KARLA H. ABATIE,
|
No. 03-55601
|
Argued and Submitted En Banc
March 23, 2006—San Francisco, California
Filed August 15, 2006
Before: Mary M. Schroeder, Chief Judge, and Alex Kozinski, Diarmuid F. O’Scannlain, Pamela Ann Rymer, Andrew J. Kleinfeld, Barry G. Silverman, Susan P. Graber, M. Margaret McKeown, Kim McLane Wardlaw, William A. Fletcher, Ronald M. Gould, Richard A. Paez, Johnnie B. Rawlinson, Jay S. Bybee, and Consuelo M. Callahan, Circuit Judges.
After Dr. Joseph Abatie died, his widow—Plaintiff Karla H. Abatie—sought life insurance benefits from Alta Health & [p.9629]
Life Insurance Company under an employee welfare benefit plan regulated by the Employee Retirement Income Security Act of 1974 (“ERISA”).
Alta, which was both the administrator and the funding source of the plan, denied benefits.
Plaintiff then brought this action. The district court upheld Alta’s decision. Plaintiff’s appeal questioned the standard of review
that the district court had used to review her claim.
We took this case en banc1 to reconsider our approach to ERISA cases in which a plan administrator denies benefits and (1) the wording of the plan confers discretion on the plan administrator and (2) the plan administrator has a conflict of interest.
In answering that question, we have returned to first principles, the Supreme Court’s opinion in Firestone Tire &
Rubber Co. v. Bruch, 489 U.S. 101 (1989).
We conclude that our earlier opinion in Atwood v. Newmont Gold Co., 45 F.3d 1317 (9th Cir. 1995), misinterpreted Firestone.
We now establish a more comprehensive approach to ERISA cases in which a conflict of interest exists.
As we will explain below, abuse of discretion review, tempered by skepticism commensurate with the plan administrator’s conflict of interest, applies here.
In addition, this case requires us to consider how a court is to review an ERISA plan administrator’s decision when the procedure that produced the decision did not follow all statutory requirements.
For the reasons that we will develop, we conclude that when a decision by an administrator utterly fails to follow applicable procedures, the administrator is not, in fact, exercising discretionary powers under the plan, and its decision should be subject to de novo review.
Lesser irregularities, like the one in this case, do not remove the decision from abuse of discretion review, but rather should be factored into the calculus of whether the administrator abused its discretion.
[p.9630]
Dr. Abatie worked for the Santa Barbara Medical Foundation Clinic from 1971 until November 1992, when he took a medical leave of absence after developing non-Hodgkin’s lymphoma.
The Clinic sponsored for its employees an employee welfare benefit plan, which provided for disability benefits, and an unfunded life insurance plan. Only life insurance is at issue in this appeal.
Shortly after taking leave from the Clinic, Dr. Abatie applied for and received disability benefits.
Dr. Abatie never returned to work and, beginning in 1993, received permanent disability benefits.
From September 1998 until April 2000, Dr. Abatie experienced a partial remission but, in June 2000, he died.
After his death, Plaintiff filed a claim for life insurance benefits.
The life insurance policy under the plan was originally issued by Home Life Financial Assurance Company.
Alta is the successor in interest to Home Life’s rights and responsibilities.
The policy requires that a beneficiary work full-time for the employer in order for insurance coverage to start.
The policy also provides that coverage ends when employment ends, unless otherwise provided for by the policy.
The policy specifies two ways in which an employee can continue to receive life insurance coverage after ending employment.
First, the insured may continue to pay premiums.
Dr. Abatie did not do that.
Second, if an insured becomes totally disabled while still covered by the policy, the insured may request what is commonly referred to as a “waiver of premium application.”
If a waiver is granted, the insured continues to receive life insurance coverage without paying premiums and without working.
The policy defines total disability as being “not able to work at all at any job or business for pay or profit due to [p.9631]
injury or sickness.”
In order to be eligible for a waiver, the insured must provide the insurer with “proof of . . . total disability within 12 months after the date [of] becom[ing] totally disabled.”
The policy further provides that, even after a waiver of premium application is granted, coverage will end if the insured is “no longer totally disabled” or fails to provide “proof of continued disability.”
Several months after Dr. Abatie’s death, the Clinic wrote to Alta requesting payment of life insurance benefits to Plaintiff.
The Clinic’s insurance broker sent a letter to Alta noting that, “due to administrative error, the waiver of premium application
was not filed.”
Despite that error, the Clinic sought “retroactive” qualification of Dr. Abatie for insurance coverage.
In March 2001, Alta denied the claim for life insurance benefits.
It did so because it concluded that Dr. Abatie had not submitted proof of his total disability within 12 months of becoming totally disabled.
Alta noted that the Clinic admitted that Dr. Abatie had never filed a waiver of premium application.
Accordingly, Alta concluded, Dr. Abatie was not covered by the life insurance plan when he died.
Plaintiff filed suit against Alta in California state court.
Alta removed the case to federal court pursuant to 28 U.S.C.§ 1441(a).
The parties conducted discovery, supplementing the administrative record.
As part of that discovery, the Clinic produced documents that suggested, contrary to its previous admission, that it had filed a waiver of premium application on behalf of Dr. Abatie, which Alta’s predecessor in interest, Home Life, had approved.
Specifically, the Clinic produced an internal document stating, Dr. Abatie’s “[p]remiums waiver requested in January, 1994. Should be receiving confirmation any day.”
A handwritten note next to that entry says, “waiver was granted 2/94.” And a separate note to the file asserts that Dr.
Abatie’s “life insurance premium is waived.” [p.9632]
In conjunction with the Clinic’s discovery of those documents, Plaintiff took depositions to shed more light on whether the waiver of premium application was in fact submitted to and approved by Alta’s predecessor.
One Clinic employee, Melissa Peter, testified in deposition that she was responsible for those notes and that she had sought and
obtained a waiver for Dr. Abatie.
However, she did not remember the circumstances under which she wrote the notes and acknowledged that it was customary to obtain official confirmation of a waiver of premium—a document not found in the Clinic’s files.
Similarly, Alta’s files contained no documentation showing that Dr. Abatie had applied for a waiver of premium, and he was not listed on the roster of employees granted waivers that Alta acquired from its predecessor, Home Life.
But a Home Life “Renewal Census Report” dated May 8, 1997, listed Joseph D. Abatie as having life insurance coverage of $500,000 under Policy No. G40612, with an effective date of August 1, 1992.
In view of this additional evidence, the parties agreed to allow Alta to conduct an additional review and render a final determination of the claim instead of proceeding directly to trial.2
On review, Alta again denied Plaintiff’s claim for life insurance benefits.
Alta concluded that there was insufficient evidence to prove that the Clinic had submitted a waiver of premium application for Dr. Abatie.
It reasoned that the [p.9634] Clinic earlier admitted that it had failed to send the application (albeit inadvertently), that neither the Clinic nor Alta had a record in its files that Dr. Abatie filed a waiver of premium application, and that the Clinic had unsubstantiated handwritten notes, but no formal documentation that an application had been submitted and approved.
In addition, Alta stated for the first time that it was denying coverage because there was insufficient evidence in the record that Dr. Abatie had remained totally disabled from the time he left work until his death, as required under the policy.
In particular, Alta questioned whether Dr. Abatie had remained totally disabled from September 1998 through April 2000—the time of his partial remission.
The parties then resumed litigation in court.
The district court conducted a bench trial.
In response to the new rationale provided by Alta for denying benefits, Plaintiff presented a declaration from Dr. Abatie’s treating physician. The declaration, created after Alta’s final determination, stated that Dr. Abatie was totally disabled and could not have engaged in any work from September 1998 through April 2000, as a result of numerous complications from lymphoma and anemia.
Following the trial, the court ruled that Alta did not abuse its discretion by denying Plaintiff’s claim.
The district court declined to decide whether, in fact, the Clinic had submitted a waiver of premium application to the plan administrator; the court simply recited some of the conflicting evidence and concluded that Alta had not abused its discretion.
The district court also refused to consider the declaration from Dr. Abatie’s treating physician in coming to its determination. Plaintiff timely appealed.
The main question before us in this case is what standard of review the district court should apply in examining a plan [p.9634]
administrator’s decision to deny ERISA benefits when the administrator labors under a conflict of interest or when the process is irregular, a question that will occupy the remainder of the opinion.
But the question of how we review the district court’s decision, in turn, is well established. We review de novo a district court’s choice and application of the standard of review to decisions by fiduciaries in ERISA cases.
Lang v. Long-Term Disability Plan of Sponsor Applied Remote Tech.,Inc., 125 F.3d 794, 797 (9th Cir. 1997).
We review for clear error the underlying findings of fact.
Friedrich v. Intel Corp., 181 F.3d 1105, 1109 (9th Cir. 1999).
A. What standard of review should the district court apply when an ERISA plan participant questions an adverse decision by the plan administrator based on a disputed interpretation of the administrative record?
When Congress enacted ERISA, it did not specify the standard of review that courts should apply when a plan participant challenges a denial of benefits. Instead, Congress expected federal courts to develop a body of common law to govern those claims and to determine the appropriate standards of review.
See Franchise Tax Bd. v. Constr. Laborers Vacation Trust, 463 U.S. 1, 24 n.26 (1983)
(“ERISA’s legislative history indicates that, in light of the Act’s virtually unique pre-emption provision, see § 514, 29 U.S.C. § 1144, ‘a body of Federal substantive law will be developed by the courts to deal with issues involving rights and obligations under private welfare and pension plans.’ ” (quoting 120 Cong. Rec. 29942 (1974) (remarks of Sen. Javits)));
Scott v. Gulf Oil Corp., 754 F.2d 1499, 1502 (9th Cir. 1985) (noting, similarly, that Congress intended the courts to develop a body of federal common law to deal with ERISA cases).
Since ERISA’s inception in 1974, Congress has not altered the statute to provide for a standard of review. Federal courts, therefore, have continued to fill the gap. [p.9635]
In 1989, the Supreme Court addressed the standard of review that courts must apply in reviewing ERISA cases in which plan administrators have denied benefits. Firestone, 489 U.S. 101.
Indeed, Firestone is the Court’s only opinion directly clarifying the nature of court review in an ERISA case.
Therefore, to determine whether the standard of review changes when a plan confers discretion but its administrator operates under a conflict of interest, we look first to the Firestone decision itself.
1. The Firestone Decision — Analyzing the Terms of the Plan
In Firestone, former employees requested severance benefits after Firestone sold to another company the plastics plants
in which they worked. 489 U.S. at 105-06.
Firestone, acting as both the administrator and the funding source of the applicable ERISA plan, denied the requests for benefits. Id. at 107.
The former employees sued, and their case reached the Supreme Court, which remanded for further proceedings. Id. at 118.
[1] The Court concluded, among other things, that general trust principles apply when considering how district courts
should review ERISA denial of benefits cases, because the plan administrator stands in a fiduciary relationship to the plan participants. Id. at 110-11.
To assess the applicable standard of review, the starting point is the wording of the plan. Id. at 111.
[2] When a plan does not confer discretion on the administrator “to determine eligibility for benefits or to construe the
terms of the plan,” a court must review the denial of benefits de novo “regardless of whether the plan at issue is funded or
unfunded and regardless of whether the administrator or fiduciary is operating under a possible or actual conflict of interest.”
Id. at 115.
de novo is the default standard of review. Id.; Kearney v. Standard Ins. Co., 175 F.3d 1084, 1089 (9th Cir.
[p.9636] 1999) (en banc).
If de novo review applies, no further preliminary analytical steps are required. The court simply proceeds to evaluate whether the plan administrator correctly or incorrectly denied benefits, without reference to whether the administrator operated under a conflict of interest.
[3] But if the plan does confer discretionary authority as a matter of contractual agreement, then the standard of review shifts to abuse of discretion. Firestone, 489 U.S. at 115.
We have held that, for a plan to alter the standard of review from the default of de novo to the more lenient abuse of discretion, the plan must unambiguously provide discretion to the administrator. Kearney, 175 F.3d at 1090.
The essential first step of the analysis, then, is to examine whether the terms of the ERISA plan unambiguously grant discretion to the administrator.
Accordingly, we first turn to the text of the plan.
[4] The plan at issue here provides:
The responsibility for full and final determinations of eligibility for benefits; interpretation of terms; determinations of claims; and appeals of claims denied in whole or in part under the HFLAC Group [Home Life] policy rests exclusively with HFLAC.(Emphasis added.) Under the applicable precedents, that provision is sufficient to confer discretion on Alta, the plan administrator and successor in interest to Home Life, even though the word “discretion” does not appear.
There are no “magic” words that conjure up discretion on the part of the plan administrator.
See Sandy v. Reliance Standard Life Ins. Co., 222 F.3d 1202, 1207 (9th Cir. 2000) (noting that “there is no magic to the words ‘discretion’ or ‘authority’ ”).
The Supreme Court has suggested that a plan grants discretion if the administrator has the “power to construe disputed or doubtful terms” in the plan. Firestone, 489 U.S. at 111;
see also Id. at 115 (noting that if a plan grants an [p.9637] administrator the right to determine eligibility for benefits or to “construe the terms of the plan,” it has discretionary authority), and
Id. at 111 (stating that Firestone cannot take advantage of the principles of discretion “for there is no evidence that under Firestone’s termination pay plan the administrator has the power to construe uncertain terms or that eligibility determinations are to be given deference”).
Moreover, we have repeatedly held that similar plan wording—granting the power to interpret plan terms and to make final benefits determinations—confers discretion on the plan administrator.
See, e.g., Bergt v. Ret. Plan for Pilots Employed by Markair, Inc., 293 F.3d 1139, 1142 (9th Cir. 2002) (holding that a plan conferred discretion because its terms granted the administrator the “power” and “duty” to “interpret the plan and to resolve ambiguities, inconsistencies and omissions” and to “decide on questions concerning the plan and the eligibility of any Employee” (internal quotation marks omitted));
Grosz-Salomon v. Paul Revere Life Ins. Co., 237 F.3d 1154, 1159 (9th Cir. 2001) (holding that a plan providing that the administrator “has the full, final, conclusive and binding power to construe and interpret the policy under the plan . . . [and] to make claims determinations” grants discretion (internal quotation marks omitted)).
Many of our sister circuits have come to a similar conclusion.
A number of cases hold that when the words give a plan administrator the authority to interpret the plan’s terms and to make final benefits determinations, discretion is unambiguously vested in the administrator.
See, e.g., McElroy v. Smith-Kline Beecham Health & Welfare Benefits Trust Plan for U.S. Employees, 340 F.3d 139, 141 (3d Cir. 2003) (holding that the following text conferred discretion: The administrator “reserves the absolute right to interpret” plan provisions and “to make determinations of facts and eligibility for benefits, and to decide any dispute that may arise.”);
Shields v. Reader’s Digest Ass’n, 331 F.3d 536, 541 n.6 (6th Cir. 2003) (holding that discretion was granted when the plan provided the administrator [p.9638] had “complete control” over the administration of the Plan, and “the power to construe” the Plan and “determine all questions” that arise under it);
Twomey v. Delta Airlines Pilots Pension Plan, 328 F.3d 27, 31 (1st Cir. 2003) (concluding that discretion was granted by the following wording: “[T]he Administrative Committee shall have such duties and powers as may be necessary to discharge its responsibilities under the Plan, including . . . decid[ing] all questions of eligibility of any Employee to participate in the Plan or to receive benefits under it, its interpretation thereof in good faith to be final and conclusive”) (second and third alterations in original);
Duhon v. Texaco, Inc., 15 F.3d 1302, 1305 (5th Cir. 1994) (holding that this sentence granted discretion: “The decisions of the Plan Administrator shall be final and conclusive with respect to every question which may arise relating to either the interpretation or administration of this Plan.”).
We have held that ERISA plans are insufficient to confer discretionary authority on the administrator when they do not grant any power to construe the terms of the plan.
For example, in our recent decision in Ingram v. Martin Marietta Long Term Disability Income Plan, 244 F.3d 1109, 1112-13 (9th Cir. 2001), we concluded that the plan merely identified the carrier as the entity that was to pay benefits and administer the plan. There, the plan provided that “[t]he carrier solely is responsible for providing the benefits under this Plan”; “[t]he carrier will make all decisions on claims”; and “the review and payment or denial of claims and the provision of full and fair review of claim denial pursuant to [ERISA] shall be vested in the carrier.” Id. at 1112. Because those provisions merely identified the plan administrator’s tasks, but bestowed no power to interpret the plan, we applied de novo review. Id. at 1113.3 [p.9639]
[5] Here, by contrast, the plan bestows on the administrator the responsibility to interpret the terms of the plan and to determine eligibility for benefits.
It goes further by giving the administrator “full and final” authority and cautions that this authority “rests exclusively” with the plan administrator.
“Discretion” means, as commonly understood, simply “the power or right to decide or act according to one’s own judgment.” Random House Unabridged Dictionary 411 (1969).
By giving the plan administrator “full and final” authority, and vesting such authority “exclusively” in the administrator, this policy clearly gave to the plan administrator the power to decide according to its own judgment.
Under Firestone, the common meaning of “discretion,”4 our own precedents, and the persuasive precedents of other circuits, this provision is sufficient to vest discretion in the plan administrator.
Accordingly, under Firestone, de novo review does not apply; abuse of discretion review does. [p.9640]
2. Abuse of Discretion Review in the Face of a Conflict of Interest
Firestone appears to provide for only two alternatives. When a plan confers discretion, abuse of discretion review applies; when it does not, de novo review applies. 489 U.S. at 115.
Abuse of discretion review applies to a discretion-granting plan even if the administrator has a conflict of interest.5
But Firestone also makes clear that the existence of a conflict of interest is relevant to how a court conducts abuse of discretion review.
In discussing abuse of discretion review, the Supreme Court cautioned that, “if a benefit plan gives discretion to an administrator or fiduciary who is operating under a conflict of interest, that conflict must be weighed as a ‘facto[r] in determining whether there is an abuse of discretion.’ Restatement (Second) of Trusts § 187, Comment d (1959).” Firestone, 489 U.S. at 115.
More recently, the Court has noted that a conflict of interest in an ERISA case can affect judicial review.
See Rush Prudential HMO, Inc. v. Moran, 536 U.S. 355, 384 n.15 (2002) (stating that, under Firestone, abuse of discretion review should “home in on any conflict of interest on the fiduciary’s part” and that “[i]t is a fair question just how deferential the review can be when the judicial eye is peeled for conflict of interest”).
We have held that an insurer that acts as both the plan administrator and the funding source for benefits operates under what may be termed a structural conflict of interest.
See
Tremain v. Bell Indus., Inc., 196 F.3d 970, 976 (9th Cir. 1999) (noting that a conflict of interest exists when an insurer both
[p.9641] administers and funds an ERISA plan).
On the one hand, such an administrator is responsible for administering the plan so that those who deserve benefits receive them.
On the other hand, such an administrator has an incentive to pay as little in benefits as possible to plan participants because the less money the insurer pays out, the more money it retains in its own coffers.
See Doe v. Group Hosp. & Med. Servs., 3 F.3d 80, 86 (4th Cir. 1993) (noting that “to the extent that [the administrator] has discretion to avoid paying claims, it thereby promotes the potential for its own profit”);
Brown v. Blue Cross & Blue Shield of Ala., Inc., 898 F.2d 1556, 1561 (11th Cir. 1990) (similarly noting that an administrator’s role as a fiduciary role lies in conflict with its role as a profitmaking entity).
As the Supreme Court indicated in Firestone, such an inherent conflict of interest, even if merely formal and unaccompanied by indicia of bad faith or self-dealing, ought to have some effect on judicial review.
The question is, what effect?
[6] This is not the first time that we have considered what standard of review to apply in ERISA conflict of interest
cases.
A little over 11 years ago, in Atwood, we held that the existence of a structural conflict of interest did not necessarily
alter the standard of review. 45 F.3d at 1322-23.
We required a plan participant to present “material, probative evidence, beyond the mere fact of the apparent conflict, tending to show that the fiduciary’s self-interest caused a breach of the administrator’s fiduciary obligations to the beneficiary.” Id. at 1323.
If the participant did so, the burden then shifted to the administrator to prove that the conflict of interest did not affect its
decision to deny benefits.
If the plan could not carry that burden, we held that the court would give no deference to the administrator’s decision to deny benefits, but would instead review the decision de novo. Id.
We have followed Atwood in a number of cases, with varying degrees of success in sorting out the burden-shifting analysis.
See, e.g., Hensley v. Nw. Permanente P.C. Ret. Plan & Trust, 258 F.3d 986, 994-95 & n.5 (9th Cir. 2001) (noting that district courts have found inconsistencies in the Ninth Circuit’s approach to conflict of interest cases);
Pinto v. Reliance Standard Life Ins. Co., 214 F.3d 377, 385 (3d Cir. 2000) (observing that Ninth Circuit precedent in conflict of interest cases is unclear).
[7] Atwood’s failure to follow Supreme Court precedent, and its placement of an unreasonable burden on ERISA plaintiffs, requires that we overrule it.
Atwood goes wrong in three ways.
First and foremost, it does not adhere to the dichotomy explicitly laid out in Firestone: Plans granting discretion to
the administrator receive abuse of discretion review for their decisions denying benefits, while plans that do not confer discretion
on the administrator have their decisions reviewed de novo.
Atwood’s back-and-forth burden shifting disobeys the Supreme Court’s guidance.
Second, and relatedly, Atwood ignores the Supreme Court’s requirement that a court weigh as a “factor” in abuse of discretion
review the conflict of interest that inheres when a plan administrator also acts as its fiduciary. Firestone, 489 U.S. at 113.
The Court’s articulated approach, abuse of discretion review that is informed by the presence of a conflict of interest, was not created arbitrarily.
The Court recognized in Firestone that “ERISA abounds with the language and terminology of trust law,” 489 U.S. at 110, and the Court therefore looked to trust law in formulating the proper standard of review.
The ERISA fiduciary is invested with the responsibilities typical of a trustee, see 29 U.S.C. § 1104, and the abuse of discretion standard for ERISA plan administrators follows directly from the review given to the discretionary actions of trustees, see Firestone, 489 U.S. at 111-12; Restatement (Second) of Trusts § 187 (1959).
As comment d to the Restatement makes clear, key factors in determining whether or not a trustee has abused discretion include
“the motives of the trustee in exercising or refraining [p.9643] from exercising [a power granted to the trustee]; [and]
the existence or nonexistence of an interest in the trustee conflicting with that of the beneficiaries.” Id. § 187 cmt. d.
Our approach under Atwood fails to consider the motives and interests of a conflicted ERISA fiduciary in denying claims to
protect its own financial interests whenever the conflict of interest is not significant enough to require de novo review.
For those cases, Atwood grants the deference due under trust law but skips the careful review that trust law demands of
actions taken by obviously conflicted parties.
At the same time, Atwood gives no deference at all to significantly conflicted administrators even when the plan grants them discretion, again contrary to trust principles and to Firestone.
Third, Atwood places on plan participants the burden of producing evidence of the plan administrator’s motives, evidence that an ERISA plan participant is much less likely to possess than is the administrator.
See Pinto, 214 F.3d at 389 (noting the inequity of requiring direct evidence of a conflict of interest to appear in the administrator’s decision).
In the absence of such “smoking gun” evidence, Atwood grants administrators highly deferential review.
That approach wrongly aligns incentives.
Instead of being encouraged affirmatively to demonstrate their impartiality and the reasonableness of their decisions, plan administrators are rewarded for suppressing dissent and denying claims with as little explanation as possible.
In view of those problems, we overrule Atwood in its entirety and, instead, adopt an approach that, we believe, more accurately reflects the Supreme Court’s instructions in Firestone.
[8] We read Firestone to require abuse of discretion review whenever an ERISA plan grants discretion to the plan administrator, but a review informed by the nature, extent, and [p.9644] effect on the decision-making process of any conflict of interest that may appear in the record.
This standard applies to the kind of inherent conflict that exists when a plan administrator both administers the plan and funds it, as well as to other forms of conflict.
Our approach is substantially similar to that adopted by several other circuits, but with a conscious rejection of their “sliding scale” metaphor.
See Stup v. Unum Life Ins. Co. of Am., 390 F.3d 301, 307 (4th Cir. 2004) (applying a slidingscale abuse of discretion review in conflict of interest cases; a court must apply less deference “ ‘to the degree necessary to neutralize any untoward influence resulting from the conflict’ ” (quoting Doe, 3 F.3d at 87));
Fought v. Unum Life Ins. Co. of Am., 379 F.3d 997, 1004 (10th Cir. 2004) (per curiam) (adopting sliding-scale abuse of discretion review, in which “ ‘the court must decrease the level of deference given to the conflicted administrator’s decision in proportion to the seriousness of the conflict’ ” (quoting Chambers v. Family Health Plan Corp., 100 F.3d 818, 825 (10th Cir. 1996))), cert. denied, 544 U.S. 1026 (2005);
Pinto, 214 F.3d at 379 (expressly adopting the sliding-scale approach, which “intensif[ies] the degree of scrutiny to match the degree of the conflict”);
Vega v. Nat’l Life Ins. Servs., Inc., 188 F.3d 287, 297 (5th Cir. 1999) (en banc) (reaffirming that the court applies a
sliding scale so that “[t]he greater the evidence of conflict on the part of the administrator, the less deferential [the] abuse of discretion standard will be”);
Woo v. Deluxe Corp., 144 F.3d 1157, 1161 (8th Cir. 1998) (adopting the sliding-scale approach, which requires a decrease in the deference given to an ERISA plan administrator’s decision in proportion to the gravity of the conflict of interest).6
[p.9645]
Insofar as those cases recognize that weighing a conflict of interest as a factor in abuse of discretion review requires a case-by-case balance, we agree.
A district court, when faced with all the facts and circumstances, must decide in each case how much or how little to credit the plan administrator’s reason for denying insurance coverage.
An egregious conflict may weigh more heavily (that is, may cause the court to find an abuse of discretion more readily) than a minor, technical conflict might.
But in any given case, all the facts and circumstances must be considered and nothing “slides,” so we find the metaphor unnecessary and potentially confusing.
A straightforward abuse of discretion analysis allows a court to tailor its review to all the circumstances before it.
See Woo, 144 F.3d at 1161 (“The abuse of discretion standard is inherently flexible, which enables reviewing courts to simply
adjust for the circumstances.”).
The level of skepticism with which a court views a conflicted administrator’s decision may [p.9646] be low if a structural conflict of interest is unaccompanied, for example, by any evidence of malice, of self-dealing, or of a parsimonious claims-granting history.
A court may weigh a conflict more heavily if, for example,
the administrator provides inconsistent reasons for denial, Lang, 125 F.3d at 799;
fails adequately to investigate a claim or ask the plaintiff for necessary evidence, Booton v. Lockheed Med. Benefit Plan,
110 F.3d 1461, 1463-64 (9th Cir. 1997);
fails to credit a claimant’s reliable evidence, Black & Decker Disability Plan v. Nord, 538 U.S. 822, 834 (2003); or
has repeatedly denied benefits to deserving participants by interpreting plan terms incorrectly or by making decisions against the weight of evidence in the record.
We recognize that abuse of discretion review, with any “conflict . . . weighed as a factor,” Firestone, 489 U.S. at 115, is indefinite.
We believe, however, that trial courts are familiar with the process of weighing a conflict of interest.
For example, in a bench trial the court must decide how much weight to give to a witness’ testimony in the face of some evidence of bias.
What the district court is doing in an ERISA benefits denial case is making something akin to a credibility determination about the insurance company’s or plan administrator’s reason for denying coverage under a particular plan and a particular set of medical and other records.
We believe that district courts are well equipped to consider the particulars of a conflict of interest, along with all the other facts and circumstances, to determine whether an abuse of discretion has occurred.
The careful, case-by-case approach that we adopt also alleviates the unreasonable burden Atwood placed on ERISA plaintiffs.
Under Atwood, we would consider the influence of the plan administrator’s conflict only if the plaintiff brought forth evidence of a “serious conflict of interest,” triggering de novo review. Gatti v. Reliance Standard Life Ins. Co., 415
F.3d 978, 985 (9th Cir. 2005) (as amended).
If the plaintiff could not make that threshold showing, we would uphold an [p.9647] administrator’s decision so long as it was “grounded on any reasonable basis.” Jordan v. Northrop Grumman Corp. Welfare Benefit Plan, 370 F.3d 869, 875 (9th Cir. 2004) (internal quotation marks omitted).
Going forward, plaintiffs will have the benefit of an abuse of discretion review that always considers the inherent conflict when a plan administrator is also the fiduciary, even in the absence of “smoking gun” evidence of conflict.
Moreover, a conflicted administrator, facing closer scrutiny, may find it advisable to bring forth affirmative evidence that any conflict did not influence its decisionmaking process, evidence that would be helpful to determining whether or not it has abused its discretion.7
3. Evidence That a Court May Consider
When a plan participant sues a plan administrator, challenging its decision to deny benefits, what evidence may a court consider in determining how deferentially to review the decision to deny the claim?
The answer depends on whether review is de novo (because the plan failed to confer discretion on the administrator) or for abuse of discretion (because the plan unambiguously conferred discretion).
Many circuits limit a district court to the administrative record when the court is reviewing a case on the merits for an abuse of discretion; consideration of new evidence is permitted only in conjunction with de novo review of a denial of benefits.
See Urbania v. Cent. States, Se. & Sw. Areas Pension Fund, 421 F.3d 580, 586 (7th Cir. 2005) (noting that “[d]eferential review of an administrative decision means review on the administrative record” (internal quotation [p.9648] marks omitted));
Kosiba, 384 F.3d at 67 n.5 (noting that, “in general, the record for arbitrary-and-capricious review of ERISA benefits denial is the record made before the plan administrator”);
Fought, 379 F.3d at 1003 (noting that courts are limited to the administrative record when reviewing for abuse of discretion);
Zervos v. Verizon N.Y., Inc., 252 F.3d 163, 173 (2d Cir. 2001) (noting that when review is for abuse of discretion, the record consists of the administrative record);
Elliott v. Sara Lee Corp., 190 F.3d 601, 608 & n.6 (4th Cir.1999) (noting that on de novo review, a court may consider
extra-judicial evidence, but stating that abuse of discretion review must be based on the evidence before the administrator);
Vega, 188 F.3d at 300 (restricting review to the administrative record when the court is considering the administrator’s factual determinations for abuse of discretion);
Buckley v. Metro. Life, 115 F.3d 936, 941 & n.2 (11th Cir. 1997) (per curiam) (holding that extra-record evidence, presented to the district court on review for abuse of discretion, was irrelevant).
Indeed, we have adhered to a similar rule.
See Jebian v.vHewlett-Packard Co. Employee Benefits Org. Income Prot.vPlan, 349 F.3d 1098, 1110 (9th Cir. 2003) (“While under anvabuse of discretion standard our review is limited to thevrecord before the plan administrator, this limitation does notvapply to de novo review.” (citation omitted));
Kearney, 175 F.3d at 1090-91 (holding that the standard of review informs the amount of evidence that a district court may consider);
Mongeluzo v. Baxter Travenol Long Term Disability Benefit Plan, 46 F.3d 938, 944 (9th Cir. 1995) (holding that the district
court has discretion to allow evidence that was not before the plan administrator “only when circumstances clearly establish that additional evidence is necessary to conduct an adequate de novo review” (internal quotation marks omitted)).
A subtler question arises when a court must decide how much weight to give a conflict of interest under the abuse of discretion standard.
In making that determination, the court[p.9649] may consider evidence outside the record.
We have held that the court may consider evidence beyond that contained in the administrative record that was before the plan administrator, to determine whether a conflict of interest exists that would affect the appropriate level of judicial scrutiny.
See Tremain, 196 F.3d at 976-77 (holding that a court may consider extrarecord evidence to determine whether the administrator was plagued by a conflict of interest);
see also Kosiba v. Merck & Co., 384 F.3d 58, 67 n.5 (3d Cir. 2004) (holding that a district court may supplement the record in order to decide whether a conflict of interest exists), cert. denied, 544 U.S. 1044 (2005).
[9] Today, we continue to recognize that, in general, a district court may review only the administrative record when
considering whether the plan administrator abused its discretion, but may admit additional evidence on de novo review.
That principle is consistent with Tremain, 196 F.3d at 976-79, which permits extrinsic evidence on the question of a conflict
of interest.
The district court may, in its discretion, consider evidence outside the administrative record to decide the nature, extent, and effect on the decision-making process of any conflict of interest; the decision on the merits, though, must rest on the administrative record once the conflict (if any) has been established, by extrinsic evidence or otherwise.
See Doe v. Travelers Ins. Co., 167 F.3d 53, 57 (1st Cir. 1999) (holding that, when deciding what record a court should use
to decide whether the administrator’s decision was reasonable, “[i]t is not clear that any single answer covers all of the variations in ERISA cases; the ‘record’ may depend on what has been decided, by whom, based on what kind of information, and also on the standard of review and the relief sought”).
B. What standard of review should the district court apply when the administrator fails to follow procedural requirements?
In the preceding sections, we have discussed how courts review a challenged denial of ERISA benefits when a plan [p.9650] participant disagrees with the administrator’s interpretation of the record or with its application of the plan’s terms to the facts.
Different concerns arise when the administrator fails to adhere to the procedural dictates of ERISA and the plan.
We must consider those issues in this case because of the plan administrator’s last-minute reliance on a new ground for denial of benefits, which afforded Plaintiff no opportunity to present relevant evidence in advance of the administrator’s final decision.
1. Procedural Violations Amounting to Failure to Exercise Discretion
Under ERISA, plan administrators must follow certain practices when processing and deciding plan participants’ claims.
For example, administrators must adhere to various procedures for giving notice, reporting, and claims processing.
See 29 U.S.C. § 1021(a) (disclosure to all plan participants); Id. § 1021(b) (reporting requirements); Id. § 1133
(claims procedures); 29 C.F.R. § 2560.503-1 (same).
We have recently held that an administrator’s failure to comply with such procedural requirements ordinarily does not alter the standard of review.
See Gatti, 415 F.3d at 985 (holding that an administrator who violates procedural requirements under ERISA usually will not be subject to a different standard of judicial review).
There are, however, some situations in which procedural irregularities are so substantial as to alter the standard of review.
In Gatti, we held that “procedural violations of ERISA do not alter the standard of review [from abuse of discretion review to de novo review] unless the violations are so flagrant as to alter the substantive relationship between the employer and employee, thereby causing the beneficiary substantive harm.” Id.
We cited Blau v. Del Monte Corp., 748 F.2d 1348 (9th Cir. 1984), abrogation on other grounds recognized by Dytrt v. Mountain State Tel. & Tel. Co., 921 F.2d 889, 894 n.4 (9th Cir. 1990), as an example of this kind of egregious act.
Gatti, 415 F.3d at 984 85.8
In Blau, the administrator had kept the policy details secret from the employees, offered them no claims procedure, and did not provide them in writing the relevant plan information; in other words, the administrator “failed to comply with virtually every applicable mandate of ERISA.” 748 F.2d at 1353.
When an administrator engages in wholesale and flagrant violations of the procedural requirements of ERISA, and thus acts in utter disregard of the underlying purpose of the plan as well, we review de novo the administrator’s decision to deny benefits.
We do so because, under Firestone, a plan administrator’s decision is entitled to deference only when the administrator exercises discretion that the plan grants as a matter of contract. 489 U.S. at 111.
Firestone directs, consistent with trust law principles, that “a deferential standard of review [is] appropriate when a trustee exercises discretionary powers.” Id. (emphasis added).
Because an administrator cannot contract around the procedural requirements of ERISA, decisions taken in wholesale violation of ERISA procedures do not fall within an administrator’s discretionary authority.
In general, we review de novo a claim for benefits when an administrator fails to exercise discretion.
See Jebian, 349 F.3d at 1106 (holding that an administrator failed to exercise its discretion when it did not make a benefits decision within the 60 days specified by the terms of the plan and the applicable regulation, so that the ultimate decision rendered was “undeserving of deference”).
Other circuits have also held that review is de novo when the plan administrator fails to exercise discretion.
See Nichols v. Prudential Ins. Co. of Am., 406 F.3d 98, 109 (2d Cir. 2005) (holding that a “deemed denied” [p.9652]
claim, in which the administrator did not issue a decision within the time required by the regulations, constituted “inaction,” which was not an exercise of discretion and which therefore was entitled to no deference; de novo review applied);
Gilbertson v. Allied Signal, Inc., 328 F.3d 625, 632 (10th Cir. 2003) (noting that “[d]eference to the administrator’s
expertise is inapplicable where the administrator has failed to apply his expertise to a particular decision”);
Gritzer v. CBS, Inc., 275 F.3d 291, 296 (3d Cir. 2002) (“Where a trustee fails to act or to exercise his or her discretion, de novo review is appropriate because the trustee has forfeited the privilege to apply his or her discretion . . . .”).
Similarly, when a plan administrator’s actions fall so far outside the strictures of ERISA that it cannot be said that the administrator exercised the discretion that ERISA and the ERISA plan grant, nodeference is warranted.
This case does not, however, fall into that rare class of cases.
Instead, we face the more ordinary situation in which a plan administrator has exercised discretion but, in doing so, has made procedural errors.
We turn, finally, to a discussion of how we are to consider such procedural errors in reviewing a denial of benefits.
2. Procedural Violations in the Course of Exercising Discretion
As noted, a procedural irregularity in processing an ERISA claim does not usually justify de novo review.
See Gatti, 415 F.3d at 985 (concluding that the district court had erred by allowing “de novo review any time a benefits administrator violates the procedural requirements in ERISA’s regulations, no matter how small or inconsequential the violation”).
That generalization does not mean, however, that procedural irregularities are irrelevant to the court’s analysis.
[10] A procedural irregularity, like a conflict of interest, is a matter to be weighed in deciding whether an administrator’s [p.9653] decision was an abuse of discretion.
See Fought, 379 F.3d at 1006 (concluding that an inherent conflict of interest, a proven conflict of interest, or a serious procedural irregularity reduces the deference owed to an administrator’s decision to deny benefits);
Woo, 144 F.3d at 1160 (noting that a conflict of interest or a procedural irregularity can heighten judicial scrutiny).
When an administrator can show that it has engaged in an “ ‘ongoing, good faith exchange of information between the administrator and the claimant,’ ” the court should give the administrator’s decision broad deference notwithstanding a minor irregularity. Jebian, 349 F.3d at 1107 (quoting Gilbertson, 328 F.3d at 635);
see also Robinson v. Aetna Life Ins. Co., 443 F.3d 389, 392-93 (5th Cir. 2006) (applying a substantial compliance standard to alleged procedural violations under ERISA).
A more serious procedural irregularity may weigh more heavily.
3. Evidence That a Court May Consider
[11] When a plan administrator has failed to follow a procedural requirement of ERISA, the court may have to consider
evidence outside the administrative record.
For example, if the administrator did not provide a full and fair hearing, as required by ERISA, 29 U.S.C. § 1133(2), the court must be in a position to assess the effect of that failure and, before it can do so, must permit the participant to present additional evidence.
We follow the Sixth Circuit in holding that, when an administrator has engaged in a procedural irregularity that has affected the administrative review, the district court should “reconsider [the denial of benefits] after [the plan participant] has been given the opportunity to submit additional evidence.” VanderKlok v. Provident Life & Accident Ins. Co., 956 F.2d 610, 617 (6th Cir. 1992).
As we noted earlier, if the plan administrator’s procedural defalcations are flagrant, de novo review applies.
And as we also noted, when de novo review applies, the court is not limited to the administrative record and may take additional evidence.
[12a] Even when procedural irregularities are smaller, though, and abuse of discretion review applies, the court may take additional evidence when the irregularities have prevented full development of the administrative record. In that way the court may, in essence, recreate what the administrative record would have been had the procedure been correct.
C. The district court erred in analyzing Plaintiff’s claim
Finally, we must consider whether the district court in the present case erred under the principles that we have established.
We conclude that the court erred in three respects in analyzing Plaintiff’s claim.
First, the court failed to examine the nature, extent, and effect on the decision-making process of Alta’s conflict of
interest in assessing whether Alta had abused its discretion; and the court followed Atwood’s burden-shifting regime.
Of course, this error is understandable because the court did not have the benefit of this opinion, which recasts the terms of the
exercise.
See Jebian, 349 F.3d at 1110 & n.10 (concluding that a remand to the district court was necessary, even if no new evidence were to be admitted, because the court had to review the evidence under a different standard of review, placing it in a different role than it had occupied originally).
Second, the district court erred by failing to make all required findings of fact.
The court conducted a bench trial, but failed to make findings of fact on all contested issues.
See Fed. R. Civ. P. 52(a);
see also Unt v. Aerospace Corp., 765 F.2d 1440, 1444 (9th Cir. 1985) (holding that factual findings made by a judge after a bench trial “must be explicit enough to give the appellate court a clear understanding of the basis of the trial court’s decision, and to enable it to determine the ground on which the trial court reached its decision” (internal [p.9655] quotation marks omitted)).
Specifically, the district court declined to decide whether or not a waiver of premium application was submitted to Alta’s predecessor on behalf of Dr. Abatie.
Instead, the court reviewed the evidence both supporting and undermining Plaintiff’s claim that a waiver application had been submitted.
The court appeared to conclude simply that the administrator did not abuse its discretion because there was evidence on both sides of the issue.
Were the court to find that a waiver application in fact had been submitted to Alta’s predecessor on Dr. Abatie’s behalf, then it is likely that the administrator abused its discretion when it denied the claim.
On the other hand, if the court were to find that neither Dr. Abatie nor the Clinic submitted a waiver application, then the administrator likely did not abuse its discretion when it relied on this reason to deny the claim.
Third, the district court neglected to consider the procedural irregularities that occurred when Alta processed Plaintiff’s claim.
Alta originally denied Plaintiff’s claim for life insurance benefits because it concluded that no waiver of premium application had been submitted on behalf of Dr. Abatie.
Later, in its final denial of Plaintiff’s claim, Alta continued to rely on that reason, but also added a second reason—that Plaintiff had provided insufficient evidence to show that Dr. Abatie had remained totally disabled from the time he left work at the Clinic until his death.
[12b] An administrator must provide a plan participant with adequate notice of the reasons for denial, 29 U.S.C.
§ 1133(1), and must provide a “full and fair review” of the participant’s claim, Id. § 1133(2);
see also 29 C.F.R. § 2560.503-1(g)(1), (h)(2).
When an administrator tacks on a new reason for denying benefits in a final decision, thereby precluding the plan participant from responding to that rationale for denial at the administrative level, the administrator violates ERISA’s procedures.9
“[S]ection 1133 requires an [p.9656] administrator to provide review of the specific ground for an adverse benefits decision.” Robinson, 443 F.3d at 393.
By requiring that an administrator notify a claimant of the reasons for the administrator’s decisions, the statute suggests that
the specific reasons provided must be reviewed at the administrative level. Id.
Moreover, a review of the reasons provided by the administrator allows for a full and fair review of the denial decision, also required under ERISA. Id.
Accordingly, an administrator that adds, in its final decision, a new reason for denial, a maneuver that has the effect of insulating the rationale from review, contravenes the purpose of ERISA.
This procedural violation must be weighed by the district court in deciding whether Alta abused its discretion.
[13] In this case, Plaintiff presented additional evidence—a declaration from Dr. Abatie’s treating doctor—to prove that
Dr. Abatie had remained totally disabled continuously from the date he left work until the date he died.
The district court declined to consider that evidence.
Under our analysis today, the district court erred by refusing to consider the additional evidence, if the court does not first find that Plaintiff’s claim is doomed by a failure to request a waiver of premiums.
REVERSED and REMANDED for further proceedings consistent with this opinion.
1 Abatie v. Alta Health & Life Ins. Co., 421 F.3d 1053 (9th Cir. 2005), reh’g en banc granted, 437 F.3d 860 (9th Cir. 2006).
2 Plaintiff sued, rather than appealing the initial adverse determination to Alta itself.
The ERISA statutes do not require exhaustion of administrative remedies before a claimant can bring an action in court, but our cases suggest that a claimant must exhaust administrative remedies first.
See Amato v. Bernard, 618 F.2d 559, 566-68 (9th Cir. 1980) (holding that, although the statutes do not specify an exhaustion requirement, the legislative history and text of the statutes show that Congress intended to grant the courts authority to apply an exhaustion requirement in ERISA cases).
Exhaustion is not an issue here, because both parties agreed to supplement the administrative record and to give Alta a second chance to review the evidence and to make a new final determination about Plaintiff’s claim for benefits.
3 The court in Ingram suggested that it would be “easy enough” for a plan to confer discretion unambiguously just by using the word “discretion” or a synonym. 244 F.3d at 1113-14.
Nevertheless, Ingram did not hold that the failure to include the very term “discretion” required application of a de novo standard; instead, the court analyzed the policy in detail, Id. at 1112-13, just as we do here.
We also note that the insurance policy in dispute in this case was drafted in 1992, nine years before we published Ingram.
Under the law of our circuit as of 1992, a plan would be held to confer discretion if it “include[d] even one important discretionary element.” Bogue v. Ampex Corp., 976 F.2d 1319, 1325 (9th Cir. 1992);
see also Eley v. Boeing Co., 945 F.2d 276, 278 & n.2 (9th Cir. 1991) (holding that a plan conferred discretion so long as it gave the company the power to determine eligibility for benefits).
The drafters of the policy in question, had they studied applicable Ninth Circuit cases, would not have doubted that the policy conferred discretion.
4 Judge Kleinfeld’s concurrence contends that the key provision of the plan is ambiguous because, in the absence of the very word “discretion,” a reasonable person could read the provision not to grant discretion.
We disagree.
If a college told its students that “the responsibility for full and final determinations of grades, interpretation of course requirements, determinations of credits, and appeals of grades or credits, rests exclusively with the college,” no reasonable student would doubt that the college has the power and right to use its judgment in good faith and to make conclusive decisions, free from de novo reconsideration by an outside body such as a court.
5 The Court did not catalogue the full range of types of conflicts of interest, but it suggested that a conflict exists when a plan administrator (which acts as a fiduciary toward the plan participants, who are beneficiaries) is also the sole source of funding for an unfunded plan; this was Firestone’s situation. Firestone, 489 U.S. at 105, 115.
6 Other circuits have developed different approaches to determine the applicable standard of review in ERISA benefits denial cases.
See, e.g., Rud v. Liberty Life Assurance Co., 438 F.3d 772, 777 (7th Cir. 2006) (holding that a structural conflict of interest, without more, does not affect the standard of review and requiring the claimant to prove that the alleged conflict affected the administrator’s decision);
Wright v. R.R. Donnelley & Sons Co. Group Benefits Plan, 402 F.3d 67, 74 (1st Cir. 2005) (holding that an inherent conflict of interest does not necessarily affect the abuse of discretion standard, and placing the burden on the claimant to demonstrate that a conflict exists);
HCA Health Servs., Inc. v. Employers Health Ins. Co., 240 F.3d 982, 993-94 (11th Cir. 2001) (applying de novo review, initially, to decide whether the claim was wrongly decided, and if an inherent conflict of interest exists, requiring the administrator to prove that its interpretation was not tainted by self-interest);
Sullivan v. LTV Aerospace & Def. Co., 82 F.3d 1251, 1255-56 (2d Cir. 1996) (requiring a claimant to show that a conflict of interest affected the reasonableness of the administrator’s decision; if the claimant carries that burden then de novo review applies).
At least one circuit has declined to establish a standard of review in conflict of interest cases.
See Wagener v. SBC Pension Benefit Plan—Non Bargained Program, 407 F.3d 395, 402 (D.C. Cir. 2005) (declining to establish the standard of review appropriate in conflict of interest cases because, in the case at hand, under any standard of review, the administrator’s actions were unreasonable).
For a detailed analysis of Firestone’s progeny and the various circuits’ approaches to this issue, see Kathryn J. Kennedy, Judicial Standards of Review in ERISA Benefit Claim Cases, 50 Am. U. L. Rev. 1083 (2001).
7 For example, the administrator might demonstrate that it used truly independent medical examiners or a neutral, independent review process; that its employees do not have incentives to deny claims; that its interpretations of the plan have been consistent among patients; or that it has minimized any potential financial gain through structure of its business (for example, through a retroactive payment system).
8 Blau pre-dated Firestone, so its analysis of the extant “arbitrary and capricious” standard of judicial review is irrelevant. But Blau illustrates the type of procedural noncompliance that, under our post-Firestone cases, allows for more stringent judicial review. Gatti, 415 F.3d at 985.
9 In Lang v. Long-Term Disability Plan of Sponsor Applied Remote Technology, Inc., 125 F.3d 794, 798-99 (9th Cir. 1997), we held that the court should review de novo the decision of a plan administrator that gave one reason in its initial denial, but changed reasons in its final denial.
Lang used the Atwood analysis and held that the administrator’s last-minute switch in the reason for denial suggested serious self-dealing. Id.
Although a change in reasoning can suggest a conflict of interest, it also can be categorized as a procedural irregularity where, as here, the plan participant is foreclosed from presenting any response to the new reason.
KLEINFELD, Circuit Judge, with whom RAWLINSON, Circuit Judge, joins, concurring in the judgment:
I concur in the judgment, but not the reasoning.
In my view, the plan does not confer discretion.
The district court should have reviewed whether the premium waiver for disability applied to Dr. Abatie de novo.
That is the default standard of review under ERISA.1
The ERISA plan at issue in this case does not say that the trustee has “discretion” to construe its terms and determine whether a person is entitled to plan benefits.
The majority concedes that the plan does not confer discretion in so many words, but says that no “magic words” are necessary.
Our en banc decision in Kearney v. Standard Ins. Co.2 held that we require the administrator be “unambiguous” in retaining discretion.
Ingram v. Martin Marietta3 applies Kearney, explaining that we “examine the text of [the] plan to determine whether it “unambiguously” states that [the administrator] has ‘discretionary authority’ in making benefits decisions.”4
The majority claims to reaffirm the holdings of Kearney and Ingram that the plan must “unambiguously” confer discretion,5 yet it finds discretion in the face of ambiguity.
This plan says that the “responsibility” for “full and final” benefits determinations rests “exclusively” upon the insurance company.
Is the “responsibility” to make a decision the same thing as discretion?
Maybe, maybe not.
One reading of the plan language is that it says who makes benefits determinations, not how they are to be made.6
If two readings are reasonable, then the language is ambiguous and the plan does not “unambiguously” confer discretion.7
The majority does not really say why we should construe a plan to confer discretion on the trustees where the plan does not plainly say so.
Calling a plain language requirement “magic words” expresses a feeling, not an argument.
The majority’s analogy to a college’s grading policy is inapposite.
People normally expect that their college grades will be determined somewhat subjectively by their professors but they do not expect that their insurance company will subjectively determine whether to pay their bills when they get sick.
So we are left with no reason not to require the plan to say “discretion” if that is what it means.
There are good reasons for requiring plain talk in this, as in so many things.
“Discretion” is not just a means by which courts can easily get rid of complicated ERISA cases.
What it means in practical affairs is that, if the administrator could reasonably decide either way, then it can decide against the
claimant and there is no recourse.
That means a lot of people who ought to get life insurance proceeds, disability benefits, or medical expense coverage will not get the coverage they should and, under a sounder reading of the evidence, would.
The power to deny claims that could reasonably be resolved either way is very significant, so we ought to require plans to say so explicitly when they reserve discretion.
And saying so is easy.
ERISA plans are not written like contracts between two lay people trying to find the words for a vague [p.9659] arrangement, like a hair salon proprietor and a person to whom she leases a chair.
Lawyers write these policies using form books, case law research, and extensive consultation.
They can use the word “retain discretion” as easily as they can use the “magic words” traditionally used in deeds, if they mean them.
If the administrator does not say that it “retains discretion,” there is probably a good marketing reason why not.
A business might not want to buy a plan that gives the administrator discretion to deny coverage whenever it is arguable.
Its employees could be left high and dry, and those employees include the executives who determine which group policies to buy.
It is easier to sell insurance on the promise that the insurance company will pay the doctor bills than if the promise is only that the insurance company will take a look at it and decide whether to exercise its discretion to pay the bill.
Forcing the administrator to say that it “retains discretion” gives the purchasers of group plans and their employees fair notice
of how much protection they have.
And an administrator might choose weasel words to evade regulation yet retain discretion for purposes of claims litigation.
States regulate insurance policies, and the National Association of Insurance Commissioners has adopted a model act8 saying that no health or disability insurance policy “may contain a provision purporting to reserve discretion.”9
Some states have adopted the scheme10 and, while California has [p.9660] not, its Department of Insurance has issued an opinion to the same effect.11
So the practical effect of today’s majority opinion is that a group insurer may avoid regulatory problems by using ambiguous terms rather than “magic words,” but still enjoy the discretionary standard of review in court.
Two questions control whether the premium waiver applies in this case:
(1) whether the company timely applied for the waiver of premiums on account of disability, and
(2) whether Dr. Abatie was disabled for the entire time between when he quit working and when he died.
If the premium waiver applies, then Dr. Abatie’s widow is entitled to $331,500 in life insurance under the group policy.
If the waiver does not apply, she is not entitled.
In this case, a reasonable adjudicator could probably go either way on whether Dr. Abatie was disabled for the entire time after he stopped working, and on whether his insurer received the form required for a premium waiver.
But if review is de novo, Dr. Abatie’s widow has established genuine issues of fact.
The absence of the form in the insurance company records in this case is weaker evidence than usual, because the policy has bounced to three different insurers.
The records therefore may not have maintained their integrity and searchability through all those changes.
I agree that the district court should make findings of fact based on the evidence to determine whether the company received the form
and whether Dr. Abatie was continuously disabled up to his death.
As with any de novo determination, the question for the district court should not be whether the insurance company went about its determination the right way, but rather whether [p.9661] the form was sent in and whether Dr. Abatie was disabled for the requisite period.
The majority’s elaborate construct for resolving cases with apparent conflicts of interest is not practical and adds unpredictability
to group insurance determinations.
Unpredictability in group insurance determinations is a very bad thing.
It means that more health care and disability money has to be spent on claims processing instead of health care and disability payments.
And it means that people fighting over amounts too small to justify hiring a lawyer will get close questions resolved against them.
Further, it is impossible as a practical matter to identify conflicts of interest in the manner the majority suggests.
It is often difficult even for the insurance company to figure out what its interest is, let alone for someone else to do it.
Claim supervisors differ sharply in their philosophies and, when marketing people are thrown into the mix, the company often finds identifying its own interest to be a conundrum.
A socalled independent administrator may have much more of an incentive to decide against claimants than an insurance company spending “its own money.”
Independent administrators may want to show how tough they are on claims to better market their services to self-insured employers.
An insurance company may have an incentive to be more liberal than is appropriate because its experience-based premiums amount to
a cost-plus contract, such that the more it spends, the more it makes.
An employer that controls the administration of its group plans may have incentives to slant its decisions in favor of coverage in close cases.
Even though that will be money out of its pocket, the employer may want to make working there attractive by means of a reputation for good medical coverage.
Or it may seek to discourage unionization by providing benefits more liberally than union plans.
Or the employer may insist upon liberal administration out of altruism.
Or because the risk management department chief has a sick child.
We in the court system will never know whether [p.9662] there is a conflict of interest in the sense addressed by the majority opinion.
Courts have fallen into the unfortunate habit in ERISA cases of focusing entirely on the standard of review.
We treat abuse of discretion review as though it means the claimant loses, which is not necessarily so.
And we treat de novo review as though it means the claimant wins, which is also not necessarily so.
The focus should be on whether the claimant is entitled to the claimed benefits.
Today’s decision adds uncertainty.
1 Firestone Tire and Rubber Co. v. Bruch, 489 U.S. 101, 115 (1989).
2 Kearney v. Standard Ins. Co., 175 F.3d 1084, 1090 (9th Cir. 1999) (en banc).
3 Ingram v. Martin Marietta, 244 F.3d 1109, 1112 (9th Cir. 2001).
6 See Ingram, 244 F.3d at 1112-13 (“An allocation of decision-making authority to [the administrator] is not, without more, a grant of discretionary authority in making those decisions.”).
7 See Kearney, 175 F.3d at 1190 (“Only by excluding alternative readings as unreasonable could we conclude that the conferral of discretion is unambiguous.”).
8 See Model Act 42 “Prohibition on the Use of Discretionary Clauses Model Act,” NAIC 42-1 (2006).
9 See Id. at § 4(A) (“No policy, contract, certificate or agreement offered or issued in this state by a health carrier to provide, deliver, arrange for, pay for or reimburse any of the costs of health care services may contain a provision purporting to reserve discretion to the health carrier to interpret the terms of the contract, or to provide standards of interpretation or review that are inconsistent with the laws of this state.”).
10 See, e.g., Me. Rev. Stat. Ann. tit. 24-A§ 4303(11).
11 See Letter Opinion per CIC § 12921.9: Discretionary Clauses (February 26, 2004), available at http://www.insurance.ca.gov/0200-industry/0300-insurers/ 0200-bulletins/bulletin-notices-commiss-Opinion/upload/ Opinion-February-26-2004.pdf (last visited July 11, 2006); see also Mitchell v. Aetna Life Ins. Co., 359 F. Supp. 2d 880, 888-89 (C.D. Cal. 2005)(describing opinion).
I concur in all of Judge Graber’s opinion, except Part III.A.2.b., and concur in the judgment.
I agree with the crux of the analysis that we should overrule Atwood, and that our review under Firestone should be for abuse of discretion, taking into account any conflict of interest.
Rather than adopt yet another approach to this problem, however, I would follow those of our sister circuits that have adopted a “sliding scale” assessment: The degree of deference given an administrator’s decision should be reduced when the administrator has a conflict of interest, and the greater the conflict, the less the deference to be given.
See Stup v. Unum Life Ins. Co. of Am., 390 F.3d 301, 307 (4th Cir. 2004);
Fought v. Unum Life Ins. Co. of Am., 379 F.3d 997, 1004 (10th Cir. 2004) (per curiam);
Pinto v. Reliance Standard Life Ins. Co., 214 F.3d 377, 390-93 (3d Cir. 2000);
Vega v. Nat’l Life Ins. Servs., Inc., 188 F.3d 287, 297 (5th Cir. 1999) (en banc);
Woo v. Deluxe Corp., 144 F.3d 1157, 1161-62 (8th Cir. 1998).
[p.9663]