Harvard Law Review Harvard Law Review Harvard Law Review

Harvard Law Review Forum

Response to Judge Kavanaugh’s Review of Judging Statutes

Response to Brett M. Kavanaugh, Fixing Statutory Interpretation

With much respect, I read Judge Kavanaugh’s review of Judging Statutes.1×1. Brett M. Kavanaugh, Fixing Statutory Interpretation, 129 Harv. L. Rev. 2118 (2016) (reviewing Robert A. Katzmann, Judging Statutes (2014)). I could not have hoped for a more thoughtful examination of the subject. Judge Kavanaugh, a rightfully highly regarded jurist and colleague, offers a measured critique that furthers discussion of how to approach the interpretive enterprise. And his fresh ideas about the use of canons open up new lines of thinking.

We may differ on how to approach legislative history, but we share the perspective that courts are not to substitute their preferences for that of the elected branches. When the language of a statute is plain and clear, our work is generally straightforward. Where Judge Kavanaugh and I differ is how to proceed when, as is often the case, the statute defies easy comprehension, when something less than clarity reigns. Because the methodology of interpretation — whether a judge should stay fixed solely on the text or should look to other sources of understanding as well — can affect the outcome of a case, what is at stake is whether the law will be construed in a manner most consistent with what Congress meant. In cases involving unclear statutes, I review the authoritative (that is, reliable) materials that circulated in Congress before the legislation’s passage. I look there to find answers, to understand the purpose behind a particular statute. By examining the history of a statute before Congress enacted it, I often find clues to guide my interpretation of an opaquely written piece of legislation.

In his disagreement with my approach to legislative history, Judge Kavanaugh argues that because committee reports are not voted on by the whole Congress, they do not have the force of law and should not be relevant to judicial interpretation. The statute’s text is the law, he states, and if Congress wants courts to look at the legislative history, they could “vote on it when voting on the statute.”2×2. Id. at 2123. Judge Kavanaugh writes: “[I]f courts tell Congress that voting on those reports is necessary, or at least necessary if Congress wants those reports to be considered authoritative by courts, then Congress could readily decide whether and when to vote on those reports.”3×3. Id. at 2124. He cites the superb scholar Professor John Manning, who Judge Kavanaugh believes “rightly emphasizes ‘Congress’s continued failure to put legislative history to a vote three decades into a textualist campaign that has put legislative history on uncertain footing in the Supreme Court.’”4×4. Id. at 2124 n.21 (quoting John F. Manning, Why Does Congress Vote on Some Texts But Not Others?, 51 Tulsa L. Rev. 559, 562 (2016) (reviewing Katzmann, supra note 1)).

I agree with Judge Kavanaugh that legislative history is not the law, but I differ with him as to whether courts should use it: I believe legislative history can shed light on what the law means, and, in fact, Congress expects that its legislative history will be respected by courts. The Constitution, after all, largely vests Congress with the authority to determine its own procedures for the introduction, consideration, and approval of bills, and that includes how the legislative branch treats legislative history. Therefore, my response to the view that Congress should have to enact legislative history into law before judges may consider it is that it is not for the courts to impose rules on the legislative branch as to how to do its business.

Judge Kavanaugh recognizes that lawmaking can be complicated, that “it is much harder to enact statutes than it is to block them.”5×5. Id. at 2133. Adding committee report commentary or analysis to existing text would increase that burden in many if not most cases. If such a substantial alteration is to be made in the dynamic of lawmaking, Congress rather than the courts should make it. And while the difficulty of enacting legislation is in part a result of constitutional design, the same Constitution also gives Congress the discretion to decide how to conduct its own work — including how to organize its legislative agenda through committees that then produce information relied on by members to help them understand the meaning of the text on which they will vote.

I wrote Judging Statutes because, in the debate about how to interpret statutes, scant consideration has been given to how Congress actually functions, how legislators signal meaning, and what they expect of those interpreting their laws. I found that Congress intends that its work should be understood through its established institutional processes and practices. An important component of those institutional processes and practices, and one that is essential to understanding statutes enacted by Congress, is legislative history — for example, the committee and conference committee reports that accompany legislative text. Each chamber has established its own rules and practices governing lawmaking, creating, in Professor James Brudney’s apt phrase, “a resultant hierarchy of internal communications.”6×6. James J. Brudney, Intentionalism’s Revival, 44 San Diego L. Rev. 1001, 1010 (2007). Those rules and procedures give particular legislators, such as committee chairs, floor managers, and party leaders, substantial control over the process by which legislation is enacted. Communications from such members as to the meaning of proposed statutes can provide reliable signals to the whole chamber as it considers a bill. And members and their staffs, who well understand that maintaining credibility with colleagues is essential to effective legislating, have every incentive to represent accurately the meaning of proposed statutes to colleagues in materials that comprise the legislative history.

Notably, agencies appreciate and are responsive to Congress’s perspective that such materials are essential to construing statutes. Judge Kavanaugh recognizes that legislative history, specifically committee reports, provides useful information to the executive branch, and serves “an important and legitimate purpose for the executive and independent agencies that must implement the statutes and exercise any discretion granted them by statute.”7×7. Kavanaugh, supra note 1, at 2126. At the same time that he acknowledges that legislative history may be appropriately used by agencies, however, Judge Kavanaugh challenges its use by courts.8×8. Id. at 2123–24. Congress has made no such distinction, and I believe that we as courts are without license to make one.

Congress, as the master of its own decisionmaking processes, is under no obligation to change its behavior to conform to modern textualist theory, whatever the merits of particular prescriptions. It is a bipartisan institutional perspective within Congress that courts should consider reliable legislative history and that failing to do so impugns Congress’s workways. In Judging Statutes, I quote Senator Grassley, currently chair of the Judiciary Committee, who has long defended courts’ use of legislative history as an interpretive aid. In 1986, at the confirmation hearing of Antonin Scalia for Supreme Court Justice, Senator Grassley expressed concern about the then–D.C. Circuit judge’s opposition to legislative history: “[A]s one who has served in Congress for 12 years, legislative history is very important to those of us here who want further detailed expression of that legislative intent.”9×9. Nomination of Judge Antonin Scalia, to Be Associate Justice of the Supreme Court of the United States: Hearing Before the S. Comm. on the Judiciary, 99th Cong. 65–66 (1986) (statement of Sen. Charles E. Grassley). He again voiced his concerns nearly two decades later during the confirmation hearing of John G. Roberts, Jr., noting:

Justice Scalia is of the opinion that most expressions of legislative history, like Committee reports or statements by the Senators on the floor, or in the House, are not entitled to great weight because they are unreliable indicators of legislative intent. Presumably, Justice Scalia believes that if the members don’t actually write a report or don’t actually vote on a report, then there is no need to defer to this expression of congressional intent.

Now, obviously, I have great regard for Justice Scalia, his intellect and legal reasoning. But, of course, as I told you in my office, I don’t really agree with his position.10×10. Confirmation Hearing on the Nomination of John G. Roberts, Jr. to Be Chief Justice of the United States: Hearing Before the S. Comm. on the Judiciary, 109th Cong. 318–19 (2005) (statement of Sen. Charles E. Grassley). A few months later, Senator Grassley would question then-Judge Samuel A. Alito, Jr., on his views of legislative history. See Confirmation Hearing on the Nomination of Judge Samuel A. Alito, Jr. to Be an Associate Justice of the Supreme Court of the United States: Hearing Before the S. Comm. on the Judiciary, 109th Cong. 503 (2006) (statement of Sen. Charles E. Grassley).

That courts should consider legislative history is further supported by a study conducted by Professors Abbe R. Gluck and Lisa Schultz Bressman.11×11. Abbe R. Gluck & Lisa Schultz Bressman, Statutory Interpretation from the InsideAn Empirical Study of Congressional Drafting, Delegation, and the Canons: Part I, 65 Stan. L. Rev. 901 (2013). The empirical findings of their study — that, in Congress, legislative history is the most important drafting and interpretive instrument apart from text;12×12. Id. at 907. that, indeed, there are members and staffs who rely more on legislative history than on the text of bills in considering their votes;13×13. Id. at 968. that the committee system is not an improper delegation of authority;14×14. Id. at 969. that dictionaries are not often used;15×15. Id. at 907. and that many canons are of limited utility16×16. Id. at 949. — all suggest quite powerfully this conclusion: When courts fail to respect the dynamics of the legislative process, they undermine their capacity to reach sound judicial decisions as to legislative meaning.

Yes, the passing of legislation can be chaotic or even gruesome, a process famously likened to making sausages. But, drawing on my political science background, I know it is worth the effort to appreciate how Congress produces its work product. With an understanding of the legislative process, a judge will be better prepared to delve into the materials behind a piece of legislation and focus on those materials that are reliable. The documents comprising legislative history may illuminate what a statute’s particular term or phrase means, what the legislators were trying to do. Of course, I recognize that not all legislative history is created equal, and judges must weigh the quality of extratextual evidence, much like they weigh the reliability of dictionary definitions, semantic canons, and any other evidence. When I write about “authoritative” legislative history — for example, conference committee reports — I do not suggest that legislative history is the law, nor that it is binding in the same way that law is. Rather, I view it as evidence of what the legislators were trying to do when they passed a statute, something that should not be ignored.

If judges exclude reliable legislative history, as pure textualists would have it, they will eliminate a useful source of information about an ambiguous law’s likely meaning — how the legislation’s congressional proponents wanted the statute to work and what problems they sought to address. When a statute is ambiguous, barring consideration of legislative history leaves a judge with words that could be interpreted in a variety of ways without contextual guidance. Ignoring such guidance increases the probability that a judge will construe a law at odds with legislative meaning, and potentially more in line with the judge’s own intuitions and policy preferences. As Professor Peter Strauss forcefully stated: “Declaring independence of these signals repudiates the faithful servant ideal — no faithful servant would insist upon ascribing his own meaning to his mistress’s words in the face of clear indications of how she understood them.”17×17. Peter Strauss, 65 J. Legal Educ. 443, 447 (2015) (reviewing Katzmann, supra note 1). Of course, in some cases, the legislative history may turn out not to be particularly helpful, but in others, it may be. Why should courts, a priori, exclude consideration of relevant legislative history materials that could be useful in understanding legislative meaning? We judges can use all the help we can get! Although textualists have helpfully shown some of the pitfalls of legislative history, they have not made the case for its categorical exclusion. The legislative record behind a bill is in truth its foundation and deserves thoughtful examination. Judicial respect for Congress and its lawmaking prerogatives in our constitutional scheme requires no less.

Judge Kavanaugh questions what I mean by “generally” when I write that “when a statute is unambiguous, resorting to legislative history is generally not necessary; in that circumstance, the inquiry ordinarily ends.”18×18. Katzmann, supra note 1, at 48; see also Kavanaugh, supra note 1, at 2128. He wonders whether the use of “generally” suggests a return to Church of the Holy Trinity v. United States.19×19. 143 U.S. 457 (1892); see also Kavanaugh, supra note 1, at 2128. It does not. My intent was not to revive that old debate, but to highlight the challenges of the interpretive task and approaches to addressing that challenge. When I construe a statute, I focus on its words. As an element of that analysis, even when the specific words under consideration appear unambiguous in isolation, I examine them in terms of the statutory structure of which the words are a part. If that inquiry does not raise questions as to the meaning of those apparently unambiguous words, then my inquiry ordinarily ends, and I do not need to consider the legislative history. But where that inquiry suggests that the meaning of the words under review is not so obvious as it appeared in isolation, I look at other possibly helpful aids, including reliable legislative history.

Judge Kavanaugh points to a case I discuss, Murphy v. Arlington Central School District Board of Education,20×20. 402 F.3d 332 (2d Cir. 2005), rev’d and remanded, 548 U.S. 291 (2006). as one where I support using a committee report to trump what he calls unambiguous language.21×21. Kavanaugh, supra note 1, at 2132. In that case, the statute said that a court may award a winning party “reasonable attorneys’ fees as part of the costs.”22×22. Murphy, 402 F.3d at 335–36 (quoting 20 U.S.C. § 1415(i)(3)(B) (2012)). The parents of a disabled child who succeeded in administrative proceedings sought compensation for the costs of expert fees incurred as part of the litigation.23×23. Id. at 336. Are those expert fees compensable under the statute? Read literally, expert fees might not appear to be covered by the provision for attorneys’ fees. But Congress, it becomes clear on examining the larger framework of the Individuals with Disabilities Education Act,24×24. 20 U.S.C. §§ 1400–1482. enacted the law to make it easier for children with disabilities and their parents to pursue claims, including by allowing parents to be accompanied during IDEA proceedings by “counsel and by individuals with special knowledge or training.”25×25. Id. § 1415(h)(1). In light of this statutory structure, it was at least a plausible question whether the fees paid to experts in the service of a winning litigation were meant to be included as “reasonable attorneys’ fees as part of the costs.” Accordingly, it made sense for the panel to deepen its inquiry to include the legislative history. In that case, the Conference Report, a particularly reliable document, which was submitted to both chambers and accepted by both, contained language relevant to our analysis. In particular, the Conference Report, consistent with the statutory framework, explicitly stated: “The conferees intend that the term ‘attorneys’ fees as part of the costs’ include reasonable expenses and fees of expert witnesses and the reasonable costs of any test or evaluation which is found to be necessary for the preparation of the . . . case.”26×26. Murphy, 402 F.3d at 336–37 (alterations in original) (quoting W. Va. Univ. Hosps., Inc. v. Casey, 499 U.S. 83, 91 n.5 (1991)). As a court that views our task as being faithful to the legislative scheme, we concluded that “reasonable attorneys’ fees as part of the costs” included expert fees.27×27. Id. at 339.

Sometimes, of course, traditional legislative history may not be conclusive or not accessible and hence not useful in resolving an interpretive puzzle. The Supreme Court’s recent decision in King v. Burwell28×28. 135 S. Ct. 2480 (2015). provides an example. In that case, the Supreme Court was faced with a statute whose broader structure suggested a reasonable interpretive question about the meaning of a seemingly unambiguous phrase: “Exchange established by the State.”29×29. Id. at 2487 (quoting 26 U.S.C. § 36B(b)–(c) (2012)). The majority acknowledged that, standing alone, this key phrase could mean that certain tax credits were available only when individuals obtained insurance through state-established exchanges, and not through those established by the federal government.30×30. Id. at 2490. But, said the Court, much like what my panel determined in Murphy, “oftentimes the ‘meaning — or ambiguity — of certain words or phrases may only become evident when placed in context’” of the whole statutory scheme.31×31. Id. at 2489 (quoting FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 132 (2000)). The majority, proving the truth of this general principle, then observed that several other provisions of the Act would make little sense if tax credits were not available for insurance bought on Exchanges created by the federal government.32×32. See id. at 2490. The Court was thus faced with an ambiguity.

Although the majority was not averse, as a general matter, to using legislative history, it concluded that it was not helpful in this case because the legislation was crafted outside the traditional legislative process.33×33. See id. at 2492. Instead, the majority looked more broadly to Congress’s underlying legislative purpose, with the Chief Justice declaring: “Congress passed the Affordable Care Act to improve health insurance markets, not to destroy them. If at all possible, we must interpret the Act in a way that is consistent with the former, and avoids the latter.”34×34. Id. at 2496. With that in mind, the Court ultimately found that the key phrase “Exchange established by the State” could refer to all exchanges, both state and federal, for purposes of the tax credits provided by the Act.35×35. Id. In King, the Court showed that it was willing to depart from what would otherwise be the most natural reading of a phrase when context and the structure of a statute require it. As this case and others show, the interpretive exercise can be complicated, and is not usefully reduced to characterizations of the Holy Trinity rubric as simplistically substituting a law’s spirit for clear text.

The Supreme Court’s debate about statutory interpretation on display in King is not limited to high-profile, politically charged cases. A prime exemplar, decided a few months before King, is Yates v. United States,36×36. 135 S. Ct. 1074 (2015). Justice Ginsburg announced the judgment of the Court in an opinion joined by Chief Justice Roberts, Justice Breyer, and Justice Sotomayor. Justice Alito concurred in the judgment, focusing on the statute’s list of nouns, list of verbs, and its title. Id. at 1089 (Alito, J., concurring in the judgment). Justice Kagan dissented, joined by Justice Scalia, Justice Kennedy, and Justice Thomas. in which the Court concluded that a statute designed to protect investors and restore trust in financial markets in the aftermath of Enron’s collapse could not be applied against a commercial fisherman accused of throwing undersized fish overboard to avoid prosecution. In the plurality opinion, Justice Ginsburg commented that,

Whether a statutory term is unambiguous . . . does not turn solely on dictionary definitions of its component words. Rather, “[t]he plainness or ambiguity of statutory language is determined [not only] by reference to the language itself, [but as well by] the specific context in which that language is used, and the broader context of the statute as a whole.”37×37. Id. at 1081–82 (plurality opinion) (second, third, and fourth alterations in original) (quoting Robinson v. Shell Oil Co., 519 U.S. 337, 341 (1997)).

Considering the phrase at issue — “tangible object” — in the context of the whole statute, the Court found that dictionary definitions of the words “tangible” and “object” were not dispositive of the statutory meaning.38×38. Id. at 1082. The Court then employed the full arsenal of statutory interpretation tools — text, dictionary, statutory context, canons, and legislative history — to reach its conclusion.39×39. See id. at 1081–87. In dissent, Justice Kagan noted her agreement with the plurality that “context matters in interpreting statutes,”40×40. Id. at 1092 (Kagan, J., dissenting). even though she ultimately reached a different result after employing the same interpretive tools, including legislative history.41×41. See id. at 1091–94.

As these cases show, statutes come in all shapes and sizes, varying in design and substance. A judge’s work takes place not on the lofty plane of grand, unified theory, but on the ground of commonsense inquiry. Statutory interpretation, in the felicitous words of Professors William N. Eskridge, Jr., and Philip P. Frickey, is an exercise in “practical reasoning.”42×42. William N. Eskridge, Jr. & Philip P. Frickey, Statutory Interpretation as Practical Reasoning, 42 Stan. L. Rev. 321, 322 (1990). Legislative history is but one of several tools at a judge’s disposal, including text, statutory structure, word usage in other relevant statutes, common law usages, agency interpretations, dictionary definitions, technical and scientific usages, lay usages, canons, common practices, and purpose. Using these tools makes it more likely that courts will interpret statutes consistent with Congress’s design. As Chief Justice Roberts wrote in King: “[I]n every case we must respect the role of the Legislature, and take care not to undo what it has done. A fair reading of legislation demands a fair understanding of the legislative plan.”43×43. 135 S. Ct. 2480, 2496 (2015).

Differing with strict textualists, a clear majority of the Supreme Court continues to recognize that courts should make use of all of the tools that might shed light on its task to understand a statute’s meaning, including authoritative legislative history. While the late Justice Scalia rightly called attention to inappropriate uses of legislative history, his suggestion that courts should exclude legislative history altogether when interpreting statutory text is not the prevailing view among his colleagues. I respectfully believe that it goes too far to claim that the textualist campaign “has put legislative history on uncertain footing in the Supreme Court.”44×44. Kavanaugh, supra note 1, at 2124 n.21 (quoting Manning, supra note 4, at 562).

In his review, Judge Kavanaugh also begins what I believe will be an important discussion of the role of canons in statutory interpretation, a subject that has generated renewed interest following the publication of Justice Scalia and Professor Bryan Garner’s significant treatise Reading Law.45×45. Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts (2012). Like Judge Kavanaugh, I often turn to canons when interpreting statutes, recognizing that canons are sometimes useful and sometimes of limited utility.46×46. See, e.g., United States v. Lockhart, 749 F.3d 148, 152–54 (2d Cir. 2014), aff’d, 136 S. Ct. 958 (2016). Frequently, different canons point in different directions and judges are without a key to determine which should prevail. In Congress, as we know from the Gluck-Bressman study, legislative staff members are familiar with some canons and not with others. Accordingly, the canons that judges often use are of uncertain value in determining what legislators had in mind in the drafting process.

Notwithstanding the limitations of canons in understanding Congress’s own work product, what Judge Kavanaugh very insightfully seeks to explore is how canons can be better employed as interpretive rules of the road, trying to settle as many of them in advance as we can, freed from an inquiry into “ambiguity.” In his graceful formulation:

A number of interpretive canons of statutory interpretation depend on an initial evaluation of whether the statutory text is clear or ambiguous. But because it is so difficult to make those clarity versus ambiguity determinations in a coherent, evenhanded way, courts should reduce the number of canons of construction that depend on an initial finding of ambiguity.”47×47. Kavanaugh, supra note 1, at 2121.

Instead, Judge Kavanaugh continues, courts should seek “the best reading of the statute by interpreting the words of the statute, taking account of the context of the whole statute, and applying the agreed-upon semantic canons.”48×48. Id. Then courts should “apply any applicable plain statement rules, and ensure that the interpretation is not absurd.”49×49. Id. at 2163. To support his prescription, Judge Kavanaugh thoughtfully reflects on the constitutional avoidance canon, legislative history canon, Chevron deference, ambiguity-dependent presumptions, plain statement rules, “mistake” and absurdity canons, and what he terms “problematic” semantic canons that should be shed (for example, the ejusdem generis canon, the anti-redundancy canon, and the consistent usage canon).50×50. See id. at 2144–62.

Judge Kavanaugh’s examination — which, he notes, is only a preliminary one51×51. Id. at 2160. — is a substantial conversation starter. Take, for example, the Chevron doctrine. Judge Kavanaugh encourages us to ask when deference is due to an agency interpretation. He proposes distinguishing between statutes using broad and open-ended terms, where courts should generally give agencies the discretion to make policy judgments, and statutes with specific terms or phrases, where courts should determine whether the agency’s interpretation is the best reading of the statutory text.52×52. Id. at 2153–54. Judge Kavanaugh raises an issue well worth considering — whether under the current Chevron framework, there is undue deference to an agency’s interpretation of a statute where the court is in as good a position as an agency to make a competent determination.

Recognizing that Judge Kavanaugh’s canon discussion is the beginning of a journey, I raise a few questions that might be helpful along the way. Judge Kavanaugh suggests that judges seek “the best reading of the statute by interpreting the words of the statute, taking account of the context of the whole statute.”53×53. Id. at 2121. But isn’t that essentially what judges already do, whatever tools they use in that effort? Inevitably, won’t a court be confronted with the “clarity versus ambiguity” issue when a statutory phrase can reasonably be read in multiple different ways? When the words do not necessarily dictate a result, won’t the “clarity versus ambiguity” question quickly come to the fore as part of a good faith exercise of judgment? In that case, won’t there be legitimate differences about what is the “best reading” under the proposed framework whenever there is ambiguity, where there are reasonable disagreements about the meaning of words? It may be that changing the standards of deference of the Chevron framework could change how cases are argued, causing counsel to focus their energies other than on whether the words are ambiguous or not. But certainly there will always be many cases where the statute can be viewed as ambiguous under any reading, where the statute is reasonably susceptible to more than one interpretation.

If the objective of Judge Kavanaugh’s proposal is to reduce the likelihood of subjective judgment, then wouldn’t eliminating reliable legislative history, to the extent that it provides guidance, increase the possibility that the decision will not reflect what Congress had in mind? Wouldn’t resorting to reliable legislative history act as a restraint on a judge’s substituting preferences for that of the legislature and thus further the “best reading” of the statute that Judge Kavanaugh supports? I am concerned that undue reliance on various semantic and linguistic sources to the exclusion of reliable legislative history or background inquiries may not reflect the “best reading” of the statute: it does not reflect appropriate deference to Congress as the institution that produced the text, to the legislature’s underlying plan. Some additional questions regarding Judge Kavanaugh’s proposal to “fix” the canons of interpretation: Who decides which canons are to be sidelined? Courts? Congress? How is that determination to be made? How do we decide between dueling canons?54×54. For a recent examination, see Anita S. Krishnakumar, Dueling Canons, 65 Duke L.J. 909 (2016).

That Judge Kavanaugh’s preliminary inquiry has generated these questions is a testament to its value. I have no doubt that in Judge Kavanaugh’s skillful hands, thinking about the canons will be usefully advanced. Just as he was kind enough to write that he learned from me, I am learning from him. I look forward to his future contributions to this continuing conversation he has so ably sparked.


* Robert A. Katzmann, Chief Judge of the U.S. Court of Appeals for the Second Circuit, is the author of Judging Statutes (2014). For their comments, I thank Elizabeth Graber Bentley, James Brudney, Celia Choy, Robert Friedman, Donald Goodson, Brian Richardson, and Sonia Steinway.