Ramos v. Banner Health

1 F.4th 769

• Court: Court of Appeals for the Tenth Circuit
• Decided: June 11, 2021
• Docket: 20-1231
• Status: Published

Opinion

FILED United States Court of Appeals Tenth Circuit

PUBLISH June 11, 2021 Christopher M. Wolpert UNITED STATES COURT OF APPEALS Clerk of Court

TENTH CIRCUIT

LORRAINE M. RAMOS; CONSTANCE R. WILLIAMSON; KAREN F. MCLEOD; ROBERT MOFFITT; CHERLENE M. GOODALE; LINDA ANN HEYRMAN; DELRI HANSON, individually and as representatives of a class of plan participants, on behalf of Banner Health Employees 401(k) Plan,

Plaintiffs-Appellants, v. No. 20-1231 BANNER HEALTH; BANNER HEALTH BOARD OF DIRECTORS; BANNER HEALTH RETIREMENT PLANS ADVISORY COMMITTEE; LAREN BATES; WILFORD A. CARDON; RONALD J. CREASMAN; GILBERT DAVILA; PETER S. FINE; SUSAN B. FOOTE; MICHAEL J. FRICK; MICHAEL GARNREITER; BARRY A. HENDIN; DAVID KIKUMOTO; LARRY S. LAZARUS; STEVEN W. LYNN; ANNE MARIUCCI; QUENTIN P. SMITH, JR.; CHRISTOPHER VOLK; CHERYL WENZINGER; BRENDA SCHAEFER; CHARLES P. LEHN; COLLEEN HALLBERG; DAN WEINMAN; DENNIS DAHLEN; ED NIEMANN, JR.; ED OXFORD; JEFF BUEHRLE; JENNIFER SHERWOOD; JULIE NUNLEY; MARGARET DEHAAN; PATRICIA K. BLOCK; PAULETTE FRIDAY; RICHARD O. SUTTON; ROBERT LUND; THOMAS R. KOELBL,

Defendants-Appellees.

APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLORADO (D.C. NO. 1:15-CV-02556-WJM-NRN)

Sean E. Soyars (Jerome J. Schlichter, Troy A Doles, and Heather Lea with him on the briefs) Schlichter, Bogard & Denton, St. Louis, Missouri, for Appellants.

Michael B. Kimberly (Margaret H. Warner, Jennifer B. Routh, and Matthew A. Waring with him on the brief), McDermott Will & Emery, Washington, D.C., for Appellees.

Before TYMKOVICH, Chief Judge, KELLY, and PHILLIPS, Circuit Judges.

TYMKOVICH, Chief Judge.

A class of employees who participated in Banner Health, Inc.’s 401(k) defined

contribution savings plan accused Banner and other plan fiduciaries of breaching duties

owed under the Employee Retirement Income Security Act. Following an eight-day

bench trial, the district court agreed in part, concluding that Banner had breached its

fiduciary duty to plan participants by failing to monitor its recordkeeping service

-2- agreement with Fidelity Management Trust Company. This failure to monitor resulted in

years of overpayment to Fidelity and corresponding losses to plan participants.

During the bench trial, the employees’ expert witness testified the plan participants

had incurred over $19 million in losses stemming from the breach. But having

determined the expert evidence on losses was not reliable, the court fashioned its own

measure of damages for the breach. The court calculated damages of about $1.6 million

and awarded prejudgment interest calculated at the Internal Revenue Service’s

underpayment rate. Also, despite finding that Banner breached its fiduciary duty, the

district court entered judgment for Banner on several of the class’s other claims: the court

found that Banner’s breach of duty did not warrant injunctive relief and that Banner had

not engaged in a “prohibited transaction” with Fidelity as defined by ERISA.

The class appeals the district court’s findings of fact and conclusions of law. The

class argues the district court adopted an improper method for calculating damages and

prejudgment interest, abused its discretion by denying injunctive relief, and erred in

entering judgment for Banner on the prohibited transaction claim. We AFFIRM the

district court in each instance.

I. Background

A. Factual Background

Banner is a large non-profit healthcare system, operating primarily in Arizona and

Colorado. As a benefit of employment, Banner sponsors and administers a 401(k)

-3- individual account, defined contribution plan for its employees (the Plan). Employees

who take part in the Plan are considered plan participants. During the period of time at

issue,1 plan participants could contribute to their individual retirement accounts and

Banner would match these contributions up to 4% of each participant’s salary. The Plan

made various investment options available to plan participants. These ranged from more

to less sophisticated options based on the level of involvement a participant wanted when

investing his funds. The value of each participant’s account is a function of contributions

and investment performance, minus the expenses associated with the Plan.

The Plan states that Banner “shall control and manage the operation and

administration of the Plan and make all decisions and determinations incident thereto.”

Aplt. App., Vol. II at 284. Banner then delegates these responsibilities to its CEO, who

appoints and supervises a Retirement Plan Advisory Committee. The Committee

oversees the Plan’s administration.

In 1999, the Committee hired Fidelity to provide recordkeeping and administrative

services to the Plan. In this role, Fidelity tracked participant contributions, maintained

investment options for participants, made service representatives available to participants,

1 The district court determined that November 20, 2009 to the date of judgment represented the relevant “Class Period.” The original complaint was filed on November 20, 2015. The statute of limitations for claims of breach of fiduciary duty under ERISA extend six years from “the date of the last action which constituted a part of the breach or violation” if the plaintiff did not have actual knowledge of the breach. 29 U.S.C. § 1113(1).

-4- and filed reports on the Plan’s performance with participants, Banner, and the

government. Until 2017, after this litigation began, Banner compensated Fidelity for its

services through an uncapped, revenue-sharing arrangement. Because the agreement was

based on revenue-sharing, Banner paid Fidelity based upon the total number of assets in

the plan. The more assets in the plan, the more Fidelity would make. And because the

agreement was uncapped, the arrangement did not set an upper limit on how much

Fidelity could make for providing services to the Plan.

Such uncapped, revenue-sharing arrangements are uncommon among Banner’s

peers. While uncapped, revenue-sharing arrangements are not unheard of in

compensating service providers, they are much less common than flat, per-participant

fees—agreements in which the compensation does not fluctuate based on the value of the

assets in a plan. Banner’s Plan is considered a “mega” plan—one with over $1 billion in

assets and 10,000 participants.2 Typically, such large plans are able to negotiate very

favorable per-participant fees. This is because the costs associated with servicing a plan

are primarily associated with the number of individual accounts rather than total assets.

And the overhead costs of servicing individual accounts levels off when a plan has many

participants. Thus, the market for service providers for mega plans was, and continues to

be, very competitive.

2 In December of 2009, the plan had 23,166 participants and $1.18 billion in assets. When Banner changed its arrangement with Fidelity in 2017, the Plan had grown to 41,416 participants and $2.25 billion in assets.

-5- Given this level of competition in the market, it is customary for plan fiduciaries to

test the market for service providers to ensure a plan is not overpaying for recordkeeping

and administrative services. This is traditionally done through a “request for proposals,”

essentially a gathering of bids from prospective service providers. One of the committee

members overseeing the Plan acknowledged that such market pricing should be done at

least every five to seven years. And yet, from 1999 onward, Banner never performed a

request for proposals or used any other market analysis tool to evaluate Fidelity’s service

fee. Rather, Banner kept the same uncapped, revenue-sharing arrangement with Fidelity

for almost two decades despite a significant increase in Plan assets and participants.

Between 2009 and 2016, the recordkeeping and administrative fees under the revenue-

sharing arrangement ranged between $52.45 and $108.29 when calculated on a per-

participant basis.

In 2012, Fidelity offered to establish a revenue credit account. This account would

be funded by reimbursements from Fidelity’s revenue-sharing proceeds and could be used

by Banner to pay expenses related to the Plan. Banner accepted this arrangement and

established the account within the next year. Fidelity and Banner agreed the amount of

the revenue credit would be “based on the Plan characteristics, asset configuration, net

cash flow, fund selection and number of participants.” Aplt. App., Vol. VIII at 2034.

Fidelity selected the initial revenue credit amount, with at least some input from Banner.

See id., Vol. IV at 978 (“[Fidelity] had offered 350,000 and we wound up getting

-6- [700,000], so apparently we asked something.”). Fidelity’s payments to the revenue

credit account continued until 2017, when Banner reached a new fee arrangement with

Fidelity at an annual rate of $42 per participant.

B. District Court Proceedings

The named plaintiffs represent a class of current and former participants in the

Plan who sued Banner and other fiduciaries of the Plan, claiming violations of ERISA, 29

U.S.C. § 1101 et seq. After bringing the initial complaint, plaintiffs sought and obtained

certification of the following class: “All participants and beneficiaries of the Banner

Health Employees 401(k) Plan from November 20, 2009 through the date of judgment,

excluding the Defendants.” Id., Vol. II at 272–73. Relevant to this appeal, the class

alleged that Banner had breached its duty of prudence by allowing Fidelity to collect

excessive recordkeeping and administrative fees, see 29 U.S.C. § 1104(a), and had

engaged in prohibited transactions with Fidelity, see id. § 1106(a). The class sought

damages for Plan losses and appropriate injunctive relief to prevent any further violations

of ERISA.

Before trial, the class presented the proposed testimony of an expert witness,

Martin Schmidt. Schmidt was to testify regarding the excessive recordkeeping fees and

also offer an estimate of corresponding losses to the Plan. Banner filed a motion to

exclude Schmidt’s testimony, which the district court denied. But the district court

explained it was retaining the ability to exclude Schmidt’s evidence later. See Aplt. App.,

-7- Vol. II at 336 (“The Court declined to exclude Schmidt’s testimony under Rule 702,

given that it retained the ability to effectively arrive at the same place after receiving all

the evidence relevant to this question at trial.”).

The class’s claims proceeded to an eight-day bench trial. After hearing from

witnesses and experts, the district court issued its findings of fact and conclusions of law.

The court concluded that Banner’s uncapped, revenue-sharing agreement with Fidelity

did not constitute a prohibited transaction under ERISA. The court did determine,

however, that Banner had breached its duty of prudence by failing to monitor Banner’s

service agreement with Fidelity and that this breach resulted in losses to the Plan.

Schmidt testified that class members had suffered $19.4 million in losses because

of Fidelity’s excessive service fee. But the district court refused to rely on Schmidt’s

proposed calculation of damages. The court declined to credit his testimony “because his

opinion is based on vague and insufficient references to his experience in the 401(k) plan

industry.” Id. at 386. Still, having found Banner breached its fiduciary duties under

ERISA, the district court concluded “it is incumbent [on the court] . . . to determine an

appropriate remedy for a breach of fiduciary duty resulting in a loss.” Id. Based on its

review of the records, the court chose to use the revenue credits Fidelity paid to Banner to

approximate the extent of the excessive recordkeeping fees. Thus, the court found no

-8- damages from 2012 to 2016, when Fidelity was paying into the revenue credit account.3

For the period of time before the revenue credit account was set up, the court used an

average of the revenue credits to project damages from 2009 to 2012. Using this average,

the district court calculated damages of $1,661,879.83.

The district court also permitted the class to recover prejudgment interest on this

amount to approximate the lost investment opportunity of funds that otherwise would

have remained in the Plan. After surveying various options for calculating interest, the

court chose to use the IRS underpayment rate as defined in 26 U.S.C. § 6621. For May of

2020, this rate was 3.25%. The court found that this rate “reasonably approximates the

lost earning investment opportunity.” Id. at 387.

II. Analysis

The class brings four arguments on appeal: (1) the district court erred by using

Fidelity’s revenue credit account payments to estimate losses; (2) the district court abused

its discretion in selecting the IRS underpayment rate to calculate prejudgment interest;

(3) the district court misinterpreted ERISA in concluding the service agreement between

Banner and Fidelity was not a prohibited transaction; and (4) the district court abused its

discretion by denying the class injunctive relief.

3 The court awarded some damages from this time period because Banner did not properly account for all of the payments made from the revenue credits back to Banner for expenses associated with maintaining the Plan. Neither party appeals this aspect of the district court’s decision.

-9- We affirm the district court on each of these issues. As explained below, the

district court operated well within its purview in calculating damages and prejudgment

interest. We also agree with the district court that a run-of-the-mill agreement for

recordkeeping services does not constitute a prohibited transaction under ERISA. And

the district court properly exercised its discretion in denying injunctive relief.

A. Standard of Review

Following a bench trial, the district court must state the basis for its findings of

fact and conclusions of law. Fed. R. Civ. P. 52(a) (“In an action tried on the facts without

a jury . . . the court must find the facts specially and state its conclusions of law

separately.”). This requirement ensures that we “ha[ve] an adequate basis for review.”

FTC v. Kuykendall, 371 F.3d 745, 763 (10th Cir. 2004). When calculating damages, the

district court must give adequate reasons to support the amount it awards. See id.

We review the district court’s findings of fact for clear error and its conclusions of

law de novo. See Sw. Stainless, LP v. Sappington, 582 F.3d 1176, 1183 (10th Cir. 2009).

A factual finding is clearly erroneous if there is no support for it in the record or “we are

left with a definite and firm conviction that a mistake has been made.” Id. (internal

quotation marks omitted). When reviewing factual findings, we must “view the evidence

in the light most favorable to the district court’s ruling and must uphold any district court

finding that is permissible in light of the evidence.” Id.

-10- We review a district court’s award of prejudgment interest, exclusion of expert

testimony, and denial of injunctive relief for abuse of discretion. See Weber v. GE Group

Life Assur. Co., 541 F.3d 1002, 1016 (10th Cir. 2008) (prejudgment interest); Att’y Gen.

of Okla. v. Tyson Foods, Inc., 565 F.3d 769, 779 (10th Cir. 2009) (expert testimony);

Combs v. Shelter Mut. Ins. Co., 551 F.3d 991, 1002 (10th Cir. 2008) (injunctive relief).

“An abuse of discretion occurs when a trial court’s decision is arbitrary, capricious,

whimsical, or manifestly unreasonable.” Jensen v. W. Jordan City, 968 F.3d 1187, 1200

(10th Cir. 2020), cert. denied, 2021 WL 1951836 (May 17, 2021) (internal quotation

marks omitted). We will not disturb the district court’s decision without a “firm

conviction that the lower court made a clear error of judgment or exceeded the bounds of

permissible choice in the circumstances.” Id. (internal quotation marks omitted).

B. ERISA

Before addressing the specific issues on appeal, a brief overview of ERISA’s

purpose and structure will help frame the disputed questions.

Through ERISA, Congress intended to “protect employees against

mismanagement of their benefits plans.” Teets v. Great-West Life & Annuity Ins. Co.,

921 F.3d 1200, 1206 (10th Cir. 2019). The statute grants this protection by

“commodiously impos[ing] fiduciary standards on persons whose actions affect the

amount of benefits retirement plan participants will receive.” Id. (internal quotation

marks omitted).

-11- To further this goal, ERISA requires that fiduciaries discharge their duties with

prudence. See 29 U.S.C. § 1104(a). And it prevents fiduciaries from engaging in

transactions that could be detrimental to Plan participants. Id. § 1106(a). To enforce

these fiduciary duties, ERISA empowers plan participants to bring civil suits against

fiduciaries who breach their duties. Id. at § 1132(a)(1). If a court finds a breach of

fiduciary duties, it has wide leeway to fashion remedies to make plan participants whole.

See id. § 1109(a).

1. Damages and Prejudgment Interest

In developing the remedies for violations of ERISA, “Congress intended the

federal courts to draw on principles of traditional trust law.” Eaves v. Penn, 587 F.2d

453, 463 (10th Cir. 1978). And trust law “provides for broad and flexible equitable

remedies in cases involving breaches of fiduciary duty.” Id. Thus, ERISA directs courts

to award damages to compensate for losses a plan sustains due to a breach. 29 U.S.C.

§ 1109(a) (explaining plan participants are entitled to “any losses to the plan resulting

from each such breach”). The statute itself does not prescribe how courts are to measure

loss, but we have previously identified several appropriate measures: “In addition to

specific remedies for recovery of profits obtained by fiduciaries by use of plan assets,

trust law provides the alternative remedy of restoring plan participants to the position in

which they would have occupied but for the breach of trust.” Eaves, 587 F.2d at 462. In

choosing the remedy, “the court has a duty to enforce the remedy which is most

-12- advantageous to the participants and most conducive to effectuating the purposes of the

trust.” Id.

But determining the damages arising from a breach of fiduciary duty can often be

difficult. For instance, there is no easy way to determine the extent of a loss from a

breach that may have been diffuse and spread out across time or the extent of a loss

resulting from lost investment opportunities. So, the plaintiff need not provide a perfect

estimate of how much loss the breach caused. See Eaves, 587 F.3d at 463 (“The law

clearly permits approximations as the extent of the damage, so long as the fact of damage

or ‘lost profit’ is certain.”).

If a district court finds a breach of fiduciary duties but rejects a plaintiff’s

proposed measure of damages, the court may fashion its own measure of loss resulting

from the breach. See Donovan v. Bierwirth, 754 F.2d 1049, 1055 (2d Cir. 1985). When a

district court undertakes such a calculation of damages, we give it considerable discretion.

See Cal. Ironworkers Field Pension Trust v. Loomis Sayles & Co., 259 F.3d 1036, 1047

(9th Cir. 2001) (“When precise calculations are impractical, trial courts are permitted

significant leeway in calculating a reasonable approximation of the damages suffered.”).

If a court determines a fiduciary breached its duty to a plan, it also has discretion to

award the plaintiffs prejudgment interest. See Weber v. GE Group Life Assur. Co., 541

F.3d 1002, 1016 (10th Cir. 2008) (internal quotation marks omitted). Prejudgment

interest is intended to make the plaintiffs whole, not to punish the fiduciary. See Caldwell

-13- v. Life Ins. Co. of N. Am., 287 F.3d 1276, 1287 (10th Cir. 2002) (“The policy that

underlies awarding prejudgment interest seeks to make persons whole for the loss

suffered because they were denied use of money to which they were legally entitled.”).

We now turn to the class’s various arguments about why the district court’s

calculation of damages is factually and legally flawed. A district court’s award of

damages is a factual finding we review for clear error. Sw. Stainless, 582 F.3d at 1183.

But we review de novo the court’s methodology in calculating damages, “such as

determining the proper elements of the award or the proper scope of recovery.” Id.

i. Schmidt’s Testimony

First, the class contends the district court abused its discretion by excluding its

expert’s testimony on damages. Schmidt acted as an expert for the class, testifying as to

the existence of a breach and the extent of damages. To estimate damages, Schmidt

projected a range of appropriate recordkeeping fees for each year in question, selected a

fee from these ranges for each year, and then measured the difference between his chosen

fee and what Banner actually paid Fidelity.

Prior to trial, Banner brought a motion to exclude Schmidt’s expert testimony.

Under Federal Rule of Evidence 702, a district court “must satisfy itself that the proposed

expert testimony is both reliable and relevant” before admitting that testimony for trial.

-14- United States v. Rodriguez-Felix, 450 F.3d 1117, 1122 (10th Cir. 2006). The district

court denied that motion and allowed Schmidt to testify.4

But, after trial, the court decided portions of Schmidt’s testimony were unreliable.5

While the court found Schmidt helpful in identifying the underlying breach, it rejected

Schmidt’s testimony on damages. In doing so, the court reasoned that “Schmidt relied

almost exclusively on his unquantifiable and non-replicable experience for his damages

estimates, a process which the Court is constrained to find as unreliable.” Aplt. App.,

Vol. II at 335.6 Schmidt’s reliance solely on prior experiences to calculate damages was

exacerbated by the fact that “the scant information Mr. Schmidt provided about these

4 In allowing Schmidt to testify, the court recognized that “a judge conducting a bench trial maintains greater leeway in admitting questionable evidence, weighing its persuasive value upon presentation.” See Tyson Foods, 565 F.3d at 780. 5 In Tyson Foods, we concluded it was not an abuse of discretion for a district court to permit an expert to testify for its own consideration and then subsequently find the expert’s testimony unreliable. See 565 F.3d at 780. 6 The parties disagree about the proper standard to apply to a district court’s decision to exclude expert evidence after a bench trial. The class argues this decision should be subject to the abuse of discretion standard traditionally applied to decisions about the admissibility of expert testimony made prior to trial. Banner, however, contends we cannot second-guess the district court’s decision to reject Schmidt’s testimony. Aple. Br. at 24–25 (“Because this aspect of the district court’s decision rests at least in part on its assessment of Schmidt’s live testimony, moreover, its conclusion that Schmidt was not reliable or credible is entitled to nearly insurmountable deference.”). Because we conclude the district court’s decision to exclude Schmidt’s testimony on damages was not an abuse of discretion, we need not determine whether expert testimony excluded after a bench trial is entirely immune from review.

-15- other experiences . . . does not allow the Court to meaningfully assess and consider

whether the quality and service of the recordkeeper services provided to these comparator

plans were on par with . . . services provided by Fidelity to the Plan.” Id. at 334–35.

The class argues the district court abused its discretion in three ways in rejecting

Schmidt’s damages testimony: (1) by applying an incorrect legal standard for assessing

Schmidt’s testimony, (2) by failing to acknowledge that Schmidt had a methodology

supporting his damages calculation, and (3) by basing its decision on inconsistent factual

findings.

The class insists the district court held Schmidt’s testimony to an unnecessarily

high standard. Because damages under ERISA are necessarily imprecise, the class argues

the district court should have deemed Schmidt’s experience-based calculations reliable.

While the class is correct that damage calculations under ERISA are often

imprecise, this principle of ERISA damages does not relieve an expert of demonstrating

that his calculations are based on a reliable methodology. The class has directed us to no

cases suggesting the standards for assessing expert evidence are relaxed in the context of

ERISA. Instead, the rules of evidence remain the same, demanding that expert testimony

be “the product of reliable principles and methods.” Fed. R. Evid. 702(c); see also

Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 593–94 (1993). Here, Schmidt

identified various factors he relied on in coming up with his estimate, including the Plan’s

size, services provided to the Plan, and the number of other recordkeepers that could

-16- perform those services. But when pressed on how he extrapolated from those factors to

the specific fees he proposed, he could only invoke vague allusions to his “experience.”

See, e.g., Aplt. App., Vol. III at 603 (“[I]n determining what the fee would be . . . I looked

at what . . . services . . . were included within the plan and then based on my experience,

and you know, what would be a part of this, these would be the fees . . . that I determined

would be charged for this particular plan.”).

The district court determined that while experience might be an adequate

methodology to support expert testimony in some instances, it was not enough to make

Schmidt’s damages calculation reliable. Instead, the district court agreed with one of

Banner’s experts, who testified that “Mr. Schmidt has not provided any support for those

numbers. So I have no bases to go back and check whether the numbers he is advocating

are correct or not.” Id., Vol. V at 1327. Without any replicable method underlying

Schmidt’s estimates, the district court concluded it could not determine whether his

approximations were reliable. The district court did not abuse its discretion in coming to

this conclusion. See Bitler v. A.O. Smith Corp., 400 F.3d 1227, 1237–38 (2005)

(“[N]othing in either Daubert or the Federal Rules of Evidence requires a district court to

admit opinion evidence that is connected to existing data only by the ipse dixit of the

expert.”) (internal quotation marks omitted).7 Though it is appropriate for the proposed

7 The court also discounted Schmidt’s testimony because, after calculating fee ranges for the years in question, Schmidt selected the lowest fee in each range (continued...)

-17- damages to be approximations, the requirements of Rule 702 are not relaxed in ERISA

cases.

The class also argues the district court overlooked Schmidt’s reliance on specific

comparators in forming his damages calculation. The class specifically refers to two

other mega plans—A and B—that Schmidt relied on in coming to his estimates. Based

on Schmidt’s reliance on these plans, the class maintains he had hard data guiding his

calculation of damages and thus the district court improperly rejected his damages

recommendation.

The district court did not ignore these other plans, though. It acknowledged that

Schmidt had presented data about Plans A and B during his testimony. But the court

determined it would not evaluate these plans as part of Schmidt’s methodology because

Schmidt himself had disclaimed using the plans to estimate the appropriate fees for

Banner’s agreement with Fidelity. See Aplt. App., Vol. XIII at 3341 (testifying “the term

sheets [from A and B] were provided as an example of my experience; not information

that I relied on to form my opinion related to the Banner Plan”). And even if the court

7 (...continued) to determine damages. The court did not abuse its discretion in doing so. Schmidt himself acknowledged any of the fees within his proposed range would be reasonable and he could not provide a specific reason for choosing the lowest possible fee to calculate damages. See Aplt. App., Vol. III at 660 (“I assigned a number that I felt was reasonable based on the characteristics of the plan from what I was looking at.”).

-18- could consider the plans, it concluded the information from just two out of over four

thousand comparable mega plans did not render Schmidt’s opinion reliable.

Finally, the class contends the district court abused its discretion by relying on

inconsistent factual findings in excluding Schmidt’s testimony. The class highlights three

examples. First, the district court found Schmidt’s testimony unreliable in part because

Schmidt did not meaningfully compare the quality of services provided by Fidelity to that

of other plans. But, according to the class, the record contained plenty of evidence to use

for comparisons. Second, the class argues it was improper for the district court to even

require such comparative evidence because the court regarded recordkeepers as fungible.

And, third, the class argues it was improper for the district court to accept Schmidt’s

testimony for one purpose (crediting his testimony as to the finding on breach) but not for

another (refusing to consider his testimony on the extent of damages flowing from that

breach).

The record simply does not bear out the class’s arguments. As explained above,

Schmidt disclaimed reliance on the two most relevant comparators—Plans A and B—in

calculating damages. Many of the other experiences Schmidt cited came in the context of

smaller plans and Schmidt made no effort to compare these plans to Banner’s Plan. Thus,

the court did not abuse its discretion in determining there was a lack of evidence to

engage in meaningful comparison across plans. And while the court did consider

recordkeepers “fungible, to a degree,” Aplt. App., Vol. II at 317, this finding did not

-19- compel the court to mechanically use other service agreements as comparators for

Banner’s agreement with Fidelity. It was entirely reasonable for the court to expect

Schmidt to explain how the other service providers measured up against Fidelity to

determine an appropriate fee range.

Finally, the court did not abuse its discretion by accepting some parts of Schmidt’s

expert testimony while rejecting others. The district court found Schmidt’s experience

adequate to make his opinion as to breach reliable but concluded the experience was

inadequate to support Schmidt’s specific damages calculation. There is nothing

inconsistent about these findings. Expert testimony is not an all-or-nothing proposition.

The fact that an expert’s opinion on one issue is admissible does not mean all his

proposed opinions are admissible. Here, the court deemed Schmidt’s experience

sufficient to render his testimony on a breach reliable, but not adequate to support his

opinion about damages. We find no abuse of discretion in the court accepting Schmidt’s

testimony in finding a breach, but not for calculating damages.

ii. Calculating damages

Having excluded Schmidt’s testimony on damages, the district court considered

“multiple alternatives measures” and ultimately “determined that the revenue credits that

Fidelity actually paid back to Banner are the best estimate of the excess fees the Plan paid

in the first instance.” Aplt. App., Vol. II at 338. The court explained “the revenue credit

may be viewed as the amount that Fidelity itself considered to be excessive recordkeeping

-20- and administrative fees, and should be allocated back to Banner for payment of

administrative Plan costs.” Id.

The class argues that even if the district court properly excluded Schmidt’s

testimony, it erred when crafting its own measure of damages. The class takes issue with

the district court’s selection of the revenue credits in two ways: (1) the district court failed

to provide sufficient reasons for choosing the revenue credits to measure damages,

leaving this court without an adequate record for review, and (2) the district court’s

choice of the revenue credit amount is factually flawed.

First, the class argues that remand is warranted because the district court fell short

of the requirements of Rule 52(a) of the Federal Rules of Civil Procedure. See

Kuykendall, 371 F.3d at 763 (explaining that under Rule 52(a) the district court must “set

forth clear reasons for its findings so this court has an adequate basis for review”).

Specifically, the class notes that while the district court claims to have considered

“alternative measures,” it did not describe what these alternatives were or why the

revenue credits better approximated Banner’s overpayments. Like many a math teacher,

the class faults the district court for not adequately showing its work.

We disagree with the class. The district court provided more than enough

reasoning for us to evaluate its decision. Specifically, the court identified that the

agreement instituting the revenue credits made clear the credits were to be “based on the

Plan characteristics, asset configuration, net cash flow, fund selection[,] and number of

-21- participants.” Aplt. App., Vol. II at 338. And, contrary to the class’s argument, the

district court was under no obligation to go through and explain why it declined to use all

other conceivable measure of damages found in the record. The court considered and

rejected the only other measure of damages the class proposed at trial—Schmidt’s

calculations based on his experience—because it was not grounded in a reliable

methodology. Left without the benefit of party presentation on any other measure of

damages, the court selected a measure of loss and explained its choice.

But the class contends that even if the district court fulfilled its obligations under

Rule 52(a), the court’s reasoning for selecting the revenue credits was unsound. The

class insists the court failed to establish a connection between the revenue credits and the

Plan’s overpayments to Fidelity. Thus, the district court’s finding that Fidelity may have

viewed the revenue credits as representing excessive fees was pure conjecture. Without

evidence of meaningful negotiation about the revenue credits, the class maintains it was

inappropriate for the court to assume Fidelity had any incentive to use the revenue credits

to make the Plan whole. The class also argues the district court’s own findings on breach

militate against using the revenue credits to measure damages. Specifically, the class

points out that the court found Banner could have leveraged the Plan’s growth to

“negotiate more favorable recordkeeping fees for the same level and quality of

recordkeeping and administrative services.” Aplt. App., Vol. II at 331. And the court

went on to explain that Banner’s failure to negotiate “resulted in Plan Participants paying

-22- excessive recordkeeping and administrative fees from the beginning of the Class Period

through December 31, 2016.” Id. The class insists the court’s finding of breach

undercuts its decision to use the revenue credits, which were part of Fidelity’s

arrangement with Banner during the time when the breach occurred, to calculate the true

measure of loss.

We conclude the district court’s decision to use the revenue credits to estimate

damages was permissible. In calculating damages, the district court was not tasked with

taking over for a class of plaintiffs that failed to provide adequate evidence and making

their case for them. Rather, the court’s sole job was to make a “reasonable

approximation” of the recordkeeping losses. Cal. Ironworkers, 259 F.3d at 1047. On

appeal, we must “view the evidence in the light most favorable to the district court’s

ruling and must uphold any district court finding that is permissible in light of the

evidence.” Sw. Stainless, 582 F.3d at 1183. The record reflects that the revenue credits

were tied to specific Plan characteristics. Viewing this evidence in the light most

favorable to the district court, the court operated well within its discretion in treating the

revenue credits as a reasonable approximation of how much Fidelity was being overpaid.

Furthermore, the district court’s decision to use the revenue credit payments was not

inconsistent with the court’s finding of breach. As the court explained, though the breach

continued from 2012 to 2016, there were no losses during this time because Fidelity used

the revenue credits to reimburse Banner.

-23- Having failed to adequately support its sole theory of damages, the class insists the

district court was obliged to scour the record in search of a more generous theory of

damages. No such obligation exists. See United States v. Singeng-Smith, 140 S. Ct.

1575, 1579 (2020) (explaining that parties, not courts, are “responsible for advancing the

facts and argument entitling them to relief”). We affirm the district court’s calculation of

damages.

iii. Selecting a prejudgment interest rate

The class raises similar arguments about the district court’s selection of the IRS

underpayment rate to calculate prejudgment interest: the court’s chosen methodology was

improper and the explanation for its choice was lacking.

After using the revenue credits to calculate damages, the district court exercised its

discretion to “permit Plaintiffs to recover prejudgment interest . . . in lieu of and to

approximate the lost investment opportunity of funds that would have otherwise remained

in the Plan.” Aplt. App., Vol. II at 344. The court first identified prejudgment interest

rates prior courts had used: Colorado’s statutory rate, the federal postjudgment interest

rate, and the IRS underpayment rate. From these options, the court ultimately selected the

IRS underpayment rate, finding that it “reasonably approximates the lost earning

investment opportunity, and is not punitive to Banner.” Id. at 387.

The class contends the district court’s selection of the IRS underpayment rate was

arbitrary and thus an abuse of discretion. While the class acknowledges that the district

-24- court cited to prior cases in which courts had used the IRS underpayment rate to calculate

prejudgment interest, it believes these cases are distinguishable because none involved a

breach of duty under ERISA. Without a specific chain of reasoning explaining why the

IRS underpayment rate makes the class whole, the class maintains the district court’s

choice cannot withstand scrutiny.8

The district court did not abuse its discretion in selecting the IRS underpayment

rate. As a reminder, to find an abuse of discretion, we must have a “firm conviction that

the lower court made a clear error of judgment or exceeded the bounds of permissible

choice in the circumstances.” Jensen, 968 F.3d at 1200. We have no such conviction

here. While the IRS underpayment rate might be lower than several of the other possible

rates the court could have chosen from, the class did not provide the court with any

definitive evidence of what the Plan’s rate of return was during the relevant time. And

the IRS underpayment rate has been previously used in ERISA cases to measure

prejudgment interest. See Russo v. Unger, 845 F. Supp. 124, 126 (S.D.N.Y. 1994)

8 The class now advocates for the Plan’s average rate of return as the appropriate prejudgment interest rate. The class raises this alternative rate for the first time on appeal. Below, the class proposed only the S&P 500 Index rate of return to calculate prejudgment interest. We will not entertain this proposal at this late hour. See Proctor & Gamble Co. v. Haugen, 222 F.3d 1262, 1271 (10th Cir. 2000) (“[W]e should not be considered a ‘second-shot’ forum . . . where secondary, back-up theories may be mounted for the first time.” (internal quotation marks omitted)).

-25- (collecting cases). The district court’s decision to use the IRS underpayment rate was

well within the realm of permissible choice.

2. Equitable Relief

The class next argues the district court abused its discretion by failing to give

injunctive relief to prevent Banner from continuing to breach its fiduciary duties.

Specifically, the class appears to be most concerned with the district court’s decision not

to require Banner to hold a request for proposals to test the market for recordkeeping and

administrative services. The class argues Banner has not meaningfully tested the market

for recordkeepers in over twenty years, so the district court should have required Banner

to engage in this long overdue process.

As previously indicated, ERISA provides courts a wide array of remedial options,

including injunctive relief. See 29 U.S.C. § 1109 (stating a fiduciary who breaches his

duties “shall be subject to such other equitable or remedial relief as the court may deem

appropriate”). Because ERISA imposes a high standard on fiduciaries, “serious

misconduct that violates statutory obligations is sufficient grounds for a permanent

injunction.” Beck v. Levering, 947 F.2d 639, 641 (2d Cir. 1991). Still, equitable relief

must be used only “to redress [a] violation or to enforce any provision” of ERISA. See 29

U.S.C. § 1132(a)(5).

Because the district court found Banner to have breached its fiduciary duties, it

considered the class’s argument for equitable relief. Ultimately, though, the court

-26- concluded the class “failed to make any showing that injunctive or equitable relief is

appropriate in the circumstances of this case.” Aplt. App., Vol. II at 402. By the time of

the court’s judgment, Banner had ended the previous uncapped, revenue-sharing

arrangement and agreed to a per-participant recordkeeping fee with Fidelity. Since

Banner had ended the prior arrangement, the court found “there is simply no evidence

from which the Court can reasonably conclude that Banner Defendants will at some point

in time resume a policy or practice of violating their duty of prudence with respect to

recordkeeping fees.” Id. at 403. Without an impending threat that Banner’s breach of

fiduciary duty would continue, the court denied the class’s request for injunctive relief.9

The class contends the court abused its discretion in denying relief because

Banner’s breach of fiduciary duties is ongoing. In finding a breach, the district court had

found it “highly significant that Banner has not undertaken a single [request for

proposals] in nearly 20 years.” Id. at 324. No market testing occurred before Banner

reached the current agreement with Fidelity, which “was proposed by Fidelity, and

9 The district court also noted that the class “did not challenge the appropriateness of the recordkeeping fees paid to Fidelity after January 1, 2017.” Aplt. App., Vol. II at 402. The class maintains they made no such concession and explicitly challenged the ongoing recordkeeping fees, even after Banner had changed its arrangement with Fidelity. See id. at 506–07 (explaining the class “would disagree . . . that our time frame ends at 2016. We have an ongoing excessive recordkeeping fee claim, even though they’ve changed . . . the payment structure”). This single error, however, does not amount to an abuse of discretion. The court gave other reasons for reaching its decision that are supported by the record.

-27- accepted without apparent negotiation by Banner.” Id. at 323. The class reasons that

because Banner has still not performed a request for proposals or otherwise tested the

market, Banner’s breach continues.

The district court did not abuse its discretion in denying injunctive relief. Banner

characterizes the court’s breach finding as stemming entirely from Banner’s failure to test

the market. That is wrong. The court did not find a breach simply because Banner had

failed to perform a request for proposals. Rather, the court found a breach of fiduciary

duty because Banner failed to adequately monitor the uncapped, revenue-sharing

agreement. Once Banner changed to the per-participant recordkeeping fee with Fidelity,

the breach the court had identified ended. And without any evidence of that specific

breach continuing, the court denied injunctive relief. Because the underlying fee

arrangement that triggered the initial finding of breach changed, we cannot say the court’s

decision to deny injunctive relief was arbitrary or manifestly unreasonable.

3. Prohibited Transaction

Finally, the class argues the district court misinterpreted ERISA’s ban on

prohibited transactions.10

10 Banner contends this issue is moot. The parties agree that the class’s entitlement to relief on the prohibited transaction and breach of fiduciary duty claims are co-extensive. Thus, Banner argues that the district court’s grant of relief to the class based on the failure to monitor means no further relief could flow from any prohibited transactions with Fidelity.

(continued...)

-28- ERISA prohibits certain transactions between fiduciaries and third parties. See

Teets, 921 F.3d at 1207. Specifically, ERISA prohibits a plan’s fiduciary from

“engag[ing] in a transaction, if he knows or should know that such transaction constitutes

a direct or indirect . . . furnishing of goods, services, or facilities between the plan and a

party in interest.” 29 U.S.C. § 1106(c). Under ERISA, a “party in interest” includes “a

person providing services to such plan.” Id. § 1002. A plan participant can sue a

fiduciary who engages in such a prohibited transaction. See id. § 1132(a)(1).

Still, ERISA provides some exemptions from the prohibited transaction rules,

thereby “allow[ing] plans to do business with parties in interest if certain conditions are

met.” Teets, 921 F.3d at 1222. Plaintiffs bear the initial burden of proving a fiduciary

engaged in a prohibited transaction. If the plaintiff meets this burden, defendants then

have the opportunity to prove the prohibited transaction qualifies for one of ERISA’s

10 (...continued) We disagree. See Mission Prod. Holdings, Inc. v. Tempnology, LLC, 139 S. Ct. 1652, 1660 (2019) (“[W]e may dismiss the case . . . only if it is impossible for a court to grant any effectual relief whatever to [the appellant] assuming it prevails.” (emphasis added)). If we were to reverse the district court’s decision on the class’s prohibited transaction claim, on remand the class would be entitled to present its arguments about why relief should be greater for Banner’s alleged violation of the prohibited transaction claim. The district court could hypothetically institute a different measure of damages based on this alternative theory of liability. And while the class could not recover damages in addition to the relief granted for Banner’s breach of fiduciary duty, the class is allowed to pursue an alternative theory of judgment because it could theoretically receive a more favorable remedy in lieu of the relief the court originally granted. Thus, the claim is not moot. Id. (“For better or worse, nothing so shows a continuing stake in a dispute’s outcome as a demand for dollars and cents.”).

-29- exemptions. See Braden v. Wal-Mart Stores, Inc., 588 F.3d 585, 601 (8th Cir. 2009). For

example, while a plan usually cannot transact with a party in interest, fiduciaries are not

prohibited from “[c]ontracting or making reasonable arrangements with a party in interest

for office space, or legal, accounting, or other services necessary for the establishment or

operation of the plan, if no more than reasonable compensation is paid therefor.” 29

U.S.C. § 1108(b)(2)(A).

Before the district court, the class argued that Fidelity was a party in interest

because Banner contracted with Fidelity to provide services to the Plan. The class further

reasoned that because Fidelity was a party in interest, Fidelity’s provision of services to

Banner constituted a prohibited transaction under ERISA. The district court disagreed. It

explained ERISA “only prohibits such service relationships with persons who are ‘parties

in interest’ by virtue of some other relations . . . . It does not prohibit a plan from paying

an unrelated party, dealt with at arm’s length, for services rendered.” Aplt. App., Vol. II

at 391 (quoting Sellers v. Anthem Life Ins. Co., 316 F. Supp. 3d 25, 34 (D.D.C. 2018)).

The district court expressed concern that treating the service agreement between Fidelity

and Banner as a prohibited transaction would “discourage employers from offering

ERISA plans” altogether. Id., Vol. II at 391 (quoting Divane v. Nw. Univ., No. 16-C-

8157, 2018 WL 2388118, at *10 (N.D. Ill. May 25, 2018)).

The class takes issue with the district court’s interpretation of prohibited

transaction under ERISA. The class argues the statute’s language is clear, categorical,

-30- and broad: “Because Fidelity is a service provider and hence a ‘party in interest,’ its

‘furnishing of’ recordkeeping and administrative services to the Plan constituted a

prohibited transaction[.]” Aplt. Br. at 52. It points to statements from this court and the

Department of Labor that support this expansive reading of what constitutes a prohibited

transaction. See Teets, 921 F.3d at 1222 (“On its face, [§ 1106] covers wide swaths of

plan activity.”); Reasonable Contract or Arrangement Under Section 408(b)(2)—Fee

Disclosure, 77 Fed. Reg. 5632, 5632 (Feb. 3, 2012) (“[A] service relationship between a

plan and a service provider would constitute a prohibited transaction, because any person

providing services to the plan is defined by ERISA to be a ‘party in interest’ to the

plan.”).11 Based on this interpretation of § 1106, the class calls on us to reverse and

11 The DOL’s passing statement suggesting that service agreements constitute prohibited transactions does not affect our analysis. Neither party invoked Chevron. So, we need not decide whether the DOL’s statement that “a service relationship between plan and a service provider would constitute a prohibited transaction” is entitled to any deference. Reasonable Contract or Arrangement Under Section 408(b)(2), 77 Fed. Reg. at 5632.

Our case law about when Chevron has been waived is, admittedly, unsettled. See Aposhian v. Barr, 958 F.3d 969, 982 n.6 (deciding Chevron was not waived because the parties invited its use but acknowledging “our cases are not entirely consistent as to whether such an invitation is necessary”). Here, there was no mention of Chevron by either party and thus any reliance on Chevron is waived. See Hays Med. Ctr. v. Azar, 956 F.3d 1247, 1264 n.18 (10th Cir. 2020) (deeming Chevron waived when the party provides a “threadbare citation” but never again mentions Chevron or applies it to the facts of the case). The class merely cited the DOL’s regulation to bolster its own interpretation of § 1106, not to suggest we owe particular deference to the agency’s interpretation of what constitutes a prohibited transaction. Therefore, we proceed with our review of the (continued...)

-31- remand for the district court to determine whether Banner has an affirmative defense of

reasonableness under § 1108(b).

Teets does not dictate our outcome here. In Teets, this court concluded a

prohibited transaction claim based on the plan’s agreement with the service provider

could not proceed because equitable relief could not be granted. 921 F.3d at 1223 (“If a

plaintiff cannot demonstrate that equitable relief is available, the suit cannot proceed.”).

The Teets panel never specifically addressed the issue we now face: whether an initial

agreement with a service provider constitutes a prohibited transaction. Though this court

generally described the statutory framework for prohibited transaction claims, we did not

address whether the service provider in that case was actually a party in interest. Our

statement that § 1106 “covers wide swaths of plan activity,” Teets, 921 F.3d at 1222, does

not mean it covers all plan activity.

Still, the class insists the service agreement constitutes a prohibited transaction.

The class’s interpretation leads to an absurd result: the initial agreement with a service

provider would simultaneously transform that provider into a party in interest and make

that same transaction prohibited under § 1106. See Sellers, 316 F. Supp. at 34 (such a

reading would be “circular reasoning: the transactions were prohibited because [the

service provider] was a party in interest, and [the service provider] was a party in interest

11 (...continued) district court’s interpretation of § 1106 de novo.

-32- because it engaged in a prohibited transaction”). Instead, we conclude that some prior

relationship must exist between the fiduciary and the service provider to make the

provider a party in interest under § 1106. ERISA cannot be used to put an end to run-of-

the-mill service agreements, opening plan fiduciaries up to litigation merely because they

engaged in an arm’s length deal with a service provider. Instead, ERISA is meant to

prevent fiduciaries from engaging in transactions with parties with whom they have pre-

existing relationships, raising concerns of impropriety. Otherwise, a plan participant

could force any plan into court for doing nothing more than hiring an outside company to

provide recordkeeping and administrative services.

The Supreme Court’s prior treatment of § 1106 supports our interpretation. The

Court has said Congress passed § 1106 “to bar categorically a transaction that [is] likely

to injure the pension plan.” Lockheed Corp. v. Spink, 517 U.S. 882, 888 (1996)

(emphasis added). What all prohibited transactions under § 1106 “have in common is that

they generally involve uses of plan assets that are potentially harmful to the plan.” Id. at

893 (emphasis added). Only certain transactions—those involving parties in interest—

raise inferences of impropriety. The class has provided no evidence to show that the

service agreement between Fidelity and Banner was anything less than an arm’s length

deal or that Fidelity had some pre-existing relationship with Banner.

In sum, we affirm the district court’s entry of judgment for Banner on the class’s

prohibited transaction claim.

-33- III. Conclusion

For the foregoing reasons, we AFFIRM the district court’s calculation of damages

and prejudgment interest, denial of injunctive relief, and entry of judgment for Banner on

the class’s prohibited transaction claim.

-34-