FILED
                                                    United States Court of Appeals
                                                            Tenth Circuit

                                                        December 19, 2017
                                         PUBLISH        Elisabeth A. Shumaker
                                                            Clerk of Court
                     UNITED STATES COURT OF APPEALS

                                   TENTH CIRCUIT



JANEEN MEDINA, individually, and
on behalf of all others similarly
situated, and on behalf of the CHI
Plans,

               Plaintiff - Appellant,
v.                                                 No. 16-1005
CATHOLIC HEALTH INITIATIVES;
GERALDINE BEDNASH;
MAUREEN COMER; RICHARD
CORRENTE; DAVID R. EDWARDS;
KATHERINE GRAY; BARBARA
HAGEDORN; JAMES HAMILL;
ANTOINETTE HARDY-WALLER;
PHYLLIS HUGHES; DONALD
JONES; ANDREA J. LEE; DAVID R.
LINCOLN; KEVIN E. LOFTON;
CHRISTOPHER R. LOWNEY;
ELEANOR F. MARTIN; MARY
MARGARET MOONEY; LILLIAN
MURPHY; MARY JO POTTER;
PATRICIA SMITH; EDWARD
SPEED; DEAN SWINDLE;
PATRICIA G. WEBB; JOHN AND
JANE DOES, 1-10, whose true names
are unknown,

               Defendants - Appellees.
---------------------------------------------
 AMERICAN ASSOCIATION OF
 RETIRED PERSONS; FREEDOM
 FROM RELIGION FOUNDATION;
 PENSION RIGHTS CENTER;
 AMERICANS UNITED FOR
 SEPARATION OF CHURCH AND
 STATE; AMERICAN CIVIL
 LIBERTIES UNION; GUIDESTONE
 FINANCIAL RESOURCES OF THE
 SOUTHERN BAPTIST
 CONVENTION; THE PENSION
 BOARDS-UNITED CHURCH OF
 CHRIST, INC.; CHURCH
 ALLIANCE; THE CATHOLIC
 HEALTH ASSOCIATION OF THE
 UNITED STATES,

             Amici Curiae.


        APPEAL FROM THE UNITED STATES DISTRICT COURT
                FOR THE DISTRICT OF COLORADO
                 (D.C. NO. 1:13-CV-01249-REB-KLM)


Ron Kilgard, Keller Rohrback L.L.P., Phoenix, Arizona (Lynn Lincoln Sarko,
Laura R. Gerber, Havila C. Unrein, and Matthew M. Gerend, Keller Rohrback
L.L.P., Seattle, Washington, and Laurie B. Ashton, Keller Rohrback L.L.P.,
Phoenix, Arizona, and Karen L. Handorf, Michelle C. Yau, and Mary J.
Bortscheller, Cohen Milstein Sellers & Toll, PLLC, Washington, D.C., with him
on the briefs), Phoenix, Arizona, for Plaintiff-Appellant

Lars C. Golumbic (Sarah M. Adams and Sean C. Abouchedid, with him on the
brief), Groom Law Group, Chartered, Washington, D.C., for Defendants-
Appellees.

William Alvarado Riveria and Mary Ellen Signorille, AARP Foundation
Litigation, Washington, D.C., on the brief for Amicus Curiae AARP.




                                      -2-
Andrew L. Seidel, Freedom from Religion Foundation, Madison, Wisconsin, on
the brief for Amicus Curiae Freedom from Religion Foundation.

Curtis L. Kennedy, Denver Colorado, Norman P. Stein, Philadelphia,
Pennsylvania, and Karen W. Ferguson, Pension Rights Center, Washington, D.C.
on the brief for Amicus Curiae Pension Rights Center.

Daniel Mach, American Civil Liberties Union Foundation, Washington, D.C., and
Richard B. Katskee and Bradley Girard, Americans United for Separation of
Church and State, Washington, D.C., on the brief for Amici Curiae Americans
United for Separation of Church and State and American Civil Liberties Union.

G. Daniel Miller, Conner & Winters, LLP, Washington, D.C., and Laurence A.
Hansen and Hugh S. Balsam, Locke Lord LLP, Chicago, Illinois, on the brief for
Amici Curiae GuideStone Financial Resources of the Southern Baptist
Convention, The Pension Boards-United Church of Christ, Inc., and The Church
Alliance.

Mark E. Chopko, Marissa Parker and Brandon Riley, Stradley Ronon Stevens &
Young, LLP, Washington, D.C., and Lisa J. Gilden, The Catholic Health
Association of the United States, Washington, D.C., on the brief for Amicus
Curiae The Catholic Health Association of the United States.



Before TYMKOVICH, Chief Judge, BACHARACH, and MORITZ, Circuit
Judges.


TYMKOVICH, Chief Judge.



      The Employee Retirement Income Security Act of 1974, 29 U.S.C. § 1001,

et seq. (ERISA), generally exempts from its requirements “church

plans”—employee-benefit plans established and maintained by churches for their

employees. ERISA also extends that church-plan exemption to so-called


                                       -3-
principal-purpose organizations. A principal-purpose organization is a church-

affiliated organization whose principal purpose is administering or funding a

benefit plan for the employees of a church or a church-affiliated nonprofit

organization.

      Catholic Health Initiatives (CHI) is a Denver-based nonprofit organization

created to carry out the Roman Catholic Church’s healing ministry. To do so,

CHI operates 92 hospitals and numerous other healthcare facilities in 18 states.

CHI offers a retirement plan for its employees, with more than 90,000 participants

and beneficiaries, and nearly $3 billion in plan assets. The CHI plan is

administered by the CHI and Affiliates Defined Benefit Plan Subcommittee

(Subcommittee), whose members are appointed and removed by CHI’s Board of

Stewardship Trustees.

      The district court held that CHI’s plan was a church plan that qualified for

the ERISA exemption. On appeal, we agree, concluding that CHI’s plan satisfies

the statutory requirements for the church-plan exemption: CHI is a tax-exempt

organization associated with a church, and the Subcommittee is a proper

principal-purpose organization that is also associated with a church. The ERISA

exemption, moreover, does not run afoul of the United States Constitution’s

Establishment Clause.




                                        -4-
                                  I. Background

      As explained above, federal law exempts church plans from certain

federally mandated reporting and funding requirements. The agencies

administering ERISA consider the CHI retirement plan to be a church plan,

because CHI and the Subcommittee are controlled by or associated with the

Catholic Church and the Subcommittee is a principal-purpose organization.

      Janeen Medina, 1 a CHI employee, filed a class action, alleging that CHI’s

retirement plan fails to satisfy the statutory criteria for the church-plan

exemption. She contends that, since the plan does not qualify for the exemption,

CHI should have complied with the reporting and funding requirements of

ERISA. Medina also argues the individual defendants who administer the plan

breached their fiduciary duties by failing to comply with ERISA. And, Medina

argues, even if the CHI plan did qualify as a church plan, the exemption would

violate the Establishment Clause of the United States Constitution.

      The district court concluded CHI’s plan satisfied the criteria for ERISA’s

church-plan exemption and dismissed her other statutory and constitutional

claims.

      While this appeal was pending, the Supreme Court resolved one of the

issues before us, holding that an employee-benefit plan need not be established by


      1
        Medina has since changed her last name to West, but we follow the parties
and refer to her as Medina.

                                          -5-
a church to qualify for ERISA’s church-plan exemption, provided it satisfies the

other statutory criteria. Advocate Health Care Network v. Stapleton, 

(2017). According to the Court, because Congress expanded “the category

of plans ‘established and maintained by a church’ to ‘include’ plans ‘maintained

by’ principal-purpose organizations, those plans . . . are exempt from ERISA’s

requirements.’” 

The fact that CHI’s plan was not established by a

church therefore does not preclude its eligibility for the church-plan exemption.

      Two issues remain for our consideration. First, Medina contends CHI’s

plan is not a church plan because it fails to satisfy the statutory criteria, which

require the plan to be maintained by a principal-purpose organization associated

with a church, for the employees of an organization associated with a church. She

also argues the district court erred in concluding there were no genuine disputes

of material fact on this point. Second, Medina contends applying the exemption

to CHI’s plan would violate the Establishment Clause of the United States

Constitution.

      We affirm the well-reasoned and thorough decision of the district court, in

an opinion that follows similar analytical lines. We find CHI’s plan satisfies the

relevant statutory criteria and qualifies as a church plan, and the district court did

not err in concluding there were no genuine disputes of material fact. Nor does

applying the exemption to CHI’s plan violate the Establishment Clause. Supreme

Court precedent allows Congress to provide a religious accommodation by

                                          -6-
exempting religious organizations from regulatory schemes, especially where

subjecting the religious organization to the regulatory scheme would raise

constitutional questions. 2

                                    II. Analysis

      Medina contends that CHI’s plan is not a church plan because it fails to

satisfy the statutory criteria, and that the district court erred in concluding there

were no genuine disputes of material fact regarding whether the plan satisfied the

statutory criteria for the exemption. She also argues applying the exemption to

CHI’s plan would violate the Establishment Clause.

      We address each argument in turn and conclude that CHI’s plan does not

violate either ERISA or the Constitution.

      A. Statutory Framework

      The Supreme Court’s recent opinion in Advocate explains the statutory

scheme at issue. “ERISA generally obligates private employers offering pension

plans to adhere to an array of rules designed to ensure plan solvency and protect

plan participants. But in enacting the statute, Congress made an important

exception. ‘[C]hurch plan[s]’ have never had to comply with ERISA’s

requirements.” 

(citation omitted) (quoting 29 U.S.C. § 1003(b)(2)).




      2
       We also grant the parties’ motions to seal Volume VIII of the Appendix
and Volume IX of the Supplemental Appendix.

                                          -7-
      As Advocate makes clear, two types of organization qualify for the

church-plan exemption: churches and so-called principal-purpose organizations.

      ERISA has always provided that “church plan” includes “a plan established

and maintained . . . for its employees . . . by a church or by a convention or

association of churches.” 29 U.S.C. § 1002(33)(A).

      A 1980 amendment to ERISA expanded that definition to include

principal-purpose organizations. That is, the church-plan exemption includes

             a plan maintained by an organization, whether a civil law
             corporation or otherwise, the principal purpose or function
             of which is the administration or funding of a plan or
             program for the provision of retirement benefits or welfare
             benefits, or both, for the employees of a church or a
             convention or association of churches, if such organization
             is controlled by or associated with a church or a
             convention or association of churches.

29 U.S.C. § 1002(33)(C)(i).

      Crucially, for purposes of this definition, an “employee of a church”

includes “an employee of an organization, whether a civil law corporation or

otherwise, which is exempt from tax under section 501 of Title 26 and which is

controlled by or associated with a church or a convention or association of

churches.” 29 U.S.C. § 1002(33)(C)(ii)(II). That is, “employee of a church” is

not limited to church employees, but includes employees of nonprofit

organizations associated with a church, such as CHI claims to be.




                                          -8-
      Advocate rejected the view that the church-plan exemption only applied to

plans established by churches. Hospital employees had argued that their

employers’ pension plans did not fall within the church-plan exemption because

those plans were not established by a church, but rather by a principal-purpose

organization. The Supreme Court disagreed, holding that “[u]nder the best

reading of the statute, a plan maintained by a principal-purpose organization . . .

qualifies as a ‘church plan,’ regardless of who established 

.

As Medina concedes, this issue is now resolved in favor of CHI. Resp. to Supp.

Auth. at 1.

      Advocate did not provide guidance on the major issue remaining in this

case: whether CHI’s internal benefits committee qualifies as a principal-purpose

organization. The Court expressly declined to address the scope of a

“principal-purpose organization,” what the requisite level of association with a

church was, or whether an internal benefits committee could 

n.2, 1658 n.3. We confront those questions today, and decide whether

CHI’s plan satisfies the statutory criteria for the church-plan exemption.

      The statute imposes a three-step inquiry for entities seeking to use the

church-plan exemption for plans maintained by principal-purpose organizations:

      1.      Is the entity a tax-exempt nonprofit organization associated with a
              church?

      2.      If so, is the entity’s retirement plan maintained by a principal-
              purpose organization? That is, is the plan maintained by an

                                          -9-
             organization whose principal purpose is administering or funding a
             retirement plan for entity employees?

      3.     If so, is that principal-purpose organization itself associated with a
             church?

Under this framework, to qualify for the church-plan exemption, CHI must

receive an affirmative answer to all three inquiries.

      Note that both the principal-purpose organization and the entity whose

employees the plan benefits must be associated with a church. First, the

principal-purpose organization must be associated with a church. The

church-plan exemption includes “a plan maintained by a [principal-purpose

organization] . . . if [the principal-purpose organization] is controlled by or

associated with a church or a convention or association of churches.” 29 U.S.C.

§ 1002(33)(C)(i).

      Second, the definition of principal-purpose organization requires that the

entity whose plan it is administering be associated with a church. A principal-

purpose organization is one whose “principal purpose or function . . . is the

administration or funding of a plan or program for the provision of retirement

benefits or welfare benefits, or both, for the employees of a church.” 29 U.S.C.

§ 1002(33)(C)(i) (emphasis added). And 29 U.S.C. § 1002(33)(C)(ii)(II) explains

that “employee of a church” includes employees of tax-exempt organizations

“controlled by or associated with a church or a convention or association of

churches.”

                                         -10-
      In short, a principal-purpose organization, by definition, is one that

administers a retirement or welfare plan for a church or a tax-exempt organization

“associated with a church.” And that principal-purpose organization can only

qualify for the church-plan exemption if the principal-purpose organization itself

is “associated with a church.”

             1. Step One: Is the Entity a Tax-Exempt Nonprofit Organization
             Associated with a Church?

      Under ERISA, “[a]n organization . . . is associated with a church or a

convention or association of churches if it shares common religious bonds and

convictions with that church or convention or association of churches.” 29 U.S.C.

§ 1002(33)(C)(iv). CHI meets this requirement.

      CHI’s connection to the Roman Catholic Church is a function of Roman

Catholic canon law. Canon law, a body of law used in the Catholic Church,

“governs the actions and rights of all Church entities.” Supp. App., Vol. I at 64.

The canon-law equivalent of a corporation is a “public juridic person.” Under

canon law, a public juridic person is an aggregate of persons or things

“constituted by competent ecclesiastical authority so that, within the purposes set

out for them, they fulfill in the name of the Church, according to the norm of the

prescripts of the law, the proper function entrusted to them in view of the public

good.” 1983 Codex Iuris Canonici c.116, § 1,

http://www.vatican.va/archive/ENG1104/__PD.HTM.


                                        -11-
      Since U.S. law does not recognize public juridic persons organized under

canon law, public juridic persons need civil-law counterparts—normally nonprofit

corporations—in order to transact civil-law business, such as holding title to

property.

      Catholic Health Initiatives is the civil-law counterpart of a canon-law

public juridic person called Catholic Healthcare Federation (CHCF). The

Catholic Church regards the two as alter egos.

      As a public juridic person organized under canon law, the Catholic Church

regards CHCF as an official part of the Catholic Church. A department of the

Vatican approved a petition to confer public juridic personality on CHCF. And

the Vatican approved CHCF’s canonical statutes, which are analogous to a

corporation’s articles of incorporation.

      CHCF is accountable to the Vatican in several ways. For example, the

Vatican must approve any change in CHCF’s purpose; any transfer of real

property above a certain amount; the dissolution of CHCF; and changes in the

canonical statutes. CHCF also makes an annual report to the Vatican.

      Several factors attest the relationship between CHI and CHCF. First, CHI’s

Articles of Incorporation provide that it is “organized and operated . . .

exclusively for the benefit of, to perform the functions of, and/or to carry out the

religious, charitable, scientific, and education purposes” of CHCF. Supp. App.,




                                           -12-
Vol. II at 482. The same people serve as Trustees of CHI and members of CHCF.

And CHI’s property is CHCF’s property under canon law.

      In addition, CHI is listed in The Official Catholic Directory, which lists

Roman Catholic institutions in the United States. The listing signifies that the

Archbishop of Denver has determined that CHI is “an official part of the Catholic

Church.” App., Vol. III at 643. That certification must be renewed annually.

Noting the significance of this listing, the IRS considers any organization listed in

The Official Catholic Directory “associated with” the Roman Catholic Church for

purposes of ERISA’s church-plan exemption. IRS General Counsel Memorandum

39007, 

at *4 (July 1, 1983).

      Given CHI’s close integration with the Roman Catholic Church, we agree

with the district court that CHI is “associated with” a church.

      Medina, however, urges us to look to the factors proposed by the Fourth

Circuit in Lown v. Continental Casualty Company, 

(4th Cir. 2001),

to reach a different conclusion. In Lown, the court analyzed three non-exclusive

factors, looking to “1) whether the religious institution plays any official role in

the governance of the organization; 2) whether the organization receives

assistance from the religious institution; and 3) whether a denominational

requirement exists for any employee or patient/customer of the organization.” 

Applying these factors, the court found the healthcare organization at

issue did not have a church plan, since it had formally disaffiliated from the

                                         -13-
Baptist Convention several years before the litigation, and no denominational

requirement existed. 

Circuit applied these factors in Chronister v. Baptist Health,

, 653 (8th Cir. 2006), also finding that the healthcare organization at

issue did not have a church plan. Here, too, the organization had formally

disaffiliated from the Baptist Convention. 

The Eighth Circuit found

that the Baptist Convention played no role in the organization’s governance; that

the Baptist Convention did not appoint or approve any of the organization’s board

members; that the organization received no financial support from the Baptist

Convention after its disassociation; and that the denominational requirement was

limited to administrators, the president/CEO, chaplains, and board members. 

This being so, there was insufficient evidence of common religious bonds

and convictions between the entities.

       Setting aside their uncertain derivation, the Lown factors cannot be the

exclusive means of determining whether an organization is “associated with a

church.” This is because the Lown factors are much narrower than the broad

language of the definition in § 1002(33)(C)(iv). Under the statute, to be

“associated with a church,” a corporation need only share “common religious

bonds and convictions with that church or convention or association of churches.”

The statute imposes no denominational requirements, corporate governance

requirements, or funding requirements. Thus, an organization could share

                                        -14-
“common religious bonds and convictions” with a church while satisfying none of

the Lown factors. Because the Lown factors are narrower than the statutory

language, satisfying the Lown factors may suffice to establish that an organization

is associated with a church. But an organization does not need to satisfy the

Lown factors in order to be associated with a church.

      We therefore do not rely on the Lown factors to reach our conclusion that

CHI is associated with a Church. CHI’s association with the Catholic Church is

not a close question given the facts we recited above. Moreover, Lown and

Chronister are easily distinguishable from the present case, since the healthcare

organizations in Lown and Chronister had both officially disaffiliated from the

Baptist Convention. Here, by contrast, CHI is regarded as an official part of the

Catholic Church.

      In sum, CHI’s many connections to the Catholic Church suffice to satisfy

the statutory requirement that CHI “share[] common religious bonds and

convictions with” the Catholic Church.

             2. Step Two: Is the Entity’s Retirement Plan Maintained by a
             Principal-Purpose Organization?

      As explained above, the church-plan exemption includes “a plan

maintained by an organization, whether a civil law corporation or otherwise, the

principal purpose or function of which is the administration or funding of” a




                                         -15-
retirement plan for employees of a church or a nonprofit associated with a church.

29 U.S.C. § 1002(33)(C)(i) (emphases added).

      The parties dispute which is the relevant “organization” for purposes of this

provision: CHI as a whole, or the Subcommittee. They also dispute the meaning

of “maintain” in this provision.

      Medina argues that even if the Subcommittee “maintained the plan,” it

cannot be an “organization” in the statutory meaning. Because the Subcommittee

is “a subcommittee of the HR Committee . . . which is a committee of the CHI

Board,” and CHI pays the Subcommittee’s expenses, indemnifies its members,

and CHI staff lead Subcommittee meetings, argues Medina, it cannot be its own

organization. Aplt. Br. at 56. It is just a part of a larger organization, not an

organization itself.

      Medina also argues the CHI plan cannot qualify for the exemption because

it is “maintained” by CHI, whose principal purpose is healthcare, not the

administration or funding of the CHI plan. Medina concedes the Subcommittee

“administers” the CHI plan, Aplt. Br. at 53, but argues CHI “maintains” it.

Medina insists that “maintain” and “administer” are different terms under ERISA,

and that “maintaining” a plan necessarily includes the power to modify or

terminate it. That being so, Medina argues only CHI can be considered to

“maintain” the plan. Finally, Medina insists that “[t]o ‘maintain’ a plan ‘means to

“continue” a plan.’” Aplt. Br. at 53 (quoting Anderson v. UNUM Provident

                                          -16-
Corp., 

, 1265 (11th Cir. 2004)). She further explains that “to

continue a plan requires the ability to modify or terminate it.” 

v. Tex. Hosp. Ass’n, 

, 449 (5th Cir. 1995)). 3

      We have not previously addressed the meaning of “maintain” under this or

any other provision of ERISA. And nowhere does ERISA define “maintain.”

“When a term goes undefined in a statute, we give the term its ordinary meaning.”

Taniguchi v. Kan Pac. Saipan, Ltd., 

, 566 (2012). Dictionary

definitions can help us determine that ordinary meaning. See, e.g., Levorsen v.

Octapharma Plasma, Inc., 

, 1231 (10th Cir. 2016).

      Black’s Law Dictionary provides several definitions for “maintain.” The

definition most relevant in this context is “[t]o care for (property) for purposes of

operational productivity or appearance; to engage in general repair and upkeep.”




      3
         In support of her position, Medina points to Anderson, from the Eleventh
Circuit, which looks to the dictionary definition of “maintain:”

      To “maintain” a plan, in the ordinary meaning of the word, simply
      means to “continue” a plan. See Webster’s Third New International
      Dictionary 813 (1961) (defining “maintain” as “1: to keep in a state of
      repair, efficiency, or validity” or “4: to provide for: bear the expense
      of”); Black’s Law Dictionary (7th ed. 1999) (defining “maintain” as “1.
      to continue”).

. She also points to Hightower, which, however, does
not define “maintain” in the way she proposes. 

. Medina thus points
to no authority explaining why “maintaining” a plan necessarily includes the
power to modify or terminate it.

                                         -17-
Black’s Law Dictionary 1039 (9th ed. 2009). Webster’s similarly defines

“maintain” as “to keep in a state of repair, efficiency, or validity” or

to provide for: bear the expense of.” Webster’s Third New International

Dictionary 1362 (2002).

      And consider the following examples, culled from across the U.S. Code:

      !      ERISA requires an employer to “maintain records with respect to
             each of his employees sufficient to determine the benefits due or
             which may become due to such employees.” 29 U.S.C. § 1059(a)(1)
             (emphasis added). This provision cannot mean that employers have
             the power to “terminate” the relevant records. It must mean, as
             Webster’s says, “to keep in a state of validity.”

      !      Federal law requires the Secretary of the Air Force to “maintain a
             retired list containing the name of each retired commissioned officer
             of the Regular Air Force.” 10 U.S.C. § 8966(a) (emphasis added).
             This provision cannot mean the Secretary of the Air Force can decide
             at will not to keep track of retired Air Force officers. It must mean,
             as Webster’s says, “to keep in a state of validity.”

      !      Federal law requires the Secretary of Homeland Security to
             “establish an importer of record program to assign and maintain
             importer of record numbers.” 19 U.S.C. § 4320(a) (emphasis added).
             This provision cannot mean that the importer of record program can
             decide not to keep track of importer of record numbers at will. It
             must mean, as Webster’s says, “to keep in a state of validity.”

We could easily multiply these examples. The point is that Medina’s view that

“maintain” must include the power to terminate or modify is contrary to the

term’s ordinary usage, as evidenced by dictionaries and by its frequent use in the

U.S. Code.




                                         -18-
      In our view, then, when ERISA says that a church plan includes a plan

“maintained” by a principal-purpose organization, 29 U.S.C. § 1002(33)(C), it

simply means the principal-purpose organization, as Black’s says, “cares for the

plan for purposes of operational productivity.” 4 And this is precisely the point of

the Subcommittee.

      In fact, the CHI plan expressly delegates the power to “maintain” the plan

to the Subcommittee. See App., Vol. IV, at 934 (“The HR Committee of the

Board of Stewardship Trustees of [CHI] has primary responsibility for the

maintenance and compliance under the Plan and has fully and completely

delegated those responsibilities to [the Subcommittee] . . .”) (emphasis added), 

of the [Subcommittee] shall be to provide for the proper

operation, administration and maintenance of the Plan . . .”) (emphasis added).

And the Subcommittee has used that power, for example, to amend the plan. See,

e.g., App., Vol. IV at 1067 (explaining that CHI has delegated power to amend

the plan to the Subcommittee, and amending the definition of “spouse” in the plan

to include spouses of the same sex as the plan participant).

      We also conclude CHI and the Subcommittee qualify as “organizations.”

Black’s defines an organization as “[a] body of persons (such as a union or a

corporation) formed for a common purpose.” Black’s Law Dictionary, at 1210.


      4
        But we take no position today on what “maintain” might mean in other
provisions of ERISA, where context may present a different answer.

                                        -19-
Webster’s defines an organization as “a group of people that has a more or less

constant membership, a body of officers, a purpose, and usu[ally] a set of

regulations.” Webster’s Third, at 1590. Medina provides no authority for her

argument that the Subcommittee cannot be an organization. See Aplt. Br. 55–57.

On the contrary, the Subcommittee satisfies the dictionary definitions. The

Subcommittee Charter describes the common purpose: “provid[ing] for the proper

operation, administration and maintenance of the Plan and the underlying Code

Section 501(a) Master Trust . . . .” App., Vol. IV at 934. It also sets forth

regulations. See App., Vol. IV at 934–940. There is a body of officers: a Chair

and four or five voting members. App., Vol. IV at 934. And nothing indicates

the membership of the Subcommittee has not been “more or less constant.”

Medina thus fails to persuade us that the Subcommittee cannot be an

“organization” within the meaning of the statute.

      Medina’s arguments are, moreover, contrary to common sense. “A

textually permissible interpretation that furthers rather than obstructs the

document’s purpose should be favored.” Antonin Scalia & Bryan A. Garner,

Reading Law 63–65 (2012). If we accept Medina’s view of the statute, virtually

no plan administered by a benefits committee or similar organization could

qualify for the church-plan exemption. Medina would only allow the exemption

for wholly independent bodies, constituted with the principal purpose of

administering or funding a retirement plan, and endowed with the power to

                                         -20-
modify or terminate that plan. Medina does not explain what this organizational

structure would look like. If the principal-purpose organization administering the

plan cannot be part of the organization whose employees the plan is intended to

benefit, a religiously-affiliated entity would have to create a separate entity just to

administer the plan. So, hypothetically, an entity called Religious Healthcare

Corp. would have a plan to benefit its employees, but Religious Healthcare Corp.

would have to create a separate, self-funded, wholly independent Pension Corp. to

administer its plan. What is more, under Medina’s theory, Religious Healthcare

Corp. could not choose the board of Pension Corp., and Pension Corp. would need

the power to terminate Healthcare Corp.’s retirement plan at any time, without

needing to consult with Healthcare Corp. There may be some organization out

there that is structured like that, but it certainly is not the most intuitive way to do

it. And it is not clear what the advantage of such a structure would be, or why

Congress would have required it.

       We see no reason to follow Medina and require principal-purpose

organizations to be organizations independent of the parent entity, endowed with

the power to terminate benefit plans. No authority compels us to bar

organizations from constituting subsidiary committees to administer their church

plans. 5

       5
         Given the lack of authority supporting Medina’s arguments, we do not
think it necessary to discuss the deference owed under Skidmore v. Swift & Co.,
                                                                    (continued...)

                                          -21-
      We therefore find the Subcommittee is a principal-purpose “organization”

in the meaning of the statute, and that it “maintains” the CHI plan.

             3. Step Three: Is the Principal-Purpose Organization Associated
             with a Church? 6

      If CHI is associated with the Catholic Church, it follows that the

Subcommittee—a subdivision of CHI—is as well.

      First, as a matter of logic, a subdivision wholly encompassed by a larger

entity shares that entity’s affiliations. Medina herself argues the Subcommittee is

so closely affiliated with CHI that it is not even a separate organization. Aplt. Br.

at 55–57. Although we disagree with her definition of “organization,” we take

her point about the close relationship between the two entities.

      Second, the Subcommittee’s bonds with the Catholic Church are made clear

by the plan documents. The plan’s operating rules state that the Subcommittee



      5
       (...continued)

(1944), to the agencies tasked with administering ERISA. As the
Supreme Court did in Advocate, we are able to resolve this question solely by
recourse to the language and internal logic of the statute.
      6
         As we noted at oral argument, Medina failed to argue in her opening brief
that the Subcommittee was not associated with a church. Oral Arg. at
23:25–24:25. Although obliquely raised in her reply brief, “issues raised by an
appellant for the first time on appeal in a reply brief are generally deemed
waived.” Wheeler v. Comm’r, 

, 1291 (10th Cir. 2008). We could
therefore resolve this issue in favor of CHI on waiver grounds. But waiver is a
discretionary doctrine, so we go ahead and more fully explain our reasoning for
the benefit of the parties. See, e.g., Abernathy v. Wandes, 

, 552
(10th Cir. 2013) (“[T]he decision regarding what issues are appropriate to
entertain on appeal in instances of lack of preservation is discretionary.”).

                                        -22-
“shall be mindful of the Employer’s Philosophy and Mission, and the teachings

and tenets of the Roman Catholic Church and of the Sponsoring Congregations of

Catholic Health Initiatives. The [Subcommittee] shares common religious bonds

with the Roman Catholic Church and the Sponsoring Congregations of Catholic

Health Initiatives.” App., Vol. V at 1149, § 12.2. Although not necessarily

dispositive, we note this is precisely the language the statute uses to define what

it means to be associated with a church. See 29 U.S.C. § 1002(33)(C)(iv) (“An

organization . . . is associated with a church or a convention or association of

churches if it shares common religious bonds and convictions with that church or

convention or association of churches.” (emphasis added)).

      Again, we need not decide today the precise scope of what it means to be

“associated with” a church—we agree with the district court that the

Subcommittee qualifies.

             4. Propriety of Summary Judgment

      Summary judgment is proper if the movant can show “there is no genuine

dispute as to any material fact and the movant is entitled to judgment as a matter

of law.” Fed. R. Civ. P. 56(a). Medina argues the existence of genuine disputes

of material fact render summary judgment improper at this stage.

      She suggests two such disputes.

      First, she suggests there is a genuine dispute of material fact as to whether

CHI is “associated with” a church. Aplt. Br. at 68–71. But this is merely a

                                         -23-
repackaging of Medina’s legal argument that CHI does not qualify for ERISA’s

church-plan exemption. For the reasons explained above, we agree with the

district court’s resolution of this issue.

      Second, she argues the district court erred in concluding that “substantially

all” plan participants were employed by a permissible entity. Aplt. Br. at 71–73.

To qualify for the church-plan exemption, “substantially all of the individuals

included” in a church plan must be deemed employees of a church. 29 U.S.C.

§ 1002(33)(B)(ii). (That is, employees of a church or employees of a tax-exempt

nonprofit associated with a church. 

Before the district court

Medina argued the plan includes “more than an insubstantial number of

employees of organizations not associated with a church,” rendering it ineligible

for the exemption. Aplt. Br. at 72. The district court disagreed, and Medina now

argues there is a genuine factual dispute here precluding summary judgment.

Aplt. Br. at 72.

      The parties agreed before the district court that “substantially all” in this

context is satisfied if at least 85% of covered employees are deemed employees of

a church. App., Vol. III at 653. They rely on Cont’l Can Co. v. Chicago Truck

Drivers, Helpers & Warehouse Workers Union (Indep.) Pension Fund, 

, 1158–60 (7th Cir. 1990), which noted that IRS regulations always interpret

“substantially all” in the Internal Revenue Code as 85%, and interpreted

“substantially all” in the Multiemployer Pension Plan Amendments Act of 1980 in

                                             -24-
the same way. Without deciding the exact meaning of “substantially all” in 29

U.S.C. § 1002(33)(B)(ii)—for example, whether 75% would be enough—we

accept Medina’s concession that 85% would satisfy the statutory requirement.

      According to Medina, over 25% of plan participants work for an

organization called Centura Health. Aplt. Br. at 72 (citing App., Vol. IV at

1016–17). Medina cites deposition testimony from CHI’s Executive Vice

President of Mission, who stated that Centura Health was primarily a

“management company” that “manages a Catholic arm and an Adventist arm.”

App., Vol. VII at 1680. Another manager stated that Centura was not required to

accept the Ethical and Religious Directives for Catholic Health Care Services.

App., Vol. VII at 1690–91.

      This argument, however, does not get Medina very far. “[A]n issue of

material fact is genuine only if the nonmovant presents facts such that a

reasonable [factfinder] could find in favor of the nonmovant.” Planned

Parenthood of the Rocky Mountains Servs. v. Owens, 

, 916 (10th Cir.

2002). As the district court explained,

      Centura is a joint venture between CHI and Adventist Health System
      Sunbelt Healthcare Corporation . . . another church-affiliated hospital
      system. Centura’s two sole corporate members are Catholic Health
      Initiatives Colorado . . . which is 100 per cent controlled by CHI, and
      PorterCare Adventist Health System (“PorterCare”), which is 100 per
      cent controlled by Adventist Health. Both CHIC’s and PorterCare’s
      facilities are tax exempt, and each operates in accordance with the
      teachings of its respective founding church.


                                          -25-
App., Vol. IV at 654. On appeal, Medina does not dispute the district court’s

findings that Centura is a joint venture between CHI and Adventist, or that both

are church-affiliated and operate in accordance with the teachings of their

respective founding churches.

      ERISA does not require that all entities covered by a church plan be

associated with or controlled by a single church. Medina argues “the statute

precludes the conclusion that entities associated with different denominations can

be included in a single church plan, since ‘associated’ requires ‘shar[ing] common

religious bonds and convictions.’” Aplt. Br. at 73 (quoting 29 U.S.C. §

1002(33)(C)(iv)). CHI, on the other hand, argues there is no such requirement in

ERISA, and that IRS regulations allow the organizational structure at issue. Aple.

Br. at 52–54.

      The district court, reviewing the record, agreed with CHI—and so do we.

      We look first at the statutory language. Section 1002(33)(B)(ii) provides

that a “church plan” does not include a plan “if less than substantially all of the

individuals included in the plan are individuals described” in 29 U.S.C.

§ 1002(33)(A) or (C)(ii). Section 1002(33)(A) refers to employees of a church or

convention or association of churches; (C)(ii) refers to employees of tax-exempt

nonprofit organizations associated with a church or a convention or association of

churches. To put the point simply, then, a church plan does not include a plan if




                                         -26-
less than substantially all those covered are deemed employees of a church or

associated organization.

      The plain language thus allows the type of structure Centura has. To

paraphrase the statute, it says “a church plan does not include a plan where fewer

than ‘substantially all’ of the covered individuals are employees of a church.”

Changing “of a church” to “of the same church” or even of “the church” might

support Medina’s reading. But the natural reading of the statute is that a plan is

acceptable as long as “substantially all” the covered employees are deemed

employees of “some” church. And the relevant question is not whether

“substantially all” the covered employees share “common religious bonds and

convictions” with each other, as Medina would have it, but whether they share

“common religious bonds and convictions” with some church.

       The IRS’s 2002 Private Letter Ruling exempting CHI noted that only

1.64% of plan participants were employed by non-tax-exempt entities. App., Vol.

VI at 1397. It is, moreover, CHI’s policy that less than 5% of the plan’s

participants work for employers that are not tax-exempt church-controlled

entities. Medina disputes neither of these points, thus waiving any potential

counterargument.

      We therefore find no genuine dispute of material fact as to whether

substantially all of the plan participants are deemed employees of a church.




                                        -27-
             5. Conclusion

      We therefore find CHI’s plan satisfies the statutory requirements for the

church-plan exemption. To reiterate, the statutory inquiry is threefold. We find

CHI is a tax-exempt organization associated with a church. We find the

Subcommittee’s principal purpose is maintaining a retirement plan for the benefit

of CHI employees. We find the Subcommittee is also associated with a church.

We find no genuine disputes of material fact precluding summary judgment. This

being so, we find CHI’s plan qualifies for the church-plan exemption.

      B. Establishment Clause Challenge

      Finally, we address Medina’s argument that granting a church-plan

exemption to CHI would violate the Establishment Clause of the First

Amendment.

      The First Amendment provides that “Congress shall make no law respecting

an establishment of religion.” U.S. Const. amend. 1. To assess an Establishment

Clause challenge, we follow the tripartite test from Lemon v. Kurtzman, 

(1971). As we have explained,

      government action does not violate the Clause if (1) it has a secular
      purpose; (2) its principal or primary effect is one that neither advances
      nor inhibits religion; and (3) it does not foster an excessive government
      entanglement with religion. We interpret the first and second prongs of
      the Lemon test in light of Justice O’Connor’s endorsement test. That
      is, we ask whether government’s actual purpose is to endorse or
      disapprove of religion, and whether, irrespective of government’s actual
      purpose, the practice under review in fact conveys a message of
      endorsement or disapproval. We evaluate the government’s actions

                                        -28-
      from the perspective of a reasonable observer who is aware of the
      history, purpose, and context of the act in question.

Fields v. City of Tulsa, 

, 1010 (10th Cir. 2014) (citations omitted)

(quotations and alterations incorporated). The test is conjunctive: failing any

prong of the Lemon test means that the government action cannot withstand

Establishment Clause scrutiny. We therefore consider each prong separately.

      In First Amendment cases, we examine “constitutional facts” de novo.

Felix v. City of Bloomfield, 

, 853 (10th Cir. 2016). “When the

dispute concerns an alleged Establishment Clause violation, ‘constitutional facts’

are the ‘district court's findings on each part of the Lemon test.’” 

v. Haskell Cty. Bd. Of Comm’rs, 

, 795–96 (10th Cir. 2009)).

We therefore owe no deference to the district court’s constitutional analysis.

             1. Purpose

      “The question presented by the first prong of the Lemon test . . . is ‘whether

the government conduct was motivated by an intent to endorse religion.’” Am.

Atheists, Inc. v. Davenport, 

, 1118 (10th Cir. 2010) (quoting

Weinbaum v. City of Las Cruces, 

, 1030 (10th Cir. 2008)). “In

deciding whether the government’s purpose was improper, a court must view the

conduct through the eyes of an ‘objective observer,’ one who takes account of the

traditional external signs that show up in the text, legislative history, and

implementation of the statute, or comparable official act.” Weinbaum, 541 F.3d


                                         -29-
at 1031 (quotations omitted). “We will not lightly attribute unconstitutional

motives to the government, particularly where we can discern a plausible secular

purpose.” 

alteration incorporated).

       There is a plausible secular purpose here, as the district court found. First,

nothing in the text evinces an intent to favor one or all religions. The legislative

history, moreover, indicates Congress wanted to avoid unnecessary entanglement

with religion when it expanded ERISA’s church-plan exemption. See 125 Cong.

Rec. 10,052 (1979) (statement of Sen. Talmadge) (“If we have enacted a statute

that may require the church plans to come under ERISA, file reports, be subject to

the examination of books and records and possible foreclosure of church property

to satisfy plan liabilities, it must be changed because we have clearly created an

excessive Government entanglement with religion.”); 125 Cong. Rec. 12,106-07

(1978) (statement of Rep. Conable) (explaining that “when we enacted [ERISA],

we exempted church plans from the provisions of the act to avoid excessive

Government entanglement with religion in violation of the [F]irst [A]mendment to

the Constitution,” and that ERISA should be amended to account for different

organizational structures).

      We share the Supreme Court’s concern that these “excerpts from committee

hearings and scattered floor statements by individual lawmakers” are “‘among the

least illuminating forms of legislative history.’” 

(quoting NLRB v. SW General, Inc., 

, 943 (2017)). But we are not

                                         -30-
historians searching for absolute truth. We find the purpose of avoiding

government entanglement with religious organizations to be plausible, which is

all that is required in this context. Moreover, Medina’s briefing does not suggest

any other purpose the statute might have. See Aplt. Br. at 59–64 (discussing the

first prong of the Lemon test). Yet it is her burden to do so. See Am. 

& n.10.

        Nor do we see any reason to conclude this explanation is so implausible

that there must actually have been a religious purpose. Am. 

.

        And we find that this purpose—avoiding entanglement with religion—is a

secular one. A contrary conclusion would be illogical—it would place the

government in a double-bind. The government could not constitutionally pass a

law to avoid entanglement, but then the entanglement would run afoul of the

Constitution. And the Supreme Court has expressly held that a purpose of

avoiding government entanglement does not violate the Establishment Clause.

See Corp. of Presiding Bishop of Church of Jesus Christ of Latter-day Saints v.

Amos, 

, 335 (1987) (“Under the Lemon analysis, it is a permissible

legislative purpose to alleviate significant governmental interference with the

ability of religious organizations to define and carry out their religious

missions.”).

        We therefore find the first prong of the Lemon test satisfied.

                                          -31-
             2. Effect

      We next consider whether the church-plan exemption violates the second

prong of the Lemon test. It is permissible government action only if its principal

or primary effect is one that neither advances nor inhibits religion. As we

explained above, our circuit interprets the effect prong of Lemon in light of

Justice O’Connor’s endorsement test. We therefore ask more specifically whether

the challenged conduct has “the effect of conveying a message that religion or a

particular religious belief is favored or preferred.” 

(quoting Bauchman for Bauchman v. W. High Sch., 

, 551 (10th Cir.

1997)). The effect prong “looks through the eyes of an objective observer,”

aware of the “purpose, context, and history” of the government action in question,

“kin to the fictitious ‘reasonably prudent person’ of tort law.” 

      Medina argues exempting religious hospital systems from ERISA conveys a

message that religion is favored. She also argues CHI’s exemption favors

religious adherents at the expense of nonadherents, because nonadhering

participants in the CHI plan do not receive the disclosures required by ERISA,

their benefits are not insured by the Pension Benefit Guaranty Corporation, and

because it takes longer for participants’ right to benefits to vest than it would in

an ERISA-compliant plan.

      But Supreme Court precedent forecloses Medina’s argument. In Amos, 

, the Supreme Court considered a challenge to a section of the Civil

                                         -32-
Rights Act that exempted religious organizations from Title VII’s prohibition

against discrimination in employment on the basis of religion. Even though the

challenged provision exempted religious organizations from an otherwise

applicable law, the Supreme Court held that it did not run afoul of the second

prong of the Lemon test. The Court explained that

      it has never indicated that statutes that give special consideration to
      religious groups are per se invalid. That would run contrary to the
      teaching of our cases that there is ample room for accommodation of
      religion under the Establishment Clause. Where, as here, government
      acts with the proper purpose of lifting a regulation that burdens the
      exercise of religion, we see no reason to require that the exemption
      comes packaged with benefits to secular entities.

(citation omitted). We do not find the present case distinguishable.

      Consider the issue from first principles. Any law of general applicability

that exempts religious organizations from its requirements could be said to

convey a message that religion is favored. Religion is, after all, being exempted

from a rule everyone else has to follow. Such an approach would mean that

Congress could never exempt religious organizations from laws that might burden

them—even when burdening religious organizations would itself run afoul of the

Constitution. But this is common practice. A number of statutes regulate wide

swathes of the American economy. And many of these statutes expressly exempt

religious organizations from various requirements. Apart from ERISA’s church-

plan exemption, examples include the Internal Revenue Code, 26 U.S.C.

§§ 6033(a)(3)(A)(i), (iii) (carving out “churches, their integrated

                                         -33-
auxiliaries, . . . conventions or associations of churches,” and “the exclusively

religious activities of any religious order” from a tax-filing requirement); Title

VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-1(a) (exempting religious

organizations from religious-discrimination claims); the Americans with

Disabilities Act of 1990, 42 U.S.C. § 12113(d) (exempting religious organizations

from anti-discrimination requirements), 

(exempting religious

organizations from public accommodation requirements); and likely many others

besides. Accepting Medina’s argument that accommodation equals favoritism

would wreak havoc to these schemes. And a unanimous Supreme Court has

cautioned that courts of appeals misconstrue Supreme Court precedent if they

read it so that “all manner of religious accommodations would fall.” Cutter v.

Wilkinson, 

, 724 (2005).

      Moreover, without the rule in Amos we would be in a position that might

require us to strike down religious-accommodation provisions in a number of

regulatory statutes, such as the ones cited in the previous paragraph from the

Internal Revenue Code, Title VII, and the Americans with Disabilities Act. And

it would entail putting Congress between a rock and a hard place: we would not

let them exempt religious organizations from regulatory requirements, but we also

would not let Congress subject organizations to onerous regulatory requirements.

See, e.g., NLRB v. Catholic Bishop, 

(1979) (construing the National

Labor Relations Act to remove the NLRB’s jurisdiction over schools operated by

                                         -34-
churches to avoid constitutional difficulties). To paraphrase Voltaire, then, if

Amos did not exist, we would have to invent it. 7

      We therefore find that ERISA’s church-plan exemption does not have the

principal or primary effect of advancing religion. Or, phrased in terms of Justice

O’Connor’s endorsement test, exempting religious organizations from complying

with a regulatory scheme does not convey an impermissible message that religion

is favored or preferred. Indeed, Justice O’Connor herself wrote separately in

Amos to express her view that an accommodation of the exercise of religion is not

a government endorsement of religion. 

(O’Connor, J.,

concurring). And that is precisely what ERISA’s church-plan exemption

provisions do.

             3. Entanglement

       Finally, the Lemon test asks whether the government action fosters an

excessive government entanglement with religion. Medina argues the church-plan

exemption entangles government in religion because it requires courts and

agencies to determine whether CHI is “associated with” a church. CHI, on the

other hand, argues applying the church-plan exemption would foster

entanglement. We agree with CHI.



      7
       Cf. Voltaire, Letter to the Author of the Book of the Three Impostors,
quoted in Bartlett’s Familiar Quotations 310 (Justin Kaplan gen. ed., 16th ed.
1992) (“If God did not exist, it would be necessary to invent him.”)

                                        -35-
      We have difficulty imagining how this area of the law would function if

merely determining whether an organization is a religious organization, or

associated with one, constituted impermissible entanglement. How would a

church exercise its many constitutional and statutory rights if agencies and courts

could not assess whether it was, in fact, a church?

      On the other hand, we are persuaded that subjecting religious organizations

to ERISA would foster impermissible entanglement.

      If we took Medina’s argument for all it is worth, then we would have to

subject religious organizations to ERISA. And that would foster even greater

entanglement. As the district court noted, “[e]nsuring ongoing ERISA

compliance by church-associated entities undoubtedly would require long-term,

continuing monitoring not involved in the relatively time-delimited analysis of a

claim of entitlement to the exemption.” App., Vol. III at 658. Complying with

ERISA’s fiduciary rules is no simple matter. A Catholic church, or entity

associated with one, might want to invest its plan assets in service of certain

social goals. But that could run afoul of ERISA, which requires diversification of

plan assets, 29 U.S.C. § 1104(a)(1)(C), and that plan assets be held for the

exclusive purpose of providing benefits and defraying reasonable plan expenses.

29 U.S.C. § 1103(c). There would, moreover, be pervasive monitoring to

determine whether the church or church-associated entity was complying with

ERISA.

                                         -36-
      If we merely invalidated the portion of ERISA extending the church-plan

exemption to entities “associated with a church,” then ERISA would require

religious organizations, in order to receive the exemption’s benefits, to adopt a

particular structure, thus interfering in the internal organization of a religious

institution. And that is precisely what Supreme Court precedent forbids: the

Religion Clauses jurisprudence “radiates . . . a spirit of freedom for religious

organizations, an independence from secular control or manipulation—in short,

power to decide for themselves, free from state interference, matters of church

government as well as those of faith and doctrine.” Hosanna-Tabor Evangelical

Lutheran Church & Sch. v. E.E.O.C., 

, 186 (2012) (quoting Kedroff

v. Saint Nicholas Cathedral of Russian Orthodox Church in North America, 

, 116 (1952)).

      We therefore find that, far from entangling the government in the affairs of

religious institutions, the church-plan exemption avoids the entanglement that

would likely occur in its absence.

                                       *    *     *

      Finding the church-plan exemption survives all three prongs of the Lemon

test, we reject Medina’s argument that exempting CHI from ERISA runs afoul of

the Establishment Clause.




                                           -37-
                             III. Conclusion

     For the reasons above, we AFFIRM the district court’s grant of summary

judgment.




                                    -38-