BOGGS, Circuit Judge. Harry C. Calcutt III, a bank executive and director, petitions for review of an order issued by the Federal Deposit Insurance Corporation (“FDIC”) that removes him from his position, prohibits him from participating in the conduct of the affairs of any insured depository institution, and imposes civil money penalties. In addition to attacking the conduct and findings in his individual proceedings, he also brings several constitutional challenges to the appointments and removal restrictions of FDIC officials.

His first hearing in these proceedings occurred before an FDIC administrative law judge (“ALJ”) in 2015. Before the ALJ released his recommended decision, the Supreme Court decided Lucia v. SEC, 138 S. Ct. 2044 (2018), which invalidated the appointments of similar ALJs in the Securities and Exchange Commission (“SEC”). The FDIC Board of Directors then appointed its ALJs anew, and in 2019 a different FDIC ALJ held another hearing in Calcutt’s matter and ultimately recommended penalties.

Broadly, Calcutt’s claims fall into two categories. First, he brings structural constitutional challenges, contending that: The FDIC Board of Directors is unconstitutionally shielded from removal by the President; the FDIC ALJs who oversee enforcement proceedings are also unconstitutionally insulated from removal; and the second hearing before a different ALJ failed to afford him a “new hearing,” as mandated by Lucia. In his second group of challenges, Calcutt attacks the procedure used and results reached in his post-Lucia adjudication. He begins by contending that the ALJ abused his discretion by curtailing cross-examination about bias of the witnesses. He then argues that the FDIC Board failed to find that he had committed misconduct that caused “effects” for Northwestern Bank, as the governing statute, 12 U.S.C. § 1818(e)(1), requires. See Dodge v. Comptroller of Currency, 744 F.3d 148, 152 (D.C. Cir. 2014).

We deny his petition. Calcutt’s challenges to the removal restrictions at the FDIC are unavailing, because even if he were to establish a constitutional violation, he has not shown that he is entitled to relief. See Collins v. Yellen, 141 S. Ct. 1761, 1789 (2021). We also conclude that his 2019 hearing satisfied Lucia’s mandate. As for the limits on cross-examination at that hearing, any error committed by the ALJ was harmless. Finally, there is substantial evidence in the record to support the FDIC Board’s findings regarding the elements of § 1818(e)(1).

I. BACKGROUND

A. Overview of FDIC Enforcement Proceedings

Among other functions, the FDIC conducts examinations and investigations to ensure banks’ safety, soundness, and compliance with statutes and regulations. See 12 U.S.C. § 1811. It has the authority to impose a range of enforcement remedies. Id. § 1818. These include removal and prohibition orders, in which the FDIC orders “an institution-affiliated party” to be removed from office or “prohibit[s] any further participation by such party, in any manner, in the conduct of the affairs of any insured depository institution.” Id. § 1818(e)(1). An institution-affiliated party includes “any director, officer, employee, or controlling stockholder (other than a bank holding company or savings and loan holding company) of, or agent for, an insured depository institution.” Id. § 1813(u)(1).

Section 8(e) of the Federal Deposit Insurance Act (“FDI Act”), 12 U.S.C. § 1818(e), as amended by the Financial Institutions Reform, Recovery, and Enforcement Act (“FIRREA”), P.L. No. 101-73, § 903, 103 Stat. 183, 453–54 (1989), provides that FDIC may remove an institution-affiliated party from office or prohibit the party from participating in conducting the affairs of any insured institution upon establishing three elements: “(1) the banker committed an improper act; (2) the act had an impermissible effect, either an adverse effect on the bank or a benefit to the actor; and (3) the act was accompanied by a culpable state of mind.” De la Fuente v. FDIC, 332 F.3d 1208, 1222 (9th Cir. 2003). First, the Board of Directors of the FDIC (“FDIC Board” or “Board”) must find that the party has committed misconduct, including engaging in “any unsafe or unsound practice in connection with any insured depository institution” or committing “any act, omission, or practice which constitutes a breach of such party’s fiduciary duty.” 12 U.S.C. § 1818(e)(1)(A)(ii)–(iii). Second, the Board must find that at least one requisite effect has occurred, i.e., that “by reason of” the party’s action, the insured depository institution “has suffered or will probably suffer financial loss or other damage,” its depositors have been or could be prejudiced, or the party has received financial gain or other benefit. Id. § 1818(e)(1)(B). Finally, the party must have had a culpable state of mind: The violation must be one that “involves personal dishonesty” or “demonstrates willful or continuing disregard by such party for the safety or soundness of such insured depository institution.” Id. § 1818(e)(1)(C).

The FDIC may also issue civil money penalties (“CMPs”) under a similar test. See 12 U.S.C. § 1818(i)(2). As relevant here, the agency may impose a “second tier” penalty of $25,000 per day of violation when a party “recklessly engages in an unsafe or unsound practice in conducting the affairs of [an] insured depository institution” or “breaches any fiduciary duty,” and that action “is part of a pattern of misconduct,” causes more than minimal loss to the institution, or benefits the institution-affiliated party. Id. § 1818(i)(2)(B).

To commence these enforcement proceedings, the FDIC first serves the party with a notice of intention to remove the party from office and/or prohibit that party from participating in other insured depository institutions. See id. § 1818(e)(1); see also id. § 1818(i)(2)(E)(i) (requiring notice for civil money penalty). The notice must contain a statement of facts establishing grounds for the removal and indicate a time and place for a hearing. Id. § 1818(e)(4). The institution-affiliated party may then appear at the hearing to contest the notice; failure to appear constitutes consent to the order. Ibid.

An ALJ conducts the adversarial hearing in accordance with the Administrative Procedure Act (“APA”), 5 U.S.C. §§ 551–559. See 12 U.S.C. § 1818(h)(1) (requiring hearings to be “conducted in accordance with the provisions of chapter 5 of Title 5”). Under the applicable regulations, an ALJ presiding over a removal proceeding has “all powers necessary to conduct a proceeding in a fair and impartial manner and to avoid unnecessary delay,” 12 C.F.R. § 308.5(a), including the power to “receive relevant evidence and to rule upon the admission of evidence and offers of proof,” id. § 308.5(b)(3); “[t]o consider and rule upon all procedural and other motions appropriate in an adjudicatory proceeding . . . ,” id. § 308.5(b)(7); and “[t]o prepare and present to the Board of Directors a recommended decision,” id. § 308.5(b)(8).

The regulations also provide that evidence that would be admissible under the Federal Rules of Evidence is admissible in adjudicatory proceedings, id. § 308.36(a)(2), and that except as otherwise provided, “relevant, material, and reliable evidence that is not unduly repetitive is admissible to the fullest extent authorized by the Administrative Procedure Act and other applicable law,” id. § 308.36(a)(1). If evidence meets this latter standard but would be inadmissible under the Federal Rules of Evidence, the ALJ may not deem the evidence inadmissible. Id. § 308.36(a)(3).

After the hearing, the ALJ must file and certify a record of the proceeding, including a recommended decision, recommended findings of fact, recommended conclusions of law, and a proposed order. Id. § 308.38(a). A party then has thirty days to file written exceptions for the FDIC Board’s review objecting to particular matters or omissions in the ALJ’s recommendations, but a failure to file an exception on a particular matter is treated as a waiver of that objection, and the Board need not consider any such objections that were not initially raised before the ALJ. Id. § 303.39.

The Board then reviews the ALJ’s recommendations and issues a final decision. Id. § 308.40. Its review is “based upon review of the entire record of the proceedings,” although it may limit its review to those arguments and exceptions that were raised by the parties. Id. § 308.40(c)(1). After the Board’s final decision, a party may petition for review in the United States Court of Appeals for the District of Columbia Circuit or the circuit in which the institution’s home office is located. ^[12] U.S.C. § 1818(h)(2).

B. FDIC Composition and Structure

The FDIC Board consists of five members: the Comptroller of the Currency, the Director of the Consumer Financial Protection Bureau (“CFPB”), and three additional directors who are appointed by the President with the advice and consent of the Senate. ^[12] U.S.C. § 1812(a)(1). The Comptroller of the Currency and the CFPB Director are also appointed by the President with Senate advice and consent. Id. § 2 (Comptroller of the Currency); id. § 5491(b)(2) (CFPB Director). The Board also incorporates a measure of partisan balancing, with a maximum of three directors permitted to be members of the same political party. Id. § 1812(a)(2).

The three members of the Board not appointed by virtue of another office serve fixed terms, and the parties agree that they are not removable at will. During the proceedings before the ALJs in this case, the CFPB Director also enjoyed for-cause protection from removal under 12 U.S.C. § 5491(c)(3); however, before the Board issued its final order, the Supreme Court held this removal restriction to be unconstitutional. Seila Law LLC v. Consumer Fin. Prot. Bureau, 140 S. Ct. 2183 (2020). The Comptroller of the Currency’s term lasts for five years “unless sooner removed by the President, upon reasons to be communicated by him to the Senate,” and Calcutt concedes that this provision provides for at-will removal. ^[12] U.S.C. § 2. In practice, however, the FDIC Board has had several vacancies during the proceedings in Calcutt’s case; additionally, at least one board member continued to serve after his term expired until a successor was appointed. See 12 U.S.C. § 1812(c)(3) (providing for continuation of service of appointed members after expiration of term before a successor is appointed).

The ALJs who hear FDIC removal and prohibition proceedings are part of a pool housed in the Office of Financial Institution Adjudication (“OFIA”), an interagency body established by FIRREA that presides over enforcement proceedings brought by the FDIC, the Office of the Comptroller of the Currency (“OCC”), the Board of Governors of the Federal Reserve System (“FRB”), and the National Credit Union Administration (“NCUA”). See FIRREA § 916, 103 Stat. 183, 486–87 (codified at 12 U.S.C. §1818 note); 12 C.F.R. § 308.3 (defining OFIA as “the executive body charged with overseeing the administration of administrative enforcement proceedings” of OCC, FRB, FDIC, and NCUA).1 These agencies signed an agreement that provides for cost-sharing and specifies that the FDIC is the “Host Agency,” responsible for the employment of an office staff consisting of ALJs and administrative employees. See Ex. L to Emergency Motion for Stay Pending Review, at 1–6. The agreement also states: “Any change to the Office Staff personnel shall be subject to the prior written approval of all Agencies.” Id. at 3. Two ALJs currently make up the pool in OFIA. See Our Judges, Office of Financial Institution Adjudication, https://www.ofia.gov/who-we-are/our-judges.html (last visited May 24, 2022).

Because Calcutt seeks relief for a constitutional violation, the Constitution provides the place to start on this remedies question. But it says almost nothing about remedies. Cf. Hernandez v. Mesa, 140 S. Ct. 735, 741–43 (2020); Armstrong v. Exceptional Child Ctr., Inc., 575 U.S. 320, 324–27 (2015). Except for a few provisions like the requirement to pay “just compensation” for a taking, see Knick v. Township of Scott, 139 S. Ct. 2162, 2171 (2019), the Constitution sets only limits on government conduct without prescribing specific relief for violations, see Alfred Hill, Constitutional Remedies, 69 Colum. L. Rev 1109, 1118 (1969). One thus will search Article II in vain for an explicit constitutional remedy that applies to an invalid removal provision.

Where else should we look? The founders enacted the Constitution against the backdrop of a preexisting legal system with preexisting causes of action and remedies. See id. at 1131–32. Before the founding, for example, this system often allowed equity courts to issue injunctions to stop “illegal executive action[.]” Armstrong, 575 U.S. at 327; Ex parte Young, 209 U.S. 123, 150–51 (1908). The Supreme Court has held that we may use these preexisting “judge-made” remedies to redress constitutional wrongs unless Congress displaces them. Armstrong, 575 U.S. at 327–28.

But courts should not take this allowance too far. The Constitution does not give us freewheeling power to adopt federal common-law remedies based on our views of wise policy. See Hernandez, 140 S. Ct. at 741–42 (citing Erie R.R. Co. v. Tompkins, 304 U.S. 64, 78 (1938)). And the Court “disfavor[s]” remedies that are rooted in legislative-like choices about the best way to deter illegal acts. Ziglar v. Abbasi, 137 S. Ct. 1843, 1857 (2017) (citation omitted).

This dichotomy points the way here. We lack an inherent power to treat the FDIC’s actions as “void” because we think it would be a good idea. See Hernandez, 140 S. Ct. at 741– 42. We instead must look to the causes of action and remedies that traditionally applied to claims like Calcutt’s—that a statutory provision related to an office was illegal and that this defect rendered the officer’s actions void. When courts traditionally chose remedies for this sort of claim, they distinguished between two types of officers: a “de facto officer” in a lawful office (whose actions were enforceable) and a “mere usurper” in an unlawful one (whose actions were void). Albert Constantineau, A Treatise on the De Facto Officer Doctrine §§ 5, 34, at 8–10, 52–53 (1910).

De Facto Officer in Lawful Office. For centuries, parties have alleged that an officer was unlawfully holding (and performing the duties of) an office. To give an example at the time of the founding, a party claimed that a sheriff could not hold that office because the sheriff had not lived in the county as long as the law required. State v. Anderson, 1 N.J.L. 318, 324–28 (N.J. 1795).

English courts channeled these claims into a specific writ (“quo warranto”) with a specific remedy (prospectively ousting the officer). See 3 William Blackstone, Commentaries[*262] –64; 2 Edward Coke, Institutes of the Laws of England 282, 494–99 (1642). American courts followed suit. Constantineau, supra, § 451, at 635 n.1; State v. Parkhurst, 9 N.J.L. 427, 437–38 (N.J. 1802). Three aspects of the quo warranto action deserve mention. For one, invalid officers caused public harms, so the government itself typically needed to sue them. See Wallace v. Anderson, 18 U.S. 291, 292 (1820). Yet private parties could sue on the government’s behalf if they showed a unique interest. See Newman v. United States ex rel. Frizzell, 238 U.S. 537, 549–51 (1915). For a second, the remedy was exclusive. Constantineau, supra, § 451, at 635. A party disputing an officer’s authority could not sue for an injunction “to restrain the exercise of official functions[.]” Floyd R. Mecham, A Treatise on the Law of Public Offices and Officers § 478, at 307 (1890). For a third, the remedy exists today. See D.C. Code § 16-3503. Parties may ask the Attorney General to seek this relief or request leave of court to seek it themselves— a process that may look “cumbersome” to modern eyes. Andrade v. Lauer, 729 F.2d 1475, 1497–98 (D.C. Cir. 1984) (Wright, J.).

Yet the process has always looked cumbersome. Rather than file a direct quo warranto suit to oust invalid officers, parties harmed by the officers’ actions have tried to collaterally attack their qualifications in suits involving the actions. Id. at 1496. Since 1431, English courts have rebuffed these attacks under the “de facto officer doctrine.” Constantineau, supra, § 5, at 8–10 (citing The Abbé de Fontaine, 1431 Y.B. ^[9] Hen. VI, fol. 32, pl. ^[3] (Eng.)); Clifford L. Pannam, Unconstitutional Statutes and De Facto Officers, 2 Fed. L. Rev. 37, 39–42 (1966). That doctrine treats the past actions of an officer with a colorable claim to office as valid whether or not the officer met all conditions to hold the office. Constantineau, supra, § 1, at 3–4. English courts introduced it “into the law as a matter of policy and necessity, to protect the interests of the public and individuals, where those interests were involved in the official acts of persons exercising the duties of an office, without being lawful officers.” State v. Carroll, 38 Conn. 449, 467 (1871).

American courts likewise adhered to the de facto officer doctrine as a corollary to the exclusive quo warranto remedy. See Cocke v. Halsey, 41 U.S. 71, 81–88 (1842); Taylor v. Skrine, 5 S.C.L. 516, 516–17 (S.C. 1815); Fowler v. Bebee, 9 Mass. 231, 234–35 (1812); People ex rel. Bush v. Collins, 7 Johns. 549, 554 (N.Y. 1811) (per curiam). Notably, these courts upheld the actions of invalid officers who did not meet constitutional conditions on their offices. An officer might not have taken an oath. Cf. Bucknam v. Ruggles, 15 Mass. 180, 182–83 (1818) (per curiam). Or the officer might have been appointed in an illegal way. Cf. Ex parte Ward, 173 U.S. 452, 454 (1899). Or the officer might have flunked an eligibility requirement. Perhaps the officer was too young. Cf. Blackburn v. State, 40 Tenn. 690, 690–91 (1859). Or maybe the officer had been in the Congress that increased the office’s salary before taking office. Cf. U.S. Const. art. I, § 6, cl. ^[2]; William Baude, The Unconstitutionality of Justice Black, 98 Tex. L. Rev. 327 (2019); In re Griffin, 11 F. Cas. ^[7], 27 (C.C.D. Va. 1869) (No. 5,815). The same rules applied even if the officer held the office by reason of an unconstitutional statute. See Constantineau, supra, §§ 192–96, at 264–70. An early decision thus upheld the acts of an officer who had been appointed by the governor under a statute authorizing this appointment, even though the state constitution had required the legislature to elect the officer. See Taylor, 5 S.C.L. at 516–17; Carroll, 38 Conn. at 474; see also State v. McMartin, 43 N.W. 572, 572 (Minn. 1889); Ex Parte Strang, 21 Ohio St. 610, 615–18 (1871); cf. Buckley v. Valeo, 424 U.S. ^[1], 142 (1976) (per curiam).

Usurper in Unlawful Office. Other times, parties have alleged that a generic office could not exist because it had been assigned “sovereign functions” that it could not possess. Mecham, supra, § 4, at 5. In one case, for example, a party alleged that a legislatively created “court” could not perform judicial duties because those duties had been vested in a wrongly abolished life-tenured court. Hildreth’s Heirs v. McIntire’s Devisee, 24 Ky. 206, 207–08 (1829); Jeffrey S. Sutton, Who Decides? States as Laboratories of Constitutional Experimentation 76–80 (2022).

Courts granted much broader relief for this type of claim. Parties affected by an illegal office did not need to sue in quo warranto to dispute the officeholder’s power to perform the challenged function. Parties instead could dispute the officer’s conduct “in any kind” of suit. Walcott v. Wells, 24 P. 367, 370 (Nev. 1890); Mecham, supra, §§ 324–26, at 216–18. And the opposing party could not defend the officer’s past acts using the de facto officer doctrine. Constantineau, supra §§ 34–36, at 51–55. The officer instead was “merely a usurper, to whose acts no validity can be attached[.]” Norton v. Shelby County, 118 U.S. 425, 449 (1886).

This rule extended to constitutional defects. The Supreme Court may have followed it as early as United States v. Yale Todd (U.S. 1794). United States v. Ferreira, 54 U.S. 40, 52–53 (1851) (note by Taney, C.J.). This unreported case addressed a law allowing pensions for disabled Revolutionary War veterans. The law ordered circuit courts to determine whether applicants were disabled and to send their findings to the Secretary of War. Circuit judges (including Supreme Court Justices) found that the law unconstitutionally gave courts executive power by making them the Secretary’s administrators. Hayburn’s Case, 2 U.S. 408, 410 n.* (1792). Given the law’s benevolent goals, though, some judges awarded pensions by claiming to act as “commissioners.” See Wilfred J. Ritz, United States v. Yale Todd (U.S. 1794), 15 Wash. & Lee L. Rev. 220, 228–29 (1958). Congress ordered the Attorney General to seek Supreme Court review of pensions granted by judges “styling themselves commissioners.” Act of Feb. 28, 1793, 1 Stat. 324, 325. In Yale Todd’s case, the Court required him to return the funds. Ritz, supra, at 228–30. As others have noted, the Court may well have found the judges’ actions void because they unconstitutionally undertook executive functions. Ferreira, 54 U.S. at 53 (note by Taney, C.J.); Keith E. Whittington, Judicial Review of Congress before the Civil War, 97 Geo. L.J. 1257, 1270–74 (2009).

Many decisions followed this remedial approach for claims that a legislative body had granted functions to an office that it could not lawfully possess. See Town of Decorah v. Bullis, 25 Iowa 12, 18–19 (1868); Hildreth’s Heirs, 24 Ky. at 207–08; G. L. Monteiro, Annotation, De Jure Office as Condition of De Facto Officer, 99 A.L.R. 294 § III(a) (1935), Westlaw (database updated 2022). When, for example, a legislature assigned local-government functions to a board of commissioners that the state constitution vested in justices of the peace, the Supreme Court treated the board’s actions as void. Norton, 118 U.S. at 441–49. It refused to apply the de facto officer doctrine because that doctrine required a valid (“de jure”) office. Id. at 444–45.

The Supreme Court’s modern cases also treat an officer’s actions as void if the generic office could “not lawfully possess” the power to take them. Collins, 141 S. Ct. at 1788. The Court thus found invalid a bankruptcy judge’s decision in a suit that an Article III court needed to resolve. See Stern v. Marshall, 564 U.S. 462, 503 (2011). And a plurality rejected the de facto officer doctrine when a party claimed that Congress assigned to Article I judges a duty (sitting on circuit courts) that Article III judges must perform. See Glidden Co. v. Zdanok, 370 U.S. 530, 535–37 (1962) (plurality opinion); cf. Bowsher v. Synar, 478 U.S. 714, 732 (1986); Young v. United States ex rel. Vuitton et Fils S.A., 481 U.S. 787, 815 (1987) (Scalia, J., concurring in the judgment).

This “long history of judicial review” has relevance for Calcutt’s request that we vacate the FDIC’s order in his case because invalid removal protections shielded two of its officers. Armstrong, 575 U.S. at 327. To begin with, the history refutes the theory that the Constitution of its own force compels courts to treat as “void” any action taken by officers whose exercise of an office does not comport with a constitutional command. That view would treat the de facto officer doctrine itself as unconstitutional. Yet it formed part of the legal backdrop against which the founders enacted the Constitution. Near the founding, judges described the doctrine as “a well settled principle of law,” Bush, 7 Johns. at 554, or “too well established to admit of a doubt,” Taylor, 5 S.C.L. at 517. Nothing in the Constitution can be read to do away with it.

This history also highlights the key inquiry for deciding whether courts may vacate an officer’s actions as a “judge-made remedy” when a statute unconstitutionally limits the President’s removal authority. Armstrong, 575 U.S. at 327. Does the unconstitutional removal provision show that Congress vested “sovereign functions” in an invalid office that cannot possess them? Mecham, supra, § 4, at 5; Norton, 118 U.S. at 449. If so, courts should treat the officer’s actions as void wherever they arise. Or is the removal provision “distinct from the provisions creating the . . . office” such that the office itself is valid “even assuming that the [removal provision] is” not? McMartin, 43 N.W. at 572; Carroll, 38 Conn. at 449. If so, courts should enforce the officer’s acts in suits involving third parties (in contrast to suits between the government and the officer).

Unfortunately for Calcutt, his claim falls on the wrong side of this divide. He does not even argue that the two executive officers (the CFPB Director and administrative law judge) sat in offices that constitutionally “could not exist” (because, for example, the Constitution vested their duties in another branch). Ashley v. Bd. of Supervisors of Presque Isle Cnty., 60 F. 55, 65 (6th Cir. 1893). Indeed, his argument’s very premise—that Congress has illegally insulated the officers from the President—assumes that they perform executive functions. Cf. Seila Law, 140 S. Ct. at 2209. So I would treat the constitutional “condition” in this case (that an officer be accountable to the President) like other constitutional conditions the violation of which does not void an officer’s acts. The condition is not much different than, say, a condition that an officer be of a certain age, see Blackburn, 40 Tenn. at 690–91, or be elected rather than appointed, see Constantineau, supra, § 192, at 264–65. If statutes departing from these mandates did not render an officer’s actions void, I fail to see why an unconstitutional removal provision would. Under traditional remedial principles, then, Calcutt could not obtain the relief that he seeks in this case.

The “lack of historical precedent” to attack removal provisions in a suit like Calcutt’s reinforces the conclusion that the provisions did not traditionally render an officer’s actions void. Seila Law, 140 S. Ct. at 2201 (citation omitted). If any private party could collaterally attack removal provisions in any suit implicating an officer’s acts, one would expect to see many of these suits. After all, Congress began to enact constitutionally dubious removal provisions shortly after the Civil War during President Johnson’s administration. See Myers, 272 U.S. at 166–73. Yet Calcutt cites no historical example in which courts evaluated removal provisions in this type of litigation. So constitutional questions about the provisions lingered for decades. Id. at 173.

Challenges to the validity of removal provisions instead arose in employment disputes. See Humphrey’s Executor, 295 U.S. at 618–19; Myers, 272 U.S. at 106; cf. Shurtleff, 189 U.S. at 311–12; Reagan v. United States, 182 U.S. 419, 424 (1901); Ex parte Hennen, 38 U.S. 230, 256– 57 (1839). A discharged officer would sue to recover a salary (or seek reinstatement) on the ground that the termination violated a tenure-protection statute. Myers, 272 U.S. at 106. The government would respond that the statute could not restrict the President’s power. Id. This different kind of suit required courts to resolve the constitutional question. Courts “almost universally recognized” that the de facto officer doctrine did not apply because the suit was between the government and the officer (not a third party) and because only valid officers could receive salaries. Constantineau, supra, § 236, at 331; 2 James Kent, Commentaries on American Law 355 n.2 (11th ed. 1867).

Modern precedent confirms my conclusion. The Supreme Court’s recent cases have all held that unconstitutional removal provisions do not render the office to which they attach invalid or require courts to find actions taken by the officers void. See Collins, 141 S. Ct. at 1787–89; Seila Law, 140 S. Ct. at 2207–11; Free Enter. Fund, 561 U.S. at 508–10. Take Free Enterprise Fund. There, accountants under investigation by the Public Company Accounting Oversight Board filed an Ex Parte Young suit seeking to enjoin all of the Board’s actions as void because of its removal protections. See 561 U.S. at 487, 491 n.2, 508. The Court agreed that various removal provisions unconstitutionally intruded on the President’s authority. Id. at 492– 98. But it refused to treat the Board’s actions as void. Id. at 508–10. It held that the Board could perform the executive functions assigned to it despite the invalid removal provisions because they were “severable from the remainder of the statute.” Id. at 508. The Court analyzed this issue in terms of “severability.” See id. at 509. But it could just as well have reasoned that the unconstitutional statutes did not render the Board’s actions void in third-party suits and so did not entitle the accountants to their requested remedy. Cf. McMartin, 43 N.W. at 572; Harrison, supra, at 73–75.

Seila Law fits a similar mold. The CFPB in that case issued a civil investigative demand seeking documents from a law firm. 140 S. Ct. at 2194. The firm refused to comply, so the CFPB filed a petition to enforce its demand. Id. The district court rejected the firm’s request to deny the CFPB’s petition on the ground that its Director’s removal protections rendered all CFPB actions void. Id. After agreeing that the protections were unconstitutional, the controlling Supreme Court opinion again held that the invalid provisions were severable and did not render all CFPB actions void. Id. at 2208–11 (opinion of Roberts, C.J.). Admittedly, the opinion did not simply reject the law firm’s remedy and affirm the enforcement of the CFPB’s demand. Rather, it remanded for the lower courts to decide whether the demand had been “validly ratified” by a Director accountable to the President. Id. at 2211. This resolution might have implied that all CFPB actions (including the investigative demand) had been void prior to the Court’s severance “remedy.” Id. at 2208. But the Court has since clarified that Seila Law did not hold that the CFPB’s prior actions were invalid and instead had left all remedy-related issues for the lower courts. See Collins, 141 S. Ct. at 1788.

Most recently in Collins, the Court expressly held that unconstitutional removal provisions do not render an officer’s past actions void in suits by third parties. Headed by a director with removal protections, the agency in Collins served as the conservator to two large mortgage-financing companies. 141 S. Ct. at 1771–72. This agency entered into agreements with the Department of Treasury requiring the companies to pay large dividends to the Treasury. Id. at 1772–74. The companies’ shareholders sued to compel the Treasury to return the dividends on the ground that the director’s removal protections were unconstitutional and that they voided the agency’s past acts (including the challenged agreements). Id. at 1775. Although the Court agreed that the removal protections were unconstitutional, id. at 1783–87, it rejected the broad remedy, id. at 1787–89. The Court found “no reason to regard any of the actions taken by the” agency “as void” simply because its head had been protected by invalid removal provisions. Id. at 1787.

All told, under traditional remedial rules, unconstitutional removal provisions do not render the offices to which they attach invalid and so do not allow courts to vacate the actions of officers as void in suits by third parties. This tradition compels me to reject Calcutt’s proposed remedy.

I end with two disclaimers about things I need not decide. Disclaimer One: Congress may generally displace judge-made remedial principles. Armstrong, 575 U.S. at 327–29. Congress, for example, has sometimes restricted a court’s power to grant Ex Parte Young’s injunctive relief for violations of federal law. See id. And Bowsher teaches that Congress may adjust the relief for structural constitutional claims too. There, the Court followed the statutory remedy once it found that Congress had illegally entrusted a legislative officer with executive duties. 478 U.S. at 734–35. Congress thus may permit courts to vacate actions taken by officers subject to unconstitutional removal protections even if traditional judge-made remedial limits would foreclose relief.

Has Congress done so here? The FDIC’s statute incorporates the APA. ^[12] U.S.C. § 1818(h)(2). It orders a court to “hold unlawful and set aside agency action” that is “contrary to constitutional right, power, privilege, or immunity[.]” 5 U.S.C. § 706(2)(B). Perhaps this text could be read to allow courts to depart from traditional limits and vacate agency “actions” if a law has structured the agency in a way that is “contrary to constitutional right” or “power.” Id.; cf. Collins, 141 S. Ct. at 1795 (Gorsuch, J., concurring in part). That the Constitution’s structural principles exist to protect individual liberty could reinforce this reading that a structural problem is “contrary to constitutional right” within the meaning of the APA. See Bond v. United States, 564 U.S. 211, 220–24 (2011).

In most structural constitutional cases, however, a private party claims that the challenged action itself is “contrary to constitutional right.” 5 U.S.C. § 706(2)(B). So parties routinely allege that a prosecution violates the Constitution because the relevant law reaches conduct that Congress may not proscribe. See, e.g., Bond, 564 U.S. at 224. Yet, as I have explained, an unconstitutional removal statute for an office would not necessarily render the officer’s “actions” void and so would not necessarily render those actions “contrary to constitutional right.” 5 U.S.C. § 706(2)(B). Perhaps the APA’s text is thus best read to incorporate—not depart from—traditional remedial limits. Cf. id. § 702; Tom C. Clark, Att’y Gen.’s Manual on the Admin. Proc. Act 108 (1947).

And even if the APA expanded the available relief, recall that it requires courts to take “due account” “of the rule of prejudicial error.” 5 U.S.C. § 706. The Court has read this text to adopt the harmless-error principles that “ordinarily apply in civil cases.” Shinseki v. Sanders, 556 U.S. 396, 406 (2009). Under those principles, constitutional errors can be harmless. See O’Neal v. McAninch, 513 U.S. 432, 440 (1995). Although Collins did not cite the APA, this harmless-error provision might be one way to understand its suggestion that third parties could seek relief for unconstitutional removal provisions if they showed that the provisions harmed them (that is, if they showed that the error was not harmless). 141 S. Ct. at 1788–89. At day’s end, I would leave these statutory questions open. The parties did not address the APA’s scope and focused only on whether the removal provisions rendered the FDIC’s order unconstitutionally void. They did not.

Disclaimer Two: The parties assume that the FDIC performs only executive functions. Our resolution should not be taken to have impliedly adopted that premise. The FDIC did not just prosecute this action. It also adjudicated the action—finding Calcutt guilty and imposing a punishment on him in the form of an end to his career and a $125,000 penalty. Once an Article III court finally enters the picture, moreover, it may review the FDIC’s factual findings only under a deferential substantial-evidence test—a test that has been called more deferential than the one governing our review of a district court’s factual findings. See Dickinson v. Zurko, 527 U.S. 150, 153 (1999).

Yet both Article III and the Due Process Clause generally require the government to follow common-law procedure (including, fundamentally, the use of a “court”) when seeking to deprive people of their private rights to property or liberty. See Stern, 564 U.S. at 482–84; Caleb Nelson, Adjudication in the Political Branches, 107 Colum. L. Rev. 559, 569–70 (2007). At first blush, one might think that the FDIC has sought to deprive Calcutt of his “core private rights” to both. B&B Hardware, Inc. v. Hargis Indus., Inc., 575 U.S. 138, 171 (2015) (Thomas, J., dissenting). According to Blackstone, Calcutt had a “property” interest in the thousands of dollars that the government seeks to take. See 1 Blackstone, supra, at[*134] –35. According to Coke, he had a “liberty” interest in continuing in his profession. See 2 Coke, supra, at 47. So perhaps the FDIC has undertaken judicial functions here—functions that the Constitution vests in courts. See Stern, 564 U.S. at 482–84. If the FDIC needed to file suit, moreover, the filing would have triggered the Seventh Amendment’s right to a jury, which Justice Brennan made clear applies to suits seeking civil penalties. See Tull v. United States, 481 U.S. 412, 422–25 (1987).

The government traditionally has responded to this call for more “process” with the defense that its action seeks to vindicate “public rights,” rights that need not be litigated in a court with a jury. See Oil States Energy Servs., LLC v. Greene’s Energy Grp., LLC, 138 S. Ct. 1365, 1373 (2018); Atlas Roofing Co. v. Occupational Safety & Health Rev. Comm’n, 430 U.S. 442, 450–51 (1977). And maybe Calcutt did not raise this argument here because a healthy amount of caselaw has accepted that defense in the banking context. See Cavallari v. Off. of Comptroller of Currency, 57 F.3d 137, 145 (2d Cir. 1995); Simpson v. Off. of Thrift Supervision, 29 F.3d 1418, 1422–24 (9th Cir. 1994). Yet this precedent predates the Court’s recent instructions in cases like Stern, which held that the adjudication of a state tort claim required an Article III court. See 564 U.S. at 487–501. And while Stern did not involve an agency, the Court “recognize[d]” that its cases may not provide “concrete guidance” on the scope of the public-rights doctrine in the administrative context. Id. at 494. Several Justices have also expressed concern with extending the doctrine too far. See Oil States, 138 S. Ct. at 1381–85 (Gorsuch, J., joined by Roberts, C.J., dissenting); B & B Hardware, 575 U.S. at 170–74 (Thomas, J., joined by Scalia, J., dissenting).

There must be some limit to the government’s ability to dissolve the Constitution’s usual separation-of-powers and due-process protections by waiving a nebulous “public rights” flag at a court. When the government indicts a person for a crime, it also vindicates “public rights” that belong to the community. Spokeo v. Robins, 578 U.S. 330, 345 (2016) (Thomas, J., concurring) (citing Ann Woolhandler & Caleb Nelson, Does History Defeat Standing Doctrine?, 102 Mich. L. Rev. 689, 695–700 (2004)). But the government cannot send people to prison using a hearing room rather than a court room or an administrative officer rather than a jury of peers. N. Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U.S. 50, 70 n.24 (1982) (plurality opinion). Why should this case be different simply because Calcutt must pay a civil penalty rather than a criminal fine? Cf. Jarkesy v. SEC, __ F.4th __, 2022 WL 1563613, at *2–7 (5th Cir. May 18, 2022). The FDIC one day must provide answers to these questions in a case that does not assume them.

C. Remedy for Appointments Clause Violation

Calcutt lastly challenges the FDIC’s remedy for an undisputed constitutional wrong. The Appointments Clause sets the ground rules for the appointment of officers. U.S. Const. art. II, § 2, cl. ^[2]. It allows Congress to vest the power to appoint inferior officers in “the President,” “Courts of Law,” or “Heads of Departments.” Id. In Lucia v. SEC, 138 S. Ct. 2044 (2018), the Court held that the SEC’s administrative law judges are inferior officers who must be appointed by the President or the Commission. Id. at 2051–55. The parties agree that the FDIC’s administrative law judges are likewise inferior officers, but Calcutt litigated his first hearing before a judge who had not been appointed by the President or FDIC. The FDIC thus granted Calcutt a “new” hearing before a different, lawfully appointed judge—the remedy that Lucia ordered. See id. at 2055. Calcutt argues that this remedy still fell short because the FDIC allowed the second judge to use records, stipulations, and orders from the invalid judge’s first hearing. According to him, the Appointments Clause required the second judge to ignore everything that occurred before.

To decide what Lucia meant by its “new hearing” remedy, my colleagues engage in a cost-benefit balance that resembles the Supreme Court’s test for whether a court should suppress evidence in a criminal trial under the Fourth Amendment’s “exclusionary rule.” Davis v. United States, 564 U.S. 229, 236–38 (2011). They point out that Calcutt’s remedy would impose heavy administrative costs (because it would require inefficient, duplicative processes). They add that it would offer few private benefits (because it is unnecessary to insulate the valid judge’s decision from the first hearing’s “taint”). Based on this prudential balancing, they reject Calcutt’s claim that the second judge had to ignore items from the first hearing. Their balance seems reasonable enough. But I would reject Calcutt’s view of Lucia based on structural grounds rooted in the best reading of the Appointments Clause and the Court’s current approach to judge-made remedies.

At the outset, I do not mean to critique my colleagues for engaging in a cost-benefit inquiry. The Supreme Court’s instructions in Appointments Clause cases may well be read to contemplate it. See Lucia, 138 S. Ct. at 2055 & nn.5–6; Ryder v. United States, 515 U.S. 177, 182–83 (1995). In Ryder, a court-martialed member of the Coast Guard had his conviction upheld by a panel that included judges whose appointments violated the Appointments Clause. 515 U.S. at 179–80. The Court of Military Appeals affirmed the panel’s conviction under the de facto officer doctrine. Id. at 180. The Supreme Court reversed and refused to apply this doctrine. It held that “one who makes a timely challenge to the constitutional validity of the appointment of an officer who adjudicates his case is entitled to a decision on the merits of the question and whatever relief may be appropriate if a violation indeed occurred.” Id. at 182–83. Did Ryder look to the “original meaning” of the Appointments Clause to adopt this remedy and reject the de facto officer doctrine? Fin. Oversight & Mgmt. Bd. for P.R. v. Aurelius Inv., LLC, 140 S. Ct. 1649, 1659 (2020). No, the Court rested on a sentence of pure policy: “Any other rule would create a disincentive to raise Appointments Clause challenges[.]” Ryder, 515 U.S. at 183. The Court summarily found the “proper” remedy to be a second appeal before a lawfully constituted panel. See id. at 188.

Lucia followed the same reasoning. It noted that Ryder called for a new hearing before a properly appointed administrative law judge. 138 S. Ct. at 2055. It then added a new requirement: an agency may not assign the case to the judge who initially heard it even if that judge had been properly appointed in the interim. Id. When responding to the claim that this “new judge” remedy was not needed to further the Appointments Clause’s purposes, the Court reasoned that its remedies in this area have been “designed not only to advance those purposes directly, but also to create ‘[]incentive[s] to raise Appointments Clause challenges.’” Id. at 2055 n.5 (quoting Ryder, 515 U.S. at 183). In both cases, therefore, the Court chose a remedy to “incentivize” these claims.

This reasoning should look familiar. The Court once expansively created judge-made remedies that would best promote the purposes of constitutional rights. Although, for example, Congress has allowed damages claims only against state officers who violate the Constitution, 42 U.S.C. § 1983, the Court felt free to create a remedy allowing parties to seek damages from federal officers who violate the Fourth Amendment. See Bivens v. Six Unknown Fed. Narcotics Agents, 403 U.S. 388, 395–96 (1971). And although the Fourth Amendment says nothing about the rules of evidence in criminal trials, the Court created the exclusionary rule to “remov[e] the incentive to disregard” its ban on unreasonable searches. Mapp v. Ohio, 367 U.S. 643, 656 (1961) (citation omitted). Ryder bears the hallmarks of Bivens and Mapp. It even discussed the exclusionary rule. The Court noted that its cases have rejected that rule when the rule’s costs (allowing criminals to go free) exceeded its benefits (incentivizing officers to obey the law). See Ryder, 515 U.S. at 185–86 (citing United States v. Leon, 468 U.S. 897 (1984)). Analogizing to this approach, Ryder foresaw no ill effects from granting an Appointments Clause remedy on direct appeal and suggested that this appellate relief would create “incentives to make such challenges.” Id. at 186.

Although Ryder might mesh well with Mapp, the Court in recent years has treated these types of judge-made innovations with a healthy dose of skepticism. See Hernandez, 140 S. Ct. at 747. The creation of remedies amounts to “lawmaking” that must balance the benefits of any remedy against its costs. Id. at 741–42. Yet the Constitution reserves this task to Congress, not the courts. See id. As a result, the Court has all but held that Bivens was wrong and has refused to extend it to any other constitutional right for some 40 years. See id. at 742–43 (citing cases); Abbasi, 137 S. Ct. at 1856–58. It has also continued to narrow the scope of the exclusionary rule, acknowledging that it is a “judicially created remedy” that must be applied cautiously only in cases of clear police misconduct. Davis, 564 U.S. at 238 (citation omitted); see, e.g., Utah v. Strieff, 579 U.S. 232, 237–38, 241 (2016); Herring v. United States, 555 U.S. 135, 140–44 (2009).

What do these principles mean for the issue that confronts us? I agree that Ryder and Lucia leave open whether a lawful judge at a “new ‘hearing’” may rely on evidence developed at the invalid hearing or on orders entered by the invalid judge. Lucia, 138 S. Ct. at 2055 (quoting Ryder, 515 U.S. at 182–83). To resolve the ambiguity, I would read the cases in a way that best comports with the Constitution’s “original meaning,” Aurelius, 140 S. Ct. at 1659, and with the Court’s recent guidance to act cautiously before expanding judge-made remedies, Hernandez, 140 S. Ct. at 747. When analyzed in that fashion, the FDIC’s remedy more than sufficed.

The Appointments Clause does not compel Calcutt’s conclusion that a valid judge must ignore all prior proceedings before an invalid one. If anything, the clause itself requires no remedy. The de facto officer doctrine broadly applied to claims like Calcutt’s that an officer had been appointed by the wrong person. See Constantineau, supra, §§ 182–86, at 248–55. An English judge who sat on the first case to enforce the doctrine in 1431 “apparently recognized” its application in this setting. Id. § 182, at 248. American courts routinely relied on it when an officer was unconstitutionally appointed by, say, the governor rather than the legislature, see Carroll, 38 Conn. at 474 (discussing Taylor, 5 S.C.L. at 516–17), or the mayor rather than the governor, see Strang, 21 Ohio St. at 615–19. And if the Constitution requires some way in which to dispute an officer’s right to an office, Congress left open the traditional (if narrow) quo warranto remedy. D.C. Code § 16-3503; cf. Henry M. Hart, Jr., The Power of Congress to Limit the Jurisdiction of Federal Courts: An Exercise in Dialectic, 66 Harv. L. Rev. 1362, 1366–67 (1953).

Ryder and Lucia thus must rest on a power to create judge-made remedies for constitutional violations. But we must act with caution when asked to expand these remedies because the weighing of the costs and benefits amounts to a legislative task, not a judicial one. See Abbasi, 137 S. Ct. at 1856–57. On the benefits side, Calcutt’s remedy would certainly promote the purposes of the Appointments Clause. See United States v. Arthrex, Inc., 141 S. Ct. 1970, 1979 (2021). But no provision—not even a constitutional one—“pursues its purposes at all costs.” Hernandez, 140 S. Ct. at 741–42 (citation omitted). And Calcutt’s remedy comes with its burdens too. It would add to the “administrative costs” already associated with the new hearings. Abbasi, 137 S. Ct. at 1856. More fundamentally, courts long recognized that permitting parties to challenge an officer’s validity at all in appeals of the officer’s actions could create “endless confusion[.]” Norton, 118 U.S. at 441–42; see Constantineau, supra, § 4, at 7. That is why they channeled these challenges into special suits that would oust officers only prospectively, not into appeals that would reverse their actions retrospectively. See Constantineau, supra, § 451, at 635–36. I see no judicial mode of analysis that can resolve this legislative weighing of interests.

All told, the Court’s cautious approach to judge-made remedies comports with traditional remedial practice governing challenges to the validity of an officer’s appointment. See Hernandez, 140 S. Ct. at 742. I thus would not read Ryder and Lucia broadly to compel administrative judges to disregard all that occurred at a prior hearing. I would instead read them literally to compel a new hearing before a properly appointed judge. Calcutt got just that.

III. Statutory Claims

In my view, Calcutt’s statutory claims fare better. The statute allowing the FDIC to bar bankers from the industry requires it to prove three things: that a banker has engaged in a listed kind of misconduct, that the misconduct will harm the bank (or benefit the banker), and that the banker acted with a culpable state of mind. 12 U.S.C. § 1818(e)(1)(A)–(C). The statute allowing the FDIC to impose penalties largely covers the same terrain. Id. § 1818(i)(2)(B). Here, Calcutt argues that the FDIC failed to prove the “misconduct” and “effect” elements. I agree that the FDIC misread these provisions and would remand for it to reconsider the case under the proper law.

A. Misconduct

To remove Calcutt from the Bank, the FDIC first must prove that he engaged in one of three types of misconduct. Id. § 1818(e)(1)(A). Specifically, the statute allows the FDIC to remove an “institution-affiliated party” if that the party “has, directly or indirectly”:

(i) violated—

(I) any law or regulation;

(II) any cease-and-desist order which has become final;

(III) any condition imposed in writing by a Federal banking agency in connection with any action on any application, notice, or request by such depository institution or institution-affiliated party; or (IV) any written agreement between such depository institution and such agency;

(ii) engaged or participated in any unsafe or unsound practice in connection with any insured depository institution or business institution; or (iii) committed or engaged in any act, omission, or practice which constitutes a breach of such party’s fiduciary duty[.]

Id. The FDIC found that Calcutt violated the second and third clauses by engaging in “unsafe or unsound practice[s]” and committing “breach[es]” of his “fiduciary duty.” App. 18–26. (It imposed the $125,000 penalty for the same reasons. See App. 35.)

1. Unsafe or Unsound Practice. The statute gives the FDIC the power to ban a banker from the profession if the banker has “engaged or participated in any unsafe or unsound practice in connection with any insured depository institution or business institution[.]” 12 U.S.C. § 1818(e)(1)(A)(ii). Regulators have long defined the key phrase—“unsafe or unsound practice”—using a two-part test that courts have generally accepted. See First Nat’l Bank of Eden v. Dep’t of Treasury, 568 F.2d 610, 611 n.2 (8th Cir. 1978). Under this test, an act qualifies as an unsafe or unsound practice if it conflicts with “generally accepted standards of prudent operation” and creates an “abnormal risk of loss or harm” to the bank. App. ^[18] (quoting Michael v. FDIC, 687 F.3d 337, 352 (7th Cir. 2012)).

This test was not intuitive to me from a review of the text, so I looked into its origins. One court transparently identified its source: “Because the statute itself does not define an unsafe or unsound practice, courts have sought help in the legislative history.” In re Seidman, 37 F.3d 911, 926 (3d Cir. 1994). The Fifth Circuit started down this path. See Gulf Fed. Sav. & Loan Ass’n v. Fed. Home Loan Bank Bd., 651 F.2d 259, 263–65 (5th Cir. 1981). Rather than seek out the ordinary meaning of “unsafe or unsound practice,” it jumped to a “lively” debate in the congressional record. Id. at 263. During this debate, the court noted, a few legislators had treated as “authoritative” a definition proposed by an agency chairman. Id. at 264. Under the chairman’s view, the phrase covered “any action” that “is contrary to generally accepted standards of prudent operation, the possible consequences of which, if continued, would be abnormal risk or loss or damage to an institution, its shareholders, or the agencies administering the insurance funds.” Id. (citation omitted). The court accepted his view as law. Id. at 264–65.

This straight-from-the-legislative-history test has spread widely since. The few courts with reasoned analysis regurgitate the same bit of legislative history. Seidman, 37 F.3d at 926. Most others, though, simply cite other precedent for this test without considering its origins. See Frontier State Bank v. FDIC, 702 F.3d 588, 604 (10th Cir. 2012); Michael, 687 F.3d at 352; Landry v. FDIC, 204 F.3d 1125, 1138 (D.C. Cir. 2000); Simpson, 29 F.3d at 1425; Doolittle v. Nat’l Credit Union Ass’n, 992 F.2d 1531, 1538 (11th Cir. 1993); Nw. Nat’l Bank v. Dep’t of Treasury, 917 F.2d 1111, 1115 (8th Cir. 1990).

I am troubled by this approach. The test springs from a mode of interpretation that no Justice on the Supreme Court would endorse today. In recent decades, the Court has given us clear marching orders: the answer to an interpretive question begins by identifying the ordinary meaning of Congress’s words when read against their context and structure. See Food Mktg. Inst. v. Argus Leader Media, 139 S. Ct. 2356, 2364 (2019); Ross v. Blake, 578 U.S. 632, 638 (2016). This “first canon” is also the “last” if the text has a clear meaning. Conn. Nat’l Bank v. Germain, 503 U.S. 249, 254 (1992). Here, however, courts have viewed the legislative history as both the beginning and the end of the analysis. Gulf Federal even claimed that the agency chairman’s proposed test had been “adopted in both Houses”—by which the court meant that it had been read into the legislative record. 651 F.2d at 264 (citation omitted). “But legislative history is not the law.” Epic Sys. Corp. v. Lewis, 138 S. Ct. 1612, 1631 (2018). And the Court has not been kind to other tests that developed in this manner. See, e.g., Food Mktg., 139 S. Ct. at 2364.

I am also troubled by this approach because courts have chosen it to create a “flexible” statute allowing regulators to address “changing business problems[.]” Seidman, 37 F.3d at 927. What does this even mean? If an agency condones a banker’s “new business model,” the agency can constrict the statute to give the banker a pass? Henson v. Santander Consumer USA Inc., 137 S. Ct. 1718, 1725–26 (2017). But if the agency disapproves of a competitor’s practice, it can expand the statute to punish the competitor? This accordion-like view of the rule of law has no place in our constitutional order—one in which the President lacks any “dispensing” prerogative. Cf. Clark v. Martinez, 543 U.S. 371, 382 (2005); McConnell, supra, at 115–19. If anything, this view has things backwards. This statute can deprive citizens of their property and livelihoods. So it would better align with our interpretive traditions if we construed the phrase “strictly” rather than flexibly. ^[1] Blackstone, supra,[*88] ; United States v. Wiltberger, 18 U.S. 76, 95 (1820). After all, the rule of lenity (the rule that we resolve ambiguities against the government) historically applied not just to criminal laws, but also to all laws considered “penal”—“that is, laws inflicting any form of punishment” like a civil penalty. Wooden v. United States, 142 S. Ct. 1063, 1086 n.5 (2022) (Gorsuch, J., concurring in the judgment). This statute fits that bill. See Proffitt v. FDIC, 200 F.3d 855, 860–62 (D.C. Cir. 2000). At the least, courts should give a phrase that affects core private rights its ordinary meaning—not a malleable one.

How might an ordinary banker interpret the phrase? The legislative history reaches any “imprudent act.” Seidman, 37 F.3d at 932; see Gulf Federal, 651 F.2d at 264. Yet this definition does not adequately account for two parts of the actual text. For starters, the statute uses the word “practice,” not “act.” 12 U.S.C. § 1818(e)(1)(A)(ii). Those words mean different things. If an otherwise conscientious banker makes a single imprudent loan to a couple down on their luck, the banker might have engaged in an unsound “act.” But nobody would say that the banker has made it a “practice” of issuing bad loans after just the one. This word includes a connotation of repetition (of habitual acts). The banker must have a habit of making bad loans (or, at the least, the bank must have that habit and the banker must “participate[] in” it). Id.; cf. Nw. Nat’l Bank, 917 F.2d at 1115. That is because a “practice” is a “habitual or customary performance,” American College Dictionary 951 (1970), or a “habitual or customary action or way of doing something,” American Heritage Dictionary of the English Language 1028 (1973).

The statute itself contemplates this distinction. One clause bars bankers from engaging in “any unsafe or unsound practice[.]” 12 U.S.C. § 1818(e)(1)(A)(ii). The next bars them from engaging in “any act, omission, or practice” that breaches their fiduciary duties. Id. § 1818(e)(1)(A)(iii) (emphases added). We presume that Congress meant different things when it used different words in clauses that sit right next to each other. See Nat’l Ass’n of Mfrs. v. Dep’t of Def., 138 S. Ct. 617, 631 (2018). So even a single act or omission “that breaches [a] fiduciary duty” suffices for punishment, but only a habit of “unsafe or unsound” actions does.

Next, the statute does not cover every unsafe or unsound practice in the abstract. Rather, the practice must be “in connection with” a bank. ^[12] U.S.C. § 1818(e)(1)(A)(ii). The Supreme Court has recognized that this phrase has an “indeterminat[e]” scope. Maracich v. Spears, 570 U.S. 48, 59–60 (2013); see Mont v. United States, 139 S. Ct. 1826, 1832 (2019). If we read it broadly here, it could cover practices with the remotest of relations to banking—such as a banker’s decision to speed to work every morning. See Maracich, 570 U.S. at 59. One regulator even thought that the phrase covered a decision to seek judicial review of the regulator’s own regulatory decision. Johnson v. Off. of Thrift Supervision, 81 F.3d 195, 202 (D.C. Cir. 1996). Could Calcutt’s decision to file a petition in this court also be an “unsound practice” because we reject his appeal? I would not read the statute this broadly. Courts instead must interpret the clause to adopt the “limiting principle” that best comports with the statute’s context and structure. See Maracich, 570 U.S. at 59–60; Chadbourne & Parke LLP v. Troice, 571 U.S. 377, 387–91 (2014).

For the reasons that a D.C. Circuit decision has explained, I would read this clause to cover only “unsafe or unsound banking practices.” Grant Thornton, LLP v. Off. of the Comptroller of the Currency, 514 F.3d 1328, 1332–33 (D.C. Cir. 2008). This definition “harmonizes” this subsection with the rest of § 1818. Id. at 1332. The section includes several other provisions that regulate unsafe or unsound practices “in conducting the business” of a bank, including one permitting the FDIC to issue cease-and-desist orders. 12 U.S.C. § 1818(b)(1). It would be odd to permit a limited remedy (a cease-and-desist order) only for unsound banking practices but a severe remedy (removal from a bank) for any unsound practice with any connection to the bank. And a definition that covered only “banking” practices would exclude, for example, an outside auditor’s deficient audit, see Grant Thornton, 514 F.3d at 1332– 33, or a decision to seek judicial review.

All of this said, courts that apply a broad legislative-history test have recognized that their reading could lead to “open-ended supervision.” Gulf Fed., 651 F.2d at 265. So they compensate by adding a limiting principle that I do not necessarily see in the text either. They have read the phrase “unsafe or unsound practice” to require that an action pose a risk of extreme harm—one that threatens the bank’s “financial stability,” Seidman, 37 F.3d at 928, or “integrity,” Johnson, 81 F.3d at 204 (quoting Gulf Fed., 651 F.2d at 267). An “unsafe” practice (one that exposes the bank to “danger or risk”) may well require a risk of some harm. ^[2] Oxford Universal Dictionary 2312 (3d ed. 1968). But the statute also covers an “unsound” practice in the disjunctive (a practice that is “not based on proven practice, established procedure, or practical knowledge”). Webster’s New International Dictionary 2511 (3d ed. 1966). Perhaps the entire phrase “unsafe or unsound” may be one of those “doublets” that Congress uses to convey a single idea (like “aid and abet” or “cease and desist”). Doe v. Boland, 698 F.3d 877, 881 (6th Cir. 2012) (citing Freeman v. Quicken Loans, Inc., 566 U.S. 624, 635–36 (2012)). Even still, I would not think that this text requires the risk of financial collapse. A loan officer at a massive bank who has followed a consistent pattern of making bad loans may have engaged in an “unsafe or unsound practice” even if the banker’s portfolio cannot threaten the bank’s existence.

Be that as it may, I would save the required financial-risk level for another appeal. When sanctioning Calcutt here, the FDIC did not apply my reading that the statute requires unsafe or unsound banking practices. I would remand for it to do so in the first instance. Most notably, the FDIC nowhere indicated that it must identify a banking “practice” as I read the phrase—i.e., a “habitual or customary action[.]” American Heritage, supra, at 1028. To the contrary, as Calcutt notes, the vast majority of its findings relied on a single loan—the Bedrock Transaction. It concluded, among other things, that Calcutt violated the Bank’s lending policies and engaged in imprudent lending by approving that transaction. App. 19–21. It is not clear that Calcutt’s actions with respect to this loan can rise to the level of an unsafe or unsound “practice.” This fact contrasts Calcutt’s case with those that the FDIC cited—which involved a pattern of bad loans. See, e.g., First State Bank of Wayne Cnty. v. FDIC, 770 F.2d 81, 82–83 (6th Cir. 1985).

2. Breach of Fiduciary Duty. The statute also gives the FDIC the authority to ban a banker from the profession if the banker has “committed or engaged in any act, omission, or practice which constitutes a breach of such party’s fiduciary duty[.]” 12 U.S.C. § 1818(e)(1)(A)(iii). The parties’ briefing on this portion of the statute raises more questions in my mind than it answers.

Start with a choice-of-law question. Citing Atherton v. FDIC, 519 U.S. 213 (1997), my colleagues and Calcutt suggest that the relevant state’s corporate-governance law supplies the rule of decision for determining whether a banker has breached a “fiduciary duty” within the meaning of § 1818(e)(1)(A)(iii). (The FDIC does not enlighten us with its position on this choice-of-law subject.) I am skeptical that their reading is correct. The relevant portion of Atherton that they cite was not interpreting federal statutory language like the “fiduciary duty” text in § 1818(e). It was rejecting the claim that purely federal common law should supply the “corporate governance standards” for federally chartered entities. See 519 U.S. at 217–26. Here, by contrast, we must determine the proper “interpretation of a federal statute,” not whether we may create federal common law. Id. at 218. And when a federal statute uses a common-law term of art, the Supreme Court generally interprets its language to adopt a uniform standard of conduct for all 50 states based on generic common-law concepts. See, e.g., Burlington Indus., Inc. v. Ellerth, 524 U.S. 742, 754–55 (1998); Cmty. for Creative Non-Violence v. Reid, 490 U.S. 730, 739–41 (1989). I might take that approach here. It would likely mean that we should interpret this phrase to codify the well-known duties of care and loyalty as they existed in this corporate-governance context at the time that Congress adopted this language in 1966. See, e.g., Harry G. Henn, Handbook of the Law of Corporations and Other Business Enterprises 362–70 (1961); William J. Grange & Thomas C. Woodbury, Corporation Law: Operating Procedures for Officers and Directors § 268, at 286–87, § 311, at 325–26 (2d ed. 1964); Dow Votaw, Modern Corporations 63–64 (1965); Harold Koontz, The Board of Directors and Effective Management 84–86 (1967).

Turn to the substantive standards. The Board held that Calcutt had breached his duty of care to the Bank by acting incompetently in his approval of the Bedrock Transaction and in his failure to manage the Nielson Loans. App. 23–24. But from my review of the FDIC’s order, I cannot even determine the substantive standards of conduct that it applied. Its order did not use the words “negligence” or “gross negligence.” And for decades, courts have debated which of these standards the statute incorporates. Julie Andersen Hill & Douglas K. Moll, The Duty of Care of Bank Directors and Officers, 68 Ala. L. Rev. 965, 986–92 (2017). The Board also neglected to mention the traditional “business-judgment rule,” the application of which is also contested. Patricia A. McCoy, A Political Economy of the Business Judgment Rule in Banking: Implications for Corporate Law, 47 Case W. Res. L. Rev. ^[1], 22–60 (1996). Yet another layer in this morass is that in the 1980s, Congress also adopted a “gross negligence” floor to govern the conduct of officers and directors in a related context. ^[12] U.S.C. § 1821(k); see Atherton, 519 U.S. at 226–28. That separate section’s implications for § 1818(e) are unclear.

Yet I would not authoritatively answer these choice-of-law or substantive questions now. As I explain below, I would remand to allow the FDIC to reconsider whether Calcutt’s misconduct was the cause of any of the claimed harms. On remand, I would give the FDIC a chance to clarify its views on these legal questions about the meaning of this fiduciary-duty statute.

B. Causation

The statute next requires the FDIC to prove either that Calcutt’s misconduct had the potential to harm the Bank or that Calcutt received a benefit from that misconduct. See 12 U.S.C. § 1818(e)(1)(B). This “effect” subparagraph provides in full:

(B) by reason of the violation, practice, or breach described in any clause of subparagraph (A)—

(i) such insured depository institution or business institution has suffered or will probably suffer financial loss or other damage;

(ii) the interests of the insured depository institution’s depositors have been or could be prejudiced; or (iii) such party has received financial gain or other benefit by reason of such violation, practice, or breach[.]

Id. The specific civil-penalty provisions on which the FDIC relied required similar “effects.” See id. § 1818(i)(2)(B)(ii)(II)–(III); App. 34–35.

The FDIC misinterpreted the causation element in this subparagraph. To show why, I start with the causation basics. The common law has long recognized two types of causation: factual (or “but for”) causation and legal (or “proximate”) causation. See William L. Prosser, Handbook of the Law of Torts §§ 45–46, at 311, 321–22 (1941). But-for causation creates a simple rule. As its name suggests, it requires a plaintiff to show that an injury would not have occurred “but for” the defendant’s wrongful conduct. See Burrage v. United States, 571 U.S. 204, 211–12 (2014); Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 347 (2013). Suppose, for example, that after a neighbor’s dam breaks and floods a plaintiff’s property, the plaintiff sues the neighbor for building the dam negligently. See Restatement (Second) of Torts § 432 illus. ^[2] (Am. L. Inst. 1965). But-for causation requires a court to ask whether the plaintiff would have suffered this injury (the flooding) in a counterfactual world in which the neighbor did not commit the wrongful act (the negligent construction). See id. § 432(1) & cmt. a. And if a once- in-a-century storm would have caused the flooding even if the neighbor had built the dam to perfection, the negligent construction did not cause the harm. See id. § 432 illus. ^[2]; Burrage, 571 U.S. at 211–12.

Proximate causation arose from the premise that a factual-cause test alone would lead to excessive liability. Prosser, supra, § 45, at 312. Courts recognized that, “[i]n a philosophical sense, the consequences of an act go forward to eternity, and the causes of an event go back to the discovery of America and beyond.” Id. They thus adopted “proximate cause” rules to cut off liability even if a defendant was a but-for cause of harm. Holmes v. Secs. Inv. Prot. Corp., 503 U.S. 258, 268 (1992). As one example, a defendant’s conduct (say, its failure to keep a ship docked) may set in motion a chain of events that leads another party to negligently cause an injury (say, the captain incompetently runs the ship aground). See Exxon Co., U.S.A. v. Sofec, Inc., 517 U.S. 830, 832–34 (1996). Under a superseding-cause test, courts will not hold the defendant liable if the other party’s negligence was unforeseeable. Id. at 837. As another example, a defendant’s misconduct (say, stock manipulation) may directly harm one person (a stockbroker who goes bankrupt) and indirectly harm third parties (the stockbroker’s creditors). See Holmes, 503 U.S. at 262–63. Under a directness test, courts will not allow the third parties to recover. Id. at 271–72.

These common-law rules have significance in this case. The Supreme Court presumes that Congress enacts statutory text with common-law concepts in mind. See Lexmark Int’l, Inc. v. Static Control Components, Inc., 572 U.S. 118, 132 (2014). It thus has long read common-law causation rules into statutes that use causal language like “because of” or “results from.” See Burrage, 571 U.S. at 213–14; Nassar, 570 U.S. at 350–52. Congress used one such phrase (“by reason of”) here. The FDIC must prove that the Bank suffered (or will likely suffer) a loss or that Calcutt received a benefit “by reason of” his misconduct. ^[12] U.S.C. § 1818(e)(1)(B). So I would interpret this statute to require both but-for and proximate causation. See Comcast Corp. v. Nat’l Ass’n of African American-Owned Media, 140 S. Ct. 1009, 1015 (2020); Holmes, 503 U.S. at 265–67.

But the FDIC has not adopted these causation rules. Its enforcement orders have all but ignored but-for cause. In fact, I have found only one such order that even used this phrase. See In re Adams, 1997 WL 805273, at *5 (F.D.I.C. Nov. ^[12], 1997). And it suggested that a “‘but for’ relationship” was not required. Id. (quoting ABKCO Music, Inc. v. Harrisongs Music Ltd., 772 F.2d 988, 995–96 (2d Cir. 1983)). The FDIC also failed to mention but-for cause in this case. It simply indicated: “An actual loss is not required; a potential loss is sufficient so long as the risk of loss to the Bank was ‘reasonably foreseeable’ to someone in [Calcutt’s] position.” App. 26 (citations omitted). The FDIC is correct that, unlike most statutes imposing liability for harm, this statute does not require a past loss. It also applies if a bank “will probably suffer” a loss in the future “by reason of” the banker’s misconduct. ^[12] U.S.C. § 1818(e)(1)(B)(i). But it incorporates but-for cause all the same. For a past loss, the FDIC must show that it “would not have occurred without” the misconduct. Nassar, 570 U.S. at 347 (citation omitted). For a future loss, the FDIC must show that the probability of loss would not have occurred without that misconduct. See id. The FDIC’s jurisprudence leaves no hint that it adheres to these first-year torts-class concepts.

The FDIC’s legal error is all the more pronounced for proximate causation. For years, it has rejected outright any need to prove this causation. See Adams, 1997 WL 802573, at *5; In re ***, 1985 WL 303871, at[*114] (F.D.I.C. Aug. ^[19], 1985). It did so in this case too, noting that “an individual respondent need not be the proximate cause of the harm to be held liable[.]” App. 26–27. Confusingly, however, the FDIC suggested that the loss needs to be “foreseeable.” App. 26, 31. Foreseeability is one component of the proximate-causation requirement that the FDIC said it was rejecting. See Hemi Grp., LLC v. City of New York, 559 U.S. ^[1], 12 (2010). If the FDIC meant to imply that the statute incorporates only proximate cause’s foreseeability element, it still erred. Proximate causation contains a group of concepts other than foreseeability. See id. So the Supreme Court has already rejected this type of argument that a federal statute contains only a foreseeability test. See Bank of Am. Corp. v. City of Miami, 137 S. Ct. 1296, 1305–06 (2017).

*

Maybe we could overlook the FDIC’s failure to identify the governing causation law if it correctly applied that law to Calcutt. But it did no such thing. The FDIC held Calcutt responsible for three injuries to the Bank and one benefit to him. The Bank incurred $6.443 million in charge-offs from the Nielson Loans. App. 27–29. It incurred a $30,000 charge-off from the $760,000 Bedrock Transaction. App. 27. And it paid its lawyers and accountants for work related to these loans. App. 29–31. Calcutt lastly received dividends from the Bank’s holding company despite the loans’ poor condition. App. 31–32. None of these “effects” sufficed.

As an initial matter, I agree with my colleagues that the FDIC failed to explain why the statute should even cover fees paid to lawyers or accountants. The statute reaches “financial loss or other damage” from Calcutt’s misconduct. ^[12] U.S.C. § 1818(e)(1)(B)(i). It would be unusual to describe the money paid for these services as “financial loss” or “other damage.” One does not normally use such terms to describe a payment of money for something of commensurate value. Cf. Summit Valley Indus. Inc. v. Loc. 112, United Brotherhood of Carpenters & Joiners of Am., 456 U.S. 717, 722–23 (1982). The payment is more naturally described as an “expense” or “cost.” Our country’s litigation traditions reinforce this view. We have long followed the “American Rule” in which a plaintiff’s legal costs are not recoverable “damages” even if the defendant’s conduct is their but-for cause. See Alyeska Pipeline Serv. Co. v. Wilderness Soc’y, 421 U.S. 240, 249–50 (1975) (citing Arcambel v. Wiseman, 3 U.S. 306 (1796)). When a statute allows a plaintiff to recover “damages,” then, courts do not read that phrase to cover attorney’s fees—or other expert fees for that matter. See Summit Valley, 456 U.S. at 722–23; cf. W. Va. Univ. Hosps., Inc. v. Casey, 499 U.S. 83, 88–92 (1991). And the Court has stuck with this rule even if a law uses a phrase (“expenses”) that is “capacious enough to include” these fees. Peter v. Nantkwest, Inc., 140 S. Ct. 365, 372 (2019). So I would not read the text “loss” or “damage” to cover them here.

That leaves the other three “effects.” The FDIC did not apply basic causation rules to any of them. Most tellingly, the FDIC held Calcutt responsible for all $6.443 million in charge-offs on the $38 million in Nielson Loans—that is, for the entire loss. App. 27–28; see id. at 6–7. But these loans were underwater in the aftermath of the Great Recession before Calcutt committed most of the identified misconduct. App. 625–26. As with my hypothetical about the negligently made dam, then, the FDIC needed to ask a “counterfactual” question: How much in charge-offs would the Bank have incurred if Calcutt had not engaged in that misconduct? Comcast, 140 S. Ct. at 1015. Suppose that the (hopefully) once-in-a-century recession would have caused $7 million in charge-offs if the Bank started collection efforts immediately because of the collapsed real-estate market. If so, a decision to enter into the Bedrock Transaction would have helped (not harmed) the Bank. And Calcutt’s misconduct (for example, the failure to undertake the usual underwriting efforts, see App. ^[19]) could not be described as a but-for cause of loss. I see nothing in the record on appeal that would help answer this critical but-for question, confirming that the FDIC did not even ask it.

The same error underlies the FDIC’s decision to hold Calcutt liable for the $30,000 charge-off for the Bedrock Transaction. App. 27. The FDIC did not consider the “counterfactual” of what would have occurred if Calcutt had not engaged in misconduct. Comcast, 140 S. Ct. at 1015. As a generic matter, the Bank suffered a total of $6.473 million in charge-offs on all Nielson Loans (including the Bedrock Transaction) and the FDIC needed to consider the amount of likely charge-offs without this transaction. Would it have lost more? Less? The FDIC did not ask these questions. More granularly, Calcutt told the FDIC that the administrative law judge had erred “by failing to tether the $30,000 charge-off (and other actual and potential losses) to specific acts of misconduct[.]” App. 27. The judge found, for instance, that Calcutt breached his fiduciary duty of candor to the Bank’s directors by failing to seek their preapproval for the Bedrock Transaction. App. 25–26. Suppose the directors would have approved the transaction even if he had done so. How could this specific misconduct have caused this harm? The FDIC responded that it was “unpersuaded” by this causation argument because the Bedrock Transaction was a “main focus” of the hearing and the judge catalogued Calcutt’s many misdeeds in approving it. App. 27. This (non)response said nothing about causation—an element distinct from misconduct.

Both but-for and proximate-cause problems undergird the FDIC’s decision that Calcutt benefited from his misconduct. He was the largest shareholder of the Bank’s holding company, and the FDIC held that his misconduct allowed him to obtain a dividend from this company. App. 31–32. Its conclusion rested on the administrative law judge’s finding that the Bank paid its own shareholder (the holding company) a $462,950 dividend in mid-2011 and that the FDIC would not have approved this payment (to the holding company) if it had known that the Nielson Loans were not performing. App. 287, 751. As a matter of but-for causation, the FDIC did not ask whether the holding company would have paid its shareholders the same dividend even if the FDIC had known of the Nielson Loans’ true condition. See Comcast, 140 S. Ct. at 1015. It cites no testimony from the company’s directors about what they would have done. And Calcutt testified that the holding company had sufficient assets to pay the dividend even if the Bank had paid it nothing. A580.

As a matter of proximate causation, the FDIC failed to consider a “directness” issue. If “by reason of” incorporates usual proximate-cause rules, it would require that Calcutt directly benefit from his misconduct. Under the FDIC’s theory, though, the holding company was the direct beneficiary that received the dividend; Calcutt was an indirect beneficiary as a shareholder of that separate company. Is this a sufficiently “direct” benefit (analogous to a larger salary)? “The general tendency” in the law has been “not to go beyond the first step.” Bank of Am., 137 S. Ct. at 1306 (citation omitted). And this theory potentially rests on the “independent” decision of the holding company. Hemi, 559 U.S. at 15. But I would leave this question for the FDIC.

All told, I would remand for the FDIC—the fact finder—to apply the correct causation rules to the two charge-offs and the dividend payment in the first instance. My colleagues recognize many of the FDIC’s legal errors but say there is no need to remand. I disagree. They first invoke the deferential substantial-evidence test. But that test governs our review of the agency’s factual findings. See Dickinson, 527 U.S. at 162. I do not quibble with those. I take issue with the FDIC’s failure to follow the proper causation law. The substantial-evidence test has nothing to say on that subject. And even the FDIC does not claim that we should defer to its legal views. See Grant Thornton, 514 F.3d at 1331; cf. Epic, 138 S. Ct. at 1629–30.

If anything, my colleagues’ analysis runs afoul of basic administrative-law principles. When an agency’s decision rests on a collapsed legal foundation, we cannot affirm the decision on the ground that the agency might have reached the right outcome under a correct legal view. We must let the agency apply the proper law in the first instance. See Gonzales v. Thomas, 547 U.S. 183, 186 (2006) (per curiam); SEC v. Chenery Corp., 318 U.S. 80, 88 (1943); Henry J. Friendly, Chenery Revisited: Reflections on Reversal and Remand of Administrative Orders, 1969 Duke L.J. 199, 209–10. But my colleagues all but find facts by applying their view of the law to the record. Recall, for example, that the FDIC held Calcutt liable for all $6.443 million in charge-offs on the Nielson Loans—a finding that leaves no doubt that the agency erred. My colleagues do not defend this finding. They nevertheless say that the FDIC “could have” found that Calcutt’s misconduct caused some unquantified percentage of the losses. Maj. Op. 48. But this “judicial judgment cannot be made to do service for an administrative judgment.” Chenery, 318 U.S. at 88.

Even if we could now find Calcutt liable for an (unknown) loss amount on a good- enough-for-government-work approach, I would still remand. The statute says that the FDIC “may” seek to remove a banker—not that it must do so—when the other requirements are met. ^[12] U.S.C. § 1818(e)(1). It thus leaves the FDIC with discretion over whether to bar Calcutt “from working in his chosen profession for the remainder of his career.” Doolittle, 992 F.2d at 1538. The amount of harm properly chargeable to Calcutt should influence its discretionary decision. The FDIC found removal proper after holding Calcutt responsible for well over $8 million (including professional fees and charge-offs). If, on remand, the FDIC were to find that Calcutt’s conduct caused a tiny fraction of this harm, it might reconsider its “draconian” sanction. Id. In fact, this logic led the Eleventh Circuit to remand a similar removal order so that a related agency could reconsider the order after the court jettisoned part of its reasoning. Id. Even a case that my colleagues cite issued this type of remand when it upheld only part of the FDIC’s order—given the “extraordinary” nature of the sanction. De la Fuente v. FDIC, 332 F.3d 1208, 1227 (9th Cir. 2003). Because the FDIC’s order is riddled with legal error, I find it inexplicable that we are not doing so here.

* * *

For these reasons, I respectfully dissent.