SUTTON, J., delivered the opinion of the court, in which McKEAGUE, J., concurred. WHITE, J. (pp. 342-52), delivered a separate dissenting opinion.
OPINION
SUTTON, Circuit Judge.
The Employee Retirement Income Security Act prohibits an employer from retaliating against an employee “because he has given information or has testified or is about to testify in any inquiry or proceeding relating to [the Act].” 29 U.S.C. § 1140. When Brian Sexton made a one[334]*334time unsolicited complaint to his employer about alleged violations of the Act, did that amount to “giv[ing] information ... in any inquiry?” We think not and hence affirm.
I.
Panel Processing makes floor panels. Brian Sexton worked as a general manager in its facility in Coldwater, Michigan, and served as a trustee for the company’s employee retirement plan. In 2011, Sexton and another trustee, Robert Karsten, campaigned on behalf of two employees running for the company’s board of directors. The employees won the election, but the board refused to seat them on the ground that it would violate the company’s bylaws, which limited the number of inside directors. At the same time, the board removed Sexton and Karsten as trustees of the retirement plan.
Two days later, Sexton emailed the chairman of the board:
I believe that your actions ... in refusing to seat [the employees] as directors of the company and removing Rob Kar-sten and me as Trustees of the [retirement plan] are violations of ERISA and the Michigan Corporations Business Act and other state and federal laws. I plan to bring these violations to the attention of the U.S. Department of Labor and Michigan Department of Licensing and Regulatory Affairs unless they are immediately remedied.
R. 9-13 at 2. Neither the chairman nor anyone else responded to the email, and Sexton took no further action. About six months later, the company fired Sexton from his job as a general manager.
Sexton sued the company in Michigan state court for violating the State’s Whis-tleblower Protection Act and for breaching his employment contract. One might think that, when a Michigan plaintiff sues a Michigan defendant under Michigan law in a Michigan court, the case would stay there. But the company invoked complete preemption, a doctrine that converts state law claims “within the scope of [ERISA’s] civil enforcement provisions” into federal claims. Metro. Life Ins. Co. v. Taylor, 481 U.S. 58, 66, 107 S.Ct. 1542, 95 L.Ed.2d 55 (1987). Under this banner, the company recharacterized Sexton’s state whistleblower claim as an ERISA claim. See 29 U.S.C. § 1140. It then removed the transfigured lawsuit to federal court.
Once in federal court, Sexton did not challenge the company’s removal of the case or its use of complete preemption. Sexton and Panel Processing instead litigated the case as though Sexton had raised a claim under ERISA. The district court granted summary judgment to the company on this ERISA claim, and it declined supplemental jurisdiction over Sexton’s breach-of-contract claim.
II.
On appeal, Sexton does not contest the district court’s application of complete preemption to Sexton’s claim or otherwise attempt to resurrect his claim under Michigan’s Whistleblower Protection Act. Under our case law, true enough, the doctrine of complete preemption goes to the subject matter jurisdiction of the court. Mikulski v. Centerior Energy Corp., 501 F.3d 555, 565 (6th Cir.2007) (en banc). But we are satisfied, just as the parties and the district court were satisfied, that the company properly invoked the doctrine here.
III.
Sexton (supported by the Secretary of Labor) instead takes aim at the district court’s rejection of his claim under ERISA. The relevant law says in relevant part: “It shall be unlawful for any person to discharge, fine, suspend, expel, or dis[335]*335criminate against any person because he has given information or has testified or is about to testify in any inquiry or proceeding relating to [the Act].” 29 U.S.C. § 1140. Both parties take as a given (for now) that the company fired Sexton because of his email to the chairman of the board.
All matters of interpretation begin with text; some end there. The text of this provision takes us well down the way of ending this dispute. No one claims that Sexton sent the email in the context of a “proceeding.” And no one claims that the email amounted to testimony—that he “testified” or was “about to testify” in an “inquiry” or “proceeding” by sending the email. That leaves the possibility that the email amounted to “givfing] information ... in any inquiry.” We have no problem with the idea that, by sending the email, Sexton was “giv[ing] information.” But, as Sexton and the Secretary concede, the statute demands not just the limitless “giv[ing]” of “information.” It requires that the information be given in an “inquiry”— “any” inquiry, to be sure, but an inquiry all the same. “Inquiry” might mean an official investigation, as in “Earl Warren led an inquiry into the assassination of President Kennedy.” Or it might mean a question or a request for information, as in “The advocate answered the judge’s inquiry.” See Bryan A. Garner, A Dictionary of Modem Legal Usage 317 (2d ed.1995). We need not decide whether § 1140 uses “inquiry” in the former sense, in the latter sense, or in both senses. Under either definition, no inquiry occurred. Sexton did not send the email in connection with an official investigation. And he did not send the email in response to a question or request for information. Even what we might call a nunc pro tunc theory of “givfing] information ... in any inquiry” would not aid Sexton. For even after he sent the email, no one asked him any questions about the subject of the email, and no investigation involving him ever occurred. As the text of the email confirms, this was nothing more than a complaint accompanied by a threat: Sexton demanded that the company change course and threatened action if it did not. The email neither asks nor answers a question. That is not “givfing] information ... in any inquiry.”
This interpretation not only respects the meaning—indeed accounts for the broadest possible meaning—of these terms, but it also respects the different ways Congress has dealt with retaliation in the work place. Congress has enacted roughly forty anti-retaliation laws, see Jon 0. Shimabukuro et al., Cong. Research Serv., R43045, Survey of Federal Whistleblower and Anti-Retaliation Laws (2013), and they tend to include two distinct types of prohibitions. The first protects employees who oppose, report or complain about unlawful practices. See, e.g., 29 U.S.C. § 218c(a) (Consumer Financial Protection Act) (“provided ... information relating to any violation”); id. § 215(a) (Fair Labor Standards Act) (“filed any complaint”); 42 U.S.C. § 2000e-3(a) (Title VII) (“opposed any ... unlawful employment practice”). The second protects employees who participate, testify or give information in inquiries, investigations, proceedings or hearings. See, e.g., 29 U.S.C. § 218c(a) (Patient Protection and Affordable Care Act) (“assisted or participated ... in ... a proceeding”); id. § 2615(b) (Family and Medical Leave Act) (“given ... any information in connection with any inquiry or proceeding”); 42 U.S.C. § 12203(a) (Americans with Disabilities Act) (“participated in any manner in an investigation, proceeding, or hearing”).
Most anti-retaliation laws include both types of clauses. See, e.g., 6 U.S.C. [336]*336§ 1142(a) (National Transit Systems Security Act); 7 U.S.C. § 26 (Commodity Exchange Act); 15 U.S.C. § 78u-6(h)(l)(A) (Securities Exchange Act); id. § 2087(a) (Consumer Product Safety Act); 29 U.S.C. § 215(a) (Fair Labor Standards Act); id. § 218c(a) (Patient Protection and Affordable Care Act); id. § 623(d) (Age Discrimination in Employment Act); id. § 660(c)(1) (Occupational Safety and Health Act); id. § 2615(b) (Family and Medical Leave Act); 30 U.S.C. § 815(c) (Federal Mine Safety and Health Act); 42 U.S.C. § 12203(a) (Americans with Disabilities Act); 42 U.S.C. § 2000e-3(a) (Title VII); 49 U.S.C. § 31105(a) (Commercial Motor Vehicle Safety Act); id. § 60129(a) (Pipeline Safety Improvement Act). Many of these laws were in existence before 1974, when Congress enacted ERISA. See the Age Discrimination in Employment Act, the Fair Labor Standards Act, the Occupational Safety and Health Act, and Title VII.
By contrast, a few laws include only the first type of clause, the sort that protects employees who report unlawful practices. See, e.g., 15 U.S.C. § 2651 (Asbestos Hazard Emergency Response Act); 46 U.S.C. § 80507(a) (International Safe Container Act). And a few laws include only the second type of clause, the sort that protects employees who participate in inquiries, proceedings or hearings. See, e.g., 33 U.S.C. § 948a (Longshore and Harbor Workers’ Compensation Act); 42 U.S.C. § 7622(a) (Clean Air Act).
In enacting this provision of ERISA, Congress included only a clause protecting people who give information or testify in inquiries or proceedings. Unlike most of the other laws just cited, ERISA thus does not contain a clause protecting people who oppose, report or complain about unlawful practices. “[Wjhere words differ as they differ here, Congress acts intentionally and purposely in the disparate inclusion or exclusion.” Burlington N. & Santa Fe Ry. Co. v. White, 548 U.S. 53, 63, 126 S.Ct. 2405, 165 L.Ed.2d 345 (2006). A contrary interpretation would parody the presumption against redundancy. See Maracich v. Spears, — U.S. -, 133 S.Ct. 2191, 2205, 186 L.Ed.2d 275 (2013). If clauses protecting involvement in investigations automatically protect even complaints outside investigations, dozens of opposing/reporting/complaining clauses across the United States Code would serve no purpose.
Sexton, the Secretary and the dissent insist that § 1140 covers an employee’s complaints to his employer about violations of the Act, no matter whether the employee speaks during an investigation or in response to an inquiry. Yet this argument spurns both plausible meanings of “inquiry,” whether the narrow one (an investigation) or the broader one (a question). This interpretation instead reads “in any inquiry or proceeding relating to [the Act]” out of the statute. Only then will the law protect every complaint about a violation of the Act, regardless of when and where the employee makes it. Accepting Sexton’s invitation thus means rejecting a foundational principle of interpretation: “It cannot be presumed that any clause in [a legal text] is intended to be without effect; and therefore such a construction is inadmissible, unless the words require it.” Marbury v. Madison, 1 Cranch 137, 174, 2 L.Ed. 60 (1803).
Subtracting words from the law is bad enough, but Sexton’s theory also requires us to add words in their place. As written, § 1140 does not distinguish reports of unlawful practices from other statements. The law covers any information or testimony, so long as it is given in an inquiry or proceeding relating to the Act. The law protects not only the witness who testifies in a criminal prosecution that corporate [337]*337executives defrauded employees of then-pensions, but it also protects the witness who testifies at a benefits hearing that an employee has a disability. By proposing a special rule for ERISA-related complaints, Sexton and the Secretary draw a distinction between information about violations of the Act and other information—a distinction found nowhere in the statute.
Sexton and the Secretary propose a textual detour around these problems: Because “complaining” sometimes means “challenging,” “challenging” sometimes means “questioning,” and “questioning” sometimes means “inquiring,” it follows that a complaint is an inquiry. But this line of reasoning could transform all manner of things into inquiries—everything from the Peloponnesian War (the Spartans challenged, hence questioned, Athenian supremacy) to the Boston Tea Party (the protesters challenged, hence questioned, British taxes). And it could transform complaints into all manner of things—everything from a whisper (complaining means grumbling, grumbling means murmuring, murmuring means whispering) to a shout (complaining means bewailing, bewailing means crying, crying means shouting). Not unlike the children’s game of telephone, this approach risks converting the ultimate message into something quite different from the original message—indeed sometimes into the opposite message. An approach to interpretation that exponentially blurs meaning is not apt to help in the fair interpretation of statutes. In the end, the overlap between “complain” and one meaning of “question” and the overlap between a different meaning of “question” and “inquiry” do not transform “complain” into “inquiry.”
Sexton persists that it would be strange for the law to protect an employee who reports a violation when asked a question in the context of an inquiry or proceeding but not an employee who reports a violation on his own initiative. Yet this result looks strange only if one assumes that Congress passed § 1140 in order to protect people who disclosed violations of the Act. If one instead starts with the premise that Congress passed § 1140 to prevent interference with inquiries and proceedings, the distinction between statements made in an inquiry and statements made outside an inquiry makes sense. As shown, Congress opted to protect employees in many ways in the context of investigations and unprompted complaints, some that may appear underinclusive and some that may appear overinclusive. When Congress picks one approach or the other in a given statute, that does not give a court license to rewrite the law—here by crossing “in any inquiry or proceeding” out of § 1140. Denying this point means condemning as absurd dozens of federal anti-retaliation clauses that limit their attention to inquiries, proceedings or investigations.
The Secretary offers a variation on this theme. Section 1140, he says, at least protects an employee whose complaint prompts an investigation, because giving information that leads to an inquiry counts as giving information “in” that inquiry. And it would be odd, he adds, for the law to deny an employee protection from retaliation merely because the employer later chooses not to conduct an investigation—meaning that § 1140 must cover all complaints, whether prompted by an inquiry or not. The first problem with this argument is that a disclosure that prompts an inquiry does not occur “in” that later inquiry. Deep Throat’s disclosures led to a congressional investigation of Watergate, but one would not say that Deep Throat gave information in Senator Er-vin’s inquiry. Otherwise, all disclosures or complaints would occur “in” an inquiry or proceeding, so long as the employee subsequently filed a lawsuit—the later “pro[338]*338ceeding” or “inquiry” in which the disclosure would occur. The second problem with this argument is that, even if we could bend the statute this far, we could bend it no farther. Saying that § 1140 covers complaints that lead to an inquiry at least leaves some role for the phrase “in any inquiry or proceeding.” Saying that it covers all complaints whatsoever— even when no inquiry results, as in this case—erases words from the statute.
The risk of unusual results, moreover, would not disappear even if we adopted the theory offered by Sexton and the Secretary. The law as written contains a sensible limitation on the information that enjoys federal protection: The information must be conveyed “in any inquiry or proceeding relating to [the Act].” Erasing that limit leaves a law that protects anyone who has “given information”—a boundless phrase that covers everything from giving tax advice to giving score updates on a World Cup soccer match. We could try to overcome this problem by limiting our reconfigured statute’s coverage to information that relates to the Act. But that does not help much, because a law protecting anyone who has “given information relating to the Act” would still protect the professor denied tenure because all of his law review articles concern ERISA. We could try to sidestep this problem too by limiting our doubly reconfigured statute’s coverage to disclosures about violations of the Act. But this takes us too far in the opposite direction, because now we have taken away federal protection from the witness who testifies at a benefits hearing about an employee’s disability. All of these problems disappear if we honor the boundaries that the statute already contains.
Sexton and the Secretary argue that their interpretation advances § 1140’s purpose of protecting whistleblowers from retaliation. One might quibble with Sexton’s self-portrait as a whistleblower: He did not disclose violations of the law to the public or to the authorities but threatened to sue his employer unless it did as he demanded. But even on its own terms, this argument falls short. A court must discern the purpose of a law “chiefly from its words,” Sturges v. Crowninshield, 4 Wheat. 122, 202, 4 L.Ed. 529 (1819), and the words of this law—particularly when examined in the context of other federal anti-retaliation laws—do not show that Congress meant to protect whistleblowers in the abstract. Section 1140 does not single out information about violations of the law; it covers information of any genre. The claim that Congress passed § 1140 to protect people who complain about violations of the law rests on nothing more than Sexton and the Secretary’s say-so.
Keep in mind, moreover, that Congress may have enacted this provision not so much to protect the jobs of whistleblowers as to protect the integrity of inquiries. Section 1140’s reference to “inquiries] or proceeding[s]” suggests that Congress had the latter objective in mind. We have attributed a similar purpose to a similar clause in another anti-retaliation law. See Booker v. Brown & Williamson Tobacco Co., 879 F.2d 1304, 1313 (6th Cir.1989) (discussing Title VII’s participation clause). A sense of how ERISA works shows why Congress would pursue that objective here. The Act requires retirement plans to provide for administrative review of employees’ claims. See 29 U.S.C. § 1133. Employees must usually exhaust internal remedies before heading to federal court, see Miller v. Metro. Life Ins. Co., 925 F.2d 979, 986 (6th Cir.1991), and courts often must defer to decisions made during the internal review process, see Firestone Tire & Rubber Co. v. Bruch, [339]*339489 U.S. 101, 111, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989). Given the centrality of hearings to the administration of retirement plans, one would expect Congress to protect these and other inquiries and proceedings. Instead of supporting Sexton’s claim, this understanding of congressional purpose defeats it.
Turning to the Congressional Record, the Secretary unearths a snippet of Senate debate that (he claims) analogizes ERISA’s prohibition against retaliation to Title VII’s prohibition. The relevant statement explains that § 1140 “would provide a remedy for any person fired such as is provided for a person discriminated against because of race or sex.” 119 Cong. Rec. 30044 (1973). The Secretary infers that, because Title VII protects complaints, so too does § 1140. But he overlooks the reality that Title VII includes both an opposition clause and a participation clause. Read alongside the text of § 1140, the legislative history at most shows that the drafter modeled § 1140’s language about inquiries on the participation clause, not the opposition clause. That does nothing to advance Sexton’s claim, because Title VTI’s participation clause does not protect unsolicited complaints. Abstract appeals to purpose and legislative history at any rate have little value in interpreting a “precise and reticulated” statute, Carter v. Welles-Bowen Realty, Inc., 736 F.3d 722, 729 (6th Cir.2013), a description that fits ERISA to a tee. In ERISA, of all statutes, we should hesitate to add a provision the statute does not contain (a clause protecting any employee who opposes allegedly unlawful practices) and to eliminate a limitation that the statute does contain (“in any inquiry or proceeding”).
Shifting from purpose to policy, Sexton and the Secretary claim that their interpretation encourages employees to report violations to their employers, and that enforcement of the Act depends on such reports. This last claim is debatable. The Act already requires an employer to keep records and file reports about retirement plans, 29 U.S.C. § 1059, and backs up these requirements with criminal penalties, id. § 1131(a). Also debatable is whether the benefits of a broader anti-retaliation rule (more reporting) outweigh the costs (more litigation and a greater likelihood that employers will refrain from even legitimate discharges for fear of getting sued). The prospect of improved enforcement at any rate does not by itself allow us to add new liabilities to a statute. Cf. Alexander v. Sandoval, 532 U.S. 275, 121 S.Ct. 1511, 149 L.Ed.2d 517 (2001); Central Bank of Denver, N.A. v. First Interstate Bank of Denver, N.A., 511 U.S. 164, 114 S.Ct. 1439, 128 L.Ed.2d 119 (1994).
Moving on to precedent, Sexton and the Secretary argue that case law requires an expansive interpretation of federal anti-retaliation laws and undercuts our focus on the text. Not so. When faced with “clear language,” the Supreme Court recently explained—in a retaliation case no less—a court may not “conclude that what Congress omitted from the statute is nevertheless within its scope.” Univ. of Texas Sw. Med. Ctr. v. Nassar, — U.S. -, 133 S.Ct. 2517, 2528, 186 L.Ed.2d 503 (2013). That is this case.
Nor does Kasten v. Saint-Gobain Performance Plastics Corp. change things. - U.S. -, 131 S.Ct. 1325, 179 L.Ed.2d 379 (2011). The Fair Labor Standards Act prohibits discharging an employee because he has “filed any complaint” alleging a violation of the Act, 29 U.S.C. § 215(a)(3), and the Court had to decide whether this provision covered oral as well as written complaints. The Court examined dictionaries, then legislation, regulations and judicial opinions using the [340]*340word “file,” then other appearances of “file” in the Fair Labor Standards Act, then the texts of other anti-retaliation provisions. 131 S.Ct. at 1331-33. Only because these indicators of meaning failed to yield a “conclusive answer” did the Court turn to “functional considerations,” such as the importance of oral complaints to the law’s administration. Id. at 1333-34. Far from licensing free-form interpretation of anti-retaliation laws, Kasten confirms the primacy of the statutory text.
Crawford v. Metropolitan Government of Nashville is of a piece. 555 U.S. 271, 129 S.Ct. 846, 172 L.Ed.2d 650 (2009). It held that Title VII’s opposition clause protects employees who oppose unlawful practices, regardless of whether they speak on their own initiative or in response to employer questions. The result does not help Sexton, because (unlike the law invoked in this case) Title VII’s opposition clause does not distinguish inquiries and proceedings from other settings. Nor does the reasoning help him, because Crawford turned on the “ordinary meaning” of the opposition clause, id. at 276, 129 S.Ct. 846, not on any special interpretive rule for anti-retaliation statutes. The dissent for its part takes up Crawford’s observation that “nothing in [Title VII’s opposition clause] requires a freakish rule protecting an employee who reports discrimination on her own initiative but not one who reports the same discrimination in the words when her boss asks a question.” Id. at 277-78, 129 S.Ct. 846. The dissent reasons that it makes no sense to distinguish complaints made during questioning from complaints made without questioning, prompting the conclusion that § 1140 protects all complaints whatsoever. But the conclusion does not follow from the premise. The dissent overlooks another possibility, one that both avoids odd results and stays within the words of the text: Interpret “inquiry” to mean “formal investigation” rather than “question.” That indeed is what the Fourth Circuit does. See King v. Marriott Int’l, Inc., 337 F.3d 421, 427 (4th Cir.2003).
Sexton, the Secretary and the dissent also cast about for support in our circuit’s precedents. The decisions they reel in, see, e.g., EEOC v. Ohio Edison Co., 7 F.3d 541 (6th Cir.1993); EEOC v. Romeo Cmty. Sch., 976 F.2d 985 (6th Cir.1992), do not help them. Each case interprets a statute that already contains an opposition clause to protect a form of opposition. None authorizes a court to create an opposition clause out of nothing.
The same is true of out-of-circuit decisions. Every relevant holding hurts Sexton’s position. The Third and Fourth Circuits have both held that § 1140 does not cover complaints like Sexton’s. See Edwards v. A.H. Cornell & Son, 610 F.3d 217, 225-26 (3d Cir.2010) (concluding that “unsolicited internal complaints are not protected under [§ 1140] based on a plain reading of that provision’s terms”); King v. Marriott Int’l, Inc., 337 F.3d 421, 428 (4th Cir.2003) (concluding that “the most compelling interpretation of the statutory language” excludes workplace complaints). The Second Circuit has reasoned the same way. Nicolaou v. Horizon Media, Inc., 402 F.3d 325, 329 (2d Cir.2005) (per cu-riam) (reasoning that § 1140 covers a complaint only if the employee made it in response to a “request for information”).
As for Sexton’s citation of cases from the Fifth, Seventh and Ninth Circuits, they offer less than advertised. In the case from the Fifth Circuit, an employee sued his employer under state law for wrongful discharge, alleging that it fired him for refusing to carry out orders violating ERISA. Anderson v. Elec. Data Sys. Corp., 11 F.3d 1311 (5th Cir.1994). The Fifth Circuit held that federal law preempts this state claim—a holding that [341]*341has nothing to do with the present debate. The court observed along the way, to be sure, that § 1140 “addresses ... discharges for providing information or testimony relating to ERISA.” Id. at 1314. But the court immediately added a footnote that quoted § 1140 in full. Id. at 1314 n. 2. The Fifth Circuit’s passing remark amounts to a shorthand description of § 1140, not a holding (or for that matter even a considered dictum) that § 1140 covers every complaint about a violation, even complaints delinked from an inquiry or proceeding.
The Ninth Circuit’s decision follows a similar pattern. Hashimoto v. Bank of Hawaii, 999 F.2d 408, 411 (9th Cir.1993). An employee sued her employer under a state whistleblower law, alleging that it fired her for complaining about violations of ERISA. The Ninth Circuit held that the doctrine of complete preemption converted the state claim into a federal ERISA claim. Id. at 409. This holding does not cover the issue at hand, which concerns the validity of a claim under federal law, not the propriety of recharacterizing a state claim as a federal one. Sexton and the Secretary persist that, according to the Ninth Circuit’s opinion, § 1140 “may be fairly construed to protect a person ... fired because she was protesting a violation of law in connection with an ERISA plan.” Id. at 411. But this statement was likely dictum. Even if it were a holding, this drive-by construction of the law—hinged on the self-fulfilling prophecy that “[t]his statute is clearly meant to protect whistle blowers,” id. at 411, and unhinged from the text of the provision—is not convincing.
That leaves the Seventh Circuit. It held that § 1140 covers any complaint about the Act that either asks or answers a question. George v. Junior Achievement of Cent. Ind., Inc., 694 F.3d 812, 817 (7th Cir.2012). But that decision does not help Sexton because his email did neither.
According to the dissent, the Fifth, Seventh and Ninth Circuit decisions prove at least that § 1140 contains an ambiguity sufficient to trigger Chevron deference and sufficient to require a liberal interpretation in favor of the employee. But as shown below and as the Secretary concedes, this case does not involve Chevron. And as just shown, the Ninth Circuit did not hold that § 1140 protects people like Sexton, and the Fifth and Seventh Circuits did not even suggest it. None of these circuits, moreover, evaluated § 1140 against the backdrop of other federal anti-retaliation laws. Not one of them acknowledges the distinction these laws make between opposition and participation clauses, and not one of them explains why § 1140 should protect all opposition to unlawful practices even though its text mentions only some participation in inquiries and proceedings. No less importantly, disagreements between judges at most suggest ambiguity. They do not prove it. If they did, the agency would win every circuit split about whether a federal law authorizes its regulation, but see, e.g., Ragsdale v. Wolverine World Wide, Inc., 535 U.S. 81, 122 S.Ct. 1155, 152 L.Ed.2d 167 (2002) (agency loses); the State would win every circuit split about whether a federal law preempts its statute, but see, e.g., Northwest, Inc. v. Ginsberg, — U.S. -, 134 S.Ct. 1422, 188 L.Ed.2d 538 (2014) (State loses); and the criminal defendant would win every circuit split about whether a federal law punishes his conduct, but see, e.g., United States v. Castleman, — U.S. -, 134 S.Ct. 1405, 188 L.Ed.2d 426 (2014) (criminal defendant loses).
The Secretary urges us to defer to his interpretation of the statute. He (correctly) does not claim deference under Chevron v. Natural Resources Defense Council, [342]*342which requires courts to accept an agency’s reasonable interpretation of a statute that the agency administers. 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). The Act does not give the Secretary any authority to administer § 1140, instead contemplating enforcement through private causes of action. Chevron has no place where Congress entrusts enforcement of a statute to the courts rather than to an agency. Adams Fruit Co. v. Barrett, 494 U.S. 638, 649-50, 110 S.Ct. 1384, 108 L.Ed.2d 585 (1990). The Secretary’s interpretation, moreover, appears not in a regulation but in a friend-of-the-court brief filed in this case and in similar briefs filed in other circuits. Briefs do not receive Chevron deference. United States v. Mead Corp., 533 U.S. 218, 238 n. 19, 121 S.Ct. 2164, 150 L.Ed.2d 292 (2001). All of this explains why the Secretary appeals instead to Skidmore v. Swift & Co., which entitles an agency’s views to respect proportionate to its persuasive power. 323 U.S. 134, 140, 65 S.Ct. 161, 89 L.Ed. 124 (1944). As shown, we find the Secretary’s arguments unpersuasive.
For what it is worth, our decision has few if any consequences beyond this Act. Most federal statutes that prohibit retaliation, like Title VII and the Fair Labor Standards Act, do include separate clauses protecting employees who complain about or oppose unlawful practices. 29 U.S.C. § 215(a); 42 U.S.C. § 2000e-3(a). Congress chose not to include a similar clause in ERISA, and we should respect that choice.
For these reasons, we affirm.
APPENDIX
29 U.S.C. § 1140. Interference with protected rights
It shall be unlawful for any person to discharge, fine, suspend, expel, discipline, or discriminate against a participant or beneficiary for exercising any right to which he is entitled under the provisions of an employee benefit plan, this subchapter, section 1201 of this title, or the Welfare and Pension Plans Disclosure Act, or for the purpose of interfering with the attainment of any right to which such participant may become entitled under the plan, this subchapter, or the Welfare and Pension Plans Disclosure Act. It shall be unlawful for any person to discharge, fine, suspend, expel, or discriminate against any person because he has given information or has testified or is about to testify in any inquiry or proceeding relating to this chapter or the Welfare and Pension Plans Disclosure Act. In the case of a multiemployer plan, it shall be unlawful for the plan sponsor or any other person to discriminate against any contributing employer for exercising rights under this chapter or for giving information or testifying in any inquiry or proceeding relating to this chapter before Congress. The provisions of section 1132 of this title shall be applicable in the enforcement of this section.