Start me up : new entry of record labels in the music industry

(1)

Start Me Up: New Entry of Record Labels in the Music Industry

Master Thesis

MSc Business Administration

Management and Entrepreneurship in the Creative Industries

Faculty of Economics and Business, Amsterdam Business School

Myra Alice Wilhelmina Ruers - 11903929

Supervisor: Dr. M. Piazzai

(2)

Statement of Originality

This document is written by Student Myra Ruers, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

The music industry is extremely vibrant, diverse, and inherently uncertain. Record labels are the main intermediaries in this industry, facilitating the connection between the creative input from musicians and the demand from the consumers. Due to many technological advancements, the barriers to entry have decreased significantly, hence making it easier to enter this industry as a new record label. Additionally, strategic integration options allow existing firms to start new ventures by acquiring or initiating subsidiaries. This research focuses on the multiple modes through which new entry can be commenced. These entry modes all have their own characteristics and consequences for the configurations of the industry. The likelihood of entry, the subsequent performance of new record labels, and the effects on the diversity of the output are the main topics that are explored. In this study, databases storing longitudinal information on musical releases are used to compile a dataset with the necessary information on new record labels. The main outcomes reveal that there are still many organizations entering the industry despite a general trend towards disintermediation, and that the widely accepted categorization that divides record labels into two groups (major and independent) is not completely accurate, because large independent labels tend to act similarly to major labels.

(4)

List of Figures and Tables

Figure 1. Music Industry Value Chain (Bockstedt et al., 2006) ... 13

Table 1. Overview of hypotheses……..………26

Table 2. Descriptives H1b ………... 32

Table 3. Results one-way ANOVA ………. 33

Figure 2. Short-term and long-term means per label type ... 35

Table 4. Descriptives Billboard songs and weeks ………37

Table 5. Results independent samples t-test ……… 38

Table 6. Descriptives H2 with original grouping ……… 38

(6)

1. Introduction

The creative industries are plagued by the tension between art and commerce (Caves, 2000). In the music industry, this distinction is also apparent. The art side, which consists of the artists and musicians who want to follow their heart and make the music they love, is contrasted against the commerce side, which consists of the businesses that are trying to fulfil corporate goals. These conflicting objectives create pressure on both parties and make fruitful collaboration often harder to achieve than in other industries. Additional factors that complicate the cooperation within the music industry, are the persistent oversupply and the inherent uncertainty that are present in the creative industries (Hirsch, 1972). This implies that there are many more songs and albums released than there is demand from consumers. This leads to extreme uncertainty about the chances of success for artists and bands (Peltoniemi, 2015).

The music industry configurations and relationships are extremely interesting to study because the environment is very vibrant. A lot of research has already been undertaken about the music industry in the past years (Burnett, 1996; Negus, 1999b; Lopes, 1992; Dowd, 2004a). Despite these efforts, literature concerning this industry from a business perspective has largely focused on the beginning and the end of the value chain, namely the artists and the consumers. The intermediaries that connect these two ends are often overlooked. The main intermediary in this industry is the record label. Record labels are extremely important actors in the music industry value chain, as they are responsible for discovering and developing great artists. They also arrange the publishing, marketing, and licensing of the musical recordings from the artists they represent (Bockstedt, Kauffman & Riggins, 2006). According to Hirsch (1972), the record label fulfils the position of cultural gatekeeper in this creative industry. This implies that they are tasked with the selection of the musical outputs that deserve coverage, by choosing those musicians and bands that are of high enough quality to be heard by a large audience. Moreover, main topics that seem to be emphasized in music business literature concern intellectual

(7)

property rights challenges, and the effects of disruptive innovations on consumer behaviour. Whilst little attention has been given to the strategic aspect concerning the intermediaries. This subject is very valuable to study since the competitive field consists of a great variety of players, for whom the stakes are high.

In the past years, many changes have had a great effect on the overall music industry, mainly due to technological innovations (Alexander, 1994b; Arditi, 2017a). There has been a transformation from an ownership model, with physical and digital formats, to an access model, in which the consumer has had to adapt to different manners of acquiring and listening to music (Wikström, 2012; Kolar, 2017). An accompanying trend is one of disintermediation and empowerment, which has changed the dynamics of the industry greatly (Fox, 2004; Hracs, 2012; Musgrave, 2017). It was expected that these developments would render record labels obsolete. However, the opposite was proven to be true, as the record label is still a very important entity in the industry (Moore, 2012). To emphasize the prominence of the record companies, it has been established that they are the primary financers in the music industry, spending large sums of money on artist and repertoire, and the marketing of records (WIN, 2017). A total revenue of $15.7 billion was generated in the global recorded music industry in 2016 (IFPI, 2017). More specifically, one of the largest record labels worldwide, Universal International Music, reported a revenue of $5.3 billion in 2016 (Schneider, 2017). This shows the size and importance of the industry and the promising returns that are being attained by successful labels.

All these factors result in an industry that is highly dynamic and very interesting to study. This thesis intends to shed more light on the competitive environment of the recording industry, with a focus on the emergence of new record labels. The strategic move of new entry is of great importance to the new entrants as well as the incumbents since it affects the levels of concentration in the industry. The complex structures of large labels with many different

(8)

subsidiaries and business units, the distinction between the types of labels, and strategic alliances that can be formed between them make for an industry that is complicated but exciting to study. The purpose of this study is to answer the research question that is formulated as follows: “Can the level of strategic integration and the structure of a company explain the likelihood of entry and performance of new record labels in the music industry?” This will be researched by thoroughly analysing music metadata on record labels, which is stored in databases that are accessible to the public.

The remainder of this thesis is structured as follows. First, the literature review will present the basis for the strategic perspective, and explain why new entry is a phenomenon that is crucial for an industry. Then, the music industry value chain will be described, and the strategic perspective will be applied to this industry by looking at diversification, integration, and types of record labels. The literature review provides the theoretical basis for the hypotheses that will be formulated subsequently. Additionally, the research method will be explained in detail and the approach taken for the analysis of the data will be clarified. This will allow the results to be presented and discussed in order to evaluate the findings that were discovered. This thesis will be ended with a conclusion, which includes the implications, the limitations, and the options for future research.

(9)

2. Literature Review

2.1 Strategy and New Entry

Companies in an industry can have different objectives they want to achieve. How they address these priorities within their corporations can vary. These differences in approach with regard to attaining their principal goals are referred to as the strategy of a firm. There has been written a lot about strategy in business and management literature in the past, and it has become a dominant subject when explaining firm survival and the configuration of the competitive environment. A strategy is defined as a coordinated set of actions designed to exploit core competencies to achieve competitive advantage (Hitt, Ireland & Hoskisson, 2012). In a strategic plan, businesses make decisions on how they want to pursue their corporate goals. Business-level strategies describe how firms aim to fulfil objectives by exploiting their core competencies to attain competitive advantages. These strategies focus mainly on specific product markets (Jones & Butler, 1988). Whereas corporate-level strategies are consciously formulated with a long-term focus that aims to specify the management of several different businesses (Pehrsson, 2006).

One of the most prevailing models in strategic management literature is the Five Forces framework by Porter (1979). In this framework, there are five competitive forces that are argued to structure competition in an industry. Porter (1979) reasons that awareness of these forces can help companies to formulate a comprehensive strategic business plan. One of these Five Forces is the threat of new entrants. This force refers to the threat that new competitors pose to the incumbents in an industry. When there is an influx of new entrants, the ability of the incumbents to achieve profitable returns could be affected. This threat can be reduced by increased barriers to entry or by the expected retaliation of the incumbents. These factors can act as a deterrent for the possible new entrants (Porter, 1979). Although there are many motivations and reasons to start a new venture, entrepreneurs must be aware of these barriers

(10)

to entry (Birley & Westhead, 1994). Barriers to entry are defined as the market conditions that raise the cost of entering the market for new firms so that incumbents can earn profits in the long run (Bain, 1956). Porter (1979) argues that there are six sources that can heighten the barriers: economies of scale, product differentiation, capital requirements, switching costs, access to distribution channels, and government policy. Depending on the industry and its characteristics, these sources can prevent new entry from being successful. Furthermore, Lambkin and Day (1989) argue that the development of a market is strongly related to the number and the types of companies that are entering an industry. However, not all incumbents are affected similarly when a new firm enters the market. This is caused by the forces that accompany new entry, and how these influence firms either in a competitive or a cooperative manner. A new entrant can, for example, intensify rivalry in an industry where heavy

competition is already waging. Whereas, new firms can also help to legitimize markets, leading to an increase in demand from consumers (Özsomer & Cavusgil, 1999).

An industry is labelled highly concentrated when most of the total market share is controlled by only a few players. High barriers to entry prevent the entrance of new actors into the market, hence causing increased concentration levels (Tremblay & Tremblay, 2012). The notion of a connection between concentration and new entry is underlined by the theories supporting the organizational ecology perspective on firm survival. More specifically, by the theory of resource partitioning and its proposed effects on industry configurations (Baum, 1996). In this theory, a distinction between generalists and specialists is made. Generalists are the large players that dominate a field and can operate in multiple types of environments. Whereas the specialists focus on certain peripheral (i.e. niche) areas as they have increased chances of survival when they can target narrow and homogeneous segments (Carroll, Dobrev & Swaminathan, 2002). Carroll (1985) researched the newspaper industry where generalists and specialists inhabit the market. This industry is known for its increased barriers to entry due

(11)

to economies of scale, leading to an environment with high levels of concentration. Economies of scale imply that large companies can produce their products at a lower price because the marginal cost of producing a unit declines as the produced quantity increases (Hitt et al., 2012). These conditions combined with consumer tastes that are heterogeneous are the assumptions for the theory of resource partitioning. This theory reasons that rivalry among generalists in the main areas of the industry leads to the opportunity for specialists to occupy the peripheral areas in the same industry without having to face generalists in direct competition (Reis, Negro, Sorenson, Perretti & Lomi, 2012). This is because the generalists focus on the mainstream markets so that they can reach a large group of potential customers, whereas the specialists focus on those domains that are more distinctive, and allow for a more targeted approach (Carroll, 1985). This theory has received wide attention and allows for application to many industries. It is related to the subject of strategic new entry because it provides an explanation as to why it is still possible and attractive for new companies to enter a market, even though the concentration levels are high.

2.2 Strategic Integration

A newly entering company is not always a completely new venture initiated by a layman. The origins of a new entrant can be diverse. If the company has already participated in the same supply or value chain, then these expansion efforts are termed horizontal or vertical integration (Caves & Porter, 1977). These actions are both undertaken to acquire more market power (Hitt et al., 2012). Horizontal integration happens when a firm acquires or starts a competing firm in the same part of the supply or value chain. This could lead to several players becoming very powerful because of the increased levels of concentration in an industry (Tarzijan, 2004). Mergers and acquisitions are expansion strategies that are pursued by many companies looking to spread their presence in an industry. Horizontal mergers and acquisitions

(12)

are not only undertaken to increase market power but are also used to reorganize resources or to achieve scale efficiencies. This could make organizations more efficient, and often also more profitable because of the reduced overhead costs (Capron, Dussauge & Mitchell, 1998). On the contrary, vertical integration entails a firm moving up or down the supply or value chain. Forward vertical integration involves a firm integrating further along the supply chain. Whereas backward vertical integration involves a firm moving towards the start of the supply chain (Harrigan, 1985).

More specifically, Mol, Wijnberg, and Chiu (2012) distinguished three modes of entry that can be pursued when entering an industry. The first one is called ‘new entry de ipso’, this means that the entrant has previous experience in the same industry and in the same part of the value or supply chain, and has therefore acquired useful knowledge and connections that make new entry significantly easier. This is similar to the notion of horizontal integration. The second entry mode is called ‘new entry de intra’, this refers to the actors, who are already part of the same value or supply chain, but have gained experience in a different task. This is similar to the notion of vertical integration. The third one is called ‘new entry de novo’, which refers to entrants who are completely new to the industry and have no experience whatsoever. Furthermore, there are several other studies that mention an additional mode of entry, that is referred to as ‘new entry de alio’. This mode entails the entry of a firm, that already has business experience in a similar market but is looking to diversify their production (Carroll, Bigelow, Seidal & Tsai, 1996; Giarratana, 2008; York & Lenox, 2014). These incumbents already have the resources and capabilities that should make diversification notably easier.

2.3 The Music Industry

In general, the music industry is referred to as a value chain rather than a supply chain. This is because distinct value is added to the product in every step of the process along the

(13)

different stages (Mol, Wijnberg & Carroll, 2005). The basic value system of the music industry is generally believed to consist of several stages. This is illustrated in figure 1, which depicts the value-adding entity and the main task for which the entity is responsible. It starts with the composition, recording, and production of music by artists and musicians. Then it moves along to the record labels that sign the artists they believe have potential for success. This entails the artists to give their musical recordings to the label. The label then coordinates the licensing, marketing, and promotion of the recordings (Graham, Burnes, Lewis & Langer, 2004). Normally, it is also the case that they arrange the manufacturing and distribution of the physical or digital end-product with other actors. The distributor then makes sure the product ends up at the correct retailers, from which the consumer can purchase the product (Bockstedt et al., 2006). The terms record company, record label, and music label are used interchangeably in the literature and this thesis. They are often referred to as the cultural intermediaries of the music industry. The term cultural intermediary was introduced by Bourdieu (1984). The main emphasis of this terminology lies in the fact that the individuals and firms that are termed cultural intermediaries are in the spots between the creative input and the commercial output. They are the middlemen between the production and consumption (Negus, 2002).

One of the main topics in music business literature in the past few years is the influence of technological innovations (Alexander, 1994a; Graham et al., 2004; Arditi, 2014a). For this study, these developments are of importance because of the impact they have had on the industry concentration. New technologies often are positive for newcomers since they are likely to reduce the barriers to entry (Hesmondhalgh, 1998; Alexander, 1994b). Three main parts of the industry have been disrupted or heavily influenced by these new technologies.

(14)

Firstly, there have been significant changes in how consumers enjoy musical recordings. There has been a transformation from a product-based to a service-based industry (Leurdijk, Slot & Nieuwenhuis, 2012). Among consumers, there has been a clear shift from individuals who owned physical items, such as vinyl and CDs, and digital items, such as MP3s, to solely access to music via streaming or subscription services (Kolar, 2017; Arditi, 2017a). Another trend that is connected to the changing consumer patterns, is the unbundling of music. Consumers now have the opportunity to ‘cherry pick’ the specific songs they want to purchase instead of having to buy complete albums (Ferraris, Auberty & Rao, 2008; Elberse, 2010).

Secondly, there are increasingly more opportunities for artists to reach their audiences due to the possibilities presented by new media channels. It is argued that this could lead to the disintermediation of the value chain and increased empowerment for musicians (Leenders, Farrell, Zwaan & ter Bogt, 2015). Digital equipment for recording, mixing, and mastering is becoming more accessible for artists since the costs of these tools have been decreasing. This in combination with the introduction of MySpace, YouTube, and Soundcloud, should have shifted the balance of power in favour of the artists (Homan & Gibson, 2007). However, many artists are still struggling to make a living as professional musicians (Hesmondhalgh, 2013). Leenders et al. (2015) concluded that even though the industry has undergone major transformations, record labels remain an important part of the value chain because of the associations they hold with global media. This notion is supported by Mol et al. (2012), who found that connections to mass media gatekeepers are vital to the success of music industry actors.

And finally, the industry is influenced by the effect of the long tail. This theory argues that because of the digitalization of content, there is an increase in the number of recordings being released (Anderson, 2006). This is made possible by the advent of the Internet and other innovations that have made the production and distribution of musical recordings less costly,

(15)

and thus more accessible. This notion is supported by research done by Bourreau, Gensollen, Moreau, and Waelbroeck (2012), who concluded that digital record labels that only release recordings through the Internet, on average produce more music than a regular record label. This suggests that two sources that normally heighten barriers to entry as argued by Porter (1979) have decreased, namely the capital requirements needed and the access to distribution channels. Thereby making new entry into the industry easier and more attractive.

2.4 Majors vs. Indies

The record label stage of the music industry value chain can be categorized as oligopolistic, meaning that there are several very strong firms that dominate the market (Burnett, 1992; Homan & Gibson, 2007). Currently, there are three major labels, namely Warner Music Group, Sony Music Entertainment, and Universal Music Group. A record label qualifies as a major, per the Association of Independent Music, when it has more than 5% of the world market’s sales of records. These record labels can provide an artist with additional services, as they often have their own distribution channels (Lindvall, 2012). This is because typically they are a part of a music group, that is owned by an even bigger conglomerate holding. These holdings have several divisions that can even be focused on non-music ventures (Rutter, 2016). Major labels are structured in a multidivisional manner, implying that the main company has many subsidiaries, frequently referred to as sublabels or semi-independent labels (Lacey, 2014; Dowd, 2004b). In appendix 1, there is a partial overview of the multidivisional structure of the three major record labels to illustrate the expansiveness and complexity of these companies. The development of these multidivisional structures took many years and consisted largely of starting, acquiring, and merging various sublabels (Scott, 1999; Shuker, 2017). This structure is related to the corporate-level strategy of related linked diversification that is often pursued in the music industry. This strategy proposes that there is a large parent company with

(16)

multiple sublabels, which share some activities, core competencies, and resources (e.g. distribution) (Tanriverdi & Lee, 2008). It was proven by Palepu (1985) that the performance of firms with a related diversification strategy is significantly better than the performance of firms with an unrelated strategy, which infers that no activities are shared among subsidiaries.

Majors are often contrasted against independent labels (indies). The main difference between the majors and the independents is that the major labels have an extremely large market share, and the opportunity to make use of their integrated distribution channels. The term ‘independent’ might be somewhat inaccurate, as these labels often do depend on one of the three major labels for the distribution of their music (Scott, 1999; Passman, 2015). For example, the Warner Music Group has a subsidiary called ‘Alternative Distribution Alliance’ that is responsible for the distribution of records for more than 100 independent labels (ADA, 2018). This is often done because a distribution network is extremely tough and costly to set up independently. It is often a lot simpler to contract another company to take over these tasks (Alexander, 1994b). Indies are regularly categorized as small niche labels, but this is not always true because independent record labels can also be large organizations with their own sublabels. They are generally known for their artist-friendly deals, and their ability to spot new trends and promising artists (McLeod, 2005). Moreover, the independent labels represented a large share (38.4%) of the recorded music market in 2016. This percentage has been steadily increasing over the past few years, with a relative growth of 0.9% from 2015 to 2016 (WIN, 2017).

Hesmondhalgh (1996) reasons that major and independent record labels can have several types of relationships. The first type is international licensing, where smaller labels negotiate a deal with a large label concerning the licensing of recordings in various countries. The second type is the distribution deal, where close cooperation between the small and the large label is required. The larger label will then take care of the distribution of records for the smaller labels. At times, this also involves a funding deal, where the small labels sell some of

(17)

their stakes to the larger label. The third type of relationship is the acquisition. This is a type of relationship that the majors have been pursuing extensively during the past 20 years, as a part of their expansion strategy. The licensing and distribution relationships are often seen as the setting stages for the oftentimes inevitable acquisition. Acquisition by a major is also an end-stage that is denoted by Wallis and Malm (1984) in their framework that depicts the possible routes of development for small record labels (see appendix 2).

Highlighting the distinction between majors and indies is essential as both types have their own advantages and disadvantages. The majors have a lot of power, resources, and connections that could prove valuable, whereas the indies have a better reputation regarding artist deals. The type of record label is of importance when looking at strategy because it has been proven that company structure affects performance negatively when a company’s strategy is not complemented with a corresponding structure (Rugman & Verbeke, 2008).

2.5 New Entry in the Music Industry

The modes of entry that were stated before, can be applied to new record labels entering the music industry (Mol et al., 2012). Firstly, the de ipso entrants are the new subsidiaries of large labels that opt to integrate horizontally by starting or acquiring new sublabels. This entry mode can be pursued by a major label or an established independent label. Secondly, the de intra entrants were identified as the entrants who have been part of the same value chain but have accumulated experience in a different task. This could be the case when artists, distributors, or manufacturers attempt to integrate vertically by starting their own record label. And thirdly, there are de novo entrants who have no previous experience in the music industry, hence they will start with no prior knowledge or connections.

Because of increased reliance on technology and the prominence of the Internet in our daily lives, the music industry value chain has undergone some transformations. Core tasks of

(18)

record labels such as the marketing of albums, now mainly take place on the Internet through social media channels (Leurdijk et al., 2012). The manufacturing stage of the value chain can be skipped completely because a wide audience can be reached with an end-product in a digital format (Wikström, 2012; Vaccaro & Cohn, 2004). These developments have lowered the barriers to entry and therefore have made it more attractive for individuals to start their own record labels (de novo entry). This is underlined by the fact that the Association of Independent Music has uploaded a booklet with guidelines for starting your own record label to their website (AIM, 2016). Moreover, due to the phenomenon known as value chain envy, entry into this stage of the music industry value chain is most attractive to prospective new entrants (Mol et al., 2005). This is because the ratio between value capture and value creation is most uneven in record label stage, hence creating more favourable prospects of attaining increased returns. Individuals that have gathered knowledge about this unequal division could experience a feeling of envy. This feeling could provide motivation to enter this exact value chain stage.

Starting a company requires a lot of effort because a company needs to be registered legally, and other organizational tasks need to be fulfilled (e.g. choosing a name, hiring personnel, setting up a website) (McKay, 2009; Reis et al., 2012). The larger labels have a lot of experience in the industry because many were established years ago. This suggests that they are aware that in order to expand their business, the acquisition of a label is a less intensive approach than starting a new label from scratch. Therefore, it could be argued that it is the strategic intention of larger labels to acquire other labels after a phase in which they had time to establish themselves. This would indicate that a strategy of horizontal integration through acquisition would be preferred over horizontal integration through starting a new subsidiary by existing major and independent labels. I expect that because of the technological advancements and the disintermediation of the value chain, more labels enter the industry without any previous experience (de novo entry) and that established labels prefer to acquire rather than

(19)

start new subsidiaries due to the effort needed, which allows the following hypothesis to be formulated:

H1a. A de novo entrant is more likely to be entering the industry than a new subsidiary

initiated by either a major or a large independent label (de ipso entrants).

A completely new record label (de novo entrant) will often focus on a niche to avoid direct competition with industry heavy-weights. This is in accordance with the theory of resource partitioning, which argues that specialists are more likely to focus on narrow and homogeneous segments, whereas generalists often concentrate on large heterogeneous segments (Carroll et al., 2002). The focus on a certain niche may entail exploring peripheral genres, which often leads to more creative and experimental releases (Negus, 2002). It was argued that these small new labels are more adept at finding talent in fringe genres than the large labels (Burnett, 1996; Hesmondhalgh, 1996). This is mainly because the structure of large labels is too complex, whereas the structure of the new label is likely to be flat, which makes communication and decision-making easier (Carzo & Yanouzas, 1969). This notion is also supported by the theory of the long tail, which reasons that targeting niche markets is more fruitful because these are often ignored by the commercially focused majors (Anderson, 2006). However, it is very hard to achieve good results because of the inherent uncertainty and continuous oversupply in the music industry (Peltoniemi, 2015). Attaining returns that are sufficient might be especially hard for the de novo entrants that focus on peripheral genres that are not well-known to the public.

These harsh circumstances are also encountered by the subsidiaries of established labels. Still, these de ipso entrants have parent companies that could supply them (if necessary) with resources, expertise, and knowledge through the related linked diversification strategy (Montgomery & Singh, 1984; Mol et al., 2012). Additionally, these companies can benefit from economies of scale, which allows them to deliver more output to the market at lower costs

(20)

which increases the margins. This is possible because of the large production size and the incremental efficiency improvements achieved through experience (Scott, 1999; Hitt et al., 2012). This enables the de ipso entrants to establish a preferential position, which will increase their chances of success, thereby improving their performance. Hence, I expect that de novo entrants will have a harder time when it comes to performing sufficiently in comparison to both types of de ipso entrants, which brings the following hypothesis:

H1b. A de novo entrant is less likely to perform well than a new subsidiary initiated by

either a major or a large independent label (de ipso entrants).

In a similar vein, a distinction can be made between the two types of de ipso entrants, the actual majors, that are a part of a large international media group, and the independent labels that are established names but are not nearly as powerful as the majors. The distinction in market power is made based on the size of the majors, the fact that they have integrated distribution systems, and more sophisticated facilities (Musgrave, 2017; Hughes & Lang, 2003). As was mentioned before, these established record labels can pursue horizontal integration by acquiring existing companies or by starting new subsidiaries. Horizontal integration through starting a new sublabel is attractive because resources and personnel can be redeployed from another label to focus on the new subsidiary (Bradley, Aldrich, Shepherd & Wiklund, 2011). This is in accordance with the concept of a related linked diversification strategy, in which core competencies and resources are shared or provided by the parent company (Tanriverdi & Lee, 2008). This approach will most likely be less costly than the acquisition of an existing company since this can be very expensive because a premium is often required to be paid on top of the fair value of the business’ assets (Li, 2011). This might not be an issue for the majors because these large corporations are expected to have more excess resources (Ingham, 2017a). To illustrate, Spinnin’ Records, a Dutch independent record label started in 1999, was acquired by Warner Music in 2017 for an amount that was estimated to be

(21)

around $100 million (Ingham, 2017b). These large amounts of money would be impossible to secure for most independent labels. Thus, this would suggest that a large independent label would pursue a strategy of horizontal integration through starting sublabels more often than a major label, causing the following hypothesis to be formulated:

H1c. A new subsidiary initiated by an established independent record label is more

likely to be entering the industry than a new subsidiary initiated by a major label.

An additional distinction between these two types of de ipso entrants is that independents are known to be more engaged and care more about musicians and bands (Hesmondhalgh, 2013). Therefore, it is more likely that these labels would engage in starting new subsidiaries that could focus on promising new genres or niches so that they can provide the best service for their artists. This follows the subsidiary’s strategy type of a specialized contributor that was first identified by White and Poynter (1984). This implies that considerable expertise in specific functions is obtained in these subsidiaries by focusing on certain markets or products. Whilst a tight connection with the parent company is maintained (Birkinshaw & Morrison, 1995). These expert subsidiaries are also proposed by the theory of resource partitioning which argues that specialists often opt to focus on niches (Carroll et al., 2002). I expect that these expansion endeavours can be pursued successfully because of this deliberate specialization, since it was proposed that specialist firms often outperform generalists, because of the explicitly developed human capital in these firms (Gompers, Kovner, Lerner & Scharfstein, 2008). For a specialized record label, this could lead to attracting more niche artists and offering them expert knowledge and resources, which should allow for a performance that is above average. This would suggest that a large independent label would pursue a strategy of horizontal integration more effectively than a major, resulting in the following hypothesis:

H1d. A new subsidiary initiated by an established independent record label is more

(22)

2.6 Diversity in the Music Industry

For a long time, increased market concentration was believed to decrease diversity and innovation in musical output. This idea was the basis of the ‘old popular music model’, in which large labels were presumed to be more interested in artists that were guaranteed to become popular and bring in a lot of revenue. The primary goal of the record label was profit maximization, which could be attained by focusing on confirming consumer tastes rather than disrupting them (Rothenbuhler & Dimmick, 1982). During the 1980s, there was a transition period, where concentration in the industry was very high, yet the diversity of output also increased significantly. One of the speculated causes of this phenomenon was the cooperation between large and small labels through licensing and distribution deals (Burnett, 1992; Lopes, 1992). Thus, the ‘old popular music model’ is not applicable anymore in the music industry (Burnett, 1996). Nowadays, large record labels produce and release music in all kinds of genres, so that no unsated demand exist among the consumers (Lopes, 1992). It was argued by Peterson and Berger (1975) that a more concentrated environment would lead to a more entrepreneurial mind-set for prospective firms. This would then induce the entry of new specialist organizations that could focus on certain niches, which is in line with the resource partitioning theory (Carroll, 1985). Therefore, when there are more record labels entering the industry, an increase in the diversity of the output is expected (Peterson & Berger, 1975).

There are several reasons why a firm would opt for a strategy where there is an emphasis on the diversification of output (Hitt et al., 2012). In the fundamentally uncertain music industry, the reduction of risk is one of the main concerns. Rothenbuhler and McCourt (2004) argue that the music industry is possibly the most ambiguous and volatile of all the creative industries, as consumer tastes are extremely fluctuant and the supply of musical output is abundant. One of the diversification strategies that is frequently used to tackle this problem, is portfolio management. This strategy allows the spreading of risks across potential sources

(23)

of income. In effect, the diversifying record label releases several records across different genres and audiences, to ensure they will always get some returns (Hitters & van de Kamp, 2010). Genres that are well-known and are loved by many consumers, are steady income sources. Whereas other genres might be riskier and have returns that are harder to predict, so the allocation of resources to these genres should be done with care (Negus, 1999a).

Diversification strategies, such as portfolio management, are related to the aforementioned mode of entry that is known as ‘new entry de alio’, which refers to existing firms that opt to diversify their output (Carroll et al., 1996). De alio entrants already have established legitimacy, expertise and resources in their industry, which should make diversification notably easier (York & Lenox, 2014). The focus of this entry mode is on the output of a record label, which are the musical recordings that are being released. For a record label, diversification could consist of adding new genres to their repertoire of releases (Hitters & van de Kamp, 2010; Negus, 1999a). Referring to the example that was given before, Spinnin’ Records started as an independent record label solely focusing on electronic music. However, after several years they decided to diversify their genre range, by releasing songs that are categorized as pop or hip-hop (Discogs, 2018b). This suggests that de alio entrants try to diversify their releases by expanding into multiple genres to hedge against risk and uncertainty.

This entry strategy can be undertaken by all types of record labels that were covered in the previous section. However, de novo entrants have to build a reputation for themselves in order to be able to attract and sign more artists. Petkova, Rindova, and Gupta (2008) argue that new ventures, that establish favourable perceptions through direct interactions in small specialized groups, can quickly build a solid reputation. Therefore, it might be more fruitful for de novo entrants to focus on a specific genre and offer specialized care for these artists. This also applies to the large independent labels, which are expected to maintain a specialized approach when establishing subsidiaries through horizontal integration. It was argued before

(24)

that independent labels are more focused on providing the best environment for their artists rather than wanting to maximize profit (Hesmondhalgh, 2013; Rudsenske & Denk, 2005). Consequently, I expect that they will start a sublabel to focus on one particular genre, which should be more attractive to artists that release music in this genre, as they would be able to provide these artists with the best resources and expertise pertaining to this genre. Conversely, it can be assumed that major labels will start sublabels with as goal to cater to a public that is as broad as possible to maximize profits (Burnett, 1992). Therefore, it is suggested that subsidiaries of major labels are started with a strategy of risk hedging in mind. This leads to the following hypothesis, in which I expect that sublabels pursued by majors use portfolio management strategies to quickly spread their releases across multiple genres to secure more returns:

H2. Subsidiaries initiated by majors tend to diversify faster into genres different from

the genre of their first release (de alio entry), in comparison to other new entrants.

2.7 Vertical Integration in the Music Industry

Another predominant type of record label in the music industry is the vanity label. This is a record label that is initiated by an established record label and is fronted by a popular artist or band. It is suggested that this artist is running this company and is therefore taking care of the day-to-day operations. However, these labels are wholly or partially owned by the original record label of that same popular artist (Canetta & Winn, 2002). An example is Shady Records, a label founded by Eminem, on which he released his own music, as well as records from other artists (Joy, 2017). Nevertheless, this record label is a subsidiary of Universal, the parent company that also owns the label Eminem signed to originally (Aftermath Entertainment). This type of label is similar to the aforementioned subsidiaries of established record labels (de ipso entrants) with the exception that vanity labels are fronted by a well-known artist. The

(25)

association with a famous artist is synergetic as it both uplifts the credibility of the newly founded label, as well as the integrity of the artists (Rutter, 2016).

Large record labels often pitch the opportunity of starting a vanity label to artists to keep them satisfied and occupied (Rutter, 2016). However, this is not always the case, as some artists choose to start their own label, completely independent of any other label (Bray, 2010). This is a form of forward vertical integration (de intra entry), as the artist moves forward along the value chain. They do not only deliver music but also take over the tasks of the record label (Harrigan, 1984). This is in line with the trend of disintermediation which was fuelled by the advancements in technology (Hesmondhalgh, 1998). An advantage that the de intra entrants might have over de novo entrants, is that they have been part of the music industry value chain for some time. Therefore, they may have been able to establish relationships with industry actors, which could make their entry easier and more successful (Hughes & Lang, 2003).

Accordingly, a distinction between the two types of artist-fronted labels can be made. Vanity labels that are initiated by established labels (de ipso entry) are posed against labels that are initiated by artists independently (de intra entry). A lot of negative publicity has been surrounding labels in the past years concerning record deals that rip off artists and contracts that include harsh conditions (i.e. ‘slave deals’) (Kaye, 2015; Messitte, 2015). The collective reputation of record labels has suffered as a result of this. Therefore, artists might be less willing to sign a deal that they deem to be exploitative when they can start a label themselves. Referring to Wallis and Malm (1984), all three conditions needed to start a new record label are present for these artists, namely dissatisfaction with the establishment, the acquisition of know-how, and accessibility to equipment. Additionally, similar to de novo entrants, artists might also experience value chain envy (Mol et al., 2005). For artists, this could have an extra dimension as they might pursue vertical integration to be able to retain more of their royalty payments. Hence, I anticipate that when an artist-fronted label is started, it is more likely to be

(26)

initiated by artists themselves without any help from an existing label, which leads to the following hypothesis:

H3a. Record labels initiated by artists independently (de intra entry) are more likely to

be entering the industry than vanity labels (de ipso entry).

However, it is not expected that these de intra entrants will be able to establish the level of performance that is attained by vanity labels. The same reasoning applies as for hypothesis 1 since the vanity label has a parent label that could provide them with back-up whereas the de intra entrant does not have access to those resources. Therefore, when artists start their own label to release their output, I expect them to perform worse than a vanity label.

H3b. Record labels initiated by artists independently (de intra entry) are less likely to

perform well than vanity labels (de ipso entry).

Table 1. Overview of hypotheses

H1a A de novo entrant is more likely to be entering the industry than a new subsidiary initiated by either a major or a large independent label (de ipso entrants).

H1b A de novo entrant is less likely to perform well than a new subsidiary initiated by either a major or a large independent label (de ipso entrants).

H1c A new subsidiary initiated by an established independent record label is more likely to be entering the industry than a new subsidiary initiated by a major label.

H1d A new subsidiary initiated by an established independent record label is more likely to perform well than a new subsidiary initiated by a major label.

H2 Subsidiaries initiated by majors tend to diversify faster into genres different from the genre of their first release (de alio entry), in comparison to other new entrants. H3a Record labels initiated by artists independently (de intra entry) are more likely to be

entering the industry than vanity labels (de ipso entry).

H3b Record labels initiated by artists independently (de intra entry) are less likely to perform well than vanity labels (de ipso entry).

(27)

3. Data and Methods

3.1 MusicBrainz Database

For this research, I opted to mainly work with the MusicBrainz database, since this database stores the most complete information on record labels (MusicBrainz, 2018a). The entity label is of importance for this study, and contains several keys that will be used as variables (see appendix 3 for an overview). This database is driven by music metadata that is generated through music information retrieval research (Futrelle & Downie, 2002). MusicBrainz is an online encyclopaedia that aims to collect metadata for any kind of musical output, based on contributions from the public (Porter, Bogdanov & Serra, 2016). It is a community-based (open-source) database, that is regarded to be of high-quality, thereby collecting information that is useful for various types of research (Vigliensoni, Burgoyne & Fujinaga, 2013; Stutzbach, 2011). All the core data in the database is available to individuals under the project’s Creative Commons license and is therefore allowed to be used for the desired purpose of this study (Hemerly, 2011; Creative Commons, 2018). However, using a database that is community-based and relies on amateur contributions could pose some limitations. It has been researched whether this could harm the reliability and validity of the data. On several bases, it was concluded that the quality of the data was not impaired by the fact that it was contributed by non-professionals, and is therefore suitable for academic research purposes (Marshall & Shipman, 2003; Zhou & Davis, 2005).

3.2 Sample

A copy of the MusicBrainz database was obtained from which a new file was compiled that contained all the information on record labels based on the record label ID. Because a lot of the data for the exact month and day of entry and exit were missing and were not important for further research, these variables were removed. The total dataset consisted of over 140,000

(28)

entries for record labels, dating back to the early 1800s. Firstly, the most crucial event in this research is the entry of a new record label, which is denoted by the first year that a label releases a recording, as from this moment the firm is an operating entity. The database was sorted on entries which had a value for entry year, the others were discarded. For several entries in the sample, this value was changed to the year of their actual first release, as in some cases these values were not congruent. Secondly, I filtered out all the label types that did not focus on producing new music, as I only want to research the record labels that release original and new music. Thus, the label types that were included in the sample are original production, production and imprints. Thirdly, the choice was made to focus on new entrants from the beginning of 2005 until the end of 2017, since the dynamics of the industry have changed significantly after the introduction of many innovative technologies. Burkeman (2009) argues that since 2005 most individuals use the Internet on a daily basis, and are able to use it for their desired purpose (e.g. setting up a MySpace or YouTube account). Also, Leenders et al. (2015) use 2005 as starting year for their research in which they measure new media adoption among artists. Then, I opted to focus on one country to control for the fact that the circumstances for starting a company might be different in various countries, thus increasing the internal validity of the design. The most appropriate country to use for the sample is the United States because per the IFPI (2017) the United States is the world’s largest recorded music market. After these restrictions, the dataset consists of 457 cases, which is sufficient and feasible for my research. Once the necessary data was collected, it was concluded that 31 entries in the dataset did have an entry year but had not released any music. These entries were discarded, which led to a total sample of 426 record labels. The main limitation of this research design is a phenomenon referred to as selection bias. This means that the selected sample is biased because only labels that actually entered the industry and are mentioned in the database are included in the final sample. This issue will be addressed extensively in the section on limitations.

(29)

3.3 Variables

3.3.1 Hypothesis 1

DV. In order to make suggestions about the performance of the record labels after new

entry, additional variables need to be compiled. Mauboussin (2012) argues that a good measure of performance is a metric that focuses on the corporate goals a firm wants to attain. The main corporate objective of a firm in an oligopolistic industry is long-term survival, according to Rothschild (1947). Long-term survival could be attained by frequently releasing new output, and thus generating continuous returns. This is because when a released song or album has had disappointing returns, it is less likely that the label will produce more output. There might not even be sufficient resources to do so, as the break-even point for the previous release has not been reached (Rothenbuhler & McCourt, 2004). Hence, I deem the number of releases produced by a label after entry, an appropriate metric to measure performance. The variable that was added, contains the total number of releases by each label after their entry until the end of 2017. This data was retrieved from a different dataset that was also acquired through the MusicBrainz database. The record label identifier was used to connect the two datasets, after which the frequency of releases was used to compile the new variable. To correct for the fact that labels entered the industry at different times, an additional variable was created which calculated the average number of releases per year after entry into the industry. This was done by dividing the value for the number of total releases by the difference in years between 2018 (or exit year) and the entry year of the record label.

IV. A framework will be used to categorize the types of labels in the database. This

classification was established by using the MusicBrainz URL and the other label identifiers to find further information on the origins of the labels. The main information sources were MusicBrainz and RateYourMusic, which provided the name of the parent company (if applicable), from which could be established whether the parent was one of the three majors or an independent label. Discogs and the official websites of the record labels were used to

(30)

check this information. The framework consists of three types of record labels. Firstly, there are new subsidiaries (de ipso entry) that are initiated by one of the three major labels. Secondly, there are new subsidiaries (de ipso entry) that are initiated by large independent labels. Thirdly, if there is a new entrant that is being established wholly independently, it is classified as a de novo entrant. For H1c and H1d, a dummy variable was created that only contained the de ipso entrants.

DV. For the second hypothesis, complementary information and variables were needed.

The new dependent variable is the time taken until the first diversification in genres by a record label. For every record label, the genre of the first release was collected, after which it was researched if there was a second genre that was pursued, and if so in which year this had happened. Discogs proved very helpful in attaining this information. Finding the correct values was done by looking up the record label, then sorting the releases on date, and checking which genre belonged to the first release. Then, the same sorting system was used to figure out whether a second genre was pursued, and in which year this was done. The values for three new variables (first genre, second genre, and year of second genre) were added to the dataset. With these values, the dependent variable (time to diversification) could be calculated by subtracting the entry year from the year that the second genre was pursued.

IV. The same classification scheme that was used for hypothesis 1, can be applied to

this hypothesis. The distinction between de ipso and de novo entrants is discarded. The only difference made is between subsidiaries of major labels and independent labels, which could be either subsidiaries of established labels or completely new ones.

(31)

DV. The measure for the performance after entry is the average number of releases until

the end of 2017 (or exit). This is the same dependent variable that was used for hypothesis 1.

IV. For the third hypothesis, additional information was needed on the origins of the record labels. This information was largely obtained from RateYourMusic, as this website provides information about the founder(s) of record labels. Any label that was founded by individuals that already had experience in the music industry value chain (e.g. producer, musician, band), was categorized as a de intra entrant. Then it was checked whether the label was associated with an established record label, if this was the case, the label was categorized as a de ipso entrant. A dummy variable was created for the two different types of artist-fronted labels, consisting of a label that is initiated by an artist independently or a vanity label that is initiated by an established record label.

(32)

4. Results

4.1 Hypothesis 1

For all tests, a statistical significance level of α= 0.05 is used. A detailed description of all test results can be found in appendix 8.4. Several tests were used to check for normality, which established that the data for the dependent variable that measures performance was not normally distributed (Ghasemi & Zahediasl, 2012). However, since the sample size is sufficiently large, there should not be any problem as parametric tests perform well even with skewed distributions (Frost, 2015). However, to confirm that these tests provide the correct outcome, complementary non-parametric tests are performed. From the sample of 426 new labels, it was found that 9.4% are sublabels of a major, 13.4% are sublabels of an established independent label, and 77.2% are completely new independent labels.

For H1a, a chi-squared goodness of fit test is performed to see if the expected frequencies differ from the observed frequencies (Keller, 2012). For this test, it is expected that the three types of record labels that are present in the sample have equal frequencies. The outcome of this analysis is significant (χ2

(2)=370.41, p=0.000). This leads to the rejection of the assumption that the three types of record labels enter the industry with equal likelihood. An additional binomial test is performed, which

verifies that de novo entrants are more likely to enter the industry than de ipso entrants, thereby confirming hypothesis 1a.

Furthermore, to determine which of these label types performs best after entering the industry, the number of average releases per year will be analysed. Table 2 displays the descriptives of the three groups that will be analysed. A one-way analysis of variance is performed, allowing the comparison of three groups with an interval dependent variable. The results in table 3 suggest that the there is a significant difference in performance between the

(33)

three groups. However, Levene’s test is not passed, meaning that the assumption of homogeneity of variances is violated, which implies that the variances are not equal. This requires additional analyses to be performed. Firstly, a Welch ANOVA is performed as a robust test of the equality of means (Moder, 2007). The results of this test confirm the outcome that was attained by the one-way ANOVA (see appendix 8.4.3). Secondly, a Games-Howell post hoc test is used to analyse the differences between the groups. This specific test was selected because of the violated assumption of homogeneity of variances (Laerd, 2018d). From this test, it can be assumed that there is a significant difference in the performance of de novo entrants in comparison to both types of de ipso entrants. Since the dependent variable is not normally distributed, I will use a non-parametric Kruskal-Wallis test to confirm these results. All conditions for this test are met (i.e. non-normal interval variable and independence of observations) (Laerd, 2018b). The test shows that there is a statistically significant difference in performance between the types of labels (H(2)=46.60, p=0.000). This means that hypothesis 1b is accepted, which leads to the conclusion that the performance of de novo entrants is worse than the performance of the other entrants based on the average number of releases.

For H1c and H1d a subset of the sample is used since there is a focus on the different types of subsidiaries (ntot=97, nsm=40, nsi=57). Firstly, a binomial test is performed to check whether it can be assumed that these two types of record labels have an equal presence in the subset of the sample (see appendix 8.4.4). For this test, a proportion of 0.50 was established, leading to a two-tailed test (Field, 2009). The outcome of this test is not statistically significant (p=0.104). This allows for the conclusion that there is no difference in size between the two groups. This means H1c is rejected, as both types of de ipso entrants are equally likely to enter.

Table 3. Results one-way ANOVA

(34)

To be able to conclude if there is a difference in the performance of these types of firms, as was predicted in H1d, an independent samples t-test is performed, for which all the conditions are met (Keller, 2012). Levene’s test is passed meaning that the variances can be considered equal. From the results of the t-test, it can be concluded that there is no significant difference in performance between sublabels (t(95)=0.265, p=0.792). Again, a non-parametric test is performed to confirm these results. A Mann-Whitney test indicates that the average number of releases after entry was not significantly different for subsidiaries from majors or subsidiaries from established independent labels (U=909, p=0.090) (Laerd, 2018c). This means that hypothesis 1d is rejected, as it cannot be inferred from the results of the statistical analyses that there is a significant difference in the performance between the two types of record labels that entered the industry horizontally.

4.1.1 Additional Analysis: Long-term and Short-term

To improve the understanding of the dynamics that may have played a role in the likelihood of new entry and the possible success of new record labels, several additional analyses are performed. Firstly, Rothschild (1947) proposed that the main objective of firms in an oligopolistic environment is long-term firm survival. However, in an environment that is as competitive and uncertain as the music industry, long-term survival might be an overly optimistic goal. It might be the case that new record labels aim at goals that are more easily attainable, such as surviving their first years in business. Hence, I will also look at the number of short-term and long-term releases per label, to see if there are significant differences between these timespans. For the short-term period, the number of releases in the first 3 years of a label’s existence will be added up. For the long-term period, the number of releases in the 4th

year up until the 6th

year will be added up. The average of these values will be calculated only for the record labels that were active for either more than 3 or 6 years. From the main sample of 426

(35)

labels, there were 392 record labels that existed for at least 3 years (ntot= 392, nsm=37, nsi=54, nni=301). Figure 2 shows that the de ipso entrants released a lot more music on average than the de novo entrants and that the subsidiaries of independent labels are slightly more productive in the first three years of their existence when compared to the subsidiaries of majors. Similar to H1b, a one-way and Welch ANOVA, and a Kruskal-Wallis test are performed respectively (see appendix 8.4.6). All the tests show that there is a statistically significant difference in the performance of the 3 types of labels in the first three years of their existence. For the long-term timespan, 52 firms were discarded because they did not exist for at least 6 years, which led to a subset consisting of 340 labels (ntot= 340, nsm=31, nsi=47, nni=262). In the three subsequent years, it is apparent that sublabels from majors got more productive, as can be seen in figure 2. Whereas the sublabels from indies and the de novo entrants reduced their number of releases. Again, a one-way and Welch ANOVA and a Kruskal-Wallis test are performed. The tests also indicate that there is a significant difference in performance between the types of labels in the long-term timespan.

(36)

4.1.2 Additional Analysis: Electronic Music

To control for the fact that some music is easier to produce and release than other types of music, I opted to rerun the analyses for hypothesis 1b with an adapted dataset. It was suggested that electronic styles of music require less effort in production than other types of recordings (e.g. classical or rock) (Colonna, Kearns & Anderson, 1993; Bourreau et al., 2012). Hence, I expect that the output of electronic-focused labels will be larger, as the production time and cost are notably lower. Therefore, all the labels that predominantly release electronic music were excluded from the sample. To acquire this subset, I used the information gathered for hypothesis 2. Every label that released a recording that was categorized as electronic as their first release was discarded, leading to a sample of 319 labels (ntot= 319, nsm=37, nsi=40, nni=242). The tests for H1b (a one-way and Welch ANOVA, and a Kruskal-Wallis test) are replicated to see if there are altered results (see appendix 8.4.7). The tests show that there still is a statistically significant difference in average output between the types of labels. Allowing for the conclusion that the performance of the record labels remains approximately the same when controlled for the fact electronic music is easier to release and produce.

4.1.3 Additional Analysis: Billboard Chart

Another method to measure the performance of a record label is to see whether their releases had great commercial success. This could be measured by the frequency with which recordings enter renowned hit charts (Dowd, 2004a). The Billboard Top 100 is a weekly chart that contains the top 100 most popular songs based on radio airplay, sales data, and streaming activities in the United States (Trust, 2013). Information from Billboard charts has been used frequently in past research on the music industry (Peterson & Berger, 1975; Lopes, 1992; Burnett, 1996; De Laat, 2014). A successful performance for a record label could entail one of their releases entering this chart, essentially because exceptionally large returns could be attained when a song occupies a top rank in this chart (Rothenbuhler & McCourt, 2004). To

(37)

gather this information, every label in the original sample was connected to their releases through various identifiers from MusicBrainz database. Then a copy of the Billboard database was obtained, which included all the Top 100s since 1958 up until 2015 (Billboard, 2018b). These information sources were combined to create two new variables. The first variable consists of the total number of unique songs per record label that entered the top 100. From the descriptives in table 4, it can be inferred that the subsidiaries of independent labels perform exceptionally well. Whereas, the de novo entrants have a much harder time, with only a very small percentage of songs able to reach the chart. The second variable consists of the total number of weeks for all songs per record label in the Top 100. Similar outcomes were attained for this variable, as independent de ipso entrants outperform the other entrants, as can be seen from table 4. For both variables, one-way and Welch ANOVAs and Kruskal-Wallis tests were performed (see appendix 8.4.8). These provide the same outcome as was offered in hypothesis

1b, namely that the

performance of the three groups is significantly different.

4.2 Hypothesis 2

For this analysis, the difference between a new subsidiary of a major label and a new independent label is researched. The new independent label could be either a subsidiary of an established independent (de ipso entrant) or a completely new label (de novo entrant). The statistical tests will be performed on a subset of the sample, to only include firms that have diversified their output (ntot=190, nsm=38, nind=152). Since this hypothesis concerns a different dependent variable, two tests were used to check for normality (see appendix 8.4.1). It was established that this dependent variable is not normally distributed. Therefore, additional non-parametric tests are used to confirm the findings. To be able to determine if there is a difference

Start me up : new entry of record labels in the music industry