Strategic Groups and Cooperation in the Dutch Hospital Industry

(1)

Strategic Groups and Cooperation in the Dutch Hospital Industry

Joint Master Thesis

MSc Business Administration - Health

MSc Business Administration - Strategic Innovation Management

Elsemiek A. E. Schulenklopper

S2466562

Van Wassenaerstraat 8a

9726 HP Groningen

Supervisors: Dr. C. Carroll & Dr. M. A. G. van Offenbeek

Faculty of Economics and Business

University of Groningen

(2)

Abstract

This study investigates the link between strategic group research and the relational view regarding the patterns of cooperation within and between strategic groups. Alliances within strategic groups could be formed in order to pool resources and exploit economies of scale. Alternatively, alliances between strategic groups could be formed based on the complementarity of capabilities and resources. These alternative sources of motivations for forming alliances are examined in the Dutch hospital industry. Three internally coherent strategic groups were found in a sample of 67 hospitals. Furthermore, six main types of alliances were characterized as either a pooling or complementary alliance, after which they were analysed on formation within or between strategic groups. The findings suggest that integration alliances, information exchange alliances, and quality agreements are formed more often between strategic groups with the aim of gaining complementary capabilities and resources. The cooperation patterns of the three other types of alliances could not be explained by strategic group research, as purchasing alliances and educational agreements occur as often within as between strategic groups. The prediction regarding non-competing region alliances could also not be supported due to missing data. Further research should therefore provide additional insights.

Keywords: strategic groups, relational view, Dutch hospital industry, pooling alliances, complementary

(3)

1. Introduction

Two hospitals in the Netherlands recently filed for bankruptcy within one week. The consequences for the parties involved were substantial: patients had to move to other hospitals, healthcare professionals lost their jobs, and large outstanding debts were not repaid. Due to this event, the current system and regulations are subject to critique. Criticism is also directed at the management of these institutions as they were not able to adequately deal with their complex environment (Sondermeijer, 2018).

Before 2006, Dutch hospitals could not file for bankruptcy and the government ensured that the institutions remained open. However, recent developments in the Dutch healthcare system and the shift towards a more conservative national government have caused a (still ongoing) transition (van de Ven & Schut, 2009). In 2006, a new Health Insurance Act was implemented which allows hospitals to negotiate with health insurers about the price and quantity of an increasing number of treatments (Maarse, Jeurissen, & Ruwaard, 2016). This development pressures hospitals to lower treatment prices and resulted in a competitive environment for hospitals (Schut & van de Ven, 2011).

(5)

Various researchers already performed a strategic group analysis in a certain hospital industry (Garcı_{́a & de Val Pardo, 2004; Ketchen et al., 1993; Marlin, Hoffman, & Lamont, 1994; Nath & Gruca,} 1997; Schreyhögg & von Reitzenstein, 2008; Short, Palmer, & Ketchen, 2002; Spencer, Peyrefitte, & Churchman, 2003). These studies are, however, rather narrow in their geographical application as well as their choice of hospital types. For instance, Schreyögg and von Reitzenstein (2008) studied strategic groups and the performance differences in German Academic Medical Centres. Moreover, Garcı_{́a and de Val Pardo} (2004) studied Spanish hospitals, but only looked at general hospitals, leaving out University Medical Centres (UMC’s). Although these studies are interesting and give valuable insights, it remains questionable whether their results are generalizable to the Dutch hospital industry as the Dutch healthcare sector contains a wide range of hospitals. It includes institutions ranging from UMC’s to small regional hospitals. Strategic group research is context-specific in only describing the industry under investigation. A thorough analysis of the Dutch industry’s different strategic groups has not yet been conducted and this study fills this gap.

As previously discussed, cooperation is still important for hospitals and different types of alliances are being formed. The purposes of forming alliances can be illustrated using the relational view (Dyer & Singh, 1998). This theory states that organizations form alliances to generate relational rents, which are advantages that cannot be achieved by either partner alone. How the relational view is linked to strategic groups remains unclear, as pointed out by Koka and Prescott (2002). Strategic group research takes place on the industry-level, whereas the relational view explores organizations on a relational level. This fundamental difference in level between strategic group research and the relational view might be an explanation why the concepts are currently not linked to each other. Rather than focusing on both concepts, studies focus on either one or the other, which leaves a gap in academic literature.

The link between these two research streams is interesting in order to investigate the source of motivation of hospitals to cooperate. On the one hand, forming alliances within strategic groups helps in pooling resources (Nohria & Garcia-Pont, 1991). This contributes to increased economies of scale, which results in lower costs. On the other hand, forming alliances between strategic groups exploits the complementarity of resources. It can be interesting to study how exactly this logic could apply to the Dutch hospital industry. This results in the following research question: Which types of alliances are mostly formed within strategic groups and which types of alliances are mostly formed between strategic groups?

(6)

hospital industry is a second, empirical, contribution. This has not yet been done in previous studies and will give managers in Dutch hospitals new insights into their industry.

For the identification of strategic groups, a cluster analysis was performed based on key strategic variables. A total of 67 Dutch hospitals were included in the test. A common clustering method is Ward’s Hierarchical Clustering technique (Ward, 1963). However, its use in strategic group analysis has been criticized for it always producing groupings even with random data (Hatten & Hatten, 1987). More specifically, the question remains to what extent it results in distinct groupings of firms (Carroll & Thomas, 2019). Hence, this study first tested whether truly internally coherent strategic groups exist in the Dutch hospital industry. After that, the different clusters were identified and described. Furthermore, it was determined which types of alliances, in terms of their purpose, occur more often between strategic groups and which ones occur more often within strategic groups. A validity check was performed to test whether the outcomes of the clustering technique correspond to the hospital classes already in place in the Netherlands. Differences between the strategic groups and the classes were analysed and described, providing an indication of the usefulness of this study.

The structure of this paper is as follows: first, the literature review is presented, which starts with a description of the Dutch hospital industry. Subsequently the literature on strategic groups and the relational view is reviewed, in which the predictions of the findings are elaborated. After this, the method is explained. Next, the results of the clustering analysis are presented, in which it is established which types of alliances are most frequently formed within strategic groups and which types of alliances are mostly formed between strategic groups. Lastly, the discussion and conclusion are presented, including the theoretical and managerial implications as well as the limitations and directions for future research.

2. The Dutch Hospital Industry

(7)

Dutch hospitals are thus indirectly significantly impacted by these reforms, which can have two consequences according to Meijer, Douven, and van den Berg (2010). Specialization becomes more important as insurers and patients are putting more emphasis on the quality and prices of the hospital. This is encouraged by the increased transparency towards patients. Hospitals are obligated to publish their prices per treatment and make this information freely available for patients. Moreover, information on the quality of care is freely available. The ‘Nederlandse Patienten Consumenten Federatie’ (NPCF) has initiated a website on which thousands of patient reviews are posted each month (ZorgkaartNederland, n.d.). This information can be used to see which hospital should deliver the best treatment. Thereby, insurers use this particular information as an assistance for their selective purchasing. Hospitals on the other hand can use this information to compare themselves with other healthcare providers. A second consequence of increased competition for the Dutch hospitals is the higher number of mergers and acquisitions. The underlying reason for this is either the increase in economies of scale, the increase in market power, or both (Chatterjee, 1986; Gaynor & Town, 2011).

The healthcare reforms were supposed to be beneficial for citizens in two ways. Firstly, citizens can freely choose an insurance company they prefer, since the barriers to switch between insurers were partly removed (Leu et al., 2009). The system assumes that citizens act freely and think rationally about the products they buy based on the information that was offered to them. Economists reason that this is key for this market competition to work (Rosenau & Lako, 2008). Insurance companies have to stay very alert in their negotiation, presumably always having patient benefits in mind. Another advantage is cheaper and more extensive policy offering, therefore it is more responsive to the patient’s needs (Schut & van de Ven, 2005). Whether these benefits are actually achieved for citizens remains a question (van de Ven & Schut, 2009). The prices for consumer premiums are increasing while patient’s perceived quality is declining (Rosenau & Lako, 2008).

As explained, hospitals are simultaneously competing and collaborating with each other to improve their quality of care and lower their prices, which forces them to think more strategically about their competitive position and the competitive position of their partners. As explained, strategic groups and the relational view will be used to analyse this. In order to do so, these two concepts have to be explained in more detail and conceptually integrated or at least related to one another.

3. Strategic groups

(8)

strategies within groups and different strategies across groups (Carroll & Thomas, 2019). Within a strategic group, organizations are likely to respond in the same manner as they resemble each other closely in terms of competition, actions and results (Hatten & Hatten, 1987). Since firms within a strategic group follow more or less the same strategy, the number of strategic groups within an industry closely resembles the number of popular strategies within that industry (Marlin et al., 1999).

A strategic group research has a practical role for managers in analysing and understanding the structure of an industry (Castle, 2003). It helps in identifying both the degree of segmentation that exists in an industry and the characteristics of the segments (Marlin et al., 1999). It offers an overview of the competitive landscape of a particular industry as it provides insights on the different competitor strategies and actions in the marketplace (Harrigan, 1985; Zinn, Aaronson, & Rosko, 1994). In other words, a strategic group analysis helps with the determination of the organization’s most important competitors (Kirby, 2012). Once strategic groups are identified, they can be a fruitful reference point for managers and organizations to use when formulating and implementing their own strategy (Fiegenbaum and Thomas, 1993; Marlin et al., 1999).

This strategy formulation and implementation is a difficult task. A method used to ease strategy formulation is by analysing and structuring the competitive environment (Porac, Thomas, & Baden-Fuller, 2011). Here, competitors are compared and are labelled the same ‘type’ if they share the same attributes and decision-makers formulate strategies that suit accordingly. The environment is often too complex to compare on a firm-level; therefore, decision-makers form cognitive groups of firms that are similar using implicit classification schemes (Tang & Thomas, 1992). These so-called cognitive strategic groups are used as a framework to make decisions (Cheng & Chang, 2009). Performing a clustering analysis to empirically identify strategic groups can help decision-makers assess their own cognitively formed strategic groups, as for example done by Spencer et al. (2003).

Hence, understanding the strategic groups in an industry helps in grasping the differences between companies within that industry and provides an identification of the relative competitive position (Fiegenbaum & Thomas, 1993). Although there is a great number of studies on strategic groups, an empirical debate is currently being held on the identification of strategic groups, and two opposing views can be identified.

3.1. Independent and Interdependent view

(9)

strategic group researchers and divided strategic group research into two opposing views, the independent view and the interdependent view (Carroll & Thomas, 2019).

The independent view states that the clustering technique is a convenient way of analysing the industry in that it provides a mapping of industry participants. Strategic clustering can be used to analyse the industry and interpret actors’ strategic choices. However, a clustering technique always provides clusters, even with random data. The organizations within groupings are not bound to one another in any way (Carroll & Thomas, 2019), so the organizations are considered as independent from each other. Hence, according to this view, strategic clustering should be seen as an atheoretical ‘analytical convenience’ (Hatten & Hatten, 1987). This view can be seen as the default unless significant clustering is observed.

In contrast, the interdependent view states that because there are a few popular strategic positions some organizations in the industry follow the same strategy. The outcomes of significant clustering show groups of firms holding similar strategic positions, or strategic groups. Within a strategic group, firms are competing both upstream and downstream. Upstream competition takes place for suppliers and downstream competition takes place for consumers (Carroll & Thomas, 2019). This means that firms within a strategic group, because they follow the same strategy, can influence each other’s performance and are therefore interdependent of one another. Organizations within a strategic group tend to be aware of their interdependency (Porter, 1979) and manage it by using both cooperative and competitive means. If the strategic groups consist of a few firms, it takes the form of an oligopoly. Within an oligopoly, ‘true group-level effects’ can be achieved by means of collusion, which takes place when rival firms cooperate for their mutual benefit (Carroll & Thomas, 2019). The decision of a few firms to collude within an oligopoly can influence the industry as a whole. Collusion is a method for dealing with interdependence.

In order to test the interdependent view empirically, a significance test will be used. Without a test for significant clustering, it is never clear if the analysis has truly discovered distinct groups or simply created arbitrary groupings (Hatten & Hatten, 1987). A significance test reports whether the variance found within the group is lower than would be expected from random data. When this is the case, internally coherent strategic groups are found in the dataset, supporting the interdependent view. Table 1 provides an overview of the two opposing views on strategic groups.

(10)

Table 1: Overview of two opposing views in strategic group research

Insights into Independent view Interdependent view Strategic group

research

Groupings of firms are solely an analytical convenience.

Internal cohesion and external isolation results in significantly distinct strategic groups.

Significant clustering Does not require the existence of significantly distinct strategic groups in the dataset.

Significantly distinct strategic groups exist in the dataset.

4. The relational view

(11)

4.1. The relational view and strategic groups

The interdependent view of strategic groups discusses cooperation as a means to manage interdependence under the conditions of an oligopoly (Carroll & Thomas, 2019). In this case, rival firms within a strategic group collude in order to gain mutual benefits. However, the cooperation patterns for larger strategic groups are not discussed. In this case it is uncertain which alliances are formed within or between strategic groups and why. The research by Nohria and Garcia-Pont (1991) provides more information on the types of alliances that might be encountered between and within strategic groups. Their research is on strategic blocks and strategic groups within the automotive industry. A strategic block is a set of firms that form more linkages with each other than with other firms in the industry. The linkages they consider in their research are both vertical and horizontal, but in all cases the partners continue to compete with each other. The reasoning behind forming strategic linkages can be traced back to the theory behind forming alliances: the relational view. For instance, strategic linkages are formed to increase economies of scale and scope, getting access to new knowledge, or enhancing existing knowledge (Kogut, 1988).

Nohria and Garcia-Pont (1991) identified two types of strategic blocks: complementary blocks composed of firms from different strategic groups, and pooling blocks composed of firms from the same strategic group. These blocks resemble the two ends of a spectrum and can therefore be referred to as two ‘ideal types’. The formation of strategic blocks depends on the desired capabilities of the firm. Firms will form linkages to gain access to those capabilities of which they perceive that it leads to an improved competitive position (Nohria & Garcia-Pont, 1991).

On the one hand, pooling blocks are strategic blocks formed between firms from the same strategic group. An important note for this argumentation is that firms or the same strategic group are likely to act similarly since they resemble each other closely in terms of competition, actions, and results (Hatten & Hatten, 1987). Hence, strategic group members are similar in terms of strategic capabilities and reactions to environmental disturbances (Emerson, 1972). Thus, forming linkages with firms in the same strategic group provides the opportunity to enhance unique capabilities the firm already possesses. Another reason for forming pooling links is increasing economies of scale resulting in the reduction of costs (Nohria & Garcia-Pont, 1991). As previously discussed, having relation-specific assets with alliance partners allow for economies of scale, especially with a high exchange volume (Dyer & Singh, 1998).

(12)

(Koka & Prescott, 2002; Duysters & Hagedoorn, 1995). Also, a complementary alliance allows for the creation of new kinds of resources and capabilities as a result of synergistic effects (Chung et al., 2000).

To sum up, the research by Nohria and Garcia-Pont (1991) provides valuable insights into the types of alliances formed within and between strategic groups. Alliances within a strategic group enables for the pooling of resources, or for flexible deployment of resources. This could possibly lead to economies of scale, for instance when investing in relation-specific assets. Alliances between strategic groups are formed to achieve synergies in terms of knowledge, resources and capabilities, resulting in superior relational rents. Table 2 provides an overview of the arguments of the relational view and the interdependent view regarding strategic cooperation. In order to examine how these arguments apply to the Dutch hospital industry, the grey quadrant will now be discussed.

Table 2: Overview of the definitions regarding strategic cooperation

Insights into Interdependent view Relational view Strategic cooperation

in an industry

Cooperation takes place under some conditions (oligopoly) to manage interdependency

Cooperation within strategic groups for pooling resources/capabilities

Cooperation between strategic groups for complementary resources/capabilities

4.2. Types of alliances between Dutch hospitals

In the Netherlands, hospitals are closely monitored when it comes to forming alliances. Hospitals that want to form an alliance are required by law to report it. The joining parties have to submit their alliance request to the NMa (The Netherlands Competition Authority), who will judge and decide if the requested cooperation is line with the ‘Competition Law’ (‘Mededingingswet’ in Dutch) and the Healthcare (Market Regulation) Act (‘Wet Marktordening Gezondheidszorg’ in Dutch). Specifically, before the new alliance is approved by the NMa, it should comply with the prohibition on cartels. There are seven main types of alliances that are permitted between Dutch hospitals. The first six are discussed by Wiggers and Eggers (2011), and a seventh type was added by the Dutch Association of Hospitals (NVZ, 2018). A table is depicted at the end of this paragraph to give an overview of the predicted findings per type of alliance (see Table 3).

1. Integration alliance. These alliances form an integrated care chain in which, within

(13)

between hospitals. This type of alliance is only permitted when there are enough other types of healthcare providers in the same integrated care chain and when there are no price or market agreements between the hospitals (Wiggers & Eggers, 2011). The ultimate goal of such an alliance is to improve patient care by better coordination of the services that are provided by the different hospitals (Shaw, Rosen & Rumbold, 2011). Integration does not only involve connecting different hospitals, but the entire care chain. It includes primary, acute, and long-term care and also housing services. In this study, however, only the integration between hospitals will be considered, like the integration between secondary and tertiary care. Integration alliances becomes increasingly important due to a recent trend in the patient population: the increase in chronic diseases (RIVM, 2018). Chronically ill patients move through different levels of the care chain to a greater extent; therefore, it becomes more important to efficiently integrate them. This kind of alliance is established because hospitals have different specialties. Certain hospitals are able to deliver highly specialized or expensive healthcare services, whereas other hospitals are not able to deliver these services. In other words, certain hospitals have certain strategic capabilities or resources that the other hospital is missing and can therefore be referred to as complementary. Consequently, integration alliances are predicted to occur more often between strategic groups than within strategic groups.

2. Information exchange alliance. This alliance is a partnership set up to form an information

exchanging network, which is focused on the efficient and transparent exchange of patient- and medical information. Sharing information can improve the quality of hospital care as specialists who see new patients can better anticipate on the medical history and allergies. Also, it reduces operational costs on the long run (Miller & Tucker, 2014). The choice of partners for this type of alliance mostly lies in the geographical proximity of the hospitals, as hospitals in the same regions mostly receive the same patients and hence have to exchange patient information (Unertl, Johnson, Gadd, & Lorenzi, 2013). It is therefore not possible to predict whether these alliance types are mostly formed within or between strategic groups, hence they are predicted to occur both between and within strategic groups.

3. Purchasing alliance. This alliance is set up to save costs by centralizing procurement. An

example of such an alliance in the Netherlands is the Inkoopalliantie Ziekenhuizen, formed by eight Dutch hospitals to centralize procurement. In 2017, 21 million euro was saved due to the alliance’s centralized procurement strategy (Benschop, 2017). According to Burns and Lee (2008), this alliance represents a pooling alliance and they are an important source of economies of scale for hospitals. Therefore, this alliance is predicted to occur more within strategic groups than between strategic groups.

4. Quality agreement. This alliance focuses on increasing the quality of care, for instance by

(14)

opportunity to improve their quality and offer a more extensive range of treatments (Ellerbeck, Bhimaraj, & Perpich, 2004). Larger hospitals also benefit as it allows them to focus more on complex treatments, because the less complex cases can be referred to other hospitals. Because smaller hospitals learn from the larger hospitals and larger hospitals can use the resources of the smaller hospitals, this type of alliance is referred to as complementary. Therefore, in this research it is predicted that a quality agreement will mostly occur between strategic groups.

5. Non-competing regions. This type of alliance is formed between hospitals in non-competing

geographical regions in the Netherlands to share capabilities and resources. These hospitals perform the same specialist treatment but cover different geographical markets, hence they do not compete on a nationwide market (Wigger & Eggers, 2011). The goal of this alliance is to pool knowledge about certain treatment procedures therefore this type of alliance is characterized as a pooling alliance and is predicted to occur more within strategic groups.

6. Exempted agreements. This is the final type of alliance mentioned by Wiggers and Eggers

(2011). Various other types of agreements are allowed by the NZa when they conform with three cumulative conditions. The alliance results in a cost reduction or an improvement in the production, distribution, or technique (1), and there is a clear advantage for the patient, which cannot be achieved by either hospital in isolation (2), and enough competition remains in place (3). Various types of agreements are exempted, like research and development (R&D) agreements, specialization agreements, and vertical agreements. This is a very diverse type of alliance; therefore, it will be left out of the analysis.

7. Educational agreement. This type of alliance is not mentioned by Wiggers and Eggers (2011),

(15)

Table 3: Predicted alliance type formation: within or between strategic groups.

Main types of alliance Goal Prediction

1. Integration Form an integrated care chain Between

2. Information exchange Exchange patient and medication information Within and between 3. Purchasing Centralize procurement for economies of scale Within

4. Quality Jointly improving the quality of care Between

5. Non-competing regions Exchange knowledge and resources Within

6. Educational Train medical students Between

5. Method

5.1. Research setting

The Dutch hospital industry was chosen as a research setting. Data for 77 hospitals from 2016 were obtained from DHD (2018), which gathers, controls, and analyses data of all Dutch hospitals. Their list of hospitals includes University Medical Centres (UMC’s), top clinical hospitals and general/regional hospitals, but does not include specialized institutes, such as cancer institutes. All other data was also gathered from 2016 for practical reasons, as the annual reports were easily accessible for that year as well as the data from the Dutch Ministry of Health, Welfare, and Sport. At the time of this study, two hospitals went bankrupt, but they are still included in this analysis as they were still in operation in 2016. Also, two hospitals merged in 2018 (AMC and VUMC), but as this was after 2016, these hospitals are considered as separate hospitals in this analysis.

5.2. Data collection

(16)

5.3. Alliance types

Alliance information of all hospitals was gathered from the websites, annual reports of the year 2016 and, if needed, additional sources (see Appendix 1). An extensive amount of asymmetry was encountered in reporting the alliances. Smaller, regional hospitals reported the cooperation with larger hospitals but this link was often not mentioned by the larger hospitals. If either hospital reported an alliance, that alliance was counted for both hospitals. Because of the asymmetry in information, it is possible that some alliances were not mentioned by any of the partners in their annual reports and are therefore not analysed. This has some implications for the reliability of this research, which will be discussed later on. Multi-partner alliances were reported as dyadic links between two hospitals. A total of 415 unique cooperation links were found in the first dataset of 77 hospitals. Excluding the hospitals with missing data resulted in 304 unique cooperation links between 67 hospitals. The alliances were classified into the main types of alliances. However, one type is very broad (exempted agreements) and contains many different subtypes, so this type of alliance is left out of the analysis. Consequently, 6 types of alliances will be analysed in this study.

5.4. Strategy variables

(17)

Table 4: Overview of the strategy variables

Variable Description

Size This is measured using the number of beds, which is a common proxy of the size and capital investments of hospitals (Harrison, Coppola, & Wakefield, 2004; Garcìa & de Val Pardo, 2004; Nath & Sudharsan, 1997; Schreyhögg & von Reitzenstein, 2008). Due to a high standard deviation, this variable was standardized by constructing a Z-score.

Employees This variable measures the total number of employees in the hospital (Claver-Corter, Molina-Azorín, & Pereira-Moliner, 2006). It was standardized by constructing a Z-score.

Scope This is measured using a proxy variable. There is a pressure on hospitals to differentiate by offering specialized quality on treatments (NOS, 2011). The number of quality marks the hospital owns can be referred to as the ‘scope of their specialization’. Hospitals in the Netherlands can acquire 11 different quality marks, each of which can be acquired individually (Independer, 2019) (see Table 5). A percentage was calculated to generated this proxy variable by dividing the number of quality marks the hospital owns by 11. One particular quality mark, namely the quality mark for child-oriented care, has three sub-quality marks, each of which can be achieved independently. This is considered in this study by taking the average of these three sub-quality marks to represent the child-oriented care quality mark. One quality mark (for colon cancer) was left out of the analysis, because of missing data.

Research Only University Medical Centres (UMC’s) have research as their primary process.

Other hospitals can participate in research, but it is not one of their primary processes (STZ, 2018). This variable is dichotomous, as a score of 1 shows that it is a UMC and a label of 0 shows that it is not.

Education UMC’s and top clinical hospitals are actively involved in education. This variable is

dichotomous, 1 if it is an educating hospital, and 0 if it is not.

DTC’s (type B) Type B Diagnose Treatment Combinations (DTC’s) are the treatments that hospitals are allowed to compete on, which are mostly common treatment procedures (Maarse et al., 2016). The total number of type B DTC’s was standardized using a Z-score.

Debt-to-asset ratio

(18)

Table 5: Quality marks for Dutch hospitals

Quality mark Number of hospitals

Varicose vein 67 Vascular 69 Senior-friendly 37 Fertility 36 Neo 7 Child-oriented care 69 - Children’s ward 64 - Day-care 48 - Maternity ward 28 Stoma care 62 Prostate cancer 59 Lymphoma 58 Breast cancer 68

Colon cancer Missing data

5.5. Data analysis

A Pearson Correlation test was conducted to test if the variables are correlated and are therefore not suitable for the clustering technique. After that, a hierarchical cluster analysis was done, using Ward’s method (1963) and squared Euclidean distance. Next, a significance test was done using a permutation test, which is a test for internal cohesion. This test creates a null distribution by randomly shuffling the data (Carroll & Thomas, 2019). Ward’s method is an agglomeration approach, which starts with each separate hospital as a cluster. In each step, different clusters are merged together such that the total within-variance in clusters is minimized. In order to find the appropriate number of clusters, the significance test, agglomeration table and a scree plot were analysed. The scree plot visually displays the variation between and within groups for different strategic group solutions. Once the number of clusters was determined, it was important to find out how they differ, which was done using a discriminant analysis and boxplots. A discriminant analysis shows the degree to which each variable helps in separating the strategic groups. The boxplots visually display the distribution of the variables per strategic group and helps in describing them in terms of strategy. Finally, the cooperation patterns per type of alliance will be analysed and described.

5.6. Validity and reliability

(19)

two formal classes. These hospitals have less beds and less equipment. They mostly deliver basic care, which implies that these hospitals have to compete with all other hospitals in the industry since all hospitals are required to deliver basic care (Castellijns, van Kollenburgh, & Oh, 2011). Hence, increasing efficiency and thereby reducing costs is therefore increasingly important for general/regional hospitals. Top clinical hospitals are specialized in one or more care areas, besides delivering basic care, and therefore receive patients from a wider geographical area. Besides that, they educational hospitals, implying that they train medical specialists. Top clinical hospitals also perform research to improve patient care, but this is not one of their primary processes (STZ, 2018). There are 28 top clinical hospitals in the Netherlands. In University Medical Centres (UMC’s), the Faculty of Medicine is merged with the Academic Hospital. Their three core responsibilities are teaching and training, basic and clinical research and (tertiary) patient care (NFU, 2008). There are 8 UMC’s in the Netherlands.

This validity check shows whether the three formal hospital classes correspond to the strategic groups found in the research. The construct validity of this research is hereby assessed. Construct validity occurs when the theoretical constructs accurately represent the real-world situations they are intended to model (Garver & Mentzer, 1999). If this is not the case, the usefulness of the clustering analysis can be examined. It could however also be the case that the formal classes that are currently in place might be obsolete and should be reassessed. Besides comparing the formal classes and strategic groups, cooperation patterns can be examined to find out whether the predications made per type of alliance will hold for the formal classes to a greater extent than for the clusters. The final step of the validity research is including the available alliance information on the hospitals with missing data in the cooperation analysis. Hereby, it can be assessed whether the outcomes hold for the whole hospital industry in the Netherlands. Consequently, the usability and validity of this research can be examined.

6. Results

6.1. Cluster analysis

(20)

Table 6: Pearson Correlation (N=67) Strategy

variables

Z-score Size

Scope Research Debt-to-asset ratio Z-score Employees Education Scope 0.367* Research 0.515** -0.096 Debt-to-asset ratio -0.194 -0.171 -0.056 Z-score Employees 0.891** 0.166 0.781** -0.127 Education 0.758** 0.313** 0.368** -0.139 0.667** Z-score DTC’s (type B) 0.712** 0.455** 0.041 -0.148 0.506** 0.696**

Note: * significant at p<0.05; ** significant at p<0.01

Table 7: Significance test for the cluster solutions

Number of Groups Ward’s Criteria Permutation test probability

(21)

Figure 1: Scree Plot for Determining the Number of Strategic Groups

(22)

Figure 2: Scatter Plot of the three strategic groups in the Dutch hospital industry

Table 8: Outcomes discriminant analysis

% of variance Z-score size Research Scope Debt-to-asset ratio

Function 1 99.7 0.930 0.146 0.141 -0.057

(23)

The discriminant analysis shows that the three strategic groups can be described by two functions. These functions differ remarkably in their contribution to the clustering, as the first function explains 99.7% of the variance and the second function only explains 0.3% of the variance. The first function is mostly correlated with the Z-score of the size and the second function is mostly correlated with the scope. Following the outcomes of this analysis, the size accounts for most of the variance between the strategic groups.

As previously discussed, the size, measured as the number of beds, was highly correlated with education, DTC’s (type B), and the number of employees, therefore these variables were dropped from the clustering analysis. However, to describe the three strategic groups in more detail, the distributions of these variables will be used to offer more elaborate strategic group descriptions. On top of that, it can be determined if the same patterns fit onto these other variables as they were highly correlated with the number of beds.

6.2. Interpreting strategic groups

Table 9 shows the averages of the strategy variables per strategic groups. Table 10 displays the hospitals that are member of each strategic group for the 3-group solution. These tables as well as the boxplots will be used to describe the strategic groups (see Figure 3).

Table 9: Descriptive statistics: averages of the strategy variables per strategic group

Variables Small Hospitals Mid-Sized Hospitals Large Hospitals

Size (Beds) 274 589 947 Scope 0.65 0.81 0.85 Research 0 0.14 0.31 Debt-to-asset ratio 0.81 0.76 0.73 Education* 0.06 0.76 1 DTC’s (type B)* 138441 261445 367597 Number of employees* 1571 3914 6952

(24)

Table 10: Strategic groups of the Dutch hospitals

Group 1: Small Hospitals Group 2: Mid-Sized Hospitals Group 3: Large Hospitals

Admiraal de Ruyterziekenhuis Albert Schweitzer Ziekenhuis AMCb

Alrijne Zorggroep Bravis Ziekenhuis Amphia Ziekenhuis

Antonius Ziekenhuis Canisius-Wilhelmina Ziekenhuis ETZ

BovenIJ Ziekenhuis Catharina Ziekenhuis Erasmus MC

Deventer Ziekenhuis Diakonessenhuis Utrecht Franciscus Gasthuis & Vlietland Elkerliek Ziekenhuis Gelre Ziekenhuizen Haaglanden MC

Flevoziekenhuis Hagaziekenhuis Isala

Groene Hart Ziekenhuis Jeroen Bosch Ziekenhuis Noordwest Ziekenhuisgroep

Havenziekenhuis Maasstad Ziekenhuis Rijnstate

IJsselland Ziekenhuis Maastricht UMC St. Antonius Ziekenhuis IJsselmeerziekenhuizena _{Martini Ziekenhuis} _UMCG

Ikazia Ziekenhuis Maxima MC UMCU

Langeland Ziekenhuis Meander MC Zuyderland

Laurentius Ziekenhuis Radboud UMC Maasziekenhuis Pantein Reinier de Graaf Groep MC Slotervaarta _{Spaarne Gasthuis}

MC Spijkenisse Tergooiziekenhuis

Ommelander Ziekenhuis Groep VieCurie MC

Rivas Zorggroep VUMCb

Rode Kruis Ziekenhuis ZGT

Saxenburgh Groep Ziekenhuis Gelderse Vallei SJG Slingeland Ziekenhuis St. Anna Zorggroep SKB Waterland Ziekenhuis Westfries Gasthuis Wilhelmina Ziekenhuis Zaans MC Ziekenhuis Amstelland Ziekenhuis Bernhoven Ziekenhuis Nij Smellinghe Ziekenhuis Rivierenland

Note: The hospitals that were not analysed due to missing data are: De Tjongerschans, Het Van Weel-Bethesdaziekenhuis, Isala Diaconessenhuis, LUMC, MC Leeuwarden, MS Twente, OLVG, Treant Zorggroep, Ziekenhuis St. Jansdal, ZorgSaam Ziekenhuis.

(25)

(26)

Figure 3 (continued)

Strategic group 1: Small Hospitals

The first strategic group is the largest as it consists of 33 hospitals. These hospitals are mostly general/regional hospitals (32). There is one top clinical hospital (Deventer Ziekenhuis) and no UMC’s in this strategic group. These hospitals are all smaller than the hospitals in other strategic groups, both in terms of the number of beds as the number of employees. The scope is highly scattered, as this group consists of both highly specialized as well as highly diversified hospitals, however the average scope is the lowest. Except for a few outliers, the number of type B DTC’s is also the lowest, so they perform the lowest number of type B treatments per year. Notably, some hospitals have a debt-to-asset ratio of above one, implying that they have negative equity on their balance sheet. This implies that for some hospitals in this strategic group it is hard to compete with other hospitals and therefore have to take on more loans to compensate on this matter.

Strategic group 2: Mid-Sized Hospitals

(27)

even though these hospitals are substantially smaller than the Large Hospitals, they are, on average, almost as diversified.

Strategic group 3: Large Hospitals

This strategic group consists of 13 hospitals. These hospitals are either UMC’s (4) or top clinical hospitals (9). These hospitals have the highest capacity, both in terms of number of beds and number of employees. The number of type B DTC’s differs highly in this strategic group, which implies that some hospitals focus on highly specialized, expensive treatments (type A DTC’s) to a relatively greater extent than other hospitals in this strategic group. Hospitals in this group have the lowest debt-to-asset ratio, meaning that their share of liability compared to equity is the lowest. This suggests that hospitals in this strategic group are better at coping with high competition and have to take on relatively less loans in comparison to smaller hospitals.

6.3. Strategic alliance analysis

After doing the cluster analysis and describing the strategic groups, the cooperation between hospitals can be analysed. A total of 304 unique cooperative links were found among the 67 hospitals. For the examination of cooperation patterns, a summarizing table was created to show the frequency of between and within strategic group cooperation (see Table 11). To analyse each type of alliance separately, a frequency table is shown in which the number of within and between strategic group alliances are shown per type of alliance (see Table 12).

Table 11: Frequency of alliances within and between strategic groups Alliance type Alliances within

(28)

Table 12: Alliance types and strategic groups

Alliance type Strategic group Small Hospitals

Mid-sized Hospitals

Large

Hospitals Total between

Integration alliances _{Small Hospitals} ₂₂ ₃₈ ₂₆ ₆₄

Mid-Sized

Hospitals 38 7 25 63

Large Hospitals 26 25 6 51

Total within 22 7 6

Information

exchange Small Hospitals 6 4 8 12

alliances Mid-Sized

Hospitals 4 3 7 11

Purchasing alliances Small Hospitals 5 12 0 12

Mid-Sized

Quality agreements Small Hospitals 8 22 14 36

Mid-Sized

Educational Small Hospitals 0 0 3 3

agreement Mid-Sized

Hospitals 0 4 5 5

Non-competing Small Hospitals 1 0 0 0

regions Mid-Sized

Hospitals 0 0 1 1

(29)

Overall, more alliances are formed between strategic groups than within strategic groups, but this differs between the strategic groups. Small Hospitals is the largest strategic group but it did not form the most alliances. The most alliances are formed by the Mid-Sized Hospitals.

When examining the cooperation patterns within and between strategic groups, one can note that the distribution of alliances formed between and within strategic groups differs per type of alliance. Integration alliances are formed between different players in the care chain to ease the movement of patients between care levels. A care chain starts at primary care and contains different levels, like secondary care (more specialized), tertiary care (highly specialized), and housing services. In this research, only the integration between two hospitals is considered. Most within-group alliances are formed by the Small Hospitals, but these hospitals did not form substantially more between-group alliances than the other strategic groups. However, overall, the findings are in line with the prediction that more integration alliances would be formed between strategic groups than within strategic groups.

Information exchange alliances are formed to exchange patient and medicine information between hospitals in the care chain. This alliance is thus formed by hospitals that are in geographical proximity of one another, therefore it was predicted that the same number of within-group alliances as between-group alliances are found. The findings, however, show that this alliance is formed more often between strategic groups. The Small Hospitals and the Large Hospitals engage in this alliance slightly more compared to the Mid-Sized hospitals.

Purchasing alliances are established to lower procurement costs by organizing purchasing on a larger scale. It was characterized as a pooling alliance; therefore, it was predicted to occur more within strategic groups. Slightly more within-group purchasing alliances are formed, and most within-group purchasing alliances are formed by the Mid-Sized Hospitals. However, the Small Hospitals also form this alliance with the Mid-Sized Hospitals, possibly to increase economies of scale by partnering up with larger hospitals in order to get favourable prices from suppliers. The Large Hospitals do not engage in this alliance in the same amount as hospitals in the other strategic groups, probably because their volume is high enough to achieve economies of scale alone and this alliance is not needed. As there are approximately the same number of purchasing alliances formed within strategic groups as between strategic groups, the findings do not support the predictions.

(30)

Non-competing regions alliances were only found twice in the dataset, once within and once between strategic groups, hence it is not possible to find support for the prediction that more non-competing region alliances occur within strategic groups. Also, the type of hospital that forms this alliance differs as the within-group alliance is formed by Small Hospitals and the between-group alliance is formed between a Large Hospital and Mid-Sized Hospital.

Finally, educational agreements are formed with the aim of educating all medical students in the Netherlands. It was classified as a complementary alliance and was therefore predicted to occur more often between strategic groups than within strategic groups. Slightly more between-group educational agreements were found. This type of alliance occurs mostly between the Mid-Sized Hospitals and the Large Hospitals. However, in these strategic groups, also within-group educational agreements occur. Although this alliance occurs more often between strategic groups in this particular dataset, the difference is so minor that the predication is not supported.

6.4. Classes analysis

A validity check was performed to determine whether the strategic groups found are useful tool for managers in Dutch hospitals to analyse their competitive landscape. In order to do so, the composition of the strategic groups was compared to the composition of the formal hospital classes already in place in the Netherlands. Then, alliance formation was analysed using the same predictions as for the strategic groups: within strategic groups or between strategic groups (see Table 13). However, the formal classes were analysed instead of the strategic groups. In the end, the missing hospitals were also included in the analysis in order to provide a complete overview of the Dutch hospital industry as well as perform a final validity check to test if the outcomes hold for the whole sample.

Table 13: Predicted alliance type formation: within or between classes.

Type of alliance Goal Prediction

1. Integration _{Form an integrated care chain} _Between

2. Information exchange

Exchange patient and medication information Within and between 3. Purchasing _{Centralize procurement for economies of scale} _Within

4. Quality _{Jointly improving the quality of care} _Between

5. Non-competing regions _{Exchange knowledge and resources} _Within

(31)

With regard to the composition of the strategic groups compared to the formal classes, some points stand out. First of all, the general/regional hospitals are almost all in the same strategic group, which confirms that these hospitals closely resemble each other in terms of the strategy variables in this research. However, five general/regional hospitals are member of the Mid-Sized Hospitals, as they are relatively larger and more diversified. Also, the top clinical hospitals and the UMC’s are not all in the same strategic groups, which is probably due to their difference in size as the number of beds was the variable that mostly explains the different strategic groups.

A table was constructed to display the number of alliances formed within classes and the number of alliances formed between classes per alliance type (see Table 14). More integration and educational agreements were formed between classes. Especially the prediction made for educational agreement is strongly confirmed, but this is logical as this alliance is formed between an UMC and a non-UMC. Quality agreements are formed as much within classes as between classes. This is not in line with the predictions that were supported by the investigation of strategic groups. Also, the two non-competing regions alliances were both formed between classes, which is not in line with the prediction that they would be formed more often within classes

Two outcomes that have not received support by the strategic group research have received support from the investigation of the classes. Information exchange alliances occur in the same amount within classes as between classes. Also, purchasing alliances are formed slightly more within classes. A table was constructed in order to give a visual representation of the strategic groups and classes outcomes (see Table 15).

Table 14: Frequency of alliances within and between classes Type of alliance Number of within classes

alliances

(32)

Table 15: Outcomes comparison of strategic groups and classes analyses

Type of alliance Prediction

Support for the predictions

Strategic group analysis Classes analysis

1. Integration Between Yes Yes

2. Information exchange Within and

between No Yes

3. Purchasing Within _No _Yes

4. Quality Between Yes No

5. Non-competing regions Within _No _No

6. Educational Between No Yes

Table 16: Frequency of alliances within and between classes with missing hospitals included Type of alliance Number of within classes

alliances

Number of between classes alliances Total 1. Integration 63 117 180 2. Information exchange 19 18 37 3. Purchasing 28 20 48 4. Quality 74 55 129 5. Non-competing regions 0 2 2 6. Educational 0 19 19 Total 184 231 415

(33)

There, quality agreements were formed approximately as often within as between classes. In the following section, a more in debt reflection on this outcome will be presented.

7. Discussion

This study has attempted to contribute to the empirical debate that has been going on in the strategic group research field. The supporters of the independent view state that strategic groups are only an analytical convenience and organizations within strategic groups are not bound to one another in any way (Hatten & Hatten, 1987). Contrarily, the interdependent view of strategic groups states that groups of firms pursue the same strategy within an industry and therefore compete for the same suppliers and buyers, hence firms within a group are considered interdependent from one another (Carroll & Thomas, 2019). The results of the permutation test in this study confirm that three strategic groups exist in the Dutch hospital industry that are more internally coherent than would be expected from random data, supporting the interdependent view. With regard to the composition of these three strategic groups, a few aspects stand out. Whereas some hospitals in the strategic group Small Hospitals have a low debt-to-asset ratio, other hospitals only in this strategic group have a debt-to-asset ratio of above one. This higher share of debts might be the result of the managed competition between hospitals in the Netherlands, introduced in 2006 (van de Ven & Schut, 2009). The two hospitals that recently became bankrupt are also both in this strategic group (Sondermeijer, 2018). These results together with the current news on the bankruptcy of these two institutions might hint that it is harder for smaller hospitals to financially deal with the fierce competition in the sector. It also suggests that there are performance differences not only between strategic groups, but also within strategic groups in the Dutch hospital industry. Further research is necessary to provide support for this potential finding.

Another notable outcome regarding the composition of the three strategic groups lies in the fact that five general/regional hospitals are member of a different strategic group than the rest of the general/regional hospitals. They are thus relatively large in size compared to the other general/regional hospitals in terms of the number of beds and employees. Hospitals are pressured to either focus on basic or complex care because of the selective contracting by insurers (Maarse et al., 2016). The fact that five general/regional hospitals are member of the Mid-Sized Hospitals might imply that they have the capacity to strive for delivering more complex or diversified care. Additional research could investigate whether making this transition would be beneficial to these hospitals.

(34)

UMC has the potential of growing large, as this is highly dependent on the population served in their geographical area. For instance, the University Medical Centre Groningen has the four Northern provinces as its response area. It receives more patients, can attain more employees and is therefore able to grow larger (Chiong Meza, van Steen, & de Jonge, 2014). In contrast, there were two UMC’s located in Amsterdam until 2018, the Vrije Universiteit Medical Centre (VUMC) and the Amsterdam Medical Centre (AMC). Since they shared a response area, the VUMC was not as large as other UMC’s. In order to increase their efficiency, as well as join their research projects, these two hospitals merged in 2018 (NOS, 2018). Not only the sizes of UMC’s differ, there are also substantial differences in other strategy variables. For example, the reported number of type B DTC’s and the scope of treatments was substantially lower for the UMCG and UMCU, implying that they focus on highly specialized, expensive treatments to a relatively greater extent than other UMC’s in their strategic group. Contrarily, other UMC’s, like the Radboud UMC, are highly diversified and treat more type B DTC’s. Thus, even though these hospitals are in the same class, the results of this research suggest that there are some strategic differences between them. Altogether, the results show that the composition of the strategic groups in this research differs on certain aspects from the formal hospital classes which are already in place in the Netherlands.

In order to study the cooperation patterns within and between strategic groups, the relational view and arguments provided by Nohria and Garcia-Pont (1991) were used. Organizations form alliances to gain access to those capabilities or resources of which they perceive it will lead to an improved competitive position. Theoretically, pooling alliances are formed within strategic groups and complementary alliances are formed between strategic groups. The outcomes of this research extend their study by suggesting that complementary alliances are formed more often in the Dutch hospital industry than pooling alliances. Although this was not predicted beforehand, it is an interesting result and can be linked back to the recent developments in the Dutch healthcare sector. Even though waiting times in Dutch hospitals have decreased over the last years, they are still substantial (Siciliani, Moran, & Borowitz, 2014). Hospitals are therefore forced to undertake measures to effectively reduce them. One way to reduce waiting times is by communicating clearly about the number of places available and referring patients to other hospitals that have shorter waiting times (Ritzen, 2017). In order to do so, larger hospitals need to ensure that patients receive the same treatment quality when referring them to smaller hospitals, which can be achieved by forming quality agreements.

(35)

Nevertheless, pooling alliances would also be expected considering the introduction of managed competition in 2006. Due to this development, numerous hospitals in the Netherlands have difficulties staying financially healthy (van Loghum, 2011). Especially general/regional hospitals find it challenging to compete, as all other hospitals are their competitors since they are all required to deliver basic care (Castellijns et al., 2011). Hence, efficiency becomes increasingly important. Smaller hospitals lack economies of scale on certain treatments and are therefore forming alliances to aid in this aspect (Blank, van Hulst & Wilschut, 2013). For instance, certain hospitals together pool their employees to simultaneously increase efficiency and improve the quality of care. Also, specialty agreements enable hospitals to split the market by together deciding to offer certain specializations only in a couple of hospitals. This enables smaller hospitals to drop treatments that are inefficient and have high costs. The few hospitals that continue to offer this treatment will receive more patients and attain more employees, offering the opportunity to increase efficiency. These two types of alliances fall under exempted agreements (Wiggers & Eggers, 2011), which were left out of this study as this category was very broad and contains many other subtypes. However, because these two types of alliances are relevant considering the current developments, future research could classify these alliances as either a pooling or a complementary alliance and look at cooperation patterns within or between strategic groups. Taking into account these alliances could also have implications for the total distribution of pooling and complementary alliances.

Two predictions with regard to alliance formation have been supported by the results. Integration alliances are formed to better coordinate services that are provided by different hospitals (Shaw et al., 2011). Integration includes primary, acute, and long-term care and also housing services, however, in this study only the integration between two hospitals is considered. In accordance with the predictions, this alliance is formed more often between strategic groups. The same result was found when analysing the cooperation pattern for this type of alliance between the formal hospital classes. The prediction concerning quality agreements has also received support since more quality agreements are formed between strategic groups than within strategic groups. These alliances are formed to jointly develop treatment protocols. Smaller hospitals can learn from larger hospitals, who in turn benefit from this alliance by the possibility of focussing more on complex cases (Ellerbeck et al., 2004). However, the outcomes from the classes analysis differ substantially from the strategic group analysis, especially including the missing hospitals.

(36)

are larger than other general/regional hospitals, it would make sense that they also form this alliance with other, smaller, hospitals within their class which are outside their strategic group. Also, the definition and function of a quality agreement in this research could be another reason for this deviating outcome. The rationale behind forming quality agreements might not necessarily lie in the size of the hospital, but more in the specialty of the hospital. Hospitals with a specific field of expertise are better in aiding other hospitals in developing a treatment protocol in their specialty. However, in order to gain more understanding of the cooperation patterns of quality agreements, further research is necessary.

Three predictions with regard to cooperation patterns within and between strategic groups have not received support from the outcomes. Although some researchers like Burns and Lee (2008) characterized a purchasing alliance as a pooling alliance, it does not occur more often within strategic groups than between strategic groups in the Dutch hospital industry. The results show that Small Hospitals also form this alliance with Mid-Sized Hospitals, probably with the aim to increase their bargaining power when they are negotiating with suppliers. Moreover, the outcomes demonstrate that the Large Hospitals do not engage in this alliance in the same amount as the other two strategic groups. This is in accordance with the aim of purchasing alliance, namely increasing economies of scale by centralizing procurement (Wiggers & Eggers, 2011). These hospitals might be large enough to already have a high bargaining power and benefit from substantial economies of scale (Wu, 2009). The outcomes of the classes analysis, however, does provide support for the prediction regarding this type of alliance.

The of the strategic group analysis also fail to support the prediction made regarding educational agreements, as they are not evidently formed more often between strategic groups. The plausible reason underlying this is that both the Mid-Sized Hospitals and the Large Hospitals contain both UMC’s and top clinical hospitals, which are educational hospitals. Therefore, educational agreements within these strategic groups are logical considering the composition of the strategic groups. In contrast, support was found by analysing the classes for this type of alliance.

(37)

of the partners in their annual reports and are therefore not analysed in this research, questioning the reliability of the results and posing a limitation on this study. To overcome missing data, future research should collect quantitative data directly from the hospital, for example by using surveys. Alternatively, interviews with managers could reveal alliances that are not mentioned in their annual reports.

Overall, the outcomes of the classes analysis and strategic groups analysis differ substantially. One prediction (regarding quality agreements) has received support from the strategic group analysis but did not receive support from the classes analysis. Contrarily, some predictions that have not been supported by the strategic group analysis have been supported by analysing the formal classes (regarding purchasing alliances, educational agreements and information exchange alliances). Including the missing hospitals in the classes analysis did not show a substantial change in these results, except for quality agreements. It should therefore be critically assessed whether a strategic group research is the appropriate method for analysing the Dutch hospital industry, because the results of the strategic group analysis differ so much from the classes analysis. Also, using different strategy variables (e.g. excluding the numbers of beds from the analysis) could give other clusters and thereby alliance analysis results that are more in line with the predictions.

8. Conclusion

This research was aimed at investigating the link between strategic group research and the relational view. The Dutch hospital industry was used as a research setting and three internally coherent strategic groups were found. On top of that, the six main types of alliances were defined and analysed on formation within or between strategic groups. The results indicate that strategic groups partly explain the cooperation patterns within the Dutch hospital industry. Integrational alliances, information exchange alliances, and quality agreements are mostly formed between strategic groups with the aim of gaining complementary resources and capabilities. The cooperation patters of purchasing alliances and educational agreement cannot be explained by strategic groups as they are formed as often within as between strategic groups. Additional research would potentially provide more insights into this matter.

8.1. Managerial implications

(38)

clarifying certain strategic choices and give the opportunity to compare strategies within and between strategic groups. For some hospitals, this research gives additional insights. Five general/regional hospitals are clustered into another strategic group than the remaining general/regional hospitals. Since these hospitals are larger and more diversified, they could evaluate whether they should pursue becoming a top clinical hospital. Additionally, this research studied the cooperation patterns of the main types of alliances formed by hospitals. The results of this analysis can aid managers in explaining the types of partner chosen for a specific alliance. Cooperation partners are mostly sought outside the strategic group for integration alliances, information exchange alliances and quality agreements.

8.2. Limitations and future research

(39)

(40)

9. References

Benschop, P. (2017, August 28). Ziekenhuizen Besparen 21 Miljoen Met Inkoopsamenwerking. Retrieved from https://www.skipr.nl/actueel/id31628-ziekenhuizen-besparen-21-miljoen-met-inkoopsamenwerking.html.

Blank, J. L., van Hulst, B., & Wilschut, J. (2013). Schaal-en synergie-effecten bij de spoedeisende hulp. Delft: Tu Delft.

Burns, L. R., & Lee, J. A. (2008). Hospital purchasing alliances: utilization, services, and performance. Health Care Management Review, 33(3), 203-215.

Carroll, C. and Thomas, H. (2019). Strategic groups and significant clustering. Unpublished manuscript. Castellijns, E., Kollenburgh, A. van & Oh, L. (2011). Second Opinion. Ziekenhuisstrategieën tegen het

licht. Utrecht: Berenschot B.V.

Castle, N. G. (2003). Strategic groups and outcomes in nursing facilities. Health Care Management Review, 28(3), 217-227.

Chatterjee, S. (1986). Types of synergy and economic value: The impact of acquisitions on merging and rival firms. Strategic Management Journal, 7(2), 119-139.

Chen, D. Q., Preston, D. S., & Xia, W. (2013). Enhancing hospital supply chain performance: A relational view and empirical test. Journal of Operations Management, 31(6), 391-408.

Cheng, S. L., & Chang, H. C. (2009). Performance implications of cognitive complexity: An empirical study of cognitive strategic groups in semiconductor industry. Journal of Business Research, 62(12), 1311-1320.

Chiong Meza, C. van Steen, J, & de Jonge, J. (2014). De Nederlandse universitair medische centra. Feiten en Cijfers 12. Den Haag: Rathenau Instituut.

Chung, S., Singh, H., & Lee, K. (2000). Complementarity, status similarity and social capital as drivers of alliance formation. Strategic management journal, 21(1), 1-22.

Strategic Groups and Cooperation in the Dutch Hospital Industry