The Importance of Alliances in the European Airline
Industry: A Strategic Groups Study
MSc Business Administration Strategic Innovation Management
Supervisor: dr. C. Carroll
Co-assessor: prof. dr. J.D.R. Oehmichen
D.W. van den Burg s2372126
Date: 21-01-2019
Abstract
Introduction
Almost 50 years ago, Hunt (1972) introduced the term “strategic groups” in a paper where he analyzed the white goods industry. A strategic group can be defined as “a grouping of organizations which pursue similar strategies with similar resources” (Hatten and Hatten, 1987. p. 329). Initially, this promising new field of study got attention from researchers because it could be used to analyze the competitive structure of industries (Barney and Hoskisson, 1990). Firms can be clustered together and this helps to describe the structure of the industry and it can further help to explain the competition (McGee and Thomas, 1986). Fiegenbaum et al. (1996) argued that analyzing strategic groups can help professionals understand the competitive structure of the market. Over time, the interest in the field decreased (Cattini et al., 2017), because researchers were not able to find a way to test the groups on their significance (McGee and Thomas, 1986; Barney and Hoskisson, 1990). A major criticism and a prime cause that many researchers abandoned the field was the inability of researchers to show the actual existence of strategic groups (Cattini et al, 2017). However, in a recent paper Carroll (2018) describes a new way to test the significance levels of strategic groups with the use of a multimethod approach. Carroll (2018) suggests that both a permutation test and a Monte Carlo test have to be performed to test the significance of the strategic group clusters. Both tests have their weaknesses, although these weaknesses do not overlap. The tests can actually complement each other well and this may be the solution for testing the significance level of strategic groups that was awaited for so long.
contrast to the aforementioned article by Carroll, this research paper will work with factor scores as variables, which allows for more variables to be included and thus a more complete analysis of the industry and the business models used.
This research will focus on the European airline industry, this is a highly competitive industry where most firms struggle to be profitable (Daft and Albers, 2015). To cope with competition, many airlines are part of a global alliance (Castiglioni et al., 2018). A global alliance is defined as “a set of firms, linked through alliances that compete in a specific business domain” (Gomes-Casseres, 2003 p. 1). The three big alliances within the airline industry (Star, SkyTeam, and Oneworld) will be analyzed and their importance within and between strategic groups will be examined. The airline industry is particularly interesting because it was also the subject of earlier strategic groups research (e.g. Kling and Smith 1995; Cappel et al., 2003; Castiglioni et al., 2018), and now with this recently introduced approach these results can be put to the test.
The results of this study will be interesting for managers within the airline industry. They learn whether interdependent strategic groups exist within the industry and thus which firms are the most important ones when it comes to both competition and cooperation. Furthermore, they can gain insight about performance differences that occur among the strategic groups as well as performance differences among the different alliances, and airlines that are operating with or without an alliance. This enables managers to make potentially better decisions towards cooperation in the future. Additionally, this research is also very relevant for scholars focusing on strategic groups. Specifically in the interdependent view, since the recently introduced multimethod approach will be tested further, this time extended by using Wilks’ lambda as a test statistic. Additionally, this exploratory study merged the strategic variables into factor scores. This has not been done before in similar multimethod studies and this will lead to interesting insights about the advantages and possibly disadvantages of using these criteria to assess clustering.
tested. Afterwards the results will be presented and their implications will be discussed. Lastly, limitations of the study and directions of future research will be provided and concluding remarks will be given.
Literature review
Strategic Groups
To understand competition strategies between firms, strategic group research is an important field of study (Fiegenbaum and Thomas, 1990). After Hunt (1972) introduced the concept, defining strategic groups as groups of firms that practice a homogeneous strategy within the group and a heterogeneous strategy compared to other groups, the field emerged and other scholars developed the concept further. Porter (1979) and Caves and Porter (1977) combined structural and strategic variables to extend Hunt’s (1972) findings. Strategic groups were later defined by Porter (1980. p. 129) as “the group of firms in an industry following the same or similar strategy along strategic dimensions”. Caves and Porter (1977. p. 251) argued that “because of their structural similarity, group members are likely to respond in the same way to disturbances from inside or outside the group, recognizing their interdependence closely and anticipating their reactions to one another’s moves quite accurately”.
The field of strategic groups was further expanded by Caves and Porter (1977) who introduced mobility barriers, this concept describes the barriers firms face when moving within and between strategic groups. High mobility barriers impact the successfulness of potential change, this could discourage groups to attempt change (Hatten and Hatten, 1987). Not only does profitability impact these strategic decisions, but firms can also face other problems created by barriers, for example significant delays and uncertainty about outcomes (McGee and Thomas, 1986). Every firm that moves between groups has to face these challenges (Carroll, 2018). However, Carroll (2018) also states that mobility barriers do not have to exist, nor do they guarantee performance differences (Cool and Schendel, 1988). Hatten and Hatten (1987) furthermore stated that barriers can be asymmetric for different firms, that larger and smaller firms can face different challenges, and that mobility barriers can vary for entry and exit.
Hatten and Hatten (1987) also came to the conclusion that strategic groups do not have to be composed out of competing companies. Hatten and Hatten (1987) are the main contributors to the independent view within the strategic groups literature, this view is based upon the idea that firms can be segmented into different groups that make it easier to analyze an industry. This can be convenient when you want to analyze what firms would do in a certain situation. Instead of analyzing what every firm would do, you can simply compare how different types of firms would behave (Carroll and Thomas, 2019). Hatten and Hatten (1987) reject the idea that firms within a group are aware of their interdependence and describe the grouping of firms as merely an invention of scholars and not a real feature within industries. Carroll and Thomas (2019. p. 4) describe this independent view as “firms within each cluster (that) are presumably not bound to each other in any meaningful way”.
In sharp contrast to the independent view stands the interdependent view. Tang and Thomas (1992) define strategic groups as subsets of firms that are interdependent and interacting with each other. These subsets reflect the social structure of rivalry within an industry. Firms that are interdependent are aware of this fact most of the time (Porter 1979). In fact, Carroll and Thomas (2019. p. 4) argued that “the conduct within strategic groups could create pockets of oligopolistic competition within the industry”. These interactions between firms could lead to true group effects. True group effects are the differences in performance that result from dynamics within the group (Carroll and van Heyningen, 2018). The interdependent view is the only view that is in critical need of a significance test (Carroll and Thomas, 2019), because without one it is never clear whether the groups really exist or are just created as an analytical convenience (Hatten and Hatten, 1987).
Development in tests
undermines the validity of the whole field and that the field should be abandoned if there was no sign of a solution to solve the problem (Barney and Hoskisson, 1990). In a recent paper, Cattani et al. (2017) disagree with this statement. They still argue that the field of research can be beneficial to professionals. However, they do agree that the validity of the findings of previous studies is uncertain. Cattani et al. (2017) conclude that the interest in the field has decreased over the last fifteen years and that therefore many issues within the field remain unanswered, so that the impact of group membership and its profitability were not clearly established.
Recently, Carroll (2018) introduced a significance test that emerged from the field of ecology and biology and this can be a possible solution for researchers to test for significance of the clusters obtained in a strategic groups analysis. This newly introduced test examines the hierarchical cluster analysis with both a permutation test and a Monte Carlo test (a more in-depth explanation is given later in this article). This significance test could support the interdependent view of strategic groups. This can help the field of study to find if strategic groups reflect the structure of competition and cooperation in an industry and, most importantly, if strategic groups exist at all or if they are just an analytical convenience.
Alliances in the airline industry
programs that result in passenger loyalty for the airlines (Castiglioni et al., 2018; Min and Joo, 2016). Entering an alliance also reduces or completely eliminates competition with partner airlines (Kleymann and Seristö, 2001). Additionally, alliances also create benefits for customers such as a more flexible schedule, increased flight frequencies, better luggage handling (if switching airlines is necessary) and the shared frequent flyer programs (Wang, 2014).
This study will focus on the three biggest global alliances within the airline industry, namely Star Alliance, SkyTeam and, Oneworld. Most global alliances in other industries are temporary and linked to a specific project, however in the airline industry alliances are long-term and more general (Gudmundsson et al., 2012). Because of the formation of alliances the competition between airlines changes to competition between groups of airlines and this has caused continuous growth in the number of airlines that are members of a strategic alliance (Gomes-Casseres, 1994).
There are certain consequences for airlines that join a global strategic alliance that need to be taken into account. First, participation in an alliance can change the role of an airline. This can be necessary to be a valuable addition for the alliance (Castiglioni et al., 2018). Some airlines need to be specialists in certain markets and others have to position themselves as link providers, so that different markets can be connected (Kleymann, 2005). Secondly, the increasing alliance size will be accompanied by diminishing additional value with each new airline because of increased overlap in partners’ networks and city duplication (Agusdinata and Klein, 2002). When this happens it can cause undesirable internal competition in the alliance. Furthermore, being part of an alliance limits an airline to join or form other alliances, because currently airlines are contractually limited to only one alliance (Corbo and Shi, 2015).
Next to the above mentioned benefits and consequences of joining a strategic alliance, there are also downsides for the airlines. For example, passengers do not seem to recognize the benefits of airline alliances such as seemingly carefree travel, extra lounge access, and transferable priority status, and therefore this does not give the allied airlines a competitive advantage over rivals (Goh and Uncles, 2003). Furthermore, joining an alliance can limit the actions of the airlines. Castiglioni et al. (2018) argue that this has prevented big airlines such as Virgin Atlantic, Emirates and Etihad from joining an alliance up to now.
that managing the alliance portfolio has a direct effect on a firm’s financial performance (Hoffmann, 2005). However, the study of Min and Joo (2016) found no significant differences in performance between airlines that are part of an alliance and airlines that are not. Moreover, they also did not find an improvement in performance before and after an airline joined an alliance. These findings are in line with the observation by Porter (1990) that indicates that strategic alliances do not guarantee success. Because of these conflicting findings, this exploratory study will test for performance differences between alliances as well as performance differences between allied and non-allied airlines.
Star Alliance, SkyTeam, and Oneworld are the biggest alliances in the airline industry. These three alliances in total offer approximately 9136 routes, which represents 36% of the total routes in the global industry (Lordan et al., 2015). These routes represent two-thirds of the total industry capacity (Gaggero and Bartolini, 2012), and have a passenger volume of almost 60% of the entire industry (Wang, 2014). Table 1 shows the airlines included in this study and to which alliance they belong. A brief overview of the three alliances mentioned above follows.
Table 1. Airlines and their alliance.
Star Alliance: This alliance was launched in 1997. Nowadays this alliance has 28 members of which five are included in this study. A study by Lordan et al. (2015) found that out of the three big alliances the Star Alliance has the most robust airline alliance route network, which means that this alliance is less vulnerable to member exits than the other alliances. By passengers, the Star alliance is the second biggest alliance with 728 million passengers per year (Star Alliance, 2018), only slightly behind Skyteam. Min and Joo (2016) discovered that
Alliance
Star Alliance Skyteam Oneworld None
Aegean Airlines Air Europa Air Berlin Aer Lingus Austrian Airlines Air France British Airways Condor
Lufthansa Alitalia Finnair Easyjet
TAP Portugal KLM Iberia Flybe
Star Alliance tended to be outperformed by the smaller alliances Oneworld and Skyteam, because the time consuming process of assimilation was unbeneficial for the big alliance.
Skyteam: Formed in 2000, this alliance was the last alliance of the three big airline alliances that was established. Skyteam has 20 members of which 4 are included in this study. With over 730 million passengers annually, this alliance is the biggest of the three in numbers of passengers (Skyteam, 2018). Behind Star Alliance, Skyteam has the second most robust route network (Lordan et al., 2015).
Oneworld: This alliance was formed in 1999 and is the smallest alliance in terms of annual passengers (527.9 million) and alliance members (14) (Oneworld, 2018). Four members are included in this study. Out of the three global alliances in the airline industry, Oneworld is the alliance with the most vulnerable network (Klophaus and Lordan, 2018; Lordan et al., 2015). Also this alliance is, because of its smaller size, more vulnerable to member exit compared to the other two alliances (Klophaus and Lordan, 2018). However, these authors also state that Oneworld has a more balanced membership compared to the other two alliances. This means that there are fewer dominant airlines that have many routes that no other member within the alliance has.
Strategic groups within the airline industry
As mentioned in the introduction, the decision to choose the airline industry for this study is based on several advantages that come along with this industry. The industry is characterized by many rivalries and well-known competitors, and the strategic groups are robust to be used in later studies (Smith et al., 1997; Peteraf 1993). There are already multiple studies performed in the airline industry to identify strategic groups. Many different approaches and methods have been used. Early studies relied on inspection (e.g. Peteraf 1993). Later, methods such as factor analysis and cluster analysis were used (Murthi et al., 2013). In most of the recent studies, cluster analysis has been the most popular analysis. This trend is in line with the rest of the strategic group field. Although every study tried to address some of the shortcomings of the analysis, no study so far has been able to come up with a solution that fixes all the limitations (Murthi et al., 2013). Thomas and Venkatraman (1988) state that diversity in methodology can be an advantage for a field, however, a downside is that it becomes more difficult to compare the results of the different studies with each other.
The article by Murthi et al. (2013) gave an overview of the most important studies concerning strategic groups within the airline industry. As one of the first, Peteraf (1993) recognized important differences between airlines and separated the industry in two strategic groups, namely formerly regulated airlines and newer airlines. Kling and Smith (1995) used a scatterplot that included the dimensions of airline cost and the quality. They identified four different strategic groups: Cost leadership, quality differentiation, focus, and stuck in the middle. Smith et al. (1997) found three strategic groups while using cluster analysis, specifically: high end, entrenched dominance, and niche seekers.
Murthi et al. (2013) used the Latent Class Regression model to identify strategic groups within the airline industry, and found four strategic groups: majors, regionals, newbies, and Southwest (a strategic group that only contains this airline). Although they try to fix the problems that cluster analysis brings, the limitation of the Latent Class Regression approach is that it can only use a single performance measure. Furthermore, because Latent Class Regression can only use panel data, cluster analysis is still the best option when there are non-repeated observations (Murthi et al., 2013).
significance test that supports the actual existence of the groups. Similar to the present study, Castiglioni et al. (2018) also paid attention to alliances within the airline industry. However, they took another approach and mainly linked the alliances to an increasing level of ‘virtualization’ of the airline industry. Other contributions to the airline industry of this article, that are relevant for this study, have been mentioned in the section above about alliances within the airline industry.
Lastly, the recent article by Carroll (2018) will be discussed. The analysis in this article is also done with cluster analysis and for significance testing the author also used a permutation test and a Monte Carlo test. These tests are complementary to each other, which means that the weaknesses of one test can be solved by another test (Carroll, 2018). This multimethod approach will also be applied in this article, however the variables are constructed differently. Furthermore, the study by Carroll used the same 2012 dataset by Daft and Albers (2015), however, the composition of airlines included is slightly different. Carroll found a two strategic groups solution, namely the low cost airlines and the full service airlines and a non-significant four group solution where the low cost airlines are separated with a group for the pure low cost carrier (LCC) and the full service carrier (FSC) are separated between larger and smaller airlines. No significant performance differences were found between both strategic groups in the significant two group solution, indicating that there are no firm level effects in the sample (Carroll, 2018).
Methods
Strategic variables
This research only works with secondary data and makes use of the dataset provided by Daft and Albers (2015). This dataset, collected in 2012, consists of 26 European airlines that are examined on 36 different variables. Because a factor analysis can only be performed successfully if the number of variables is lower than the number of observations (Arrindell and van der Ende, 1985), the most relevant variables had to be selected. The selected variables will be merged into factor scores, which will be presented in the results section.
Performance variables
Some data was not available here, but could be retrieved from the annual reports of the airlines. Nevertheless, because this was not possible for all the airlines there is some missing data: only 23 of the original 26 airlines that were analyzed in the Daft and Albers (2015) dataset could be used for the analysis. The profit margin for Germanwings was not available in the aforementioned sources, and this airline is not active anymore at this point in time which limits the possibilities to obtain this information in a different way. The ROE could not be calculated for Swiss Air, therefore this airline was also not included. Furthermore, the data of AS Scandinavian Airlines could not be included, because their 2012 financial data only included the period January - October. It is worth noting that there are more airlines whose performance measurements were calculated in October, instead of the end of the year, but since these calculations were based on a 12 month period, these airlines are still included in the analyses. Because of the missing data, the final dataset contains 23 airlines.
Analyses
Factor analysis. When the measures loading on a particular factor have very different scales, all variables should be standardized before being summed into factor scores (Floyd and Widaman, 1995). Therefore, the independent variables were all transformed into z-scores, this way all variables contributed equally in the analyses.
Thereafter, a factor analysis was performed for data reduction using the principal factor analysis. This is the best analysis method to use in most cases with simple patterns (De Winter and Dodou, 2012). The Direct Oblimin was used as rotation method, because this is the method that assumes that the variables are correlated (Corner, 2009). The six factor scores that were created from this analysis are entered into the cluster analysis. Using these scores has multiple advantages over the original variables. Floyd and Widaman (1995. p. 294) explain that the scores are more reliable and “the use of fewer variables means fewer chances for shared measurement variances that cannot be sufficiently stipulated by theory”.
Wilks’ lambda (close to 0) means that the groups are well separated”. There are previous studies on strategic groups that include Wilks’ lambda (e.g. Tokman et al., 2016; Ferguson et al., 2000). Tokman et al. (2016) state that Wilks’ lambda is an especially good indicator in studies with smaller sample sizes, which is the case for this study.
Two tests for significant clustering were used, a permutation test and a Monte Carlo test, both using 999 iterations. In the permutation test, distribution by randomly shuffling the data generates a null distribution. The variables are all permuted independently to destroy any clustering in order to generate the null distribution for the cluster analysis. The Monte Carlo creates data that mimics the original data while ensuring that observations are not clustered. Although both tests have their weaknesses, they do not have identical flaws and form a strong method when used in combination (Carroll, 2018). However, the complementarity goes deeper than non- overlapping weaknesses. The Monte Carlo test can be regarded as a validity check for the permutation test and vice versa, combining the tests explicitly rules out the key threats to their individual validity (Carroll, 2018).
The permutation test can generate significant results due to correlations among variables in absence of clustering (Carroll, 2018). The Monte Carlo test generates a null distribution of the clustering statistic in this scenario and when this test comes out significant, it is unlikely that the result was generated by correlations without clustering (Carroll, 2018). In addition to the significance tests, this study makes use of a scree plot to find the most appropriate cluster solution. An ‘elbow’ in a scree plot is an indication of a potentially interesting solution (Carroll, 2018).
Results and Discussion
Variables and factor scores
The correlations table of the selected variables can be found in Table 2. The variables are merged into factor scores. An overview of the six different factors can be found in Table 3. The factor scores with an eigenvalue higher than 1 were selected for this analysis. In Table 3 it can be observed that the first four factors are the most important in this analysis, the fifth and sixth factors only explain a small percentage of the variance.
Cluster significance tests
The results of the significance tests on both the Ward’s method (total within group variance) and the Wilks’ lambda method are shown in Table 4. First, looking at the permutation test of the Ward’s method, it can be seen that cluster solutions 7 to 11 provide significant results. However, the corresponding Monte Carlo test shows no significant results for any of the cluster solutions. This can be caused by the fact that the simulated data copies the correlation matrix of the original data, which are the factor scores in this study. The simulated data replicates both the factor scores and the correlations between these factors scores, this can cause the original results to not stand out.
Table 3. Factor scores analysis
Factors
I II III IV V VI
Variables Service level Operations Scope Wage
Ownership
& IFE Relations
Route design 0,87 -0,15 0,13 0,02 0,08 0,59 Lounge access 0,86 -0,25 0,07 0,32 -0,04 0,55 Lobbying in associations 0,81 -0,12 -0,07 0,07 0,09 0,20 Sales promotion 0,76 0,09 -0,02 0,18 0,18 0,70 Bundling concept 0,75 -0,20 -0,37 0,07 0,27 0,63 Seat pitch 0,58 -0,11 -0,25 0,33 0,36 0,42 Labor intensity 0,52 -0,34 0,51 0,13 -0,13 0,32 Aircraft financing -0,12 0,86 0,06 0,01 0,28 -0,12
Flight crew skills 0,32 -0,74 0,25 0,30 -0,02 0,30
Routes offered -0,51 -0,57 -0,21 0,21 -0,15 -0,30 Owning facilities 0,55 -0,54 -0,15 -0,16 -0,11 0,53 National scope -0,04 0,09 0,82 0,37 0,19 -0,17 Aircraft utilization -0,01 -0,12 -0,59 0,31 -0,09 -0,04 Spatial scope 0,30 0,12 -0,51 -0,01 0,28 0,47 Wage policy -0,08 0,10 -0,02 -0,76 0,06 0,08 In-flight entertainment 0,21 0,26 -0,19 -0,24 0,84 0,04 Executive ownership 0,00 0,18 0,21 0,05 0,81 0,15 Cooperation policy 0,62 -0,35 0,06 -0,17 0,06 0,88 Target passenger 0,54 -0,19 -0,29 0,08 0,03 0,83
Access to primary airports 0,23 -0,04 0,04 -0,18 0,08 0,77
Brand presentation 0,37 -0,35 -0,13 0,20 0,12 0,61
Fleet modernity 0,30 0,12 -0,31 -0,20 -0,34 0,51
Scatter plot
Now the canonical discriminant functions will be analyzed. This scatter plot gives a visual representation of the clusters and the distances between them. The scatter plot in Figure 3 shows that some clusters are well separated from the others, which would indicate that these groups do not react to the actions of each other (Carroll, 2018). However, some groups are clustered closely together in the middle and this could be caused by “a convergence trend in airline business models to the mainstream middle” (Daft and Albers, 2015. p. 3). Although this trend might exists there was still, as concluded in the paragraph above, a significant seven-cluster solution. Therefore important differences must also occur between groups that are grouped more closely together. Furthermore, the scatter plot only displays the differences on factor one and two, that are explaining two-thirds of the variance in the sample, differences on other factors are not shown in this figure. In Table 5 the means on the factors scores are shown. Here it can also be seen where the biggest differences between strategic groups occur. The first four factor scores are the most important because they explain the majority of the variance.
Strategic groups
Figure 3. Canonical Discriminant Functions for the seven cluster solution Table 5. Cluster means on factor scores for each strategic group
Strategic group 1: Major global carriers Part of this strategic group are: Air France, Austrian Airlines, British Airways, Lufthansa, Iberia and KLM. These airlines are (former) flag carriers with a global hub-and-spoke network. The airlines in this group are in both a joint venture and an alliance. Moreover, all the airlines are involved with another airline in this group. Air France and KLM (both Skyteam) merged in 2004, Lufthansa group bought Austrian Airlines (both Star Alliance) in 2009 and British Airways and Iberia (both Oneworld) merged in 2010. This group scores above average on the service factor, operations factor, and scope factor. This is also what can be expected of these major global carriers (Castiglioni et al., 2018).
Strategic group 2: Regional (former) flag carriers This strategic group contains: Aer Lingus, Air Berlin, Finnair, TAP Portugal and Turkish airlines, which all have a hub-and-spoke network. With the exception of Air Berlin, all the airlines are (former) flag carriers. In this group, only Aer Lingus is without an alliance, this is also the airline out of this group that scores the lowest on most of the full service carrier (FSC) characteristic variables. Following the ideas of Reger and Huff (1993), it could be argued that Aer Lingus is a secondary group member. This is a firm that implements the characteristics of the strategic group less steadily than the other firms in the group. The average service level of this group is the highest of all the strategic groups, the airlines within the group can therefore also be described as FSC.
Strategic group 3: Smaller FSC The airlines in the first strategic group (Aegean Airlines, Air Europe, Alitalia, Virgin Atlantic Airways) are smaller full service carriers. This group is characterized by the highest percentage of executive ownership, the highest level of in-flight entertainment (IFE) and the least amount of routes offered. Furthermore, this group has the characteristics that are expected of a full service carrier, scoring above average on most of the factor 1 (service level) variables. Three of the four airlines are part of one of the three big alliances. The exception is Virgin Atlantic Airways, however, this airline is connected with other airlines with both a joint venture and a codeshare agreement.
airports. Furthermore they stand out with, by far, the most routes offered of any strategic group. The characteristics that define this group are in line with the characteristics that were found by Agusdinata (2002) and Castiglioni et al. (2018).
Strategic group 5: LCC (modified) The last strategic group consists of two airlines: Norwegian Air Shuttle and Transavia. Both of these airlines follow a low cost strategy and are not part of an alliance. This group stands out on the wage factor: they have the highest wage expenses out of all the strategic groups, which can be caused by the locations of the airlines as both are located in high-wage countries. This group differentiates slightly from the pure low cost carriers, with in-flight entertainment and also slightly better other services. This way they do not have to compete solely on prize with pure low cost carriers. Furthermore the route design is different from the pure low cost carriers, as both airlines in this strategic group offer transfers.
Strategic group 6: Moderate low cost (charter) This strategic group (Condor, Monarch Airlines and Vueling Airlines) is the group that has the most average scores of all the groups. None of the airlines are part of an alliance, and only Vueling has a codeshare agreement. Both Condor and Monarch have their focus on leisure flights and are (former) charter airlines, Vueling has a high percentage of national flights. This group has moderate levels of service. These airlines are clustered together in the same group because they each do not belong to the groups that follow a FSC strategy, nor do they follow a pure low cost strategy. The group is different from other groups on operations level, having the highest percentage of leased aircrafts and low scores on routes offered, owning facilities and flight crew skills.
Table 6. Strategic groups composition and the alliances
Group Airlines Star Alliance Skyteam Oneworld None
1. Major global carriers Austrian x
Lufthansa x
Air France x
KLM x
British x
Iberia x
2. Regional flag carriers TAP Portugal x
Turkish x Air Berlin x Finnair x Aer Lingus x 3. Smaller FSC Aegean x Air Europa x Alitalia x Virgin x 4. Pure LCC Ryanair x Easyjet x 5. LCC (modified) Norwegian x Transavia x
6. Moderate low cost Condor x
Monarch x
Vueling x
7. Local carrier Flybe x
Performance differences
MANOVA was performed on the seven-cluster solution, where profit margin and ROE serve as dependent variables. The results of this MANOVA did not show that any significant performance differences occur between the strategic groups. Looking closer at the boxplots, shown in Figure 4, it can be seen that there are differences between the strategic groups.
The best performing group on both dependent variables is the pure LCC group (group 4). A reason for this could be that the airlines in this group are offering the lowest prices, this makes the airlines less vulnerable within the economic cycle (Agusdinata et al., 2002). That these pure low cost airlines go a long way to cut costs can be illustrated by a statement made by the CEO of Ryanair. Michael O’Leary famously said that he would rather cut off his own hand than sign a deal with unions (Irish Independent, 2018). Customers will be more price sensitive during economic downturns and in 2012 the 2007-2009 recession still had its impact on the airline industry, negatively influencing others more than the pure low-cost airlines. Carroll (2018) suggested that a pure low cost strategy could lead to firm-level advantage over airlines that differentiate slightly from that low-cost formula, because they can offer lower prices and capture more market share.
Figure 4. Boxplots of the performance differences between the strategic groups
Figure 5. Boxplots of the performance differences between the alliances
Secondly another MANOVA was performed to test if there were performance differences between the allied airlines and the non-allied airlines. To be able to test if there were performance differences all the airlines that are part of an alliance were grouped together. Also this MANOVA did not bring up performance differences between the two groups. Additionally, analyzing the boxplots, shown in Figure 6, did not bring up any big differences between the two groups.
Figure 6. Boxplots of the performance differences between the allied and non-allied group of airlines
operating in with a point-to-point network are non-allied. It is expected that airlines that operate within a hub-and-spoke network need an alliance to operate otherwise there is a chance that the performance will drop significantly, because then they will not be able to fly from anywhere to anywhere in the world and they need an alliance to be able to fly the many routes with lower traffic.
An airline using a point-to-point strategy can however function independently because of the route model which relies on a “cherry picking” strategy whereby they only choose the most profitable routes. The independent airlines are designed in a way that they do not need an alliance to improve the performance. This is in line with the conclusion made by Castiglioni et al. (2018 p. 144.) that “the option to remain independent requires an operational strategy consistent with that choice and the resources available to the airline”.
Implications
This study found seven different strategic groups within the European airline industry using a multimethod approach. A major implication of this approach is that the seven-cluster solution has identified clusters that all have unique features and the different groups make sense within the industry, which supports the face validity of the clusters (Carroll, 2018).
The results of this study differ from the results of Carroll (2018) who only found a significant two-cluster solution within the same dataset. However there are many differences between both studies that can have caused this difference. First, this study uses different and more strategic variables found in Daft and Albers’ (2015) dataset, providing a more complete picture of the business model configurations. Second, this study uses the output of the factor analysis, the factor scores, as the strategic variables used in the analysis. These ‘summarizing’ scores all represent a different aspect of the airlines. Third, this study extended the method used by Carroll (2018) by also including the Wilks’ lambda in addition to the Ward’s method.
however, this seven-cluster solution divided the groups further on differences in business models.
Another major implication of this study is that there are no performance differences between allied and non-allied airlines, nor are there performance differences between the different alliances. This is a very important finding for airline executives. It means that joining a strategic alliance alone does not necessarily improve the financial performance of an airline. This finding is in line with the findings of Min and Joo (2016), who also found that joining an alliance does not guarantee financial success for an airline. An important practical guideline is that the alliance should be used to improve route networks, frequent flyer programs, and compatibility between the services of partners.
Another implication is that factor scores can successfully be used in strategic group research, and especially the combination with the Wilks’ lambda test statistic can be very effective. This study expanded the multimethod approach introduced by Carroll (2018) by adding the Wilks’ lambda to test if the clusters are well separated. Using factor scores gave a more complete view of the industry and this led to very clearly distinguished strategic groups.
Limitations and future research
Then there are also some limitations of this study that create interesting issues for other studies to explore. The histograms of the output of the permutation test show that using factor scores puts some of the clustering on a single factor. This issue did not fit the scope of this study and this is therefore an interesting matter to explore further in methodological studies. The last limitation and also a suggestion for future research is the influence of factor scores on the Monte Carlo test. Although the non-significant findings in the Monte Carlo test are an expected limitation of the use of factor scores, the obtained solution should still be interpreted with slight cautiousness. Analyzing this issue and finding a possible solution for this can be an interesting topic for future studies.
Conclusion
Strategic group research has lost momentum over the last decades. However, with a more reliable way to analyze the significance of the identified clusters, the field can have great potential to analyze the competition and cooperation structures of an industry. This exploratory study used the recently introduced multimethod approach to significantly test whether there are interdependent strategic groups within the airline industry. An extra test statistic was added, namely the Wilks’ lambda. The positive results on the Wilks’ lambda created even more confidence in the found cluster solution. The results of the analysis show that there is a seven-cluster solution that is significant on the permutation test of the Ward’s method and both the permutation test and the Monte Carlo test of the Wilks’ lambda. That this approach had mostly successful results can help the field of strategic groups move forward to a new approach to test the significance of the cluster solutions.
This study wanted to get an in-depth representation of the different business models used within the airline industry. Therefore factor scores were used as input for the cluster analysis. The use of factor scores offers great potential for future strategic group research using the multimethod approach because this study shows that these scores can be very effective in mapping out an industry since more variables can successfully be included in the analysis. By using factor scores, this study was able to provide a more detailed representation of the different business models used within the industry, compared to earlier strategic group studies on the airline industry.
References
Agusdinata, B., & De Klein, W. (2002). The dynamics of airline alliances. Journal of Air Transport Management, 8(4), 201-211.
Arrindell, W. A., & Van der Ende, J. (1985). An empirical test of the utility of the observations-to-variables ratio in factor and components analysis. Applied Psychological Measurement, 9(2), 165-178.
Barney, J. B., & Hoskisson, R. E. (1990). Strategic groups: Untested assertions and research proposals. Managerial and Decision Economics, 11(3), 187-198.
Cappel, S. D., Pearson, T. R., & Romero, E. J. (2003). Strategic group performance in the commercial airline industry. Journal of Management Research (09725814), 3(2).
Carroll, C. (2018). Significant Clustering: Implications for Strategic Groups Research. Australian Academy of Business and Economics Review, 4(2), 60-73.
Carroll, C., & van Heyningen, K. (2017). Assessing the Disprof Test for Significant Clustering: Is It Suitable for Strategic Groups Research?. Australian Academy of Business and Economics Review, 4(1), 29-42.
Carroll, C. & Thomas, H. (2019). Strategic groups and significant clustering. Unpublished manuscript.
Castiglioni, M., Gallego, Á., & Galán, J. L. (2018). The virtualization of the airline industry: A strategic process. Journal of Air Transport Management, 67, 134-145.
Cattani, G., Porac, J. F., & Thomas, H. (2017). Categories and competition. Strategic Management Journal, 38(1), 64-92.
Caves, R. E., & 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. & Schendel, D. (1988). Performance differences among strategic group members. Strategic Management Journal, 9(3), 207-223.
Corner, S. (2009). Choosing the right type of rotation in PCA and EFA. JALT testing & evaluation SIG newsletter, 13(3), 20-25.
Czipura, C., & Jolly, D. R. (2007). Global airline alliances: sparking profitability for a troubled industry. Journal of business Strategy, 28(2), 57-64.
Daft, J., & Albers, S. (2015). An empirical analysis of airline business model convergence. Journal of Air Transport Management, 46, 3-11.
De Winter, J. C., & Dodou, D. (2012). Factor recovery by principal axis factoring and maximum likelihood factor analysis as a function of factor pattern and sample size. Journal of Applied Statistics, 39(4), 695-710.
Devlin, M. (2018, July 28). The sky won’t fall in for O’Leary and Ryanair if they just learn to work with their employees. Irish Independent. Retrieved from www.pressreader.com/ireland/irish-independent/20180728/282020443094545
Doz, Y. L., Olk, P. M., & Ring, P. S. (2000). Formation processes of R&D consortia: Which path to take? Where does it lead?. Strategic Management Journal, 239-266.
Dyer, J. H., Kale, P., & Singh, H. (2001). How to make strategic alliances work. MIT Sloan management review, 42(4), 37.
Ferguson, T. D., Deephouse, D. L., & Ferguson, W. L. (2000). Do strategic groups differ in reputation?. Strategic Management Journal, 21(12), 1195-1214.
Fiegenbaum, A., Hart, S., & Schendel, D. (1996). Strategic reference point theory. Strategic Management Journal, 219-235.
Fiegenbaum, A., & Thomas, H. (1990). Strategic groups and performance: the US insurance industry, 1970–84. Strategic Management Journal, 11(3), 197-215.
Floyd, F. J., & Widaman, K. F. (1995). Factor analysis in the development and refinement of clinical assessment instruments. Psychological assessment, 7(3), 286.
Gaggero, A. A., & Bartolini, D. (2012). The determinants of airline alliances. Journal of Transport Economics and Policy (JTEP), 46(3), 399-414.
Goh, K., & Uncles, M. (2003). The benefits of airline global alliances: an empirical assessment of the perceptions of business travelers. Transportation Research Part A: Policy and Practice, 37(6), 479-497.
Gomes-Casseres, B. (2003). Constellation strategy: Managing alliance groups. Ivey Business Journal, 67(5), 1-6.
Gudmundsson, S. V., Lechner, C., & van Kranenburg, H. (2012). Multilevel embeddedness in multilateral alliances: A conceptual framework. In: Das, T.K. Research in Strategic Alliances Series: Interpartner Dynamics in Strategic Alliances. Information Age Publishing, USA (forthcoming)
Hatten, K. J., & Hatten, M. L. (1987). Strategic groups, asymmetrical mobility barriers and contestability. Strategic Management Journal, 8(4), 329-342.
Hunt, M. S. (1972). Competition in the major home appliance industry, 1960-1970. Harvard University.
Ketchen, D. J., & Shook, C. L. (1996). The application of cluster analysis in strategic management research: an analysis and critique. Strategic management journal, 17(6), 441-458.
Kleymann, B. (2005). The dynamics of multilateral allying: a process perspective on airline alliances. Journal of Air Transport Management, 11(3), 135-147.
Kleymann, B., & Seristö, H. (2001). Levels of airline alliance membership: balancing risks and benefits. Journal of Air Transport Management, 7(5), 303-310.
Kline R. B., (2004). Beyond Significance Testing: Reforming Data Analysis Methods in Behavioral Research. APA Books: Washington, DC.
Klophaus, R., & Lordan, O. (2018). Codesharing network vulnerability of global airline alliances. Transportation Research Part A: Policy and Practice, 111, 1-10.
Kling, J. A., & Smith, K. A. (1995). Identifying strategic groups in the US airline industry: an application of the Porter model. Transportation Journal, 26-34.
Lordan, O., Sallan, J. M., Simo, P., & Gonzalez-Prieto, D. (2015). Robustness of airline alliance route networks. Communications in nonlinear science and numerical simulation, 22(1-3), 587-595.
McGee, J., & Thomas, H. (1986). Strategic groups: Theory, Research and Taxonomy. Strategic Management Journal, 7(2), 141-160.
Murthi, B. P. S., Rasheed, A. A., & Goll, I. (2013). An empirical analysis of strategic groups in the airline industry using latent class regressions. Managerial and Decision Economics, 34(2), 59-73.
Oneworld (2018) Oneworld at a glance. Network and operations Retrieved from https://www.oneworld.com/news-information/oneworld-fact-sheets/oneworld-at-a-glance#network
Patel, S., & Bhavsar, C. D. (2013). Analysis of pharmacokinetic data by wilk’s lambda (An important tool of manova). International Journal of Pharmaceutical Science
Invention, 2(1), 36-44.
Peteraf, M. A. (1993). The cornerstones of competitive advantage: A resource‐ based view. Strategic Management Journal, 14(3), 179-191.
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.
Porter, M. E. (1990). The competitive advantage of nations. Competitive Intelligence Review, 1(1), 14-14.
Reger, R. K., & Huff, A. S. (1993). Strategic groups: A cognitive perspective. Strategic management journal, 14(2), 103-123.
Skyteam (2018) About us: Facts and figures Retrieved from https://www.skyteam.com/en/about/ on 23-11-2018
Smith, K. G., Grimm, C. M., Wally, S., & Young, G. (1997). Strategic groups and rivalrous firm behavior: Towards a reconciliation. Strategic Management Journal, 149-157.
Star Alliance (2018) Star alliance facts and figures. Retrieved from https://portal.staralliance.com/employees/about/bottom-left/the-star-alliance-network on 23-11-2018
Tang M-J, Thomas H. (1992). The concept of strategic groups: theoretical construct or analytical convenience. Managerial and Decision Economics 13(4): 323–329.
Tokman, M., Richey, R. G., & Deitz, G. D. (2016). A Strategic Choice Theory Taxonomy of Retailers’ Strategic Orientations. Journal of Marketing Theory and Practice, 24(2), 186-208.
