National Religious Broadcasters Noncommercial Music License Committee v. CRB

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued February 17, 2023 Decided July 28, 2023

No. 21-1243

NATIONAL RELIGIOUS BROADCASTERS NONCOMMERCIAL
MUSIC LICENSE COMMITTEE,
APPELLANT

v.

COPYRIGHT ROYALTY BOARD AND LIBRARIAN OF CONGRESS,
APPELLEES

GOOGLE LLC, ET AL.,
INTERVENORS

Consolidated with 21-1244, 21-1245

On Appeals from a Final Determination
of the Copyright Royalty Board

Samir Deger-Sen argued the cause for appellant National
Association of Broadcasters. With him on the briefs were
Joseph R. Wetzel, Andrew M. Gass, Sarang V. Damle, and
Blake E. Stafford.
 2
Karyn K. Ablin argued the cause and filed the briefs for
appellant National Religious Broadcasters Noncommercial
Music License Committee. John J. Bursch, Rory T. Gray, and
Erin M. Hawley entered appearances.

Matthew S. Hellman argued the cause and filed the briefs
for appellant SoundExchange, Inc. Previn Warren entered an
appearance.

Jennifer L. Utrecht, Attorney, U.S. Department of Justice,
argued the cause for appellees. With her on the brief were
Michael D. Granston, Deputy Assistant Attorney General, and
Daniel Tenny, Attorney.

David P. Mattern argued the cause for intervenors Google
LLC, et al. in support of appellees. With him on the brief were
Sarang V. Damle, Blake E. Stafford, Kenneth L. Steinthal,
Joseph R. Wetzel, Andrew M. Gass, Samir Deger-Sen, Joshua
N. Mitchell, and Karyn K. Ablin. John J. Bursch, Jason B.
Cunningham, Rory T. Gray, and Erin M. Hawley entered
appearances.

Matthew S. Hellman was on the brief for intervenor
SoundExchange, Inc. in support of appellees. Previn Warren
entered an appearance.

Before: MILLETT, WILKINS, and PAN, Circuit Judges.

Opinion for the Court filed PER CURIAM.

Every five years, the Copyright Royalty Board (the
“Board”) issues a statutory license that establishes the terms
and rates under which certain entities that stream copyrighted
songs over the internet make royalty payments to the songs’
copyright owners. The “webcasters” that are subject to the
 3
license are “noninteractive” — i.e., they stream music without
letting their listeners choose songs on demand. This appeal
challenges on various grounds the Board’s most recent
noninteractive webcaster license Final Determination,
covering calendar years 2021 through 2025. We sustain the
Board’s Final Determination in all respects.

I

The Copyright Act, 17 U.S.C. § 101 et seq., provides the
statutory framework for regulating copyrights. Under that
framework, a recorded song has two components with distinct
rights: (1) the “musical work,” which is the song’s underlying
composition (i.e., the lyrics and melody); and (2) the “sound
recording,” which is a recorded version of the song. See
SoundExchange, Inc. v. Copyright Royalty Board, 904 F.3d 41,
46 (D.C. Cir. 2018).

Historically, the owner of a musical work had an exclusive
right of public performance but the owner of a sound recording
did not. SoundExchange, 904 F.3d at 46. Thus, an FM radio
station could broadcast a sound recording without permission
from its copyright owner. But in 1995, Congress amended the
Copyright Act to grant sound-recording owners the exclusive
right of public performance “by means of a digital audio
transmission.” Digital Performance Right in Sound
Recordings Act of 1995, Pub. L. No. 104-39, § 2, 109 Stat. 336,
336 (codified at 17 U.S.C. § 106(6)). Under the amended
statute, a webcaster cannot stream a sound recording without
paying royalties to its copyright owner.

In defining the scope of this new right, Congress
distinguished between webcasters (also known as “digital
audio services”) that are “interactive” and “noninteractive.”
Interactive services let users choose the particular songs they
 4
want to listen to on demand, e.g., Spotify, while noninteractive
services do not, e.g., Pandora. See 17 U.S.C. § 114(j)(7).
Interactive webcasters must contract directly with copyright
owners to obtain public performance rights for their sound
recordings. Id. § 114(d)(2)(A)(i). By contrast, Congress
tasked the Copyright Royalty Board with creating a
compulsory license covering the use of sound recordings by all
noninteractive webcasters. Id. § 114(f)(1). The license is
“compulsory” because copyright owners cannot opt out of it
unless they negotiate individual settlement agreements with
noninteractive webcasters. Id. §§ 114(f)(1)–(2). Royalties
under the compulsory license are paid to a “nonprofit
collective,” which distributes the funds to performing artists or
other copyright owners. Id. § 114(g)(2). Meanwhile,
traditional AM/FM radio, also known as terrestrial or over-the-
air radio, still plays by the old rules: those radio stations pay
no royalties to broadcast songs to listeners, and copyright
owners instead treat AM/FM radio as a promotional
opportunity.

The Board must set the rates and terms of the compulsory
license for noninteractive webcasters every five years.
17 U.S.C. § 114(f)(1)(A). Interested parties may negotiate
settlement agreements amongst themselves to opt out of the
compulsory license. Id. § 114(f)(2). If a particular record label
and a webcaster negotiate a settlement agreement that sets
terms for the webcaster’s use of the record label’s copyrighted
sound recordings, that agreement controls instead of the
Board’s compulsory license. Non-settling parties are subject
to the license, and the Board holds an evidentiary proceeding
to determine the applicable terms and rates under that license.
SoundExchange, 904 F.3d at 46–47. Noninteractive
webcasting produces hundreds of billions of streams per year,
the vast majority of which are covered by the compulsory
license rather than by a settlement.
 5

Congress set forth instructions for the Board’s compulsory
license determinations in 17 U.S.C. § 114(f)(1)(B). The statute
directs the Board to “distinguish among the different types of
[webcasting] services then in operation” based on, among other
factors, the “quantity and nature of the use of sound recordings
and the degree to which use of the service may substitute for or
may promote the purchase of phonorecords by consumers.” Id.
Applying that standard, the Board has previously distinguished
between commercial and noncommercial webcasting services
and between subscription-based and nonsubscription-based
commercial services. See Determination of Royalty Rates and
Terms for Ephemeral Recording and Webcasting Digital
Performance of Sound Recordings (Web IV), 81 Fed. Reg.
26,316, 26,409 (May 2, 2016). For each different type of
service, the Board must establish rates and terms that represent
what “would have been negotiated in the marketplace between
a willing buyer and a willing seller.” 17 U.S.C. § 114(f)(1)(B).
This is called the “willing buyer/willing seller” standard.
SoundExchange, 904 F.3d at 56. In so doing, the Board must
consider factors including the effect of the license’s rates and
terms on other sources of sound recording revenue, such as
whether a service tends to boost or deflate interactive streaming
royalties. 17 U.S.C. § 114(f)(1)(B)(i)(I). The Board may also
consider voluntary license agreements negotiated for
comparable services as “benchmarks” that provide reference
points in its analysis. SoundExchange, 904 F.3d at 47; see
17 U.S.C. § 114(f)(1)(B)(ii). And the Board’s rates and terms
must “include a minimum fee” that each webcaster must pay to
use the compulsory license. 17 U.S.C. § 114(f)(1)(B). As we
have made clear, “the statute does not require that the
[hypothetical] market assumed by the [Board] achieve
metaphysical perfection.” Intercollegiate Broad. Sys., Inc. v.
Copyright Royalty Board (Intercollegiate II), 796 F.3d 111
(D.C. Cir. 2015).
 6

This appeal concerns the Board’s fifth noninteractive
webcaster rate Final Determination, which set the rates and
terms of the statutory license for calendar years 2021 through
2025. Determination of Rates and Terms for Digital
Performance of Sound Recordings and Making of Ephemeral
Copies To Facilitate Those Performances (Web V), 86 Fed.
Reg. 59,452 (Oct. 27, 2021). The Board’s previous four
noninteractive webcaster rate determinations were reviewed
and largely upheld by this court. See SoundExchange, 904 F.3d
41 (reviewing Web IV); Intercollegiate II, 796 F.3d 111
(reviewing Web III); Intercollegiate Broad. Sys., Inc. v.
Copyright Royalty Board (Intercollegiate I), 574 F.3d 748
(D.C. Cir. 2009) (reviewing Web II); Beethoven.com LLC v.
Librarian of Cong., 394 F.3d 939 (D.C. Cir. 2005) (reviewing
Web I).1

The Web V evidentiary hearing lasted from August 4,
2020, to September 9, 2020. Ten parties participated,
including the appellants and intervenors in this consolidated
case: (1) the National Association of Broadcasters (the
“NAB”), an association of radio and television stations; (2) the
National Religious Broadcasters Noncommercial Music
License Committee (the “Committee”), an arm of a trade
association that represents religious radio and television
stations; (3) SoundExchange, Inc., a collective management
organization that represents sound-recording copyright holders

1
For the underlying Board determinations, see Web IV, 81 Fed.
Reg. 26,316; Determination of Royalty Rates for Digital
Performance Right in Sound Recordings and Ephemeral Recordings
(Web III), 79 Fed. Reg. 23,102 (April 25, 2014); Digital Performance
Right in Sound Recordings and Ephemeral Recordings (Web II), 72
Fed. Reg. 24,084 (May 1, 2007); Determination of Reasonable Rates
and Terms for the Digital Performance of Sound Recordings and
Ephemeral Recordings (Web I), 67 Fed. Reg. 45,240 (July 8, 2002).
 7
and artists; and (4) Google LLC, a technology company. The
Board heard oral testimony from thirty-three witnesses and
received written testimony from eight, which together included
thirteen qualified experts. The Board admitted 748 exhibits
into evidence, comprising more than 900,000 pages of
documents. After the hearing, the parties submitted proposed
findings and conclusions, and responses thereto, and made
closing arguments on November 19, 2020. The Librarian of
Congress published the Board’s Final Determination on
October 27, 2021.

In its Final Determination, the Board identified three
relevant categories of webcasters: commercial subscription
webcasters, commercial nonsubscription webcasters, and
noncommercial webcasters. See Web V, 86 Fed. Reg. at
59,589. Commercial subscription webcasters are services like
Pandora Plus that collect payments from their listeners. See
SoundExchange, 904 F.3d at 48. Commercial nonsubscription,
i.e., “ad-supported,” webcasters are services like Free Pandora
that collect payment from advertisers rather than listeners. See
id. at 48, 58. Noncommercial webcasters are services owned
by a government entity or a nonprofit, such as National Public
Radio (“NPR”) and certain religious webcasters. See Web V,
86 Fed. Reg. at 59,593; 17 U.S.C. § 114(f)(4)(E)(i). Besides
noncommercial, educational, and public webcasters, all other
webcasters are commercial. Web V, 86 Fed. Reg. at 59,592.

For all webcasters, the Board set a minimum fee of $1,000
per channel or station. Web V, 86 Fed. Reg. at 59,589.
Commercial webcaster license fees were capped at $100,000.
Id. at 59,589. Payment of the minimum fee grants a webcaster
access to the compulsory license. See 17 U.S.C.
§ 114(f)(1)(B). Each licensee can have multiple channels, but
with the $100,000 cap, a large commercial webcaster licensee
pays the minimum fee only for its first one hundred channels.
 8
This provision doubled the prior minimum-fee payment—
which was $500 per channel and capped at $50,000 per
licensee. See Web IV, 81 Fed. Reg. at 26,409; Web III, 79 Fed.
Reg. at 23,132.

Beyond the minimum fee, when setting royalty rates for
all webcasters, the Board puts forward an amount to be paid
“per performance.” One copyrighted song heard by one
listener is a performance. Web V, 86 Fed. Reg. at 59,593. So,
for instance, if the Board set a royalty rate at $0.002 per
performance, and if a webcaster subject to that rate streamed
two copyrighted songs to one thousand listeners each, it would
have to pay for two thousand performances, amounting to
$4.00 total.

For commercial subscription webcasters, the Board set a
2021 royalty rate of $0.0026 per performance, adjusted
annually for inflation. Web V, 86 Fed. Reg. at 59,589. For all
commercial webcasters, the minimum fee of $1,000 covers a
service’s first $1,000 in royalty payments, id., or about 385,000
performances for commercial subscription webcasters in 2021.

For commercial nonsubscription webcasters, the Board set
a 2021 royalty rate of $0.0021 per performance, adjusted
annually for inflation. Web V, 86 Fed. Reg. at 59,589. The
minimum fee of $1,000 covered roughly 475,000
performances for commercial nonsubscription webcasters
in 2021.

For noncommercial webcasters, the Board set a payment
structure under which the webcaster receives a monthly
allowance of 159,140 aggregate tuning hours (“ATH”) by
paying the minimum fee; and pays a 2021 royalty rate of
$0.0021 per performance above that threshold, adjusted
annually for inflation—the same rate that applies to
 9
commercial nonsubscription webcasters. Web V, 86 Fed. Reg.
at 59,589. ATH is, essentially, the cumulative time spent
listening to copyrighted songs. See id. at 59,592. For instance,
if 1,000 individuals each listened to one hour of copyrighted
songs, that would amount to 1,000 ATH. See id.

Four aspects of the Board’s decision are challenged on
appeal. First, the NAB argues that the Board should have
adopted its proposal to distinguish simulcasters from other
commercial nonsubscription webcasters. Simulcasters are
traditional AM/FM stations that simultaneously stream their
programming on the internet. The NAB sought a lower rate for
those stations. Second, the NAB and the Committee
(collectively, the “Services”) argue that the Board should have
rejected SoundExchange’s proposal to double the minimum
fee to $1,000 per channel and $100,000 per licensee. The
Services proposed keeping the incumbent minimum fee
structure instead. Third, the Committee argues that the Board
should have set a lower rate for noncommercial webcasters,
based on a settlement agreement between SoundExchange,
NPR, and the Corporation for Public Broadcasting (“CPB”)
that the Committee proffered as a benchmark. And fourth,
SoundExchange argues that the Board should have set a higher
commercial nonsubscription rate, contending that the Board’s
rate is lower than copyright owners’ opportunity costs.

The NAB, the Committee, and SoundExchange timely
appealed the aforementioned aspects of the Board’s Final
Determination under 17 U.S.C. § 803(d)(1). SoundExchange
intervened on behalf of the government in the appeals brought
by the Services, while the Services and Google intervened on
behalf of the government in SoundExchange’s appeal.
 10
II

We review the Board’s rate determinations under Section
706 of the Administrative Procedure Act. See 17 U.S.C.
§ 803(d)(3). We uphold the results of the Board’s proceedings
“unless they are arbitrary, capricious, contrary to law, or not
supported by substantial evidence.” Intercollegiate I, 574 F.3d
at 755. Our “[r]eview of administratively determined rates is
‘particularly deferential’ because of their ‘highly technical’
nature.” Id. (quoting East Ky. Power Coop. v. FERC, 489 F.3d
1299, 1306 (D.C. Cir. 2007)). Applying that standard, we
sustain the Board’s Final Determination against the appellants’
challenges.

III

A

As an association of radio and television stations, the NAB
represents hundreds of simulcasters nationwide. Its members
range in size from larger broadcasters, such as iHeartMedia—
a company operating around 850 radio stations—to smaller
broadcasters, such as individuals operating only a handful of
stations. Focusing on the three identified categories of
webcasters, the NAB contests the Board’s decision to place
simulcasters in the broad commercial nonsubscription
webcaster category, thus subjecting them to the same rate as
what the NAB argues are fundamentally different custom radio
services. Custom radio refers to services like Pandora, which
allow users to skip songs and to “curate the listening
experience.” Web V, 86 Fed. Reg. at 59,547. By contrast,
simulcasters are traditional AM/FM stations that
simultaneously stream their programming on the internet
without allowing for customization.
 11
During the Board’s proceedings, the NAB put forth a rate
structure under which simulcasters would pay $0.0008 per
play, and other eligible commercial nonsubscription
webcasters would pay $0.0016 per play. If adopted, the Board
would have distinguished simulcasters from other webcasters
for the first time. Web V, 86 Fed. Reg. at 59,547. According
to the NAB, the Board’s statutory obligation to distinguish
between different services, see 17 U.S.C. § 114(f)(1)(B),
required it to adopt this proposal because simulcasting is
critically distinct from other types of commercial webcasting.
As support, the NAB offered various voluntary agreements as
benchmarks, including “[d]irect license agreements between
sound recording rights owners and webcaster iHeart and
license agreements for musical compositions between
performing rights organizations and webcasters Pandora and
iHeart.” Web V, 86 Fed. Reg. at 59,547. Ultimately, the Board
found that a new distinction was unwarranted based on the
record, and more specifically, that “significant evidence”
showed “simulcasters and other commercial webcasters
compete in the same submarket and therefore should be subject
to the same rate.” Id. at 59,565.

A second point of contention arose regarding the
statutorily mandated minimum fee. See 17 U.S.C.
§ 114(f)(1)(B). Having maintained the same $500 minimum
fee since 2006, the Board considered SoundExchange’s
proposal to double the fee to $1,000 in order “at least to cover
[its] administrative cost.” Web V, 86 Fed. Reg. at 59,579
(internal quotation marks omitted). The Services collectively
challenged SoundExchange’s request, arguing that, because
the fee is solely meant to cover “incremental administrative
costs”—meaning fees associated with administering the
webcasting license—SoundExchange’s average administrative
cost was “irrelevant.” Id. at 59,580 (emphasis omitted). In the
Services’ view, what the Board accepted as “average
 12
administrative cost” in fact encompasses SoundExchange’s
“total costs,” including fees unrelated to license administration.
NAB Opening Br. 17–18 (emphasis omitted); see Committee
Opening Br. 51–52. Thus, the Services asked the Board to
adopt their narrower view of the minimum fee. But finding no
statutory basis that supported it doing so, the Board rejected the
Services’ proposal and explained why the record justified
doubling the minimum fee.

On appeal, the NAB advances a two-part theory, arguing
the Board’s determination is arbitrary, capricious, or otherwise
contrary to law. First, the NAB challenges the Board’s refusal
to distinguish simulcasters from other nonsubscription
commercial services as violating 17 U.S.C. § 114(f)(1)(B)’s
plain language. It also argues that the Board’s analysis
justifying its decision was arbitrary and capricious. Second,
the Services challenge the Board’s decision to double the
minimum fee in consideration of SoundExchange’s average
administrative costs.

Unpersuaded by either theory, we affirm both aspects of
the Board’s determination.

1

Looking first to the NAB’s categorization-related
arguments, we uphold the Board’s determination, finding this
record failed to establish that simulcasters warrant a different
royalty rate than other commercial nonsubscription services.
According to the NAB, the Board violated its statutory
obligation to distinguish services when it acknowledged that
simulcasters differ from custom radio, yet still subjected both
groups to the same rate. After reviewing this record, however,
we confirm that the Board reasonably evaluated the NAB’s
differentiation evidence and appropriately exercised its
 13
discretion in declining to set a separate, lower rate for
simulcasters.

Recall that the Board “shall distinguish among the
different types of services then in operation” when
“establish[ing] rates and terms that most clearly represent”
what “would have been negotiated in the marketplace between
a willing buyer and a willing seller.” 17 U.S.C. § 114(f)(1)(B).
The Copyright Act also instructs the Board to base its decision
on criteria such as “the quantity and nature of the use of sound
recordings and the degree to which use of the service may
substitute for or may promote the purchase of phonorecords by
consumers.” Id.

Here, the Board satisfied 17 U.S.C. § 114(f)(1)(B) by
maintaining the preexisting rate categories and distinguishing
royalty rates for (1) commercial subscription services;
(2) commercial nonsubscription services; and (3)
noncommercial services. When setting rates, we have
explained that the Board has discretion in determining what to
use as a starting point, so long as it explains itself. Music
Choice v. Copyright Royalty Board, 774 F.3d 1000, 1012 (D.C.
Cir. 2014) (finding that the Board “did not err when [it] used
the prevailing rate as the starting point of [its] analysis,” given
“the lack of creditable benchmarks in the record” and the
Board’s “reasoned explanation”). Furthermore, Section
114(f)(1)(B) contemplates the Board will “make adjustments
to the prevailing rate” and also “consider prior determinations”
in its decisionmaking. Music Choice, 774 F.3d at 1012
(internal quotation marks omitted).

The Board has never set a lower rate for simulcasters. So
its decision not to here is not an “unexplained presumption in
favor of uniform rates.” NAB Opening Br. 29. Rather, the
Board was justified in relying on its three preexisting rate
 14
category distinctions to at least determine a starting point. Web
V, 86 Fed. Reg. at 59,547; see also Web I, 67 Fed. Reg. at
45,252 (adopting a single rate for commercial webcasters);
Web II, 72 Fed. Reg. at 24,095 (refusing to establish a separate
rate for simulcasters); Web IV, 81 Fed. Reg. at 26,323
(rejecting arguments for a separate simulcaster rate). It was
thus up to the NAB to establish a record showing why, and
how, this starting point should be altered to reflect the willing
buyer/willing seller standard. See Web IV, 81 Fed. Reg. at
26,320 (“As the proponent of a rate structure that treats
simulcasters as a separate class of webcasters, the NAB bears
the burden of demonstrating not only that simulcasting differs
from other forms of commercial webcasting, but also that it
differs in ways that would cause willing buyers and willing
sellers to agree to a lower royalty rate in the hypothetical
market.”); Web I, 67 Fed. Reg. at 45,254 (referring to “the
burden of proof on the broadcasters to present evidence to
distinguish between the direct transmission of their programs
over the Internet and the retransmission of the same
programming made by a third-party”).

Our reasoning here also resolves the NAB’s secondary
challenges to the Board’s decision, rejecting the new simulcast
distinction. First, the NAB argues that the Board improperly
made a presumption in favor of uniform rates, which required
the NAB to present contrary evidence to rebut the Board’s
presumption. The NAB asserts that the Board instead should
have supported its decision with substantial evidence. This
argument gets things backwards. The Board is allowed to
consider its prior determinations, and the NAB failed to meet
its burden to show this record warranted something different.
Second, the NAB argues that the Board’s analysis discussing
competition between simulcasters and other commercial
webcasters was “far too generalized to have any relevance” and
rendered the determination arbitrary and capricious. NAB
 15
Opening Br. 37–40. We disagree because, as explained
throughout this section, the Board appropriately exercised its
discretion in finding the differentiation evidence failed to
support a new simulcast rate category under the willing
buyer/willing seller standard.

The Board reasonably declined to interpret the NAB’s
evidence as supporting a separate rate for simulcasters. The
NAB presented benchmark agreements that it claimed were
evidence that simulcasters should be subject to a lower rate
than custom radio. The Board rejected the NAB’s iHeart/Indie
Agreements as benchmarks because they only covered “a small
portion of the sound recordings performed by iHeart, and an
even smaller portion of the entire market for simulcast, custom
radio, and internet radio performances.” Web V, 86 Fed. Reg.
at 59,549 (emphasis added). The NAB also introduced survey
evidence that it argued showed simulcasters should be subject
to a lower rate. Importantly, the Board also declined to rely on
the NAB’s Hauser Survey—a survey intended to reveal the
“percentage of respondents that, in the absence of simulcasts,
would consume content from” other alternative activities. Id.
at 59,551. Doing so was permissible given the Board’s
concerns with the survey’s design, such as its failure to include
services like internet radio services covered by the statutory
license. Id. at 59,563–59,565.

Nevertheless, the NAB insists that the differences between
simulcasts and custom radio show the Board was required to
distinguish a separate, simulcaster rate. But the Board acted
reasonably in taking issue with the fact that “the bulk” of the
NAB’s evidence “regarding differentiated use of music versus
non-music content” was limited to comparing simulcasts
against custom radio services, when the NAB’s proposal was
to establish a simulcast rate separate from “the full scope of
 16
noninteractive webcasting[.]” Web V, 86 Fed. Reg. at 59,551
(emphasis added).

Critically, the NAB also failed to show that internet
simulcasters constitute a distinct market segment. The absence
of distinct market segmentation evidence is detrimental to the
NAB’s case because the statutory distinction requirement
“mean[s] that distinct segments of webcasters—such as
noncommercial services—receive their own rates and terms.”
SoundExchange, 904 F.3d at 47 (emphasis added). This
emphasis on evidence of competition is rooted in the willing
buyer/willing seller standard. It is unclear whether a willing
buyer and a willing seller would agree to a new rate for a
service, falling under a preexisting rate category, absent
evidence showing that the service constitutes a distinct
segment of webcasters. In SoundExchange, we shed light on
the distinction requirement when discussing how to determine
if a service constitutes a distinct segment of webcasters. There,
we affirmed the Board’s decision to distinguish between ad-
based and subscription-based services because the record
showed each service “appeal[ed] to a different segment of the
market[.]” Id. at 58. But the NAB never argues that any such
market segmentation would actually result in a lower rate for
simulcasters based on a rate that a willing buyer and a willing
seller would accept.

Our SoundExchange decision also addressed the Board’s
prior determinations more generally and upheld the process
through which it determines how to distinguish between
services. 904 F.3d at 58–59. The Board’s process began in
Web II, when it established that “the key question in
ascertaining the propriety of differentiation is whether the
services occupy ‘distinct segment[s]’ of the market or instead
compete for listeners.” Id. at 58 (quoting Web II, 72 Fed. Reg.
at 24,097–24,098). To answer this question, the Board
 17
assesses service competition by “examin[ing] whether the
services compete with each other for listeners, or whether one
service instead operate[d] in a submarket separate from and
noncompetitive with the other[.]” Id. (internal quotation marks
and citation omitted). And as for market segmentation, the
Board considers various factors, “including whether
comparable agreements have been negotiated in which one
service paid a lower rate than the other.” Id. In short, our
precedent counsels that when evaluating categorization
challenges under 17 U.S.C. § 114(f)(1)(B), we consider
whether the record shows the Board reasonably distinguished
between the “types of services then in operation[,]” id., and
established rates for each “distinct segment[,]”
SoundExchange, 904 F.3d at 58, under the willing
buyer/willing seller standard. See Intercollegiate II, 796 F.3d
at 128 (emphasizing that the minimum fee distinction
requirement applies to each “type of service,” 17 U.S.C.
§ 114(f)(1)(B)—“not for each individual webcaster”).

Instead of structuring its position around why simulcasters
constitute a distinct segment of the market from all other
commercial nonsubscription services, the NAB dedicated most
of its argument to distinguishing simulcasters from a mere
subset of other commercial webcasters—custom internet radio.
In doing so, the NAB argued that the Board failed to justify its
refusal to establish a separate rate for simulcasters despite its
acknowledgment of certain differences between simulcasters
and custom radio. But the NAB does not argue that
simulcasting is different from commercial webcasting more
generally—only that it differs from custom radio. There are
other types of commercial webcasting that may not have the
same features as custom radio, but the NAB does not argue that
a willing buyer and a willing seller would agree to a lower rate
for simulcasting than for any other types of commercial
webcasting. Although such an argument could succeed if the
 18
record showed that willing buyers and willing sellers would
agree to lower royalty rates for simulcasting than for other
types of commercial webcasting, the NAB did not do so here.
Moreover, we underscore the “technical nature” of rate
determinations and find the NAB failed to meet its burden here.
Intercollegiate II, 796 F.3d at 127 (quoting Intercollegiate I,
574 F.3d at 755). We acknowledge, however, that future
records may warrant new rate category distinctions.
SoundExchange, 904 F.3d at 58 (describing evidence
appropriately distinguishing between ad-based and
subscription-based services).

2

Together, the Services contest the minimum fee, arguing
the Board erroneously raised the fee to account for costs other
than the incremental cost of administering webcasting licenses.
As explained below, we reject the Services’ challenge and
uphold the $1,000 minimum fee as reasonable.

Although Congress requires the Board to establish a
minimum fee for each service under 17 U.S.C. § 114(f)(1)(B),
it never enumerated a list of specific costs that such a fee
should cover. Starting in 2006, the Board has continuously
maintained the annual minimum fee at $500 for each channel,
including an aggregate cap of $50,000 per commercial
webcaster. Here, the Board accepted SoundExchange’s
proposal to raise the minimum fee to $1,000 per channel and
raise the aggregate cap to $100,000. In doing so, it reasonably
relied upon three evidentiary findings.

Most importantly, the Board was persuaded by
SoundExchange’s evidence demonstrating an increase in its
average administrative costs. SoundExchange calculated this
increased average “by dividing its total administrative costs by
 19
its total number of licensees” and dividing that amount “by the
estimated number of channels or stations per licensee.” Web V,
86 Fed. Reg. at 59,582. This calculation revealed that
SoundExchange’s estimated average administrative cost per
channel “increased from approximately $1,900 to
approximately $4,448 between 2013 and 2018, an increase of
2.34 times.” Id. While acknowledging that this was an
estimate, the Board concluded that the “relative increase in
average administrative costs”—134%—“would yield a
minimum fee of $1170[,]” and noted that this amount exceeded
SoundExchange’s proposed $1000 minimum fee. Id.

The other two points that the Board found supported
increasing the minimum fee included: (1) a settlement between
SoundExchange and College Broadcasters, Inc. (“CBI”),
agreeing to a minimum fee of $750 by 2025; and (2) inflation.
As relevant here, the Board concluded that the CBI settlement
generally indicated willing buyers and willing sellers would
agree to a higher minimum fee. And to the second point, the
record utilized the Bureau of Labor Statistics’ Consumer Price
Index for All Urban Consumers to show that a minimum fee of
$656.77 would be necessary to account for general inflation
since the minimum fee was set at $500 in 2006. Web V, 86
Fed. Reg. at 59,583. The Board considered this evidence in
finding that the record supported a “zone of reasonable
minimum fees” from $656.77 to $1,170. Id. Ultimately, the
Board adopted the proposed $1,000 minimum fee because it
was most persuaded by SoundExchange’s average
administrative cost evidence, and because the proposed fee fell
within the reasonable zone. Id. at 59,583–59,584.

On appeal, the Services again contest the Board’s
consideration of SoundExchange’s average administrative
costs, arguing that the minimum fee should be limited to the
 20
incremental cost of administering the webcasting license. We
disagree.

The Services point to nothing in either the statutory
scheme, or in the Board’s prior determinations, that requires
the Board to adopt such a restrictive view of 17 U.S.C.
§ 114(f)(1)(B). Indeed, the statute only instructs that the
statutory rates and terms “shall include a minimum fee for each
such type of service,” and that such a fee will reflect the willing
buyer/willing seller standard. Id. § 114(f)(1)(B); see
Intercollegiate II, 796 F.3d at 128 (“[T]he Board must set a fee
that both a willing buyer and a willing seller would negotiate,
not just one that is acceptable to the buyer (the webcaster).”)
(emphasis omitted). And as the Board explained, it has
consistently rejected the interpretations of the minimum fee as
limited to incremental administrative costs. Web V, 86 Fed.
Reg. at 59,581 (“To be sure, the Services have made that
[incremental administrative costs] argument consistently since
Web I. However, the Judges and their predecessors have never
embraced it.”).

Not only does the Services’ minimum fee challenge lack
support based on the statute, as well as from the Board’s prior
determinations, but the Services’ position also conflicts with
our precedent under which we have held that the minimum fee
may reflect average—as opposed to incremental—
administrative costs. Intercollegiate II, 796 F.3d at 131 (noting
that “the Board did not set the [minimum] fee based solely on
SoundExchange’s administrative costs” but also “relied on the
evidence of industry-wide average administrative cost”). In
Intercollegiate II, we concluded that while “[e]vidence of
average cost may not be perfect,” nothing prohibited us from
upholding its use. Id. (“This court’s task is ‘only [to] assess
the reasonableness of the [Board’s] interpretation of the
 21
inherent ambiguity’ in Congress’ directive.”) (quoting
Intercollegiate I, 574 F.3d at 757). So too here.

The Services have demonstrated no reason to bar the
Board’s consideration of SoundExchange’s increased average
administrative cost, and the Board’s determination comports
with our precedent as well as with the Board’s prior
determinations. Given the evidence of (1) SoundExchange’s
estimated increase in average administrative cost, (2) a
voluntary agreement to a higher minimum fee between
SoundExchange and CBI, and (3) general inflation since 2006,
we find the Board had substantial evidence to support its
decision that a $1,000 minimum fee reasonably satisfied the
willing buyer/willing seller standard. Thus, we uphold the
increased minimum fee.

B

The Committee is the arm of a trade association that
“represent[s] the interests of religious noncommercial radio
stations in issues of music licensing.” Committee Opening
Br. v. “Many of the [radio] stations represented by the
[Committee] simultaneously transmit their broadcast
programming online” under the terms of the statutory license
at issue in this appeal. Committee Opening Br. v. The
Committee challenges the Board’s rate determination for all
noncommercial webcasters, including the nonprofit religious
stations that are its members.

The Committee proposed two alternative rate structures
that would have lowered the rates paid by noncommercial
webcasters. Web V, 86 Fed. Reg. at 59,567. The Board,
however, rejected the Committee’s proposals and instead
accepted SoundExchange’s recommendation to essentially
maintain the incumbent rate structure that had been in effect
 22
from 2006 to 2020. Id. at 59,579. In support of its decision,
the Board noted that “SoundExchange relies on the same
reasoning adopted by the Judges in webcasting proceedings
going back to Web II to support its proposed rate structure.” Id.
at 59,573. The Board adopted that long-standing reasoning in
the absence of “persuasive counterarguments” from the
Committee. Id. at 59,573, 59,579.

Under the preexisting rate structure, a noncommercial
webcast station paid the minimum fee to gain access to 159,140
ATH of monthly usage. See Web IV, 81 Fed. Reg. at 26,405–
26,406. Above that usage threshold, noncommercial
webcasters paid the same rates as commercial nonsubscription
webcasters. See id. The Committee proposed alternative rates
that (1) would have maintained the same usage allowance and
minimum fee, while allowing noncommercial webcasters to
pay one-third of the commercial rate for above-threshold
usage; or (2) would have allowed certain Committee-
designated noncommercial webcasters to pay a lump sum for
an aggregate usage allowance. Web V, 86 Fed. Reg. at 59,567.2

2
The Committee’s proposed rate structures for noncommercial
webcasters were deemed “Alternative 1” and “Alternative 2.”
Web V, 86 Fed. Reg. at 59,567. Under Alternative 1, noncommercial
webcasters would pay an annual minimum fee of $500 that would
entitle them to 1,909,680 ATH of usage per year. For usage above
that threshold, noncommercial webcasters would pay one-third of
the rate for commercial webcasters for the same type of transmission
(subscription versus nonsubscription). Under Alternative 2, the
Committee would pay a flat annual fee of $1.2 million to
SoundExchange, for a group of up to 795 noncommercial religious
radio stations designated by the Committee. Those stations would
have an aggregate usage cap of 540 million ATH in 2021, increasing
by 15 million ATH per year. The proposal set no terms for usage
above that cap. Meanwhile, stations not designated for inclusion in
that group would be subject to the terms of Alternative 1. Id.
 23

The Committee’s rate proposals were based on a
settlement agreement between NPR, CPB, and
SoundExchange (the “NPR Agreement”) that covered the
years 2021 through 2025. The Committee sought to introduce
the NPR Agreement as a benchmark to support its rate
proposals for noncommercial webcasters. Web V, 86 Fed. Reg.
at 59,567–59,569. But the Board rejected the NPR Agreement
as a benchmark and rejected the Committee’s rate proposals.
The Board provided three primary reasons for its decision:
(1) the Committee neglected to offer any expert testimony to
establish that the NPR Agreement was “comparable” to a
compulsory license for noncommercial webcasters; (2) the
Committee’s rate proposals failed to make adjustments for
economically significant aspects of the NPR Agreement; and
(3) one aspect of the Committee’s proposal was based on dated
information that the Board was statutorily barred from
considering. See id. at 59,569–59,573.

On appeal, the Committee argues that the Board
(1) violated the APA by arbitrarily and capriciously rejecting
the NPR Agreement as a benchmark; (2) violated the APA by
arbitrarily and capriciously setting the noncommercial
webcaster royalty rate too high; and (3) violated the federal
Religious Freedom Restoration Act (“RFRA”), 42 U.S.C.
§ 2000bb et seq., and the First Amendment by setting a rate
that was less favorable to large, predominantly religious
webcasters than the rate enjoyed by secular NPR affiliates
under the NPR Agreement.3

3
The Committee also joined the NAB’s challenge to the Board’s
minimum fee, and essentially relied on the NAB’s briefing and oral
argument. Its only additional argument was that the Board should
have adjusted the $500 minimum fee for inflation from 2020 rather
 24

We sustain the Board’s rate determination for
noncommercial webcasters and its rejection of the
Committee’s proposals.

1

The Board’s decision to reject the NPR Agreement as a
benchmark, as well as the Committee’s rate proposals that were
based on the NPR Agreement, was reasonable and supported
by substantial evidence. We note that

appellants face[] an uphill battle in challenging the
Board’s selection of its benchmarks. We have
repeatedly recognized that it is “within the discretion
of the [Board] to assess evidence of an agreement’s
comparability and to decide whether to look to its
rates and terms for guidance.” The Board’s “broad
discretion” encompasses its selection or rejection of
benchmarks, as well as its adjustment of benchmarks
to “render them useful.”

SoundExchange, 904 F.3d at 50–51 (first quoting
Intercollegiate I, 574 F.3d at 759; then quoting Music Choice,
774 F.3d at 1009). Here, the Board properly exercised its
discretion.

than from 2006, because the $500 figure was last set in 2020. That
argument did not affect the Board’s ultimate determination of the
minimum fee. See Web V, 86 Fed. Reg. at 59,583 (using the
inflation-adjusted minimum fee as a lower bound of the “zone of
reasonable minimum fees” and relying primarily on
SoundExchange’s rising administrative costs to set a new minimum
fee). The NAB’s proposal to maintain the lower minimum fee is
addressed Part III.A, supra.
 25
First, the Board reasonably concluded that the Committee
presented insufficient evidence to establish the NPR
Agreement’s comparability to the compulsory license for
noncommercial webcasters. In a benchmark analysis, the
Board “may consider the rates and terms for comparable types
of audio transmission services and comparable circumstances
under voluntary license agreements.” 17 U.S.C.
§ 114(f)(1)(B)(ii) (emphasis added). Yet, the Committee
presented no expert testimony on comparability; it addressed
the issue only in arguments made by its attorneys in post-
hearing briefing. Web V, 86 Fed. Reg. at 59,570. Moreover,
SoundExchange disputed whether the NPR Agreement was
comparable. The Board reasonably decided that expert
testimony was necessary to establish comparability. See id.
Requiring expert testimony in this case was consistent with the
“highly technical nature” of administrative rate determinations.
SoundExchange, 904 F.3d at 50 (quoting Intercollegiate II, 796
F.3d at 127). Although the Committee argues that the Board
has never explicitly announced an expert-testimony
requirement, that does not render the Board’s decision in this
case arbitrary, particularly where the Board has previously
demanded expert testimony in an analogous situation. See Web
IV, 81 Fed. Reg. at 26,327 (rejecting lay testimony and
considering only expert testimony to determine whether to
adjust a benchmark).

Second, the Board reasonably rejected the Committee’s
rate proposals due to their failure to account for economically
significant aspects of the NPR Agreement. When a party
derives a proposed rate from a benchmark agreement, it is
required to account for economically significant, non-rate
aspects of the benchmark. See SoundExchange, 904 F.3d at 47.
For instance, if a benchmark contains a relatively low royalty
rate but imposes other costs on webcasters, the rate derived
from that benchmark should be adjusted upward to capture
 26
those costs. The Board properly placed the burden on the
Committee to make the appropriate adjustments to its rate
proposals derived from the NPR Agreement. See Music
Choice, 774 F.3d at 1009 (“While the Judges might have made
further adjustments to [a proponent’s] benchmarks to render
them useful, the Judges were not required to do so.” (citation
omitted)); see also id. at 1012 (“The Judges were under no
obligation to salvage benchmarks they found to have
fundamental problems.”). The Board faulted the Committee
for failing to account for the following aspects of the NPR
Agreement that benefited one or both of the settling parties but
were not reflected in the Committee’s proposed rates: (1) the
avoidance of litigation costs by the parties to the NPR
Agreement; (2) the value of NPR’s advance, lump-sum
payments to SoundExchange; and (3) NPR’s consolidated
reporting of data from individual stations to SoundExchange.
See Web V, 86 Fed. Reg. at 59,570, 59,572–59,573. The
Board’s determination that those adjustments were necessary
to adequately capture the value of the Agreement reflected
rational economic reasoning, even though the Board did not
determine the precise amount by which each of these factors
distorted the Agreement’s pricing.4 The Committee argues that

4
See Web V, 86 Fed. Reg. at 59,570 (“[S]ettlement agreements,
unlike voluntary agreements reached outside the context of
litigation, are not ‘free from trade-offs motivated by litigation cost,
as distinguished from the underlying economics of the transaction.’”
(quoting Determination of Royalty Rates and Terms for Making and
Distributing Phonorecords (Phonorecords III), 84 Fed. Reg. 1918,
1935 (Feb. 5, 2019))); id. at 59,572 (The NPR Agreement’s “rate
reflects * * * [a] discount that reflects the administrative
convenience to [SoundExchange] of receiving annual lump sum
payments that cover a large number of separate entities, as well as
the protection from bad debt that arises from being paid in
advance.”); id. at 59,573 (“The record reflects that consolidated
 27
the Board acted contrary to agency precedent and the statutory
scheme, but its citations are all inapposite.5

Third, the Board appropriately concluded that it was
statutorily barred from considering the royalty rates contained
in the “NPR Analysis,” an internal SoundExchange document
created in 2015. See Web V, 86 Fed. Reg. at 59,570–59,572.
In the NPR Analysis, SoundExchange performed calculations
using a royalty-rate structure that included per-performance
payments at one-third the commercial rate. The Board found,

reporting has value to SoundExchange. * * * ‘[O]ne of the things
that NPR does is it collects together the messy data of the individual
stations and reports it as part of the agreement.’” (quoting 8/17/20
Tr. 2232 (Professor Catherine Tucker))).
5
For example, the Committee misstates that Board precedent
required SoundExchange to make the necessary adjustments.
Although Web IV required the challenger of an “otherwise proper
and reasonable benchmark” to quantify further proposed
adjustments, Web IV, 81 Fed. Reg. at 26,387, that case is
distinguishable because the NPR Agreement was not an “otherwise
proper and reasonable” benchmark. Moreover, the Committee notes
that Web IV accepted a settlement-based benchmark without
litigation cost adjustments; but Web IV accepted that benchmark only
as “support for some elements of SoundExchange’s rate proposal”
and “not for the proposed rate for usage beyond the ATH threshold,”
id. at 26,394 (emphasis added), which is exactly what the Committee
attempted here. The Committee’s remaining citations, see
Committee Opening Br. 29–32, are similarly off point. See Web III,
79 Fed. Reg. at 23,123–23,124; Determination of Rates and Terms
for Preexisting Subscription Services and Satellite Digital Audio
Radio Services (SDARS II), 78 Fed. Reg. 23,054, 23,068–23,069
(April 17, 2013); Digital Millennium Copyright Act, Pub. L. No.
105-304, § 405, 112 Stat. 2860, 2895–2896 (1998); 17 U.S.C.
§ 801(b)(7)(A).
 28
however, that the rate structure came from an old settlement
agreement negotiated pursuant to the Webcaster Settlement
Act (“WSA”) of 2009, Pub. L. No. 111-36, 123 Stat. 1926. See
Web V, 86 Fed. Reg. at 59,572. The Board properly determined
that it was statutorily barred from considering that rate
structure under 17 U.S.C. § 114(f)(4)(C), which prohibits
“admi[tting] as evidence or otherwise tak[ing] into account”
the “provisions of any agreement entered into pursuant to [the
WSA], including any rate structure * * * set forth therein.” Id.
The Committee argued that the Board could rely on the rate
structure because the NPR Analysis was prepared for use in
future, non-WSA settlement negotiations. But the Board was
not required to accept the Committee’s inference that the rates
were actually used in the non-WSA settlement agreement
relied upon by the Committee. See Web V, 86 Fed. Reg. at
59,571. We disagree with the Committee’s contention that the
Board’s decision contradicted a binding opinion issued by the
Register of Copyrights, as set forth in Scope of the Copyright
Royalty Judges’ Continuing Jurisdiction, 80 Fed. Reg. 58,300
(Sept. 28, 2015). Although the Register’s opinion allows the
Board to consider voluntary license agreements that
incorporate WSA settlement terms, as well as the effect of the
WSA on private-settlement negotiations, it does not require or
even allow the Board to consider documents like the NPR
Analysis. See id. at 58,305. The Register found that, if parties
incorporate terms from their WSA settlements into subsequent
agreements, those are fair game for the Board’s consideration.
But the NPR analysis does not fall into that category; it
documents WSA rates that may have been used to propose
terms for a subsequent agreement. It does not document any
post-WSA terms. Without the NPR Analysis, the Committee
lacked support for a discounted above-threshold rate for
noncommercial webcasters, which became a critical flaw in its
proposal.
 29
2

We are unpersuaded by the Committee’s argument that the
Board arbitrarily and capriciously set the noncommercial
webcaster rate. In the absence of acceptable benchmarks, the
Board properly used the incumbent rate structure as the starting
point in its analysis. See Music Choice, 774 F.3d at 1012
(“[G]iven the lack of creditable benchmarks in the record, the
Judges did not err when they used the prevailing rate as the
starting point of their Section 801(b) analysis.”). Moreover,
substantial evidence supported the Board’s decision to
maintain the prevailing rate structure.

The incumbent rate structure originated in Web II. It
reflected an “economic insight” that noncommercial
webcasters that compete with the commercial market tend to
be larger, so that size is “a ‘proxy’ for determining when a
noncommercial webcaster poses a competitive threat[.]” Web
V, 86 Fed. Reg. at 59,565–59,566 (“[L]arger noncommercial
webcasters have the same or similar competitive impact in the
marketplace as similarly sized commercial webcasters.”).
Because large noncommercial webcasters can divert listeners
away from commercial webcasters, the Board found that
copyright holders would not willingly license sound recordings
to large noncommercial webcasters at a discount, because that
would decrease overall royalty revenue by cannibalizing
commercial royalty revenue. See Web II, 72 Fed. Reg. at
24,097–24,100. The noncommercial webcasters that exceed
the ATH threshold tend to be larger ones, and they are charged
the same rates as commercial webcasters for above-threshold
usage. In the instant rate-setting proceeding, the Board relied
on expert testimony to conclude that the same competitive
dynamics remained in effect and justified retaining the pre-
existing rate structure. See Web V, 86 Fed. Reg. at 59,575
 30
(discussing testimony of Mr. Jon Orszag, Professor Joseph
Cordes, and Professor Richard Steinberg).6

The Board appropriately rejected the Committee’s
attempts to undermine the analysis that justified the previous
rate structure. The Committee argued that cannibalization was
unlikely because noncommercial webcasters’ missions
differed from those of commercial webcasters. But the Board
reasonably declined to rely on the webcasters’ motivations in
evaluating market dynamics. See Web V, 86 Fed. Reg. at
59,575 (“The concerns about cannibalization that the Judges
articulated in past webcasting proceedings focus on potential
displacement in listenership from commercial to
noncommercial webcasters and is independent of
noncommercial webcasters’ motivations.”). Moreover, the
Board permissibly relied on an “overlap study” and other
evidence to reject the Committee’s argument that
noncommercial webcasting would not cannibalize commercial
webcasters because they each offer different programming.
The overlap study compared the songs played by commercial
and noncommercial Christian Adult Contemporary radio
stations. Id. at 59,576. It revealed that “commercial and
noncommercial stations broadcasting in the Christian [Adult
Contemporary] format play many of the same songs.” Id.7 The

6
The Committee’s argument that this expert testimony is mere “ipse
dixit” that cannot constitute substantial evidence to maintain the
incumbent rate, Committee Opening Br. 40, understates the Board’s
reasoned analysis of the expert testimony and overlooks our
precedent allowing the Board to start with the incumbent rate in the
absence of a suitable benchmark. See Web V, 86 Fed. Reg. at
59,575–59,578; Music Choice, 774 F.3d at 1012.
7
The Committee argues that the overlap study should not have been
admitted under 37 CFR § 351.10(e). As an initial matter, it is unclear
 31
Board also pointed to competition between two Atlanta-based
commercial and noncommercial religious radio stations. See
id. at 59,575.8 The Board permissibly inferred from that
evidence that noncommercial and commercial webcasters’
programming did not differ enough that cannibalization was
unlikely.

Finally, the Board reasonably rejected the Committee’s
argument that the incumbent rate structure failed to consider
noncommercial webcasters’ lower willingness to pay, thus
violating the Copyright Act’s willing buyer/willing seller
standard. According to the Committee, the Board overlooked
whether willing buyers would negotiate “above-threshold
commercial-level rates,” and accounted for only what willing
sellers would negotiate. Committee Opening Br. 40. That
argument, however, unduly focuses on the above-threshold
rates without considering the entire rate structure. The Board
noted that the minimum-fee rate structure gives

whether the Committee actually moved to exclude the study, or only
the testimony of SoundExchange’s witnesses about the study. See
Committee Mot. to Strike Test. Written Rebuttal Test. Related to
Mediabase Study. In any event, the Board reasonably concluded that
both the study and the witnesses’ testimony were admissible. See
Order Denying Committee Mot. to Strike 2–3 (noting that the study
simply compiled data “that industry participants rely on and that the
Judges have relied upon in past proceedings when presented by lay
witnesses,” and relying on the Board’s discretion to admit hearsay
testimony from witnesses about such data).
8
The Committee’s arguments against this evidence are meritless.
Contrary to the Committee’s suggestions, an expert witness need not
have “first-hand knowledge of the asserted facts,” Committee
Opening Br. 50, to testify about an issue, and the evidence fell well
within the proper scope of rebuttal. Moreover, the Board expressly
acknowledged the anecdotal nature of the evidence and treated it
with appropriate restraint. See Web V, 86 Fed. Reg. at 59,575.
 32
noncommercial webcasters of all sizes a hefty discount through
the monthly ATH allowance. See Web V, 86 Fed. Reg. at
59,566 (discussing testimony of Professor Catherine Tucker);
id. at 59,574 (discussing effective discount for noncommercial
webcasters). For instance, if the average copyrighted song is
four minutes long, see id. at 59,570–59,571, then 159,140 ATH
per month would let a noncommercial webcaster play roughly
28.5 million performances in a year for the $1,000 minimum
fee. That same number of performances would cost a
commercial subscription webcaster roughly $75,000 and a
commercial nonsubscription webcaster roughly $60,000.
Thus, the significant benefit conferred on noncommercial
webcasters by their access to the ATH allowance takes account
of their unwillingness to pay the same rates as commercial
webcasters. The Board justifiably concluded that the
incumbent rate structure adequately balanced noncommercial
webcasters’ lower willingness to pay with the risk of large
noncommercial webcasters cannibalizing copyright owners’
royalty revenue from commercial webcasters, thereby
satisfying the willing buyer/willing seller standard. See id. at
59,573–59,574.

3

The Committee argues that the Board’s rate determination
violates the First Amendment’s Free Exercise Clause and the
federal RFRA, because the terms that it adopts for
noncommercial religious webcasters are less favorable than the
terms enjoyed by NPR, a noncommercial secular webcaster.
We disagree.

The Free Exercise Clause provides that “Congress shall
make no law * * * prohibiting the free exercise” of religion.
U.S. CONST. Amend. I. Under that clause, a government entity
may not “burden[] [a person’s] sincere religious practice
 33
pursuant to a policy that is not ‘neutral’ or ‘generally
applicable’” unless the action is “narrowly tailored in pursuit
of” a “compelling state interest.” Kennedy v. Bremerton Sch.
Dist., 142 S. Ct. 2407, 2422 (2022) (quoting Employment Div.,
Dep’t of Hum. Res. of Ore. v. Smith, 494 U.S. 872, 881 (1990)).
“But the First Amendment is not the only potential refuge for
[a litigant’s] religious claim—the RFRA offers religious
exercise greater protection from intrusion by religion-neutral
federal laws.” Kaemmerling v. Lappin, 553 F.3d 669, 677
(D.C. Cir. 2008). Under the RFRA, federal government action
cannot “substantially burden a person’s exercise of religion
even if the burden results from a rule of general applicability,”
unless it is “the least restrictive means of furthering [a]
compelling governmental interest.” 42 U.S.C. §§ 2000bb-
1(a)–(b). The statute’s protection reaches “any exercise of
religion, whether or not compelled by, or central to, a system
of religious belief.” Id. § 2000cc-5(7)(A).

The Committee’s RFRA and Free Exercise arguments are
premised on a factual assertion that the rate for noncommercial
webcasters under the compulsory license is higher than the rate
enjoyed by NPR under the NPR Agreement. See Committee
Opening Br. 53–54. But there is no record finding to support
that assertion. As we have explained, the Board reasonably
rejected the NPR Agreement as a benchmark for
noncommercial webcasters and faulted the Committee’s
proposals based on the NPR Agreement for failing to make
necessary adjustments to account for economically significant
features of the Agreement. The Board also appropriately
refused to consider the NPR Analysis proffered by the
Committee, thereby eliminating the evidence that enabled the
Board to compare the above-threshold per-play royalty rates of
the compulsory license with the NPR Agreement’s lump-sum
payments. The record therefore contains no basis for the
Board, or this court, to effectively determine whether
 34
noncommercial webcasters subject to the compulsory license
are paying higher rates than the NPR stations covered by the
NPR Agreement. Without making that initial showing of
unfavorable treatment of religious webcasters, the Committee
cannot establish a violation of the RFRA or the First
Amendment.

We note that the Committee’s arguments are also
problematic in other respects. For example, the Committee
attempts to challenge the rates paid by the religious
broadcasters that are members of its trade association, but the
compulsory license applies to all noncommercial webcasters.
Even if the above-threshold noncommercial webcasters are
“almost exclusively religious,” as the Committee asserts,
Committee Opening Br. 8, the Committee does not explain
why it does not have to consider the overall rate structure,
which applies to all noncommercial webcasters. Indeed, the
Committee fails to cite any precedent that would give the
Board the power to impose a statutory license that, like the
Committee’s Alternative 2, would be available only to select
members of a particular trade organization, rather than to a
category of webcasters. See Web V, 86 Fed. Reg. at 59,567.
We are aware of none. Nor did the Committee argue to the
Board that the religious broadcasters it represents form a
distinct market segment for purposes of the 2021–2025
proceeding. We need not address that substantial and open
question, given the absence of any factual basis to compare the
rates at issue.

C

Finally, SoundExchange challenges as arbitrary and
capricious the royalty rate the Board set for commercial,
nonsubscription “ad-supported” webcasters, i.e.,
noninteractive streaming services like Free Pandora that collect
 35
payment from advertisers. Specifically, SoundExchange
argues that the Board acted arbitrarily and capriciously by
making an express finding that the opportunity cost for sellers
was a particular value, but then adopting a royalty rate that falls
below that opportunity cost without further explanation.

The very premise of SoundExchange’s argument is wrong.
The Board never found as fact the opportunity cost value on
which SoundExchange hangs its argument. As a result, its
arbitrary and capricious challenge fails before it even starts.

1

To understand SoundExchange’s argument and the
Board’s conclusions, some background on the Board’s
analytical process for setting royalty rates.

The Board’s statutory obligation is to “establish rates and
terms that most clearly represent the rates and terms that would
have been negotiated in the marketplace between a willing
buyer and a willing seller.” 17 U.S.C. § 114(f)(1)(B). The
Board views this hypothetical market to be one that is
“effectively competitive.” SoundExchange, 904 F.3d at 56.

To determine rates under that standard, the Board has
relied on various modes of economic analysis to best
approximate the price at which a willing seller would sell and
at which a willing buyer would buy. See Intercollegiate I, 574
F.3d at 757; Music Choice, 774 F.3d at 1010. One approach is
to consider “the rates and terms negotiated for comparable
services and ‘comparable circumstances under voluntary
license agreements.’” SoundExchange, 904 F.3d at 47
(quoting 17 U.S.C. § 114(f)(2)(B)). To that end, parties in a
proceeding may submit examples of voluntary license
agreements that they argue are sufficiently analogous to inform
 36
the appropriate statutory royalty rate for ad-supported, non-
interactive services. Because those proposed benchmarks
come from real-world markets that may not map perfectly onto
the Board’s hypothetical market, the parties commonly submit
expert economic analyses with their proposed benchmarks that
propose adjustments to account for any potential variance.

The Board then evaluates each license agreement’s
economic merits to determine which proposals, if any, could
be a useful touchstone in setting the statutory rate. In some
instances, the Board has accepted multiple proffered
benchmarks, and used the array to establish a “zone of
reasonableness” into which the final rate should fall. Web III,
79 Fed. Reg. at 23,110–23,115.

Another approach the Board uses is economic modeling
submitted by parties and their experts that are designed to
produce an appropriate rate. These models often employ game
theory to make and justify certain assumptions about the
hypothetical market, comb relevant data to calculate inputs,
and essentially solve for what parties contend is an appropriate
royalty rate. See, e.g., Web V, 86 Fed. Reg. at 59,522–59,546
(Evaluation of Game Theoretic Modelling Evidence); Johnson
v. Copyright Royalty Board, 969 F.3d 363, 372 (D.C. Cir.
2020) (discussing the Board’s use of specific game theoretic
model known as Shapley Analysis).9

9
Game theory “uses equations to model the behavior of decision-
makers whose choices affect one another.” Peter H. Huang,
Strategic Behavior and the Law, 36 JURIMETRICS J. 99, 100 (1995)
(quoting Rob Norton, A New Tool to Help Managers, FORTUNE, May
30, 1994, at 136). Most relevant here, game theory can help model
the likely, hypothetical behavior of both negotiating buyers and
sellers in an effectively competitive market.
 37
In assessing economic models and their outputs, the Board
evaluates a model’s utility as well as its flaws, and decides
whether a model can reasonably assist the Board’s calculation
of the statutory willing buyer/willing seller rate. See, e.g.,
Phonorecords III, 84 Fed. Reg. at 1947–1950.

2

In this case, the Board considered both benchmarking
analyses and economic models utilizing game-theory concepts.

With respect to benchmarking analyses, the Board
reviewed benchmarks submitted by SoundExchange, Pandora,
and Google, along with supplemental submissions by the
parties’ experts. See Web V, 86 Fed. Reg. at 59,491–59,522.
Making relevant adjustments, the Board found that both
SoundExchange’s and Pandora’s analyses yielded a royalty
rate of $0.0023 per play, while Google’s analysis yielded a rate
of $0.0021 per play. Id. at 59,589.

The Board also discussed the “Game Theoretic Modelling
Evidence” submitted by SoundExchange through its expert,
Professor Willig, and by the Services through their expert,
Professor Shapiro. As relevant here, the Board evaluated
Professor Willig’s “Shapley Value Model,” a game-theoretic
model that focused on “how to apportion among the members
of a multi-party bargaining group the surplus created by their
productive cooperation with each other.” Web V, 86 Fed. Reg.
at 59,522 (internal quotation marks omitted). That is, when the
parties get together and work out a deal, there is an independent
value derived from that agreement that is greater than the sum
of what every party could get on its own. The Willig model
sought to divide up this surplus. The Willig model also
assumed a small number of actors would get together to split
 38
up market share, presupposing a state of limited competition
otherwise known as an oligopoly.

One of the inputs that the Willig model used was the
“fallback value,” or the money a record company could make
with its repertoire through avenues beyond entering into a
voluntary licensing agreement with the streaming services.
Web V, 86 Fed. Reg. at 59,522. That fallback value was the
party’s “opportunity cost”: The party forgoes these alternative
money-making options when it enters into a specific licensing
agreement. “The opportunity cost of anything of value is what
you must give up to get it[.]” Id. at 59,522 n.220 (quoting JOHN
QUIGGIN, ECONOMICS IN TWO LESSONS: WHY MARKETS
WORK SO WELL, AND WHY THEY CAN FAIL SO BADLY 15
(2019)) (internal quotation marks omitted). Because
opportunity cost represents what a party gives up in taking a
particular option, that opportunity cost necessarily sets a floor
for the statutory rate under the willing buyer/willing seller
standard. That is because no willing buyer or seller would
agree to a rate below the cost borne for the choice made.

As part of its overall review of Professor Willig’s model,
the Board evaluated his opportunity cost figure, and found
salient Professor Shapiro’s criticism that the Willig model’s
opportunity cost input was artificially inflated. Web V, 86 Fed.
Reg. at 59,538–59,539. As a result, the Board adjusted the
Willig model estimate of the opportunity cost down, and
recalculated a proposed royalty rate using that new number.
Id.10

Even with those adjustments, however, the Board
ultimately found that Professor Willig’s model was fatally

10
The precise numbers are sealed as proprietary commercial
information.
 39
flawed because it baked in certain oligopoly power of the major
record labels. Oligopolies, of course, are not present in
effectively competitive markets. So a model premised on
oligopoly power by definition did not produce a royalty rate
reflective of what a willing buyer and seller would agree to in
an effectively competitive environment. As a result, the Board
concluded that the Willig model and the rate it produced could
only serve as a “limited guidepost” in determining a statutory
royalty rate. Web V, 86 Fed. Reg. at 59,540 (“[T]he evidentiary
record only allows the Judges to state with regard to the royalty
rates they have determined [from Professor Willig’s model]
that those 2021 rates * * * exceed an effectively competitive
rate by an indeterminate amount.”).

The Board likewise rejected Professor Shapiro’s game
theoretic modeling. The Board found that “new evidence”
Professor Shapiro relied on in his model was fundamentally
flawed in multiple ways and some key information was absent.
Web V, 86 Fed. Reg. at 59,540–59,546. Because the missing
evidence was the “sine qua non” of Professor Shapiro’s
modelling, its absence made Professor Shapiro’s models
“unusable.” Id. at 59,546.

Having found both game theoretic models to be
fundamentally unsound, the Board determined that the separate
benchmarking analyses submitted by the parties provided more
reliable evidence of an appropriate royalty rate on this record,
and that Google’s proposed benchmark in particular was the
most convincing. The Board emphasized that Google’s
benchmark provided “more granular, [record] label-specific,
analysis,” as well as a more reliable application of adjustments
to account for known concerns, all of which ultimately lent
more weight to Google’s proposed rate. Web V, 86 Fed. Reg.
at 59,589.
 40
On that basis, the Board set the commercial ad-supported,
noninteractive royalty rate at Google’s proposed adjusted
benchmark of $0.0021 per play. Web V, 86 Fed. Reg. at
59,589. In so finding, the Board noted that its chosen royalty
rate was only slightly below that produced by Professor
Willig’s adjusted model, which served as a relevant, though
quite limited, guidepost. That further buttressed the Board’s
judgment that its final royalty rate accurately reflected the
willing buyer/willing seller standard in a hypothetical,
effectively competitive market.

3

SoundExchange’s entire argument is premised on the
contention that the Board necessarily erred because it
specifically adopted the Willig model’s adjusted opportunity
cost for the ad-supported, noninteractive market, and yet set a
royalty rate that fell below that amount.

But the factual premise on which SoundExchange’s
argument rests is no fact at all. The Board never made a
definitive finding that the true opportunity cost was the
adjusted Willig value. In fact, the Board ultimately rejected
the Willig model, and the game theoretic models more
generally, for use in calculating the appropriate royalty rate,
opportunity cost and all.

Remember, the Board rejected Professor Willig’s
opportunity cost figure as inflated. The Board then noted
Professor Shapiro’s proposed downward adjustment of the
Willig model’s number. The Board took that into account and
concluded that, “[b]ased on the foregoing adjustments accepted
by the Judges, Professor Willig’s opportunity cost calculation
must be adjusted, as set forth [below in Figure 8].” Web V, 86
Fed. Reg. at 59,538.
 41

SoundExchange argues that this one sentence renders the
Board’s entire rate-setting arbitrary and capricious. But all the
Board said was that it accepted Professor Shapiro’s specific
adjustments to Professor Willig’s calculations in the broader
context of Professor Willig’s game theoretic model. The Board
did not go further and make a definitive determination that the
adjusted Willig model number was, in fact, the actual
opportunity cost for sellers that would govern the entire rate-
setting procedure. Especially not since the Board ultimately
abandoned altogether the use of economic models like
Professors Shapiro’s and Willig’s as a reliable basis for
calculating the royalty rate.

On top of that, the statement on which SoundExchange
relies only addressed one of the Board’s multiple critiques of
the Willig model’s initial opportunity cost calculation. The
Board, for instance, repeatedly objected to Professor Willig’s
failure to factor opportunity benefits into his opportunity cost
calculation. See Web V, 86 Fed. Reg. at 59,523, 59,537.

Notably, SoundExchange lodges no criticism of the
Board’s decision to use, instead, the benchmarking process to
set the royalty rate. That benchmarking process itself
accounted for opportunity cost, as the voluntary agreements
naturally involve a party’s own estimation of its opportunity
cost. And the Board considered that opportunity cost where
relevant in adjusting the benchmark proffered. See Web V, 86
Fed. Reg. at 59,496–59,497.

In sum, because the Board never found as fact that the
opportunity cost input to Professor Willig’s model, even as
adjusted, represented the record companies’ true opportunity
cost, the Board’s decision to set a royalty rate that was slightly
below the Willig model’s flawed opportunity-cost measure is
 42
neither here nor there. And so this court has no occasion to
decide, as SoundExchange urges, whether it would, as a matter
of law, violate the willing buyer/willing seller standard for the
Board to set a royalty rate below some definitive measure of
opportunity cost.

IV

For the foregoing reasons, we affirm the Final
Determination of the Copyright Royalty Board.

So ordered.