Copyright Board Canada

Commission du droit d’auteur Canada

 

Date	2012-10-05
Citation	Files: Public Performance of Musical Works; Reproduction of Musical Works
Regime	Collective Administration of Performing Rights and of Communication Rights Copyright Act, subsections 68(3), 70.15(1)
Members	Mr. Justice William J. Vancise Mr. Claude Majeau Mrs. Jacinthe Théberge
Proposed Tariff(s) Considered	SOCAN Tariff 22.A (Internet – Online Music Services), 2007-2010 and CSI Online Music Services Tariff, 2008-2010

statements of royalties to be collected by socan and cmrra/sodrac inc. for the communication to the public by telecommunication or the reproduction, in canada, of musical works

Reasons for decision

I. INTRODUCTION 1

II. PREVIOUS DECISIONS DEALING WITH ONLINE MUSIC SERVICES 2

III. PARTIES AND PROPOSED RATES 5

IV. EVIDENCE AND POSITIONS OF THE PARTIES 7

A. CSI 7

B. SOCAN 8

C. The Objectors 9

V. ANALYSIS 12

A. Permanent downloads 12

i. CSI 12

a. Dr. Barker’s proposed models 14

b. Dr. Chipty’s models 16

ii. SOCAN 18

B. Limited downloads 20

C. On-demand streaming 21

D. Video-clips 23

E. Rate base 24

F. Minimum fees 25

i. The parties’ proposals 25

ii. Permanent and limited downloads 26

iii. On-demand streaming 27

iv. Video-clips 28

G. The 10 per cent discount 28

H. Total royalties 29

VI. Wording, Terms and Conditions of the tariff 29

A. Separate Tariffs 29

B. Technology 29

C. A New Tariff for Free On-Demand Streams 29

D. Reporting and Accounting 30

E. Challenges to CSI Ownership Claims 32

F. Consequences of a Failure to Provide a Report 32

G. Confidentiality 32

H. The SOCAN Tariff 33

I. Transitional Provisions 33

I. INTRODUCTION

[1] On March 31, 2006, March 30, 2007, March 31, 2008 and March 27, 2009, the Society of Composers, Authors and Music Publishers of Canada (SOCAN) filed, pursuant to subsection 67.1(1) of the Copyright Act [1] (the “Act”), proposed statements of royalties for the communication to the public by telecommunication of musical works by the same services for 2007, 2008, 2009 and 2010.

[2] On March 30, 2007, March 31, 2008 and March 28, 2009, CMRRA/SODRAC Inc. (CSI) filed pursuant to subsection 70.13(1) of the Act, proposed statements of royalties for the reproduction of musical works by online music services for 2008, 2009 and 2010.

[3] The proposed statements were published in the Canada Gazette. Prospective users and their representatives were informed on each occasion of their right to object. Bell Canada, Rogers Communications Inc., Telus Communications Company and Videotron Ltd. (jointly the “Cable/Telcos”), Apple Canada Inc. and the Canadian Recording Industry Association (CRIA, now Music Canada) filed objections.

[4] On January 10, 2008, CSI requested the Board set in motion the examination of its tariff. The Objectors opposed the application. They requested that CSI’s tariff be heard jointly with that of SOCAN, which was not then ready to proceed. On February 21, 2008, the Board denied CSI’s request and ruled that it would examine both online music tariffs at the same time.

[5] On August 25, 2009, CSI again requested the Board set in motion the examination of its tariff, this time jointly with SOCAN. The objectors again opposed the application on the basis that, among others, the Board should not proceed until the Federal Court of Appeal had disposed of applications for judicial review of SOCAN’s tariffs. [2] The Board decided not to await the outcome of the proceedings and granted the request. [3] It agreed with SOCAN that previews, which do raise issues under review that might have a direct impact on the tariff, could be segregated by leaving in abeyance anything having to do with them.

[6] The matter was heard over eight days in June of 2010 and the record of the proceedings was closed on July 7, 2010, when the parties made their oral arguments. This decision deals with the tariffs up to 2010. On May 26, 2011, the Board ruled that the examination of the proposed Online Music Services Tariffs of CSI and SOCAN for the years 2011 and 2012 as well as of the proposed SODRAC Tariff 6 (Online Music Services – Music Videos) for 2010-2012 be merged. The Board added that the process is to be triggered as soon as the tariffs for the instant case are certified. This was reiterated on July 11, 2012.

[7] This decision deals once again with three types of products, namely permanent downloads, limited downloads and on-demand streaming. [4] A permanent download is a file that contains a sound recording of a musical work sent to and first stored on the device (computer, cell phone, Smartphone, iPod, etc.) used to purchase it. The person who receives the download can listen to it indefinitely. Limited downloads are offered as part of a subscription service and can only be used as long as the user’s subscription is paid. The digital rights management (“DRM”) software attached to the music file prevents further use once the subscription expires. On-demand streams are not downloads. The user does not receive the complete file containing the sound recording. The service only transmits or streams enough data to allow the user to listen to the recording at the time of the transmission and the user is prevented from copying the recording onto a recording medium or device. This decision also deals, for the first time, with video-clips, of which more will be said in due course.

II. PREVIOUS DECISIONS DEALING WITH ONLINE MUSIC SERVICES

[8] An online music service operating in Canada must account for as many as four copyrights and two remuneration rights. [5] Only two are at issue in these proceedings: the rights to communicate and to reproduce musical works. Both have been the subject of earlier proceedings before the Board and of subsequent decisions of the courts.

[9] In 2007, the Board established for the first time the royalties payable by online services for their reproductions of musical works. [6] The decision based the rate for permanent downloads on the amount in pennies per track (the “penny rate”) record labels pay to reproduce musical works onto prerecorded CDs. It then converted the penny rate into a percentage of the average price of a download, yielding a rate of 8.8 per cent. The rate for limited downloads was established at two-thirds of the rate for permanent downloads. The rate for on-demand streams was established somewhere between the rate for limited downloads and the negotiated rate for digital pay audio (“DPA”) services. [7] The Board applied a discount of 10 per cent to account for the fact that the industry was still in its infancy and had low profit margins. It also determined minimum rates to ensure that the collectives would not subsidize the services’ business models.

[10] Later in 2007, the Board, for the first time, set the royalties payable by online music services for their communication of musical works. [8] The decision first determined a combined price (the “bundled rate”) for the right to reproduce and to communicate a musical work (the “bundle of rights”) by increasing the amount transferred to rights holders for those works based on the increased profitability of the record labels for online sales as compared to CD sales. The bundled rate was set at 12.2 per cent. Since the Board had previously set the price for the reproduction right at 8.8 per cent, the rest, or 3.4 per cent, was attributed to SOCAN. The Board also decided the bundled rate for limited downloads and on-demand streams should be the same. The proportionate share of each right was varied depending on the relative importance of those rights in each type of offering. As in CSI – Online Music Services (2007) and for the same reasons, the Board applied a 10 per cent discount and set minimum rates.

[11] As a result of the Board’s decisions, the applicable rates were as follows: for permanent downloads, 8.8 (CSI) and 3.4 (SOCAN) per cent of the amount paid by consumers, with minimum fees of $0.059 and $0.023, respectively, for single tracks and $0.045 and $0.017, respectively, for tracks sold as part of a bundle of 13 or more tracks; for limited downloads, 5.9 (CSI) and 6.3 (SOCAN) per cent of the amount paid by subscribers, with minimum fees of $0.057 and $0.0609 per subscriber, respectively, if the subscription authorizes copying a music file on a portable device (portable limited download) and $0.0374 and $0.0399, respectively, if it does not; for on-demand streams, 4.6 (CSI) and 7.6 (SOCAN) per cent of the amount paid by subscribers, with minimum fees of $0.0291 and $0.0441, respectively.

[12] The CSI tariff was not challenged before the Federal Court of Appeal. The SOCAN tariff triggered five applications for judicial review. The issues raised included whether an Internet transmission is a communication to the public, whether the services’ dealings with previews constitute fair dealing for the purpose of research, some evidentiary issues as well as the methodology used by the Board to set the tariff. The applications were dismissed in four separate decisions. [9] Leave to appeal to the Supreme Court of Canada was granted in three cases. [10]

[13] On July 12, 2012, the Supreme Court of Canada rendered its decisions in the three cases. [11] These decisions ruled (a) that the Internet delivery of a permanent copy of a file containing a musical work does not involve a communication of that work, (b) that a communication to the public can result from a series of non-simultaneous transmissions of a musical work (as long as this does not result in the delivery of a permanent copy of a file containing the work), and (c) that dealings in previews of downloads constituted fair dealing for the purposes of research.

[14] On July 20, the Board issued an Order which read in part as follows.

The decisions the Supreme Court of Canada issued […] require that the Board modify, sometimes significantly, the tariffs and reasons it was soon to release in [this] matter.

In particular, the Board is of the preliminary view that SOCAN is no longer entitled to a tariff for permanent downloads or limited downloads and that listening to previews constitutes fair dealing for the purpose of research in 2007-2010 just as it did in 1996-2006. Some other adjustments might also be necessary.

The Board is considering three ways to deal with the consequences of the Decisions […]

1) The Board can proceed to adjust the reasons and tariffs without further input from the parties. […] The Board would then proceed without delay with the examination of the tariffs for the years 2011 and following, at which time the full impact of the Decisions would be accounted for.

2) The Board can seek further input from the parties. […]

3) The Board can fully re-open the matter and join it with the examination of the tariffs for the years 2011 and following. […]

[…]

The Board favours option 1.

Parties are invited to communicate to the Board their views on all these issues […]

[15] The parties chose option 1. None of them challenged our preliminary conclusions. These reasons are drafted accordingly. We did not revisit the approaches we had rejected earlier except to the extent that ESA, Rogers or Bell made it absolutely necessary. We reviewed the approaches we intended to use to make them comply as best we could with these decisions. As a result we set no SOCAN tariff for permanent or limited downloads or for previews. Any further fine tuning is left to the next round of proceedings.

[16] Other editorial choices we made as to what to remove and what to leave are admittedly somewhat arbitrary, and sometimes dictated by the desire to issue these reasons in the least possible amount of time. For example, some descriptions of the parties’ arguments and evidence that could have been removed are left in to allow the reader to better follow the train of thought of a witness or party. We can only hope that this will not make the reading of what follows unnecessarily awkward.

III. PARTIES AND PROPOSED RATES

[17] CSI is a collective society created by two other collectives, the Canadian Mechanical Reproduction Rights Agency (CMRRA) and the Society for the Reproduction Rights of Authors, Composers and Publishers in Canada and SODRAC 2003 Inc. (SODRAC). These collectives have granted CSI an exclusive mandate to licence the reproduction of musical works in their repertoires for certain uses, including those targeted in the proposals under examination. Initially, CSI proposed the following rates: 9.9 per cent for permanent downloads, subject to a minimum of $0.044 per download in a bundle of 15 or more, and of $0.066 in all other cases; 9.9 per cent for limited downloads, subject to a minimum of $0.96 per subscriber if portable limited downloads are allowed, and $0.63 if not; and 6.8 per cent for on-demand streams, subject to a minimum fee of $0.43 per subscriber. For permanent downloads, the proposed rate base was the amount paid by consumers in 2008 and 2009 and gross revenue in 2010; for limited downloads and on-demand streams, it was gross revenue for the whole period.

[18] In its statement of case, CSI proposed that the rate base for permanent downloads be the amount paid by consumers for all of the period under examination. Otherwise, its proposal remained unchanged.

[19] At the hearing, CSI modified its application in two respects. First, it proposed that the rate base for limited downloads and on-demand streams be subscription revenues. Second, it lowered the proposed rate for on-demand streams to 5.9 per cent for 2008 and 2009 and 6.1 per cent for 2010, with minimums of $0.43 and $0.44 per subscriber respectively. Finally, CSI requested that the initial discount used by the Board to allow for a phase-in of the tariff be discontinued.

[20] SOCAN administers the right to communicate virtually all protected musical works. For 2007 and 2008, it proposed 25 per cent of the greater of gross revenue or operating expenses, with a minimum monthly fee of $200. Thereafter, it requested different rates for different products. The proposed rates for permanent and limited downloads are no longer relevant. For on-demand streams, SOCAN proposed 6.8 per cent for 2009, subject to a minimum of $0.433 per subscriber. For 2010, it proposed 15.2 per cent, subject to a minimum of $0.962 per subscriber. For free streams, SOCAN proposed a royalty of $0.046 per streamed file.

[21] SOCAN proposed that the rate base be the amounts paid by subscribers in 2009 and gross revenue in 2010.

[22] In its statement of case, SOCAN proposed rates for on-demand streams of 8.1, 8.4, 8.3 and 8.5 per cent for each year from 2007 to 2010, subject to a minimum fee of $0.962 per subscriber per month, and $0.046 per free stream. It proposed gross revenue as rate base.

[23] SOCAN proposed that the tariff apply to video-clips and that the rates be the same as for audio files. It also argued for the elimination of the initial discount. CSI did not propose a rate for music video-clips.

[24] Apple sells permanent downloads through its online music store iTunes. The cell phone customers of Bell, Rogers and Telus can purchase permanent downloads and subscribe to limited download services. Groupe Archambault Inc., a Quebecor Media Company, offers permanent downloads through Zik.ca. It also sells CDs both over the Internet and in its bricks-and-mortar stores.

[25] Apple and the Cable/Telcos proposed that the combined royalties payable to SOCAN and CSI for all audio offerings be fixed in a range between 7.5 and 10 per cent for 2008 and 2009 and between 7.5 to 10.3 per cent for 2010. They proposed using the amount paid by consumers as rate base. For video-clips, they suggested that the SOCAN royalties be set between 0.7 to 2.2 per cent. They proposed a new structure for minimum rates and requested that the phase-in discount be maintained.

[26] CRIA is a not-for-profit corporation that represents the interests of Canadian companies that create, manufacture and market sound recordings. Neither CRIA nor any of its members are prospective users pursuant to the proposed tariffs. However, as no one objected to its participation, it was allowed to participate fully.

[27] CRIA contended that the combined rate for the bundle of rights should be set at 6.96 per cent of the price paid by consumers. With respect to the communication right for video-clips, CRIA agreed with the proposal put forward by Apple and the Cable/Telcos. Finally, CRIA opposed minimum fees regardless of level and requested that the phase-in discount be maintained.

IV. EVIDENCE AND POSITIONS OF THE PARTIES

A. CSI

[28] Mr. Paul Audley, Paul Audley & Associates Ltd., and Mr. Douglas Hyatt, Professor at the Rotman School of Management and at the Centre for Industrial Relations, University of Toronto filed a report [12] that updates the Board’s calculations in CSI – Online Music Services (2007). They proposed using the same approach to set the rate for permanent downloads. As the penny rate has increased since 2007, those calculations, after adjusting for free downloads, now yield rates of 10.0 per cent for 2008 and 2009 and 10.3 per cent for 2010.

[29] In 2007, the Board set the rate for limited downloads at two-thirds of the rate for permanent downloads. Messrs. Audley and Hyatt contended that the CSI rate should be the same for both, by reason that both involve the same uses of the reproduction and communication rights. In both cases, a single communication of a music file is required; subsequent plays occur from the reproduction made on the user’s media or device.

[30] In 2007, the Board set the rate for on-demand streams between the negotiated DPA rate and the rate for limited downloads. Messrs. Audley and Hyatt proposed to apply the same formula, with one adjustment. In 2009, the Board used the DPA rate as proxy to set the rate for satellite radio services but reduced it by half to account for the fact that the DPA rate base is a wholesale price (what broadcasting distribution undertakings pay to DPAs) while the satellite radio rate base is a retail price (what subscribers pay to satellite radio services). [13] Messrs. Audley and Hyatt proposed this correction should also be applied here. As a result, they recommended rates of 5.9 per cent for 2008 and 2009, and 6.1 per cent for 2010.

[31] Messrs. Audley and Hyatt filed a further report in response to the evidence presented by the objectors’ experts. [14] To the extent necessary, it is described and discussed below.

[32] Ms. Caroline Rioux, Vice president, Operations, CMRRA, Ms. Lori Ellis, Director of Operations, CMRRA, and Mr. Joël Martin, Manager, IT Services, SODRAC discussed the challenges CSI has encountered in applying the tariff, especially in identifying, among musical works used by online services, those that are in the repertoire of CSI.

[33] Dr. Marcel Boyer, Emeritus Professor of Economics, Université de Montréal and Research Fellow, Department of Economics, École Polytechnique de Paris, reviewed and commented [15] on the reports prepared by the Objectors’ experts, Dr. Tasneem Chipty and Dr. George Barker. Dr. Boyer disagreed with their use of certain economic principles. Among other things, he contended that Dr. Chipty misapplied the Nash bargaining solution when using it to determine the appropriate rate, partly by reason that she did not include a fourth party, the record labels, in applying this methodology. In Dr. Boyer’s opinion, this failure to include one of the four parties who provide an essential input to the online services distorted the outcome. Dr. Boyer also argued that Dr. Barker misunderstood the methodology applied by the Board in 2007.

[34] In Dr. Boyer’s opinion, both experts failed to take into account that iTunes is a loss leader; in other words, Apple accepts making only a modest profit from its online music sales because those sales allow it to make more profit on complementary devices such as iPods. He therefore cautioned in using iTunes’ accounting data as an indication of its profitability.

B. SOCAN

[35] Dr. Stanley J. Liebowitz, Economics Professor at the University of Texas, filed a report [16] that updated the calculations found in SOCAN 22.A (2007). He argued that increases in the penny rate and the further-enhanced profitability of the digital market relative to the proxy CD market warrant an increase in the rates. Dr. Liebowitz also was of the opinion that music is as important in video-clips as in audio-only files. He therefore argued that the rate should be the same.

[36] Dr. Liebowitz also argued that the services’ advertising revenues are as dependent on the ability to use music as revenues generated by the sale or rental of music. For this reason, he proposed that all revenues be included in the rate base. Finally, he contended that since digital sales and revenues have increased exponentially since 2006, the online music industry no longer requires a phase-in discount.

[37] Dr. Liebowitz also criticized the reports of Dr. Barker and Dr. Chipty. In his opinion, neither offered a basis for the rate reductions they proposed. He contended among others things that Dr. Chipty failed to recognize that the reproduction and communication rights are distinct and that each right requires compensation. He rejected Dr. Barker’s claim that the application of new data to the Board’s methodology yielded a lower profitability figure than that put forward by SOCAN, arguing that Dr. Barker failed both to apply the Board’s methodology properly and to use adequate data. In his opinion, Dr. Barker was also incorrect in his claim that the Board had made mistakes in translating the penny rate into a percentage of retail revenues for permanent downloads. Finally, in response to these experts’ assertion that the certified rates were unsustainable, he noted that both failed to acknowledge that since the royalties paid by online music services for the use of the record labels’ sound recordings are the same in Canada as in the United States, they are presumptively too high in Canada, for at least two reasons. First, the licence American labels provide includes the right to reproduce the musical work, when in Canada, it does not. Second, no one is required to pay royalties for the communication of music in the United States because such a right does not exist; in Canada, that right does exist and must be paid for. [17]

[38] Mr. Marc Paquette, Team leader, Media/Customer Operations/Licensing Department, SOCAN, spoke about his experience in administering the current tariff. He explained that since the tariff for on-demand streams is set at a percentage of subscription fees, it does not account for on-demand streams that are available for free from services that are not subscription-based. To address this, he proposed that where a service offers free streams, the tariff provide a set fee per file streamed for free.

C. The Objectors

[39] Apple and the Cable/Telcos commissioned Dr. Tasneem Chipty, Vice President of Charles River Associates, to provide an economic analysis of the appropriate tariffs to be paid for the sale of permanent downloads and of video-clips. [18] Dr. Chipty concluded that the royalties for the bundle of rights in permanent downloads should be no higher than what the record labels pay to reproduce musical works on CDs.

[40] In reaching her conclusions, Dr. Chipty relied on two approaches. First, she applied competitive principles to the CD proxy, arguing that, if competitive prices apply, the combined rate for reproduction and communication for online music could not exceed the reproduction rate for CDs. Second, she used the Nash model of economic bargaining to develop a rate as if CSI and SOCAN had bargained with the online music services. These approaches yielded combined rates ranging between 7.3 and 8.8 per cent of the retail price of permanent downloads. [19] To derive a rate for the communication of musical works in video-clips, she adjusted the same rate for permanent audio downloads downwards to take into account the non-music contributions (choreography, acting, script, etc.) that add value to video-clips, which are typically sold at a higher price than audio files. She concluded that the rate should be between 0.7 and 2.2 per cent of a clip’s retail price.

[41] In a further report, [20] Dr. Chipty analyzed the reports of Messrs. Audley and Hyatt and of Dr. Liebowitz. She contended that the authors of the Audley/Hyatt report made an error when they translated the new penny rate into a percentage of the average price of a permanent download. She also argued that Dr. Liebowitz overstated the percentage of retail revenue that Apple pays to the record labels. She also updated the figures in her initial report to reflect the rates in force since 2008. She finally proposed that the rate base remain the same and that the phase-in discount be maintained.

[42] Mr. Eddie Cue, Vice President of Internet Services at Apple, testified that the online distribution of music is a low margin business. Significant costs are involved, many of which do not exist in the physical CD world. [21] He denied that iTunes is a loss leader. He agreed that iTunes and iPods are complementary, that one cannot exist without the other but maintained that this does not of itself support a loss leader strategy. He testified that his mandate is to operate at a profit and that the price of iTunes downloads is a profit-maximizing price. Mr. Cue also testified on the impact of the current minimum fee structure on the effective rates Apple pays, which end up being higher than the certified rates. He proposed an alternative formula for the calculation of minimum rates as they apply to bundles.

[43] The Cable/Telcos presented a panel consisting of Mr. Nauby Jacob, Vice President, User Experience & Content, Bell Mobility, Mr. Upinder Saini, Vice President, Product Development, Rogers Communications and Mr. Justin Jamieson, Product manager, Telus. Each witness provided a general description of the products his company offers, how customers use their handsets to access music and the prices at which music is sold. Each also provided financial information which demonstrated that online music is a profitable business line for Bell, Rogers and Telus. Each denied that music sales can be used as a loss leader for the service he represents. Music sales cannot increase profits made on the sale of handsets, since the price of handsets is subsidized. Neither does music drive the sale of data plans: customers buy data plans for uses that are essentially unrelated to music (e-mails, browsing on the Internet, etc.).

[44] The Cable/Telcos also filed the witness statement of Mr. Alain Maynard, Vice President Administration and Finance, Groupe Archambault. The statement dealt with the financial performance of the Zik.ca music service, which has not been profitable since its creation. Regarding the possibility of selling music files as a loss leader, the statement noted that since Archambault does not sell devices, it has no interest in pricing its online music below cost, to the detriment of Archambault’s own physical CD business.

[45] Testifying for CRIA, Dr. George Barker, Director of the Centre for Law and Economics at the Australian National University presented a report [22] whose purpose was to determine the appropriate rates for the reproduction and communication rights in permanent downloads. Dr. Barker described the broad economic approaches that may be used to determine the permanent download rate. He concluded that a comparison with foreign rates, which he labelled the Comparable Actual Market Sales (“CAMS”) methodology, was the most suitable way to determine how much online services should pay for the bundle of rights. He then used a modified version of the Board’s methodology to derive a rate for the reproduction right and then applied the difference as a rate for the communication right. As an alternative to calculate the SOCAN rate, Dr. Barker proposed to use the Board’s methodology based on increased profitability, even though he considered that this methodology was inappropriate and contained several errors.

[46] Dr. Barker filed an additional report [23] in reply to the Audley/Hyatt and Liebowitz reports and refined certain statements contained in his original report.

[47] Mr. Graham Henderson, president of CRIA, described the current state of the Canadian music industry, spoke to the impact of unauthorized peer-to-peer file sharing and explained CRIA’s continuing efforts and investments in trying to create a viable market framework for online music.

[48] Finally, Mr. Kevin Dale, Vice President of Finance for EMI Music Canada, Mr. Daryl Short, Vice President of Finance for Sony Music Entertainment Canada, Mr. Mark Jones, Senior Vice President for Universal Music Canada and Mr. Neil Kerr, Vice President of Finance for Warner Music Canada provided explanations on the financial data they provided to Dr. Barker for the purposes of his report.

V. ANALYSIS

A. Permanent downloads

i. CSI

[49] In CSI – Online Music Services (2007), the Board used as a proxy for the permanent download rate the penny rate record labels pay to reproduce musical works onto prerecorded CDs, converted to a percentage of the retail price of a download (the “MLA model”). Messrs. Audley and Hyatt propose using the same approach with updated figures.

[50] The first step in this approach requires computing an average rate using both the rate contained in the Mechanical Licensing Agreement (“MLA”) between CRIA and CMRRA, and the rates paid to SODRAC by CRIA members, by members of l’Association québécoise de l’industrie du disque, du spectacle et de la vidéo (ADISQ) and by record labels that are not members of either association. [24] The average of these rates, weighted by revenues (the “blended penny rate”), is $0.082 for 2008 and 2009 and $0.084 for 2010.

[51] The next step involves converting the blended penny rates to percentage rates. This is achieved by dividing the rates by the average price per track paid by purchasers of permanent downloads. The average price is calculated by dividing the total amounts paid by consumers for all tracks downloaded in 2009 by the number of tracks downloaded; CSI can derive these figures using the information supplied to it pursuant to the CSI Online Music Services Tariff, 2005-2007. The average price per track of $0.7818 yields rates of 10.0 per cent for 2008 and 2009 and 10.3 per cent for 2010.

[52] To decide whether to reuse the MLA model requires us to answer two questions. The first is whether this approach is still sound. It is not because a proxy served to set a tariff of first impression that the proxy must necessarily continue to be used thereafter. Once set, a price can gain a life of its own. [25] In this instance, this reservation does not apply. The MLA model uses as a proxy the price paid for the same input (the reproduction right) in a competitive market (retail CDs). It seeks to treat the physical and digital worlds equally. This has a sound basis in economic theory. In a competitive equilibrium, the price of downloads and the price of CDs are such that the demand for downloads equals the supply of downloads and the demand for CDs equals the supply for CDs. Competitive equilibrium also has implications for the prices of the inputs used in both markets. In particular, since the right to reproduce a musical work is an input to both a CD and a download, its price, which is a cost from the perspective of both suppliers of CDs and downloads, will tend to be the same in both markets.

[53] According to economic theory, the equalization of costs should occur in dollar terms. This is what the Board expressed in concise, layman’s language in CSI – Online Music Services (2007): “[w]e want the tariff to generate the same amount of royalties as if services paid 7.8¢ per track.” [26] Still, there are at least two advantages to converting this penny rate to a percentage. First, it avoids having to make short term corrections to account for short term market fluctuations. Second, this is what the parties asked for both in 2007 [27] and now, and we see no need to interfere with this choice unless doing so entails serious negative consequences to the parties or to others. These advantages remain as relevant today as they were in 2007. Thus, the approach is still sound.

[54] Second, we must ask whether the data used in the formula ought to be updated. Not every change in the data used to determine a tariff should occasion a change in the tariff rate. Adjusting royalties requires balancing a number of considerations, as can be seen from decisions where the Board declined to do so. [28] No adjustment should be made if the relevant data has a tendency to fluctuate seemingly randomly around a fixed figure. [29] Tariffs should reflect longer-term market trends, not short terms fluctuations, so as to provide stability and predictability and to avoid uncertainty and adjustment costs. [30] Updates should not be made if fluctuations in the proxy market price are unrelated to conditions in the market under consideration.

[55] In this instance, we are asked to update both the numerator (blended penny rate for mechanical reproductions) and the denominator (average price per track downloaded). There are sound reasons to do so.

[56] The numerator is a price expressed in pennies per track, not in percentage terms. Such prices naturally tend to rise, if only to keep pace with inflation. The increase in the blended penny rate is about two per cent per year, which is consistent with inflation trends in Canada. Failing to adjust prices expressed in pennies or dollars for inflation erodes the value of music. [31]

[57] The denominator is also a price expressed in pennies per track. However, the reason to update it is different. In 2007, the Board computed the average price per track using several assumptions [32] based on the evidence filed by the parties, but not on actual sales data. [33] In this case, we know actual sales as reported to CSI for 2009. [34] This implies an overall average price per track of $0.818 [($1.0401 × 0.5018) + ($0.5939 × 0.4982)].

[58] The MLA model remains appropriate, however, both the numerator and the denominator must be updated, the first to reflect an important trend in the pricing of mechanical royalties and the second because we prefer using actual data over assumptions, especially if these assumptions are contradicted by those data. CSI requested 9.9 per cent, as found in its proposed statement of royalties, even though the model generates 10 per cent in 2008 and 2009 and 10.3 per cent in 2010. [35] The rate we certify is 9.9 per cent. [36] This is an increase of 12.5 per cent compared to the previously undiscounted rate of 8.8 per cent decided by the Board. This rate, as well as all other rates we certify in this decision, can be found in the Table in the Appendix.

[59] In our opinion, for the reasons that follow, none of the Objectors’ proposed models is reasonable.

a. Dr. Barker’s proposed models

[60] Dr. Barker considered that foreign rates should be used to cap the royalties paid for the bundle of rights. However, for CSI specifically, both of the rates he proposed were based on the MLA model. He accepted starting with the average rate computed by Messrs. Audley and Hyatt for the reproduction of music on CDs. He then proposed making two adjustments to this rate.

[61] The first would account for an apparent paradox in the way the Board converted the penny rate into a percentage rate: while the numerator (mechanical royalties) comes from the physical market, the denominator (average price per track downloaded) comes from the digital market. Dr. Barker expressed the opinion that in using data from different markets in the numerator and denominator, the Board either had made a mistake [37] or should have better explained how and why it used these different bases. To account for this, Dr. Barker proposed to discount the penny rate by the relative price of digital albums to physical albums. The difference being 29 per cent, the discount yielded a putative penny rate of $0.059 per track. Dr. Barker then applied the average price per track of $0.89 to obtain a rate of 6.6 per cent for CSI.

[62] This approach is dependent on the proposition that in CSI – Online Music Services (2007), the Board should have discounted the mechanical royalties for CDs to account for the fact that digital tracks sell for less than physical tracks. Messrs. Audley and Hyatt explained why this proposition is unsupported on the evidence or by the text of the 2007 decision: the Board wished to set the same price in pennies, not in percentages. [38] We agree, [39] but would add two comments.

[63] First, the proposed approach appears to assume that the rate set in the MLA is implicitly related to the retail price of music. The fact that this rate has continued to rise while the price of CDs has fallen is evidence against this assumption.

[64] Second, this approach assumes that the price paid for the reproduction right should be a constant share of the retail price of the file or medium used to deliver music. In fact, the penny rate for the reproduction right has always been the same for all types of media. [40] Furthermore, in many other countries to which Drs. Barker and Chipty referred to, the price paid for the reproduction right is not a constant share of the retail price. We agree with Dr. Boyer. If two apartments have the same size but different values, it does not make any sense to charge an amount to paint these apartments proportional to their value. This analogy applies to the MLA rate as well. [41]

[65] Accordingly, we reject this adjustment.

[66] Dr. Barker proposed a further adjustment to reflect his opinion that the rate set in the MLA is a non-competitive rate negotiated:

[…] in “the shadow” cast by the Copyright Act, i.e. CMRRA has the option of applying to the Copyright Board to have a rate imposed. On this basis a negotiated rate in Canada must be treated as an outcome that is likely to have been seen by CMRRA as ensuring a rate that was higher than could have been achieved in the regulatory process – net of regulatory costs including regulatory risk. [42]

To account for this, Dr. Barker proposed assuming that mechanical royalties should have remained a constant fraction of CD price over time. This, he stated, would ensure that royalties fell with CD prices, thereby better reflecting market trends, including the emergence of piracy. [43]

[67] There are two difficulties with this approach. First, it assumes that the general regime in the Act (the “shadow” Dr. Barker refers to) inherently tends to favour collectives operating under it. However, this is offset because users have the same option as collectives to ask the Board to set prices under the arbitration regime. Furthermore, all the facts that would justify a lower rate in Dr. Barker’s opinion were public information, including the fact that CD prices were falling. Accepting that this information was not taken into account in setting the mechanical royalties is hard to believe given the record labels’ highly sophisticated negotiation skills. Second, a negotiated price normally is somewhere in the range between the minimum price the seller is willing to accept and the maximum price the buyer is willing to pay. It is possible that such a range did not include the decreasing series of prices proposed by Dr. Barker.

[68] Thus, we also reject this adjustment.

b. Dr. Chipty’s models

[69] In 1950, Dr. John Nash described the axiomatic foundation of a very simple bargaining rule. If the (two) parties to the bargain are equal in all respects (preferences, patience, and bargaining power), the amount to be shared should be divided equally. Further refinements of this model allowed for the possibility that parties could remove costs from the amount to be shared and that the bargain or game could involve more than one player. Dr. Chipty proposed to set the tariff rates using this formal, mathematical model of bargaining.

[70] Dr. Chipty’s bargaining model involves three players – SOCAN, CSI, and online services who offer permanent downloads for sale. The amount to be shared is the revenue of the services associated with permanent downloads, after removing four amounts: what the services pay to the record labels for the “master rights”; [44] the services’ other costs of operation (such as operating servers, producing and distributing gift cards); the services’ normal profit; and an amount to compensate CSI for mechanical royalties lost when a digital track is sold instead of a physical CD. [45] As in Nash’s original model, the surplus is split equally among the three players.

[71] Dr. Chipty fixed the four amounts just mentioned using data from Apple and making assumptions as to what normal profits ought to be (either 0 or 7 per cent) and the proportion of CD sales that would be lost to digital sales (either 33 or 50 per cent). [46]

[72] We have decided not to use this model, for six reasons: two relate to the need for the record labels to be one of the players in the bargaining game; two relate to other issues in the design of the game; and two relate to data and how the game is parameterised. The fact that SOCAN no longer is part of the model as a result of ESA [47] has no effect on our analysis.

[73] First, according to the model as specified, the surplus could be negative from time to time, even for reasonable values of the model’s parameters. This entails a zero tariff. A properly specified Nash bargaining model cannot yield negative benefits for any of the players: any player confronted with the certainty of a negative profit will simply walk away. [48] Therefore Dr. Chipty’s model necessarily omits one or more important variables: put more technically, the model is per se misspecified. One source of this misspecification is evident: the record labels are not included in the game. [49]

[74] Second, the omission of the record labels from the game is unfair to CSI. Since the labels are absent, the amount they receive is treated as a cost to the services and subtracted before any division of surplus; as a result, CSI receives smaller royalty payments, as Dr. Chipty readily admitted when questioned by the Chairman of the Board. [50]

[75] Third, to remove an amount corresponding to CSI’s lost mechanical royalties from the surplus to be shared is an oversimplification. Mechanical royalties are negotiated between CSI and the record labels. The MLA “game” is connected to the “game” for online music royalties. Treating the two as separate is inconsistent [51] and leads to an incorrect tariff. Indeed, CSI could play each game with an eye on the other: for example, it could seek a higher penny rate in the MLA in exchange for concessions that are relevant to CDs but not to permanent downloads. [52]

[76] Fourth, it is inconsistent to remove the fixed (or quasi-fixed) costs of the services from the surplus to be shared but not those of CSI. The reason for removing fixed costs is that they have to be incurred regardless of how many permanent downloads are sold, thus they should not affect the sharing rule.

[77] Fifth, removing normal profits raises significant practical issues. The concept of normal profit is not controversial for economists. However, measures of normal profit can be, as one can see from the fact that Dr. Chipty presented calculations based on two amounts of normal profit. To the extent that normal profit should be removed from the surplus to be shared, far more substantial effort needs to be made to obtain a reliable value.

[78] Sixth, it is not appropriate to treat CSI as a residual claimant, who receives nothing if the industry is not sufficiently profitable. Profit data are inherently suspect, since online music services operate as business lines within larger corporations. In addition, data from business lines are never audited; they can thus be subject to manipulation for the purpose of minimizing the liability under the tariff. Treating CSI as a residual claimant gives an incentive to the services to increase the costs attributed to online music (instead of other business lines), so as to decrease their profits.

[79] Dr. Chipty proposed a second approach which relies on a similar assumption to that used by Dr. Barker. She proposed that the sum of the CSI and SOCAN rates not exceed the rate obtained by using the MLA model without specifying how to allocate royalties between the two collectives, based on her opinion that competitive market forces would drive the royalties payable in the digital market to those payable in the physical market. Put another way, she assumes that what online services pay for the use of musical works should be the same without regard to the number of (copy)rights being used.

[80] This approach ignores a principle the Board has expressed on many occasions: all things being equal, using two rights should cost more than using only one. [53] As a result of ESA, [54] however, this no longer is relevant in this instance.

ii. SOCAN

[81] To calculate the SOCAN rate for permanent downloads, SOCAN proposed using the same approach as in SOCAN 22.A (2007) using updated data. This model (the “22.A model”) uses two figures. The first is the difference in the record labels’ profits between the digital download and CD markets (“profit differential”). The second is the “gross costs”: these are the sum of the net costs (the record labels’ digital download costs less profits), a “normal return” and the mechanical licensing royalties. The SOCAN rate is the profit differential multiplied by the mechanical licensing royalties and divided by the gross costs. CRIA would accept this approach, using different data.

[82] Were SOCAN entitled to royalties on permanent downloads, which is not the case, we would need to address several issues to determine whether we should reuse the 22.A model: a) whether the model is appropriate for determining a rate for SOCAN; b) whether the logic of the model is internally consistent; and finally c), whether the data requirements of the model are overly burdensome.

[83] The 22.A model calculates the price to be paid for the bundle of rights and then allocates that price between SOCAN and CSI. [55] Since SOCAN has no copyright in respect of permanent downloads, no bundle exists and the 22.A model is therefore no longer applicable. Had the contrary been true, we would have found that using the 22.A model generally, and a bundled approach specifically, would have been inappropriate, for the following reasons.

[84] First, a bundled approach is easier to justify when the Board is able to deal with all the relevant rights at the same time, as is always the case with retransmission. This allows the Board to determine not only what a fair price for the user is, but also what a fair allocation among copyright owners is. In the case at hand, the absence of some of the players makes this difficult or impossible. Without Re:Sound Music Licensing Company and ArtistI being involved in the proceedings, it is difficult to decide how much should be left on the table for them to eventually receive a fair compensation. Because the record labels have chosen to licence their rights directly to the services, a fair allocation of royalties among copyright owners would be impossible if we were to conclude (for example) that the record labels are being paid too much. In essence, the same reasons that lead us to conclude that Dr. Chipty’s bargaining model is misspecified [56] lead us to conclude that a bundled approach is unhelpful in this instance.

[85] Second, the 22.A model uses the record labels’ accounting profits to calculate the price to be paid for the bundle or rights. As we noted in paragraphs 77 and 78, using profits to determine the rate for a collective society is inherently problematic. Profits are an artifice of accounting conventions and can be manipulated. Large corporations, including the record labels, recognize revenues and costs based on complex accounting rules, making profits appear greater or smaller than they would be on a strict cash basis. Using data for unaudited business lines increases the problem. So does using data from different corporate entities whose accounting conventions, while legitimate, may be vastly different. Therefore, to the extent that the formula used to set the bundled rate is of questionable value, the need to use a bundled approach also can be questioned.

[86] Furthermore, tying payments by the services to record labels’ profits is troublesome. These profits vary according to a number of factors, including some which, though unrelated to the value of the underlying rights, would nevertheless influence the price paid for those rights. Demand for online music is likely less income-elastic than demand for CDs, since one can purchase online music in individual tracks. In a recession, as consumer income falls, so will record labels’ profits. Because of the different elasticities, profits will decline more for CDs than for online music. In the 22.A model, this implies that the rate for online music royalties increases even though online music sales decrease.

[87] That the Board used record labels’ profits in 2007 as a proxy to set the communication rates is understandable; it was the best information the Board had at the time. The 2007 model may have been helpful to capture an industry snapshot; we find it unhelpful in following trends, to the extent they exist. Accordingly, we do not feel that it would have been the best way to account for the value of the communication right today.

[88] Third, the question of normal profits, as noted, is not difficult in theory but challenging in practice. To measure them properly, one needs to assess the outside opportunities available to those in the recording industry, as well as the implicit risks being taken by operating in that industry. This calculation requires a considerable amount of information, from inside and outside the music industry. Furthermore, some of that information is confidential, which creates further difficulties we outline below. [57]

[89] Again, given ESA, [58] there is no need to comment further on Dr. Barker’s proposed CAMS methodology, the purpose of which was to derive a SOCAN rate for rights that the Supreme Court has ruled do no exist.

B. Limited downloads

[90] In 2007, the rate for limited downloads was set at two-thirds of the rate for permanent downloads at the request of CSI. CSI now contends that it should receive the same rate for permanent and limited downloads. It relies on the opinion of Messrs. Audley and Hyatt, who find no difference in the relative importance of the reproduction and communication rights in limited and permanent downloads. [59]

[91] In 2007, the Board expressed concerns with respect to the discounted CSI rate for limited downloads:

We remain concerned with the possibility that this approach may lead to some level of double discounting. The lower price charged for limited downloads already reflects a lower value, compared with permanent downloads. A lower royalty rate, intended to reflect a lower value for the right, might be an additional reduction to account for the same lower value. [60]

[92] Conceptually, the main difference between permanent and limited downloads from the perspective of the consumer is that the first represents an acquisition, while the second is equivalent to a rental. In both cases, however, only one transmission of a music file is required. All subsequent plays occur from the copy found on the consumer’s devices, whether or not the reproduction can be played without time limit. As a rule, the market should reflect the lower value of the rental through a lower price, resulting in lower royalties even with a constant rate. Consequently, there is some merit to both propositions that the relative importance of each right is the same in both instances and that the reproduction right should be paid at the same rate.

[93] Absent ESA, [61] two reasons might have led us to again set the rate for limited downloads at two-thirds of the rate for permanent downloads. Both no longer apply.

[94] First, extra value that may otherwise have been attributable to the reproduction right may be removed from the table if it is remitted to SOCAN. SOCAN now will get nothing.

[95] Second, the proposition of Messrs. Audley and Hyatt did not reflect the existing practice in many other jurisdictions. As the Board previously noted, the ratio of reproduction to communication royalties recommended by the International Confederation of Societies of Authors and Composers (CISAC) is generally not the same for both rights. [62] There can be no ratio of reproduction to communication royalties if there is no communication right.

[96] We share the concerns the Board expressed in 2007. Furthermore, as a result of ESA [63] and absent compelling evidence or argument to the contrary, we agree with Messrs. Audley and Hyatt: the rate should be the same for permanent and limited downloads, especially since the reasons that lead us to increase the CSI rate for permanent downloads by 12.5 per cent are just as relevant to limited downloads. Therefore, we make the rate the same as for permanent downloads, increasing it from 5.9 to 9.9 per cent.

C. On-demand streaming

[97] CSI contends that the rate for on-demand streaming should be set somewhere between the rate for limited downloads and half the rate CSI gets from DPAs. This is similar to what the Board did in 2007, with one modification. CSI would discount the DPA rate by half to reflect the Board’s conclusion in 2009 that the pay audio services rate base is a wholesale price while the satellite radio rate base is a retail price. [64] SOCAN seeks the same percentage of the bundle as before (62 per cent) based on the argument that the communication right is more important in streams than the reproduction right. Here again, no objector takes a position on how the royalties should be split between CSI and SOCAN, but agrees with one another and with SOCAN that the bundled rate should be the same for on-demand streaming as for permanent downloads.

[98] Several reasons lead us to conclude that we should not reapply the formula used in 2007 to set the CSI rate for on-demand streams.

[99] First, what DPAs pay to CSI is the result of private negotiations between it and the only two Canadian DPAs. Understandably, CSI treated that information as highly confidential when it filed it with the Board for these hearings. Using these numbers on an ongoing basis to set the tariff for on-demand streams would make it impossible for online services to forecast the amount that they will have to pay for this. Where possible, we wish to set rates such that users will have a good idea as to their future path. [65]

[100] Second, we find it difficult to accept that the market dynamics of a wholesale market and a retail market operate similarly. Wholesale profits are usually higher than retail profits, and wholesale prices typically grow slower than retail prices. Therefore, while comparisons between wholesale and retail markets may be helpful in setting a tariff of first impression, they are less so once a tariff has been certified.

[101] In a recent decision, the Board rejected a formula it had used in the past, but accepted as a starting point a figure coming from that formula:

The significant changes mentioned earlier make it inappropriate, for the reasons already stated, to use the 1991 formula today. That being said, we see no reason to conclude that the application of the formula in 1991 yielded an unfair result. As such, the amount can be used as the starting point for our analysis. [66]

[102] We find the same logic can be applied to CSI’s streaming rate. Thus, the starting point for that rate is 4.6 per cent. In addition, the reasons that lead us to increase the CSI rate for downloads by 12.5 per cent are just as relevant to streams. Therefore, we increase the CSI rate from 4.6 to 5.18 per cent.

[103] The SOCAN streaming rate of 7.6 per cent was previously set as part of a bundled analysis. Whereas the CSI streaming rate was set using a formula we now find inappropriate, the SOCAN streaming rate was set using a formula that is now meaningless. Put simply, there is no bundle because there is only one right when musical works are being downloaded, as the Supreme Court ruled in ESA. [67]

[104] None of the parties offered convincing evidence or argument that the SOCAN rate should be changed. More specifically, SOCAN offered no evidence or argument in support of the proposition that the 62:38 ratio in its favour should be maintained. We also find that none of the reasons that justify increasing the rate for CSI – that is, a change in the penny rate and the availability of actual data on the sale of singles and albums, – are relevant to SOCAN. Therefore, applying the CBC Radio rationale quoted in paragraph 101, we see no reason to adjust the SOCAN rate up or down. The questions of the appropriate rate for SOCAN and the appropriate ratio for the streaming rates can be re-examined in later proceedings.

[105] Absent any relevant evidence, it is not possible to determine whether the non-existence of the communication right for downloads may influence the price of the communication right for streams. For example, what if the ability to transmit downloads and streams were “joint products” in the economic sense of the term, i.e. that costs are shared in developing the ability to communicate? Economic theory suggests that if the market for one of the joint products is eliminated (arguably, declaring that a product does not exist eliminates the market for the product), the price of all other joint products should rise, all other things being equal. We leave this and other valuation issues to later proceedings.

D. Video-clips

[106] A video-clip (or music video) is a short movie that integrates a specific sound recording with imagery. Originally produced for promotional purposes, video-clips have become an object of commerce. Only SOCAN applied for a tariff for the music used in video-clips (only for 2010); CSI did not. SODRAC filed a tariff for video-clips starting in 2010. It will be considered at a later date. CMRRA has not filed a tariff for video-clips because it “by and large does not license music videos in the ordinary course.” [68]

[107] SOCAN argues that the rate for video-clips should be the same as the rate for audio tracks. It maintains that the video-clip is nothing more than a sophisticated presentation of the audio track. Because of the sophistication, the services are able to charge a higher price for the video-clip, effectively price discriminating between those who are price sensitive and those who want a sophisticated presentation of the audio track.

[108] Apple and the Cable/Telcos argue that the entirety of the price differential between the audio track and the video-clip is attributable to non-audio inputs. They propose that the price of a video-clip be set at the ratio of the mid-price audio track to the mid-price video-clip. We agree with Apple and the Cable/Telcos.

[109] As we stated in paragraphs 52 and 53, economic theory teaches that the price in dollar terms paid for the same input will be the same in closely related markets: if, as Dr. Liebowitz suggests, a video-clip is but a sophisticated presentation of an audio file, the two markets are indeed closely related. Furthermore, the comparison Dr. Liebowitz attempted to draw between video-clips and audio files on the one hand, and hardcover and paperback books on the other, is not appropriate for two reasons. First, video-clips and audio files usually are marketed at the same time; hardcover and paperback books are not. Second, it would seem to us that visual inputs are inherently more numerous and important to a video-clip than are “hardcover” inputs to such a book.

[110] Following ESA, there can no longer be a SOCAN rate for permanent downloads. From an economic perspective however, the approach proposed by Apple and the Cable/Telcos is as valid for on-demand streams as it is for permanent downloads. Consequently, we set the on-demand streaming video rates identically, at a proportion of 0.66 of the streaming rate of 7.6 per cent, or 5.02 per cent. That rate will apply to video-only services. Services that offer both audio and video files will pay the same rate as audio-only services.

E. Rate base

[111] By the end of the proceedings, CSI and the Objectors proposed that the rate base remain the amounts subscribers pay for their downloads and streams, while SOCAN asked that the base be expanded to include non-user revenues.

[112] In the long run, all the services’ income linked to the provision of music should be included in the rate base. A correlation almost certainly exists between non-user revenues and users’ enjoyment of music. Viewed another way, online business models are sufficiently diversified that user revenues do not always account for the full value of the music used. That being said, we will leave the rate base as it is for the time being, for the following reasons.

[113] First, non-user revenues appear to accrue largely to sites that offer on-demand streams at very low or zero costs to the user. The minimum fees we certify address this issue for the time being, if somewhat imperfectly.

[114] Second, non-user revenues appear to remain relatively modest. In fact, the evidence suggests that several of Canada’s largest services receive no revenue outside of their sales and subscriptions.

[115] Third, determining the role music plays in attracting non-user revenues is far from easy. Yet doing so is essential in order to ensure that services do not over-pay royalties.

[116] Fourth, the tariff we certify applies to past events. Changing the basis on which past transactions are tariffed may be necessary when important sums are at play. In this instance, however, the amounts involved appear relatively modest, and the administrative burden created by changing the rate base retroactively probably would outweigh any gains made by doing so.

[117] That being said, online services’ revenue sources will no doubt change over time, and the Board will no doubt expect in the future that the parties will provide it with fresh evidence to assess the relevance, significance and feasibility of adding non-user revenues to the rate base.

F. Minimum fees

i. The parties’ proposals

[118] CSI proposes that the minimum fee for permanent downloads for singles and bundles smaller than 15 tracks be two-thirds of the royalties for a single track sold at the average price. It proposes that the minimum fee for permanent downloads for bundles of 15 or more tracks be two-thirds of the per-track royalties for a 15-track album sold at the average price. The average prices that CSI uses are $0.99 for single tracks and $9.99 for albums.

[119] CSI proposes that the minimum fee for limited downloads be two-thirds of the royalties payable on the monthly subscription price. The monthly subscription prices that CSI uses are $10 for limited downloads without portability and $15 with portability.

[120] CSI proposes that the minimum fee for on-demand streams be the ratio of the streaming rate to the limited download rate without portability multiplied by the minimum fee for limited downloads. All of these proposals correspond almost exactly [69] to the approach the Board used in CSI – Online Music Services (2007).

[121] In 2007, the minimum rates for SOCAN were calculated in the same way as for CSI. SOCAN agrees with this approach in all but one respect. For free streams, SOCAN asks for $0.046 per streamed file.

[122] Apple and the Cable/Telcos propose minimum fees only for permanent downloads. They are: for single tracks, 65 per cent of the royalties for a track sold at $0.99; for albums with 2 to 10 tracks, 72 per cent of the single track minimum, multiplied by the number of tracks; for albums with 11 to 26 tracks, 10 times the minimum for the album with 2 to 10 tracks; for albums with 27 or more tracks, 28 per cent of the single track minimum, multiplied by the number of tracks.

[123] CRIA opposed minimum fees regardless of level for the reason that online services have to compete with illegal downloads and that the pricing of online music is continuing to evolve.

ii. Permanent and limited downloads

[124] In 2007, the Board stated that minimum fees are necessary to ensure that rights holders do not subsidize the services’ business models. [70] We agree. We will only add that Apple’s complaint that the effective rates it pays are higher because of minimum fees is quite illogical. Higher effective rates are an inherent characteristic of minimum fees. Indeed, such fees are set because applying the general rate would result in too low a price.

[125] The approach Apple and the Cable/Telcos propose for permanent downloads is intriguing but ultimately deficient. As presented, it suffers from ambiguity. Albums with 11 to 26 tracks (which should be the majority of albums) are to be charged at 10 times the minimum fee for albums for 2 to 10 tracks. But the minimum fee for 2 to 10 tracks depends on how many tracks are on the album. If there are 2 tracks, the minimum fee is $0.062; if there are 10 tracks, the minimum fee is $0.31. One way to resolve that ambiguity is to assume that Apple mean to imply that the fee is 10 times the minimum minimorum, or the fee for albums of two tracks. Even this small fix to the proposal leaves it simply too complex.

[126] For permanent downloads, CSI’s proposal is conceptually simpler and corresponds to the formula used in CSI – Online Music Services (2007). We see no reason to vary the formula for minimum fees from its previous incarnation.

[127] In 2007, the Board set a lower minimum fee for bundles of 13 tracks or more. The collectives now ask that the threshold be raised to 15. While we suspect that 15 reflects the average number of tracks per CD sold, [71] no evidence was offered in support of the proposed change. Mr. Audley, who first proposed that number in the context of the private copying hearings, simply alluded to the new number without offering an explanation for it. Absent any evidence on the issue, we find no reason to effect the change. We are comforted in our decision by an analysis we asked Board staff to perform of iTunes data, which shows that moving the threshold from 13 to 15 results in 20 per cent or more of bundles currently eligible to the lower minimum no longer being so.

[128] CSI’s proposal to retain the same formula for minimum fees for limited downloads is unopposed by the Objectors. As such, we see no reason to propose any change. We compute minimum fees by taking two-thirds of the amount generated by applying the rate to the average price. For permanent downloads, the average prices are $1.0401 per track for a single and $0.5939 per track in a bundle. For limited downloads, these averages are $15 per month for portable subscriptions and $10 per month for non-portable subscriptions.

[129] The minimum fees we certify for CSI are: 3.92¢ per file for permanent downloads if in a bundle and 6.86¢ if not; 66¢ per subscriber for limited non-portable downloads and 99¢ per subscriber for limited portable downloads. We certify these figures even though some exceed what CSI asked for. While the services’ liability for certain transactions may be higher, we have no doubt that their overall liability does not exceed what it would have been if we had granted CSI all it asked for.

iii. On-demand streaming

[130] Currently, minimum fees for on-demand streams are based on those for non-portable limited downloads, adjusted by the ratio between the percentage rate for on-demand streams and limited downloads. This approach is no longer possible for SOCAN, who is not entitled to royalties for limited downloads. Under the circumstances, and absent other evidence, we find it preferable to calculate minimum royalties for on-demand streams using the same formula as for limited downloads.

[131] Minimum royalties for limited downloads depend in part on the average monthly subscription price. We do not know that price for on-demand streams. The formula the Board used in 2007 to calculate the monthly minimum for on-demand streams [(4.6 / 5.9) x $0.374 = $0.292] implies a monthly subscription fee of $9.50 [$0.292 = (2/3) (4.6% x $9.50)]. This is necessarily the same amount the Board used in 2007 to calculate the minimum fee for non-portable limited downloads. Thus we will assume, for the purposes of these proceedings, that a monthly subscription for non-portable limited downloads and on-demand streaming services cost the same. As we noted in paragraph 119, CSI suggested that the average monthly subscription price for non-portable limited downloads now is $10. No one challenged this. As a result, and subject to what follows, we set the minimum royalties for on-demand streaming at $0.3453 [(5.18% x $10) x (2/3)] for CSI and at $0.5067 [(7.6% x $10) x (2/3)] for SOCAN.

[132] On-demand streams sometimes are offered free to the user. Mr. Paquette explained the challenges this creates for SOCAN. Even if visitors are tracked, [72] a SOCAN minimum fee of over 50 cents per subscriber per month certainly is excessive. Furthermore, it may be difficult for a service to identify each unique visitor.

[133] There are two terms used in the tariff whose definition becomes less meaningful in the presence of free streams. The definition of revenue as subscription revenue does not capture the scope of economic activity if most or all streams are offered for free. The definition of subscriber does not fit the business model where free-streaming sites have regular and irregular visitors.

[134] We address this concern in two ways. First, we structure minimum fees to account for both the presence of free streaming and the heterogeneity of the free-streaming experience. Second, we replace the term “subscriber” with the term “visitor”. This allows the tariff sufficient flexibility to account not only for purely free sites, but also for sites that are free up to a consumption threshold and charge a subscription fee thereafter.

[135] We start from the proposition that free streaming services should pay a royalty per stream that reflects the average number of streams per visitor, per month; over that number, the royalties should be capped at the minimum royalty other services pay, as set out in paragraph 131. We assume, in the absence of any other evidence about the number of files streamed per month, that the average visitor streams an album per day or approximately 390 files per month. [73] Dividing $0.3453 by 390 yields a CSI price of $0.0009 per file streamed. Dividing $0.5067 by 390 yields a SOCAN price of $0.0013 per file streamed. With these fees in place, a free site with some visitors that stream a few files and some visitors that stream considerably more would not be unduly prejudiced by the structure of the minimum fee.

iv. Video-clips

[136] The amounts paid to SOCAN for video-clips should be the same as those paid for audio files. This is best achieved by applying the same minimums for the former as for the latter.

G. The 10 per cent discount

[137] In 2007, the Board applied a 10 per cent discount to the rates it set with the intention it would apply only for the life of this tariff. [74] We see no reason to apply that discount this time. While some users have claimed low profitability, the evidence suggests that the online music market has matured and its structure has stabilized.

[138] The practice of the Board is to discount its tariffs when applied to a new industry. However, the Board does not typically do so when an existing industry brings forward a new product. The sellers of audio files are the same as the sellers of video-clips. Accordingly, we do not provide a discount for the video-clip royalties.

H. Total royalties

[139] In these reasons, we outline three key changes relative to the previously certified tariffs. First, as a result of the Supreme Court decisions in ESA and Rogers, SOCAN receives nothing for permanent and limited downloads. Second, we eliminate the 10 per cent, first-tariff discount. Finally, we increase the rates for CSI by 12.5 per cent, applying the MLA model. We estimate that the combined effect of these three changes is to reduce total royalties by 10 per cent relative to the first tariffs.

[140] We have not included in our estimates the effect of the new tariff for video-clips and the new tariff for free streams. The former will increase royalties while the latter will decrease them. However, since we do not have information precise enough about the size of either, we ignored both in our calculations.

VI. Wording, Terms and Conditions of the tariff

[141] The tariffs we certify largely reflect the ones they replace. In the following paragraphs, we outline some of the main differences and we comment some changes the parties proposed.

A. Separate Tariffs

[142] The SOCAN and CSI tariffs are worded identically to the extent possible, so as to minimize any interpretive issues between the two. That being said, the tariff structure and reporting requirements are sufficiently different that we decided to certify separate tariffs for the time being.

B. Technology

[143] Some changes from the previous tariffs are driven by technological evolution. For example, the emergence of smart phones led us, at CSI’s request, to amend the definition of “download” to reinforce the fact that a download can be delivered to a device as well as to a medium.

C. A New Tariff for Free On-Demand Streams

[144] Other changes were required because of the new, separate tariff for free on-demand streams, including the introduction of definitions for “free subscription”, “free on-demand stream” and “unique visitor” and the replacement of the concept of consumer with that of end user.

[145] The tariff is now articulated around three types of online service clients: end users, that is anyone with whom an online music service deals; unique visitors, who receive the free on-demand streams for which royalties are paid pursuant to subsection 5(3) of the CSI tariff; and subscribers, that is everyone involved with a service on a non-à-la-carte basis.

[146] “Unique visitor” is defined as “each end user, excluding a subscriber, who receives a free on-demand stream from an online music service in a month.” CSI wished to specify that unique visitors are identified by a unique IP address. The notion of unique visitor is a common web analytic metric. However, rarely does it rely only on an IP address, due in part to dynamic IP addressing: the same address is assigned to different users in different sessions. A further identifier is needed; this can be a cookie, user agent data or something else. Web sites count unique visitors in different ways, not all of which are known to us. In this instance, the need for consistency is trumped by market reality: we cannot impose a metric that is not universal. Consequently, the tariff relies on industry practices. However, in order that the collectives understand how each service operates, the tariff requires that services disclose their metric for determining the number of unique visitors in a month.

[147] The term “end user” will remain undefined, since we use it in its ordinary meaning.

[148] The previous CSI tariff specified that previews of no more than 30 seconds did not attract royalties. This provision is removed, since such uses almost certainly involve fair dealing.

D. Reporting and Accounting

[149] The existing tariff requires services to report to CSI only files that require a CSI licence. Also, reporting is required whether or not a file attracts sales. By contrast, services actually report all sales of downloads and streams. The parties agreed to reflect that practice in the tariff. We agree that this is the more practical and rational approach.

[150] Making this change allowed us to substantially simplify information reporting, which will continue to occur in three stages. First, only basic, tombstone information will be reported before a service starts operating. Music use information will be provided as part of the second stage. Second, services will report all their sales, but will not provide information about available tracks that have not recorded any sales in a given reporting period. As a result, a track’s existence will be reported to the collectives only after it attracts a first sale. Third, as before, CSI will provide a report that will identify each file as known to be in CSI’s repertoire, known not to be in repertoire or unable to confirm whether in repertoire. The report also will contain a detailed calculation of the royalties payable per file.

[151] The nature and extent of the information services must report remains more or less the same.

[152] Currently, some information is provided only where available. The Objectors asked that this continue to be so. According to them, they already provide everything they can reasonably obtain either themselves or through others, including the record labels. CSI already has this information or is in a better position to obtain it. Making the provision of information mandatory when some is impossible to obtain in practice would turn all services into infringers.

[153] CSI relies on the record of these proceedings to conclude that all the information being asked is essential to the accurate identification of musical works, and that services are highly deficient in complying with the tariff’s requirements: at the time of the proceedings, CSI had been able to identify only 21 per cent of all reported files.

[154] The Board’s position on this matter is clear: if information is essential, a service ought to provide it even if this means getting it from someone else. [75] At some point in time, services that do not supply to CSI all the information it requires to efficiently collect and distribute royalties will be asked to bear the additional costs arising out of this deficiency. Making all reporting requirements compulsory is not the only possible solution, as Apple pointed out. Some foreign collectives offer a discount to users that provide all the information they wish to receive.

[155] That being said, and with some reluctance, this is not the time to change the status quo in this respect, for three reasons. First, the fact that a large number of files remain unidentified may not be as serious in practice as would appear: the 21 per cent of reported files that CSI managed to identify accounted for 59 per cent of all files generating one or more sales, and for 93 per cent of sales by volume. Second, since CSI will now be allowed to cancel the licence of a service who fails to report all required information, increasing the information requirements would turn services into permanent infringers. Third, the question raises too many complex issues about too many tariffs for this to be the proper forum to resolve it.

[156] Consequently, we leave the reporting requirements as they currently are. However, we ask that the Secretariat initiate a consultation process with a view to obtaining reliable, complete information allowing the Board to assess, for all online uses of music, which information is helpful and which is necessary, which can reasonably be obtained from users, where the balance of convenience lies and how to ensure an optimum flow of information between users and collectives.

[157] We rejected CSI’s request to secure through the tariff information not needed for tariff administration that it would have found helpful to argue changes in the rate base in future proceedings.

E. Challenges to CSI Ownership Claims

[158] Currently, a service must challenge a CSI claim that a work is in the repertoire within 20 days of receiving such a claim. The tariff does not specify what happens if a challenge is filed late. We inquired with CSI about the reason for this; for one thing, we wondered what would occur if a service received later on information from a third party confirming that the required rights had been otherwise secured, making the licence unnecessary. CSI argued that removing the time limitation would give services unlimited time to dispute CSI’s representation of the work, which could create serious administrative issues. We see no reason why a service should pay for what is already licensed. We have removed the time limitation. Any difficulty created by this decision can be documented and addressed when the Board next deals with this tariff.

F. Consequences of a Failure to Provide a Report

[159] CSI asked that a failure to provide a timely report have two consequences. First, CSI would be allowed to terminate the licence. Second, the due date for royalties would remain that which it would be had the report been filed in time, thereby triggering interest payments earlier. Objectors took issue only with the second.

[160] According to CSI, the second measure is meant to address an unintended consequence of the existing tariff wording. Royalties are not due until CSI is able to invoice them; as a result, a service can artificially defer the obligation to pay royalties by failing to provide the reports necessary to calculate them. The Objectors argued that this is punitive, unnecessary and duplicative in light of the existing obligation to pay interest on late royalty payments.

[161] The scenario CSI describes should be avoided. In our view, this is best achieved by providing that the invoice sent following a late report is deemed to have been received in time by the service as long as CSI sends the invoice within 20 days of receiving the late report (the delay within which CSI is supposed to send the invoice when the service’s report is received in time).

G. Confidentiality

[162] CSI asked to be allowed to share information received pursuant to the tariff with SOCAN. The Objectors argued that this would be unprecedented and would jeopardize commercially-sensitive information. For the reasons set out in a recent decision of the Board with respect to Re:Sound Tariff 5, [76] CSI will be allowed to share information with SOCAN for the purposes of royalty collection and tariff enforcement.

[163] CSI also asked to prevent services from sharing with others information they receive from CSI pursuant to the tariff. The Objectors did not understand which information CSI may supply pursuant to the tariff that may need to be treated in confidence. CSI answered that section 8 of the tariff requires it to provide potentially sensitive information about the extent of its repertoire.

[164] All that section 8 requires is that CSI indicate: what it knows to be in its repertoire and if so, to what extent; what it knows not to be in its repertoire; and why CSI is unable to determine whether a work is or not in its repertoire. In our opinion, this is information CSI would be required to provide pursuant to section 70.11 of the Act and as such cannot be confidential.

[165] CSI sometimes provides other information that what section 8 of the tariff requires, such as the name of the publisher who owns a share of a work that is in CSI’s repertoire only for part. Assuming that this information is confidential, it is not provided pursuant to the tariff. It is up to CSI to require the services to treat the information in confidence if they wish to receive it.

[166] CSI also fears that since the repertoire is dynamic, online services may share information that is out of date. That is not a sufficient reason to impose an obligation of confidentiality on the services pursuant to the tariff.

H. The SOCAN Tariff

[167] The SOCAN tariff was redesigned to align its wording as much as possible on CSI’s. Three reasons explain the main differences between the tariffs. First, SOCAN is not entitled to royalties in respect of downloads. Second, SOCAN represents virtually all the eligible repertoire, and CSI does not: thus, payment dates are different because CSI invoices the services, while SOCAN does not. Third, SOCAN, not CSI, will collect royalties for music videos (albeit only for streams).

I. Transitional Provisions

[168] The tariff contains transitional provisions made necessary because the tariffs takes effect on January 1, 2007 and 2008, while they are being certified much later.

[169] What CSI and SOCAN proposed was significantly different from the previous certified tariffs. No doubt in part because of this, CSI proposed a complex and lengthy transitional process. Each service would have been required to pay an interim royalty top-up for the period from January 1, 2008 to the date the tariff was certified. During an interim period, royalties would have been invoiced at different rates, depending on whether uses occurred before or after the tariff was certified. A final invoice would have been sent a year or so later.

[170] The Objectors considered the proposal too complex. Transitional provisions should be straightforward. They should provide parties with a sufficient opportunity to make the necessary adjustments to royalties already paid, while placing the onus on the parties to do it once, and to do it correctly.

[171] For CSI, the rates are increased but the tariff structure is unchanged. Online services can calculate themselves how much more they owe for past periods, based on the reports already received from CSI. Ninety days should be sufficient to perform the necessary calculations. There is no reason to impose different delays with respect to SOCAN.

[172] CSI proposed that past free on-demand streams be invoiced in a separate process. The proposal triggered no comments on the part of the Objectors. This is what we have certified in this respect.

[173] Services who file reports pursuant to the previous tariffs before September 30, 2012 will not be asked to file additional information with respect to the corresponding reports required by the new tariffs.

[174] A table sets out interest factors to be used on sums owed in a given quarter and not already payable on an interim basis pursuant to the previous tariffs. Our reasons for setting such factors need not be repeated here. [77] The factors are slightly higher for SOCAN than for CSI, since SOCAN is entitled to its royalties 50 days earlier than CSI.

[175] This time again, arriving at the wording of the tariff we certify proved to be a complicated task. We thank the parties for their considerable assistance.

Gilles McDougall

Secretary General

APPENDIX: TABLE

Royalties to be paid

	Royalties to be paid to CSI (2008-2010)	Royalties to be paid to SOCAN (2007-2010)
Permanent Downloads	9.90% of the amount paid by the consumer   Minimum fee 3.92¢ per file in a bundle of 13 tracks or more 6.86¢ per file otherwise	n/a
Limited Downloads	9.9% of the amount paid by subscribers   Minimum fee 99¢ per month, per subscriber, if portable limited downloads are allowed 66¢ if not	n/a
On-Demand Streams	5.18% of the amount paid by subscribers   Minimum fee Free streaming: 0.09¢ per file streamed per visitor, up to a maximum of 34.53¢ per visitor per month Otherwise 34.53¢ subscriber per month	7.60% of the amount paid by subscribers   Minimum fee Free streaming: 0.13¢ per file streamed per visitor, up to a maximum of 50.67¢ per visitor per month Otherwise, 50.67¢ per subscriber per month
Video-Clips (2010 only) On-Demand Streams	n/a	5.02% of the amount paid by subscribers   Minimum fee Free streaming: 0.13¢ per file streamed per visitor, up to a maximum of 50.67¢ per visitor per month Otherwise, 50.67¢ per subscriber per month