The Relevant Market for Music Streaming: Market Definition and Measurement Challenges
Working Paper. DOI 10.5281/zenodo.6088844
2022-02-15
Executive Summary
In the last decade, global online platforms have fundamentally changed the music business. Streaming music has become a very important distribution channel on various online platforms. It is also the most important music discovery location for the audiences. As music streaming became one of the major forms of music distribution, the disruptive technologies of streaming platforms have raised new business strategy and public policy questions.
Regardless if we take an antitrust, copyright policy or business strategy point of view, in our experience it is impossible to understand music streaming outside of the context of the broader music markets and the music industry. While online platforms have indeed disrupted the music value chain, they did not fundamentally change how music is brought to the market. The revenue streams from music streaming royalties largely follow long-established practices and regulations concerning the valuation of intellectual property, accounting for royalties, and regulating the administration of music copyrights and neighbouring rights. Music streaming competes with other uses of music, such as public performances in television and radio broadcasting, or digital downloads. It cannot be analysed in isolation from the volume and price trends of public performance, mechanical licensing, and home copying.
In our experience, it is impossible to understand the business of music streaming outside the context of the music market or the music industry. Music streaming royalty rates and distribution methods largely follow long-established practices and regulations concerning valuing intellectual property, accounting for royalties, and regulating the administration of music copyrights and neighbouring rights. Music streaming competes with other uses of music, such as public performances in television and radio broadcasting, or digital downloads. It cannot be analysed in isolation from the volume and price trends of public performance, mechanical licensing, and home copying.
What makes understanding music markets so difficult is that they are seemingly in the plain sight of the public and policy makers, but most of the market transactions are invisible. In many advanced music markets more than half of the music uses, i.e., the market volumes are not paid for by the user. Radio stations or the popular UUC platforms like YouTube, TikTok or SoundCloud, but even licensed music streaming platforms like Spotify offer zero-price services for the users. While there is a growing body of consensus and practice on how to incorporate zero price transactions into competition practice, but in our experience the empirical observation of these markets is very challenging. When consumers pay a price, both the quantity of the sales and the price is recorded on the invoice, and this information is translated into tax returns and financial statements. Zero price transactions have no invoices, and no accounting trail.
If we also consider the fact that the supply side of music is made of freelancers, micro- or small enterprises, we must realize that even the aggregated data sources that we can usually take for granted in the analysis of various other economic sectors, such as governmental statistics created from financial statements, mandatory statistical filings or tax returns, is missing. Micro- and small enterprises do not participate in full business statistics surveys, make simplified financial reports, and file simplified tax returns. They usually do not even possess the data processing or accounting possibilities that would allow them to provide information to a competition inquiry with the same level of precision or detail that markets authorities would take for granted in other sectors.
In the past years, there has been an increased interest in analysing online platforms where the consumers do not pay a price. Watching a music video on YouTube, or reading amusing content on Facebook, finding accommodation on AirBnB does not cost anything for the buyer money, though the buyer’s time in viewing advertisement, the buyer’s data footprint, or a commission from a two-sided product (the supply side of a rented room) offers a compensation for the platform. As a recent CMA study pointed out, such practices are probably originated from the use of music and audiovisual content on radio and television (Competition & Markets Authority 2020).
While there is a growing body of consensus and practice on how to incorporate zero price transactions into competition practice, the data collection required for such an analysis requires a strong focus on the underlying, relevant market, and unusually sophisticated data collection plans. Such data collection can be a daunting task if we want to consider all aspects of competition:
music creators are competing with each other for the audiences on platforms;
platforms are competing for subscribers with each others;
platforms are also competing for advertising revenue with other media platforms, with each other, and ad-supported zero-price offerings are competing with user-paid, subscription or other models.
The methodology that we introduce here briefly, in a work-in-progress document, was designed in competition law-based evidence gathering and analysis and provides adequate estimates for various forms of zero-price transaction, including the use of UUC platforms, radio, and home copying.
Our dynamic The Relevant Market for Music Streaming: Market Definition and Measurement Challenges document is a versioned work-in-progress publication, available from the Zenodo open science repository. It is built upon the actual application of competition law in non-UK jurisdictions (Antal 2017a, 2019a), and it is intended for further research and publication. After reviewing some methods to empirically collect data about quantities, prices, market shares and market power in the music markets, we present some ideas on the definition of music markets. After reviewing some methods to empirically collect data about quantities, prices, market shares and market power in the music markets, we present some ideas on the definition of music markets in the context of the music value chain.
The statement of scope for the CMA Music and streaming market study (Competition & Markets Authority 2022) intends to explore a range of questions that require empirical economic evidence. The gathering of such evidence requires a formal market definition (in simple terms: “Who, which sellers and which buyers should be contacted for evidence?”). This is very challenging for two reasons: a) the majority of the uses has a zero-price and no accounting trail b) the sales are global and most music enterprises do not have a sophisticated accounting practice to report their revenues for geographic segments.
The scoping statement claims that “music streaming services are now the predominant means of music consumption, supplanting traditional physical media such as CDs and vinyl.” We believe that this statement should be seen as a starting hypothesis, and not as fact. It is not evident how it can be tested against empirical evidence, and it also appears to contradict how the music value chain is understood in the United States (Hull et al. 2011) or in Europe (Leurdijk et al. 2012; Leurdijk and Ottilie 2012). If we apply the UK version of the relevant market definition (Office of Fair Trading 2004), starting with music streaming as the focal product, it would be very difficult to justify the exclusion of radio and television broadcasting and retransmission as potential substitutes for the use of music streaming services or physical media. We believe that there is sufficient demand- and supply side substitution taking place among licensed streaming, radio broadcasting and retransmission at least that pricing practices of the industry treat these alternative sales channels as market comparators for each other. On the other hand, while the scoping report states the ambition to investigate the potential ‘value gap’ towards UUC platforms and their role in the music market, we would like to weigh in with the evidence of jurisprudence, business practices and empirically observed prices that UUC is rather dissimilar to licensed “music services … and traditional physical media such as CDs and vinyl.”
As the scoping statement correctly describes, the monetization of UUC is fundamentally different from the other licensing models that are governed by international copyright and neighbouring rights law. The UUC service providers, at least, in the United Kingdom, can negotiate with rightsholders after (rather than before) the music is used; opt to take down specific instances of music as an alternative to paying for its use; and do not bear the costs of any unidentified music rights being used.
In the previous decade, the various representative associations of composers (GESAC 2015a, 2015b; CISAC 2017), performers (AEPO-ARTIS 2016, 2017), and producers (Moore 2016) in Europe were all strongly calling for a change in these practices which were made. A re-regulation of UUC in the European Union is under way with the recent EU copyright directive (EUR-Lex 2019), which, after Brexit, is not applicable in the UK—and at this point there is no empirical evidence available from these European markets about the re-regulation (some EU countries have not even adopted the new measures into their copyright law.) However, exactly because there is no regulatory mechanism that would bring the UUC segment closer to “normally” licensed streaming, and one could hypothesise that radio and retransmission is a closer substitute of music streaming than UUC. In our view, any re-visiting of the “value gap” debate in the UK, as foreseen in the scoping report, is a different topic from the rest of the market analysis, or it requires a far broader analysis in all forms of zero-price music use, regardless of their legal basis.
We would also urge for caution with the starting point of the CMA scoping statement that “music streaming services have helped to restore growth in the sector and made it easier for new artists to share their music.” As we show in our paper, the idea of growth is very difficult to verify empirically, and again, it should be formulated more as a hypothesis than as a statement. Our empirical analysis found that growth is located in narrow segments of music—a general devaluation of the music, falling prices may lead to flat or falling revenues even with growing quantities used, and a large part of the music ecosystem may not experience any growth at all. The current heated debate around the re-regulation of music streaming is due to the fact that in large pockets of the relevant market growth has not been present in the last decade.
Our paper proposes our full-market comparators model for the analysis of the UK music markets the use of, which gives a broader view of the music industry, and which has been applied in competition law-based analysis as well as in price setting in other jurisdictions (relevant geographical markets.) It is based on the widely accepted value chain of the three income streams model (Hull et al. 2011; Leurdijk et al. 2012; Leurdijk and Ottilie 2012) and actual pricing practice in the industry, which are in turn based on more general pricing models of intellectual property (PwC 2008; IFRS 2011; Flignor and Orozco 2006; Puca and Zyla 2019). Yet even the application of this comprehensive model is difficult when the ambition of understanding market power is considered. In the recent year, it has been pointed out many times that the traditional market definition of the relevant market is difficult at best with two-sided platforms like YouTube or when users are facing zero prices. The SSNIP test can be, theoretically, applied in the cases of YouTube, radio stations, or ad-supported streaming, however, the actual analysis is very difficult, because the implicit price used must be based on the advertisement revenues (OECD 2018; Hausemer et al. 2021). Our full-market comparator does exactly this, but the estimation problems are considerable.
In the market where we have conducted detailed surveys, the biggest volume of music was used in zero-price forms–mainly in radio, followed by UUC streaming, home copying, but also ad-supported music streaming had measurable market shares. The music use overlaps with commercialized forms of music consumption, and participation on non-commercialized or even illegal uses (such a liturgical music, authentic folk music, and some no longer protect classical works, or home copying, which is exempted from copyright law in most European markets, but not in the United Kingdom.)
When consumer prices are zero, the implied prices must be calculated based on the revenues collected by the platform to finance this supply and the volume of the consumption. One problem may be that the relevant revenue is not available–for example, YouTube is not calculating or making available its advertising revenues for most national jurisdictions, because those revenues fall below the threshold where such disclosure would be mandatory under its accounting standards.
As one can see from our graphic figures, the music business has a particularly complicated value chain, because music is a copyright-based intellectual property with a very complex licensing system. Each music recording is a bundle of copyrights and neighbouring rights, which are often remunerated separately. Furthermore, in terms of the largest volume of use, the user is not paying for it directly. This makes the empirical analysis particularly challenging.
Following best statistical practice, we use the ICET model and the so-called cultural access and participation surveys to capture the use of paid and not paid, commercial and non-commercial uses of music in a systematic way (Bína, Vladimir et al. 2012; Haan and Adolfsen 2008; Haan and Broek 2012). While CAP surveys give a biased measurement of use hours, we do not see evidence that they give a biased estimate of use ratios, which is necessary for our models’ weights. With almost no exception, in the music markets we could observe or estimate revenues, but not prices. We always gave a priority to the use of observable market transactions (IFRS 2011; EUR-Lex 2012), but these transaction were usually not available in full detail. Our model’s understanding of price (which often cannot be directly observed, only calculated from revenues and volumes) was partly based on Valuing the use of recorded music, an excellent methodological guide created by PriceWaterhouseCoopers for IFPI (PwC 2008), and the various globally applicable WIPO and IFRS standards (IFRS 2011; Flignor and Orozco 2006; Puca and Zyla 2019) on valuing copyrights in more practical terms for the music industry.
We assume from the language of the scoping report that the DCSM’s working definition is that of a UK national relevant market. In our experience with competition law-based market analysis in other jurisdictions, this is indeed the best point of view, but poses its own challenges. As we have shown in the Music Creator Earnings’ Project initiated by the UKIPO, the British music industry is so interwoven with the national music industry, that it is next to impossible to obtain relevant market figures for the United Kingdom alone.
The new platforms that make music available for literally billions of people are global: YouTube can be accessed almost anywhere on Earth, and Spotify is selling subscriptions in more than 200 countries and territories (Spotify 2022). They are competing with radio and music television broadcasting and retransmission which is licensed territorially, and very often to less than national territories—for example, Ofcom licenses commercial radio stations to as small territories as “West Central Scotland.” From an economic or competition law point of view, zero prices are difficult to work with. A geographically unrestricted access makes the empirical observation or estimation of prices very challenging.
Most undertakings of the music industry, wherever they are located in the value chain, are micro- or small enterprises, and therefore enjoy the benefits of simplified tax returns, exemption from most mandatory statistical filings, and the disclosures of simplified financial statements. Unfortunately, these are the actual sources of official government economic statistics, which, therefore, exclude the music industry. The best-known music industry reports are created by membership-based trade organizations with surveying these microenterprises. The management information systems and accounting systems of these microenterprises are usually not sophisticated enough to record revenues by geographic segments, or to apply an auditable use of currency translations — in fact, most music enterprises are so small that they do not audit their disclosures. In business practice, often a mixture of global income is reported with an unknown currency weighting. Some revenues are earned in GBP, others in the major currencies, i.e., USD, EUR, JPY, but even a small UK entity may have some revenues that are earned in more than a hundred other national currencies.
As we have stated above, price observation is directly not possible for most of the uses. However, even the indirect calculation, after cleaning the exchange rate effects, requires further special royalty accounting know-how. The way royalties are accounted for and earned is different in the various sales formats of music. Mechanical royalties are paid upfront in lump sum, public performance royalties are paid per annum, and streaming royalties are accrued until reaching a threshold. In practice this means that many UK music business reported streaming income very difficult to connect to a consistent timeframe, geographic segment, and a proper weight of exchange rates. We there are many arguments for analysing music streaming within the geographical scope of a national UK market, we demonstrate that such an analysis is easier said than done. In our paper we draw the attention to computer simulations that we made for the Music Creators’ Earning report to show various pitfalls, particularly not considering the exchange rate effects of rightsholder income.
In our experience, most music undertakings and their trade associations are not equipped with the data processing, accounting, or statistical capacity to adequately report revenues by geographical segments. This means that there is no easily verifiable source of the domestic, national and export income, and reliance on the industry’s internal reports, which are clearly not compiled for a competition law-based inquiry, is risky.
Some of these problems are solved by the Digital Music Observatory, a working demo of the planned European Music Observatory is solving. It grew out of the Central & Eastern European Music Industry Databases (CEEMID) initiative in 2014, in which righthsolders from three countries attempted to solve this problem (Artisjus et al. 2014; Antal 2017b). By 2019, CEEMID had collected information on 20 European markets, including the United Kingdom, and processed data on far more markets—this data has been used in various competition law-based analysis and modelling outside of the United Kingdom.
The idea of this observatory was brought to the UK policy debate on music streaming by the observatory’s only (former) British users, via the Written evidence submitted by The state51 Music Group to the Economics of music streaming review of the DCMS Committee (state51 Music Group 2020; Antal 2020). The state51 music group, through its distribution arm, has been supporting the creation of the largest ever European market report, the Central European Music Industry Report, and supported the creation of the CEEMID-CI indexes, which, for the first time provided a stock-index type of view from an individual rightsholder’s perspective on volume and price movements in the UK and in other countries (Antal 2020). The state51 music group drew attention to the observatory approach and this work in the Digital, Culture, Media and Sport Committee (DCMS) Select Committee of the British House of Commons. The Music Creator Earnings’ Project, that created the Intellectual Property Office’s Music Creator Earnings’ in the Digital Era report individually contacted the Digital Music Observatory (successor of CEEMID) and state51, and eventually with the permission of state51, the project commissioned our report that re-uses the CEEMID-CI indexes. The MCE project also committed to share data in the Digital Music Observatory.
This paper grew out of the analysis we have provided for the Music Creator Earnings’ Project, hat created the Intellectual Property Office’s Music Creator Earnings’ in the Digital Era report (Antal 2021b). We were tasked with providing longitudinal analysis of earnings development and relating our findings to equitable remuneration. The starting point of our work was centred around a very broadly defined problem: how much money music creators (rightsholders) earn from streaming, how these earnings are distributed, and how the earnings and their distribution have developed during the last decade. Eventually that analysis only got a single paragraph and a footnote in the Music creators’ earnings in the digital era report and was published as an accompanying document independently of the main report, An Empirical Analysis of Music Streaming Revenues and Their Distribution.
We hope that our paper will be a useful contribution to understanding the music markets of the United Kingdom, and contribute to several important questions:
We hope that our full market comparator model is a useful tool to map all the quantities, revenues and implied prices of the music value chain, and understand the potential of supply- and demand side substitution in the market;
Provide appropriate estimation techniques where not accounting or transactional information is available due to the application of zero prices for both indirect prices and quantities;
Provide estimation techniques for correctly attributing the revenues of small undertakings to a consistent timeframe and geographical segment and applying the correct exchange rates.
We hope that our analytical framework can bring much clarity to the relative performance of the UK undertakings domestic and export markets.
This document is a work in progress. Please cite this version as:
Daniel Antal. (2022). The Relevant Market for Music Streaming: Market Definition and Measurement Challenges (Version 20220215). Zenodo. 10.5281/zenodo.6088844
Our refer to the Github repo: https://github.com/dataobservatory-eu/music-competition