Cultural Industries: Product–Market Characteristics, Management Challenges and Industry Dynamics

During the last decade, cultural industries have grown in economic importance, and research interest in them has increased. Despite prolific research, there is a lack of a comprehensive view on the subject. The purpose of the present paper is to offer a reconceptualization of cultural industries by tracing their boundaries, their features and the dynamics that follow from these features. This is achieved through a review of 314 cultural industries studies, whereby a classification system of three main and six sub‐categories is constructed. On the basis of the review, a framework for future research is presented. Most importantly, future research should examine selection criteria and selection performance, and explore the relationships between tastes, sales, diversity and quality. In this way, researchers might be able to create some order in the extreme uncertainty that cultural industries managers face.


Introduction
Research on cultural industries has become increasingly popular. This has been motivated by the growth of the economic importance of such sectors (e.g. United Nations 2010), by the introduction of cultural and creative industries policies (e.g. DCMS 2008) and perhaps also by researchers' passion for and interest in sectors such as film, music and video games. Despite increasing popularity, we lack a comprehensive view of the subject. Some cultural industries researchers emphasize the selection system through which cultural goods need to pass to reach audiences (e.g. Hirsch 1972), while others highlight the inherent uncertainty concerning the success potential of any cultural good when they do so (e.g. De Vany 2004). For yet others cultural goods are defined by their non-utilitarian value (e.g. , while many have expressed interest in the conflicts between artistically motivated creatives and financially motivated managers (e.g. Eikhof and Haunschild 2007). Despite a vast and varied body of work, it remains unclear what characteristics make industries cultural, which industries belong under the cultural umbrella, and what phenomena are specific to them. The purpose of the present paper is to offer a reconceptualization of cultural industries by tracing their boundaries, their features and the dynamics that follow from these features. Based on this, a framework for future research is presented.
Based on a review of 314 cultural industries studies, this paper proposes the following definition: Cultural industries are those that produce experience goods with considerable creative elements and aim these at the consumer market via mass distribution. The creative elements consist of stories and styles, and they serve the purposes of entertainment, identity-building and social display. Mass distribution refers to storage and delivery where economies of scale play an important role. On the basis of this definition, the activities belonging under the cultural industries concept include film, music, book and magazine publishing, TV and radio, fashion and video games.
There are two features that are specific to cultural industries. First, there is a persistent oversupply of creative labour, which is independent of economic cycles (e.g. Hesmondhalgh 2002;Throsby 2001), i.e. there are many more aspiring singers, actors and fashion designers than the respective markets can support. Secondly, there is extreme uncertainty regarding the success potential of any specific product (e.g. De Vany 2004;Pratt 2008) because, prior to consumption, consumers cannot have complete information about the product (e.g. Towse 2003b). These two features result in a system of continuous overproduction, where a variety of gatekeepers restricts the access of cultural products to audiences (e.g. Hirsch 1972). At product level the objective of cultural industries research is to explain why certain products reach the consumer market, while others do not, and also to explain the differential sales of those that do. At organization level, the objective is to explore the organizational interactions taking place in cultural production, and the effects they have on the end product. At industry level, the objective is to study the effects that industry structure has on creative outcomes. Taken together, these research directions comprise the study of how the existing system of cultural production enables and constrains creativity.
In the framework for future research, it is proposed that research attention should be directed towards (1) selection performance as a function of selection criteria taking shape through firm strategy, resources and sensemaking, and (2) the relationships between consumer tastes, diversity, quality and sales. The framework also highlights several additional research areas, including industry structure and the division of labour in selection, which in turn influence the diversity of outputs, the balance of artistic and profit motives, and the role of revenue-sharing arrangements in distributing incentives.

Cultural industries definitions
Research on cultural industries has been conducted in several academic disciplines. Hesmondhalgh and Baker (2011, pp. 53-54) make a distinction between three approaches: (1) the political economy of culture, concerned with how cultural production is financed and organized, and the effects of this on power and social justice; (2) organizational, business and management studies on the specifics of cultural production including innovation, organizational structures and the management of creative labour; and (3) cultural studies aiming to study popular culture as a social phenomenon. This review focuses on the second stream of research, i.e. organizational, business and management studies on cultural industries. Thus, it excludes normative research traditions, including political economy, critical studies and policy research. Moreover, it excludes economic geography because it forms a wide body of work deserving its own review and because it is closely connected to policy issues.
Many cultural industries definitions begin from the nature of the value of cultural goods. Such goods offer a low level of utilitarian value, and a high level of aesthetic (Hirsch 1972;Power 2002), symbolic (DeFillippi et al. 2007;Throsby 2001), social meaning (Markusen et al. 2008) and social display value (Bourdieu 1984;Scott 1999a). These cater to the needs of entertainment and identity-building. Music, films, games and novels serve the purposes of entertainment: they are activities in which people engage for amusement and to experience emotions that daily life may lack. They also serve as social markers and signal identity. Therefore, preferences for particular types of films, literature or fashion apparel are affected by choices made by others in the social network (Potts et al. 2008). Moreover, cultural goods are experiential, meaning that they are judged on their ability to offer fun, enjoyment and pleasure, and, prior to consumption, consumers cannot know whether the goods will succeed in delivering these . This also means that innovation plays a special role in cultural industries: innovation does not aim to improve performance or functionality, but to constantly introduce different types of goods to please diverse and fickle tastes (Aksoy and Robins 1992;Hirschman and Holbrook 1982;Power and Scott 2004).
While many classifications are used for cultural industries, a strong consensus exists regarding the inclusion of film, music, book and magazine publishing, theatre and opera, TV and radio, and the fine arts (Cowen 2000;Hirsch 1972;Throsby 1994). Many authors include advertising, architecture, design, new media, heritage and crafts, fashion, video games and photography (Hesmondhalgh 2002;Scott 1999a;Towse 2003a). Some also include festivals, jewellery, furniture, tourism, toys and perfume (DeFillippi et al. 2007;Power 2002). All these products contain some

42
M. Peltoniemi symbolic content, but for the purposes of this review, only products whose value stems primarily from symbolic content are included. I propose a definition of cultural industries that incorporates industries producing experience goods with considerable creative elements and aiming these at the consumer market via mass distribution. The creative elements consist of stories and styles, and they serve the purposes of entertainment, identity-building and social display. Mass distribution refers to storage and delivery, where economies of scale are important. 1 On the basis of this definition, the activities belonging under the cultural industries concept include film, music, book and magazine publishing, TV and radio, fashion and video games. During the last decade, research on cultural industries has been complemented by the concept of 'creative industries'. Many authors use these terms interchangeably (Peterson and Anand 2004;Thompson et al. 2007;Towse 2003a). In the present review, studies using cultural or creative industries conceptualizations are included because they do not form independent populations. However, those studies that explore activities outside my definition are excluded.

Method
First, a search was conducted in ISI Web of Science, using the search terms 'film industry' OR 'music industry' OR 'book publishing industry' OR 'magazine publishing industry' OR 'fashion industry' OR 'games industry' OR 'game industry' OR 'cultural industries' OR 'creative industries'. The search was limited to the fields of management, business, sociology and economics. Book reviews, editorial material and review papers were excluded. A total of 682 results were obtained, and those that (1) were in English, (2) had been published in a journal, (3) contained empirical work, and (4) acknowledged either cultural or creative industries as a meaningful entity were selected for further analysis. A list of 196 articles was the result. While analysing these papers, citing paths both forwards and backwards were tracked for several steps, in order to achieve comprehensive coverage. Regarding citing and cited papers, articles fulfilling conditions (1)-(4) above were included. The second step resulted in a list of 118 additional studies. The resulting 314 empirical studies form the backbone of the present review.
The next phase involved coding the selected studies according to research themes and empirical contexts. After careful analysis, three main themes emerged. The first concerns research on creative products and how consumers perceive them, the second focuses on people engaged in creative work, and the third discusses how cultural industries are organized and how this affects creative outcomes. Under the first theme, research on consumer tastes, product popularity and gatekeeping mechanisms is reviewed. The second theme covers the tension between artistic and profit goals, interaction between artists and managers, and labour issues specific to creative work. The third theme includes research on interactions between firms within cultural industries, and industry structure and its effect on creativity. The empirical contexts consisted of seven classes. Table 1 shows the classification. Detailed tables on studies concerning each theme are available from the author.

Taste and popularity
Four interrelated assumptions are made in cultural industries research concerning demand. First, cultural goods are experiential, and consumers have a desire for novelty . Secondly, most cultural industries operate under increasing returns. While producing a film, for example, may be extremely expensive, further copies can be made at negligible cost. Thirdly, there is extreme variance in sales, i.e. products diverge into hits and misses (see Hesmondhalgh 2002). Finally, demand is unpredictable in the extreme ('nobody knows anything' 2 ), and therefore, it is impossible to identify hits in advance (Caves 2000). Research relating to these assumptions diverges into research on tastes and popularity. The former looks at the preferences that consumers report, and the latter at the sales that goods with particular characteristics achieve. 1 Sectors that rely mainly on live performances are excluded from the review, because such sectors have fundamentally different dynamics. In many works, the performing arts are included under the cultural industries umbrella, but the author believes that they would deserve a review of their own. 2 'Nobody knows anything' is a catch-phrase within the film industry describing the unpredictability of the consumers' response to any film. It originates from screenwriter William Goldman (Goldman 1983).

Cultural Industries 43
Taste. Taste in aesthetic and hedonic consumption remains a relatively under-researched area (Hoyer and Stokburger-Sauer 2012). So far, most attention has been targeted at social class, cumulative learning and various demographic variables. In the spirit of Bourdieu, the upper classes have been found to have more highbrow tastes, and the lower classes more popular tastes (Denisoff and Levine 1972;Favaro and Frateschi 2007;Hall and Blau 1987;Ragot 2010;Redondo and Holbrook 2010). A specific topic of interest has been the hypothesis of 'omnivorousness', which states that social elites enjoy many types of cultural products, while lower classes tend to enjoy only a specific type (Katz-Gerro et al. 2009;Lizardo and Skiles 2009;Peterson 1992;Peterson and Kern 1996). In addition to taste being affected by social class, taste is conceptualized as a cumulative variable that evolves through consumption and a taste for a particular type of music, for example, is acquired rather than pre-existent (Blaug 2001;Cowen 1989;McCain and Towse 2003;Throsby 2001). This is supported by the tendency of consumers to become more critical in their assessments over time and of consumers specializing in particular genres to rate such products more highly (Moon et al. 2010). Several studies have also shown a nostalgia effect in tastes. Mature consumers tend to prefer musical styles that were popular in their youth (Holbrook 1993;Holbrook and Schindler 1989;Schindler and Holbrook 2003). The impressionable years vary according to sector: for books, the preferences stabilize later than for music (Janssen et al. 2007). Youngsters also inherit preferences from their parents (ter Bogt et al. 2011).
Contrary to the acquisition of cultural taste view, there are scholars who take the view that preferences are rather innate, pre-dating consumption (Kivetz et al. 2008) and evolve in their dormant state (Simonson 2008). This view is supported by the ability of the neural responses of individuals to predict future sales of music (Berns and Moore 2012) and consumers reporting relatively stable genre preferences over time (Mulder et al. 2010). Miesen (2004) has shown that readers reporting genre preferences also prefer certain types of texts, implying that genres have distinct attributes that cause them to be chosen. Alderighi and Lorenzini (2012) reconcile the contradiction of cumulative tastes and inherent tastes by dividing consumers into two populations: those interested in accumulating cultural capital by exploring in order to enjoy future cultural offerings more, and those interested in immediate enjoyment. It is, however, important to note that taste and demand are separate variables: demand forms as a combination of taste, price, information and other factors, whereas taste either forms cumulatively through consumption or is inherent.
In addition to studying the emergence of cultural taste, there are some studies on consumer preferences for specific product characteristics. Consumers seem to prefer films with younger, opposite-gender leading stars, older directors and unfamiliar temporal settings (Addis and Holbrook 2010). This means that star power comes from romantic attraction, there is a reverence for more mature directors, and films help viewers to escape from ordinary life. Contrary to these results, Millar (2008) finds that there is a bias in artist preferences towards male artists, especially by male consumers. Musical preferences have been found to be based on various functions, such as physiological arousal, self-reflection, communication and coming together with like-minded people The 'Cultural, creative' class includes studies that contain data relating to a large number of industries without specifying each of them. b Some studies cover more than one theme and industry. They are calculated only once in the totals.

44
M. Peltoniemi Sedlmeier 2009, 2010). In terms of international comparisons, some nationalities have been found to be more susceptible to particular genres and to social influence (d' Astous et al. 2005), and to be more consistent than others in their preferences (Wilken et al. 2011). However, there are reportedly greater differences between individuals than between nationalities in responding to different types of cultural goods (Ko et al. 2007). Both cumulative taste formation and inherent tastes indicate that changes in preferences are incremental rather than radical. This contradicts the assumption of extremely unpredictable demand. These findings suggest that the source of the unpredictability does not necessarily lie in tastes, but rather in the ways in which information diffuses among consumers or in the uncertain outcomes of the production processes in which goods are created. Moreover, research on tastes has largely ignored the question of identity-building. For example, whether tastes are affected by social class or by the will to belong to a particular social class is not specified.
Popularity. Cultural goods tend to diverge into hits and misses and this is confirmed by research on sales distributions (Black et al. 2007;Walls 2005Walls , 2009a. This has encouraged researchers to explore various factors explaining sales. The highest level of support concerns the logic that sales result from information cascades through which consumers inform each other about products' qualities (Beck 2007;De Vany and Lee 2001;De Vany and Walls 1996;Gaffeo et al. 2008;Gazley et al. 2011;Giles 2007;Godart and Mears 2009;Moul 2007;Salganik et al. 2006;Salganik and Watts 2008). The intention to purchase is also important information circulating among consumers (Broekhuizen et al. 2011). Some consumers may become trendsetters whose opinions are considered especially valuable (Suzuki and Best 2003). Producers try to create positive word of mouth by reporting box office performance (Malhotra and Helmer 2012) and investing in advertising (Bruce et al. 2012). In research on DVD film sales, the information cascade view has received both support (McKenzie 2010) and opposition (Walls 2010).
The second factor predicting popularity is signals, i.e. superstars and exotic locations (Brewer et al. 2009;Elberse 2007;Gazley et al. 2011;Ravid 1999). According to Nelson and Glotfelty's (2012) analysis, replacing an average actor with a superstar increased film revenue by $16,618,570. The superstar phenomenon is also seen in the distribution of royalties to members of Performing Rights Organizations (Pitt 2010) and in concert tour revenues (Black et al. 2007). But what makes someone a superstar? Giles (2006) concludes that small differences in ability translate into large differences in success. Unfortunately, the additional sales generated by superstars tend to be smaller than their cost, rendering their films poorer financial performers (De Vany and Walls 2004;Walls 2009b). Moreover, De Vany and Walls (1999) find no connection between superstars and sales, and Hadida (2010) finds that stars increase sales only if the production budget is large.
The third explanation for popularity rests on the balance between novelty and familiarity (Alvarez et al. 2005;Cillo et al. 2010); consumers need a level of familiarity to understand a product and a level of novelty to enjoy it. In consequence, book-based films perform better than others during the opening weekend (Joshi and Mao 2012), and greatest hits albums achieve better sales the more recently the artist has been active (Ordanini and Rubera 2004). Sequels usually outperform other films Brewer et al. 2009;Chang and Ki 2005;Dhar et al. 2012;Walls 2009b) and are more likely to be issued on DVD (Nelson et al. 2007). This could result from only successful original concepts generating sequels, but also from consumers preferring cultural products with which they are already familiar. In a similar manner, a cultural product's loyalty to established genres has been found to increase its appeal (Hsu 2006b). Cultural familiarity has also been found to increase box office success (Meiseberg and Ehrmann 2013), but not rental success (Hennig-Thurau et al. 2006b). Despite mounting evidence of consumers' preference for familiarity, the exceptionally successful products tend to be highly innovative (Hsu et al. 2012), and entry into new content areas where competitors are not particularly good tends to lead to financial success for the firm (Shamsie et al. 2009).
The final group of explanations for product popularity involves distribution (Brewer et al. 2009;Chang and Ki 2005;Eliashberg et al. 2006), price (Clements and Ohashi 2005;Gazley et al. 2011;Mixon and Ressler 2000;Orbach and Einav 2007), production budget (Brewer et al. 2009;Chang and Ki 2005;Gemser et al. 2012), MPAA 3 rating (Ravid and Basuroy 2004) and timing in relation to season (Beck 3 Motion Picture Association of America (MPAA) rates films as to whether the content is suitable for children. 45 2007;Brewer et al. 2009;Cartier and Liarte 2012;Chang and Ki 2005;Clerides 2002;Einav 2007Einav , 2010Krider and Weinberg 1998). It is a common finding that the sales of cultural goods would benefit from a more balanced timing over the year, as opposed to concentrating on the Christmas and spring seasons. The findings on popularity raise three concerns relating to the unpredictability of demand. The assumption of the extreme unpredictability of future sales is supported by the lack of correlation between popularity and quality (Hamlen 1994;Holbrook and Addis 2008). This begs the question whether there is a correlation between taste and quality. Perhaps taste correlates with quality, but there are limiting factors (price, lack of information) that complicate the relationship between quality and sales. Secondly, the idea of information cascades, where consumers discover what they like and inform their friends about such products (De Vany 2004) assumes that friends have similar tastes. This research stream could be complemented with data on tastes and the content of the information cascades. Thirdly, are more familiar products less sensitive to information cascades?

Gatekeeping
Gatekeeping is a central concept in the production of culture view (Hirsch 1972;Peterson and Anand 2004). Owing to persistent oversupply, the experiential nature of the products and consumers' limited sensemaking resources, a filtering system is required to reduce the number of products reaching the market. Moreover, because of extreme variance in sales and heavy upfront investment, production companies are compelled to reject most aspiring artists. Selectors referred to in the literature include (1) upstream selectors, i.e. the companies with whom creative artists make contracts (recording labels, film studios) and who finance creative projects; and (2) downstream selectors, including media, retailers, charts, critics and award committees. Here we see upstream selectors as those who decide whether a product reaches completion, and downstream selectors as those exerting influence on whether a finished product reaches an audience.
Upstream selectors. Upstream selectors may be divided into large generalists and small specialists, who focus on certain styles or local tastes (Hitters and van de Kamp 2010). Despite their smaller financial resources, specialists may succeed due to their superior cultural capital (Scott 2012) and dense information-sharing networks (Foster et al. 2011). There is evidence that artists pre-selected by specialists are slower to succeed, but their success is longer lasting than that of artists selected by generalists (Ordanini 2006). According to studies with fieldwork from selection meetings, selections are made on the basis of an artist conforming to a genre, not imitating other artists, having a good reputation (Mauws 2000), and conforming to a creative prototype (Elsbach and Kramer 2003). A seemingly non-biased method of selection in the music industry is to use success in the singles market to determine whether an artist is qualified to enter the more cost-intensive albums market . There is also evidence that different funders apply different criteria in their selection processes as to whether artistic or financial reputation is emphasized Wijnberg 2012a, 2012b). Moreover, the power to select may transfer between actors in a value chain (Grantham and Kaplinsky 2005).
The bulk of the quantitative research on upstream selectors has focused on advertising and its effect on sales. Large advertising budgets increase sales ), but the returns on advertising depend on the product: for films of average quality the returns tend to be negative (Elberse and Anand 2007) and for sequels positive . Hennig-Thurau et al. (2006a) show that the returns on advertising are modest making it very expensive to save a bad film. Similarly, McKenzie and Walls (2013) find that heavily promoted local films earn less than imported ones with smaller advertising expenditure. Logically, advertising budgets are allocated to products that have already succeeded in other markets (Elliott and Simmons 2008). Moreover, investors evaluate firms on the basis of the ratio of sales to advertising and Therefore, exaggerated hype runs the risk of lowering company value (Joshi and Hanssens 2009).
Research on upstream selectors has so far mostly omitted selection performance, i.e. the ability to maximize sales by selecting artists and goods closely corresponding to consumer tastes. What types of selectors achieve the best performance? Do specialists outperform generalists? What is the relationship between selection criteria and selection performance? Do certain criteria lead to better performance? What resources are required of selectors for adopting a particular criterion? Some firms are more successful than others, indicating that they are not merely 'throwing mud at the wall and seeing what 46 M. Peltoniemi sticks' (as articulated by Hesmondhalgh 2002). Are there persistent performance differences between firms, and can they be explained by different resources and strategies? Empirical work concentrates on advertising decisions. Little research has been conducted on the selection rounds, such as budget allocations and decisions to axe projects already in production occurring in production organizations before products enter the market. Are resources allocated to high-quality projects? Or do the projects allocated with the largest resources showcase higher output quality? Moreover, research on selectors has ignored the processes through which the power to select is assigned. The transfer of power between selectors has been documented (Grantham and Kaplinsky 2005), but there is a lack convincing explanations. Does it follow from changes in selection performance?
Downstream selectors. Empirical work on downstream selectors focuses on how critics affect sales. Positive critical reviews increase sales (Brewer et al. 2009;Eliashberg and Shugan 1997;Gazley et al. 2011;McKenzie 2009), and this effect grows with product age (Eliashberg and Shugan 1997), tending to be greater for less familiar (Desai and Basuroy 2005) or highbrow and 'artsy' goods (Gemser et al. 2007;Reinstein and Snyder 2005). Critics, however, define products as either highbrow or lowbrow (Zuckerman and Kim 2003). These classifications also influence export sales (Lee 2009). The findings on whether positive or negative reviews are more influential are mixed. Negative reviews have been found to influence consumer expectations more (Suarez-Vazquez 2011), whereas positive reviews have been found to influence sales more (Hennig-Thurau et al. 2012). Negative reviews may also contain information conducive to buying (Boatwright et al. 2007), and sales may be aided by extremely contradictory reviews (Clement et al. 2007). Moreover, personality traits have an effect on whether a consumer listens to a critic (d' Astous and Colbert 2002) and there are national differences regarding the influence that critics exert (d' Astous et al. 2005). In addition to critical reviews, awards, such as the Oscar for the best picture, have been shown to increase sales (Brewer et al. 2009;Zhuang et al. 2011), and awards have different effects on consumer behaviour, depending on whether they are granted by consumers, peers or experts (Gemser et al. 2008).
Apart from professional evaluations, consumers can access information from lay critics posting their reviews online. Lay critics are more influential concerning niche products (Zhu and Zhang 2010), and consumers interested in many genres and styles are likely to listen to both professional and lay critics (Verboord 2010). Information provided by lay critics does not necessarily differ from professional offerings: the more experienced lay critics are, the more their reviews concur with those of experts (Plucker et al. 2009). The effect of lay reviews has been tested in terms of valence and volume. For mainstream products, volume appears more important, and valence for niche goods in influencing sales (Yang et al. 2012). At national level, volume has been found more important, and valence in local settings (Chintagunta et al. 2010). Concerning blog posts, volume seems to be the crucial variable affecting sales (Dhar and Chang 2009).
Owing to the impact of critics' and other selectors' assessments, there is interest in possible bias. Critics have indeed been found to prefer products from certain producers (Hsu 2006a;Ravid et al. 2006), and distributors prefer earlier collaborators despite their poorer sales (Sorenson and Waguespack 2006). Moreover, MPAA tends to give more lenient ratings to films by their members than to independent films (Waguespack and Sorenson 2011). Also, spending on advertising in a particular magazine increases the coverage of a firm's products in it (Rinallo and Basuroy 2009). However, producers also employ a range of strategies to influence the selectors' decision-making. Ample attention has been paid to payola 4 , even though empirical studies claim that the effects of payola were mild (Rossman et al. 2008). Furthermore, following the criminalization of payola, major record companies discovered other ways of exerting influence, such as fraternizing with downstream selectors (Fairchild 2012;Mol and Wijnberg 2007). In addition to trying to influence downstream selections, upstream selectors make production decisions that anticipate preferences among downstream selectors (Negus 1993). Downstream selectors may also signal their preferences in ways designed to affect decisions made by upstream selectors (Readman and Grantham 2006). 4 Payola means 'incentives -monetary or otherwise -offered by selected (in our case, the record companies) to selectors (in our case, the DJs) to cause the selectors to favor the products of these selected' (Mol and Wijnberg 2007).

Cultural Industries 47
Research shows that certain downstream selectors are better in tune with consumer tastes and more powerful than others in influencing purchasing behaviour. However, research on the connections between critical reviews and sales risks comparing apples with oranges: the tastes of one group (critics) being compared to the purchasing behaviour of another group (consumers) neglects important moderating variables. Critical reviews as expressions of taste may have the power to change consumer tastes, which in turn influence consumer demand. However, critics are under pressure to appear knowledgeable and Therefore, try to align with consumer tastes. Therefore, there may be a two-way relationship between critic and consumer tastes.

Art for art's sake and art for profit
The tension between artistic ambition and financial gain often features in the cultural industries literature. Hirsch (1972) already noted that the production process shapes artistic expression because artists and distributors have different critical standards.  also point out that managers need to reconcile artistic ambitions and the economics of mass entertainment. Therefore, creative individuals need to conform simultaneously to both artistic and financial logics in cultural production (Durand and Jourdan 2012).
Reconciling artistic and profit goals. Whether researchers perceive art and profit as mutually exclusive or complementary goals is controversial. In the former perspective, the consensus is that the profit motive is absent in sectors where profit opportunities are minimal (Bayma 1995;Chaston 2008;Eikhof and Haunschild 2006), and artists prefer wide distribution to profit (Gayer and Shy 2006). In terms of company strategy, managers choose between approaches driven more by artistic intent, and approaches focusing on market considerations (Cillo and Verona 2008). Based on findings by Tschang (2007), firms are not actually free to make such choices, but forced to conform to a rational commercial strategy. However, firms may create separate artistic and commercial product lines with different branding (Phillips and Kim 2009).
Research on art and profit as complementary goals finds that artists value a combination of artistic freedom and pecuniary rewards (Cowen and Tabarrok 2000;Throsby 2010). The supply of artistic work has also been found to respond to economic incentives (Hui and Png 2002), and even though money is not palpably the primary motivation, it can be a validation for creative skills (Taylor 2012). Moreover, research on authenticity finds that there are financial penalties associated with appearing to pursue quick profits and therefore firms should encourage artists to offer distinct approaches to be profitable (Jones and Smith 2005;Peterson 1997;Shusterman 1999;Svejenova 2005).
These two views of art and profit as mutually exclusive and complementary goals hint at the existence of varying levels of domination of these goals, and varying combinations ending up with different results. An artist with a pure art motive teamed up with a manager with a pure profit motive is hardly likely to perform well on either dimension. Research is scarce on locating the goals or strategies of both parties (artists and managers) somewhere on a continuum between art and profit, with the outcomes measured in terms of the levels achieved on each dimension. This would enable analysis of the extent to which these goals complement or contradict each other.
'Suits' and creatives. The unique management challenges in cultural industries follow from the relentless creation of new genres, formats and products (DeFillippi et al. 2007), virtually unlimited aesthetic choices (Thompson et al. 2007), and difficulties in monitoring and controlling such processes . Additional management challenges may also result from the ongoing convergence of different cultural sectors (Gandia 2013). Hesmondhalgh (2002) states that symbol creators are granted considerable autonomy because of the need for originality and novelty. Empirical findings confirming this are limited to raising the project size (Baker and Faulkner 1991) and the complexity and non-linearity of creative projects (Baba and Tschang 2001;Tschang 2005) transferring power from corporations to creatives. Managers and corporations have the power to both redirect and block creative activity. This is done by giving very little leeway and firing underperforming creatives (Peterson and Berger 1971), practising portfolio management (Negus 1998) and redefining creative work as R&D (Roberts 2010). Such control mechanisms decrease autonomy, which creatives see as a marker of 'bad work' (Hesmondhalgh and Baker 2011). Creatives

48
M. Peltoniemi may undertake identity work strategies, such as distancing, to maintain artistic integrity in work assigned by management that does not meet their quality standards (Wei 2012). In companies engaged in cultural production, artists and managers must collaborate. Artists are dependent on support personnel, and this constrains their range of artistic options (Becker 1974). Research on teamwork in the cultural domain focuses on integration and negotiation. The integration of artistic and financial goals has been found to enable both creative outcomes (Cappetta and Cillo 2008;Lingo and O'Mahony 2010) and financial success (Cohendet and Simon 2007;Ordanini et al. 2008;Yamashita and Masaru 2006). Such integration takes place via continuous microcommunications between professional groups within organizations (Mora 2006). The growth in creative activity can be handled via dual leadership, i.e. a combination of artistic manager and production manager (Hotho and Champion 2011), or by recruiting managers with creative backgrounds. This tends to lead to higher innovative performance for the firm than would be achieved via recruiting managers with management background (Camelo-Ordaz et al. 2012). When creative firms grow, the ratio of management to creative staff tends to increase (Autier and Picq 2005). Despite the controlling mechanisms, executives experiment with new content types, especially in the early stages of their careers. Unfortunately, experimentation is often less successful early on and improves with tenure (Miller and Shamsie 2001).
The power relations between managers and artists offer a moderating variable for the analysis of the relationships between art and profit motives and related outcomes. Artists and managers enjoy varying levels of power in different firms. Does the level of power exercised by each party moderate the relationship between art and profit goals and art and profit outcomes? Such a research setting would enable us to examine the value of control. The cultural industries literature treats control exercised by managers as something inherently negative. But can control also add value to the creative process? There is some preliminary evidence of the importance of managerial capabilities (Chaston and Sadler-Smith 2012) and of the advantages of a certain level of control for creative performance (Barkey and Godart 2013). Do the benefits of control increase with manager tenure?

Creative labour
Inferior working conditions. Owing to the permanent oversupply of creative labour, employment in the cultural industries often means low wages, long working hours and structured job insecurity (Hesmondhalgh 2002;Miège 1989). Even successful creatives may make only a negligible income, especially in certain fields such as poetry (Dubois 2010). Formal training in the cultural professions does not predict individual financial success as it does in many other professions (Throsby 2001). Moreover, the majority of profitable creative work is performed by a minority of artists (De Vany 2004;Menger 1999). Consequently, Oakley (2006) and McRobbie (2002) claim that the creative industries exacerbate existing patterns of economic inequality, and many creative graduates fail to become such agents of knowledge spillovers as policy-makers expect (Comunian et al. 2010). Many people working in artistic occupations are willing to accept the downsides, because creative work is not a disutility for them (Caves 2003). Even though the poetry profession, for example, is not lucrative, it is considered prestigious and attractive (Dubois 2006). Being recognized, seen as interesting and glamorous, and the ability to develop one's talent become the basis of exploitation (Hesmondhalgh and Baker 2011). However, creatives tolerate uncertainty and recurring unemployment, because they are repeatedly drawn back to the industry by financial, creative, social and emotional rewards (Rowlands and Handy 2012). This dynamic results in fragile careers for creatives (Hesmondhalgh and Baker 2011).
Quantitative research on creative labour confirms that employees in cultural industries suffer from long working hours and structured job insecurity (Blair et al. 2001;Blair and Rainnie 2000;Storper and Scott 1990). Compared with similar positions in other industries, employees in entertainment are paid less (Wetzels 2008). In addition, employees who are between jobs are expected to work unpaid for latent organizations in the hope of future employment (Ebbers and Wijnberg 2009a). They also suffer from stress, anxiety and poor employment relationships (Hesmondhalgh and Baker 2008). A specific shortfall is the absence of mentors: when everyone is freelancing, old-timers have no motivation to support newcomers (Grugulis and Stoyanova 2011). In contrast to the consensus regarding inferior working conditions, there is also some evidence that creative industries are expanding employers who offer Cultural Industries 49 relatively well-paid jobs (Dolfman et al. 2007), and in terms of unemployment and job security offer above-average jobs (Creigh-Tyte et al. 2003).
Copyright is a labour issue of crucial importance in the cultural industries. The majority of artists earn relatively little from copyright, while the majority of earnings goes to corporations and a small number of superstars (Macmillan 2002;Taylor and Towse 1998;Towse 1999). Therefore, copyright mostly benefits artists who would be able to bargain even without such legislation. Despite copyright law securing publishers' income rather than artists', most artists comply with the law because copyright gives them other benefits, such as status and prestige (Towse 2001). Most importantly, the copyright legislation plays a role in ensuring the continuing dominance of major corporations in cultural industries, which may affect the degree of creativity allowed in the market (Towse 2003a;Tschmuck 2009).
Owing to the widespread adoption of peer-to-peer networks, piracy of copyright material has become an important research topic, especially relating to music and film. However, piracy is also an issue in fashion, where low-quality knockoffs and highquality overruns 5 are common (Barrere and Delabruyere 2011;Hilton et al. 2004). While piracy reduces sales (Ahn and Yoon 2009;De Vany and Walls 2007;Klein and Slonaker 2010;Liebowitz 2008;Zentner 2006), the effect is not necessarily purely negative for the company or the artist. Consumers may want to hear songs before buying and, therefore, illegal downloading enables them to make better-informed decisions and may actually increase sales (Andersen and Frenz 2010;Molteni and Ordanini 2003;Peitz and Waelbroeck 2006). Such additional information increases the chart-survival of superstars and female artists, and decreases the predominance of major labels (Bhattacharjee et al. 2007). Peer-to-peer downloading also increases artist income through increased audiences at live performances (Curien and Moreau 2009;Gayer and Shy 2006). Similarly, Smith and Telang (2009) found that giving away content in one channel stimulates sales in a paid channel, given that that there is some differentiation between the free-of-charge and the paid content.
These findings highlight the role of contracts in distributing incentives in creative work. What determines the revenue-sharing percentages between artists and corporations? Which of these require stronger incentives (via copyright legislation, for example) to ensure high-quality outputs? Would more artist-friendly contracts lead to higher quality? Or should an artist accept a less favourable contract in order to motivate the firm to invest in other resources, such as advertising, to ensure a better outcome? Game-theoretic research on the contracting of artistic work is largely missing from the literature.
Work allocation. Research on work allocation highlights the importance of informal networks. Artists form latent networks, which activate when projects are launched (Starkey et al. 2000). Teams are formed from known collaborators who have received similar earnings and awards (Faulkner and Anderson 1987) and positive critical reviews (Ebbers and Wijnberg 2010). Repeated collaboration allows creatives to exploit and stretch existing capabilities (Manning and Sydow 2011), and mutual trust allows an industry to operate without formal organizations (Johnsen 2011). Performanceoutcome learning through film credits, for example, directs employment opportunities towards the same group of people (Schwab 2009;Schwab and Miner 2008). New team members tend to be chosen when an existing team mostly consists of either highstatus or low-status members (Perretti and Negro 2006). To increase their chances of being selected, artists and their agents undertake reputation work (Zafirau 2008).
Work allocation through path-dependent social networks has been characterized either positively as inclusion, or negatively as exclusion. In an insecure job market, risks are offset by networking (Coulson 2012), and by building semi-permanent groups, which offer artists a source of competitive advantage and collaboration opportunities (Antcliff et al. 2007;Blair 2001Blair , 2003Blair et al. 2001;Lee 2011;Starkey et al. 2000). Creatives have been found to adhere to community obligations and social imperatives that limit the pursuit of personal gain (Banks 2006). The other side of this coin is that such exclusive project ecologies are difficult to enter (Johns 2010;Lee 2011), there are asymmetrical power relations between network members (Blair 2003;Ekinsmyth 2002), and creatives are encouraged to develop only mobile capabilities of value 5 Overruns are garments manufactured by the subcontractor with the original pattern and fabric of the designer after the number ordered by the fashion house has been completed. They are distributed without the authorization or brand of the fashion house (Hilton et al. 2004).

50
M. Peltoniemi also outside the current organization (Ebbers and Wijnberg 2009b). Zwaan and co-workers (2009) found that career success is predicted by contacts with professionals rather than talent. Artists have also reported feelings of anxiety and isolation when social networking with professional competitors becomes an obligation (Hesmondhalgh and Baker 2010;Lange 2011;Lee 2011). Moreover, creative industries are fraught with gender inequality and discrimination: women are more often typecast (Bielby and Bielby 1996), and producers tend to offer breakthrough opportunities to individuals whose gender matches their own (Skilton 2008). In contrast, transferring from bureaucratic to decentralized decisionmaking has benefited women through more numerous and diverse decision-makers (Dowd et al. 2005;Smith-Doerr 2010). What effect does the exclusive network have on product performance? Previous co-productions successful in terms of audience size increase the likelihood of success of future collaborations (Soda et al. 2004;Zaheer and Soda 2009). However, such repeated collaborations may lead to diminished innovativeness, which eventually decreases consumer interest (Soda and Bizzi 2012). From the viewpoint of individual artists, an intermediate position between core and periphery is beneficial for creative outcomes . As regards product sales, combining old-timers and newcomers is beneficial (Ferriani et al. 2009). Closed networks have been found to perform better in domestic markets and worse in international markets (Andersen 2013).
The processes of work allocation treated as either inclusion or exclusion omit the findings showing that repeated post-success collaborations have high levels of performance. This plausibly explains the willingness of artists to stick with old collaborators. However, there is no evidence whether collaborators on failed projects also tend to repeat joint work. If this is the case, it suggests that collaborators are not chosen on the basis of expected rewards, but perhaps owing to personal preferences or inability to recruit alternative collaborators. Therefore, such a test would shed additional light on the decisionmaking processes shaping the networks. A further avenue of research is modelling the sales lost due to exclusion. Current research compares the numbers of male and female artists being selected for distribution, but does not model this process as losses caused by lower average quality of products reaching the audiences. Do exclusive networks decrease the overall sales of cultural goods by impairing product quality or diversity?

Cultural industries as communities
Research on cultural industries as communities has adopted the concepts of industry system (Hirsch 1972;Miège 1987) and organizational field (Anand and Peterson 2000;Anand and Watson 2004;Scott 1999b). Such research focuses on sensemaking in these communities and the interdependence of firms.
Sensemaking within the community. Sensemaking research has focused on popularity charts, events and award rituals. The history of US popular music offers an interesting example of results of changing the method of compiling popularity information. In 1991, Billboard charts adopted the SoundScan method, i.e. point-of-sale information, replacing the previous method of sampling stores for sales data. The change had a significant effect on managers' sensemaking by providing more accurate information about the popularity of different genres and the timing of sales in relation to release dates. This encouraged increasing fragmentation in terms of styles offered within the sector (Anand and Peterson 2000;McCourt and Rothenbuhler 1997). Similar outcomes have been identified with the adoption of the BookScan system in the publishing industry (Andrews and Napoli 2006). Apart from popularity information, other ways of making sense of the field include industry events and award rituals. Trade fairs and fashion weeks generate visibility for the field and also create a continuing global circuit (Entwistle and Rocamora 2006;Klein 2011), while award rituals distribute prestige within the sector and attract attention (Anand and Jones 2008;Anand and Watson 2004). Industry-level sensemaking appears to encourage homogenization within the sector. Styles tend to diffuse rapidly as firms rely on numerous sources to catch winning trends in good time (Aage and Belussi 2008). This type of cohesion is apparent in the evolution of genres, which progress from avant-garde to traditionalist as they pervade the system (Lena and Peterson 2008).
In addition to the SoundScan case, little research is available on changes in sensemaking methods and their consequences. This case showed that a change in sensemaking changes the selection criteria employed as upstream selectors strive to better align Cultural Industries 51 with consumer preferences. Therefore,, sensemaking is directly linked to the selection criteria and selection performance discussed above. Sensemaking should therefore be included in research on selection performance.
Interdependence in communities. Studies on interdependence in communities focus on the actions and outcomes of complementors and competitors in industry creation (Mezias and Kuperman 2001), and the dependence of producers on distributors ) and platform providers (Gallagher and Park 2002;Schilling 2003;Shankar and Bayus 2003). Such effects may also take place between industries, such as the dependence of the Japanese video game industry on creative resources developed in animation and cartoons Izushi 2003, 2004). Local audiences also play a role in industry creation: less-developed tastes enable industry emergence, whereas developed tastes block such opportunities (Barnard and Tuomi 2008). Research on value chain strategies has concluded that vertical integration may result from the need to buffer against competitive pressures (Negro and Sorenson 2006), or 'value-chain-envy', i.e. one stage in a value chain capturing more than its fair share of the profit (Mol et al. 2005). In video game development, however, developers choose which value chain they want to be part of by selecting an appropriate platform (Venkatraman and Lee 2004).
Interdependence between companies may lead to homogeneity in production capabilities and organizational structures within an industry (Huygens et al. 2001). Cappetta et al. (2006) found that convergent and divergent eras alternate. In times of greater turbulence, uncertainty over changing market shares encourages stylistic variation (Miller and Shamsie 1999). Stylistic change may also be precipitated by socio-technical transitions (Geels 2007) or purposeful group action (Bjorn 1981). However, Tschang and Szczypula (2006) state that novelty is never absolute, as new ideas are merely combinations of old ideas already existing within the field.
The interdependence of firms within creative communities centres on similarity and difference. Firms must find an optimum level of differentiation in order to offer a unique product, but at the same time remain sufficiently similar in order to remain legitimate. Such an optimization process is a function of what is already on the market and what resources the firm has. Furthermore, firms often need to anticipate other firms' next season's offerings when optimizing their own level of differentiation. Level of differentiation therefore adds a dimension to the selection criteria discussed above. Consumers require a level of familiarity to understand a cultural product and a level of novelty to enjoy it (Alvarez et al. 2005), but the tastes of different consumer groups may vary widely regarding this balance. Therefore, exploring the level of differentiation would offer an interesting research avenue relating to tastes and selection performance.

Industry structure and creative outcomes
Research on the relationship between industry structure and creative outcomes focuses on three specific phenomena. First, cultural industries are inhabited by majors and independents. Majors tend to go with proven solutions while independents handle specialized styles and new artists (Brunet and Gornostaeva 2006;Burnett 1996;Dubois 2012). Independents have relational assets valuable in identifying new artists and contracting venues which make them superior in supporting early careers (Montanari and Mizzau 2007). Secondly, sociologists have studied the relationship between cultural diversity and industry concentration. The assumption is that majors prefer secure cash flows over artistic values and Therefore, concentration results in cultural homogeneity (see DiMaggio 1977). Thirdly, flexible specialization offers producers access to highly specialized inputs. While this creates opportunities for specialized independents, major corporations still control access to audiences. Finally, digitalization and its effects on industry structure and creativity are attracting increasing research attention. The interesting question is whether digital distribution enables artists to by-pass corporate control, thereby increasing the diversity of cultural goods on offer.
Majors and independents. The existence of majors and independents, i.e. skewed distribution in company size, has been confirmed by several empirical studies (Scott 1999b(Scott , 2002(Scott , 2004. The growing number of small firms shows in firm numbers growing faster than employment (Power 2002) and vertical disintegration also contributes to firm numbers (Storper and Christopherson 1987). However, the majors are growing via acquisitions and partnerships (Huygens et al. 2001), and the level of horizontal integration is rising as entertainment conglomerates acquire and partner with companies 52 M. Peltoniemi from various cultural industries (Arsenault and Castells 2008;Chon et al. 2003). Empirical studies provide three primary explanations for the structure of cultural industries. First, artistic and business resources are placed in different companies to avoid cross-contamination (Gander et al. 2007;Rieple 2002, 2004). Secondly, the requirement for highly differentiated inputs encourages vertical disintegration (Gil and Spiller 2007;Storper 1989Storper , 1993. Such resource partitioning has been documented in the film industry (Mezias and Mezias 2000). Finally, an industry structure of this type allows majors to cut costs and control markets (Aksoy and Robins 1992;Sedgwick 2002). The film industry especially has been found to suffer from a chronic absence of profit, which makes such cost-cutting strategies inescapable (Leaver 2010). Companies tend to be disbanded when projects end and company survival is not a key goal (DeFillippi and Arthur 1998). Historical analysis shows that independent production is not cheaper than studio films, but independent films offer on average superior, yet volatile, financial results (Robins 1993).
The amount of attention paid in the cultural industries literature to the skewed size distribution of firms suggests that it is a feature that differentiates cultural industries from others. However, no comparisons between cultural and non-cultural industries are available to support this view. Most firms tend to be small independent of industry context (e.g. Axtell 2001). Is the size distribution more skewed in cultural than in other industries? Are there differences between various cultural sectors, and what might explain these?
The division of labour between generalist majors and specialist independents suggests that they have differing selection performance as a function of the distance from the mainstream. This is probably not only a question of selection expertise. Majors are better equipped to invest in the development of mainstream offerings and such investments may be recouped through sales in large markets. Independents have less capital to invest, and sales in smaller niches also remain lower. Therefore, they need to rely on their selection expertise rather than investing in production and promotion. Exploring selectionbased and production-based strategies would offer an additional avenue for future research. This is related to the division of consumers into those exploring to accumulate cultural capital and those interested in immediate enjoyment (Alderighi and Lorenzini 2012). How do firm strategies differ when targeting these different populations? Do these populations offer different income potential or creative opportunities for creatives?
Concentration and diversity. Research on industry concentration and product diversity began with Peterson and Berger's (1975) study concluding that eras of high levels of concentration correspond to low levels of diversity and vice versa. In later studies, the consensus has been that the increasing use of semi-independent departments, i.e. an open system of production, has increased diversity even during eras of high levels of concentration (Alexander 1996(Alexander , 1997Benhamou and Peltier 2007;Burnett 1992;Dowd 2004;Dowd and Blyler 2002;Lee 2004;Lopes 1992;Peterson and Berger 1996). Moreover, because the cultural goods of a particular sector do not only compete with each other, but music may compete with books, for example, companies have an incentive to offer heterogeneous products despite high levels of concentration (Greco 1999). However, innovative entry is rare in industries dominated by a few large firms (Crane 1997). Ahlkvist and Fisher (2000) argue that standardization is not a matter of ownership, but a result of large markets. Methods of measuring diversity have become a subject of debate (see Alexander 1996;Peterson and Berger 1996), but most studies still rely on the numbers of artists and products for determining the level of diversity of the goods in the market.
Flexible specialization. Changing industry structures have also been studied from the viewpoint of flexible specialization, i.e. a type of industrial organization where specialized firms involved in production are frequently changed in order to create a variety of outputs (Christopherson and Storper 1989). Prior to the Paramount decision, 6 Hollywood operated via integrated production and distribution. Thereafter, disintegration and flexible specialization diffused to other cultural industries. However, pre-Paramount decision, Hollywood benefited greatly from the flexibility afforded by integrated production and distribution (Hanssen 2010), and even nowadays integrated distribution is preferred for products whose sales are especially difficult to predict (e.g. 6 The Paramount decision given by the US Supreme Court in 1948 outlawed the ownership of cinemas by film studios. This led to the disintegration of the studio system where creatives had long-term contracts with studios, and studios controlled the distribution channels.
In response to the increasing need for flexibility, different organizational outcomes have emerged in different countries (Djelic and Ainamo 1999). Under flexible specialization, firm capabilities required to succeed have changed from property-based and transformational into knowledge-based and mobilizing (Lampel and Shamsie 2003;Miller and Shamsie 1996). In such temporary organizations, work is organized around constant role systems (Bechky 2006). Flexible specialization is related to many of the downsides of creative work described above. It has motivated labour unions to emerge as entrepreneurial actors controlling access to work (Christopherson and Storper 1989;Murphy 1997;Paul and Kleingartner 1994). This exacerbates the problems of exclusion in creative work. An extreme case of flexible specialization is the inclusion of consumers into cultural production in order to innovate (Arakji and Lang 2007;Jeppesen and Molin 2003), or to test and formulate FAQs (Burger-Helmchen and Cohendet 2011).
Digital distribution. The emergence of digital distribution has implications for concentration and diversity by enabling more varied styles to reach consumers, and for flexible specialization by by-passing the corporations as decision-makers. As a result of digitization, a variety of business models have emerged (Bourreau et al. 2012). The opportunities of the internet for digital distribution have been hailed for the power they offer creatives to by-pass corporations and reach audiences directly (e.g. Chaney 2012). However, empirical investigations have concluded that digital distribution offers a very narrow loophole to escape corporate control (Alexander 2002;Benghozi and Benhamou 2010;McLean et al. 2010;Regner and Barria 2009), because having access to institutional gatekeepers is critical for financial performance, even in the new channels (Broekhuizen et al. 2013;Mol et al. 2012). There is, however, evidence that highly specialized styles may benefit from digital distribution and escape corporate control (Hardaker and Graham 2008), but that this effect is slight (Benghozi and Benhamou 2010). Weeds (2012) finds that both superstars and 'long tail' artists benefit from digital distribution, but Karniouchina (2011) shows that even Hollywood overestimates the sales potential of the long tail. Digital cultural goods also still need to find a way to fulfil consumer needs for collecting, ownership, ritual for buying and the enjoyment from the look, feel and smell of the original product (Chen and Granitz 2012).
Digital distribution may enable market access, but the costs of production first need to be covered. There are adherents of the fan-funded model (e.g. Chaney 2010), but Sorensen (2012) shows that crowds funding the initiatives are sensitive to size, name and brand in their decisions, implying that gatekeeping mechanisms also operate in crowdfunding. Another alternative is the ad-funded model, but it has not proved competitive in relation to corporate portals such as iTunes (Papies et al. 2011). Observers also report that majors have been slow and reluctant to adopt digital distribution (Currah 2006;Elberse 2010;Moreau 2013;Shang et al. 2008).
The main challenge in studying the relationship between industry structure and product variety lies in the quantification of variety. Using the number of artists as a proxy for diversity includes the assumption that all artists make equal contributions to diversity, and that no two artists are identical. This is a problematic assumption, because injecting a rock band into a market populated by jazz bands increases diversity more than would adding another jazz band. Using content characteristics to determine levels of diversity has also been attempted (Alexander 1996;Cappetta et al. 2006;van Venrooij 2009). Further applications of content analysis could help to capture innovation and diversity more efficiently than artist-based measures. Proxies for innovation could include new content components and new combinations of such components. More advanced methods for measuring diversity and creativity would enable further exploration of the relationships between industry structure, sales, product quality and product content. Does a higher level of concentration lead to poorer selection performance as a result of a diminished number of decision-makers and, therefore, poorer fit with consumer tastes? Can vertical disintegration offset this effect? What effect does the growth of multi-industry entertainment conglomerates have on product diversity? Can digital distribution undermine the position of the majors or will it remain a narrow outlet for fringe styles? Historical lessons from the entry of desktop publishing systems suggest the former (Cox and Mowatt 2008), while those from the microradio movement suggest the latter (Greve et al. 2006).

Framework for future research
Even though cultural industries research is abundant and answers many research questions, several areas remain for future research (see Figure 1). The concepts presented in the figure are all present in existing cultural industries research. The connections that have not been studied so far are presented with arrows containing question marks, whereas the connections that have already been tested at least to some extent are presented without question marks. This framework centres on the selection processes taking place in cultural production. Many cultural industries scholars assume that it is impossible to assess the success potential of any good in advance, and that it is purely a matter of luck. However, the existence of selectors suggests that they have some value, and some selectors perform better than others. Therefore, the selection performance of upstream selectors is an important avenue for future research.
Upstream selectors adopt various selection criteria that influence their selection performance. Moreover, upstream selectors require resources to enforce particular selection criteria -not every criterion is possible for all selectors. Those resources are probably affected by industry structure. Small specialists and large generalists vary in terms of how specialized their knowledge is. Moreover, vertically integrated and disintegrated selection processes vary as to how many decision-makers affect the final outcome. Production-based and selection-based strategies are an important avenue for future research. Can heavy investments in artist and product development compensate for less information in selection, or do the outcomes from these two strategies form separate product populations?
In addition to resources, selection criteria are affected by the firm's differentiation strategy as a function of distance from the mainstream. In the mainstream, sales may be easier to forecast and, therefore, choosing a strategy further from it signals a preference for risk. In order to implement a particular differentiation strategy and to choose their selection criteria, the firms need to make sense of the field in terms of what types of offerings there are and what types of sales they are achieving. Research on changes in sensemaking methods is still scarce outside the SoundScan case. The selection system consisting of resources, sensemaking, differentiation strategy, selection criteria and industry structure together set the level of differentiation of the outputs. This process has so far been studied either as a relationship between industry structure and product variety or between sensemaking and selection criteria. Therefore, there is ample room for research on various relationships between these variables. Moreover, the transfer of selection power from one actor to another, and the role of selection performance in explaining it, has received scant attention.
In addition to the levels of skill applied by selectors, sales are dependent on consumer tastes. According to existing research, these develop in a cumulative and path-dependent manner. Therefore it is reasonable to assume a level of consistency in such tastes. Research has concluded that sales of cultural goods accumulate through information cascades, whereby consumers inform each other about the products they like. Tastes may likewise evolve through such a process. It has been shown that critic tastes can have an effect on sales to consumers, but there is no research on whether or how consumer tastes influence critic tastes. Critics are under professional pressure to appear knowledgeable and credible in their field, and therefore need to stay in touch with the latest developments in consumer preferences. Do critics change their tastes as a result of consumer interest in emerging styles? Moreover, it is important to make the analytical distinction between sales and tastes.
Even though novelty and innovation are vital for cultural industries, the empirical research offers only limited findings on the audience response to innovation. Are innovative goods more popular? Can consumers be divided into pro-and anti-innovation populations? Do innovative goods need more advertising? Are innovative goods more sensitive to endorsement by critics? Are innovative products more sensitive to information cascades? It would be reasonable to assume that innovative products benefit more from increased information among consumers, because they entail a higher level uncertainty of consumer satisfaction. Moreover, legitimacy has so far been analysed only in the literature on categories. What is the nature of the trade-off between legitimacy and innovation in the context of cultural industries?
Quality is a problematic concept when it comes to cultural goods. However, there are studies where quality, either technical or artistic, is included. The relationship between quality and sales is an interesting area for research, which might reveal characteristics that produce a consistent sales advantage. Sales have an effect on the remuneration that artists receive through various revenue-sharing arrangements. However, other factors affecting remuneration have not been explored. Do past successes, past collaborations or quality influence the revenue-sharing percentages? Or is revenue-sharing simply a means of targeting incentives at those parties needing the strongest motivations? Would contracts more favourable to creatives improve the quality or sales of cultural goods?
Moreover, quality may be affected by the balance of art and profit motives between artists and managers. Earlier research concludes that these two goals are either mutually exclusive or complementary. Future studies could explore various combinations in the levels of these goals in the production process and related outcomes in terms of artistic achievements and profitability in order to identify beneficial combinations. Moreover, the level of control exercised by managers and artists could be added as control variables in determining such outcomes. The research has treated the control exercised by managers as inherently negative, but the analysis described above might reveal that control has some value.
Selection bias has been studied from the viewpoint of box office sales and gender distribution while knowledge on exclusion from creative networks rests on interviews. The next step would be to show how such practices affect product characteristics. If they lead to lower sales or lower customer satisfaction, by what mechanism? Do such practices impair the average quality of products? Or do they reduce the diversity of products leading to consumers whose tastes are not satisfied abandoning cultural consumption altogether? Digital distribution may enable more diverse creatives and products to enter the market. However, the challenge of informing consumers of the qualities of the goods on offer persists even with digital distribution. So far research has treated selection as gatekeeping, i.e. corporations having the power to keep many artists out of the market. However, selection may also be seen as the power to 56 M. Peltoniemi choose artists for the market. Research from that perspective might reveal that the selectors are valuable for the consumer. Further research on digital distribution might also help to explore the value of selectors. Three reasons might explain the extreme unpredictability of demand. Consumer tastes may be unknown, information concerning the products may not reach the consumers, or the complexity of the production processes may render intended outcomes difficult to achieve. So far, studies have not combined these sources of uncertainty to analyse their relative importance. Research on tastes concludes that they are relatively stable, there is ample research on the effects of information cascades, advertising and critical reviews, and the complexities of creative processes have been documented. Research combining these phenomena could offer useful insights for firms operating in cultural industries.

Conclusions
This review of 314 empirical studies has shown that the field of cultural industries has developed specific topics of interest relating to products, consumers, creatives and industrial organization. These phenomena are connected and result from a constant overproduction of product titles and extreme uncertainty in terms of their success potential. On the basis of this, the review presented a reconceptualization of cultural industries as sectors where firms introduce experience goods with considerable creative elements and aim these at the consumer market via mass distribution. These products contain various stories and styles, and cater to the needs of entertainment and social display. Mass distribution refers to storage and delivery where economies of scale are important. Even though many goods, such as cars, computers and mobile phones contain cultural elements, they also comprise profound functionality and technological performance value. Therefore, it is important to make a distinction between truly cultural goods and goods with certain cultural features. For cultural industries research to offer meaningful findings, the definition should not be all-encompassing.
The review shows that, despite extensive research, many important areas remain for future research. The selection system being at the heart of the cultural industries literature, future research would benefit from examining selection criteria and selection performance in greater detail. Moreover, exploring the relationships between tastes, sales and quality might resolve some of the extreme uncertainty taken for granted in much of the existing research. This review is an invitation to study the relationships presented in the framework. Many questions remain open, but the growing popularity of cultural industries research makes me confident that they will be answered in the near future.