Upload
others
View
11
Download
0
Embed Size (px)
Citation preview
Public Version
1
Before the UNITED STATES COPYRIGHT ROYALTY JUDGES
Washington, D.C.
In the Matter of: Determination of Rates and Terms for Digital Performance of Sound Recordings and Making of Ephemeral Copies to Facilitate those Performances (Web V)
Docket No. 19-CRB-0005-WR (2021-2025)
SOUNDEXCHANGE’S UNOPPOSED MOTION TO SUBMIT THE CORRECTED
WRITTEN DIRECT TESTIMONY OF ROBERT WILLIG
SoundExchange respectfully moves for permission to submit the Corrected Written Direct
Testimony of Robert Willig (the “Corrected Testimony”) for the purpose of (a) correcting minor
computational errors in Professor Willig’s Shapley Value calculations, and (b) making minor
corrections to certain citations, including a change to the title and date of an unpublished article
referenced in Professor Willig’s curriculum vitae. The other participants in the proceeding have
consented to the submission of the Corrected Testimony, subject to Movants’ agreement to provide
certain additional discovery related to the corrections. The Corrected Testimony is appended to
this motion as Exhibit A (clean version) and Exhibit B (redline showing changes).
SoundExchange discovered the information forming the basis of the corrections after the
submission of their written direct testimony, as it was preparing the written rebuttal testimony of
Professor Willig. Specifically, Professor Willig’s team discovered an inadvertent error in his
computation of the royalty rates that record companies earn from Sirius XM. The Written Direct
Testimony estimated this royalty rate at [ ] per subscriber (computed as [ ] in
Sirius XM royalties divided by 33.7 million Sirius XM subscribers). However, the formula that
produced this number inadvertently summed only 9 months of royalties, instead of the full 12
months. The corrected calculation is [ ] of royalties divided by 33.7 million Sirius
Electronically FiledDocket: 19-CRB-0005-WR (2021-2025)
Filing Date: 12/12/2019 11:32:13 AM EST
Public Version
2
SOUNDEXCHANGE’S UNOPPOSED MOTION TO SUBMIT THE CORRECTED WRITTEN DIRECT TESTIMONY OF ROBERT WILLIG
XM subscribers, yielding a royalty rate of [ per subscriber. This change produces a de
minimis upward effect on the royalty rates for ad-supported and subscription noninteractive
services that result from Professor Willig’s Shapley Value analysis.
SoundExchange respectfully request that the Judges exercise their discretion under 17
U.S.C. § 801(c) to accept the Corrected Testimony, and respectfully submit that such relief is
legally and factually justified. See Docket No. 19-CRB-0005-WR (2021-2025 (Web V), Order
Granting Sirius XM Radio Inc.’s and Pandora Media, LLC’s Unopposed Motion to Submit the
Corrected Written Direct Testimonies of David Reiley and Carl Shapiro at 1 (finding “that good
cause exists” to grant unopposed motion to correct testimony); Docket No. 14-CRB-0001-WR
(2016-2020) (Web IV), Order Denying Licensee Services’ Motion to Strike SoundExchange’s
“Corrected” Written Rebuttal Testimony of Daniel Rubinfeld and Section III.E of the Written
Rebuttal Testimony of Daniel Rubinfeld, and Granting Other Relief (“Web IV Rubinfeld Order”)
at 7 (Apr. 2, 2015) (“While acceptance after the due date of a correction to an otherwise timely
filing would appear to fall squarely within the discretion afforded to the Judges under Section
801(c), the appropriate procedure for asking the Judges to exercise that discretion is to file a motion
stating the relief requested and the legal and factual bases for the Judges to grant the relief.”).
First, the other participants have consented to submission of the Corrected Testimony.
Movants notified those participants as soon as they discovered this issue. Movants have already
provided to opposing counsel updated underlying data files related to the correction. Second, the
correction introduces only modest changes to Professor Willig’s Shapley Value analysis.
Specifically, the correction results in the analysis finding that $0.0029 (rather than $0.0028) is a
reasonable per-play royalty rate for ad-supported noninteractive webcasting beginning in 2021.
Similarly, the analysis now finds that $1.96 (rather than $1.95) is a reasonable per-subscriber
Public Version
3
SOUNDEXCHANGE’S UNOPPOSED MOTION TO SUBMIT THE CORRECTED WRITTEN DIRECT TESTIMONY OF ROBERT WILLIG
monthly royalty rate for subscription noninteractive webcasting beginning in 2021. Third, as the
Judges have ruled in past proceedings, it is in their and the participants’ interest to have a full,
correct, and complete record. The Judges have accepted corrected filings in this proceeding and
past proceedings in furtherance of that interest. See, e.g., Web IV Rubinfeld Order at 9 (“Consistent
with their prior rulings, the Judges are disinclined in this peculiar instance to allow their procedural
rules to prevent them from obtaining a full evidentiary record. The Judges conclude that the
interests of justice are served by examination of a more complete, informed expert record. . . .
[T]he Judges invoke section 801(c) of the Act, which provides them with discretion to ‘make any
necessary procedural or evidentiary rulings in any proceeding under this chapter.’”).
CONCLUSION
For the foregoing reasons, SoundExchange respectfully requests that the Judges grant its
Unopposed Motion to Submit the Corrected Written Direct Testimony of Professor Willig.
Respectfully submitted,
Dated: December 12, 2019 JENNER & BLOCK LLP
By: /s/ Previn Warren Previn Warren (D.C. Bar No. 1022447)
[email protected] 1099 New York, Ave., NW, Suite 900 Washington, DC 20001 Tel.: 202-639-6000
Counsel for SoundExchange, Inc., American Federation of Musicians of the United States and Canada, Screen Actors Guild-American Federation of Television and Radio Artists, American Association of Independent Music, Sony Music Entertainment, UMG Recordings, Inc., Warner Music Group Corp., and Jagjaguwar Inc.
EXHIBIT A
Public Version
Before the UNITED STATES COPYRIGHT ROYALTY JUDGES
Washington, D.C.
In re
Determination of Rates and Terms for Digital Performance of Sound Recordings and Making of Ephemeral Copies to Facilitate those Performances (Web V)
Docket No. 19-CRB-0005-WR (2021-2025)
CORRECTED WRITTEN DIRECT TESTIMONY OF
December, 2019
Robert Willig Professor of Economics and Public Affairs, Emeritus
Princeton University
Table of Contents
I. QUALIFICATIONS ................................................................................................................ 1
II. SUMMARY OF TESTIMONY .............................................................................................. 2
A. Introduction ...................................................................................................................... 2
B. Assignment and Summary of Conclusions ...................................................................... 4
III. SHAPLEY VALUES AND THEIR APPLICATION TO § 114 OBJECTIVES .................... 5
IV. A SHAPLEY VALUE MODEL FOR RECORD COMPANIES AND NONINTERACTIVE WEBCASTERS ..................................................................................................................... 12
A. Defining the Model ........................................................................................................ 12
B. Empirical Inputs ............................................................................................................. 16
i. Royalties for Other Forms of Distribution ............................................................. 17
ii. Noninteractive Distributors’ Audience Sizes ......................................................... 20
iii. Diversion Ratios ..................................................................................................... 21
iv. Record Company Play Shares ................................................................................ 23
v. Noninteractive Distributor Fixed Costs and Marginal Profit Rates ....................... 24
C. Summary of Results ....................................................................................................... 26
D. Alternative Shapley Value Calculations Using “Share of Ear” Data ............................. 32
E. Alternative Calculations Using a Nash-in-Nash Bargaining Model .............................. 34
V. CONCLUSION ..................................................................................................................... 37
Appendix A: Curriculum Vitae and List of Prior Testimony …………….…………………... A-1
Appendix B: Materials Relied On …………………………………………………………….. B-1
Appendix C: Shapley Value Model …………………………………………………………… C-1
Appendix D: Data Inputs ……………………………………………………………………… D-1
Appendix E: Use of Zauberman Survey Results to Generate Model Parameters ……………... E-1
Appendix F: Shapley Values with “Share of Ear” Data ………………………………………. F-1
Appendix G: Alternative Nash-in-Nash Bargaining Model ………………………………….. G-1
Public Version
1
I. QUALIFICATIONS
1. My name is Robert D. Willig. I am Professor of Economics and Public Affairs
Emeritus at Princeton University, where I have held a joint appointment in the Economics
Department and at the Woodrow Wilson School of Public and International Affairs since 1978.
Previously, I was a Supervisor in the Economics Research Department of Bell Laboratories. My
teaching and research have specialized in the fields of industrial organization, government-
business relations, and social-welfare theory. I served as Deputy Assistant Attorney General for
Economics in the Antitrust Division of the U.S. Department of Justice from 1989 to 1991, and in
that capacity served as the Division’s Chief Economist. I am the author of Welfare Analysis of
Policies Affecting Prices and Products, Contestable Markets and the Theory of Industry Structure
(with William Baumol and John Panzar), and some eighty articles and chapters in the professional
literatures of economics and law. I am also a co-editor of The Handbook of Industrial
Organization, which summarizes the state of economic thinking on the structure of industries and
the nature of competition among firms, and have served on the editorial boards of the American
Economic Review, the Journal of Industrial Economics, and the MIT Press Series on Regulation.
I am an elected Fellow of the Econometric Society and was an associate of The Center for
International Studies.
2. I have appeared as an expert witness before Congress, federal and state courts,
federal administrative agencies, and state public utility commissions on subjects involving
intellectual property rights, competition, regulation, and antitrust. In particular, I previously
appeared as an expert witness before the Copyright Royalty Judges in the SDARS III proceeding.
I have also served as a consultant to the Federal Trade Commission, the U.S. Department of Justice,
and many leading corporations on antitrust, regulation, intellectual property, and policy issues.
Public Version
2
3. I have spent a significant portion of my career studying, consulting, and testifying
as an expert on many different aspects of the economics of unregulated and regulated pricing, the
economics of intellectual property, and the distribution of rights to its access. I have substantial
consulting experience working on music industry issues including the SDARS III proceeding, the
merger of BMG and Sony, other potential control transactions involving major record companies
and music publishers, and the regulatory treatment of musical work performing rights
organizations.
4. My full curriculum vitae and a listing of my prior testimony are included in
Appendix A. The rate charged by Compass Lexecon for my work on this matter is $1,450 per
hour. I have a financial interest in the overall profitability of Compass Lexecon, but I have no
financial interest in the outcome of this case.
II. SUMMARY OF TESTIMONY
A. Introduction
5. On January 24, 2019, the Copyright Royalty Judges announced the commencement
of a proceeding (“Web V”) to determine reasonable rates and terms for statutory licenses under 17
U.S.C. § 114, allowing for the digital transmission of sound recordings over the internet, and under
17 U.S.C. § 112, allowing for the making of ephemeral recordings to facilitate such digital
transmissions.1 The rates and terms set in this proceeding will apply to the five year period from
January 1, 2021 to December 31, 2025.2 I understand the relevant statutory licenses are
compulsory for rights owners and are available to commercial and noncommercial noninteractive
webcasters.
1 Federal Register, Vol. 84, No. 16, Jan. 24, 2019, Notices, at 359. 2 Id.
Public Version
3
6. It is my understanding that the standard in this Web V proceeding requires the
Copyright Royalty Judges to set 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.”3
In past proceedings, I understand the relevant marketplace has been defined as a “hypothetical
marketplace, free of the influence of compulsory, statutory licenses.”4 Moreover, this hypothetical
marketplace is one that is “effectively competitive,” meaning that any “holdout” value associated
with a seller’s property that exists because it “is a necessary complement” to the property of other
sellers (i.e., “complementary oligopoly” value) should not be included in the statutory rates
determined by the Judges.5
7. I further understand that the relevant statute requires the Copyright Royalty Judges
to “base their decision on economic, competitive, and programming information presented by the
parties, including” (1) whether the use of the noninteractive service “may substitute for or may
promote the sales of phonorecords or otherwise may interfere with or may enhance the sound
recording copyright owner’s other streams of revenue from its sound recordings,” and (2) the
“relative roles of the copyright owner and the transmitting entity in the copyrighted work and the
service made available to the public with respect to relative creative contribution, technological
contribution, capital investment, cost, and risk.”6 Moreover, I understand the Judges are permitted,
3 17 U.S.C § 114(f)(2)(B). 4 Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms for Ephemeral Recording and Webcasting Digital Performance of Sound Recordings (Web IV), Docket No. 14-CRB-0001-WR (“Web IVDetermination”), at 2. 5 Id. at 40. 6 17 U.S.C § 114(f)(2)(B).
Public Version
4
but not required, to “consider the rates and terms for comparable types of digital audio transmission
services and comparable circumstances under voluntary license agreements.”7
8. At the conclusion of the Web IV proceeding, after applying these criteria, the
Copyright Royalty Judges established a rate for commercial noninteractive ad-supported services
in 2016 of $0.0017 per performance and a rate for commercial noninteractive subscription services
in 2016 of $0.0022 per performance, with these rates adjusting in each year from 2017 to 2020 to
reflect changes in the Consumer Price Index.8 These continue to be the statutory commercial
webcasting rates that apply today, with 2019 rates of $0.0018 per play for noninteractive ad-
supported services and $0.0023 per play for noninteractive subscription services.
B. Assignment and Summary of Conclusions
9. I have been retained by counsel for SoundExchange to explain and apply the
economics of bargaining theory to estimate reasonable royalty rates for the digital transmission of
sound recordings by commercial noninteractive webcasters operating under a statutory license.
Below, I present a multi-party bargaining approach that can be used to determine these rates,
known in the literature as the “Shapley Value.” I then bring data to bear in order to compute these
Shapley Values and the concomitant royalty rates at issue.9
10. In conducting my analyses and formulating my conclusions, I reviewed and
analyzed data concerning the various modes of music distribution, including relative audience
sizes, profitability, and royalty payments made to the owners of copyrights on sound recordings.
I examined and analyzed the results of a survey of consumers of music distribution that was
7 Id.
8 Web IV Determination at 1.9 I am aware that statutory royalty rates will also be set in this proceeding for noncommercial webcasting. Analysis of those rates is beyond the scope of my assignment and testimony.
Public Version
5
recently conducted under the supervision of Professor Gal Zauberman and produced in this matter,
as well as a survey conducted outside of this proceeding by Edison Research, a source of industry
data and research. I reviewed the written direct testimony of executives at major record companies
who are centrally involved with the licensing of sound recordings to digital music distribution
services. In addition, I relied on my career-long experience with economic theory, economic
literature, empirical study, and the regulated and unregulated practices of pricing products,
services, and assets, including intellectual property. Finally, in applying this background to the
matter at hand, I also benefitted from knowledge I have acquired from many years consulting on
music industry issues.
11. A complete list of materials I relied on in formulating the opinions in this report
can be found at Appendix B.
12. Based on my analysis, I conclude that Shapley Values are an appropriate tool for
assessing rates that would be negotiated in the hypothetical marketplace for noninteractive
webcasting. The approach is consistent with the objectives laid out in the relevant legal statute
and has been accepted by the Copyright Royalty Judges in prior rate-setting proceedings. My
Shapley Value analysis finds that a reasonable royalty rate for ad-supported noninteractive
webcasting beginning in 2021 is $0.0029 per play and for subscription noninteractive webcasting
is $0.0030 per play (or $1.96 per subscriber per month). These Shapley Value results are robust
to numerous alternative model specifications, different input data, and even an alternative multi-
party bargaining framework known as Nash-in-Nash.
III. SHAPLEY VALUES AND THEIR APPLICATION TO § 114 OBJECTIVES
13. Cooperative bargaining games, in general, are focused on how “surplus” is divided
by the parties to a negotiation. Surplus is the extra value that is jointly created by a group of
Public Version
6
cooperating parties over and above what those parties could create on their own. Specifically,
each party to a negotiation has a “fallback value” which is equal to the value that party could create
on its own without an agreement with the other parties. The sum of these fallback values is the
total value the parties could generate without an agreement. The extra value the parties could
generate with an agreement is the surplus.
14. Shapley Values are a generalized solution to the problem of how to apportion
among the members of a multi-party bargaining group the surplus created by their productive
cooperation with each other. This solution divides up the surplus according to each party’s
incremental contributions to the total amount of value created.10 These contributions are assessed
as increments to every possible combination of unilateral, bilateral, and multilateral deals that may
be struck by the different parties, and then averaged across all such combinations. In the recent
Phonorecords III proceeding, the Copyright Royalty Judges credited a Shapley Value analysis as
one way of addressing concerns about complementary oligopoly power, noting that the analysis
performed in that proceeding “eliminate[d] this ‘walk away’ power by valuing all possible
orderings of the players’ arrivals.”11
15. The concept of a Shapley Value is best understood by reference to a simple analogy.
Imagine that parties A, B, and C are negotiating a deal in person. Party C can be the first, the
10 See Lloyd S. Shapley, “A value for n-person games” (1953), in Alvin E. Roth (Ed.), The Shapley value, Essays in honor of Lloyd S. Shapley, Cambridge University Press (1988): 31-40. See alsoEyal Winter, “The Shapley Value,” in R.J. Aumann and S. Hart (Eds.), Handbook of Game Theory, Vol. 3, Chapter 53, Elsevier Science (2002): 2025-2054, pp. 2027-2030; Andreu Mas-Colell, Michael D. Whinston, and Jerry R. Green, Microeconomic Theory, Oxford University Press (1995), pp. 679-684. 11 Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms for Making and Distributing Phonorecords (Phonorecords III), Docket No. 16-CRB-0003-PR (“Phonorecords III Determination”) at 33 n.69.
Public Version
7
second, or the third to arrive in the room. The value it brings to the bargaining table may be
contingent on the order in which it arrives. For example, if Party C is last to the negotiation it may
have more bargaining power as a result of its ability to hold up or frustrate consummation of a deal
to which Parties A and B are otherwise amenable. When C is first to the negotiation, it has no
bargaining power over the others. Shapley analysis takes into account all such possible differences
in Party C’s bargaining power that are contingent on its order of arrival to the negotiation. It does
so by taking the average of each “incremental value” created by Party C in each possible sequence
of arrivals. As such, Party C’s Shapley Value will only be high relative to the other parties’
Shapley Values if, on average, it brings a relatively high incremental value to all possible orderings
and sub-orderings of Parties A, B, and C.
16. As the foregoing reveals, Shapley Values incorporate principles of fairness in the
allocation of the value created by the cooperation of multiple parties.12 No party’s allocation is
affected by its preferential position in the negotiations. This is so because each party’s allocation
is determined by an averaging over all possible sequences of its incremental impacts on the group’s
12 Meinhardt (2014) formally describes these principles of fairness: “The four original principles (axioms) as introduced by Shapley (1953) can be described verbally as follows: 1. (Efficiency): the solution should distribute the maximal total payoff; 2. (Symmetry) any two players who contribute the same input should obtain the same payoff (equal treatment of equals); 3. (Dummy player): any player who contributes nothing to any coalition should obtain his value; 4. (Additivity): adding the solution of two games together produces the solution of the sum of these games (in this sense, the solution is invariant against an arbitrary decomposition of the game). The Shapley value is the unique scheme of the game v that satisfies these principles.” Holger Ingmar Meinhardt, The Pre-Kernel as a Tractable Solution for Cooperative Games, An Exercise in Algorithmic Game Theory, Springer-Verlag (2014), p. 18 (note omitted). See also, e.g., Andreu Mas-Colell, Michael D. Whinston, and Jerry R. Green, Microeconomic Theory, Oxford University Press (1995), pp. 679-680 (“[The Shapley Value] attempts to describe a reasonable, or ‘fair,’ way to divide the gains from cooperation, taking as a given the strategic realities captured by the characteristic form.”) (emphasis in the original); Hervé Moulin, Fair Division and Collective Welfare, MIT Press (2003), p. 11 (“The Shapley value is a systematic formula used to divide a joint cost or a jointly produced output. It offers a reasonable definition and computation of the share of cost or surplus for which a user of the commons is deemed responsible.”).
Public Version
8
value. The Shapley Value accorded to a party rests on the value that it brings to the group’s
cooperation, taking into account all the subsets of the group to which it can join.
17. In analyzing the outcome of bargaining among cooperating parties, therefore, it is
necessary to assess and delineate the value that can result from the cooperation of any subset of
the overall cooperating group. The tool that delineates this value for a particular subset of the
group is called the “characteristic function” of the subset. Each combination of parties (including
subsets of the cooperating group that have only one, or even zero, parties) has its own characteristic
function value.
18. To further illustrate the intuition and mechanics behind Shapley Values, consider a
more complex example—a negotiation between the owner of patentA, the owner of patentB, and
the manufacturer of deviceD. Both patentA and patentB are necessary for the manufacture and
sale of deviceD. If deviceD is not on the market, the owners of patentA and patentB would each
still make a profit of $5 (their “fallback value”), through already committed agreements with rival
manufacturers. If a cooperative deal were to be made between the parties, the market profits from
the sales of deviceD would be $16. However, because deviceD would cannibalize the sales of
competing devices, the individual net returns of the owners of patentA and patentB from licensing
to other rival devices would decrease from $5 to $3. This $2 decline in the net returns of the
owners of patentA and patentB is their “opportunity cost” from entering a deal with the
manufacturer of deviceD.
19. For the above example, the characteristic function values for each possible subset
of the cooperating group are as follows:
Public Version
9
Figure 1: Example of Characteristic Function
Notice that the characteristic function is zero if there are no parties in the subset. But it is also
zero for the group of just deviceD, because that device cannot be marketed without both patentA
and patentB. The characteristic function is $5 for a subset comprised of just the owner of patentA
or just the owner of patentB, because again that is the fallback value for each of these parties. For
the same reason, the characteristic function for the subset comprised of both patent owners, but
not the device manufacturer, is $10—the sum of each patent owner’s fallback value. And, because
deviceD cannot be marketed without both patents, the characteristic function of the subset of
deviceD and patentA, and also the subset of deviceD and patentB, is just the $5 that arises from
the royalties earned by the patent owner from rival devices. The final row displays the
characteristic function of the full group of bargaining parties. This $22 is composed of the $16
from the market profits of deviceD plus the $3 in lowered royalties earned by each of the owners
of patentA and patentB from other devices when deviceD is in the market.
20. Calculation of the Shapley Value for the manufacturer of deviceD entails assessing
the incremental value that would be created by deviceD when it joins the group of parties that
arrived at the metaphorical bargaining table before it did (the “predecessor group”), in each of the
Subset
CharacteristicFunction Value
for Subset
No parties $0
deviceD manufacturer ("D") $0
patentA owner ("A") $5
patentB owner ("B") $5
A and B $10
A and D $5
B and D $5
A, B, and D $22
Public Version
10
six possible orderings of the parties’ arrivals.13 The incremental value of deviceD is the
characteristic function of the predecessor group with deviceD added less the characteristic function
of the predecessor group alone. The figure below displays the incremental value of deviceD for
each arrival ordering:
Figure 2: Example of Shapley Value for deviceD
21. As this figure illustrates, the Shapley Value of deviceD is the average of its
incremental values across all the possible orders of arrival, namely ($12 + $0 + $12 + $0 + $0 +
$0) / 6 = $4). Analogous calculations for patentA and patentB (which, in the interest of brevity, I
do not reproduce here) show that their Shapley Values are each $9.
22. Having solved for the Shapley Values of each party to the negotiation, it is now
possible to determine the royalties that the manufacturer of deviceD must pay out. In order to
receive its Shapley Value of $9, the owner of patentA must receive royalties of $6 from the
manufacturer of deviceD, which are added to its net return of $3 in royalties from rival devices.
13 In Shapley Value analysis there are always N! (i.e., N factorial) different arrival orderings, where N is the number of negotiating parties. For example, with three negotiating parties, there are 3 x 2 x 1 = 6 different arrival orderings.
Orderof
Arrival
SubsetPrior to
Arrival of D
CharacteristicFunction Value
for SubsetPrior to D
SubsetAfter
Arrival of D
CharacteristicFunction Value
for Subset with D Added
IncrementalValue of D
A, B, D A, B $10 A, B, D $22 $12
A, D, B A $5 A, D $5 $0
B, A, D B, A $10 B, A, D $22 $12
B, D, A B $5 B, D $5 $0
D, A, B None $0 D $0 $0
D, B, A None $0 D $0 $0
Average of Incremental Values of D (“Shapley Value”): $4
Public Version
11
Likewise, the owner of patentB must also receive $6 in royalties from the manufacturer of deviceD.
That means the manufacturer must pay out a total of $12 in royalties from its market profits of
$16. Importantly, this results in the manufacturer of deviceD receiving its Shapley Value of $4.
23. Note that, in this illustration, each party is better off with an agreement than without
one. Absent an agreement, each patent owner obtains only its $5 fallback value, whereas with an
agreement where surplus value is distributed according to Shapley Values each patent owner
receives a total value of $9. Furthermore, the $6 of royalties each patent owner receives from the
manufacturer of deviceD well exceeds their opportunity costs of $2 from the cannibalization of
other income resulting from the presence of deviceD in the market. If these conditions did not
hold, the patent owners would prefer not to enter into a deal with the maker of deviceD. And
finally, observe that a deal with the owners of patentA and patentB leaves the device manufacturer
better off—without a deal, deviceD cannot be brought to market and the manufacturer makes $0.
24. As this example makes clear, Shapley Values fit within the requirements of the
relevant legal statute, which calls on the Judges to consider copyright owners’ opportunity costs.
In the course of calculating the surplus from a deal between sound recording copyright owner and
noninteractive webcasting distributor, the Shapley Value model takes into account the extent to
which such a deal might enhance or cannibalize revenue from other sources, for either party. As
such, an appropriately specified Shapley Value model takes into account the extent to which
noninteractive distributors “substitute for or may promote the sales of phonorecords or otherwise
may interfere with or may enhance the sound recording copyright owner’s other streams of revenue
from [the copyright owner’s] sound recordings.”14 Furthermore, the model’s assessment of
fallback values necessarily takes account of the “relative roles of the copyright owner and the
14 17 U.S.C § 114(f)(2)(B).
Public Version
12
transmitting entity in the copyrighted work and the service made available to the public with
respect to relative creative contribution, technological contribution, capital investment, cost, and
risk.”15 Thus, it is my opinion that Shapley Values are an appropriate approach for assessing rates
that would be negotiated in the hypothetical marketplace for noninteractive webcasting.
IV. A SHAPLEY VALUE MODEL FOR RECORD COMPANIES AND NONINTERACTIVE WEBCASTERS
A. Defining the Model
25. Shapley Values depend upon how the parties to the hypothetical negotiation are
delineated. My representation of the hypothetical negotiation between record companies on the
one hand and noninteractive webcasting distributors on the other involves four upstream record
companies and two downstream noninteractive webcasting distributors. The first three record
companies represent each of the three major record companies (Universal Music Group, Sony
Music Entertainment, and Warner Music Group) and the fourth represents the combination of all
other record companies (i.e., the indies). The two noninteractive webcasting distributors represent
the combination of all ad-supported (i.e., free) noninteractive distributors and the combination of
all subscription noninteractive distributors, respectively.
26. It is my opinion that defining the hypothetical negotiations among these six players
is an appropriate elaboration of a negotiation between record companies on the one hand and
noninteractive webcasters on the other. This specification strikes a balance between offering a
granular and realistic depiction of the hypothetical market and maintaining enough simplicity
around the number of entities being modeled such that the model can be readily solved and
necessary data inputs can be estimated. First, delineating each of the major record companies
15 Id.
Public Version
13
individually along with the amalgam of indies accounts for the potentially different negotiating
positions of each. Second, delineating ad-supported noninteractive distributors separately from
subscription noninteractive distributors allows for the assessment of whether each form of
distribution should yield different royalty rates, as was found to be the case by the Judges in the
previous Web IV proceeding.16 Moreover, grouping all ad-supported noninteractive distributors
together and all subscription noninteractive distributors together is conservative from the
perspective of SoundExchange (i.e., produces lower royalty rates) because this modeling choice,
in effect, simplifies away rivalry among the various extant ad-supported noninteractive distributors
and among the various extant subscription noninteractive distributors. By eliminating
consideration of competition within these groups of distributors, their respective market power in
the negotiation is elevated.17 At the same time, the specification carefully models competition
between the collective ad-supported noninteractive distributor and the collective subscription
noninteractive distributor.
27. As noted, the characteristic function defines the value created by each possible
cooperating grouping of these six parties to the hypothetical negotiation (i.e., the four record
companies and two noninteractive distributors). Next, I describe considerations specific to
determining the value that each party or set of parties brings to the hypothetical negotiation.
28. The value brought by each record company is a function of both the costs it incurs
and the revenue it could generate by licensing its sound recordings to distributors other than the
noninteractive services (the record company’s “fallback value” as in the illustrations above).
There are a variety of other forms of music distributors, which are not involved in this hypothetical
16 Web IV Determination at 1. 17 Phonorecords III Determination at 66-67.
Public Version
14
negotiation, that pay royalties that are determined outside the ambit of this matter. These include:
ad-supported interactive (i.e., “on demand”) music and video services (e.g., free Spotify and free
YouTube); subscription interactive (i.e., “on demand”) music and video services (e.g., paid
Spotify, Apple Music, Amazon Music, YouTube Premium, etc.); SiriusXM satellite radio; and
physical/digital music purchases (e.g., digital downloads and CD album sales). These “outside”
distributors are not modeled as parties to the hypothetical bargaining process that will arrive at the
royalty rates to be determined in this matter. Each of their respective royalty rates are taken as
they actually are or are expected to be. Also in the background are the options of listening to
broadcast AM/FM radio or not listening to music, which are modeled realistically as not producing
any royalties for the record companies.
29. Determining a record company’s fallback value—which, again, is a key
component of the value it brings to the hypothetical negotiation—also requires evaluating what
would happen if each noninteractive distributor did not have access to that record company’s
music. If the hypothetical deal is not consummated, how much of each noninteractive distributor’s
audience would divert to other music listening options (including to the other noninteractive
distributor)? To answer this question, my analysis needs to consider the size of the audience of
each noninteractive distributor. My analysis then needs diversion parameters that represent the
proportion of these audiences that would divert to each alternative mode of distribution. And
finally, because the value each record company brings to the negotiation varies depending on its
relative share of overall music listening, my analysis needs an empirical input on each record
company’s respective share of plays.
30. The value brought by the noninteractive distributors to the hypothetical negotiation
depends on their ability to generate profits (before paying any sound recording royalties), which
Public Version
15
subtracts out various costs including the copyright royalties for musical works that distributors
must pay to publishers. Distributor profits are measured on a per-play basis.
31. The profits the noninteractive distributors create vary depending on which record
companies are present in each grouping of negotiating parties. Specifically, each of the three major
record companies is taken to be a “must-have,” meaning that the sound recording collection of
each is necessary for a noninteractive distributor to operate. To reference the illustration from
above, the major record companies are like the owners of patentA and patentB—without their
agreement to license, the manufacturer of deviceD cannot bring its product to market. This “must-
have” specification follows from the Copyright Royalty Judges’ conclusion in the Web IV
proceeding that “[t]here appears to be a consensus that the repertoire of each of the three Majors
is a ‘must have’ in order for a noninteractive service to be viable.”18 Hence, within the Shapley
Value model, without access to the sound recordings of all three of the major record companies, a
noninteractive distributor does not operate and contributes zero profits to the rest of the subset of
the bargaining parties. In contrast, the record company representing the indies does not have a
must-have repertoire. Instead, the model assumes that without access to the music of indies, a
distributor’s profits decline not to zero, but to an extent determined by listeners’ preferences for
the content carried by indies.
32. The empirical parameters described above are listed in Appendix C. These
parameters enable the algebraic specification of the characteristic function, which is derived and
displayed in that same appendix.
33. Once the characteristic function is specified, Shapley Values can be derived for
each party to the hypothetical negotiation, which is also accomplished in Appendix C. Recall
18 Web IV Determination at 133.
Public Version
16
Figure 2 above. The difference between the characteristic function for a subset of the parties
without the distributor and the characteristic function for that subset with the distributor added is
the “incremental value” that the distributor brings to the negotiation. By definition, the Shapley
Value for a distributor is the weighted average of the incremental values that the distributor brings
to every possible predecessor subset of the other bargaining parties.19
34. Finally, once Shapley Values are derived, the corresponding royalties from the two
noninteractive distributors to the record companies can be computed. These are the payments that
result in each party’s bottom line equaling its Shapley Value. For each distributor, the total royalty
payments it makes to the record companies must equal the difference between its profits from its
market operations and its Shapley Value. For each record company, the total royalty payments it
receives must equal the difference between its Shapley Value and the total compensation it receives
from its other sources of distribution, less its costs of operation.
B. Empirical Inputs
35. Under this Shapley Value analysis, royalty rates for both ad-supported and
subscription noninteractive distributors are a function of the following data inputs: (i) royalty rates
that record companies earn from other forms of music distribution; (ii) noninteractive distributors’
audience sizes; (iii) diversion ratios reflecting the amount of a noninteractive distributor’s audience
that would switch to other forms of music distribution and generate royalties if that noninteractive
distributor were unavailable; (iv) record company play shares; and (v) noninteractive distributors’
fixed costs and marginal profit rates. Because the hypothetical negotiation will set royalty rates
for the 2021-2025 period, this is the relevant time period for assessing these parameter values. In
19 The weighted average accords equal weight to the incremental value of the distributor to the group of parties that arrived before the distributor arrives in any one of the possible sequences of arrivals.
Public Version
17
other words, these parameters should reflect forward looking estimates to the greatest extent
possible. I discuss the estimation of each of these input parameters below. Additional details of
these data estimates are contained in Appendix D.
i. Royalties for Other Forms of Distribution
36. Rather than speculate about the path that royalty rates on other forms of music
distribution will take in the future, I use currently observable sound recording royalty rates as a
proxy for the sound recording royalty rates that will prevail during the relevant 2021-2025 period.
These rates are estimated using a variety of sources.
37. First, royalty rates for subscription on-demand streaming music and video services
(e.g., Spotify Premium, Apple Music, Amazon Music, YouTube Premium, etc.) are determined
from royalty statements that detail the royalty payments these services paid to various record
companies for each month over the 12-month period ending March 2019.20 The weighted average
amount of monthly royalties these services paid was approximately [ ] per subscriber. See
Appendix D at Ex. D.1.
38. Second, royalty rates for ad-supported on-demand streaming music and video
services (e.g., Spotify Free and free YouTube) are similarly determined from royalty statements.
The weighted average amount of royalties these services paid over the 12-month period ending
March 2019 was approximately [ ] per play. See Appendix D at Ex. D.2.
39. Third, royalty rates for SiriusXM satellite radio broadcasts are determined from (i)
Statements of Account provided by SiriusXM to SoundExchange detailing the total amount of
statutory royalties paid, over the 12-month period ending March 2019, for performances on
20 This is the most recent period of four consecutive calendar quarters for which royalty statements were consistently available for all services.
Public Version
18
satellite radio and from (ii) SiriusXM’s SEC Forms 10-K and 10-Q which detail its subscriber
counts. Specifically, SoundExchange reported total SDARS royalties of ] for the
12-month period ending March 2019.21 SiriusXM’s 10-K and 10-Q filings reported the quarter-
end subscriber counts shown below.
Figure 3: SiriusXM Average Subscribers
From these amounts I compute quarterly average subscriber counts, and an average subscriber
count of 33.7 million for the full 12-month period ending March 2019, as shown in Figure 3.
Converting the ] of annual royalties to an average monthly amount and dividing by
the 33.7 million subscriber count results in SiriusXM monthly royalties per subscriber of [ ].
Importantly, given the data currently available to me, these royalty rates are artificially
conservative, as they do not account for royalty payments that I understand are paid by SiriusXM
outside of SoundExchange, for example through directly negotiated licenses with certain record
companies. Nor do these royalty rates account for payments that I understand SiriusXM must now
make for its performance of sound recordings fixed prior to February 15, 1972 (which, until very
21 SoundExchange Exhibits 32-33.
Ending AverageSubscribers Subscribers
Quarter (millions) (millions)1Q18 33.12Q18 33.5 33.33Q18 33.7 33.64Q18 34.0 33.91Q19 34.2 34.1
Average: 33.7
Source: SiriuxXM 10-K and 10-Q filings.
Public Version
19
recently, were not royalty-bearing under federal law).22
40. Fourth, royalty rates for physical music purchases (e.g., CDs and vinyl records) and
digital downloads are determined from 2018 wholesale and retail sales data from the Recording
Industry Association of America (“RIAA”) and from a 2018 Annual Music Study by industry
research firm MusicWatch prepared for the RIAA. According to MusicWatch, in 2018, CD
purchasers spent an average of [ ] on CDs, vinyl purchasers spent an average of [ ],
and digital download purchasers spent an average of [ ].23 Moreover, as shown in Appendix
D at Ex. D.3, RIAA data indicate that, in 2018, sound recording royalties totaled approximately
[ ] of the retail prices of CDs, vinyl, and digital downloads, respectively.
Multiplying these data and converting annual amounts to monthly amounts produces average
monthly royalties per purchaser of [ ] for CDs, [ ] for vinyl, and [ ] for digital
downloads, or a weighted average of [ ] across all forms of physical and digital download
distribution. See Appendix D at Ex. D.3.
41. Fifth, royalty rates for AM/FM broadcasts (and all other forms of outside music
distribution not listed above) are assumed to be zero.
42. Figure 4 below summarizes all of these royalty rates.
22 Until recently, I understand performances of sound recordings fixed before February 15, 1972 were not royalty-bearing under federal law. It is my understanding that changed late in 2018, when President Trump signed into law the Music Modernization Act. Given that the royalty statements I reviewed covered a time period when performances of pre-1972 recordings were not compensable under federal law, that data understates the revenue that SiriusXM can be expected to pay going forward. 23 SoundExchange Exhibit 42 at 19.
Public Version
20
Figure 4: Royalty Rates for Outside Distributors (RESTRICTED)
ii. Noninteractive Distributors’ Audience Sizes
43. Audience sizes for the noninteractive distributors are measured as total numbers of
plays per month. For Pandora, I am able to estimate projected audience sizes from its publicly
reported financial projections (which are discussed in more detail at ¶¶ 49-50 infra and Appendix
D at Ex. D.6). Further, I assume Pandora will have the same share of the noninteractive ad-
supported and subscription markets in 2021-2025 as it did over the recent 12-month period ending
March 2019. Royalty statements and data from SoundExchange reveal that Pandora has an
approximately [ play share of the ad-supported noninteractive market and an [ ] play
share of the subscription noninteractive market. See Appendix D at Ex. D.4. Using those shares,
I am able to gross up the Pandora audience size amounts to reflect the total size of the audience
for all noninteractive distributors.
44. Accordingly, projected audience sizes can be estimated as follows:
Public Version
21
Figure 5: Estimated Average Audience Size (2021-2025) (RESTRICTED)
45. Last, my analysis requires converting these play counts into per-user metrics for
interaction with the per-user information derived from the survey results described infra. To do
so, I compare projected average monthly play counts for Pandora’s ad-supported tier to the
projected number of active users of Pandora’s ad-supported service. Similarly, for Pandora’s
subscription noninteractive tier, I compare projected average monthly play counts on that tier to
the projected number of subscribers. Again, these figures are derived from Pandora’s public
financial projections. See Appendix D at Ex. D.6. I find that users of Pandora’s ad-supported
service are projected to listen to approximately ] plays per month and subscribers to Pandora’s
subscription noninteractive service (i.e., Pandora Plus) are projected to listen to approximately
[ ] plays per month over the 2021-2025 period.
iii. Diversion Ratios
46. Diversion ratios to other forms of music distribution if a noninteractive distributor
were no longer available in the marketplace are estimated from the results of a survey conducted
under the direction of Gal Zauberman and thoroughly described in his report (“Zauberman
Public Version
22
Survey”).24 The survey obtained responses from hundreds of users of both ad-supported and
subscription noninteractive services. Survey respondents were first asked how they currently listen
to music and then how they would listen to music under hypothetical scenarios where
noninteractive services are no longer available. Some pertinent details of the survey design and
responses are provided in Appendix E. In addition, Appendix E details how I use these survey
responses to derive diversion ratios between noninteractive distributors and from noninteractive
distributors to outside forms of music distribution, including subscription and ad-supported
interactive streaming music and video services, satellite radio, physical album purchases and
digital downloads, and broadcast AM/FM radio.
47. In Figure 6 below, I summarize these diversion ratios in Panel A, which shows the
percentage of survey respondents that would respond to the absence of noninteractive services by
obtaining new music subscriptions, beginning to purchase CDs/digital downloads, or consuming
additional monthly plays on ad-supported streaming services. Panel B shows the corresponding
levels of royalty compensation that the record companies would earn as a result from these various
forms of music distribution. Finally, Panel C multiplies the diversion rates in Panel A by the
royalty rates in Panel B to obtain an estimate of the opportunity cost that record companies
experience by licensing to noninteractive distributors instead of only licensing to all the outside
forms of music distribution. (AM/FM radio is not shown below as it generates zero royalties and
thus zero opportunity cost.)
24 See Written Direct Testimony of Professor Gal Zauberman, Sept. 23, 2019 (“Zauberman WDT”).
Public Version
23
Figure 6: Diversion Ratios and Opportunity Cost (RESTRICTED)
iv. Record Company Play Shares
48. Music content shares for each of the major record companies and for the indies
together as a group are estimated from the royalty statements that music streaming and video
services provide to record companies when operating under directly negotiated license
agreements.25 These monthly royalty statements denote both the total plays on that service in any
given month as well as the number of those plays that were copyrighted recordings of the record
company. I have no basis to believe record company shares are anticipated to be substantially
different during the 2021-2025 period than they are currently. Thus, I use the most recent monthly
25 Play counts for services operating under statutory noninteractive licenses are not included. However, all of Pandora’s noninteractive play counts and iHeart’s subscription noninteractive play counts are included as these services currently operate under directly negotiated licenses and, accordingly, report their total play counts to the major record companies.
Public Version
24
royalty statements available from interactive streaming music and video services and
noninteractive services, for the 12-month period ending March 2019.26 These statements allow
me to calculate the proportion of total plays of Universal-, Sony-, Warner-, and indie-owned
content, respectively. Specifically, I compute each major record company’s play shares, and then
compute the play share for the combination of all indies as a group as 100% less the sum of the
major record companies’ shares.27 See Appendix D at Ex. D.5. The resulting play shares are
summarized in the figure below:
Figure 7: Estimated Play Shares (RESTRICTED)
v. Noninteractive Distributor Fixed Costs and Marginal Profit Rates
49. As noted above, the value brought by the noninteractive distributors to the
hypothetical negotiation depends on their ability to generate profits (before paying any sound
recording royalties), which subtract out from revenues various costs including the copyright
26 I evaluate all tiers of service on the following services: Apple Music, Amazon Music Unlimited, Amazon Prime, Google Play, iHeart (both interactive and noninteractive tiers), Pandora (both interactive and noninteractive tiers), Napster, Spotify, Vevo, and YouTube. Play share data from SiriusXM, and from physical retail and digital downloads, were not available to me. I have no reason to think the content of any of the record companies is played with more or less frequency on these distribution methods, when compared to the distribution methods (interactive and noninteractive streaming) for which I did have data. Thus, I have no reason to believe this additional data would materially change the respective shares of plays among Universal, Sony, Warner, and the indies. 27 InGrooves and The Orchard, wholly owned subsidiaries of Universal and Sony, respectively, are treated as part of the group of indies. Treating them instead as part of their affiliated must-have major record companies has a de minimis effect on the royalty rates derived from my Shapley Value model.
Public Version
25
royalties for musical works that distributors must pay to publishers. See supra ¶ 30. To estimate
fixed costs, variable or marginal costs and the associated marginal profit rates of noninteractive
distributors, I rely on Pandora financials as a proxy. Pandora is the only noninteractive service for
which I have been able to obtain financial data that allow for forward-looking estimates of ad-
supported and subscription noninteractive service profit rates.28 Furthermore, Pandora is the
largest noninteractive distributor in the market, accounting for more than [ ] of total plays in
the noninteractive market (see Appendix D at Ex. D.4).
50. I use the forward-looking estimates of Pandora’s costs and profits for the relevant
2021-2025 period disclosed in the company’s recent Merger Proxy.29 These projections were
disclosed to Pandora shareholders prior to their vote to approve SiriusXM’s acquisition of the
company and were utilized by Pandora’s investment bankers as an input into their fairness
opinions. As described in greater detail in Appendix D, using these projections I estimate the
following monthly fixed costs and marginal profit rates:
28 Pandora now offers a fully interactive subscription tier, Pandora Premium, whose data must be removed from Pandora’s financial results to obtain estimates of ad-supported and subscription noninteractive services’ profit rates. 29 As explained in Appendix D, estimating marginal profit rates from Pandora’s recent historical financials would result in figures with an inappropriate downward bias with respect to the hypothetical negotiation because the company’s recent financial results show substantial operating losses. Notwithstanding these recent losses, it is clear that Pandora is anticipated to be profitable in the future, as demonstrated by the approximately $3.5 billion purchase price paid by SiriusXM to acquire the company, as well as Pandora’s substantial market capitalization of approximately $2.4 billion immediately prior to the announcement of the SiriusXM acquisition. Thus, Pandora’s recent history of operating losses does not accurately reflect expectations about the incremental value Pandora is anticipated to bring to the hypothetical negotiation concerning royalty rates for the 2021-2025 period.
Public Version
26
Figure 8: Estimated Fixed Costs and Marginal Profit Rates (2021-2025)30
C. Summary of Results
51. Using the formulas derived in Appendix C and the data inputs discussed above, I
derive Shapley Values and corresponding royalty rates for ad-supported and subscription
noninteractive distributors. These results are summarized below:
30 As explained in Appendix D, certain key inputs to the Pandora projections were not disclosed in Pandora’s proxy statements (e.g., projected ad-supported user and subscriber counts, projected plays, and a breakdown of subscription revenue into its underlying Pandora Plus and Pandora Premium component parts). Accordingly, certain allocation assumptions were required to estimate key parameters from Pandora’s projected financial information. Estimates derived from these projections may require amendment following the completion of discovery.
Ad-SupportedNoninteractive
SubscriptionNoninteractive
Fixed Costs (per month) $40.4 million $8.9 million
Marginal Profit Rate $0.0042 per play $0.0048 per play
Public Version
27
Figure 9: Estimated Shapley Values and Royalty Rates for Noninteractive Distributors31
(amounts are monthly except per play rates) (RESTRICTED)
Notably, the resulting Shapley Value for the ad-supported noninteractive distributor is near zero.
This is the case because, as Figure 9 shows, the record companies’ opportunity costs are high
relative to the total projected profits of those distributors (before paying the royalties at issue).32
31 I calculate a 95 percent confidence interval for these royalty rates using a bootstrap procedure on the survey results. The details of this computation are described in Appendix E. The resulting 95 percent confidence interval ranges from $0.00291 to $0.00299 for the ad-supported noninteractive royalty rate and from $0.00299 to $0.00317 for the subscription noninteractive royalty rate. 32 Note that the total opportunity costs shown in Figure 9 exceed the opportunity cost subtotals from Figure 6 because they include not only the portion of opportunity cost associated with diversion to outside distributors, but also the additional portion of opportunity cost associated with diversion to the other noninteractive distributor. (The record companies’ total opportunity cost associated with licensing sound recordings to the ad-supported noninteractive distributor depends, in part, on the amount by which the market operations of ad-supported noninteractive distributors cannibalize the audience of subscription noninteractive distributors, and vice versa.) These additional amounts of opportunity cost are computed as the diversion of plays from one
Public Version
28
Hence, the vast majority of those profits are necessary to compensate the record companies for the
ad-supported noninteractive distributors’ cannibalization of listeners that would otherwise
consume music via other compensatory forms of music distribution.33
noninteractive distributor to the other (as reflected in Panel A of Figure 6) times the relevant per play royalty rates shown in Figure 9. 33 This finding does not exclude the possibility that the record companies and distributors of ad-supported noninteractive services with multiple service tiers would find it mutually advantageous to enter into individual agreements that provide lower royalty rates than those available under the statute in exchange for commitments to convert listeners to their affiliated subscription tiers that pay higher royalty rates.
For example, Spotify, an on-demand streaming service, has shown an ability to convert users from its ad-supported on-demand tier to its subscription on-demand tier. As explained by its CFO in September 2018, approximately half of the users of its ad-supported service that remain engaged with the service ultimately become paying subscribers. See SoundExchange Exhibit 66 (Transcript of Goldman Sachs Communacopia Conference, Sept. 14, 2018). For this ad-supported tier, Spotify has been able to negotiate paying record companies an effective royalty rate of approximately [ ] per play (on average over the 12-month period ending March 2019). See Appendix D at Ex. D.2.
[
Even though many ad-supported noninteractive distributors do not have subscription service tiers, I nonetheless test whether Pandora’s financial projections imply significant conversions of Pandora Free users to subscribers, and whether such conversion rates would affect the royalty rates specified by my model. Using Pandora’s public projections for its streaming services for 2021 to 2025, see Appendix D at Ex. D.6, I conservatively assume that any differences in the projected growth rates of Pandora Plus and Pandora Premium as compared to Pandora Free are due solely to the conversion of Pandora Free users to one of Pandora’s subscription services, and that this conversion would not occur if Pandora Free were not operating as a result of failing to reach a deal in the hypothetical negotiation with one or more of the must-have record companies. In effect, such conversions offset and reduce the opportunity cost to the record companies associated with the market operations of Pandora’s ad-supported service. I find that Pandora’s financial projections imply a relatively low rate of conversion of ad-supported listeners to paying subscribers (as compared to that implied by the Spotify CFO’s comments above). Accounting for
Public Version
29
52. The finding that the record companies’ opportunity costs are high relative to the
profits of noninteractive distributors is perhaps not surprising in light of internal documents from
the services that [
].35, 36
53. The finding that ad-supported noninteractive services, on their own, currently bring
little (if any) incremental value to the hypothetical negotiation is also consistent with the testimony
of record company executives. For example, Mark Piibe (Executive Vice President for Global
Business Development and Digital Strategy at Sony Music Entertainment) explains:
As far as I know, [
this conversion in a dynamic Shapley Value model does not substantially impact royalty rates for ad-supported noninteractive distributors. 34
. 35 [
)]. 36 [
.
Public Version
30
.37
Similarly, Aaron Harrison (Senior Vice President, Business & Legal Affairs, Digital, UMG
Recordings, Inc.) explains: “Ordinarily, UMG only sees value in a standalone ad-supported audio
streaming service where the rates are sufficiently high on their own to justify licensing our content.
To that end, [
].”38
Moreover, in describing the value brought by subscription noninteractive services, Mr. Harrison
states that “even if mid-tier subscription services succeed in drawing some consumers away from
poorly-monetized ad-supported streaming services, there is also a danger that they could
cannibalize the premium on-demand subscription services” and thus [
]39
54. Finally, it is notable that streaming services themselves describe the ad-supported
37 Written Direct Testimony of Mark Piibe, Sept. 23, 2019. 38 Written Direct Testimony of Aaron Harrison, Sept. 23, 2019. 39 Id.
Public Version
31
tier not as an independent generator of meaningful revenue, but as a “funnel” to higher-priced
subscription tiers. Spotify’s 2018 annual report states that “[o]ur Ad-Supported Service serves as
a funnel, driving more than 60% of our total gross added Premium Subscribers since we began
tracking this data in February 2014.”40 Similarly, [
].42
55. In summary, my Shapley Value model indicates a reasonable royalty rate for ad-
supported noninteractive distributors of $0.0030 per play over the 2021-2025 period, and for
subscription noninteractive distributors of $0.0031 per play (or $2.04 per subscriber per month)
over that period. Notably, these royalty rates are estimated using projected data for the 2021-2025
period, and as such, reflect “on average” rates for the entire five-year period. To the extent it is
appropriate to express a rate as of the beginning of this period (such that the rate can be adjusted
over the 2021-2025 period based on increases or decreases in the general price level, as is the case
for the current statutory rates that were determined in the Web IV proceeding), these rates can be
discounted back two years (from the mid-point of 2023 to the mid-point of 2021) using a discount
rate equal to the rate of inflation. A two percent estimate of inflation results in a discount factor
40 SoundExchange Exhibit 71 (Spotify 2018 Form 20-F) at 38. 41 [
].
42 [.
Public Version
32
of 0.96117.43 Multiplying this discount factor by the above rates produces a 2021 royalty rate of
$0.0029 per play for ad-supported noninteractive services and $0.0030 for subscription
noninteractive services (or $1.96 per subscriber per month).
D. Alternative Shapley Value Calculations Using “Share of Ear” Data
56. As an alternative to the Shapley Value computations presented above, I also
consider an alternative source of data from which diversion ratios may be estimated. For this
alternative calculation I use data from Edison Research, an industry research firm, instead of the
data from the Zauberman Survey.
57. Edison Research publishes a quarterly study, known as “Share of Ear,” that
analyzes the share of time Americans spend listening to all different forms of music distribution.
The study is based on a survey that asks several thousand respondents aged 13 and over who own
smartphones to track their music listening behavior.44 I have obtained the most recent Q2 2019
Share of Ear research report.
58. I apply a logit demand model to the data from this report, with the implication that
if ad-supported or subscription noninteractive distributors were not available in the marketplace,
their audiences would divert to other forms of music listening in the same relative proportions as
those reflected in the share of listening time figures reported in the Share of Ear study. Because
43 The U.S. Federal Open Market Committee’s inflation rate forecast for 2021 is two percent. See SoundExchange Exhibit 22. 0.96117 = 1 / (1+.02)^2.44 The Share of Ear study also tracks the U.S. population more broadly by including non-smartphone owners, but I rely on the smartphone owner subset (approximately [ ] of the total population) as this subset of respondents seems to more closely represent the behavior of the relevant group of noninteractive streaming users than would the total U.S. population at large. Regardless, my alternative Shapley Value calculations using Share of Ear Data do not differ substantially if the total population of Share of Ear respondents is used instead of the smartphone owner subset.
Public Version
33
these diversion ratios are specified in terms of listening time, instead of the number of listeners as
in the Zauberman Survey, I modify my Shapley Value model accordingly to allow for the
incorporation of these alternative diversion input parameters. Appendix F provides a complete
description of how I estimate diversion ratios from the Edison Research 2Q 2019 Share of Ear
study and revise my Shapley Value model to incorporate these data.
59. Applying the Share of Ear diversion parameters produces the following Shapley
Value results:45
Figure 10: Alternative Royalty Rates Using Share-of-Ear-Based Diversion Rates (amounts are monthly except per play rates) (RESTRICTED)
60. As can be seen in the table above, the royalty rates determined by the Shapley Value
model using the alternative Share of Ear data are highly consistent with those determined from the
data generated by the Zauberman Survey. Incorporating the discount factor discussed above to
45 Appendix F describes the robustness of these results to various alternative specifications.
Public Version
34
convert these amounts to 2021 values produces an ad-supported noninteractive rate of $0.0028 per
play and a subscription noninteractive rate of $0.0030 per play ($1.94 per subscriber per month).
E. Alternative Calculations Using a Nash-in-Nash Bargaining Model
61. Shapley Value analysis is not the only approach embraced by the economics
literature for modeling a multi-party negotiation. “Nash-in-Nash bargaining” is another such
framework recently articulated as a model of market outcomes without reliance on fairness. I
apply this alternative approach to further test the sensitivity of my Shapley Value results.
62. In Nash-in-Nash bargaining, the parties negotiate in pairs, and each pair’s
agreement may affect the equilibrium outcomes of the other negotiating pairs.46 Determination of
the Nash-in-Nash solution involves two distinct steps. In the first step, the bargaining solution
between each negotiating pair is modeled with the “Nash Bargaining Solution”47 concept, whereby
each party is accorded its fallback value plus one half of the surplus created by the deal.48 During
this negotiation, the two parties assume that all other pairs of parties have reached (or will reach)
an equilibrium agreement. Thus, this first step determines an equation representing the Nash
Bargaining Solution for each negotiating pair, where the resulting solution is a function of the
46 See, e.g., Allan Collard-Wexler, Gautam Gowrisankaran, and Robin Lee, “‘Nash-in-Nash’ Bargaining: A Microfoundation for Applied Work,” Journal of Political Economy, Vol. 127, No. 1 (2019): 163-195, p. 165. 47 The Nash Bargaining Solution is a widely-accepted theory used to characterize the outcomes of bilateral bargaining between willing buyers and willing sellers. In 1994, John Nash was awarded the Nobel Prize in economics for his work. See, e.g., John Nash, “The Bargaining Problem,” Econometrica, Vol. 18, No. 2 (1950): 155-162. See also Henrick Horn and Asher Wolinsky, “Bilateral monopolies and incentives for merger,” RAND Journal of Economics, Vol. 19, No. 3 (1988): 408-419, pp. 411-412; Ian McDonald and Robert Solow, “Wage Bargaining and Employment,” American Economic Review, Vol. 71, No. 5 (1981): 896-908, p. 905. 48 It is these fallback payoffs that represent and cause disparities in bargaining power notwithstanding the even split of the surplus, as discussed below.
Public Version
35
solutions from every other negotiating pair. In the second step, a unique solution, known as a Nash
Equilibrium, is found where there is no negotiating pair with an incentive to change its
agreement.49 This is done by simultaneously solving the system of equations generated in the first
step.
63. In short, the Nash-in-Nash bargaining solution models each potential pair of
bargainers by assuming that each bargaining pair correctly anticipates that all other pairs have each
reached or will reach their equilibrium agreements, i.e., agreements that will not change with the
outcome of the current bargaining pair’s negotiation. In multi-party Nash-in-Nash bargaining, the
parties’ relative bargaining strengths in their negotiations are set by their fallback values, which
are generally affected by their own and other parties’ proposed deals in the marketplace, while the
surplus generated by each pair of negotiating parties relative to their fallback values is split on a
50/50 basis. For this reason, empirical work in economics often uses Nash-in-Nash bargaining to
model competitive outcomes in multi-party settings.50
64. The 50/50 split of surplus that is applied in Nash-in-Nash bargaining has been
mischaracterized by some as effecting an overall “50/50 split” between the parties to a negotiation.
However, this view fails to capture the very real sense in which the Nash Bargaining Solution can
result in an unequal allocation of value between two parties, depending on their respective
bargaining positions. Again, the relative bargaining positions of the parties are reflected in their
49 See, e.g., Jean Tirole, The Theory of Industrial Organization, MIT Press (1988), p. 206. 50 See, e.g., Gautam Gowrisankaran, Aviv Nevo, and Robert Town, “Mergers When Prices Are Negotiated: Evidence from the Hospital Industry,” American Economic Review, Vol. 105, No. 1 (2015): 172-203. See also Gregory S. Crawford, Robin S. Lee, Michael D. Whinston, and Ali Yurukoglu, “The Welfare Effects of Vertical Integration in Multichannel Television Markets,” Econometrica, Vol. 86, No. 3 (2018): 891-954, p. 910. (The paper cites multiple studies that have employed the “Nash-in-Nash” bargaining solution.).
Public Version
36
often disparate fallback values. When fallback values are not the same, the Nash Bargaining
Solution employed in Nash-in-Nash will result in something other than a 50/50 split of the total
value generated by the cooperating parties. Only the surplus, which requires both parties’ equal
cooperation, is split on a 50/50 basis, which is economically appropriate for the simple reason that
no party could obtain any of the surplus without equal participation by the other.
65. In Appendix G, I provide an illustration of the Nash-in-Nash bargaining approach
and then algebraically derive my Nash-in-Nash bargaining model using the same parameters as
employed in the Shapley Value model.
66. Application of the same input data used in the Shapley Value model based on the
Zauberman Survey produces the quantitative solution to the Nash-in-Nash model shown below.
Figure 11: Estimated Royalty Rates Using Alternative Nash-in-Nash Model (amounts are monthly except per play rates) (RESTRICTED)
67. As can be seen in this figure, the Nash-in-Nash model produces similar royalty rates
to those obtained from the Shapley Value model. Incorporating the discount factor discussed
above to convert these amounts to 2021 values produces an ad-supported noninteractive rate of
Public Version
37
$0.0029 per play and a subscription noninteractive rate of $0.0029 per play ($1.89 per subscriber
per month).
V. CONCLUSION
68. As explained in this report, it is my conclusion that Shapley Values are an
appropriate tool for assessing rates that would be negotiated in the hypothetical marketplace for
noninteractive webcasting. The approach is consistent with the objectives laid out in the relevant
legal statute and has been accepted by the Copyright Royalty Judges in prior rate-setting
proceedings. My Shapley Value analysis finds that a reasonable royalty rate for ad-supported
noninteractive webcasting beginning in 2021 is $0.0029 per play and for subscription
noninteractive webcasting is $0.0030 per play (or $1.96 per subscriber per month). These Shapley
Value results are robust to numerous alternative model specifications, input data, and bargaining
frameworks.
Public Version
I declare under penalty of perjury that the foregoing testimony is true and correct.
,*", (<tlrqobert Willig
Public Version
A-1
Appendix A: Curriculum Vitae and Testimony List
September 23, 2019Curriculum Vitae
Name: Robert D. Willig
Address: 220 Ridgeview Road, Princeton, New Jersey 08540
Birth: 1/16/47; Brooklyn, New York
Marital Status: Married, four children
Education: Ph.D. Economics, Stanford University, 1973 Dissertation: Welfare Analysis of Policies
Affecting Prices and Products. Advisor: James Rosse
M.S. Operations Research, Stanford University, 1968.
A.B. Mathematics, Harvard University, 1967.
Professional Positions:
Lecturer with the rank of Professor, Woodrow Wilson School of Public and International Affairs, 2/2017 – 6/2020.
Professor of Economics and Public Affairs, Emeritus, Princeton University, 7/2016 –
Professor of Economics and Public Affairs, Princeton University, 7/1978 - 6/2016.
Principal External Advisor, Infrastructure Program, Inter-American Development Bank, 6/97- 8/98.
Deputy Assistant Attorney General, U.S. Department of Justice, l989-1991.
Supervisor, Economics Research Department, Bell Laboratories, 1977-1978.
Visiting Lecturer (with rank of Associate Professor), Department of Economics and Woodrow Wilson School, Princeton University, 1977-78 (part time).
Public Version
A-2
Economics Research Department, Bell Laboratories, 1973-77.
Lecturer, Economics Department, Stanford University, 1971-73.
Other Professional Activities
ABA Section of Antitrust Law Economics Task Force, 2010-2012
Advisory Committee, Compass Lexecon 2010 -
OECD Advisory Council for Mexican Economic Reform, 2008 - 2009
Senior Consultant, Compass Lexecon, 2008 -
Director, Competition Policy Associates, Inc., 2003-2005
Advisory Bd., Electronic Journal of I.O. and Regulation Abstracts, 1996-2008.
Advisory Board, Journal of Network Industries, 2004-2010.
Visiting Faculty Member (occasional), International Program on Privatization and Regulatory Reform, Harvard Institute for International Development, 1996-2000.
Member, National Research Council Highway Cost Allocation Study Review Committee, 1995-98.
Member, Defense Science Board Task Force on the Antitrust Aspects of Defense Industry Consolidation, 1993-94.
Editorial Board, Utilities Policy, 1990-2001.
Leif Johanson Lecturer, University of Oslo, November 1988.
Member, New Jersey Governor's Task Force on Market-Based Pricing of Electricity, 1987-89.
Co-editor, Handbook of Industrial Organization, 1984-89.
Associate Editor, Journal of Industrial Economics, 1984-89.
Director, Consultants in Industry Economics, Inc., 1983-89, 1991-94.
Fellow, Econometric Society, 1981-.
Organizing Committee, Carnegie-Mellon-N.S.F. Conference on Regulation, 1985.
Public Version
A-3
Board of Editors, American Economic Review, 1980-83.
Nominating Committee, American Economic Association, 1980-1981.
Research Advisory Committee, American Enterprise Institute, 1980-1986.
Editorial Board, M.I.T. Press Series on Government Regulation of Economic Activity, 1979-93.
Program Committee, 1980 World Congress of the Econometric Society.
Program Committee, Econometric Society, 1979, 1981, 1985.
Organizer, American Economic Association Meetings: 1980, 1982.
American Bar Association Section 7 Clayton Act Committee, 198l.
Principal Investigator, NSF grant SOC79-0327, 1979-80; NSF grant 285-6041, 1980-82; NSF grant SES-8038866, 1983-84, 1985-86.
Aspen Task Force on the Future of the Postal Service, 1978-80.
Organizing Committee of Sixth Annual Telecommunications Policy Research Conference, 1977-78.
Visiting Fellow, University of Warwick, July 1977.
Institute for Mathematical Studies in the Social Sciences, Stanford University, 1975.
Published Articles and Book Chapters:
“Economic Foundations for 21st Century Freight Rail Rate Regulation,” (with John W. Mayo), in U.S. Freight Rail Economics and Policy: Are We on the Right Track?, Macher, J.T. and J.W. Mayo (eds.), Routledge, 2019.
“Is Professor Salop Right That Judge Leon Bungled United States v. AT&T?” (with Janusz Ordover and J. Gregory Sidak), The Criterion Journal on Innovation, v. 3, 249-263, 2018.
“The Role of the Circle Principle in Market Definition,” (with Bryan Keating and Jonathan Orszag), The Antitrust Source, April 2018, 1-6.
“Economy-wide and Sectoral Impacts on Workers of Brazil’s Internet Rollout” (with Mark A. Dutz, Lucas Ferreira Mation, and Stephen D. O’Connell), Forum for Social Economics, 46:2, 160-
Public Version
A-4
177, 2017.
“Two-Sided Market Definition and Competitive Effects for Credit Cards After United States v. American Express” (with J. Gregory Sidak), The Criterion Journal on Innovation, v. 1, 1301, 2016.
"Unilateral Competitive Effects" (with Bryan Keating), in The Oxford Handbook on International Antitrust Economics, (Roger D. Blair and D. Daniel Sokol, eds.), Oxford University Press, 2014.
“Activating Actavis: A More Complete Story” (with Barry C. Harris, Kevin M. Murphy, and Matthew B. Wright), Antitrust, vol. 28, No. 2 (Spring), 2014.
“'Reverse Payments' in Settlements of Patent Litigation: Split Opinions on Schering-Plough’s K- Dur (2005 and 2012)" (with John P. Bigelow), in The Antitrust Revolution (Sixth Edition), (J. Kwoka and Laurence White, eds.), Oxford University Press, 2013.
"The Delta-Northwest Merger: Consumer Benefits from Airline Network Effects (2008)" (with Mark Israel, Bryan Keating and Daniel Rubinfeld), in The Antitrust Revolution (Sixth Edition), (J. Kwoka and Laurence White, eds.), Oxford University Press, 2013.
"Airline Network Effects and Consumer Welfare" (with Bryan Keating, Mark Israel and Daniel Rubinfeld), Review of Network Economics, published online November 2013.
"The Liftoff of Consumer Benefits from the Broadband Revolution" (with Mark Dutz and JonOrszag), Review of Network Economics (2012) vol. 11, issue 4, article 2.
"Competition and innovation-driven inclusive growth" (with Mark Dutz, Ioannis Kessides and Stephen O’Connell), in Promoting Inclusive Growth: Challenges and Policies, Luiz de Mello and Mark Dutz (eds.), OECD, 2011.
"Unilateral Competitive Effects of Mergers: Upward Pricing Pressure, Product Quality, andOther Extensions," Review of Industrial Organization (2011) 39:19–38.
“Antitrust and Patent Settlements: The Pharmaceutical Cases,” (with John Bigelow) in The Antitrust Revolution (Fifth Edition), John Kwoka and Lawrence White (eds.), 2009.
“The 1982 Department of Justice Merger Guidelines: An Economic Assessment,” (with J. Ordover) reprinted in Economics of Antitrust Law, Benjamin Klein (ed.), Edward Elgar, 2008.
“On the Antitrust Treatment of Production Joint Ventures,” (with Carl Shapiro) reprinted in Economics of Antitrust Law, Benjamin Klein (ed.), Edward Elgar, 2008.
“Consumer’s Surplus Without Apology,” reprinted in Applied Welfare Economics, Richard Just, Darrel Hueth and Andrew Schmitz (eds.), Edward Elgar, 2008; reprinted in Readings in Social Welfare: Theory and Policy, Robert E. Kuenne (ed.), Blackwell, 2000, pp. 86-97;
Public Version
A-5
reprinted in Readings in Microeconomic Theory, M. M. La Manna (ed.), Dryden Press, 1997, pp. 201-212.
“The Risk of Contagion from Multi-Market Contact,” (with Charles Thomas), The International Journal of Industrial Organization, Vol. 24, Issue 6 (Nov. 2006), pp 1157 – 1184.
“Pareto-Superior Nonlinear Outlay Schedules,” reprinted in The Economics of Public Utilities, Ray Rees (ed.), Edward Elgar, 2006; reprinted in The Economics of Price Discrimination, G. Norman, (ed.), Edward Elgar, 1999.
“Economic Effects of Antidumping Policy,” reprinted in The WTO and Anti-Dumping, Douglas Nelson (ed.), Edward Elgar, 2005.
“Merger Analysis, Industrial Organization Theory and the Merger Guidelines,” reprinted in Antitrust and Competition Policy, Andrew Kleit (ed.) Edward Elgar, 2005
“Antitrust Policy Towards Agreements That Settle Patent Litigation,” (with John Bigelow), Antitrust Bulletin, Fall 2004, pp. 655-698.
“Economies of Scope,” (with John Panzar), reprinted in The Economics of Business Strategy, John Kay (ed.), Edward Elgar, 2003.
“Panel on Substantive Standards for Mergers and the Role of Efficiencies,” in International Antitrust Law & Policy, Barry E. Hawk (ed.), Juris Publishing, 2003.
“Practical Rules for Pricing Access in Telecommunications,” (with J. Ordover) in Second Generation Reforms in Infrastructure Services , F. Basanes and R. Willig (eds.), Johns Hopkins Press, 2002.
“Comments on Antitrust Policy in the Clinton Administration,” in American Economic Policy in the 1990s, J. Frankel and P. Orszag (eds.), MIT Press, 2002.
“Entrepreneurship, Access Policy and Economic Development: Lessons from Industrial Organization,” (with M. Dutz and J. Ordover), European Economic Review, (44)4-6 (2000), pp. 739-747.
"Public Versus Regulated Private Enterprise," reprinted in Privatization in Developing Countries, P. Cook and C. Kirkpatrick (eds.), Edward Elgar, 2000.
“Deregulation v. the Legal Culture: Panel Discussion,” in Is the Telecommunications Act of 1996 Broken?, G. Sidak (ed.), AEI Press, 1999.
“Economic Principles to Guide Post-Privatization Governance,” in Can Privatization Deliver? Infrastructure for Latin America, R. Willig co-editor, Johns Hopkins Press, 1999.
Public Version
A-6
“Access and Bundling in High-Technology Markets,” (with J. A. Ordover), in Competition, Innovation and the Microsoft Monopoly: Antitrust in the Digital Marketplace, J. A. Eisenach and T. Lenard (eds.), Kluwer, 1999.
“Competitive Rail Regulation Rules: Should Price Ceilings Constrain Final Products or Inputs?,” (With W. J. Baumol), Journal of Transport Economics and Policy, Vol. 33, Part 1, pp. 1-11.
“Economic Effects of Antidumping Policy,” Brookings Trade Forum: 1998, 19-41.
“Interview With Economist Robert D. Willig,” Antitrust , Vol. 11, No. 2, Spring 1997, pp.11-15.
“Parity Pricing and its Critics: A Necessary Condition for Efficiency in Provision of Bottleneck Services to Competitors,” (with W. J. Baumol and J. A. Ordover), Yale Journal on Regulation, Vol. 14, No. 1, Winter 1997, pp. 145-164.
“Restructuring Regulation of the Rail Industry,” (with Ioannis Kessides), in Private Sector, Quarterly No. 4, September 1995, pp. 5 - 8. Reprinted in Viewpoint, October, 1995, The World Bank. Reprinted in Private Sector, special edition: Infrastructure, June 1996.
“Competition and Regulation in the Railroad Industry,” (with Ioannis Kessides), in Regulatory Policies and Reform: A Comparative Perspective, C. Frischtak (ed.), World Bank, 1996.
"Economic Rationales for the Scope of Privatization," (with Carl Shapiro), reprinted in The Political Economy of Privatization and Deregulation, E. E. Bailey and J. R. Pack (eds.), The International Library of Critical Writings in Economics, Edward Elgar Publishing Co., 1995, pp. 95-130.
"Weak Invisible Hand Theorems on the Sustainability of Multi-product Natural Monopoly," (with W. Baumol and E. Bailey), reprinted in The Political Economy of Privatization and Deregulation, E. E. Bailey and J. R. Pack (eds.), The International Library of Critical Writings in Economics, Edward Elgar Publishing Co., 1995, pp. 245-260.
“Economists’ View: The Department of Justice Draft Guidelines for the Licensing and Acquisition of Intellectual Property,” (with J. Ordover), Antitrust, V. 9, No. 2 (spring 1995), 29- 36.
"Public Versus Regulated Private Enterprise," in Proceedings of the World Bank Annual Conference on Development Economics 1993, L. Summers (ed.), The World Bank, 1994.
"Economics and the 1992 Merger Guidelines: A Brief Survey," (with J. Ordover), Review of Industrial Organization, V. 8, No. 2, (1993), pp. 139-150.
"The Role of Sunk Costs in the 1992 Guidelines' Entry Analysis," Antitrust, V. 6, No. 3 (summer 1992).
Public Version
A-7
"Antitrust Lessons from the Airlines Industry: The DOJ Experience," Antitrust Law Journal, V. 60, No. 2 (1992).
"William J. Baumol," (with E. E. Bailey), in New Horizons in Economic Thought: Appraisals of Leading Economists, W. J. Samuels (ed.), Edward Elgar, 1992.
"Anti-Monopoly Policies and Institutions," in The Emergence of Market Economies in Eastern Europe, Christopher Clague and Gordon Rausser (eds.), Basil Blackwell, 1992.
"Economics and the 1992 Merger Guidelines," (with Janusz Ordover), in Collaborations Among Competitors: Antitrust Policy and Economics, Eleanor Fox and James Halverson (eds.), American Bar Association, 1992.
"On the Antitrust Treatment of Production Joint Ventures," (with Carl Shapiro), reprinted in Collaborations Among Competitors: Antitrust Policy and Economics, Eleanor Fox and James Halverson (eds), American Bar Association, 1992.
"Merger Analysis, Industrial Organization Theory, and Merger Guidelines," Brookings Papers on Economic Activity -- Microeconomics 1991, pp. 281-332.
"On the Antitrust Treatment of Production Joint Ventures," (with C. Shapiro), Journal of Economic Perspectives, Vol. 4, No. 3, Summer 1990, pp. 113-130.
"Economic Rationales for the Scope of Privatization," (with Carl Shapiro), in The Political Economy of Public Sector Reform and Privatization, E.N. Suleiman and J. Waterbury (eds.), Westview Press, Inc., 1990, pp. 55-87.
"Contestable Market Theory and Regulatory Reform," in Telecommunications Deregulation: Market Power and Cost Allocation, J.R. Allison and D.L. Thomas (eds.), Ballinger, 1990.
"Address To The Section," Antitrust Law Section Symposium, New York State Bar Association, 1990.
"Price Caps: A Rational Means to Protect Telecommunications Consumers and Competition," (with W. Baumol), Review of Business, Vol. 10, No. 4, Spring 1989, pp. 3-8.
"U.S.-Japanese VER: A Case Study from a Competition Policy Perspective," (with M. Dutz) in The Costs of Restricting Imports, The Automobile Industry. OECD, 1988.
"Contestable Markets," in The New Palgrave: A Dictionary of Economics, J. Eatwell, M. Milgate, and P. Newman (eds.), 1987.
"Do Entry Conditions Vary Across Markets: Comments," Brookings Papers on Economic Activity, 3 - 1987, pp. 872-877.
Public Version
A-8
"Railroad Deregulation: Using Competition as a Guide," (with W. Baumol), Regulation, January/February 1987, Vol. 11, No. 1, pp. 28-36.
"How Arbitrary is 'Arbitrary'? - or, Toward the Deserved Demise of Full Cost Allocation," (with W. Baumol and M. Koehn), Public Utilities Fortnightly, September 1987, Vol. 120, No. 5, pp. 16-22.
"Contestability: Developments Since the Book," (with W. Baumol), Oxford Economic Papers, December 1986, pp. 9-36.
"The Changing Economic Environment in Telecommunications: Technological Change and Deregulation," in Proceedings from the Telecommunications Deregulation Forum; Karl Eller Center; 1986.
"Perspectives on Mergers and World Competition," (with J. Ordover), in Antitrust and Regulation, R.E. Grieson (ed.), Lexington, 1986.
"On the Theory of Perfectly Contestable Markets," (with J. Panzar and W. Baumol), in New Developments in The Analysis of Market Structure, J. Stiglitz and F. Mathewson (eds.), MIT Press, 1986.
"InterLATA Capacity Growth and Market Competition," (with C. Shapiro), in Telecommunications and Equity: Policy Research Issues, J. Miller (ed.), North Holland, 1986.
"Corporate Governance and Market Structure," in Economic Policy in Theory and Practice, A. Razin and E. Sadka (eds.), Macmillan Press, 1986.
"Antitrust for High-Technology Industries: Assessing Research Joint Ventures and Mergers," (with J. Ordover), Journal of Law and Economics, Vol 28(2), May 1985, pp. 311-334.
"Non-Price Anticompetitive Behavior by Dominant Firms Toward the Producers of Complementary Products," (with J. Ordover and A. Sykes), in Antitrust and Regulation, F.M. Fisher (ed.), MIT Press, 1985.
"Telephones and Computers: The Costs of Artificial Separation," (with W. Baumol), Regulation, March/April 1985.
"Transfer Principles in Income Redistribution," (with P. Fishburn), Journal of Public Economics, 25 (1984), pp. 1-6.
"Market Structure and Government Intervention in Access Markets," in Telecommunications Access and Public Policy, A. Baughcam and G. Faulhaber (eds.), 1984.
"Pricing Issues in the Deregulation of Railroad Rates," (with W. Baumol), in Economic
Public Version
A-9
Analysis of Regulated Markets: European and U. S. Perspectives, J. Finsinger (ed.), 1983.
"Local Telephone Pricing in a Competitive Environment," (with J. Ordover), in Telecommunications Regulation Today and Tomorrow, E. Noam (ed.), Harcourt Brace Jovanovich, 1983.
"Economics and Postal Pricing Policy," (with B. Owen), in The Future of the Postal Service, J. Fleishman (ed.), Praeger, 1983.
"Selected Aspects of the Welfare Economics of Postal Pricing," in Telecommunications Policy Annual, Praeger, 1987.
"The Case for Freeing AT&T" (with M. Katz), Regulation, July-Aug. 1983, pp. 43-52.
"Predatory Systems Rivalry: A Reply" (with J. Ordover and A. Sykes), Columbia Law Review, Vol. 83, June 1983, pp. 1150-1166. Reprinted in Corporate Counsel's Handbook - 1984.
"Sector Differentiated Capital Taxation with Imperfect Competition and Interindustry Flows," Journal of Public Economics, Vol. 21, 1983.
"Contestable Markets: An Uprising in the Theory of Industry Structure: Reply," (with W.J. Baumol and J.C. Panzar), American Economic Review, Vol. 73, No. 3, June 1983, pp. 491-496.
"The 1982 Department of Justice Merger Guidelines: An Economic Assessment," (with J. Ordover), California Law Review, Vol. 71, No. 2, March 1983, pp. 535-574. Reprinted in Antitrust Policy in Transition: The Convergence of Law and Economics, E.M. Fox and J.T. Halverson (eds.), 1984.
"Intertemporal Failures of the Invisible Hand: Theory and Implications for International Market Dominance," (with W.J. Baumol), Indian Economic Review, Vol. XVI, Nos. 1 and 2, January-June 1981, pp. 1-12.
"Unfair International Trade Practices," (with J. Ordover and A. Sykes), Journal of International Law and Politics, Vol. 15, No. 2, winter 1983, pp. 323-337.
"Journals as Shared Goods: Reply," (with J. Ordover), American Economic Review, V. 72, No. 3, June 1982, pp. 603-607.
"Herfindahl Concentration, Rivalry, and Mergers," (with J. Ordover and A. Sykes), Harvard Law Review, V. 95, No. 8, June 1982, pp. 1857-l875.
"An Economic Definition of Predation: Pricing and Product Innovation," (with J. Ordover), Yale Law Journal, Vol. 90: 473, December 1981, pp. 1-44.
"Fixed Costs, Sunk Costs, Entry Barriers, and the Sustainability of Monopoly," (with W.
Public Version
A-10
Baumol), Quarterly Journal of Economics, Vol. 96, No. 3, August 1981, pp. 405-432.
"Social Welfare Dominance," American Economic Review, Vol. 71, No. 2, May 1981, pp. 200-204.
"Economies of Scope," (with J. Panzar), American Economic Review, Vol. 72, No. 2, May 1981, pp. 268-272.
"Income-Distribution Concerns in Regulatory Policymaking," (with E.E. Bailey) in Studies in Public Regulation (G. Fromm, ed.), MIT Press, Cambridge, 1981, pp. 79-118.
"An Economic Definition of Predatory Product Innovation," (with J. Ordover), in Strategic Predation and Antitrust Analysis, S. Salop (ed.), 1981.
"What Can Markets Control?" in Perspectives on Postal Service Issues, R. Sherman (ed.), American Enterprise Institute, 1980.
"Pricing Decisions and the Regulatory Process," in Proceedings of the 1979 Rate Symposium on Problems of Regulated Industries, University of Missouri-Columbia Extension Publications, 1980, pp. 379-388.
"The Theory of Network Access Pricing," in Issues in Public Utility Regulation, H.M. Trebing (ed.), MSU Public Utilities Papers, 1979.
"Customer Equity and Local Measured Service," in Perspectives on Local Measured Service, J. Baude, etal. (ed.), 1979, pp. 71-80.
"The Role of Information in Designing Social Policy Towards Externalities," (with J. Ordover), Journal of Public Economics, V. 12, 1979, pp. 271-299.
"Economies of Scale and the Profitability of Marginal-Cost Pricing: Reply," (with J. Panzar), Quarterly Journal of Economics, Vol. 93, No. 4, Novmber 1979, pp. 743-4.
"Theoretical Determinants of the Industrial Demand for Electricity by Time of Day," (with J. Panzar) Journal of Econometrics, V. 9, 1979, pp. 193-207.
"Industry Performance Gradient Indexes," (with R. Dansby), American Economic Review, V. 69, No. 3, June 1979, pp. 249-260.
"The Economic Gradient Method," (with E. Bailey), American Economic Review, Vol. 69, No. 2, May 1979, pp. 96-101.
"Multiproduct Technology and Market Structure," American Economic Review, Vol. 69, No. 2, May 1979, pp. 346-351.
Public Version
A-11
"Consumer's Surplus Without Apology: Reply," American Economic Review, Vol. 69, No. 3, June 1979, pp. 469-474.
"Decisions with Estimation Uncertainty," (with R. Klein, D. Sibley, and L. Rafsky), Econometrica, V. 46, No. 6, November 1978, pp. 1363-1388.
"Incremental Consumer's Surplus and Hedonic Price Adjustment," Journal of Economic Theory, V. 17, No. 2, April 1978, pp. 227-253.
"Recent Theoretical Developments in Financial Theory: Discussion, "The Journal of Finance, V. 33, No. 3, June 1978, pp. 792-794.
"The Optimal Provision of Journals Qua Sometimes Shared Goods," (with J. Ordover), American Economic Review, V. 68, No. 3, June 1978, pp. 324-338.
"On the Comparative Statics of a Competitive Industry With Infra-marginal Firms," (with J. Panzar), American Economic Review, V. 68, No. 3, June 1978, pp. 474-478.
"Pareto Superior Nonlinear Outlay Schedules," Bell Journal of Economics, Vol. 9, No. 1, Spring 1978, pp. 56-69.
"Predatoriness and Discriminatory Pricing," in The Economics of Anti-Trust: Course of Study Materials, American Law Institute-American Bar Association, 1978.
"Economies of Scale in Multi-Output Production," (with J. Panzar), Quarterly Journal of Economics, V. 91, No. 3, August 1977, pp. 481-494.
"Weak Invisible Hand Theorems on the Sustainability of Multi-product Natural Monopoly," (with W. Baumol and E. Bailey), American Economic Review, V. 67, No. 3, June 1977, pp. 350-365.
"Free Entry and the Sustainability of Natural Monopoly," (with J. Panzar), Bell Journal of Economics, Spring 1977, pp. 1-22.
"Risk Invariance and Ordinally Additive Utility Functions," Econometrica, V. 45, No. 3, April 1977, pp. 621-640.
"Ramsey-Optimal Pricing of Long Distance Telephone Services," (with E. Bailey), in Pricing in Regulated Industries, Theory and Application, J. Wenders (ed.), Mountain State Telephone and Telegraph Co., 1977, pp. 68-97.
"Network Externalities and Optimal Telecommunications Pricing: A Preliminary Sketch," (with R. Klein), in Proceedings of Fifth Annual Telecommunications Policy Research Conference, Volume II, NTIS, 1977, pp. 475-505.
Public Version
A-12
"Otsenka ekonomicheskoi effektivnosti proizvodstvennoi informatsii" ["The Evaluation of the Economic Benefits of Productive Information"] in Doklady Sovetskikh i Amerikanskikh Spetsialistov Predstavlennye na Pervyi Sovetsko-Amerikanskii Simpozium po Ekonomicheskoi Effektivnosti Informat sionnogo Obsluzhivaniia [Papers of Soviet and American Specialists Presented at the First Soviet- American Symposium on Costs and Benefits of Information Services], All Soviet Scientific Technical Information Center, Moscow, 1976.
"Vindication of a 'Common Mistake' in Welfare Economics," (with J. Panzar), Journal of Political Economy, V. 84, No. 6, December 1976, pp. 1361-1364.
"Consumer's Surplus Without Apology," American Economic Review, V. 66, No. 4, September 1976, pp. 589-597.
Books
Second Generation Reforms in Infrastructure Services, F. Basanes and R. Willig (eds.), Johns Hopkins Press, 2002.
Can Privatization Deliver? Infrastructure for Latin America, R. Willig co-editor, Johns Hopkins Press, 1999.
Handbook of Industrial Organization, (edited with R. Schmalensee), North Holland Press, Volumes 1 and 2, 1989.
Contestable Markets and the Theory of Industry Structure, (with W.J. Baumol and J.C. Panzar), Harcourt Brace Jovanovich, 1982. Second Edition, 1989.
Welfare Analysis of Policies Affecting Prices and Products, Garland Press, 1980.
Unpublished Papers and Reports:
“Brief Amici Curiae of 37 Economists, Antitrust Scholars, and Former Government Antitrust Officials In Support of Appellees and Supporting Affirmance,” In the United States Court of Appeals for the District of Columbia Circuit; USCA Case #18-5214; United States of America, Plaintiff-Appellant, v. AT&T, Inc.; DirecTV Group Holdings, LLC; and TimeWarner Inc., Defendants-Appellees; On appeal from a final judgment of the U.S. District Court for the District of Columbia, Hon. Richard J. Leon, No. 1:17-cv-2511; 9/26/2018.
“Brief for Amici Curiae J. Gregory Sidak and Robert D. Willig in Support of Respondents,” In the Supreme Court of the United States; In the Supreme Court of the United States; State of Ohio, et al., v. American Express Company, et al., On Writ Of Certiorari To The United States Court Of Appeals For The Second Circuit; No. 16-1454; January 23, 2018.
Public Version
A-13
“Brief of Leading Economists as Amici Curiae in Support of Respondents,” In the Supreme Court of the United States; Douglas R. M. Nazarian, et al, v. PPL Energyplus, LLC, et al. and CPV Maryland, LLC, v. PPL Energyplus, LLC, et al.; On Writ of Certiorari to the US Court of Appeals for the Fourth Circuit; Nos. 14-614, 14-623; January 19, 2016.
“Technological change and labor market segmentation in the developing world: Evidence from Brazil,” (with Dutz, Mark, Lucas Ferreira-Mation, and Stephen O’Connell), 2015 Background Paper for the 2016 World Bank’s World Development Report.
“Brief for Amici Curiae J. Gregory Sidak, Robert D. Willig, David J. Teece, and Keith N. Hylton, Scholars and Experts in Antitrust Economics in Support of Defendants-Appellants and Supporting Reversal,” 15-1672 In the United States Court of Appeals for the Second Circuit; United States of America, et al., v. American Express Company, et al., 8/10/2015.
"Economics at the FTC: Hospital Mergers, Authorized Generic Drugs, and Consumer Credit Markets" (with Bryan Keating, Paolo Ramezzana, Margaret Guerin-Calvert, and Nauman Ilias), 2012.
"Recommendations for Excessive-Share Limits in the Surfclam and Ocean Quahog Fisheries" (with Glenn Mitchell and Steven Peterson), Report to National Marine Fisheries Service and the Mid-Atlantic Fishery Management Council, 5/23/2011.
"Public Comments on the 2010 Draft Horizontal Merger Guidelines," paper posted to Federal Trade Commission website, 6/4/2010
“An Economic Perspective on the Antitrust Case Against Intel,” (with Jon Orszag and Gilad Levin), 2009.
"An Econometric Analysis of the Matching Between Football Student-Athletes and Colleges," (with Yair Eilat, Bryan Keating and Jon Orszag)
Supreme Court Amicus Brief Regarding Morgan Stanley Capital Group Inc. v. Public Utility District No. 1 of Snohomish County, Washington, (co-authored), AEI-Brookings Joint Center Brief No. 07-02, 12/2/07
“(Allegedly) Monopolizing Tying Via Product Innovation,” statement before the Department of Justice/Federal Trade Commission Section 2 Hearings, November 1, 2006.
“Assessment of U.S. Merger Enforcement Policy,” statement before the Antitrust Modernization Commission, 11/17/05.
“Investment is Appropriately Stimulated by TELRIC,” in Pricing Based on Economic Cost, 12/2003.
Public Version
A-14
“Brief of Amici Curiae Economics Professors, re Verizon v. Trinko, In the Supreme Court of the U.S.,” (with W.J. Baumol, J.O. Ordover and F.R. Warren-Boulton), 7/25/2003.
“Stimulating Investment and the Telecommunications Act of 1996,” (with J. Bigelow, W. Lehr and S. Levinson), 2002.
“An Economic Analysis of Spectrum Allocation and Advanced Wireless Services,” (with Martin N. Baily, Peter R. Orszag, and Jonathan M. Orszag), 2002
“Effective Deregulation of Residential Electric Service,” 2001
“Anticompetitive Forced Rail Access” (with W. J. Baumol), 2000
“The Scope of Competition in Telecommunications” (with B. Douglas Bernheim), 1998 “Why
Do Christie and Schultz Infer Collusion From Their Data? (with Alan Kleidon), 1995.
"Demonopolization," (with Sally Van Siclen), OECD Vienna Seminar Paper, 1993.
"Economic Analysis of Section 337: The Balance Between Intellectual Property Protection and Protectionism," (with J. Ordover) 1990.
"The Effects of Capped NTS Charges on Long Distance Competition," (with M. Katz).
"Discussion of Regulatory Mechanism Design in the Presence of Research Innovation, and Spillover Effects," 1987.
"Industry Economic Analysis in the Legal Arena," 1987.
"Deregulation of Long Distance Telephone Services: A Public Interest Assessment," (with M. Katz).
"Competition-Related Trade Issues," report prepared for OECD.
"Herfindahl Concentration Index," (with J. Ordover), Memorandum for ABA Section 7 Clayton Act Committee, Project on Revising the Merger Guidelines, March 1981.
"Market Power and Market Definition," (with J. Ordover), Memorandum for ABA Section 7 Clayton Act Committee, Project on Revising the Merger Guidelines, May 1981.
"The Continuing Need for and National Benefits Derived from the REA Telephone Loan Programs - An Economic Assessment," 1981.
"The Economics of Equipment Leasing: Costing and Pricing," 1980.
Public Version
A-15
"Rail Deregulation and the Financial Problems of the U.S. Railroad Industry," (with W.J. Baumol), report prepared under contract to Conrail, 1979.
"Price Indexes and Intertemporal Welfare," Bell Laboratories Economics Discussion Paper, 1974.
"Consumer's Surplus: A Rigorous Cookbook," Technical Report #98, Economics Series, I.M.S.S.S., Stanford University, l973.
"An Economic-Demographic Model of the Housing Sector," (with B. Hickman and M. Hinz), Center for Research in Economic Growth, Stanford University, 1973.
Invited Conference Presentations:
NYU Stern and Concurrence Global Antitrust Economics Conference “Pharmaceutical Industry Antitrust” -- moderator 2019
FTC Competition and Consumer Protection Hearings "The State of US Antitrust Law" 2018
Portuguese Competition Authority Public Seminar Series “Ups and Downs of Horizontal and Vertical Mergers” 2017
World Bank Workshop on Digital Technology Adoption, Skills, Productivity and Jobs in Latin America
“Discussion of Models of Firm Heterogeneity” 2016
George Mason Law Review Annual Antitrust Symposium: Antitrust in an Interconnected World “GUPPI and the Safe Harbor” 2016
Competition Law & Policy Institute of New Zealand Annual Workshop “Merger Analysis Keynote” 2015
Economic Studies at Brookings: Railroads, Policy and the Economy “The Industry Perspective” 2015
Georgetown University McDonough School of Business Railroad Economics Symposium “The Role of Economic Theory in the ‘Deregulated’ Rail Industry” 2015
Brazilian School of Economics and Finance (FGV EPGE) Seminario “Public Interest Regulation: Lessons from Railroads” 2015
Public Version
A-16
NYU School of Law Conference on the Fiftieth Anniversary of United States v. Philadelphia National Bank: The Past, Present and Future of Merger Law “Discussion with Agency Economists” 2013
Brookings Institution Conference on The Economics of the Airline Industry "Airline Network Effects and Consumer Welfare" 2012
AGEP Public Policy Conference on Pharmaceutical Industry Economics, Regulation and Legal Issues; Law and Economics Center, George Mason University School of Law "Pharmaceutical Brand-Generic Disputes" 2012
U.S.-EU Alliance Study Peer Review Conferences "Review of Cooperative Agreements in Transatlantic Airline Markets" 2012 "The Research Agenda Ahead" 2012
Antitrust in the High Tech Sector Conference "Developments in Merger Enforcement" 2012
Georgetown Center for Business and Public Policy, Conference on the Evolution of Regulation "Reflections on Regulation" 2011
Antitrust Forum, New York State Bar Association "Upward Price Pressure, Market Definition and Supply Mobility" 2011
American Bar Association, Antitrust Section, Annual Convention "The New Merger Guidelines' Analytic Highlights" 2011
OECD and World Bank Conference on Challenges and Policies for Promoting Inclusive Growth "Inclusive Growth From Competition and Innovation" 2011
Villanova School of Business Executive MBA Conference "Airline Network Effects, Competition and Consumer Welfare" 2011
NYU School of Law Conference on Critical Directions in Antitrust "Unilateral Competitive Effects" 2010
Conf. on the State of European Competition Law and Enforcement in a Transatlantic Context "Recent Developments in Merger Control" 2010
Center on Regulation and Competition, Universidad de Chile Law School "Economic Regulation and the Limits of Antitrust Law" 2010
Center on Regulation and Competition, Universidad de Chile Law School "Merger Policy and Guidelines Revision" 2010
Public Version
A-17
Faculty of Economics, Universidad de Chile "Network Effects in Airlines Markets" 2010
Georgetown Law Global Antitrust Enforcement Symposium "New US Merger Guidelines" 2010
FTI London Financial Services Conference "Competition and Regulatory Reform" 2010
NY State Bar Association Annual Antitrust Conference “New Media Competition Policy” 2009
Antitrust Law Spring Meeting of the ABA “Antitrust and the Failing Economy Defense” 2009
Georgetown Law Global Antitrust Enforcement Symposium “Mergers: New Enforcement Attitudes in a Time of Economic Challenge” 2009
Phoenix Center US Telecoms Symposium “Assessment of Competition in the Wireless Industry” 2009
FTC and DOJ Horizontal Merger Guidelines Workshop “Direct Evidence is No Magic Bullet” 2009
Northwestern Law Research Symposium: Antitrust Economics and Competition Policy "Discussion of Antitrust Evaluation of Horizontal Mergers" 2008
Inside Counsel Super-Conference "Navigating Mixed Signals under Section 2 of the Sherman Act" 2008
Federal Trade Commission Workshop on Unilateral Effects in Mergers "Best Evidence and Market Definition" 2008
European Policy Forum, Rules for Growth: Telecommunications Regulatory Reform “What Kind of Regulation For Business Services?” 2007
Japanese Competition Policy Research Center, Symposium on M&A and Competition Policy “Merger Policy Going Forward With Economics and the Economy” 2007
Federal Trade Commission and Department of Justice Section 2 Hearings “Section 2 Policy and Economic Analytic Methodologies” 2007
Pennsylvania Bar Institute, Antitrust Law Committee CLE “The Economics of Resale Price Maintenance and Class Certification” 2007
Public Version
A-18
Pennsylvania Bar Institute, Antitrust Law Committee CLE “Antitrust Class Certification – An Economist’s Perspective” 2007
Fordham Competition Law Institute, International Competition Economics Training Seminar “Monopolization and Abuse of Dominance” 2007
Canadian Bar Association Annual Fall Conference on Competition Law “Economic Tools for the Competition Lawyer” 2007
Conference on Managing Litigation and Business Risk in Multi-jurisdiction Antitrust Matters “Economic Analysis in Multi-jurisdictional Merger Control” 2007
World Bank Conference on Structuring Regulatory Frameworks for Dynamic and Competitive South Eastern European Markets “The Roles of Government Regulation in a Dynamic Economy” 2006
Department of Justice/Federal Trade Commission Section 2 Hearings “(Allegedly) Monopolizing Tying Via Product Innovation” 2006
Fordham Competition Law Institute, Competition Law Seminar “Monopolization and Abuse of Dominance” 2006
Practicing Law Institute on Intellectual Property Antitrust “Relevant Markets for Intellectual Property Antitrust” 2006
PLI Annual Antitrust Law Institute “Cutting Edge Issues in Economics” 2006
World Bank’s Knowledge Economy Forum V “Innovation, Growth and Competition” 2006
Charles University Seminar Series “The Dangers of Over-Ambitious Antitrust Regulation” 2006
NY State Bar Association Antitrust Law Section Annual Meeting “Efficient Integration or Illegal Monopolization?” 2006
World Bank Seminar “The Dangers of Over-Ambitious Regulation” 2005
ABA Section of Antitrust Law 2005 Fall Forum “Is There a Gap Between the Guidelines and Agency Practice?” 2005
Hearing of Antitrust Modernization Commission “Assessment of U.S. Merger Enforcement Policy” 2005
Public Version
A-19
LEAR Conference on Advances in the Economics of Competition Law “Exclusionary Pricing Practices” 2005
Annual Antitrust Law Institute “Cutting Edge Issues in Economics” 2005
PRIOR Symposium on States and Stem Cells “Assessing the Economics of State Stem Cell Programs” 2005
ABA Section of Antitrust Law – AALS Scholars Showcase “Distinguishing Anticompetitive Conduct” 2005
Allied Social Science Associations National Convention “Antitrust in the New Economy” 2005
ABA Section of Antitrust Law 2004 Fall Forum “Advances in Economic Analysis of Antitrust” 2004
Phoenix Center State Regulator Retreat “Regulatory Policy for the Telecommunications Revolution” 2004
OECD Competition Committee “Use of Economic Evidence in Merger Control” 2004
Justice Department/Federal Trade Commission Joint Workshop “Merger Enforcement” 2004
Phoenix Center Annual U.S. Telecoms Symposium “Incumbent Market Power” 2003
Center for Economic Policy Studies Symposium on Troubled Industries “What Role for Government in Telecommunications?” 2003
Princeton Workshop on Price Risk and the Future of the Electric Markets “The Structure of the Electricity Markets” 2003
2003 Antitrust Conference “International Competition Policy and Trade Policy” 2003
International Industrial Organization Conference “Intellectual Property System Reform” 2003
ABA Section of Antitrust Law 2002 Fall Forum “Competition, Regulation and Pharmaceuticals” 2002
Public Version
A-20
Fordham Conference on International Antitrust Law and Policy “Substantive Standards for Mergers and the Role of Efficiencies” 2002
Department of Justice Telecom Workshop “Stimulating Investment and the Telecommunications Act of 1996” 2002
Department of Commerce Conference on the State of the Telecom Sector “Stimulating Investment and the Telecommunications Act of 1996” 2002
Law and Public Affairs Conference on the Future of Internet Regulation “Open Access and Competition Policy Principles” 2002
Center for Economic Policy Studies Symposium on Energy Policy “The Future of Power Supply” 2002
The Conference Board: Antitrust Issues in Today’s Economy “The 1982 Merger Guidelines at 20” 2002
Federal Energy Regulatory Commission Workshop “Effective Deregulation of Residential Electric Service” 2001
IPEA International Seminar on Regulation and Competition “Electricity Markets: Deregulation of Residential Service” 2001 “Lessons for Brazil from Abroad” 2001
ABA Antitrust Law Section Task Force Conference “Time, Change, and Materiality for Monopolization Analyses” 2001
Harvard University Conference on American Economic Policy in the 1990s “Comments on Antitrust Policy in the Clinton Administration” 2001
Tel-Aviv Workshop on Industrial Organization and Anti-Trust “The Risk of Contagion from Multimarket Contact” 2001
2001 Antitrust Conference “Collusion Cases: Cutting Edge or Over the Edge?” 2001 “Dys-regulation of California Electricity” 2001
FTC Public Workshop on Competition Policy for E-Commerce “Necessary Conditions for Cooperation to be Problematic” 2001
HIID International Workshop on Infrastructure Policy “Infrastructure Privatization and Regulation” 2000
Villa Mondragone International Economic Seminar “Competition Policy for Network and Internet Markets” 2000
Public Version
A-21
New Developments in Railroad Economics: Infrastructure Investment and Access Policies “Railroad Access, Regulation, and Market Structure” 2000
The Multilateral Trading System at the Millennium “Efficiency Gains From Further Liberalization” 2000
Singapore – World Bank Symposium on Competition Law and Policy “Policy Towards Cartels and Collusion” 2000
CEPS: Is It a New World?: Economic Surprises of the Last Decade “The Internet and E-Commerce” 2000
Cutting Edge Antitrust: Issues and Enforcement Policies “The Direction of Antitrust Entering the New Millennium” 2000
The Conference Board: Antitrust Issues in Today’s Economy “Antitrust Analysis of Industries With Network Effects” 1999
CEPS: New Directions in Antitrust “Antitrust in a High-Tech World” 1999
World Bank Meeting on Competition and Regulatory Policies for Development “Economic Principles to Guide Post-Privatization Governance” 1999
1999 Antitrust Conference “Antitrust and the Pace of Technological Development” 1999 “Restructuring the Electric Utility Industry” 1999
HIID International Workshop on Privatization, Regulatory Reform and Corporate Governance “Privatization and Post-Privatization Regulation of Natural Monopolies” 1999
The Federalist Society: Telecommunications Deregulation: Promises Made, Potential Lost?
“Grading the Regulators” 1999
Inter-American Development Bank: Second Generation Issues In the Reform Of Public Services
“Post-Privatization Governance” 1999 “Issues Surrounding Access Arrangements” 1999
Economic Development Institute of the World Bank -- Program on Competition Policy “Policy Towards Horizontal Mergers” 1998
Twenty-fifth Anniversary Seminar for the Economic Analysis Group of the Department of
Public Version
A-22
Justice “Market Definition in Antitrust Analysis” 1998
HIID International Workshop on Privatization, Regulatory Reform and Corporate Governance “Infrastructure Architecture and Regulation: Railroads” 1998
EU Committee Competition Conference – Market Power “US/EC Perspective on Market Definition” 1998
Federal Trade Commission Roundtable “Antitrust Policy for Joint Ventures” 1998
1998 Antitrust Conference “Communications Mergers” 1998
The Progress and Freedom Foundation Conference on Competition, Convergence, and the Microsoft Monopoly
Access and Bundling in High-Technology Markets 1998
FTC Program on The Effective Integration of Economic Analysis into Antitrust Litigation The Role of Economic Evidence and Testimony 1997
FTC Hearings on Classical Market Power in Joint Ventures Microeconomic Analysis and Guideline 1997
World Bank Economists --Week IV Keynote Making Markets More Effective With Competition Policy 1997
Brookings Trade Policy Forum Competition Policy and Antidumping: The Economic Effects 1997
University of Malaya and Harvard University Conference on The Impact of Globalisation and Privatisation on Malaysia and Asia in the Year 2020
Microeconomics, Privatization, and Vertical Integration 1997
ABA Section of Antitrust Law Conference on The Telecommunications Industry Current Economic Issues in Telecommunications 1997
Antitrust 1998: The Annual Briefing The Re-Emergence of Distribution Issues 1997
Inter-American Development Bank Conference on Private Investment, Infrastructure Reform and Governance in Latin America & the Caribbean
Economic Principles to Guide Post-Privatization Governance 1997
Public Version
A-23
Harvard Forum on Regulatory Reform and Privatization of Telecommunications in the Middle East
Privatization: Methods and Pricing Issues 1997
American Enterprise Institute for Public Policy Research Conference Discussion of Local Competition and Legal Culture 1997
Harvard Program on Global Reform and Privatization of Public Enterprises “Infrastructure Privatization and Regulation: Freight” 1997
World Bank Competition Policy Workshop “Competition Policy for Entrepreneurship and Growth” 1997
Eastern Economics Association Paul Samuelson Lecture “Bottleneck Access in Regulation and Competition Policy” 1997
ABA Annual Meeting, Section of Antitrust Law “Antitrust in the 21st Century: The Efficiencies Guidelines” 1997
Peruvian Ministry of Energy and Mines Conference on Regulation of Public Utilities “Regulation: Theoretical Context and Advantages vs. Disadvantages” 1997
The FCC: New Priorities and Future Directions “Competition in the Telecommunications Industry” 1997
American Enterprise Institute Studies in Telecommunications Deregulation “The Scope of Competition in Telecommunications” 1996
George Mason Law Review Symposium on Antitrust in the Information Revolution “Introduction to the Economic Theory of Antitrust and Information” 1996
Korean Telecommunications Public Lecture “Market Opening and Fair Competition” 1996
Korea Telecommunications Forum “Desirable Interconnection Policy in a Competitive Market” 1996
European Association for Research in Industrial Economics Annual Conference “Bottleneck Access: Regulation and Competition Policy” 1996
Harvard Program on Global Reform and Privatization of Public Enterprises “Railroad and Other Infrastructure Privatization” 1996
Public Version
A-24
FCC Forum on Antitrust and Economic Issues Involved with InterLATA Entry “The Scope of Telecommunications Competition” 1996
Citizens for a Sound Economy Policy Watch on Telecommunications Interconnection “The Economics of Interconnection” 1996
World Bank Seminar on Experiences with Corporatization “Strategic Directions of Privatization” 1996
FCC Economic Forum on the Economics of Interconnection Lessons from Other Industries 1996
ABA Annual Meeting, Section of Antitrust Law The Integration, Disintegration, and Reintegration of the Entertainment Industry 1996
Conference Board: 1996 Antitrust Conference How Economics Influences Antitrust and Vice Versa 1996
Antitrust 1996: A Special Briefing Joint Ventures and Strategic Alliances 1996
New York State Bar Association Section of Antitrust Law Winter Meeting Commentary on Horizontal Effects Issues 1996
FTC Hearings on the Changing Nature of Competition in a Global and Innovation-Driven Age Vertical Issues for Networks and Standards 1995
Wharton Seminar on Applied Microeconomics Access Policies with Imperfect Regulation 1995
Antitrust 1996, Washington D.C. Assessing Joint Ventures for Diminution of Competition 1995
ABA Annual Meeting, Section of Antitrust Law Refusals to Deal -- Economic Tests for Competitive Harm 1995
FTC Seminar on Antitrust Enforcement Analysis Diagnosing Collusion Possibilities 1995
Philadelphia Bar Education Center: Antitrust Fundamentals Antitrust--The Underlying Economics 1995
Vanderbilt University Conference on Financial Markets
Public Version
A-25
Why Do Christie and Schultz Infer Collusion From Their Data? 1995
ABA Section of Antitrust Law Chair=s Showcase Program Discussion of Telecommunications Competition Policy 1995
Conference Board: 1995 Antitrust Conference Analysis of Mergers and Joint Ventures 1995
ABA Conference on The New Antitrust: Policy of the '90s Antitrust on the Super Highways/Super Airways 1994
ITC Hearings on The Economic Effects of Outstanding Title VII Orders "The Economic Impacts of Antidumping Policies" 1994
OECD Working Conference on Trade and Competition Policy "Empirical Evidence on The Nature of Anti-dumping Actions" 1994
Antitrust 1995, Washington D.C. "Rigorous Antitrust Standards for Distribution Arrangements" 1994
ABA -- Georgetown Law Center: Post Chicago-Economics: New Theories - New Cases?
"Economic Foundations for Vertical Merger Guidelines" 1994
Conference Board: Antitrust Issues in Today's Economy "New Democrats, Old Agencies: Competition Law and Policy" 1994
Federal Reserve Board Distinguished Economist Series "Regulated Private Enterprise Versus Public Enterprise" 1994
Institut d'Etudes Politiques de Paris "Lectures on Competition Policy and Privatization" 1993
Canadian Bureau of Competition Policy Academic Seminar Series, Toronto. "Public Versus Regulated Private Enterprise" 1993
CEPS Symposium on The Clinton Administration: A Preliminary Report Card "Policy Towards Business" 1993
Columbia Institute for Tele-Information Conference on Competition in Network Industries, New York, NY
"Discussion of Deregulation of Networks: What Has Worked and What Hasn't"
World Bank Annual Conference on Development Economics1993
"Public Versus Regulated Private Enterprise" 1993
Public Version
A-26
Center for Public Utilities Conference on Current Issues Challenging the Regulatory Process "The Economics of Current Issues in Telecommunications Regulation" 1992 "The Role of Markets in Presently Regulated Industries" 1992
The Conference Board's Conference on Antitrust Issues in Today's Economy, New York, NY "Antitrust in the Global Economy" 1992 "Monopoly Issues for the '90s" 1993
Columbia University Seminar on Applied Economic Theory, New York, NY "Economic Rationales for the Scope of Privatization" 1992
Howrey & Simon Conference on Antitrust Developments, Washington, DC "Competitive Effects of Concern in the Merger Guidelines" 1992
Arnold & Porter Colloquium on Merger Enforcement, Washington, DC "The Economic Foundations of the Merger Guidelines" 1992
American Bar Association, Section on Antitrust Law Leadership Council Conference, Monterey, CA
"Applying the 1992 Merger Guidelines" 1992
OECD Competition Policy Meeting, Paris, France "The Economic Impacts of Antidumping Policy" 1992
Center for Public Choice Lecture Series, George Mason University Arlington, VA "The Economic Impacts of Antidumping Policy" 1992
Brookings Institution Microeconomics Panel, Washington, DC, "Discussion of the Evolution of Industry Structure" 1992
AT&T Conference on Antitrust Essentials "Antitrust Standards for Mergers and Joint Ventures" 1991
ABA Institute on The Cutting Edge of Antitrust: Market Power "Assessing and Proving Market Power: Barriers to Entry" 1991
Second Annual Workshop of the Competition Law and Policy Institute of New Zealand "Merger Analysis, Industrial Organization Theory, and Merger Guidelines" 1991 "Exclusive Dealing and the Fisher & Paykel Case" 1991
Special Seminar of the New Zealand Treasury "Strategic Behavior, Antitrust, and The Regulation of Natural Monopoly" 1991
Public Version
A-27
Public Seminar of the Australian Trade Practices Commission "Antitrust Issues of the 1990's" 1991
National Association of Attorneys General Antitrust Seminar "Antitrust Economics" 1991
District of Columbia Bar's 1991 Annual Convention "Administrative and Judicial Trends in Federal Antitrust Enforcement" 1991
ABA Spring Meeting "Antitrust Lessons From the Airline Industry" 1991
Conference on The Transition to a Market Economy - Institutional Aspects "Anti-Monopoly Policies and Institutions" 1991
Conference Board's Thirtieth Antitrust Conference "Antitrust Issues in Today's Economy" 1991
American Association for the Advancement of Science Annual Meeting "Methodologies for Economic Analysis of Mergers" 1991
General Seminar, Johns Hopkins University "Economic Rationales for the Scope of Privatization" 1991
Capitol Economics Speakers Series "Economics of Merger Guidelines" 1991
CRA Conference on Antitrust Issues in Regulated Industries "Enforcement Priorities and Economic Principles" 1990
Pepper Hamilton & Scheetz Anniversary Colloquium "New Developments in Antitrust Economics" 1990
PLI Program on Federal Antitrust Enforcement in the 90's "The Antitrust Agenda of the 90's" 1990
FTC Distinguished Speakers Seminar "The Evolving Merger Guidelines" 1990
The World Bank Speakers Series "The Role of Antitrust Policy in an Open Economy" 1990
Seminar of the Secretary of Commerce and Industrial Development of Mexico "Transitions to a Market Economy" 1990
Public Version
A-28
Southern Economics Association "Entry in Antitrust Analysis of Mergers" 1990 "Discussion of Strategic Investment and Timing of Entry" 1990
American Enterprise Institute Conference on Policy Approaches to the Deregulation of Network Industries
"Discussion of Network Problems and Solutions" 1990
American Enterprise Institute Conference on Innovation, Intellectual Property, and World Competition
"Law and Economics Framework for Analysis" 1990
Banco Nacional de Desenvolvimento Economico Social Lecture "Competition Policy: Harnessing Private Interests for the Public Interest" 1990
Western Economics Association Annual Meetings "New Directions in Antitrust from a New Administration" 1990 "New Directions in Merger Enforcement: The View from Washington" 1990
Woodrow Wilson School Alumni Colloquium "Microeconomic Policy Analysis and Antitrust--Washington 1990" 1990
Arnold & Porter Lecture Series "Advocating Competition" 1991 "Antitrust Enforcement" 1990
ABA Antitrust Section Convention "Recent Developments in Market Definition and Merger Analysis" 1990
Federal Bar Association "Joint Production Legislation: Competitive Necessity or Cartel Shield?" 1990
Pew Charitable Trusts Conference "Economics and National Security" 1990
ABA Antitrust Section Midwinter Council Meeting "Fine-tuning the Merger Guidelines" 1990 "The State of the Antitrust Division" 1991
International Telecommunications Society Conference "Discussion of the Impact of Telecommunications in the UK" 1989
The Economists of New Jersey Conference "Recent Perspectives on Regulation" 1989
Public Version
A-29
Conference on Current Issues Challenging the Regulatory Process "Innovative Pricing and Regulatory Reform" 1989 "Competitive Wheeling" 1989
Conference Board: Antitrust Issues in Today's Economy "Foreign Trade Issues and Antitrust" 1989
McKinsey & Co. Mini-MBA Conference "Economic Analysis of Pricing, Costing, and Strategic Business Behavior" 1989
1994
Olin Conference on Regulatory Mechanism Design "Revolutions in Regulatory Theory and Practice: Exploring The Gap" 1989
University of Dundee Conference on Industrial Organization and Strategic Behavior "Mergers in Differentiated Product Industries" 1988
Leif Johanson Lectures at the University of Oslo "Normative Issues in Industrial Organization" 1988
Mergers and Competitiveness: Spain Facing the EEC "Merger Policy" 1988 "R&D Joint Ventures" 1988
New Dimensions in Pricing Electricity "Competitive Pricing and Regulatory Reform" 1988
Program for Integrating Economics and National Security: Second Annual Colloquium "Arming Decisions Under Asymmetric Information" 1988
European Association for Research in Industrial Economics "U.S. Railroad Deregulation and the Public Interest" 1987 "Economic Rationales for the Scope of Privatization" 1989 "Discussion of Licensing of Innovations" 1990
Annenberg Conference on Rate of Return Regulation in the Presence of Rapid Technical Change "Discussion of Regulatory Mechanism Design in the Presence of Research, Innovation, and Spillover Effects" 1987
Special Brookings Papers Meeting "Discussion of Empirical Approaches to Strategic Behavior" 1987 "New Merger Guidelines" 1990
Deregulation or Regulation for Telecommunications in the 1990's "How Effective are State and Federal Regulations?" 1987
Public Version
A-30
Conference Board Roundtable on Antitrust "Research and Production Joint Ventures" 1990 "Intellectual Property and Antitrust" 1987
Current Issues in Telephone Regulation "Economic Approaches to Market Dominance: Applicability of Contestable Markets" 1987
Harvard Business School Forum on Telecommunications "Regulation of Information Services" 1987
The Fowler Challenge: Deregulation and Competition in The Local Telecommunications Market
"Why Reinvent the Wheel?" 1986
World Bank Seminar on Frontiers of Economics "What Every Economist Should Know About Contestable Markets" 1986
Bell Communications Research Conference on Regulation and Information "Fuzzy Regulatory Rules" 1986
Karl Eller Center Forum on Telecommunications "The Changing Economic Environment in Telecommunications: Technological Change and Deregulation" 1986
Railroad Accounting Principles Board Colloquium "Contestable Market Theory and ICC Regulation 1986
Canadian Embassy Conference on Current Issues in Canadian -- U.S. Trade and Investment "Regulatory Revolution in the Infrastructure Industries" 1985
Eagleton Institute Conference on Telecommunications in Transition "Industry in Transition: Economic and Public Policy Overview" 1985
Brown University Citicorp Lecture "Logic of Regulation and Deregulation" 1985
Columbia University Communications Research Forum "Long Distance Competition Policy" 1985
American Enterprise Institute Public Policy Week "The Political Economy of Regulatory Reform" 1984
MIT Communications Forum "Deregulation of AT&T Communications" 1984
Public Version
A-31
Bureau of Census Longitudinal Establishment Data File and Diversification Study Conference "Potential Uses of The File" 1984
Federal Bar Association Symposium on Joint Ventures "The Economics of Joint Venture Assessment" 1984
Hoover Institute Conference on Antitrust "Antitrust for High-Technology Industries" 1984
NSF Workshop on Predation and Industrial Targeting "Current Economic Analysis of Predatory Practices" 1983
The Institute for Study of Regulation Symposium: Pricing Electric, Gas, and Telecommunications Services Today and for the Future
"Contestability As A Guide for Regulation and Deregulation" 1984
University of Pennsylvania Economics Day Symposium "Contestability and Competition: Guides for Regulation and Deregulation" 1984
Pinhas Sapir Conference on Economic Policy in Theory and Practice "Corporate Governance and Market Structure" 1984
Centre of Planning and Economic Research of Greece "Issues About Industrial Deregulation" 1984 "Contestability: New Research Agenda" 1984
Hebrew and Tel Aviv Universities Conference on Public Economics "Social Welfare Dominance Extended and Applied to Excise Taxation" 1983
NBER Conference on Industrial Organization and International Trade "Perspectives on Horizontal Mergers in World Markets" 1983
Workshop on Local Access: Strategies for Public Policy "Market Structure and Government Intervention in Access Markets" 1982
NBER Conference on Strategic Behavior and International Trade "Industrial Strategy with Committed Firms: Discussion" 1982
Columbia University Graduate School of Business, Conference on Regulation and New Telecommunication Networks
"Local Pricing in a Competitive Environment" 1982
International Economic Association Roundtable Conference on New Developments in the Theory of Market Structure
Public Version
A-32
"Theory of Contestability" 1982 "Product Dev., Investment, and the Evolution of Market Structures" 1982
N.Y.U. Conference on Competition and World Markets: Law and Economics "Competition and Trade Policy--International Predation" 1982
CNRS-ISPE-NBER Conference on the Taxation of Capital "Welfare Effects of Investment Under Imperfect Competition" 1982
Internationales Institut fur Management und Verwalturg Regulation Conference "Welfare, Regulatory Boundaries, and the Sustainability of Oligopolies" 1981
NBER-Kellogg Graduate School of Management Conference on the Econometrics of Market Models with Imperfect Competition
"Discussion of Measurement of Monopoly Behavior: An Application to the Cigarette Industry" 1981
The Peterkin Lecture at Rice University "Deregulation: Ideology or Logic?" 1981
FTC Seminar on Antitrust Analysis "Viewpoints on Horizontal Mergers 1982 "Predation as a Tactical Inducement for Exit" 1980
NBER Conference on Industrial Organization and Public Policy "An Economic Definition of Predation" 1980
The Center for Advanced Studies in Managerial Economics Conference on The Economics of Telecommunication
"Pricing Local Service as an Input" 1980
Aspen Institute Conference on the Future of the Postal Service "Welfare Economics of Postal Pricing" 1979
Department of Justice Antitrust Seminar "The Industry Performance Gradient Index" 1979
Eastern Economic Association Convention "The Social Performance of Deregulated Markets for Telecom Services" 1979
Industry Workshop Association Convention "Customer Equity and Local Measured Service" 1979
Symposium on Ratemaking Problems of Regulated Industries "Pricing Decisions and the Regulatory Process" 1979
Public Version
A-33
Woodrow Wilson School Alumni Conference "The Push for Deregulation" 1979
NBER Conference on Industrial Organization "Intertemporal Sustainability" 1979
World Congress of the Econometric Society "Theoretical Industrial Organization" 1980
Institute of Public Utilities Conference on Current Issues in Public Utilities Regulation "Network Access Pricing" 1978
ALI-ABA Conference on the Economics of Antitrust "Predatoriness and Discriminatory Pricing" 1978
AEI Conference on Postal Service Issues "What Can Markets Control?" 1978
University of Virginia Conference on the Economics of Regulation "Public Interest Pricing" 1978
DRI Utility Conference "Marginal Cost Pricing in the Utility Industry: Impact and Analysis" 1978
International Meeting of the Institute of Management Sciences "The Envelope Theorem" 1977
University of Warwick Workshop on Oligopoly "Industry Performance Gradient Indexes" 1977
North American Econometric Society Convention"Intertemporal Sustainability" 1979"Social Welfare Dominance" 1978"Economies of Scope, DAIC, and Markets with Joint Production" 1977
Telecommunications Policy Research Conference "Transition to Competitive Markets" 1986"InterLATA Capacity Growth, Capped NTS Charges and LongDistance Competition" 1985"Market Power in The Telecommunications Industry" 1984"FCC Policy on Local Access Pricing" 1983"Do We Need a Regulatory Safety Net in Telecommunications?" 1982"Anticompetitive Vertical Conduct" 1981"Electronic Mail and Postal Pricing" 1980"Monopoly, Competition and Efficiency": Chairman 1979
Public Version
A-34
"A Common Carrier Research Agenda" 1978"Empirical Views of Ramsey Optimal Telephone Pricing" 1977"Recent Research on Regulated Market Structure" 1976"Some General Equilibrium Views of Optimal Pricing" 1975
National Bureau of Economic Research Conference on Theoretical Industrial Organization "Compensating Variation as a Measure of Welfare Change" 1976
Conference on Pricing in Regulated Industries: Theory & Application "Ramsey Optimal Pricing of Long Distance Telephone Services" 1977
NBER Conference on Public Regulation "Income Distributional Concerns in Regulatory Policy-Making" 1977
Allied Social Science Associations National Convention"Merger Guidelines and Economic Theory" 1990Discussion of "Competitive Rules for Joint Ventures" 1989"New Schools in Industrial Organization" 1988"Industry Economic Analysis in the Legal Arena" 1987"Transportation Deregulation" 1984Discussion of "Pricing and Costing of Telecommunications Services" 1983Discussion of "An Exact Welfare Measure" 1982"Optimal Deregulation of Telephone Services" 1982"Sector Differentiated Capital Taxes" 1981"Economies of Scope" 1980"Social Welfare Dominance" 1980"The Economic Definition of Predation" 1979Discussion of "Lifeline Rates, Succor or Snare?" 1979"Multiproduct Technology and Market Structure" 1978"The Economic Gradient Method" 1978"Methods for Public Interest Pricing" 1977Discussion of "The Welfare Implications of New Financial Instruments" 1976"Welfare Theory of Concentration Indices" 1976Discussion of "Developments in Monopolistic Competition Theory" 1976"Hedonic Price Adjustments" 1975"Public Good Attributes of Information and its Optimal Pricing" 1975"Risk Invariance and Ordinally Additive Utility Functions" 1974"Consumer's Surplus: A Rigorous Cookbook" 1974
University of Chicago Symposium on the Economics of Regulated Public Utilities "Optimal Prices for Public Purposes" 1976
American Society for Information Science "The Social Value of Information: An Economist's View" 1975
Public Version
A-35
Institute for Mathematical Studies in the Social Sciences Summer Seminar
Public Version
A-36
"The Sustainability of Natural Monopoly" 1975
U.S.-U.S.S.R. Symposium on Estimating Costs and Benefits of Information Services "The Evaluation of the Economic Benefits of Productive Information" 1975
NYU-Columbia Symposium on Regulated Industries "Ramsey Optimal Public Utility Pricing" 1975
Research Seminars:
Bell Communications Research (2) University of California, San Diego
Bell Laboratories (numerous) University of Chicago
Department of Justice (3) University of Delaware
Electric Power Research Institute University of Florida
Federal Reserve Board University of Illinois
Federal Trade Commission (4) University of Iowa (2)
Mathematica Universite Laval
Rand University of Maryland
World Bank (4) University of Michigan
Carleton University University of Minnesota
Carnegie-Mellon University University of Oslo
Columbia University (4) University of Pennsylvania (3)
Cornell University (2) University of Toronto
Georgetown University University of Virginia
Harvard University (2) University of Wisconsin
Hebrew University University of Wyoming
Johns Hopkins University (2) Vanderbilt University
M. I. T. (4) Yale University (2)
New York University (4) Princeton University (many)
Northwestern University (2) Rice University
Norwegian School of Economics and Stanford University (5)
Business Administration S.U.N.Y. Albany
Public Version
A-37
Expert Testimony of Robert Willig
September 23, 2015 to September 23, 2019
In re: Domestic Drywall Antitrust Litigation, In the United States District Court for the Eastern District of Pennsylvania, MDL No. 2437 13-MD-2437, Expert Report 03/13/15; Deposition 4/9/15, 4/10/15; Expert Report 01/26/18; Expert Report 02/26/18; Deposition 05/31/18.
Methodist Health Services Corporation v. OSF Healthcare System, In the United States District Court for the Central District of Illinois, Peoria Division, Case No.: 13-cv-1054, Expert Report 8/14/2015, Deposition 10/8/2015, Reply Report, 9/2016.
Application of the National Railroad Passenger Corporation Under 49 U.S.C. § 24308(a) – Canadian National Railway Company; Before the Surface Transportation Board, Docket No. FD 35743; Verified Statement, 9/4/2015; Rebuttal Verified Statement, 9/14/2017.
BRFHH Shreveport, LLC d/b/a University Health Shreveport and Vantage Health Plan, Inc. v. Willis-Knighton Medical Center, d/b/a Willis-Knighton Health System, In the United States District Court for the Western District of Louisiana, Shreveport Division, Case No.: 5:15-CV-02057, Joint Declaration of Margaret E. Guerin-Calvert and Robert D. Willig 9/8/2015, Expert Report 3/23/2017, Deposition 5/12/2017.
Australian Consumer and Competition Commission v. Informed Sources Pty Ltd & Ors, Before the Federal Court of Australia, Victoria Registry, File number VID450/2014, Expert Report 11/24/15.
Clark R. Huffman, Brandi K. Winters, Patricia L. Grantham, and Linda M. Pace, Individually and on behalf of all others similarly situated vs. The Prudential Insurance Company of America, In the United States District Court for the Eastern District of Pennsylvania, Civ. No. 2:10-cv-05135-EL, Expert Report 2/25/16, Deposition 4/5/16.
Maxon Hyundai Mazda, et al., vs. Carfax Inc., In the United States District Court for the Southern District of New York, Case No.: 13 CV 2680 (AJN) (RLE), Expert Report 2/26/16, Deposition 4/21/16.
Federal Trade Commission and Commonwealth of Pennsylvania vs. Penn State Hershey Medical Center and Pinnacle Health System, In the United States District Court for the Middle District of Pennsylvania, Civil Action No. 1:15-cv-02362, Expert Report 3/7/16, Deposition 3/25/16, Trial Testimony 4/15/16.
In the Matter of Business Data Services in an Internet Protocol Environment; Special Access for Price Cap Local Exchange Carriers; AT&T Corporation Petition for Rulemaking to Reform Regulation of Incumbent Local Exchange Carrier Rates for Interstate Special Access Services; Before the FCC; WC Docket No. 16-143; WC Docket No. 05-25; RM-10593; Declaration 8/8/16.
Public Version
A-38
United States et al. v. Anthem Inc. and Cigna Corp., In the United States District Court for the District of Columbia, Civil Action No. 16-cv-01493 (ABJ), Expert Report 10/7/16, Rebuttal and Supplemental Expert Report 10/28/16, Deposition 11/9/2016, Trial Testimony 12/2/16 and 1/3/17.
In the Matter of: Determination of Royalty Rates and Terms for Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Before the United States Copyright Royalty Judges, Washington, D.C., Docket No. 16-CRB-0001 SR/PSSR (2018-2022), Written Direct Testimony 10/19/16, Written Rebuttal Testimony 2/13/2017, Deposition 03/31/17, Trial Testimony 05/02/17 and 05/04/17.
Petition for Rulemaking to Adopt Revised Competitive Switching Rules, Before the Surface Transportation Board, Docket Number EP 711 (Sub-No. 1), Verified Statement 10/26/16.
In re: Evanston Northwestern Healthcare Corporation Antitrust Litigation, In the United States District Court Northern District of Illinois Eastern Division, Master File No. 07-CV-4446, Expert Report 10/26/16, Deposition 1/12/17.
Independent Power Producers of New York v. New York Independent System Operator, Before the Federal Energy Regulatory Commission, Docket No. EL13-62-002, Declaration 1/26/17.
Calpine Corporation et. al. v. PJM Interconnection, L.L.C., Before the Federal Energy Regulatory Commission, Docket No. EL16-49-000, Declaration 1/30/17.
In re: LIBOR-Based Financial Instruments Antitrust Litigation, Mayor and City Council of Baltimore, et. al. v. Credit Suisse Group AG, et. al., in the United States District Court Southern District of New York, MDL No. 2262, Master File No. 1:11-md-2262-NRB, 11-cv-5450 (NRB), Expert Report 4/3/17, Deposition 5/19/17.
In re: LIBOR-Based Financial Instruments Antitrust Litigation, Metzler Investment GmbH, et. al. v. Credit Suisse Group AG, et. al., in the United States District Court Southern District of New York, MDL 2262, 11 Civ. 2613, Master File No. 1:11-md-2262-NRB, 11-cv-5450 (NRB), Expert Report 4/3/17, Deposition 6/8/17.
In re: LIBOR-Based Financial Instruments Antitrust Litigation, Berkshire Bank, Government Development Bank v. Bank of America Corp., et. al., in the United States District Court Southern District of New York, MDL No. 2613, Master File No. 1:11-md-2262-NRB, 12-cv-5723-NRB, Expert Report 4/3/17, Deposition 4/21/17.
The Coca-Cola Company & Subs. v. Commissioner of Internal Revenue, in the United States Tax Court, Docket No. 31183-15, Expert Report 6/29/2017, Rebuttal Report 9/28/17, Deposition 12/19/17, Trial Testimony 4/5/18 and 4/6/18.
In re: Automotive parts antitrust litigation -- Bearings Cases; in the United States District Court for the Eastern District of Michigan Southern Division, Master File No. 12-md-02311; 2:12-cv-00501-MOB-MKM; Declaration in Support of Defendants’ Opposition to Direct Purchaser Plaintiffs’
Public Version
A-39
Motion for Class Certification and Appointment of Class Counsel, 7/26/2017, Deposition 10/13/2017; Declaration in Support of Defendants’ Reply in Support of Defendants’ Joint Motion to Exclude the Proposed Expert Reports and Testimony of Drs. McClave and Langenfeld, 12/14/17.
In re: General Motors LLC Ignition Switch Litigation, in the U.S. District Court for the Southern District of New York, No. 14-MD-2543 (JMF), Expert Report 2/23/2018, Expert Report 4/20/2018, Supplemental Tables, 5/7/2018, Deposition 5/8/2018 and 5/9/2018.
CPV Power Holdings et. al. v. PJM Interconnection, L.L.C., Before the Federal Energy Regulatory Commission, Docket No. EL18-169-000, Declaration 06/20/18.
In Re Thalomid and Revlimid Antitrust Litigation, in the U.S, District Court for the District of New Jersey, Civil No. 14-6997 (MCA) (MAH), Expert Report 08/27/18, Deposition 10/25/2018.
In Re Djeneba Sidibe et al, v. Sutter Health, Case No. 3: 12-cv-4854-LB, in the U.S. District Court for the Northern District of California, Expert Declaration 9/21/2018, Deposition 11/8/2018, Sur-Reply Declaration 1/7/2019, Expert Report 6/21/2019, Deposition 7/24/2019.
In Re UFCW & Employers Benefit Trust, et al v. Sutter Health, et al., Case No. CGC-14-538451, Consolidated with Case No. CGC-18-565398, People of the State of California, ex rel. et al v. Sutter Health, before the Superior Court of the State of California for the City and County of San Francisco, Expert Report 10/29/2018, Deposition 12/17-18/2018, Addendum to Expert Report 3/26/2019, Response to Supplemental Merits Expert Declaration of Jeffrey J. Leitzinger 4/16/2019, Response to Second Supplemental Expert Declaration of Jeffrey J. Leitzinger 6/10/2019.
In Re Christopher Dicesare et al v. The Charlotte-Mecklenburg Hospital Authority, d/b/a Carolinas Healthcare System, in the General Court of Justice State of North Carolina Superior Court Division County of Mecklenburg, Case No. 16-CVS-16404, Expert Report 12/7/2018, Deposition 3/8/2019.
In Re Epipen (Epinephrine Injection, USP) Marketing, sales practices and antitrust litigation; Sanofi-Aventis US, LLC v. Mylan Inc., et al.; before the United States District Court for the District of Kansas, Case No. 2:17-md-2785 and Case No. 2:17-cv-2452; expert report March 25, 2019; Deposition May 19, 2019.
Public Version
B-1
Appendix B: Materials Relied On
Academic Articles • Allan Collard-Wexler, Gautam Gowrisankaran, and Robin Lee, “‘Nash-in-Nash’
Bargaining: A Microfoundation for Applied Work,” Journal of Political Economy, Vol. 127, No. 1 (2019): 163-195
• Carl Shapiro, “Mergers with Differentiated Products,” Antitrust, No. 10 (1996): 23-30 • Gautam Gowrisankaran, Aviv Nevo, and Robert Town, “Mergers When Prices Are
Negotiated: Evidence from the Hospital Industry,” American Economic Review, Vol. 105, No. 1 (2015): 172-203
• Gregory S. Crawford, Robin S. Lee, Michael D. Whinston, and Ali Yurukoglu, “The Welfare Effects of Vertical Integration in Multichannel Television Markets,” Econometrica, Vol. 86, No. 3 (2018): 891-954
• Henrick Horn and Asher Wolinsky, “Bilateral monopolies and incentives for merger,” RAND Journal of Economics, Vol. 19, No. 3 (1988): 408-419
• Ian McDonald and Robert Solow, “Wage Bargaining and Employment,” American Economic Review, Vol. 71, No. 5 (1981): 896-908
• John Nash, “The Bargaining Problem,” Econometrica, Vol. 18, No. 2 (1950): 155-162
Textbooks • Andreu Mas-Colell, Michael D. Whiston, and Jerry R. Green, Microeconomic Theory,
Oxford University Press (1995)• Eyal Winter, “The Shapley Value,” In R.J. Aumann and S. Hart (Eds.), Handbook of
Game Theory, Vol. 3, Chapter 53, Elsevier Science (2002) • Hervé Moulin, Fair Division and Collective Welfare, MIT Press (2003)• Holger Ingmar Meinhardt, The Pre-Kernel as a Tractable Solution for Cooperative
Games, An Exercise in Algorithmic Game Theory, (Springer-Verlag: 2014) • Jean Tirole, The Theory of Industrial Organization, (MIT Press: 1988) • Lloyd S. Shapley, “A value for n-person games,” (1953) In Alvin E. Roth (Ed.), The
Shapley Value, Essays in honor of Lloyd S. Shapley, Cambridge University Press (1988)• Michael Carter, “Cooperative Games,” In Hal R. Varian (Ed.), Economic and Financial
Modeling with Mathematica, Springer-Verlag (1993) • Robert Willig, “Merger Analysis, Industrial Organization Theory, and Merger
Guidelines,” Brookings Papers: Microeconomics (1991)
CRB Documents • Determination, United States Copyright Royalty Judges, The Library of Congress, In re
Determination of Royalty Rates and Terms for Ephemeral Recording and Webcasting Digital Performance of Sound Recordings (Web IV), Docket No. 14-CRB-0001-WR
• Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms For Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Docket No. 16-CRB-0001 SR/PSSR
Public Version
B-2
• Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms for Making and Distributing Phonorecords (Phonorecords III), Docket No. 16-CRB-0003-PR
• Federal Register, Vol. 84, No. 16, January 24, 2019
Written Direct Testimony • Written Direct Testimony of Jason Gallien, September 23, 2019 • Written Direct Testimony of Aaron Harrison, September 23, 2019 • Written Direct Testimony of Mark Piibe, September 23, 2019 • Written Direct Testimony of Professor Gal Zauberman, September 23, 2019 • Written Direct Testimony of Robert Willig, October 19, 2016, In the Matter of
Determination of Royalty Rates and Terms for Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Docket No. 16-CRB-0001 SR/PSSR (2018-2022)
Data • “662009_Final_AllStarts_091712.xlsx” Survey Data Results from Professor Gal
Zauberman (SoundExchange Exhibit 29) • “bootstrap_q1_wide.csv” Survey Data Results from Professor Gal Zauberman
(SoundExchange Exhibit 30) • “Demographics.xlsx” Survey Data Results from Professor Gal Zauberman
(SoundExchange Exhibit 31) • Sirius XM SDARS December 2018.pdf (SoundExchange Exhibit 32) • Sirius XM SDARS April 2019.pdf (SoundExchange Exhibit 33) • RIAA US Wholesale revenue summary 2012 – 2018 CONFIDENTIAL.xlsx
(SoundExchange Exhibit 34) • 2018 RIAA shipments data update.xlsx (SoundExchange Exhibit 35) • 2019-8-6 SOA and Allocations 2016 to 2019 for Web V.xlsm (SoundExchange Exhibit
36) • 2019-08-08 Performances iHeart Pandora Rhapsody.xlsx (SoundExchange Exhibit 37) • Pandora Market Capitalization from Bloomberg (SoundExchange Exhibit 38)
Royalty Statements • Amazon Music Unlimited-UMG Royalty Statements Covering the Period April 2018 to
March 2019 • Amazon Music Unlimited-SME Royalty Statements Covering the Period April 2018 to
March 2019 • Amazon Music Unlimited-WMG Royalty Statements Covering the Period April 2018 to
March 2019 • Amazon Prime-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Amazon Prime-SME Royalty Statements Covering the Period April 2018 to March 2019 • Amazon Prime-WMG Royalty Statements Covering the Period October 2018 to March
2019 • Apple Music-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Apple Music-SME Royalty Statements Covering the Period April 2018 to March 2019
Public Version
B-3
• Apple Music-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Google-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Google-SME Royalty Statements Covering the Period April 2018 to March 2019 • Google-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Google-Merlin Royalty Statements Covering the Period April 2018 to March 2019 • iHeart Free-WMG Royalty Statements Covering the Period April 2018 to March 2019 • iHeart Subscription-UMG Royalty Statements Covering the Period April 2018 to March
2019 • iHeart Subscription-SME Royalty Statements Covering the Period April 2018 to March
2019 • iHeart Subscription-WMG Royalty Statements Covering the Period April 2018 to March
2019 • iHeart Subscription-Merlin Royalty Statements Covering the Period April 2018 to March
2019 • Pandora-UMG Royalty Statements Covering the Period January 2018 to March 2019 • Pandora Premium-SME Royalty Statements Covering the Period April 2018 to March
2019 • Pandora Radio Plus-SME Royalty Statements Covering the Period January 2018 to
March 2019 • Pandora Radio-SME Royalty Statements Covering the Period January 2018 to March
2019 • Pandora Premium-WMG Royalty Statements Covering the Period April 2018 to March
2019 • Pandora Radio Plus-WMG Royalty Statements Covering the Period January 2018 to
March 2019 • Pandora Radio-WMG Royalty Statements Covering the Period January 2018 to March
2019 • Pandora-Merlin Royalty Statements Covering the Period January 2018 to March 2019 • Napster (Rhapsody)-UMG Royalty Statements Covering the Period April 2018 to March
2019 • Napster (Rhapsody)-SME Royalty Statements Covering the Period April 2018 to March
2019 • Napster (Rhapsody)-WMG Royalty Statements Covering the Period April 2018 to March
2019 • Spotify-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Spotify-SME Royalty Statements Covering the Period April 2018 to March 2019 • Spotify-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Spotify-Merlin Royalty Statements Covering the Period April 2018 to March 2019 • Vevo-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Vevo-SME Royalty Statements Covering the Period April 2018 to March 2019 (except
May 2018) • Vevo-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Vevo-Merlin Royalty Statements Covering the Period April 2018 to March 2019
Public Version
B-4
• YouTube Streams-UMG Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Streams-SME Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Streams-WMG Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Subscription-UMG Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Subscription-SME Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Subscription-WMG Royalty Statements Covering the Period April 2018 to March 2019
License Agreements • [
] (SoundExchange Exhibit 39) • [
(SoundExchange Exhibit 40) • [ ]
(SoundExchange Exhibit 41)
Industry Reports • MusicWatch, “Annual Music Study 2018: Report to RIAA,” April 2019 (SoundExchange
Exhibit 42) • Edison Research, “Share of Ear,” Q3 2018 (SoundExchange Exhibit 43) • Edison Research, “Share of Ear,” Q2 2019 (SoundExchange Exhibit 44) • Q2 2019 Share of Ear Survey Methodology Statement (SoundExchange Exhibit 72) • JMP Securities, “SiriusXM Announces All-Stock Agreement to Acquire Pandora,”
September 24, 2018 (SoundExchange Exhibit 45)
SEC Filings • Pandora Media, Inc. Form 10-K for the Fiscal Year Ended December 31, 2017
(SoundExchange Exhibit 46) • Pandora Media, Inc. Form 8-K dated September 23, 2018 (SoundExchange Exhibit 47)• Pandora Media, Inc. Schedule 14A Definitive Proxy Statement, December 20, 2018
(SoundExchange Exhibit 48) • Sirius XM Holdings Inc., Form 10-K for the Fiscal Year Ended December 31, 2018
(SoundExchange Exhibit 49) • Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended March 31, 2018
(SoundExchange Exhibit 50) • Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended June 30, 2018
(SoundExchange Exhibit 51) • Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended September 30, 2018
(SoundExchange Exhibit 52)
Public Version
B-5
• Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended March 31, 2019 (SoundExchange Exhibit 53)
• Spotify Technology S.A., Form 20-F for the Fiscal Year Ended December 31, 2018 (SoundExchange Exhibit 71)
Bates-Stamped Documents • GOOG-WEBV-00036425 (SoundExchange Exhibit 54) • GOOG-WEBV-00036688 (SoundExchange Exhibit 55) • GOOG-WEBV-00036873 (SoundExchange Exhibit 56) • PANWEBV_00002985 (SoundExchange Exhibit 57) • PANWEBV_00003357 (SoundExchange Exhibit 58) • PANWEBV_00003425 (SoundExchange Exhibit 59) • PANWEBV_00003868 (SoundExchange Exhibit 60) • PANWEBV_00004139 (SoundExchange Exhibit 61) • PANWEBV_00004249 (SoundExchange Exhibit 62) • PANWEBV_00004335 (SoundExchange Exhibit 63) • SXMWEBV_00004542 (SoundExchange Exhibit 64) • SXMWEBV_00005082 (SoundExchange Exhibit 65)
Other • 17 U.S.C § 114(f)(2)(B) • Transcript of Spotify Technology SA at Goldman Sachs Communacopia Conference,
September 14, 2018 (SoundExchange Exhibit 66) • “AAPOR Guidance on Reporting Precision for Nonprobability Samples,” American
Association for Public Opinion Research, available at: https://www.aapor.org/getattachment/Education-Resources/For-Researchers/AAPOR_Guidance_Nonprob_Precision_042216.pdf.aspx (SoundExchange Exhibit 67)
• Merlin: http://www.merlinnetwork.org/whowelicense (SoundExchange Exhibit 68) • Federal Reserve Bank of St. Louis data: https://fred.stlouisfed.org/series/JCXFECTM
(SoundExchange Exhibit 22)
Public Version
C-1
Appendix C: Shapley Value Model
I. Model
1. This appendix provides a detailed description of my Shapley Value model. This
model involves two noninteractive streaming distributors (a free ad-supported distributor and a
paid subscription distributor), three “must-have” record companies (representing the majors) and
one “not-must-have” record company (representing all the indies). In constructing this Shapley
Value model, I utilize the following notation:
Figure C-1: Notation
D1 and D2The ad-supported and subscription noninteractive streaming distributors, respectively.
LA, LB, and LC The three must-have major record companies.
LD
An amalgam of not-must-have indie record companies grouped together without loss of generality. Referred to hereafter as the not-must-have record company.
SharekRecord company k’s share of plays, which is assumed to be the same across all distributors.
CkRecord company k’s costs of producing sound recordings.
��� and ��
�D1’s marginal profit per play and D2’s marginal profit per play, respectively, before accounting for any sound recording royalties that they pay.
F1 and F2 D1’s and D2’s fixed costs, respectively.
Dj
where � ∈ {3, …, N}
Distributors whose royalty rates are set outside of the current proceedings, such as ad-supported and subscription interactive music and video streaming services, Sirius XM satellite radio, and music purchases (i.e., CDs, vinyl, digital downloads).
���
The royalty rate paid by outside distributor j per royalty-bearing unit � ∈ {play, subscriber, CD/vinyl-record/download}.
���
Di’s audience, measured in Di’s royalty bearing units t (e.g., plays for D1), when all other distributors are operating.
���The average number of monthly plays per user of D1’s service.
���The average number of monthly plays per user of D2’s service.
��,��
A diversion ratio from service i to j: Per current user of noninteractive streaming distributor i’s service, the change in royalty-bearing units t on distributor j’s service if Di were to stop operating while all other distributors continued operating.
Public Version
C-2
� = � ���� ∗ ��
���
���
The four record companies’ combined royalties from outside distributors when all distributors are operating.
�� = � ���� ∗ ��,�
� ��
���
Per current user of noninteractive streaming distributor i’s service, the change in the four record companies’ combined royalties from outside distributors if Di were to stop operating while all other distributors continued operating.
�′�,��
Another diversion ratio from service i to j: The fraction of current users of noninteractive distributor i’s service who would divert to outside distributor Dj’s service if service i were to stop operating, while the other noninteractive distributor were also not operating. This diversion is measured as the change in royalty-bearing units t on service jper then-current user of service i.
��� = ����
� ∗ �′�,�� �
�
���
Per then-current user of noninteractive distributor i’s service, the change in the four record companies’ combined royalties from outside distributors if noninteractive distributor Di were to stop operating while the other non-interactive distributor were also not operating.
2. The Shapley Value solution concept requires calculating the average incremental
contribution to the collective surplus that each record company or distributor creates, based on all
possible orderings of the parties. The first step in this process is to calculate the total surplus each
hypothetical coalition of noninteractive streaming distributors and record companies would
generate if they acted in a manner that maximized their collective surplus. These values define a
function �[�] of all those coalitions � ∈ �(�), called the “characteristic function,” where �(�) is
the set of subsets of the players � = {��,��, ��, ��, �� , ��}.
3. In the present case, the characteristic function can be determined recursively. An
empty coalition and coalitions comprised of only the distributors create zero payoffs. Hence:
�[{}] = 0;
�[{��}] = 0;
�[{��}] = 0;
�[{��,��}] = 0.
4. Each record company, when it is alone, generates its current profits from outside
Public Version
C-3
distributors plus the additional profits from outside distributors caused by diversion from
distributors 1 and 2 to the outside distributors. Hence:1
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��.
5. Since LA, LB, and LC are must-haves, only when all three are present in a coalition
can the distributor(s) begin making profits. When only these three record companies are present
in a coalition, three cases can arise. First, if only D1 is in a coalition, and if it increases the
coalition’s collective surplus, diversion occurs from D2 to both D1 and the outside distributors
(because D2 is not operating) and from D1 to outside distributors (because D1 does not have a
license to LD’s content). Second, if only D2 is in a coalition, and if it increases the coalition’s
collective surplus, diversion occurs from D1 to both D2 and the outside distributors (because D1 is
not operating) and from D2 to outside distributors (because D2 does not have a license to LD’s
content). And third, if both distributors are in a coalition, and if they increase the coalition’s
collective surplus, no diversion between D1 and D2 occurs, but there is still diversion from D1 and
D2 to the outside distributors (because they do not have licenses to LD’s content).2
1 I assume that it is profitable for the must-have record companies to operate. 2 In keeping with the Judges’ conclusion in the Web IV proceeding that “the repertoire of each of the three Majors is a ‘must have’ in order for a noninteractive service to be viable,” I assume that a distributor could not operate without agreements with all three of the major record companies. See Web IV Determination at 133. Conversely, I assume that if a distributor has access to the must-have record companies’ content but does not have access to LD’s content, the distributor loses �ℎ���� of its plays (or subscribers).
Public Version
C-4
6. Each coalition is assumed to act in the manner that maximizes the collective surplus
of the coalition.3 Therefore:
�[{��,��,��,��}]
= ��� �(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗��,��
����+ � + ��
�∗������− (�� + �� + ��)
− ��,�[{��,�� ,��}]� ; 4
�[{��,��,�� ,��}]
= ��� �(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗��,��
���� + � + ��
�∗������− (�� + �� + ��)
− ��,�[{��,��,��}]� ;
�[{��,��,��,��,��}]
= ����(1 − �ℎ����) ∗ ����∗ ��
�+ ��
�∗ ��
�+ ��− (�� + �� + ��)− ��
− ��,�[{��,��,��,��}],�[{��,��,�� ,��}]�.
7. When LD joins a coalition, it adds to the coalition’s surplus by contributing its
royalties from outside distributors, and also by licensing one or both distributors in the coalition
(if such a license would increase the coalition’s collective surplus). Thus:
�[{��,��,��,�� ,��}]
= ��� ��� + ���∗�����
+ ���∗ ���
�+ ��
�∗��,��
�����− (�� + �� + �� + ��)
− ��,�[{��,��,�� ,��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
3 In some cases, the collective surplus of a coalition may be higher if one or both distributors do not operate.4 As explained below, �[{��,�� ,��}] is the combined profits of the coalition that includes only the three must-have record companies. �[{��,��,��}] = �[{��}] + �[{��}] + �[{��}].
Public Version
C-5
�[{��,��,�� ,�� ,��}]
= ��� ��� + ���∗�����
+ ���∗ ���
�+ ��
�∗��,��
�����− (�� + �� + �� + ��)
− ��,�[{��,��,��, ��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
�[{��,��,��,��,�� ,��}]
= ��� ��� + ���∗ ��
�+ ��
�∗ ��
��− (�� + �� + �� + ��)− ��
− ��,�[{��,��,��,�� ,��}],�[{��,��,�� ,�� ,��}],�[{��,��,��,��,��}]
+ �[{��}],� + ���∗ ���
�+ �ℎ���� ∗ ��
�∗��,��
����− �� − (�� + �� + �� + ��)
+ (1 − �ℎ����) ∗���∗ ��
�+ �ℎ���� ∗ ���
�∗������− ��,� +��
�
∗ ����
+ �ℎ���� ∗ ���∗��,��
����− �� − (�� + �� + �� + ��) + (1 − �ℎ����) ∗��
�
∗ ���
+ �ℎ���� ����∗������− ���.
8. In any other case not defined above, the noninteractive distributors do not carry one
or more of the must-have record companies and cannot operate. In such cases, all the profits
generated by those coalitions are just the sum of the �[{��}] defined above. For example,
�[{��,��}] = �[{��}] + �[{��}],�[{��,��,��,��}] = �[{��}] + �[{��}], etc.
9. Based on this characteristic function, the Shapley Value algorithm can be
implemented. Specifically, I evaluate every possible arrival ordering and for each determine the
incremental value contributed by each noninteractive distributor relative to the value created by
the parties that preceded its arrival. Then I average all of these marginal contributions across all
arrival orderings. With six players, there are 6! (i.e., 720) arrival orderings. Algebraically, this
can be expressed as follows. For every possible subset � of � that includes i,
��� = ���(�[�] − �[� − {�}])
�⊂�
,
where
Public Version
C-6
�� =(|�| − 1)! ∗ (� − |�|)!
�!,
for n equal to the number of elements in � (six in this case) and |�| the number of elements in
subset �.5
10. Following this formula, I calculate the Shapley Values for both noninteractive
distributors, based on the parameters estimated as described in Appendices D and E. The total
royalty payments to be made by D1 and D2 can be calculated using these Shapley Values as
follows:
�� = ��� ∗ ��
� − �1 − ���;
�� = ��� ∗ ��
� − �2 − ���.
In other words, total royalties are equal to the pre-royalty profits of each of the noninteractive
distributors less their respective Shapley Values. In this way, the distributor will be left with its
Shapley Value after it has collected its profits from operations and paid out the royalties to the
record companies.
5 See Michael Carter, “Cooperative Games,” in Hal R. Varian (Ed.), Economic and Financial Modeling with Mathematica, Springer-Verlag (1993), pp. 185-186.
Public Version
D-1
Appendix D: Data Inputs
1. This appendix provides further detail on the data inputs used in my analysis of
Shapley Values. The construction of these data inputs is discussed in the body of the report. I
elaborate here on one specific data input—noninteractive distributor fixed costs and marginal
profit rates. In addition, backup computations for data inputs discussed in the report appear as
exhibits to this appendix.
2. As discussed in the body of the report, I estimate noninteractive distributor fixed
costs and marginal profit rates using Pandora financials as a proxy. Because my analysis models
the hypothetical negotiation over royalty rates for the 2021-2025 period, this is the relevant time
period for analyzing Pandora financials. Moreover, estimating marginal profit rates from
Pandora’s recent historical financials would result in figures with an inappropriate downward bias
with respect to the hypothetical negotiation. Specifically, Pandora’s recent financial results
demonstrate that the company has incurred substantial operating losses. [
]1 However, the approximately $3.5 billion purchase price paid by SiriusXM
to acquire the company, as well as Pandora’s substantial market capitalization of approximately
$2.4 billion immediately prior to the announcement of the SiriusXM acquisition demonstrate that
Pandora is anticipated to be profitable in the future.2 Thus, Pandora’s recent history of operating
1 See SoundExchange Exhibit 57. Operating losses for 2016 and 2017 totaled $318.8 million and $492.9 million, respectively. SoundExchange Exhibit 46 (Pandora 2017 Form 10-K) at 66. Pandora did not file a Form 10-K for the fiscal year ended December 31, 2018 due to its merger with SiriusXM. 2 SoundExchange Exhibit 38; SoundExchange Exhibit 47 (Pandora Form 8-K, September 23, 2018) at Ex. 99.1.
Public Version
D-2
losses does not accurately reflect expectations about the incremental value Pandora is anticipated
to bring to the hypothetical negotiation concerning royalty rates for the 2021-2025 period.
3. For these reasons, I use the forward-looking estimates of Pandora’s costs and profits
for the relevant 2021-2025 period disclosed in the company’s recent Merger Proxy. These
projections were disclosed to Pandora shareholders prior to their vote to approve SiriusXM’s
acquisition of the company and were utilized by Pandora’s investment bankers as an input into
their fairness opinions.3 Pandora disclosed two sets of projections in its Proxy Statement, a lower
“Scenario 1a” projection and a higher “Scenario 2” projection.4 I utilize the Scenario 2 projections
in my analysis because Pandora’s investment bankers prepared discounted cash flow valuation
analyses using these Scenario 2 projections, which produced valuations in-line with the $3.5 billion
market price paid by SiriusXM to acquire the company. By contrast, the Scenario 1a projections
produced valuations substantially below this $3.5 billion market price.5
4. As shown in Exhibit D.6 at the end of this appendix, using these projections I
estimate annual fixed costs for Pandora Free and Pandora Plus of $397 million and $85 million,
3 SoundExchange Exhibit 48 (Pandora Definitive Proxy Statement, December 20, 2018) at 55-60. 4 I understand Pandora also recently produced additional projections that it prepared for these proceedings
See SoundExchange Exhibit 57. I do not use these projections in my analysis as I view the projections Pandora prepared for purposes of evaluating its recent $3.5 billion merger with SiriusXM that were fully disclosed to shareholders in the company’s SEC filings in late-2018 as more reliable than those prepared for purposes of these proceedings. 5 Specifically, when using the Scenario 1 forecasts, Centerview produced a DCF valuation range of $4.50 to $7.00 per Pandora share and LionTree produced a DCF valuation range of $5.53 to $7.81 per share. When using the Scenario 2 forecasts, Centerview and LionTree produced DCF valuation ranges of $8.50 to $11.75 and $9.41 to $12.22, respectively. The per share consideration implied by the $3.5 billion acquisition price was $10.05 per share, which is in-line with the DCF values prepared using the Scenario 2 forecasts. See SoundExchange Exhibit 48 (Pandora Definitive Proxy Statement, December 20, 2018) at 65-66, 73-74.
Public Version
D-3
respectively. Expressing these amounts on an average monthly basis, and then grossing them up
so that they are reflective of the total for all noninteractive distributors, results in monthly fixed
costs of $40.4 million and $8.9 million for all ad-supported noninteractive and subscription
noninteractive distributors, respectively.6 Exhibit D.6 also shows how I estimate a marginal ad-
supported noninteractive profit rate of $0.0042 per play and a subscription noninteractive marginal
profit rate of $0.0048 per play.7
6 I gross up these fixed cost amounts using the same approach used to gross up the audience size figures, which is discussed in the body of the report. 7 The Pandora projections on which these estimates are based do not disclose certain key inputs that were used to create the projections. For instance, the projections do not include a breakdown of subscription revenue into the portions related to its Pandora Plus noninteractive and Pandora Premium on-demand services, respectively, and therefore require an allocation assumption to exclude Pandora Premium revenue and costs from the analysis. Moreover, the projections do not include the projected subscriber counts, active user counts, and play counts underlying the projections, requiring these figures to be derived so that profit rates can be computed. Accordingly, the assumptions required to estimate key parameters for use in my Shapley Value model may need to be updated following the completion of discovery.
Public Version
Exhibit D.1 (RESTRICTED)Subscription On-Demand Music and Video Services: Weighted Average Royalty per Subscriber
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.2 (RESTRICTED)Ad-Supported On-Demand Music and Video Services: Weighted Average Royalty per Play
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.3 (RESTRICTED)CD/Vinyl/Digital Download Royalties per Purchaser
2018
Public Version
Exhibit D.4 (RESTRICTED)Noninteractive Audience and Fixed Cost Adjustment Factors
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.5 (RESTRICTED)Record Company Play Shares
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.6 (RESTRICTED)Pandora Media, Inc. Scenario 2 Management Projections from Merger Proxy
With Allocations to Free, Plus, and Premium Services(Millions of Dollars, Except Per Play and Per Subscriber Amounts)
Page 1 of 3
Public Version
Exhibit D.6 (RESTRICTED)Pandora Media, Inc. Scenario 2 Management Projections from Merger Proxy
With Allocations to Free, Plus, and Premium Services(Millions of Dollars, Except Per Play and Per Subscriber Amounts)
Page 2 of 3
Public Version
Exhibit D.6 (RESTRICTED)Pandora Media, Inc. Scenario 2 Management Projections from Merger Proxy
With Allocations to Free, Plus, and Premium Services(Millions of Dollars, Except Per Play and Per Subscriber Amounts)
Page 3 of 3
Public Version
E-1
Appendix E: Use of Zauberman Survey Results to Generate Model Parameters
1. This appendix describes how I used the results of the survey conducted by Professor
Zauberman to generate parameters employed in my analysis of Shapley Values, as well as the
computation of confidence intervals. I begin with a high-level description of the survey structure,1
then describe my processing of the survey responses to generate inputs for my analysis of Shapley
Values, and finally describe how I use these survey responses to compute inputs for my calculation
of confidence intervals around my royalty rate results.
I. Overview of Zauberman Survey Structure
2. The Zauberman Survey begins by asking respondents to identify the ways that they
currently listen to music. Specifically, it asks them whether or not they have listened to music
through each of the following modes of music distribution in the last 30 days:
• Subscription interactive (“on-demand”) services, such as Spotify Premium or
Apple Music (“Paid-OD”);
• Ad-supported on-demand services, such as free Spotify or free YouTube (“Free-
OD”);
• Subscription noninteractive services, such as Pandora Plus (“Paid-
Noninteractive”);
• Ad-supported noninteractive services, such as Pandora Free (“Free-
Noninteractive”);
1 This is intended to be an overview that provides the context for my use of the survey, not a complete and precise description of all elements of the survey. For such details, see Zauberman WDT.
Public Version
E-2
• SiriusXM satellite radio over-the-air (“SXM-Air”)2;
• AM/FM (terrestrial) radio over-the-air (“AM/FM-Air”)3;
• Physical and digital music, such as CDs, vinyl records, and mp3 files, that they own
(“CD/V/MP3”).
3. The respondents who indicate that they “currently listen” (i.e., have listened in the
last 30 days) to a noninteractive service are then assigned to one of two hypotheticals:4
• NoFree: this hypothetical postulates that all Free-Noninteractive services are no
longer available;
• NoPaid: this hypothetical postulates that all Paid-Noninteractive services are no
longer available.
4. Respondents who are assigned to the NoFree hypothetical are then asked what they
would do instead of listening to their Free-Noninteractive service (which is no longer available).
This questioning is done in two steps. First, the respondents are asked to identify the categories
of music listening they would use instead of their Free-Noninteractive service—the choices of
categories are the ones listed above (Paid-OD, Free-OD, etc.), except that Free-Noninteractive is
removed and “Do something else” (i.e., do not listen to music and, hence, listen to less music
2 This category only includes listening to SiriusXM through a satellite radio receiver (e.g., in a car); streaming of SiriusXM radio through the internet is included in the Paid-Noninteractive category. 3 This category only includes listening to AM/FM radio through a traditional receiver (e.g., in a car); streaming of AM/FM stations through the internet is included in the Free-Noninteractive category. 4 Because some people currently listen to both Paid-Noninteractive and Free-Noninteractive services, a specific approach was developed for assigning respondents to the hypotheticals to ensure that both samples are representative (i.e., include the appropriate mix of people with just one versus both types of noninteractive services). See further discussion in Zauberman WDT.
Public Version
E-3
overall) is added. Second, the respondents are asked to specify the percentage of their
Free-Noninteractive listening time that would be allocated to each of the alternatives chosen in the
first step.5,6 The survey requires that these percentages sum up to 100.
5. The respondents assigned to the NoPaid hypothetical are given similar questions,
except that Free-Noninteractive is one of the available substitutes and Paid-Noninteractive is not.7
II. Processing of Zauberman Survey Results – Overview
6. In this section, I describe how I processed the Zauberman Survey results to generate
inputs for my modeling. Professor Zauberman provided an Excel spreadsheet that contained the
“raw” survey responses, along with the corresponding “datamap” that defines how the responses
are coded.8 There are a total of 989 qualified responses for the NoFree hypothetical and 563
qualified responses for the NoPaid hypothetical. From these qualified responses, I computed the
5 The framing of the two steps is somewhat different. The first step is framed as a general question about services that would be substituted for Free-Noninteractive listening, without reference to a specific time period. The second step is framed as a specific question about services that would be substituted for Free-Noninteractive listening on the same day next week.6 This percentage question is not asked in three circumstances: (1) if the respondent specifies only one option in response to the switching question, in which case, 100 percent of the time is assigned to the option chosen; (2) if the respondent indicates that he/she would not switch to Free-OD services; or (3) if the respondent indicates that he/she would not listen to the Free-Noninteractive service on the same day next week (or is unsure about such listening). In the latter case, zero switching time is assigned to all categories, which is conservative with respect to calculating opportunity costs.7 In the NoPaid hypothetical, the percentage allocation question is not asked in three circumstances: (1) if the respondent specifies only one option in response to the switching question, in which case, 100 percent of the time is assigned to the option chosen; (2) if the respondent indicates that he/she would not switch to either Free-OD or Free-Noninteractive services; or (3) if the respondent indicates that he/she would not listen to the Paid-Noninteractive service on the same day next week (or is unsure about such listening), in which case zero switching time is again assigned to all categories.8 See SoundExchange Exhibit 29.
Public Version
E-4
utilized data parameters.
7. I computed the following from the survey responses: (a) the opportunity cost per
respondent (denominated in dollars per respondent per month) associated with diversion from the
noninteractive services to other forms of music distribution (i.e., O1 and O2); and (b) diversion
from one type of noninteractive service to the other (Free-to-Paid in the NoFree hypothetical and
Paid-to-Free in the NoPaid hypothetical).
8. The pertinent details of the diversion depend on the structure of royalty payments
for the various alternative modes of music distribution, as described in more detail below. For
example, if royalty payments in the alternative mode of distribution are based on subscriptions,
then I derive results from the survey for incremental subscriptions added. On the other hand, if
the royalty payments in the alternative mode of distribution are based on listening time (plays)9,
then I derive results from the survey for incremental plays. These diversion results from the survey
are then combined with the royalty rates paid by outside distributors (discussed in the body of the
report) to determine the opportunity cost associated with forms of music distribution other than
noninteractive streaming.
9. The overall results for both the NoFree and NoPaid hypotheticals are shown in
Figure 6 in the body of the report. The remainder of this appendix describes the calculations used
to generate these results. I begin with the NoFree hypothetical and follow with the NoPaid
hypothetical. The calculations for the latter are largely the same as those for the former.
III. Processing of Zauberman Survey Results – NoFree Hypothetical
10. For the NoFree hypothetical, there are four components of the opportunity cost
associated with diversion from Free-Noninteractive to other modes of music distribution
9 In the absence of specific data, I assume that plays are proportional to listening time.
Public Version
E-5
(excluding Paid-Noninteractive).
11. The first is Paid-OD. Diversion to Paid-OD listening generates incremental
royalties for the record companies if the respondent would acquire a new Paid-OD subscription.
Additional use of an existing Paid-OD subscription does not generate additional royalties, [
].10 Acquisitions of new
Paid-OD subscriptions are identified in the survey results as respondents who both: (i) do not
currently listen to a Paid-OD service, and (ii) would listen to a Paid-OD service instead of the
Free-Noninteractive service in the hypothetical. In this case, the amount of time that they would
listen (specified in the percentage allocation question) is not relevant. The survey results show
that 9.1 percent of the respondents would add a new Paid-OD subscription. As shown in Figure 4
of the body of the report, the average royalty per Paid-OD subscription is [ ] per month, so
the opportunity cost for this component is then 9.1 percent multiplied by [ ], or [ ] per
Free-Noninteractive user per month.
12. The second component of opportunity cost is SXM-Air. As with Paid-OD,
royalties are typically generated when new SXM-Air subscriptions are added, but not when
listening to an existing subscription is increased.11 As such, SXM-Air subscription additions are
10 Specifically, [
]
11 In SDARS III, the Judges set sound recording royalty rates for SiriusXM satellite radio broadcasts at 15.5 percent of gross revenue, including both subscription and advertising revenue. Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms For Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Docket No. 16-CRB-0001 SR/PSSR, p. 2. However, according to SiriusXM’s most recent annual report, its “Subscriber revenue” was
Public Version
E-6
also identified as respondents who both: (i) do not currently listen to SXM-Air, and (ii) would
listen to SXM-Air instead of Paid-Noninteractive in the hypothetical. The survey results show
that 5.5 percent of the respondents would add a new SXM-Air subscription. The opportunity cost
for this component is this percentage with additional subscriptions multiplied by the average
royalty rate per SXM-Air subscription of [ ] per month reflected in Figure 4 of the body of
the report, which is [ ] per Free-Noninteractive user per month.
13. The third component of opportunity cost is CD/V/MP3. For CD/V/MP3, royalties
are generated when additional recordings are purchased; the amount of listening time does not
affect royalties in this category. For purposes of my analysis, I estimate incremental royalties from
diversion to CD/V/MP3 in the same way as for Paid-OD and SXM-Air. That is, a respondent is
considered to “add” CD/V/MP3 spending if he/she (i) does not currently listen to CD/V/MP3, and
(ii) would listen to CD/V/MP3 instead of his/her Free-Noninteractive service in the hypothetical.
In this case, I assume that the respondent spends the same amount of money on purchases of
CD/V/MP3 as an average CD/V/MP3 purchaser, with the corresponding royalties equal to a
percentage of such expenditures. As shown in Figure 4 of the body of the report, this results in
average monthly royalties of [ ] for each new CD/V/MP3 purchaser.12 The survey results
show that 14.8 percent of the respondents would “add” CD/V/MP3 listening. The opportunity cost
for this component is then this percentage of additions multiplied by the average monthly royalties
$4,593.8 million, while its “Advertising revenue” was only $187.6 million. SoundExchange Exhibit 49 (Sirius XM Holdings 2018 Form 10-K) at F-4. 12 This assumes both that people who do not currently engage in CD/V/MP3 listening also do not currently engage in any CD/V/MP3 spending, and that people who do currently engage in CD/V/MP3 listening would not spend more on such products even if they increased their listening time to CD/V/MP3 in the hypothetical.
Public Version
E-7
of [ ] per CD/V/MP3 purchaser, which is [ ] per Free-Noninteractive user per month.
14. The fourth component of opportunity cost is Free-OD. In contrast to the prior three
categories, the amount of time spent listening to Free-OD services does affect the amount of
royalties generated.13 As such, I utilize the responses to the time allocation (percentage) question
in the survey in determining the opportunity cost for this category. Specifically, determining that
opportunity cost for this category involves five steps, as described below.
15. First, I determine the average plays per month for people who currently listen to
Free-Noninteractive services. As shown in Panel A of Figure E-1 below, this amounts to
approximately [ ] plays per month.
13 For example,
]
Public Version
E-8
Figure E-1: Current Plays per User/Subscriber (April 2018 – March 2019) (RESTRICTED)
16. Second, for each respondent who indicates that he/she would listen to Free-OD
instead of Free-Noninteractive, I multiply the average plays per month (above) by the percentage
of time allocated to Free-OD in the survey response. For example, if someone would allocate 20
percent of his/her time to Free-OD, that person would contribute 20 percent x [ ] = [ ] Free-
OD plays. If someone would not switch to Free-OD or assigns zero percent to Free-OD listening,
Public Version
E-9
then he/she does not contribute any Free-OD plays.14
17. Third, I sum the added Free-OD plays (above) across all respondents and divide by
the number of respondents. This amount is the average incremental Free-OD plays per respondent
per month. The survey results show that this average is [ ] plays per respondent.
18. Fourth, I make an adjustment to the number of incremental Free-OD plays
calculated above for respondents who indicate all of the following:
• He/she does not currently listen to Paid-OD, but would listen to Paid-OD instead
of Free-Noninteractive in the hypothetical (i.e., adds a new Paid-OD subscription);
and
• He/she does currently listen to Free-OD, but would not listen to Free-OD instead
of Free-Noninteractive in the hypothetical.
In the above circumstance, I assume that the respondent would replace his/her Free-OD listening
(during times other than when he/she is listening to Free-Noninteractive) with his/her new
Paid-OD subscription. This results in a reduction in Free-OD plays.
19. To calculate this reduction in Free-OD plays, I use the current average number of
plays per month for Free-OD listening. As shown in the figure below, based on Spotify royalty
statements, I estimate average Free-OD listening of [ ] plays per user per month.
14 As previously discussed, if a respondent indicates that his/her only switching option is Free-OD, then 100 percent of the time is allocated to Free-OD. In addition, if the respondent indicates that he/she would not listen to his/her Free-Noninteractive service on the same day next week (or is unsure), then zero percent time is assigned to Free-OD.
Public Version
E-10
Figure E-2: Current Plays per Spotify User (April 2018 – March 2019) (RESTRICTED)
I multiply these [ ] plays per user per month by the 1.6 percent of respondents that the survey
indicates would “drop” their Free-OD listening in the hypothetical, which equates to a reduction
to Free-OD diversion of [ ] plays per respondent per month.
20. Finally, I calculate the net effect of the above – the increased listening less the
“drops.” The survey results reflect a net increase of [ ] Free-OD plays per respondent per month.
This net increase is then multiplied by the royalty rate for Free-OD of [ ] per play (as shown
in Figure 4 of the body of the report) to obtain the opportunity cost, which is [ ] per Free-
Noninteractive user per month.
21. Combining the above components, the total opportunity cost associated with
diversion from Free-Noninteractive to modes of music distribution other than Paid-Noninteractive
is $0.95 per Free-Noninteractive user per month. This amount is the O1 Shapley Value input
parameter defined in Appendix C.
22. The final Shapley Value input coming from the NoFree hypothetical is diversion
from Free-Noninteractive to Paid-Noninteractive (the ��,�� parameter defined in Appendix C). The
calculations for diversion from Free-Noninteractive to Paid-Noninteractive are analogous to the
Public Version
E-11
calculations for diversion to Free-OD services described earlier.15 That is, the survey results
provide the number of additional Paid-Noninteractive plays per respondent based on the
percentage allocation questions.16 Here, the survey indicates that a diversion of [ ] Paid-
Noninteractive plays would be generated per respondent per month.
IV. Processing of Zauberman Survey Results – NoPaid Hypothetical
23. Turning to the NoPaid hypothetical, the calculations for the O2 parameter (i.e.,
opportunity cost associated with diversion to modes of music distribution other than to Free-
Noninteractive) are the same as described above for the NoFree hypothetical. The only difference
is that the average number of plays for Paid-Noninteractive services is [ ] per month, as opposed
to [ ] per month for Free-Noninteractive, as shown in Figure E-1 above. Combining all of the
components, O2 is $1.24 per Paid-Noninteractive user per month.
24. The calculations for diversion from Paid-Noninteractive to Free-Noninteractive
(i.e., ��,�� ) are analogous to the calculations described above for diversion to Free-OD. That is, I
use the time allocation (percentage) responses from the survey to determine the number of
incremental Free-Noninteractive plays generated for each respondent. The result is that, under the
NoPaid hypothetical, an average of [ ] Paid-Noninteractive plays are diverted to Free-
15 The current statutory license requires royalties from subscription noninteractive distributors to be paid out to record companies on a per-play basis.16 As discussed above, the time allocation question is not asked in three circumstances, which are also relevant here: (1) if the respondent specifies only one option in response to the switching question, in which case, 100 percent of the time is assigned to the option chosen; (2) if the respondent indicates that he/she would not switch to Free-OD services; or (3) if the respondent indicates that he/she would not listen to the Free-Noninteractive service on the same day next week (or is unsure about such listening), in which case zero switching time is assigned to all categories. In the case of (2), for instances when the respondent indicated that he/she would switch to Paid-Noninteractive listening in the hypothetical, but not Free-OD, zero switching time is assigned to Paid-Noninteractive – this again is conservative with respect to calculating opportunity costs.
Public Version
E-12
Noninteractive services per respondent per month.
V. Calculation of “O-Prime” Parameters
25. This section describes how the additional Shapley Model parameters��� and ��� are
computed from the parameters discussed in the prior sections. As described in Appendix C, these
“O-prime” parameters represent the opportunity cost from diversion of listening activity if both
ad-supported noninteractive and subscription noninteractive services were not available. These
parameters are calculated by combining the results from both hypothetical scenarios, as described
in detail below.
26. Relatedly, ��,��� and ��,�
�� measure diversion rates of listening time from the ad-
supported and subscription noninteractive distributors, respectively, to outside distributor Dj in the
situation where the other noninteractive distributors are not operating.17 ��,��� and ��,�
�� are linked
by the following relationship:18
��,��� = ��,�
� + ��,��� ∗ ��,�
This equation implies that diversion to an outside distributor from ad-supported noninteractive
services when subscription noninteractive services are unavailable is given by the diversion when
the subscription noninteractive services are available (��,�� ) plus some additional diversion. The
additional diversion is the diversion from D1 that would have gone to D2 had it still been available,
17 Specifically, ��,�′� measures the fraction of current users of noninteractive distributor i’s service who would divert to outside distributor j’s service if service i were to stop operating, while the other noninteractive distributor were also not operating. This diversion is measured as the change in royalty-bearing units t on service j per then-current user of service i. 18 In the absence of more specific data, d1,2 is taken here to be the average fraction of listening time diverted to Paid-Noninteractive services per user of Free-Noninteractive services if they were not available, which is [ ]. Similarly, in the absence of more specific data, d2,1 is taken here to be the average fraction of listening time diverted to Free-Noninteractive services per user of Paid-Noninteractive services if they were not available, which is [ ].
Public Version
E-13
and then would have in turn diverted from D2 to Dj upon the exit of D2, in the situation where D1
was not available.
27. Similarly,
��,��� = ��,�
� + ��,��� ∗ ��,�.
28. Then, I solve these two equations simultaneously for ��,��� and ��,�
�� to obtain:
��,��� =
��,�� + ��,�
� ∗ ��,�
1 − ��,� ∗ ��,�
��,��� =
��,�� + ��,�
� ∗ ��,�
1 − ��,� ∗ ��,�.
29. With these �′�,�� parameters, I can then calculate ��� and ��� using the royalty rates
for each outside distributor – [ ] for Paid-OD, etc. – as previously discussed. This results in
an ��� of $1.01 and an ��� of $1.42.19
19 In accordance with Professor Zauberman’s testimony, the results discussed throughout this section do not incorporate any demographic weighting of the survey responses. See Zauberman WDT. Nevertheless, Professor Zauberman has provided demographic weights, which I have used to test the sensitivity of my results. See SoundExchange Exhibit 31. The demographic weights are based on geography (region), gender, and age. Incorporating these demographic weights has no material impact on the results produced by my Shapley Value model.
Public Version
E-14
VI. Calculation of Confidence Intervals
30. As discussed in the body of the report, I use a bootstrap procedure to calculate
confidence intervals for the ad-supported noninteractive and the subscription noninteractive
royalty rates computed from my Shapley Value model.20
31. For the bootstrapping analysis, I utilize 1,000 simulated sets of survey respondents
and calculate the Shapley Value analysis parameters for each of those simulated sets of
respondents. Each of the simulated sets of survey respondents is based on N random draws from
the overall actual population of qualified responses, where N is the number of qualified responses
to the main survey (989 for the NoFree hypothetical and 563 for the NoPaid hypothetical). The
sampling for each simulated set of respondents is done with replacement, so any individual actual
responses to the main survey can be included more than once (or not at all) in any individual
simulated set of survey respondents. The specific responses included in each simulated set of
survey respondents were provided by Professor Zauberman.21
32. With each of these 1,000 simulated sets of respondents, I apply the same approach
as previously described to calculate the corresponding Shapley Value analysis parameters. From
each of these 1,000 simulated groups of input parameters, I calculate 1,000 royalty rates for the
ad-supported noninteractive distributor and 1,000 royalty rates for the subscription noninteractive
distributor using my Shapley Value model described in Appendix C. From both sets of 1,000
royalty rates I calculate two statistics. First, I compute the 2.5 percent quantile (i.e., the number
20 For a general discussion of the use of bootstrapping, see SoundExchange Exhibit 67 (“AAPOR Guidance on Reporting Precision for Nonprobability Samples,” American Association for Public Opinion Research). 21 See SoundExchange Exhibit 30. My understanding is that these are the same samples that Professor Zauberman used in his bootstrapping procedures to compute confidence intervals for the individual survey responses (see Zauberman WDT).
Public Version
E-15
below which 2.5 percent of the 1,000 royalty rates fall). Second, I compute the 97.5 percent
quantile (i.e., the number below which 97.5 percent of the 1,000 royalty rates fall). These two
statistics define the upper and lower bounds of the 95 percent confidence interval for the royalty
rates determined by my Shapley Value model.
Public Version
F-1
Appendix F: Shapley Values with “Share of Ear” Data
1. This appendix provides a detailed description of my Shapley Value model
calibrated with data from Edison Research’s Q2 2019 Share of Ear research report. Like the model
detailed in Appendix C, this model involves two noninteractive streaming distributors (a free ad-
supported distributor and a paid subscription distributor), three “must-have” record companies
(representing the majors), and one “not-must-have” record company (representing the indies).
I. Model
2. In constructing this Shapley Value model, I utilize the following notation:
Figure F-1: Notation
D1 and D2The ad-supported and subscription noninteractive streaming distributors, respectively.
LA, LB, and LC The three must-have major record companies.
LD
An amalgam of not-must-have indie record companies grouped together without loss of generality. Referred to hereafter as the not-must-have record company.
SharekRecord company k’s share of plays, which is assumed to be the same across all distributors.
CkRecord company k’s costs of producing sound recordings.
��� and ��
�D1’s and D2’s marginal profit per play, respectively, before accounting for any sound recording royalties that they pay.
F1 and F2 D1’s and D2’s fixed costs, respectively.
Dj
where � ∈ {3, …, N}
Distributors whose royalty rates are set outside of the current proceedings, such as ad-supported and subscription on-demand music and video streaming services, SiriusXM satellite radio, and music purchases (i.e., CDs, vinyl, digital downloads).
��� The per play royalty rate paid by outside distributor
j.
��� Di’s audience, measured in plays, when all other
distributors are operating.
��,�
A diversion ratio from service i to j: The percentage of noninteractive distributor Di’s plays that would divert to distributor j’s service if Di
were to stop operating while all other distributors continued operating.
� = � ����∗ ��
��
�
���
The four record companies’ combined royalties from outside distributors when all distributors are operating.
Public Version
F-2
���
= � ����∗ ��,��
�
���
Per play of noninteractive streaming distributor Di’s service, the change in the four record companies’ combined royalties from outside distributors if Di were to stop operating while all other distributors continued operating.
�′�,�
Another diversion ratio from service i to j: The percentage of noninteractive distributor Di’s plays that would divert to outside distributor Dj’s service if service i were to stop operating, while the other noninteractive distributor were also not operating.
����
= �����∗ �′�,��
�
���
Per play of noninteractive streaming distributor Di’s service, the change in the four record companies’ combined royalties from outside distributors if noninteractive distributor Di were to stop operating, while the other non-interactive distributor were also not operating.
3. The Shapley Value solution concept requires calculating the average marginal
contribution to the collective surplus that each record company or distributor creates, based on all
possible orderings of the parties. The first step in this process is to calculate the total surplus each
hypothetical coalition of noninteractive streaming distributors and record companies would
generate if they acted in a manner that maximized their collective surplus. These values define a
function �[�] of all those coalitions � ∈ �(�), called the “characteristic function,” where �(�) is
the set of subsets of the players � = {��,��, ��, ��, �� , ��}.
4. In the present case, the characteristic function can be determined recursively.
An empty coalition, and coalitions comprised of only the distributors, create zero payoffs. Hence:
�[{}] = 0;
�[{��}] = 0;
�[{��}] = 0;
�[{��,��}] = 0.
Public Version
F-3
5. Each label, when it is alone, generates its current profits from outside distributors
plus the additional profits from outside distributors caused by diversion of the plays of
noninteractive distributors 1 and 2 to the outside distributors. Hence:1
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��.
6. Since LA, LB, and LC are must-have, only when all three are present in a coalition
can the distributor(s) operate and make profits. When only these three record companies are
present in the coalition, three cases can arise. First, if only D1 is in a coalition, and if it increases
the coalition’s collective surplus, diversion occurs from D2 to both D1 and the outside distributors
(because D2 is not operating) and from D1 to outside distributors (because D1 does not have a
license to LD’s content). Second, if only D2 is in a coalition, and if it increases the coalition’s
collective surplus, diversion occurs from D1 to both D2 and the outside distributors (because D1 is
not operating) and from D2 to outside distributors (because D2 does not have a license to LD’s
content). And third, if both distributors are in a coalition, and if they increase the coalition’s
collective surplus, no diversion between D1 and D2 occurs, but there is still diversion from D1 and
D2 to the outside distributors (because they do not have licenses to LD’s content).2
1 I assume that it is profitable for the must-have record companies to operate.2 In keeping with the Judges’ conclusion in the Web IV proceeding that “the repertoire of each of the three Majors is a ‘must-have’ in order for a noninteractive service to be viable,” I assume that a distributor could not operate without agreements with all three of the major record companies. See Web IV Determination at 133. Conversely, I assume that if a distributor has access to the must-have record companies’ content but does not have access to LD’s content, the distributor loses �ℎ���� of its plays.
Public Version
F-4
7. Each coalition is assumed to act in the manner that maximizes its collective
surplus.3 Therefore:
�[{��,��,��,��}]
= ����(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗ ��,�� + � + ��
�∗ �2
��− (�� + �� + ��)
− ��,�[{��,�� ,��}]�; 4
�[{��,��,�� ,��}]
= ����(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗ ��,��+ � + ��
�∗ �1
�� − (�� + �� + ��)
− ��,�[{��,��,��}]�;
�[{��,��,��,��,��}]
= ����(1 − �ℎ����) ∗ ����∗ ��
�+ ��
�∗ ��
�+ ��− (�� + �� + ��)− ��
− ��,�[{��,��,��,��}],�[{��,��,�� ,��}]�.
8. When LD joins a coalition, it adds to the coalition’s surplus by contributing its
royalties from outside distributors, and also by licensing one or both distributors in the coalition
(if such a license would increase the coalition’s collective surplus). Thus:
�[{��,��,��,�� ,��}]
= ��� ��� + ���∗ �2
�+ ��
�∗ ���
�+ ��
�∗ ��,�
���− (�� + �� + �� + ��)
− ��,�[{��,��,�� ,��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
�[{��,��,�� ,�� ,��}]
= ��� ��� + ���∗ �1
�+ ��
�∗ ���
�+ ��
�∗ ��,�
���− (�� + �� + �� + ��)
− ��,�[{��,��,��, ��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
�[{��,��,��,��,�� ,��}]
= ������ + ���∗ ��
�+ ��
�∗ ��
��− (�� + �� + �� + ��) − ��
− ��,�[{��,��,��,�� ,��}],�[{��,��,�� ,�� ,��}],�[{��,��,��,��,��}]
+ �[{��}],� + ���∗ ���
�+ �ℎ���� ∗ ��
�∗ ��,��− �� − (�� + �� + �� + ��)
+ (1 − �ℎ����) ∗���∗ ��
�+ �ℎ���� ∗ ���
�∗ �2
��− ��,� + ��
�∗ (��
�+ �ℎ����
∗ ���∗ ��,�) − �� − (�� + �� + �� + ��) + (1 − �ℎ����) ∗��
�∗ ��
�
+ �ℎ��������∗ �1
��− ���.
3 In some cases, the collective surplus of a coalition may be higher if one or both distributors do not operate.4 As explained below, �[{��, ��, ��}] is the combined profits of the coalition that includes only the three must-have record companies. �[{��, ��, ��}] = �[{��}] + �[{��}] + �[{��}].
Public Version
F-5
9. In any other case not defined above, the noninteractive distributors do not carry
one or more of the must-have record companies and cannot operate. In such cases, all the profits
generated by those coalitions are just the sum of the �[{��}] defined above. For example,
�[{��,��}] = �[{��}] + �[{��}],�[{��,��,��,��}] = �[{��}] + �[{��}], etc.
10. Based on this characteristic function, the Shapley Value algorithm can be
implemented. Specifically, I evaluate every possible arrival ordering and for each determine the
incremental value contributed by each noninteractive distributor relative to the value created by
the parties that preceded its arrival. Then I average all of these incremental contributions across
all arrival orderings. With six players, there are 6! (i.e., 720) arrival orderings. Algebraically, this
can be expressed as follows. For every possible subset � of � that includes i,
��� = ���(�[�] − �[� − {�}])
�⊂�
,
where
�� =(|�| − 1)! ∗ (� − |�|)!
�!,
for n equal to the number of elements in � (six in this case) and |�| the number of elements in
subset �.5
11. Following this formula, I calculate the Shapley Values for both noninteractive
distributors, based on the parameters estimated as described in Appendices D and E. The total
royalty payments to be made by D1 and D2 can be calculated using these Shapley Values as
follows:
5 See Michael Carter, “Cooperative Games,” in Hal R. Varian (Ed.), Economic and Financial Modeling with Mathematica, Springer-Verlag (1993), pp. 185-186.
Public Version
F-6
�� = ��� ∗ ��
� − �� − ���;
�� = ��� ∗ ��
� − �� − ���.
In other words, total royalties are equal to the pre-royalty profits of each of the noninteractive
distributors less their respective Shapley Values. In this way, the distributor will be left with its
Shapley Value after it has collected its profits from operations and paid out the royalties to the
record companies.
II. Calibration
12. As described in the body of the report, as an alternative approach, I estimate
diversion ratios from data in the Edison Research Q2 2019 Share of Ear research report.6 Using
these data, I obtained the following shares of time spent listening to different modes of music
distribution:
Figure F-2: Share of Time Spent Listening to Music, Smartphone Owners 13+ (RESTRICTED)
6 See SoundExchange Exhibit 44 (Edison Research, “Share of Ear,” Q2 2019). For further details on Edison Research’s Share of Ear survey methodology, see SoundExchange Exhibit 72 (Q2 2019 Share of Ear Methodology Statement).
Public Version
F-7
13. The Share of Ear data also provide the following decomposition of time spent
listening to streaming music services.
Figure F-3: Share of Streaming Music, Smartphone Owners 13+ (RESTRICTED)
14. The Share of Ear data, however, do not provide a way to split the streaming
services’ audiences into different tiers of service. For instance, the data do not provide shares for
Pandora’s noninteractive Pandora Radio and Pandora Radio Plus tiers or its on-demand Pandora
Premium tier. Hence, to further split each streaming service’s audience into individual tiers of
service, I use play count information from the 12-month period ending March 2019 contained in
royalty statements reported to the record companies. This breakdown is summarized in the figure
below.
Figure F-4: Share by Streaming Service (RESTRICTED)
Public Version
F-8
15. By combining the data in the three figures above, I calculate the total share of
listening time for each tier of services. For example, the share of total listening on the free Pandora
Radio tier is equal to [
] Next,
I group the ad-supported noninteractive distributors (Pandora Radio, iHeart Non-Simulcast
Webcast, iHeart Internet Simulcast, and AM/FM Streaming) together and calculate their share.
These entities combined represent the ad-supported noninteractive distributor D1 in the Shapley
model. I also calculate the share of the subscription noninteractive distributors (Pandora Radio
Plus, iHeart Radio Plus, and Napster UnRadio). These entities combined represent the
subscription noninteractive distributor D2 in the Shapley model.
16. Then, I follow a logit demand model to estimate diversions based on shares,
under the assumption that if either the ad-supported or subscription noninteractive distributors
were to lose access to sound recordings, users would divert their noninteractive plays to other
distributors in a way that is proportional to these distributors’ current shares of listening time.7,8
The resulting diversions are shown in the figure below.
7 See, e.g., Robert D. Willig, “Merger Analysis, Industrial Organization Theory, and Merger Guidelines,” Brookings Papers: Microeconomics (1991): 281- 332, p. 302. See also, Carl Shapiro, “Mergers with Differentiated Products,” Antitrust, No. 10 (1996): 23-30, p. 25. As an example of how this works, assume the shares of product A, B, and C are 10%, 20%, and 70%, respectively. If product A exits the market, diversion to product B is 20% / (100% - 10%) = 22.22% and diversion to product C is 70% / (100% - 10%) = 77.78%. 8 In the absence of specific data, I assume that plays are proportional to time.
Public Version
F-9
Figure F-5: Share-Based Diversion (RESTRICTED)
17. Figure F-5 shows that, for instance, AM/FM Broadcast, which has a share of ear
of [ ], would receive diversion of the plays on ad-supported noninteractive distributors (d1) of
[ ] if those distributors were to exit the market, and diversion of the plays on subscription
noninteractive distributors (d2) of [ ] if subscription noninteractive distributors were to exit
the market. Correspondingly, distributors with lower shares receive less diversion.
18. Relatedly, d1' and d2' measure diversion rates from the ad-supported and
subscription noninteractive distributors, respectively, in the situation where the other
noninteractive distributors are not operating. As explained in Appendix E, d1' and d2' are linked
by the following relationship (for a given outside distributor D0):
�′�,� = ��,� + �′�,� ∗ ��,�
Public Version
F-10
This equation implies that diversion to an outside distributor from ad-supported noninteractive
services when subscription noninteractive services are unavailable is given by the diversion when
the subscription noninteractive services are available (��,�) plus some additional diversion. The
additional diversion is the diversion from D1 that would have gone to D2 had it still been available,
and then would have in turn diverted from D2 to D0 upon the exit of D2, in the situation where D1
was not available.
19. Similarly,
�′�,� = ��,� + �′�,� ∗ ��,�.
20. Then I solve these two equations simultaneously for �′�,� and �′�,� to obtain
�′�,� =��,� + ��,� ∗ ��,�
1 − ��,� ∗ ��,�
�′�,� =��,� + ��,� ∗ ��,�
1 − ��,� ∗ ��,�.
For example, in the case where D0 is AM/FM Broadcast, �′�,� = [ ]
21. Next, with these diversion rates and the per play royalty rates shown in the figure
below, I compute the opportunity cost associated with each form of outside distribution. For
example, the amounts shown in Figure F-6 in the column labeled “d1 * Royalty” show the
opportunity cost of the record companies associated with the diversion that would occur to each
source of music distribution if noninteractive distributor D1 were not in the market. The “Total”
line shows the sums of the opportunity costs associated with all such outside music distributors,
and these are the computations of the model input parameters ���
, ���
, ���� and ��
�� found in Figure
F-7.
Public Version
F-11
Figure F-6: Royalties and Calculation of Opportunity Cost (RESTRICTED)
Figure F-7: Opportunity Cost and Diversion Rates with Share of Ear
22. The additional diversion parameters d1,2 and d2,1 are also shown in Figure F-7.
All other parameters used to calibrate the Shapley model are described in the body of the report
and Appendix D. The resulting Shapley Values and royalties are presented in Figure 10 in the
body of the report.9
9 I tested the robustness of these results by performing a number of sensitivities. First, I used the same Share of Ear data, but from an earlier Q3 2018 Share of Ear report. See SoundExchange Exhibit 43 (Edison Research, “Share of Ear,” Q3 2018). Second, I used the total population data in the Q2 2019 Share of Ear study rather than limiting the data to only smartphone users. Third, rather than utilize the share of streaming music breakdowns shown in Figure F-3, I used play count information from royalty statements to allocate all Streaming and YouTube listening amounts from Figure F-2 among the various services and service tiers. Last, instead of share of listening time from Edison Research, I employed similar share of listening time data from MusicWatch. See
d12 O1p O1
p' d21 O2p O2
p'
1.1% $0.00234 $0.00237 9.6% $0.00214 $0.00237
Public Version
F-12
SoundExchange Exhibit 42 (MusicWatch, “Annual Music Study 2018: Report to RIAA,” April 2019) at 27, 39. None of these sensitivity checks substantially affects my quantitative results.
Public Version
G-1
Appendix G: Alternative Nash-in-Nash Bargaining Model
I. Introduction
1. As explained in the body of the report, Nash-in-Nash bargaining is a method
derived from economic theory to determine the outcome of a multi-party negotiation.
2. To provide a simple example, consider a bargaining situation involving one
copyright owner and two noninteractive distributors, D1 and D2. In step one, the Nash Bargaining
Solution resulting from the negotiation between the copyright owner and noninteractive distributor
D1 is identified, which depends on the Nash Bargaining Solution to the negotiation between the
copyright owner and noninteractive distributor D2. Likewise, the Nash Bargaining Solution
resulting from the negotiation between the copyright owner and distributor D2 is identified, which
depends on the outcome of the negotiation between the copyright owner and distributor D1.
3. At the end of step one, the two bargaining solutions are characterized, with each
depending on the solution to the other. Mathematically, this analysis is represented as two
equations relating the two unknown values of the two bargaining solutions. In step two, this system
of two equations and two unknowns is simultaneously solved to determine the terms of the
agreements between the copyright owner and each of distributors D1 and D2.
4. In the above example, if the copyright owner reaches agreements with
noninteractive distributors D1 and D2, it may earn less profits by licensing its sound recordings to
other forms of music distribution. The extent to which that is the case will depend on the extent
to which D1 and D2 can successfully lure away the listening audiences of the other distributors.
This is the copyright owner’s opportunity cost of licensing to D1 and D2. In order for the copyright
owner to have any incentive to enter into deals with D1 and D2, it must receive royalties that at
least fully offset the profits it will be giving up from other distributors by virtue of entering those
Public Version
G-2
deals. No rational copyright owner would enter into deals with noninteractive distributors D1 and
D2 unless it recovered this amount, plus at least some share of the surplus jointly created by those
deals.
5. An adaptation and continuation of the numerical example from the body of the
report further illustrates the concept of the Nash-in-Nash bargaining solution. Here, two patents,
patentA and patentB, are both necessary for the manufacture and sales of deviceD. The fallback
values for the owners of patentA and patentB are each $5 if deviceD is not on the market. If
deviceD is brought to market, then the individual net returns that the owners of patentA and
patentB will realize from licensing to other rival devices will decrease to $3, due to cannibalization
from deviceD.
6. To analyze the bargaining between the owners of patentA and deviceD, suppose
that the owners of patentB and deviceD have already committed to a deal that provides a royalty
payment of RB from the owner of deviceD to the owner of patentB if the patent is utilized. Given
this deal, the fallback value for the owner of patentA is $5 and the fallback value for the owner of
deviceD is zero. The total value to the owners of patentA and deviceD from an agreement between
them, which would enable the manufacture and sale of deviceD due to the deal already struck with
the owner of patentB, is the $3 from the net return to the owner of patentA from other licensing
activity plus the net return to the owner of deviceD of $16 minus RB. The surplus created by this
deal is the total value created from the agreement minus the sum of the parties’ fallback values,
which is $3 + $16 – RB – $5, or $14 – RB.
7. The corresponding Nash Bargaining Solution splits that surplus, leaving the owner
of deviceD with .5 x ($14 – RB). Since the owner of deviceD earns $16 of market profit, pays out
RB to the owner of patentB, and pays out RA to the owner of patentA, the resulting relationship
Public Version
G-3
is $16 – RB – RA = .5 x ($14 – RB).
8. To find the Nash-in-Nash bargaining equilibrium, it is necessary to also analyze the
bargaining between the owners of patentB and deviceD, while assuming that the owners of patentA
and deviceD have already committed to a deal that provides a royalty payment of RA from the
owner of deviceD to the owner of patentA if the patent is utilized. Given the symmetry of
assumptions for patentA and patentB, the resulting Nash Bargaining solution to this negotiation is
similar to the one identified above, namely: $16 – RA – RB = .5 x ($14 – RA). Solving these two
equations simultaneously with simple algebra yields that the royalties indicated by the Nash-in-
Nash solution are RA = RB = $6.
II. Six-Player Nash-in-Nash Model
9. This section provides a detailed description of the Nash-in-Nash bargaining model
that involves two noninteractive distributors (an ad-supported distributor and a subscription
distributor), three “must-have” record companies, and one “not-must-have” record company.
10. In the Nash-in-Nash model, I analyze the royalty rates that would prevail if: (i) each
record company and each noninteractive distributor negotiated bilaterally a lump sum royalty that
the distributor would pay the record company for its distribution of that record company’s content;
and (ii) each of these bilateral negotiations proceeded under the assumption by both parties that
every other distributor-record company pair had agreed or would agree on a lump sum royalty.
From this model’s construction I identify the Nash Equilibrium royalty payments, which are the
set of royalty payments from which no pair of negotiators has an incentive to deviate unilaterally.
In constructing this Nash-in-Nash bargaining model, I utilize the following notation:
Figure G-1: Notation
D1 and D2The ad-supported and subscription noninteractive streaming distributors, respectively.
LA, LB, and LC The three must-have major record companies.
Public Version
G-4
LD
An amalgam of not-must-have indie record companies grouped together without loss of generality. Referred to hereafter as the not-must-have record company.
SharekRecord company k’s share of plays, which is assumed to be the same across all distributors.
CkRecord company k’s costs of producing sound recordings.
��� and ��
�D1’s marginal profit per play and D2’s marginal profit per play, respectively, before accounting for any sound recording royalties that they pay.
F1 and F2 D1’s and D2’s fixed cost, respectively.
Ri,kThe dollar amount of royalties Di agrees to pay Lk, if they reach an agreement.
Dj
where � ∈ {3, …, N}
Distributors whose royalty rates are set outside of the current proceedings, such as ad-supported and subscription interactive music and video streaming services, Sirius XM satellite radio, and music purchases (i.e., CDs, vinyl, digital downloads).
���
The royalty rate paid by outside distributor j per royalty-bearing unit � ∈ {play, subscriber, CD/vinyl-record/download}.
���
Di’s audience, measured in Di’s royalty bearing units t (e.g., plays for D1), when all other distributors are operating.
���The average number of monthly plays per user of D1’s service.
���The average number of monthly plays per user of D2’s service.
��,��
A diversion ratio from service i to j: Per current user of noninteractive streaming distributor i’s service, the change in royalty-bearing units t on distributor j’s service if Di were to stop operating while all other distributors continued operating.
� = � ���� ∗ ��
���
���
The four record companies’ combined royalties from outside distributors when all distributors are operating.
�� = � ���� ∗ ��,�
� ��
���
Per current user of noninteractive streaming distributor i’s service, the change in the four record companies’ combined royalties from outside distributors if Di were to stop operating while all other distributors continued operating.
11. I assume that if an agreement between a record company and a noninteractive
distributor increases their combined profits (assuming all other pairs of record companies and
noninteractive distributors reach an agreement), the two parties will reach the Nash Bargaining
Solution by agreeing on the lump sum royalty that splits their combined incremental profits
Public Version
G-5
equally.
12. The first step in identifying this royalty is to define each record company’s and
each noninteractive distributor’s profits when they reach an agreement and when they do not,
assuming in both cases that all other pairs of record companies and noninteractive distributors
reach an agreement.
13. If all pairs of record companies and noninteractive distributors reach agreements,
the profits of each of these parties can be expressed as follows:1
π�{���} = ��,� + ��,� + (Share� ∗ E) − ��, where � ∈ {�,�,�,�};
π�{���} = ���
∗ ���
− �� − ∑ ��,�� ;
π�{���} = ���
∗ ���
− �� − ∑ ��,�� .
14. If a given record company and noninteractive distributor do not reach an
agreement, while all other pairs of record companies and noninteractive distributors do reach
agreements, the profits of the record company and distributor that do not reach an agreement can
be expressed as follows:2
π�{~(D�, L�)} = ��,� + (�ℎ���� ∗ E) + ���
��1∗ �ℎ���� ∗ ��
�� − ��, where � ∈ {�,�,�,�};
π�{~(D�, L�)} = 0, where � ∈ {�,�,�};
π�{~(D�, L�)} = ����
∗ (1 − �ℎ����) ∗ ���
− ��� − ∑ ��,���� ;
π�{~(D�, L�)} = ��,� + (�ℎ���� ∗ E) + ���
��2∗ �ℎ���� ∗ ��
�� − ��, where � ∈ {�,�,�,�};
1 I assume that each record company’s share of the combined royalties from outside distributors when all distributors are operating (E) is equal to its share of plays. 2 In keeping with the Judges’ conclusion in the Web IV proceeding that “the repertoire of each of the three Majors is a ‘must have’ in order for a noninteractive service to be viable,” I assume that a distributor could not operate without agreements with all three of the major record companies. See Web IV Determination at 133. Conversely, I assume that if a distributor has access to the must-have record companies’ content but loses access to LD’s content, the distributor would lose �ℎ����of its plays (or subscribers).
Public Version
G-6
π�{~(D�, L�)} = 0; where � ∈ {�,�,�};
π�{~(D�, L�)} = ����
∗ (1 − �ℎ����) ∗ ���
− ��� − ∑ ��,���� .
15. Using the above definitions, each record company’s and each non-interactive
distributor’s incremental profits from reaching an agreement with each other, when all other
pairs of record companies and non-interactive distributors reach an agreement, can be expressed
as follows:3
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ��,� − ���
��1∗ �ℎ���� ∗ ��
�� , where � ∈ {�,�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ���
∗ ���
− �� − ∑ ��,�� , where � ∈ {�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ����
∗ �ℎ���� ∗ ����− ��,�;
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ��,� − ���
��2∗ �ℎ���� ∗ ��
�� , where � ∈ {�,�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ���
∗ ���
− �� − ∑ ��,�� , where � ∈ {�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ����
∗ �ℎ���� ∗ ����− ��,�.
16. Since the Nash bargaining solution dictates that the parties split equally the
combined incremental profits from an agreement, the royalty that results from the bilateral
negotiation between Di and Lk is the one that equalizes both parties’ incremental profits, i.e., that
leads to ∆π�{(D� , L�)} = ∆π�{(D�, L�)}. This yields a system of eight equations with eight
unknown royalty payments.
17. Solving this system of equations yields the following total royalty payments for
each distributor:
�� = ���,�
�
=�(2 − �ℎ����) ∗
����1
+ ���
∗ (6 + �ℎ����)� ∗ ���
− 6 ∗ ��
8;
3 ∆π�{(D�, L�)} = π�{���} − π�{~(D� , L�)} and ∆π�{(D� , L�)} = π�{���} − π�{~(D� , L�)}represent Lk’s and Di’s incremental profits, respectively, from reaching an agreement with each other when all other pairs of record companies and non-interactive distributors reach an agreement.
Public Version
G-7
�� = ���,�
�
=�(2 − �ℎ����) ∗
����2
+ ���
∗ (6 + �ℎ����)� ∗ ���
− 6 ∗ ��
8.
Public Version
EXHIBIT B
Table of Contents
I. QUALIFICATIONS ................................................................................................................ 1
II. SUMMARY OF TESTIMONY .............................................................................................. 2
A. Introduction ...................................................................................................................... 2
B. Assignment and Summary of Conclusions ...................................................................... 4
III. SHAPLEY VALUES AND THEIR APPLICATION TO § 114 OBJECTIVES .................... 5
IV. A SHAPLEY VALUE MODEL FOR RECORD COMPANIES AND NONINTERACTIVE WEBCASTERS ..................................................................................................................... 12
A. Defining the Model ........................................................................................................ 12
B. Empirical Inputs ............................................................................................................. 16
i. Royalties for Other Forms of Distribution ............................................................. 17
ii. Noninteractive Distributors’ Audience Sizes ......................................................... 20
iii. Diversion Ratios ..................................................................................................... 21
iv. Record Company Play Shares ................................................................................ 23
v. Noninteractive Distributor Fixed Costs and Marginal Profit Rates ....................... 24
C. Summary of Results ....................................................................................................... 26
D. Alternative Shapley Value Calculations Using “Share of Ear” Data ............................. 32
E. Alternative Calculations Using a Nash-in-Nash Bargaining Model .............................. 34
V. CONCLUSION ..................................................................................................................... 37
Appendix A: Curriculum Vitae and List of Prior Testimony …………….…………………... A-1
Appendix B: Materials Relied On …………………………………………………………….. B-1
Appendix C: Shapley Value Model …………………………………………………………… C-1
Appendix D: Data Inputs ……………………………………………………………………… D-1
Appendix E: Use of Zauberman Survey Results to Generate Model Parameters ……………... E-1
Appendix F: Shapley Values with “Share of Ear” Data ………………………………………. F-1
Appendix G: Alternative Nash-in-Nash Bargaining Model ………………………………….. G-1
Public Version
1
I. QUALIFICATIONS
1. My name is Robert D. Willig. I am Professor of Economics and Public Affairs
Emeritus at Princeton University, where I have held a joint appointment in the Economics
Department and at the Woodrow Wilson School of Public and International Affairs since 1978.
Previously, I was a Supervisor in the Economics Research Department of Bell Laboratories. My
teaching and research have specialized in the fields of industrial organization, government-
business relations, and social-welfare theory. I served as Deputy Assistant Attorney General for
Economics in the Antitrust Division of the U.S. Department of Justice from 1989 to 1991, and in
that capacity served as the Division’s Chief Economist. I am the author of Welfare Analysis of
Policies Affecting Prices and Products, Contestable Markets and the Theory of Industry Structure
(with William Baumol and John Panzar), and some eighty articles and chapters in the professional
literatures of economics and law. I am also a co-editor of The Handbook of Industrial
Organization, which summarizes the state of economic thinking on the structure of industries and
the nature of competition among firms, and have served on the editorial boards of the American
Economic Review, the Journal of Industrial Economics, and the MIT Press Series on Regulation.
I am an elected Fellow of the Econometric Society and was an associate of The Center for
International Studies.
2. I have appeared as an expert witness before Congress, federal and state courts,
federal administrative agencies, and state public utility commissions on subjects involving
intellectual property rights, competition, regulation, and antitrust. In particular, I previously
appeared as an expert witness before the Copyright Royalty Judges in the SDARS III proceeding.
I have also served as a consultant to the Federal Trade Commission, the U.S. Department of Justice,
and many leading corporations on antitrust, regulation, intellectual property, and policy issues.
Public Version
2
3. I have spent a significant portion of my career studying, consulting, and testifying
as an expert on many different aspects of the economics of unregulated and regulated pricing, the
economics of intellectual property, and the distribution of rights to its access. I have substantial
consulting experience working on music industry issues including the SDARS III proceeding, the
merger of BMG and Sony, other potential control transactions involving major record companies
and music publishers, and the regulatory treatment of musical work performing rights
organizations.
4. My full curriculum vitae and a listing of my prior testimony are included in
Appendix A. The rate charged by Compass Lexecon for my work on this matter is $1,450 per
hour. I have a financial interest in the overall profitability of Compass Lexecon, but I have no
financial interest in the outcome of this case.
II. SUMMARY OF TESTIMONY
A. Introduction
5. On January 24, 2019, the Copyright Royalty Judges announced the commencement
of a proceeding (“Web V”) to determine reasonable rates and terms for statutory licenses under 17
U.S.C. § 114, allowing for the digital transmission of sound recordings over the internet, and under
17 U.S.C. § 112, allowing for the making of ephemeral recordings to facilitate such digital
transmissions.1 The rates and terms set in this proceeding will apply to the five year period from
January 1, 2021 to December 31, 2025.2 I understand the relevant statutory licenses are
compulsory for rights owners and are available to commercial and noncommercial noninteractive
webcasters.
1 Federal Register, Vol. 84, No. 16, Jan. 24, 2019, Notices, at 359. 2 Id.
Public Version
3
6. It is my understanding that the standard in this Web V proceeding requires the
Copyright Royalty Judges to set 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.”3
In past proceedings, I understand the relevant marketplace has been defined as a “hypothetical
marketplace, free of the influence of compulsory, statutory licenses.”4 Moreover, this hypothetical
marketplace is one that is “effectively competitive,” meaning that any “holdout” value associated
with a seller’s property that exists because it “is a necessary complement” to the property of other
sellers (i.e., “complementary oligopoly” value) should not be included in the statutory rates
determined by the Judges.5
7. I further understand that the relevant statute requires the Copyright Royalty Judges
to “base their decision on economic, competitive, and programming information presented by the
parties, including” (1) whether the use of the noninteractive service “may substitute for or may
promote the sales of phonorecords or otherwise may interfere with or may enhance the sound
recording copyright owner’s other streams of revenue from its sound recordings,” and (2) the
“relative roles of the copyright owner and the transmitting entity in the copyrighted work and the
service made available to the public with respect to relative creative contribution, technological
contribution, capital investment, cost, and risk.”6 Moreover, I understand the Judges are permitted,
3 17 U.S.C § 114(f)(2)(B). 4 Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms for Ephemeral Recording and Webcasting Digital Performance of Sound Recordings (Web IV), Docket No. 14-CRB-0001-WR (“Web IVDetermination”), at 2. 5 Id. at 41-4240. 6 17 U.S.C § 114(f)(2)(B).
Public Version
4
but not required, to “consider the rates and terms for comparable types of digital audio transmission
services and comparable circumstances under voluntary license agreements.”7
8. At the conclusion of the Web IV proceeding, after applying these criteria, the
Copyright Royalty Judges established a rate for commercial noninteractive ad-supported services
in 2016 of $0.0017 per performance and a rate for commercial noninteractive subscription services
in 2016 of $0.0022 per performance, with these rates adjusting in each year from 2017 to 2020 to
reflect changes in the Consumer Price Index.8 These continue to be the statutory commercial
webcasting rates that apply today, with 2019 rates of $0.0018 per play for noninteractive ad-
supported services and $0.0023 per play for noninteractive subscription services.
B. Assignment and Summary of Conclusions
9. I have been retained by counsel for SoundExchange to explain and apply the
economics of bargaining theory to estimate reasonable royalty rates for the digital transmission of
sound recordings by commercial noninteractive webcasters operating under a statutory license.
Below, I present a multi-party bargaining approach that can be used to determine these rates,
known in the literature as the “Shapley Value.” I then bring data to bear in order to compute these
Shapley Values and the concomitant royalty rates at issue.9
10. In conducting my analyses and formulating my conclusions, I reviewed and
analyzed data concerning the various modes of music distribution, including relative audience
sizes, profitability, and royalty payments made to the owners of copyrights on sound recordings.
I examined and analyzed the results of a survey of consumers of music distribution that was
7 Id.
8 Web IV Determination at 1.9 I am aware that statutory royalty rates will also be set in this proceeding for noncommercial webcasting. Analysis of those rates is beyond the scope of my assignment and testimony.
Public Version
5
recently conducted under the supervision of Professor Gal Zauberman and produced in this matter,
as well as a survey conducted outside of this proceeding by Edison Research, a source of industry
data and research. I reviewed the written direct testimony of executives at major record companies
who are centrally involved with the licensing of sound recordings to digital music distribution
services. In addition, I relied on my career-long experience with economic theory, economic
literature, empirical study, and the regulated and unregulated practices of pricing products,
services, and assets, including intellectual property. Finally, in applying this background to the
matter at hand, I also benefitted from knowledge I have acquired from many years consulting on
music industry issues.
11. A complete list of materials I relied on in formulating the opinions in this report
can be found at Appendix B.
12. Based on my analysis, I conclude that Shapley Values are an appropriate tool for
assessing rates that would be negotiated in the hypothetical marketplace for noninteractive
webcasting. The approach is consistent with the objectives laid out in the relevant legal statute
and has been accepted by the Copyright Royalty Judges in prior rate-setting proceedings. My
Shapley Value analysis finds that a reasonable royalty rate for ad-supported noninteractive
webcasting beginning in 2021 is $0.00298 per play and for subscription noninteractive webcasting
is $0.0030 per play (or $1.965 per subscriber per month). These Shapley Value results are robust
to numerous alternative model specifications, different input data, and even an alternative multi-
party bargaining framework known as Nash-in-Nash.
III. SHAPLEY VALUES AND THEIR APPLICATION TO § 114 OBJECTIVES
13. Cooperative bargaining games, in general, are focused on how “surplus” is divided
by the parties to a negotiation. Surplus is the extra value that is jointly created by a group of
Public Version
6
cooperating parties over and above what those parties could create on their own. Specifically,
each party to a negotiation has a “fallback value” which is equal to the value that party could create
on its own without an agreement with the other parties. The sum of these fallback values is the
total value the parties could generate without an agreement. The extra value the parties could
generate with an agreement is the surplus.
14. Shapley Values are a generalized solution to the problem of how to apportion
among the members of a multi-party bargaining group the surplus created by their productive
cooperation with each other. This solution divides up the surplus according to each party’s
incremental contributions to the total amount of value created.10 These contributions are assessed
as increments to every possible combination of unilateral, bilateral, and multilateral deals that may
be struck by the different parties, and then averaged across all such combinations. In the recent
Phonorecords III proceeding, the Copyright Royalty Judges credited a Shapley Value analysis as
one way of addressing concerns about complementary oligopoly power, noting that the analysis
performed in that proceeding “eliminate[d] this ‘walk away’ power by valuing all possible
orderings of the players’ arrivals.”11
15. The concept of a Shapley Value is best understood by reference to a simple analogy.
Imagine that parties A, B, and C are negotiating a deal in person. Party C can be the first, the
10 See Lloyd S. Shapley, “A value for n-person games” (1953), in Alvin E. Roth (Ed.), The Shapley value, Essays in honor of Lloyd S. Shapley, Cambridge University Press (1988): 31-40. See alsoEyal Winter, “The Shapley Value,” in R.J. Aumann and S. Hart (Eds.), Handbook of Game Theory, Vol. 3, Chapter 53, Elsevier Science (2002): 2025-2054, pp. 2027-2030; Andreu Mas-Colell, Michael D. Whinston, and Jerry R. Green, Microeconomic Theory, Oxford University Press (1995), pp. 679-684. 11 Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms for Making and Distributing Phonorecords (Phonorecords III), Docket No. 16-CRB-0003-PR (“Phonorecords III Determination”) at 33 n.69.
Public Version
7
second, or the third to arrive in the room. The value it brings to the bargaining table may be
contingent on the order in which it arrives. For example, if Party C is last to the negotiation it may
have more bargaining power as a result of its ability to hold up or frustrate consummation of a deal
to which Parties A and B are otherwise amenable. When C is first to the negotiation, it has no
bargaining power over the others. Shapley analysis takes into account all such possible differences
in Party C’s bargaining power that are contingent on its order of arrival to the negotiation. It does
so by taking the average of each “incremental value” created by Party C in each possible sequence
of arrivals. As such, Party C’s Shapley Value will only be high relative to the other parties’
Shapley Values if, on average, it brings a relatively high incremental value to all possible orderings
and sub-orderings of Parties A, B, and C.
16. As the foregoing reveals, Shapley Values incorporate principles of fairness in the
allocation of the value created by the cooperation of multiple parties.12 No party’s allocation is
affected by its preferential position in the negotiations. This is so because each party’s allocation
is determined by an averaging over all possible sequences of its incremental impacts on the group’s
12 Meinhardt (2014) formally describes these principles of fairness: “The four original principles (axioms) as introduced by Shapley (1953) can be described verbally as follows: 1. (Efficiency): the solution should distribute the maximal total payoff; 2. (Symmetry) any two players who contribute the same input should obtain the same payoff (equal treatment of equals); 3. (Dummy player): any player who contributes nothing to any coalition should obtain his value; 4. (Additivity): adding the solution of two games together produces the solution of the sum of these games (in this sense, the solution is invariant against an arbitrary decomposition of the game). The Shapley value is the unique scheme of the game v that satisfies these principles.” Holger Ingmar Meinhardt, The Pre-Kernel as a Tractable Solution for Cooperative Games, An Exercise in Algorithmic Game Theory, Springer-Verlag (2014), p. 18 (note omitted). See also, e.g., Andreu Mas-Colell, Michael D. Whinston, and Jerry R. Green, Microeconomic Theory, Oxford University Press (1995), pp. 679-680 (“[The Shapley Value] attempts to describe a reasonable, or ‘fair,’ way to divide the gains from cooperation, taking as a given the strategic realities captured by the characteristic form.”) (emphasis in the original); Hervé Moulin, Fair Division and Collective Welfare, MIT Press (2003), p. 11 (“The Shapley value is a systematic formula used to divide a joint cost or a jointly produced output. It offers a reasonable definition and computation of the share of cost or surplus for which a user of the commons is deemed responsible.”).
Public Version
8
value. The Shapley Value accorded to a party rests on the value that it brings to the group’s
cooperation, taking into account all the subsets of the group to which it can join.
17. In analyzing the outcome of bargaining among cooperating parties, therefore, it is
necessary to assess and delineate the value that can result from the cooperation of any subset of
the overall cooperating group. The tool that delineates this value for a particular subset of the
group is called the “characteristic function” of the subset. Each combination of parties (including
subsets of the cooperating group that have only one, or even zero, parties) has its own characteristic
function value.
18. To further illustrate the intuition and mechanics behind Shapley Values, consider a
more complex example—a negotiation between the owner of patentA, the owner of patentB, and
the manufacturer of deviceD. Both patentA and patentB are necessary for the manufacture and
sale of deviceD. If deviceD is not on the market, the owners of patentA and patentB would each
still make a profit of $5 (their “fallback value”), through already committed agreements with rival
manufacturers. If a cooperative deal were to be made between the parties, the market profits from
the sales of deviceD would be $16. However, because deviceD would cannibalize the sales of
competing devices, the individual net returns of the owners of patentA and patentB from licensing
to other rival devices would decrease from $5 to $3. This $2 decline in the net returns of the
owners of patentA and patentB is their “opportunity cost” from entering a deal with the
manufacturer of deviceD.
19. For the above example, the characteristic function values for each possible subset
of the cooperating group are as follows:
Public Version
9
Figure 1: Example of Characteristic Function
Notice that the characteristic function is zero if there are no parties in the subset. But it is also
zero for the group of just deviceD, because that device cannot be marketed without both patentA
and patentB. The characteristic function is $5 for a subset comprised of just the owner of patentA
or just the owner of patentB, because again that is the fallback value for each of these parties. For
the same reason, the characteristic function for the subset comprised of both patent owners, but
not the device manufacturer, is $10—the sum of each patent owner’s fallback value. And, because
deviceD cannot be marketed without both patents, the characteristic function of the subset of
deviceD and patentA, and also the subset of deviceD and patentB, is just the $5 that arises from
the royalties earned by the patent owner from rival devices. The final row displays the
characteristic function of the full group of bargaining parties. This $22 is composed of the $16
from the market profits of deviceD plus the $3 in lowered royalties earned by each of the owners
of patentA and patentB from other devices when deviceD is in the market.
20. Calculation of the Shapley Value for the manufacturer of deviceD entails assessing
the incremental value that would be created by deviceD when it joins the group of parties that
arrived at the metaphorical bargaining table before it did (the “predecessor group”), in each of the
Subset
CharacteristicFunction Value
for Subset
No parties $0
deviceD manufacturer ("D") $0
patentA owner ("A") $5
patentB owner ("B") $5
A and B $10
A and D $5
B and D $5
A, B, and D $22
Public Version
10
six possible orderings of the parties’ arrivals.13 The incremental value of deviceD is the
characteristic function of the predecessor group with deviceD added less the characteristic function
of the predecessor group alone. The figure below displays the incremental value of deviceD for
each arrival ordering:
Figure 2: Example of Shapley Value for deviceD
21. As this figure illustrates, the Shapley Value of deviceD is the average of its
incremental values across all the possible orders of arrival, namely ($12 + $0 + $12 + $0 + $0 +
$0) / 6 = $4). Analogous calculations for patentA and patentB (which, in the interest of brevity, I
do not reproduce here) show that their Shapley Values are each $9.
22. Having solved for the Shapley Values of each party to the negotiation, it is now
possible to determine the royalties that the manufacturer of deviceD must pay out. In order to
receive its Shapley Value of $9, the owner of patentA must receive royalties of $6 from the
manufacturer of deviceD, which are added to its net return of $3 in royalties from rival devices.
13 In Shapley Value analysis there are always N! (i.e., N factorial) different arrival orderings, where N is the number of negotiating parties. For example, with three negotiating parties, there are 3 x 2 x 1 = 6 different arrival orderings.
Orderof
Arrival
SubsetPrior to
Arrival of D
CharacteristicFunction Value
for SubsetPrior to D
SubsetAfter
Arrival of D
CharacteristicFunction Value
for Subset with D Added
IncrementalValue of D
A, B, D A, B $10 A, B, D $22 $12
A, D, B A $5 A, D $5 $0
B, A, D B, A $10 B, A, D $22 $12
B, D, A B $5 B, D $5 $0
D, A, B None $0 D $0 $0
D, B, A None $0 D $0 $0
Average of Incremental Values of D (“Shapley Value”): $4
Public Version
11
Likewise, the owner of patentB must also receive $6 in royalties from the manufacturer of deviceD.
That means the manufacturer must pay out a total of $12 in royalties from its market profits of
$16. Importantly, this results in the manufacturer of deviceD receiving its Shapley Value of $4.
23. Note that, in this illustration, each party is better off with an agreement than without
one. Absent an agreement, each patent owner obtains only its $5 fallback value, whereas with an
agreement where surplus value is distributed according to Shapley Values each patent owner
receives a total value of $9. Furthermore, the $6 of royalties each patent owner receives from the
manufacturer of deviceD well exceeds their opportunity costs of $2 from the cannibalization of
other income resulting from the presence of deviceD in the market. If these conditions did not
hold, the patent owners would prefer not to enter into a deal with the maker of deviceD. And
finally, observe that a deal with the owners of patentA and patentB leaves the device manufacturer
better off—without a deal, deviceD cannot be brought to market and the manufacturer makes $0.
24. As this example makes clear, Shapley Values fit within the requirements of the
relevant legal statute, which calls on the Judges to consider copyright owners’ opportunity costs.
In the course of calculating the surplus from a deal between sound recording copyright owner and
noninteractive webcasting distributor, the Shapley Value model takes into account the extent to
which such a deal might enhance or cannibalize revenue from other sources, for either party. As
such, an appropriately specified Shapley Value model takes into account the extent to which
noninteractive distributors “substitute for or may promote the sales of phonorecords or otherwise
may interfere with or may enhance the sound recording copyright owner’s other streams of revenue
from [the copyright owner’s] sound recordings.”14 Furthermore, the model’s assessment of
fallback values necessarily takes account of the “relative roles of the copyright owner and the
14 17 U.S.C § 114(f)(2)(B).
Public Version
12
transmitting entity in the copyrighted work and the service made available to the public with
respect to relative creative contribution, technological contribution, capital investment, cost, and
risk.”15 Thus, it is my opinion that Shapley Values are an appropriate approach for assessing rates
that would be negotiated in the hypothetical marketplace for noninteractive webcasting.
IV. A SHAPLEY VALUE MODEL FOR RECORD COMPANIES AND NONINTERACTIVE WEBCASTERS
A. Defining the Model
25. Shapley Values depend upon how the parties to the hypothetical negotiation are
delineated. My representation of the hypothetical negotiation between record companies on the
one hand and noninteractive webcasting distributors on the other involves four upstream record
companies and two downstream noninteractive webcasting distributors. The first three record
companies represent each of the three major record companies (Universal Music Group, Sony
Music Entertainment, and Warner Music Group) and the fourth represents the combination of all
other record companies (i.e., the indies). The two noninteractive webcasting distributors represent
the combination of all ad-supported (i.e., free) noninteractive distributors and the combination of
all subscription noninteractive distributors, respectively.
26. It is my opinion that defining the hypothetical negotiations among these six players
is an appropriate elaboration of a negotiation between record companies on the one hand and
noninteractive webcasters on the other. This specification strikes a balance between offering a
granular and realistic depiction of the hypothetical market and maintaining enough simplicity
around the number of entities being modeled such that the model can be readily solved and
necessary data inputs can be estimated. First, delineating each of the major record companies
15 Id.
Public Version
13
individually along with the amalgam of indies accounts for the potentially different negotiating
positions of each. Second, delineating ad-supported noninteractive distributors separately from
subscription noninteractive distributors allows for the assessment of whether each form of
distribution should yield different royalty rates, as was found to be the case by the Judges in the
previous Web IV proceeding.16 Moreover, grouping all ad-supported noninteractive distributors
together and all subscription noninteractive distributors together is conservative from the
perspective of SoundExchange (i.e., produces lower royalty rates) because this modeling choice,
in effect, simplifies away rivalry among the various extant ad-supported noninteractive distributors
and among the various extant subscription noninteractive distributors. By eliminating
consideration of competition within these groups of distributors, their respective market power in
the negotiation is elevated.17 At the same time, the specification carefully models competition
between the collective ad-supported noninteractive distributor and the collective subscription
noninteractive distributor.
27. As noted, the characteristic function defines the value created by each possible
cooperating grouping of these six parties to the hypothetical negotiation (i.e., the four record
companies and two noninteractive distributors). Next, I describe considerations specific to
determining the value that each party or set of parties brings to the hypothetical negotiation.
28. The value brought by each record company is a function of both the costs it incurs
and the revenue it could generate by licensing its sound recordings to distributors other than the
noninteractive services (the record company’s “fallback value” as in the illustrations above).
There are a variety of other forms of music distributors, which are not involved in this hypothetical
16 Web IV Determination at 1. 17 Phonorecords III Determination at 66-67.
Public Version
14
negotiation, that pay royalties that are determined outside the ambit of this matter. These include:
ad-supported interactive (i.e., “on demand”) music and video services (e.g., free Spotify and free
YouTube); subscription interactive (i.e., “on demand”) music and video services (e.g., paid
Spotify, Apple Music, Amazon Music, YouTube Premium, etc.); SiriusXM satellite radio; and
physical/digital music purchases (e.g., digital downloads and CD album sales). These “outside”
distributors are not modeled as parties to the hypothetical bargaining process that will arrive at the
royalty rates to be determined in this matter. Each of their respective royalty rates are taken as
they actually are or are expected to be. Also in the background are the options of listening to
broadcast AM/FM radio or not listening to music, which are modeled realistically as not producing
any royalties for the record companies.
29. Determining a record company’s fallback value—which, again, is a key
component of the value it brings to the hypothetical negotiation—also requires evaluating what
would happen if each noninteractive distributor did not have access to that record company’s
music. If the hypothetical deal is not consummated, how much of each noninteractive distributor’s
audience would divert to other music listening options (including to the other noninteractive
distributor)? To answer this question, my analysis needs to consider the size of the audience of
each noninteractive distributor. My analysis then needs diversion parameters that represent the
proportion of these audiences that would divert to each alternative mode of distribution. And
finally, because the value each record company brings to the negotiation varies depending on its
relative share of overall music listening, my analysis needs an empirical input on each record
company’s respective share of plays.
30. The value brought by the noninteractive distributors to the hypothetical negotiation
depends on their ability to generate profits (before paying any sound recording royalties), which
Public Version
15
subtracts out various costs including the copyright royalties for musical works that distributors
must pay to publishers. Distributor profits are measured on a per-play basis.
31. The profits the noninteractive distributors create vary depending on which record
companies are present in each grouping of negotiating parties. Specifically, each of the three major
record companies is taken to be a “must-have,” meaning that the sound recording collection of
each is necessary for a noninteractive distributor to operate. To reference the illustration from
above, the major record companies are like the owners of patentA and patentB—without their
agreement to license, the manufacturer of deviceD cannot bring its product to market. This “must-
have” specification follows from the Copyright Royalty Judges’ conclusion in the Web IV
proceeding that “[t]here appears to be a consensus that the repertoire of each of the three Majors
is a ‘must have’ in order for a noninteractive service to be viable.”18 Hence, within the Shapley
Value model, without access to the sound recordings of all three of the major record companies, a
noninteractive distributor does not operate and contributes zero profits to the rest of the subset of
the bargaining parties. In contrast, the record company representing the indies does not have a
must-have repertoire. Instead, the model assumes that without access to the music of indies, a
distributor’s profits decline not to zero, but to an extent determined by listeners’ preferences for
the content carried by indies.
32. The empirical parameters described above are listed in Appendix C. These
parameters enable the algebraic specification of the characteristic function, which is derived and
displayed in that same appendix.
33. Once the characteristic function is specified, Shapley Values can be derived for
each party to the hypothetical negotiation, which is also accomplished in Appendix C. Recall
18 Web IV Determination at 133.
Public Version
16
Figure 2 above. The difference between the characteristic function for a subset of the parties
without the distributor and the characteristic function for that subset with the distributor added is
the “incremental value” that the distributor brings to the negotiation. By definition, the Shapley
Value for a distributor is the weighted average of the incremental values that the distributor brings
to every possible predecessor subset of the other bargaining parties.19
34. Finally, once Shapley Values are derived, the corresponding royalties from the two
noninteractive distributors to the record companies can be computed. These are the payments that
result in each party’s bottom line equaling its Shapley Value. For each distributor, the total royalty
payments it makes to the record companies must equal the difference between its profits from its
market operations and its Shapley Value. For each record company, the total royalty payments it
receives must equal the difference between its Shapley Value and the total compensation it receives
from its other sources of distribution, less its costs of operation.
B. Empirical Inputs
35. Under this Shapley Value analysis, royalty rates for both ad-supported and
subscription noninteractive distributors are a function of the following data inputs: (i) royalty rates
that record companies earn from other forms of music distribution; (ii) noninteractive distributors’
audience sizes; (iii) diversion ratios reflecting the amount of a noninteractive distributor’s audience
that would switch to other forms of music distribution and generate royalties if that noninteractive
distributor were unavailable; (iv) record company play shares; and (v) noninteractive distributors’
fixed costs and marginal profit rates. Because the hypothetical negotiation will set royalty rates
for the 2021-2025 period, this is the relevant time period for assessing these parameter values. In
19 The weighted average accords equal weight to the incremental value of the distributor to the group of parties that arrived before the distributor arrives in any one of the possible sequences of arrivals.
Public Version
17
other words, these parameters should reflect forward looking estimates to the greatest extent
possible. I discuss the estimation of each of these input parameters below. Additional details of
these data estimates are contained in Appendix D.
i. Royalties for Other Forms of Distribution
36. Rather than speculate about the path that royalty rates on other forms of music
distribution will take in the future, I use currently observable sound recording royalty rates as a
proxy for the sound recording royalty rates that will prevail during the relevant 2021-2025 period.
These rates are estimated using a variety of sources.
37. First, royalty rates for subscription on-demand streaming music and video services
(e.g., Spotify Premium, Apple Music, Amazon Music, YouTube Premium, etc.) are determined
from royalty statements that detail the royalty payments these services paid to various record
companies for each month over the 12-month period ending March 2019.20 The weighted average
amount of monthly royalties these services paid was approximately [ ] per subscriber. See
Appendix D at Ex. D.1.
38. Second, royalty rates for ad-supported on-demand streaming music and video
services (e.g., Spotify Free and free YouTube) are similarly determined from royalty statements.
The weighted average amount of royalties these services paid over the 12-month period ending
March 2019 was approximately [ ] per play. See Appendix D at Ex. D.2.
39. Third, royalty rates for SiriusXM satellite radio broadcasts are determined from (i)
Statements of Account provided by SiriusXM to SoundExchange detailing the total amount of
statutory royalties paid, over the 12-month period ending March 2019, for performances on
20 This is the most recent period of four consecutive calendar quarters for which royalty statements were consistently available for all services.
Public Version
18
satellite radio and from (ii) SiriusXM’s SEC Forms 10-K and 10-Q which detail its subscriber
counts. Specifically, SoundExchange reported total SDARS royalties of ] for
the 12-month period ending March 2019.21 SiriusXM’s 10-K and 10-Q filings reported the
quarter-end subscriber counts shown below.
Figure 3: SiriusXM Average Subscribers
From these amounts I compute quarterly average subscriber counts, and an average subscriber
count of 33.7 million for the full 12-month period ending March 2019, as shown in Figure 3.
Converting the ] of annual royalties to an average monthly amount and
dividing by the 33.7 million subscriber count results in SiriusXM monthly royalties per subscriber
of [ ]. Importantly, given the data currently available to me, these royalty rates are
artificially conservative, as they do not account for royalty payments that I understand are paid by
SiriusXM outside of SoundExchange, for example through directly negotiated licenses with
certain record companies. Nor do these royalty rates account for payments that I understand
SiriusXM must now make for its performance of sound recordings fixed prior to February 15, 1972
21 SoundExchange Exhibits 32-33.
Ending AverageSubscribers Subscribers
Quarter (millions) (millions)1Q18 33.12Q18 33.5 33.33Q18 33.7 33.64Q18 34.0 33.91Q19 34.2 34.1
Average: 33.7
Source: SiriuxXM 10-K and 10-Q filings.
Public Version
19
(which, until very recently, were not royalty-bearing under federal law).22
40. Fourth, royalty rates for physical music purchases (e.g., CDs and vinyl records) and
digital downloads are determined from 2018 wholesale and retail sales data from the Recording
Industry Association of America (“RIAA”) and from a 2018 Annual Music Study by industry
research firm MusicWatch prepared for the RIAA. According to MusicWatch, in 2018, CD
purchasers spent an average of [ ] on CDs, vinyl purchasers spent an average of [ ],
and digital download purchasers spent an average of [ ].23 Moreover, as shown in Appendix
D at Ex. D.3, RIAA data indicate that, in 2018, sound recording royalties totaled approximately
[ of the retail prices of CDs, vinyl, and digital downloads, respectively.
Multiplying these data and converting annual amounts to monthly amounts produces average
monthly royalties per purchaser of [ ] for CDs, [ ] for vinyl, and [ ] for digital
downloads, or a weighted average of [ ] across all forms of physical and digital download
distribution. See Appendix D at Ex. D.3.
41. Fifth, royalty rates for AM/FM broadcasts (and all other forms of outside music
distribution not listed above) are assumed to be zero.
42. Figure 4 below summarizes all of these royalty rates.
22 Until recently, I understand performances of sound recordings fixed before February 15, 1972 were not royalty-bearing under federal law. It is my understanding that changed late in 2018, when President Trump signed into law the Music Modernization Act. Given that the royalty statements I reviewed covered a time period when performances of pre-1972 recordings were not compensable under federal law, that data understates the revenue that SiriusXM can be expected to pay going forward. 23 SoundExchange Exhibit 42 at 19.
Public Version
20
Figure 4: Royalty Rates for Outside Distributors (RESTRICTED)
ii. Noninteractive Distributors’ Audience Sizes
43. Audience sizes for the noninteractive distributors are measured as total numbers of
plays per month. For Pandora, I am able to estimate projected audience sizes from its publicly
reported financial projections (which are discussed in more detail at ¶¶ 49-50 infra and Appendix
D at Ex. D.6). Further, I assume Pandora will have the same share of the noninteractive ad-
supported and subscription markets in 2021-2025 as it did over the recent 12-month period ending
March 2019. Royalty statements and data from SoundExchange reveal that Pandora has an
approximately [ ] play share of the ad-supported noninteractive market and an [ ] play
share of the subscription noninteractive market. See Appendix D at Ex. D.4. Using those shares,
I am able to gross up the Pandora audience size amounts to reflect the total size of the audience
for all noninteractive distributors.
44. Accordingly, projected audience sizes can be estimated as follows:
Public Version
21
Figure 5: Estimated Average Audience Size (2021-2025) (RESTRICTED)
45. Last, my analysis requires converting these play counts into per-user metrics for
interaction with the per-user information derived from the survey results described infra. To do
so, I compare projected average monthly play counts for Pandora’s ad-supported tier to the
projected number of active users of Pandora’s ad-supported service. Similarly, for Pandora’s
subscription noninteractive tier, I compare projected average monthly play counts on that tier to
the projected number of subscribers. Again, these figures are derived from Pandora’s public
financial projections. See Appendix D at Ex. D.6. I find that users of Pandora’s ad-supported
service are projected to listen to approximately [ ] plays per month and subscribers to Pandora’s
subscription noninteractive service (i.e., Pandora Plus) are projected to listen to approximately
[ ] plays per month over the 2021-2025 period.
iii. Diversion Ratios
46. Diversion ratios to other forms of music distribution if a noninteractive distributor
were no longer available in the marketplace are estimated from the results of a survey conducted
under the direction of Gal Zauberman and thoroughly described in his report (“Zauberman
Public Version
22
Survey”).24 The survey obtained responses from hundreds of users of both ad-supported and
subscription noninteractive services. Survey respondents were first asked how they currently listen
to music and then how they would listen to music under hypothetical scenarios where
noninteractive services are no longer available. Some pertinent details of the survey design and
responses are provided in Appendix E. In addition, Appendix E details how I use these survey
responses to derive diversion ratios between noninteractive distributors and from noninteractive
distributors to outside forms of music distribution, including subscription and ad-supported
interactive streaming music and video services, satellite radio, physical album purchases and
digital downloads, and broadcast AM/FM radio.
47. In Figure 6 below, I summarize these diversion ratios in Panel A, which shows the
percentage of survey respondents that would respond to the absence of noninteractive services by
obtaining new music subscriptions, beginning to purchase CDs/digital downloads, or consuming
additional monthly plays on ad-supported streaming services. Panel B shows the corresponding
levels of royalty compensation that the record companies would earn as a result from these various
forms of music distribution. Finally, Panel C multiplies the diversion rates in Panel A by the
royalty rates in Panel B to obtain an estimate of the opportunity cost that record companies
experience by licensing to noninteractive distributors instead of only licensing to all the outside
forms of music distribution. (AM/FM radio is not shown below as it generates zero royalties and
thus zero opportunity cost.)
24 See Written Direct Testimony of Professor Gal Zauberman, Sept. 23, 2019 (“Zauberman WDT”).
Public Version
23
Figure 6: Diversion Ratios and Opportunity Cost (RESTRICTED)
iv. Record Company Play Shares
48. Music content shares for each of the major record companies and for the indies
Public Version
24
together as a group are estimated from the royalty statements that music streaming and video
services provide to record companies when operating under directly negotiated license
agreements.25 These monthly royalty statements denote both the total plays on that service in any
given month as well as the number of those plays that were copyrighted recordings of the record
company. I have no basis to believe record company shares are anticipated to be substantially
different during the 2021-2025 period than they are currently. Thus, I use the most recent monthly
royalty statements available from interactive streaming music and video services and
noninteractive services, for the 12-month period ending March 2019.26 These statements allow
me to calculate the proportion of total plays of Universal-, Sony-, Warner-, and indie-owned
content, respectively. Specifically, I compute each major record company’s play shares, and then
compute the play share for the combination of all indies as a group as 100% less the sum of the
major record companies’ shares.27 See Appendix D at Ex. D.5. The resulting play shares are
summarized in the figure below:
25 Play counts for services operating under statutory noninteractive licenses are not included. However, all of Pandora’s noninteractive play counts and iHeart’s subscription noninteractive play counts are included as these services currently operate under directly negotiated licenses and, accordingly, report their total play counts to the major record companies. 26 I evaluate all tiers of service on the following services: Apple Music, Amazon Music Unlimited, Amazon Prime, Google Play, iHeart (both interactive and noninteractive tiers), Pandora (both interactive and noninteractive tiers), Napster, Spotify, Vevo, and YouTube. Play share data from SiriusXM, and from physical retail and digital downloads, were not available to me. I have no reason to think the content of any of the record companies is played with more or less frequency on these distribution methods, when compared to the distribution methods (interactive and noninteractive streaming) for which I did have data. Thus, I have no reason to believe this additional data would materially change the respective shares of plays among Universal, Sony, Warner, and the indies. 27 InGrooves and The Orchard, wholly owned subsidiaries of Universal and Sony, respectively, are treated as part of the group of indies. Treating them instead as part of their affiliated must-have major record companies has a de minimis effect on the royalty rates derived from my Shapley Value model.
Public Version
25
Figure 7: Estimated Play Shares (RESTRICTED)
v. Noninteractive Distributor Fixed Costs and Marginal Profit Rates
49. As noted above, the value brought by the noninteractive distributors to the
hypothetical negotiation depends on their ability to generate profits (before paying any sound
recording royalties), which subtract out from revenues various costs including the copyright
royalties for musical works that distributors must pay to publishers. See supra ¶ 30. To estimate
fixed costs, variable or marginal costs and the associated marginal profit rates of noninteractive
distributors, I rely on Pandora financials as a proxy. Pandora is the only noninteractive service for
which I have been able to obtain financial data that allow for forward-looking estimates of ad-
supported and subscription noninteractive service profit rates.28 Furthermore, Pandora is the
largest noninteractive distributor in the market, accounting for more than [ ] of total plays in
the noninteractive market (see Appendix D at Ex. D.4).
50. I use the forward-looking estimates of Pandora’s costs and profits for the relevant
2021-2025 period disclosed in the company’s recent Merger Proxy.29 These projections were
28 Pandora now offers a fully interactive subscription tier, Pandora Premium, whose data must be removed from Pandora’s financial results to obtain estimates of ad-supported and subscription noninteractive services’ profit rates. 29 As explained in Appendix D, estimating marginal profit rates from Pandora’s recent historical financials would result in figures with an inappropriate downward bias with respect to the hypothetical negotiation because the company’s recent financial results show substantial operating losses. Notwithstanding these recent losses, it is clear that Pandora is anticipated to be profitable in the future, as demonstrated by the approximately $3.5 billion purchase price paid by SiriusXM to acquire the company, as well as Pandora’s substantial market capitalization of approximately $2.4 billion immediately prior to the announcement of the SiriusXM acquisition. Thus, Pandora’s recent history of operating losses does not accurately reflect expectations about the incremental
Public Version
26
disclosed to Pandora shareholders prior to their vote to approve SiriusXM’s acquisition of the
company and were utilized by Pandora’s investment bankers as an input into their fairness
opinions. As described in greater detail in Appendix D, using these projections I estimate the
following monthly fixed costs and marginal profit rates:
Figure 8: Estimated Fixed Costs and Marginal Profit Rates (2021-2025)30
C. Summary of Results
51. Using the formulas derived in Appendix C and the data inputs discussed above, I
derive Shapley Values and corresponding royalty rates for ad-supported and subscription
noninteractive distributors. These results are summarized below:
value Pandora is anticipated to bring to the hypothetical negotiation concerning royalty rates for the 2021-2025 period. 30 As explained in Appendix D, certain key inputs to the Pandora projections were not disclosed in Pandora’s proxy statements (e.g., projected ad-supported user and subscriber counts, projected plays, and a breakdown of subscription revenue into its underlying Pandora Plus and Pandora Premium component parts). Accordingly, certain allocation assumptions were required to estimate key parameters from Pandora’s projected financial information. Estimates derived from these projections may require amendment following the completion of discovery.
Ad-SupportedNoninteractive
SubscriptionNoninteractive
Fixed Costs (per month) $40.4 million $8.9 million
Marginal Profit Rate $0.0042 per play $0.0048 per play
Public Version
27
Figure 9: Estimated Shapley Values and Royalty Rates for Noninteractive Distributors31
(amounts are monthly except per play rates) (RESTRICTED)
31 I calculate a 95 percent confidence interval for these royalty rates using a bootstrap procedure on the survey results. The details of this computation are described in Appendix E. The resulting 95 percent confidence interval ranges from $0.002910 to $0.00299 for the ad-supported noninteractive royalty rate and from $0.00299 to $0.003176 for the subscription noninteractive royalty rate.
Public Version
28
Notably, the resulting Shapley Value for the ad-supported noninteractive distributor is near zero.
This is the case because, as Figure 9 shows, the record companies’ opportunity costs are high
relative to the total projected profits of those distributors (before paying the royalties at issue).32
Hence, the vast majority of those profits are necessary to compensate the record companies for the
ad-supported noninteractive distributors’ cannibalization of listeners that would otherwise
32 Note that the total opportunity costs shown in Figure 9 exceed the opportunity cost subtotals from Figure 6 because they include not only the portion of opportunity cost associated with diversion to outside distributors, but also the additional portion of opportunity cost associated with diversion to the other noninteractive distributor. (The record companies’ total opportunity cost associated with licensing sound recordings to the ad-supported noninteractive distributor depends, in part, on the amount by which the market operations of ad-supported noninteractive distributors cannibalize the audience of subscription noninteractive distributors, and vice versa.) These additional amounts of opportunity cost are computed as the diversion of plays from one noninteractive distributor to the other (as reflected in Panel A of Figure 6) times the relevant per play royalty rates shown in Figure 9.
Public Version
29
consume music via other compensatory forms of music distribution.33
52. The finding that the record companies’ opportunity costs are high relative to the
profits of noninteractive distributors is perhaps not surprising in light of internal documents from
33 This finding does not exclude the possibility that the record companies and distributors of ad-supported noninteractive services with multiple service tiers would find it mutually advantageous to enter into individual agreements that provide lower royalty rates than those available under the statute in exchange for commitments to convert listeners to their affiliated subscription tiers that pay higher royalty rates.
For example, Spotify, an on-demand streaming service, has shown an ability to convert users from its ad-supported on-demand tier to its subscription on-demand tier. As explained by its CFO in September 2018, approximately half of the users of its ad-supported service that remain engaged with the service ultimately become paying subscribers. See SoundExchange Exhibit 66 (Transcript of Goldman Sachs Communacopia Conference, Sept. 14, 2018). For this ad-supported tier, Spotify has been able to negotiate paying record companies an effective royalty rate of approximately [ ] per play (on average over the 12-month period ending March 2019). See Appendix D at Ex. D.2.
[
.
Even though many ad-supported noninteractive distributors do not have subscription service tiers, I nonetheless test whether Pandora’s financial projections imply significant conversions of Pandora Free users to subscribers, and whether such conversion rates would affect the royalty rates specified by my model. Using Pandora’s public projections for its streaming services for 2021 to 2025, see Appendix D at Ex. D.6, I conservatively assume that any differences in the projected growth rates of Pandora Plus and Pandora Premium as compared to Pandora Free are due solely to the conversion of Pandora Free users to one of Pandora’s subscription services, and that this conversion would not occur if Pandora Free were not operating as a result of failing to reach a deal in the hypothetical negotiation with one or more of the must-have record companies. In effect, such conversions offset and reduce the opportunity cost to the record companies associated with the market operations of Pandora’s ad-supported service. I find that Pandora’s financial projections imply a relatively low rate of conversion of ad-supported listeners to paying subscribers (as compared to that implied by the Spotify CFO’s comments above). Accounting for this conversion in a dynamic Shapley Value model does not substantially impact royalty rates for ad-supported noninteractive distributors.
Public Version
30
the services that [
].35, 36
53. The finding that ad-supported noninteractive services, on their own, currently bring
little (if any) incremental value to the hypothetical negotiation is also consistent with the testimony
of record company executives. For example, Mark Piibe (Executive Vice President for Global
Business Development and Digital Strategy at Sony Music Entertainment) explains:
As far as I know, [
34 .
35 [
. 36 [
.
Public Version
31
.37
Similarly, Aaron Harrison (Senior Vice President, Business & Legal Affairs, Digital, UMG
Recordings, Inc.) explains: “Ordinarily, UMG only sees value in a standalone ad-supported audio
streaming service where the rates are sufficiently high on their own to justify licensing our content.
To that end, [
].”38
Moreover, in describing the value brought by subscription noninteractive services, Mr. Harrison
states that “even if mid-tier subscription services succeed in drawing some consumers away from
poorly-monetized ad-supported streaming services, there is also a danger that they could
cannibalize the premium on-demand subscription services” and thus [
]39
54. Finally, it is notable that streaming services themselves describe the ad-supported
tier not as an independent generator of meaningful revenue, but as a “funnel” to higher-priced
subscription tiers. Spotify’s 2018 annual report states that “[o]ur Ad-Supported Service serves as
a funnel, driving more than 60% of our total gross added Premium Subscribers since we began
37 Written Direct Testimony of Mark Piibe, Sept. 23, 2019. 38 Written Direct Testimony of Aaron Harrison, Sept. 23, 2019. 39 Id.
Public Version
32
tracking this data in February 2014.”40 Similarly, [
].42
55. In summary, my Shapley Value model indicates a reasonable royalty rate for ad-
supported noninteractive distributors of $0.0030 per play over the 2021-2025 period, and for
subscription noninteractive distributors of $0.0031 per play (or $2.043 per subscriber per month)
over that period. Notably, these royalty rates are estimated using projected data for the 2021-2025
period, and as such, reflect “on average” rates for the entire five-year period. To the extent it is
appropriate to express a rate as of the beginning of this period (such that the rate can be adjusted
over the 2021-2025 period based on increases or decreases in the general price level, as is the case
for the current statutory rates that were determined in the Web IV proceeding), these rates can be
discounted back two years (from the mid-point of 2023 to the mid-point of 2021) using a discount
rate equal to the rate of inflation. A two percent estimate of inflation results in a discount factor
of 0.96117.43 Multiplying this discount factor by the above rates produces a 2021 royalty rate of
$0.00298 per play for ad-supported noninteractive services and $0.0030 for subscription
40 SoundExchange Exhibit 71 (Spotify 2018 Form 20-F) at 38. 41 [
].
42 [.
43 The U.S. Federal Open Market Committee’s inflation rate forecast for 2021 is two percent. See SoundExchange Exhibit 22. 0.96117 = 1 / (1+.02)^2.
Public Version
33
noninteractive services (or $1.965 per subscriber per month).
D. Alternative Shapley Value Calculations Using “Share of Ear” Data
56. As an alternative to the Shapley Value computations presented above, I also
consider an alternative source of data from which diversion ratios may be estimated. For this
alternative calculation I use data from Edison Research, an industry research firm, instead of the
data from the Zauberman Survey.
57. Edison Research publishes a quarterly study, known as “Share of Ear,” that
analyzes the share of time Americans spend listening to all different forms of music distribution.
The study is based on a survey that asks several thousand respondents aged 13 and over who own
smartphones to track their music listening behavior.44 I have obtained the most recent Q2 2019
Share of Ear research report.
58. I apply a logit demand model to the data from this report, with the implication that
if ad-supported or subscription noninteractive distributors were not available in the marketplace,
their audiences would divert to other forms of music listening in the same relative proportions as
those reflected in the share of listening time figures reported in the Share of Ear study. Because
these diversion ratios are specified in terms of listening time, instead of the number of listeners as
in the Zauberman Survey, I modify my Shapley Value model accordingly to allow for the
incorporation of these alternative diversion input parameters. Appendix F provides a complete
44 The Share of Ear study also tracks the U.S. population more broadly by including non-smartphone owners, but I rely on the smartphone owner subset (approximately [ ] of the total population) as this subset of respondents seems to more closely represent the behavior of the relevant group of noninteractive streaming users than would the total U.S. population at large. Regardless, my alternative Shapley Value calculations using Share of Ear Data do not differ substantially if the total population of Share of Ear respondents is used instead of the smartphone owner subset.
Public Version
34
description of how I estimate diversion ratios from the Edison Research 2Q 2019 Share of Ear
study and revise my Shapley Value model to incorporate these data.
59. Applying the Share of Ear diversion parameters produces the following Shapley
Value results:45
Figure 10: Alternative Royalty Rates Using Share-of-Ear-Based Diversion Rates (amounts are monthly except per play rates) (RESTRICTED)
45 Appendix F describes the robustness of these results to various alternative specifications.
Public Version
35
60. As can be seen in the table above, the royalty rates determined by the Shapley Value
model using the alternative Share of Ear data are highly consistent with those determined from the
data generated by the Zauberman Survey. Incorporating the discount factor discussed above to
convert these amounts to 2021 values produces an ad-supported noninteractive rate of $0.00287
per play and a subscription noninteractive rate of $0.0030 per play ($1.943 per subscriber per
month).
E. Alternative Calculations Using a Nash-in-Nash Bargaining Model
61. Shapley Value analysis is not the only approach embraced by the economics
literature for modeling a multi-party negotiation. “Nash-in-Nash bargaining” is another such
framework recently articulated as a model of market outcomes without reliance on fairness. I
apply this alternative approach to further test the sensitivity of my Shapley Value results.
62. In Nash-in-Nash bargaining, the parties negotiate in pairs, and each pair’s
Public Version
36
agreement may affect the equilibrium outcomes of the other negotiating pairs.46 Determination of
the Nash-in-Nash solution involves two distinct steps. In the first step, the bargaining solution
between each negotiating pair is modeled with the “Nash Bargaining Solution”47 concept, whereby
each party is accorded its fallback value plus one half of the surplus created by the deal.48 During
this negotiation, the two parties assume that all other pairs of parties have reached (or will reach)
an equilibrium agreement. Thus, this first step determines an equation representing the Nash
Bargaining Solution for each negotiating pair, where the resulting solution is a function of the
solutions from every other negotiating pair. In the second step, a unique solution, known as a Nash
Equilibrium, is found where there is no negotiating pair with an incentive to change its
agreement.49 This is done by simultaneously solving the system of equations generated in the first
step.
63. In short, the Nash-in-Nash bargaining solution models each potential pair of
bargainers by assuming that each bargaining pair correctly anticipates that all other pairs have each
reached or will reach their equilibrium agreements, i.e., agreements that will not change with the
46 See, e.g., Allan Collard-Wexler, Gautam Gowrisankaran, and Robin Lee, “‘Nash-in-Nash’ Bargaining: A Microfoundation for Applied Work,” Journal of Political Economy, Vol. 127, No. 1 (2019): 163-195, p. 165. 47 The Nash Bargaining Solution is a widely-accepted theory used to characterize the outcomes of bilateral bargaining between willing buyers and willing sellers. In 1994, John Nash was awarded the Nobel Prize in economics for his work. See, e.g., John Nash, “The Bargaining Problem,” Econometrica, Vol. 18, No. 2 (1950): 155-162. See also Henrick Horn and Asher Wolinsky, “Bilateral monopolies and incentives for merger,” RAND Journal of Economics, Vol. 19, No. 3 (1988): 408-419, pp. 411-412; Ian McDonald and Robert Solow, “Wage Bargaining and Employment,” American Economic Review, Vol. 71, No. 5 (1981): 896-908, p. 905. 48 It is these fallback payoffs that represent and cause disparities in bargaining power notwithstanding the even split of the surplus, as discussed below. 49 See, e.g., Jean Tirole, The Theory of Industrial Organization, MIT Press (1988), p. 206.
Public Version
37
outcome of the current bargaining pair’s negotiation. In multi-party Nash-in-Nash bargaining, the
parties’ relative bargaining strengths in their negotiations are set by their fallback values, which
are generally affected by their own and other parties’ proposed deals in the marketplace, while the
surplus generated by each pair of negotiating parties relative to their fallback values is split on a
50/50 basis. For this reason, empirical work in economics often uses Nash-in-Nash bargaining to
model competitive outcomes in multi-party settings.50
64. The 50/50 split of surplus that is applied in Nash-in-Nash bargaining has been
mischaracterized by some as effecting an overall “50/50 split” between the parties to a negotiation.
However, this view fails to capture the very real sense in which the Nash Bargaining Solution can
result in an unequal allocation of value between two parties, depending on their respective
bargaining positions. Again, the relative bargaining positions of the parties are reflected in their
often disparate fallback values. When fallback values are not the same, the Nash Bargaining
Solution employed in Nash-in-Nash will result in something other than a 50/50 split of the total
value generated by the cooperating parties. Only the surplus, which requires both parties’ equal
cooperation, is split on a 50/50 basis, which is economically appropriate for the simple reason that
no party could obtain any of the surplus without equal participation by the other.
65. In Appendix G, I provide an illustration of the Nash-in-Nash bargaining approach
and then algebraically derive my Nash-in-Nash bargaining model using the same parameters as
employed in the Shapley Value model.
50 See, e.g., Gautam Gowrisankaran, Aviv Nevo, and Robert Town, “Mergers When Prices Are Negotiated: Evidence from the Hospital Industry,” American Economic Review, Vol. 105, No. 1 (2015): 172-203. See also Gregory S. Crawford, Robin S. Lee, Michael D. Whinston, and Ali Yurukoglu, “The Welfare Effects of Vertical Integration in Multichannel Television Markets,” Econometrica, Vol. 86, No. 3 (2018): 891-954, p. 910. (The paper cites multiple studies that have employed the “Nash-in-Nash” bargaining solution.).
Public Version
38
66. Application of the same input data used in the Shapley Value model based on the
Zauberman Survey produces the quantitative solution to the Nash-in-Nash model shown below.
Figure 11: Estimated Royalty Rates Using Alternative Nash-in-Nash Model (amounts are monthly except per play rates) (RESTRICTED)
67. As can be seen in this figure, the Nash-in-Nash model produces similar royalty rates
to those obtained from the Shapley Value model. Incorporating the discount factor discussed
above to convert these amounts to 2021 values produces an ad-supported noninteractive rate of
Public Version
39
$0.0029 per play and a subscription noninteractive rate of $0.0029 per play ($1.898 per subscriber
per month).
V. CONCLUSION
68. As explained in this report, it is my conclusion that Shapley Values are an
appropriate tool for assessing rates that would be negotiated in the hypothetical marketplace for
noninteractive webcasting. The approach is consistent with the objectives laid out in the relevant
legal statute and has been accepted by the Copyright Royalty Judges in prior rate-setting
proceedings. My Shapley Value analysis finds that a reasonable royalty rate for ad-supported
noninteractive webcasting beginning in 2021 is $0.00298 per play and for subscription
noninteractive webcasting is $0.0030 per play (or $1.965 per subscriber per month). These
Shapley Value results are robust to numerous alternative model specifications, input data, and
bargaining frameworks.
Public Version
I declare under penalty of perjury that the foregoing testimony is true and correct.
,*", (<tlrqobert Willig
Public Version
A-1
Appendix A: Curriculum Vitae and Testimony List
September 23, 2019Curriculum Vitae
Name: Robert D. Willig
Address: 220 Ridgeview Road, Princeton, New Jersey 08540
Birth: 1/16/47; Brooklyn, New York
Marital Status: Married, four children
Education: Ph.D. Economics, Stanford University, 1973 Dissertation: Welfare Analysis of Policies
Affecting Prices and Products. Advisor: James Rosse
M.S. Operations Research, Stanford University, 1968.
A.B. Mathematics, Harvard University, 1967.
Professional Positions:
Lecturer with the rank of Professor, Woodrow Wilson School of Public and International Affairs, 2/2017 – 6/2020.
Professor of Economics and Public Affairs, Emeritus, Princeton University, 7/2016 –
Professor of Economics and Public Affairs, Princeton University, 7/1978 - 6/2016.
Principal External Advisor, Infrastructure Program, Inter-American Development Bank, 6/97- 8/98.
Deputy Assistant Attorney General, U.S. Department of Justice, l989-1991.
Supervisor, Economics Research Department, Bell Laboratories, 1977-1978.
Visiting Lecturer (with rank of Associate Professor), Department of Economics and Woodrow Wilson School, Princeton University, 1977-78 (part time).
Public Version
A-2
Economics Research Department, Bell Laboratories, 1973-77.
Lecturer, Economics Department, Stanford University, 1971-73.
Other Professional Activities
ABA Section of Antitrust Law Economics Task Force, 2010-2012
Advisory Committee, Compass Lexecon 2010 -
OECD Advisory Council for Mexican Economic Reform, 2008 - 2009
Senior Consultant, Compass Lexecon, 2008 -
Director, Competition Policy Associates, Inc., 2003-2005
Advisory Bd., Electronic Journal of I.O. and Regulation Abstracts, 1996-2008.
Advisory Board, Journal of Network Industries, 2004-2010.
Visiting Faculty Member (occasional), International Program on Privatization and Regulatory Reform, Harvard Institute for International Development, 1996-2000.
Member, National Research Council Highway Cost Allocation Study Review Committee, 1995-98.
Member, Defense Science Board Task Force on the Antitrust Aspects of Defense Industry Consolidation, 1993-94.
Editorial Board, Utilities Policy, 1990-2001.
Leif Johanson Lecturer, University of Oslo, November 1988.
Member, New Jersey Governor's Task Force on Market-Based Pricing of Electricity, 1987-89.
Co-editor, Handbook of Industrial Organization, 1984-89.
Associate Editor, Journal of Industrial Economics, 1984-89.
Director, Consultants in Industry Economics, Inc., 1983-89, 1991-94.
Fellow, Econometric Society, 1981-.
Organizing Committee, Carnegie-Mellon-N.S.F. Conference on Regulation, 1985.
Public Version
A-3
Board of Editors, American Economic Review, 1980-83.
Nominating Committee, American Economic Association, 1980-1981.
Research Advisory Committee, American Enterprise Institute, 1980-1986.
Editorial Board, M.I.T. Press Series on Government Regulation of Economic Activity, 1979-93.
Program Committee, 1980 World Congress of the Econometric Society.
Program Committee, Econometric Society, 1979, 1981, 1985.
Organizer, American Economic Association Meetings: 1980, 1982.
American Bar Association Section 7 Clayton Act Committee, 198l.
Principal Investigator, NSF grant SOC79-0327, 1979-80; NSF grant 285-6041, 1980-82; NSF grant SES-8038866, 1983-84, 1985-86.
Aspen Task Force on the Future of the Postal Service, 1978-80.
Organizing Committee of Sixth Annual Telecommunications Policy Research Conference, 1977-78.
Visiting Fellow, University of Warwick, July 1977.
Institute for Mathematical Studies in the Social Sciences, Stanford University, 1975.
Published Articles and Book Chapters:
“Economic Foundations for 21st Century Freight Rail Rate Regulation,” (with John W. Mayo), in U.S. Freight Rail Economics and Policy: Are We on the Right Track?, Macher, J.T. and J.W. Mayo (eds.), Routledge, 2019.
“Is Professor Salop Right That Judge Leon Bungled United States v. AT&T?” (with Janusz Ordover and J. Gregory Sidak), The Criterion Journal on Innovation, v. 3, 249-263, 2018.
“The Role of the Circle Principle in Market Definition,” (with Bryan Keating and Jonathan Orszag), The Antitrust Source, April 2018, 1-6.
“Economy-wide and Sectoral Impacts on Workers of Brazil’s Internet Rollout” (with Mark A. Dutz, Lucas Ferreira Mation, and Stephen D. O’Connell), Forum for Social Economics, 46:2, 160-
Public Version
A-4
177, 2017.
“Two-Sided Market Definition and Competitive Effects for Credit Cards After United States v. American Express” (with J. Gregory Sidak), The Criterion Journal on Innovation, v. 1, 1301, 2016.
"Unilateral Competitive Effects" (with Bryan Keating), in The Oxford Handbook on International Antitrust Economics, (Roger D. Blair and D. Daniel Sokol, eds.), Oxford University Press, 2014.
“Activating Actavis: A More Complete Story” (with Barry C. Harris, Kevin M. Murphy, and Matthew B. Wright), Antitrust, vol. 28, No. 2 (Spring), 2014.
“'Reverse Payments' in Settlements of Patent Litigation: Split Opinions on Schering-Plough’s K- Dur (2005 and 2012)" (with John P. Bigelow), in The Antitrust Revolution (Sixth Edition), (J. Kwoka and Laurence White, eds.), Oxford University Press, 2013.
"The Delta-Northwest Merger: Consumer Benefits from Airline Network Effects (2008)" (with Mark Israel, Bryan Keating and Daniel Rubinfeld), in The Antitrust Revolution (Sixth Edition), (J. Kwoka and Laurence White, eds.), Oxford University Press, 2013.
"Airline Network Effects and Consumer Welfare" (with Bryan Keating, Mark Israel and Daniel Rubinfeld), Review of Network Economics, published online November 2013.
"The Liftoff of Consumer Benefits from the Broadband Revolution" (with Mark Dutz and JonOrszag), Review of Network Economics (2012) vol. 11, issue 4, article 2.
"Competition and innovation-driven inclusive growth" (with Mark Dutz, Ioannis Kessides and Stephen O’Connell), in Promoting Inclusive Growth: Challenges and Policies, Luiz de Mello and Mark Dutz (eds.), OECD, 2011.
"Unilateral Competitive Effects of Mergers: Upward Pricing Pressure, Product Quality, andOther Extensions," Review of Industrial Organization (2011) 39:19–38.
“Antitrust and Patent Settlements: The Pharmaceutical Cases,” (with John Bigelow) in The Antitrust Revolution (Fifth Edition), John Kwoka and Lawrence White (eds.), 2009.
“The 1982 Department of Justice Merger Guidelines: An Economic Assessment,” (with J. Ordover) reprinted in Economics of Antitrust Law, Benjamin Klein (ed.), Edward Elgar, 2008.
“On the Antitrust Treatment of Production Joint Ventures,” (with Carl Shapiro) reprinted in Economics of Antitrust Law, Benjamin Klein (ed.), Edward Elgar, 2008.
“Consumer’s Surplus Without Apology,” reprinted in Applied Welfare Economics, Richard Just, Darrel Hueth and Andrew Schmitz (eds.), Edward Elgar, 2008; reprinted in Readings in Social Welfare: Theory and Policy, Robert E. Kuenne (ed.), Blackwell, 2000, pp. 86-97;
Public Version
A-5
reprinted in Readings in Microeconomic Theory, M. M. La Manna (ed.), Dryden Press, 1997, pp. 201-212.
“The Risk of Contagion from Multi-Market Contact,” (with Charles Thomas), The International Journal of Industrial Organization, Vol. 24, Issue 6 (Nov. 2006), pp 1157 – 1184.
“Pareto-Superior Nonlinear Outlay Schedules,” reprinted in The Economics of Public Utilities, Ray Rees (ed.), Edward Elgar, 2006; reprinted in The Economics of Price Discrimination, G. Norman, (ed.), Edward Elgar, 1999.
“Economic Effects of Antidumping Policy,” reprinted in The WTO and Anti-Dumping, Douglas Nelson (ed.), Edward Elgar, 2005.
“Merger Analysis, Industrial Organization Theory and the Merger Guidelines,” reprinted in Antitrust and Competition Policy, Andrew Kleit (ed.) Edward Elgar, 2005
“Antitrust Policy Towards Agreements That Settle Patent Litigation,” (with John Bigelow), Antitrust Bulletin, Fall 2004, pp. 655-698.
“Economies of Scope,” (with John Panzar), reprinted in The Economics of Business Strategy, John Kay (ed.), Edward Elgar, 2003.
“Panel on Substantive Standards for Mergers and the Role of Efficiencies,” in International Antitrust Law & Policy, Barry E. Hawk (ed.), Juris Publishing, 2003.
“Practical Rules for Pricing Access in Telecommunications,” (with J. Ordover) in Second Generation Reforms in Infrastructure Services , F. Basanes and R. Willig (eds.), Johns Hopkins Press, 2002.
“Comments on Antitrust Policy in the Clinton Administration,” in American Economic Policy in the 1990s, J. Frankel and P. Orszag (eds.), MIT Press, 2002.
“Entrepreneurship, Access Policy and Economic Development: Lessons from Industrial Organization,” (with M. Dutz and J. Ordover), European Economic Review, (44)4-6 (2000), pp. 739-747.
"Public Versus Regulated Private Enterprise," reprinted in Privatization in Developing Countries, P. Cook and C. Kirkpatrick (eds.), Edward Elgar, 2000.
“Deregulation v. the Legal Culture: Panel Discussion,” in Is the Telecommunications Act of 1996 Broken?, G. Sidak (ed.), AEI Press, 1999.
“Economic Principles to Guide Post-Privatization Governance,” in Can Privatization Deliver? Infrastructure for Latin America, R. Willig co-editor, Johns Hopkins Press, 1999.
Public Version
A-6
“Access and Bundling in High-Technology Markets,” (with J. A. Ordover), in Competition, Innovation and the Microsoft Monopoly: Antitrust in the Digital Marketplace, J. A. Eisenach and T. Lenard (eds.), Kluwer, 1999.
“Competitive Rail Regulation Rules: Should Price Ceilings Constrain Final Products or Inputs?,” (With W. J. Baumol), Journal of Transport Economics and Policy, Vol. 33, Part 1, pp. 1-11.
“Economic Effects of Antidumping Policy,” Brookings Trade Forum: 1998, 19-41.
“Interview With Economist Robert D. Willig,” Antitrust , Vol. 11, No. 2, Spring 1997, pp.11-15.
“Parity Pricing and its Critics: A Necessary Condition for Efficiency in Provision of Bottleneck Services to Competitors,” (with W. J. Baumol and J. A. Ordover), Yale Journal on Regulation, Vol. 14, No. 1, Winter 1997, pp. 145-164.
“Restructuring Regulation of the Rail Industry,” (with Ioannis Kessides), in Private Sector, Quarterly No. 4, September 1995, pp. 5 - 8. Reprinted in Viewpoint, October, 1995, The World Bank. Reprinted in Private Sector, special edition: Infrastructure, June 1996.
“Competition and Regulation in the Railroad Industry,” (with Ioannis Kessides), in Regulatory Policies and Reform: A Comparative Perspective, C. Frischtak (ed.), World Bank, 1996.
"Economic Rationales for the Scope of Privatization," (with Carl Shapiro), reprinted in The Political Economy of Privatization and Deregulation, E. E. Bailey and J. R. Pack (eds.), The International Library of Critical Writings in Economics, Edward Elgar Publishing Co., 1995, pp. 95-130.
"Weak Invisible Hand Theorems on the Sustainability of Multi-product Natural Monopoly," (with W. Baumol and E. Bailey), reprinted in The Political Economy of Privatization and Deregulation, E. E. Bailey and J. R. Pack (eds.), The International Library of Critical Writings in Economics, Edward Elgar Publishing Co., 1995, pp. 245-260.
“Economists’ View: The Department of Justice Draft Guidelines for the Licensing and Acquisition of Intellectual Property,” (with J. Ordover), Antitrust, V. 9, No. 2 (spring 1995), 29- 36.
"Public Versus Regulated Private Enterprise," in Proceedings of the World Bank Annual Conference on Development Economics 1993, L. Summers (ed.), The World Bank, 1994.
"Economics and the 1992 Merger Guidelines: A Brief Survey," (with J. Ordover), Review of Industrial Organization, V. 8, No. 2, (1993), pp. 139-150.
"The Role of Sunk Costs in the 1992 Guidelines' Entry Analysis," Antitrust, V. 6, No. 3 (summer 1992).
Public Version
A-7
"Antitrust Lessons from the Airlines Industry: The DOJ Experience," Antitrust Law Journal, V. 60, No. 2 (1992).
"William J. Baumol," (with E. E. Bailey), in New Horizons in Economic Thought: Appraisals of Leading Economists, W. J. Samuels (ed.), Edward Elgar, 1992.
"Anti-Monopoly Policies and Institutions," in The Emergence of Market Economies in Eastern Europe, Christopher Clague and Gordon Rausser (eds.), Basil Blackwell, 1992.
"Economics and the 1992 Merger Guidelines," (with Janusz Ordover), in Collaborations Among Competitors: Antitrust Policy and Economics, Eleanor Fox and James Halverson (eds.), American Bar Association, 1992.
"On the Antitrust Treatment of Production Joint Ventures," (with Carl Shapiro), reprinted in Collaborations Among Competitors: Antitrust Policy and Economics, Eleanor Fox and James Halverson (eds), American Bar Association, 1992.
"Merger Analysis, Industrial Organization Theory, and Merger Guidelines," Brookings Papers on Economic Activity -- Microeconomics 1991, pp. 281-332.
"On the Antitrust Treatment of Production Joint Ventures," (with C. Shapiro), Journal of Economic Perspectives, Vol. 4, No. 3, Summer 1990, pp. 113-130.
"Economic Rationales for the Scope of Privatization," (with Carl Shapiro), in The Political Economy of Public Sector Reform and Privatization, E.N. Suleiman and J. Waterbury (eds.), Westview Press, Inc., 1990, pp. 55-87.
"Contestable Market Theory and Regulatory Reform," in Telecommunications Deregulation: Market Power and Cost Allocation, J.R. Allison and D.L. Thomas (eds.), Ballinger, 1990.
"Address To The Section," Antitrust Law Section Symposium, New York State Bar Association, 1990.
"Price Caps: A Rational Means to Protect Telecommunications Consumers and Competition," (with W. Baumol), Review of Business, Vol. 10, No. 4, Spring 1989, pp. 3-8.
"U.S.-Japanese VER: A Case Study from a Competition Policy Perspective," (with M. Dutz) in The Costs of Restricting Imports, The Automobile Industry. OECD, 1988.
"Contestable Markets," in The New Palgrave: A Dictionary of Economics, J. Eatwell, M. Milgate, and P. Newman (eds.), 1987.
"Do Entry Conditions Vary Across Markets: Comments," Brookings Papers on Economic Activity, 3 - 1987, pp. 872-877.
Public Version
A-8
"Railroad Deregulation: Using Competition as a Guide," (with W. Baumol), Regulation, January/February 1987, Vol. 11, No. 1, pp. 28-36.
"How Arbitrary is 'Arbitrary'? - or, Toward the Deserved Demise of Full Cost Allocation," (with W. Baumol and M. Koehn), Public Utilities Fortnightly, September 1987, Vol. 120, No. 5, pp. 16-22.
"Contestability: Developments Since the Book," (with W. Baumol), Oxford Economic Papers, December 1986, pp. 9-36.
"The Changing Economic Environment in Telecommunications: Technological Change and Deregulation," in Proceedings from the Telecommunications Deregulation Forum; Karl Eller Center; 1986.
"Perspectives on Mergers and World Competition," (with J. Ordover), in Antitrust and Regulation, R.E. Grieson (ed.), Lexington, 1986.
"On the Theory of Perfectly Contestable Markets," (with J. Panzar and W. Baumol), in New Developments in The Analysis of Market Structure, J. Stiglitz and F. Mathewson (eds.), MIT Press, 1986.
"InterLATA Capacity Growth and Market Competition," (with C. Shapiro), in Telecommunications and Equity: Policy Research Issues, J. Miller (ed.), North Holland, 1986.
"Corporate Governance and Market Structure," in Economic Policy in Theory and Practice, A. Razin and E. Sadka (eds.), Macmillan Press, 1986.
"Antitrust for High-Technology Industries: Assessing Research Joint Ventures and Mergers," (with J. Ordover), Journal of Law and Economics, Vol 28(2), May 1985, pp. 311-334.
"Non-Price Anticompetitive Behavior by Dominant Firms Toward the Producers of Complementary Products," (with J. Ordover and A. Sykes), in Antitrust and Regulation, F.M. Fisher (ed.), MIT Press, 1985.
"Telephones and Computers: The Costs of Artificial Separation," (with W. Baumol), Regulation, March/April 1985.
"Transfer Principles in Income Redistribution," (with P. Fishburn), Journal of Public Economics, 25 (1984), pp. 1-6.
"Market Structure and Government Intervention in Access Markets," in Telecommunications Access and Public Policy, A. Baughcam and G. Faulhaber (eds.), 1984.
"Pricing Issues in the Deregulation of Railroad Rates," (with W. Baumol), in Economic
Public Version
A-9
Analysis of Regulated Markets: European and U. S. Perspectives, J. Finsinger (ed.), 1983.
"Local Telephone Pricing in a Competitive Environment," (with J. Ordover), in Telecommunications Regulation Today and Tomorrow, E. Noam (ed.), Harcourt Brace Jovanovich, 1983.
"Economics and Postal Pricing Policy," (with B. Owen), in The Future of the Postal Service, J. Fleishman (ed.), Praeger, 1983.
"Selected Aspects of the Welfare Economics of Postal Pricing," in Telecommunications Policy Annual, Praeger, 1987.
"The Case for Freeing AT&T" (with M. Katz), Regulation, July-Aug. 1983, pp. 43-52.
"Predatory Systems Rivalry: A Reply" (with J. Ordover and A. Sykes), Columbia Law Review, Vol. 83, June 1983, pp. 1150-1166. Reprinted in Corporate Counsel's Handbook - 1984.
"Sector Differentiated Capital Taxation with Imperfect Competition and Interindustry Flows," Journal of Public Economics, Vol. 21, 1983.
"Contestable Markets: An Uprising in the Theory of Industry Structure: Reply," (with W.J. Baumol and J.C. Panzar), American Economic Review, Vol. 73, No. 3, June 1983, pp. 491-496.
"The 1982 Department of Justice Merger Guidelines: An Economic Assessment," (with J. Ordover), California Law Review, Vol. 71, No. 2, March 1983, pp. 535-574. Reprinted in Antitrust Policy in Transition: The Convergence of Law and Economics, E.M. Fox and J.T. Halverson (eds.), 1984.
"Intertemporal Failures of the Invisible Hand: Theory and Implications for International Market Dominance," (with W.J. Baumol), Indian Economic Review, Vol. XVI, Nos. 1 and 2, January-June 1981, pp. 1-12.
"Unfair International Trade Practices," (with J. Ordover and A. Sykes), Journal of International Law and Politics, Vol. 15, No. 2, winter 1983, pp. 323-337.
"Journals as Shared Goods: Reply," (with J. Ordover), American Economic Review, V. 72, No. 3, June 1982, pp. 603-607.
"Herfindahl Concentration, Rivalry, and Mergers," (with J. Ordover and A. Sykes), Harvard Law Review, V. 95, No. 8, June 1982, pp. 1857-l875.
"An Economic Definition of Predation: Pricing and Product Innovation," (with J. Ordover), Yale Law Journal, Vol. 90: 473, December 1981, pp. 1-44.
"Fixed Costs, Sunk Costs, Entry Barriers, and the Sustainability of Monopoly," (with W.
Public Version
A-10
Baumol), Quarterly Journal of Economics, Vol. 96, No. 3, August 1981, pp. 405-432.
"Social Welfare Dominance," American Economic Review, Vol. 71, No. 2, May 1981, pp. 200-204.
"Economies of Scope," (with J. Panzar), American Economic Review, Vol. 72, No. 2, May 1981, pp. 268-272.
"Income-Distribution Concerns in Regulatory Policymaking," (with E.E. Bailey) in Studies in Public Regulation (G. Fromm, ed.), MIT Press, Cambridge, 1981, pp. 79-118.
"An Economic Definition of Predatory Product Innovation," (with J. Ordover), in Strategic Predation and Antitrust Analysis, S. Salop (ed.), 1981.
"What Can Markets Control?" in Perspectives on Postal Service Issues, R. Sherman (ed.), American Enterprise Institute, 1980.
"Pricing Decisions and the Regulatory Process," in Proceedings of the 1979 Rate Symposium on Problems of Regulated Industries, University of Missouri-Columbia Extension Publications, 1980, pp. 379-388.
"The Theory of Network Access Pricing," in Issues in Public Utility Regulation, H.M. Trebing (ed.), MSU Public Utilities Papers, 1979.
"Customer Equity and Local Measured Service," in Perspectives on Local Measured Service, J. Baude, etal. (ed.), 1979, pp. 71-80.
"The Role of Information in Designing Social Policy Towards Externalities," (with J. Ordover), Journal of Public Economics, V. 12, 1979, pp. 271-299.
"Economies of Scale and the Profitability of Marginal-Cost Pricing: Reply," (with J. Panzar), Quarterly Journal of Economics, Vol. 93, No. 4, Novmber 1979, pp. 743-4.
"Theoretical Determinants of the Industrial Demand for Electricity by Time of Day," (with J. Panzar) Journal of Econometrics, V. 9, 1979, pp. 193-207.
"Industry Performance Gradient Indexes," (with R. Dansby), American Economic Review, V. 69, No. 3, June 1979, pp. 249-260.
"The Economic Gradient Method," (with E. Bailey), American Economic Review, Vol. 69, No. 2, May 1979, pp. 96-101.
"Multiproduct Technology and Market Structure," American Economic Review, Vol. 69, No. 2, May 1979, pp. 346-351.
Public Version
A-11
"Consumer's Surplus Without Apology: Reply," American Economic Review, Vol. 69, No. 3, June 1979, pp. 469-474.
"Decisions with Estimation Uncertainty," (with R. Klein, D. Sibley, and L. Rafsky), Econometrica, V. 46, No. 6, November 1978, pp. 1363-1388.
"Incremental Consumer's Surplus and Hedonic Price Adjustment," Journal of Economic Theory, V. 17, No. 2, April 1978, pp. 227-253.
"Recent Theoretical Developments in Financial Theory: Discussion, "The Journal of Finance, V. 33, No. 3, June 1978, pp. 792-794.
"The Optimal Provision of Journals Qua Sometimes Shared Goods," (with J. Ordover), American Economic Review, V. 68, No. 3, June 1978, pp. 324-338.
"On the Comparative Statics of a Competitive Industry With Infra-marginal Firms," (with J. Panzar), American Economic Review, V. 68, No. 3, June 1978, pp. 474-478.
"Pareto Superior Nonlinear Outlay Schedules," Bell Journal of Economics, Vol. 9, No. 1, Spring 1978, pp. 56-69.
"Predatoriness and Discriminatory Pricing," in The Economics of Anti-Trust: Course of Study Materials, American Law Institute-American Bar Association, 1978.
"Economies of Scale in Multi-Output Production," (with J. Panzar), Quarterly Journal of Economics, V. 91, No. 3, August 1977, pp. 481-494.
"Weak Invisible Hand Theorems on the Sustainability of Multi-product Natural Monopoly," (with W. Baumol and E. Bailey), American Economic Review, V. 67, No. 3, June 1977, pp. 350-365.
"Free Entry and the Sustainability of Natural Monopoly," (with J. Panzar), Bell Journal of Economics, Spring 1977, pp. 1-22.
"Risk Invariance and Ordinally Additive Utility Functions," Econometrica, V. 45, No. 3, April 1977, pp. 621-640.
"Ramsey-Optimal Pricing of Long Distance Telephone Services," (with E. Bailey), in Pricing in Regulated Industries, Theory and Application, J. Wenders (ed.), Mountain State Telephone and Telegraph Co., 1977, pp. 68-97.
"Network Externalities and Optimal Telecommunications Pricing: A Preliminary Sketch," (with R. Klein), in Proceedings of Fifth Annual Telecommunications Policy Research Conference, Volume II, NTIS, 1977, pp. 475-505.
Public Version
A-12
"Otsenka ekonomicheskoi effektivnosti proizvodstvennoi informatsii" ["The Evaluation of the Economic Benefits of Productive Information"] in Doklady Sovetskikh i Amerikanskikh Spetsialistov Predstavlennye na Pervyi Sovetsko-Amerikanskii Simpozium po Ekonomicheskoi Effektivnosti Informat sionnogo Obsluzhivaniia [Papers of Soviet and American Specialists Presented at the First Soviet- American Symposium on Costs and Benefits of Information Services], All Soviet Scientific Technical Information Center, Moscow, 1976.
"Vindication of a 'Common Mistake' in Welfare Economics," (with J. Panzar), Journal of Political Economy, V. 84, No. 6, December 1976, pp. 1361-1364.
"Consumer's Surplus Without Apology," American Economic Review, V. 66, No. 4, September 1976, pp. 589-597.
Books
Second Generation Reforms in Infrastructure Services, F. Basanes and R. Willig (eds.), Johns Hopkins Press, 2002.
Can Privatization Deliver? Infrastructure for Latin America, R. Willig co-editor, Johns Hopkins Press, 1999.
Handbook of Industrial Organization, (edited with R. Schmalensee), North Holland Press, Volumes 1 and 2, 1989.
Contestable Markets and the Theory of Industry Structure, (with W.J. Baumol and J.C. Panzar), Harcourt Brace Jovanovich, 1982. Second Edition, 1989.
Welfare Analysis of Policies Affecting Prices and Products, Garland Press, 1980.
Unpublished Papers and Reports:
“Brief Amici Curiae of 37 Economists, Antitrust Scholars, and Former Government Antitrust Officials In Support of Appellees and Supporting Affirmance,” In the United States Court of Appeals for the District of Columbia Circuit; USCA Case #18-5214; United States of America, Plaintiff-Appellant, v. AT&T, Inc.; DirecTV Group Holdings, LLC; and TimeWarner Inc., Defendants-Appellees; On appeal from a final judgment of the U.S. District Court for the District of Columbia, Hon. Richard J. Leon, No. 1:17-cv-2511; 9/26/2018.
“Brief for Amici Curiae J. Gregory Sidak and Robert D. Willig in Support of Respondents,” In the Supreme Court of the United States; In the Supreme Court of the United States; State of Ohio, et al., v. American Express Company, et al., On Writ Of Certiorari To The United States Court Of Appeals For The Second Circuit; No. 16-1454; January 23, 2018.
Public Version
A-13
“Brief of Leading Economists as Amici Curiae in Support of Respondents,” In the Supreme Court of the United States; Douglas R. M. Nazarian, et al, v. PPL Energyplus, LLC, et al. and CPV Maryland, LLC, v. PPL Energyplus, LLC, et al.; On Writ of Certiorari to the US Court of Appeals for the Fourth Circuit; Nos. 14-614, 14-623; January 19, 2016.
“Technological change and labor market segmentation in the developing world: Evidence from Brazil,” (with Dutz, Mark, Lucas Ferreira-Mation, and Stephen O’Connell), 2015 Background Paper for the 2016 World Bank’s World Development Report.
“Brief for Amici Curiae J. Gregory Sidak, Robert D. Willig, David J. Teece, and Keith N. Hylton, Scholars and Experts in Antitrust Economics in Support of Defendants-Appellants and Supporting Reversal,” 15-1672 In the United States Court of Appeals for the Second Circuit; United States of America, et al., v. American Express Company, et al., 8/10/2015.
"Commentary on Economics at the FTC: Hospital Mergers, Authorized Generic Drugs, and Consumer Credit Markets" (with Nauman Ilias, Bryan Keating, and Paolo Ramezzana, Margaret Guerin-Calvert, and Nauman Ilias), 20126.
"Recommendations for Excessive-Share Limits in the Surfclam and Ocean Quahog Fisheries" (with Glenn Mitchell and Steven Peterson), Report to National Marine Fisheries Service and the Mid-Atlantic Fishery Management Council, 5/23/2011.
"Public Comments on the 2010 Draft Horizontal Merger Guidelines," paper posted to Federal Trade Commission website, 6/4/2010
“An Economic Perspective on the Antitrust Case Against Intel,” (with Jon Orszag and Gilad Levin), 2009.
"An Econometric Analysis of the Matching Between Football Student-Athletes and Colleges," (with Yair Eilat, Bryan Keating and Jon Orszag)
Supreme Court Amicus Brief Regarding Morgan Stanley Capital Group Inc. v. Public Utility District No. 1 of Snohomish County, Washington, (co-authored), AEI-Brookings Joint Center Brief No. 07-02, 12/2/07
“(Allegedly) Monopolizing Tying Via Product Innovation,” statement before the Department of Justice/Federal Trade Commission Section 2 Hearings, November 1, 2006.
“Assessment of U.S. Merger Enforcement Policy,” statement before the Antitrust Modernization Commission, 11/17/05.
“Investment is Appropriately Stimulated by TELRIC,” in Pricing Based on Economic Cost, 12/2003.
Public Version
A-14
“Brief of Amici Curiae Economics Professors, re Verizon v. Trinko, In the Supreme Court of the U.S.,” (with W.J. Baumol, J.O. Ordover and F.R. Warren-Boulton), 7/25/2003.
“Stimulating Investment and the Telecommunications Act of 1996,” (with J. Bigelow, W. Lehr and S. Levinson), 2002.
“An Economic Analysis of Spectrum Allocation and Advanced Wireless Services,” (with Martin N. Baily, Peter R. Orszag, and Jonathan M. Orszag), 2002
“Effective Deregulation of Residential Electric Service,” 2001
“Anticompetitive Forced Rail Access” (with W. J. Baumol), 2000
“The Scope of Competition in Telecommunications” (with B. Douglas Bernheim), 1998 “Why
Do Christie and Schultz Infer Collusion From Their Data? (with Alan Kleidon), 1995.
"Demonopolization," (with Sally Van Siclen), OECD Vienna Seminar Paper, 1993.
"Economic Analysis of Section 337: The Balance Between Intellectual Property Protection and Protectionism," (with J. Ordover) 1990.
"The Effects of Capped NTS Charges on Long Distance Competition," (with M. Katz).
"Discussion of Regulatory Mechanism Design in the Presence of Research Innovation, and Spillover Effects," 1987.
"Industry Economic Analysis in the Legal Arena," 1987.
"Deregulation of Long Distance Telephone Services: A Public Interest Assessment," (with M. Katz).
"Competition-Related Trade Issues," report prepared for OECD.
"Herfindahl Concentration Index," (with J. Ordover), Memorandum for ABA Section 7 Clayton Act Committee, Project on Revising the Merger Guidelines, March 1981.
"Market Power and Market Definition," (with J. Ordover), Memorandum for ABA Section 7 Clayton Act Committee, Project on Revising the Merger Guidelines, May 1981.
"The Continuing Need for and National Benefits Derived from the REA Telephone Loan Programs - An Economic Assessment," 1981.
"The Economics of Equipment Leasing: Costing and Pricing," 1980.
Public Version
A-15
"Rail Deregulation and the Financial Problems of the U.S. Railroad Industry," (with W.J. Baumol), report prepared under contract to Conrail, 1979.
"Price Indexes and Intertemporal Welfare," Bell Laboratories Economics Discussion Paper, 1974.
"Consumer's Surplus: A Rigorous Cookbook," Technical Report #98, Economics Series, I.M.S.S.S., Stanford University, l973.
"An Economic-Demographic Model of the Housing Sector," (with B. Hickman and M. Hinz), Center for Research in Economic Growth, Stanford University, 1973.
Invited Conference Presentations:
NYU Stern and Concurrence Global Antitrust Economics Conference “Pharmaceutical Industry Antitrust” -- moderator 2019
FTC Competition and Consumer Protection Hearings "The State of US Antitrust Law" 2018
Portuguese Competition Authority Public Seminar Series “Ups and Downs of Horizontal and Vertical Mergers” 2017
World Bank Workshop on Digital Technology Adoption, Skills, Productivity and Jobs in Latin America
“Discussion of Models of Firm Heterogeneity” 2016
George Mason Law Review Annual Antitrust Symposium: Antitrust in an Interconnected World “GUPPI and the Safe Harbor” 2016
Competition Law & Policy Institute of New Zealand Annual Workshop “Merger Analysis Keynote” 2015
Economic Studies at Brookings: Railroads, Policy and the Economy “The Industry Perspective” 2015
Georgetown University McDonough School of Business Railroad Economics Symposium “The Role of Economic Theory in the ‘Deregulated’ Rail Industry” 2015
Brazilian School of Economics and Finance (FGV EPGE) Seminario “Public Interest Regulation: Lessons from Railroads” 2015
Public Version
A-16
NYU School of Law Conference on the Fiftieth Anniversary of United States v. Philadelphia National Bank: The Past, Present and Future of Merger Law “Discussion with Agency Economists” 2013
Brookings Institution Conference on The Economics of the Airline Industry "Airline Network Effects and Consumer Welfare" 2012
AGEP Public Policy Conference on Pharmaceutical Industry Economics, Regulation and Legal Issues; Law and Economics Center, George Mason University School of Law "Pharmaceutical Brand-Generic Disputes" 2012
U.S.-EU Alliance Study Peer Review Conferences "Review of Cooperative Agreements in Transatlantic Airline Markets" 2012 "The Research Agenda Ahead" 2012
Antitrust in the High Tech Sector Conference "Developments in Merger Enforcement" 2012
Georgetown Center for Business and Public Policy, Conference on the Evolution of Regulation "Reflections on Regulation" 2011
Antitrust Forum, New York State Bar Association "Upward Price Pressure, Market Definition and Supply Mobility" 2011
American Bar Association, Antitrust Section, Annual Convention "The New Merger Guidelines' Analytic Highlights" 2011
OECD and World Bank Conference on Challenges and Policies for Promoting Inclusive Growth "Inclusive Growth From Competition and Innovation" 2011
Villanova School of Business Executive MBA Conference "Airline Network Effects, Competition and Consumer Welfare" 2011
NYU School of Law Conference on Critical Directions in Antitrust "Unilateral Competitive Effects" 2010
Conf. on the State of European Competition Law and Enforcement in a Transatlantic Context "Recent Developments in Merger Control" 2010
Center on Regulation and Competition, Universidad de Chile Law School "Economic Regulation and the Limits of Antitrust Law" 2010
Center on Regulation and Competition, Universidad de Chile Law School "Merger Policy and Guidelines Revision" 2010
Public Version
A-17
Faculty of Economics, Universidad de Chile "Network Effects in Airlines Markets" 2010
Georgetown Law Global Antitrust Enforcement Symposium "New US Merger Guidelines" 2010
FTI London Financial Services Conference "Competition and Regulatory Reform" 2010
NY State Bar Association Annual Antitrust Conference “New Media Competition Policy” 2009
Antitrust Law Spring Meeting of the ABA “Antitrust and the Failing Economy Defense” 2009
Georgetown Law Global Antitrust Enforcement Symposium “Mergers: New Enforcement Attitudes in a Time of Economic Challenge” 2009
Phoenix Center US Telecoms Symposium “Assessment of Competition in the Wireless Industry” 2009
FTC and DOJ Horizontal Merger Guidelines Workshop “Direct Evidence is No Magic Bullet” 2009
Northwestern Law Research Symposium: Antitrust Economics and Competition Policy "Discussion of Antitrust Evaluation of Horizontal Mergers" 2008
Inside Counsel Super-Conference "Navigating Mixed Signals under Section 2 of the Sherman Act" 2008
Federal Trade Commission Workshop on Unilateral Effects in Mergers "Best Evidence and Market Definition" 2008
European Policy Forum, Rules for Growth: Telecommunications Regulatory Reform “What Kind of Regulation For Business Services?” 2007
Japanese Competition Policy Research Center, Symposium on M&A and Competition Policy “Merger Policy Going Forward With Economics and the Economy” 2007
Federal Trade Commission and Department of Justice Section 2 Hearings “Section 2 Policy and Economic Analytic Methodologies” 2007
Pennsylvania Bar Institute, Antitrust Law Committee CLE “The Economics of Resale Price Maintenance and Class Certification” 2007
Public Version
A-18
Pennsylvania Bar Institute, Antitrust Law Committee CLE “Antitrust Class Certification – An Economist’s Perspective” 2007
Fordham Competition Law Institute, International Competition Economics Training Seminar “Monopolization and Abuse of Dominance” 2007
Canadian Bar Association Annual Fall Conference on Competition Law “Economic Tools for the Competition Lawyer” 2007
Conference on Managing Litigation and Business Risk in Multi-jurisdiction Antitrust Matters “Economic Analysis in Multi-jurisdictional Merger Control” 2007
World Bank Conference on Structuring Regulatory Frameworks for Dynamic and Competitive South Eastern European Markets “The Roles of Government Regulation in a Dynamic Economy” 2006
Department of Justice/Federal Trade Commission Section 2 Hearings “(Allegedly) Monopolizing Tying Via Product Innovation” 2006
Fordham Competition Law Institute, Competition Law Seminar “Monopolization and Abuse of Dominance” 2006
Practicing Law Institute on Intellectual Property Antitrust “Relevant Markets for Intellectual Property Antitrust” 2006
PLI Annual Antitrust Law Institute “Cutting Edge Issues in Economics” 2006
World Bank’s Knowledge Economy Forum V “Innovation, Growth and Competition” 2006
Charles University Seminar Series “The Dangers of Over-Ambitious Antitrust Regulation” 2006
NY State Bar Association Antitrust Law Section Annual Meeting “Efficient Integration or Illegal Monopolization?” 2006
World Bank Seminar “The Dangers of Over-Ambitious Regulation” 2005
ABA Section of Antitrust Law 2005 Fall Forum “Is There a Gap Between the Guidelines and Agency Practice?” 2005
Hearing of Antitrust Modernization Commission “Assessment of U.S. Merger Enforcement Policy” 2005
Public Version
A-19
LEAR Conference on Advances in the Economics of Competition Law “Exclusionary Pricing Practices” 2005
Annual Antitrust Law Institute “Cutting Edge Issues in Economics” 2005
PRIOR Symposium on States and Stem Cells “Assessing the Economics of State Stem Cell Programs” 2005
ABA Section of Antitrust Law – AALS Scholars Showcase “Distinguishing Anticompetitive Conduct” 2005
Allied Social Science Associations National Convention “Antitrust in the New Economy” 2005
ABA Section of Antitrust Law 2004 Fall Forum “Advances in Economic Analysis of Antitrust” 2004
Phoenix Center State Regulator Retreat “Regulatory Policy for the Telecommunications Revolution” 2004
OECD Competition Committee “Use of Economic Evidence in Merger Control” 2004
Justice Department/Federal Trade Commission Joint Workshop “Merger Enforcement” 2004
Phoenix Center Annual U.S. Telecoms Symposium “Incumbent Market Power” 2003
Center for Economic Policy Studies Symposium on Troubled Industries “What Role for Government in Telecommunications?” 2003
Princeton Workshop on Price Risk and the Future of the Electric Markets “The Structure of the Electricity Markets” 2003
2003 Antitrust Conference “International Competition Policy and Trade Policy” 2003
International Industrial Organization Conference “Intellectual Property System Reform” 2003
ABA Section of Antitrust Law 2002 Fall Forum “Competition, Regulation and Pharmaceuticals” 2002
Public Version
A-20
Fordham Conference on International Antitrust Law and Policy “Substantive Standards for Mergers and the Role of Efficiencies” 2002
Department of Justice Telecom Workshop “Stimulating Investment and the Telecommunications Act of 1996” 2002
Department of Commerce Conference on the State of the Telecom Sector “Stimulating Investment and the Telecommunications Act of 1996” 2002
Law and Public Affairs Conference on the Future of Internet Regulation “Open Access and Competition Policy Principles” 2002
Center for Economic Policy Studies Symposium on Energy Policy “The Future of Power Supply” 2002
The Conference Board: Antitrust Issues in Today’s Economy “The 1982 Merger Guidelines at 20” 2002
Federal Energy Regulatory Commission Workshop “Effective Deregulation of Residential Electric Service” 2001
IPEA International Seminar on Regulation and Competition “Electricity Markets: Deregulation of Residential Service” 2001 “Lessons for Brazil from Abroad” 2001
ABA Antitrust Law Section Task Force Conference “Time, Change, and Materiality for Monopolization Analyses” 2001
Harvard University Conference on American Economic Policy in the 1990s “Comments on Antitrust Policy in the Clinton Administration” 2001
Tel-Aviv Workshop on Industrial Organization and Anti-Trust “The Risk of Contagion from Multimarket Contact” 2001
2001 Antitrust Conference “Collusion Cases: Cutting Edge or Over the Edge?” 2001 “Dys-regulation of California Electricity” 2001
FTC Public Workshop on Competition Policy for E-Commerce “Necessary Conditions for Cooperation to be Problematic” 2001
HIID International Workshop on Infrastructure Policy “Infrastructure Privatization and Regulation” 2000
Villa Mondragone International Economic Seminar “Competition Policy for Network and Internet Markets” 2000
Public Version
A-21
New Developments in Railroad Economics: Infrastructure Investment and Access Policies “Railroad Access, Regulation, and Market Structure” 2000
The Multilateral Trading System at the Millennium “Efficiency Gains From Further Liberalization” 2000
Singapore – World Bank Symposium on Competition Law and Policy “Policy Towards Cartels and Collusion” 2000
CEPS: Is It a New World?: Economic Surprises of the Last Decade “The Internet and E-Commerce” 2000
Cutting Edge Antitrust: Issues and Enforcement Policies “The Direction of Antitrust Entering the New Millennium” 2000
The Conference Board: Antitrust Issues in Today’s Economy “Antitrust Analysis of Industries With Network Effects” 1999
CEPS: New Directions in Antitrust “Antitrust in a High-Tech World” 1999
World Bank Meeting on Competition and Regulatory Policies for Development “Economic Principles to Guide Post-Privatization Governance” 1999
1999 Antitrust Conference “Antitrust and the Pace of Technological Development” 1999 “Restructuring the Electric Utility Industry” 1999
HIID International Workshop on Privatization, Regulatory Reform and Corporate Governance “Privatization and Post-Privatization Regulation of Natural Monopolies” 1999
The Federalist Society: Telecommunications Deregulation: Promises Made, Potential Lost?
“Grading the Regulators” 1999
Inter-American Development Bank: Second Generation Issues In the Reform Of Public Services
“Post-Privatization Governance” 1999 “Issues Surrounding Access Arrangements” 1999
Economic Development Institute of the World Bank -- Program on Competition Policy “Policy Towards Horizontal Mergers” 1998
Twenty-fifth Anniversary Seminar for the Economic Analysis Group of the Department of
Public Version
A-22
Justice “Market Definition in Antitrust Analysis” 1998
HIID International Workshop on Privatization, Regulatory Reform and Corporate Governance “Infrastructure Architecture and Regulation: Railroads” 1998
EU Committee Competition Conference – Market Power “US/EC Perspective on Market Definition” 1998
Federal Trade Commission Roundtable “Antitrust Policy for Joint Ventures” 1998
1998 Antitrust Conference “Communications Mergers” 1998
The Progress and Freedom Foundation Conference on Competition, Convergence, and the Microsoft Monopoly
Access and Bundling in High-Technology Markets 1998
FTC Program on The Effective Integration of Economic Analysis into Antitrust Litigation The Role of Economic Evidence and Testimony 1997
FTC Hearings on Classical Market Power in Joint Ventures Microeconomic Analysis and Guideline 1997
World Bank Economists --Week IV Keynote Making Markets More Effective With Competition Policy 1997
Brookings Trade Policy Forum Competition Policy and Antidumping: The Economic Effects 1997
University of Malaya and Harvard University Conference on The Impact of Globalisation and Privatisation on Malaysia and Asia in the Year 2020
Microeconomics, Privatization, and Vertical Integration 1997
ABA Section of Antitrust Law Conference on The Telecommunications Industry Current Economic Issues in Telecommunications 1997
Antitrust 1998: The Annual Briefing The Re-Emergence of Distribution Issues 1997
Inter-American Development Bank Conference on Private Investment, Infrastructure Reform and Governance in Latin America & the Caribbean
Economic Principles to Guide Post-Privatization Governance 1997
Public Version
A-23
Harvard Forum on Regulatory Reform and Privatization of Telecommunications in the Middle East
Privatization: Methods and Pricing Issues 1997
American Enterprise Institute for Public Policy Research Conference Discussion of Local Competition and Legal Culture 1997
Harvard Program on Global Reform and Privatization of Public Enterprises “Infrastructure Privatization and Regulation: Freight” 1997
World Bank Competition Policy Workshop “Competition Policy for Entrepreneurship and Growth” 1997
Eastern Economics Association Paul Samuelson Lecture “Bottleneck Access in Regulation and Competition Policy” 1997
ABA Annual Meeting, Section of Antitrust Law “Antitrust in the 21st Century: The Efficiencies Guidelines” 1997
Peruvian Ministry of Energy and Mines Conference on Regulation of Public Utilities “Regulation: Theoretical Context and Advantages vs. Disadvantages” 1997
The FCC: New Priorities and Future Directions “Competition in the Telecommunications Industry” 1997
American Enterprise Institute Studies in Telecommunications Deregulation “The Scope of Competition in Telecommunications” 1996
George Mason Law Review Symposium on Antitrust in the Information Revolution “Introduction to the Economic Theory of Antitrust and Information” 1996
Korean Telecommunications Public Lecture “Market Opening and Fair Competition” 1996
Korea Telecommunications Forum “Desirable Interconnection Policy in a Competitive Market” 1996
European Association for Research in Industrial Economics Annual Conference “Bottleneck Access: Regulation and Competition Policy” 1996
Harvard Program on Global Reform and Privatization of Public Enterprises “Railroad and Other Infrastructure Privatization” 1996
Public Version
A-24
FCC Forum on Antitrust and Economic Issues Involved with InterLATA Entry “The Scope of Telecommunications Competition” 1996
Citizens for a Sound Economy Policy Watch on Telecommunications Interconnection “The Economics of Interconnection” 1996
World Bank Seminar on Experiences with Corporatization “Strategic Directions of Privatization” 1996
FCC Economic Forum on the Economics of Interconnection Lessons from Other Industries 1996
ABA Annual Meeting, Section of Antitrust Law The Integration, Disintegration, and Reintegration of the Entertainment Industry 1996
Conference Board: 1996 Antitrust Conference How Economics Influences Antitrust and Vice Versa 1996
Antitrust 1996: A Special Briefing Joint Ventures and Strategic Alliances 1996
New York State Bar Association Section of Antitrust Law Winter Meeting Commentary on Horizontal Effects Issues 1996
FTC Hearings on the Changing Nature of Competition in a Global and Innovation-Driven Age Vertical Issues for Networks and Standards 1995
Wharton Seminar on Applied Microeconomics Access Policies with Imperfect Regulation 1995
Antitrust 1996, Washington D.C. Assessing Joint Ventures for Diminution of Competition 1995
ABA Annual Meeting, Section of Antitrust Law Refusals to Deal -- Economic Tests for Competitive Harm 1995
FTC Seminar on Antitrust Enforcement Analysis Diagnosing Collusion Possibilities 1995
Philadelphia Bar Education Center: Antitrust Fundamentals Antitrust--The Underlying Economics 1995
Vanderbilt University Conference on Financial Markets
Public Version
A-25
Why Do Christie and Schultz Infer Collusion From Their Data? 1995
ABA Section of Antitrust Law Chair=s Showcase Program Discussion of Telecommunications Competition Policy 1995
Conference Board: 1995 Antitrust Conference Analysis of Mergers and Joint Ventures 1995
ABA Conference on The New Antitrust: Policy of the '90s Antitrust on the Super Highways/Super Airways 1994
ITC Hearings on The Economic Effects of Outstanding Title VII Orders "The Economic Impacts of Antidumping Policies" 1994
OECD Working Conference on Trade and Competition Policy "Empirical Evidence on The Nature of Anti-dumping Actions" 1994
Antitrust 1995, Washington D.C. "Rigorous Antitrust Standards for Distribution Arrangements" 1994
ABA -- Georgetown Law Center: Post Chicago-Economics: New Theories - New Cases?
"Economic Foundations for Vertical Merger Guidelines" 1994
Conference Board: Antitrust Issues in Today's Economy "New Democrats, Old Agencies: Competition Law and Policy" 1994
Federal Reserve Board Distinguished Economist Series "Regulated Private Enterprise Versus Public Enterprise" 1994
Institut d'Etudes Politiques de Paris "Lectures on Competition Policy and Privatization" 1993
Canadian Bureau of Competition Policy Academic Seminar Series, Toronto. "Public Versus Regulated Private Enterprise" 1993
CEPS Symposium on The Clinton Administration: A Preliminary Report Card "Policy Towards Business" 1993
Columbia Institute for Tele-Information Conference on Competition in Network Industries, New York, NY
"Discussion of Deregulation of Networks: What Has Worked and What Hasn't"
World Bank Annual Conference on Development Economics1993
"Public Versus Regulated Private Enterprise" 1993
Public Version
A-26
Center for Public Utilities Conference on Current Issues Challenging the Regulatory Process "The Economics of Current Issues in Telecommunications Regulation" 1992 "The Role of Markets in Presently Regulated Industries" 1992
The Conference Board's Conference on Antitrust Issues in Today's Economy, New York, NY "Antitrust in the Global Economy" 1992 "Monopoly Issues for the '90s" 1993
Columbia University Seminar on Applied Economic Theory, New York, NY "Economic Rationales for the Scope of Privatization" 1992
Howrey & Simon Conference on Antitrust Developments, Washington, DC "Competitive Effects of Concern in the Merger Guidelines" 1992
Arnold & Porter Colloquium on Merger Enforcement, Washington, DC "The Economic Foundations of the Merger Guidelines" 1992
American Bar Association, Section on Antitrust Law Leadership Council Conference, Monterey, CA
"Applying the 1992 Merger Guidelines" 1992
OECD Competition Policy Meeting, Paris, France "The Economic Impacts of Antidumping Policy" 1992
Center for Public Choice Lecture Series, George Mason University Arlington, VA "The Economic Impacts of Antidumping Policy" 1992
Brookings Institution Microeconomics Panel, Washington, DC, "Discussion of the Evolution of Industry Structure" 1992
AT&T Conference on Antitrust Essentials "Antitrust Standards for Mergers and Joint Ventures" 1991
ABA Institute on The Cutting Edge of Antitrust: Market Power "Assessing and Proving Market Power: Barriers to Entry" 1991
Second Annual Workshop of the Competition Law and Policy Institute of New Zealand "Merger Analysis, Industrial Organization Theory, and Merger Guidelines" 1991 "Exclusive Dealing and the Fisher & Paykel Case" 1991
Special Seminar of the New Zealand Treasury "Strategic Behavior, Antitrust, and The Regulation of Natural Monopoly" 1991
Public Version
A-27
Public Seminar of the Australian Trade Practices Commission "Antitrust Issues of the 1990's" 1991
National Association of Attorneys General Antitrust Seminar "Antitrust Economics" 1991
District of Columbia Bar's 1991 Annual Convention "Administrative and Judicial Trends in Federal Antitrust Enforcement" 1991
ABA Spring Meeting "Antitrust Lessons From the Airline Industry" 1991
Conference on The Transition to a Market Economy - Institutional Aspects "Anti-Monopoly Policies and Institutions" 1991
Conference Board's Thirtieth Antitrust Conference "Antitrust Issues in Today's Economy" 1991
American Association for the Advancement of Science Annual Meeting "Methodologies for Economic Analysis of Mergers" 1991
General Seminar, Johns Hopkins University "Economic Rationales for the Scope of Privatization" 1991
Capitol Economics Speakers Series "Economics of Merger Guidelines" 1991
CRA Conference on Antitrust Issues in Regulated Industries "Enforcement Priorities and Economic Principles" 1990
Pepper Hamilton & Scheetz Anniversary Colloquium "New Developments in Antitrust Economics" 1990
PLI Program on Federal Antitrust Enforcement in the 90's "The Antitrust Agenda of the 90's" 1990
FTC Distinguished Speakers Seminar "The Evolving Merger Guidelines" 1990
The World Bank Speakers Series "The Role of Antitrust Policy in an Open Economy" 1990
Seminar of the Secretary of Commerce and Industrial Development of Mexico "Transitions to a Market Economy" 1990
Public Version
A-28
Southern Economics Association "Entry in Antitrust Analysis of Mergers" 1990 "Discussion of Strategic Investment and Timing of Entry" 1990
American Enterprise Institute Conference on Policy Approaches to the Deregulation of Network Industries
"Discussion of Network Problems and Solutions" 1990
American Enterprise Institute Conference on Innovation, Intellectual Property, and World Competition
"Law and Economics Framework for Analysis" 1990
Banco Nacional de Desenvolvimento Economico Social Lecture "Competition Policy: Harnessing Private Interests for the Public Interest" 1990
Western Economics Association Annual Meetings "New Directions in Antitrust from a New Administration" 1990 "New Directions in Merger Enforcement: The View from Washington" 1990
Woodrow Wilson School Alumni Colloquium "Microeconomic Policy Analysis and Antitrust--Washington 1990" 1990
Arnold & Porter Lecture Series "Advocating Competition" 1991 "Antitrust Enforcement" 1990
ABA Antitrust Section Convention "Recent Developments in Market Definition and Merger Analysis" 1990
Federal Bar Association "Joint Production Legislation: Competitive Necessity or Cartel Shield?" 1990
Pew Charitable Trusts Conference "Economics and National Security" 1990
ABA Antitrust Section Midwinter Council Meeting "Fine-tuning the Merger Guidelines" 1990 "The State of the Antitrust Division" 1991
International Telecommunications Society Conference "Discussion of the Impact of Telecommunications in the UK" 1989
The Economists of New Jersey Conference "Recent Perspectives on Regulation" 1989
Public Version
A-29
Conference on Current Issues Challenging the Regulatory Process "Innovative Pricing and Regulatory Reform" 1989 "Competitive Wheeling" 1989
Conference Board: Antitrust Issues in Today's Economy "Foreign Trade Issues and Antitrust" 1989
McKinsey & Co. Mini-MBA Conference "Economic Analysis of Pricing, Costing, and Strategic Business Behavior" 1989
1994
Olin Conference on Regulatory Mechanism Design "Revolutions in Regulatory Theory and Practice: Exploring The Gap" 1989
University of Dundee Conference on Industrial Organization and Strategic Behavior "Mergers in Differentiated Product Industries" 1988
Leif Johanson Lectures at the University of Oslo "Normative Issues in Industrial Organization" 1988
Mergers and Competitiveness: Spain Facing the EEC "Merger Policy" 1988 "R&D Joint Ventures" 1988
New Dimensions in Pricing Electricity "Competitive Pricing and Regulatory Reform" 1988
Program for Integrating Economics and National Security: Second Annual Colloquium "Arming Decisions Under Asymmetric Information" 1988
European Association for Research in Industrial Economics "U.S. Railroad Deregulation and the Public Interest" 1987 "Economic Rationales for the Scope of Privatization" 1989 "Discussion of Licensing of Innovations" 1990
Annenberg Conference on Rate of Return Regulation in the Presence of Rapid Technical Change "Discussion of Regulatory Mechanism Design in the Presence of Research, Innovation, and Spillover Effects" 1987
Special Brookings Papers Meeting "Discussion of Empirical Approaches to Strategic Behavior" 1987 "New Merger Guidelines" 1990
Deregulation or Regulation for Telecommunications in the 1990's "How Effective are State and Federal Regulations?" 1987
Public Version
A-30
Conference Board Roundtable on Antitrust "Research and Production Joint Ventures" 1990 "Intellectual Property and Antitrust" 1987
Current Issues in Telephone Regulation "Economic Approaches to Market Dominance: Applicability of Contestable Markets" 1987
Harvard Business School Forum on Telecommunications "Regulation of Information Services" 1987
The Fowler Challenge: Deregulation and Competition in The Local Telecommunications Market
"Why Reinvent the Wheel?" 1986
World Bank Seminar on Frontiers of Economics "What Every Economist Should Know About Contestable Markets" 1986
Bell Communications Research Conference on Regulation and Information "Fuzzy Regulatory Rules" 1986
Karl Eller Center Forum on Telecommunications "The Changing Economic Environment in Telecommunications: Technological Change and Deregulation" 1986
Railroad Accounting Principles Board Colloquium "Contestable Market Theory and ICC Regulation 1986
Canadian Embassy Conference on Current Issues in Canadian -- U.S. Trade and Investment "Regulatory Revolution in the Infrastructure Industries" 1985
Eagleton Institute Conference on Telecommunications in Transition "Industry in Transition: Economic and Public Policy Overview" 1985
Brown University Citicorp Lecture "Logic of Regulation and Deregulation" 1985
Columbia University Communications Research Forum "Long Distance Competition Policy" 1985
American Enterprise Institute Public Policy Week "The Political Economy of Regulatory Reform" 1984
MIT Communications Forum "Deregulation of AT&T Communications" 1984
Public Version
A-31
Bureau of Census Longitudinal Establishment Data File and Diversification Study Conference "Potential Uses of The File" 1984
Federal Bar Association Symposium on Joint Ventures "The Economics of Joint Venture Assessment" 1984
Hoover Institute Conference on Antitrust "Antitrust for High-Technology Industries" 1984
NSF Workshop on Predation and Industrial Targeting "Current Economic Analysis of Predatory Practices" 1983
The Institute for Study of Regulation Symposium: Pricing Electric, Gas, and Telecommunications Services Today and for the Future
"Contestability As A Guide for Regulation and Deregulation" 1984
University of Pennsylvania Economics Day Symposium "Contestability and Competition: Guides for Regulation and Deregulation" 1984
Pinhas Sapir Conference on Economic Policy in Theory and Practice "Corporate Governance and Market Structure" 1984
Centre of Planning and Economic Research of Greece "Issues About Industrial Deregulation" 1984 "Contestability: New Research Agenda" 1984
Hebrew and Tel Aviv Universities Conference on Public Economics "Social Welfare Dominance Extended and Applied to Excise Taxation" 1983
NBER Conference on Industrial Organization and International Trade "Perspectives on Horizontal Mergers in World Markets" 1983
Workshop on Local Access: Strategies for Public Policy "Market Structure and Government Intervention in Access Markets" 1982
NBER Conference on Strategic Behavior and International Trade "Industrial Strategy with Committed Firms: Discussion" 1982
Columbia University Graduate School of Business, Conference on Regulation and New Telecommunication Networks
"Local Pricing in a Competitive Environment" 1982
International Economic Association Roundtable Conference on New Developments in the Theory of Market Structure
Public Version
A-32
"Theory of Contestability" 1982 "Product Dev., Investment, and the Evolution of Market Structures" 1982
N.Y.U. Conference on Competition and World Markets: Law and Economics "Competition and Trade Policy--International Predation" 1982
CNRS-ISPE-NBER Conference on the Taxation of Capital "Welfare Effects of Investment Under Imperfect Competition" 1982
Internationales Institut fur Management und Verwalturg Regulation Conference "Welfare, Regulatory Boundaries, and the Sustainability of Oligopolies" 1981
NBER-Kellogg Graduate School of Management Conference on the Econometrics of Market Models with Imperfect Competition
"Discussion of Measurement of Monopoly Behavior: An Application to the Cigarette Industry" 1981
The Peterkin Lecture at Rice University "Deregulation: Ideology or Logic?" 1981
FTC Seminar on Antitrust Analysis "Viewpoints on Horizontal Mergers 1982 "Predation as a Tactical Inducement for Exit" 1980
NBER Conference on Industrial Organization and Public Policy "An Economic Definition of Predation" 1980
The Center for Advanced Studies in Managerial Economics Conference on The Economics of Telecommunication
"Pricing Local Service as an Input" 1980
Aspen Institute Conference on the Future of the Postal Service "Welfare Economics of Postal Pricing" 1979
Department of Justice Antitrust Seminar "The Industry Performance Gradient Index" 1979
Eastern Economic Association Convention "The Social Performance of Deregulated Markets for Telecom Services" 1979
Industry Workshop Association Convention "Customer Equity and Local Measured Service" 1979
Symposium on Ratemaking Problems of Regulated Industries "Pricing Decisions and the Regulatory Process" 1979
Public Version
A-33
Woodrow Wilson School Alumni Conference "The Push for Deregulation" 1979
NBER Conference on Industrial Organization "Intertemporal Sustainability" 1979
World Congress of the Econometric Society "Theoretical Industrial Organization" 1980
Institute of Public Utilities Conference on Current Issues in Public Utilities Regulation "Network Access Pricing" 1978
ALI-ABA Conference on the Economics of Antitrust "Predatoriness and Discriminatory Pricing" 1978
AEI Conference on Postal Service Issues "What Can Markets Control?" 1978
University of Virginia Conference on the Economics of Regulation "Public Interest Pricing" 1978
DRI Utility Conference "Marginal Cost Pricing in the Utility Industry: Impact and Analysis" 1978
International Meeting of the Institute of Management Sciences "The Envelope Theorem" 1977
University of Warwick Workshop on Oligopoly "Industry Performance Gradient Indexes" 1977
North American Econometric Society Convention"Intertemporal Sustainability" 1979"Social Welfare Dominance" 1978"Economies of Scope, DAIC, and Markets with Joint Production" 1977
Telecommunications Policy Research Conference "Transition to Competitive Markets" 1986"InterLATA Capacity Growth, Capped NTS Charges and LongDistance Competition" 1985"Market Power in The Telecommunications Industry" 1984"FCC Policy on Local Access Pricing" 1983"Do We Need a Regulatory Safety Net in Telecommunications?" 1982"Anticompetitive Vertical Conduct" 1981"Electronic Mail and Postal Pricing" 1980"Monopoly, Competition and Efficiency": Chairman 1979
Public Version
A-34
"A Common Carrier Research Agenda" 1978"Empirical Views of Ramsey Optimal Telephone Pricing" 1977"Recent Research on Regulated Market Structure" 1976"Some General Equilibrium Views of Optimal Pricing" 1975
National Bureau of Economic Research Conference on Theoretical Industrial Organization "Compensating Variation as a Measure of Welfare Change" 1976
Conference on Pricing in Regulated Industries: Theory & Application "Ramsey Optimal Pricing of Long Distance Telephone Services" 1977
NBER Conference on Public Regulation "Income Distributional Concerns in Regulatory Policy-Making" 1977
Allied Social Science Associations National Convention"Merger Guidelines and Economic Theory" 1990Discussion of "Competitive Rules for Joint Ventures" 1989"New Schools in Industrial Organization" 1988"Industry Economic Analysis in the Legal Arena" 1987"Transportation Deregulation" 1984Discussion of "Pricing and Costing of Telecommunications Services" 1983Discussion of "An Exact Welfare Measure" 1982"Optimal Deregulation of Telephone Services" 1982"Sector Differentiated Capital Taxes" 1981"Economies of Scope" 1980"Social Welfare Dominance" 1980"The Economic Definition of Predation" 1979Discussion of "Lifeline Rates, Succor or Snare?" 1979"Multiproduct Technology and Market Structure" 1978"The Economic Gradient Method" 1978"Methods for Public Interest Pricing" 1977Discussion of "The Welfare Implications of New Financial Instruments" 1976"Welfare Theory of Concentration Indices" 1976Discussion of "Developments in Monopolistic Competition Theory" 1976"Hedonic Price Adjustments" 1975"Public Good Attributes of Information and its Optimal Pricing" 1975"Risk Invariance and Ordinally Additive Utility Functions" 1974"Consumer's Surplus: A Rigorous Cookbook" 1974
University of Chicago Symposium on the Economics of Regulated Public Utilities "Optimal Prices for Public Purposes" 1976
American Society for Information Science "The Social Value of Information: An Economist's View" 1975
Public Version
A-35
Institute for Mathematical Studies in the Social Sciences Summer Seminar
Public Version
A-36
"The Sustainability of Natural Monopoly" 1975
U.S.-U.S.S.R. Symposium on Estimating Costs and Benefits of Information Services "The Evaluation of the Economic Benefits of Productive Information" 1975
NYU-Columbia Symposium on Regulated Industries "Ramsey Optimal Public Utility Pricing" 1975
Research Seminars:
Bell Communications Research (2) University of California, San Diego
Bell Laboratories (numerous) University of Chicago
Department of Justice (3) University of Delaware
Electric Power Research Institute University of Florida
Federal Reserve Board University of Illinois
Federal Trade Commission (4) University of Iowa (2)
Mathematica Universite Laval
Rand University of Maryland
World Bank (4) University of Michigan
Carleton University University of Minnesota
Carnegie-Mellon University University of Oslo
Columbia University (4) University of Pennsylvania (3)
Cornell University (2) University of Toronto
Georgetown University University of Virginia
Harvard University (2) University of Wisconsin
Hebrew University University of Wyoming
Johns Hopkins University (2) Vanderbilt University
M. I. T. (4) Yale University (2)
New York University (4) Princeton University (many)
Northwestern University (2) Rice University
Norwegian School of Economics and Stanford University (5)
Business Administration S.U.N.Y. Albany
Public Version
A-37
Expert Testimony of Robert Willig
September 23, 2015 to September 23, 2019
In re: Domestic Drywall Antitrust Litigation, In the United States District Court for the Eastern District of Pennsylvania, MDL No. 2437 13-MD-2437, Expert Report 03/13/15; Deposition 4/9/15, 4/10/15; Expert Report 01/26/18; Expert Report 02/26/18; Deposition 05/31/18.
Methodist Health Services Corporation v. OSF Healthcare System, In the United States District Court for the Central District of Illinois, Peoria Division, Case No.: 13-cv-1054, Expert Report 8/14/2015, Deposition 10/8/2015, Reply Report, 9/2016.
Application of the National Railroad Passenger Corporation Under 49 U.S.C. § 24308(a) – Canadian National Railway Company; Before the Surface Transportation Board, Docket No. FD 35743; Verified Statement, 9/4/2015; Rebuttal Verified Statement, 9/14/2017.
BRFHH Shreveport, LLC d/b/a University Health Shreveport and Vantage Health Plan, Inc. v. Willis-Knighton Medical Center, d/b/a Willis-Knighton Health System, In the United States District Court for the Western District of Louisiana, Shreveport Division, Case No.: 5:15-CV-02057, Joint Declaration of Margaret E. Guerin-Calvert and Robert D. Willig 9/8/2015, Expert Report 3/23/2017, Deposition 5/12/2017.
Australian Consumer and Competition Commission v. Informed Sources Pty Ltd & Ors, Before the Federal Court of Australia, Victoria Registry, File number VID450/2014, Expert Report 11/24/15.
Clark R. Huffman, Brandi K. Winters, Patricia L. Grantham, and Linda M. Pace, Individually and on behalf of all others similarly situated vs. The Prudential Insurance Company of America, In the United States District Court for the Eastern District of Pennsylvania, Civ. No. 2:10-cv-05135-EL, Expert Report 2/25/16, Deposition 4/5/16.
Maxon Hyundai Mazda, et al., vs. Carfax Inc., In the United States District Court for the Southern District of New York, Case No.: 13 CV 2680 (AJN) (RLE), Expert Report 2/26/16, Deposition 4/21/16.
Federal Trade Commission and Commonwealth of Pennsylvania vs. Penn State Hershey Medical Center and Pinnacle Health System, In the United States District Court for the Middle District of Pennsylvania, Civil Action No. 1:15-cv-02362, Expert Report 3/7/16, Deposition 3/25/16, Trial Testimony 4/15/16.
In the Matter of Business Data Services in an Internet Protocol Environment; Special Access for Price Cap Local Exchange Carriers; AT&T Corporation Petition for Rulemaking to Reform Regulation of Incumbent Local Exchange Carrier Rates for Interstate Special Access Services; Before the FCC; WC Docket No. 16-143; WC Docket No. 05-25; RM-10593; Declaration 8/8/16.
Public Version
A-38
United States et al. v. Anthem Inc. and Cigna Corp., In the United States District Court for the District of Columbia, Civil Action No. 16-cv-01493 (ABJ), Expert Report 10/7/16, Rebuttal and Supplemental Expert Report 10/28/16, Deposition 11/9/2016, Trial Testimony 12/2/16 and 1/3/17.
In the Matter of: Determination of Royalty Rates and Terms for Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Before the United States Copyright Royalty Judges, Washington, D.C., Docket No. 16-CRB-0001 SR/PSSR (2018-2022), Written Direct Testimony 10/19/16, Written Rebuttal Testimony 2/13/2017, Deposition 03/31/17, Trial Testimony 05/02/17 and 05/04/17.
Petition for Rulemaking to Adopt Revised Competitive Switching Rules, Before the Surface Transportation Board, Docket Number EP 711 (Sub-No. 1), Verified Statement 10/26/16.
In re: Evanston Northwestern Healthcare Corporation Antitrust Litigation, In the United States District Court Northern District of Illinois Eastern Division, Master File No. 07-CV-4446, Expert Report 10/26/16, Deposition 1/12/17.
Independent Power Producers of New York v. New York Independent System Operator, Before the Federal Energy Regulatory Commission, Docket No. EL13-62-002, Declaration 1/26/17.
Calpine Corporation et. al. v. PJM Interconnection, L.L.C., Before the Federal Energy Regulatory Commission, Docket No. EL16-49-000, Declaration 1/30/17.
In re: LIBOR-Based Financial Instruments Antitrust Litigation, Mayor and City Council of Baltimore, et. al. v. Credit Suisse Group AG, et. al., in the United States District Court Southern District of New York, MDL No. 2262, Master File No. 1:11-md-2262-NRB, 11-cv-5450 (NRB), Expert Report 4/3/17, Deposition 5/19/17.
In re: LIBOR-Based Financial Instruments Antitrust Litigation, Metzler Investment GmbH, et. al. v. Credit Suisse Group AG, et. al., in the United States District Court Southern District of New York, MDL 2262, 11 Civ. 2613, Master File No. 1:11-md-2262-NRB, 11-cv-5450 (NRB), Expert Report 4/3/17, Deposition 6/8/17.
In re: LIBOR-Based Financial Instruments Antitrust Litigation, Berkshire Bank, Government Development Bank v. Bank of America Corp., et. al., in the United States District Court Southern District of New York, MDL No. 2613, Master File No. 1:11-md-2262-NRB, 12-cv-5723-NRB, Expert Report 4/3/17, Deposition 4/21/17.
The Coca-Cola Company & Subs. v. Commissioner of Internal Revenue, in the United States Tax Court, Docket No. 31183-15, Expert Report 6/29/2017, Rebuttal Report 9/28/17, Deposition 12/19/17, Trial Testimony 4/5/18 and 4/6/18.
In re: Automotive parts antitrust litigation -- Bearings Cases; in the United States District Court for the Eastern District of Michigan Southern Division, Master File No. 12-md-02311; 2:12-cv-00501-MOB-MKM; Declaration in Support of Defendants’ Opposition to Direct Purchaser Plaintiffs’
Public Version
A-39
Motion for Class Certification and Appointment of Class Counsel, 7/26/2017, Deposition 10/13/2017; Declaration in Support of Defendants’ Reply in Support of Defendants’ Joint Motion to Exclude the Proposed Expert Reports and Testimony of Drs. McClave and Langenfeld, 12/14/17.
In re: General Motors LLC Ignition Switch Litigation, in the U.S. District Court for the Southern District of New York, No. 14-MD-2543 (JMF), Expert Report 2/23/2018, Expert Report 4/20/2018, Supplemental Tables, 5/7/2018, Deposition 5/8/2018 and 5/9/2018.
CPV Power Holdings et. al. v. PJM Interconnection, L.L.C., Before the Federal Energy Regulatory Commission, Docket No. EL18-169-000, Declaration 06/20/18.
In Re Thalomid and Revlimid Antitrust Litigation, in the U.S, District Court for the District of New Jersey, Civil No. 14-6997 (MCA) (MAH), Expert Report 08/27/18, Deposition 10/25/2018.
In Re Djeneba Sidibe et al, v. Sutter Health, Case No. 3: 12-cv-4854-LB, in the U.S. District Court for the Northern District of California, Expert Declaration 9/21/2018, Deposition 11/8/2018, Sur-Reply Declaration 1/7/2019, Expert Report 6/21/2019, Deposition 7/24/2019.
In Re UFCW & Employers Benefit Trust, et al v. Sutter Health, et al., Case No. CGC-14-538451, Consolidated with Case No. CGC-18-565398, People of the State of California, ex rel. et al v. Sutter Health, before the Superior Court of the State of California for the City and County of San Francisco, Expert Report 10/29/2018, Deposition 12/17-18/2018, Addendum to Expert Report 3/26/2019, Response to Supplemental Merits Expert Declaration of Jeffrey J. Leitzinger 4/16/2019, Response to Second Supplemental Expert Declaration of Jeffrey J. Leitzinger 6/10/2019.
In Re Christopher Dicesare et al v. The Charlotte-Mecklenburg Hospital Authority, d/b/a Carolinas Healthcare System, in the General Court of Justice State of North Carolina Superior Court Division County of Mecklenburg, Case No. 16-CVS-16404, Expert Report 12/7/2018, Deposition 3/8/2019.
In Re Epipen (Epinephrine Injection, USP) Marketing, sales practices and antitrust litigation; Sanofi-Aventis US, LLC v. Mylan Inc., et al.; before the United States District Court for the District of Kansas, Case No. 2:17-md-2785 and Case No. 2:17-cv-2452; expert report March 25, 2019; Deposition May 19, 2019.
Public Version
B-1
Appendix B: Materials Relied On
Academic Articles • Allan Collard-Wexler, Gautam Gowrisankaran, and Robin Lee, “‘Nash-in-Nash’
Bargaining: A Microfoundation for Applied Work,” Journal of Political Economy, Vol. 127, No. 1 (2019): 163-195
• Carl Shapiro, “Mergers with Differentiated Products,” Antitrust, No. 10 (1996): 23-30 • Gautam Gowrisankaran, Aviv Nevo, and Robert Town, “Mergers When Prices Are
Negotiated: Evidence from the Hospital Industry,” American Economic Review, Vol. 105, No. 1 (2015): 172-203
• Gregory S. Crawford, Robin S. Lee, Michael D. Whinston, and Ali Yurukoglu, “The Welfare Effects of Vertical Integration in Multichannel Television Markets,” Econometrica, Vol. 86, No. 3 (2018): 891-954
• Henrick Horn and Asher Wolinsky, “Bilateral monopolies and incentives for merger,” RAND Journal of Economics, Vol. 19, No. 3 (1988): 408-419
• Ian McDonald and Robert Solow, “Wage Bargaining and Employment,” American Economic Review, Vol. 71, No. 5 (1981): 896-908
• John Nash, “The Bargaining Problem,” Econometrica, Vol. 18, No. 2 (1950): 155-162
Textbooks • Andreu Mas-Colell, Michael D. Whiston, and Jerry R. Green, Microeconomic Theory,
Oxford University Press (1995)• Eyal Winter, “The Shapley Value,” In R.J. Aumann and S. Hart (Eds.), Handbook of
Game Theory, Vol. 3, Chapter 53, Elsevier Science (2002) • Hervé Moulin, Fair Division and Collective Welfare, MIT Press (2003)• Holger Ingmar Meinhardt, The Pre-Kernel as a Tractable Solution for Cooperative
Games, An Exercise in Algorithmic Game Theory, (Springer-Verlag: 2014) • Jean Tirole, The Theory of Industrial Organization, (MIT Press: 1988) • Lloyd S. Shapley, “A value for n-person games,” (1953) In Alvin E. Roth (Ed.), The
Shapley Value, Essays in honor of Lloyd S. Shapley, Cambridge University Press (1988)• Michael Carter, “Cooperative Games,” In Hal R. Varian (Ed.), Economic and Financial
Modeling with Mathematica, Springer-Verlag (1993) • Robert Willig, “Merger Analysis, Industrial Organization Theory, and Merger
Guidelines,” Brookings Papers: Microeconomics (1991)
CRB Documents • Determination, United States Copyright Royalty Judges, The Library of Congress, In re
Determination of Royalty Rates and Terms for Ephemeral Recording and Webcasting Digital Performance of Sound Recordings (Web IV), Docket No. 14-CRB-0001-WR
• Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms For Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Docket No. 16-CRB-0001 SR/PSSR
Public Version
B-2
• Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms for Making and Distributing Phonorecords (Phonorecords III), Docket No. 16-CRB-0003-PR
• Federal Register, Vol. 84, No. 16, January 24, 2019
Written Direct Testimony • Written Direct Testimony of Jason Gallien, September 23, 2019 • Written Direct Testimony of Aaron Harrison, September 23, 2019 • Written Direct Testimony of Mark Piibe, September 23, 2019 • Written Direct Testimony of Professor Gal Zauberman, September 23, 2019 • Written Direct Testimony of Robert Willig, October 19, 2016, In the Matter of
Determination of Royalty Rates and Terms for Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Docket No. 16-CRB-0001 SR/PSSR (2018-2022)
Data • “662009_Final_AllStarts_091712.xlsx” Survey Data Results from Professor Gal
Zauberman (SoundExchange Exhibit 29) • “bootstrap_q1_wide.csv” Survey Data Results from Professor Gal Zauberman
(SoundExchange Exhibit 30) • “Demographics.xlsx” Survey Data Results from Professor Gal Zauberman
(SoundExchange Exhibit 31) • Sirius XM SDARS December 2018.pdf (SoundExchange Exhibit 32) • Sirius XM SDARS April 2019.pdf (SoundExchange Exhibit 33) • RIAA US Wholesale revenue summary 2012 – 2018 CONFIDENTIAL.xlsx
(SoundExchange Exhibit 34) • 2018 RIAA shipments data update.xlsx (SoundExchange Exhibit 35) • 2019-8-6 SOA and Allocations 2016 to 2019 for Web V.xlsm (SoundExchange Exhibit
36) • 2019-08-08 Performances iHeart Pandora Rhapsody.xlsx (SoundExchange Exhibit 37) • Pandora Market Capitalization from Bloomberg (SoundExchange Exhibit 38)
Royalty Statements • Amazon Music Unlimited-UMG Royalty Statements Covering the Period April 2018 to
March 2019 • Amazon Music Unlimited-SME Royalty Statements Covering the Period April 2018 to
March 2019 • Amazon Music Unlimited-WMG Royalty Statements Covering the Period April 2018 to
March 2019 • Amazon Prime-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Amazon Prime-SME Royalty Statements Covering the Period April 2018 to March 2019 • Amazon Prime-WMG Royalty Statements Covering the Period October 2018 to March
2019 • Apple Music-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Apple Music-SME Royalty Statements Covering the Period April 2018 to March 2019
Public Version
B-3
• Apple Music-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Google-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Google-SME Royalty Statements Covering the Period April 2018 to March 2019 • Google-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Google-Merlin Royalty Statements Covering the Period April 2018 to March 2019 • iHeart Free-WMG Royalty Statements Covering the Period April 2018 to March 2019 • iHeart Subscription-UMG Royalty Statements Covering the Period April 2018 to March
2019 • iHeart Subscription-SME Royalty Statements Covering the Period April 2018 to March
2019 • iHeart Subscription-WMG Royalty Statements Covering the Period April 2018 to March
2019 • iHeart Subscription-Merlin Royalty Statements Covering the Period April 2018 to March
2019 • Pandora-UMG Royalty Statements Covering the Period January 2018 to March 2019 • Pandora Premium-SME Royalty Statements Covering the Period April 2018 to March
2019 • Pandora Radio Plus-SME Royalty Statements Covering the Period January 2018 to
March 2019 • Pandora Radio-SME Royalty Statements Covering the Period January 2018 to March
2019 • Pandora Premium-WMG Royalty Statements Covering the Period April 2018 to March
2019 • Pandora Radio Plus-WMG Royalty Statements Covering the Period January 2018 to
March 2019 • Pandora Radio-WMG Royalty Statements Covering the Period January 2018 to March
2019 • Pandora-Merlin Royalty Statements Covering the Period January 2018 to March 2019 • Napster (Rhapsody)-UMG Royalty Statements Covering the Period April 2018 to March
2019 • Napster (Rhapsody)-SME Royalty Statements Covering the Period April 2018 to March
2019 • Napster (Rhapsody)-WMG Royalty Statements Covering the Period April 2018 to March
2019 • Spotify-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Spotify-SME Royalty Statements Covering the Period April 2018 to March 2019 • Spotify-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Spotify-Merlin Royalty Statements Covering the Period April 2018 to March 2019 • Vevo-UMG Royalty Statements Covering the Period April 2018 to March 2019 • Vevo-SME Royalty Statements Covering the Period April 2018 to March 2019 (except
May 2018) • Vevo-WMG Royalty Statements Covering the Period April 2018 to March 2019 • Vevo-Merlin Royalty Statements Covering the Period April 2018 to March 2019
Public Version
B-4
• YouTube Streams-UMG Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Streams-SME Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Streams-WMG Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Subscription-UMG Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Subscription-SME Royalty Statements Covering the Period April 2018 to March 2019
• YouTube Subscription-WMG Royalty Statements Covering the Period April 2018 to March 2019
License Agreements • [
] (SoundExchange Exhibit 39) • [
(SoundExchange Exhibit 40) • [ ]
(SoundExchange Exhibit 41)
Industry Reports • MusicWatch, “Annual Music Study 2018: Report to RIAA,” April 2019 (SoundExchange
Exhibit 42) • Edison Research, “Share of Ear,” Q3 2018 (SoundExchange Exhibit 43) • Edison Research, “Share of Ear,” Q2 2019 (SoundExchange Exhibit 44) • Q2 2019 Share of Ear Survey Methodology Statement (SoundExchange Exhibit 72) • JMP Securities, “SiriusXM Announces All-Stock Agreement to Acquire Pandora,”
September 24, 2018 (SoundExchange Exhibit 45)
SEC Filings • Pandora Media, Inc. Form 10-K for the Fiscal Year Ended December 31, 2017
(SoundExchange Exhibit 46) • Pandora Media, Inc. Form 8-K dated September 23, 2018 (SoundExchange Exhibit 47)• Pandora Media, Inc. Schedule 14A Definitive Proxy Statement, December 20, 2018
(SoundExchange Exhibit 48) • Sirius XM Holdings Inc., Form 10-K for the Fiscal Year Ended December 31, 2018
(SoundExchange Exhibit 49) • Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended March 31, 2018
(SoundExchange Exhibit 50) • Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended June 30, 2018
(SoundExchange Exhibit 51) • Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended September 30, 2018
(SoundExchange Exhibit 52)
Public Version
B-5
• Sirius XM Holdings Inc., Form 10-Q for the Quarterly Period Ended March 31, 2019 (SoundExchange Exhibit 53)
• Spotify Technology S.A., Form 20-F for the Fiscal Year Ended December 31, 2018 (SoundExchange Exhibit 71)
Bates-Stamped Documents • GOOG-WEBV-00036425 (SoundExchange Exhibit 54) • GOOG-WEBV-00036688 (SoundExchange Exhibit 55) • GOOG-WEBV-00036873 (SoundExchange Exhibit 56) • PANWEBV_00002985 (SoundExchange Exhibit 57) • PANWEBV_00003357 (SoundExchange Exhibit 58) • PANWEBV_00003425 (SoundExchange Exhibit 59) • PANWEBV_00003868 (SoundExchange Exhibit 60) • PANWEBV_00004139 (SoundExchange Exhibit 61) • PANWEBV_00004249 (SoundExchange Exhibit 62) • PANWEBV_00004335 (SoundExchange Exhibit 63) • SXMWEBV_00004542 (SoundExchange Exhibit 64) • SXMWEBV_00005082 (SoundExchange Exhibit 65)
Other • 17 U.S.C § 114(f)(2)(B) • Transcript of Spotify Technology SA at Goldman Sachs Communacopia Conference,
September 14, 2018 (SoundExchange Exhibit 66) • “AAPOR Guidance on Reporting Precision for Nonprobability Samples,” American
Association for Public Opinion Research, available at: https://www.aapor.org/getattachment/Education-Resources/For-Researchers/AAPOR_Guidance_Nonprob_Precision_042216.pdf.aspx (SoundExchange Exhibit 67)
• Merlin: http://www.merlinnetwork.org/whowelicense (SoundExchange Exhibit 68) • Federal Reserve Bank of St. Louis data: https://fred.stlouisfed.org/series/JCXFECTM
(SoundExchange Exhibit 22)
Public Version
C-1
Appendix C: Shapley Value Model
I. Model
1. This appendix provides a detailed description of my Shapley Value model. This
model involves two noninteractive streaming distributors (a free ad-supported distributor and a
paid subscription distributor), three “must-have” record companies (representing the majors) and
one “not-must-have” record company (representing all the indies). In constructing this Shapley
Value model, I utilize the following notation:
Figure C-1: Notation
D1 and D2The ad-supported and subscription noninteractive streaming distributors, respectively.
LA, LB, and LC The three must-have major record companies.
LD
An amalgam of not-must-have indie record companies grouped together without loss of generality. Referred to hereafter as the not-must-have record company.
SharekRecord company k’s share of plays, which is assumed to be the same across all distributors.
CkRecord company k’s costs of producing sound recordings.
��� and ��
�D1’s marginal profit per play and D2’s marginal profit per play, respectively, before accounting for any sound recording royalties that they pay.
F1 and F2 D1’s and D2’s fixed costs, respectively.
Dj
where � ∈ {3, …, N}
Distributors whose royalty rates are set outside of the current proceedings, such as ad-supported and subscription interactive music and video streaming services, Sirius XM satellite radio, and music purchases (i.e., CDs, vinyl, digital downloads).
���
The royalty rate paid by outside distributor j per royalty-bearing unit � ∈ {play, subscriber, CD/vinyl-record/download}.
���
Di’s audience, measured in Di’s royalty bearing units t (e.g., plays for D1), when all other distributors are operating.
���The average number of monthly plays per user of D1’s service.
���The average number of monthly plays per user of D2’s service.
��,��
A diversion ratio from service i to j: Per current user of noninteractive streaming distributor i’s service, the change in royalty-bearing units t on distributor j’s service if Di were to stop operating while all other distributors continued operating.
Public Version
C-2
� = � ���� ∗ ��
���
���
The four record companies’ combined royalties from outside distributors when all distributors are operating.
�� = � ���� ∗ ��,�
� ��
���
Per current user of noninteractive streaming distributor i’s service, the change in the four record companies’ combined royalties from outside distributors if Di were to stop operating while all other distributors continued operating.
�′�,��
Another diversion ratio from service i to j: The fraction of current users of noninteractive distributor i’s service who would divert to outside distributor Dj’s service if service i were to stop operating, while the other noninteractive distributor were also not operating. This diversion is measured as the change in royalty-bearing units t on service jper then-current user of service i.
��� = ����
� ∗ �′�,�� �
�
���
Per then-current user of noninteractive distributor i’s service, the change in the four record companies’ combined royalties from outside distributors if noninteractive distributor Di were to stop operating while the other non-interactive distributor were also not operating.
2. The Shapley Value solution concept requires calculating the average incremental
contribution to the collective surplus that each record company or distributor creates, based on all
possible orderings of the parties. The first step in this process is to calculate the total surplus each
hypothetical coalition of noninteractive streaming distributors and record companies would
generate if they acted in a manner that maximized their collective surplus. These values define a
function �[�] of all those coalitions � ∈ �(�), called the “characteristic function,” where �(�) is
the set of subsets of the players � = {��,��, ��, ��, �� , ��}.
3. In the present case, the characteristic function can be determined recursively. An
empty coalition and coalitions comprised of only the distributors create zero payoffs. Hence:
�[{}] = 0;
�[{��}] = 0;
�[{��}] = 0;
�[{��,��}] = 0.
4. Each record company, when it is alone, generates its current profits from outside
Public Version
C-3
distributors plus the additional profits from outside distributors caused by diversion from
distributors 1 and 2 to the outside distributors. Hence:1
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗���
���+ ��
�∗���
���) − ��.
5. Since LA, LB, and LC are must-haves, only when all three are present in a coalition
can the distributor(s) begin making profits. When only these three record companies are present
in a coalition, three cases can arise. First, if only D1 is in a coalition, and if it increases the
coalition’s collective surplus, diversion occurs from D2 to both D1 and the outside distributors
(because D2 is not operating) and from D1 to outside distributors (because D1 does not have a
license to LD’s content). Second, if only D2 is in a coalition, and if it increases the coalition’s
collective surplus, diversion occurs from D1 to both D2 and the outside distributors (because D1 is
not operating) and from D2 to outside distributors (because D2 does not have a license to LD’s
content). And third, if both distributors are in a coalition, and if they increase the coalition’s
collective surplus, no diversion between D1 and D2 occurs, but there is still diversion from D1 and
D2 to the outside distributors (because they do not have licenses to LD’s content).2
1 I assume that it is profitable for the must-have record companies to operate. 2 In keeping with the Judges’ conclusion in the Web IV proceeding that “the repertoire of each of the three Majors is a ‘must have’ in order for a noninteractive service to be viable,” I assume that a distributor could not operate without agreements with all three of the major record companies. See Web IV Determination at 1334. Conversely, I assume that if a distributor has access to the must-have record companies’ content but does not have access to LD’s content, the distributor loses �ℎ���� of its plays (or subscribers).
Public Version
C-4
6. Each coalition is assumed to act in the manner that maximizes the collective surplus
of the coalition.3 Therefore:
�[{��,��,��,��}]
= ��� �(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗��,��
����+ � + ��
�∗������− (�� + �� + ��)
− ��,�[{��,�� ,��}]� ; 4
�[{��,��,�� ,��}]
= ��� �(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗��,��
���� + � + ��
�∗������− (�� + �� + ��)
− ��,�[{��,��,��}]� ;
�[{��,��,��,��,��}]
= ����(1 − �ℎ����) ∗ ����∗ ��
�+ ��
�∗ ��
�+ ��− (�� + �� + ��)− ��
− ��,�[{��,��,��,��}],�[{��,��,�� ,��}]�.
7. When LD joins a coalition, it adds to the coalition’s surplus by contributing its
royalties from outside distributors, and also by licensing one or both distributors in the coalition
(if such a license would increase the coalition’s collective surplus). Thus:
�[{��,��,��,�� ,��}]
= ��� ��� + ���∗�����
+ ���∗ ���
�+ ��
�∗��,��
�����− (�� + �� + �� + ��)
− ��,�[{��,��,�� ,��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
3 In some cases, the collective surplus of a coalition may be higher if one or both distributors do not operate.4 As explained below, �[{��,�� ,��}] is the combined profits of the coalition that includes only the three must-have record companies. �[{��,��,��}] = �[{��}] + �[{��}] + �[{��}].
Public Version
C-5
�[{��,��,�� ,�� ,��}]
= ��� ��� + ���∗�����
+ ���∗ ���
�+ ��
�∗��,��
�����− (�� + �� + �� + ��)
− ��,�[{��,��,��, ��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
�[{��,��,��,��,�� ,��}]
= ��� ��� + ���∗ ��
�+ ��
�∗ ��
��− (�� + �� + �� + ��)− ��
− ��,�[{��,��,��,�� ,��}],�[{��,��,�� ,�� ,��}],�[{��,��,��,��,��}]
+ �[{��}],� + ���∗ ���
�+ �ℎ���� ∗ ��
�∗��,��
����− �� − (�� + �� + �� + ��)
+ (1 − �ℎ����) ∗���∗ ��
�+ �ℎ���� ∗ ���
�∗������− ��,� +��
�
∗ ����
+ �ℎ���� ∗ ���∗��,��
����− �� − (�� + �� + �� + ��) + (1 − �ℎ����) ∗��
�
∗ ���
+ �ℎ���� ����∗������− ���.
8. In any other case not defined above, the noninteractive distributors do not carry one
or more of the must-have record companies and cannot operate. In such cases, all the profits
generated by those coalitions are just the sum of the �[{��}] defined above. For example,
�[{��,��}] = �[{��}] + �[{��}],�[{��,��,��,��}] = �[{��}] + �[{��}], etc.
9. Based on this characteristic function, the Shapley Value algorithm can be
implemented. Specifically, I evaluate every possible arrival ordering and for each determine the
incremental value contributed by each noninteractive distributor relative to the value created by
the parties that preceded its arrival. Then I average all of these marginal contributions across all
arrival orderings. With six players, there are 6! (i.e., 720) arrival orderings. Algebraically, this
can be expressed as follows. For every possible subset � of � that includes i,
��� = ���(�[�] − �[� − {�}])
�⊂�
,
where
Public Version
C-6
�� =(|�| − 1)! ∗ (� − |�|)!
�!,
for n equal to the number of elements in � (six in this case) and |�| the number of elements in
subset �.5
10. Following this formula, I calculate the Shapley Values for both noninteractive
distributors, based on the parameters estimated as described in Appendices D and E. The total
royalty payments to be made by D1 and D2 can be calculated using these Shapley Values as
follows:
�� = ��� ∗ ��
� − �1 − ���;
�� = ��� ∗ ��
� − �2 − ���.
In other words, total royalties are equal to the pre-royalty profits of each of the noninteractive
distributors less their respective Shapley Values. In this way, the distributor will be left with its
Shapley Value after it has collected its profits from operations and paid out the royalties to the
record companies.
5 See Michael Carter, “Cooperative Games,” in Hal R. Varian (Ed.), Economic and Financial Modeling with Mathematica, Springer-Verlag (1993), pp. 185-186.
Public Version
D-1
Appendix D: Data Inputs
1. This appendix provides further detail on the data inputs used in my analysis of
Shapley Values. The construction of these data inputs is discussed in the body of the report. I
elaborate here on one specific data input—noninteractive distributor fixed costs and marginal
profit rates. In addition, backup computations for data inputs discussed in the report appear as
exhibits to this appendix.
2. As discussed in the body of the report, I estimate noninteractive distributor fixed
costs and marginal profit rates using Pandora financials as a proxy. Because my analysis models
the hypothetical negotiation over royalty rates for the 2021-2025 period, this is the relevant time
period for analyzing Pandora financials. Moreover, estimating marginal profit rates from
Pandora’s recent historical financials would result in figures with an inappropriate downward bias
with respect to the hypothetical negotiation. Specifically, Pandora’s recent financial results
demonstrate that the company has incurred substantial operating losses. [
]1 However, the approximately $3.5 billion purchase price paid by SiriusXM
to acquire the company, as well as Pandora’s substantial market capitalization of approximately
$2.4 billion immediately prior to the announcement of the SiriusXM acquisition demonstrate that
Pandora is anticipated to be profitable in the future.2 Thus, Pandora’s recent history of operating
1 See SoundExchange Exhibit 57. Operating losses for 2016 and 2017 totaled $318.8 million and $492.9 million, respectively. SoundExchange Exhibit 46 (Pandora 2017 Form 10-K) at 66. Pandora did not file a Form 10-K for the fiscal year ended December 31, 2018 due to its merger with SiriusXM. 2 SoundExchange Exhibit 38; SoundExchange Exhibit 47 (Pandora Form 8-K, September 23, 2018) at Ex. 99.1.
Public Version
D-2
losses does not accurately reflect expectations about the incremental value Pandora is anticipated
to bring to the hypothetical negotiation concerning royalty rates for the 2021-2025 period.
3. For these reasons, I use the forward-looking estimates of Pandora’s costs and profits
for the relevant 2021-2025 period disclosed in the company’s recent Merger Proxy. These
projections were disclosed to Pandora shareholders prior to their vote to approve SiriusXM’s
acquisition of the company and were utilized by Pandora’s investment bankers as an input into
their fairness opinions.3 Pandora disclosed two sets of projections in its Proxy Statement, a lower
“Scenario 1a” projection and a higher “Scenario 2” projection.4 I utilize the Scenario 2 projections
in my analysis because Pandora’s investment bankers prepared discounted cash flow valuation
analyses using these Scenario 2 projections, which produced valuations in-line with the $3.5 billion
market price paid by SiriusXM to acquire the company. By contrast, the Scenario 1a projections
produced valuations substantially below this $3.5 billion market price.5
4. As shown in Exhibit D.6 at the end of this appendix, using these projections I
estimate annual fixed costs for Pandora Free and Pandora Plus of $397 million and $85 million,
3 SoundExchange Exhibit 48 (Pandora Definitive Proxy Statement, December 20, 2018) at 55-60. 4 I understand Pandora also recently produced additional projections that it prepared for these proceedings
See SoundExchange Exhibit 57. I do not use these projections in my analysis as I view the projections Pandora prepared for purposes of evaluating its recent $3.5 billion merger with SiriusXM that were fully disclosed to shareholders in the company’s SEC filings in late-2018 as more reliable than those prepared for purposes of these proceedings. 5 Specifically, when using the Scenario 1 forecasts, Centerview produced a DCF valuation range of $4.50 to $7.00 per Pandora share and LionTree produced a DCF valuation range of $5.53 to $7.81 per share. When using the Scenario 2 forecasts, Centerview and LionTree produced DCF valuation ranges of $8.50 to $11.75 and $9.41 to $12.22, respectively. The per share consideration implied by the $3.5 billion acquisition price was $10.05 per share, which is in-line with the DCF values prepared using the Scenario 2 forecasts. See SoundExchange Exhibit 48 (Pandora Definitive Proxy Statement, December 20, 2018) at 65-66, 73-74.
Public Version
D-3
respectively. Expressing these amounts on an average monthly basis, and then grossing them up
so that they are reflective of the total for all noninteractive distributors, results in monthly fixed
costs of $40.4 million and $8.9 million for all ad-supported noninteractive and subscription
noninteractive distributors, respectively.6 Exhibit D.6 also shows how I estimate a marginal ad-
supported noninteractive profit rate of $0.0042 per play and a subscription noninteractive marginal
profit rate of $0.0048 per play.7
6 I gross up these fixed cost amounts using the same approach used to gross up the audience size figures, which is discussed in the body of the report. 7 The Pandora projections on which these estimates are based do not disclose certain key inputs that were used to create the projections. For instance, the projections do not include a breakdown of subscription revenue into the portions related to its Pandora Plus noninteractive and Pandora Premium on-demand services, respectively, and therefore require an allocation assumption to exclude Pandora Premium revenue and costs from the analysis. Moreover, the projections do not include the projected subscriber counts, active user counts, and play counts underlying the projections, requiring these figures to be derived so that profit rates can be computed. Accordingly, the assumptions required to estimate key parameters for use in my Shapley Value model may need to be updated following the completion of discovery.
Public Version
Exhibit D.1 (RESTRICTED)Subscription On-Demand Music and Video Services: Weighted Average Royalty per Subscriber
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.2 (RESTRICTED)Ad-Supported On-Demand Music and Video Services: Weighted Average Royalty per Play
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.3 (RESTRICTED)CD/Vinyl/Digital Download Royalties per Purchaser
2018
Public Version
Exhibit D.4 (RESTRICTED)Noninteractive Audience and Fixed Cost Adjustment Factors
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.5 (RESTRICTED)Record Company Play Shares
Based on April 2018 - March 2019 Monthly Royalty Statements
Public Version
Exhibit D.6 (RESTRICTED)Pandora Media, Inc. Scenario 2 Management Projections from Merger Proxy
With Allocations to Free, Plus, and Premium Services(Millions of Dollars, Except Per Play and Per Subscriber Amounts)
Page 1 of 3
Public Version
Exhibit D.6 (RESTRICTED)Pandora Media, Inc. Scenario 2 Management Projections from Merger Proxy
With Allocations to Free, Plus, and Premium Services(Millions of Dollars, Except Per Play and Per Subscriber Amounts)
Page 2 of 3
Public Version
Exhibit D.6 (RESTRICTED)Pandora Media, Inc. Scenario 2 Management Projections from Merger Proxy
With Allocations to Free, Plus, and Premium Services(Millions of Dollars, Except Per Play and Per Subscriber Amounts)
Page 3 of 3
Public Version
E-1
Appendix E: Use of Zauberman Survey Results to Generate Model Parameters
1. This appendix describes how I used the results of the survey conducted by Professor
Zauberman to generate parameters employed in my analysis of Shapley Values, as well as the
computation of confidence intervals. I begin with a high-level description of the survey structure,1
then describe my processing of the survey responses to generate inputs for my analysis of Shapley
Values, and finally describe how I use these survey responses to compute inputs for my calculation
of confidence intervals around my royalty rate results.
I. Overview of Zauberman Survey Structure
2. The Zauberman Survey begins by asking respondents to identify the ways that they
currently listen to music. Specifically, it asks them whether or not they have listened to music
through each of the following modes of music distribution in the last 30 days:
• Subscription interactive (“on-demand”) services, such as Spotify Premium or
Apple Music (“Paid-OD”);
• Ad-supported on-demand services, such as free Spotify or free YouTube (“Free-
OD”);
• Subscription noninteractive services, such as Pandora Plus (“Paid-
Noninteractive”);
• Ad-supported noninteractive services, such as Pandora Free (“Free-
Noninteractive”);
1 This is intended to be an overview that provides the context for my use of the survey, not a complete and precise description of all elements of the survey. For such details, see Zauberman WDT.
Public Version
E-2
• SiriusXM satellite radio over-the-air (“SXM-Air”)2;
• AM/FM (terrestrial) radio over-the-air (“AM/FM-Air”)3;
• Physical and digital music, such as CDs, vinyl records, and mp3 files, that they own
(“CD/V/MP3”).
3. The respondents who indicate that they “currently listen” (i.e., have listened in the
last 30 days) to a noninteractive service are then assigned to one of two hypotheticals:4
• NoFree: this hypothetical postulates that all Free-Noninteractive services are no
longer available;
• NoPaid: this hypothetical postulates that all Paid-Noninteractive services are no
longer available.
4. Respondents who are assigned to the NoFree hypothetical are then asked what they
would do instead of listening to their Free-Noninteractive service (which is no longer available).
This questioning is done in two steps. First, the respondents are asked to identify the categories
of music listening they would use instead of their Free-Noninteractive service—the choices of
categories are the ones listed above (Paid-OD, Free-OD, etc.), except that Free-Noninteractive is
removed and “Do something else” (i.e., do not listen to music and, hence, listen to less music
2 This category only includes listening to SiriusXM through a satellite radio receiver (e.g., in a car); streaming of SiriusXM radio through the internet is included in the Paid-Noninteractive category. 3 This category only includes listening to AM/FM radio through a traditional receiver (e.g., in a car); streaming of AM/FM stations through the internet is included in the Free-Noninteractive category. 4 Because some people currently listen to both Paid-Noninteractive and Free-Noninteractive services, a specific approach was developed for assigning respondents to the hypotheticals to ensure that both samples are representative (i.e., include the appropriate mix of people with just one versus both types of noninteractive services). See further discussion in Zauberman WDT.
Public Version
E-3
overall) is added. Second, the respondents are asked to specify the percentage of their
Free-Noninteractive listening time that would be allocated to each of the alternatives chosen in the
first step.5,6 The survey requires that these percentages sum up to 100.
5. The respondents assigned to the NoPaid hypothetical are given similar questions,
except that Free-Noninteractive is one of the available substitutes and Paid-Noninteractive is not.7
II. Processing of Zauberman Survey Results – Overview
6. In this section, I describe how I processed the Zauberman Survey results to generate
inputs for my modeling. Professor Zauberman provided an Excel spreadsheet that contained the
“raw” survey responses, along with the corresponding “datamap” that defines how the responses
are coded.8 There are a total of 989 qualified responses for the NoFree hypothetical and 563
qualified responses for the NoPaid hypothetical. From these qualified responses, I computed the
5 The framing of the two steps is somewhat different. The first step is framed as a general question about services that would be substituted for Free-Noninteractive listening, without reference to a specific time period. The second step is framed as a specific question about services that would be substituted for Free-Noninteractive listening on the same day next week.6 This percentage question is not asked in three circumstances: (1) if the respondent specifies only one option in response to the switching question, in which case, 100 percent of the time is assigned to the option chosen; (2) if the respondent indicates that he/she would not switch to Free-OD services; or (3) if the respondent indicates that he/she would not listen to the Free-Noninteractive service on the same day next week (or is unsure about such listening). In the latter case, zero switching time is assigned to all categories, which is conservative with respect to calculating opportunity costs.7 In the NoPaid hypothetical, the percentage allocation question is not asked in three circumstances: (1) if the respondent specifies only one option in response to the switching question, in which case, 100 percent of the time is assigned to the option chosen; (2) if the respondent indicates that he/she would not switch to either Free-OD or Free-Noninteractive services; or (3) if the respondent indicates that he/she would not listen to the Paid-Noninteractive service on the same day next week (or is unsure about such listening), in which case zero switching time is again assigned to all categories.8 See SoundExchange Exhibit 29.
Public Version
E-4
utilized data parameters.
7. I computed the following from the survey responses: (a) the opportunity cost per
respondent (denominated in dollars per respondent per month) associated with diversion from the
noninteractive services to other forms of music distribution (i.e., O1 and O2); and (b) diversion
from one type of noninteractive service to the other (Free-to-Paid in the NoFree hypothetical and
Paid-to-Free in the NoPaid hypothetical).
8. The pertinent details of the diversion depend on the structure of royalty payments
for the various alternative modes of music distribution, as described in more detail below. For
example, if royalty payments in the alternative mode of distribution are based on subscriptions,
then I derive results from the survey for incremental subscriptions added. On the other hand, if
the royalty payments in the alternative mode of distribution are based on listening time (plays)9,
then I derive results from the survey for incremental plays. These diversion results from the survey
are then combined with the royalty rates paid by outside distributors (discussed in the body of the
report) to determine the opportunity cost associated with forms of music distribution other than
noninteractive streaming.
9. The overall results for both the NoFree and NoPaid hypotheticals are shown in
Figure 6 in the body of the report. The remainder of this appendix describes the calculations used
to generate these results. I begin with the NoFree hypothetical and follow with the NoPaid
hypothetical. The calculations for the latter are largely the same as those for the former.
III. Processing of Zauberman Survey Results – NoFree Hypothetical
10. For the NoFree hypothetical, there are four components of the opportunity cost
associated with diversion from Free-Noninteractive to other modes of music distribution
9 In the absence of specific data, I assume that plays are proportional to listening time.
Public Version
E-5
(excluding Paid-Noninteractive).
11. The first is Paid-OD. Diversion to Paid-OD listening generates incremental
royalties for the record companies if the respondent would acquire a new Paid-OD subscription.
Additional use of an existing Paid-OD subscription does not generate additional royalties, [
].10 Acquisitions of new
Paid-OD subscriptions are identified in the survey results as respondents who both: (i) do not
currently listen to a Paid-OD service, and (ii) would listen to a Paid-OD service instead of the
Free-Noninteractive service in the hypothetical. In this case, the amount of time that they would
listen (specified in the percentage allocation question) is not relevant. The survey results show
that 9.1 percent of the respondents would add a new Paid-OD subscription. As shown in Figure 4
of the body of the report, the average royalty per Paid-OD subscription is [ ] per month, so
the opportunity cost for this component is then 9.1 percent multiplied by [ ], or [ ] per
Free-Noninteractive user per month.
12. The second component of opportunity cost is SXM-Air. As with Paid-OD,
royalties are typically generated when new SXM-Air subscriptions are added, but not when
listening to an existing subscription is increased.11 As such, SXM-Air subscription additions are
10 Specifically, [
]
11 In SDARS III, the Judges set sound recording royalty rates for SiriusXM satellite radio broadcasts at 15.5 percent of gross revenue, including both subscription and advertising revenue. Final Determination, United States Copyright Royalty Judges, The Library of Congress, In re Determination of Royalty Rates and Terms For Transmission of Sound Recordings by Satellite Radio and “Preexisting” Subscription Services (SDARS III), Docket No. 16-CRB-0001 SR/PSSR, p. 2. However, according to SiriusXM’s most recent annual report, its “Subscriber revenue” was
Public Version
E-6
also identified as respondents who both: (i) do not currently listen to SXM-Air, and (ii) would
listen to SXM-Air instead of Paid-Noninteractive in the hypothetical. The survey results show
that 5.5 percent of the respondents would add a new SXM-Air subscription. The opportunity cost
for this component is this percentage with additional subscriptions multiplied by the average
royalty rate per SXM-Air subscription of [ ] per month reflected in Figure 4 of the body of
the report, which is [ ] per Free-Noninteractive user per month.
13. The third component of opportunity cost is CD/V/MP3. For CD/V/MP3, royalties
are generated when additional recordings are purchased; the amount of listening time does not
affect royalties in this category. For purposes of my analysis, I estimate incremental royalties from
diversion to CD/V/MP3 in the same way as for Paid-OD and SXM-Air. That is, a respondent is
considered to “add” CD/V/MP3 spending if he/she (i) does not currently listen to CD/V/MP3, and
(ii) would listen to CD/V/MP3 instead of his/her Free-Noninteractive service in the hypothetical.
In this case, I assume that the respondent spends the same amount of money on purchases of
CD/V/MP3 as an average CD/V/MP3 purchaser, with the corresponding royalties equal to a
percentage of such expenditures. As shown in Figure 4 of the body of the report, this results in
average monthly royalties of [ ] for each new CD/V/MP3 purchaser.12 The survey results
show that 14.8 percent of the respondents would “add” CD/V/MP3 listening. The opportunity cost
for this component is then this percentage of additions multiplied by the average monthly royalties
$4,593.8 million, while its “Advertising revenue” was only $187.6 million. SoundExchange Exhibit 49 (Sirius XM Holdings 2018 Form 10-K) at 9 & F-4. 12 This assumes both that people who do not currently engage in CD/V/MP3 listening also do not currently engage in any CD/V/MP3 spending, and that people who do currently engage in CD/V/MP3 listening would not spend more on such products even if they increased their listening time to CD/V/MP3 in the hypothetical.
Public Version
E-7
of [ ] per CD/V/MP3 purchaser, which is [ ] per Free-Noninteractive user per month.
14. The fourth component of opportunity cost is Free-OD. In contrast to the prior three
categories, the amount of time spent listening to Free-OD services does affect the amount of
royalties generated.13 As such, I utilize the responses to the time allocation (percentage) question
in the survey in determining the opportunity cost for this category. Specifically, determining that
opportunity cost for this category involves five steps, as described below.
15. First, I determine the average plays per month for people who currently listen to
Free-Noninteractive services. As shown in Panel A of Figure E-1 below, this amounts to
approximately [ ] plays per month.
13 For example,
]
Public Version
E-8
Figure E-1: Current Plays per User/Subscriber (April 2018 – March 2019) (RESTRICTED)
16. Second, for each respondent who indicates that he/she would listen to Free-OD
instead of Free-Noninteractive, I multiply the average plays per month (above) by the percentage
of time allocated to Free-OD in the survey response. For example, if someone would allocate 20
percent of his/her time to Free-OD, that person would contribute 20 percent x [ ] = [ ] Free-
OD plays. If someone would not switch to Free-OD or assigns zero percent to Free-OD listening,
Public Version
E-9
then he/she does not contribute any Free-OD plays.14
17. Third, I sum the added Free-OD plays (above) across all respondents and divide by
the number of respondents. This amount is the average incremental Free-OD plays per respondent
per month. The survey results show that this average is [ ] plays per respondent.
18. Fourth, I make an adjustment to the number of incremental Free-OD plays
calculated above for respondents who indicate all of the following:
• He/she does not currently listen to Paid-OD, but would listen to Paid-OD instead
of Free-Noninteractive in the hypothetical (i.e., adds a new Paid-OD subscription);
and
• He/she does currently listen to Free-OD, but would not listen to Free-OD instead
of Free-Noninteractive in the hypothetical.
In the above circumstance, I assume that the respondent would replace his/her Free-OD listening
(during times other than when he/she is listening to Free-Noninteractive) with his/her new
Paid-OD subscription. This results in a reduction in Free-OD plays.
19. To calculate this reduction in Free-OD plays, I use the current average number of
plays per month for Free-OD listening. As shown in the figure below, based on Spotify royalty
statements, I estimate average Free-OD listening of [ ] plays per user per month.
14 As previously discussed, if a respondent indicates that his/her only switching option is Free-OD, then 100 percent of the time is allocated to Free-OD. In addition, if the respondent indicates that he/she would not listen to his/her Free-Noninteractive service on the same day next week (or is unsure), then zero percent time is assigned to Free-OD.
Public Version
E-10
Figure E-2: Current Plays per Spotify User (April 2018 – March 2019) (RESTRICTED)
I multiply these [ ] plays per user per month by the 1.6 percent of respondents that the survey
indicates would “drop” their Free-OD listening in the hypothetical, which equates to a reduction
to Free-OD diversion of [ ] plays per respondent per month.
20. Finally, I calculate the net effect of the above – the increased listening less the
“drops.” The survey results reflect a net increase of [ ] Free-OD plays per respondent per month.
This net increase is then multiplied by the royalty rate for Free-OD of [ ] per play (as shown
in Figure 4 of the body of the report) to obtain the opportunity cost, which is [ ] per Free-
Noninteractive user per month.
21. Combining the above components, the total opportunity cost associated with
diversion from Free-Noninteractive to modes of music distribution other than Paid-Noninteractive
is $0.953 per Free-Noninteractive user per month. This amount is the O1 Shapley Value input
parameter defined in Appendix C.
22. The final Shapley Value input coming from the NoFree hypothetical is diversion
from Free-Noninteractive to Paid-Noninteractive (the ��,�� parameter defined in Appendix C). The
calculations for diversion from Free-Noninteractive to Paid-Noninteractive are analogous to the
Public Version
E-11
calculations for diversion to Free-OD services described earlier.15 That is, the survey results
provide the number of additional Paid-Noninteractive plays per respondent based on the
percentage allocation questions.16 Here, the survey indicates that a diversion of [ ] Paid-
Noninteractive plays would be generated per respondent per month.
IV. Processing of Zauberman Survey Results – NoPaid Hypothetical
23. Turning to the NoPaid hypothetical, the calculations for the O2 parameter (i.e.,
opportunity cost associated with diversion to modes of music distribution other than to Free-
Noninteractive) are the same as described above for the NoFree hypothetical. The only difference
is that the average number of plays for Paid-Noninteractive services is [ ] per month, as opposed
to [ ] per month for Free-Noninteractive, as shown in Figure E-1 above. Combining all of the
components, O2 is $1.240 per Paid-Noninteractive user per month.
24. The calculations for diversion from Paid-Noninteractive to Free-Noninteractive
(i.e., ��,�� ) are analogous to the calculations described above for diversion to Free-OD. That is, I
use the time allocation (percentage) responses from the survey to determine the number of
incremental Free-Noninteractive plays generated for each respondent. The result is that, under the
NoPaid hypothetical, an average of [ ] Paid-Noninteractive plays are diverted to Free-
15 The current statutory license requires royalties from subscription noninteractive distributors to be paid out to record companies on a per-play basis.16 As discussed above, the time allocation question is not asked in three circumstances, which are also relevant here: (1) if the respondent specifies only one option in response to the switching question, in which case, 100 percent of the time is assigned to the option chosen; (2) if the respondent indicates that he/she would not switch to Free-OD services; or (3) if the respondent indicates that he/she would not listen to the Free-Noninteractive service on the same day next week (or is unsure about such listening), in which case zero switching time is assigned to all categories. In the case of (2), for instances when the respondent indicated that he/she would switch to Paid-Noninteractive listening in the hypothetical, but not Free-OD, zero switching time is assigned to Paid-Noninteractive – this again is conservative with respect to calculating opportunity costs.
Public Version
E-12
Noninteractive services per respondent per month.
V. Calculation of “O-Prime” Parameters
25. This section describes how the additional Shapley Model parameters��� and ��� are
computed from the parameters discussed in the prior sections. As described in Appendix C, these
“O-prime” parameters represent the opportunity cost from diversion of listening activity if both
ad-supported noninteractive and subscription noninteractive services were not available. These
parameters are calculated by combining the results from both hypothetical scenarios, as described
in detail below.
26. Relatedly, ��,��� and ��,�
�� measure diversion rates of listening time from the ad-
supported and subscription noninteractive distributors, respectively, to outside distributor Dj in the
situation where the other noninteractive distributors are not operating.17 ��,��� and ��,�
�� are linked
by the following relationship:18
��,��� = ��,�
� + ��,��� ∗ ��,�
This equation implies that diversion to an outside distributor from ad-supported noninteractive
services when subscription noninteractive services are unavailable is given by the diversion when
the subscription noninteractive services are available (��,�� ) plus some additional diversion. The
additional diversion is the diversion from D1 that would have gone to D2 had it still been available,
17 Specifically, ��,�′� measures the fraction of current users of noninteractive distributor i’s service who would divert to outside distributor j’s service if service i were to stop operating, while the other noninteractive distributor were also not operating. This diversion is measured as the change in royalty-bearing units t on service j per then-current user of service i. 18 In the absence of more specific data, d1,2 is taken here to be the average fraction of listening time diverted to Paid-Noninteractive services per user of Free-Noninteractive services if they were not available, which is [ ]. Similarly, in the absence of more specific data, d2,1 is taken here to be the average fraction of listening time diverted to Free-Noninteractive services per user of Paid-Noninteractive services if they were not available, which is [ ].
Public Version
E-13
and then would have in turn diverted from D2 to Dj upon the exit of D2, in the situation where D1
was not available.
27. Similarly,
��,��� = ��,�
� + ��,��� ∗ ��,�.
28. Then, I solve these two equations simultaneously for ��,��� and ��,�
�� to obtain:
��,��� =
��,�� + ��,�
� ∗ ��,�
1 − ��,� ∗ ��,�
��,��� =
��,�� + ��,�
� ∗ ��,�
1 − ��,� ∗ ��,�.
29. With these �′�,�� parameters, I can then calculate ��� and ��� using the royalty rates
for each outside distributor – [ ] for Paid-OD, etc. – as previously discussed. This results in
an ��� of $1.010.99 and an ��� of $1.421.38.19
19 In accordance with Professor Zauberman’s testimony, the results discussed throughout this section do not incorporate any demographic weighting of the survey responses. See Zauberman WDT. Nevertheless, Professor Zauberman has provided demographic weights, which I have used to test the sensitivity of my results. See SoundExchange Exhibit 31. The demographic weights are based on geography (region), gender, and age. Incorporating these demographic weights has no material impact on the results produced by my Shapley Value model.
Public Version
E-14
VI. Calculation of Confidence Intervals
30. As discussed in the body of the report, I use a bootstrap procedure to calculate
confidence intervals for the ad-supported noninteractive and the subscription noninteractive
royalty rates computed from my Shapley Value model.20
31. For the bootstrapping analysis, I utilize 1,000 simulated sets of survey respondents
and calculate the Shapley Value analysis parameters for each of those simulated sets of
respondents. Each of the simulated sets of survey respondents is based on N random draws from
the overall actual population of qualified responses, where N is the number of qualified responses
to the main survey (989 for the NoFree hypothetical and 563 for the NoPaid hypothetical). The
sampling for each simulated set of respondents is done with replacement, so any individual actual
responses to the main survey can be included more than once (or not at all) in any individual
simulated set of survey respondents. The specific responses included in each simulated set of
survey respondents were provided by Professor Zauberman.21
32. With each of these 1,000 simulated sets of respondents, I apply the same approach
as previously described to calculate the corresponding Shapley Value analysis parameters. From
each of these 1,000 simulated groups of input parameters, I calculate 1,000 royalty rates for the
ad-supported noninteractive distributor and 1,000 royalty rates for the subscription noninteractive
distributor using my Shapley Value model described in Appendix C. From both sets of 1,000
royalty rates I calculate two statistics. First, I compute the 2.5 percent quantile (i.e., the number
20 For a general discussion of the use of bootstrapping, see SoundExchange Exhibit 67 (“AAPOR Guidance on Reporting Precision for Nonprobability Samples,” American Association for Public Opinion Research). 21 See SoundExchange Exhibit 30. My understanding is that these are the same samples that Professor Zauberman used in his bootstrapping procedures to compute confidence intervals for the individual survey responses (see Zauberman WDT).
Public Version
E-15
below which 2.5 percent of the 1,000 royalty rates fall). Second, I compute the 97.5 percent
quantile (i.e., the number below which 97.5 percent of the 1,000 royalty rates fall). These two
statistics define the upper and lower bounds of the 95 percent confidence interval for the royalty
rates determined by my Shapley Value model.
Public Version
F-1
Appendix F: Shapley Values with “Share of Ear” Data
1. This appendix provides a detailed description of my Shapley Value model
calibrated with data from Edison Research’s Q2 2019 Share of Ear research report. Like the model
detailed in Appendix C, this model involves two noninteractive streaming distributors (a free ad-
supported distributor and a paid subscription distributor), three “must-have” record companies
(representing the majors), and one “not-must-have” record company (representing the indies).
I. Model
2. In constructing this Shapley Value model, I utilize the following notation:
Figure F-1: Notation
D1 and D2The ad-supported and subscription noninteractive streaming distributors, respectively.
LA, LB, and LC The three must-have major record companies.
LD
An amalgam of not-must-have indie record companies grouped together without loss of generality. Referred to hereafter as the not-must-have record company.
SharekRecord company k’s share of plays, which is assumed to be the same across all distributors.
CkRecord company k’s costs of producing sound recordings.
��� and ��
�D1’s and D2’s marginal profit per play, respectively, before accounting for any sound recording royalties that they pay.
F1 and F2 D1’s and D2’s fixed costs, respectively.
Dj
where � ∈ {3, …, N}
Distributors whose royalty rates are set outside of the current proceedings, such as ad-supported and subscription on-demand music and video streaming services, SiriusXM satellite radio, and music purchases (i.e., CDs, vinyl, digital downloads).
��� The per play royalty rate paid by outside distributor
j.
��� Di’s audience, measured in plays, when all other
distributors are operating.
��,�
A diversion ratio from service i to j: The percentage of noninteractive distributor Di’s plays that would divert to distributor j’s service if Di
were to stop operating while all other distributors continued operating.
� = � ����∗ ��
��
�
���
The four record companies’ combined royalties from outside distributors when all distributors are operating.
Public Version
F-2
���
= � ����∗ ��,��
�
���
Per play of noninteractive streaming distributor Di’s service, the change in the four record companies’ combined royalties from outside distributors if Di were to stop operating while all other distributors continued operating.
�′�,�
Another diversion ratio from service i to j: The percentage of noninteractive distributor Di’s plays that would divert to outside distributor Dj’s service if service i were to stop operating, while the other noninteractive distributor were also not operating.
����
= �����∗ �′�,��
�
���
Per play of noninteractive streaming distributor Di’s service, the change in the four record companies’ combined royalties from outside distributors if noninteractive distributor Di were to stop operating, while the other non-interactive distributor were also not operating.
3. The Shapley Value solution concept requires calculating the average marginal
contribution to the collective surplus that each record company or distributor creates, based on all
possible orderings of the parties. The first step in this process is to calculate the total surplus each
hypothetical coalition of noninteractive streaming distributors and record companies would
generate if they acted in a manner that maximized their collective surplus. These values define a
function �[�] of all those coalitions � ∈ �(�), called the “characteristic function,” where �(�) is
the set of subsets of the players � = {��,��, ��, ��, �� , ��}.
4. In the present case, the characteristic function can be determined recursively.
An empty coalition, and coalitions comprised of only the distributors, create zero payoffs. Hence:
�[{}] = 0;
�[{��}] = 0;
�[{��}] = 0;
�[{��,��}] = 0.
Public Version
F-3
5. Each label, when it is alone, generates its current profits from outside distributors
plus the additional profits from outside distributors caused by diversion of the plays of
noninteractive distributors 1 and 2 to the outside distributors. Hence:1
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��;
�[{��}] = �ℎ���� ∗ (� + ���∗ �1
�′+ ��
�∗ �2
�′) − ��.
6. Since LA, LB, and LC are must-have, only when all three are present in a coalition
can the distributor(s) operate and make profits. When only these three record companies are
present in the coalition, three cases can arise. First, if only D1 is in a coalition, and if it increases
the coalition’s collective surplus, diversion occurs from D2 to both D1 and the outside distributors
(because D2 is not operating) and from D1 to outside distributors (because D1 does not have a
license to LD’s content). Second, if only D2 is in a coalition, and if it increases the coalition’s
collective surplus, diversion occurs from D1 to both D2 and the outside distributors (because D1 is
not operating) and from D2 to outside distributors (because D2 does not have a license to LD’s
content). And third, if both distributors are in a coalition, and if they increase the coalition’s
collective surplus, no diversion between D1 and D2 occurs, but there is still diversion from D1 and
D2 to the outside distributors (because they do not have licenses to LD’s content).2
1 I assume that it is profitable for the must-have record companies to operate.2 In keeping with the Judges’ conclusion in the Web IV proceeding that “the repertoire of each of the three Majors is a ‘must-have’ in order for a noninteractive service to be viable,” I assume that a distributor could not operate without agreements with all three of the major record companies. See Web IV Determination at 1334. Conversely, I assume that if a distributor has access to the must-have record companies’ content but does not have access to LD’s content, the distributor loses �ℎ���� of its plays.
Public Version
F-4
7. Each coalition is assumed to act in the manner that maximizes its collective
surplus.3 Therefore:
�[{��,��,��,��}]
= ����(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗ ��,�� + � + ��
�∗ �2
��− (�� + �� + ��)
− ��,�[{��,�� ,��}]�; 4
�[{��,��,�� ,��}]
= ����(1 − �ℎ����) ∗ ����∗ ���
�+ ��
�∗ ��,��+ � + ��
�∗ �1
�� − (�� + �� + ��)
− ��,�[{��,��,��}]�;
�[{��,��,��,��,��}]
= ����(1 − �ℎ����) ∗ ����∗ ��
�+ ��
�∗ ��
�+ ��− (�� + �� + ��)− ��
− ��,�[{��,��,��,��}],�[{��,��,�� ,��}]�.
8. When LD joins a coalition, it adds to the coalition’s surplus by contributing its
royalties from outside distributors, and also by licensing one or both distributors in the coalition
(if such a license would increase the coalition’s collective surplus). Thus:
�[{��,��,��,�� ,��}]
= ��� ��� + ���∗ �2
�+ ��
�∗ ���
�+ ��
�∗ ��,�
���− (�� + �� + �� + ��)
− ��,�[{��,��,�� ,��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
�[{��,��,�� ,�� ,��}]
= ��� ��� + ���∗ �1
�+ ��
�∗ ���
�+ ��
�∗ ��,�
���− (�� + �� + �� + ��)
− ��,�[{��,��,��, ��}] + �[{��}],�[{��,�� ,�� ,��}]� ;
�[{��,��,��,��,�� ,��}]
= ������ + ���∗ ��
�+ ��
�∗ ��
��− (�� + �� + �� + ��) − ��
− ��,�[{��,��,��,�� ,��}],�[{��,��,�� ,�� ,��}],�[{��,��,��,��,��}]
+ �[{��}],� + ���∗ ���
�+ �ℎ���� ∗ ��
�∗ ��,��− �� − (�� + �� + �� + ��)
+ (1 − �ℎ����) ∗���∗ ��
�+ �ℎ���� ∗ ���
�∗ �2
��− ��,� + ��
�∗ (��
�+ �ℎ����
∗ ���∗ ��,�) − �� − (�� + �� + �� + ��) + (1 − �ℎ����) ∗��
�∗ ��
�
+ �ℎ��������∗ �1
��− ���.
3 In some cases, the collective surplus of a coalition may be higher if one or both distributors do not operate.4 As explained below, �[{��, ��, ��}] is the combined profits of the coalition that includes only the three must-have record companies. �[{��, ��, ��}] = �[{��}] + �[{��}] + �[{��}].
Public Version
F-5
9. In any other case not defined above, the noninteractive distributors do not carry
one or more of the must-have record companies and cannot operate. In such cases, all the profits
generated by those coalitions are just the sum of the �[{��}] defined above. For example,
�[{��,��}] = �[{��}] + �[{��}],�[{��,��,��,��}] = �[{��}] + �[{��}], etc.
10. Based on this characteristic function, the Shapley Value algorithm can be
implemented. Specifically, I evaluate every possible arrival ordering and for each determine the
incremental value contributed by each noninteractive distributor relative to the value created by
the parties that preceded its arrival. Then I average all of these incremental contributions across
all arrival orderings. With six players, there are 6! (i.e., 720) arrival orderings. Algebraically, this
can be expressed as follows. For every possible subset � of � that includes i,
��� = ���(�[�] − �[� − {�}])
�⊂�
,
where
�� =(|�| − 1)! ∗ (� − |�|)!
�!,
for n equal to the number of elements in � (six in this case) and |�| the number of elements in
subset �.5
11. Following this formula, I calculate the Shapley Values for both noninteractive
distributors, based on the parameters estimated as described in Appendices D and E. The total
royalty payments to be made by D1 and D2 can be calculated using these Shapley Values as
follows:
5 See Michael Carter, “Cooperative Games,” in Hal R. Varian (Ed.), Economic and Financial Modeling with Mathematica, Springer-Verlag (1993), pp. 185-186.
Public Version
F-6
�� = ��� ∗ ��
� − �� − ���;
�� = ��� ∗ ��
� − �� − ���.
In other words, total royalties are equal to the pre-royalty profits of each of the noninteractive
distributors less their respective Shapley Values. In this way, the distributor will be left with its
Shapley Value after it has collected its profits from operations and paid out the royalties to the
record companies.
II. Calibration
12. As described in the body of the report, as an alternative approach, I estimate
diversion ratios from data in the Edison Research Q2 2019 Share of Ear research report.6 Using
these data, I obtained the following shares of time spent listening to different modes of music
distribution:
Figure F-2: Share of Time Spent Listening to Music, Smartphone Owners 13+ (RESTRICTED)
6 See SoundExchange Exhibit 44 (Edison Research, “Share of Ear,” Q2 2019). For further details on Edison Research’s Share of Ear survey methodology, see SoundExchange Exhibit 72 (Q2 2019 Share of Ear Methodology Statement).
Public Version
F-7
13. The Share of Ear data also provide the following decomposition of time spent
listening to streaming music services.
Figure F-3: Share of Streaming Music, Smartphone Owners 13+ (RESTRICTED)
14. The Share of Ear data, however, do not provide a way to split the streaming
services’ audiences into different tiers of service. For instance, the data do not provide shares for
Pandora’s noninteractive Pandora Radio and Pandora Radio Plus tiers or its on-demand Pandora
Premium tier. Hence, to further split each streaming service’s audience into individual tiers of
service, I use play count information from the 12-month period ending March 2019 contained in
royalty statements reported to the record companies. This breakdown is summarized in the figure
below.
Figure F-4: Share by Streaming Service (RESTRICTED)
Public Version
F-8
15. By combining the data in the three figures above, I calculate the total share of
listening time for each tier of services. For example, the share of total listening on the free Pandora
Radio tier is equal to [
] Next,
I group the ad-supported noninteractive distributors (Pandora Radio, iHeart Non-Simulcast
Webcast, iHeart Internet Simulcast, and AM/FM Streaming) together and calculate their share.
These entities combined represent the ad-supported noninteractive distributor D1 in the Shapley
model. I also calculate the share of the subscription noninteractive distributors (Pandora Radio
Plus, iHeart Radio Plus, and Napster UnRadio). These entities combined represent the
subscription noninteractive distributor D2 in the Shapley model.
16. Then, I follow a logit demand model to estimate diversions based on shares,
under the assumption that if either the ad-supported or subscription noninteractive distributors
were to lose access to sound recordings, users would divert their noninteractive plays to other
distributors in a way that is proportional to these distributors’ current shares of listening time.7,8
The resulting diversions are shown in the figure below.
7 See, e.g., Robert D. Willig, “Merger Analysis, Industrial Organization Theory, and Merger Guidelines,” Brookings Papers: Microeconomics (1991): 281- 332, p. 302. See also, Carl Shapiro, “Mergers with Differentiated Products,” Antitrust, No. 10 (1996): 23-30, p. 25. As an example of how this works, assume the shares of product A, B, and C are 10%, 20%, and 70%, respectively. If product A exits the market, diversion to product B is 20% / (100% - 10%) = 22.22% and diversion to product C is 70% / (100% - 10%) = 77.78%. 8 In the absence of specific data, I assume that plays are proportional to time.
Public Version
F-9
Figure F-5: Share-Based Diversion (RESTRICTED)
17. Figure F-5 shows that, for instance, AM/FM Broadcast, which has a share of ear
of [ ], would receive diversion of the plays on ad-supported noninteractive distributors (d1) of
[ ] if those distributors were to exit the market, and diversion of the plays on subscription
noninteractive distributors (d2) of [ ] if subscription noninteractive distributors were to exit
the market. Correspondingly, distributors with lower shares receive less diversion.
18. Relatedly, d1' and d2' measure diversion rates from the ad-supported and
subscription noninteractive distributors, respectively, in the situation where the other
noninteractive distributors are not operating. As explained in Appendix E, d1' and d2' are linked
by the following relationship (for a given outside distributor D0):
�′�,� = ��,� + �′�,� ∗ ��,�
Public Version
F-10
This equation implies that diversion to an outside distributor from ad-supported noninteractive
services when subscription noninteractive services are unavailable is given by the diversion when
the subscription noninteractive services are available (��,�) plus some additional diversion. The
additional diversion is the diversion from D1 that would have gone to D2 had it still been available,
and then would have in turn diverted from D2 to D0 upon the exit of D2, in the situation where D1
was not available.
19. Similarly,
�′�,� = ��,� + �′�,� ∗ ��,�.
20. Then I solve these two equations simultaneously for �′�,� and �′�,� to obtain
�′�,� =��,� + ��,� ∗ ��,�
1 − ��,� ∗ ��,�
�′�,� =��,� + ��,� ∗ ��,�
1 − ��,� ∗ ��,�.
For example, in the case where D0 is AM/FM Broadcast, �′�,� = [ ]
21. Next, with these diversion rates and the per play royalty rates shown in the figure
below, I compute the opportunity cost associated with each form of outside distribution. For
example, the amounts shown in Figure F-6 in the column labeled “d1 * Royalty” show the
opportunity cost of the record companies associated with the diversion that would occur to each
source of music distribution if noninteractive distributor D1 were not in the market. The “Total”
line shows the sums of the opportunity costs associated with all such outside music distributors,
and these are the computations of the model input parameters ���
, ���
, ���� and ��
�� found in Figure
F-7.
Public Version
F-11
Figure F-6: Royalties and Calculation of Opportunity Cost (RESTRICTED)
Public Version
F-12
Figure F-7: Opportunity Cost and Diversion Rates with Share of Ear
22. The additional diversion parameters d1,2 and d2,1 are also shown in Figure F-7.
All other parameters used to calibrate the Shapley model are described in the body of the report
and Appendix D. The resulting Shapley Values and royalties are presented in Figure 10 in the
body of the report.9
9 I tested the robustness of these results by performing a number of sensitivities. First, I used the same Share of Ear data, but from an earlier Q3 2018 Share of Ear report. See SoundExchange Exhibit 43 (Edison Research, “Share of Ear,” Q3 2018). Second, I used the total population data in the Q2 2019 Share of Ear study rather than limiting the data to only smartphone users. Third, rather than utilize the share of streaming music breakdowns shown in Figure F-3, I used play count information from royalty statements to allocate all Streaming and YouTube listening amounts from Figure F-2 among the various services and service tiers. Last, instead of share of listening time from Edison Research, I employed similar share of listening time data from MusicWatch. See SoundExchange Exhibit 42 (MusicWatch, “Annual Music Study 2018: Report to RIAA,” April 2019) at 27, 39. None of these sensitivity checks substantially affects my quantitative results.
d12 O1p O1
p' d21 O2p O2
p'
1.1% $0.00229 $0.00231 9.6% $0.00209 $0.00231
d12 O1p O1
p' d21 O2p O2
p'
1.1% $0.00234 $0.00237 9.6% $0.00214 $0.00237
Public Version
G-1
Appendix G: Alternative Nash-in-Nash Bargaining Model
I. Introduction
1. As explained in the body of the report, Nash-in-Nash bargaining is a method
derived from economic theory to determine the outcome of a multi-party negotiation.
2. To provide a simple example, consider a bargaining situation involving one
copyright owner and two noninteractive distributors, D1 and D2. In step one, the Nash Bargaining
Solution resulting from the negotiation between the copyright owner and noninteractive distributor
D1 is identified, which depends on the Nash Bargaining Solution to the negotiation between the
copyright owner and noninteractive distributor D2. Likewise, the Nash Bargaining Solution
resulting from the negotiation between the copyright owner and distributor D2 is identified, which
depends on the outcome of the negotiation between the copyright owner and distributor D1.
3. At the end of step one, the two bargaining solutions are characterized, with each
depending on the solution to the other. Mathematically, this analysis is represented as two
equations relating the two unknown values of the two bargaining solutions. In step two, this system
of two equations and two unknowns is simultaneously solved to determine the terms of the
agreements between the copyright owner and each of distributors D1 and D2.
4. In the above example, if the copyright owner reaches agreements with
noninteractive distributors D1 and D2, it may earn less profits by licensing its sound recordings to
other forms of music distribution. The extent to which that is the case will depend on the extent
to which D1 and D2 can successfully lure away the listening audiences of the other distributors.
This is the copyright owner’s opportunity cost of licensing to D1 and D2. In order for the copyright
owner to have any incentive to enter into deals with D1 and D2, it must receive royalties that at
least fully offset the profits it will be giving up from other distributors by virtue of entering those
Public Version
G-2
deals. No rational copyright owner would enter into deals with noninteractive distributors D1 and
D2 unless it recovered this amount, plus at least some share of the surplus jointly created by those
deals.
5. An adaptation and continuation of the numerical example from the body of the
report further illustrates the concept of the Nash-in-Nash bargaining solution. Here, two patents,
patentA and patentB, are both necessary for the manufacture and sales of deviceD. The fallback
values for the owners of patentA and patentB are each $5 if deviceD is not on the market. If
deviceD is brought to market, then the individual net returns that the owners of patentA and
patentB will realize from licensing to other rival devices will decrease to $3, due to cannibalization
from deviceD.
6. To analyze the bargaining between the owners of patentA and deviceD, suppose
that the owners of patentB and deviceD have already committed to a deal that provides a royalty
payment of RB from the owner of deviceD to the owner of patentB if the patent is utilized. Given
this deal, the fallback value for the owner of patentA is $5 and the fallback value for the owner of
deviceD is zero. The total value to the owners of patentA and deviceD from an agreement between
them, which would enable the manufacture and sale of deviceD due to the deal already struck with
the owner of patentB, is the $3 from the net return to the owner of patentA from other licensing
activity plus the net return to the owner of deviceD of $16 minus RB. The surplus created by this
deal is the total value created from the agreement minus the sum of the parties’ fallback values,
which is $3 + $16 – RB – $5, or $14 – RB.
7. The corresponding Nash Bargaining Solution splits that surplus, leaving the owner
of deviceD with .5 x ($14 – RB). Since the owner of deviceD earns $16 of market profit, pays out
RB to the owner of patentB, and pays out RA to the owner of patentA, the resulting relationship
Public Version
G-3
is $16 – RB – RA = .5 x ($14 – RB).
8. To find the Nash-in-Nash bargaining equilibrium, it is necessary to also analyze the
bargaining between the owners of patentB and deviceD, while assuming that the owners of patentA
and deviceD have already committed to a deal that provides a royalty payment of RA from the
owner of deviceD to the owner of patentA if the patent is utilized. Given the symmetry of
assumptions for patentA and patentB, the resulting Nash Bargaining solution to this negotiation is
similar to the one identified above, namely: $16 – RA – RB = .5 x ($14 – RA). Solving these two
equations simultaneously with simple algebra yields that the royalties indicated by the Nash-in-
Nash solution are RA = RB = $6.
II. Six-Player Nash-in-Nash Model
9. This section provides a detailed description of the Nash-in-Nash bargaining model
that involves two noninteractive distributors (an ad-supported distributor and a subscription
distributor), three “must-have” record companies, and one “not-must-have” record company.
10. In the Nash-in-Nash model, I analyze the royalty rates that would prevail if: (i) each
record company and each noninteractive distributor negotiated bilaterally a lump sum royalty that
the distributor would pay the record company for its distribution of that record company’s content;
and (ii) each of these bilateral negotiations proceeded under the assumption by both parties that
every other distributor-record company pair had agreed or would agree on a lump sum royalty.
From this model’s construction I identify the Nash Equilibrium royalty payments, which are the
set of royalty payments from which no pair of negotiators has an incentive to deviate unilaterally.
In constructing this Nash-in-Nash bargaining model, I utilize the following notation:
Figure G-1: Notation
D1 and D2The ad-supported and subscription noninteractive streaming distributors, respectively.
LA, LB, and LC The three must-have major record companies.
Public Version
G-4
LD
An amalgam of not-must-have indie record companies grouped together without loss of generality. Referred to hereafter as the not-must-have record company.
SharekRecord company k’s share of plays, which is assumed to be the same across all distributors.
CkRecord company k’s costs of producing sound recordings.
��� and ��
�D1’s marginal profit per play and D2’s marginal profit per play, respectively, before accounting for any sound recording royalties that they pay.
F1 and F2 D1’s and D2’s fixed cost, respectively.
Ri,kThe dollar amount of royalties Di agrees to pay Lk, if they reach an agreement.
Dj
where � ∈ {3, …, N}
Distributors whose royalty rates are set outside of the current proceedings, such as ad-supported and subscription interactive music and video streaming services, Sirius XM satellite radio, and music purchases (i.e., CDs, vinyl, digital downloads).
���
The royalty rate paid by outside distributor j per royalty-bearing unit � ∈ {play, subscriber, CD/vinyl-record/download}.
���
Di’s audience, measured in Di’s royalty bearing units t (e.g., plays for D1), when all other distributors are operating.
���The average number of monthly plays per user of D1’s service.
���The average number of monthly plays per user of D2’s service.
��,��
A diversion ratio from service i to j: Per current user of noninteractive streaming distributor i’s service, the change in royalty-bearing units t on distributor j’s service if Di were to stop operating while all other distributors continued operating.
� = � ���� ∗ ��
���
���
The four record companies’ combined royalties from outside distributors when all distributors are operating.
�� = � ���� ∗ ��,�
� ��
���
Per current user of noninteractive streaming distributor i’s service, the change in the four record companies’ combined royalties from outside distributors if Di were to stop operating while all other distributors continued operating.
11. I assume that if an agreement between a record company and a noninteractive
distributor increases their combined profits (assuming all other pairs of record companies and
noninteractive distributors reach an agreement), the two parties will reach the Nash Bargaining
Solution by agreeing on the lump sum royalty that splits their combined incremental profits
Public Version
G-5
equally.
12. The first step in identifying this royalty is to define each record company’s and
each noninteractive distributor’s profits when they reach an agreement and when they do not,
assuming in both cases that all other pairs of record companies and noninteractive distributors
reach an agreement.
13. If all pairs of record companies and noninteractive distributors reach agreements,
the profits of each of these parties can be expressed as follows:1
π�{���} = ��,� + ��,� + (Share� ∗ E) − ��, where � ∈ {�,�,�,�};
π�{���} = ���
∗ ���
− �� − ∑ ��,�� ;
π�{���} = ���
∗ ���
− �� − ∑ ��,�� .
14. If a given record company and noninteractive distributor do not reach an
agreement, while all other pairs of record companies and noninteractive distributors do reach
agreements, the profits of the record company and distributor that do not reach an agreement can
be expressed as follows:2
π�{~(D�, L�)} = ��,� + (�ℎ���� ∗ E) + ���
��1∗ �ℎ���� ∗ ��
�� − ��, where � ∈ {�,�,�,�};
π�{~(D�, L�)} = 0, where � ∈ {�,�,�};
π�{~(D�, L�)} = ����
∗ (1 − �ℎ����) ∗ ���
− ��� − ∑ ��,���� ;
π�{~(D�, L�)} = ��,� + (�ℎ���� ∗ E) + ���
��2∗ �ℎ���� ∗ ��
�� − ��, where � ∈ {�,�,�,�};
1 I assume that each record company’s share of the combined royalties from outside distributors when all distributors are operating (E) is equal to its share of plays. 2 In keeping with the Judges’ conclusion in the Web IV proceeding that “the repertoire of each of the three Majors is a ‘must have’ in order for a noninteractive service to be viable,” I assume that a distributor could not operate without agreements with all three of the major record companies. See Web IV Determination at 1334. Conversely, I assume that if a distributor has access to the must-have record companies’ content but loses access to LD’s content, the distributor would lose �ℎ���� of its plays (or subscribers).
Public Version
G-6
π�{~(D�, L�)} = 0; where � ∈ {�,�,�};
π�{~(D�, L�)} = ����
∗ (1 − �ℎ����) ∗ ���
− ��� − ∑ ��,���� .
15. Using the above definitions, each record company’s and each non-interactive
distributor’s incremental profits from reaching an agreement with each other, when all other
pairs of record companies and non-interactive distributors reach an agreement, can be expressed
as follows:3
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ��,� − ���
��1∗ �ℎ���� ∗ ��
�� , where � ∈ {�,�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ���
∗ ���
− �� − ∑ ��,�� , where � ∈ {�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ����
∗ �ℎ���� ∗ ����− ��,�;
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ��,� − ���
��2∗ �ℎ���� ∗ ��
�� , where � ∈ {�,�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ���
∗ ���
− �� − ∑ ��,�� , where � ∈ {�,�,�};
∆π�{(D�, L�)} = π�{���} − π�{~(D�, L�)} = ����
∗ �ℎ���� ∗ ����− ��,�.
16. Since the Nash bargaining solution dictates that the parties split equally the
combined incremental profits from an agreement, the royalty that results from the bilateral
negotiation between Di and Lk is the one that equalizes both parties’ incremental profits, i.e., that
leads to ∆π�{(D� , L�)} = ∆π�{(D�, L�)}. This yields a system of eight equations with eight
unknown royalty payments.
17. Solving this system of equations yields the following total royalty payments for
each distributor:
�� = ���,�
�
=�(2 − �ℎ����) ∗
����1
+ ���
∗ (6 + �ℎ����)� ∗ ���
− 6 ∗ ��
8;
3 ∆π�{(D�, L�)} = π�{���} − π�{~(D� , L�)} and ∆π�{(D� , L�)} = π�{���} − π�{~(D� , L�)}represent Lk’s and Di’s incremental profits, respectively, from reaching an agreement with each other when all other pairs of record companies and non-interactive distributors reach an agreement.
Public Version
G-7
�� = ���,�
�
=�(2 − �ℎ����) ∗
����2
+ ���
∗ (6 + �ℎ����)� ∗ ���
− 6 ∗ ��
8.
Public Version
Proof of Delivery
I hereby certify that on Thursday, December 12, 2019, I provided a true and correct copy of
the SoundExchange's Unopposed Motion to Submit the Corrected Written Direct Testimony of
Robert Willig to the following:
Sirius XM Radio Inc., represented by Bruce Rich, served via Electronic Service at
Corporation for Public Broadcasting, represented by Kenneth L Steinthal, served via
Electronic Service at [email protected]
Google Inc., represented by Kenneth L Steinthal, served via Electronic Service at
Radio Paradise Inc., represented by David Oxenford, served via Electronic Service at
Educational Media Foundation, represented by David Oxenford, served via Electronic
Service at [email protected]
National Public Radio, Inc., represented by Kenneth L Steinthal, served via Electronic
Service at [email protected]
Pandora Media, LLC, represented by Bruce Rich, served via Electronic Service at
circle god network inc d/b/a david powell, represented by david powell, served via Electronic
Service at [email protected]
National Religious Broadcasters Noncommercial Music License Committee, represented by
Karyn K Ablin, served via Electronic Service at [email protected]
iHeartMedia, Inc., represented by Julia L Haines, served via Electronic Service at
National Association of Broadcasters, represented by Andrew Gass, served via Electronic
Service at [email protected]
College Broadcasters, Inc., represented by David D Golden, served via Electronic Service
Signed: /s/ Previn Warren