• Nenhum resultado encontrado

Data and methods for understanding industry dynamics

99

4 Data and methods

The objective of the empirical enquiry is to examine the applicability of the concepts and propositions of the industry life-cycle theory in the analysis of the games industry. The review of the literature on cultural and creative industries gives an indication as to what can be expected to be found. Tschang (2007, p. 990) suggests that it may be that particular creative industries do not settle down to a dominant design followed by the era of incremental innovations as such industries are driven by the ongoing tension of creative innovative and rational business interests. Whether this is the case within the games industry, or something more complex, remains to be seen.

The empirical enquiry comprises two analytical steps. The first one aims at understanding the dynamics of game hardware and software sectors through the concepts and variables of industry life-cycle theory. The second step aims at understanding the micromechanisms of the game development industry through interview data and qualitative systems dynamics modelling. This way the observations of the industry dynamics can be explained through the micromechanisms that produce them. Comparable two-step approach has been used by Huygens et al. (2001), whose study on the music industry evolution comprised a historical study based on secondary sources and a multiple-case study based on interview data.

100

variables of interest and the occurrence of other industry events that have played a role in industry evolution. For example, Agarwal and Bayus (2002) report using data from the Thomas Register of American Manufacturers, industry associations, Government Statistics Bureaus, trade press, product directories and price guides, historical accounts and personal communications with experts.

Similarly, Anderson and Tushman (1990) used directories, historical accounts, company reports, company installation censuses and trade press. Abernathy et al. (1983) used popular accounts on the industry, government reports, company histories, prior academic research, trade press and publications by industry associations specifically to trace innovations. Also patent data has been used to trace innovative activity (e.g. Agarwal 1998). Relevant performance metrics are often seemingly easily identified. For example, Anderson and Tushman (1990) determined square feet per hour and bottles per minute as the performance criteria for the glass industry, barrels per day for the cement kiln industry and microseconds per CPU cycle for minicomputers. Abernathy et al. (1983) applied a different approach as they determined performance criteria based on which features were advertised in car ads.

The data sources used in the present study were also various. The purpose of constructing a historical narrative on the evolution of the games industry was to understand the industry context and to be able to trace changes in variables of interest and the occurrence of innovations. The data have been collected from books, reports and websites. The books include monographs, such as Kent (2001), DeMaria and Wilson (2004), Forster (2005), Fox (2006) and Newman and Simons (2007), which are popular accounts on the history of the games industry. The reports are by governmental organisations, such as the British Department of Trade and Industry and the Australian House of Representatives, as well as industry organisations, such as the Entertainment Software Association (ESA) and the Entertainment & Leisure Software Publishers Association (ELSPA). Websites covering data on game hardware and software, such as gamefaqs.com, gamerankings.com, mobygames.com, gamedevmap.com, ultimateconsoledatabase.com, fan sites and company websites, have been used to gather data on game devices, games and their influence, genre classifications, developers, publishing dates and to corroborate details. Games industry related news websites, such as gamasutra.com, gamespot.com and developmag.com, have been used to gather information especially on recent events. Also prior academic research has been used to elicit facts.

The available data allow the tracing of many core industry variables. However, there are limitations and some assumptions have had to be made. First of all, innovations are tracked separately for hardware and software. For hardware the innovations are technological and they are followed through device generations. This approach is common in games industry related literature and press.

The division of devices into generations is done following Kent (2001), DeMaria and Wilson (2004), Forster (2005) and Fox (2006). The tracking of innovations and changes in performance metrics is done with the help of a proprietary database of 117 home game devices introduced between 1972 and 2006. The database includes data for each device concerning producer, launch date, generation, whether the device was a commercial success, country of origin, game storage media, significant innovations introduced by the device and performance in three performance metrics introduced below. The database excludes unauthorised clone machines, such as Coleco Gemini, which is an Atari 2600 clone and Chitendo Vii, which is an inferior Nintendo Wii clone.

101

The performance metrics followed include CPU bit capacity, number of colours that the device can display and graphics rendering capability in polygons per second. The identification of these performance metrics was relatively easy as they have been often mentioned in games industry related literature as well as in advertisements for the devices.

Software innovations were classified into technological and stylistic following Cappetta et al. (2006). Technological innovations include the introduction of 3D graphics and motion capture, for example. Technological innovations were divided into major and minor. This kind of classification has previously been used in empirical work by Gort and Klepper (1982), for example.

It is a simpler version of the classification used by Abernathy et al. (1983) where a seven-point scale is used to judge between innovations with very little impact and innovations that are major disruptions. In the present study, innovations introducing new functions are classified based on whether they are technologically trivial or technologically ambitious. For example, high score is a technologically trivial innovation whereas angles and zoom is an ambitious one. The trivial ones are deemed minor and ambitious ones major innovations. Innovations relating to graphics and visual realism are major when they introduce a new process into game development. Minor innovations are refined versions of doing something that has also been done before.

Stylistic innovations are to a considerable extent tied to genres. The aim is to identify an introduction innovation and a refinement innovation for each genre. The concept of refinement innovation has been used previously, for example, by Clark (1985) and Jovanovic and MacDonald (1994). In Clark‘s (1985) model refinement innovations are abundant and reinforce commitments to the existing dominant design. In the Jovanovic and MacDonald (1994) model an invention is followed by a single refinement innovation that commercialises the original idea. Here the analysis is done in the spirit of Jovanovic and MacDonald (1994), as for each genre an introduction invention and a refinement are identified. The introduction innovation corresponds to the game that first introduces the kind of gameplay that the genre defines. The refinement innovation corresponds to the game that is identified as having popularised the genre or created standards for the genre.

The classification of game related innovations is challenging. Such classification has been attempted by Sapsed et al. (2007) who defined game ideas presented by developers either disruptive or sustaining. In their classification game ideas for PC and consoles end up on the sustaining pile and ideas for mobile, online and DVDi games end up on the disruptive pile. Thus their classification is based on hardware choice rather than on software characteristics. Here the aim is to identify stylistic innovations on the basis of content characteristics rather than technological choices. Both technological and stylistic innovations in software are tracked and the classification into minor and major innovations is used for technology and into introduction and refinement innovations for game content.

In addition to classifying innovations as described above, the objective is to assess the interconnections of the innovation frequency in different innovation classes. This cannot prove

102

causality but will yield the order in which innovative activity has taken place in different innovation classes.

In terms of output, the available data allow the consideration of changes in console game sales in both units and dollars. Based on these, changes in price can be calculated. However, such data is available only for the period 1996-2007 and limited to the US market.

The hardware database allows the examination of entries and exits by generation. Thus, rather than the actual entry and exit timings of the firms, the analysis is based on the introduction of the first device by the firm (entry) and the emergence of the first generation in which the firm does not introduce any devices (exit). As international data for the entries and exits in game software are not available, Finnish firms are used as a sample of the international games industry. The data is gathered one firm at a time from the Business Register of Statistics Finland. The list of firms was constructed based on the listing of members at the Neogames7 website, newspaper and magazine articles, Tekes project participant listings and trade expo participant listings8. The goal was to include all firms existing at some point since 1990. The firms were included in the data set if they claimed to be developing and/or publishing games or technology specifically for games. Thus firms that have not yet shipped a finished game were also included. It is probable that each and every firm did not end up in the dataset. However, this may be offset by the fact that some of the included firms had also other than game business and two of the firms each constituted of three legal entities that showed up as separate firms in the data set. Also, for this reason those legal entities that did not have any game business, despite the other entities of the firm engaged in games, were excluded.

Even though Finland represents a small percentage of international game production, exit and entry trends in Finland reflect the ability of the international game market to absorb new entrants and their products as all Finnish firms aim at the international market.

The analysis on entries, exits and firm numbers is complemented with an analysis on industry concentration in game development. This is done with the help of the Herfindahl index and the four-firm concentration index. The Herfindahl index is commonly used to assess market concentration. It is the sum of the squares of the market shares of firms populating the market.

H denotes the Herfindahl index, n the number of firms and s the market share of a firm. It ranges from 1/n to 1. Smaller values indicate lower level of concentration whereas values close to 1 indicate a high level of concentration.

n

i

si

H

1 2

7 Neogames is a Finnish state funded organisation that specialises in promoting game business, research and development. See www.neogames.fi. Its member listing can be seen as an industry directory.

8 This means firms that have taken part in Game Developers‘ Conference or Electronic Entertainment Expo listed by Tekes, Neogames or conference organisers.

103

The reciprocal version of the Herfindahl index is also used because its interpretation is more convenient. It can be interpreted as the number of firms that would populate the market if all firms were the size of the largest ones. It ranges from 1 to ∞. The smaller the index the more concentrated and thus the more mature the industry.

n

i

si

H

1 2

* 1

The four-firm concentration is a simpler index as it is the percentage of sales that the four largest firms in the market control.

The data used to calculate the values of these indexes come from the Develop 100 rankings that the Develop magazine has been publishing since 2004. These rankings list the most successful game developers according to their UK retail revenue. These rankings are only available for 2007 and 2006 covering the sales figures and thus the index values can be calculated only for 2006 and 2007 (see French and Walbank 2007; French et al. 2008). As it is not possible to observe a long-term trend, the values need to be compared to suitable signposts. For the Herfindahl index the comparison is made with the US Department of Justice guidelines and the maturation process of the early car manufacturing industry. The index values are calculated in four varieties. As many of the studios in the ranking are owned by other companies the values are first calculated by treating development studios as independent firms and secondly by assigning the sales of each studio to their owner. As the 100 firms in the ranking do not cover the entire market the index values are calculated firstly by treating the sales of the 100 studios as the entire market and then by filling the market with small firms according to the actual market size. The size of the market is retrieved for 2006 from the Games Investor Consulting Ltd. (2007, p. 45) report. For 2007 the UK market size is approximated based on the US sales for 2007 and the relationship that US and UK sales had in the 2000s. The US sales figures come from the ESA (2008). According to the ELSPA/TIGA (2005, p. 15) UK is the third largest game market after the US and Japan. In Europe the UK is the largest market with 32.8% of European sales followed by Germany with 17.9%, France with 15.3%

and Scandinavia with 7.7 % (ibid.). For its size and because of its mainstream flavour the UK market can serve adequately as a proxy for the global game market.