Jacqueline Halbig v. Sylvia Mathews Burwell

Opinion for the Court filed by Circuit Judge GRIFFITH.

Concurring opinion filed by Senior Circuit Judge RANDOLPH.

Dissenting opinion filed by Senior Circuit Judge EDWARDS.

GRIFFITH, Circuit Judge:

Section 36B of the Internal Revenue Code, enacted as part of the Patient Protection and Affordable Care Act (ACA or the Act), makes tax credits available as a form of subsidy to individuals who purchase health insurance through marketplaces — known as “American Health Benefit Exchanges,” or “Exchanges” for short — that are “established by the State under section 1311” of the Act. 26 U.S.C. § 36B(c)(2)(A)(i). On its face, this provision authorizes tax credits for insurance purchased on an Exchange established by one of the fifty states or the District of Columbia. See 42 U.S.C. § 18024(d). But the Internal Revenue Service has interpreted section 36B broadly to authorize the subsidy also for insurance purchased on an Exchange established by the federal government under section 1321 of the Act. See 26 C.F.R. § 1.36B-2(a)(i) (hereinafter “IRS Rule”).

Appellants are a group of individuals and employers residing in states that did not establish Exchanges. For reasons we explain more fully below, the IRS’s interpretation of section 36B makes them subject to certain penalties under the ACA that they would rather not face. Believing that the IRS’s interpretation is inconsistent with section 36B, appellants challenge the regulation under the Administrative Procedure Act (APA), alleging that it is not “in accordance with law.” 5 U.S.C. § 706(2)(A).

On cross-motions for summary judgment, the district court rejected that challenge, granting the government’s motion and denying appéllants’. See Halbig v. Sebelius, No. 13 Civ. 623(PLF), — F.Supp.3d -, 2014 WL 129023 (D.D.C. Jan. 15, 2014). After resolving several threshold issues related to its jurisdiction, the district court held that the ACA’s text, structure, purpose, and legislative history make “clear that Congress intended to make premium tax credits available on both state-run and federally-facilitated Exchanges.” Id. at -, 2014 WL 129023 at *18. Furthermore, the court held that even if the ACA were ambiguous, the IRS’s regulation would represent a permissible construction entitled to deference under Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984).

' Appellants timely appealed the district court’s orders, and we have jurisdiction under 28 U.S.C. § 1291. Our review of the orders is de novo, and “[o]n an independent review of the record, we will uphold an agency action unless we find it to be ‘arbitrary, capricious, an abuse of discretion, or otherwise not in accordance *394with law.’ ” Holland v. Nat’l Mining Ass’n, 309 F.3d 808, 814 (D.C.Cir.2002) (quoting 5 U.S.C. § 706(2)(A)). Because we conclude that the ACA unambiguously restricts the section 36B subsidy to insurance purchased on Exchanges “established by the State,” we reverse the district court and vacate the IRS’s regulation.

I

Congress enacted the Patient Protection and Affordable Care Act in 2010 “to increase the number of Americans covered by health insurance and decrease the cost of health care.” Nat’l Fed’n of Indep. Bus. v. Sebelius (NFIB), — U.S. -, 132 S.Ct. 2566, 2580, 183 L.Ed.2d 450 (2012). The ACA pursues these goals through a complex network of interconnected policies focused primarily on helping individuals who do not receive coverage through an employer or government program to purchase affordable insurance directly. Central to this effort are the Exchanges. 42 U.S.C. § 18031(b)(1). Exchanges are “governmental agenc[ies] or nonprofit entities]” that serve as both gatekeepers and gateways to health insurance coverage. See id. § 18031(d)(1). Among their many functions as gatekeepers, Exchanges determine which health plans satisfy federal and state standards, and they operate websites that allow individuals and employers to enroll in those that do. See id. § 18031(b)(1), (d)(i)(d)(4). Section 1311 of the ACA delegates primary responsibility for establishing Exchanges to individual states. See id. § 18031(b)(1) (providing that “[e]ach State shall, not later than January 1, 2014, establish an American Health Benefit Exchange (referred to in this title as an ‘Exchange’) for the State”). However, because Congress cannot require states to implement federal laws, see Printz v. United States, 521 U.S. 898, 904-05, 935, 117 S.Ct. 2365, 138 L.Ed.2d 914 (1997), if a state refuses or is unable to set up an Exchange, section 1321 provides that the federal government, through the Secretary of Health and Human Services (HHS), “shall ... establish and operate such Exchange within the State.” 42 U.S.C. § 18041(c)(1). As of today, only fourteen states and the District of Columbia have established Exchanges. The federal government has established Exchanges in the remaining thirty-six states, in some cases with state assistance but in most cases not. See Richard Cauchi, State Actions To Address Health Insurance Exchanges, Nat’l Conference of State Legislatures (May 9, 2014), http:// www.ncsl.org/research/health/state-aetions-to-implement-the-health-benefit.aspx.

Under section 36B, Exchanges also serve as the gateway to the refundable tax credits through which the ACA subsidizes health insurance. See 26 U.S.C. § 36B(a). Generally speaking, section 36B authorizes credits for “applicable taxpayer[s],” id., defined as those with household incomes between 100 and 400 percent of the federal poverty line, id. § 36B(c)(l)(A). But section 36B’s formula for calculating the credit works further limits on who may receive the subsidy. According to that formula, the credit is to equal the sum of the “premium assistance amounts” for each “coverage month.” Id. § 36B(b)(Z). The “premium assistance amount” is based on the cost of a “qualified health plan ... enrolled in through an Exchange established by the State under [section] 1311 of the [ACA].” Id. § 36B(b)(2); see also 42 U.S.C. §§ 18021(a)(1), 18031(c)(1) (establishing requirements for “qualified health plans”). Likewise, a “coverage month” is a month for which, “as of the first day of such month the taxpayer ... is covered by a qualified health plan ... that was enrolled in through an Exchange established by the State under section 1311 of the [ACA].” 26 *395U.S.C. § 36B(c)(2)(A)(i). In other words, the tax credit is available only to subsidize the purchase of insurance on an “Exchange established by the State under section 1311 of the [ACA].”

But, in a regulation promulgated on May 23, 2012, the IRS interpreted section 36B to allow credits for insurance purchased on either a state- or federally-established Exchange. Specifically, the regulation provided that a taxpayer may receive a tax credit if he “is enrolled in one or more qualified health plans through an Exchange,” 26 C.F.R. § 1.36B-2(a)(l), which the IRS defined as “an Exchange serving the individual market for qualified individuals ..., regardless of whether the Exchange is established and operated by a State (including a regional Exchange or subsidiary Exchange) or by HHS.” 45 C.F.R. § 155.20 (emphasis added); see 26 C.F.R. § 1.36B-l(k) (incorporating the definition in 45 C.F.R. § 155.20 by reference). In promulgating this broader rule, the IRS acknowledged that “[c]ommentators disagreed on whether the language in section 36B(b)(2)(A) limits the availability of the premium tax credit only to taxpayers who enroll in qualified health plans on State Exchanges,” but asserted without elaboration that “[t]he statutory language of section 36B and other provisions of the [ACA],” as well as “the relevant legislative history,” supported its view. Health Insurance Premium Tax Credit, 77 Fed.Reg. 30,377, 30,378 (May 23, 2012).

This broader interpretation has major ramifications. By making credits more widely available, the IRS Rule gives the individual and employer mandates' — key provisions of the ACA — broader effect than they would have if credits were limited to state-established Exchanges. The individual mandate requires individuals to maintain “minimum essential coverage” and, in general, enforces that requirement with a penalty. See 26 U.S.C. § 5000A(a)-(b). The penalty does not apply, however, to individuals for whom the annual cost of the cheapest available coverage, less any tax credits, would exceed eight percent of their projected household income. See id. § 5000A(e)(l)(A)-(B). By some estimates, credits will determine on which side of the eight-percent threshold millions of individuals fall. See Br. of Economic Scholars in Support of Appellees 18. Thus, by making tax credits available in the 36 states with federal Exchanges, the IRS Rule significantly increases the number of people who must purchase health insurance or face a penalty.

The IRS Rule affects the employer mandate in a similar way. Like the individual mandate, the employer mandate uses the threat of penalties to induce large employers — defined as those with at least 50 employees, see 26 U.S.C. § 4980H(c)(2)(A)— to provide their full-time employees with health insurance. See generally id. § 4980H(a). Specifically, the ACA penalizes any large employer who fails to offer its full-time employees suitable coverage if one or more of those employees “enroll[s] ... in a qualified health plan with respect to which an applicable tax credit ... is allowed or paid with respect to the employee.” Id. § 4980H(a)(2); see also id. § 4980H(b) (linking another penalty on employers to employees’ receipt of tax credits). Thus, even more than with the individual mandate, the employer mandate’s penalties hinge on the availability of credits. If credits were unavailable in states with federal Exchanges, employers there would face no penalties for failing to offer coverage. The IRS Rule has the opposite effect: by allowing credits in such states, it exposes employers there to penalties and thereby gives the employer mandate broader reach.

*396II

Before we can turn to the merits of the parties’ dispute, we must first address the government’s argument that all appellants lack standing and that, even if they have standing, the APA does not provide them with a cause of action to challenge the IRS Rule. Because we find that appellant David Klemencic has standing and a cause of action under the APA, we do not reach the issue of our jurisdiction over the remaining appellants’ claims. See Mountain States Legal Found. v. Glickman, 92 F.3d 1228, 1232 (D.C.Cir.1996) (explaining that as long as one plaintiff has standing for a claim, “we need not consider the standing of the other plaintiffs to raise that claim”).

A

The “ ‘irreducible constitutional minimum’ ” a plaintiff must show to establish standing is (1) an injury in fact (2) fairly traceable to the alleged conduct of the defendant (3) that is likely to be redressed by the relief the plaintiff seeks. Sprint Commc’ns Co. v. APCC Servs., Inc., 554 U.S. 269, 273-74, 128 S.Ct. 2531, 171 L.Ed.2d 424 (2008) (quoting Lujan v. Defenders of Wildlife, 504 U.S. 555, 560-61, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992)). The district court determined that at least one of the appellants, David Klemencic, has standing. Klemencic resides in West Virginia, a state that did not establish its own Exchange, and expects to earn approximately $20,000 this year.1 He avers that he does not wish to purchase health insurance and that, but for federal credits, he would be exempt from the individual mandate because the unsubsidized cost of coverage would exceed eight percent of his income. The availability of credits on West Virginia’s federal Exchange therefore confronts Klemencic with a choice he’d rather avoid: purchase health insurance at a subsidized cost of less than $21 per year or pay a somewhat greater tax penalty.

The government primarily questions whether Klemencic has suffered an injury in fact. An injury in fact is “a concrete and particularized invasion of a legally protected interest.” Sprint Commc’ns Co., 554 U.S. at 273, 128 S.Ct. 2531 (internal quotation marks omitted). The government characterizes Klemencic’s injury as purely ideological and hence neither concrete nor particularized. But, although Klemencic admits to being at least partly motivated by opposition to “government handouts,” he has established that, by making subsidies available in West Virginia, the IRS Rule will have quantifiable economic consequences particular to him. See Clapper v. Amnesty Int’l USA, — U.S. -, 133 S.Ct. 1138, 1147, 185 L.Ed.2d 264 (2013) (explaining that a “threatened injury” that is “certainly impending” may “constitute injury in fact” (emphasis and internal quotation marks omitted)). Those consequences may be small, but even an “ ‘identifiable trifle’ ” of harm may establish standing. Chevron Natural Gas v. FERC, 199 Fed.Appx. 2, 4 (D.C.Cir.2006) (quoting United States v. Students Challenging Regulatory Agency Procedures, 412 U.S. 669, 689 n. 14, 93 S.Ct. 2405, 37 L.Ed.2d 254 (1973)); see Bob Jones Univ. v. United States, 461 U.S. 574, 581-82, 103 S.Ct. 2017, 76 L.Ed.2d 157 (1983) (noting that Bob Jones University *397sued for a tax refund of $21.00). Klemen-cic thus satisfies the requirement of establishing an injury in fact, and because that injury is traceable to the IRS Rule and redressable through a judicial decision invalidating the rule, we find that he has standing to challenge the rule. We therefore proceed to consider whether Klemen-cic may mount his challenge under the APA.

B

The APA provides a cause of action to challenge final agency action “for which there is no other adequate remedy in a court.” 5 U.S.C. § 704. The government argues that even if Klemencic has standing to challenge the IRS Rule, he cannot do so under the APA because he has an adequate alternative remedy in the form of a tax-refund suit: Klemencic could violate the individual mandate, pay the penalty, and then sue for a refund, raising the same arguments he makes here. See 28 U.S.C. § 1346(a)(1); see also 26 U.S.C. § 7422(a). Such a remedy is adequate, the government contends, because if Klemencic were successful, the suit would make him financially whole.

The APA “embodies the basic presumption of judicial review” of agency action. Abbott Labs. v. Gardner, 387 U.S. 136, 140, 87 S.Ct. 1507, 18 L.Ed.2d 681 (1967). Therefore, in determining whether an alternative remedy is adequate, we must give the APA’s “generous review provisions” a “hospitable interpretation,” such that “only upon a showing of clear and convincing evidence of a contrary legislative intent should the courts restrict access to judicial review.” Id. at 141, 87 S.Ct. 1507 (internal quotation marks omitted); see Garcia v. Vilsack, 563 F.3d 519, 523 (D.C.Cir.2009). Under this standard, “[a]n alternative remedy will not be adequate ... if the remedy offers only ‘doubtful and limited relief ” Garcia, 563 F.3d at 522 (quoting Bowen v. Massachusetts, 487 U.S. 879, 901, 108 S.Ct. 2722, 101 L.Ed.2d 749 (1988)). Although “the alternative remedy need not provide relief identical to relief under the APA,” it must “offer[] relief of the ‘same genre.’ ” Id. at 522 (quoting El Rio Santa Cruz Neighborhood Health Ctr. v. U.S. Dep’t of Health & Human Servs., 396 F.3d 1265, 1272 (D.C.Cir.2005)).

In arguing that a tax refund suit provides an adequate alternative remedy, the government emphasizes Klemencic’s ability to recover any assessed overpayment, plus interest. But that backward-looking relief differs in kind from the prospective relief Klemencic could obtain under the APA. See Bowen, 487 U.S. at 904-05, 108 S.Ct. 2722 (rejecting as “unprecedented” the government’s argument that a suit for monetary damages is an adequate alternative to prospective relief under the APA). Specifically, requiring Klemencic to proceed via refund suit would deprive him of the opportunity to obtain a “certificate of exemption.” See 45 C.F.R. § 155.605(g)(2). Such certificates are a form of safe harbor, allowing an individual to obtain an exemption from the mandate’s penalty on the basis of projected income, “notwithstanding any [subsequent] change in an individual’s circumstances.” Id. § 155.605(g)(2)(vi). Unlike the “prospeetive[]” assurance such certificates offer, id., a refund suit would require Klemencic to violate the law as it now stands, pay a penalty, and only then challenge the assessment of the penalty for that previous year based on his actual income. And even if Klemencic were to prevail, his relief — financial restitution — would be backwards looking, meaning that Klemencic would have to repeat the cycle the following year. The government offers no sug-*398gestión that he could obtain a certificate of exemption through a refund action.

Furthermore, it is not clear that Klem-encic could obtain any prospective relief through a refund action, let alone that which he seeks under his APA claim— namely, a declaration that the IRS Rule is invalid and an injunction barring its implementation. As we explained in Cohen v. United States, the provision authorizing refund suits “does not, at least explicitly, allow for prospective relief.” 650 F.3d 717, 732 (D.C.Cir.2011) (en banc); see 26 U.S.C. § 7422(a) (setting forth requirements applicable to any “suit or proceeding ... for the recovery ... of any penalty claimed to have been collected without authority” (emphasis added)). And the government here does not suggest that it implicitly allows such relief, maintaining instead the studied silence as to the availability of non-monetary relief that, in Cohen, we interpreted as a concession of the limited nature of the remedies a refund suit under section 7422 offers. See Cohen, 650 F.3d at 732. (noting that, by being “agnostic concerning the availability of broad equitable remedies as part of a refund suit,” the IRS “unknowingly concedes” that an action under section 7422 does not offer prospective relief). We must therefore conclude that a tax refund suit is inadequate as an alternative remedy: it is “doubtful” that it offers prospective relief at all, and the monetary relief it does offer is clearly not “of the same genre” as the relief available to appellants under the APA. See Garcia, 563 F.3d at 522 (internal quotation marks omitted). Because a tax refund suit thus offers Klemencie only “doubtful and limited relief,” Bowen, 487 U.S. at 901, 108 S.Ct. 2722, we hold that the APA provides him with a cause of action to challenge the IRS Rule and turn to the merits of his claim.

Ill

On the merits, this ease requires us to determine whether the ACA permits the IRS to provide tax credits for insurance purchased through federal Exchanges. To make this determination, we begin by asking “whether Congress has directly spoken to the precise question at issue,” for if it has, we must give effect to its unambiguously expressed intent. Chevron U.S.A., Inc. v. Natural Res. Def. Council, 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). The text of section 36B is only the starting point of this analysis. That provision is but one piece of a vast, complex statutory scheme, and we must consider it both on its own and in relation to the ACA’s interconnected provisions and overall structure so as to interpret the Act, if possible, “as a symmetrical and coherent scheme.” See FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 133, 120 S.Ct. 1291, 146 L.Ed.2d 121 (2000) (internal quotation marks omitted); Wolf Run Mining Co. v. Fed. Mine Safety & Health Review Comm’n, 659 F.3d 1197, 1200 (D.C.Cir.2011).

Although both appellants and the government argue that the ACA, read in its totality, evinces clear congressional intent, they dispute what that intent actually is. Appellants argue that if taxpayers can receive credits only for plans enrolled in “through an Exchange established by the State under section 1311 of the [ACA],” then the IRS clearly cannot give credits to taxpayers who purchased insurance on an Exchange established by the federal government. After all, the federal government is not a “State,” see 42 U.S.C. § 18024(d) (defining “State” to “mean[ ] each of the 50 States and the District of Columbia”), and its authority to establish Exchanges appears in section 1321 rather than section 1311, see id. § 18041(c)(1). The government counters that appellants *399take a blinkered view of the ACA and that sections 1311 and 1321 of the Act establish complete equivalence between state and federal Exchanges, such that when the federal government establishes an Exchange, it does so standing in the state’s shoes. Furthermore, the government argues, whereas appellants’ construction of section 36B renders other provisions of the ACA absurd, its own view brings coherence to the statute and better promotes the purpose of the Act.

We conclude that appellants have the better of the argument: a federal Exchange is not an “Exchange established by the State,” and section 36B does not authorize the IRS to provide tax credits for insurance purchased on federal Exchanges. We reach this conclusion by the following path: First, we examine section 36B in light of sections 1311 and 1321, which authorize the establishment of state and federal Exchanges, respectively, and conclude that section 36B plainly distinguishes Exchanges established by states from those established by the federal government. We then consider the government’s arguments that this construction generates absurd results but find that it does not render other provisions of the ACA unworkable, let alone so unreasonable as to justify disregarding section 36B’s plain meaning. Finally, turning to the ACA’s purpose and legislative history, we find that the government again comes up short in its efforts to overcome the statutory text. Its appeals to the ACA’s broad aims do not demonstrate that Congress manifestly meant something other than what section 36B says.

A

The crux of this case is whether an Exchange established by the federal government is an “Exchange established by the State under section 1311 of the [ACA].” We therefore begin with the provisions authorizing states and the federal government to establish Exchanges. Section 1311 provides that states “shall” establish Exchanges. 42 U.S.C. § 18031(b)(1). But, as the parties agree, despite its seemingly mandatory language, section 1311 more cajoles than commands. A state is not literally required to establish an Exchange; the ACA merely encourages it to do so. And if a state elects not to (or is unable to), such that it “will not have any required Exchange operational by January 1, 2014,” section 1321 directs the federal government, through the Secretary of Health and Human Services, to “establish and operate such Exchange within the State.” Id. § 18041(c)(1) (emphasis added).

The phrase “such Exchange” has twofold significance. First, the word “such”— meaning “aforementioned,” see Black’s Law Dictionary 1473 (8th ed.2004); Webster’s Third Int’l Dictionary 2283 (1981)— signifies that the Exchange the Secretary must establish is the “required Exchange” that the state failed to establish. In other words, “such” conveys what a federal Exchange is: the equivalent of the Exchange a state would have established had it elected to do so. The meaning of “Exchange” in the ACA reinforces and builds on this sense. The ACA defines an “Exchange” as “an American Health Benefit Exchange established under [section 1311 of the ACA].” 42 U.S.C. § 300gg-91(d)(21). If we import that definition into the text of section 1321, the provision directs the Secretary to “establish ... such American Health Benefit Exchange established under [section 1311 of the ACA] within the State.” This suggests not only that the Secretary is to establish the type of exchange described in section 1311, but also that when she does so, she acts under section 1311, even though her authority *400appears in section 1321. Thus, section 1321 creates equivalence between state and federal Exchanges in two respects: in terms of what they are and the statutory authority under which they are established.

The problem confronting the IRS Rule is that subsidies also turn on a third attribute of Exchanges: who established them. Under section 36B, subsidies are available only for plans “enrolled in through an Exchange established by the State under section 1311 of the [ACA].” 26 U.S.C. § 36B(c)(2)(A)(i) (emphasis added); see also id. § 36B(b)(2)(A). Of the three elements of that provision — (1) an Exchange (2) established by the State (3) under section 1311 — federal Exchanges satisfy only two: they are Exchanges established under section 1311. Nothing in section 1321 deems federally-established Exchanges to be “Exchanged] established by the State.” This omission is particularly significant since Congress knew how to provide that a non-state entity should be treated as if it were a state when it sets up an Exchange. In a nearby section, the ACA provides that a U.S. territory that “elects ... to establish an Exchange ... shall be treated as a State.”2 42 U.S.C. § 18043(a)(1). The absence of similar language in section 1321 suggests that even though the federal government may establish an Exchange “within the State,” it does not in fact stand in the state’s shoes when doing so. See NFIB, 132 S.Ct. at 2583 (“Where Congress uses certain language in one part of a statute and different language in another, it is generally presumed that Congress acts intentionally.” (citing Russello v. United States, 464 U.S. 16, 23, 104 S.Ct. 296, 78 L.Ed.2d 17 (1983))).

The dissent attempts to supply this missing equivalency by pointing to section 1311(d)(1), which provides: “An Exchange shall be a governmental agency or nonprofit entity that is established by a State.” 42 U.S.C. § 18031(d)(1). According to the dissent, (d)(1) means that an Exchange established under section 1311 is, by definition, established by a state. Therefore, the dissent argues, because federal Exchanges are established under section 1311, they too, by definition, are established by a state.

The premise that (d)(1) is definitional, however, does not survive examination of (d)(l)’s context and the ACA’s structure. The other provisions of section 1311(d) are operational requirements, setting forth what Exchanges must (or, in some cases, may) do.3 See generally 42 U.S.C. § 18031(d)(2)-(7) (listing “Requirements”). Read in keeping with that theme, (d)(1) would simply require that an Exchange operate as either a governmental agency or nonprofit entity. But the dissent would have us construe (d)(1) differently. In its view, (d)(1) plays a definitional role unique among section 1311(d)’s otherwise operational provisions, creating a legal fiction that any Exchange is, by definition, established by a state, even when, as a matter of fact, it is not. That reading, however, would render (d)(1) the odd man out twice over: both within section 1311(d) and among the ACA’s other definitional provi*401sions, which, unlike (d)(1), employ the (unmistakably definitional) formula of “The term ‘X’ means.... ” See, e.g., 42 U.S.C. §§ 300gg-91, 18024; see also 26 U.S.C. § 4980H(c).

The dissent’s reading would also require us to overlook the fact that section 1311(d) would be a strange place for Congress to have buried such a legal fiction. Section 1311, after all, concerns Exchanges that are established by states in fact; the legal fiction the dissent urges would matter only to Exchanges established by the federal government. To accept the dissent’s construction would therefore transform (d)(1) into the proverbial elephant in the mouse-hole — the “ancillary provision! ]” that “alter[s] the fundamental details of a regulatory scheme.” Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 468, 121 S.Ct. 903, 149 L.Ed.2d 1 (2001). The Supreme Court has repeatedly held that Congress does not legislate in this manner, see id.; accord Gonzales v. Oregon, 546 U.S. 243, 267, 126 S.Ct. 904, 163 L.Ed.2d 748 (2006), and we see no evidence that it did so here.4 Indeed, we are particularly loath to accept the dissent’s construction given that there are far more natural locations to place this fiction, such as section 1321 or the provision defining the term “Exchange,” 42 U.S.C. § 300gg-91(d)(21).

The dissent’s construction of (d)(1) also ignores the structural relationship between sections 1311 and 1321. Just as section 1311(b)(1) assumes that states will establish Exchanges in general, see 42 U.S.C. § 18031(b)(1), section 1311(d) assumes that states will carry out the specific requirements Exchanges must meet. But if those assumptions prove wrong, section 1321 assigns the federal government responsibility both to establish the Exchange and to ensure that it satisfies the particulars of section 1311(d). See id. § 18041(c) (directing the Secretary to “establish and operate such Exchange” and to “take such actions as are necessary to implement such other requirements” pertaining to Exchanges). In other words, section 1321 creates a limited scheme of substitution: the requirements assigned to states by 1311(d) are transferred to the federal government if a state fails to establish an Exchange. The specific requirement that (d)(1) assumes each state will fulfill is to establish an Exchange in the form of “a governmen*402tal agency or nonprofit entity.” So if a state elects not to participate in the creation of an Exchange, section 1321 directs the federal government that it must create “a governmental agency or nonprofit entity” to serve as the Exchange. Crucially, this construction does not entail ignoring the plain meaning of “established by a State” in section 1311(d)(1); here, section 1321 tells us to substitute the federal government for the state under a certain scenario. But there is nothing comparable with respect to section 36B: no analogue to section 1321 says that section 36B should be read to encompass federally-established Exchanges. Accordingly, we reject the dissent’s argument that, because federal Exchanges are established under section 1311, they are by definition “established by a State.”

Instead, sections 1311 and 1321 lead us to interpret section 36B essentially as appellants do. Those provisions, to be sure, establish some degree of equivalence between state and federal Exchanges— enough, indeed, that if section 36B had authorized credits for insurance purchased on an “Exchange established under section 1311,” the IRS Rule would stand. But section 36B actually authorizes credits only for coverage purchased on an “Exchange established by the State under section 1311,” 26 U.S.C. § 36B(c)(2)(A)(i), and the government offers no textual basis — in sections 1311 and 1321 or elsewhere — for concluding that a federally — established Exchange is, in fact or legal fiction, established by a state. Moreover, as we have noted, that absence is especially glaring given that the ACA elsewhere provides that a federal territory that establishes an Exchange “shall be treated as a State,” 42 U.S.C. § 18043(a), clearly demonstrating that Congress knew how to deem a non-state entity to be a “State.” Thus, at least in light of sections 1311 and 1321, the meaning of section 36B appears plain: a federal Exchange is not an “Exchange established by the State.”

B

The government argues that we should not adopt the plain meaning of section 36B, however, because doing so would render several other provisions of the ACA absurd. Our obligation to avoid adopting statutory constructions with absurd results is well-established. See Pub. Citizen v. U.S. Dep’t of Justice, 491 U.S. 440, 454-55, 109 S.Ct. 2558, 105 L.Ed.2d 377 (1989). Under this principle, we will not give effect to a statute’s literal meaning when doing so would “render[ the] statute nonsensical or superfluous or ... create[ ] an outcome so contrary to perceived social values that Congress could not have intended it.” United States v. Cook, 594 F.3d 883, 891 (D.C.Cir.2010) (internal quotation marks omitted). But we do not disregard statutory text lightly. The Constitution assigns the legislative power to Congress, and Congress alone, see U.S. Const, art. I, § 1, and legislating often entails compromises that courts must respect. See Barnhart v. Sigmon Coal Co., 534 U.S. 438, 461, 122 S.Ct. 941, 151 L.Ed.2d 908 (2002). See generally John F. Manning, The Absurdity Doctrine, 116 Harv. L.Rev. 2387, 2434-2435 (2003) (warning that an overbroad application of the absurdity doctrine “contradicts the rule-of-law objectives implicit in the Constitution’s strict separation of lawmaking from judging”). We therefore give the absurdity principle a narrow domain, insisting that a given construction cross a “high threshold” of unreasonableness before we conclude that a statute does not mean what it says. Cook, 594 F.3d at 891. A provision thus “may seem odd” without being “absurd,” and in such instances “it is up to Congress rather than the courts to fix it,” even if it “may have been an unintentional drafting gap.” Exx*403on Mobil Corp. v. Allapattah Servs., Inc., 545 U.S. 546, 565, 125 S.Ct. 2611, 162 L.Ed.2d 502 (2005) (internal quotation marks omitted); see also Sierra Club v. EPA, 294 F.3d 155, 161 (D.C.Cir.2002) (“Because our role is not to ‘correct’ the text so that it better serves the statute’s purposes, we will not ratify an interpretation that abrogates the enacted statutory text absent an extraordinarily convincing justification.” (internal quotation marks and citation omitted)).

i

The government first argues that we must uphold the IRS Rule to avoid rendering language in 26 U.S.C. § 36B(f) superfluous. Titled “Reconciliation of credit and advance credit,” section 36B(f) requires the IRS to reduce a taxpayer’s end-of-year credit by the amount of any advance payments made by the government to the taxpayer’s insurer to offset the cost of monthly premiums. Id. § 36B(f)(i); see 42 U.S.C. § 18082(c)(2)(A) (authorizing such advance payments). As relevant here, section 36B(f) also requires “each Exchange” — i.e., both state and federal Exchanges — to report certain information to the government. With respect to any health plan it provides, an Exchange must report:

(A) The level of coverage ... and the period such coverage was in effect.
(B) The total premium for the coverage without regard to the credit under this section or cost-sharing reductions under section 1402 of [the ACA],
(C) The aggregate amount of any advance payment of such credit or reductions ....
(D) The name, address, and [taxpayer identification number (TIN) ] of the primary insured and the name and TIN of each other individual obtaining coverage under the policy.
(E) Any information provided to the Exchange, including any change of circumstances, necessary to determine eligibility for, and the amount of, such credit.
(F) Information necessary to determine whether a taxpayer has received excess advance payments.

26 U.S.C. § 36B(f)(3). The government contends that these reporting requirements assume that credits are available on federal Exchanges, and it argues that the requirements would be superfluous, even nonsensical, as applied to federal Exchanges if we were to reject that assumption.

Not so. Even if credits are unavailable on federal Exchanges, reporting by those Exchanges still serves the purpose of enforcing the individual mandate — a point the IRS, in fact, acknowledged in promulgating a recent regulation, 26 C.F.R. § 1.6055-1(d)(l). That regulation exempts insurers from 26 U.S.C. § 6055, which otherwise would require that, for each policy they issue, insurers report to the IRS such information as “the name, address, and TIN of the primary insured,” the dates of coverage, and the “amount (if any) or any advance payment ... or of any premium tax credit under section 36B with respect to such coverage.” 26 U.S.C. § 6055(b)(1)(B). The IRS justified the exemption for insurers on the ground that “Exchanges must report on this coverage under section 36B(f)(3).” Information Reporting of Minimum Essential Coverage, 79 Fed.Reg. 13,220, 13,221 (Mar. 10, 2014); see 26 C.F.R. § 1.6055-1(d)(1).5 The gov*404ernment’s claim that section 36B(f)(3)’s reporting requirement serves no purpose other than reconciling credits is therefore simply not true.6

Furthermore, holding that credits are unavailable on federal Exchanges would not convert the specific reporting requirements concerning credits into an “ ‘empty gesture.’ ” Gov’t Br. 28 (quoting Fund for Animals, Inc. v. Kempthorne, 472 F.3d 872, 878 (D.C.Cir.2006)). Those requirements would still allow the reconciling of credits on state Exchanges; as applied to federal Exchanges, they would simply be over-inclusive. Over-inelusiveness, however, remains a problem even if we were to agree that section 36B allows credits on federal Exchanges. Section 36B(f)(3), after all, mandates reporting “with respect to any health plan provided through the Exchange,” 26 U.S.C. § 36B(f)(3) (emphasis added), even though only plans purchased by taxpayers with incomes between 100 and 400 percent of the federal poverty line may be subsidized, see id. § 36B(a), (c)(1)(A). A weakness common to both views of the availability of credits hardly serves as a basis for choosing between them.

ii

The government next points to the supposedly absurd consequences appellants’ interpretation of section 36B would have for section 1312 of the ACA, which defines the rights of “qualified individuals.” See 42 U.S.C. § 18032. The term “ ‘qualified individual’ means, with respect to an Exchange, an individual who- — (i) is seeking to enroll in a qualified health plan in the individual market offered through the Exchange; and (ii) resides in the State that established the Exchange.” Id. § 18032(f)(1)(A). If this provision is given its plain meaning, then the 36 states with federal Exchanges (that, obviously, the states did not establish) have no qualified individuals. That outcome is absurd, the government argues, because in its view section 1312 restricts access to Exchanges to qualified individuals alone. See 45 C.F.R. § 155.20. The absence of qualified individuals would mean that federal Exchanges have no customers and therefore no purpose. The government urges us to avoid this outcome by construing section 1321 to authorize the federal government to establish Exchanges “on behalf of” states that decline to do so. Gov’t Br. 21 (internal quotation marks omitted).

The government, however, tilts at windmills. It assumes that when section 1312(a) states that “[a] qualified individual may enroll in any qualified health plan available to such individual and for which such individual is eligible,” 42 U.S.C. § 18032(a)(1), it means that only a qualified individual may enroll in such a plan. The obvious flaw in this interpretation is that the word “only” does not appear in the provision. We have repeatedly emphasized that it is “not our role” to “engage in a statutory rewrite” by “inserting] the word ‘only’ here and there.” Adirondack Med. Ctr. v. Sebelius, 740 F.3d 692, 699-700 (D.C.Cir.2014); see Lamie v. U.S. Tr., 540 U.S. 526, 538, 124 S.Ct. 1023, 157 L.Ed.2d 1024 (2004) (rejecting an interpretation that “would have [the Court] read *405an absent word into the statute” because such an interpretation “would result ‘not [in] a construction of [the] statute, but, in effect, an enlargement of it by the court’ ” (second and third alterations in original) (quoting Iselin v. United States, 270 U.S. 245, 251, 46 S.Ct. 248, 70 L.Ed. 566 (1926))); Pub. Citizen, 533 F.3d at 817 (“Congress knows well how to say that disclosures may be made only under specified provisions or circumstances, but it did not do so here.” (footnote omitted)). Section 1312(a)’s actual language simply establishes the right of a qualified individual to enroll in any qualified health plan, at any level of coverage.7 On this reading, giving the phrase “established by the State” its plain meaning creates no difficulty, let alone absurdity. Federal Exchanges might not have qualified individuals, but they would still have customers— namely, individuals who are not “qualified individuals.”8

Several other provisions in section 1312 imply that not only “qualified individuals” may participate in an Exchange. Take, for example, the provision concerning incarcerated convicts. Section 1312(f)(1)(B) states that “[a]n individual shall not be treated as a qualified individual if, at the time of enrollment, the individual is incarcerated, other than incarceration pending the disposition of charges.” 42 U.S.C. § 18032(f)(1)(B) (emphasis added). By implying that an incarcerated convict may enroll in coverage through an Exchange despite not being a “qualified individual,” this provision suggests that participation in an Exchange does not depend on “qualified individual” status. That proposition gains further strength from section 1312(d)(3), which states, first, that “[nothing in this title shall be construed to restrict the choice of a qualified individual to enroll or not to enroll in a qualified health plan or to participate in an Exchange,” 42 U.S.C. § 18032(d)(3)(A), and, second, that “[n]othing in this title shall be construed to compel an individual to enroll in a qualified health plan or to participate in an Exchange,” id. § 18032(d)(3)(B). The second provision, which speaks of “individuals]” generally, would be wholly unnecessary if only “qualified individuals” were eligible to participate in the Exchanges.9

iii

The government also claims that a plain meaning reading of section 36B would have peculiar effects under 42 U.S.C. § 1396a(gg)(i). That provision states that, as a condition of receiving Medicaid funds, a State may not tighten its Medicaid eligibility standards for adults until “the *406date on which the Secretary determines that an Exchange established by the State under [section 1311] is fully operational.” 42 U.S.C. § 1396a(gg)(l). If a federally-established Exchange is not one “established by the State,” the government argues, states with federal Exchanges “would never be relieved of th[is] ... requirement,” transforming an “interim measure” into a “perpetual obligation.” Gov’t Br. at 33. But appellants propose a logical explanation for why the ACA might establish this rule: to preserve Medicaid benefits for the impoverished residents of states where, as a result of having federally-established Exchanges, subsidies are unavailable. Cf. Pub. Citizen, 533 F.3d at 817 (adopting a reasonable explanation of a provision’s purpose despite not being able to “know for certain what purpose Congress had in mind”). In this light, the results produced by giving section 36B its plain meaning seem sensible, not absurd.10

iv

The government urges us, in effect, to strike from section 36B the phrase “established by the State,” on the ground that giving force to its plain meaning renders other provisions of the Act absurd. But we find that the government has failed to make the extraordinary showing required for such judicial rewriting of an act of Congress. Nothing about the imperative to read section 36B in harmony with the rest of the ACA requires interpreting “established by the State” to mean anything other than what it plainly says.

C

This conclusion places us at a fork in our precedent. One line of cases instructs us to cease our inquiry and give effect to the statute’s unambiguous language. See Coal. for Responsible Regulation, Inc. v. EPA 684 F.3d 102, 137 (D.C.Cir.2012) (per curiam) (noting, in the Chevron context, that “ ‘[w]hen the words of a statute are unambiguous ... judicial inquiry is complete’ ”) (ellipsis in original) (quoting Conn. Nat’l Bank v. Germain, 503 U.S. 249, 254, 112 S.Ct. 1146, 117 L.Ed.2d 391 (1992)), aff'd in relevant part sub nom. Util. Air Regulatory Grp. v. EPA (UARG), — U.S. -, 134 S.Ct. 2427, 2448, 189 L.Ed.2d 372 (2014); accord Dep’t of Housing & Urban Dev. v. Rucker, 535 U.S. 125, 132-33, 122 S.Ct. 1230, 152 L.Ed.2d 258 (2002); Hughes Aircraft Co. v. Jacobson, 525 U.S. 432, 438, 119 S.Ct. 755, 142 L.Ed.2d 881 (1999) (“As in any case of statutory construction, our analysis begins with the language of the statute. And where the statutory language provides a clear answer, it ends there as well.” (emphasis added) (internal quotation marks and citation omitted)); see also Am. Fed’n of Gov’t Emps. v. Shinseki, 709 F.3d 29, 33 (D.C.Cir.2013). Another tells us to wade into the legislative history in the hope of glimpsing “new light on congressional intent.” Sierra Club v. EPA, 551 F.3d 1019, 1027 (D.C.Cir.2008). But, though we recognize that our *407decision about which path to travel implicates substantial theoretical questions of statutory interpretation, its practical consequences are less momentous here because both paths lead to the same destination. Therefore, assuming arguendo that it is proper to consult legislative history when the statutory text is clear, we consider what light the ACA’s history offers.

We begin by clarifying the role the ACA’s legislative history might play in our analysis. Legislative history is a means to an end, to be consulted for evidence of congressional intent. See, e.g., Sierra Club, 551 F.3d at 1027. But legislative history is not the sole, or even the primary, source of such evidence. Rather, “[t]he most reliable guide to congressional intent is the legislation the Congress enacted.” Sierra Club, 294 F.3d at 161; see also Cal. Indep. Sys. Operator Corp. v. FERC, 372 F.3d 395, 400 (D.C.Cir.2004) (“[W]e assume ‘that the legislative purpose is expressed by the ordinary meaning of the words used.’” (quoting Sec. Indus. Ass’n v. Bd. of Governors of Fed. Reserve Sys., 468 U.S. 137, 149, 104 S.Ct. 2979, 82 L.Ed.2d 107 (1984))); Engine Mfrs. Ass’n, 88 F.3d at 1088 (noting that the “most traditional tool” for “determining] Congressional intent” is “to read the text”). Where used, legislative history plays a distinctly secondary role. Its purpose is not to confirm already clear text; clear text speaks for itself and requires no “amen” in the historical record. See, e.g., Harrison v. PPG Indus., Inc., 446 U.S. 578, 592, 100 S.Ct. 1889, 64 L.Ed.2d 525 (1980) (“[I]t would be a strange canon of statutory construction that would require Congress to state in committee reports or elsewhere in its deliberations that which is obvious on the face of a statute.”). Instead, only when “apparently plain language compels an ‘odd result’ ” might we look to legislative history to ensure that the “ ‘literal application of a statute will [not] produce a result demonstrably at odds with the intentions of its drafters.’ ” Engine Mfrs. Ass’n, 88 F.3d at 1088 (quoting Public Citizen, 491 U.S. at 454, 109 S.Ct. 2558, and United States v. Ron Pair Enters., Inc., 489 U.S. 235, 242, 109 S.Ct. 1026, 103 L.Ed.2d 290 (1989)). Thus, accepting for the sake of argument the government’s contention that the results of appellants’ construction of section 36B are odd, our inquiry into the ACA’s legislative history is quite narrow. In the face of the statute’s plain meaning — a federal Exchange is not an “Exchange established by the State”— we ask only whether the legislative history provides evidence that this literal meaning is “demonstrably at odds with the intentions” of the ACA’s drafters. Unless evidence in the legislative record establishes that it is, we must hew to the statute’s plain meaning, even if it compels an odd result. See id. (“[T]here must be evidence that Congress meant something other than what it literally said before a court can depart from plain meaning.”); accord Garcia v. United States, 469 U.S. 70, 75, 105 S.Ct. 479, 83 L.Ed.2d 472 (1984) (noting that “only the most extraordinary showing of contrary intentions ... would justify a limitation on the ‘plain meaning’ of the statutory language”); Bldg. & Constr. Trades Dep’t, AFL-CIO v. U.S. Dep’t of Labor Wage Appeals Bd., 932 F.2d 985, 990 (D.C.Cir.1991).

Here, the scant legislative history sheds little light on the precise question of the availability of subsidies on federal Exchanges. The government points, for example, to a Congressional Budget Office report from November 2009, before the ACA’s adoption, that calculated the cost of subsidies based on the assumption that they would be available in all states. But that assumption is as consistent with an expectation that all states would cooperate (i.e., establish their own Exchanges) as *408with an understanding that subsidies would be available on federal Exchanges as well. Cf. Robert Pear, U.S. Officials Brace for Huge Task of Operating Health Exchanges, N.Y. TIMES, at A17 (Aug. 5, 2012) (“When Congress passed legislation to expand coverage two years ago, Mr. Obama and lawmakers assumed that every state would set up its own exchange.... ”). Equally unilluminating are floor statements by Senate sponsors of the ACA touting the availability and benefits of premium tax credits in general, but not addressing the precise issue of whether they would be available on federal Exchanges.

The government and its amici are thus left to urge the court to infer meaning from silence, arguing that “during the debates over the ACA, no one suggested, let alone explicitly stated, that a State’s citizens would lose access to the tax credits if the State failed to establish its own Exchange.” Br. of Amici Members of Congress and State Legislatures 8. The historical record, however, belies this claim. The Senate Committee on Health, Education, Labor, and Pensions (HELP) proposed a bill that specifically contemplated penalizing states that refused to participate in establishing “American Health Benefit Gateways,” the equivalent of Exchanges, by denying credits to such states’ residents for four years. See Affordable Health Choices Act, S. 1679, 111th Cong. § 3104(a), (d)(2) (2009). This is not to say that section 36B necessarily incorporated this thinking; we agree that inferences from unenacted legislation are too uncertain to be a helpful guide to the intent behind a specific provision. See Village of Barrington v. Surface Transp. Bd., 636 F.3d 650, 666 (D.C.Cir.2011). But the HELP Committee’s bill certainly demonstrates that members of Congress at least considered the notion of using subsidies as an incentive to gain states’ cooperation.

In any case, even if the historical record were silent, that silence is unhelpful to the government. For the court to depart from the ACA’s plain meaning, which favors appellants, “there must be evidence that Congress meant something other than what it literally said,” from which the court can conclude that applying the statute literally would be “ ‘demonstrably at odds with the intentions of [the ACA’s] drafters.’ ” Engine Mfrs. Ass’n, 88 F.3d at 1088 (quoting Ron Pair Enters., 489 U.S. at 242, 109 S.Ct. 1026) (emphases added). As Chief Justice Marshall wrote, “it is incumbent on those who oppose” a statute’s plain meaning “to shew an intent varying from that which the words import.” United States v. Fisher, 6 U.S. (2 Cranch) 358, 386, 2 L.Ed. 304 (1805). Nothing the government or its amici cite demonstrates what that precise intent was. And “[i]n the absence of such evidence, the court cannot ignore the text by assuming that if the statute seems odd to us, i.e., the statute is not as we would have predicted beforehand that Congress would write it, it could be the product only of oversight, imprecision, or drafting error.” Engine Mfrs. Ass’n, 88 F.3d at 1088-89; see also id. at 1091 (“With such a meager record of what happened in conference, the court is unable to reconstruct the legislative compromises that were made. Even if the final product might strike us as unexpected ... the court could not make the leap from such an impression to the certainty that such a result was unintentional”).

The government, together with the dissent, also leans heavily on a more abstract form of legislative history — Congress’s broad purpose in passing the ACA — urging the court to view section 36B through the lens of the ACA’s economic theory and ultimate aims. They emphasize that to achieve the goals of “near universal coverage” and “lowering] health insurance premiums,” 42 U.S.C. § 18091(2)(D), (F), the *409ACA relies on three interrelated policies: insurance market reforms prohibiting insurers from denying coverage or charging higher premiums based on an individual’s health status, see, e.g., id. § 300gg (community rating requirement); id. § 300gg-l (guaranteed issue requirement); the individual mandate, see 26 U.S.C. § 5000A; and subsidies to individuals purchasing insurance in the individual market, see id. § 36B. These policies, the government and dissent explain, are like the legs of a three-legged stool; remove any one, and the ACA will collapse. The insurance market reforms are necessary to expand the availability of insurance. The individual mandate is necessary to avoid the adverse selection that would result if people could exploit the insurance market reforms to wait to purchase insurance until they were sick. And subsidies are necessary both to make the mandated insurance affordable and, in so doing, to expand the reach of the individual mandate by reducing the cost of insurance below the threshold-eight percent of household income-at which taxpayers are exempt from the mandate’s penalty. See 26 U.S.C. § 5000A(e)(l)(A)-(B). Given this structure, the government and dissent argue that it is “inconceivable” to think Congress would have risked the ACA’s stability by making subsidies conditional on states establishing Exchanges.11 Dissenting Op. at 2.

*410Yet the supposedly unthinkable scenario the government and dissent describe — one in which insurers in states with federal Exchanges remain subject to the community rating and guaranteed issue requirements but lack a broad base of healthy customers to stabilize prices and avoid adverse selection — is exactly what the ACA enacts in such federal territories as the Northern Mariana Islands, where the Act imposes guaranteed issue and community rating requirements without an individual mandate. See 26 U.S.C. § 5000A(f)(4) (exempting residents of such federal territories as Puerto Rico and the Northern Mariana Islands from the individual mandate by providing that they are automatically treated as having “minimum essential coverage”); 42 U.S.C. § 201(f) (providing that the Public Health Service Act, where the guaranteed issue and community rating requirements appear, applies to residents of such territories). This combination, predictably, has thrown individual insurance markets in the territories into turmoil. See Sarah Kliff, Think Your State Has Obamacare Problems? They’re Nothing Compared to Guam, Wash. Post (Dec. 19, 2013), http://www. washingtonpost.com/blogs/wonkblog/wp/ 2013/12/19/think-your-statehas-obamacare-problems-theyre-nothing-compared-to-guam/. But HHS has nevertheless refused to exempt the territories from the guaranteed issue and community rating requirements, recognizing that, “[h]owever meritorious” the reasons for doing so might be, “HHS is not authorized to choose which provisions of the [ACA] might apply to the territories.” Letter from Gary Cohen, Director, Center for Consumer Information and Insurance Oversight, HHS, to Sixto K. Igisomar, Secretary of Commerce, Commonwealth of the Northern Mariana Islands (July 12, 2013), available at http://www.doi.gov/oia/ igia/upload/12-3-HHS-CMS-CNMI-Letterigisomar7-12-13.pdf

Moreover, the territories are not the only instance where the ACA did the unimaginable. A separate title of the ACA, known as the Community Living Assistance Services and Supports (CLASS) Act, see ACA, Pub.L. No. 111-148, §§ 8001-8002, 124 Stat. 119, 828-47 (2010), required the Secretary of HHS to establish a long-term care insurance program subject to guaranteed issue and community rating requirements but unaided by an individual mandate or premium subsidies, see 124 Stat. at 834. This recipe for adverse selection risk never materialized only because Congress, in response to actuarial analyses predicting that the CLASS Act would be fiscally unsustainable, repealed the provision in 2013.12 See American Taxpayer *411Relief Act of 2012, Pub.L. No. 112-240, § 642, 126 Stat. 2313, 2358 (2013); Sarah Kliff, The Fiscal Cliff Cuts $1.9 Billion from Obamacare. Here’s How, Wash. Post (Jan. 2, 2013), http://www. washingtonpost.com/ blogs/wonk-blog/wp/2013/01/02 /the-fiscal-cliff-cuts-1-9-billion-from-obamacareheres-how/.

The CLASS Act and the provisions applicable to the territories attest that Congress twice did exactly what the government and the dissent insist it never would: introduce significant adverse selection risk to insurance markets. This is not to say that as Congress did in the CLASS Act and territories, so too must it have done in section 36B; perhaps Congress was willing to tolerate risks in those corners of the insurance market that it never would tolerate at its core. But perhaps not. The point is that we don’t know, and in asking us to ignore the best evidence of Congress’s intent — the text of section 36B — in favor of assumptions about the risks that Congress would or would not tolerate— assumptions doubtlessly influenced by hindsight — the government and dissent in effect urge us to substitute our judgment for Congress’s. We refuse. As the Supreme Court explained just this term, “an agency may not rewrite clear statutory terms to suit its own sense of how the statute should operate.” UARG, 134 S.Ct. at 2446. And neither may we. “The role of th[e] [c]ourt is to apply the statute as it is written — even if we think some other approach might ‘accor[d] with good policy.’ ” Burrage v. United States, — U.S. -, 134 S.Ct. 881, 892, 187 L.Ed.2d 715 (2014) (quoting Comm’r v. Lundy, 516 U.S. 235, 252, 116 S.Ct. 647, 133 L.Ed.2d 611 (1996)) (third alteration in original); see also Lewis v. City of Chicago, 560 U.S. 205, 217, 130 S.Ct. 2191, 176 L.Ed.2d 967 (2010) (“[I]t is not our task to assess the consequences of each approach [to interpreting a statute] and adopt the one that produces the least mischief. Our charge is to give effect to the law Congress enacted.”); United States v. Locke, 471 U.S. 84, 95, 105 S.Ct. 1785, 85 L.Ed.2d 64 (1985) (“[T]he fact that Congress might have acted with greater clarity or foresight does not give courts a carte blanche to redraft statutes in an effort to achieve that which Congress is perceived to have failed to do.”).

More generally, the ACA’s ultimate aims shed little light on the “precise question at issue,” Chevron, 467 U.S. at 842-, 104 S.Ct. 2778namely, whether subsidies are available on federal Exchanges because such Exchanges are “established by the State.” As the Supreme Court has repeatedly warned, “it frustrates rather than effectuates legislative intent simplistically to assume that whatever furthers the statute’s primary objective must be the law” because “no legislation pursues its purposes at all costs.” Rodriguez v. United States, 480 U.S. 522, 525-26, 107 S.Ct. 1391, 94 L.Ed.2d 533 (1987) (per curiam); see also Pension Benefit Guar. Corp. v. LTV Corp., 496 U.S. 633, 646-47, 110 S.Ct. 2668, 110 L.Ed.2d 579 (1990); MetroPCS Cal., LLC v. FCC, 644 F.3d 410, 414 (D.C.Cir.2011) (“ ‘The Act must do everything necessary to achieve its broad purpose’ is the slogan of the enthusiast, not the analytical tool of the arbiter.”). Thus, if legislative intent is to be our lodestar, we cannot assume, as the government does, that section 36B sin*412gle-mindedly pursues the ACA’s lofty goals.

The fact is that the legislative record provides little indication one way or the other of congressional intent, but the statutory text does. Section 36B plainly makes subsidies available only on Exchanges established by states. And in the absence of any contrary indications, that text is conclusive evidence of Congress’s intent. Cf. Ethyl Corp. v. EPA, 51 F.3d 1053, 1063 (D.C.Cir.1995) (“At best, the legislative history is cryptic, and this surely is not enough to overcome the plain meaning of the statute.”). To hold otherwise would be to say that enacted legislation, on its own, does not command our respect — an utterly untenable proposition. Accordingly, applying the statute’s plain meaning, we find that section 36B unambiguously forecloses the interpretation embodied in the IRS Rule and instead limits the availability of premium tax credits to state-established Exchanges.

IV

We reach this conclusion, frankly, with reluctance. At least until states that wish to can set up Exchanges, our ruling will likely have significant consequences both for the millions of individuals receiving tax credits through federal Exchanges and for health insurance markets more broadly. But, high as those stakes are, the principle of legislative supremacy that guides us is higher still. Within constitutional limits, Congress is supreme in matters of policy, and the consequence of that supremacy is that our duty when interpreting a statute is to ascertain the meaning of the words of the statute duly enacted through the formal legislative process. This limited role serves democratic interests by ensuring that policy is made by elected, politically accountable representatives, not by appointed, life-tenured judges.

Thus, although our decision has major consequences, our role is quite limited: deciding whether the IRS Rule is a permissible reading of the ACA. Having concluded it is not, we reverse the district court and remand with instructions to grant summary judgment to appellants and vacate the IRS Rule.

. Although West Virginia actually passed legislation authorizing the establishment of an Exchange, see W. VA. CODE § 33-16G-1 et seq., it subsequently decided to allow the federal government to establish the Exchange, in partnership with the state, due to cost concerns, see Nat’l Conference of State Legislatures: Health Insurance Exchanges or Marketplaces: State Action — May 2014, http:// www.ncsl.Org/Portals/l/Documents/HeaIth/ Health_Insurance_Exchanges_State_Profiles. pdf#page=49 (last visited June 12, 2014).

. Specifically, the ACA permits territories to be treated as states for the limited purposes of sections 1311, 1312, and 1313. See 42 U.S.C. § 18043(a).

. Although we attach little weight to section titles, the title of section 1321(c) — "Failure to establish Exchange or implement requirements” — reinforces this interpretation. See Gorman v. Nat’l Transp. Safety Bd., 558 F.3d 580, 588 n. 5 (D.C.Cir.2009) (recognizing that "headings ‘are of use ... when they shed light on some ambiguous word or phrase' " (ellipsis in original) (quoting Bhd. of R.R. Trainmen v. Balt. & O.R. Co., 331 U.S. 519, 529, 67 S.Ct. 1387, 91 L.Ed. 1646 (1947))).

. The government makes its own elephants-in-mouseholes argument, asserting that the formula for calculating tax credits (located in section 36B(b)) is an odd place to insert a condition that the states must establish their own Exchanges if they wish to secure tax credits for their citizens. The more natural location, the government suggests, would have been section 36B(a), which authorizes the credit in the first place. See 26 U.S.C. § 36B(a). But even under the government's reading of section 36B(b), the statutory formula houses an elephant: namely, the rule that subsidies are only available for plans purchased through Exchanges. Given that this other crucial limitation on the availability of subsidies is found only in section 36B’s formula, the government’s contention that the formula is a mere mousehole is unpersuasive.

Equally unpersuasive is the dissent's suggestion that section 36B cannot mean what it plainly says because Congress did not use an "if/then” formula to signify that credits are available only on state-established Exchanges. The dissent cites no authority for requiring such magic words, and we perceive none. Section 36B(b) also does not employ an "if/1 then” construction for the requirement that credit-eligible coverage be purchased through an Exchange, yet neither the government nor dissent disputes that requirement. It is simply not the case that Congress expresses conditions only through such language. Indeed, in 26 U.S.C. § 35, which establishes a tax credit to offset the cost of health insurance for certain workers displaced by foreign competition, Congress made the availability of the credit turn, in part, on state cooperation without employing "if/then” language, simply through its definition of the phrase "eligible coverage month.” See 26 U.S.C. § 35(e)(2)(A).

. Appellants also suggest that the information collected from federal Exchanges could be useful for the "Study on Affordable Coverage” *404mandated by the ACA in that same section. See ACA § 1401(c), 124 Stat, at 220.

. The dissent takes a slightly different tack, emphasizing that the "principal purpose” of the reporting requirement is to reconcile advance and end-of-year payments. Dissenting Op. at 423-24. We agree but fail to see how this helps the government. Reporting by state-established Exchanges still would serve this purpose, while reporting by federally-established Exchanges would serve the secondary purpose implicitly recognized by 26 C.F.R. § 1.6055 — 1(d)(1).

. Under the ACA, qualified health plans may offer four different levels of coverage: bronze, silver, gold, and platinum. The level of coverage reflects the percentage of the insured’s medical costs that the plan’s benefits are designed to cover. See 42 U.S.C. § 18022(d)(1). Lower levels of coverage have higher deductibles and thus higher out-of-pocket costs and, as a general matter, lower premiums. See id.; see also id. § 18032(a)(2) (providing that qualified employers may "select[] any level of coverage under section 18022(d) ... to be made available to employees through an Exchange”).

. The government warns that interpreting section 1312(a) as a non-discrimination provision would allow undocumented aliens to shop on Exchanges. Gov’t Br. at 31. But section 1312 specifically addresses that concern, providing that aliens not "lawfully present in the United States ... may not be covered under a qualified health plan in the individual market that is offered through an Exchange.” 42 U.S.C. § 18032(f)(3).

.We note that section 1312’s heading, "Consumer Choice,” and subsection 1312(a)'s heading, "Choice,” also suggest that the purpose of section 1312(a) is primarily to protect choice among levels of coverage, not restrict access to Exchanges.

. In a footnote, the government identifies another set of provisions that supposedly embodies the assumption that federal Exchanges are Exchanges “established by the State”: 42 U.S.C. § 1397ee(d)(3)(B)-(C). Those provisions instruct states to enroll children in coverage "offered through an Exchange established by the State under section [1311]” in the event of a funding shortfall in a state's Children’s Health Insurance Program. See id. § 1397ee(d)(3)(B). Although we recognize the oddity of requiring some states and not others to take this step, we do not see how it makes the statute nonsensical or otherwise meets the high threshold of absurdity. The statute remains workable, and nothing suggests that in states with federal Exchanges, the federal government could not step in and perform the same service for uninsured children. The government's bare citation to the provisions thus hardly demonstrates absurdity-

. Appellants do not challenge the government’s account of the economic theory behind the ACA, but they contend that the theory must be understood through the lens of political reality. In their telling, section 36B is the product of legislative compromise to secure the support of Nebraska Senator Ben Nelson, the crucial sixtieth vote needed to avoid a filibuster. Nelson opposed House plans for a national, federally-run exchange, fearing that it would set the United States down a path to a single-payer system. See Carrie Budoff Brown, Nelson: National Exchange a Deal-breaker, Politico (Jan. 25, 2010), http://www. politico. com/livepulse/0110/Nelson_National_ exchange_a_dealbreaker.html. To gain Nelson’s support, proponents of the ACA scrapped the national exchange in favor of establishing exchanges on a state-by-state basis. This change, in turn, required Congress to devise means of inducing states to take on the politically and technologically challenging task of establishing exchanges. Congress's solution, appellants maintain, was a package of “carrots” and "sticks” for states. The carrots included federal grants to states for “activities (including planning activities) related to establishing an [Exchange].” 42 U.S.C. § 18031(a)(3). The sticks included the prohibition against tightening Medicaid eligibility requirements imposed on states that do not create their own Exchanges. See id. § 1396a(gg). The most important incentive of all, appellants argue, was the provision at issue here: making premium tax credits available only for individual coverage purchased through state-established Exchanges. According to appellants, the ACA's supporters believed no state would refuse so good an offer — and, appellants add, perhaps no state would have had the IRS not eliminated this incentive by proposing and promulgating the IRS Rule, making subsidies available regardless of which entity established an Exchange, before states had to elect whether to establish Exchanges. See Health Insurance Premium Tax Credit, 77 Fed.Reg. 30,377, 30,378 (May 23, 2012); Health Insurance Premium Tax Credit, 76 Fed.Reg. 50,931, 50,934 (Aug. 17, 2011).

Like the government, however, appellants fail to marshal persuasive evidence (apart from the statutory text, that is) in support of their theory. Senator Nelson may have opposed a single, national exchange, but it does not necessarily follow that he opposed making subsidies available on federal fallback Exchanges in uncooperative states. Similarly, the fact that the ACA contained some incentives to states does not necessarily mean that section 36B is one of them. Nor does the fact that Congress has conditioned federal benefits on state cooperation in other contexts shed light on the precise question of whether Congress did so in section 36B. Thus, the most that can be said of appellants’ theory is that it is plausible. But we need not endorse appellants’ historical account to agree with their construction of section 36B. “Where the statutory language is clear and unambiguous, we need neither accept nor reject a particular ‘plausible’ explanation for why Congress *410would have written a statute [as it did].” Barnhart, 534 U.S. at 460, 122 S.Ct. 941.

. The dissent attempts to distinguish the market targeted by the CLASS Act from the individual insurance market by pointing out that the CLASS Act contains no individual mandate. In the dissent's view, the omission "of a tool [Congress] knew to be important to preventing adverse selection merely indicates that Congress had a substantially higher tolerance for the risk of adverse selection” in peripheral markets than in the core market. Dissenting Op. at 422. This argument, however, assumes the very conclusion at issue, taking for granted that the mandate in the individual market indeed is as broad as it must be to eliminate all adverse selection risk. But the plain language of section 36B suggests that it is not. If section 36B limits the availability of subsidies and thus curtails the reach of the individual mandate, this is evidence that Congress was tolerant of adverse selection risk in the core markets, although Congress might not have expected the risk to materialize.

We recognize that, from an economic standpoint, such adverse selection risk bodes ill for individual insurance markets. But it made no more sense economically in the CLASS Act. Congress may simply have miscalculated the consequences of omitting a *411mandate, as its decision to repeal the CLASS Act suggests. In any event, whether by error or design, the CLASS Act in clear terms created a significant adverse selection risk, which, as Congress and the government recognized, could be undone only by subsequent legislation, not administrative fiat. Cf. UARG, 134 S.Ct. at 2445 ("An agency has no power to 'tailor' legislation to bureaucratic policy goals by rewriting unambiguous statutory terms.'').