CHAD A. READLER, Circuit Judge.

Appeal from the United States District Court for the Southern District of Ohio at Columbus. No. 2:20-cv-00945—Algenon L. Marbley, Chief District Judge. ARGUED: Tony C. Merry, LAW OFFICES OF TONY C. MERRY, Worthington, Ohio, for Appellant. Byrne J. Decker, OGLETREE DEAKINS NASH SMOAK & STEWART, P.C., Portland, Maine, for Appellee. ON BRIEF: Tony C. Merry, LAW OFFICES OF TONY C. MERRY, Worthington, Ohio, for Appellant. Byrne J. Decker, OGLETREE DEAKINS NASH SMOAK & STEWART, P.C., Portland, Maine, Ann-Martha Andrews, OGLETREE DEAKINS NASH SMOAK & STEWART, P.C., Phoenix, Arizona, for Appellee. Before: COLE, NALBANDIAN, and READLER, Circuit Judges. READLER, J., delivered the opinion of the court in which COLE and NALBANDIAN, JJ., joined. NALBANDIAN, J. (pp. 316-24), delivered a separate concurring opinion.

OPINION

CHAD A. READLER, Circuit Judge.

Vickie Tranbarger's quality of life declined precipitously after a routine surgery. She later left her job and claimed disability benefits, which her insurer denied. Tranbarger challenged that decision in district court. Resolution of the case turned on Tranbarger's ability to demonstrate complete and continuous disability during the six months following her resignation. After reviewing the administrative record, the district court concluded that Tranbarger failed to prove as much. Accordingly, the court granted judgment to the insurer. We now affirm.

I.

For years, Vickie Tranbarger led an active lifestyle. An avid cyclist, she rode 10 miles a day. She also traveled extensively. Unfortunately, those activities were curtailed following an operation to remove her gallbladder. After her surgery, Tranbarger began suffering from a host of medical conditions, including physical pain and chronic fatigue syndrome. At work, Tranbarger continued in her role as an accounts receivable manager, a primarily sedentary position. Her supervisor modified some of her responsibilities to accommodate her reduced capacity. But even with these modifications, Tranbarger eventually resigned her post in July 2016, citing pain and fatigue.

Through her employer, Tranbarger was enrolled in a disability insurance plan operated by Lincoln Life & Annuity Company of New York. About 14 months after resigning, Tranbarger filed a claim for long-term disability benefits with Lincoln. Under her plan, Tranbarger was entitled to benefits if she could show "total disability" such that she was "unable to perform each of the [m]ain [d]uties of . . . her [o]wn [o]ccupation"— an accounts receivable manager—during a six-month "Elimination Period" following her resignation. Tranbarger presented various forms of evidence demonstrating her physical limitations, including a Social Security ruling in her favor, doctors' notes, and statements from individuals otherwise familiar with her condition.

Following its review, Lincoln denied Tranbarger's claim. She responded with this lawsuit. Her complaint alleged that Lincoln's denial of benefits violated the Employee Retirement Income Security Act of 1974 (ERISA). With cross-motions for judgment on the administrative record before it, the district court sided with Lincoln, prompting this appeal.

II.

Did Tranbarger shoulder her burden of demonstrating a continuous inability to perform the main duties of an accounts receivable manager during the six months following her resignation? That is the question at the heart of today's case. The bar set by the plan's requirement of "continuous" disability, it bears mentioning, is a high one. Even one day of partial work ability during the Elimination Period is enough to defeat Tranbarger's claim. See Frazier v. Life Ins. Co. of N. Am., 725 F.3d 560, 568-69 (6th Cir. 2013) (finding no continuous disability where the petitioner "was cleared to return to work" "at least at several points during the [Elimination Period]"). We, like the district court, look only to the administrative record for evidence of Tranbarger's functional capacity. Okuno v. Reliance Standard Life Ins. Co., 836 F.3d 600, 607 (6th Cir. 2016). In accordance with settled precedent, the district court evaluated that record de novo because Tranbarger's benefits plan did not vest Lincoln with discretionary authority. See Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 114-15, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989); Wilkins v. Baptist Healthcare Sys., Inc., 150 F.3d 609, 613 (6th Cir. 1998). The standard of review we apply to the district court's decision in this setting, however, is less established. See Wallace v. Oakwood Healthcare, Inc., 954 F.3d 879, 889-90 (6th Cir. 2020) (discussing both de novo and clear error review as possibilities). As we have done before, we decline the parties' invitation to resolve the issue definitively because Tranbarger's claim fails under any standard. See, e.g., id.; Hutson v. Reliance Standard Life Ins. Co., 742 F. App'x 113, 117-18 (6th Cir. 2018).

We begin with a point of agreement: neither party suggests (nor do we discern) any error in the district court's assessment that Tranbarger's main duties "required minimal physical activity, moderate to high aptitude abilities," and "the ability to talk frequently." With those duties in mind, we ask what evidence exists in the record of Tranbarger's functional capacity during the six-month Elimination Period. The answer, it seems, is middling. Begin with the notes from Tranbarger's visit to the Mayo Clinic following her resignation. Tranbarger's consulting doctor diagnosed her with fibromyalgia and chronic fatigue syndrome. Her self-reported pain was at level 8 out of 10, with fatigue at a full 10. Under "Functional Status," the report stated: "Patient describes . . . being very limited in being able to carry out activities of daily living."

Next is a physical therapist's write-up from an October 2016 consultation with Tranbarger. The therapist characterized Tranbarger's gait distance as "limited by fatigue," per Tranbarger's own account, and noted she "report[ed] constant pain and fatigue." As for the level of pain Tranbarger claimed to be experiencing, she said it varied between 5 and 10 (out of a maximum of 10), depending on the day and activity. But the therapist went on to rate Tranbarger's rehabilitation potential as "[f]air," giving her a score of 55/56 on a balance test. The passage of time seemingly bore out the therapist's prediction. Following a later, 45-minute visit—which entailed backward, forward, and lateral walking exercises—the therapist's notes quoted Tranbarger as stating she had "no pain," even though she continued to experience fatigue after swimming and on rest days. Other post-visit entries are much to the same end.

Four other statements generated after the period's expiration describe Tranbarger's condition during the Elimination Period. The first is a questionnaire completed by Tranbarger's former supervisor shortly after Tranbarger submitted her disability claim in September 2017. The supervisor stated that after Tranbarger fell ill, "it was very hard for her to do all the tasks required by her position . . . [so] we re-distributed her responsibilities among other workers . . . [but s]he was still responsible for phone calls and keeping track of collections agencies." The second is an affidavit submitted by Tranbarger's caretaker. Without referencing specific dates, the affidavit details a gradual and severe decline in Tranbarger's condition in the period between her 2015 surgery and May 2019. The final two statements come from Tranbarger herself. One describes her condition to a representative from Lincoln shortly after submitting her claim. The other is an undated summary of her symptoms. Both detail considerable struggles. Yet neither describes Tranbarger as being "totally disabled."

All things considered, this evidence fails to meet the rigorous standard Tranbarger's benefits plan establishes: proof of a continuous, 180-day inability to perform the main duties of her former position. See Frazier, 725 F.3d at 568-69. Ample evidence suggests she could perform some work in some instances. Consider the report from Tranbarger's physical therapy sessions, the only description of her physical capacities made during the Elimination Period. The report describes her moving well and suffering—in her own words—no pain, at the time. To the same end, the statement from Tranbarger's supervisor suggests that she could still perform some of her main duties. Taken as a whole, the record leaves little doubt that Tranbarger suffered considerable pain. But it does not satisfy the plan's rigid eligibility requirements, standards we are not at liberty to relax. See Wallace, 954 F.3d at 890-91.

True, the Mayo Clinic diagnosed Tranbarger with fibromyalgia and chronic pain syndrome. Those general diagnoses, however, do not answer the more granular question about her ability to work during the period in question. See Huffaker v. Metro. Life Ins. Co., 271 F. App'x 493, 502 (6th Cir. 2008); Vance v. Comm'r of Soc. Sec'y, 260 F. App'x 801, 806 (6th Cir. 2008); Boardman v. Prudential Ins. Co. of Am., 337 F.3d 9, 16 n.5 (1st Cir. 2003) ("While the diagnoses of chronic fatigue syndrome and fibromyalgia may not lend themselves to objective clinical findings, the physical limitations imposed by the symptoms of such illnesses do lend themselves to objective analysis."). Much the same can be said of the fact that Tranbarger reported experiencing pain and discomfort during the Elimination Period. That evidence likewise offers little basis for evaluating whether she could perform her job duties. Nor does her caretaker's affidavit allow us to reach a finding of disability—it lacks any particularity about Tranbarger's ability during the relevant period.

Tranbarger also directs our attention to consultations with Drs. James Wedner, Sarah Duda, and Nancy Russell. Those evaluations yielded declarations that Tranbarger was "totally disabled from working in any occupation," was limited to "20 minutes of effort" daily, faced a "progressive, unrelenting disease," and had to contend with severe physical restrictions. Setting aside the fact that these assessments were made years after the Elimination Period closed, none speaks to the critical issue of whether Tranbarger could work during that period. Rather, they detail her condition at the time of (much later) treatment. Javery v. Lucent Techs., Inc. Long Term Disability Plan, 741 F.3d 686, 690 n.1 (6th Cir. 2014) ("[W]e only consider [evidence from outside the Elimination Period] to the extent that it speaks to Plaintiff's condition during the relevant time period.").

So too for the March 2019 administrative law judge decision finding Tranbarger totally disabled for Social Security purposes since the beginning of the Elimination Period. Disability decisions of the Social Security Administration do not bind us in resolving ERISA claims. Cox v. Standard Ins. Co., 585 F.3d 295, 303 (6th Cir. 2009). Nor, in any event, do we find the decision here persuasive. After all, the ALJ primarily relied on evidence we have already said does not prove Tranbarger's claim, as well as post-Elimination Period evidence that does not speak to the timeframe at hand. See Javery, 741 F.3d at 690 n.1.

* * * * *

Two final matters. One, Tranbarger faults Lincoln's claims representative for improperly limiting the record it amassed before evaluating her claim, the same record later reviewed by the district court. Assuming that error did occur, the standard practice to remedy the deficiency is to remand to the plan administrator to amplify the record. See Card v. Principal Life Ins. Co., 17 F.4th 620, 626-29 (6th Cir. 2021) (Murphy, J., concurring). Tranbarger, however, did not raise the issue of remand with the district court. So she has forfeited the argument. Sheet Metal Workers' Health & Welfare Fund of N.C. v. Law Off. of Michael A. DeMayo, LLP, 21 F.4th 350, 357 (6th Cir. 2021) (citation omitted). Separately, we find it beside the point that Lincoln depended on its own paid doctor's report in evaluating her file, as it was Tranbarger's burden, remember, to prove disability.

We affirm the decision of the district court.

CONCURRENCE

NALBANDIAN, Circuit Judge, concurring.

I agree with the majority that we should affirm Lincoln's denial of benefits to Tranbarger. But I think our caselaw requires de novo review of the district court's factfinding in this case—even if I wouldn't choose that standard as a matter of first impression. So I write separately.

I.

We've recently expressed some confusion over the standard of review for a district court's factfinding in a case like Tranbarger's. See Wallace v. Oakwood Healthcare, Inc., 954 F.3d 879, 890 (6th Cir. 2020). But that confusion is unwarranted based on our caselaw. First, I'll describe the posture of Tranbarger's case and why under our current caselaw we review both the district court's factfinding and legal conclusions de novo. Then, I'll explain why I would apply clear-error review if we were writing on a blank slate and suggest that perhaps the Supreme Court should revisit how courts review ERISA cases.

A.

The mechanics of an ERISA appeal inform our review of the district court's factfinding. Take Tranbarger's case as an example. After receiving an adverse benefits determination from Lincoln, Tranbarger challenged that decision under 29 U.S.C. § 1132(a)(1)(B) in federal district court. The district court reviewed de novo the same administrative record that the plan administrator reviewed and determined that Tranbarger was not entitled to benefits. See Wilkins v. Baptist Healthcare Sys., Inc., 150 F.3d 609, 619 (6th Cir. 1998) (Gilman, J., concurring) (writing for the majority) (explaining that unless there is a procedural challenge to the plan administrator's decision the district court's review is limited to the record that was before the plan administrator).

Tranbarger appealed. And the record before us includes the decisions of both the plan administrator and the district court. And the question is whether we defer at all to the district court's factfinding on the administrative record. As I'll explain, we've answered that question in the negative.

In Firestone Tire & Rubber Co. v. Bruch, the Supreme Court held that "a denial of benefits challenged under § 1132(a)(1)(B) is to be reviewed under a de novo standard unless the benefit plan gives the administrator or fiduciary discretionary authority to determine eligibility for benefits or to construe the terms of the plan." 489 U.S. 101, 115, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989). Neither party argues that the administrator had discretionary authority, so we—and the district court sitting as a court of review in this posture—review the "denial of benefits" de novo. Id.

After Firestone came down, we explained that "the Supreme Court expressly limited its discussion to the issue of 'the appropriate standard of review in [29 U.S.C.] § 1132(a)(1)(B) actions challenging denials of benefits based on plan interpretations.' " Rowan v. Unum Life Ins. Co. of Am., 119 F.3d 433, 436 (6th Cir. 1997) (alteration in original) (citation omitted). That meant the Supreme Court "arguably[ ] left open the question of the appropriate standard of review of factual determinations by plan administrators." Id.

In Rowan, we answered that question. We decided that "factual determinations of plan administrators in actions brought under 29 U.S.C. § 1132(a)(1)(B) are subject to de novo review." Id. at 435. In doing so, we rejected the view the Fifth Circuit held at the time, that an abuse-of-discretion standard applied to a plan administrator's factfinding. See Pierre v. Conn. Gen. Life Ins., 932 F.2d 1552, 1558 (5th Cir. 1991), overruled by Ariana M. v. Humana Health Plan of Tex., Inc., 884 F.3d 246, 257 (5th Cir. 2018) (remanding for the district court to apply de novo review to the plan administrator's factfinding). The Fifth Circuit adopted the abuse-of-discretion standard for district-court review of plan-administrator factfinding because it is customary for "courts [to] accord [deference] to fact finding by administrative bodies and [ ] appellate courts [to] accord [deference] to district court factual determinations due to the difficulty and uncertainty involved in reviewing a 'cold record.' " Rowan, 119 F.3d at 436 (quoting Pierre, 932 F.2d at 1558).

But we didn't buy the Fifth Circuit's approach. We explained that deferring "under th[o]se circumstances is very different [from] deferring to a plan administrator, who will be one party to a private dispute . . . . The reason for treating the two situations differently is obviously that one party to a contract has an incentive to find facts not in a neutral fashion, but in the manner that is most advantageous to its own interests." Id. (citation omitted).

And we rejected the Fifth Circuit's alternative argument that "failure to defer to plan administrators' factual findings will lead to a flood of litigation[.]" Id. That's because "drafters of ERISA plans can avoid de novo review of plan administrators' factual determinations in this context by careful[ly] drafting" ERISA plans that "vest fact-finding discretion in the plan administrator." Id. So in Rowan, we adopted a de novo standard of review for district courts reviewing a plan administrator's factfinding.

But we never explicitly said that we, as the court of appeals, review the district court's factfinding in these cases de novo too. We'd clear that up soon. Extending (and citing to) Rowan, we explained in Wilkins v. Baptist Healthcare Systems, Inc., that "both the district court and this court review de novo the plan administrator's denial of ERISA benefits," and that "[t]his de novo standard of review applies to the factual determinations as well as to the legal conclusions of the plan administrator." 150 F.3d at 613. In other words, we said that we and the district court review the plan administrator's decision. And we both apply a de novo standard of review to the factfinding and legal conclusions of the plan administrator based on the administrative record.

It follows from Wilkins then that if we, as the court of appeals, apply de novo review to the plan administrator's factfinding, then we cannot defer to the district court's factfinding. Whether we call it de novo review or "not deferring" to district-court factfinding, we look at the administrative record with fresh eyes. And that makes sense under Wilkins. In Wilkins, we said that the "district court should conduct a de novo review based solely upon the administrative record" unless there is "a procedural challenge to the administrator's decision." Id. at 619 (Gilman, J., concurring) (writing for the majority on this point). That rule means that unless there is a procedural challenge to the decision, we're reviewing the same record that was before the district court and the plan administrator. So under Wilkins, we review the facts just like the district court—the first reviewing court—did.

In Wallace v. Oakwood Healthcare, Inc., we recently said that this "Court's precedent conflicts as to the standard of review we apply to a district court's factual findings" in these cases. 954 F.3d at 890. That panel pointed out that while we have reviewed a district court's factual findings de novo, we've also "reviewed a district court's factual findings for clear error." Id. Wallace cited Moore v. Lafayette Life Insurance, 458 F.3d 416 (6th Cir. 2006), as an example of where we had used clear error to review a district court's factfinding in an ERISA case like Tranbarger's. Id.

True, Moore said that this "Court reviews the district court's decisions on matters of law in an ERISA benefits action de novo and its factual findings for clear error." Moore, 458 F.3d at 438. It cited an unpublished case, Stoll v. Western & Southern Life Insurance, 64 F. App'x 986, 990 (6th Cir. 2003), for that proposition. But that unpublished case took its standard-of-review section from Sanford v. Harvard Industries, Inc., 262 F.3d 590, 594 (6th Cir. 2001). Sanford in turn cited Davies v. Centennial Life Insurance, a case in which the district court had actually conducted a bench trial. See 128 F.3d 934, 938 (6th Cir. 1997), overruled on other grounds by UNUM Life Ins. v. Ward, 526 U.S. 358, 119 S.Ct. 1380, 143 L.Ed.2d 462 (1999); Fed. R. Civ. P. 52. But in Wilkins, we held that ERISA cases should not be adjudicated like a bench trial, so Davies is inapplicable. 150 F.3d at 618 (Gilman, J., concurring) (writing for the majority on this point) ("[W]e are satisfied that a district court should not adjudicate an ERISA action as if it were conducting a standard bench trial under Rule 52."). In other words, the record is generally limited to what was in front of the plan administrator.

Long story short, if you trace back the case Moore cites for the clear-error standard of review, you arrive at a case that isn't in the same posture as ours. And none of these cases grapple with Wilkins's holding on the standard of review.

That's a problem because of the prior-panel precedent rule. Ward v. Holder, 733 F.3d 601, 608 (6th Cir. 2013). "Forced to choose between conflicting precedents, we must follow the first one[.]" United States v. Jarvis, 999 F.3d 442, 445-46 (6th Cir. 2021). Our first clear pronouncement on the issue is Rowan, followed by Wilkins. That means the de novo standard of review for district-court factfinding controls.

B.

That's not to say that I would choose a de novo standard of review if I were writing on a blank slate. Instead, I'd adopt the Fourth Circuit's view in Tekmen v. Reliance Standard Life Insurance Co. See 55 F.4th 951, 962 (4th Cir. 2022). Tekmen held that clear-error review applies to a district court's factfinding in these cases, even when the district court is not taking new evidence but instead simply reviewing the administrative record that was before the plan administrator. Id.

Tekmen acknowledged that under Firestone we must review de novo the legal conclusions of the district court but noted that Firestone left open the question of how to review factual findings in these cases. So Tekmen leaned on the district court's institutional role as factfinder to land on clear-error review, noting that these cases require a "a careful examination of the often-voluminous administrative record . . . [and] may involve assessing credibility and determining the appropriate weight to assign evidence." Id. at 961.

While "Courts of Appeals are also capable of reviewing cold records," Tekmen held that clear-error review ensured that "the district court's resolution of the facts" would be treated as the "main event" rather than a "tryout on the road" to our review. Id. (quoting Anderson v. City of Bessemer City, 470 U.S. 564, 575, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985)).

That logic makes sense to me. We've said time and time again that the district court is the trier of fact—not us. See, e.g., O'Bryan v. County of Saginaw, 722 F.2d 313, 314 (6th Cir. 1983) ("An appellate court is not a trier of fact. In the federal system, the district court is normally the trier of fact in the first instance and initially hears evidence on disputed questions."). And under my view, the district court would still review the plan administrator's factual findings and legal determinations de novo, eliminating the concern that the federal court could be bound by a private party's factfinding. See Rowan, 119 F.3d at 436 (explaining that deferring "to a plan administrator, who will be one party to a private dispute" is inappropriate because the plan administrator "has an incentive to find facts not in a neutral fashion, but in the manner that is most advantageous to its own interests" (citing Perez v. Aetna Life Ins., 96 F.3d 813, 824 (6th Cir. 1996), reh'g en banc granted, vacated on other grounds, 106 F.3d 146 (6th Cir. 1997))).

I acknowledge that the institutional argument is weakened in our Circuit in that unless there is a procedural challenge to the plan administrator's decision, we don't allow the district court to look at new evidence or weigh testimony in the same way that it would in a bench trial under Rule 52. Wilkins, 150 F.3d at 618 (Gilman, J., concurring) (writing for the majority on this point). But I can't think of another context where we take over the role as factfinder, and the district court is specialized in that task. So I think the equities weigh heavily in favor of reviewing the facts in these cases for clear error. And that's what I'd advocate for if we were writing on a clean slate.

C.

I also think it's worth pointing out why our approach to § 1132 potentially conflicts with both the Federal Rules of Civil Procedure and Supreme Court precedent. I start with the language of § 1132. Then I explain how we developed review of these cases, first in Perry v. Simplicity Engineering, and later in Wilkins v. Baptist Healthcare Systems, Inc. And finally, I explain why this is a problem under the Court's recent decision in United States v. Tsarnaev.

1.

Start with the text of the statute. The only guidance Congress gave us in the text is that "[a] civil action may be brought . . . by a participant or beneficiary . . . to recover benefits due to him under the terms of his plan, to enforce his rights under the terms of the plan, or to clarify his rights to future benefits under the terms of the plan[.]" 29 U.S.C. § 1132(a)(1)(B). And the Federal Rules of Civil Procedure typically govern "all civil actions and proceedings in the United States district courts, except as stated in Rule 81." Fed. R. Civ. P. 1; see Reid v. Sears, Roebuck & Co., 790 F.2d 453, 459 (6th Cir. 1986) (explaining that "[t]he Federal Rules of Civil Procedure [including discovery and motions practice] are the rules of practice which apply to civil actions in the federal courts"). The only exception to that is when a "specific statutory provision prevails over [the Federal Rule] which is a general rule of procedure." Pension Benefit Guar. Corp. v. Alloytek, Inc., 924 F.2d 620, 626 (6th Cir. 1991).

Rule 81 doesn't say anything about ERISA, see Fed. R. Civ. P. 81, and § 1132 doesn't provide a more specific rule than the Federal Rules of Civil Procedure. So by negative implication, the Federal Rules should apply to ERISA civil actions. See Eaton Corp. & Subsidiaries v. Comm'r, 47 F.4th 434, 444 (6th Cir. 2022) ("After all, under the expressio unius canon, '[t]he expression of one thing implies the exclusion of others.' " (quoting Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts 107 (2012))). If Rule 1's plain language applied this way, we would allow the parties to build up a record with discovery under Rule 26, move for summary judgment under Rule 56, and conduct a bench trial under Rule 52.

2.

After Firestone handed down its de novo standard of review for denial-of-benefits decisions, we considered how district courts would conduct de novo review in this context. Perry v. Simplicity Eng'g, 900 F.2d 963, 966 (6th Cir. 1990). There were two options. We said that de novo could refer either "to review of the decision below based only on the record below [or] to review based on the record below plus any additional evidence received by the reviewing court." Id. "The Supreme Court in [Firestone] did not indicate which meaning it had in mind." Id.

We chose the first option. Perry established an atextual regime—by which we "review[ ] the plan administrator's decision to deny" benefits, rather than adjudicate the case as a civil action. Id. at 967. Looking at ERISA's context, we explained that "[n]othing in the legislative history suggests that Congress intended that federal district courts would function as substitute plan administrators, a role they would inevitably assume if they received and considered evidence not presented to administrators concerning an employee's entitlement to benefits." Id. at 966. And we added that "[s]uch a procedure would frustrate the goal of prompt resolution of claims by the fiduciary under the ERISA scheme." Id.; see id. at 967 ("Permitting or requiring district courts to consider evidence from both parties that was not presented to the plan administrator would seriously impair" the "primary goal of ERISA . . . to provide a method for workers and beneficiaries to resolve disputes over benefits inexpensively and expeditiously.").

In short, we told district courts not to consider new evidence. And that directive was based on a policy rationale and (some might say questionable) congressional intent. In doing so, we failed to consider the text of the statute—the only clear indication of what Congress wanted. Cf. Card v. Principal Life Ins., 17 F.4th 620, 627 (6th Cir. 2021) (Murphy, J., concurring) (explaining that our policy of remanding these cases to plan administrators has no support in the text of ERISA); Wallace, 954 F.3d at 900 (Thapar, J., concurring) (finding it "troubling" that our circuit has added an exhaustion requirement to ERISA cases where no such requirement exists in the text of the statute).

Section 1132 refers to "civil actions"—a term that necessarily incorporates the Federal Rules of Civil Procedure and the introduction of evidence like in any other case. See Fed. R. Civ. P. 1. But Perry chose intent over the clear text.

3.

Later, we extended Perry's logic in Wilkins. There, we discussed how the Perry approach worked in practice. We explained that a district court's de novo review standard "does not neatly fit under either Rule 52 or Rule 56, but is a specially fashioned rule designed to carry out Congress's intent under ERISA." Wilkins, 150 F.3d at 618 (Gilman, J., concurring) (writing for the majority here).

We reasoned that Rule 52 was inapposite because "[s]uch a proceeding would inevitably lead to the introduction of testimonial and/or other evidence that the administrator had no opportunity to consider," which would contradict Perry. Id. At the same time, we rejected Rule 56 as inappropriate because summary judgment necessarily implies that there could be a genuine dispute of fact for trial, which would also contradict Perry. Id.

So, instead of allowing for traditional litigation under the Federal Rules of Civil Procedure, we proposed "suggested guidelines" that the district court "should employ . . . in adjudicating an ERISA action": (1) The district court "should conduct a de novo review based solely upon the administrative record, and render findings of fact and conclusions of law accordingly"; (2) "The district court may consider evidence outside of the administrative record only if that evidence is offered in support of a procedural challenge to the administrator's decision, such as an alleged lack of due process afforded by the administrator or alleged bias on its part"; (3) "[T]he summary judgment procedures set forth in Rule 56 are inapposite to ERISA actions and thus should not be utilized in their disposition." Id. at 619.

Judge Cole did not join this part of Wilkins, pointing out that these suggested guidelines "ha[d] neither been mandated by Congress nor recognized by any other court." Id. at 617. And he was right. Our approach, which eschews both Rule 56 and Rule 52, rejects the application of the Federal Rules of Civil Procedure to § 1132's "civil action." Again, that means we are sidestepping Rule 1—that the Federal Rules "govern the procedure in all civil actions . . . except as stated in Rule 81." See Fed. R. Civ. P. 81 (not mentioning ERISA); compare Wilkins, 150 F.3d at 618-20, with Tekmen, 55 F.4th at 961 & n.6 (explaining that "district courts in [the Fourth] Circuit may allow new evidence that was not before the plan administrator" only when "necessary to conduct an adequate de novo review of the benefit decision" (citation omitted)).

4.

So, as a matter of civil procedure, I think we've got a problem with our current approach to ERISA claims. My concern is only heightened by the Supreme Court's recent decision in United States v. Tsarnaev. See — U.S. —, 142 S. Ct. 1024, 212 L.Ed.2d 140 (2022). That case analyzed a prophylactic rule the First Circuit had created that required district courts to ask about prospective jurors' media exposure before jury selection. Id. at 1035-36. The Supreme Court held that the First Circuit's rule conflicted with Supreme Court precedent holding that "a district court enjoys broad discretion to manage jury selection, including what questions to ask prospective jurors," and that the district court's management of jury selection could only be reviewed for "abuse of discretion." Id. at 1036 (collecting cases).

The Supreme Court explained that while a supervisory authority may "inhere[ ] in federal courts," "supervisory rules cannot conflict with or circumvent a constitutional provision or federal statute." Id. at 1035-36 (citing Thomas v. Arn, 474 U.S. 140, 148, 106 S.Ct. 466, 88 L.Ed.2d 435 (1985)). "Nor can they conflict with or circumvent a Federal Rule." Id. at 1036 (citing Carlisle v. United States, 517 U.S. 416, 426, 116 S.Ct. 1460, 134 L.Ed.2d 613 (1996)). And "lower courts cannot create prophylactic supervisory rules that circumvent or supplement legal standards set out in decisions of [the Supreme Court]." Id.

Bottom line, Tsarnaev prevents lower courts from creating prophylactic rules that contradict federal statutes or rules—like the Federal Rules of Civil Procedure. So it seems like Perry and Wilkins could be a problem, not just under the terms of the Federal Rules of Civil Procedure, but also under Tsarnaev.

To make things even more complicated, I think the Supreme Court perhaps has a problem under Tsarnaev too. That's because Tsarnaev and Firestone appear to be in tension. Firestone implicitly condones the arbitrary-and-capricious standard of review for ERISA actions where the plan administrator has discretionary authority to administer the plan. The Supreme Court grounded that part of its analysis in trust law. Firestone, 489 U.S. at 111, 109 S.Ct. 948 ("Trust principles make a deferential standard of review appropriate when a trustee exercises discretionary powers." (citing Restatement (Second) of Trusts § 187 (1959))). But the Court only found trust law applicable to the extent that the plan administrator had discretionary authority. Id. at 112-13, 109 S.Ct. 948. After explaining that the federal courts have been tasked with "develop[ing] a 'federal common law of rights and obligations under ERISA-regulated plans,' " id. at 110, 109 S.Ct. 948 (citation omitted), the Court held that de novo review applied to decisions of plan administrators who did not possess discretionary authority, id. at 115, 109 S.Ct. 948; see id. at 112-13, 109 S.Ct. 948 (explaining that when a plan administrator does not have discretionary authority federal courts should "construe terms" in the plan like "contractual provisions" without "deferring to either party's interpretation").

So the Supreme Court, in building up ERISA's federal common law, drew its distinction for the standard of review based on how closely a plan tracks trust law. If the plan creates a trustee-like relationship between the plan administrator and benefits seeker, then arbitrary-and-capricious review applies to the plan administrator's decision. If the plan creates a traditional contractual agreement, then federal courts must review the benefits determination de novo.

In my mind, the problem with this distinction is that the Supreme Court is endorsing a quasi-administrative-law review regime for a "civil action" under § 1132. In a broad sense, the Supreme Court's federal common law for ERISA is eschewing the Federal Rules of Civil and Appellate Procedure in favor of something like agency review—at least in cases where the plan administrator has discretionary authority to interpret the plan. Cf. Williams v. Int'l Paper Co., 227 F.3d 706, 712 (6th Cir. 2000) ("This Court has noted that the arbitrary and capricious standard is the least demanding form of judicial review of administrative action.").

And the Supreme Court doing so seems to run afoul of Tsarnaev—again broadly construed. One might argue that the Court-created ERISA regime conflicts with Rule 1, which says that the Federal Rules of Civil Procedure govern all "civil actions." To the extent that Firestone establishes an administrative-law regime that blesses the arbitrary-and-capricious review of a plan administrator's decisions by both the district court and court of appeals, it seems that federal common law is in tension with our standard review of civil actions. See Tsarnaev, 142 S. Ct. at 1035-36 (holding that prophylactic, judge-made rules cannot supersede the Federal Rules of Civil Procedure); see also Buckeye Sugars, Inc. v. Commodity Credit Corp., 744 F.2d 1240, 1244 (6th Cir. 1984) (explaining that "[c]ourts should not search for gaps in legislation in order to fashion and apply federal common law").

That's because if the Federal Rules apply, then the mechanics of the introduction of evidence, summary judgment, and bench trials would also apply. And we'd apply de novo review to legal conclusions and clear-error review to factual findings. Monasky v. Taglieri, — U.S. —, 140 S. Ct. 719, 730, 206 L.Ed.2d 9 (2020) ("Absent a treaty or statutory provision . . . [g]enerally, questions of law are reviewed de novo and questions of fact, for clear error[.]").

At least there is tension if the district court conducts a bench trial under Rule 52 in the case where the plan administrator had discretionary authority to administer the plan. That's because we'd still be bound to review the plan administrator's decision under the arbitrary-and-capricious standard, even as we reviewed the district court's factual findings under the clear-error standard Rule 52 supplies. Fed. R. Civ. P. 52(a)(6) (explaining that reviewing courts may only set aside bench trial factual findings if those findings are clearly erroneous).

But Firestone may also be construed as a routine standard-of-review case. We routinely assign standards of review based on whether the question is legal or factual, whether proper objections were raised below, and whether we or the district court are better situated to address the issue. And setting aside narrow exceptions, like Rule 52, standards of review aren't baked into the Federal Rules. So in that light, if Firestone is understood as a standard-of-review case only, it would appear to be generally consistent with Tsarnaev.

In any event, because the majority reaches the right outcome, I respectfully concur.