groups: an analysis in the automotive industry
Thesis MSc Business Administration
Spec. Strategic Innovation Management
MSc Business Administration
University of Groningen, Faculty of Economics & Business
Duisenberg Building, Nettelbosje 2
9747 AE Groningen, The Netherlands
Supervisor: dr. C. Carroll
Second supervisor: P.J. Steinberg
18 February 2019
Michel Roo
Oude Ebbingestraat 51a
9712 HC Groningen
Michelroo92@gmail.com
s2015145
2
ABSTRACT
This study represents an explorative research into the strategic groups concept from an industrial organization economy and strategy theory perspective with the structure-conduct-performance-link as underlying model. This study is done within the global automotive industry with a sample consisting of 24 manufacturing conglomerations. Research interest into strategic group has decreased last decades and is at a low level at this moment. This is due to a lack of tests for significance clustering. A multi-method approach with a Permutation test and Monte Carlo test is used in this study to assure significant clustering. A cluster analysis divided the industry into strategic groups. Analyses came up with a three and six cluster solution as best solutions for distinct strategic groups, whereby is switched back and forth to expose differences between strategic groups on both units of analysis. Also, an explorative view is taken on collaborative relationship patterns within and between strategic groups by means of strategic alliances. The main findings are that significant clusters can be proved by the multi-method approach used, and that return on assets as performance measure differs significantly between strategic groups on both units of analysis. Another finding is that, from an overall view, firms in the automotive industry prefer to form alliances within strategic groups compared with alliances between strategic groups. Automotive firms prefer to pool their resources to create economies of scale rather than form complementary alliances between strategic groups. An in-depth look shows that there were more cooperating alliances formed between strategic groups where more joint ventures were expected.
3
TABLE OF CONTENTS
ABSTRACT ... 2 1. INTRODUCTION ... 4 2. STRATEGIC GROUPS ... 6 3. STRUCTURE-CONDUCT-PERFORMANCE MODEL ... 7 4. AUTOMOTIVE INDUSTRY ... 95. STRATEGIC GROUP FORMING ... 9
6. STRATEGIC GROUPS AND FINANCIAL PERFORMANCE ... 12
7. COLLABORATION PATTERNS IN STRATEGIC GROUPS ... 15
8. METHODOLOGY ... 19
9. RESULTS ... 23
10. DISCUSSION ... 35
11. LIMITATIONS AND FURTHER RESEARCH ... 37
12. CONCLUSION ... 39
4
1. INTRODUCTION
In many industries there is a lot of competition to get the best strategic position relative to the rest of the firms operating in the same industry. An industry exists of firms who carry out the same activities, uses the same kind of resources, creating the same type of products or provide a similar service towards their customers. Tracking each rival in the industry will be overwhelming. The strategic groups concept can help firms to identify groups of firms. By grouping firms based on dimensions like vertical integration, level of fixed costs, breadth of product line, distribution arrangements, geographically served markets and so on (Porter, 1979). He also argued that these groups of firms follow a similar strategy in terms of these dimensions. Strategies between groups differ. With this reasoning, a strategic group is homogeneous within groups and heterogeneous between groups in terms of strategy (Carroll, 2018). Analyzing an industry on the strategic group level helps managers to shape their interpretations of their business environment in which they operate (Reger and Palmer, 1996). For firms, separating an industry into strategic groups will create a more manageable overview of firm types to keep track of. Instead of figuring out what all firms are doing, a firm only has to find out what types of firms will do. Also, firms can use industry separation into groups as a reference point to decide which firms to cooperate with to expand their businesses. Firms in the same group will have similar resources to pool together by collaboration to create economies of scale advantages, and collaboration with firms of different groups can be done to complement resources to create advantages. A way to collaborate is by forming strategic alliances.
5 The aim of this research from a theoretical perspective is to extend existing literature on strategic groups. When strategic groups are established, potential performances differences between strategic groups can be measured. This study will also extend existing research on strategic groups by analyzing collaboration patterns on an alliance level within and between strategic groups.
From a methodological perspective, this study will help to give an impulse to research in strategic groups. Up to now, cluster analysis to identify strategic groups has been missing significance testing (Carroll, 2018). For years the interest of scholars in strategic group’s research was very flourishing, and the impact of these groups on firm’s performance was a long time one of the dominant areas of empirical research in strategic management literature (Barney and Hoskisson, 1990). The inability of a significance test for cluster analysis is hindering the research into strategic groups which led to a decline in interest (Carroll, 2018). He underlined this by stating: “The field of strategic group research offers enormous potential for exploring the social structure of rivalry within an industry, but it has lost momentum in the recent decades” (Carroll, 2018, p. 71). Barney and Hoskisson (1990) even argued that research in strategic groups should be abandoned when there will not be developed a significance test for cluster analysis. Because this issue is still unresolved, Cattani, Porac and Thomas (2017) argued that results of empirical studies from over decades have been equivocal. But recently a multimethod approach including two significance tests is suggested by Carroll (2018). This study will take an exploratory view into these significance tests to see if discrete strategic groups can be formed supported by these tests. Significant clustering outcomes can help as support to create new interests into the phenomenon of strategic groups, revitalize the interest into strategic groups research and makes this research stream flourishing again.
From a practical perspective, this study provides insights for managers on the strategic group level. The strategic group level is important because managers’ understandings of groups serve as strategic reference points (Fiegenbaum, Hart and Schendel, 1996; Fiegenbaum and Thomas, 1995). Insights in group formations within an industry based on strategic variables can help managers to focus their attention on the most important rivals in the industry (Carroll, 2018), and set a future course for their firm based on this.
6 to that, patterns of collaboration are an interesting topic to analyze on the strategic group concept. Therefore, the following research questions are composed.
1. Do discrete strategic groups exist in a given industry?
2. How are strategic groups related to possible differences in performance outcomes? 3. How are patterns of collaboration related to strategic groups?
The automotive industry is analyzed in this study to answer the research questions. The automotive industry is an appropriate industry for these kinds of research questions because it is a global industry in which firms use collaboration mechanisms like strategic alliances to expand businesses and gain new activities to increase profits. Data relating to the automotive industry is collected by secondary data retrieved from multiple databases.
The remainder of this study will consist of an extensive analysis of the strategic group concept, the SCP model, the automotive industry, performance and alliance literature that will serve as a basis for answering the research questions. After that, the used methodology and results of the study will be treated. This paper will end with a discussion and conclusion of the results.
2. STRATEGIC GROUPS
The term strategic groups was first used by Hunt (1972). He observed that there were three sources of asymmetry between firms within the appliance industry: 1) the extent of vertical integration, 2) degree of product diversification, and 3) differences in product differentiation (McGee and Thomas, 1986). Hunt (1972) argued based on these sources that a strategic group is a group of firms in an industry that, along a set of strategic dimensions, follow the same or similar strategy. In the years following, scholars came up with different extensions on the concept whereby the view of strategic group research widened. Newman (1973, 1978) defined strategic groups by their vertical integration within the industry. He demonstrated that the existence of strategic groups impairs expected collusion among firms (Cool and Schendel, 1987). Reger and Huff (1993, p. 116) suggest that “a strategic group might be best conceptualized as a core group of firms that define the group position and secondary firms that are aligned with core firms in many essential respects, though they also make some unique strategic decisions”.
7 made that firms are homogeneous in the way they act. On the other hand, from a strategic theory perspective, research is done at individual firm-level and business-unit level within a firm. Conclusions retrieved from this view are that assumptions can be made that firms are idiosyncratic in strategically important ways (Porter, 1981). Based on this, strategy scholars use a degree of intra-industry firm heterogeneity as a key component to analyze their strengths (Barney and Hoskisson, 1990). These two perspectives can be combined for a clear explanation of strategic groups whereby firms are homogeneous within, and heterogeneous across strategic groups in an industry.
So, a strategic group can be seen as a collection of similar firms within an industry that together systematically differs from firms outside their group, measured along certain strategic dimensions (Hunt, 1972; Porter 1979). In terms of these strategic dimensions, firms of a strategic group follow similar strategies (Porter, 1979). Next, to the fact these firms follow a similar strategy, they also react in the same way to external events. As stated by Porter (1979), firms of a strategic group respond in the same way to disturbances, recognize their mutual dependence, and are able to anticipate each other’s reactions quite accurately. In the base, firms within a strategic group should react in a similar way to external events, and firms from different groups should react differently to the same external events. This is substantiated by Schroeder (1990) mentioning that the appearance of new production technology will have a different effect on each strategic group but a similar effect within a group. Thus, a strategic group consists of firms which use the same or similar strategy, have the same kind of resources to their disposal and react in the same way to external events. The strategic group classification can be used in research as a device to segment an industry into sets of companies whose competitors, actions and results are relevant to each other (Hatten & Hatten, 1987). This segmentation should be done by analysis along certain strategic dimensions.
3. STRUCTURE-CONDUCT-PERFORMANCE MODEL
8 strategic management, and in the area of strategic groups to do competitive analysis (Panagiotou, 2006).
The concepts in the SCP model are used widely in the area of strategic groups (Panagiotou, 2006). The model examines how the structure of an industry is related to the conduct, and conduct to the performance of firms (Lipczynski and Wilson, 2004). In this model, “structure” reflects the factors that determine the structure and competitiveness in industries. These factors correspond to the strategic dimensions mentioned earlier. “Conduct” is the behavior of the firms operating under investigation. In the SCP model, there is the assumption that there is a stable and causal relationship between the structure of an industry, the behavior of a firm, and the industry performance (Panagiotou, 2006). Strategic groups have an influence on how firms act, interact and react to each other. This conduct of acting leads to performance outcomes of the firms. However, this does not mean that the model is just a chain of two causal relations. All three components have an impact on each other (Panagiotou, 2006).
While it was always assumed that industry’ structure determines the conduct of firms and firms’ conduct determines firms’ performance (Carroll and Thomas, 2019), in later research the “conduct” component disappeared from the model. This led to the structure-performance link (S-P model). Scholars assumed that by transitivity the structure concept also influences the performance concept (Cool and Schendel, 1987). With that, it is assumed that the strategy concept determines performance directly (Caves and Porter, 1978). The S-P model is criticized in the paper of Carroll and Thomas (2019). They mentioned that strategic groups can be seen as groups of oligopolistic competition and that the S-P model is not a suitable model for generalization to cases involving oligopolies.
9
4. AUTOMOTIVE INDUSTRY
The automotive industry has ideal characteristics for the analysis performed in this study. The automotive industry is a mature, and globalized industry (Sturgeon et al., 2008). The automotive industry is also capital-intensive and has significant effects on related industries (Veloso and Soto, 2001; Chu, 1993). Furthermore, the automotive industry is of economic significance to many countries and is one of the largest manufacturing activity in the world (Sánchez and Pérez Pérez, 2005). Next to that, it is a high-precision, technology-intensive and integrated industry (Jan and Hsiao, 2004). The automotive industry has very high barriers for new competitors to enter the market, making it a high concentrate market (Zapata and Nieuwenhuis, 2010). The manufacturing of cars and trucks is cost-intensive, and therefore they have a high purchase price for customers. The automotive industry is also a suitable industry to analyze firm collaboration patterns within and between strategic groups. The automotive industry is described by Nohria and Garcia-Pont (1991 p. 110) as: “it is widely regarded as being exemplar of a global industry with networks of strategic alliances (Womack, 1988; Ohmae, 1989)”. They also stated that firms operating in the automotive industry enter a wide variety of international strategic alliances since the 1980s triggered by two major oil shocks that happened in the 1970s.
5. STRATEGIC GROUP FORMING
Strategic group research can be used as a tool to analyze the structure of industries. It has been found that group structure is stable and predictable over periods of time (Fiegenbaum and Thomas, 1990). Cool and Schendel (1987, 1988) stated that there are debates about how strategic groups should be identified. Following the SCP model, there needs to be decided what the structure is, cq. how are strategic groups identified, to do follow-up analyses on the strategic groups in this study. It is already mentioned that the structure of strategic groups can be measured along certain strategic dimensions. These strategy dimensions can differ from industry to industry. Therefore, it is essential to analyze an industry in-depth to come up with relevant strategy dimensions before proceeding with analyses to separate an industry in different strategic groups. In this section different strategic dimensions that can be used in the automotive industry will be discussed. These dimensions will be operationalized in the methodology section.
10 and Porter (1977) as a structural attribute of a strategic group that makes it difficult and costly for a firm that is not a member of that group to move into it. Mobility barriers as strategic dimensions cannot be imitated by firms outside the group without substantial costs, significant elapsed time, or uncertainty about the outcome of those decisions (McGee and Thomas, 1986, p 150). Firms that want to move between groups must absorb those costs, delays, and/or uncertainties along the way (Carroll and Thomas, 2019, p. 6). When an industry consists of multiple strategic groups, a mobility barrier is not directly in play for the whole industry. It is possible that a mobility barrier is only applicable to one- or some strategic groups. Therefore, mobility barriers can be seen as proper strategic dimensions for separating an industry into different strategic groups. From a strategic group view the barriers not only protect for new entrants entering the industry, but they also protect a strategic group from entry by firms from another strategic group (Porter, 1979). So, a mobility barrier can be seen as an entry barrier that protects an industry from the entry of other firms into their industry, and in the same time mobility barriers are applied to groups within an industry. Caves and Porter (1977) further say that firms have mobility barriers based on interfirm differences that limit the movement between strategic groups.
There are three different sources of mobility barriers: 1) market-related strategies, 2) supply characteristics in the industry and 3) firm characteristics (McGee and Tomas, 1986). Table 1 shows examples of the different sources that McGee and Thomas (1986) highlighted in their study. A thorough explanation of the examples can be found in their paper.
Table 1: Sources of mobility barriers, extracted from McGee and Thomas (1986)
Market-related strategies Industry supply characteristics Characteristics of firms
Product line Economies of scale: Ownership
User technologies - production Organization structure Market segmentation - marketing Control systems Distribution channels - administration Management skills Brand names Manufacturing processes Boundaries of firms Geographic coverage R&D capability - diversification Selling systems Marketing and distibution systems - vertical integration
Firm size
11 Firms of a strategic group will conduct similar strategies that arise from mobility barriers shaping the structure within the industry. The strategies used will differ relative sharply between strategic groups within an industry. Outcome examples of mobility barriers are scale economies and access to distribution channels (Barney and Hoskisson, 1990).
Firms do not always match with a strategic group as well as other firms do. While some firms match more with a strategic group than other firms do (Reger and Huff, 1993), based on the mobility barriers at play in the specific industry, a firm will not switch quickly to another strategic group.
Based on the literature discussed and the expectation that mobility barriers are present in every kind of industry, it is expected in this study that industries are dividable into strategic groups based on mobility barriers that act as strategy variables. Below strategy variables are discussed that are useable as dimensions to separate the automotive industry into strategic groups.
A first analysis resulted in a selection of six variables. Results of a correlation test made three of the variables to be excluded. This was because of correlations that were too high. Strong correlations imply that they measure the same concept and are therefore not suitable for cluster analysis. These variables were selling, general and administrative (SGA) expenses, total stock, and capital expenditures. A second analysis came up with new variables that show good results on the correlation test. The description of the chosen variables in this study will be given below.
Firm size
The automotive industry is an industry with huge amounts of assets that reflects the size of firms. The relative size of a firm within an industry demonstrates the extent to which that firm has economies of scale and scope. Economies of scale and scope are important factors in the automotive industry (Nohria and Garcia-Pont, 1991). Because the industry is technology and capital-intensive, it is difficult for new competitors to enter the industry. This corresponds to mobility barriers. Therefore, firm size is a good measurement for competition comparison between firms in the automotive industry. Firm size will be measured by the total assets of firms.
Brands
12 customers. These have all their preferences that determine which market segment they belong to. The number of brands a corporate group exists of is a way to see the differentiation in market segment presence a corporate group has.
R&D intensity
The automotive industry is an industry that is characterized as an industry that always faces changes by technological developments (Lavie, 2006). These developments are often based on changing preferences of customers or other global influences like the pressure to reduce CO2 emissions and fuel consumptions. The creation of new car models and improving existing products have a lot of research and development costs. Therefore, it is essential for firms operating in the automotive industry that they invest in research and development to go along with the trends within the industry. Because it involves high costs, it acts as high barriers to industry entry.
Inventory levels
Lean management is very important within the automotive industry. A principle within lean management is just-in-time deliveries. Firms aim to reduce inventory levels and related costs by achieving just-in-time deliveries (Larsson, 2002), whereby the inventory levels are kept as low as possible. Inventory of a company exists of raw materials, work-in-progress, and finished goods. For this study, there is chosen to prioritize vertical integration at the supplier side of the supply chain and therefore focus on raw materials inventory levels that represent the link with suppliers.
Account receivables
Account receivables represent payment arrangements that act as selling systems. Firms sell autos on credit rather than requiring immediate cash payments, generating accounts receivable (Mian, Clifford and Smith, 1992). “A strong dominant player within a supply chain could help weak customers by adjusting the payment periods and credit terms (Saranga, 2009)” (Lind et al., 2012). A customer of the automotive industry can buy a product with a payment arrangement whereby the product does not have to be paid in once. This is very interesting from a strategic view for companies in the automotive industry where the products sold are relatively expensive products compared to other industries.
6. STRATEGIC GROUPS AND FINANCIAL PERFORMANCE
13 performance has also become one of the dominant areas of research in strategic management literature (Barney and Hoskisson, 1990). The possibility to link strategic group membership to financial performance has generated enthusiasm among researchers (Perryman and Rivers, 2011). Many scholars have advocated strategic groups as an important mechanism for understanding strategic behavior, competition, and differential firm performance in an industry (Hatten and Hatten, 1987; Porac et al., 1995; Porter, 1979)” (Gómez, Orcos and Palomas, 2017, p. 383). Analyzing strategic groups as the structure of an industry and how the firm of these groups behave within and between groups can help to explain and understand potential performance differences amongst groups.
Primarily, strategic groups have been used to explore differences in profitability among firms within an industry (McGee and Thomas, 1986). But, in existing research, there is still a lot of debate about the performance consequences of strategic group membership (Cool and Schendel, 1987, 1988). Results of prior research on financial performance differences have quite mixed results (Cattani et al., 2017). Sometimes evidence for performance differences was found and sometimes it was not found. Despite the fact that it is often an assumed fact that there are performance differences between strategic groups, this is not always the case.
14 automatically to performance differences. But when there exist differences, mobility barriers merely protect these differences (Cool and Schendel, 1988).
Next to the existence of mobility barriers, to create increased performance within a group the number of members cannot be too big, so that perfect competition dynamics will not unfold (Barney and Hoskisson, 1990). This is supported by Porter (1979, p. 218) who mentioned that “the profitability of a strategic group will be influenced by the degree to which firms within the group compete among each other”. When a group increases in size, it can be more difficult to develop a collaboration between group members that following the same strategy avoiding costly competition of firms from other groups.
Because industries can exist of multiple strategic groups, being present within a specific group has a potential impact on firms’ profitability compared with firms of another strategic group (York and Miree, 2016). Caves and Porter (1977) argue that group membership of a strategic group affects, to some extent, the performance of individual firms in that group. Hatten and Hatten (1987) even argue that different strategic groups within an industry have different performance levels. Firms within a strategic group will act in a similar way towards their customers. Also, they will probably use the same kind of suppliers to acquire the necessary resources. This homogeneity of firm strategies within a strategic group could lead to the same kind of performance. Firms that are a member of different strategic groups would likely have different sets of strategies to cope with their suppliers and customers. The use of different strategies could lead to performance differences.
15 firms in the industry and how this relates to the performance of different strategic groups will be analyzed.
7. COLLABORATION PATTERNS IN STRATEGIC GROUPS
Industrial organization models with the focus on the SCP model showed that the more firms operating in an industry, the greater the levels of competition (Barney, 1986). In terms of industry structures, this corresponds with a perfect competition industry form (Stigler, 1957). In general, perfect competition will lead to lower profits. Within a perfect competition industry firms do not have market power to set the prices for their products. Firms should look for ways to reduce the “damage” of lower profits. From a strategic groups point of view, firms fighting for the same customers and resources like skills and technological input can be separated into groups that use different strategies to deal with the issues of perfect competition. Instead of competing with each other, firms can choose to collaborate at the same time to reduce “the damage” of perfect competition. Competing and cooperate at the same time is called coopetition (Nalebuff and Brandenburger, 1997; Walley, 2007). A way to compete and collaborate at the same time is by entering strategic alliances with other firms. Strategic alliances are defined as “interfirm cooperative arrangements aimed at achieving the strategic objectives of the partners” (Das and Teng, 1998, p. 491).
Cooperation between international firms exists already for a very long time, national as well as internationally. “In earlier years, most international alliances were formed between a multinational and a local partner (Fagre and Wells, 1982; Lecraw, 1984; Beamish, 1985)” (Garcia-Pont, 1992). These collaborations were often “tactical”. Globalization of an industry changes the structure within an industry. It can enhance the value existing capabilities and diminishes the value of other capabilities, and at the same time there will be a need for new capabilities (Nohria-Garcia-Pont, 1991). Nowadays the collaborations are more “strategic” of nature and exists among multinational competitors. In the last decades, technological developments and globalization of competition result in rapid growth of alliances (Das and Teng, 2002).
16 alliances involve less commitment because the participating firms do not have to form a new separate entity (Das, Sen and Sengupta, 1998). Alliances can be formed for multiple purposes in different business areas like production, sales, marketing, distribution channels and research and development.
There are different reasons why firms choose to enter into a strategic alliance with another firm. Alliances have been seen by firms as an attractive tool which they can use to grow and expand their business (Gulati and Singh, 1998). Coopers and Lybrand (1997) rate growth strategies together with entering new markets among the top reasons for forming alliances based on a survey they conducted. As competition increases and becomes more global, many firms will use alliances to enter new markets, obtain new skills, and share risks and resources (Inkpen and Beamish, 1997). Other benefits can be added value, secure contacts, improved productivity and quality, access to raw materials (Meyer, 1998). Wheelen and Hungar (2000) added obtaining specific technology, reduction of financial and political risks, and achieve and ensure competitive advantage as benefits of forming strategic alliances. Possessing intellectual property rights could also be a reason to cooperate with other firms. By forming alliances, firms can accomplish new projects quicker and cheaper than when they will do it on their own. Alliances have the intentions to improve profits for the involved firms, but they should also lead to better product and services for the customer (Walley, 2007).
The balance of coopetition is determined by relative payoffs (Oxley and Sampson, 2004). This is also the case for firms in strategic groups (Dranove et al., 1998). It is predicted by Garcia-Pont and Nohria (2002) that firms in the same strategic group will behave in a similar way when they face industry changes. Cooperation between strategic group members can lead to collective strategies where they work towards a common goal, and by pursuing these collective strategies a strategic group can be able to manipulate industry’ structure to their advantage (Carroll, Porac, and Thomas, 1993). Firms that collaborate with firms of another strategic group can provide each other with complementary resources, skills, and knowledge resulting in better results for all members of that alliance. Nohria and Garcia-Pont (1991) gave as examples that firms can gain market access or transfer existing technologies to each other. Osborn and Hagedoorn (1997, p. 262) mentioned based on earlier literature that “R&D alliances can have beneficial effects on the larger economy and its innovative potential”.
17 alliances are alliances that describes cooperation between a firm and its downstream and upstream partners like suppliers and distributors. Another typology is of equity and non-equity alliances. By an equity alliance, the participating firms acquire an equity stake of each other, and non-equity alliances are alliances with a verbal agreement or a simple contract. Mitchell, Dussauge and Garette (2002) made the distinction between “link” and “scale” alliance. Hereby a scale alliance is formed to contribute similar resources to achieve scale advantages (Mitchell et al., 2002). They mean with link alliances that firms contribute complimentary resources to achieve differential advantages. Nohria and Garcia-Pont (1991) defined alliances within a strategic group as pooling alliances and alliances between firms of different groups as complementary alliances. Based on these typologies, this study will use a separation into three different types of alliances that are relevant for this study: 1) cooperating alliances, 2) joint ventures and 3) R&D alliances. These types will be discussed below.
With cooperating alliances, this study means a bundle that consists of horizontal alliance types that have the purpose to act in one of the business areas like production and marketing as explained above which have a non-equity character by an oral agreement or a simple form of a contract. These alliances are the most flexible because there is no equity exchange involved (Stafford, 1994). A flexible contract is the easiest do terminate when the objectives of the alliance are met. A joint venture is a form of alliances whereby a new company will be established by the firms that form the joint venture. A joint venture has a relatively high level of hierarchical control where both partners share control by equity sharing (Harrigan, 1985; Williamson, 1996). R&D alliances are collaborations whereby technological input, knowledge, and skills are combined to come up with new products or services. R&D alliances can be formed both in formal and informal forms.
18 the partner firms will put together to create economies of scale will have mostly a similar character. Because of the similarities of the resources, involved firms will be less inclined to protect their resources against copying by their partner. Therefore, less control is needed to assure the objectives of all the firms to cooperate whereby this objective will stay their priority. Therefore, it is expected in this study that the strategic alliances formed within a strategic group will, for the most part, consists of cooperating alliances instead of joint ventures.
Where strategic groups are formed based on strategy characteristics, it is expected that firms of different strategic groups differ in their conduct. Link alliances are expected to be formed more between firms of different strategic groups with complementary assets. The assets of firms within strategic groups are expected similar, and different between strategic groups. It can be expected that they use link alliances with different firms to create leverage for their technological and economic performances (Duysters and Hagedoorn, 1995). For example, it is not possible for a firm to have all the necessary resources in-house to perform the most successful. Because the resources are complementary to each other, the resources of one firm will not be in possession of the other partners, and they want to keep that. Joint ventures are a good way to protect firms’ resources for copying by partners. Therefore, it is expected that link alliances formed between firms from different firms will consist more of joint ventures instead of cooperating alliances.
19
8. METHODOLOGY
This study is an exploratory study about the research question that will answer what the effects of strategic groups are on performance outcomes, and how collaboration patterns are related to strategic groups in an industry. An empirical study was conducted in the automotive industry (SIC-3711). The sample consists of 24 automotive firms.
Data collection
In this research, only secondary data are used. Multiple databases were used to collect the necessary data for the analysis. The analyses in this research are divided into three steps. First, a cluster analysis is done to set up strategic groups within the industry. To divide an industry into clusters, strategic variables are needed to use in the statistical test. Therefore, data from the databases Orbis and COMPUSTAT, and annual reports were retrieved and combined. For the second part of the analysis on performances differences, financial performance data is retrieved from the Orbis database. For the third part about the role of alliances, data was retrieved from Thomson Reuters’ database SDC platinum complemented by data from annual reports. Additional, all firms’ annual reports were analyzed to cross-check the data and to fill in missing data from the databases.
20 For the first part of cluster analysis, strategic variables were needed. With these variables, the companies in the dataset could be clustered into subsets of firms that act homogeneously. Taking the SCP model in mind, there needed to be decided what variables will cover the structure of the industry to examine in this study. The automotive industry was extensively examined in this study to come up with strategy variables suitable to distinguish strategic groups. The strategy variables mentioned in the “strategic group forming” section are used as input for the cluster analysis, and relevant information was retrieved from Orbis and COMPUSTAT. The missing values were added after examining the annual reports of 2012 for those firms.
The second part of the analysis is about the performance hypothesis. For this, data from Orbis is retrieved. Some firms had missing data for the financial performance variables. COMPUSTAT and annual report were consulted to add these missing values.
The third part of the analysis consists of alliances data collected from a relatively long timeframe to ensure that there is a proper set of data to perform the necessary analysis. Because the Thomson Reuters’ database SDC platinum most recent year is 2012, the period 2008-2012 is chosen as observation period. There is chosen for a five-year period because this is assumed as the average duration of an alliance (Sampson, 2007).
Variable descriptions
Strategy variables
The strategy variables used in this study come forth of the strategy dimensions as discussed in the “Strategic group forming” section and will be operationalized below.
Total assets: The size of the firms in this study is operationalized as the total assets. Because
the automotive industry is an industry with a huge amount of assets, total assets is a good measurement for competition comparison between firms in this industry.
Number of brands: How firms differentiate to customers of the automotive industry is in this
study operationalized as the number of brands that exist within a corporate group. The number of brands a corporate group consists of shows their relative market share.
R&D intensity: R&D intensity is operationalized in this study as a ratio from the R&D
21
Inventory levels: For this study, there is chosen to look at the raw materials because this
represents the link with vertical integration at the supplier side of the supply chain. It is operationalized as the total value of all raw materials in stock at the firms.
Account receivables: In this study, there is chosen to operationalize account receivables in a
ratio. With this it can be seen how high the percentage of account receivables is of the net income of a firm, that is gathered by outstanding earnings due to favorable payment schemes and leasing contracts. The account receivable ratio is measured by the values of the end of the fiscal year. Seasonal fluctuations are modest in the automotive industry, and therefore the account receivable levels of the end of the fiscal year represent the need of account receivable well (Lind et al., 2012).
Performance variables
The dependent variable in this study is financial performance. There are many different measures that can be used to assess profitability. In this study there is chosen to measure the construct of financial performance by the following three measurement variables: 1) Profit margin, 2) Return on assets (ROA) and 3) Return on equity (ROE).
Profit margin: Profit margin is suggested in existing literature (Faems et al., 2010) as a
valuable measure for financial performance. Profit margin is measured as a percentage of the firms’ net profit/loss divided by firms’ revenue.
Return on Assets (ROA): Return on assets is a ratio that measures how much profit a firm can
generate from their assets at their disposal. Return on assets reflects the asset utilization of the firm (Griffin and Mahon, 1997). Because the automotive industry is a capital-intensive industry, measuring the utilization of assets is a good measurement of performance. It is measured as a percentage of the net income of a firm divided by their average total assets. How higher the percentage, the more efficient a firm uses his assets to generate profits.
Return on Equity: Return on equity is used to compare the performance of firms operating in
22 equity reflects the profitability of a firm by measuring the investors’ return (Griffin and Mahon, 1997). Return on assets is measured as a percentage of the firms’ net income divided by firms’ average total equity.
Alliance descriptions
The operationalization of the strategic alliance types come forth from the descriptions made in the “Collaborations patterns in strategic groups” section and are explained below.
Cooperating alliances: Within this study cooperating alliances are a collection of alliances
with different purposes. The alliances are common in the way they are arranged by an oral agreement or a simple form of contract. The purposes of the alliances are joint production, supply agreements, joint marketing and licensing agreements.
Joint ventures: A joint venture in this study is meant with an alliance between partners
whereby a new entity is established. A joint venture has a contractual agreement that creates a new legal entity in which the “parent firms” hold their ownerships under conditions and provisions that are specified by a legal document (Murray and Siehl, 1989). Most joint ventures do not last indefinitely and have a duration time taken within the contract.
R&D alliances: R&D alliances within this study contains all alliances, cooperating alliances
as well as joint ventures, that exists with the objective to research and develop new products and/or services intended to introduce towards the customers of the automotive industry.
Descriptive analysis
23 the outcome validity of the research in increased. Both tests have their strengths and weaknesses. The two tests seem to complement each other (Carroll and Heyningen, 2017). An extensive explanation of the Permutation test and Monte Carlo test can be found in the paper of Carroll and Heyningen (2017) wherein they explain both tests in detail, discusses the strength and weaknesses of both tests, and how these tests can compensate each other for their weaknesses. “If both techniques yield similar results, that would support for the convergent validity of the significance tests and the results of the cluster analysis could be accepted with more confidence (Brewer and Hunter, 2006)” (Carroll and Heyningen, 2017, p. 40).
After the strategic groups were identified, performance differences between groups were tested with (M)ANOVA tests. The strategic groups originated from the cluster analysis were used as input and tested on the financial performance measurements profit margin, return on assets and return on equity. Lastly, collaboration patterns of strategic alliances within and between strategic groups were analyzed. Frequency tables were created to create an overview of the strategic alliances formed. Frequency tables are created for the total alliances, cooperating alliances, joint ventures, and R&D alliances.
9. RESULTS
The results are separated in three parts in; cluster analysis, performance measures and collaboration. Table 3 presents the variables that are used in this study for the cluster analysis, tests for performance differences and collaboration pattern analysis.
Strategic groups
24 point where the solid line crosses the dotted line are the places of the “elbow” kinks in the solid line.
Table 2: Ward’s cluster analysis, Permutation test, and Monte Carlo test outcomes
Multiple solutions are identified by the tests as good cluster solutions to separate the automotive industry in strategic groups. The result of the cluster analysis is combined with the results of the permutation test and the Monte Carlo test to find an appropriate cluster solution. This gives the 6 clusters solution as the most appropriate solution. Despite the 3 clusters solution has no significance for the Monte Carlo test, this solution will be kept as a unit of analysis together with the 6 clusters solution. This is because it can be interesting to go back and forth between the 3 and 6 clusters solution.
The next step was to perform a canonical discriminant function within SPSS. With this, all the firms are plotted in a scatter plot. With the scatter plot the distances of each case to the strategic group centroid can be observed. The scatter plots relating to this study can be found in Figures 5 to 8.
Number of
clusters Ward’s Criteria
25
Table 3: Descriptive statistics and Pearson’s correlations
Variable Min. Max. Mean St. D. 1 2 3 4 5 6 7 8
1 Total Assets in th 844,467 309,644,000 77,810,723 85,152,908 1
2 Inventory raw materials in th -128,392 8,160,000 1,650,947 1,988,834 0.567 ** 1
3 R&D intensity 0.804 66.296 5.663 12.996 -0.154 -0.135 1
4 Number of brands 1.000 13.000 4.583 3.500 0.516 ** 0.680 ** -0.213 1
5 Account receivables ratio 0.540 41.060 8.826 10.057 -0.099 -0.108 0.615 ** -0.289 1
6 Profit margin -95.843 16.459 1.362 21.911 0.184 0.036 -0.950 ** 0.128 -0.584 ** 1
7 Return on assets -35.561 13.372 2.674 9.084 0.097 0.008 -0.903 ** 0.050 -0.539 ** 0.942 ** 1
8 Return on equity -317.733 35.198 -1.083 69.129 0.201 0.131 -0.970 ** 0.180 -0.519 ** 0.962 ** 0.953 ** 1
Figure 1: Permutation significant test Figure 2: Monte Carlo significant test
26 With a deep-dive into both the solutions, it is found that in both solutions Tesla is separate in a group that is far away of the rest of the groups in the scatter plot. This can be seen in Figures 5 and 7. Another analysis is that looking at the 3 cluster solution in Figure 6 the firms which are part of strategic group 1 are really scattered. The 6 cluster solution groups in Figure 8 are internally more ‘clumped’ together. A remark hereby is that half of the clusters exists only of 1 firm. Those firms, as already explained for Tesla, can be identified as outliers of the strategic groups in the 3 clusters solution. This makes it interesting to choose both cluster solution as the scope for analyses in the remainder of this study where will be switched between both solutions as unit of analysis. Hereby the 3 clusters solution will have a more general view for strategic
Figure 5: Scatterplot of the 3 cluster solution with Tesla Figure 6: Scatterplot of the 3 cluster solution without Tesla
27 group 1, where the 6 clusters solution will take a deeper look into the differences between the four strategic groups that forms together strategic group 1 in the 3 clusters solution. The following section will go more deeply into the different strategic groups and the reasoning how these groups are separated.
To analyze and compare the strategic groups with each other to find support for the separation in strategic groups opted above, boxplots per strategy variable are created. The descriptive statistics related to the boxplots for the 3 clusters solution are summarized in Table 4 and the boxplots for the 6 clusters solution are summarized in Table 5.
Table 4: Descriptive statistics 3 clusters solution, th= thousand
Table 5: Descriptive statistics 6 clusters solution, th = thousand
The 3 cluster solution
First, the strategic groups that are separated with the 3 clusters solution will be discussed. Thereafter will be zoomed in on the 6 clusters solution that will go into more detail about the differences between the four groups that makes together the “Large firms” group in the 3 clusters solution. Strategic group 3 of the 3 cluster solution exists of group 3.1; 3.2; 3.3 and 3.4 in the 6 cluster solution.
Strategic group 1: Tesla
This strategic group exists of one firm; Tesla. Tesla is different on all strategy variables compared with the other groups. As generally know, Tesla is a firm that is founded in 2003 and is a very young firm compared with all other companies within the industry. They entered the automotive industry in 2008 with the well-known Model S, a complete revolutionary product
Strategic
group Total assets in th
Inventory raw
materials in th Number of brands R&D intensity Account receivables
1 844,467 137,494 1.00 66.30 36.72
2 14,373,178 239,964 2.40 2.39 7.59
3 132,529,317 2,852,737 6.54 3.50 7.63
Strategic
Groups Total assets in th
Inventory raw
materials in th Number of brands R&D intensity Account receivables
28 within the industry by which they prove it is possible to create a fully electric car that is not inferior in performance and safety to the current car standard of that moment. They only produce in one factory located in Fermont resulting in very low total assets compared with the rest of the group. Their R&D intensity is sky high compared with the rest of the strategic groups. This is a result of all the developments that are needed to create a completely new car together with all the new things that are needed to facilitate this. With their products, they serve only a small niche within the automotive industry: electrical vehicles whereby they offer revolutionary innovations. Thus, it can be said that Tesla follows a focus strategy.
Strategic group 2: Small firms
This group consists of 10 firms. Apart from Tesla, the average of total assets is way lower than the other groups. Where Tesla is just one brand, the average of different brands is lower than the “large firms” group. They choose to have less different brand under their care. They focus on specific market segments and try to perform best in these. Their account receivable ratio is a little bit lower than the “large firms” group. But taking a look into the 6 clusters solution, it can be seen that they have a higher ratio than the “big established firms”, “medium-sized firms” and “market-differentiation firm” groups. This could be explained by vertical integration they try to integrate towards their customer to gain market values. As said, in this industry the products that are for sale are expensive for most of the potential customers. Using a payment term or leasing opportunities that are more interesting compared with competitors, they can reach to a group of potential customers who otherwise cannot afford that kind of cars.
Strategic group 3: Large firms
When we look at the 3 clusters solution this group is almost ten times bigger than the “small firms” group, and more than 150 times bigger than Tesla looking to the average of total assets. This difference is also displayed in the value of the inventory of raw materials. Their size is also related to the number of average brands owned. The R&D intensity and account receivables are a little bit higher compared with small firms. Because this strategic group exists of 13 firms, the averages shown can be biased through outliers that neutralizes differences on the total ‘spectrum’.
The 6 cluster solution
29
Strategic group 3.1: Big established firms
This group exists of the two major companies within the industry: Volkswagen and Toyota. The average total assets of these two firms is way bigger than the total assets of the other strategic groups. This group has a similar score for their R&D intensity as the “medium-sized firms” group. Considering all the groups on R&D intensity, they have an average score. Except for “market-differentiation firm”, this group has at average the highest number of different brands under their umbrella. The same remark can be made for the account receivable variable where they also have an average score. Taking a deeper look into this shows that both firms’ number of brands are far from the average of the strategic group. With this in mind, it can be said that these strategic variables have no meaningful influence on the strategy of this group that differentiates them from the other strategic groups. Therefore, it can be assumed that their size, expressed in this study by total assets, will influence their strategy where they can achieve economies of scale through their size compared with the other strategic groups.
Strategic group 3.2: Medium-sized firms
This group exists of nine firms. This group does not have the highest score on any of the variables. On each of the variables, they have an average score. At all variables there are strategic groups that outperform them. With this fact, this group can be seen as the base of the “large firms” group. It is not their score on a specific variable that makes them unique compared with the other strategic groups composing together the “large firms” group. Other firms with outlier scores on variables make them unique of this group and therefore are separated from this group.
Strategic group 3.3: Customer-focused firm
30
Strategic group 3.4: Market-differentiation firm
This strategic group also consist of one firm that is an outlier of the “large firms” group within the 3 clusters solution. What is immediately noticeable is that their inventory level of raw materials is extremely high compared with the other groups, and the number of brands they have. The high level of stock can be a result of the merger that just happened between Fiat and Chrysler, where the capital of the two companies are put together. Expected economies of scale and efficiency are not yet accomplished. Compared with the other strategic groups their number of different brands is much larger. This indicates that they want to serve all kinds of market segments by differentiation with multiple brands. Their ratio for account receivables is the lowest compared with the other strategic groups. They also have by far the lowest R&D intensity compared with the other strategic groups. This implies that this firm differentiates towards their customers with their already existing range of products they offer instead of new innovative products and services.
Strategic group financial performance differences
To see if performance differences exist between strategic groups, (M)ANOVA tests are performed. The dependent variables that are used in this analysis are profit margin, return on assets and return on equity. The (M)ANOVA tests are conducted for both the 3 and 6 clusters solutions. The results from the MANOVA tests are shown in Table 6 and 7. The results of the one-way ANOVA tests are shown in table 8 and 9. The MANOVA tests show all significance for the F-scores at a p<0.05 level. The one-way ANOVA tests show that all three performance measures for both cluster solutions are significance at a p<0.05 level. This means that there are significant performance differences between the strategic groups in this study.
Table 6: MANOVA test for 3 cluster solution
Value F h. df e. df Sig.
Pillai's Trace 1.272 11.636 6 40 0.000
Wilks' Lambda 0.023 35.685 6 38 0.000
Hotelling's Trace 30.065 90.196 6 36 0.000
31
Table 7: MANOVA test for 6 cluster solution
Table 8: one-way ANOVA for 3 clusters solution
Table 9: one-way ANOVA for 6 clusters solution
A more in-depth analysis of the performances of the firms in the automotive industry shows that Tesla scores low on all three performance measures. These performances can be seen by
Value F h. df e. df Sig.
Pillai's Trace 1.404 3.168 15 54.000 0.001
Wilks' Lambda 0.013 11.526 15 44.570 0.000
Hotelling's Trace 46.704 45.666 15 44.000 0.000
Roy's Largest Root 46.045 165.761 5 18.000 0.000
Sum of
squares df
Mean
Square F Sig. Profit Margin Between groups 9948,112 2 4974,056 95,463 0,000
Within groups 1094,201 21 52,105 Total 11042,313 23
Return on Assets Between groups 1615,619 2 807,809 60,061 0,000 Within groups 282,447 21 13,450
Total 1898,066 23
Return on Equity Between groups 104943,488 2 52471,744 221,721 0,000 Within groups 4969,782 21 236,656 Total 109913,217 23 Sum of squares df Mean Square F Sig. Profit Margin Between groups 10035,166 5 2007,033 35,870 0,000
Within groups 1007,147 18 55,953 Total 11042,313 23
Return on Assets Between groups 1632,119 5 326,424 22,093 0,000 Within groups 265,947 18 14,775
Total 1898,066 23
Return on Equity Between groups 105900,653 5 21180,131 95,011 0,000 Within groups 4012,617 18 222,923
32 comparing the min and mean values in the descriptive statistics in Table 3. It is interesting to see what the influence is of the performance values of Tesla on the significance levels for the performance variables. Additional analyses are performed to see what the exclusion of Tesla will do on outcomes of the ANOVA tests. Because in both cluster solutions there are groups consisting of one firm, it is not possible to perform post-hoc tests in SPSS to see if the significance differs between the ties between different strategic groups. Therefore is chosen to do ANOVA tests whereby Tesla is excluded as case. The results of these tests are shown in Table 10 and 11. The tests gave different results as the ANOVA tests did with Tesla included in the sample. ROA was the only significant outcome.
Table 10: one-way ANOVA for 3 clusters solution without Tesla as case
Table 11: one-way ANOVA for 6 clusters solution without Tesla as case
Sum of
squares df
Mean
Square F Sig. Profit Margin Between groups 88,409 1 88,409 1,697 0,207
Within groups 1094,201 21 52,105
Total 1182,61 22
Return on Assets Between groups 90,142 1 90,142 6,702 0,017 Within groups 282,447 21 13,45
Total 372,589 22
Return on Equity Between groups 316,822 1 316,822 1,339 0,260 Within groups 4969,782 21 236,656 Total 5286,604 22 Sum of squares df Mean Square F Sig. Profit Margin Between groups 175,463 4 43,866 0,784 0,550
Within groups 1007,147 18 55,953
Total 1182,61 22
Return on Assets Between groups 106,642 4 26,661 1,804 0,172 Within groups 265,947 18 14,775
Total 372,589 22
Return on Equity Between groups 1273,987 4 318,497 1,429 0,265 Within groups 4012,617 18 222,923
33
Strategic group collaboration patterns
This section will discuss the collaboration patterns of firms in the automotive industry which is separated in strategic groups. Frequency tables are created to examine the collaborations between firms within the industry. Table 12 gives the total overview of alliances per type within and between strategic groups in the automotive industry. Table 13 shows the number of alliances within and between groups per strategic group. The gray shaded numbers illustrate alliances within the strategic groups. Table 14, 15 and 16 will specify the number of alliances per type of alliance: cooperating alliances, joint ventures and R&D alliances.
Table 12: Overview within and between alliances
Table 13: Total strategic alliances in strategic groups
The total amount of alliances analyzed in this study is 99. These alliances are divided by 44 non-equity alliances, 34 joint ventures and 21 R&D alliances (Table 14). 68 of the alliances formed are alliances that are formed between firms within the same strategic group. 31 of the total alliances are alliances that are formed between firms of different strategic groups. So,
Within strategic groups Between strategic groups "scale alliances" "link alliances''
Cooperating alliances 29 15 44
Joint ventures 27 7 34
R&D alliances 12 9 21
total 68 31 99
Total Strategic alliance type
Strategic group number of firms in
strategic group 1 2 3 Total
1 1 0
2 10 0 15
3 13 4 27 53
34 generally speaking, there is a substantial difference comparing within and between strategic groups alliances. Now, a closer look at the specified types of alliances will be taken.
Table 14: Cooperating alliances in strategic groups
Table 15: Joint ventures in strategic groups
Table 16: R&D alliances in strategic groups
Table 14 shows that more cooperating alliances are formed within strategic groups than between strategic groups. There are 29 cooperating alliances formed within a strategic group relative to 15 cooperating alliances between members of different strategic groups. Table 15 shows that joint ventures are substantially more formed within strategic groups, 27 joint ventures versus 7 joint ventures between strategic groups. Most of the joint ventures are formed within the “large firms” group. As can be seen in Table 16, in total 21 R&D alliances were established. These are almost equally distributed. There were 12 R&D alliances formed within strategic groups and 9 alliances between firms of different strategic groups. Herewith, it cannot
Strategic group number of firms in
strategic group 1 2 3 Total
1 1 0
2 10 0 9
3 13 2 13 20
total 24 44
Strategic group number of firms in
strategic group 1 2 3 Total
1 1 0
2 10 0 3
3 13 0 7 24
total 24 34
Strategic group number of firms in
strategic group 1 2 3 Total
1 1 0
2 10 0 3
3 13 2 7 9
35 be said that there is a substantial difference between R&D alliances formed within or between strategic group members.
10. DISCUSSION
Within this study, the concept of strategic groups is tested in the automotive industry by assigning industry conglomerations into strategic groups, and tests on differences in financial performance and examining collaboration patterns on the concept of alliances within and between strategic groups. Now the findings of this study will be discussed followed by limitations and future research, and by the conclusion where the research questions stated in the introduction will be answered.
Strategic groups of the automotive industry served as input for the performance differences tests, and the collaboration patterns within and between strategic groups. In line with the assumption made in this study that industries are dividable into a different set of strategic groups, the automotive industry was separated into strategic groups successfully supported with significance outcomes. The strategy variables used in this study turns out to be useful to use as input for cluster analysis in the automotive industry where conduct related to the variables shapes the strategies used in the strategic groups.
As discussed through this paper the use of cluster analysis was criticized by authors like Barney and Hoskisson (1990) and Cattani et al. (2017) due to a lack of significance tests. With this study, the permutation test and Monte Carlo test opted by Carroll (2018) were used from an explorative point of view to see if the Permutation test and Monte Carlo test can prove significance for cluster analysis. The results of both tests in this study show significance for a part of the clusters solutions. This supports more confidence in the findings.
36 contain numerous firms. They say that when perfect competition unfolds in strategic groups, this would approximately result in no differences between strategic groups. There was only a significant difference for return on assets measure in the 3 cluster solution. Without Tesla, this solution exists of two strategic groups, “large firms” and “small firms”. One of the groups has created a competitive advantage with their utilization of assets relative to the other group. A deeper look into the data showed that “small firms” scored better on return on assets than “large firms”. The “small firms” group creates more profits out of their assets than large firms do. Caves and Pugel (1980) also found evidence in their study that small firms were more profitable in some of the industries they studied. The automotive industry is an asset-intensive industry, and by the efficient deployment of assets, a firm can create a competitive advantage. Overall, the prediction made in this study that performance differences exist between groups can only be confirmed for return on assets. This study cannot confirm that performance differences exist between strategic groups for profit margin and return on equity.
37 When taking a more in-depth look into the distribution of alliances formed for the different types of alliances, some interesting differences occur in the types of alliances and their frequencies of occurrence. In total, more alliances were formed within strategic groups. It was expected in this study that of the alliances formed within a strategic group, these would be more formed on a cooperating alliance base. The results provide that this is true, but cooperating alliances and joint ventures come close to equal numbers of formed partnerships. This does not reflect the idea that members of a strategic group pool their similar resources to create economies of scale whereby joint ventures are not deemed necessary. It could be that firms within strategic groups also cooperate with firms that possess resources and skills not that common within their strategic group. A firm that possesses those resources or knowledge would be more careful with allying and will rather opt for a joint venture instead of a cooperating alliance.
It was expected that most of the total between strategic groups alliances formed would be joint ventures. The results show that this is not the case. More cooperating alliances were formed. Therefore, the prediction that more joint ventures would be formed between strategic groups cannot be confirmed. Findings show that there were not that much R&D alliances formed and the total of R&D alliances within and between strategic groups does not differ exceptionally. So, no conclusions can be linked to this. A sample with more R&D alliances is needed to draw conclusions.
11. LIMITATIONS AND FURTHER RESEARCH
38 complete and detailed enough. Because of the short timeframe of this study, it was not possible to gather the necessary data of these firms by contacting them.
Because only the global automotive industry is studied, the findings of this study need to be interpreted keeping its context in mind. The findings are therefore not automatically applicable to other industries. As stated earlier in this study, the selection of strategy variables is essential to analyze the industry under investigation to see what variables are most applicable to use. Because it depends on the industry under investigation which variables are the best variables to use, the findings of this study are not for granted generalizable to other industries.
Secondly, a limitation is the size of the available alliance data. This could be due to the year the data is used of. The data used for this research is from 2008 to 2012. The crisis of 2008 had a big influence on the automotive industry. Big losses were suffered and some companies remained standing with government support. It took years to recover from the crisis. In this period there was no high focus to form alliances. For example, research and development investments will turn down because this is not most important to continue to exist in the short term. This is reflected in the number of R&D alliances formed in this period. Because the automotive industry is a technological industry, it was expected to have more R&D alliances formed. The year 2012 was chosen because this was the latest year available in the SDC Platinum database. For future research, it would be interesting to analyze the collaboration between firms based on alliances with data from a more recent period where it will be expected that companies have entered into more alliances.
Thirdly, as mentioned in the methodology section, when operationalizing strategic groups, it is important to use knowledge and preliminary analyses of the industry to identify variables related to profitability (Harrigan, 1985; McGee and Thomas, 1986). The first analyses in this study came up with six variables distinctive to the automotive industry. Three of them needed to be excluded. These were selling, general and administrative (SGA) expenses, total stock, and capital expenditures. For future research, it would be interesting to deepen more into these variables and see if they are measurable in other ways than they were with the data available for this study.
