On Appeal from the United States District Court for the Southern District of New York

ARGUED: MAY 20, 2024 DECIDED: FEBRUARY 24, 2026

Before: WESLEY and MENASHI, Circuit Judges . †

Page 2Broadcast Music, Inc. ('BMI') licenses musical works in exchange for a fee. Because of BMI’s large share of the music licensing market, it is subject to an antitrust consent decree that requires BMI to set reasonable licensing fees and authorizes a court to determine a reasonable fee if BMI and the prospective licensee are unable to agree. North American Concert Promoters Association ('NACPA') has historically purchased blanket licenses covering all the musical works in BMI’s repertory so that artists may perform the works at concerts. During the last negotiation cycle, BMI and NACPA were unable to agree on a rate and revenue base, and for the first time in the history of the parties’ relationship, BMI petitioned the district court to resolve the impasse. In 2023, the district court accepted BMI’s rate quote for the 2014-2018 period and set a rate of 0.5 percent of NACPA’s gross revenues for the 2018-2022 period. The district court also expanded the definition of 'gross revenues' to cover items not previously included in the revenue base. NACPA appealed both decisions, and BMI cross-appealed the denial of its motion for prejudgment interest.

We conclude that the district court imposed unreasonable rates. First, the district court adopted a definition of the revenue base that had no precedent in the history of the industry without a compelling reason. The definition included revenue streams that do not reflect the fair market value of the music and that involve significant administrative costs without a corresponding benefit. Second, the district court set a rate that depends disproportionately on less comparable benchmark agreements. The district court relied on agreements with unaffiliated individual promoters even though NACPA has historically obtained significantly lower rates from the Page 3same counterparties. And it identified no change in economic circumstances that would justify a rate more than double what NACPA has historically paid to BMI and to the American Society of Composers, Authors, and Publishers. Third, while we conclude that the district court did not abuse its discretion by denying the motion for prejudgment interest, the denial may have depended on the erroneous rates. Accordingly, we vacate the judgment of the district court and remand for further proceedings consistent with this opinion.

SCOTT A. EDELMAN (Atara Miller, Andrew L. Porter, on the brief ), Milbank LLP, New York, New York, for Petitioner-Appellee-Cross-Appellant.

ANDREW GASS, Latham & Watkins LLP, San Francisco, California (Joseph R. Wetzel, Latham & Watkins LLP, San Francisco, California; Jennifer L. Giordano, Sarang V. Damle, Blake E. Stafford, Latham & Watkins LLP, Washington, D.C.; Benjamin E. Marks, Sarah Sternlieb, David Yolkut, Weil, Gotshal & Manges LLP, New York, New York; Samir Deger-Sen, Latham & Watkins LLP, New York, New York, on the brief ), for RespondentAppellant-Cross-Appellee .

Frank P. Scibilia, Donald S. Zakarin, Katie E. Garber, Pryor Cashman LLP, New York, New York, for Amicus Curiae National Music Publishers' Association.

Jay Cohen, Hallie S. Goldblatt, Paul, Weiss, Rifkind, Wharton & Garrison LLP, New York, New York; Clara Kim, Richard H. Reimer, Jackson Wagener, American Society of Composers, Authors and Publishers, New Page 4Anne M. Voigts, King & Spalding LLP, Palo Alto, California; David P. Mattern, King & Spalding LLP, Washington, D.C.; Kenneth Steinthal, King & Spalding LLP, San Francisco, California, for Amici Curiae Motion Picture Association, Inc., Radio Music License Committee, National Association of Broadcasters, Digital Media Association, Exhibitions & Conferences Alliance, and International Association of Venue Managers .

MENASHI, Circuit Judge :

This case involves the licensing of copyrighted musical works to concert promoters for live performance. Petitioner-Appellee-CrossAppellant Broadcast Music, Inc. (“BMI”) represents over one million songwriters, composers, and music publishers, and it licenses their works to music users in exchange for a fee. Because of BMI’s large share of the music licensing market, it is subject to an antitrust consent decree that requires BMI to set reasonable licensing fees and authorizes a court to determine a reasonable fee if BMI and the prospective licensee are unable to agree. Respondent-AppellantCross-Appellee North American Concert Promoters Association (“NACPA”) has historically purchased blanket licenses covering all the musical works in BMI’s repertory so that artists may perform the works at concerts.

During the last negotiation cycle, BMI and NACPA were unable to agree on a rate and revenue base, and for the first time in the history of the parties' relationship, BMI petitioned the district court to resolve the impasse. In 2023, the district court accepted BMI’s Page 5rate quote for the 2014-2018 period (the “Retroactive Period”) and set a rate of 0.5 percent of NACPA's gross revenues for the 2018-2022 period (the “Current Period”). The district court also expanded the definition of “gross revenues” to cover items not previously included in the revenue base. NACPA appealed both decisions, and BMI crossappealed the denial of its motion for prejudgment interest.

We conclude that the district court imposed unreasonable rates. First, the district court adopted a definition of the revenue base that had no precedent in the history of the industry without a compelling reason. The definition included revenue streams that do not reflect the fair market value of the music and that involve significant administrative costs without a corresponding benefit. Second, the district court set a rate that depends disproportionately on less comparable benchmark agreements. The district court relied on agreements with unaffiliated individual promoters even though NACPA has historically obtained significantly lower rates from the same counterparties. And the district court identified no change in economic circumstances that would justify a rate more than double what NACPA has historically paid to BMI and to the American Society of Composers, Authors, and Publishers ('ASCAP'). Third, while we conclude that the district court did not abuse its discretion by denying the motion for prejudgment interest, the denial may have depended on the erroneous rates. Accordingly, we vacate the judgment of the district court and remand for further proceedings consistent with this opinion.

The owner of the copyright in a musical work has the exclusive right to perform the work publicly. See 17 U.S.C. § 106(4). The copyright owner may authorize another to perform the work by Page 6granting a license. See id. § 201(d)(2). In the United States, the owner of a musical work confers on a Performing Rights Organization (“PRO”) the right to license performances of the work, and the PRO then issues licenses, collects fees, and distributes royalties to the copyright owner. See BMI v. Columbia Broad. Sys., Inc. , 441 U.S. 1, 4-5 (1979). The purchasers of performance licenses include 'concert promoters, commercial radio stations, broadcast television stations, broadcast and cable television networks, internet websites, commercial music services, concert halls, universities, airlines, restaurants, bars and nightclubs.” Agreed Findings of Fact ¶ 2, BMI v. NACPA , No. 18-CV-8749 (S.D.N.Y. Oct. 10, 2022), ECF No. 124-1. The PRO typically issues a blanket license that allows the licensee to perform any of the songs in the PRO’s repertory.

Four PROs control the market for music performance licenses: BMI; ASCAP; the Society of European Stage Authors and Composers ('SESAC'); and Global Music Rights ('GMR'). The district court estimated that BMI, ASCAP, SESAC, and GMR hold repertories that account for about 45.4 percent, 46.5 percent, 3.6 percent, and 4.5 percent of the total market, respectively. See BMI v. NACPA , 664 F. Supp. 3d 470, 474 (S.D.N.Y. 2023).

Because of the dominant shares of the license market that the entities control, BMI and ASCAP operate under antitrust consent decrees. See Final Judgment, United States v. BMI , No. 64-CV-3787 (S.D.N.Y. Nov. 18, 1994) (“BMI Consent Decree”); Second Amended Final Judgment, United States v. ASCAP , No. 41-CV-1395 (S.D.N.Y. June 11, 2001) ('ASCAP Consent Decree').¹ The BMI Consent Decree Page 7requires BMI “to offer any prospective music user a license to publicly perform the songs in its repertoire.” BMI , 664 F. Supp. 3d at 474.² If BMI and a prospective licensee cannot agree on a rate, then either party may petition the district court to determine a “reasonable” rate. BMI Consent Decree § XIV(A).

NACPA, formed in 1998, is an association of concert promoters. “To qualify for NACPA membership, a promoter must stage at least 60 shows in a year, including at least 5 club shows, 5 theater shows and 5 shows in venues with more than 10,000 seats.” Agreed Findings of Fact ¶ 11. In addition to “execut[ing] every aspect of producing [an] artist’s concert,” promoters are responsible for “negotiat[ing] with an artist or their representative for their live performance rights, either Page 8for an individual engagement or a series of engagements or a tour.'³ According to NACPA, “because concert promoters typically do not know all the songs the artist will perform, or which PRO (or PROs) administers the license for each of those songs, promoters must in practice secure blanket licenses from each of the four principal domestic PROs … to avoid the risk of copyright infringement.' Appellant’s Br. 13.

The structure of the concert promotion industry has changed in the past few decades. Prior to the mid-1990s, the industry consisted primarily of small, independent promoters, each of which 'exclusively function[ed] within [its] own market.'⁴ Today, two parent companies—Live Nation Entertainment, Inc. (“Live Nation”) and AEG Presents, LLC (“AEG”)—control approximately 75 percent of NACPA’s members. Live Nation, AEG, and affiliated entities pay about 90 percent of the fees under NACPA’s license with BMI. The business of Live Nation and AEG includes not only concert promotion but also ticket servicing and venue ownership. Live Nation and AEG own the ticket servicing companies Ticketmaster and AXS, respectively. Ticketmaster and AXS charge service fees on most concert tickets sold; the fees typically amount to about 20 percent of the face value of the ticket.

BMI has issued blanket licenses to NACPA since NACPA was founded. The first license, issued in 1998, required NACPA to pay a Page 9fee to BMI equal to 0.3 percent of 'Gross Ticket Revenues' for concerts in venues with under 10,000 seats and 0.15 percent for concerts in venues with over 10,000 seats. App'x 543. The license defined 'Gross Ticket Revenues' as 'the total monies received, directly or indirectly, by [NACPA’s] Licensed Individual Member(s) or their authorized representatives from all ticket sales per attraction,” with exclusions for taxes, venue fees, ticket servicing fees, and parking fees. Id . at 540. The license also gave NACPA a 10 percent administrative discount provided that at least 80 percent of its members signed on to the license. See id. at 539. “Blending the 0.15% and 0.30% rates on a weighted basis based on fees paid to BMI, NACPA paid an effective rate of 0.21% under the BMI historical licenses, before the application of the administrative discount.' Agreed Findings of Fact ¶ 72. BMI and NACPA entered into a new license in 2006 on the same terms as the 1998 license, except that the 2006 license added a separate rate for music festivals. According to BMI, festivals were not covered by the 1998 license. The 2006 license was renewed several times before BMI provided notice in August 2013 that it would terminate effective December 31, 2013. “Since then, the parties have been on an interim agreement and have engaged in protracted negotiations for a new license without success.' BMI , 664 F. Supp. 3d at 476.

BMI’s current proposal, which it quoted to NACPA in an email dated December 9, 2020, sets different terms for the Retroactive Period and the Current Period. For the Retroactive Period, BMI proposes keeping the historical revenue base—face value of tickets sold—and charging a varying percentage fee based on the size of the venue, as follows:

Page 10Venue Size Rate 0 to 2,500 seats 0.8% 2,501 to 3,500 seats 0.6% 3,501 to 5,000 seats 0.4% 5,001 to 9,999 seats 0.3% 10,000 seats or more 0.15%

App'x 438. For the Current Period, BMI proposes a unitary rate of 0.8 percent of "Gross Revenues," which is defined more broadly than Was "Gross Ticket Revenues" in the previous licenses. The proposed definition of "Gross Revenues" includes:

(1) Revenues from primary ticket sales paid or payable to the licensee or a contractually-related third party; (2) Revenues from service, administration, ,and/or handling charges paid or payable to the licensee in connection with primary ticket sales; (3) Revenues from the sale of VIP packages and box suites for concerts paid or payable to the licensee or a contractually-related third party; (4) Revenues from sponsorships and other forms of advertising paid or payable to the licensee; and (5) Revenues from direct-to-secondary market ticket sales paid or payable to the licensee, meaning tickets whose initial distribution to the public is on the secondary market.

BMI, 664 F. Supp. 3d at 476; see App'x 437. BMI's proposal would also eliminate the 10 percent administrative discount for both the Retroactive Period and the Current Period. See App'x 438.

NACPA's counterproposal would continue the same terms as the 2006 license, including the historical revenue base and the Page 11administrative discount. NACPA argued that the rate should be converted from a tiered structure to a simple unitary rate of 0.21 to 0.275 percent of gross ticket revenues. See BMI , 664 F. Supp. 3d at 477.

When the parties failed to reach agreement, BMI petitioned the district court to determine a reasonable rate. Pursuant to a 2014 agreement, during the pendency of the negotiations and trial, NACPA paid BMI an interim rate of 0.3 percent of the face value of concert tickets for venues with fewer than 10,000 seats and 0.15 percent for venues with more than 10,000 seats.

The district court issued its opinion on March 28, 2023, following a five-week trial. The district court first addressed the Current Period. It explained that BMI had “attempt[ed] to justify its proposed rate through the use of sixteen benchmarks which can be put into three groups': (1) licenses issued by BMI or ASCAP to NACPA or non-NACPA promoters; (2) licenses issued by SESAC and GMR to NACPA, non-NACPA promoters, or individual NACPA members (Live Nation and AEG); and (3) licenses issued by foreign PROs. Id. at 479.

To facilitate a comparison between the benchmark agreements and BMI's proposal, BMI's economic expert, Professor Catherine Tucker, applied several adjustments to the rate of each benchmark agreement. First, Tucker converted all of the benchmark rates into unitary rates: “[W]hile BMI’s proposed license stipulates a single rate of 0.8 percent for all live concerts and festivals, certain benchmark licenses contain a fixed fee per ticket, tiered rates based on venue capacity, tiered rates based on event revenue, and/or different license Page 12rates for festivals and concerts.'⁵ 'Therefore, to make these different rate structures comparable to BMI’s single proposed rate for all live concerts and festivals,” Tucker “convert[ed] all fixed fee, tiered, or event-specific rates into a single, percentage-of-revenue rate.'⁶ She did so “by weighting … each tier’s or event type’s unique rate by its share of industry revenue and then calculating the weighted average across these rates.'⁷

Second, because “the proposed license covers songs in BMI’s repertoire,” Tucker “convert[ed] the rates that promoters negotiated with other PROs to reflect differences in the monetized use of BMI’s repertoire in live concerts relative to the monetized use of benchmark PROs' repertoires.'⁸ Tucker reasoned that 'if a promoter were willing to pay $5 for access to ASCAP’s repertoire for a concert that brought in $100 in revenue, and if the same concert used twice as many songs in BMI's repertoire, other things equal, that promoter should be willing to pay BMI $10.'⁹

Third, because BMI’s proposed revenue base is broader than the revenue base in the benchmark agreements-which uniformly define gross revenue as the face value of tickets sold-Tucker adjusted each rate downward to reflect royalty payments as a percentage of the total face value of tickets sold, revenues from tickets sold directly into the secondary market, fees above the face value of Page 13the ticket paid by the consumer, box suite and VIP package revenues, and sponsorship revenues.

Thus, for each benchmark agreement, Tucker calculated two rates. The first rate, which the district court called the "primary rate," reflects the first two adjustments: It is the rate for each benchmark agreement, converted to a unitary rate and multiplied by the ratio of the relevant PRO's market share to BMI's. The second rate, which the se a rd st , p, t ps o ps downward to account for the larger revenue base that BMI proposed. d e s s for each of the sixteen benchmark agreements that BMI identified as well as the 2006 BMI-NACPA agreement:

Period Primary Implied License Covered Rate Rate GMR-Live 2015-2016 0.63% 0.54% Nation GMR-AEG 2015-2016 0.63% 0.54% GMR-APE 2016-2018 0.63% 0.54% GMR-JAM 2015-2018 0.63% 0.54% GMR-MEMI 2016-2018 0.63% 0.54% GMR2015-2016 0.63% 0.54% Nederlander SESAC2019-2024 0.40% 0.34% NACPA SESAC-non2021 0.59% 0.51% NACPA

Page 14BMI2006 0.21%10 0.21%11 NACPA BMI-non2017 0.42% 0.36% Regulated NACPA Domestic PROs ASCAP2020-2021 0.27% 0.23% NACPA ASCAP2021 0.43% 0.37% non-NACPA SOCAN 2015-2017 1.33% 1.15% June 2020PRS 1.81% 2.12% present Foreign PROs IMRO 2021 2.13% 1.90% APRA AU November 2020 0.94% 0.81% APRA NZ December 2021 0.86% 0.74%

Page 15With respect to the revenue base, the district court held that “sponsorship and advertising contributions must be excluded because they do not reflect the music or its performance.' BMI , 664 F. Supp. 3d . at 479. But the district court included the other four revenue categories that BMI proposed. First, the district court concluded that “[d]efining the revenue base to include tickets sold for the first time in a secondary market is a natural clarification of what some promoters think the base already subsumes.' Id. at 484. Second, the district court concluded that box suite and VIP package revenues should be included because, although those sales 'may include charges in addition to the seat ticket, like for food or merchandise,” the purchaser still “must pay the total price if she wishes to sit in these premium seats and hear the music.” Id . Third, the district court held that ticketing service fees should be included because the consumer pays those fees when she purchases a ticket and “[t]here is virtually no way to opt out of paying the fees.” Id .

The district court acknowledged NACPA's concern that “calculating the [revenue base] in this manner is unworkable because promoters do not have access to all the revenue information, specifically revenues for boxes and VIP packages.' Id. at 485. The district court believed, however, that “[l]imiting the revenues to those received by the promoters or a contractually related third-party helps to alleviate those concerns, even if it does not produce perfectly efficient administration.” Id.

To determine the reasonable rate to apply to this expanded revenue base, the district court first concluded that BMI’s proposed rate of 0.8 percent was unreasonable. Cf. BMI Consent Decree § XIV(A) (“[D]efendant shall have the burden of proof to establish the reasonableness of the fee requested by it.”). The district court noted that “[w]hen looking at the secondary implied rates, only the foreign Page 16licensing agreements have rates above or close to BMI’s proposed one of 0.8%.” BMI , 664 F. Supp. 3d at 483. For that reason, “the reasonableness of BMI’s proposed rate of 0.8% rests on the foreign licensing agreements.' Id. The district court decided that 'these foreign licenses cannot be relied upon as valid benchmarks because BMI failed to show that the foreign parties are similar to BMI and NACPA and that the agreements were negotiated in similar economic circumstances.” Id.

Instead, the district court concluded that '[a]ll the BMI and ASCAP licenses are proper benchmarks” regardless of whether those licenses were negotiated with NACPA or with non-NACPA promoters. Id. at 486. The district court observed that the licenses issued to non-NACPA promoters “arose in a market that reflects the same degree of competition and economic circumstances, and they cover the same rights.” Id . Moreover, “[a]lthough NACPA and nonNACPA promoters differ in the scope of their business operations, they are direct competitors for shows hosted in smaller venues.” Id.

The district court also concluded that “[t]he SESAC and GMR licenses are appropriate benchmarks” because “[c]ompared to [the] proposed license, the SESAC and GMR licenses are between similar parties, for similar rights, and were negotiated in similar economic circumstances.” Id. at 487. The district court explained that, because SESAC and GMR are not subject to a consent decree that prevents those entities from refusing a license to any music user, their negotiations 'approximate[] the dynamics of a direct licensing negotiation between a music user and the individual music publisher.' Id . The district court determined that “the fact that SESAC and GMR operate free of a consent decree does not mean they are extracting supercompetitive rates.” Id. at 488.

Page 17Based on this analysis, the district court considered eleven benchmarks, with the “range of the implied rates accounting for the expansion in the gross revenue base includ[ing] rates of 0.21%, 0.23%, 0.34%, 0.36%, 0.37%, 0.51%, and 0.54%.' Id . Given the benchmark rates, the district court determined that '[a] rate of 0.5% of the expanded gross revenue base is … a reasonable one.” Id. The district court decided that “[i]ncreasing the rate of a license for live concerts better reflects the fair market value placed on licenses in music intensive industries,” such as “commercial radio stations or virtual live concert streaming services.' Id . The district court additionally decided that BMI could not eliminate the 10 percent administrative discount because NACPA had already performed the administrative services that the discount was intended to remunerate. Id. at 489.

The district court adopted BMI’s proposal for the Retroactive Period. The district court explained that this “quotation would align the NACPA rates with the rates paid to BMI by non-NACPA promoters during that time.' Id. “Having already found that the BMInonNACPA benchmark is valid,' the district court concluded that BMI’s proposal was reasonable. Id. At the same time, it disallowed BMI’s proposal to eliminate the 10 percent administrative discount for the Retroactive Period. Id. 12

Following the entry of judgment, BMI moved for an award of pre- and post-judgment interest. The district court granted the motion for post-judgment interest but denied the motion for prejudgment interest. The district court reasoned that the interim fee rate was “correct for the time it [was] in effect” and for that reason BMI had Page 18not been wrongfully deprived of the time value of money. BMI v. NACPA , 674 F. Supp. 3d 70, 73 (S.D.N.Y. 2023). The district court clarified that its “determination of the final fee amount was designed to be the entire embodiment of what constitutes a reasonable fee.” Id . This appeal followed.

In its appeal, NACPA argues that the rates the district court adopted for the Current Period and the Retroactive Period are unreasonably high. NACPA also contends that the district court erred by expanding the definition of the revenue base for the Current Period. In BMI's cross-appeal, BMI argues that the district court abused its discretion by denying its motion for prejudgment interest. NACPA maintains that it did not.

We agree with NACPA that the district court erred by expanding the definition of the revenue base and set unreasonable rates. While we conclude that the district court did not abuse its discretion by denying the motion for prejudgment interest, the denial may have depended on the rates the district court set. As a result, we vacate the judgment of the district court and remand for further proceedings consistent with this opinion.

“We review the rate set by the District Court for reasonableness.” United States v. BMI ( Music Choice IV ), 426 F.3d 91, 96 (2d Cir. 2005). This standard requires us to ask (1) whether the rate is substantively reasonable such that it is not 'based on any clearly erroneous findings of fact,” and (2) whether the rate is procedurally reasonable such that “the setting of the rate, including the choice and adjustment of a benchmark,” is not “based on legal errors.” Id.

The parties disagree about how we should evaluate the decisions of the district court to rely on the SESAC and GMR licenses as benchmarks, to expand the revenue base for the Current Period, and to depart from the rates previously negotiated by NACPA with BMI and ASCAP. NACPA argues that such decisions are subject to de novo review, and BMI maintains that we should review for clear error.

According to NACPA, “this [c]ourt has applied de novo review to every one of the determinations challenged by NACPA in this case,' including the 'selection of benchmarks,' the 'weighing of benchmarks,' and the 'revenue base.' Appellant's Reply Br. 6. In United States v. ASCAP , we said that “[d]eterminations by the district court that particular benchmarks are comparable and particular factors are relevant are questions of law reviewed de novo ' but the 'factual findings as to each factor under consideration or those underlying a proposed benchmark agreement, as well as findings with respect to fair market value, are reviewed for clear error.' 627 F.3d 64, 76 (2d Cir. 2010). In United States v. BMI ( Music Choice II ), we said that the district court 'made a fundamental error' in concluding that “retail price was not a good indicator of fair market value” and should not be used as the revenue base. 316 F.3d 189, 195 (2d Cir. 2003). And in ASCAP v. Showtime/The Movie Channel, Inc. ( Showtime II ), we said that “whether the price paid by a buyer may be given less weight as a ‘comparable sale’ either because the price was higher than would have obtained in a more competitive market or was perceived by the buyer to be higher is a matter of law.” 912 F.2d 563, 569-70 (2d Cir. 1990).

In response to these authorities, BMI argues that the district court made 'factual determinations' when it decided to 'adopt[]Page 20 SESAC and GMR licenses as benchmarks,” to “order[] an expansion of the revenue base to include the full retail cost of attending a concert,” and to “set[] a rate at the high end of the benchmark range that does not mirror the ASCAP/NACPA (and historical BMI/NACPA) rate.' Appellee’s Br. 41. BMI observes that in Showtime II we said that “[f]air market value is a factual matter, albeit a hypothetical one,' and that the selection of an appropriate benchmark is “a determination analogous to an evidentiary ruling that would have occurred if the price of the [relevant] license had been offered at a jury trial.” Showtime II , 912 F.2d at 569-71. According to BMI, “NACPA does not identify the legal errors” in the opinion of the district court “because there are none.” Appellee’s Br. 41. The district court instead 'applied the correct four-factor legal framework articulated in Music Choice IV .” Id.

We conclude that we properly review these issues de novo . In Showtime II , we analogized the determination of fair market value to an evidentiary ruling but explained that we would nevertheless accord plenary review to the legal issues involved in the determination:

[T]he factual component of the issue before the [district judge] does not render all aspects of his decision-making subject to review under the “clearly erroneous” standard. In making a factual determination, a decisionmaker might rely on legally impermissible factors, fail to give consideration to legally relevant factors, apply incorrect legal standards, or misapply correct legal standards. In jury trials, such matters are normally resolved by rulings on admissibility of evidence and by jury instructions. In court trials, where the functions of fact-finding and exposition of law are performed by the same person, the line between the functions is not always

Page 21distinct. For example, the line between admissibility of evidence (law) and evaluation of the persuasive force of evidence (fact) is often blurred. Nevertheless, an appellate court is obliged to observe the law/fact distinction as best it can and accord plenary review to any aspect of a trial court's decision that can fairly be isolated as determining an issue of law.

912 F.2d at 569. We have articulated legal standards that govern the selection and weighing of benchmarks and the definition of an appropriate revenue base.¹³ The arguments that NACPA raises on appeal concern whether the district court “fail[ed] to give consideration to legally relevant factors” and “misappl[ied] correct legal standards' when it adopted the SESAC and GMR licenses as benchmarks, expanded the revenue base, and set a rate at the high end of the benchmark range. Showtime II , 912 F.2d at 569. Because these decisions “can fairly be isolated as determining an issue of law,” we undertake a plenary review. Id.

If BMI does not meet its burden of proof on the reasonableness of its requested rate, the district court must “determine a reasonable fee based upon all the evidence.' Music Choice II , 316 F.3d at 194 (quoting BMI Consent Decree § XIV(A)). It must make that Page 22determination by 'establishing the fair market value of the music rights, in other words, ‘the price that a willing buyer and a willing seller would agree to in an arm's length transaction.'' Music Choice IV , 426 F.3d at 95 (quoting Music Choice II , 316 F.3d at 194).

In attempting to apply that precedent, district courts have lamented “the problematic nature of the analysis envisioned and the enigmatic task thrust upon a court that must undertake this evaluation.” United States v. ASCAP ( ABC/CBS ), 831 F. Supp. 137, 14344 (S.D.N.Y. 1993). As one district court has explained:

[T]he market for blanket licenses appears to be one whose natural consequence is the lack of broad-based competition. To postulate what prices would prevail were such a market 'competitive' is perplexing in theory, impractical in practice, and dubious in outcome given the efficiencies obtained due to the aggregating nature of the service rendered. Little can be adduced as to the expected behavior of such a market were it populated by multiple sellers. It is questionable whether such a market could sustain many sellers; whether the market would function efficiently and the price levels at which its supply and demand would converge are even more uncertain. In addition, limited evidence is discernible as to the [marginal] costs [of issuing blanket licenses] facing [the PROs]. Consequently, any ratesetting standard that calls, in the abstract, for a theoretic construct of a competitive market in blanket licensing must confront the reality that there exists minimal evidence as to what that market would look like, much less the prices it would yield.

Id. at 144 (footnotes omitted). Scholars have similarly argued that, from an economic perspective, there is no 'indisputably correct solution” to the problem of determining the fair market value of a Page 23blanket license because “[i]n the noncompetitive market of musical performance rights … a competitive price will never emerge, and there is no economically meaningful method of determining a competitive price.'¹⁴

The district court faces three primary obstacles to completing the task assigned to it under the consent decrees. First, because “there is no competitive market in music rights,” the parties and the court “lack any economic data that may be readily translated into a measure of competitive pricing for the rights in question.' Showtime II , 912 F.2d at 577.

Second, in a competitive market, the price of a product or service should equal the marginal cost to the seller of producing an additional unit of that product or service. But “the marginal cost of additional consumption of musical compositions is zero or even negative.” Sobel, supra note 14, at 33-34.¹⁵ As a result, 'according to economic theory, the ‘competitive’ price for music [licenses] would be zero (or less).' Sobel, supra note 14, at 34; see also Cirace, supra note 14, at 283-84 ('Because marginal cost is zero in the market of musical Page 24performance rights … the competitive price theoretically would also be zero.”). Setting BMI’s price at zero, however, would conflict with the goal of compensating copyright owners for the right to use their musical works.

Third, BMI and NACPA both have substantial market power, so a negotiation between BMI and NACPA involves a bilateral monopoly. See Cirace, supra note 14, at 281. “One distinctive feature of a bilateral monopoly is that … price and output are indeterminate.” Id. at 283; see also id. at 304 (“Given the wide range of possible prices in bilateral monopoly bargaining, the court, if it did set a price, would probably choose the historical price with minor adjustments.').

In light of these features of the relevant market, a district court cannot identify a unique “market value” for a BMI blanket license. “Rather than speak in terms of competitive market pricing, when such a term carries vague significance within the context of the blanket license market,” district courts have considered it “more instructive to view the [c]ourt's role as a moderating influence on [BMI or] ASCAP that serves 'to minimize the likelihood that [its] evident market leverage may be exerted to obtain unacceptably inflated price levels for its license[s].'' ABC/CBS , 831 F. Supp. at 144-45 (quoting Showtime II , 912 F.2d at 576). The district court will therefore 'consider previous agreements voluntarily entered between the parties, or those similarly situated, as the starting point of its analysis” and then “account for alterations in the economic conditions confronting the parties, as well as appraise variations in the nature and value of the rights at issue.' Id . The district court will not 'merely endorse as appropriate for today the terms of compromises concluded yesterday,' but it also cannot 'ignore the history of the parties' preferences as expressed in their prior agreements.' Id. at 145. In fact, the “prices negotiated voluntarily in an arms-length transaction offer Page 25the only palpable point from which to proceed towards an estimation of fair value for later periods.” Id.

The price of a music license equals a percentage rate of the revenue derived from the licensee’s use of the music. In this case, the parties disagree as to both the rate and the definition of the gross revenue base.

Historically, music license agreements between PROs and concert promoters have defined the gross revenue base as the face value of concert tickets sold. BMI seeks to expand the revenue base for the Current Period to include, in addition to the face value of tickets sold, (1) “revenues received by the promoter from any tickets sold in the first instance directly onto the secondary market (including any amounts above the face value of the ticket),' (2) 'any ticket service, handling, or other fees above the face value of the ticket paid by the consumer if received by the promoter,” (3) “box suite and VIP package revenues attributable to live concerts and paid to the promoter or a venue or artist with which the promoter has a contractual relationship,” and (4) “sponsorship revenues attributable to live concerts and paid to the promoter.” BMI , 664 F. Supp. 3d at 478.

BMI recognizes that “[h]istorically, the revenue base for concert promoter PRO licenses included only the face value of the tickets sold,” but it argues that this “expanded revenue base” is necessary “[t]o account for changes in the concert industry.” Appellee’s Br. 31. NACPA maintains that the gross revenue base should continue to be defined as the face value of concert tickets sold. According to NACPA, 'every agreement in the history of the industry' has adopted that definition and it would be impracticable for concert Page 26promoters to calculate gross revenue under the new definition that BMI proposes. Appellant's Br. 54.

The district court began its analysis with the proposition that the gross revenue base should equal the 'fair market value of the music' measured by '[w]hat the consumer pays to attend the concert.” BMI , 664 F. Supp. 3d at 479. The district court proceeded from that proposition to the conclusion that the gross revenue base for the Current Period must be “enlarged to include not only the face value of tickets, but also the value of tickets sold in the primary instance directly onto the secondary market, ticket servicing fees received by the promoter, and revenues from box suites and VIP package[s] that are attributable to live concerts and paid to the promoter or a contractually related third-party.' Id. at 485. The district court explained that '[e]ach of those categories reflect[s] payments to ‘receive the product or service in question,’ a live concert, which is 'an excellent indicator of its fair market value,' while controlling for revenues that never flow to the promoters.” Id. at 484 (citation omitted) (quoting Music Choice II , 316 F.3d at 195).

The district court acknowledged NACPA's argument 'that calculating the gross in this manner is unworkable because promoters do not have access to all the revenue information, specifically revenues for boxes and VIP packages.' Id. at 485. But the district court reasoned that '[l]imiting the revenues to those received by the promoters or a contractually related third-party helps to alleviate those concerns, even if it does not produce perfectly efficient administration.” Id .

We conclude that the decision of the district court to expand the definition of the revenue base for the Current Period was unreasonable for three reasons. First, the district court did not provide Page 27an adequate justification for adopting a definition that has no precedent among the benchmark agreements that the parties identified. The district court recognized that calculating the expanded revenue base would create administrative costs for NACPA, but it did not identify any corresponding benefits to the parties that would offset the costs. The record indicates that there are no such benefits. Second, even accepting the premise that the revenue base should be measured by what the consumer pays to attend the concert—rather than by the face value of concert tickets sold-the district court included revenue categories that do not reflect what the consumer pays for the concert. Third, the district court did not resolve the objection that the expanded revenue base would be commercially impracticable.

The district court considered twelve benchmark agreements— including agreements by both regulated and unregulated domestic PROs with NACPA and non-NACPA promoters as well as with individual members of NACPA such as Live Nation and AEG. All of the benchmark agreements defined the revenue base in the same way: the face value of concert tickets sold with “customary deductions for ticket servicing fees, taxes, facility fees, and parking.'¹⁶

The district court did not address the fact that none of the domestic benchmark agreements the parties identified used anything other than the face value of tickets sold to measure the revenue base. Even assuming that the newly defined revenue base captured the “fair market value of the music”—which, as explained in Part II.B., it Page 28did not—the district court need not define the revenue base according to an abstract notion of 'what retail customers pay to receive the product or service in question” if there is a “valid reason for using a different measure.” Music Choice II , 316 F.3d at 195. In this case, there was not only a valid but a compelling reason to define the revenue base as the face value of tickets sold: every domestic benchmark agreement that the district court considered defined the revenue base that way.

Benchmarks play the paramount role in rate-setting proceedings because it is impossible to identify an indisputably correct market price based on economic principles. When the district court adopts a rate, a revenue base, or a substantive license term that has no precedent among the benchmark agreements it considered, it must have a compelling reason for doing so.

The primary role of the district court in a rate-setting proceeding is to serve “as a moderating influence” on the PRO and to “minimize the likelihood that [its] evident market leverage may be exerted to obtain unacceptably inflated price levels for its license[s].” ABC/CBS , 831 F. Supp. at 144-45. If the industry has settled on a particular definition of gross revenue—or another substantive license term—the role of the district court will presumptively be confined to determining whether the percentage rate quoted by the PRO is reasonable, holding the revenue base definition and other license terms fixed.

The economic context supports the conclusion that a district court which aims to curb the regulated PRO’s excessive bargaining power must ordinarily decide only the price of a music license while holding the substantive terms, such as the definition of gross revenue, fixed. As contract law scholars have explained:

Page 29It is widely believed that parties exercise bargaining power by requiring weaker contracting partners to take unfavorable [substantive] terms. … But when bargaining power is determined prior to contract formation, as is common in business contexts, these views are incorrect. Bargaining power instead is exercised in the division of the surplus, which is determined by the price term. Parties jointly choose the contract terms so as to maximize the surplus, which the price may then divide unequally.¹⁷

In other words, both PROs and promoters will prefer the set of substantive license terms-including the revenue base definitionthat maximizes the total value created by the blanket license. The party with the greater bargaining power will then seek a greater share of that value by negotiating for a higher percentage rate. It is in neither party's interest to adopt a revenue base definition that imposes administrative costs without adding to the total surplus that the parties share. Consistent with these dynamics, PROs and concert promoters have uniformly adopted a simple and easily administrable revenue base definition—the face value of concert tickets sold—and have negotiated varying royalty rates according to the relative bargaining power of the parties.¹⁸

Page 30The parties that negotiated the twelve benchmark agreements concluded that the benefits, if any, of a broader or more complex revenue base definition did not outweigh the administrative costs. The district court offered no reason to believe that this negotiation between BMI and NACPA should be the first to reach a different conclusion. The district court admitted that its broader revenue base definition “does not produce perfectly efficient administration.” BMI , 664 F. Supp. 3d at 485. But the district court did not identify any corresponding benefits to justify these administrative costs or any other reason to expect that broadening the revenue base will increase the surplus from the licensing agreement.

BMI argues on appeal that failing to include all revenue streams that flow to Live Nation and AEG in the gross revenue base would allow those entities to play a 'revenue shell game' to minimize payments to PROs. Appellee’s Br. 34. According to BMI and some amici, Live Nation and AEG have increased “junk” ticketing fees— and thereby boosted revenue to their ticketing affiliates Ticketmaster and AXS—without proportionately increasing the face value of the tickets. In other words, Live Nation and AEG each seeks “to pretend that songwriters should only be paid out of its unilaterally determined promoter pocket while it stuffs revenues into its other pockets.” Brief of Amicus Curiae National Music Publishers’ Association 12.

The district court did not adopt this “shell game” argument or even mention the argument in its opinion. In any event, the argument does not justify the departure from the uniform substantive terms of the benchmark agreements. If Live Nation is manipulating revenues in this way, the expanded revenue base definition may prevent Live Page 31Nation from transferring value to itself and away from BMI and its artists. But it would not increase the joint surplus from the blanket license agreement; to the contrary, it would reduce the surplus because it is costly to administer. BMI could preserve the larger surplus by retaining the simpler definition and still receive compensation for any losses that result from the promoters’ “shell game' by insisting on a higher percentage rate. See supra note 18. Adjusting both the rate and the base, by contrast, would allow another sort of shell game that enables BMI to capture revenue beyond the appropriate rate at the cost of an overall destruction of value.

BMI argues that the expanded revenue base is necessary “[t]o account for changes in the concert industry.” Appellee’s Br. 31. The primary change is the advent of large and powerful promoters. Even if the emergence of Live Nation and AEG has increased NACPA's bargaining power vis-à-vis BMI—which, as explained in Part III.C., seems unlikely—the relative bargaining power of the parties does not ordinarily affect substantive contract terms such as the revenue base definition; it affects only the price term, which in this case is the royalty rate. See Schwartz & Scott, supra note 17, at 554.

In a real-world, arm’s-length negotiation, the PRO would have no incentive to insist on a definition of the revenue base that increases the administrative cost to music users without a corresponding benefit-and that thereby reduces the total surplus value of the licensing agreement. Meanwhile, if an expanded revenue base would increase the total value of the deal, then a music user would agree to it regardless of its bargaining power. But there is no apparent justification for the district court to impose an unprecedented revenue base against a party’s will. The district court acted unreasonably by Page 32departing from the industry-standard revenue base definition without a compelling reason.

Even if it were appropriate to define the revenue base not by reference to historical standards but according to the principle of “[w]hat the consumer pays to attend the concert,” BMI , 664 F. Supp. 3d at 479, the definition adopted here does not conform to that principle. The district court included categories of revenue— from box suites and VIP packages-that do not reflect what the consumer pays to attend the concert. The district court reasoned that while “[b]ox suites and VIP packages may include charges in addition to the seat ticket, like for food or merchandise,” a consumer “must pay the total price if she wishes to sit in these premium seats and hear the music.” Id. at 484.

In reaching that conclusion, the district court misinterpreted our decision in Music Choice II . That case involved a company in “the so-called residential music service industry, consisting of companies providing music to cable and satellite TV subscribers.' Music Choice II , 316 F.3d at 191. The district court had held 'that retail revenues of the cable / satellite operators did not properly reflect the fair market value of the music because in paying the retail price the subscriber was paying for materials and services not provided by the author of the music, such as the machinery of transmission and delivery to the subscriber's home.' Id. at 194. For that reason, the district court “concluded that the true value of the music is expressed at the earlier stage where it is incorporated into Music Choice's programs and sold wholesale to the cable / satellite operators.' Id. (internal quotation marks omitted). We vacated the judgment of the district court and remanded for further proceedings. In doing so, we Page 33recognized that 'to make the music available to its customers, the retail seller must incur expenses for various processes and services not provided by the owner of the music, such as the laying of cable, the establishment of satellite systems, etc.' Id. at 195. But we said that these retail expenses were 'in no way incompatible with the proposition that retail revenues derived from the sale of the music fairly measure the value of the music” because “[t]he customer pays the retail price because the customer wants the music, not because the customer wants to finance the laying of cable or the launching of satellites.” Id.

In this case, the district court decided that the perks of a box suite or a VIP package resemble “the machinery of transmission and delivery to the subscriber's home' that we described in Music Choice II . Id. at 194. The district court explained that while the retail price of a box suite or a VIP package covers items whose 'only relationship to the music is in connection with its delivery to market,” the price still “reasonably reflects the value of the music because the customer is willing to pay it to hear the music.” BMI , 664 F. Supp. 3d at 484.

The analogy to Music Choice II is misplaced. The customer who buys a cable television package with a music channel “pays the retail price because the customer wants the music, not because the customer wants to finance the laying of cable or the launching of satellites.” Music Choice II , 316 F.3d at 195. Those retail expenses represent the cost of delivery to the consumer. The customer who buys a VIP concert package, by contrast, pays the premium retail price precisely because the customer does want to finance the delivery of additional services, such as food and drinks, access to a box suite, a backstage pass, or other benefits. The VIP customer pays more than other customers to listen to the same music because the price of a VIP ticket Page 34involves a payment not only to attend the concert but also to receive these additional benefits. The retail expenses associated with producing a concert—which would reflect the cost of delivery to the consumer—are already reflected in the price of a regular ticket.

BMI suggests that '[d]efining the revenue base to explicitly include VIP revenues is also necessary to ensure consistency in payment among promoters and across concerts' because '[e]ven though NACPA … instructed its members in 2016 that such VIP revenues should be included in the revenue base, concert promoters took inconsistent positions on the inclusion of such revenues.' Appellee's Br. 57 (citation omitted); see also BMI , 664 F. Supp. 3d at 484 (noting that “[r]evenues from both [box suites and VIP packages] are in some instances already being reported in the base').¹⁹ BMI and NACPA may freely agree to include VIP revenues in the gross revenue base. But even on the assumption that the district court was correct to theorize about an ideal revenue base that reflects the principle of “[w]hat the consumer pays to attend the concert,” BMI ,Page 35 664 F. Supp. 3d at 479, the inclusion of VIP revenues violates that principle.

NACPA identifies another problem with the inclusion of box suite revenues: box suites are “sold mostly through annual contracts for all events hosted at a venue, including sporting events.” Appellant’s Reply Br. 35. The district court purported to include only those box suite revenues that are “attributable to live concerts.” BMI , 664 F. Supp. 3d at 485. But NACPA objects that “there is no way of knowing how much of [the annual box suite] revenue is attributable … to live music events, let alone to a particular concert.” Appellant’s Br. 55-56. BMI responds that the trial evidence “showed that box suites can be sold for individual concerts.” Appellee’s Br. 58. Even if that is possible, however, the trial evidence established that most box suite sales are made on an annual basis. See, e.g. , App’x 203 (testimony of the CEO of AEG that “out of 300 suites, probably 294 of them are on annual contracts'). And the venues rather than the promoters apparently control whether box suites are sold on an individual or annual basis.

The important point here is that the district court did not even address the potential administrative difficulties in disaggregating live concert revenue from annualized box suite sales. And a rate term cannot be reasonable if the district court considered only its correspondence to an abstract principle without regard to whether its implementation would be administratively feasible and economically rational.

NACPA argues that the expanded revenue base itself is 'commercially impracticable.' Appellant's Br. 55. According to NACPA, '[u]nrebutted testimony at trial made clear that NACPAPage 36 promoters do not have access to the revenue data that would be necessary to apply all of these expanded revenue categories to the rate period at issue.” Id. The district court addressed this problem in a single sentence: 'Limiting the revenues to those received by the promoters or a contractually related third-party helps to alleviate those concerns, even if it does not produce perfectly efficient administration.” BMI , 664 F. Supp. 3d at 485.

That conclusion indicates that the district court sought to conform the revenue base definition to an abstract notion of what properly counts as revenue regardless of whether that definition would reduce the surplus value available to the parties. But there is no single correct definition of the revenue base; the primary task of the district court is to reasonably approximate the licensing agreement that would result from an arm's-length transaction between a willing seller and buyer in which the exertion of “market leverage' is 'moderat[ed]' to avoid 'unacceptably inflated price levels.” ABC/CBS , 831 F. Supp. at 145. As explained above, it would not be rational for the parties to agree to incur administrative costs that have no corresponding benefit, so an agreement that imposes such costs cannot be reasonable.

Even on its own terms, moreover, the statement of the district court is unresponsive to the concern of commercial impracticability. NACPA explains that its promoters do not have access to data about revenues received by contractually related third parties rather than by NACPA promoters. 'The fact that the non-promoter entities receiving such revenues may have a ‘contractual relationship’ with promoters does not solve the promoters' absence-of-data problem when the promoters are not actually receiving the revenues.' Appellant's Reply Br. 36. BMI's economic expert 'opined that promoters could include revenue reporting obligations in their Page 37contracts” to “alleviat[e] the practical difficulty.” Appellee’s Br. 33 n.8. But that means the new revenue base definition requires the promoters not only to modify industry-standard contracts but also to ensure that venues comply with the new reporting obligations and to verify the accuracy of the revenue figures the venues report. It would not be rational for the parties to agree to a licensing agreement that imposed such extensive costs as the rewriting of third-party contracts and the establishment of a new data collection bureaucracy. It is therefore unreasonable for a district court to impose it.

We conclude that the district court erred by expanding the revenue base. First, the district court did not justify its unprecedented departure from the industry-standard definition of “gross revenue,” and the record indicates that no justification is available. Second, the district court prioritized adherence to an abstract principle of “[w]hat the consumer pays to attend the concert” but its definition did not even conform to that principle. Third, the district court did not merely fail to address the objection that the expanded revenue base would be commercially impracticable but even acknowledged that it was imposing inefficiencies without an economic justification.

The district court further erred in the benchmark analysis that led it to adopt a rate of 0.5 percent. First, the district court failed to assign weights to the benchmarks and implicitly accorded greater weight to the SESAC and GMR benchmarks than to the BMI and ASCAP benchmarks. Second, the district court erroneously considered license agreements with non-NACPA promoters. Third, the district court failed to identify any significant changes in the live concert industry since the BMI and ASCAP benchmark agreements were negotiated that would justify a higher rate.

Page 38The district court “must determine the degree of comparability' between the parties that negotiated each of the benchmark agreements, on the one hand, and the parties before the court, on the other. Music Choice IV , 426 F.3d at 95 (internal quotation marks omitted). That comparability analysis will determine the weight it assigns to each benchmark agreement. And the district court may thereby “explain how it reached a particular rate sufficiently to permit our review of the rate for reasonableness.” Id. at 99.

In this case, the district court did not indicate the relative weights it assigned to the benchmark agreements it considered; nor did it explain how it derived the rate of 0.5 percent from those benchmarks. It listed the seven different implied rates reflected in the twelve benchmarks it considered: 0.21 percent, 0.23 percent, 0.34 percent, 0.36 percent, 0.37 percent, 0.51 percent, and 0.54 percent. And it 'therefore' decided that a rate of 0.5 percent would be 'a reasonable one.” BMI , 664 F. Supp. 3d at 488.

An average of the implied rates for the twelve benchmark agreements yields a rate of 0.44 percent. An average of the seven different percentage rates represented among the twelve benchmarks yields a rate of 0.37 percent. The BMI and ASCAP benchmarks all had implied rates between 0.23 and 0.37 percent. To arrive at a rate of 0.5 percent based on the benchmarks, the district court must have implicitly assigned greater weight to the benchmarks with higher implied rates: the SESAC and GMR benchmarks.

If so, that was unreasonable. The district court correctly recognized that 'fairly negotiated prior agreements are the proper starting point from which to determine reasonable fees for subsequent periods.” Id. at 486 (quoting United States v. ASCAP , 157Page 39 F.R.D. 173, 195 (S.D.N.Y. 1994)). And it correctly determined that “ASCAP is indisputably BMI’s closest comparator” and that “[t]he rights in question in the ASCAP agreement are essentially the same as those NACPA seeks to license from BMI.” Id. In the agreements “[f]or those similar rights, NACPA has paid ASCAP and BMI at near parity in prior agreements.' Id. Based on these conclusions, one would expect the district court to assign greater weight to the BMI and ASCAP benchmarks.

The district court, however, explained that “unconstrained by a consent decree, GMR and SESAC have been able to preserve the legal monopoly power granted by the Copyright Act that rate-setting is intended to incorporate” and that “[t]he analogous market in which GMR and SESAC operate thus reflects the level of competition that would be inherent in a direct licensing negotiation.” Id. at 487. Those considerations-and the apparent decision of the district court to assign greater weight to the SESAC and GMR benchmarks—suggest that it adopted the argument that these benchmarks “provide the best evidence of the price that a willing licensee and a willing licensor would agree to in an arm’s-length transaction in a competitive free market” because “SESAC and GMR have no antitrust consent decree mandating that those PROs grant a license to music users upon request.' Brief of Amicus Curiae ASCAP 4.

This argument implies that—far from serving “to minimize the likelihood” that the “evident market leverage” of the regulated PROs will produce 'unacceptably inflated price levels'-the consent decrees have instead kept BMI’s and ASCAP’s rates artificially low. Showtime II , 912 F.2d at 576. But we have said that “rate-setting courts must take seriously the fact that they exist as a result of monopolists exercising disproportionate power over the market for music rights.” Music Choice IV , 426 F.3d at 96; see also Showtime II , 912 F.2d at 570Page 40 (“Though the rate court’s existence does not mean that ASCAP has violated the antitrust law, the court need not conduct itself without regard to the context in which it was created.”). As long as the consent decrees remain in place, it is not within the discretion of the district court to decide that BMI's market power no longer warrants the 'moderating influence' of the rate-setting framework. ABC/CBS , 831 F. Supp. at 145.

The district court suggested that the market shares of SESAC and GMR 'are more comparable to those of the large music publishers that music users would have to negotiate with directly in the absence of PROs.” BMI , 664 F. Supp. 3d at 487. But a publisher with a catalogue similar in size to those of SESAC and GMR would not necessarily charge a competitive price for a blanket license. NACPA claims that “SESAC and GMR have repeatedly faced claims that they have extracted supracompetitive fees in violation of the antitrust laws.” Appellant’s Br. 40; see also id. at 11 & n.2. The district court decided that the “realities” of extensive “negotiations between SESAC, GMR, and promoters cannot be reconciled with NACPA’s claim that the PROs are extracting rates that are anti-competitive.” BMI , 664 F. Supp. 3d at 488. The fact that a PRO engaged in extensive negotiations, however, does not show that it lacks market power. A negotiation might begin with a demand for a higher supracompetitive price and result in a lower but still supracompetitive price.²⁰ The district court said that '[t]he fact that promoters elected to buy a blanket license for their own business goals and convenience does not Page 41establish that they did so under compulsion.” BMI , 664 F. Supp. 3d at 488. But a lack of compulsion does not establish that the seller is not charging a supracompetitive price. A monopolist can raise its price as long as the alternative to its product is sufficiently costly or inconvenient.²¹

Other district courts have noted that “[t]he SESAC license has historically been a benchmark of limited value because the public knows little about the size of the SESAC repertoire” and therefore “it is difficult to adjust a SESAC license rate to arrive with confidence at an implied ASCAP rate.” In re Pandora Media, Inc. , 6 F. Supp. 3d 317, 362 (S.D.N.Y. 2014).²² Moreover, 'SESAC's small size, when compared to ASCAP and BMI, not only amplifies any error in a projection, but also reduces the incentive to resist SESAC's rate requests' because 'the cost associated with resistance may not be justified when a license fee is relatively small.' Pandora Media , 6 F. Supp. 3d at 362.²³

Page 42In light of these considerations, it was unreasonable for the district court to assign greater weight to the SESAC and GMR licenses than to the prior BMI and ASCAP agreements.

The district court additionally erred by treating agreements with promoters unaffiliated with NACPA as comparable benchmarks. The district court reasoned that “[c]ompared to the BMINACPA license,” the agreements between BMI or ASCAP and nonNACPA promoters “arose in a market that reflects the same degree of competition and economic circumstances, and they cover the same rights.” BMI , 664 F. Supp. 3d at 486. The district court added that “[a]lthough NACPA and non-NACPA promoters differ in the scope of their business operations, they are direct competitors for shows hosted in smaller venues.' Id. The district court did not expressly address the licenses between SESAC or GMR and non-NACPA promoters, but BMI maintains that the district court implicitly applied the same reasoning to those licenses. See Appellee's Br. 54.

That reasoning was inadequate to establish comparability. The ASCAP/NACPA and ASCAP/non-NACPA licenses covered approximately the same period—2020-2021 and 2021, respectively— but the implied rates differed by a significant amount: 0.14 percent. See BMI , 664 F. Supp. 3d at 482. Similarly, the SESAC/NACPA and SESAC/non-NACPA licenses were issued around the same timecovering 2019-2024 and 2021, respectively—but the rates differed byPage 43 0.17 percent. Id. The difference in the rates indicates that the nonNACPA promoters are not similarly situated to NACPA when negotiating with PROs.²⁴ NACPA explains that BMI has sought to “realign” NACPA and non-NACPA promoter rates “since 2013” but “has not been able to achieve that result through arm's-length negotiations.' Appellant's Reply Br. 31. It is not surprising that NACPA—which negotiates on behalf of a large group of promoters— has more leverage than other promoters in negotiations with PROs. “NACPA is itself an aggregator” and “represents the interests of the largest promoters.” Brief of Amicus Curiae National Music Publishers’ Association 8. That difference makes the agreements with non-NACPA promoters inappropriate benchmarks.

The parties do not address whether, if agreements with nonNACPA promoters are invalid benchmarks, GMR’s agreements with Live Nation and AEG remain valid benchmarks. Live Nation, AEG, and their affiliates compose about 75 percent of NACPA’s membership and pay about 90 percent of NACPA's licensing fees under its agreement with BMI. According to one amicus, “[f]or all practical purposes, NACPA is Live Nation and AEG.” Id. at 14. All of the GMR agreements that the district court considered—including the agreements with Live Nation and AEG—have the same rates and rate structure, which may suggest that the agreements are non-negotiated form licenses. Yet witnesses from Live Nation and AEG testified that the GMR licenses were heavily negotiated.²⁵ Roux originally Page 44recommended that Live Nation negotiate a license agreement with GMR “through NACPA” and agree to “GMR[’s] original ask” with respect to the percentage rate. Trial Transcript (10/31/22) at 497. On the one hand, the testimony suggests that the licenses between GMR and Live Nation and AEG reflected a negotiated rate and that the rates would have been similar if the licenses had been negotiated through NACPA. On the other hand, NACPA negotiated a significantly lower rate with SESAC—a PRO of similar size to GMR— than the non-NACPA promoters obtained.

We leave this question for the district court to address in the first instance. The district court should determine whether the rate and rate structure in the GMR licenses with Live Nation and AEG reflect what NACPA would have achieved in a negotiation with GMR. If NACPA would have negotiated a significantly lower rate with GMR, then the two agreements would not be comparable benchmarks. If the rate and rate structure in the two licenses reflect what NACPA would have obtained in a negotiation with GMR, then the agreements would be comparable benchmarks. This exercise is admittedly hypothetical. Apparently, GMR was unwilling to negotiate with NACPA because of NACPA’s 6 percent administrative discount. See Agreed Findings of Fact ¶ 62 (“GMR does not license with NACPA. GMR refused to negotiate with NACPA and instead engaged in direct negotiations with individual promoters.'). But Page 45projecting the outcome of a hypothetical negotiation is the purpose of the rate-setting proceeding. See Showtime II , 912 F.2d at 569 ('Fair market value is a factual matter, albeit a hypothetical one.”).

Assuming that the GMR licenses with Live Nation and AEG are appropriate benchmarks, the GMR licenses with non-NACPA promoters remain inappropriate even if the licenses include the same rate. In any event, because all six GMR licenses cover a similar period and contain identical quantitative terms, it is unnecessary to include all six as benchmarks. Doing so would risk introducing bias into the rate determination. Even the two possible GMR benchmarks-the agreements with Live Nation and AEG—deserve less weight relative to the BMI and ASCAP benchmarks.

The district court did not identify any relevant changes in economic circumstances that justify its departure from the prior rates negotiated by NACPA with BMI and ASCAP. The district court gestured toward changes in the live concert industry because of the emergence of Live Nation and AEG, but it did not explain why those changes justified a higher rate. Moreover, the BMI and ASCAP benchmarks mostly postdate the emergence of Live Nation and AEG. The district court expressly found that “[t]here is no evidence in the record that the economic circumstances in 2018” when the ASCAP/NACPA license was negotiated “are significantly different from the present day.” BMI , 664 F. Supp. 3d at 486.

On appeal, BMI does not offer a persuasive reason why the emergence of Live Nation and AEG as dominant players in the concert promotion industry would justify a departure from the historical rate. The argument is that '[f]or all practical purposes, NACPA is Live Nation and AEG, and those entities all have Page 46substantial market and negotiating power.” Brief of Amicus Curiae National Music Publishers' Association 14. But as the bargaining power of BMI’s counterparty increases, the counterparty should be able to insist on a lower rate in an arm’s-length negotiation. In fact, it appears that the emergence of Live Nation and AEG did not enable NACPA to obtain a lower rate in its last negotiation with ASCAP. BMI’s economic expert testified that the 2018 ASCAP/NACPA license reflected 'quite a large increase' from prior years. Supp. App'x 185.²⁶ The record does not establish that the consolidation of the live concert industry afforded NACPA additional bargaining power. It may be that, even before the emergence of Live Nation and AEG, NACPA already represented more or less all concert promoters, so BMI and ASCAP already had to negotiate rates with a monopsonist.

BMI argues that-in 2018 when the ASCAP/NACPA license was negotiated—ASCAP knew “that BMI was likely to commence a rate proceeding against NACPA. Accordingly, ASCAP negotiated for an early termination right. By securing this right, ASCAP positioned itself to benefit if BMI was successful in litigation, without incurring the substantial costs that BMI would incur.' Appellee's Br.²³ (citations omitted). The implication is that ASCAP did not necessarily believe in 2018 that the rate to which it agreed represented a fair rate. The district court did not adopt this theory; to the contrary, it decided that '[t]he ASCAP-NACPA license is a valid benchmark.' BMI , 664 F. Supp. 3d at 486. Moreover, BMI conceded before the district court that the 2018 ASCAP/NACPA rate reflected “what the parties Page 47thought Judge Cote would do in a rate court proceeding.'²⁷ That concession conflicts with the contention that ASCAP expected BMI to obtain a higher rate in its rate-setting proceeding.

Moreover, the argument does not make sense from an economic perspective. If ASCAP believed that the rate in the 2018 ASCAP/NACPA agreement was unfairly low and that BMI would obtain a higher rate in its rate-setting proceeding, then ASCAP would have insisted on—and likely would have obtained—a higher rate in its own negotiation. That outcome not only would have allowed ASCAP to collect more revenue; it also would have increased the likelihood that BMI would obtain a higher rate in the rate-setting proceeding by providing a higher benchmark rate.

The district court suggested that '[i]ncreasing the rate of a license for live concerts better reflects the fair market value placed on licenses in music intensive industries.” BMI , 664 F. Supp. 3d at 488. The district court explained that '[l]icenses in music intensive industries, like commercial radio stations or virtual live concert streaming services, have higher rates because music is at the heart of the product being offered by the business.” Id. By contrast, “in a lessmusic intensive industry, like television sporting events or talk radio, the music is supplemental to the consumers’ primary purpose and thus the rate owed to BMI's affiliates for their contributions is smaller.” Id. at 488-89. Because 'it is indisputable that music is essential to a live concert,” the district court concluded that “[t]hose who contribute to the musical compositions should be compensated accordingly.” Id. at 489. Yet the commercial conditions in the “music intensive industries' cannot be comparable to the live concert Page 48industry because music users such as radio stations and streaming services pay rates of up to 2.5 percent—far higher than what even the regulated PROs are able to charge for performance licenses.²⁸

Based on the considerations we have identified, the comparable benchmark agreements are the BMI/NACPA license, the ASCAP/NACPA license, the SESAC/NACPA license, and possibly the GMR/Live Nation and GMR/AEG licenses. The primary rates for these agreements are 0.21 percent, 0.27 percent, 0.40 percent, and 0.63 percent, respectively. An average of these four rates yields a rate of 0.38 percent. If the BMI and ASCAP benchmarks were to receive double the weight of the SESAC and GMR benchmarks, then the resulting rate would be 0.33 percent.

Accordingly, a well-supported rate for the new BMI/NACPA license would be significantly lower than the rate imposed by the district court, but it may be higher than the rate for the preexisting BMI/NACPA license. To arrive at such a rate, the district court “must determine the degree of comparability' of the benchmarks and “explain how it reached a particular rate sufficiently to permit our review of the rate for reasonableness.' Music Choice IV , 426 F.3d at 95, 99 (internal quotation marks omitted). We remand for the district court to do so.

The district court adopted BMI’s rate quote for the Retroactive Period that extends from January 1, 2014, to June 30, 2018. The parties Page 49do not devote as much argument to that decision on appeal, presumably because the parties’ proposed rates do not differ much. NACPA’s economic expert estimated that BMI’s rate quote for the Retroactive Period implied a unitary rate of 0.28 percent, and NACPA itself sought a rate of 0.23 percent. See App'x 372. In addition, the district court left the revenue base for the Retroactive Period “defined to include only the face value of the ticket (exclusive of all fees and other charges).' BMI , 664 F. Supp. 3d at 477. The district court reasoned that BMI's rate quote for the Retroactive Period was reasonable because it “would align the NACPA rates with the rates paid to BMI by non-NACPA promoters during that time,” and the district court had “already found that the BMI-nonNACPA benchmark is valid.” Id. at 489; see also Appellee’s Br. 66 (“For over a decade, BMI licensed NACPA and non-NACPA promoters at the same rates. BMI’s rate quote for the Retroactive Period returns that alignment.') (citation omitted).

As explained above, the district court unreasonably relied on licenses with promoters unaffiliated with NACPA. The comparable benchmarks indicate that an arm's-length negotiation would not result in an alignment of NACPA rates with those negotiated by nonNACPA promoters. And it is unsurprising that NACPA, which represents many concert promoters, would have the market power to negotiate a lower rate. It was unreasonable for the district court to adopt a rate quote for the Retroactive Period on the ground that it would align NACPA rates with non-NACPA rates.

BMI argues in its cross-appeal that the district court erred by denying its motion for prejudgment interest. We have held that “[t]he decision to award prejudgment interest is governed by the equities, Page 50reflecting 'considerations of fairness' rather than 'a rigid theory of compensation.’” SEC v. Contorinis , 743 F.3d 296, 308 (2d Cir. 2014) (quoting Blau v. Lehman , 368 U.S. 403, 414 (1962)). The decision to award prejudgment interest “should be a function of (i) the need to fully compensate the wronged party for actual damages suffered, (ii) considerations of fairness and the relative equities of the award, (iii) the remedial purpose of the statute involved, and/or (iv) such other general principles as are deemed relevant by the court.' Wickham Contracting Co. v. Local Union No. 3 , 955 F.2d 831, 833-34 (2d Cir. 1992). “Awards of prejudgment interest must not result in overcompensation of the plaintiff,' however, and are disfavored when 'the statute itself fixes damages deemed fully compensatory as a matter of law.” Id. at 834.

The BMI Consent Decree provides that during the pendency of any rate negotiation or rate-setting proceeding, “the applicant shall have the right to use any, some or all of the compositions in [BMI’s] repertory … without payment of any fee or other compensation.' BMI Consent Decree § XIV(A). However, 'either the applicant or [BMI] may apply to [the district court] to fix an interim fee pending final determination of what constitutes a reasonable fee.' Id. § XIV(B). If the district court fixes an interim fee, BMI must then issue a license “providing for the payment of a fee at such interim rate from the date the applicant requested a license.” Id. Once the district court makes its decision, 'the reasonable fee finally determined by [the district court] shall be retroactive to the date the applicant requested a license.” Id. The consent decree does not specify whether BMI is entitled to prejudgment interest if the reasonable fee the district court ultimately sets is higher than the interim fee.

Rather than petitioning the district court to fix an interim fee, the parties in this case agreed in 2014 to an interim fee of 0.3 percent Page 51of gross revenue-defined as the face value of tickets sold-for concerts held in venues seating under 10,000 and to an interim fee of 0.15 percent for concerts held in venues seating over 10,000. See BMI , 674 F. Supp. 3d at 72. The 2014 agreement does not specify whether BMI would be entitled to prejudgment interest if the reasonable fee ultimately exceeds the interim fee.

The district court decided that even though the parties had negotiated an interim fee agreement rather than petitioned to fix an interim fee, there was “no reason to ignore the relationship between the interim fee and final fee as outlined in the Consent Decree.” Id. at 73. It explained that “Article XIV clearly states that the interim fee controls until the final fee is determined and can be retroactively applied. That calls for a simple substitution of the interim rate by the new rate, not a new rate plus interest, as though the new higher rate had been known since back then.” Id. In other words, “[t]he interim fee rate is correct for the time it is in effect: it is not a down payment on some larger amount.” Id.

The district court concluded that prejudgment interest was not warranted because prejudgment interest compensates a plaintiff for the lost time value of money a defendant wrongfully retains, and it was not wrongful for NACPA to pay only the interim fee until the district court determined a reasonable fee. Cf. Waterside Ocean Nav. Co. v. Int'l Nav. Ltd. , 737 F.2d 150, 154 (2d Cir. 1984) (explaining that an 'award of pre-judgment interest' allows 'a person wrongfully deprived of his money [to] be made whole for the loss”). The district court clarified that its 'determination of the final fee amount was designed to be the entire embodiment of what constitutes a reasonable fee, not merely some figure to which additional embellishments, such as interest, could be attached.” BMI , 674 F. Supp. 3d at 73. For that reason, “[a]n award of prejudgment Page 52interest at this juncture would overcompensate BMI at NACPA's expense.” Id.

We conclude that the district court did not abuse its discretion when it denied the motion for prejudgment interest. That the “reasonable fee” the district court ultimately sets is higher than the interim fee does not necessarily mean that the interim fee was unreasonable; there may be a range of reasonable fees. If the interim fee was reasonable, then no money has been wrongfully withheld from BMI, and it has not suffered an injury that requires compensation through prejudgment interest. The district court did not expressly state that the interim fee was reasonable in retrospect, but it did explain that its final fee determination was the 'entire embodiment” of a reasonable fee. In other words, the district court determined that the final fee, applied retroactively, provided BMI with reasonable compensation for NACPA's use of the licensed performance rights during the pendency of the negotiation and the rate-setting proceeding without the addition of prejudgment interest.

The explanation of the district court made clear that it denied prejudgment interest because either (1) the interim fee was reasonable in retrospect, or (2) some component of the final 'reasonable fee' would compensate BMI for the time value of money. Either way, the addition of prejudgment interest would overcompensate BMI. Those reasons provided a sufficient justification for the denial.

We do not hold that the consent decree requires the district court to follow the approach of retroactively substituting the final fee for the interim fee. District courts have awarded prejudgment interest in other cases. See, e.g. , ABC/CBS , 831 F. Supp. at 166. But the district court did not abuse its discretion here.

Page 53Because we vacate the judgment based on the unreasonableness of the rates the district court adopted, however, the district court may reconsider whether to award prejudgment interest on remand after it arrives at a final fee determination.

There is no single correct way to “establish[] the fair market value” of a blanket license to the repertory of a PRO. Music Choice IV , 426 F.3d at 95. Nevertheless, the district court must exercise 'a moderating influence' that avoids 'unacceptably inflated price levels' and 'account[s] for alterations in the economic conditions confronting the parties.” ABC/CBS , 831 F. Supp. at 145. It must set a fee that is 'reasonable' and 'based upon all the evidence.' BMI Consent Decree § XIV(A). To do so, the district court will rely on prior agreements negotiated by the same or similarly situated parties unless changed economic circumstances compellingly show that those agreements are obsolete. And the district court will “explain how it reached a particular rate sufficiently to permit our review of the rate for reasonableness.' Music Choice IV , 426 F.3d at 99.

In this case, the district court adopted a revenue base definition that had no precedent in the history of the industry without a compelling reason. It included revenue streams that do not reflect the fair market value of the music and that involve significant administrative costs without a corresponding benefit. The district court did not expressly assign weights to the benchmark agreements it considered but implicitly accorded greater weight to less comparable benchmarks. It relied on agreements with unaffiliated individual promoters even though NACPA has historically obtained significantly lower rates from the same counterparties. And it identified no change in economic circumstances that would justify a Page 54rate more than double what NACPA has historically paid to BMI and ASCAP.

Under these circumstances, we conclude that the rates the district court adopted were unreasonable. The district court did not abuse its discretion by denying the motion for prejudgment interest, but the denial may have depended on its prior determination of the rates. We vacate the judgment of the district court and remand for further proceedings consistent with this opinion.