Counsel’s Manual on Drafting Style was originally prepared by Ward Hussey in 1989.126 Other interviewees suggested (without referring to formal style guides) that there is just a certain style common to the tax law, which has perpetuated over time, and that part of what Legislative Counsel did was conform new tax law to this style. When pressed, some interviewees appeared to have a hard time articulating the components of that drafting style. But a few interviewees indicated that that style includes, generally, a general rule, followed by exceptions, with defined terms later in the statutory provision, and effective dates at the end. Finally, several interviewees suggested that certain choices by Legislative Counsel merely reflected the style of particular drafters, and that styles differ between different individuals at different times.

ii.     Deference to Legislative Counsel on Drafting Matters

In light of Legislative Counsel’s clear, official role as the drafting arm of Congress, it was not surprising that we received the resounding response that Legislative Counsel makes drafting decisions. What was surprising was the Legislative Counsel’s extent of control over drafting decisions, and conversely the general lack of participation of other counsels and participants in making, questioning, or revising drafting and language choices made in this process. The participation of each of these other participants is described below.


Beginning first with Members, our interviewees repeatedly reported that Congress Members themselves did not generally look at or consider statutory language.127 Rather, Members generally only dealt with the substantive ideas underlying the legislation, leaving it to Legislative Counsel and other professional tax staff (discussed more below) to turn those ideas into language. They would only pay attention to specific language if some constituency or interested party complained about it. This was the case even after the shift in the House Ways and Means Committee to markup of actual statutory language, described below.128 As one interviewee put it:

A Member would never read the legislative language. They would probably never read the conceptual mark. They’d read the reading and they’d be briefed on it. I guess it really depends on the Member . . . but yeah it was the responsibility of staff to make sure that it reflected what the Member wanted and not for the Member to make sure that it reflected what the Member wanted.129

Thus, Members would generally rely on their own professional staffs to ensure that what the Members believed was going to be enacted into law was enacted into law. The general consensus was that, as long as there was no question that the Members’ substantive intent was enacted into law, the Members would not quarrel regarding the particular way a tax statute was drafted. One interviewee noted that even the most intellectually inclined Member was very unlikely to focus on legislative language. Others suggested that it would be a waste of Members’ time to focus on legislative language, or even that they lacked the ability to do so in a meaningful way. 

In addition, one interviewee alluded to the fact that staff will sometimes deliberately keep the language from the Members by design. This is done to prevent Members from leaking bill text to lobbyists and outside interests, which would likely doom proposed legislation. This description paints a picture of a Congress not only not involved in the creation of the actual statutory text, but a Congress that is intentionally left out of the loop in order to improve the likelihood of success.

Other interviewees affirmed this dynamic between Members and outside interests and the benefits of keeping language from Members. One noted:

A Member of Congress doesn’t really pay much attention to the statutory language. But what they do is they receive feedback on the statutory language from the people who are pressing them. And so if there’s a draft out that the, you know, the people who are pushing the idea don’t like, they say you know what, if you conclude this, this really kills the whole idea of what we want to do. Then the Member of Congress who’s going to get into the act and say you know, let’s get rid of this thing, right? And that person typically will go to his or her top staffer and then which may be a committee staffer, personnel staffer. And then they will eventually try to translate and get the-get the revised draft out of that.130

This description suggests that feedback from outside interests may actually prompt reconsideration or removal of proposed language, and it gives credence to the idea that keeping language from the Members may reduce the likelihood of roadblocks other Members and outside interests may erect. Along similar lines, another interviewee suggested that the drafting process might be improved if Members and their staff were kept out of the process, because their presence and their focus on discrete provisions detracted from getting things right.131 That interviewee also suggested, however, that keeping Members and staff out of the process would likely be politically unacceptable.132

Other Government Counsels

Beyond Members of Congress, interviewees suggested that there was also usually a clear demarcation of control between Legislative Counsel and other tax professionals and counsels involved in the drafting process. These other tax counsels involved in the drafting process include staff of the Joint Committee on Taxation, the staff members that serve on the congressional committees (for instance, majority and minority counsel for House Ways and Means or Senate Finance and their respective staffs), certain executive officials (such as staffers and counsels from the Treasury Department’s Office of Tax Policy), and others. These other tax professionals, in particular the JCT staff, would heavily participate in discussions of how to turn the Members’ agreements into law. However, their input tended to focus on broader policy, design, and implementation decisions, rather than statutory language itself.133

JCT staff, for instance, would help think through what potential loopholes a new statutory provision might create and how best to address the loopholes. Or they would be responsible for thinking through how to take an idea the Members agreed to and make it work in a particular industry (all under the watch of the Members’ and committees’ own staff, to ensure the decisions accorded with the broad-based agreements Members reached). But, as to how to actually craft these decisions into statutory language, JCT staff, Treasury staff, and others in the room generally took a back seat. These formulation decisions (for instance, whether to use a new defined term or not or to what extent a new rule should be integrated into an old rule) were typically left to Legislative Counsel. When asked whether JCT or the other professionals in the room, outside of Legislative Counsel, could weigh in on these decisions, many interviewees suggested that any attempt to provide input into this formulation process would be viewed as nitpicking and a waste of time and resources. The other professional staff left these decisions to Legislative Counsel.

Drafting Major Legislation

Some interviewees suggested that this delineation of responsibility broke down somewhat when drafting and passing major legislation. Due to the technical difficulty, time pressure constraints, and complexity of major legislation, drafting major legislation was described as more “collaborative,” with Legislative Counsel drafters and JCT lawyers working together on major drafting projects.134 One interviewee noted:

[W]hen you were dealing, at the end of a [legislative] session, with a gigantic piece of legislation, [the usual] process basically fell apart. And what happened would be that several people from each of the representatives’ staffs would be assigned to draft, and they would draft and the Leg. Counsel [i.e., Legislative Counsel] would sort of give it a cursory look, and it would go on this pile of crap that was on the Senate floor or the House Floor, and that’s what got passed.135

This account indicates Legislative Counsel ceded some control over drafting in cases of passing major legislation under time constraints.

In a similar vein, another interviewee described how drafting occurred during the lengthy 1986 legislation markup process.136 Because it was not feasible to have drafting take place only after the markup, drafting occurred during the markup process, often in the middle of the night due to time pressures. Once the day’s markup was over, the person responsible would go and draft. If policy issues or roadblocks came up during drafting, the staff would come back to the Ways and Means Committee, JCT, and Treasury and raise questions and obtain guidance. In this way, drafting during the 1986 legislation markup was characterized as “a dynamic process where [participants] recognized that neither the markup process or the actual drafting process necessarily would reflect the final decisions.”137

Thus, despite the repeated story of Legislative Counsel control over drafting, some interviewee accounts provided a flavor of how the drafting process might differ in cases of time crunched, major legislation.

Ward Hussey

While almost all interviewees emphasized Legislative Counsel control over drafting and language, interviewees who served during the 1980s emphasized the role of one Legislative Counsel in particular: Ward Hussey. Hussey began serving as House Legislative Counsel in 1946 and served for 43 years, retiring in 1989.138 He rose to be head of House Legislative Counsel and served as Legislative Counsel for all tax legislation, with the assistance of his deputy, John Buckley.139

Interviewees who worked during Hussey’s tenure described him as an experienced and brilliant figure. We were told that Hussey was rumored at one point to have drafted 90 percent of the Code.140 When asked “What made Hussey so good?”, interviewees told us that he was a master of how the Code was written and understanding the interrelationships between different parts of the Code. Interviewees described both his deeply ingrained understanding of the Code’s patterns and how such patterns interrelated. He could tell how a change made in one part of the Code would reverberate in other parts of the Code. In the view of almost all of the interviewees who worked during his tenure, this made Hussey a real giant of tax law drafting. 

Interviewees who participated in the tax legislative drafting process during Hussey’s tenure also described how he exerted extensive control over the process. After a conceptual markup of tax legislation by Congress, the tax professional staff would assemble in a drafting room.141 The room contained a large drafting table (and, according to the recollection of some, no windows). Hussey would conduct a Socratic method questioning of the professional staff, asking them questions about how they thought Congress’s intent should be implemented in the form of law. In this process, the nonpartisan JCT staff was relied upon as the principal source of information for Hussey’s examination of the issues. He would push JCT experts on why they thought certain decisions should be made (for example, why a certain rule should be used to implement a particular policy decision by Congress, what the ramifications and problems would be, etc.). 

After the questioning and discussion was done, Hussey and Buckley would take principal control over turning the substance into legislative language. The other professional staff could raise problems with the proposed language, but any suggestions regarding formulation decisions (for example, decisions about whether to use a broad general rule plus exceptions or a narrower general rule with defined terms) were generally deemed to be the purview of Legislative Counsel. Likewise, several interviewees who worked with Hussey explained that no one would bring alternative legislative language to Hussey.142 Some interviewees referred to the possibility that Hussey might throw a drafting participant out of the room if he thought a participant had overstepped his or her bounds. One interviewee explained, “Basically what would happen is Ward would throw the person out of the room. Literally. Literally. Ward said, get out of here. We’re going to do it my way.”143

Several interviewees suggested that the idea that Hussey would accept proposed language from a lobbyist or even from a Member’s office, even as a first draft, was unimaginable. Interviewees speculated that Hussey would have just ripped such language up.144 Hussey was clearly the tax drafting expert and the person responsible for formulating the legislative language and thinking about to what extent such legislative language should interact with, or change, other aspects of the Code.

iii.     Shift to More Amorphous Control of Drafting

The story told by more recent participants in the tax legislation drafting process reflected more amorphous control over legislative language than during the 1980s. As noted, counsels and staff who participated in the legislative drafting process during Ward Hussey’s tenure—and who in particular were involved in the drafting of the 1986 Code—repeatedly mentioned the prominent role Hussey and Buckley played in the drafting and creation of the 1986 Code. And, as noted, they indicated that there was widespread deference to Hussey and Buckley in matters of statutory drafting, even if other participants were involved in questions of substantive tax policy design. In contrast, interviewees who participated in the process in later years pointed less frequently to the role and influence of one or two particular individuals.145

When speaking with more recent counsels, there was a sense that a greater number of individuals may have had a hand in the drafting of statutory language, even if Legislative Counsel still ultimately held the pen. Interviewees indicated that staffers in Members’ offices may work with Legislative Counsel earlier in the process to formulate draft statutory language, which might then be circulated either to other Members or through the Senate Finance Committee or House Ways and Means Committee. Interviewees who were recent legislative process participants also indicated that lobbyists sometimes also offer actual legislative language, which a particular Member may promote.146 Several interviewees gave the caveat that participants in the drafting process would not take too seriously any legislative language that lobbyists clearly offered. One interviewee did suggest that lobbyists might have expertise with particular industries that might be valuable to counsels in crafting legislative language and policy.147 

One interviewee suggested that, unlike in times past (and in contrast to Hussey’s tenure), the tax legislative process now lacks a clear locus of control.148 Various counsels and staff including, for instance, JCT staff, Senate Finance Committee counsel, and House Ways and Means counsel, jockey for control. Who has the most dominant personality in a given room at a given time plays an outsize role in how the legislation was drafted.

Despite the stated involvement of a greater number of individuals in the tax legislation drafting process, and despite the change in the House markup process described below, even the more recently active counsels agreed that Members themselves would almost never focus on the actual legislative language and may even be intentionally left in the dark regarding the proposed legislation until the last minute, in order to prevent the Members’ ability to leak the legislation.

iv.     Recycling of Legislative Language

Importantly, several interviewees also indicated that drafters might reuse old proposed legislation that had previously been drafted when ideas and policy reforms that were formerly proposed became re-proposed in later years. Thus, drafters often do not start from scratch when drafting legislation, but rather draw on language that had previously been drafted.

v.     Shift in House Ways and Means Committee Markup Process

One notable shift that more recent participants in the legislative drafting process described was the change in the markup process of the House Ways and Means Committee over time. This shift may relate to the turn to more amorphous control over legislative language discussed above. 

Historically, and certainly in the time of Ward Hussey, Member markup of tax legislation in both the Ways and Means and the Senate Finance committees was purely conceptual, meaning that Members only fleshed out, and agreed upon, the substantive and policy aspects of tax legislation in markup sessions, rather than actual legislative language. In the mid-1990s, under the chairmanship of Bill Archer, the House Ways and Means Committee started requiring actual legislative language for markup sessions.

As a result of this change, Legislative Counsel had to draft language prior to the markup session, with the idea that the Ways and Means Committee markup sessions would work off and be informed by the proposed legislative language. Interviewees mentioned that, even in the Senate, in which the markup session still remains—in theory—conceptual, Legislative Counsel formulates actual statutory language and the language exists prior to the markup session.149

The upshot of this shift is that the Ways and Means Committee now has legislative language in front of it during markup. According to some interviewees, the shift has made the process more tedious and longer, has resulted in staff making more legislative language decisions in advance of guidance from Members, and has created more work for staff, with the result that more drafting help is sought from and outsourced to lobbyists. Interviewees indicated that despite the shift in markup practices, Members themselves still do not focus on actual statutory language, relying instead on the staff. Some interviewees explained that it was unrealistic to expect Members themselves to understand or engage with language.

One interviewee suggested that the Members’ offices (though, again, almost never the Members themselves) have become engaged in fiddling with legislative language for a variety of purposes. For instance, a Member’s office may add certain language to proposed legislative language in the form of a heading, so as to emphasize how the legislation would help a certain constituency. The language would not necessarily have substantive effect, but could be used to make a messaging point. This might be the case, for example, with vanity bills drafted to make a point or suggest support for a certain platform, even if that bill has no chance of passage. Likewise, a Member’s office may repeat something that is already stated by cross-reference in a statutory provision to gain messaging benefits, even though the repetition (perhaps with slight alteration in specific words) may ultimately create confusion when courts have to interpret the provision.150 

vi.     Move Away from Executive Branch Input

Despite indications that more actors are now involved in the crafting of legislative language relative to the past, interviewees also suggested that there may be less executive branch input than previously. Interviewees observed that the Treasury Department and, in particular, the IRS used to have a more prominent seat at the drafting table. Some interviewees did note that Treasury’s involvement at a given time would depend on whether the same political party controlled Congress and the White House (thus leading to the possibility of more Treasury involvement when the parties were aligned). There was, nevertheless, a sense that, over time and all other things being equal, the Treasury and the IRS were less engaged in legislating.151 Representatives from the IRS Legislation and Regulations Division of the office of Chief Counsel used to routinely offer comments on draft language.152 This division no longer exists.153 Interviewees indicated that IRS participation in the process is more muted now. The IRS may participate in decisions that implicate tax administration directly, but the IRS does not necessarily have a seat at the drafting table, and is not necessarily routinely consulted about how a new provision will be administered, or how particular drafting decisions might affect tax administration.

Interviewees offered various theories for why this shift has occurred. One interviewee suggested that particular Treasury or IRS officials overstepping their bounds and weighing in on policy decisions might explain not including them. Others suggested that by nature the Treasury Department is going to have less influence in the statutory than the regulatory process, and that thus their time is best spent elsewhere. Another interviewee suggested that the advent of email consultation meant there was less need for the IRS to be physically in the room.

vii.     Drafting Choices Intertwined with Policy Decisions

Finally, it is important to emphasize that although the focus of our study was how drafting choices are made and the implications of such choices, some interviewees were careful to note that drafting choices are often intertwined with policy decisions. Putting pen to paper to draft the law often raises implementation issues that cannot be foreseen when Members and their staffs agree to high-level policy. As a result, those who actually draft the law can make implementation decisions with significant substantive content.

As just one anecdote of this, one interviewee described how Hussey (and his Legislative Counsel deputy, Buckley) came up with the content to carry out the rehabilitation tax credit. The interviewee described how, tasked with drafting the tax credit, the drafters had to figure out what rehabilitating a building actually meant. As the interviewee described, “They have this inherently murky concept. It’s their job to come up with something clear that you can actually apply. So bingo, [they came up with the rule that a rehabilitation means that you keep] three walls,” a solution that on one level seems arbitrary, but on another “was genius by Ward or John Buckley,” because it took the “inherently murky concept” of a rehabilitation and made it “enforceable.”154 As another interviewee remarked generally, “The technical issues are so enormous . . . so . . . a lot gets delegated to you, a lot of very, very significant stuff gets delegated to you.”155

IV.     Insights and Implications

Our interview findings have important implications. First, our study has implications for statutory interpretation in general. Second, it has implications for the tax system in particular. Third, it underscores several important questions about the legislative process.

A.     Statutory Interpretation

1.     Affirmation of Prior Research

Our study adds to and extends the growing legal literature regarding the implications of statutory drafting practices for statutory interpretation. While political scientists have long been attentive to some of the inner-workings of Congress,156 until recently legal scholars have paid surprisingly little attention to who does what in crafting legislation.157 Recently, a number of scholars have begun to rectify this deficit. Notable contributions include those by Professors Gluck and Bressman, Shobe, and Nourse and Schacter.158 While their extensive work has produced many findings, not all susceptible to distillation, some of their most important findings have included the following: Members of Congress themselves do not write the text of statutes, nor do they focus in particular on the text.159 Rather, Members of Congress focus on the concepts that will be translated into text.160 Professional drafters within Congress actually draft the statutory text.161 These drafters do not view themselves as having a principal relationship with courts. Rather, they principally view themselves in conversation with agencies that will implement the legislation.162 Partly as a result, the drafters pay varying level of attention to judicial canons of construction,163 although the greater professionalization of drafters over time may mean that current drafters pay more attention to judicial canons and legislative drafting issues than occurred in the past.164

This new research has explored how drafting dynamics should impact judicial interpretation.165 For instance, in one article, Gluck and Bressman problematize common justifications for the use of judicial canons and explore what role canons can appropriately play in facilitating courts’ interpretive relationship with Congress.166 In another article, they explore how courts can root their interpretive theories more faithfully in the realities of the legislative process.167 Gluck and Bressman’s research builds on earlier work by Nourse and Schacter, who argued that their findings about the legislative process (which were similar to many by Gluck and Bressman) raise problems for existing judicial theories of interpretation and textualism in particular.168 Shobe in part responds to Gluck and Bressman by arguing that the professionalization of legislative drafting over time suggests that certain approaches to judicial interpretation (in particular textualism) may be more appropriate as applied to newer statutes.169

Our study in some ways affirms the findings of these prior studies.170 In particular, we too found that Members only engage in drafting at a policy level, rather than engaging in the actual drafting of statutory text.171 Subject matter experts engage in the drafting and, when they do so, most are writing with other subject matter experts and intermediaries (including agency officials) in mind.172 They do not see themselves as being in a primary interpretive partnership with courts.173 In this way, our study bolsters some of the questions raised by the earlier work about whether existing theories of judicial interpretation enable courts to be faithful agents. For instance, like in the work of Gluck and Bressman, as well as Nourse and Schacter, our finding that Members are not closely engaged in the actual drafting of statutory text suggests some problems with judicial approaches that focus primarily or exclusively on the text of a statute as the best way to be faithful interpretive partners.174

2.     Deeper Challenges for Textualism

But our study also pushes further than prior work. While the work of Gluck and Bressman and Nourse and Schacter suggests that attention to textual details may not be the best indicator of what Members meant to convey in the statute, their work left open the possibility that textual details may be the best indicator of the meaning that the actual drafters and professional staff meant to convey. Under some views, that may be enough to buoy a textualist approach as the best means of making meaning of statutes.175

Our study challenges the notion that even the drafters and staff themselves necessarily mean to convey significant meaning through small textual details. For instance, interviewees told us that textual decisions can be largely “idiosyncratic,”176 reflecting, for instance, the specifics of how a particular piece of legislation has passed through the legislative process, what staffer has taken ownership of it, the particular personalities that dominated textual choices, attempts to follow old patterns simply for the sake of following patterns, and other path-dependent factors. Perhaps the most whimsical example of this is the fact that, as one interviewee pointed out, the Code contains the use of the word “thru” in numerous places, such as in the phrase “pass-thru,” even though the correct spelling should be “through,” as in “pass-through.”177 As this interviewee explained, “thru” permeates the Code when it really should say “through” simply because Ward Hussey liked using “thru,” a usage that drafters then seemingly simply replicated even after Hussey was gone.178

But staff indifference to particular textual choices extended beyond whimsical word choices, drafting idiosyncrasies, and inertial patterns. As Part III discussed, the government counsels we interviewed, many of whom were closely involved in the crafting and passage of tax legislation, generally did not think that particular textual choices mattered much, as long as the statute conveyed the agreed-upon substantive content. As one interviewee poignantly stated, “I don’t remember writing usually being the issue. That would often be like nitpicking and everyone being stressed out and being like; why are you nitpicking? No . . . real people are going to read the statue.”179

As this statement indicates, at least in the tax arena, many government counsels closely involved in the creation of tax legislation view statutes as a set of instructions that will be mechanically transmitted to its subjects via I.R.S. publications, experts, and software providers, rather than a text that will be carefully parsed.180 These counsels generally do not contemplate that small textual choices in the statute will be ascribed meaning; they simply do not pay much attention to these choices, focusing on substantive policy instead. The counsels’ own perceptions of how meaning will be made of statutes thus sharply contrasts with the judicial practice of textualism.

This incompatibility both deepens the general challenges to textualism mounted by recent literature181 and provides a counterweight against some arguments for textualism in the tax context.182 Aggressive tax planning, including tax shelters, often turns on claims that the Code should be applied in a literal way, with extensive attention being paid to small details that lead to unexpectedly taxpayer favorable results.183 Our study reveals that this type of aggressive tax planning, which relies on hyper-literal interpretations of the Code, is inconsistent with the mindset of Members as well as government counsels. Rather than focusing on small textual choices, Members pay little attention to the Code’s actual text. Moreover, while government counsels do focus on the text, they regard small textual choices as generally unimportant and certainly do not intend for these choices to be used in a way that subverts Congress’s policy.

Thus, hyper-literal text-based analysis claiming that the text indicates Member and counsel intent misconstrues how textual decisions are often made in the actual process of legislating. Of course, there may be other arguments that support reliance on the literal text of the statute—for example, some might argue that the ability of the governed to rely on the legislated word is consistent with rule of law values. But our study does suggest that arguments for textualism that focus on text as a manifestation of Member and counsel intent are not generally reflective of how they actually approach statutory language choices.

3.     Beyond Prior Work: Canons of Construction

Our study also supports and extends some of the findings of recent statutory drafting literature regarding canons of construction. In particular, Gluck and Bressman cast doubt on the underpinnings of many canons of construction, revealing that there is often a disconnect between judicial reliance on such canons and legislative drafting realities.184 Our study supports some of their findings.

For instance, Gluck and Bressman find that none of the publicly stated justifications hold for the rule against superfluities, an interpretive canon that attempts to avoid rendering parts of a statute redundant.185 Despite the fact that courts often rely on this canon, Gluck and Bressman found that drafters often actually intend to inject redundancy into the statute for a variety of reasons, such as to make sure they have captured all possibilities, or because they are using redundancy to satisfy particular stakeholders.186

Our interviewees described similar dynamics, which support the critique of the rule against superfluities. For instance, we heard that staffers in a Member’s office may tinker with statutory language, including by adding unnecessary headings or language that is duplicative of a cross reference, in order to get messaging benefits or satisfy a particular constituency.187 Like in Gluck and Bressman’s study, this description undermines many of the underpinnings of a canon that tries to avoid finding redundant the very redundancies that drafters themselves admit to intentionally using.

However, by examining the way that the tax law develops over time, rather than focusing on counsel perspectives in different areas at a particular point in time, our study also undermines even some of the canons for which Gluck and Bressman found substantial support. Take, as an example, the expressio unius canon, which suggests that the inclusion of specific items on a list suggests the intent to exclude others.188 Gluck and Bressman’s interviewees generally agreed that using a list signals the intent to exclude items not included on the list, thereby supporting the underpinnings of the canon.189

Our study suggests that, even supposing that a list signals an intent to be exclusive at a given point in time, it may be unlikely to do so over time. In describing why they make particular drafting choices, our interviewees described how such choices are influenced by the law’s evolution. Taking a longitudinal view, the existence of a statutory list may not reflect an intent for a current statutory list to be exclusive. Rather, a list may better reflect the statute’s historical development and the difficulty in making changes over time.

For example, the fact that the Code defines a capital asset through the broad, inclusive definition of “property held by the taxpayer,” followed by a list of specific items that are excluded from the definition of capital assets190 does not necessarily mean that the list of exclusions is meant to be exclusive. In originally drafting the provision, the drafters were likely following the format that many interviewees described to us as the Code’s “style”—having a general rule followed by exceptions. Even if the section’s original drafters considered, at the time, all the potential exclusions they could think of, it is unlikely that the retention of this statutory format signals that the list of exclusions is meant to remain exclusive. Given the difficulty in changing statutory formulation,191 it is unlikely that modern drafters have updated the list to include every exclusion from the definition of capital assets that should exist, based on all of the potential classes of assets that exist today.192 Nor would maintaining the general rule plus exception format be likely to signal a considered desire for the list of exclusions to continue to be considered exclusive. Rather, keeping the old format of general rule followed by the exclusions is much more likely to have occurred simply because that is the way the section has always been written.193 Understanding the patterns and inertial tendencies that actually drive the drafting of tax legislation over time thus undermines a court’s ability to rely on the expressio unius canon to reach strong conclusions about the reach of a statute that evolved over time.

The extent to which our study causes us to question the expressio unius canon is just one example of the insights that may be derived from a longitudinal examination of how drafters have actually made drafting decisions over time.194 By uncovering some of the factors that drive how statutory language accretes and evolves, our study takes an important step in that direction.

B.     Implications for Our Tax System

Our study also has implications for the tax system outside of the courts. It illuminates the important distributional tradeoffs inherent in expert-centered drafting, raises questions about how our tax system allocates responsibility for compliance, and suggests the need for strategies to more effectively help taxpayers comply with the law (including through use of technology and artificial intelligence).

1.     Expert-Centered Drafting and Legal Complexity: Distributional Tradeoffs

We set out to study why particular drafting choices are made, for example, why a drafter might use a general rule followed by exceptions, rather than a narrower general rule. However, despite our prodding, few interviewees were able to articulate, or were even particularly interested in articulating, the impact of particular drafting choices, viewing these questions as not very significant. This view derived in part from the belief that expert intermediaries, not ordinary taxpayers, were the audience for tax statutes, along with the unstated assumption that drafting choices by and large did not matter for such experts. Interviewees seemed to imagine a frictionless transmission of statutory content to tax experts of boundless capacity, who would in turn transmit such content to the law’s ultimate subjects. This led interviewees to view the benefits of cleaning up the Code’s formulation to be low. Meanwhile, they simultaneously perceived the costs of such cleanup to be high. Interviewees described how the process of enacting and absorbing new language was difficult and costly,195 politically risky,196 and could unsettle existing court interpretations and upend the expertise of experienced practitioners.197

This expert-centric approach, and the other drafting dynamics that accompany it, leads tax law to have strong inertial tendencies, whereby existing formulations are likely to be preserved and whereby the law accrues in a layer-cake fashion over time and becomes ever more convoluted. These inertial tendencies are a significant but underappreciated contributor to legal complexity. Scholars from a variety of fields have observed and analyzed how the law seems to become increasingly lengthy and complex over time.198 Our study provides unique insights into the process by which drafting can lead to increasing statutory complexity.199 While focusing on substantive rather than formulation issues may be the least costly way of minimizing substantive mistakes, this approach also tends to lead drafters to tolerate increasingly convoluted formulations that accrete over time.

The mortgage interest deduction, which underwent significant modification in the 2017 tax reform, is one example of this phenomenon.200 The 2017 legislation significantly lowered the amount of debt the interest on which is eligible for the deduction.201 This, when paired with the increase in the standard deduction, has drastically reduced the extent to which taxpayers will use the mortgage interest deduction. And yet, in terms of formulation, the 2017 reform left in place the broad general rule stating that all interest is deductible, followed by the exception disallowing the deduction for personal interest, followed by the exception to that exception, which ultimately permits the mortgage interest deduction.202 The new limitation on deductibility was simply layered on top of this already very complex framework in the form of a new “special rule,” (I.R.C. § 163(h)(3)(F)), which imposes the new dollar limitation and only applies from years 2018 through 2025.203 There was no accompanying reworking of the statute’s existing language or structure. The result, over time, is an enormously complex statute that newcomers will likely struggle to understand.204

An increasingly convoluted Code has underappreciated distributional tradeoffs. Most strikingly, while our interviewees highlighted the benefits of law accruing in patchwork fashion, namely not upsetting interpretations of and reliance on current formulations,205 they did not seem to appreciate the costs of not reformulating to anywhere near the same extent. While making patchwork changes may benefit existing parties who have reliance interests on existing Code formulations, an increasingly convoluted and hard-to-read Code creates barriers-to-entry to outsiders, including new entrants to the tax profession and other professional advisors, which may thus yield higher fees for tax advice. A convoluted Code may also create interpretive problems for generalist courts. In addition, the increasing convolution of the Code has often been used to advocate for tax simplification, which may well be a cover for tax deregulation.206

In short, by focusing on the benefits of keeping formulation the same for existing users, drafters tolerate increasing statutory complexity and may be privileging entrenched interests relative to newcomers and laypersons, who may struggle with the impenetrable nature of the tax law. The distributional consequences of these choices—whether unintended or deliberate—may be profound.

2.     Allocating Responsibility in an Expert-Centric System

The fact that the Code is complex and geared towards expert intermediaries raises questions about how our tax system assigns responsibility for tax compliance. Arguably, the way responsibility is currently allocated appears inconsistent with the drafters’ vision of how the law they draft will be used. Our tax system treats taxpayers themselves, rather than experts and software companies, as primarily responsible for accurately filling out returns.207 A taxpayer must sign her tax return attesting that, under penalties of perjury, she has examined her tax return, accompanying schedules and statements, and to the best of her knowledge and belief, they are true and correct.208 Various penalties apply to taxpayers for inaccurate returns.209 There are defenses to penalties such as the “reasonable cause and good faith” exception.210But there is uncertainty regarding the scope of these defenses, particularly for errors that may result in part from the actions of sophisticated intermediaries, such as tax return software companies,211and for bad behaviors that may have been induced by experts’ advice.212In short, there is a risk that our tax system assigns ordinary taxpayers disproportionate responsibility for complying with laws that most drafters do not expect them to understand and that they do not, in fact, understand.

How significant is this risk? Certain factors, if present in our tax system, might reduce it. First, if IRS forms and instructions, tax prep software, and the advice of experts are accessible and comprehensive enough to educate taxpayers, then perhaps it is not so problematic that taxpayers aren’t meant to understand the statutes themselves. Second, if penalties for errors or certain failures to comply were shifted to experts where appropriate, this would arguably bring the allocation of filer responsibility in the tax system in line with drafters’ expectations.213 Finally, if the law was simple and intuitive or the behaviors that constitute wrongdoing were obvious, we might be more inclined to view ignorance of the law as no excuse.214

The problem is that these offsetting factors, as they exist in our tax system, are likely insufficient to offset the risk of disproportionate allocations of responsibility to taxpayers. First of all, the Code is so complex that it is unrealistic to expect that tax forms and instructions can capture every single rule, exception, and nuance of the tax law.215 A taxpayer who just reads forms and instructions risks missing rules and exceptions that may apply. Furthermore, the mere process of communicating financial and other facts from a taxpayer who might not understand the law to an advisor or computer program can result in errors due to misunderstandings about which facts are legally relevant. To be sure, some taxpayers, such as W-2 wage earners, may find forms and instructions sufficient and compliance easy despite complex statutes. However, others (e.g., small business owners,216 taxpayers with cross-border income,217 and taxpayers who are eligible for the Earned Income Tax Credit,218 to name just a few categories) may face greater challenges in interfacing with IRS or expert guidance.

Second, our tax system does not allocate liability to experts and sophisticated intermediaries to anywhere near the same degree as that placed on taxpayers. While preparers also must attest to the accuracy of a tax return they prepare (taking into account the information they actually knew)219 and are subject to certain penalties,220there remains a significant gap between the responsibilities and penalties that apply to preparers, which are generally quite light,221 and those that apply to taxpayers.222Some might argue that penalties are actually imposed on ordinary taxpayers so sparingly that the fact they are unlikely to understand the law is not worrisome. But this response is inadequate. A large swath of taxpayers finds the prospect of IRS audit and potential penalties frustrating and anxiety-provoking,223a phenomenon that the government uses to support its compliance efforts.224Taxpayers’ beliefs that they may be subject to negative consequences for inaccurate application of law they do not necessarily understand is thus relevant, even if the likelihood of penalty is low.

Finally, our tax laws are complex and often unclear and non-intuitive. Furthermore, tax is an area in which the government places an affirmative obligation on the entire taxpaying population to engage in elaborate steps to comply with these detailed and often non-intuitive laws. Thus, the maxim that ignorance of the law is no excuse should arguably be more bounded in the case of tax statutes.225 Indeed, the greater concern with respect to tax statutes may be how to square difficult-to-decipher tax statutes, and taxpayers’ obligations to apply them, with the rule of law principle that the government must make the law accessible to the governed.226 Like the allegory of Emperor Nero placing the law high on tablets where it is difficult to read, which was a classic affront to the rule of law227 drafting tax statutes for experts while leaving compliance responsibility on taxpayers may threaten fundamental rule of law values.

There are potential ways to reform the tax system’s allocation of responsibility to better align with the expert-driven system envisioned and implemented by drafters. However, the system’s very reliance on experts, in addition to other factors, makes these reforms difficult. For instance, one possibility is that rather than starting with a default in which taxpayers are primarily on the hook, perhaps penalties should instead be allocated based on whether the wrong facts or the wrong law were applied. If the wrong facts were applied, the penalty could presumptively be imposed on the taxpayer (who is typically in possession of the underlying facts, such as the amount of income earned in a given year). If, instead, the law is applied incorrectly to the facts, the penalty could presumptively be imposed on the preparer, in cases where a preparer (whether human or software) was responsible for the wrong law. Here, the revised penalty might be based on the amount of underpayment, rather than the current set amount. Such a system may better allocate responsibility to the correct parties and may incentivize preparers to ask taxpayers the right questions and give accurate advice, effectively enlisting preparers as enforcers of the tax law.228

But there are difficulties with this approach. Some may worry that any weakening of penalties on taxpayers themselves may reduce sorely needed incentives for taxpayers to report honestly.229 And perhaps the new allocation of responsibilities would create gaps that could be exploited by taxpayers, preparers, or some combination thereof, thereby undermining deterrence.230

Relatedly, such an approach may present line-drawing issues as between mistakes of fact and law. In some cases, it is not obvious whether a fact is relevant if one does not know the law. Another problem is that if the risk imposed on preparers is too large compared to the economic upside of preparing the return, preparers may refuse to prepare certain returns (for example, tax returns presenting complicated or uncertain tax issues).

More fundamentally, the very dependence of taxpayers on preparers may make it difficult to ensure that preparers actually bear the full burden of any penalties imposed on them. If new penalties are placed on preparers, they may charge high premiums to cover their risk, or they may create contracts that require taxpayers to indemnify the preparers. Preparer resistance to increased penalties, buoyed by a tradition of penalties being placed on taxpayers, would also likely be strong, potentially dooming a significant increase in responsibilities being shifted onto preparers. All of this suggests that any proposed penalty reform would need to be carefully tailored to achieve the desired outcome, and that any such reform might face significant objections from software companies and tax preparers.

3.     Implications for Technology in Tax Preparation

Our study suggests the need for strategies to more effectively help taxpayers comply with law for which they are not the primary audience. Here, it may be useful to consider the interaction between the drafting practices we uncovered and the role of technology in tax preparation. For instance, if taxpayers are not the intended audience of the tax law, and if existing guidance is inadequate, this may suggest that the government should take a more active role filling out taxpayers’ returns for them (for example, by electronically pre-populating tax returns with taxpayer data).231 Along the same lines, perhaps in the future artificial intelligence could play a greater role in helping taxpayers comply with tax law.232 While the tax system already contains some of these features to some degree, the fact that drafters are not writing the tax law intending it to be useable by ordinary taxpayers may suggest the need for a more complete paradigm shift. Such a shift would have the benefit of lessening the burden imposed on taxpayers to “get it right” themselves even if they do not understand the law.

As was the case with penalties, however, there are challenges associated with these types of reforms. For instance, with respect to artificial intelligence, Sarah Lawsky has recently examined how the Code could be reformulated using logical reasoning to make it easier to program for AI.233 However, our findings suggest that drafters are not currently consciously optimizing tax law drafting for the purposes of computer programming. Even though some interviewees mentioned the importance of implementation by TurboTax, most understood the primary audiences for tax statutes to be humans (including IRS form and instruction writers and sophisticated tax advisors), not machines.234 Interviewees repeatedly referenced Code patterns that have existed since the 1950s and did not indicate that drafting practices are designed to optimize computer programming capacities in any way.235 Moreover, as discussed previously, drafting practices tend to be sticky, with drafters continuing to use old models and existing patterns and practices. Existing drafting practices are based on approaches and styles created in drafting manuals that date back to 1989, which certainly were not created with optimal programming considerations in mind.236 All of this suggests that, should we decide that a more complete shift to use of artificial intelligence is warranted, that shift would likely require a more conscious reappraisal and reform of existing drafting practices.

There are also other challenges associated with having the government fill out taxpayers’ returns for them or turning to AI to prepare tax returns. In addition to the privacy issues raised by a significantly greater turn to AI or computers,237 the very power of preparers in the existing system makes some of these reforms difficult. For instance, tax preparation companies on whom the current system relies have strongly lobbied against the government using information already in its possession to automate preparation of taxpayers’ returns.238 These lobbying efforts have impeded more extensive government preparation of taxpayer returns.

Fundamentally, we have a tax system that is geared towards experts and preparers but that still places principal responsibility on taxpayers themselves. There may be good reasons for such a system—for example, the existence of different types of taxpayers and aggressive tax planning behavior may demand complex laws, which then require experts to serve as interpreters and intermediaries.239 But this system also has real costs. It puts power in the hands of experts and intermediaries, potentially raising compliance costs for taxpayers, while simultaneously making it difficult to impose penalties on the experts themselves for mistakes. It also may lead at least some taxpayers to be filled with anxiety about signing tax returns (subject to penalties) that they do not understand.240 The fact that drafters generally know and intend that ordinary taxpayers will not understand the tax law calls for a more conscious reappraisal of these costs.

C.     Implications for the Tax Legislative Process

Our study also provides important insights about the process by which tax legislation is enacted. For instance, while our study did not specifically set out to study the 2017 reform, the data we gathered is useful in helping us contextualize and understand the criticisms of the tax reform process.241 This contextualized understanding calls for a deeper appraisal of the legitimacy of the legislative process.

1.     Member Engagement with Statutory Text: Myth vs. Reality

A subset of criticisms of the recent 2017 tax reform—one of the most significant tax reforms in recent history242—centered on Democratic Members’ lack of opportunity to read the legislation.243 Critics charged that much of the process seemed to happen under cover of night, with little time for the Congress Members (Senators in particular) to read, much less digest, the bill.244 Howls of protest were heard on the Internet, Twitter, and major news sources.245 Democrats in Congress, academics, and ordinary citizens alleged that the fact that Senators did not have time to read the actual text of the bill prior to voting on it (as epitomized by the handwritten scrawls that remained on the final copy) represented a significant breakdown in the legislative process and called into question the bill’s legitimacy.246

It is possible that these protests were meant to communicate something more subtle. For instance, perhaps Democratic Members’ protests about lack of time to read the bill really meant to communicate the lack of time for their staffers to write briefs of the bill for them, or for Members to read their staffers’ briefings. In other words, perhaps Democratic complaints meant to communicate anger at the lack of opportunity to read legislation prior to voting and the lack of opportunity to engage staffers in the process of evaluating it. But Democratic Members did not actually make these more nuanced arguments. Rather, they focused on their own inability to read the bill. For instance, Senate Minority Leader Chuck Schumer protested, “Not a single member of this chamber has read the bill. . . . It would be impossible.”247

These complaints, which focus specifically on Members themselves reading the bill, are simply out of step with the realities of legislating. Our research, as well as other recent research,248 suggests that Congress Members not actually reading the 2017 tax legislation prior to voting was nothing new. Not only did the counsels we interviewed overwhelmingly agree that Members generally do not read the text of proposed legislation even prior to actually voting on it, interviewees also indicated that in passing major tax legislation, it is not uncommon for changes to the statutory text to be made hastily at the last minute on the Senate floor.249

To be sure, the process by which the 2017 tax legislation was passed was remarkable on several fronts. For example, it was characterized by strong partisanship, with Republicans passing the legislation using the reconciliation process, without any Democrat votes, and with minimal opportunity for hearings and the undermining of budget estimators.250 However, outrage surrounding the lack of opportunity for Members to read the bill language papers over the reality that Members had generally not paid attention to the actual statutory language long before the 2017 reform. Put differently, complaints about the lack of opportunity to read statutory language perpetuates a myth that Congress Members are closely engaged with evaluating the statutory text as part of the process of passing legislation.

At the core, public outrage regarding the purported lack of opportunity for Members to read legislation suggests a potential disconnect between the public’s understandings and expectations of how Congress Members ought to consider proposed legislation, and the process by which this actually happens. Member complaints in the popular press regarding the lack of opportunity to read the legislation may have been mere political posturing but may also suggest a desire to perpetuate the myth that Members read statutory language before passage. Commentary by mainstream journalists on this same lack of opportunity may reflect expectations and understandings of the legislative process that are out of sync with reality.

2.     Who Controls Drafting? And Who Should?

All of this begs the question: Is there a legitimacy problem with the process by which legislation is passed? Is the fact that Members do not read drafted legislative language and even government counsels pay relatively little attention to formulation decisions problematic? If professional drafters are really controlling these decisions, what legitimizes their doing so?

As we examined in Section IV.A, formulation decisions can affect how accessible the law is and to whom, and thus have distributive consequences on a number of dimensions.251 Moreover, as we described in Section III.B.3.vii, those who have power to draft the law have power that extends beyond formulation choices, to include significant substantive decisions about the law.252 Yet, Members do not pay attention to drafted statutory text. Indeed, some interviewees even alluded to the benefit of intentionally excluding Members from the drafting process, because Member involvement could slow down or even upend potential tax legislation.253 The upshot is that, over a long period of time, through inattention as well as design, Members have ceded a good deal of control over drafting. In doing so, Members have not only ceded formulation decisions; they have also ceded power over policy decisions that are inextricably intertwined with formulation decisions.254 

So, should we be sanguine about this ceding of control? The answer depends, in part, on who control is being ceded to and how such control is being wielded. This may have changed over the years. During the creation of the 1986 Code, Legislative Counsel drafters—most prominently Ward Hussey and John Buckley—exercised extensive control over the drafting of statutory language.255 But counsels active in later periods described a move toward a more amorphous process, with input from more private sector players, fuzzier lines of control, and a less clear dominance by one or two experienced tax drafters.256

This reduction in expert Legislative Counsel control and turn to a more diffuse and potentially chaotic drafting process may mirror the increasingly “unorthodox legislation” noted by some scholars that has occurred at the congressional level.257 Unorthodox legislation is characterized, among other things, by an erosion of committee control, more diffuse and political lines of control, and the increasing chaos that accompanies the dispersion of power among many sources.258 It is possible, though not inevitable, that the control of expert drafters (such as the tax House Legislative Counsel) would also erode as the power of committees with which they have the closest working relationships eroded (such as the House Ways and Means Committee).259

These types of shifts raise questions regarding how power should be allocated among Congressional counsels and staff, and how allocations of power at this level affect legitimacy.260 Recent literature, most notably the work of Gluck and Bressman, has only very obliquely addressed the question of what legitimates different staffers’ involvement in different parts of the legislative drafting process. For instance, Gluck and Bressman respond to the contention that legislative history is drafted by renegade staff by arguing that the staff that drafts legislative history is actually more directly accountable to Members than Legislative Counsel.261 Gluck and Bressman point to the fact that individual Members cannot fire Legislative Counsel and that Legislative Counsel do not lose their jobs as a result of a shift of control in Congress.262

This begs, but does not answer, the question, of whether and how different staffers’ roles in the legislative process need to be legitimated. One possibility is to look to the extensive discussion of legitimacy in the context of the administrative state.263 The above discussion suggests, in particular, that perhaps we should deploy some sort of analogue of the political control model from the administrative law context, whereby staffers’ roles are legitimated based on their control by politically accountable Members.264 This might suggest that we should look favorably upon a shift in drafting control away from Legislative Counsel and toward political counsels who are more directly subject to Members’ control.

But our findings suggest that there are risks to such a shift. First, shifting power closer to Member staff may understate the value of having a highly skilled expert drafter at the helm who is able to ensure the technical integrity of a complex statute by accurately translating policy into statutory language. More problematically, arguments in favor of such a shift may overestimate the extent to which a lesser degree of expert control will be replaced by an accountable alternative. What we instead heard from interviewees was that a decline in Legislative Counsel dominance has coincided with more private sector interests providing legislative language and more of a personality-driven jockeying for power by other staffers.265 Meanwhile, Members themselves still do not read the language, and even high-level counsels pay more attention to policy than statutory language.266 This suggests the risk that a loss of centralizing, expert control over drafting may be replaced by a less disciplined alternative, rather than a more accountable system.

This analysis suggests that perhaps we need to look elsewhere in thinking about legitimating staffers’ roles in the legislative process.267 Or, to the extent that we think that staffers’ roles need no legitimating because they act as “mere transmission belt[s]”268 for Congress, perhaps we need to be more explicit about and prepared to defend such a theory. Or perhaps some combination of theories is the most fruitful approach. We do not seek to resolve the question in this Article, but rather underscore fundamental questions about the legitimacy of drafting practices that merit deeper examination than exists in the current literature.

3.     The Future of Tax Reform

Understanding long-standing drafting dynamics may also help us better predict the future of the recent tax reform. For instance, in evaluating the 2017 tax reform, many commentators have suggested that extensive technical corrections will be needed to clean up mistakes that were made by hasty drafting, and, indeed, prominent lawmakers themselves have promised that technical corrections will be used to clean up the legislation.269 Our study shows how, while not impossible, such technical corrections will have to overcome significant barriers. As previously discussed, our interviewees explained that technical corrections are often quite difficult to accomplish.270 This is due in part to Member and staff suspicion that technical corrections may not be merely technical, and also due to the reluctance to use scarce congressional resources to fix past mistakes.271 The difficulty in passing technical corrections may be even greater for controversial legislation passed through a divided Congress.272

D.     Future Research

Beyond the implications we discussed above, various themes that arose in our semi-structured interviews suggested avenues for future exploration. For instance, some interviewees identified a lesser presence of executive branch officials in drafting over time. Interviewees pointed to various explanations, such as conflicts between Congress and the administration, executive branch officials’ desires to stay away, the elimination of an IRS division dedicated to legislation and regulations, and the advent of email. These findings suggest that much may be learned by exploring the relationship between congressional committees and their executive branch counterparts, and by investigating how such relationships have shifted over time and how this shift may affect the law’s creation and administration.

Another theme that arose in our interviews was a shift in partisanship in the process over time. Some interviewees lamented a move away from a less partisan process of creating the law to a more partisan process in which the minority party’s views have less of a place.273 But other recent drafters suggested that, even though Congress itself may be more partisan, staffers would try to work across party lines in good faith. This dynamic, if it exists, raises the intriguing prospect of a less partisan congressional staffer deep state, the inner workings of which merit exploration.

In addition, many interviewees referred to and sometimes lamented the shift in markup practices in the House to a markup of actual legislative language (as opposed to a policy markup) and offered various reasons for and thoughts about this shift. The implications of this shift are worth further investigation. For instance, did the change in markup practice affect the shape of legislation? What would the answer to this question tell us about how drafting practices should ideally be structured?

Finally, though some interviewees spoke about the choice to delegate drafting decisions to Treasury Regulation writers, this Article has not probed how drafting choices interact with regulatory choices. The implications of how these decisions are made could be profound, and we explore this subject in other work.274

These findings and others we uncovered from our conversations with our interviewees were not the principal subject of our inquiry, so we were not able to develop them either empirically or theoretically in this Article. We hope to investigate some of these topics in future work.

V.     Conclusion

In this Article, we investigated how and why drafting choices are made in the formulation of tax statutes, and we analyzed the implications of these choices. We conducted extensive interviews with government counsels who were involved in the creation of tax legislation over the past four decades. These interviews produced a wealth of information about how tax law is made and, in particular, how discrete drafting decisions are made as part of that process. We explored how these findings have important implications for statutory interpretation, the tax system, and the legislative process.

Fundamentally, this Article uncovers an aspect of legislative drafting that is generally hidden from view: how and why decisions are made to put particular words in the statute books. We found that these decisions both reflect and perpetuate particular visions of lawmaking and administration. Understanding them is thus essential for legal analysis and scholarship.

Appendix: Interview Script for Semi-Structured Interviews

These questions are to be used as a guide to begin conversation.

Opening/Setup: We are conducting research on the tax legislative process. Specifically, we are interested in how drafting choices are made in the tax legislative process and what, if any, the effects of those choices might be. We are conducting interviews with counsel and staff who have been involved with the drafting and creation of tax legislation to investigate these issues. 

1.     Could you tell us a bit of background about your experience with the tax legislative drafting process?

a.     In what capacity and on what committee did you serve?

b.     For what years (how long)?

2.     How often did your committee get primary drafting responsibility for drafting legislation?

a.     Or, did legislation arrive at your feet already drafted?

b.     If the legislation arrived on your desk already drafted, who did the drafting?

3.     Once a certain policy or scheme has been agreed upon in principle, who makes decisions regarding the drafting of the tax statutes? Where/in what committee/by what person in such committee do these decisions get made?

a.     Does the process of figuring out how to draft tax legislation change the substantive legal rule?

4.     What was the role of congressional staff persons (i.e., staff of individual congresspersons) and/or lobbyists and/or executive branch officials in the drafting of tax legislation? How often was legislation drafted by those actors, relative to being solely drafted by congressional drafting committees?

5.     We are particularly interested in how decisions about presentation of the rule in the statute get made. How, in your experience were drafting decisions made when there were multiple ways to present the same legal content?

6.     How, if at all, did the process of drafting tax legislation differ if the legislation changed an existing tax statute (for instance, by adding an additional distinction or exception to an already existing statutory provision) versus if the legislation addressed new subject matter not already covered by tax statutes?

7.     We have noticed that general rules combined with exceptions (“exception-based drafting”) can achieve the same substantive goals as a general rule that encompasses definitions or commonly understood terms (“definition-based drafting”).

a.     Is this phenomenon something you have observed or thought about?

b.     If so, what are some examples that come to mind?

c.     Based on your observations and work at the [insert committee(s)], why do you think some statutes are drafted using “exception-based drafting” rather than “definition-based drafting”? Are there certain factors or considerations that explain the choice the between the two?

d.     Did you ever have to make choices regarding whether to use “exception-based drafting” or “definition-based drafting”? If so, please tell us about the choice, what you chose, why you made the choice, and what, if any, consequences you think flowed from that choice.

e.     Looking back on it, what consequences, if any, do you think flowed from the choice to use “definition-based drafting” versus “exception-based drafting?” Does your answer depend on context and, if so, how? 

8.     We have noticed that sometimes in the Code exceptions appear to “swallow the general rule.”

a.     Is this phenomenon something you have observed or thought about?

b.     If so, what are some examples that come to mind?

c.     In these situations, wouldn’t it be possible to draft the statute to create a narrower general rule (rather than having a giant exception)?

d.     Based on your observations and work at the [insert committee(s)], why do you think some statutes use “exception swallows the rule drafting” rather than “narrower general rule” drafting?

e.     Based on your observations and work at the [insert committee(s)], are there general principles that you believe explain when “exception swallows the rule” drafting is used, rather than “narrower general rule drafting”? Are there certain factors or considerations that explain the choice between the two?

f.     Did you ever have to make choices regarding whether to use “exception-swallows the rule drafting” or “narrower general rule drafting”? If so, please tell us about the choice, what you chose, why you made the choice, and what, if any, consequences you think flowed from that choice.

g.     Looking back on it, what consequences do you think might have flowed from the choice to use “exception swallows the rule drafting” versus “narrower general rule drafting”? Does your answer depend on context and, if so, how? 

9.     We have noticed that exceptions to a rule are often contained and addressed with the same rule. We refer to this as “same section general rule and exceptions.” But we have also noticed that sometimes the exception is not contained within the same section and may not even be referred to explicitly as an exception to the prior rule. We refer to this as “apparently independent rules” drafting. 

a.     Is this phenomenon something you have observed or thought about?

b.     If so, what are some examples that come to mind?

c.     Based on your observations and work at the [insert committee(s)], why do you think some provisions are presented as “apparently independent rules,” rather than as “same section general rule and exceptions”? Are there certain factors or considerations that explain the choice between the two?

d.     Based on your observations and work at the [insert committee(s)], are there general principles that you believe explain when “apparently independent rules” drafting is used, rather than “same section general rule and exceptions”?

e.     Did you ever have to make choices regarding whether to use “apparently independent rules” or “same section general rule and exceptions”? If so, please tell us about the choice, what you chose, why you made the choice, and what, if any, consequences you think flowed from that choice.

f.     Looking back on it, what consequences do you think may have flowed from the choice to use “apparently independent rules” versus “same section general rule and exceptions”? Does your answer depend on context and, if so, how? 

10.     We have noticed that sometimes a particular code section will have its own definition for a specific term (for instance: income). Other times, a code section would rely on a definition offered in another provision (cross-references). Other times, a code section would use a word that could have been defined but offer no definition, either within the section or via cross reference.

a.     Is this phenomenon something you have observed or thought about?

b.     If so, what are some examples that come to mind?

c.     Based on your observations and work at the [insert committee(s)], are there general principles that you believe explain when these approaches are used?

d.     Did you ever have to make choices regarding whether to use “within section definitions,” “cross-referenced definitions,” or “no definition drafting”? If so, please tell us about the choice, what you chose, why you made the choice, and what, if any, consequences you think flowed from that choice.

e.     Looking back on it, what consequences do you think may have flowed from the choice to use “within section definitions,” “cross-referenced definitions,” or “no definition drafting”? Does your answer depend on context and, if so, how? 

11.     Did prior drafting choices for an existing Code section or sections in any way affect your ability to draft new legal content?

12.     Does the way tax statutes are actually written (i.e., the form) matter? In general, who is the audience for drafted statutes?

13.     What do you think the key drafters of tax legislation might say about complexity/charges that the Code is too complex?

14.     Are there any other questions you think we need to be asking to understand the drafting choices that are made for presenting legal content? Are we asking the right questions? What other interesting features of the technical statutory drafting process do you think we should be asking about?

15.     Might you have suggestions for other people who were involved in legislative drafting process we should be talking to? 


  1. [1]. Thomas Kaplan, House Gives Final Approval to Sweeping Tax Overhaul, N.Y. Times
    (Dec. 20, 2017),; Damian Paletta & Jeff Stein, Sweeping Tax Overhaul Clears Congress, Wash. Post (Dec. 20, 2017),; Deirdre Walsh et al., White House, GOP Celebrate Passing Sweeping Tax Bill, CNN (Dec. 20, 2017, 5:37 PM),

  2. [2]. Erica Werner, Precision Sacrificed for Speed as GOP Rushes Ahead on Taxes, Wash. Post
    (Dec. 10, 2017),

  3. [3]. See David Kamin et al., The Games They Will Play: Tax Games, Roadblocks, and Glitches Under the 2017 Tax Legislation, 103 Minn. L. Rev. 1439, 1442–44 (2019).

  4. [4]. Adam Looney, How the New Tax Bill Encourages Tax Avoidance, Brookings Inst.: Unpacked (Dec. 14, 2017),

  5. [5]. Edward J. McCaffery, The Tax Bill’s Loopholes for the Wealthy Benefit Trump and Corker, CNN (Dec. 18, 2017, 4:41 PM),; Matthew Wisner, Tax Bill Loopholes Still Exist for Wall Street Fat Cats: Trish Regan, FOXBusiness (Dec. 19, 2017),

  6. [6]. Josh Keefe, Senator Bob Corker Said he Hasn’t Read the Tax Bill, Denies Changing his Vote in Exchange for Personal Tax Breaks, Int’l Bus. Times (Dec. 16, 2017, 6:50 PM), https://www.ib (criticizing Members not having time to read the bill before voting on it); Jim Tankersley & Alan Rappeport, A Hasty, Hand-Scribbled Tax Bill Sets Off an Outcry, N.Y. Times
    (Dec. 1, 2017),; Z. Byron Wolf, The Senate Voted on a Tax Bill Pretty Much Nobody Had Read, CNN (Dec. 2, 2017,
    5:20 AM),

  7. [7]. John Cassidy, The Passage of the Senate Republican Tax Bill Was a Travesty, New Yorker (Dec. 2, 2017),

  8. [8]. For discussion of the tax legislative process, see generally Ronald A. Pearlman, The Tax Legislative Process: 1972–1992, 57 Tax Notes 939 (1992); George K. Yin, Codification of the Tax Law and the Emergence of the Staff of the Joint Committee on Taxation (Va. Law & Econs. Research Paper No. 2017-20, 2017), A significant empirical legal literature has begun to develop regarding the legislative process generally. Lisa Schultz Bressman & Abbe R. Gluck, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part II, 66 Stan. L. Rev. 725, 731–33 (2014) [hereinafter Gluck & Bressman II]; Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpretation from the Inside—An Empirical Study of Congressional Drafting, Delegation, and the Canons: Part I,
    65 Stan. L. Rev. 901, 924–48 (2013) [hereinafter Gluck & Bressman I]; Victoria F. Nourse
    & Jane S. Schacter, The Politics of Legislative Drafting: A Congressional Case Study, 77 N.Y.U. L. Rev. 575, 577–78 (2002); Jarrod Shobe, Intertemporal Statutory Interpretation and the Evolution of Legislative Drafting, 114 Colum. L. Rev. 807, 811 (2014).

  9. [9]. For an early exception, see Michael Livingston, Congress, the Courts, and the Code: Legislative History and the Interpretation of Tax Statutes, 69 Tex. L. Rev. 819, 822–23 (1991) (discussing the realities of the tax legislative process, informed by the author’s experience as a Legislation Attorney with the Joint Committee on Taxation, and exploring the implications for interpretation of tax statutes and use of legislative history in particular).

  10. [10]. See Office of the Legislative Counsel: U.S. House of Representatives, https://legcounsel. (last visited Oct. 24, 2018)
    (describing the role and function of the House Office of Legislative Counsel); Office of the Legislative Counsel: United States Senate, (last visited Oct. 24, 2018) (describing the role and function of the Senate Office of Legislative Counsel).

  11. [11]. Gluck & Bressman I, supra note 8, at 912–14; Gluck & Bressman II, supra note 8, at 784–90 (exploring impact of findings on textualism and other theories of judicial interpretation); Nourse & Schacter, supra note 8, at 616–21 (exploring implications of findings for theories of judicial interpretation, in particular raising problems for textualism and originalism); Livingston, supra note 9, at 836–38, 872–86 (discussing significance of drafting for statutory interpretation); Shobe, supra note 8, at 851–60 (examining textualism in light of findings); see also BJ Ard, Comment, Interpreting by the Book: Legislative Drafting Manuals and Statutory Interpretation, 120 Yale L.J. 185, 194–200 (2010) (examining drafting manuals and the puzzle they pose for textualists); Grace E. Hart, Note, State Legislative Drafting Manuals and Statutory Interpretation, 126 Yale L.J. 438, 443 (2016) (surveying drafting manuals used by drafters in state legislatures to inform statutory interpretation).

  12. [12]. See, e.g., Gluck & Bressman I, supra note 8, at 940 (finding that Members interact with abstract legislative policy rather than the granular text); Nourse & Schacter, supra note 8, at 585 (finding that legislative text is mostly drafted by staff); see also Livingston, supra note 9, at 833–36 (asserting that the same is true in the tax legislative process). While our study focuses on how drafting choices are made for given substantive content, which was not the focus of any of these prior studies, at times our interviews touched on similar issues (such as, for instance, Congress’s general lack of involvement in the details of statutory drafting). We note any replication of findings. See, e.g., Leonard E. Burman et al., A Call for Replication Studies, 38 Pub. Fin. Rev. 787, 787–88 (2010) (pointing to the importance of replication in empirical research).

  13. [13]. See infra Part II.

  14. [14]. See infra Part III.

  15. [15]. See infra Part IV.

  16. [16]. See infra Part V.

  17. [17]. Indeed, this statement is so clearly true that cites could include gigantic swaths of tax scholarship. To pick one issue, scholars have engaged in extensive debate about the merits of an income tax versus a consumption tax. See generally, e.g., William D. Andrews, A Consumption-Type or Cash Flow Personal Income Tax, 87 Harv. L. Rev. 1113 (1974) (conceptualizing a cash flow personal income tax); Joseph Bankman & David A. Weisbach, The Superiority of an Ideal Consumption Tax over an Ideal Income Tax, 58 Stan. L. Rev. 1413 (2006) (arguing that an ideal consumption tax is superior to an ideal income tax); Daniel Shaviro, Beyond the Pro-Consumption Tax Consensus, 60 Stan. L. Rev. 745 (2007) (challenging the case for the consumption tax by examining relationship to the permanent income hypothesis); Joseph Bankman & David Weisbach, Reply, Consumption Taxation Is Still Superior to Income Taxation, 60 Stan. L. Rev. 789 (2007) (addressing the arguments raised by Shaviro).

  18. [18]. See infra Sections II.B–.C.

  19. [19]. A full survey of the tax legislative process is beyond the scope of this article. For examples of scholarship studying the tax legislative process, see generally Harry L. Gutman, Reflections on the Process of Enacting Tax Law, 86 Tax Notes 93 (2000) (discussing various facets of the tax legislative process in addition to areas in the tax code in need of rationalization); Pearlman, supra note 8 (discussing changes to the tax legislative process driven by budgetary concerns); George K. Yin, The Evolving Legislative Process: Implications for Tax Reform, 114 Tax Notes 313 (2007) (discussing how trends in the legislative process affect tax reform); Yin, supra note 8 (describing the role of the Joint Committee on Taxation in the tax legislative process).

  20. [20]. See generally Barbara Sinclair, Unorthodox Lawmaking: New Legislative Processes in the U.S. Congress (5th ed. 2017) (describing updated, foundational work regarding unorthodox legislation); Abbe R. Gluck et al., Unorthodox Lawmaking, Unorthodox Rulemaking,
    155 Colum. L. Rev. 1789 (2015) (developing an account of unorthodox legislation and unorthodox rulemaking and linking the two); Rebecca M. Kysar, Tax Law and the Eroding Budget Process, 81 Law & Contemp. Probs. 61 (2018) (describing how pressure to enact recent tax reform has eroded traditional budget processes). In some ways, our findings in Part III about how formulation decisions are made interact with and capture how the actual legislative process differs from this textbook description. In future work, one of us will explore further the problems of legislative drafting errors that flow from unorthodox legislation and how best to fix them. Leigh Osofsky, Post-Legislation Gridlock (working paper) (on file with author).

  21. [21]. See, e.g., infra notes 125–30 and accompanying text (outlining differences in the case of major legislation).

  22. [22]. U.S. Const. art. 1, § 7, cl. 1.

  23. [23]. About, U.S. House of Representatives Comm. on Ways & Means, https://waysand (last visited Oct. 25, 2018).

  24. [24]. Boris I. Bittker & Lawrence Lokken, Federal Taxation of Income, Estates and Gift ¶ 116A.1 (2018).

  25. [25]. Id. ¶ 116A.2; Full Committee, U.S. House of Representatives Comm. On Ways & Means, (last visited Oct. 25, 2018).

  26. [26]. In 1985, the Ways and Means Committee held 28 days of hearings on the bill that became the Tax Reform Act of 1986. Bittker & Lokken, supra note 24, ¶ 116A.2.

  27. [27]. Id.

  28. [28]. Id.

  29. [29]. See Ways & Means Comm., Markup of the Tax Cuts and Jobs Act, YouTube (Nov. 6, 2017),; Ways & Means Comm., Markup of the Tax Cuts and Jobs Act, Day 2, YouTube (Nov. 7, 2017),; Ways & Means Comm., Markup of the Tax Cuts and Jobs Act, Day 3, YouTube (Nov. 8, 2017),; Ways
    & Means Comm., Markup of the Tax Cuts and Jobs Act, Day 4, YouTube (Nov. 9, 2017),

  30. [30]. Joint Comm. on Taxation, U.S. Cong., About the Joint Committee on Taxation 1, (last visited October 25, 2018).

  31. [31]. Bittker & Lokken, supra note 24, ¶ 116A.2.

  32. [32]. Id. However, in 2017, Treasury did not release an assessment of the House tax plan. It did release a widely panned one-page analysis of the Senate Finance Committee tax bill suggesting that the tax reform would pay for itself. See Alan Rappeport & Jim Tankersley, Treasury Defends Tax Plan Cost with One-Page Analysis, N.Y. Times (Dec. 11, 2017),

  33. [33]. Bittker & Lokken, supra note 24, ¶ 116A.2.

  34. [34]. See FAQ: Committee Hearings, U.S. Senate Committee on Fin.,
    about/faq (click on “What’s the difference between a hearing and an open executive session?”) (last visited Oct. 25, 2018) (noting the difference between hearings and “open executive sessions” or markups; hearings are held “to gather information and opinions on proposed legislation” while markups are “to debate, amend and rewrite proposed legislation before reporting the bill to the full Senate for consideration”).

  35. [35]. Bittker & Lokken, supra note 24, ¶ 116A.2.

  36. [36]. Id.

  37. [37]. Id.

  38. [38]. For more details, see Joint Committee Role in the Tax Legislative Process, Joint Committee on Tax’n (1970), (explaining the role the Joint Committee on Taxation plays in creating tax legislation).

  39. [39]. I.R.C. §§ 8001–8005 (2012); id. §§ 8021–8023; see also Joint Comm. on Taxation, supra note 30, at 1 (providing an overview of the role of the JCT). See generally Yin, supra note 8 (providing a general overview of the tax legislative process). In this Article, we use the shorthand “Members” to refer to Members of Congress (in other words, elected congresspersons themselves).

  40. [40]. I.R.C. § 8022; see also Yin, supra note 8 (manuscript at 4–5 & n.17) (citing E.W. Kenworthy, Colin F. Stam: A Study in Anonymous Power, in Adventures in Public Service 105, 113–14 (Delia Kuhn & Ferdinand Kuhn, eds. 1963)) (discussing the prospect that JCT staff were directed to participate in tax legislation development).

  41. [41]. See Joint Comm. on Taxation, supra note 30, at 1; Bittker & Lokken, supra note 24, ¶ 116A.2.

  42. [42]. 2 U.S.C. § 601(f) (2012). See generally Bernard M. “Bob” Shapiro, The Last 50 Years: The Evolving Role of the Joint Tax Committee, 151 Tax Notes 1125 (2016) (discussing various aspects JCT’s role, including revenue estimates); Randall D. Weiss & James W. Wetzler, The Evolution of Economics at the Joint Committee on Taxation, Part 2, 151 Tax Notes 1113 (2016) (discussing revenue estimation role of the JCT).

  43. [43]. See, e.g., supra note 19.

  44. [44]. Frederick Schauer, Exceptions, 58 U. Chi. L. Rev. 871, 872 (1991).

  45. [45]. Id. For an example of some of the follow-on work to Schauer, see, for example, Glanville Williams, The Logic of “Exceptions, 47 Cambridge L.J. 261 (1988).

  46. [46]. Schauer, supra note 44, at 878–79.

  47. [47]. Or, alternatively: Sex is illegal, except sex with one’s spouse is legal.

  48. [48]. Schauer examined this phenomenon generally, but claimed that existing language and societal understandings dictated whether an exception needed to be created or not. Schauer, supra note 44, at 872. Others suggested that an exception may do more. See, e.g., Claire Oakes Finkelstein, When the Rule Swallows the Exception, 19 Quinnipiac L. Rev. 505, 530–37 (2000); Jeremy B. Stein, Note, The Necessary Language of Exceptions: A Response to Frederick Schauer’s “Exceptions, 63 N.Y.U. Ann. Surv. Am. L. 99, 102 (2007).

  49. [49]. I.R.C. § 121(b)(1) (2012).

  50. [50]. Id. § 121(a). The rule is more complicated than this. For instance, the exclusion can be increased in the case of married taxpayers filing a joint return. Id. § 121(b)(2). If a taxpayer does not meet the 2 out of 5 year rule, under certain circumstances the taxpayer can qualify for a lesser exclusion. Id. § 121(c)(1). The point here, and elsewhere, is not to provide a comprehensive explanation of the tax law, but rather to highlight particular choices that are made in drafting a statute.

  51. [51]. Another possible exception framing is: “Gain from the sale or exchange of property is taxable, except that up to $250,000 of gain on the sale or exchange of property is not taxable if the taxpayer owns and uses the property as a principal residence for 2 of last 5 years.”

  52. [52]. I.R.C. § 61(a).

  53. [53]. Id. § 101(a)(2) (providing the transfer for valuable consideration exception).

  54. [54]. Id. § 102.

  55. [55]. Id. § 103.

  56. [56]. Id. § 163.

  57. [57]. Id. § 163(h)(1). The exception for personal interest is only one of the exceptions to I.R.C. § 163. For example, the 2017 tax reform introduced a new limitation on deductible business interest, which is another significant exception. Id. § 163(j).

  58. [58]. Id. § 163(h)(2)(D). For the 2018 to 2025 tax years, the definition of “qualified residence interest” in I.R.C. § 163(h)(3) has been narrowed to exclude “home equity indebtedness” and to cap the amount of acquisition indebtedness upon which the mortgage interest deduction is allowable. Id. § 163(h)(3)(F).

  59. [59]. Id. § 163(h)(2).

  60. [60]. Cf. Schauer, supra note 44, at 878–80 (stating that the ability to craft general rules depends on existence of terms with underlying social meaning).

  61. [61]. I.R.C. §§ 957(a), 163(i).

  62. [62]. See id. §§ 351, 721.

  63. [63]. See, e.g., id. § 351.

  64. [64]. Id. § 351(a).

  65. [65]. Id. § 351(b).

  66. [66]. Id. § 351.

  67. [67]. Id. § 721.

  68. [68]. Id.

  69. [69]. Id. § 721(a).

  70. [70]. Id. §§ 707(a)(2)(B), 731(a), 752(b).

  71. [71]. See id. §§ 731, 752(b).

  72. [72]. Id. § 108(d)(1).

  73. [73]. Id. § 351(a).

  74. [74]. Id. §§ 304(c)(3), 318(a)(2)(C).

  75. [75]. Id. § 199A(d).

  76. [76]. Id. § 7701.

  77. [77]. Id.

  78. [78]. See, e.g., id. § 162.

  79. [79]. Id. § 61.

  80. [80]. Id. § 108(d)(1).

  81. [81]. Id. § 82. The 2017 tax reform suspends this exclusion for years 2018–2025. H.R. Rep. No. 115-466, at 278–77 (2017).

  82. [82]. I.R.C. § 132.

  83. [83]. See Lee Anne Fennell & Richard H. McAdams, Inversion Aversion, 86 U. Chi. L. Rev. (forthcoming 2019) (manuscript at 1–2) (discussing how sometimes the value of foundational theories is as a foil).

  84. [84]. Cf. Gail Levin Richmond, Federal Tax Research 47 (9th ed. 2014) (explaining that related sections referring to each other “is the ideal situation,” whereas the “worst-case scenario” is when neither related section refers to the other).

  85. [85]. It is not clear that even experts are immune to convolution in a Code. While experts are perhaps in a better position to digest more convoluted statutory language due to training or experience, convoluted language is nonetheless likely to be more costly and difficult to parse, relative to simpler alternatives. Experts are also not immune to cognitive biases that may result from different drafting choices. See, e.g., Jon D. Hanson & Douglas A. Kysar, Taking Behavioralism Seriously: The Problem of Market Manipulation, 74 N.Y.U. L. Rev. 630, 661 (1999) (explaining that “when the pertinent events are not easily predictable and the feedback is not unambiguous, experts tend to be even more overconfident than laypersons”).

  86. [86]. Barbara H. Karlin, Tax Research 63, 65–67 (4th ed. 2008) (explaining that “[l]earning to read and interpret the [Code] is definitely a formidable challenge” and exploring the more intensive search processes new users of the Code should use, relative to the quicker processes that experienced users can rely upon); Richmond, supra note 84, at 48 (exploring how differing levels of knowledge of the Code require different types of searches in order to find issues like hidden connections between different provisions).

  87. [87]. For example, the growing use of pass-through entities in domestic and cross-border tax planning since the issuance of the “check-the-box” regulations in 1997 has meant that understanding the partnership provisions of the Tax Code has become increasingly important for tax lawyers. But the partnership provisions are exceedingly complex, with the result that the group of lawyers with expertise in partnership law is small. See, e.g., Andrea Monroe,Hidden in Plain Sight: IRS Publications and a New Path to Tax Reform, 21 Fla. Tax Rev. 81, 126–27 (2017) (describing both prevalence and complexity of tax partnerships).

  88. [88]. Evidence regarding the impact of professional preparers on compliance is mixed. Some studies have found a correlation between professional preparers and lower compliance. Brian Erard, Taxation with Representation: An Analysis of the Role of Tax Practitioners in Tax Compliance,
    52 J. Pub. Econ. 163, 194 (1993). Other studies have found that professional preparers reduce compliance when there is legal ambiguity, but increase compliance when the law is clear. Steven Klepper & Daniel Nagin, The Role of Tax Preparers in Tax Compliance, 22 Pol’y Sci. 167, 168 (1989).

  89. [89]. Joshua D. Blank & Leigh Osofsky, Simplexity: Plain Language and the Tax Law, 66 Emory L.J. 189, 192 (2017).

  90. [90]. See id. at 238–41.

  91. [91]. See Victoria F. Nourse, A Decision Theory of Statutory Interpretation: Legislative History by the Rules, 122 Yale L.J. 70, 90 (2012) (“Purposivists begin with text, and textualists look to text to find purpose.”).

  92. [92]. See Susan C. Morse & Leigh Osofsky, Regulating by Example, 35 Yale J. on Reg. 127, 137 (2018).

  93. [93]. United States v. Barnes, 222 U.S. 513, 518 (1912).

  94. [94]. I.R.C. § 1221 (2012).

  95. [95]. For example, I.R.C. § 62, which defines “adjusted gross income,” contains a list of “above the line” deductions (deductions that can be taken in addition to the standardized deduction in computing gross income), a list that has grown over time.

  96. [96]. See generally Schauer, supra note 44 (discussing the role the exception has played in our legal system).

  97. [97]. See Edward A. Zelinsky, The Defined Contribution Paradigm, 114 Yale L.J. 451, 513–22 (2004) (exploring how the defined contribution retirement paradigm has made the Code a consumption tax for middle-class taxpayers). See generally Andrews, supra note 17 (making the case for a consumption tax).

  98. [98]. Joseph Bishop-Henchman, How Many Words Are in the Tax Code?, Tax Found. (Apr. 15, 2014),; Scott Greenberg, Federal Tax Laws and Regulations Are Now over 10 Million Words Long, Tax Found. (Oct. 8, 2015),

  99. [99]. See, e.g., David F. Bradford, Untangling the Income Tax 266–67 (1986) (exploring different types of tax law complexity).

  100. [100]. See Sharan B. Merriam & Elizabeth J. Tisdell, Qualitative Research: A Guide to Design and Implementation 98 (4th ed. 2015) (describing snowball sampling as using early participants to refer other participants).

  101. [101]. Although almost every person we interviewed was a lawyer, some were not.

  102. [102]. By way of comparison, in Gluck and Bressman’s study of statutory interpretation, in which they interviewed counsels with responsibility for legislative drafting working across all areas of legislation (not just tax), the researchers conducted surveys of 137 counsels (out of a maximum possible 650). See generally Gluck & Bressman I, supra note 8 (explaining the method used in Gluck and Bressman’s study). Shobe’s study of intertemporal statutory interpretation included seven interviews of congressional staff, confirming research, and Shobe’s experience as an intern for a period in the Office of Legislative Counsel. Shobe, supra note 8, at 817–18. Nourse and Schacter’s study regarding statutory drafting included interviews with 16 staffers on the Senate Judiciary Committee and two lawyers from Senate Legislative Counsel. Nourse & Schacter, supra note 8, at 578–79.

  103. [103]. Merriam & Tisdell, supra note 100, at 97–98.

  104. [104]. See, e.g., Anonymous Interview Collection, Interviewee 3 (on file with author) (“I didn’t think that it mattered that much whether a lay person turned to the Code [and] what the provisions looked like. I didn’t think that was going to deter the audience.”); id. Interviewee 12 (“I’ve never thought about it. I don’t know why things are drafted one way one time and another way another time.”); id. Interviewee 20 (“I would say that as I can remember we gave very little thought to questions like that [i.e., questions like whether to use broad general rule followed by exceptions or whether to use cross references or defined terms]. . . . I don’t remember consciously thinking about that at all.”); id. Interviewee 21 (“I think those decisions are largely by House Legislative Counsel and are not important.”).

  105. [105]. See, e.g., id. Interviewee 21 (“[W]e as a staff wanted to make life easier for taxpayers... . So we created a whole regime ... that was really complicated, no member of Congress would understand what we’re doing and what the words meant. But the end user, the taxpayer, had their lives simplified immensely. They got a number from [a third party], they put the number on the return, and then went to sleep. Or they imported stuff from TurboTax... . Yes, there’s a level of complexity to the actual statute, and that complexity is necessary in often cases... . And the real question, though, is how can that complexity be translated in a way so that the end user, the taxpayer, can complete the return and comply without going through undue expenditures and efforts, and confusion.”).

  106. [106]. Id. Interviewee 13.

  107. [107]. This was consistent with Gluck and Bressman’s finding that counsels in their study generally did not intend for courts to be their interpretive partners. Gluck & Bressman II, supra note 8, at 765–77; see also Nourse & Schacter, supra note 8, at 600 (finding lack of attention by counsels to courts’ interpretive methods); Shobe, supra note 8, at 831 (finding the same).

  108. [108]. Anonymous Interview Collection, supra note 104, Interviewee 7.

  109. [109]. Id. Interviewee 20.

  110. [110]. For discussion, see infra Section III.B.3.

  111. [111]. See Yin, supra note 8 (manuscript at 46 & n.240) (noting designation of JCT as official scorekeeper beginning in 1974). See generally Pearlman, supra note 8 (discussing major tax legislative developments from 1972 to 1992).

  112. [112]. Anonymous Interview Collection, supra note 104, Interviewee 22.

  113. [113]. See, e.g., id. Interviewee 7 (“You are having everybody change all of their systems that for one reason or another have relied on words that were used before. So you just don’t do that unless you have a very strong reason. You don’t do it just to improve marginally. You have to [have] an overriding reason to change wording that has been in the Code for a long time in some structural way.”).

  114. [114]. See, e.g., id. Interviewee 14.

  115. [115]. It is well understood that the legislative process is designed to make change difficult. See, e.g., Maxwell L. Stearns, The Misguided Renaissance of Social Choice, 103 Yale L.J. 1219, 1272 (1994) (explaining that “numerous negative legislativecheckpointsrender passage of proposed legislation difficult by design” (footnote omitted)). However, to our knowledge, no one has yet observed or examined how inertial tendencies apply to the formulation, as opposed to the substance, of the law.

  116. [116]. Anonymous Interview Collection, supra note 104, Interviewee 25.

  117. [117]. For more information on technical corrections, see generally Marc J. Gerson, Technically Speaking: The Art of Tax Technical Corrections, 2007 Tax Notes 927 (explaining what tax technical corrections are and the process for getting them enacted).

  118. [118]. Anonymous Interview Collection, supra note 104, Interviewee 8.

  119. [119]. See, e.g., id. Interviewee 9 (“[T]here may be an exception to this, but I think very, very rarely, maybe I could say almost never, would that be a cleanup of a provision that was not the subject of recent legislation.”).

  120. [120]. Id. Interviewee 10.

  121. [121]. There is both an Office of Legislative Counsel of the U.S. House of Representatives and an Office of Legislative Counsel of the United States Senate. See supra note 10.

  122. [122]. See generally Gluck & Bressman II, supra note 8 (describing the results of a survey of legislation drafters).

  123. [123]. See infra Section III.B.3.iii.

  124. [124]. Office of the Legislative Counsel: U.S. House of Representatives, supra note 10.

  125. [125]. Office of the Legislative Counsel: United States Senate, supra note 10.

  126. [126]. See, e.g., Office of the Legislative Counsel: U.S. House of Representatives, House Legislative Counsel’s Manual on Drafting Style (1995 ed.),

  127. [127]. Cf. Gluck & Bressman I, supra note 8, at 940 (finding that Members interact with abstract legislative policy rather than the granular text); Nourse & Schacter, supra note 8, at 585–86 (finding that legislative text is mostly drafted by staff); see also Livingston, supra note 9, at 836 (asserting that “[t]he fact is that members of Congress do not write tax statutes or committee reports, except in a very general sense”).

  128. [128]. See infra Section III.B.3.iv.

  129. [129]. Anonymous Interview Collection, supra note 104, Interviewee 10.

  130. [130]. Id. Interviewee 2.

  131. [131]. Id. Interviewee 26.

  132. [132]. Id.

  133. [133]. Cf. Gluck & Bressman II, supra note 8, at 740 (describing Legislative Counsel as main drafters of text and other staffers, including those on committees, as policy drafters).

  134. [134]. See, e.g., Anonymous Interview Collection, supra note 104, Interviewee 2.

  135. [135]. Id. Interviewee 8.

  136. [136]. The House Ways and Means Committee markup of the 1986 legislation took 26 days and occurred between September 1985 and December 3, 1985. See KPMG, Understanding the Tax Reform Process: FAQs (Updated for What’s Happened Lately) 6 (2017), https://home.

  137. [137]. Anonymous Interview Collection, supra note 104, Interviewee 9.

  138. [138]. Jasper L. Cummings, Jr. & Alan J.J. Swirski, Interview, Ward M. Hussey, ABA Sec. Tax’n Newsquarterly 4 (2013),

  139. [139]. John Buckley served in the House Office of Legislative Counsel from 1973–1993, in the House Ways and Means Committee as Minority Chief Tax Counsel from 1995–2006, as Majority Chief Tax Counsel from 2007–2010, and as Chief of Staff of the Joint Committee on Taxation from 1993–1994. See John Buckley, Joint Committee on Tax’n, (last visited Nov. 7, 2018). Interviewees active in the 1980s also repeatedly mentioned Buckley’s role in drafting tax legislation.

  140. [140]. It should be noted that this account is potentially at odds with the suggestion of some interviewees that drafting was more collaborative and less tightly controlled when passing major legislation. For discussion, see supra notes 134–37 and accompanying text.

  141. [141]. The shift to markup of actual legislative language did not occur until later. For discussion, see infra Section III.B.3.v.

  142. [142]. Again, it is unclear how this account intersects with the suggestion that more individuals participated in drafting of major legislation. For discussion, see supra notes 134–37 and accompanying text.

  143. [143]. Anonymous Interview Collection, supra note 104, Interviewee 8.

  144. [144]. In an interview he gave when he departed the Legislative Counsel’s office, Hussey stated that once he had gained some experience as Legislative Counsel, lobbyists knew better than to try to influence the drafting process. See Ellin Rosenthal, After 42 Years, Behind-the-Scene Taxwriter Departs, 43 Tax Notes 249, 250 (1989).

  145. [145]. While interviewees mentioned the expertise and importance of John Buckley and another drafter in the Senate, James Fransen, interviewees did not repeatedly refer to their control over the process in the same way they did with Hussey. For more background regarding James Fransen, see 160 Cong. Rec. S6584 (daily ed. Dec. 11, 2014) (tribute to James Fransen).

  146. [146]. See Shobe, supra note 8, at 847–49 (describing anecdotal evidence of the increased involvement of lobbyists in the drafting process, including offering draft bills); cf. Gluck & Bressman II, supra note 8, at 747 (mentioning Legislative Counsel being presented with text drafted by lobbyists, which Legislative Counsel may have limited leeway to change).

  147. [147]. Cf. Nourse & Schacter, supra note 8, at 611–12 (describing view among congressional staffers that lobbyists’ input adds value to the legislative process, including, for instance, lobbyists providing initial drafts of legislation); Shobe, supra note 8, at 847–48 (describing “qualified praise” some staffers had for lobbyists’ involvement in the drafting process).

  148. [148]. Anonymous Interview Collection, supra note 104, Interviewee 26.

  149. [149]. This shift in the Senate makes a certain amount of sense, given that Senate consideration follows after the legislation is passed by the House. For discussion, see supra Section II.A.

  150. [150]. Cf. Gluck & Bressman I, supra note 8, at 954 (referencing drafters’ description that they may use redundant language to please stakeholders). For discussion, see infra Section IV.A.

  151. [151]. A variety of anecdotal evidence suggests that this was certainly true of the 2017 tax reform. Despite the fact that both Congress and the White House were under Republican control, insiders reported that the Treasury was less engaged in the process than previously. See, e.g., Thomas D. Greenaway, Interview with Dana Trier, 37 ABA Tax Times, Spring 2018, at 6, 8 (providing remarks of Dana Trier, who served as the Deputy Assistant Treasury for Tax Policy in the U.S. Treasury Department during the 2017 tax reform including, for instance, Trier’s observation that “[c]ompared to earlier periods, Treasury’s on-the-ground role in this tax legislation was circumscribed; it was not large”).

  152. [152]. This squares with what we know about the subject from other sources. See I.R.S., Internal Revenue Service Highlights of 1986, at 16 (1986).

  153. [153]. See Staff of Joint Comm. on Taxation, 106th Cong., Study of Present-Law Taxpayer Confidentiality and Disclosure Provisions as Required by Section 3802 of the Internal Revenue Service Restructuring and Reform Act of 1998, at 100 (Comm. Print 2000) (mentioning “[t]he former IRS Office of Chief Counsel’s Legislation and Regulations Division”).

  154. [154]. Anonymous Interview Collection, supra note 104, Interviewee 3.

  155. [155]. Id. Interviewee 14.

  156. [156]. See generally, e.g., Richard F. Fenno, Jr., Congressmen in Committees (1973) (analyzing impact of congressional committees); Keith Krehbiel, Information and Legislative Organization (1991) (setting forth an informational theory of how legislatures work); Sinclair, supra note 20 (examining a turn to unorthodox procedures in the legislative process).

  157. [157]. See Gluck & Bressman I, supra note 8, at 911 (explaining that “[t]he foundational scholarship of federal legislation has, for the most part, been based on a generic and stylized account of statutory drafting”).

  158. [158]. See supra note 8.

  159. [159]. See Nourse & Schacter, supra note 8, at 584–90.

  160. [160]. See id.

  161. [161]. See id. at 588–90. Interestingly, in their study of the Senate Judiciary Committee, Nourse and Schacter found that Legislative Counsel had a non-mandatory role in drafting. Id. at 588. In their work, Gluck and Bressman focused much more on the role of Legislative Counsel, who play a more significant role in most other areas of legislation. See Gluck & Bressman II, supra note 8, at 739 (explaining the difference).

  162. [162]. Gluck & Bressman II, supra note 8, at 767.

  163. [163]. See Gluck & Bressman I, supra note 8, at 924–64.

  164. [164]. Shobe, supra note 8, at 854–60.

  165. [165]. See, e.g., Gluck & Bressman I, supra note 8, at 908 (“[B]y way of making the stakes clear, the findings have potential relevance for virtually all of the major interpretive debates, both at the canon-specific level and also more broadly at the theoretical level.”).

  166. [166]. See id. at 942–46.

  167. [167]. See Gluck & Bressman II, supra note 8, at 779–91.

  168. [168]. See Nourse & Schacter, supra note 8, at 616–21.

  169. [169]. See Shobe, supra note 8, at 813–15.

  170. [170]. See Burman et al., supra note 12, at 787–88 (underscoring the importance of confirmation studies).

  171. [171]. See supra Section III.B.1.i.

  172. [172]. See supra Section III.B.1, .3.

  173. [173]. See supra Sections III.B.3.i.–.ii.

  174. [174]. See, e.g., Abner S. Greene, The Missing Step of Textualism, 74 Fordham L. Rev. 1913, 1916 (2006) (describing textualism as a theory that attempts to promote faithful agency by “cordon[ing] off much material outside of statutory text”).

  175. [175]. To some extent, Shobe takes this view when he points to the increasing professionalization of statutory drafting staffers as a justification for textualism. See Shobe, supra note 8, at 853. Note that there are other defenses of textualism that we do not address. For instance, as John Manning has explained, “[t]extualists often rely on the formal claim that bicameralism andpresentmentmandatetextualism because the enacted text alone has survived the legislative process requirement ofArticle I, Section 7.” John F. Manning,Textualism and the Equity of the Statute, 101 Colum. L. Rev. 1, 71 (2001).

  176. [176]. Anonymous Interview Collection, supra note 104, Interviewee 4.

  177. [177]. See, e.g., I.R.C. § 1(h)(10) (2012).

  178. [178]. Anonymous Interview Collection, supra note 104, Interviewee 16.

  179. [179]. Id. Interviewee 10.

  180. [180]. To the extent that interviewees did consider how text would be parsed, they expressed the value they had placed on legislative history. See, e.g., id. Interviewee 3. For instance, in describing legislative history, as drafted by the JCT, one interviewee explained:

  181. [181]. Gluck & Bressman II, supra note 8, at 784–85; Nourse & Schacter, supra note 8, at 619–21.

  182. [182]. See, e.g., N.Y. State Bar Ass’n Tax Section, Tax Report No. 797, Report on the Proposed Partnership Anti-Abuse Rule 2 (1994) (making textualist case against partnership anti-abuse rule).

  183. [183]. See, e.g., Joseph Bankman,Tax Enforcement: Tax Shelters, the Cash Economy, and Compliance Costs, 31 Ohio N. U. L. Rev. 1, 3 (2005) (tying threat of tax shelters to literal interpretations of the Code); Noël B. Cunningham & James R. Repetti,Textualism and Tax Shelters, 24 Va. Tax Rev. 1, 3–4 (2004) (exploring how the rise of textualism facilitated the rise of tax shelters by legitimizing them); Tanina Rostain,Sheltering Lawyers: The Organized Tax Bar and the Tax Shelter Industry, 23 Yale J. on Reg. 77, 94 (2006) (explaining that tax textualism “gain[ed] currency” alongside the rise of tax shelters).

  184. [184]. Gluck & Bressman I, supra note 8, at 907.

  185. [185]. See id. at 930, 954.

  186. [186]. Id. at 934.

  187. [187]. See supra note 150 and accompanying text.

  188. [188]. See, e.g., United States v. Oberle, 136 F.3d 1414, 1423–24 (10th Cir. 1998) (applying the expressio unius canon to conclude that the inclusion of only one section indicates Congress’s intent not to include others).

  189. [189]. Gluck & Bressman I, supra note 8, at 933. Interestingly, their interviewees generally expressed that they reached this result based on “intuiti[on],” and thus did not need to formally employ a canon. Id.

  190. [190]. I.R.C. § 1221 (2012).

  191. [191]. See supra notes 115–16 and accompanying text.

  192. [192]. See generally, e.g., Jay A. Soled, The Sale of Donors’ Eggs: A Case Study of Why Congress Must Modify the Capital Asset Definition, 32 U.C. Davis L. Rev. 919 (1999) (using donated eggs as an example of how the default treatment as capital assets gets the wrong results for certain new classes of assets).

  193. [193]. See supra Section III.B.2.i (discussing statutory inertia).

  194. [194]. Gluck and Bressman acknowledge as much by emphasizing that their extensive study is merely an introduction into the importance of legislative practice for statutory interpretation. Gluck & Bressman II, supra note 8, at 800–01.

  195. [195]. See supra Section III.B.2.iii.

  196. [196]. See supra Section III.B.2.ii.

  197. [197]. See supra Section III.B.2.i. It is worth noting that there may also be a more psychological, and less conscious explanation, for the phenomenon of statutory inertia: a status quo bias. As researchers have explored, individuals may exhibit the bias as a result of “convenience, habit or inertia.” William Samuelson & Richard Zeckhauser, Status Quo Bias in Decision Making, 1 J. Risk
    & Uncertainty 7, 10 (1988). Our interviewees’ descriptions of how they make drafting decisions appeared consistent with status quo bias. Indeed, when asked why certain drafting choices were made, interviewees often suggested that drafting decisions were done to keep the existing style of the Code, or because new drafts would simply work off old models, even when those models were not a great fit for the problem at hand. See supra Section III.B.2.i.

  198. [198]. See, e.g., Bayless Manning, Hyperlexis and the Law of Conservation of Ambiguity: Thoughts on Section 385, 36 Tax Law. 9, 9–12 (1982); Mila Sohoni, The Idea of “Too Much Law, 80 Fordham L. Rev. 1585, 1601–22 (2012). Some have defended the ever-growing legal rules. Sohoni, supra, at 1622–31 (critically analyzing the claim that the proliferation of the law reflects democratic failure).

  199. [199]. As noted above, complexity of the written statute is only one dimension of tax complexity. See supra note 99 and accompanying text.

  200. [200]. I.R.C. § 163(h)(3)(F) (2012); H.R. Rep. No. 115-466, § 11043, at 34–35 (2017).

  201. [201]. Kathryn Vasel, Homeowners: Here’s What’s in the Tax Bill for You, CNN (Dec. 17, 2017, 12:17 PM),

  202. [202]. I.R.C. § 163.

  203. [203]. Id. § 163(h)(3)(F); H.R. Rep. No. 115-466, § 11043, at 34–35.

  204. [204]. One interviewee who suggested that drafting choices do not matter because “[n]o real people are going to read the statute” also offered that teaching this same statute to tax students is laborious because of how difficult the statute is to read, precisely because of its drafting. Anonymous Interview Collection, supra note 104, Interviewee 10. Some interviewees thereby indirectly confirm that there are costs from certain drafting choices, even as they professed that such choices did not matter. Id.

  205. [205]. See supra Section III.B.2.i.

  206. [206]. See, e.g., Emily Cauble, Superficial Proxies for Simplicity in Tax Law, 53 U. Rich. L. Rev. 329, 334–38 (2019) (discussing confusion between simplification efforts and deregulation); Steven A. Dean, Attractive Complexity: Tax Deregulation, the Check-the-Box Election, and the Future of Tax Simplification, 34 Hofstra L. Rev. 405, 419–20 (2005) (citing calls for tax reform that rely on Code complexity); Steven A. Dean, Tax Deregulation, 86 N.Y.U. L. Rev. 387, 410–13 (2011) (explaining the difference between tax simplification and tax deregulation).

  207. [207]. Cf. Mark D. Janis & Timothy R. Holbrook, Patent Law’s Audience, 97 Minn. L. Rev. 72, 86–89 (2012) (examining problems with a complex patent law, which is incomprehensible to the general public, being directed toward the general public).

  208. [208]. I.R.S., Form 1040 (2017),

  209. [209]. I.R.C. § 6662 (2012).

  210. [210]. Id. § 6664.

  211. [211]. See, e.g., Rodney P. Mock & Nancy E. Shurtz, The TurboTax Defense, 15 Fla. Tax Rev. 443, 532 (2014) (“[T]he Tax Court has missed almost every opportunity to clarify and expand upon exactly when a TurboTax defense is justifiable.”).

  212. [212]. See, e.g., Bedrosian v. United States, No. 15-5853, 2017 WL 1361535, at *2, *5 (E.D. Pa. Apr. 13, 2017) (appealing from government attempt to impose 50% willful penalty for failure to list bank account in FBAR; District held that government failed to prove that taxpayer’s action was willful). Taxpayers may not rely on IRS publications to assert the amount of tax owed, though such reliance might be considered in determining penalties. See Emily Cauble, Detrimental Reliance on IRS Guidance, 2015 Wis. L. Rev. 421, 438.

  213. [213]. See, e.g., I.R.C. § 6664(d)(4) (applying penalty to preparers for understatement due to unreasonable positions).

  214. [214]. See, e.g., United States v. Baker, 197 F.3d 211, 218 (6th Cir. 1999) (“Even those not versed in the law recognize the centuries-old maxim that ‘ignoranceof the law is no excuse.’”).

  215. [215]. See Blank & Osofsky, supra note 89, at 20434 (highlighting ways that IRS publications deviate from the underlying tax law); see also Cauble, supra note 212, at 438 (noting that taxpayers may not rely on I.R.S. publications to assert amount of tax owed).

  216. [216]. Stacy Cowley, Why the I.R.S. Fails to Crack the Small-Business Tax Nut, N.Y. Times (June 15, 2016), (noting that many small business owners do not fully understand the extent of their tax obligations, including, for instance, estimated reporting and the self-employment tax, in part as a result of “impenetrable” tax laws).

  217. [217]. See, e.g., Hearing on the Need for Comprehensive Tax Reform to Help American Companies Compete in the Global Market and Create Jobs for American Workers Before H. Comm. on Ways and Means, 112th Cong. (2011) (statement of Dirk J.J. Suringa, Partner, Covington & Burling) (practitioner testimony before Congress on the exceedingly complex foreign tax credit rules); Letter from Troy K. Lewis, Chair, AICPA Tax Exec. Comm., to John A. Koskinen, Comm’r, I.R.S., William J. Wilkins, Chief Counsel, I.R.S., Robert Stack, Deputy Assistant Sec’y, Int’l Tax Affairs, Dep’t of the Treasury, Danielle Rolfes, Int’l Tax Counsel, Dep’t of the Treasury (Aug. 15, 2016) (recommending simplified tax compliance procedures for Americans living abroad on the grounds that “it is unfair to subject these taxpayers to a complicated, time-consuming, and potentially expensive reporting requirement for information available to the IRS from alternative sources”).

  218. [218]. See, e.g., Leslie Book, The IRS’s EITC Compliance Regime: Taxpayers Caught in the Net,
    81 Or. L. Rev. 351, 352 (2002) (“The EITC is excessively complicated in its application and can have significant and sometimes unforeseen consequences on the lives of those who are the provision’s beneficiaries, largely the working poor, of whom great numbers are racial minorities and women.”).

  219. [219]. I.R.S., Form 1040, supra note 208.

  220. [220]. I.R.C. § 6694(a) (2012) (applying penalty to preparers for understatement due to unreasonable positions).

  221. [221]. See, e.g., id. § 6694(a)(1) (setting tax return preparer penalty equal to the greater of $1,000 or 50% of the income derived by the tax preparer from the return).

  222. [222]. See, e.g., Haynes v. United States, No. EP-16-CV-112-KC, 2017 WL 2895438, at *5–6 (W.D. Tex. June 15, 2017), on appeal, No. 17-50816 (5th Cir.) (holding that couple whose return was e-filed by their CPA did not have reasonable cause defense to late filing penalties for e-filing error due to software malfunction). For example, the preparer penalties for “understatements due to unreasonable positions” are limited only to the greater of $1,000 or the “income derived (or to be derived) by the tax return preparer,” and therefore may not adequately incentivize sophisticated intermediaries to protect against inadvertent errors they make in translating the tax law. I.R.C. § 6694(a).

  223. [223]. See, e.g., Stacey Vanek Smith, Stanford Professor Loses Political Battle to Simplify Tax Filing Process, Nat’l Pub. Radio (Mar. 29, 2017, 4:41 PM),
    54033/stanford-professor-loses-political-battle-to-simplify-tax-filing-process (“Filing taxes each year, it’s one of the most important interactions you’ve got with your government. And how does it make you feel? It makes you pissed. You can’t understand it. You’re anxious. You’re worried about screwing up.” (quoting Joseph Bankman’s description of the tax filing process)).

  224. [224]. See generally Joshua D. Blank & Daniel Z. Levin, When Is Tax Enforcement Publicized?,
    30 Va. Tax Rev. 1 (2010) (using empirical evidence to surmise that the IRS deliberately uses tax enforcement publicity to influence taxpayers’ perceptions of enforcement).

  225. [225]. See generally Dan M. Kahan, Ignorance of Law Is an Excuse—But Only for the Virtuous,
    96 Mich. L. Rev. 127 (1997) (examining how the maxim does or does not operate in certain contexts and focusing on tax as an area in which it does not operate, albeit in the criminal context).

  226. [226]. Antonin Scalia, The Rule of Law as a Law of Rules, 56 U. Chi. L. Rev. 1175, 1179 (1989).

  227. [227]. Id.

  228. [228]. Cf. Rachelle Y. Holmes, The Tax Lawyer as Gatekeeper, 49 U. Louisville L. Rev. 185, 190–201 (2010) (examining potential role for tax advisor in gatekeeper capacity).

  229. [229]. See IRS, Tax Gap: Tax Gap Estimates for Tax Years 2008–2010,
    newsroom/the-tax-gap (last visited Nov. 14, 2018) (estimating that the average annual tax gap, the amount of taxes owed but not paid, for years 2008–2010 was $458 billion).

  230. [230]. See Michael G. Allingham & Agnar Sandmo, Income Tax Evasion: A Theoretical Analysis,
    1 J. Pub. Econ. 323, 332–37 (1972) (introducing classic deterrence model of tax compliance). A vast literature has followed regarding tax penalties. See generally, e.g., Joshua D. Blank, Collateral Compliance, 162 U. Pa. L. Rev. 719 (2014) (examining ways non-tax consequences may bolster tax penalties); Mark P. Gergen, Uncertainty and Tax Enforcement: A Case for Moderate Fault-Based Penalties, 64 Tax L. Rev. 453 (2011) (justifying fault-based penalties on deterrence grounds).

  231. [231]. See, e.g., Letter Re: Support for the Tax Filing Simplification Act of 2016 (Apr. 18, 2016), (providing a letter of 54 tax professors supporting the Tax Filing Simplification Act, a bill that would among other things “allow[] taxpayers with simple returns to request a pro-forma return pre-populated with taxpayers’ data, completed tax calculations, and the amount of taxes or refund due”); Farhad Manjoo, Would You Let the I.R.S. Prepare Your Taxes?, N.Y. Times (Apr. 15, 2015), (discussing potential for government creation of new tax preparation method to reduce costs of tax filing, which would include I.R.S. pre-calculating tax liability for many taxpayers).

  232. [232]. Sarah B. Lawsky, A Logic for Statutes, 21 Fla. Tax Rev. 60, 77–79 (2017) [hereinafter Lawsky, A Logic for Statutes] (arguing that default logic may be beneficial for artificial intelligence); see also Sarah B. Lawsky, Formalizing the Code, 70 Tax L. Rev. 377, 394–99 (2017) (examining how the Code could be reformulated using logical reasoning).

  233. [233]. Lawsky, A Logic for Statutes, supra note 232, at 77–79.

  234. [234]. For discussion, see supra Section III.B.1.i.

  235. [235]. See supra text accompanying note 125.

  236. [236]. See supra note 125 and accompanying text.

  237. [237]. See, e.g., Ryan Calo, Artificial Intelligence Policy: A Primer and Roadmap, 51 U.C. Davis L. Rev. 399, 420–25 (2017).

  238. [238]. See, e.g., Jay A. Soled & Kathleen DeLaney Thomas, Regulating Tax Return Preparation,
    58 B.C. L. Rev. 152, 182
    83 (2017) (discussing strength of preparer lobby as employed against efforts to turn toward government preparation of tax returns).

  239. [239]. See generally, e.g., Louis Kaplow, A Model of the Optimal Complexity of Legal Rules,
    11 J.L. Econ. & Org. 150 (1995) (exploring the tradeoff between complexity and accuracy whereby complexity helps more accurately meet certain goals, including redistribution).

  240. [240]. See supra note 226 and accompanying text (giving a description of how the complex statutes can cause this type of stress). As just an anecdote, WebMD even has a webpage dedicated to how to cope with Tax Filing Season Stress. How to Cope With Tax-Time Stress, WebMD, (last visited Nov. 16, 2018).

  241. [241]. See, e.g., supra note 2 and accompanying text.

  242. [242]. See, e.g., supra notes 1–7 (characterizing legislation as sweeping and/or problematic).

  243. [243]. See, e.g., supra note 6 and accompanying text.

  244. [244]. See, e.g., supra note 6 and accompanying text; see also Stephen E. Shay, Tax Reform
    —Process Failures, Loopholes and Wealth Windfalls 1–2 (Nov. 21, 2017) (unpublished manuscript), (noting process failures in the enactment of the 2017 tax reform).

  245. [245]. See, e.g., supra notes 44–47 (critiquing the passage of the tax reform on both various process and substantive grounds).

  246. [246]. See, e.g., Senator Jon Tester (@SenatorTester), Twitter (Dec. 1, 2017, 6:06 PM), (“I was just handed a 479-page tax bill a few hours before the vote. One page literally has hand scribbled policy changes on it that can’t be read. This is Washington, D.C. at its worst. Montanans deserve so much better.”); Senator Bob Menendez (@SenatorMenendez), Twitter (Dec. 1, 2017, 5:30 PM), https://
    (“So much for regular order! Oh yeah sure, we’ll have plenty of time to read the final GOP tax bill—if you can make out the scribbled handwriting snuck into this massive, unsearchable PDF. #GOPTaxScam”). For more examples of tweets by Congress Members and others, see Thea Glassman, Tweets About Scribbles on the Tax Reform Bill Prove People are Disappointed, Elite Daily (Dec. 2, 2017),
    p/tweets-about-scribbles-on-the-tax-reform-bill-prove-people-are-disappointed-6764083. These types of concerns have also been echoed in subsequent legislation. See, e.g., Senator Rand Paul (@RandPaul), Twitter (Mar. 22, 2018, 10:02 AM),
    (“Well here it is, all 2,232 budget-busting pages. The House already started votes on it. The Senate is expected to soon. No one has read it. Congress is broken...”).

  247. [247]. Kathryn Watson, Democrats Outraged Over Senate GOP’s Last-Minute Tax Bill Reveal, CBS News (Dec. 1, 2017, 10:21 PM),

  248. [248]. For discussion, see supra Section IV.A.1.

  249. [249]. See supra text accompanying note 127.

  250. [250]. Rebecca M. Kysar & Linda Sugin, Opinion, The Built-In Instability of the G.O.P.’s Tax Bill, N.Y. Times (Dec. 19, 2017), (remarking on the instability of the legislation in part because it was passed without any Democratic votes); Kysar, supra note 20, at 62–63.

  251. [251]. See supra Section IV.A.

  252. [252]. See supra Section III.B.3.vii.

  253. [253]. See supra text accompanying note 129.

  254. [254]. Gluck et al., supra note 20, at 1832–34 (identifying participation of outsiders, such as interest groups, and congressional offices, such as the Congressional Budget Office (“CBO”), in the drafting of legislation as a type of unorthodox lawmaking and hypothesizing that unorthodox drafting could lead to unorthodox delegation of legislative power from Congress, depending on who is influencing the drafting); see also Hanah Metchis Volokh, A Read-the-Bill Rule For Congress, 76 Mo. L. Rev. 135, 140–48 (2011) (arguing that lawmakers are constitutionally required to exercise their judgment when voting on legislation, and that a bill should be passed to prevent Members of Congress from delegating away their responsibility to read bills).

  255. [255]. See supra Section III.B.3.ii.

  256. [256]. See supra Section III.B.3.iii.

  257. [257]. See supra note 20 and accompanying text.

  258. [258]. Sinclair, supra note 20 (describing these changes in the legislative process and the effects of such changes); see also Elizabeth Garrett,The Congressional Budget Process: Strengthening the Party-in-Government, 100 Colum. L. Rev. 702, 708–11 (2000) (describing decline in committee power and control that is both more diffuse and political).

  259. [259]. It is also theoretically possible that the erosion of power of congressional committees could create a decentralization of control at the congressional level that would allow Legislative Counsel to wield more influence. Our interviews did not suggest that this possibility had come to pass, but more research would be useful in any event regarding how professional drafters’ roles have shifted in response to unorthodox legislation.

  260. [260]. See generally Shobe, supra note 8 (arguing that increased professionalization of drafters over time, including that of Legislative Counsel, bolsters the case for textualism). But see Edward Kleinbard, Senators Picked Americans’ Pockets via Degraded Tax Policy Process, Hill (Dec. 4, 2017, 10:00 AM), that JCT and Treasury expertise were ignored in passing the 2017 tax bill).

  261. [261]. Gluck & Bressman II, supra note 8, at 741 (contrasting ordinary staff who have “greater direct accountability to the members” with “Legislative Counsels []who ... cannot be fired by individual members, and do not lose their jobs when control over Congress changes” (footnotes omitted)); see also id. at 794 (exploring potential benefits of “transferring more text-writing power to policymakers and away from non-policy-oriented drafting staff”).

  262. [262]. Id. at 741.

  263. [263]. The literature is extensive. See generally Gerald E. Frug,The Ideology of Bureaucracy in American Law, 97 Harv. L. Rev. 1276 (1984) (setting forth an early influential discussion of various theories).

  264. [264]. See Joel D. Aberbach, Keeping a Watchful Eye: The Politics of Congressional Oversight 14, 34–37 (1990) (discussing the expansion of congressional oversight generally); Alexander M. Bickel, The Least Dangerous Branch: The Supreme Court at the Bar of Politics 16–23 (2d ed. 1986) (describing judicial review as counter-majoritarian, which helped spur a majoritarian paradigm of democratic legitimacy and therefore the accountability approach to legitimate administrative agencies); Lisa Schultz Bressman, Judicial Review of Agency Inaction: An Arbitrariness Approach, 79 N.Y.U. L. Rev. 1657, 1677 (2004) (describing dominance of the “presidential control model” as a means of legitimating administrative agencies); Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2331–39 (2001) (making foundational accountability argument for presidential control of administrative agencies).

  265. [265]. See supra Section III.B.3.iii. Another key characteristic of the 2017 tax reform was the speed of enactment, which reduced the opportunity of many constituencies to carefully consider the legislation. Focusing on this factor suggests that, had the 1986 Act been pushed through Congress as quickly as the 2017 tax reform was, it is possible that no centralization of expert control would have been able to overcome the time pressures. There may be some truth to this. However, time pressure is not new. Counsels active during the 1980s as well as more recently active counsels both mentioned time pressure as a constant force and a factor leading to drafting imperfections. Similarly, counsels active in the 1980s also described the last-minute drafting frenzy that accompanies major legislation. Moreover, computerization, email, and data storage are all factors that make the process more efficient and may mean that the same sort of deliberation can happen during a much shorter time today.

  266. [266]. See supra Section III.B.3.ii.

  267. [267]. See, e.g., Peter L. Strauss,Speech, From Expertise to Politics: The Transformation of American Rulemaking, 31 Wake Forest L. Rev. 745, 755–56 (1996) (discussing earlier view of agencies as apolitical experts as well as the more recent turn away from this model toward political view).

  268. [268]. See, e.g., Richard B. Stewart, The Reformation of American Administrative Law,88 Harv. L. Rev. 1667, 1675(1975) (describing early transmission belt model in the context of the administrative state).

  269. [269]. See, e.g., Stephen K. Cooper, Brady Says to Expect Passthrough Changes, Technical Corrections, 2017 Tax Notes Today 237-1.

  270. [270]. See supra Section III.B.2.ii.

  271. [271]. See supra Section III.B.2.ii.

  272. [272]. See, e.g., Stephen K. Cooper, Democrats Unlikely to Support Tax Technical Corrections in 2018, 2017 Tax Notes Today 242-4 (describing how Democrats are not in the mood to support technical corrections in 2018); Jonathan Bernstein, Why the Tax Bill’s Errors Won’t Be Fixed, Bloomberg (Dec. 12, 2017, 4:00 AM), (explaining why it will be very difficult to get technical corrections passed).

  273. [273]. See George K. Yin, Crafting Structural Tax Legislation in a Highly Polarized Congress (Va. Pub. Law & Legal Theory Research Paper No. 2017-58; Va. Law and Econs. Research. Paper No. 2017-25, 2018) (manuscript 5–6), (documenting recent polarization in Congress in general, and in the House Ways and Means Committee and Senate Finance Committee in particular).

  274. [274]. See Shu-Yi Oei & Leigh Osofsky, Post-Enactment Politicking: Agency Rulemaking After Hasty Legislation, (working paper) (on file with author); cf. Christopher J. Walker, Legislating in the Shadows, 165 U. Pa. L. Rev. 1377, 1382–87 (2017) (exploring agency involvement in statutory drafting).


Professor of Law & Dean’s Distinguished Scholar, Boston College Law School.


Professor of Law, University of North Carolina School of Law.

We thank the participants of the Boston College Law School Faculty Workshop, the University of Virginia School of Law Tax Roundtable, the Boston College-Tulane Tax Roundtable, the 2017 National Tax Association Annual Conference on Taxation, the 2018 Law & Society Annual Meeting, and the 2017 Southeastern Association of Law Schools Annual Conference for helpful comments. We are particularly grateful to Lily Batchelder, William Blatt, Andrew Dawson, Kathleen DeLaney Thomas, Heather Field, Michael Froomkin, Abbe Gluck, Rebecca Kysar, Roberta Mann, Ruth Mason, Patricia McCoy, Shannon Weeks McCormack, Alex Raskolnikov, James Repetti, Diane Ring, Daniel Shaviro, Natalya Shnitser, Ariel Stevenson, David Weisbach, George Yin, and Larry Zelenak for feedback on drafts.