Faculty of Economics & Business
Department of Innovation Management & Strategy
MASTER THESIS
Master in Business Administration, spec. Strategic Innovation Management
Integrating Acquired Alliances:
The Impact of Acquired Alliances on Innovation Performance
Moderated by the Post-Acquisition Strategy
Jurjen M. Lelifeld
S1974971
J.M.Lelifeld@student.rug.nl
First Supervisor: A.A. Oleksiak, MSc. Second Supervisor: Dr. P.M.M. de Faria
Word count: 12472
The Impact of Acquired Alliances on Innovation Performance
Moderated by the Post-Acquisition Strategy
Jurjen M. Lelifeld
University of Groningen, The Netherlands
ABSTRACT
Acquirers often encounter considerable challenges when integrating an acquired firm. These challenges also affect the alliance portfolio once the acquired firm’s alliances have to be incorporated into the alliance portfolio. Nonetheless, the current literature is limited to alliances directly initiated by the focal firm, yet alliances entering the firm’s alliance portfolio through acquisitions may have a substantive effect on the integral portfolio. Therefore, in this paper, I hypothesize that acquired alliances negatively impact the innovation performance of the firm. Moreover, I hypothesize that the applied post-acquisition strategy after the acquisition is crucial to minimize these negative consequences. Using a panel dataset of 22 firms in the ICT service sector containing data for a period of six years, I find that the share of acquired alliances indeed has a negative impact on innovation performance and that the post-acquisition strategy is essential to consider in order to minimize negative implications. This suggests that acquired alliances, as well as the corresponding post-acquisition strategy, are important to consider in the alliance portfolio literature and management.
Key words: strategic alliances, alliance portfolio, acquired alliances, acquisitions,
INTRODUCTION ... 4
LITERATURE REVIEW ... 5
Strategic Alliances and Alliance Portfolios ... 5
The Share of Acquired Alliances ... 7
Post-Acquisition Strategy ... 10 METHODS ... 15 Data Collection ... 16 Sample ... 18 Measures ... 18 Analytical Method ... 22 RESULTS ... 24 Descriptive Statistics ... 24 Regression Results ... 26 DISCUSSION ... 29 Theoretical Implications ... 29 Managerial Implications ... 31
Limitations and Future Research ... 32
CONCLUSION ... 35
REFERENCES ... 36
APPENDIX A: REGRESSION COMPARISON ... 44
4
INTRODUCTION
The focus in the academic field of the innovation paradigm shifted from a closed to an open model (Chesbrough, Vanhaverbeke and West, 2006). An essential part of this open model are the alliances the focal firm possesses. The collection of a focal firm’s alliances, in which the firm is engaged simultaneously, is defined in the literature as an alliance portfolio (e.g. Wassmer, 2010). These portfolios are seen as an efficient strategy that contribute to the acquisition of external resources and therefore many structural (e.g. Duysters, Heimeriks, Lokshin, Meijer and Sabidussi, 2012; Jiang, Tao and Santoro, 2010; Lahiri and Narayanan, 2013) and managerial (e.g. Faems, Janssens and Neyens, 2012) aspects have already been researched. However, the current literature is limited to alliances directly initiated by the focal firm.
5
To test these hypotheses, I have used a fixed-effects panel regression of 22 firms from the ICT services sector between 2000 and 2005. The Thompson Securities Data Corporation Platinum Strategic Alliances and the Mergers & Acquisitions databases have been used, as well as LexisNexis, Orbis and Compustat to obtain the relevant data. The findings partially support the first hypothesis, yet the second hypothesis is not supported. Nonetheless, an opposite effect has been found, implying that structural integration is the more suitable strategy in case of a high share of acquired alliances. The paper adds to the acquisition literature by better explaining the complexities that arise after acquisitions. Moreover, it gives new opportunities for portfolio design research by incorporating acquired alliances into the studies, thereby enhancing insights. The study also adds to the post-acquisition strategy literature by giving detailed insights into the moderating effects of the used strategy.
This paper is structured as follows. First, I review existing literature on alliance portfolios and the post-acquisition strategies and build the hypotheses. Subsequently, I describe the applied methodology in detail after which the analyses and results will be presented. Finally, I discuss the theoretical and managerial implications, outline the main limitations and I identify opportunities for future research. The thesis ends with a conclusion.
LITERATURE REVIEW
This section discusses the relevant concepts of acquired alliances and post-acquisition strategies. First, the definitions of strategic alliances and alliance portfolios are given and the research gap is identified. Subsequently, the hypotheses will be developed based on their theoretical backgrounds.
Strategic Alliances and Alliance Portfolios
6
include contributions by partners of capital, technology, or firm-specific assets.” The strategic benefits of alliances, through the access to unique external resources, encourages firms to form multiple alliances. This allows firms to access multiple bases of resources, thereby reducing risk and uncertainty and adding learning benefits. The result of this collection of alliances is an alliance portfolio, defined by Lavie and Miller (2008: 623) as “a firm’s collection of immediate alliance partners”, which creates strategic advantages to enhance competitiveness (Wassmer, 2010). So far, research has offered many findings on the different modes and characteristics of the alliance portfolio, for instance the quantities of partners (Duysters et al., 2012); roles of partners (Brouthers, Brouthers and Wilkinson, 1995); and structural characteristics (Dyer, Singh and Kale, 2008). These characteristics of portfolios may have a significant impact on innovation performance (Wassmer, 2010) and are thus important to take into account when starting a new alliance to manage the portfolio effectively. Well-managed portfolios can result in performance increases, especially when they provide access to diverse information and capabilities without too much redundancy and complexity (Baum, Calabrese and Silverman, 2000).
7
to be of a different nature than the original alliances. It is therefore important to research the effects these alliances have on the firm.
An important aspect of these alliances are the underlying acquisitions. Wiklund and Shepherd (2009: 195) define acquisitions as “transactions in which one firm buys controlling interest in another firm and the acquired business becomes a subsidiary of the acquirer’s portfolio.” Acquisitions, like alliances, also provide access to complementary resources which can result in growth and competitive advantages. However, higher risks are involved due to higher levels of uncertainty (Wang and Zajac, 2007) and information asymmetries (Reuer, 2005). These high risk levels are also demonstrated by the high failure rates that have been reported on acquisitions (e.g. Chang and Tsai, 2013). Therefore it is expected that these risks are also translated to the acquired alliances that enter the alliance portfolio of the acquiring firm. The different nature of the acquired alliances vis-à-vis the original alliances may impose extra risks and challenges and the acquired alliances are therefore expected to have a different effect on performance compared to original alliances.
The Share of Acquired Alliances
8
These insights from the resource-based view are also applicable to acquired alliances and show that they can be potentially performance increasing as they give firms access to new (potentially unique) resources. Nonetheless, careful selection is of crucial importance as indicated by Lavie (2009: 26) who emphasizes that “the success of an alliance portfolio depends (…) on the nature of partners with whom a company decides to ally”. This is especially relevant for acquired alliances, since the nature of these alliances differs from the nature of original alliances. In the case of original alliances a process is in place that gives the firm full control and allows it to look for a fit of the alliance in the alliance portfolio (Laursen and Salter, 2014), which is not possible with acquired alliances.
9
Moreover, engaging in R&D alliances creates a restrain on managerial costs and complexity, which reduces innovation performance (Duysters and Lokshin, 2011). When looking at the coordination perspective (regarding portfolio management), acquired alliances will increase this restrain even more, since all acquired alliances enter the portfolio at once. This high degree of complexity will have a negative effect on the portfolio performance and thereby the innovation performance of the firm (Hoang, 2001).
Lastly, the coordination perspective also shows that acquired alliances impose problems due to the difficulty of linking the firm’s capabilities (Puranam, Singh and Chaudhuri, 2009). Since the acquired firm negotiated the alliances (instead of the acquiring firm) it likely has a trusted relationship with these partners and experiences low levels of insecurity. Failing to properly integrate the acquired firm will therefore make it increasingly difficult to integrate their alliances as well (with higher insecurity and lower trust as a result), since the relational capital is not properly transferred.
Overall, these points stress the challenges acquired alliances pose in addition to the normal challenges faced by a firm concerning its alliance portfolio. When the focal firm’s portfolio is expanded through acquired acquisition, the entire portfolio may change, which has consequences for all partners involved. These new partners result in insecurity for the entire portfolio; not only the focal firm has to make sure there is a successful cooperation, but all partners are now dealing with the new party involved (Lavie, 2006). It is thus not only the strain on managerial resources that reduces relational capital, relations also suffer due to the addition of new alliances that are unknown to the current alliance partners which creates tensions, uncertainty and lowers trust (Sarkar et al., 2001).
10
sharing and thus decrease access to resources. On top of that coordination difficulties further reduce these attributes. Overall this will hamper the innovation performance of the firm. Consequently, the higher the amount of acquired alliances that enter the portfolio (resulting in a higher share of acquired alliances), the more severe the aforementioned effects will be. Therefore, I hypothesize that:
H1: The relationship between the share of acquired alliances in a firm’s portfolio and
innovation performance is negative.
Post-Acquisition Strategy
Based on Porter and Millar (1985), Teixeira et al. (2012: 73) defined integration as “the collaboration and linkages between and across organizational functions as well as organizational partners, including customers and suppliers”. It is thus important to achieve effective coordinating activities across the divisions within the firm (Puranam et al., 2009) and create unique organizational structures within firms to work toward a firm’s unique competitive advantage (as previously described with the resource-based view). This works towards the creation of an organizational fit as it impacts management styles, administrative practices, cultural practices and personal characteristics (Jemison and Sitkin, 1986).
11
autonomy since they are mostly influenced by the different post-acquisition strategies (Angwin and Meadows, 2015; Puranam and Srikanth, 2007; Ranft, 2006).
First of all, it is often difficult to transfer tacit knowledge after an acquisition. Therefore efficient communication between the acquiring and acquired firms to facilitate knowledge transfer is crucial. This often requires in-depth face to face communication, like interfirm meetings (Ranft and Lord, 2000). However, it is important that the acquiring firm respects the culture of the acquired firm and is supportive (Schweizer and Patzelt, 2012). Moreover, the management of the acquired firm also plays a crucial role. Compatibility of management styles is important, but often creates difficulties due to conflicts and discord between both firms (Ranft, 2006). Changes will lead to complexity and uncertainty as well as increased tensions with managers (Datta, 1991). Control and monitoring will lead to high employee turnover, which negatively affects performance (Walsh, 1988). Acquired firms often react defensively by clinging to their own beliefs to protect their identity and reduce uncertainty. It is therefore important for an acquiring firm to use the right post-acquisition strategy to effectively tackle these difficulties and restrictions on performance.
12
distort the target firm’s innovative capabilities since it does not have autonomy over its processes any longer. This may result in lower motivation and productivity, and it increases free-riding (Baker, 2002). Also the changed formal procedures in the acquired firm can undermine the innovative capabilities (Benner and Tushman, 2003).
Conversely, structural preservation grants the acquired organization “a high degree of autonomy, typically being positioned within the acquiring organization as [a] stand-alone subsidiary” (Schoenberg, 2004: 155). It involves high levels of autonomy and limited contact between management (Haspeslagh and Jemison, 1991). In this case, the acquired firm will be relatively unaffected by the acquirer’s organizational climate (Schoenberg, 2004). However, drawbacks to this strategy are lower amounts of knowledge transfer due to limited contact and groupthink (Ranft, 2006) as well as difficulties with strategy setting and financial reporting due to different management styles (Schoenberg, 2004). It is important to link these two strategies to the previously mentioned concepts from the acquisition literature to show the effect each strategy has.
13
organizations can be limited reducing conflict (Ranft, 2006). However, autonomy reduces the disruptive effect regarding the management of the acquired firm. It creates value by keeping capabilities that are difficult to integrate, like value systems and firm culture, which makes sure key employees and managers stay with the firm (Ranft and Lord, 2000).
These points are similarly important in the case of acquired alliances. As discussed in the previous subsection, a high share of acquired alliances in the portfolio results in a lack of trust and relational capital as well as high uncertainty due to disruption. Moreover, acquisitions and acquired alliances create constraints on managerial resources. In the case of a high share of acquired alliances, the usage of the integration strategy leads to more disruption for the firms in the alliance portfolio since all the alliances have to be integrated, thereby increasing distrust and uncertainty (Empson, 2001). This leads to less knowledge sharing, less cooperation between the firms and the unique resources that created the value in the first place may now be lost. Also, due to the complexity of the integration process, the firm does not have time to establish a trusted relationship with the acquired alliance to position the alliance properly in the portfolio and create new goals and procedures (Jemison and Sitkin, 1986). A high share of acquired alliances can make it even harder to achieve the trusted relationship with all acquired alliances, thereby increasing managerial conflict. However, a positive effect of the integration strategy is the creation of new common goals and procedures with the alliance, which will be in line with the goals and procedures of the other alliances in the portfolio as well (providing this is realized despite the complexity of the integration process). These common goals will increase cooperation and coordination (Puranam et al., 2009), thereby also increasing knowledge transfer. Yet taken together, these properties will decrease the innovation performance of the firm due to the overall increased disruptions and uncertainty.
14
with the original alliances (Puranam and Srikanth, 2007). Adding many more acquired alliances in the case of a high share does therefore not directly increase the impact on the original portfolio. Moreover, the management structure will remain, thereby providing an already familiar interface keeping knowledge transfer, common goals and procedure at a high level, which reduces coordination difficulties that can become extremely complex in the case of many acquired alliances (Ranft and Lord, 2000). This will result in fewer communication, trust and relational issues, thereby reducing uncertainty. However, issues with the preservation strategy can occur due to constraints on managerial resources in the focal firm (Puranam et al., 2009). The autonomy of the acquired firm and thereby the contacts with the alliances mean that even more managerial attention is needed to effectively transfer knowledge. Also more control is needed. Yet overall, by keeping the original alliances structures in place, these issues will be less significant and reduce the negative impacts of a high share of acquired alliances on innovation performance.
Based on the above, I expect that a low level of integration (i.e. structural preservation), which limits negative impacts like loss of autonomy, organizational disruptions, process complexity and implementation costs also limits the negative effects of the share of acquired alliances on innovation performance of the firm. However, the level of integration should be enough to allow for efficient knowledge transfer and cooperation (i.e. the acquired firm and alliances should not be left alone completely; Zollo and Singh, 2004). It is especially important to leave management intact to protect against organizational disruption (Zollo and Singh, 2004). Therefore, I hypothesize that:
H2: The relationship between the share of acquired alliances in a firm’s portfolio
15
The overall conceptual model, presented in figure 1, outlines both the main effect of the share of acquired alliances on innovation performance, as presented in hypothesis 1, as well as the moderating effect of the post-acquisition strategy on the relationship between the share of acquired alliances and innovation performance, as presented in hypothesis 2. In the next section, I outline the applied methodology of this study.
METHODS
16
The ICT services sector has been chosen because of the rapid pace of technology developments (Pisano, Russo and Teece, 1998), thereby forcing firms to constantly innovate. Moreover, patents are important in this sector, since they enable firms to capture the value they have created through their innovations (Hall and Ziedonis, 2001) and play a central role in innovation efforts in the industry (Howells, 2000). Moreover, to spread risks and to remain competitive with the constant shifting changes in technology, firms in the ICT services sector are also constantly seeking outside knowledge through alliances and acquisitions (Lambe and Spekman, 1997). Thus, the ICT services sector is well-suited to research the proposed effects.
Data Collection
As mentioned in the introduction of this section, the focal firms of the sample are constructed based on a fixed list with the largest industry companies (see table B1 in appendix B for the specific list). Using the Thomson Securities Data Corporation (SDC) database on Strategic Alliances, all alliance announcements were obtained by searching on the ultimate parent name of the firm for each focal firm. In case of subsidiaries or large conglomerates that are active in multiple – unrelated – industries, irrelevant announcements were filtered out. In the case of subsidiaries only announcements that specifically specified they belong to the subsidiary are kept.
17
However, the acquired alliance will not be counted in the portfolio as long as it does not reach the acquisition year – for the reason that before this year, the alliance will still be in another firm’s portfolio. For example, if an alliance is formed in an acquired firm in the year 2000 it will last until 2004. However, the firm is only acquired in the year 2003. Therefore, this acquired alliance will only be counted from 2003 until its ending date (in this instance 2004).
After the retrieval of alliance data of the focal firms, the acquisition data has been obtained from the SDC Platinum database on Mergers & Acquisitions. Subsequently, through each firm’s acquisition data, the alliance data of the acquired firms has been obtained. For this the same method has been used as described for the focal firms. However, in many cases only subsidiaries or specific divisions were acquired. For example, Cisco acquired CAIS Software Solutions, which was part of CAIS Internet Inc. (the ultimate parent). In this case, only the alliances from the Software Solutions divisions are retained, whereas alliance data from the ultimate parents (that were not specifically linked to the Software division) or alliance data from other divisions has been filtered out.
To retrieve the data for the moderator variable (post-acquisition strategy), the LexisNexis database has been used. Following Puranam, Singh and Zollo (2006), a completely disappearance of the firm two years after the acquisition was interpreted as structural integration (and coded as 0) whereas a continued appearance two years after the acquisition indicated that structural preservation had occurred (coded as 1). To find the evidence on preservation, all mentions on the firm were searched from the LexisNexis database dated two years after the acquisition (Puranam et al., 2009). When company activities were found, this was counted as preservation, whereas a lack of results indicated a structural integration strategy.
18
number of previous acquisitions conducted by a firm, the data from the SDC M&A database has been used. The industrial diversity SIC codes have obtained from the SDC Strategic Alliances database. Lastly, the number of employees has been obtained from the Compustat database, which derives the data from annual reports. In case of missing data, the data has been complemented by manually searching for the firm’s annual reports.
Finally, all data has been merged into a single datasheet, containing alliances, acquired alliances, patents and the control variables. Subsequently the statistical software package STATA14 has been used to conduct the statistical analysis.
Sample
All obtained alliances have been manually reviewed and processed for the original 30 firms. After filtering out irrelevant alliance announcements, the focal firm’s portfolios contained 4824 alliances from 1996 to 2005, of which 4353 are original alliances and 471 acquired alliances (obtained through 356 acquisitions). The final sample consists of 22 firms (see table B1 in appendix B), since 8 firms have been omitted in the regression analysis due to missing values. In total this results in 116 observations during the 2000-2005 time period. Out of these 22 firms, 13 firms have both original and acquired alliances in their portfolios. The other 9 firms have only original alliances in their portfolios, even though in total 18 firms conducted acquisitions during the time period. This shows that 5 firms acquired firms without alliances.
Measures
This subsection will comprehensively describe the variables and their respective measurements used in this study.
Dependent variable. Number of patents. Innovation performance is measured by
19
usually a stronger indicator of R&D performance than the measure of R&D expenditures (Griliches, 1990). The count will be based on the total number of patents issued by the focal firm within the specific time period and is measured per specific year. For each firm the patens will be counted based on a four year window with a one year time lag (Samspon, 2007). For example, to measure the effect of the share of acquired alliance in 2000, the number of patents is constructed for the time period 2001 – 2004. The lag is included to account for the time it takes to transfer knowledge and start efficient cooperation between the focal firm and the alliance partner before innovations can be created and patents can be published (Hausman, Hall and Griliches, 1984). The publication date is used to determine the corresponding year.
Independent variable. Share of acquired alliances in the total portfolio. The share of
acquired alliances is created since it is expected to have a negative effect on a firm’s innovation performance. It is measured per focal year and is calculated from the number of original
alliances and number of acquired alliances that have been counted for that specific year. The
20
𝑆𝐴𝑡= 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑎𝑐𝑞𝑢𝑖𝑟𝑒𝑑 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠𝑡
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑎𝑐𝑞𝑢𝑖𝑟𝑒𝑑 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠𝑡+ 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠𝑡 where t ∈ {2000, …, 2005}.
Post-acquisition strategy (moderator). As explained under data collection subsection,
the number of preserved and the number of integrated alliances have been obtained and counted for each year. The post-acquisition strategy was then measured per focal year and calculated as the fraction of preserved alliances on the total number of acquired alliances. The result can be any positive decimal number ranging from 0 to 1, where a value of 0 would indicate that all firms have been integrated and a value of 1 indicates that all firms have been preserved. Thus, to compute the post-acquisition strategy (PAS) of acquired alliances for a given firm in a given year, t, the formal notation would be,
𝑃𝐴𝑆𝑡 =
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑟𝑒𝑠𝑒𝑟𝑣𝑒𝑑 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠𝑡
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑟𝑒𝑠𝑒𝑟𝑣𝑒𝑑 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠𝑡+ 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠𝑡
where t ∈ {2000, …, 2005}.
Control variables. I controlled for several firm, relational and acquiring characteristics
that could possibly influence the innovation performance of the firm.
Firm size. The size of the firm may influence specific firm attributes. Specifically, larger
21
Acquirer acquisition experience. Experience with acquisitions can increase the focal
firm’s ability to integrate the firms or to acquire the knowledge. Also, the experience can reduce the number of disruptions that arise. It can help avoid a depraved relationship between firms (Puranam et al., 2009) and it allows the firms to better manage their alliance activities (Anand and Khanna, 2000). On the other hand, when a firm performs many acquisitions it can constrain resources and management, thus becoming very costly (Riviezzo, 2013). In line with Puranam et al. (2009), the acquisition experience of a firm will be measured as the total number of prior acquisitions carried out by the acquirer since the start of the alliance time period 1996, to allow the measure to take into account the intensity before the focal period. It is measured per specific year and can be any positive integer.
Industrial diversity. Diversity of a portfolio impacts performance of the firm, where high
and low diversity are negatively correlated with performance and moderate amounts of diversity are positively related to performance (Jiang et al., 2010). This is attributable to the novelty value of the portfolio compared to the absorptive capacity, where low levels of diversity are likely to lead to low levels of novelty and high levels are likely to lead to difficulties concerning the absorptive capacity of the firm (Gilsing, Nooteboom, Vanhaverbeke, Duysters and Van den Oord, 2008).
22
diversity, the degree of diversity in firm i's alliance portfolio is measured as one minus the sum of squared portions of each category j’s occurrences in the alliance portfolio,
𝐵𝐼𝑉𝑖 = 1 − ∑ 𝑃𝑖𝑗2
𝑁
𝑗
where i is the firm, N is the total number of categories, j the category and…
𝑃𝑖𝑗 = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠 𝑓𝑟𝑜𝑚 𝑓𝑖𝑟𝑚 𝑖 𝑖𝑛 𝑐𝑎𝑡𝑒𝑔𝑜𝑟𝑦 𝑗
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑎𝑙𝑙𝑖𝑎𝑛𝑐𝑒 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑜𝑓 𝑓𝑖𝑟𝑚 𝑖
The value of BIV can range anywhere from 0 (indicating a perfectly homogenous portfolio) to (𝑁−1)
𝑁 , where the maximum score is thus dependent on the total number of
categories N. In this case the maximum score is equal to 0.75, since the total number of categories is equal to 4. After the calculation of the BIV value, the Index of Qualitative Variation (IQV) correction is applied to standardize the values and make them better interpretable by creating a positive decimal number ranging from 0 to 1 regardless of the number of categories (Agresti and Agresti, 1978). The IQV is equal to 𝑁
(𝑁−1), in this instance
equaling to IQV = 1.25. This value is then applied to the BIV scores by multiplication.
Analytical Method
23
Therefore the negative binomial regression should be used as the method of analysis (Katila, 2002), which can be done in STATA with the nbreg command.1
However, due to the small sample size when the moderator is included, STATA indicated that the nbreg method is not able to converge. Consequently the models yield no usable (valid) results and this method cannot be applied to the dataset. Therefore the only possible course of action to produce valid results (although with caution) is the usage of a linear regression for panel data (which can be done in STATA with the xtreg command). Nonetheless, this method violates the assumptions for count data, since it is in fact created for testing continuous dependent variables.
Therefore, to tackle this problem, the dependent count variable should be transformed logarithmically thereby making the linear regression method usable for this count panel dataset, albeit with certain limitations.2 After transformation the count variable is closer to a continuous variable and thus more suitable for the linear regression analysis.3
Finally, the linear fixed-effects regression analysis has been used for hypotheses testing, since it is seen as more appropriate than random-effects modeling if the data cannot pass the Hausman test (Jiang et al., 2010). The conducted Hausman test indicated that the coefficient estimations delivered by the random-effects test were significantly different to the estimations from the fixed-effects test in one model (Model 1: χ2 = 16.87, p = 0.0008). Furthermore, the random effects model assumes no correlation between the errors and regressors, whereas the fixed-effects model explores the variation within entities. Since the correlation terms are
1 The xtnbreg command, which has been based on the Hausman et al. (1984) model and is specifically created for
negative binomial models with panel data, should not be used due to flaws in the model. As Allison and Waterman (2002) have shown the model is not a truly fixed effect models and does not control for all stable predictors. Therefore they recommend to use the nbreg command that does not have these limitations.
2 The transformation of a count variable to continuous variable by calculating the logarithmic values is not flawless.
The logarithmic value can only take positive values, whereas normally a linear regression can also have negative values.
3 Please refer to appendix A for a detailed comparison between the transformed and non-transformed regression
24
significantly different from 0 in all cases, it is better to use the fixed-effects model. This shows that the less efficient but safer fixed-effect regression model should be used, since it accounts for unobserved heterogeneity between firms (Noseleit and De Faria, 2013).
RESULTS
This section discusses the results of the statistical analysis. First the descriptive statistics will be discussed followed by the regression results for the hypotheses.
Descriptive Statistics
This subsection discusses the core statistics of the variables as well as the correlations between them, as indicated in table 1. The sample in the regression analysis consists of 22 firms with 116 observations. However, when the moderating variable is included the number of observations goes down to 53 at a total of 12 firms. This is due to the nature of the moderating variable (post-acquisition strategy) which only holds values when the firm has acquired alliances in the alliance portfolio. In other words, without acquired alliances there is no post-acquisition strategy used by the focal firm in the alliance portfolio and it thus holds no value.
25
Moreover, in 2000 the average firm had a total portfolio size of 120.91 alliances, which was almost equal to the 114.91 alliances in 2005. As shown in figure B3 (appendix B), the average portfolio size shows a concave relation. The average share of acquired alliances in the total portfolio is 5.67%. When breaking this number down in yearly averages, it can be seen that it sharply increased from 0.89% in 2000 to 8.87% in 2005 (see figure B4 in appendix B for specific development details).
Regarding the moderating variable, the average number of the post-acquisition strategy variable is 0.2654, which shows that – on average – the sample firms make more use of the structural integration strategy than the structural preservation strategy. Out of the 13 firms using a post-acquisition strategy, 4 made use solely of the integration strategy, 9 firms used a mixture of the strategies and none of the firms used only the preservation strategy. When breaking the number down in yearly averages, it can be seen that in 2000 the average post-acquisition strategy is more focused on the integration strategy (0.40) and in the year 2005 this has gradually consolidated (see figure B5 in appendix B for specific the change over time) resulting in the use of even more structural integration (0.24).
Figure 2 below shows that the distribution of patents is highly skewed to the right, where the majority of firms obtained only a small amount (or even zero) patents. This is in line with existing research on patents (Schoenmakers and Duysters, 2010). Moreover, the control variables gives some interesting insights. The average firm has 95472 employees, reflecting the use of large firms in the sample. Correspondingly, an average firm conducted 12.47 acquisitions (measured from 1996), showing fairly experienced firms regarding acquisitions in the sample. Lastly, the diversity of the sample is fairly moderate with an industrial diversity score of 0.598.
4 A value of 0 indicates a firm uses only structural integration and a value of 1 indicates a firm uses only structural
26
Finally, a correlation test (as reported in table 1) and a Variance Inflection Factors (VIF) test have been conducted in STATA to rule out multicollinearity issues. The mean VIF value for the variables in the model are equal to 1.34 with a highest value of 1.56 for the moderating variable. Since none of the variables exceed the critical multicollinearity value of 10.00 the VIF test yields no problems (see table B2 in appendix B). Moreover, none of the correlation values exceed the threshold value of 0.70. Therefore these results confirm that we can rule out multicollinearity issues (Grewal, Cote and Baumgartner, 2004).
Regression Results
27
since the variable can only hold values when a firm has acquired alliances. In this model the value of share of acquired alliances is again negative and significant on the 10% level. The model also has a significant F-test score. Based on these findings I conclude that hypothesis 1 can be partially supported. The findings show that the share of acquired alliances indeed has a negative linear impact on the innovation performance of a firm (β = -0.120, p ≤ 0.1), which is in particular visible in a sample consisting of firms having access to acquired alliances. However, the results should be interpreted with care since both models show only limited support.
28
strategy, which is used to test the moderating effect of hypothesis 2, is negative and significant (β = -0.408, p ≤ 0.05). Interestingly the share of acquired alliances became insignificant in this model. This points towards a cross-over interaction effect (Keppel, 1991). This effect can indeed be observed, as shown in figure 3. This plot shows that low post-acquisition strategy (i.e., structural integration) reduces the negative effect of the share of acquired alliances on the firm’s innovation performance, whereas the high post-acquisition strategy (i.e., structural preservation) is better in the case of a low share of acquired alliances. Since this is the opposite effect of the hypothesized one, hypothesis 2 cannot be supported, yet the effect is significant.
29
These results show that both the share of acquired alliances in the alliance portfolio and the post-acquisition strategy matter regarding a firm’s innovation performance. It shows that managing acquired alliances is difficult and that is important for management to consider the strategies after the acquisition of a company, also regarding the alliances. It is therefore also important to account for acquired alliances when researching alliances. The results will be more elaborately discussed in the next section.
DISCUSSION
In this final section, the theoretical implications of the findings will be discussed. Subsequently, the managerial implications will be considered. Finally, the main limitations as well as the opportunities for future research of the study are presented.
Theoretical Implications
To date, research in the field of alliances portfolios has mainly focused on improving alliance portfolio performance (e.g. Duysters et al., 2012; Jiang et al., 2010; Neyens and Faems, 2013), thereby assuming that these portfolios consist only of original alliances. I supplement this research by adding the notion of acquired alliances, thereby analyzing how these alliances affect firms, how firms can best deal with the new alliances strategically and their impact on innovation performance. These concepts should help the current alliance portfolio research to better understand the alliance portfolios and their impact on performance.
30
share of acquired alliances (however the opposite effect of the hypothesized one). These results thus indicate that acquired alliances are relevant for future research. Nonetheless, these results should be interpreted with caution due to limitations in the statistical model, which will be elaborately discussed in the last subsection.
This study reveals that a bigger share of acquired alliances in a firm’s alliance portfolio has a negative impact on a firm’s innovation performance, thereby completing previous research. This result helps to advance the knowledge on value creation through acquisitions (e.g. Wiklund and Shepherd, 2009) with the complexity that has been found in current literature. For example, the coordinating of resources during and after acquisitions has been found to contribute to the complexity of acquisitions (e.g. Reuer, 2005; Wang and Zajac, 2007). The findings reveal that alliances add to the complexities acquiring firms face as well, which could be linked to current concepts in the acquisition literature as managerial involvement (Faems et al., 2012) and lack of trust (Doz, 1996, Sarkar et al., 2001) thereby creating more insights. Moreover, current alliance portfolio research shows the many challenges a firm faces regarding their portfolio, like diversity (e.g. Laursen and Salter, 2014), strategic misfit (Nielsen, 2010), coordination (e.g. Duysters and Lokshin, 2011) and trust (e.g. Sarkar et al., 2001). This shows that a proper portfolio design is important (e.g. Faems et al., 2012; Neyens and Faems, 2013). The concept of acquired alliances adds to this research and allows for new opportunities to explore why the acquired firms add to difficulties precisely and how this can be overcome.
31
preservation (e.g. Ranft and Lord, 2000; Schweizer and Patzelt, 2012) mostly finding negative consequences and difficulties for firms.
This paper helps to advance the existing literature by showing that the post-acquisition strategy also has a moderating effect in the case of acquired alliances and is thus not limited to merely direct effects (e.g. Datta and Grant, 1990; Puranam et al., 2009). Especially interesting is the choice between coordination and autonomy, where the benefits from coordination obtained through structural integration dominate the costs of the disruption that has been caused by the loss of autonomy (Puranam et al., 2006). This suggests that the shaping of organizational processes like (tacit) knowledge transfer and the creation of common goals as well as (informal) procedures and goals likely leads to a coordination effect (Puranam et al., 2009). This effect occurs due to underlying activities between the acquirer’s and target’s capabilities that are dependent on each other, which is not present in the case of autonomy.
On the other hand, the effect also suggest that the structural preservation strategy seems to be most suitable in the case of a small number of acquired alliances. This could be explained by the fact that the acquiring firm’s managerial resources are less constrained (e.g. Ranft, 2006). This makes it possible for the firm to take care of both the coordination and autonomy difficulties. Due to the high levels of autonomy the firm experiences more difficulties regarding coordination, yet more managerial resources are available to make sure successful implementation possible (Jemison and Sitkin, 1986). In the case of a high share of acquired alliances there are not enough resources to properly implement the coordination properties (Datta, 1991). Therefore the firm should choose the importance of the coordination effect over autonomy, which is achieved through the structural integration strategy.
Managerial Implications
32
better understanding. Based on this study’s findings, specific guidelines can be provided to help with the management of acquired alliances in the alliance portfolio. The data suggests that integrating acquired alliances in the alliance portfolio is difficult and can be negative for innovation performance. This is on top of the many difficulties managers already face with acquisitions. At the same time, if managers do decide to acquire a firm with alliances, it is important to make sure the right post-acquisition strategy is applied. The study suggests that managers could best choose a strategy of structural preservation if the current portfolio consists of only a small number of acquired alliances whereas a structural integration strategy seems to be better suitable in case of many acquired alliances. Of course, it is important to keep in mind the specific characteristics of each acquired firm and alliance thereby taking into account the individual characteristics to optimize their specific needs. In general, the study puts forward that it is important to treat carefully when acquiring firms due to the negative effects on innovation performance and firms should not neglect the current partnerships of the potential takeover candidate.
Limitations and Future Research
33
different results that are more suitable to acquired alliances, thereby optimizing the implications for managers.
Secondly, the chosen time period of 2000-2005 in the ICT service sector is characterized by the dot-com bubble where the equity prices of – mainly – ICT firms peaked and rapidly dropped during 1998 to 2001 (Clarida, 2007). This may have impacted the investments made by firms in the industry, thereby impacting acquisition and alliance deals. It would therefore be interesting to see if the results differ during another time-period not characterized by such a macroeconomic event or in an industry that has not been affected by this event.
Furthermore, due to time limitations, this study uses Sampson’s (2007) assumption that each alliance lasts five years. However, real world alliance duration is not fixed and alliances even create value over time (e.g. Gulati, 1999). Therefore it is important to take all partnering years into account and it would be interesting to see if models become stronger when alliances remain active in the portfolio until evidence is found that it was canceled by the firms. This could be achieved by finding the alliance ending dates via LexisNexis.
34
adding the use of weighted patents could make the patent data and thereby the overall model stronger (Sampson, 2007).
Moreover, the SDC Strategic Alliances database creates limitations since it does not hold complete data. One of the main disadvantages of this SDC database (even though it is often seen as the most comprehensive and elaborate database regarding alliances), is the bias towards English language sources (Schilling, 2009). This results in missing information from non-US firms, especially Asian firms. Since the sample used in this study also contains various non-US firms (see table B1 in appendix B) it may therefore be interesting to see if the use of LexisNexis to obtain the alliance data creates stronger models.
The last limitation is primarily an empirical one caused by the size of the dataset due to time restrictions. The small size of the dataset made it impossible to use the negative binomial regression, which would be the most suitable regression model in case of overdispersed count data. The small number of observations is mainly caused by the moderating variable, since it only holds values when a firm’s portfolio contains acquired alliances. Future research could solve this problem by including more firms in the sample and by increasing the time-period.
The researched effect on innovation performance also allows for future research opportunities. I could be interesting to research the effect in different industries. At this moment the effect has been researched in a high-tech industries only. It would thus be interesting to see if the results also hold in different (low- and medium-technology) industries (Santamaría, Nieto and Barge-Gil, 2009). Lastly, it would be interesting to investigate the issues of portfolio management in more detail to investigate possibilities to lessen the negative impact of the share of acquired alliances (Neyens and Faems, 2013).
35
Important with M&A activities is the aspect of knowledge transfer to make sure the acquired firms are successfully incorporated (Vaara, Sarala, Stahl and Björkman, 2012) thereby also allowing the acquired alliances to be transferred successfully. Moreover, employee retention is crucial to successfully integrate, also regarding the management area (Ahammad, Tarba, Liu and Glaister, 2016) and possibly helps to integrate acquired alliances as well. These kinds of insights will help the organization of M&A activities of firms, thereby possibly making them more successful, also in the case of their alliances.
CONCLUSION
36
REFERENCES
Agresti, A., & Agresti, B. F. (1978). Statistical analysis of qualitative variation. Sociological
methodology, 9, 204-237.
Ahammad, M. F., Tarba, S. Y., Liu, Y., & Glaister, K. W. (2016). Knowledge transfer and cross-border acquisition performance: The impact of cultural distance and employee retention. International Business Review, 25(1), 66-75.
Albrecht, M. A., Bosma, R., van Dinter, T., Ernst, J. L., van Ginkel, K., & Versloot-Spoelstra, F. (2010). Quality assurance in the EPO patent information resource. World Patent
Information, 32(4), 279-286.
Allison, P. D., & Waterman, R. P. (2002). Fixed–effects negative binomial regression models.
Sociological methodology, 32(1), 247-265.
Anand, B. N., & Khanna, T. (2000). Do firms learn to create value? The case of alliances.
Strategic management journal, 21(3), 295-315.
Angwin, D. N., & Meadows, M. (2015). New integration strategies for post-acquisition management. Long Range Planning, 48(4), 235-251.
Baker, G. (2002). Distortion and risk in optimal incentive contracts. Journal of human
resources, 728-751.
Barney, J. (1991). Firm resources and sustained competitive advantage. Journal of
management, 17(1), 99-120.
Baum, J. A., Calabrese, T., & Silverman, B. S. (2000). Don't go it alone: Alliance network composition and startups' performance in Canadian biotechnology. Strategic management
journal, 21(3), 267-294.
37
Brouthers, K. D., Brouthers, L. E., & Wilkinson, T. J. (1995). Strategic alliances: Choose your partners. Long range planning, 28(3), 2-25.
Camerer, C., & Knez, M. (1996). Coordination, organizational boundaries and fads in business practices. Industrial and Corporate Change, 5(1), 89-112.
Chang, S. C., & Tsai, M. T. (2013). The effect of prior alliance experience on acquisition performance. Applied Economics, 45(6), 765-773.
Chesbrough, H., Vanhaverbeke, W., & West, J. (Eds.). (2006). Open innovation: Researching
a new paradigm. Oxford university press.
Clarida, R. H. (Ed.). (2007). G7 Current Account Imbalances: Sustainability and Adjustment. University of Chicago Press.
Comanor, W. S., & Scherer, F. M. (1969). Patent statistics as a measure of technical change.
The Journal of Political Economy, 392-398.
Datta, D. K. (1991). Organizational fit and acquisition performance: Effects of post‐acquisition integration. Strategic management journal, 12(4), 281-297.
Datta, D. K., & Grant, J. H. (1990). Relationships between type of acquisition, the autonomy given to the acquired firm, and acquisition success: An empirical analysis. Journal of
Management, 16(1), 29-44.
Datta, D. K., Guthrie, J. P., & Wright, P. M. (2005). Human resource management and labor productivity: does industry matter?. Academy of management Journal, 48(1), 135-145. Doz, Y. L. (1996). The evolution of cooperation in strategic alliances: Initial conditions or
learning processes?. Strategic management journal, 17(S1), 55-83.
Duysters, G., & Lokshin, B. (2011). Determinants of Alliance Portfolio Complexity and Its Effect on Innovative Performance of Companies*. Journal of Product Innovation
38
Duysters, G., Heimeriks, K. H., Lokshin, B., Meijer, E., & Sabidussi, A. (2012). Do Firms Learn to Manage Alliance Portfolio Diversity? The Diversity‐Performance Relationship and the Moderating Effects of Experience and Capability. European Management Review, 9(3), 139-152.
Dyer, J. H., Singh, H., & Kale, P. (2008). Splitting the pie: rent distribution in alliances and networks. Managerial and Decision Economics, 29(2‐3), 137-148.
Empson, L. (2001). Fear of exploitation and fear of contamination: Impediments to knowledge transfer in mergers between professional service firms. Human relations, 54(7), 839-862. Faems, D., Janssens, M., & Neyens, I. (2012). Alliance portfolios and innovation performance
connecting structural and managerial perspectives. Group & Organization Management,
37(2), 241-268.
Faems, D., Van Looy, B., & Debackere, K. (2005). Interorganizational collaboration and innovation: toward a portfolio approach. Journal of product innovation management, 22(3), 238-250.
Galbraith, J. R. (1977). Organization design: An information processing view. Organizational
Effectiveness Center and School, 21, 21-26.
Gilsing, V., Nooteboom, B., Vanhaverbeke, W., Duysters, G., & van den Oord, A. (2008). Network embeddedness and the exploration of novel technologies: Technological distance, betweenness centrality and density. Research policy, 37(10), 1717-1731.
Goldhaber, G. M., Dennis, H. S., Richetto, G. M., & Wiio, O. A. (1984). Information Strategies:
New pathways to management productivity. New York. Ablex
39
Griliches, Z. (1990). Patent Statistics as Economic Indicators: A Survey. Journal Of Economic
Literature, 28(4), 1661-1707.
Gulati, R. (1999). Network location and learning: The influence of network resources and firm capabilities on alliance formation. Strategic management journal, 20(5), 397-420.
Gulati, R., & Singh, H. (1998). The architecture of cooperation: Managing coordination costs and appropriation concerns in strategic alliances. Administrative science quarterly, 781-814. Hall, B. H., & Ziedonis, R. H. (2001). The patent paradox revisited: an empirical study of patenting in the US semiconductor industry, 1979-1995. RAND Journal of Economics, 101-128.
Haspeslagh, P. C., & Jemison, D. B. (1991). Managing acquisitions: Creating value through
corporate renewal (Vol. 416). New York: Free Press.
Hausman, J., Hall, B. H., & Griliches, Z. (1984). ECONOMETRIC MODELS FOR COUNT DATA WITH AN APPLICATION TO THE PATENTS-R&D RELATIONSHIP.
Econometrica, 52(4), 909-938.
Hoang, H. (2001). The impact of organizational and alliance-based complexity on the
development of alliance capacity. INSEAD.
Howells, J. (2000). Innovation & services: New conceptual frameworks. Centre for Research on Innovation and Competition, The University of Manchester.
Jemison, D. B., & Sitkin, S. B. (1986). Corporate acquisitions: A process perspective. Academy
of Management Review, 11(1), 145-163.
Jiang, R. J., Tao, Q. T., & Santoro, M. D. (2010). Alliance portfolio diversity and firm performance. Strategic Management Journal, 31(10), 1136-1144.
Katila, R. (2002). New product search over time: past ideas in their prime?. Academy of
Management Journal, 45(5), 995-1010.
40
Lahiri, N., & Narayanan, S. (2013). Vertical integration, innovation, and alliance portfolio size: Implications for firm performance. Strategic Management Journal, 34(9), 1042-1064. Lambe, C. J., & Spekman, R. E. (1997). Alliances, external technology acquisition, and
discontinuous technological change. Journal of product innovation management, 14(2), 102-116.
Laursen, K., & Salter, A. J. (2014). The paradox of openness: Appropriability, external search and collaboration. Research Policy, 43(5), 867-878.
Lavie, D. (2006). The competitive advantage of interconnected firms: An extension of the resource-based view. Academy of management review, 31(3), 638-658.
Lavie, D. (2009). Capturing value from alliance portfolios. Organizational dynamics, 38(1), 26-36.
Lavie, D., & Miller, S. R. (2008). Alliance portfolio internationalization and firm performance.
Organization Science, 19(4), 623-646.
Neyens, I., & Faems, D. (2013). Exploring the impact of alliance portfolio management design on alliance portfolio performance. Managerial and Decision Economics, 34(3-5), 347-361. Nielsen, B. B. (2010). Strategic fit, contractual, and procedural governance in alliances. Journal
of Business Research, 63(7), 682-689.
Noseleit, F., & de Faria, P. (2013). Complementarities of internal R&D and alliances with different partner types. Journal of Business Research, 66(10), 2000-2006.
Penrose, E. T. (1959). The Theory of the Growth of the Firm. New York: Wiley.
Pisano, G. P., Russo, M. V., & Teece, D. J. (1998). Joint Ventures and Collaborative Arrangements in the Telecommunications Equipment Industry. In D. J. Teece (Ed.) , The
41
Porter, M. E., & Millar, V. E. (1985). How information gives you competitive advantage.
Harvard Business Review, 63(4), 149-160.
Puranam, P., & Srikanth, K. (2007). What they know vs. what they do: How acquirers leverage technology acquisitions. Strategic Management Journal, 28(8), 805-825.
Puranam, P., Singh, H., & Chaudhuri, S. (2009). Integrating acquired capabilities: When structural integration is (un) necessary. Organization Science, 20(2), 313-328.
Puranam, P., Singh, H., & Zollo, M. (2006). Organizing for innovation: Managing the coordination-autonomy dilemma in technology acquisitions. Academy of Management
Journal, 49(2), 263-280.
Ranft, A. (2006). Knowledge preservation and transfer during post-acquisition integration.
Advances in mergers and acquisitions, 5(1), 51-67.
Ranft, A. L., & Lord, M. D. (2000). Acquiring new knowledge: The role of retaining human capital in acquisitions of high-tech firms. The Journal of High Technology Management
Research, 11(2), 295-319.
Reuer, J. J. (2005). Avoiding lemons in M&A deals. MIT Sloan Management Review, 46(3), 15-17.
Riviezzo, A. (2013). Acquisitions in knowledge-intensive industries: Exploring the distinctive characteristics of the effective acquirer. Management Research Review, 36(2), 183-212. Sampson, R. C. (2007). R&D alliances and firm performance: The impact of technological
diversity and alliance organization on innovation. Academy of Management Journal, 50(2), 364-386.
42
Sarkar, M. B., Echambadi, R., Cavusgil, S. T., & Aulakh, P. S. (2001). The influence of complementarity, compatibility, and relationship capital on alliance performance. Journal
of the academy of marketing science, 29(4), 358-373.
Schilling, M. A. (2009). Understanding the alliance data. Strategic Management Journal, 30(3), 233-260.
Schoenberg, R. (2004). Dimensions of management style compatibility and cross-border acquisition outcome. Advances in mergers and acquisitions, 3, 149-175.
Schoenmakers, W., & Duysters, G. (2010). The technological origins of radical inventions.
Research Policy, 39(8), 1051-1059.
Schweizer, L., & Patzelt, H. (2012). Employee commitment in the post-acquisition integration process: The effect of integration speed and leadership. Scandinavian journal of
management, 28(4), 298-310.
Teixeira, R., Koufteros, X., & Peng, X. D. (2012). Organizational structure, integration, and manufacturing performance: A conceptual model and propositions. Journal of Operations
and Supply Chain Management, 5(1), 70-81.
Thompson, V. A. (1965). Bureaucracy and Innovation. Administrative Science Quarterly,
10(1), 1-20.
Vaara, E., Sarala, R., Stahl, G. K., & Björkman, I. (2012). The impact of organizational and national cultural differences on social conflict and knowledge transfer in international acquisitions. Journal of Management Studies, 49(1), 1-27.
Van Kooten, M. (2010, November 10). Services Top 100: The World’s Largest IT Services Companies (2010). Retrieved March 9, 2016, from http://www.servicestop100.org/services-top-100-the-worlds-largest-it-services-companies-2010.php
Walsh, J. P. (1988). Top management turnover following mergers and acquisitions. Strategic
43
Wang, L., & Zajac, E. J. (2007). Alliance or acquisition? A dyadic perspective on interfirm resource combinations. Strategic Management Journal, 28(13), 1291-1317.
Wassmer, U. (2010). Alliance Portfolios: A Review and Research Agenda. Journal Of
Management, 36(1), 141-171.
Wiklund, J., & Shepherd, D. A. (2009). The effectiveness of alliances and acquisitions: The role of resource combination activities. Entrepreneurship Theory and Practice, 33(1), 193-212.
Wuyts, S., & Dutta, S. (2014). Benefiting From Alliance Portfolio Diversity The Role of Past Internal Knowledge Creation Strategy. Journal of Management, 40(6), 1653-1674.
Zollo, M., & Singh, H. (2004). Deliberate learning in corporate acquisitions: post‐acquisition strategies and integration capability in US bank mergers. Strategic Management Journal,
44
APPENDIX A: REGRESSION COMPARISON
As indicated under the analytical method section, the number of observations becomes too little as soon as the moderating variable is included. This small number of observations makes it impossible to use the negative binomial regression (nbreg in STATA) since convergence cannot be reached and therefore the models yields no usable results. To resolve this problem, a normal linear regression (xtreg in STATA) has been conducted to find the results. However, a normal linear regression is usually not the appropriate model when the dependent variable is a count variable, since it violates the assumption of this method. To minimize the violations of the assumption it is important to transform the count variable into a continuous one – as far as possible. To achieve this, the dependent variable has been logarithmically transformed. This transformation creates values that are closer to the ones of a true continuous variable, however limitations still exists (i.e., the values will always be positive even though a real continuous variable should also be able to take negative values). Due to these limitations the results have to be interpreted with extreme caution. However, by comparing different regression and showing that the results are rather similar among all models, it becomes possible to more confidently interpret the results of the model. The results of these comparisons can be seen in table A1 and table A2 below.
45
comparison frame, the results can still produce valuable insights and give the transformed linear regression results more strength. Moreover, since the nbreg model is the method that should normally be used, it can provide great confidence in backing up the transformed linear regression results of the first two models.
As shown in the table A1, the results of the regression for both model 1 and 2 are very similar. The coefficients of the results are the same for all models (except for the constant variable, which can be expected, since this value only indicates the intersection point of the line of the model). More importantly, the significance levels of the results are also comparable. Most notably in fact are the results of the used transformed linear regression model that seem to be the most strict (i.e., they show the lowest significance levels overall). This indicates that the results for model 1 and 2 can be interpreted with more confidence, since other models show similar results or even higher significance.
Moreover, as shown in table A2, the results are a bit more difficult to interpret. In this case the results of the normal linear regression are the least significant. On the other hand, the results of the negative binomial regression (xtnbreg) are more significant, while the used transformed linear regression is somewhere in the middle. This indicates that the results for model 3, 4 and 5 can also be interpreted more confidently due to the fairly similar results, however these results should be interpreted with more caution than the model 1 and 2 results, since there are some more differences between the models.
48
APPENDIX B: SUPPLEMENTARY STATISTICS
49
Table B2
50
Figure B2
51
Figure B4
