Regulations and strategic groups

(1)

2018

Regulations and strategic groups

An US pharmaceutical industry outlook

Master’s Thesis MSc. Business Administration – July 2018

Strategic Innovation Management

Word count main text: 11.303

First supervisor: Dr. C. Carroll Co-assessor: Dr. W.W.M.E. Schoenmakers

Manasseh Struijck S3259927 Populierenlaan 1-181

(2)

1

Abstract

(3)

2

1. Introduction

Ever since Hunt (1972) first created the term strategic group, the phenomena has captured the attention of strategic management scholars and other scholars alike. Especially the possibility of linking financial performance to strategic group membership has generated enthusiasm among researchers (Perryman and Rivers, 2011). Consequently, this has led to a plethora of research regarding the existence and dynamics of strategic groups (Barney and Hoskisson, 1990).

Caves and Porter (1977) stated that strategic groups are groups of firms in a given industry that pursue the same or similar strategies, a definition commonly used to define strategic groups. After Hunt gave birth to the concept, Caves and Porter (1977) extended the concept by introducing the term mobility barriers. These barriers were described as forces that make it difficult or sometimes even impossible for firms to change their competitive position. High capital expenditure for research and development can for example be regarded as a mobility barrier. This extension of the field furthermore introduced the notion that strategic groups are a rather stable phenomenon in industries. McGee and Thomas (1986), in their key review of the field, extended this definition by stating the following:

‘A firm within a group makes strategic decisions that cannot readily be imitated by firms outside the group without substantial costs, significant elapsed time, or uncertainty about the outcome

of those decisions’ (p. 150)’.

(5)

4

example performance between groups. Less research has been carried out on the causes of group formation.

Simultaneously to the development of the strategic group concept, considerable debate and attention has been given on the premise if strategic groups exist or that scholars have merely created groups based on statistical convenience (Carroll, 2018b). This debate was mainly caused due to the lack of appropriate methods available to establish whether an industry is composed of distinct strategic groups or not.

Within these methods, researchers can resort to multivariate statistics as a way to identify how strategic groups are formed, with the most frequently used technique being the cluster analysis. In such a cluster analysis Ward’s method and squared Euclidean distance are commonly used (Carroll, 2018b). Ward’s method can be used to identify areas where many firms are positioned. However, as argued by Carroll (2018b) while homogeneity within groups can be established with this method, heterogeneity across groups cannot. One could not rule out the possibility that there are areas in which firms are less clumped together. Therefore, it is impossible to determine whether actual strategic groups exists, or that they are merely classified by the researcher. The missing piece within this method is a significant test(s). The debate on the validity of methods, along with other factors, ultimately led to a decline in strategic group research (Cattani, Porac and Thomas, 2017).

(6)

5

By adopting an evolutionary perspective Cool (1985);Bogner, Thomas and McGee (1996) the assumption is created that strategic group structures, if existent, are dynamic. Implying that composition of strategic groups as well as number of distinct groups differ across points in time. Herein incorporating the notion of mobility barriers. These acts as barriers, inferring a level of stability in the strategic groups. These mobility barriers are susceptible to, if not created by, regulatory policies. The concept of mobility barriers will be elaborated in more detail in the next sections of this research. For now, it is important to consider that this research views strategic groups as a dynamic phenomenon shaped by environmental factors.

(7)

6

research will contribute to further understanding of how external factors can cause the formation of strategic groups

To further Tywoniak’s et al. (2007) work and to advance the field of strategic management, this thesis builds upon the newly developed and earlier discussed advancements in hierarchical cluster analysis methodology for strategic group research. As it is now possible to establish whether distinct clustering has been formed in a given industry, new research avenues have been opened. This thesis will use these new techniques to empirically assess the following research question:

How is strategic group composition in a given industry related to regulatory policies in the

form of government regulations?

Herein, an analysis of the most important regulatory policies is conducted, which will act as the basis for assumptions regarding the influence of regulations on strategic groups. These assumptions are empirically tested with a hierarchical cluster analysis using the newly developed tests.

(8)

7

2. Theoretical background and prior academic contributions

The causes of formation of strategic groups, and group formation in general, is of critical importance for this study. New economic institutionalism is furthermore discussed in the context of strategic group analysis. Lastly, this section covers prior strategic group analysis studies conducted in the pharmaceutical industry.

2.1 Causes of strategic group formation

Research surrounding the topic of strategic groups has been abundant over the years, despite

the methodological issues. However, the formation of strategic groups has been given less attention than the existence of strategic groups. Caves and Porter (1977) argue that firms erect different mobility barriers, based on interfirm differences and thus form the basis for strategic group forming. They argue that differences in risk posture and differences in skills and assets drive the formation of strategic groups. Lee (2003) showed that strategic choices differ between firms and can be at the basis of strategic group forming. In his study, the (strategic) choice to participate in a government funded initiative led to a two-cluster solution, one group with participants of this program and one with non-participants. Several researchers support the notion that firms differentiate themselves based on strategic choices and are at the basis of strategic group forming (Oster, 1982; Harrigan, 1985).

(9)

8

determined by relocation cost. Therefore, Tang and Thomas (1992) argue that the number of groups and the mobility barriers shaping the groups are subjected to the basic conditions of an industry. Following this assumption, numerous studies have identified dimensions on which firms cluster. Hunt (1972) first did this on vertical integration, Cool and Schendel (1987) on geographical and product scope and (Porter, 1979) on firm size. Following spatial competition the “relocation cost” can be seen as a product of the specificity and reversibility of the investment in one of those dimensions (Tang and Thomas, 1992).

Carroll (2018a) proposes a different view on the notion of mobility barriers. He argues that mobility barriers do not necessarily have to be present in order for strategic groups to exist. And that mobility barriers are not defined by for example significant capital cost, but can be seen as the “undulating terrain” of the industry. Herein strategic groups are re-envisioned as interdependent rival firms acting collectively in an effort to create an advantage and explore the terrain.

During the 1980s, another stream of research studied the formation of strategic groups through the concept of cognitive taxonomy. Herein, the decision making process and the principles of categorization can be used to explain the grouping of firms (Rosch and Mervis, 1975; Rosch, 1978). Porac et al. (1987) applied cognitive taxonomy in a complex industry, proposing that in such an industry firms tend to create an industry structure in order to frame the competitive landscape and to construct differentiation strategies.

(10)

9

Several explanations thus exists that could drive group formation. However, next to these potential causes highlighted above numerous other factors could drive the formation of strategic group. For example economic factors such as economies of scale. It is therefore important I provide a demarcation in the next paragraph, with the acknowledgement that my demarcation is deliberate and limits the study.

In this thesis, the assumption is made that next to strategic choices, firms are forced to be similar or different, to some degree caused by the institutional environment (regulations). Herein, the institutional environment furthermore impacts formation of strategic groups. This is similar to the previously introduced paper of Tywoniak et al. (2007) which found that regulations significantly altered strategic choices firms make and the competitive landscape. Additionally, the view of Carroll (2018a) is used in defining mobility barriers, where regulations can impact the interdependence and collective action that firms might undertake. For example the introduction of a regulation causing stricter patent laws, might push one firm to increase their patent applications. Other firms following similar strategies, and thus belonging to the same group, might be inclined to follow similar patent application strategies, in order to also gain the competitive advantage. Herein, the formation, memberships and number of firms of strategic groups are affected by regulations in the sense that regulations shape the competitive landscape or undulated terrain as Carroll (2018a) puts it. Regulations are regarded as part of the “terrain”.

2.2 New institutional economics (NIE)

(11)

10

According to Williamson (2000), the theory of NIE can be viewed as an analytical framework. This is best illustrated by the four levels of social analysis of economic institutions that exemplifies the complexity of the relationships between economic actors (Williamson, 2000). But before this framework is discussed more in detail, it is perhaps wise to first explain the cornerstones of the NIE theory. Williamson (1998) argues that what makes NIE “new” is that is does not only consider all institutions to be important, as hypothesized by early institutionalists, but that these institutions are susceptible to analysis. As discussed before, Williamson (1998, 2000) argued that the complex relationships between economic actors could be explained by a four level of social analysis of economic institutions. As depicted in Table 1, level one contains the embeddedness, and is comprised of customs, traditions and religions. Level two is seen as the institutional environment containing the “formal rules of the game” such as property rights. Furthermore, level three contains governance, which can be seen as “the play of the game”. Lastly, level four is the resource allocation and employment.

2.3 Combining NIE and strategic group analysis

NIE theory differs significantly from strategic group analysis. While NIE focuses on

(12)

11

following model based on the earlier discussed four levels analysis framework of Williamson (2000), an overview of these analogies can be found in Table 1.

Table 1: Adapted from Tywoniak et al. (2007) depicting analogies between level of analysis in NIE and strategic group research

New institutional economics (NIE)

Strategic groups research

Level 1: Embeddedness “spirit of the game”

Norms, customs, behavioral rules

Institutional isomorphism, communities of practice, based around accepted norm, customs etc.

Level 2: Formal rules “rules of the game”

Laws, property rights Industry regulations,

government policies, industry standards, codes of conduct Level 3: Plays of the game Governance structures Strategic

direction/configuration Level 4: Resource allocation

“play execution”

Adjustments in price, output and employment in response to market conditions

Firm-level effects: the results of deploying resources in particular ways

(13)

12

While significant differences thus exist within the two fields of literature, using analogical reasoning to convey several concepts from NIE to strategic management analysis can be of significant addition to a strategic groups analysis. Especially, the formation and stability of strategic groups can be further understood with the incorporation of NIE concepts into a strategic group analysis. For this study, the framework constructed by Tywoniak will be used. First, a level two analysis will be conducted, herein the most important regulations that affected the industry will be analyzed. The level two analysis will form the basis for the subsequent conducted level three analysis. Herein, firms will be clustered based on their strategic choices. By combining the two analyses, one can infer whether regulations have in fact shaped strategic group membership and structure.

2.4 Strategic group analysis on the pharmaceutical industry

(14)

13

3. The institutional analysis; patent and price regulations

The motivation behind analyzing the regulatory factors in a given industry are two-fold. First, the analysis can act as a basis on which in-depth knowledge regarding the studied industry can be obtained, this knowledge can then be used as a foundation for a more thorough empirical analysis (Tywoniak et al., 2007). Secondly, the impact of regulations has been widely unrecognized in strategic group research (Porter, 1980). Yet studies have shown that public policies and regulations could influence strategic group composition. (Lee, 2003). Lastly, by incorporating both an institutional analysis and an empirical analysis, this research tries to answer the call for multi-dimensional segmentations in strategic management studies. (Porter, 1980; Tywoniak et al., 2007). A multi-dimensional segmentation provides a more thorough analysis and might lead to less equivocal results.

The reason why regulations could influence strategic groups can be found within the theory of new institutional economics (NIE). One of the pillars of NIE is that it assumes that not only firms influence the competitive environment but also other institutions such as governments, and that they are susceptible for analysis, as discussed earlier. (Williamson, 1998). Derived from this argumentation a justification can be made for the use of an institutional analysis. Herein, the institutional environment pertaining for example to regulations and policies of an industry is analyzed in order to assess the impact of those institutions. However, for strategic group research one must make the assumption that these policies and regulations affect firms differently and can thus be causes for strategic group formation (Parker, 1999).

(15)

14

historical review will be on two main regulatory concepts that shaped the pharmaceutical industry: patents and pricing regulations (Caves, Whinston, Hurwitz, Pakes and Temin, 1991; Malerba and Orsenigo, 2015). Additionally, the pharmaceutical industry is one of the few industries in which patent exclusivity is extremely effective in protecting products (Lehman, 2003). Consequently, regulations affecting patens can therefore significantly impact the pharmaceutical industry, and thus its composition.

The history and development of these two main concepts enables the construction of assumption regarding the existence and composition of the current US pharmaceutical industry’s strategic groups. Information regarding legislations will be extracted from prior academic historical reviews. The data will then be triangulated with the use of corresponding news articles found in the Lexis Nexus database. Lastly, the review ranges from 1940s till 2010, as 1940s are cited as being the birth years of the modern day US pharmaceutical industry (Lee, 2003)

However a cautionary note is placed. Causes of strategic group formation essentially requires historical data starting from the inception of an industry. Unfortunately such data does not available . A viable alternative is to discuss how regulations throughout history have altered the present day composition of strategic groups. Therefore, this historical review discusses the main regulations impacting the industry and what groups might have originated from these regulations. Another note should be placed here, as regulations are not the sole factor driving formation. It should be stated that other factors, such as economies of scale or market developments could have impacted the formation of strategic groups to an even higher degree.

3.1 Era 1: Golden age: 1940s to the mid-1970s

(16)

15

programs had were provided with a large number of resources. These programs were initiated by the government to aid the war efforts of the Second World War. Lee (2003) conducted a study on the formation of strategic groups in the US pharmaceutical industry between 1920-1960 and found that a large numbers of firms belonging to the same group (innovators) participated in these programs.

Another important regulatory mechanism that aided the development of the US pharmaceutical industry was the strong US patent law. The pharmaceutical industry has traditionally been an industry where patents prove to be effective against imitators (Schilling, 2010). This became even more so when the US patent office in 1946 granted patents to naturally occurring substances (Malerba and Orsenigo, 2015). This led to an increased tightness of the appropriability regime and that in turn led to less effective price competition.

In 1962 amendments to the 1938 Food, Drug and Cosmetics Act severely impacted the competitive environment in which pharmaceutical firms had to compete (Peltzman, 1973). This amendment required firms to prove that new drug were both safe and effective. Cool and Schendel (1987) argue that these amendments were regarded as a major exogenous shock to the industry, as firms now had to be more selective in their R&D commitments. This could led to smaller firms in the industry adopting a more specialized strategy, as R&D spending would otherwise be too high. This could be observed in present day strategic groups, where it is assumed that a strategic group exists that adopts a specialized strategy.

(17)

16

3.2 Era 2: the biotechnology revolution: 1970s-2000

In the beginning of the 1970s, the increased in public funding for healthcare associated research began to positively affect the industry (Malerba and Orsenigo, 2015). Numerous advances led to a transformation in the pharmaceutical industry. In the US, this was partly due to the radical changes in the IPR regime. In the US various government actions created more favorable patent laws for the pharmaceutical industry.

Another important area of government actions that shaped the industry was cost containment considerations. As development cost began to rise in this “era”, an increase in prices of drugs became noticeable too. To contain the costs the US government shifted their price controls to systems of reference pricing. This was followed by a critical legislation, the so-called Waxman-Hatch act in 1984 (Malerba and Orsenigo, 2015; Tancer and Mosseri‐Marlio, 2002). This act reduced the previously in place safety controls regarding generic bio-equivalents of branded products. It allowed pharmacists to retail these ”generics”. Which had significant implications for the competitive environment of the US pharmaceutical industry. Mainly due to the fact that firms arose that now solely focused on the development and sale of these generic drugs. It is therefore plausible this act played a role, at least partly, in the birth of a strategic group that follow similar strategies focused on generic drugs (e.g. lower R&D spending). It is furthermore assumed that these regulations were impactful enough to be noticeable in the industry nowadays. Therefore it is assumed that a current day strategic group exists that focused on these generic drugs.

3.3 Era 3: Winter of discontent: 2000-2010

(18)

17

industry. This is turn sparks the debates of pros and cons regarding the current strong patent law.

Evidently, government regulations pertaining patents and prices have significantly altered the competitive environment of the modern US pharmaceutical industry. It would be improbable to assume that these regulations impacted all strategic groups equally. For example, the 1962 amendments required firms to prove that their drugs were both safe and effective. As a result, researchers argue that firms had to alter their product strategies and R&D commitments (Cool and Schendel, 1987). However larger firms in the industry would have been less affected as their budgets for R&D spending would have been significantly larger than for example their smaller counterparts. This does not mean that the amendments did not impact these firms, but the degree should logically vary.

In a similar vein, the patents laws regarding both generic and prescriptive drugs significantly impacted the industry. The strong US patent law gave rise to power positions of larger established firms, whereas any relaxation of patent laws could favorably affect smaller firms, more focused on a specific generic drug. Furthermore, a firm, regardless of size, pursuing a strategy focused on patens can be negatively affected by a relaxation in patent regulation. Whereas, a firm pursuing a strategy focused on generic could be positive, or less negatively affected by a relaxation of patent law.

(19)

18

Throughout history, US pharmaceutical firms have been repeatedly subjugated to changes in regulation regarding patents and pricing. The presence of patent law significantly altered competitive positions of firms throughout history. Combined with regulations pertaining to the pricings of drugs. These regulations affect the composition of current day strategic groups in several ways. The evolution of patent law led to strong positions of large incumbents. Creating the assumption that a distinct group exist with large incumbents that follow similar patent strategies and R&D orientation. Furthermore the regulations regarding the patent expiration, has contributed to the creation of firms following a different strategic direction, those that likely focus on generic drugs

4. Methodology

An empirical analysis was conducted to statistically test the assumptions that arose from the institutional analysis. A hierarchical cluster analysis was conducted using Ward’s method and squared Euclidean distance. This type of analysis was chosen as it can statistically measure the strategic groups composition in a given industry. Moreover, due to his appropriate nature, Ward’s method has become the de facto choice for a strategic group analysis (Carroll, 2018a). The analysis will be supplemented with two significance tests to increase the validity of the tests. Additionally, with these significance tests it can be empirically established whether firms within groups are in fact homogenous and if clusters (strategic groups) are in fact isolated islands (Carroll, 2018). Furthermore, the sample, strategic variables and validity of the study will be discussed.

4.1 Selected industry of research

(20)

19

data regarding this industry was abundant. Thirdly, although prior literature discussed the existence of distinct strategic groups in the US pharmaceutical industry, regulatory influences on the formation and existence of these groups remains widely unrecognized. Furthermore, Lee (2003) argued that strategic divergence (i.e. existence of strategic groups) would likely to occur in more R&D intensive industry, a characteristic frequently attributed to the pharmaceutical industry. Lastly, several extensive strategic group studies have been conducted in this industry (Cool and Schendel, 1987; Fiegenbaum et al., 1987; Lee, 2003) providing a sound theoretical foundation.

4.2 Data collection

(21)

20

(FDA) database was used for the data regarding the drug applications and type of drug applications. The FDA is the US governmental agency responsible for the quality, inspection and verification of all food and drug sold in the United States. The agency has a database where it records the numerous drug applications made by firms. The database is updated regularly and text files containing the records can be downloaded publicly. For this research, the text files were downloaded on 13 March 2018. Subsequently the text files containing data on the drug applications and type of submission were merged to create a database with the necessary information to construct the corresponding strategic variables. The United States Patent and Trademark Office (USPTO) was used to gather the patent count data of each firm. The patent data was gathered on the 13th of March 2018, similar to the FDA data mentioned earlier. Finally, the Lexis Nexis database was used to triangulate information regarding the regulations described in the institutional analysis. Unfortunately, the database only provides articles from 1980 until today. Therefore, information regarding regulations that predate 1980 were not triangulated. In Table 2 an overview of these databases and their corresponding variable is given

Table 2: Overview of the strategy variables and their data sources

Variable Literature Data source

R&D spending Cool and Schendel (1987) _{Lee (2003)} Orbis/Compustat

Expenditure by sales Fiegenbaum, Sudharshan _{and Thomas (1990)} Orbis/Compustat

Number of US patents _{Martens (1988)}Katila (2000) USPTO

Patents by employees

Katila (2000) Martens (1988)

USPTO/Orbis

Approved applications FDA Cool and Schendel (1987) FDA database

(22)

21

Table 3: List of US pharmaceutical firms, primary business line, main products services and cluster membership

Cluster Name of firm Primary business line* Main products/services*

1 Johnson &

Johnson

A multi-national manufacturer of pharmaceutical, diagnostic, therapeutic, surgical, biotechnology and personal hygiene products, as well as a provider of related services, for the consumer, pharmaceutical, and medical devices and diagnostics markets

Pharmaceutical, diagnostic, therapeutic, surgical, biotechnology and personal hygiene products

1 Pfizer Engaged in the production and distribution of

pharmaceutical products

Pharmaceutical Segment offers products for diseases in cardiovascular and metabolism, such as Lipitor for blood cholesterol; Norvasc for hypertension; Caduet for cardiovascular effects; Chantix/Champix for cure of smoking; and Exubera, the insulin inhaled by human for glycemic check; Animal Health Segment offers chemicals such as anti-inflammatory, parasiticides, antibiotics, vaccines and other related medicines

1 Merck & Co.,

Inc.

Engaged in the manufacture and sale of medicines, vaccines, biologic therapies, and consumer and animal products

Medicines, vaccines, biologic therapies, and consumer and animal products

1 Abbvie Inc. Research-based pharmaceuticals business that is

engaged in the discovery, development,

manufacture, and sale of a line of pharmaceuticals and biologics worldwide

Pharmaceuticals and biologics

2 Abbot

Laboratories

Global, diversified health care firm engaged in the discovery of new medicines, technologies and ways to manage health

Pharmaceutical, nutritional, diagnostic and medical products

1 Eli Lilly and

Company

Global pharmaceutical company engaged in the discovery, development, manufacture, and sale of pharmaceutical products

Pharmaceutical products

1 Bristol-Myers

Squibb Company

Engaged in the manufacture of prescription pharmaceuticals

Prescription pharmaceuticals

1 Celgene Corp. Operates as an American integrated

biopharmaceutical firm that specializes in the discovery, development and commercialization of innovative therapies for the treatment of cancer and immunological diseases through regulation of genomic and proteomic targets

Specializes in the discovery, development and commercialization of innovative therapies for the treatment of cancer and immunological diseases through regulation of genomic and proteomic targets

2 Regeneron

Pharmaceuticals Inc.

Biopharmaceutical company that discovers, develops, manufactures, and commercializes medicines for the treatment of serious medical conditions in the United States

Medicines for the treatment of serious medical conditions

2 Zoetis Inc. Engaged in the discovery, development,

production, and commercialization of animal health medicines and vaccines

Animal health medicines and vaccines

2 Alexion

Biopharmaceutical firm, develops and commercializes life-transforming therapeutic products

(23)

22

3 Vertex

Engaged in the production and distribution of pharmaceutical and ancillary products

Pharmaceutical and ancillary products

2 Catalent Inc. Engaged in the provision of advanced delivery

technologies and development solutions for drugs, biologics, and consumer health products

Advanced delivery technologies and development solutions for drugs, biologics, and consumer health products

2 United

Therapeutics Corp.

Operates as a biotechnology company focused on the development and commercialization of unique products for patients with chronic and life-threatening cardiovascular, cancer and infectious diseases

Cardiovascular health, including developing analogs of the endogenous hormone prostacyclin for the treatment of pulmonary arterial hypertension and critical limb ischemia and a variety of telemedicine services for patients with an array of possible cardiac arrhythmias; Oncology and has engaged in the pivotal trial of a potential medicine for preventing the recurrence of ovarian cancer; Infectious disease targeting hepatitis C and other diseases with unique glycobiology

compounds; Remodulin; ProstaRex

2 Biomarin

Develops, manufactures and commercializes innovative biopharmaceuticals for serious diseases and medical conditions

Innovative biopharmaceuticals for serious diseases and medical conditions

2 Akorn Inc. Engaged in the development, manufacture and

marketing of generic and branded prescription pharmaceuticals as well as animal and consumer health products

Sterile and non-sterile dosage forms including: Ophthalmics, injectables, oral liquids, otics, topicals, inhalants, and nasal sprays

2 Phibro Animal

Health Corp.

A diversified animal health and mineral nutrition company

Animal Health, Mineral Nutrition and Performance Products

2 Lannet

Company Inc.

Manufactures and distributes pharmaceutical products sold under generic names

2 Emergent

Biosolutions Inc.

A biopharmaceutical firm that develops,

manufactures, and commercializes immunobiotics

Immunobiotics (vaccines and immune globulins that assist the body's immune system)

2 Cambrex Corp. Engaged in the manufacture and wholesale

distribution of pharmaceutical products

2 Ionis

Engaged in the development of ribonucleic acid (RNA)-based therapeutics

Ribonucleic acid (RNA)-based therapeutics

2 Ironwood

Discovers, develops, and commercializes medicines targeting human therapeutic needs principally in the United States

Linaclotide; Phase 1 pain drug candidate and multiple preclinical programs

2 Tesaro Inc. An oncology-focused biopharmaceutical firm that

identifies, acquires, develops, and commercializes oncology therapeutics and supportive care product candidates in the United States, Europe, and other international markets

Rolapitant; Niraparib; TSR-011

2 Intercept

Developmental stage biopharmaceutical firm that is engaged in the development and

commercialization of therapeutics to treat chronic liver disease

Therapeutics to treat chronic liver disease

3 Alnylam

A biopharmaceutical company developing novel therapeutics based on a breakthrough in biology known as RNA interference

RNAi

(24)

23

4.3 Hierarchical cluster analysis

In order to use a cluster analysis in a strategic group analysis one needs to identify the important indicators of strategy in a given industry (Cool and Schendel; 1987). Ketchen and Shook (1996) proposed a process pertaining to three critical issues that researchers have to consider when selecting appropriate variables for the cluster analysis.

4.3.1 Selection of variables

The first step is to consider how the variables will be selected. Herein, Ketchen, Thomas and Snow (1993) distinguish between inductive, deductive and cognitive approaches. The inductive approach is closely linked to exploratory research, and pertains to the inclusion of as much variables as one can find, whereas the deductive approach suggests inclusion of variables with a strong (prior) theoretical foundation (Ketchen and Shook, 1996). Lastly, the cognitive approach, albeit similar to the inductive approach, suggests using industry experts when selecting appropriate variables. Ketchen and Shook (1996) argue that the approach should fit the research. Further arguing that studies that aim to measure the extent of key constructs should opt for a deductive approach. As this study aims to link strategic directions to group formation, the deductive approach was adopted. Moreover, given the design of this study, inclusion of variables with a strong theoretical foundation would yield the most reliable results. Opposed to an inductive approach or a cognitive approach with a large qualitative methodology (i.e. interviews with industry experts). Lastly, the (US) pharmaceutical industry has long been of interested for strategic management researchers, increasing the likelihood of finding variables with a strong theoretical foundation.

4.3.2 Standardization of variables

(25)

24

phenomenon of interest (Hair, Anderson, Tatham and Black, 1992). Several researchers argue that standardization could solve this problem (Hair et al., 1992; Harrigan, 1985). However, other researchers argue that standardization, in some cases, could have little to no effect (Edelbrock, 1979; Milligan and Sokol, 1980). The variables in this study will be subject to standardization, as the composition of data demands the use of a form of standardization.

4.3.3 Multicollinearity of variables

The last issue affecting any selected variables is the mulicollinearity. Substantial correlation between clustering variables can lead to “overweighting” the underlying constructs (Ketchen and Shook, 1996). This can in turn lead to unreliable assumptions derived from the corresponding cluster analysis. In the current study design, it might negatively impact the significance tests used. I have subjugated the selected variables to a Pearson correlation coefficient test (Table 4), in order to account for occurring multicollinearity.

4.3.4 Strategic variables

After a review of prior literature studying strategic group in pharmaceutical industries (Cool: 1985; Cool and Schendel, 1987; Fiegenbaum et al., 1987; Bogner, et al., 1996; Lee, 2003; Leask and Parker, 2006).) And taking into account the above discussed issues. A selection of six strategic variables was made for us of the hierarchical cluster analysis.

(26)

25

2000; Martens 1988). Additionally, the decision was made to focus mainly on R&D and innovation. Partly due to this dimension being of great importance, partly due to availability of data. For the current study it is simply not possible to obtain the, often sensitive data, needed for a similar study such as Cool and Schendel (1987). Nevertheless give the strategy variables used an accurate, albeit focused, reflection of strategy.

The following six strategic variables have been constructed for the hierarchical cluster analysis.

(1) R&D expenditure

Total firm R&D spending of 2017, illustrating the intensity of the current R&D spending. This variable reflects resource commitments (Cool and Schendel, 1987; Lee, 2003). This variable is used to sketch an imagine of the size of R&D activities of firms from the sample.

(2) R&D spending by sales

Total firm R&D spending of 2017 divided by worldwide total sales of 2017 to get a weighted percentage of the total R&D spending. Illustrating the intensity of the current R&D spending as proposed by Cool and Schendel (1987).

(3) Total number of US patents

Total number of US patents is used to establish effectiveness of patent strategies. According to Katila (2000), patents are a frequently used proxy variable to measure the innovation activity of a firm. Due to the scope of this research, only US patents will be included.

(4) US Patents by employees

(27)

26

(5) Approved applications FDA

The total number of NDAs approved in the year 2017 will be used to give a reflection of the current R&D commitment. (Cool and Schendel, 1987). An NDA is the mandatory application process that every firms willing to sell drugs in the US has to abide by (FDA, 2018). It can therefore be used to give an accurate description of a firms innovation activities.

(6) R&D orientation

The cumulative number of novel drug applications containing new chemical entities (NCEs) will be divided with the cumulative number of new drug applications (NDA). Both data from NCEs and NDAs include the cumulative count until 30 April 2018. This will indicate how new the drugs are, or if they are merely a recombination of existing formulas. This variable reflects R&D orientation as suggested by Cool and Schendel (1987). An NCE is a NDA that contains a new entity and can therefore be used as a measure of a firms innovation activity (Cool and Schendel, 1987). Furthermore Cool and Schendel (1987) suggests a ratio to give an accurate reflection of the R&D orientation of firm.

(28)

27

**. Correlation is significant at the 0.01 level (2-tailed).

4.4 Performances measures

In order to give meaningful conclusions regarding the membership of firms to a certain strategic group, performance measures will be included in this study. Carroll (2018a) argues that performance differences between groups do not have to occur. Nevertheless he argues that performance is an interesting issues as it gives insight in how groups are performing. Performance is tested using a MANOVA test. The performance measures focus on financial performance, to give an indicating whether belonging to a certain group implicates differences in performance. Return on assets (ROA) was chosen as this measure gives an indication of how the company is using the assets to generate money. It is therefore a suitable measure for firm performance. Secondly profit margin was used, as this ratio is a frequent used ratio to measure firm performance. Data for the first measures (ROA) was gathered from the Orbis database, using the year 2017. Unfortunately, data regarding profit margin was found to be incomplete.

(1) R&D spending (2) R&D spending by sales (3) Total number of US patents (4) US Patents by employees (5) Approved applications FDA (6) R&D orientation Pearson Correlation 1 -.189 _,756** _.070 ,585** -.244 Sig. (2-tailed) .365 .000 .740 .002 .240 N 25 25 25 25 25 25 Pearson Correlation -.189 1 -.156 _,518** _-.198 _.199 Sig. (2-tailed) .365 .457 .008 .342 .339 N 25 25 25 25 25 25 Pearson Correlation ,756** -.156 1 .192 ,523** -.208 Sig. (2-tailed) .000 .457 .358 .007 .318 N 25 25 25 25 25 25 Pearson Correlation .070 _,518** _.192 ₁ _-.012 _-.124 Sig. (2-tailed) .740 .008 .358 .956 .555 N 25 25 25 25 25 25 Pearson Correlation _,585** _-.198 ,523** -.012 1 -.219 Sig. (2-tailed) .002 .342 .007 .956 .293 N 25 25 25 25 25 25 Pearson Correlation -.244 .199 -.208 -.124 -.219 1 Sig. (2-tailed) .240 .339 .318 .555 .293 N 25 25 25 25 25 25 (4) US Patents by employees (5) Approved applications FDA (6) R&D orientation (1) R&D spending (2) R&D spending by sales (3) Total number of US patents

(29)

28

Therefore, the profit margin measure was constructed, based on Horrigan (1966) with the following formula: Profit margin= Net profit / Net sales. The data for both the variables net profit and net sales used in this formula were gathered from Orbis using the year 2017. Fortunately, data for these variables was complete.

5. Discussion of results

5.1 Hierarchical cluster analysis

A hierarchical cluster analysis was conducted, combined with a permutation test and a Monte Carlo test. The cluster analysis used six standardized strategic variables. The standardization method was a conversion of all the variables to Z-scores, in order to account for “overweighing” larger numbers. Hierarchical simply indicates that this type of cluster analysis aims to make a hierarchy of clusters. The cluster method used was Ward’s method and the measure squared Euclidean distance. The permutation test generates a null distribution and randomly shuffles the data. It does so to ensure the data used in the cluster analysis is based on random chance (Carroll, 2018a). Additionally, the Monte Carlo test used 999 iterations and generated a null distribution as well.

(30)

29

solutions. Whereas the Monte Carlo test displays significance for cluster solutions with 4-12 groups.

Table 5: Values of the two significant tests

Number

of clusters Ward’s Criteria

Permutation test probability Simulation test Probability 12 5.206 0.023 0.001 11 6.904 0.020 0.001 10 8.672 0.012 0.001 9 11.101 0.010 0.001 8 13.822 0.003 0.001 7 17.699 0.003 0.001 6 24.309 0.004 0.001 5 34.309 0.006 0.005 4 47.607 0.003 0.046 3 71.281 0.009 0.377 2 101.219 0.016 0.647

(31)

30

Figure 1: Scree plot depicting the coefficient changes from the agglomeration schedule

The canonical discriminant functions allows to observe the distances of each case to the belonging centroid. Furthermore, showing how firms are separated in the strategic space (Carroll, 2018a). The plots allow for visual inspection of the clusters and their positions in the strategic space. Upon closer inspection of the canonical discriminant functions, a three-cluster solution was determined to be the most fitting solution. This appears counterintuitive, as the simulation test probability is not significant. However, Figures 2, 3 and 4 below indicate one of the flaws of the Monte Carlo test in determining distinct clustering.

(32)

31

Figure 2: Canonical discriminant functions depicting the positions of the strategic groups in a three-cluster solution

(33)

32

Now the cluster solution with five groups (Figure 4) is deemed significant according to both the permutation and Monte Carlo tests. But by observing the plots one can see that there are a couple of groups closely grouped (groups 1, 3 and 4). This is notable too in a four-group solution, where groups 1 and 2 seem to be one large cluster. An increase in groups, from four to five, led to an extra strategic group in this densely populated part of the strategic space. Although the significant tests show that these solutions are significant, the groups indicate that these groups have similar characteristics. Thus belonging to, for lack of better description, one large “sausage shaped” cluster. This is why the three-cluster solution is deemed more appropriate, as a classification of three different types of strategy is likely to be a more appropriate description of the industry. Now in this occurrence the permutation test has a “liberal” approach to correlations and the Monte Carlo a too “conservative” approach. It is possible that this causes the Monte Carlo tests to be insignificant for cluster solutions with 2

(34)

33

and 3 groups. Even though these can be regarded as potentially interesting clusters, as shown by the canonical discriminant functions. I included solutions with four and five groups to illustrate the decision process. Solutions with more than five groups, that are deemed significant according to the significance tests, are showing an increase in the densely populated part of the strategic space as just described.

One important assumption made in the institutional analysis was that the (US) pharmaceutical industry is in fact characterized by distinct clustering of firms following different strategies. From the results illustrated above this assumption is plausible. As distinct clustering is observable in a multitude of canonical discriminant functions and significant tests.

Furthermore, in line with that assumption is that numerous studies have confirmed the existence of strategic groups in various pharmaceutical industries (Cool: 1985; Cool and Schendel, 1987; Fiegenbaum et al., 1987; Bogner et al., 1996; Lee, 2003; Leask and Parker, 2006). A nuance is placed by Fiegenbaum et al. (1987) which argues that firms in the drug industry have a degree of homogeneity in their strategic orientation. While differences have been observed in this study, other significant cluster solutions (e.g. solution with five groups), show clusters close together in the strategic space. This could in part be attributed to the homogeneity suggested by Fiegenbaum et al. (1987). In addition, from the data it appears that big firms exists that are active in many segments and offer many products. This is similar to results found by Cool and Schendel (1987), whereas smaller firms operate in less segments with less products, which Cool and Schendel also identified.

(35)

34

minimum, first quartile, mean, third quartile and maximum, differences in strategic groups on the strategic variables can quickly be observed, and were used to analyze the strategic groups.

Figure 5: Boxplot for the variable “R&D expenditure” for each strategic group in a three-cluster solution

(36)

35

Figure 7: Boxplot for the variable “Total number of US patents” for each strategic group in a three-cluster solution

Figure 8: Boxplot for the variable “US patents by employees” for each strategic group in a three-cluster solution

(37)

36

Table 6: Overview of variable averages per strategic group in a three-cluster solution

Variables Group 1 Group 2 Group 3

R&D spending 6990257 480658 857630 Expenditure by sales 0.231 0.366 2.438 Number of US patents 3652 310 584 Patents by employees 0.074 0.034 0.358 Approved applications FDA 44.2 4.4 3.5 R&D orientation 0.023 0.103 0.032

Group 1: The big guys

The group with the largest companies in the industry. Characterized by high research and development spending. The R&D orientation signals a low focus on the new drug development. However, the high number of patents suggests the presence of a few key brands protected by a

(38)

37

large amount of patents. The high number of approved applications seems contrasting with the low ratio of R&D orientation. However, the R&D orientation outcome might be due to high the multitude of operations, which would explain the low ratio of expenditure by sales. Simply put, they could have a focus on new drug development and sustaining key brands at the same time. Meaning that they have a high count of approvals to protect and improve their key brands and have new drugs containing new chemical entities. This would led to a seemingly lower R&D orientation. Nevertheless, it could furthermore indicate that there a few brands in the group that are large and focus on new drug development and firms that are solely focusing on generic drugs.

Group 2: The large specialists

The largest group comprised of 16 members. Noticeable is the significant less R&D spending as compared to group one and three. The low R&D spending could indicate that these firms are specialized and have been focusing on developing in single or low number of areas. The relative low expenditure by sales could indicate that research and development is effective.

Moreover, the R&D orientation of this group indicates the specialized nature of this group, as a large portion of NDAs submitted seem to include NCEs. While the average total number of applications is relatively low.

Group 3: The small specialists

The smallest group, consisting out of only two members from the sample. This group has seemingly low R&D spending, but also contains smaller companies. However, R&D spending is significantly higher than for group 2. This could indicate that these firms have been developing new drugs and have the “starts-up costs” that group 2 does not have any more to that degree.

(39)

38

towards highly specialized firms focused on a small area within the market, which requires significant capital expenditure. But less successful and developed as firms in group 2, that seems to have a larger R&D orientation and thus more successful developments with new ingredients.

5.2 Further assumptions regarding the relations between regulations and strategic groups

The significance test bolsters the results from previous studies on strategic groups, as well as the current results of this study. One important addition is the assumption that these observed distinct clustering were partly due to the development and existence of regulations. Something that is hinted towards, but not empirically validated by previous studies (Cool and Schendel, 1987; Fiegenbaum, et al., 1987; Bogner et al., 1996; Lee, 2003). One clear example from the data is the existence of a distinct group that appears to hold a high number of patents. Second to that, this group has a relatively low expenditure by sales. Now while a causal connection cannot be established within this study design, it is plausible that the strong patents law enforced by the US government led to a strong position for these firms. This strong position enables them to have high R&D spending, while simultaneously have high sales. The effective use of patents, in part caused by the government, could have enabled the low expenditure on sales, while still having high R&D spending. The assumption made regarding the existence of a group with large incumbents following similar patent strategies can therefore be seen as plausible as well. Previous research such as Cool and Schendel (1987) have also found groups that were R&D intensive or hinted towards a focus on generic drugs. As their research was founded on the variables verified by industry insiders, the face validity of classification among dimensions such as R&D spending and R&D orientation is strengthened by these results.

(40)

39

group one and three have lower R&D orientation, that is a low ratio NCEs. However, for group one it is arguably due to the high number of NDAs, moreover these firms would have a large range of operations, not solely focusing on generic drugs. The last group on the other hand contains companies that specialize in a certain area, indicating the low ratio was derived from a specialized strategy. This does not mean that no group or firms follow strategies solely on generic drugs, however in a three-group solution, based on these strategic variables and the sample, a distinct group of solely that strategic direction could not have been found. Cool and Schendel (1987) found evidence of strategic grouping that were characterized by the lack of research and development, which hint towards a focus on generic drugs. The existence of those firms would fall within the notion of this study that regulations helped shape distinct strategic groups.

It is arguable that (1) composition of strategic groups, as well as the formation of strategic groups is influenced by regulations in the form of patents and price regulations and (2) these regulations can be deemed impactful enough to influence firm’s strategic choices by altering the terrain (mobility barriers) that firms have to explore. Based on Carroll’s (2018a) theory of mobility barriers, regulations could therefore alter the capacity of firms to overcome mobility barriers and affect the interdependence between firms in strategic groups. However, the exact degree requires alternative study designs as will be explained in subsequent section.

5.3 Performance differences

(41)

40

study. It however remains that any occurring significant performance differences between groups could provide important implications for managers and scholars. Even so, if performance differences are not found, then this is an important finding regardless. Therefore, a MANOVA tests have been run on two, previously selected performance measures. This was done since it cannot be assumed that these variables are independent of one another. Table 7 shows that the F value is in fact significant (0.016) for the profit variable, therefore we can assume the means of the three groups differ significantly on that variable. ROA does not have a significant F value.

Table 8 provides the overview of the MANOVA test. In terms of performance, group three clearly shows a significant difference profit margin performance as compared to group one (0.013) and two (0.023). For return on assets (ROA), no significant intergroup differences were observable.

Table 7: Tests of Between-Subjects Effects

Source Dependent Variable

Type III Sum of

Squares df Mean Square F Sig.

Corrected Model Profit 125765.756a 2 62882.878 5.056 .016

ROA 570.730b 2 285.365 .605 .555 Intercept Profit 111933.469 1 111933.469 9.000 .007 ROA 99.508 1 99.508 .211 .650 ward3 Profit 125765.756 2 62882.878 5.056 .016 ROA 570.730 2 285.365 .605 .555 Error Profit 273612.465 22 12436.930 ROA 10375.470 22 471.612 Total Profit 429557.554 25 ROA 11089.003 25

Corrected Total Profit 399378.221 24

ROA 10946.201 24

(42)

41

Table 8: Multiple Comparisons MANOVA test

Tukey HSD

Dependent Variable (I) Ward Method (J) Ward Method

Mean

Difference (I-J) Std. Error Sig.

95% Confidence Interval

Lower Bound Upper Bound

Profit 1 2 40.0516 50.53724 .711 -86.9011 167.0043 3 281.0529* _89.41571 _.013 _56.4349 _505.6708 2 1 -40.0516 50.53724 .711 -167.0043 86.9011 3 241.0013* _83.64074 _.023 _30.8904 _451.1121 3 1 -281.0529* _89.41571 _.013 _-505.6708 _-56.4349 2 -241.0013* _83.64074 _.023 _-451.1121 _-30.8904 ROA 1 2 10.0021 9.84119 .575 -14.7195 34.7238 3 13.6971 17.41205 .715 -30.0430 57.4373 2 1 -10.0021 9.84119 .575 -34.7238 14.7195 3 3.6950 16.28748 .972 -37.2202 44.6102 3 1 -13.6971 17.41205 .715 -57.4373 30.0430 2 -3.6950 16.28748 .972 -44.6102 37.2202

(43)

42

6. Conclusion

Through the theoretical lens of new economic institutionalism, I have tried to unravel to what degree regulations impact the formation and composition of strategic groups. Based on Williamson’s (2000) four level framework this study included both an institutional analysis (level 2) and empirical analysis (level 3). Assumptions derived from the analysis of the most important regulations (level 2) were deemed plausible after the empirical analysis (level 3) indicated distinct clustering. Moreover, the characteristics (i.e. cumulative number of patents) of these groups further contributed to the plausibility that regulations have in fact shaped the industry.

The study has several important implications, both for managers and scholars. First by illustrating the applicability of the new methodology, it becomes clear that the potential use is wide. Researchers should continue to develop and improve the methodology in order to advance the field. Secondly, the framework by Tywoniak et al. (2007) proved to be a useful tool in structuring the analyses, and could provide a useful tool in future strategic group analyses. Especially when intensive knowledge of the industry is required for a thorough analysis. Lastly, the preliminary theory of Carroll (2018a) regarding mobility barriers and interdependence is extended by proposing how regulations might fit into this theory. While this extension should be nuanced due to this study design, it could provide a stepping-stone for future developments of this preliminary theory.

(44)

43

It was my wish to incorporate a longitudinal study design, similar to Cool and Schendel (1987) and Fiegenbaum et al. (1987). Conducting a study over longer periods would have drawn important implications regarding the stability and (perceived) impact of discussed regulations on formation and composition of strategic groups. Herein, one could for example study how strategic group composition is altered by introduction of a regulation by observing strategic group composition before and after introduction of said regulation. This would simultaneously shed light on both the impact of regulations and the stability of strategic group composition. Due to data constraints, it was only possible to conduct a cross-sectional study. Fortunately, with such a study one is still able to derive meaningful implications but one is limited to observe effects of regulations on a snapshot of data, in this case years after the regulation was introduced. Albeit it can be made plausible that present day strategic groups are in part shaped by these regulations the study design should be noted as a limitation of this study.

Future research could include the adoption of a longitudinal design in an effort to shed light on the above mentioned impact of regulations and the stability of strategic group composition. This research can then act as a stepping-stone once more.

(45)

44

composition, therefore governments must take in account the effect that their regulations could have on firms (financial) performances. Especially, since at the end of the chain, performance of these firms could affect national economy. Future research could design a multi-country study, similar to Martens (1988), aimed towards the implications that differences in country regulations could have on strategic group formation, composition, stability and profitability of strategic group membership.

Due scope considerations, the selection process of firms chose to exclude firms under 500 employees. This decision was made in the consideration that the bulk of firms that make up the market share were firms larger than 500 employees. However, it is entirely possible that different strategies are employed by these smaller firms. Especially when it comes to handling the high development costs. Future research could be aimed at the inclusion of these smaller firms.

(46)

45

7.

References

Barney, J. B., and Hoskisson, R. E. (1990). Strategic groups: Untested assertions and research proposals. Managerial and decision Economics, 11(3), 187-198.

Carroll, C. (2018a). Unpublished manuscript. University of Groningen.

Carroll, C. (2018b). Significant Clustering: Implications for Strategic Groups Research. Australian Academy of Business and Economics Review, 4(2), 60-73.

Caves, R. E., Whinston, M. D., Hurwitz, M. A., Pakes, A., and Temin, P. (1991). Patent expiration, entry, and competition in the US pharmaceutical industry. Brookings papers

on economic activity. Microeconomics, 1991, 1-66.

Caves, R. E., and Porter, M. E. (1977). From entry barriers to mobility barriers: Conjectural decisions and contrived deterrence to new competition. The quarterly journal of

economics, 241-261.

Cool, K. O. (1985). Strategic group formation and strategic group shifts: a longitudinal

analysis of the US pharmaceutical industry, 1963-1982. Purdue University.

Cool, K. O., and Schendel, D. (1987). Strategic group formation and performance: The case of the US pharmaceutical industry, 1963–1982. Management science, 33(9), 1102-1124.

DiMaggio, P., & Powell, W. W. (1983). The iron cage revisited: Collective rationality and institutional isomorphism in organizational fields. American sociological review, 48(2), 147-160.

(47)

46

Faulkner, D. O., and Campbell, A. (Eds.). (2006). The Oxford handbook of strategy: a strategy

overview and competitive strategy (Vol. 1). Oxford University Press.

Fiegenbaum, A., Sudharshan, D., & Thomas, H. (1987). The concept of stable strategic time periods in strategic group research. Managerial and Decision Economics, 8(2), 139-148.

Fiegenbaum, A., & Thomas, H. (1990). Strategic groups and performance: the US insurance

industry, 1970–84. Strategic Management Journal, 11(3), 197-215.

Hair, J. F., Anderson, R. E., Tatham, R. L., and Black, W. C. (1992). Multivariate Data Analysis New York: Macmillan.

Harrigan, K. R. (1985). Vertical integration and corporate strategy. Academy of Management

journal, 28(2), 397-425.

Horrigan, J. O. (1966). The determination of long-term credit standing with financial ratios. Journal of Accounting Research, 44-62.

Hunt, M. S. (1972). Competition in the major home appliance industry, 1960-1970. Harvard University.

Katila, R. (2000). Using patent data to measure innovation performance. International Journal

of Business Performance Management, 2(1-3), 180-193.

Ketchen Jr, D. J., and Shook, C. L. (1996). The application of cluster analysis in strategic management research: an analysis and critique. Strategic management journal, 441-458.

Ketchen Jr, D. J., Thomas, J. B., and Snow, C. C. (1993). Organizational configurations and performance: A comparison of theoretical approaches. Academy of management

(48)

47

Leask, G., and Parker, D. (2006). Strategic group theory: review, examination and application in the UK pharmaceutical industry. Journal of Management Development, 25(4), 386-408.

Lee, J. (2003). Innovation and strategic divergence: An empirical study of the US pharmaceutical industry from 1920 to 1960. Management Science, 49(2), 143-159.

Lehman, B. (2003). The pharmaceutical industry and the patent system. International

Intellectual Property Institute, 1-14.

Malerba, F., and Orsenigo, L. (2015). The evolution of the pharmaceutical industry. Business

History, 57(5), 664-687.

Martens, R. (1988). Strategic group formation and performance. the case of the pharmaceutical industry in five EC countries 1978-1985. Faculty of Applied Economics, 402

Mehra, A. (1996). Resource and market based determinants of performance in the US banking industry. Strategic Management Journal, 307-322.

Milligan, G. W., and Sokol, L. M. (1980). A two-stage clustering algorithm with robust recovery characteristics. Educational and psychological measurement, 40(3), 755-759.

Nalebuff, B. J., Brandenburger, A., and Maulana, A. (1996). Co-opetition. London: HarperCollinsBusiness.

Numagami, T. (1998). Perspective—the infeasibility of invariant laws in management studies: A reflective dialogue in defense of case studies. Organization science, 9(1), 1-15.

Oster, S. (1982). Intraindustry structure and the ease of strategic change. The Review of

(49)

48

Parker, R. (1999). From national champions to small and medium sized enterprises: changing policy emphasis in France, Germany and Sweden. Journal of Public Policy, 19(1), 63-89.

Peltzman, S. (1973). An evaluation of consumer protection legislation: the 1962 drug amendments. Journal of political economy, 81(5), 1049-1091.

Perryman, M. M., and Rivers, P. A. (2011). Strategic groups in health care: a literature review. Health services management research, 24(3), 151-159.

Porac, J. F., Thomas, H., and Baden‐Fuller, C. (1989). Competitive groups as cognitive communities: The case of Scottish knitwear manufacturers. Journal of Management

studies, 26(4), 397-416.

Porter, M. E. (1973). Consumer Behavior, Retailer Power, and Manufacturing Strategy in Consumer Goods Industries. Unpublished doctoral dissertation, Harvard University.

Porter, M. E. (1979). The structure within industries and companies' performance. The review

of economics and statistics, 214-227.

Porter, M. E. (1980). Competitive strategy: Techniques for analyzing industries and competition. New York, 300, 28.

Punj, G., and Stewart, D. W. (1983). Cluster analysis in marketing research: Review and suggestions for application. Journal of marketing research, 134-148.

Rosch, E., and Lloyd, B. B. (Eds.). (1978). Cognition and categorization.

(50)

49

Schilling, M. A. (2010). Strategic management of technological innovation. Tata McGraw-Hill Education.

Tancer, R. S., and Mosseri‐Marlio, C. (2002). Evolution of pharmaceutical regulations and their consequences in the European Union and the United States. Thunderbird International

Business Review, 44(2), 263-281.

Tang, M. J., and Thomas, H. (1992). The concept of strategic groups: Theoretical construct or analytical convenience. Managerial and Decision Economics, 13(4), 323-329.

Tywoniak, S., Galvin, P., and Davies, J. (2007). New institutional economics' contribution to strategic groups analysis. Managerial and Decision Economics, 28(3), 213-228.

Williamson, O. E. (1998). Transaction cost economics: how it works; where it is headed. De

economist, 146(1), 23-58.

Regulations and strategic groups