This article originally appeared in the May 31, 2021, issue of Tax Notes Federal.]
Victor Thuronyi was lead tax counsel for the IMF.
In this article, Thuronyi argues that the IRC should be redrafted, and he proposes specific style changes, noting that the scope of the rewrite should be modest.
Copyright 2021 Victor Thuronyi.
All rights reserved.
Major tax changes are again being contemplated by Congress. In this context, is it possible to make the IRC easier to read and understand? While the tax code is notoriously complex, is this complexity the result of rules that are too complex or a style that is too impenetrable? As I explain below, the vast bulk of the problem is the complexity of the rules. Simplifying those rules should be a priority. That said, the drafting of the IRC can also be improved. A project to rewrite the code should be undertaken, but its scope should be limited so that it can be accomplished in a reasonable period and without too much controversy or disruption.
Experience of Other Countries
In deciding whether to undertake a comprehensive redraft, the United States can look to the experience of other English-speaking countries with very complex tax laws that have been rewritten, namely the United Kingdom, Australia, and New Zealand.1 In these countries, the tax laws before they were rewritten were poorly organized and encrusted with outdated and confusing terminology. The rewrites were painful and produced laws that are not necessarily very user-friendly, although they are probably better than before. By contrast, although the IRC is not often cited as a model of user-friendliness, it is in fact quite well organized, and its terminology is consistent, with the result that any rewrite need not involve reorganization of code sections or major changes. Accordingly, a rewrite of the IRC would be less involved than the redrafting of the tax laws in the aforementioned countries.
Organization of the IRC
The IRC is organized into subtitles, chapters, and subchapters (several subchapters are further organized into parts and subparts), and that organization has been maintained over the years as sections have been added. Unlike the United Kingdom, which was plagued by legislation that was not properly incorporated on enactment, new U.S. legislation is routinely incorporated into the proper section of the code (located in the appropriate subchapter — and part or subpart where applicable) on enactment.2 When the section numbers did not allow a particular provision to fit, a new section number with a capital letter was created (for example, section 453A) so that the new provision went into an appropriate place. The result is that a tax code rewrite in the United States can be much less jarring than in the United Kingdom, Australia, or New Zealand.
A rewrite of the IRC can keep existing section numbers. There might be a few rearrangements of subsections or paragraphs within existing sections. As a result, a practitioner who is familiar with the existing code will be able to look at the rewritten code with no need to learn where provisions can be found. From a practitioner’s point of view, a redraft would hardly be noticeable. Upon going to IRC sections with which a practitioner is familiar, they might find the language a bit more user-friendly, but there should be very few surprises or a need to learn new law. The amount of relearning that a practitioner will have to do should be much less than for a typical major tax reform bill.
Use of ‘Shall’
What kind of changes might a redrafting entail? In terms of the number of words that would have to be changed, the most extensive change that might be considered is in the use of “shall.” According to the House Office of the Legislative Counsel (HOLC): “The term ‘shall’ means that an action is required; the term ‘may’ means that it is permitted but not required.”3 This usage is followed in the code, but the code also inconsistently uses the present indicative and “shall” in the same sense as the present indicative (that is, not requiring any action). This makes the code sound like legalese and de-emphasizes the main meaning of “shall” when that word is used to require an action. If the present indicative were used (when appropriate) instead, the IRC would become a bit easier to read, and it would have slightly fewer words.4
The use of the word “shall” throughout the code is often in the passive voice. When the passive voice is used, it is not clear which actor is in mind — is it the taxpayer, the IRS, or perhaps the courts in applying the statute? Or maybe there is really no actor at all when, as in most cases, what is involved is largely a concept (is a particular deduction allowed or is a particular item included in income? does a particular section apply?). Using the present indicative would be more straightforward. Some fixes are easy; for example, changing “this section shall apply” to “this section applies.” In other cases, switching from “shall” to the present indicative might prompt other small changes. Consider this example from section 453A (and my proposed rewrite, shown in italics and strike-through):
(d) Pledges, etc., of installment obligations
(1) In general
For purposes of section 453, if any indebtedness (hereinafter in this subsection referred to as “secured indebtedness”) is secured by an installment obligation to which this section applies, the net proceeds of the secured indebtedness (up to the limit described in paragraph (2)) are shall be treated as a payment received on such installment obligation as of the later of —
(A) the time the indebtedness becomes secured indebtedness, or
(B) the time the proceeds of such indebtedness are received by the taxpayer.
(2) Limitation based on total contract price
The limit referred to in the amount treated as received under paragraph (1) is by reason of any secured indebtedness shall not exceed the excess (if any) of —
What this rewrite makes clear is that there is no mysterious actor deciding how much of the amount to treat as received under paragraph (1). Instead, paragraph (1) defines how much is treated as received: namely, the amount in paragraph (1) but no more than the limit defined in paragraph (2). In other words, it is the smaller of these two amounts, and that would be an alternative way of putting it. The exercise of eliminating the word “shall” led me to think about what the expression “the amount treated as received under paragraph (1) shall not exceed” was all about. What became clear is that this amount is a limit. This is a clearer way of stating things and allows the interaction between paragraph (1) and paragraph (2) to be made more transparent. What this exercise shows is that if the drafter is forced to eliminate “shall,” this exercise can lead to ideas for a clearer statement of the rule. Eliminating “shall” therefore serves as an occasion for minor edits that improve readability.
In the example above, the difference between the existing and rewritten provisions is minor. The revised version is a bit more clear and uses fewer words. This kind of change should not unsettle practitioners. The meaning is the same, only a few words are changed, and the existing numbering of the statute is not changed. It is just slightly easier to read. I used the words “up to” instead of “not in excess of” in paragraph (1) because they are more user-friendly, but the latter wording would also be an option.
Another example (section 23(b)(1)): “The aggregate amount of qualified adoption expenses which may be taken into account under subsection (a) for all taxable years with respect to the adoption of a child by the taxpayer shall not exceed $10,000.” I would rewrite this as: “The aggregate amount of qualified adoption expenses taken into account under subsection (a) for all taxable years with respect to the adoption of a child by the taxpayer is limited to $10,000.” Same meaning, three words shorter, and I would argue a bit clearer.
Removing the inappropriate use of “shall” would involve changing a lot of words and would lead to a significant reduction in the number of words in the code, given that in most cases one would be replacing expressions like “shall apply” with “applies” and so forth. It would be thousands of words in total, even if this is a small percentage of the number of words in the IRC. At the same time, some clarity and user-friendliness would be gained.
The word “shall” would not be changed when it directs someone to do something. For example, “The Secretary shall prescribe regulations” would be left unchanged. This approach will also render the IRC less ambiguous. If “shall” is used only in the sense that someone is required to do something, then its use will provide a signal to the reader that that is what is going on. I am skeptical that artificial intelligence can be used to read the tax code,5 but eliminating the word “shall” so that its usage is unambiguous should help AI (and especially humans) read the code more easily.
The IRC would be more user-friendly and clearer if “shall” were changed to the present indicative when there is no mandatory action. The current code is inconsistent in its use of “shall.” The code often does use the present indicative. Definitions typically say “means” rather than “shall mean.” Section 1 says, “There is hereby imposed . . . a tax” (not “a tax shall be imposed”). Section 49(a)(1)(B) says “this paragraph applies” instead of “this paragraph shall apply.” Sections 101, 102, 103, 104, and 106 say “gross income does not include” instead of “shall not include.” But section 119 says “there shall be excluded from gross income.” Section 121 says “gross income shall not include.” Section 130 says “shall not be included in gross income.” Why not change all these to “gross income does not include”?
Another set of changes would make the code gender-neutral. Readers may be surprised to hear that there are actually relatively few uses of the pronoun “his” in the code — a search identified 432 code sections. Drafters of the IRC have already been using a gender-neutral style to a large extent. In most cases, drafters have avoided use of the pronoun by using a noun instead, and it should be relatively straightforward to do so in these 432 sections as well.6 In a few instances, it might be necessary to change the word order of the sentence to make this work. With this approach, not only are gender-specific terms avoided, but the code becomes easier to read because the reader does not have to hunt for the pronoun’s antecedent.
The word “husband” appears even more infrequently — in only about 30 code sections — and usually together with the word “wife.” These terms can be replaced with terms such as “married couple” or “spouse,” so that they are gender-neutral. Again, the code is not consistent, and the word “spouse” is already used frequently, appearing in 178 code sections.
Consistency in Use of ‘Includes’
Another expression for which better and more consistent usage can be applied is “includes.” According to the HOLC drafting guide, “the phrase ‘includes, but is not limited to’ is redundant. In fact, using it in some places out of an abundance of caution could cause a limitation to be read into places where it is not used.”7 A search of the IRC shows 16 instances of “but not limited to” and six instances of “but is not limited to,” in connection with “includes.” Cleaning these up will make for better usage and will also make the code a few words shorter.
Use of the Singular
“In general, provisions should be drafted in the singular to avoid the ambiguity that plural constructions can create.”8 The IRC follows this guideline pretty well. For example, the code usually refers to a taxpayer rather than “taxpayers,” but there are some exceptions when “taxpayers” is used but there doesn’t seem to be a good reason for doing so (for example, section 6081(a)). I am not sure in how many other instances the code uses the plural when the singular would be appropriate; this would be one thing that should be checked as part of a rewrite.
Another situation in which a rewriting would be warranted involves code provisions that are simply impenetrable. Reading the tax code should not be a puzzle. Yes, sometimes the rules are complex, and so reading the code is not easy, but that is not a drafting problem: It is a problem of the rules being too complex.
A good example of a statutory provision that is a puzzle is the last paragraph of section 509(a): “For purposes of paragraph (3), an organization described in paragraph (2) shall be deemed to include an organization described in section 501(c)(4), (5), or (6) which would be described in paragraph (2) if it were an organization described in section 501(c)(3).” This sentence is not very long, and that is what ends up being the problem. It is possible to restate the provision so that it is easier to follow, but that ends up requiring more words. Here is a rewrite:
For purposes of determining whether an organization is described in paragraph (3), another organization is considered to be described in paragraph (2) if such other organization:
Maybe to specialists in the tax-exempt area the meaning of the current text is obvious, but the rewrite strikes me as being a bit more clear, even if still a bit of a puzzle. Notice that the assumed audience is still made up of experts, and that the remaining opacity flows from the complexity of the rule, not its drafting.
There are likely several other provisions in the IRC that contain mistakes9 or language that is ambiguous or difficult to understand. Fixing these is likely to be a significant task, but not a huge one. The way to proceed would be for whatever body is doing the rewrite to solicit input from interested parties to identify provisions that they think are hard to understand or ambiguous, or contain mistakes, preferably explaining why.
Formulating the General Rule
IRC sections are typically organized by stating the general rule in subsection (a).10 The subsections that follow may define elements of the general rule, may provide special rules, and may provide exceptions. Definitions for terms used in the section are often at the end, and at the very end there may be cross-references to other related code sections. Shu-Yi Oei and Leigh Osofsky identify the problem of exception-swallows-the-rule drafting situations in which a section states a general rule but the exceptions to the rule are so extensive that they effectively make the general rule an exception.11 I am not sure that this is a significant problem in the IRC. While the general rule can be stated in different ways, it is often best to state it in a way that accurately summarizes the entire section. When there are exceptions to the general rule, the general rule can usually be formulated in a way that explicitly or implicitly identifies them. Implicit identification would be to say something like: “except as otherwise provided in this section.” But that statement, while accurate, does not help the reader understand the rule. A clearer way might be to say something like: “except in the cases described in subsections (b), (c), and (d).” Another technique is to use terms like “qualified taxpayer” and “qualified expenditures,” which gives the reader a clue that not all taxpayers or expenditures are affected by the rule and they need to read the definitions of these terms. In summary, it is usually best to encapsulate the rule of the section in subsection (a) so that the reader has a roadmap for navigating the rest of the section and understanding the rule.
Sometimes the general rule is articulated in a way that is too complex. For example, the notoriously lengthy section 509(a) could be condensed to read as follows:
For purposes of this title, the term “private foundation” means a domestic or foreign organization described in section 501(c)(3) other than —
an organization described in section 170(b)(1)(A) other than in clauses (vii) and (viii);
a publicly supported organization; or
a supporting organization.
This would give the reader a roadmap. Obviously, this would require lengthy definitions of “publicly supported organization” and “supporting organization,” which could be added at the end of the section. Besides making section 509 easier to follow, this approach would have the advantage of creating defined terms that could then be used instead of rather abstract references like “an organization described in section 509(a)(3)(B)(iii).”
Sections Degraded by Amendment
As mentioned earlier, in most cases in which amendments in the form of new code sections have been made, the sections have been added in appropriate places, thereby preserving the logical structure of the IRC. In a few cases, however, amendments have been tacked on to the end of existing code sections, with the result that the section has lost its logical structure and therefore is more difficult to parse than it has to be.
An example of this is section 501. In its original 1954 version, this section consisted of subsections (a) through (d), and subsection (e) with a cross-reference. Many of the subsections that were later added to section 501 do belong there, because they have to do with the definition of organizations that qualify for exempt status. However, subsections (h) and (p) should be in separate code sections because they concern different subjects (lobbying and suspension of tax-exempt status of terrorist organizations, respectively), although the decision that they fit within the scope of section 501 is not clearly wrong. It is more that putting them into separate code sections would make section 501 easier to follow. Even if these two subsections were moved, section 501 would remain complex, but the complexity is not the result of poor drafting; it is because Congress has enacted detailed rules under which different kinds of organizations qualify for tax exemption.
Another example (discussed by Oei and Osofsky12) is section 163. This section has been amended many times. The general rule that today exists in section 163(a) has never been amended. In 1954, section 163 consisted only of two subsections, namely the general rule in subsection (a) and a special rule for unstated interest on installment sales of personal property in subsection (b). Each time section 163 was amended with a new limitation, this was tacked on at the end, with the result that the logical order of the section has been degraded. No legislative drafter in their right mind would draft section 163 the way it appears today if starting from scratch. For one thing, subsection (b), which was a special rule for installment sales, should not be located immediately following the general rule.
In a sense, section 163 is a microcosm of the degradation of organization that motivated other countries to rewrite their tax laws. The good news is that in the United States — because the 1954 code had a good logical organization when it was enacted,13 which was maintained over time — the problem does not apply to the code generally but just to a few sections that have received many significant amendments. Section 163 is therefore a candidate for cleaning up — to improve its organization — and maybe there are a few other sections, but my sense is that there are not many sections in which the benefits of substantially rewriting and reorganizing them would warrant the disruptiveness that the rewriting would cause.
When it is helpful to read two code sections together to understand them, the technique adopted in the IRC is to include cross-references at the end of the relevant sections. This is not always done.14 An IRC rewrite could usefully include identifying places in which cross-sections would be helpful and adding them. This is also a situation in which public input would be helpful. Because these types of cross-references do not have legal effect,15 it is pretty easy to add them. In other words, the drafter does not have to agonize over it for fear of making a mistake. If two sections are related, a cross-reference can be added to make the code more user-friendly. A search reveals 214 cross-references at the end of code sections. This number should most likely be substantially increased, thereby making the code more user-friendly. In a few cases, cross-references could be added within a section. For example, section 501(c)(5) could include a cross-reference to section 501(g), which defines a term used in subsection (c)(5) (yes, the better approach is to revise all of subchapter F, which has become badly organized, but pending a revision, inserting a few cross-references to make it easier to use would not hurt).
The expression “for purposes of this section” is used 686 times in the IRC. It may be possible to get rid of this language in several places. In the case of defined terms, it may be that the restriction is not needed if the term is not used elsewhere in the code in a different sense.
For example, section 664(g)(4) starts: “For purposes of this section, the term ‘qualified employer securities’ means.” This could be shortened to: “‘Qualified employer securities’ means.” Why? First, the words “the term” are not needed. If something is in quotes, and followed by “means,” it is pretty clear that one is defining a term. (This revision could be made throughout the IRC.) Second, “for purposes of this section” is not needed. It turns out that “qualified employer securities” is used only in section 664 and in a few other sections that make a cross-reference to section 664. One option would be to include “qualified employer securities” as a defined term for the whole code, but even if this is not done, it is not necessary to restrict the definition of this term to section 664 because the term is not used anywhere else in the code with a different meaning. If this were done, a general provision could be included in section 7806 that the definitions of terms made in this title apply only for purposes of this title.
The advantage of the current approach of explicit cross-referencing for some defined terms that are used infrequently, rather than including them in section 7701, is that section 7701 might otherwise become overloaded.
The rewritten income tax law of New Zealand includes at the end of each section a list of defined terms that are used in that section. This adds to the length of the document and may not be so helpful for words that are used frequently (for example, “taxpayer”).
My general conclusion on definitions is that the general approach used in the IRC makes sense. It includes in section 7701 those terms that are used frequently and keeps definitions of terms used less frequently (or only in one code section) in the particular section in which the term is used. What might be done is to (1) review whether each defined term in a section that is defined “for purposes of this section” only needs to be so limited, and (2) make sure that, for those defined terms in section 7701 that are used only in a few places in the code, the sections using those terms include a cross-reference to the section 7701 definition, so that the reader is alerted to look there.
The terms defined in section 7701(a) should be rearranged in alphabetical order so that defined terms are easier to find.
Moving a bit further, one can identify particular provisions as deadwood. These are rules that no longer likely apply to anyone, or perhaps apply so rarely and with so little policy justification that they can be deleted. Of course, if the provisions still apply to some taxpayers and confer a benefit, those taxpayers may disagree, so there is a policy or political judgment involved. One approach for provisions that do still apply to some taxpayers would be to repeal them with a prospective effective date to get political buy-in.
Just to identify one at random: section 512(b)(15) concerning unrelated business taxable income and an exception for a religious order. It seems that this provision was written to benefit one taxpayer (I don’t have any inside knowledge — I just stumbled across it). One question is whether this taxpayer still exists and makes use of this provision. If so, maybe it could be repealed with a delayed effective date. If the provision is no longer used, then its repeal should not be controversial. Provisions like this are not without cost from the point of view of the complexity of the code because they contribute to the tangle of rules that people need to wade through. Deadwood provisions can be hard to identify, but one advantage of a redrafting exercise is that those engaged in it could check a box identifying suspected deadwood provisions in the code sections that they are rewriting. If a provision is truly deadwood, in that it applies to no taxpayer, then perhaps including it in a rewrite exercise would not slow it down much and could increase the political attractiveness of the exercise because it would involve reducing the length of the code. However, for those provisions that might not fully be deadwood in that they still have some application, it may be sufficient to put them on a list for future attention.
Going beyond deadwood, we come to the substantive simplification of the code. This is the most difficult lift politically, but it also provides the greatest benefit. The best option may be for those involved in redrafting to check a box identifying where simplification can be undertaken. Even if there is the political will to engage in simplification, this is likely to be a multiyear process, but a starting point would be a checklist of items that are candidates. Perhaps some simplification of rules can be combined with a redrafting exercise, but even if most of the simplification is put off for future work, at least it can be identified.
As an example of simplification possibilities, if capital gains and ordinary income rates are unified, a multitude of code provisions can be simplified because there are many rules policing the distinction between capital gains and ordinary income to prevent taxpayers from arranging transactions to exploit the distinction. This simplification opportunity was available in 1986, but those involved in that year’s tax reform found it too daunting to get rid of these ancillary rules and therefore did not attempt the exercise. Such an exercise is likely to take a substantial amount of work, and so it could not be done in short order.
Recent decades have seen a growing interest in what might be called “plain English” drafting, in which the drafting is done in modern language, avoiding legalese as much as possible, using shorter sentences, and making similar style choices to make the law as readable as possible.16 Redrafting the IRC would be in line with those initiatives, and is long overdue. Drafting of the IRC should not assume that ordinary taxpayers will be reading the law, although a few might. Redrafting to improve readability is needed even if one assumes that the main audience is tax professionals because, even for most tax professionals, the IRC has become too difficult to read.
Scope of Redrafting Project
While it is possible to envision a project to redraft the IRC that goes so far as to include some simplification, the key to success would be to keep the scope of the project manageable. There will be opportunity for rule simplification in the future. Moreover, in terms of drafting style, restraint in making changes is likely to make the project more manageable. Keeping the existing code sections and keeping changes to a minimum is likely to be most successful both in terms of the acceptability of the project and getting it done in a reasonable time frame.
The redrafting project should include:
removing the word “shall” except when a person is directed to do something (and performing associated light editing);
making the code gender-neutral (eliminating masculine pronouns and the terms “husband” and “wife”);
checking the use of the plural and changing to the singular when appropriate;
deleting “not limited to” as used in conjunction with “includes”;
redrafting provisions that are impenetrable and correcting mistakes;
reorganizing and rewriting the handful of code sections that have been frequently amended and have lost their structure;
deleting “for purposes of this section” when not needed;
adding helpful cross-references at the end of sections when they are missing; and
rearranging the defined terms in section 7701 in alphabetical order and, because quotes suffice, removing the words “the term” from definitions.
This work plan is modest. It would not change any existing code section numbers, and it would substantially rewrite only a few sections. It would make the code a bit easier to read and reduce its length. It would also identify provisions that could be eliminated in the future by way of repealing deadwood and simplifying. The modesty of this approach is justified by a concern not to take on too much, lest the project take far too long.
As Oei and Osofsky point out, there is not a lot of political weight behind efforts to make the IRC more readable. This is one reason that a project with a modest scope might stand a better chance of getting political backing than a more ambitious alternative would. A tax code rewrite project can only be undertaken, however, if the senators and representatives on the taxwriting committees believe it is worth the effort and likely to succeed in the current political environment. Bipartisan buy-in would seem to be a requirement. Given its modest scope, it would be important for the rewrite not to involve substantive changes to tax policy, but including some technical corrections could provide some sweetener to make the project more attractive.
The effort would also require those in charge of drafting legislation, namely the House and Senate legislative counsels, to get on board. One possible objection is that style changes, in particular the use of “shall,” should arguably be made throughout the U.S. Code rather than just in title 26. The rebuttal to this is that the current IRC is inconsistent on this point: The present indicative is already used in several sections,17 and there is no rule mandating the use of “shall” when the present indicative can alternatively be used. If the style of title 26 is cleaned up, the same can later be done for other titles, and a rewrite of title 26 can be considered as a pilot.
1 The U.K. Tax Law Rewrite Project was set up in 1996, with the goal of making the U.K.’s direct tax legislation clearer and easier to use. See Ipsos Mori, “Review of Rewritten Income Tax Legislation,” Social Research Institute (2011). New Zealand’s tax law rewrite project resulted in the Income Tax Act 2007. The rewrite projects in these countries have taken 15 to 20 years.
2 We probably have Ward Hussey to thank for much of this, including the original organization of the IRC. See infra note 13. A recent interview with someone who knew Hussey well, and who himself started work at Treasury in 1961, confirms that Hussey was most likely substantially involved in any aspect of the 1954 code that was significant. According to H. Res. 97, 101st Cong., 1st Sess. (Mar. 1, 1989) (“Ward M. Hussey has been the principal draftsman of all the federal income tax laws beginning before the enactment of the Internal Revenue Code of 1954.”).
3 HOLC, “Guide to Legislative Drafting.”
4 Use of the present indicative instead of “shall” unless someone is being directed to do something is also supported by the Uniform Law Commission, Drafting Rules and Style Manual (2021).
5 See generally Sarah Lawsky, “Formalizing the Code,” 70 Tax L. Rev. 377 (2017).
7 HOLC, supra note 3.
9 A couple of examples of drafting mistakes, together with proposed solutions, are discussed in Lawsky, supra note 5, at 392-394.
11 Oei and Osofsky, “Constituencies and Control in Statutory Drafting: Interviews With Government Counsels,” 104 Iowa L. Rev. 1291 (2019).
13 The Internal Revenue Code of 1939 was a good start, and one can see the embryonic organization of the 1954 code there, but its income tax provisions had less than 160 sections. The 1954 code had a more detailed and logical organizational structure, which has withstood the test of time. It also helped that one of the principal authors of the 1954 code, Hussey, who started working in 1946 for the House Legislative Counsel, remained in that office until the late 1980s, and was able to keep the organization of the code together.
14 See Oei and Osofsky, supra note 11.
16 E.g., International Monetary Fund Legal Department, Plain English Tax Law Drafting (2008); Uniform Law Commission, supra note 4.
17 See discussion above.