Acquisition experience and acquisition performance : the moderating effect of performance feedback

Acquisition Experience and Acquisition Performance: The Moderating Effect of Performance Feedback

Master’s Thesis

Msc. Business Administration - Strategy Track

University of Amsterdam - Faculty of Economics and Business

Student: Harm Doddema (11152257)

Supervisor: Bernardo Silveira Barbosa Correia Lima, Msc.

STATEMENT OF ORIGINALITY

This document is written by student Harm Doddema, who declares to take full responsibility for the contents of this document:

“I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.”

Harm Doddema 11152257

TABLE OF CONTENTS

ABSTRACT ... 4

I. INTRODUCTION ... 5

II. LITERATURE REVIEW ... 8

Experiential learning ... 9

Acquisition experience ... 10

Performance feedback ... 13

III. HYPOTHESES DEVELOPMENT ... 15

Acquisition experience and performance ... 15

Performance feedback, experience and performance ... 16

Conceptual Framework ... 19

IV. DATA AND METHOD ... 20

Research setting ... 20

Dependent variable: acquisition performance ... 21

Independent variable: acquisition experience ... 23

Moderating variable: performance feedback ... 24

Control variables ... 25

Analysis ... 28

V. RESULTS ... 29

Descriptive statistics & correlation analysis ... 29

Regression analysis ... 31

Post-hoc analyses ... 39

VI. DISCUSSION, LIMITATIONS AND FUTURE RESEARCH ... 41

Discussion ... 41

Major findings ... 41

Contributions and managerial implications ... 43

Limitations and future research opportunities ... 45

Conclusion ... 47

Acknowledgements ... 48

VII. REFERENCES ... 48

VIII. APPENDICES ... 58

Appendix 1: Variable calculations and explanation ... 58

ABSTRACT

Studies regarding acquisition experience and subsequent acquisition performance have provided mixed empirical results and differing relationships between the two constructs. Building on the work of scholars in the field on management research, the moderating effect of performance feedback is added to this complex relationship, to identify a potential source for the mixed empirical results in the past. Performance feedback is examined based on firm performance, compared to social and historical aspirational levels. Furthermore, a distinction is made between performance above and performance below these aspirational goals. Based on an extensive literature review, there are several hypotheses proposed that require empirical testing. The two main hypotheses are that performance below aspiration will weaken the negative effect of acquisition experience on the performance of the acquisition, whilst performance above aspiration will strengthen the negative effect of acquisition experience on subsequent acquisition performance. The most prominent result of analysing 21.830 acquisitions by public firms from the US between 1985 and 2014 is that performance below aspiration significantly moderates the relationship of acquisition experience and performance, by weakening the negative effect of experience. This main finding suggests that firms performing below their aspirational goal apply acquisition experience differently in an acquisition, depending on their performance feedback, which leads to significantly different acquisition performance. Future research can use this study to advance and further conjoin the fields of experiential learning and performance feedback theory, as performance feedback could provide a possible contingency for the mixed empirical results in the past.

Key words: Acquisition experience; acquisition performance; performance feedback; experiential learning

I. INTRODUCTION

Experiential learning has been examined extensively in the past decades, as many scholars are interested in the way organizations learn and apply the obtained knowledge that stems from experience (Kolb, 2014). Several studies focussed on the direct or indirect effects of experience on organizational or acquisition performance (Baum & Dahlin, 2007; Ellis, Reus, Lamont, & Ranft, 2011; Haleblian & Finkelstein, 1999; Haunschild & Sullivan, 2002; Hayward, 2002; Jiménez-Jiménez & Sanz-Valle, 2011; Zollo, 2009). The claim that experience should enhance acquisition performance is largely supported and is also the central tenet in numerous studies regarding the performance perspective (Luo & Peng, 1999).

Outside of traditional industrial settings, the effects of organizational experience on performance have proved to be a complex relationship (Muehlfeld, Rao Sahib, & Van Witteloostuijn, 2012). This is particularly true for the effects of an organizations prior acquisition experience on an acquisition’s subsequent performance (Haleblian & Finkelstein, 1999). Linking acquisition experience and performance is complex, since even though the number of acquisitions increases every year and the knowledge about acquisitions elaborates, they still fail on a regular basis (Hayward, 2002). This has caused a debate in literature with mixed empirical results regarding the relationship between acquisition experience and the subsequent performance of the acquisition. The relationship is posed to be positive according to Barkema, Bell, & Pennings (1996), stating that prior entries of a firm in the same country positively influence the longevity and performance of acquisitions. However, other studies found non-significant (e.g. Zollo & Singh, 2004) and negative relationships (e.g. Uhlenbruck, Hitt, & Semadeni, 2006) between acquisition

experience and acquisition performance. Furthermore, there is also evidence for U-shaped (e.g. Haleblian & Finkelstein, 1999) and inverted U-U-shaped (Hayward, 2002) relationships between experience and performance. Given these mixed findings, it can be misleading to extrapolate outcomes from the manufacturing industry to the field of acquisitions (Hayward, 2002).

Even though there is an obvious debate in the literature, it is unclear which specific contingencies cause these mixed empirical outcomes and therefore, more research is needed (Barkema & Schijven, 2008). Haleblian and Finkelstein (1999) state that the mixed empirical outcomes are due to the finding that experience could seemingly be treated as a heterogeneous concept. This is crucial in determining the effect of for instance acquisition experience on subsequent acquisition performance. Even though experience heterogeneity is seen as a valuable contribution in explaining the mixed empirical findings, it could be missing an important contingency factor. Baum and Dahlin (2007) state that the role of organizational performance (e.g. improving performance beyond current levels) is implicit in explaining these differences in acquisition performance. Furthermore, the patterns of learning and subsequent actions of firms depend on the extent to which the firm performance differs from their aspiration levels (Cyert & March, 1963; Greve, 2003). These aspirational levels are divided in historical aspirations (e.g. a firms prior performance and social aspirations (e.g. performance of firms in the same industry) used by firms to compare themselves to others and assess performance (Gaba & Joseph, 2012). Performance feedback could therefore influence acquisition performance, as it is focused on the outcome of prior behaviour (e.g. experience) and the influence on future behaviours by organizations (Haleblian, Kim, & Rajagopalan, 2006). Since there are different empirical outcomes in terms of the effect of experience on

performance, this could indicate a moderating variable (Wu & Zumbo, 2008). Moreover, performance feedback should causally interact with the relationship between experience and performance, because it significantly affects the learning of firms and their behaviour (Iyer & Miller, 2008). Adding performance feedback variables as an influencer of this relationship should provide a possible contingency that will help explain the mixed results of studies in the past. By examining the interaction effect of performance feedback variables on the acquisition experience and subsequent performance relationship, it extends the work of Haleblian and Finkelstein (1999) to provide deeper knowledge how applying past experience affect acquisition performance. In sum, this has lead to the following central research question:

What is the influence of performance feedback on the relationship between organizational acquisition experience and subsequent acquisition performance?

This paper makes two main contributions to the existing literature. The first contribution of this study is that it helps to entangle which contingencies cause the mixed empirical findings regarding experience and performance, by incorporating knowledge from the field of performance feedback. Performance feedback could influence the relationship between acquisition experience and performance, since firms have different responses in respect to the feedback they receive (Greve, 2011). Subsequently, the potential moderating influence of performance above and below aspirational levels are tested separately for social and historical aspiration. By initiating a first move towards bridging two fields of research (e.g. experiential learning and performance feedback), the complex relationship between experience and performance could be better defined, as well as the opportunity to provide new research opportunities. The second contribution of this study is that, in keeping with

past research, it is testing the direct the effect of acquisition experience on subsequent acquisition performance, to add to the previous empirical findings of those studies.

This paper is organized in several sections, following this introduction. The literature review section will provide an overview of relevant literature and from this, the central concepts will be explained. In the subsequent sections, the development of the hypotheses will explained, the data and method will be described and the results are presented. In the final chapter, the major findings, limitations and opportunities for future research will be discussed.

II. LITERATURE REVIEW

In this section of the research paper, the relevant empirical papers in the selected fields of research and the current debates and definitions in these papers will be discussed. The first part is explaining organizational learning from experience (experiential learning), followed by the main findings on the effect of experience and the specific empirical setting of acquisitions is discussed. In conclusion, the rationale of studying the influence of performance feedback as a moderator variable is set forth.

Even though the relationship between experience and performance intuitively feels beneficial, the empirical outcomes and proposed types of the relationship are very divergent (Ellis et al., 2011). Since multiple scholars have all found a wide variety of relationship types, there must be other contingencies at hand (Barkema & Schijven, 2008). However, before going into these contingencies it is vital to understand the development of organizational learning-by-doing since the inception of the concept, since this is the foundation for studying the effect of experience.

Experiential learning

The way people, organizations and even industries learn or transfer knowledge has been a widely researched topic since the late 19th century (Ingram, 2002). This interest in the way learning and gaining experience works is therefore certainly not a new phenomenon, however even in this day and age, there is still a lot of research to be done in order to understand this principle more thoroughly. In experiential learning theory, learning is defined as the process whereby knowledge is created through the transformation of experience. Knowledge is a result of the combination of grasping and transforming experience (Kolb, 2014).

In the literature regarding experiential learning, there have been several approaches to examine this concept. For instance, early investigations on experiential learning focused on the behavior of certain subjects in an organizational context (Argote, Beckman, & Epple, 1990). These investigations revealed that the time that was required to perform a specific task declined at a decreasing rate as experience with the task increased (Graham & Gagne, 1940; Thorndike, 1898; Thurstone, 1919). This effect of a learning curve, observed for individuals and small groups is hereafter also found at the other levels of analysis (Alchian, 1963; Hirsch, 1956; Wright, 1936). Experience, in general, should therefore benefit the organization and should improve organizational performance, as early research on this topic has shown in multiple studies on the learning curve of an organization (Anzai & Simon, 1979; Gick & Holyoak, 1987). Experiential learning theory is highly interdisciplinary, as it addresses learning from experience in several fields of research (Kolb, Sternberg & Zhang, 2014). However, this study focuses on the implications for management research.

One of the implications of experiential learning theory is that organizations can learn, not only directly from their own experience, but also indirectly from the experience of other organizations (Argote, 2012). Moreover, organizational learning is a dynamic process, with a learning-by-doing approach, found in for instance cumulated production, and this is claimed to be the purest example of learning from direct (organizational) experience (Levitt & March, 1988). In empirical terms, positive returns to the accumulation of operating experience are one of the most robust and consistent findings in organizational learning (Argote, 2012). Outside of production and industrial settings, learning from experience is much more difficult to predict, and applying this experience is also more complicated (Haleblian & Finkelstein, 1999). This is due to the finding that unstructured tasks (for instance acquisitions) impede reinforcement learning and exacerbate the identification of links between current actions and observed outcomes (Denrell, Fang, & Levinthal, 2004). Applying experience outside of manufacturing settings in the complex field of acquisitions is therefore not (fully) explained by using experiential learning theory. Therefore, the effects of experience in acquisitions and other contingencies that influence this process should be further examined.

Acquisition experience

Organizations gain experience by engaging in acquisitions, transfer the learning taken out of these acquisitions and build on this learning in future acquisitions (Argote, 2012). More specifically, the experience gained from acquisitions can be defined as the principal mechanism by which firms learn from previous experience and apply this in following acquisitions (Hayward, 2002; Ravenscrarft & Scherer, 1987).

Since acquisitions are going more and more beyond industries, countries and even continents, there are a lot of differences within these acquisitions and the experience gained from it (Brouthers, 2013). This implies that there is a difference between one experience compared to the next experience, as acquisitions have a diverging nature (Haleblian & Finkelstein, 1999). Later research by March (2011) confirms that experience is in fact heterogeneous and that it can vary from ambiguous in nature to easily interpretable. However, if experience, and more specifically acquisition experience can be treated as heterogeneous, this has implications in the way organizational learning takes place and affects within a firm, therefore this needs more examination, since this heterogenetic aspect is not completely explained. Organizations that possess heterogeneous experiences should be in a better position to benefit from the variety of experience within the organization and this benefit should be reflected in higher-quality (acquisition) decision-making (Beckman & Haunschild, 2002).

There are multiple scholars in organizational learning and experience literature that build on this heterogeneity aspect of experience. For instance, the papers of Haleblian and Finkelstein (1999), Haleblian et al. (2006) and later the work of Ellis et al. (2011). In the paper of Haleblian and Finkelstein (1999), the transfer theory of learning is used as the explanation to deal with this heterogeneity in experience and tries to explain when organizations should apply or disregard past experience. Their paper more importantly implies that there is in fact the case of heterogeneity in experience, based on their findings. Furthermore, having no experience is posed to be better than having some experience, as you don’t make a mistake in the distinction between similar and dissimilar experience. In their explanation, the best performers

are either firms without any or with a significant amount of experience in order to appropriately discriminate between past acquisitions.

Acquisitions by firms provide an empirical setting to test the effect of potential influence factors on the relationship between organizational acquisition experience and subsequent acquisition performance. This is because of the findings that organizations learn from experience and that this influences subsequent acquisitions by firms (Hayward, 2002). When examining these acquisitions, the empirical picture with regards to the relationship between acquisition experience and acquisition performance is mixed (Agrawal, Jaffe, & Mandelker, 1992; Hayward, 2002; Loughran & Vijh, 1997; Meschi & Metais, 2006). Therefore, Barkema & Schijven (2008) stated that this relationship, especially in an acquisition setting, needs more research, as there is a lack of understanding on the contingencies that cause different effects of prior acquisition experience on acquisition performance. The heterogeneity in acquisition experience and the way this is applied is therefore a potential explanation of the mixed findings on this topic.

However, the performance feedback of acquiring firms is not taken into account in some previous papers (e.g. Ellis et al., 2011; Haleblian & Finkelstein, 1999) or is hasn’t directly been linked it to acquisition performance and acquisition experience as Haleblian and Finkelstein (1999) proposed (e.g. Haleblian et al., 2006). This is noteworthy, since performance feedback affects the behavior of organizations significantly (Iyer & Miller, 2008). Both fields of research should be connected as literature on organizational learning suggests that organizational behavior is guided both by routines that stem from experience and by performance feedback (Greve, 2011; Haleblian et al., 2006; Nelson & Winter, 1982). While the effect of both experience and performance feedback on performance have been examined

individually, the combined effect has rarely been tested in an empirical setting (Haleblian et al., 2006). The combined effect of (heterogeneous) acquisition experience and performance feedback could explain the mixed empirical findings in the past. Hayward (2002) stated that when managers of firms experienced great losses they are more likely to search for new acquisitions, instead of stop making acquisitions. This could also be argued for organisational performance, especially when it is falling below aspiration or is rising highly above aspiration. Therefore, the next paragraph will be used to define performance feedback, as well as a more detailed explanation why it needs to be linked with acquisition experience and subsequent acquisition performance.

Performance feedback

Performance feedback stems form the behavioural theory of the firm from Cyert and March (1963) and March and Simon (1958). In this behavioural theory of the firm specific goal variables that manager strive for are discussed, alongside of the feedback that is received in reaching this goal and an aspiration level to assess if the performance is in fact satisfactory (Greve, 2003). Performance feedback is defined as environmental and internal feedback on aspirational goals, set by the organization (Cyert & March, 1963; Greve, 2003). An aspirational level can be seen as the smallest possible outcome of a decision or action that would be perceived as a success by the respective decision maker(s) (Schneider, 1992). Performance is evaluated against these aspiration levels, which may be determined by the recent history of performance of the firm (e.g. historical aspiration levels) or by the performance of similar other firms (e.g. social aspiration levels) in a similar industry (Audia & Greve, 2006; Cyert & March, 1963; Greve, 1998). Managers and individuals within the firm use and react

to this performance feedback in their decision making process. Therefore it seems intuitive that performance feedback influences the application of acquisitions experience. Moreover, besides the differences of organizational experience regarding acquisitions, performance feedback can also potentially explain diversity in acquisition performance (Haleblian et al., 2006).

However, the combined effect of both experience, the transfer of this experience and performance feedback lacks empirical weighing (Haleblian et al., 2006). This is noteworthy, as firms are thought to demonstrate different responses to good and poor performance outcomes (March, 1981). The performance feedback approach is focused on the outcome of prior behaviours and their influence on future behaviours by organizations (Haleblian et al., 2006). Haleblian et al. (2006) propose that the theoretical arguments and findings on performance feedback developed in the broader strategic literature are also applicable to the context of acquisitions; and state that this needs to be examined further. Hayward (2002) poses acquisitions as a suitable setting to test certain claims. For these reasons, more research on acquisition experience, performance feedback and its actual influence on acquisition performance is needed (Haleblian et al., 2006; Muehlfeld et al., 2012). Extending previous research by incorporating the potential interaction between acquisition experience and performance feedback, makes it a valuable contribution to the literature. Furthermore, the theoretical predictions of Haleblian et al. (2006) should be tested in a different setting and time period, as their claim is that this could help develop a more generalizable theory of organizational learning. The next section discusses the proposed hypotheses based on this literature review and displays a conceptual framework.

III. HYPOTHESES DEVELOPMENT

Acquisition experience and performance

Barkema & Schijven (2008) state that there is a lack of understanding on the effect of (heterogeneity in) prior acquisition experience on acquisition performance. Thus, the effect of experience on performance still needs to be examined further, in keeping with past research. For instance, the study of Zollo (2009) found that acquisition experience weakened the effect of past success on the performance of a focal acquisition. In line with Haleblian and Finkelstein (1999), the paper of Zollo (2009) states that learning from experience becomes harder when experience is heterogeneous. Based on the assumption that acquisitions are in fact heterogeneous, it is possible that experience in acquisitions has a negative effect on acquisition performance. This is primarily due to the fact that heterogeneous experience could reduce the likelihood of “superstitious learning” and furthermore it can enhance learning by proving a deeper understanding of the parts that contributes to successful performance (Argote, 2012). This concept of superstitious learning occurs, according to McGrath (2011), when the connection between the cause of an action and the outcomes experienced aren’t clear, or are misattributed. Misapplying previous experience can therefore lead to lower firm performance, if the approach if the behavior and antecedent conditions don’t match. Moreover, Uhlenbruck et al. (2006) found that the change in market value of acquiring firms varies closely around zero on average and that many acquirers experience negative returns on their acquisitions. Therefore, it can be argued that acquisition experience has a negative impact on acquisition performance:

Hypothesis 1a: Acquisition experience will be negatively related to acquisition performance

In contrast to the negative view on the effect of acquisition experience on performance, there are also studies that propose positive effects of experience on performance. The study of Barkema et al. (1996) have found that prior experience of a firm positively influences the performance of acquisitions. Furthermore, when acquirers appropriately use routines and experience developed through prior acquisitions, the effect on the focal acquisition will be positive (Cormier & Hagman, 2014; Ellis et al., 2011). Fowler and Schmidt (1989) found that acquisition performance improved significantly for organizations that had previous acquisition experience. Moreover, acquisitions by firms that pursue a strategy where they actively seek to learn from experience can outperform the acquisitions of firms that follow less deliberate strategies (Lubatkin, 1983). Therefore, a contrary hypothesis is proposed, to test if acquisition experience is either positively or negatively related to acquisition performance:

Hypothesis 1b: Acquisition experience will be positively related to acquisition performance

Performance feedback, experience and performance

The performance feedback of a firm stems from comparing current performance to either social or historical aspirational levels. Firms can therefore perform above aspirational levels, or perform below these aspirational levels. This is important as firms respond differently to performing above or below aspirational goals (Greve, 2003). Categorizing these performance outcomes as successes and failures based on aspirational levels affects decision makers’ willingness to learn and act (Baum & Dahlin, 2007). In particular, this willingness depends on whether performance is distant from or near aspiration levels and above or below aspirations

(March & Shapira, 1992). Therefore, there can be argued that there is a distinction in the way past experience is applied in acquisitions for firms performing above and below relative aspirations (Baum & Dahlin, 2007).

Performance below aspiration can drive organizations and the policy makers toward strategic changes (Haleblian et al., 2006). Moreover, organizations performing below aspirations are expected to endeavor more vigorously to improve (Baum & Dahlin, 2007). Furthermore, Iyer and Miller (2008) examined that when performance drops below aspiration, the probability of acquisitions increases, indicating the search for strategic change. Haleblian and Finkelstein (1999) state that is best to have either a lot of experience or no experience at all to make acquisitions perform well. Therefore it can be argued that firms performing below their aspirational goal are likely to change their acquisition strategy and look for new acquisitions as a way to accomplish this.

Based on this finding, performing below aspiration will drive acquirers to approach types of firms or markets where they have no experience in, which is beneficial to acquisition performance (Haleblian & Finkelstein, 1999). In others words, performance below aspiration will affect the way that acquisition experience is applied. Acquisition experience is seen as a heterogeneous concept, and if performance drops below aspiration this will have a positive effect on the way firms apply and regard past experiences, given that they consciously regard the new acquisition as heterogeneous and thus dissimilar to past experience (Haleblian & Finkelstein, 1999). Concluding, performance below aspiration will weaken the effect of experience when it is found to be negative, or strengthen the effect of experience on acquisition performance when positive. This leads to the following hypothesis:

Hypothesis 2: Performance below aspiration will positively moderate the relationship between acquisition experience and acquisition performance

However, initial success with an activity or approach, leads to an increased likelihood of repetition of this activity or approach (Burgelman, 1991). This occurs due to the fact that the organization has gained the skills and capabilities associated with it over time and finds it less risky and more rewarding to repeat it than to try alternatives with which they have no organizational experience (Levitt & March, 1988). Therefore, firms performing above aspirational levels have little incentive to change their behavior or strategy (Iyer & Miller, 2008). This tendency is also know as choosing exploitation over exploration and is seen in many high performing organizations (Burgelman, 1991).

When a firm performs above aspiration, they therefore tend to repeat the previous activity, even though they might not have a lot of experience, because it was successful in the past. Moreover, since the activity was successful, the experience could regarded as similar when in fact is it dissimilar. This increases the likelihood of misapplying experience, as it is stated before that firms either need to have no experience or a large amount of experience to correctly distinct between experiences (Haleblian & Finkelstein, 1999). Hence the so called “hazard of acquisitions” increased for firms performing up until above aspirational performance (Iyer & Miller, 2008, p. 818). This also builds on the premise that acquiring firms that are similar from the acquirer create less value on short-term and perform lower compared to acquisitions where the target is not similar to the acquirer (Park, 2003). Therefore having a high performance relative to aspiration will negatively influence the way that experience is applied in acquiring:

Hypothesis 3: Performance above aspiration will negatively moderate the relationship between acquisition experience and acquisition performance

Conceptual Framework

The first hypotheses are in keeping with past research and state that acquisition experience is negatively (H1a) or positively (H1b) related to acquisition performance. The effect of experience on performance could vary due to for instance the heterogeneity of the experience and perceived difficulties in correctly applying it (Haleblian & Finkelstein, 1999).

Hypothesis 2 is building on past research by incorporating performance feedback variables into the way experience is applied when acquiring. In this hypothesis, there is argued that performance below social and historical aspiration, will positively moderate the relationship between acquisition experience and subsequent performance. This is mainly due to the premise that underachieving organizations will try to upturn this and therefore potentially acquire firms where they lack experience, which is beneficial to acquisition performance according to Haleblian and Finkelstein (1999). To provide a visual representation, these hypotheses are all displayed in Figure 1: Conceptual Framework.

The final hypothesis is also building on the potential influence of performance on the way experience is applied in acquisitions. Based on a lack of incentive to change behavior for high performing firms, performing above aspirational levels will lead to a stronger negative effect on the way experience is applied in acquisitions. This is due to the central tendency of exploitation, seen in most well performing organizations (Burgelman, 1991). This behavior increases the opportunity of falsely regarding experience as similar, where it is rather heterogeneous and dissimilar

(Haleblian & Finkelstein, 1999). In conclusion, performance above aspiration has a negative impact of on the effect of acquisitions experience on performance.

- (H1a) + (H1b)

+ (H2) - (H3)

Figure 1: Conceptual Framework

IV. DATA AND METHOD

Research setting

The paper of Haleblian and Finkelstein (1999) examined completed majority acquisitions in the manufacturing setting and therefore, so will this paper to correctly build on their work. The sample in this paper consists of 21.830 observations of completed acquisitions by public firms originating from the United States. The acquisitions are completed between 1985 and 2014 by the acquiring firm. Each acquisition in the sample is from a company, operating in the manufacturing industry, with SIC codes ranging from 2000 – 3999. The deal value is ranging from under $1 million to over $90 billion; there are no restrictions in deal size, as too many observations would be lost. The sample consists of firms operating in the whole manufacturing industry, to test the relationship between acquisition experience and performance across multiple industries (Hayward, 2002).

Acquisition experience Acquisition Performance Performance below aspiration Performance above aspiration

The sample is selected in line with the empirical criteria of Haleblian and Finkelstein (1999), as this paper tries to build on their work in the field of organizational acquisition experience and subsequent performance. The focus is on completed majority acquisitions; therefore the acquisitions in the sample are limited to those that are likely to make a strong and noticeable impact on the market value evaluation (Chatterjee & Lubatkin, 1990). The data is collected via two databases: Securities Data Corporation (SDC) data on acquisitions and Compustat financial data for measuring the financial aspects from the organizations in the sample. All the data comes from firms in the before mentioned SIC 2000 – 3999 manufacturing industries, as this is a suitable empirical setting (Hayward, 2002). Via GVKEY identifiers in Compustat, the information on each unique firm is acquired, in SDC the data is obtained through historical CUSIP company identifiers. Since SDC data is in calendar years and Compustat data in fiscal years, adjustments were made to harmonize the data set. All relevant constructs for the analysis will be explained in the upcoming paragraphs, as well as several control variables to stabilize other influences.

Dependent variable: acquisition performance

The performance of an event like an acquisition can be assessed by the stock price change during a period surrounding the event (Haleblian & Finkelstein, 1999). This price change is sometimes called an abnormal return and is calculated as the difference between the observed return for a security and the predicted or normal return for the same security (Finkelstein & Haleblian, 2002). Hence, the impact of an event is measured by the part of the return that is unanticipated by an economic model of the so called anticipated normal returns (Haleblian & Finkelstein, 1999).

In the setting of acquisitions, the conventional event study methodology is to measure the cumulative abnormal returns (CAR) (Hayward, 2002). In this study CAR are calculated over an observation window of five trading day before through five trading days after the announcement of the acquisition event (Hutzschenreuter, Kleindienst, & Schmitt, 2014; Finkelstein & Haleblian, 2002). This time span of five days prior and five days post the focal acquisition provides information on the stock market reaction of this acquisition and its anticipated effects. In the literature, ex ante measures of acquirer CAR have a high correlation with ex post measures of acquisition performance, demonstrating that event study methodology has predictive validity (Haleblian et al., 2006). Furthermore early empirical work in the financial sector has found a strong positive relationship between the pre-acquisition announcements and the post-acquisition increase in operating cash flows of the organization (Healy, Palepu, & Ruback, 1992). In later research, Rappaport & Sirower (1999) found that acquirer returns at the time of an acquisition announcement were representative of long-term performance. Furthermore, Finkelstein & Haleblian (2002) also indicate that CAR are valid indicators of acquisition performance.

Even though there are other measures available for acquisition performance in management research (e.g. various forms of return on assets), CAR is posed to be the most suitable for this paper, as it is extending the study of Haleblian and Finkelstein (1999). Furthermore, the effect and outcome of an acquisition are not in the financial statements of an acquirer directly, with time-frames ranging from six months up to to three years before the acquirer truly realizes the effects of a certain acquisition (Haleblian et al., 2006). Moreover, Hutzschenreuter, et al. (2014) state that organizations are therefore posed to acquire firms for non-financial reasons (e.g. research and development focused, portfolio building or geographical expansion).

These types of acquisitions could undermine the acquirers' performance and thus may be considered failures from an accounting standpoint, but they are not if the acquirers achieve their strategic objectives (Haleblian, Devers, McNamara, Carpenter, & Davison, 2009). In conclusion, CAR have been the most suited and frequently used analytical approach for measuring acquisition performance (e.g. Haleblian et al., 2006; Capron & Pistre, 2002; Finkelstein & Haleblian, 2002; King, Dalton, Daily, & Covin, 2004; Rhoades, 1994)

Independent variable: acquisition experience

Acquisition experience is a commonly researched topic in the field of management research. Therefore, consistent with Ingram and Baum (1997), Haleblian and Finkelstein (1999), Finkelstein and Haleblian (2002) and Ellis et al. (2011), this study measures organizations' acquisition experience by the number of prior acquisitions the firm made up to the focal acquisition. Since the benefits of prior acquisition experience may not increase monotonically with the amount of experience that an organization accumulates because old experience becomes less useful over time, this needs to be taken into account in the operationalization (Argote, 1999; Barnett, Greve, & Park, 1994; Hayward, 2002). Therefore, in order to specify and operationalize the construct, acquisition experience is defined as the total number of acquisitions made by a firm from the five years prior to the acquisition of interest (Cording, Christmann, & King, 2008).

The data is collected through for the time period 1980 – 2014 from the SDC merger database to calculate this variable. This way of measuring and operationalizing the variable is used extensively, posing it to be a valid method of operationalizing acquisition experience (Field & Mkrtchyan, 2017). In conclusion,

following prior studies that have examined acquisition experience (e.g. Haleblian & Finkelstein, 1999; Haunschild & Sullivan, 2002; Hayward, 2002), this construct is measured as the total number of prior acquisitions that an organization made in the five years up to the focal acquisition.

Moderating variable: performance feedback

There are various financial performance measures that may be relevant to assess the performance of organizations relative to aspiration of the organization, for instance return on assets (ROA), return on sales (ROS) and return on equity (ROE) can be used (Iyer & Miller, 2008). However, ROA is chosen over other available financial indicators in most empirical papers (Gaba & Joseph, 2012). This is due to the influence of a firms’ mix of equity and debt and the deliberate sales strategy on ROE and ROS (Iyer & Miller, 2008). The ROA of an organization is defined as the profit (e.g. net income) as a percentage of total assets and can be compared internally or externally (Tallon, 2010). In line with Greve (2003), two different proxies for assessing performance to aspiration are used in this study: historical and social aspiration. Historical aspiration is based on the prior performance of the organization and social aspiration is based on the performance of the average firm in the same industry on a 3-digit SIC-code level (Iyer & Miller, 2008). The decision to prefer the 3-digit SIC-code rather than the 4-digit SIC-code level is made, to provide better comparable observations in the analysis. Since all industries are used between the 2000 - 3999 SIC-codes, the 4-digit SIC level would provide too many small groups to properly analyse the performance compared to social aspirational levels.

Performance is operationalized through the net income divided by the total assets of the organization in the year of an acquisition. The social aspiration is the

average performance of a firm in the same industry from the 3 years prior to the acquisition. For historical aspiration, the financial performance of the organization in the 3 years prior to the year of the focal acquisition is used to compare current performance to. Performance feedback is computed using the performance of the firm minus the aspirational level based on the average firm performance in the same industry (social aspiration) or its own prior performance in the past three years (historical aspiration).

All the financial data that is used to compute these two variables is obtained from the Compustat database. In line with Audia & Greve (2006) and to overcome the situation where organizations vary in the performance below or above aspiration in the two variables, a separate analysis is executed for both variables.

Control variables

The main variables and their hypothesized effects are displayed in the conceptual model of this paper; however there still could be other variables that influence acquisition performance. Therefore, to overcome the outside effects of these other factors, there is controlled for attitude, relatedness between target and acquirer, acquirer slack (e.g. potential and observed) and period effects in line with Haleblian and Finkelstein (1999), Hayward (2002), Haleblian et al. (2006) and Ellis et al. (2011).

Attitude - The way acquirers approach the target firm can differ per

acquisition. The attitude of an acquisition can vary from friendly, neutral or hostile. Acquisitions are defined as friendly when the management of the target agree to the acquisition offer of the acquiring firm (Capron & Pistre, 2002). In hostile acquisitions,

potential targets are not agreeing to the offer of the acquiring party, or they might not even be given notice of the attempted acquisition. In this case, the target may take actions, such as adopting a certain defence strategy or they try to arrange an acquisition by a third party. This strategy is chosen to make it less likely for acquirers to succeed (Brickley, Coles, & Terry, 1994; Haleblian & Finkelstein, 1999; Mallette & Fowler, 1992). Moreover, if acquisitions are hostile, this may have a negative effect on the returns of the acquirer, since the share price and premiums might increase due to the interest by multiple bidders (Uhlenbruck et al., 2006). Thus, in conclusion, hostile acquisitions may be inversely related to acquisition success (Haleblian & Finkelstein, 1999). Two dummy variables are created to control for acquisition attitude, by coding either hostile of friendly acquisitions as 1 and the other variables as 0. Attitude did not significantly influence one of the final models in the analysis, therefore it is not reported in the final analysis section.

Relatedness of Target to Acquirer – The similarity between the target and the

acquirer is included as a control variable, since it measures if the target operates in a similar or dissimilar industry. The outcome of this variable is computed as a dummy variable, containing a simple comparison of the 4-digit SIC code of the target and the acquirer. Thus, if the SIC code of the acquirer matches the SIC code of the target, the acquisition is coded as 1, if the SIC codes do not correspond, they are coded as 0. It is important to distinguish between acquiring a related firm compared of a unrelated firm, since there is evidence that unrelated acquisitions create less value and perform lower compared to related acquisitions (Park, 2003).

Acquirer slack - Hitt, Harrison, Ireland, & Best (1993) argued for the

existence of various potential influences that slack (resources) can have on acquisition performance. They argued that with greater amounts of acquirer slack, not only is less

financing necessary but also debt financing is less cost-intensive and therefore easier to obtain. Consistent with the expectations in their model, they found that slack is associated with successful acquisitions, thus this should be taken into consideration (Haleblian & Finkelstein, 1999). Accordingly, acquirer slack resources is being controlled for, which was measured as the ratio of an acquirer's cash on hand that was not being used for operations to the total assets of the acquirer (Haleblian et al., 2006). Moreover, there is also controlled for potential slack of the acquiring firm, since this influences their ability to get additional loans in the financial market (Bourgeois & Singh, 1983). The potential slack is operationalized as the total debt to equity ratio of the acquiring firm.

Period effects - In line with Haleblian and Finkelstein (1999), period effects

need to be controlled for, as there are potential effects of macroeconomic conditions on acquisition activity. This is executed by entering the years as a set of dummy variables into the models. Dummy variables were created for twenty-nine of the thirty years in the sample (= 1 for a particular acquisition year and = 0 for all other years); 1985 was the first year in the sample and therefore the omitted category. These dummy variables were all not significant, and results were substantially unchanged when they were included in the analysis. Therefore, the control variable period effect isn’t included in the final analysis and the regression models.

In Appendix 1: Variable calculations and explanations there is an overview provided of the calculations for the financial variables that are used in this study. Furthermore, the rationales behind all the variables that are used in this study are explicated to provide more insight in the variables.

Analysis

Before the results can be displayed, several preliminary checks have to be executed, to inspect the quality of the data. Therefore data is checked for outliers and missing values, by plotting and summarizing the main variables in the model. This needs to be done, since outliers can cause the models to be biased because they affect the values of the estimated regression coefficients (Field, 2013).

Based on recommendations by Aiken, West, & Reno (1991) and following other empirical papers (e.g. Jansen, Van Den Bosch, & Volberda, 2006; Lim & McCann, 2013) the main variables used for the interaction are mean-centered before creating the regression variables, to mitigate any potential multicollineartity problems. This was a historically valid method to overcome multicollineartity, but is currently under debate (Disatnik & Sivan, 2016). However, since it helps interpreting the outcomes and provides a clearer display in the regression outputs, this method is chosen. To examine potential multicollineartity, the variance inflation factors (VIF) were calculated for each of the regression models. The maximum mean VIFs of all the regression models (8.02) was below the rule-of-thumb cut-off of point of 10 (Jansen et al., 2006; Neter, Wasserman, & Kutner, 1990).

Since the data is pooled cross-sectional and the dependent variable is a continuous variable, ordinary least squares (OLS) regression analysis is used as the estimation procedure to test the hypotheses. Robust standard errors are used since Angrist & Pischke (2008) and others report that this could help overcome the influence of heteroskedasticity or unequal variances in the standard errors and the analysis.

V. RESULTS

In this section, the results of this research are reported. Starting with the descriptive statistics for the variables of the study. These variables are presented to provide an overview of the data and to review the number of observations, their mean and the respective standard deviations. Subsequently, a correlation analysis is performed to display and report the (significant) correlations among the variables. Finally, several regression analyses are carried out to test the hypotheses that were formulated in the theoretical part of this paper, followed by a plotted representation of some of these results.

Descriptive statistics & correlation analysis

Table 1: Descriptive statistics & correlation presents the main statistics and

correlations for the variables that are used in the sample of this study. To test the relationships between the variables, pairwise correlations coefficients are used. In the first column of Table 1, the relationships between the dependent variable (e.g. acquisition performance) and the independent, moderating and control variables are displayed. After which this is done for the independent, moderating variables and control variables themselves.

The sample consisted of 21.830 observations from 1985 to 2014. As discussed in the method section of this paper, there are firms from a range of manufacturing industries between SIC 2000 and 3199. In the sample there are four major two-digit industry segments: Industrial and Commercial Machinery and Computer Equipment (17,10%), Electronic and other Electrical Equipment and Components, except Computer Equipment (16,73%), Chemicals and Allied Products (14,42%) and

Measuring, Analysing and Controlling Instruments (14,14%). All other two-digit industry segments are represented below 10% in the sample; see Appendix 2: Industry

Segments for a complete overview of those segments.

When assessing the correlation between the main dependent and independent variables of acquisition performance and acquisition experience several things stand out. The correlation is negative and significant, as expected beforehand, yet the strength of the correlation is weak (r=-.038). However it must be emphasized that although some reported correlations might appear relatively low, in finance studies in which abnormal returns are the dependent variable, low R-squares and correlations are typical (Haleblian & Finkelstein, 1999). The correlations that are displayed in Table 1 have four other noteworthy points, which are worth further clarification. Considering these correlations must be done with caution, since outside influence factors on these respective correlations are not taken into consideration (Disatnik & Sivan, 2016).

Firstly, looking at the correlations between the acquisition performance and the control variables, only one of the control variables has a significant effect: firm size (r = -.09). Even though is has a significant effect, it a weak (<. 30) and negative effect on the dependent variable. In contrast, the similarity of the target and acquirer and both the potential and the observed slack variables did not have a significant correlation with acquisition performance.

Secondly, the correlation between acquisition performance and the moderating variables of performance feedback above aspiration is both social (r = -.022) and historical (r = .036) weakly correlated with the dependent variable, with a high significance (p < 0.01). Performance below aspiration is not significantly correlated to acquisition performance, neither socially nor historically.

Thirdly, acquisition experience is significantly (p < 0.01) correlated to the acquisition performance, but the relationship is weak and negative (r = -.045). Acquisition experience is furthermore significantly correlated to the performance feedback variables, except for below historical aspiration (p = .116). Moreover, all control variables are significantly (p < 0.01) correlated to acquisition experience, except for potential slack, which is not significant. The strongest correlation of acquisition experience is a medium correlation with firm size (r = .4350).

Finally, the moderating variables for performance feedback are significantly correlated (mostly p < 0.01, some p < 0.05) with almost all control variables and each other. Therefore, the interaction between historical and social aspiration and acquisition experience are tested in separate models to avoid distorted parameter estimates and to test for the earlier hypothesized effects (Hutzschenreuter et al., 2014; Iyer & Miller, 2008). For a full display of correlations between all variables, see

Table 1: Descriptive statistics & correlation.

Regression analysis

In Table 2: Regression model for acquisition performance, an overview of the estimation results is presented. Model 1 is a control model containing only control variables and their effect on acquisition performance. Model 2 is a base model that consists of only the independent variable (acquisition experience) and the control variables. Since including performance above and below aspiration in one model for social and for historical aspirational levels does not influence the coefficients or significance levels, they are presented in a combined model, instead of separately.

Model 3 examines the relationship between acquisition experience and performance only taking into account performance above and below social aspiration. Building on this, Model 4 incorporates the interaction between acquisition experience and respective performance feedback on both above and below social aspiration. Model 5 is similar to Model 3, but it makes use of the performance compared to historical aspiration variables. Moreover, Model 6 is the final model and similar to Model 4; incorporating interaction effects between acquisition experience and performance feedback variables based performance below and above historical aspiration.

Hypothesis 1a states that acquisition experience will have a negative effect on acquisition performance. Model 2 displays that the direct relationship between acquisition experience and performance is negative (t = -.4183) and significant (p < 0.01). This suggests that if acquisition experience increases, acquisition performance decreases, even though it is a slight decrease, it is significant. Therefore, hypothesis 1a is supported in the overall analysis. Subsequently, Hypothesis 1b states that acquisition experience will have a positive effect on acquisition performance. Based on preliminary outcomes and the discussed negative coefficient of Model 2 (t = -.4183, p < 0.01), Hypothesis 1b did not find support in the overall analysis. Thus, these findings indicate that acquisition experience is negatively related to acquisition performance.

Hypothesis 2 postulates that performance below aspiration will have a positive influence on the relationship between acquisition experience and acquisition performance. To find support for this hypothesis, the regression outputs need to be analysed. In Model 3, the coefficient of performance below social aspiration is highly significant (p < 0.01) and positively related to acquisition performance (t = 1.308).

When considering Model 5, the coefficient of performance below historical aspiration is also highly significant (p < 0.01) and positive (t = 1.544), providing preliminary evidence for a significant effect on acquisition performance. To correctly assess the moderating effect, the interaction coefficient of performance below social aspiration and acquisition experience needs to be examined. When considering Model 4 and there is a significant (p < 0.05) positive (t = 1.741) coefficient. Moreover, in Model 6 there is also a significant (p < 0.05) and even stronger positive interaction coefficient (t = 3.427) for performance below historical aspiration and acquisition experience. Furthermore, the negative effect of acquisition experience on acquisition performance decreases when performance below aspiration variables are included in the model. Thus, incorporating performance below both historical and social aspirational levels has a significant effect on the relationship between acquisition experience and performance. This can be concluded from both the interaction coefficients with acquisition experience and the less negative direct effect of acquisitions experience on subsequent acquisition performance. In conclusion, Hypothesis 2 is supported in the overall analysis, showing that performance below aspiration is in fact a moderator of the relationship between acquisition experience and performance, since it weakens the negative effect of acquisition experience on performance.

To aid in interpreting the results of the analysis, Figure 2 and Figure 3 provide the interaction effects of performance feedback based on both social and historical aspiration and their effect on the relationship between the acquisition experience and subsequent performance. Since Hypothesis 2 only uses the performance below the aspirational levels, the spline variable is used, that reports values only below either social or historical aspiration. The plots are made based on Aiken and West simple slope analysis test, to assess if low and high simple slopes of performance feedback

differ from 0. Low values of performance feedback are operationalized as one standard deviation (SD) below the mean (e.g. performance far away from the aspirational level) and high values of performance feedback are defined as one SD above the mean (e.g. performance close to the aspirational level) of both performances minus social (Figure 2) and historical (Figure 3) aspiration.

For social aspiration, the effect on acquisition experience is negative and significant for low values of performance below the social aspiration level (t = -.8175; p < 0.01). In line with this, the high values of performance below the social aspiration level are less negative and also significant (t = -.6023; p < 0.01)

Moreover, when repeating this analysis for historical aspiration, the effect on acquisition experience is also negative and significant for the low values of performance under the historical aspiration level (t = -.8171; p < 0.01). For high values of performance below historical aspiration, the effect is becoming less negative and significant (t = -.6553; p < 0.01). Therefore, when the simple slopes of both Figure 2 and Figure 3 are examined, the higher values of performance below aspiration improve the effect of acquisition experience on performance. The figures display that there is a significant impact of performance feedback on the relationship between acquisition experience and performance. However, for historical aspiration, the effect is stronger, compared to social aspiration.

Figure 2: Moderating effect of performance feedback based on social aspiration on the relationship between acquisition experience and acquisition performance

Figure 3: Moderating effect of performance feedback based on historical aspiration on the relationship between acquisition experience and acquisition performance

10,7 10,8 10,9 11 11,1 11,2

Low Acq. Experience High Acq. Experience

A cq u is iti on P er for man ce

Social Aspiration

Low Acq. Experience High Acq. Experience

A cq u is iti on P er for man ce

Historical Aspiration

Finally, Hypothesis 3 stated that performance above aspiration would negatively influence the relationship between acquisition experience and acquisition performance. Therefore, since the relationship between acquisition experience and subsequent acquisition performance is posed to be negative in line with Hypothesis 1a, it would lead to a more negative effect of previous experience on performance.

When considering Model 3, the coefficient of performance above social aspiration is significant (p < 0.05) and negatively related to acquisition performance (t = -11.59). In Model 5, performance above historical aspiration does not have a significant effect on acquisition performance directly. To find support for this hypothesis, the interactions between acquisition experience and performance above social and historical aspirations are the decisive indicators. In Model 4, the interaction coefficient of performance above social aspiration and acquisition experience is not significant. Furthermore, in Model 6 there is also no significant interaction effect based on the three significance intervals for performance above historical aspiration and acquisition experience. Therefore, incorporating performance above both historical and social aspirational levels does not have a significant (moderating) effect on the relationship between acquisition experience and performance. However, it is noteworthy that performance above social aspiration had a significant negative effect on acquisition performance directly.

All in all, Hypothesis 3 is not supported based on the different models in the analysis. Since performance above aspiration does not seem to causally interact with the acquisition experience and subsequent performance relationship, there is no need for plotting the interactions.

Table 1: Descriptive statistics & correlation

N Mean S.D. 1 2 3 4 5 6 7 8 9

1. Acquisition performance 21600 .0092 .1143

2. Acquisition experience 17411 6.116 7.843 -.0381**

3. Performance above aspiration (social) 21621 .1154 .1507 -.0224** .0426**

4. Performance below aspiration (social) 21621 -.0407 .7081 .0017 .0165* .0423**

5. Performance above aspiration (historical) 21169 .0319 .1176 .0360** -.0597** .0667** -.0092

6. Performance below aspiration (historical) 21169 .6132 .7068 .0053 .0119 .0273** .9897** .0184**

7. Firm size 21618 8.313 2.016 -.0900** .4350** .0240** .0874** -.2023** .0695**

8. Relatedness of target to acquirer 21830 .2409 .4276 .0049 -.1075** .0455** -.0132 .0182** -.0162* -.0771**

9. Acquirer observed slack 21039 .0924 .1117 .0015 -.0721** .1238** -.0368** .1807** -.0286** -0.3249** .0497**

10. Acquirer potential slack 21782 .5853 19.44 -.0026 -.0018 -.0076 .0021 -.0041 .0028 0.0140* -.0086 -.0056

** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed) + Correlation is significant at the 0.10 level (2-tailed

Table 2: Regression model for acquisition performance1

Social Aspiration Historical aspiration

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

Variables Control Base Main Interaction Main Interaction

1. Acquisition experience -.4183** _{(.1028) (.1053)} -.1901 _(.1235) -.1647 _(.1029) -.1147 _(.1944) -.1707

2. Performance above aspiration -11.59* _(5.715) -12.35* _{(5.734) (10.93)} -10.85 _(8.527) 12.08

3. Performance below aspiration 1.308** _{(.3706) (7.557)} 10.56+ 1.544** _(.3580) 18.75** _(6.902)

4. Performance below aspiration x acquisition experience _(1.490) 1.741* _(1.357) 3.427*

5 Performance above aspiration x acquisition experience _(.8740) -.1053 _(1.809) -.1017

6. Firm size -5.697** _(.6520) -4.427** _(.6513) -4.376** _(.6551) -4.436** _(.6558) -4.539** _(.6580) -4.504** _(.6526)

7. Relatedness of target to acquirer _(1.966) -.1483 _(1.961) .2369 _(1.971) .6038 _(1.972) .6256 _(1.961) .2787 _(1.959) .3322

8. Acquirer observed slack _(12.88) -30.93 _(13.16) -3.812 _(13.51) -1.177 _(13.50) -4.724 _(13.43) -.1959 _(13.14) -3.346

9. Acquirer potential slack -.00076 _{(.0222) (.0234)} -.0008 -.00013 _(.0228) 0.00011 _(.0229) -.00010 _(.0233) 0.00010 _(.0236)

N 20609 16580 16419 16419 16576 16576

R-square .0060 .0080 .0080 .0079 .0078 .0084

Model F 19.82** 22.58** 12.17** 10.63** 12.14** 10.08**

** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed) + Correlation is significant at the 0.10 level (2-tailed) Standard errors are reported in parentheses

Post-hoc analyses

Besides testing for correlations among variables and test the coefficients in multiple regression analyses, there are additional post-hoc analyses conducted to improve the robustness of the results.

Firstly the analysis was replicated using a robust regression with standard errors clustered at the firm level by using the GVKEY company identifier. After clustering, the robust linear regression was conducted for the same 6 models as those used in the regression analysis in the paragraphs before. This type of post-hoc analysis is done, since residuals can be correlated across firms or across time, and therefore the OLS SEs can potentially be biased (Petersen, 2009). In the robust regression with clustered standard errors, the models did not report any significant chances in terms of coefficients; R-square and F-value of the model. The SEs were clustered at the firm level, varying from a maximum of 3210 clusters and a minimum of 2212 unique clusters, when re-performing the analysis. However, no significant changes in the outcomes are reported, which contributes to the robustness of the analysis.

Secondly, the different possible windows for computing CAR where assessed. Building on Haleblian and Finkelstein (1999) 5 days prior and 5 days post the focal acquisition is used as the window in the analysis. Other papers (e.g. Hayward 2002, Finkelstein & Haleblian 2006) use different windows for computing abnormal returns. This is defendable, since a longer window for computing CAR, could cause other factors to influence these returns as well (Capron & Pistre, 2002). Therefore, the analysis is repeated, using CAR windows of 0 to 1 day post acquisition, 0 to 2 days post acquisition and 2 days prior until 2 days post acquisition.