THE EFFECT OF FAULTLINES WITHIN CORPORATE BOARDS ON COMMERCIAL INNOVATION PERFORMANCE: ARE FAULTLINES GOOD OR BAD?
University of Groningen Faculty of Economics and Business
by
Robbert Mos – S3249468
MScBA Thesis Strategic Innovation Management
Date
14-07-2018 Word Count: 14316
Supervisor: Dr. Thijs L.J. Broekhuizen
Co-Assessor: Dr. Pedro M.M. de Faria
2 Abstract
In this research, we study the effect of faultline strength on commercial innovation performance post-M&A. Faultlines are hypothetical dividing lines dividing a group into sub-groups, which are hypothesized to reduce commercial innovation performance via hindering cross-sub-group communication. Next, contextual conditions, like board size and board independence, are hypothesized to impact the degree to which faultlines reduces commercial innovation performance. The methodology section introduces the ratio between Tobin’s Q (investor expectations about a firm’s innovation potential) and R&D’s expenditures (R&D intensity) as a new measure for commercial innovation performance. The results show a negative impact of faultlines on commercial innovation performance. Additionally, results showed that contextual factors moderate this relation in some cases. As board size increases the negative effect of faultline strength on commercial innovation performance at low levels of faultline strength. But when faultline strength increases, smaller boards experience a stronger negative effect of faultline strength on commercial innovation performance. Additionally we found that board independence attenuates the negative effect, although only in the case of small boards. This study adds to the existing literature by assessing the effect of faultlines in deviant context, boardroom, compared to most other upper-echelon studies. With the addition of the measurement of contextual factors, board size and board independence. And lastly, we measure the dependent variable, innovation performance, with a newly created measure.
Keywords: Faultlines, Diversity, Commercial Innovation Performance, Tobin’s Q, R&D intensity, Boards;
3
Table of content
Abstract ... 2
1. Introduction ... 4
2. Theoretical framework ... 6
2.1. Conceptual Model ... 6
2.2. Literature review ... 7
3. Methodology ... 11
3.1. Research setting... 11
3.2. Research method and sample ... 12
3.3. Measures ... 13
3.4. Preparing dataset... 18
4. Results ... 21
5. Discussion ... 24
5.1 Theoretical implications ... 25
5.2 Managerial implications ... 27
5.3 Limitations and further research ... 28
6. References ... 31
Appendix ... 36
Appendix A – Calculations faultline strength ... 36
Appendix B – outcome Winsorization ... 44
Appendix C – Winsorize output ... 45
Appendix D – Normality test output ... 48
Appendix E - Variance Inflator Factor test results ... 51
4
1. Introduction
Innovation is highly important for firm survival (Bocquet, Le Bas, Mothe & Poussing, 2017). The driving force behind innovation is the top management team, who are responsible for strategic decision making and governing the firm. The management team´s decision making process is thus of great importance (Flanagan, Kreuze & Smith, 2004; Harrison, Hart & Oler, 2014; Makri, Hitt & Lane, 2010). Because of their power, they set the fortune of the firm (Bezrukova, Jehn, Zanutto & Thatcher, 2009; Costanzo & Di Domenico, 2015; Mizruchi, 1983).
The composition of a board affects the performance of a firm (Bantel and Jackson, 1989;
Carpenter, Geletkaycz & Sanders, 2004), as the diversity in the composition can result in conflicts or other communication problems. This makes the composition interesting to study when looking for an explanation for innovation performance.
While some theorists argue that group diversity has a positive effect on performance, a large majority argues that diversity, especially when it is high, can have negative effects on performance (De Clercq, Thongpapanl, & Dimov, 2009; Homan, Van Knippenberg, Van Kleef
& De Dreu, 2007; Marcel, Barr, & Duhaime, 2011; Siciliano, 1996). The opposing perspectives mark two streams in research: diversity and faultlines.
The faultline perspective is part of the upper-echelon theory and assumes that differences in the socio-demographic composition of team members to determine performance.
The term faultlines originates from geography, where it refers to the intersection of two tectonic plates (Gu, Nakagawa, Saito & Yamaga, 2012). Faultlines are “hypothetical dividing lines that may split a group into sub-groups based on one or more attributes” (Lau & Murningham, 1998). Faultlines mark cracks in the composition of a group, where they potentially split or cause a collision to occur when the two sides do not move together. When activated, faultlines cause sub-groups to emerge within a team (Meyer, Glenz, Antino, Rico & González-Romá, 2014; Lau & Murnighan, 1998). The upper-echelons perspective assumes that an individual board member’s objective diversity is associated with their socio-demographic characteristics (Carpenter et al., 2004). As such, research mostly uses differences in socio-demographic attributes: gender, age, education, nationality etc.; as an explanation for opposing perspectives (Lau & Murnighan, 1998).
Faultline theorists have used group composition to explain difference in team performance (Bezrukova et al., 2009; Carpenter, Geletkaycz & Sanders, 2004; Siciliano, 1996;
Van Knippenberg, Dawson, West & Homan, 2011). When the level of diversity rises, so does the diversity of the objectives (Siciliano, 1996). Opposing objectives can lead to problems in
5
decision making (De Clercq et al,., 2009). So, the ability to make decisions and perform as a team, or in this case as a board, highly depends on the socio-demographic diversity of the board composition (Van Knippenberg et al., 2011).
Bezrukova et al. (2009) distinguish faultlines into two subgroups: social category and information-based faultlines. While social category faultlines relate to differences in age, gender and nationality, information-based faultlines relate to differences in the level of knowledge members may have, for instance, by measuring differences in education level.
Previous research has established the effect of faultlines on financial firm performance (Ndofor, Sirmon & He, 2015; Vandebeek, Voordeckers, Lambrechts, & Huybrechts, 2016;
Carter, D'Souza, Simkins & Simpson, 2010), but has not yet investigated their impact on commercial innovation performance. This research links faultlines to commercial innovation performance, defined as the ratio between the market valuation of a firm (Tobin’s Q) and its intensity of innovation activities (R&D intensity). Innovation performance has become an increasingly important measure for firm performance over the years (Bocquet, Le Bas, Mothe,
& Poussing, 2017).
Rather than using the commonly used measure for innovation, patents (Chang, 2016;
Laursen, & Salter, 2006; Midavaine et al., 2016), this research develops and introduces a new measure for innovation performance: commercial innovation performance (CIP). Therewith contributes to the innovation research, by introducing an improved measure. CIP represents the in- and output of the innovation process. Where the input side is represented by R&D intensity, and at the output side by Tobin’s Q. The second contribution of this research is that it examines the moderating role of board size and board independence; hence, it establishes insights on whether the effect of faultlines varies with contextual conditions. Finally, this study differs from existing research by looking at faultlines in boards just after the establishment of an M&A.
Most studies focus on existing teams, where this research looks at newly created teams. Their ability to cooperate is particularly relevant for the post-M&A performance. Hence, this study provides insight onto the effect of faultlines within newly merged boards. An M&A has a clear effect on the board composition, as the two boards are often formed into one new board. This new setting and composition may clearly impact the diversity of objectives, capabilities and performance of the team (Makri et al. 2010). For example, when diversity in objectives rise to a point where relational conflicts arise between members, which reduces their decision-making speed and ability. In this case conflicts are not focused on tasks or processes but shifts towards relationship conflicts (De Clercqet al., 2009; Marcel et al., 2011). And in that case the relational conflicts negatively impact the firms’ innovative process.
6
The paper is structured as follows: the next section introduces the theoretical framework, which includes a literature study on the relevant concepts within the post-M&A and faultline theory. This results in the development of hypotheses. Followed by the methodology section, including the measures and the data gathering and preparation. Third, the results of the analysis and the robustness check of the results. Lastly the discussion, including theoretical- and managerial implications. Followed by the limitations and future research.
2. Theoretical framework
The research question that initiated this study is the following: “What is the effect of FLS on CIP, and how is this relation affected by contextual factors?”. This section discusses the relevant theoretical concepts that are used to answer the research question. This chapter starts with a brief introduction of the conceptual model, after which this model will be further explained by a literature review. Subsequently, the hypotheses will be developed based on the findings from the literature.
2.1. Conceptual Model
Faultline theory assumes that faultline strength (FLS) impacts firm performance. This study develops a conceptual model (see Figure 1) that aims to explain the link between FLS and the newly developed measure CIP. To measure CIP, we measure the in- and output of the innovation process. More specifically, the market valuation of a firm (Tobin’s Q - output) divided by its intensity of innovation activities (R&D intensity - input). A greater score implies that investors are more positive about how well firms can convert current knowledge investments (R&D intensity) into firm value (Tobin’s Q).
Scholars found that the relationship between FLS and performance is not the same for all firms, and that specific conditions can attenuate or aggravate the negative effect of FLS on performance. By analogy, we conjecture that contextual factors have a moderating effect on the FLS-CIP relationship. A further explanation on the effect of faultlines and the contextual factors can be found below.
7
Figure 1 Conceptual model
2.2. Literature review 2.2.1. Board of directors
This research focuses on faultlines within boards, an introduction is made to clarify the concept of this focus group. The board consists of a group of people who are responsible for the firm’s strategy and governance. The members are appointed or elected, depending on the statutes of the firm, and can come from in- and outside of the firm (Goergen, Manjon &
Renneboog, 2008; Thomsen & Conyon, 2012).
Board members possess the power to set the fortune of the firm (Bezrukova, Jehn, Zanutto & Thatcher, 2009; Costanzo & Di Domenico, 2015; Mizruchi, 1983), and may differ in their socio-demographic background. This combination of power and diversity is of interest for this study. As it can create faultlines that affect firm performance (Bantel and Jackson, 1989;
Carpenter, Geletkaycz & Sanders, 2004). Faultlines are particularly relevant for boards, as faultlines may have stronger negative effects in teams with a high degree of autonomy (Rico, Molleman, Sánchez-Manzanares & Van der Vegt, 2007).
2.2.2. Faultline strength and commercial innovation performance
Teams provide the building blocks for organizations. The ability to work together on tasks creates the ability to process more information and perform better than an individual would. With combining individuals in a team structure comes diversity (Meyer et al. 2014).
Team diversity has been subject to a wide range of research, where both negative and positive aspects come to light (De Clercq et al. 2009; Homan, Van Knippenberg, Van Kleef & De Dreu, 2007; Marcel et al. 2011).
8
Two distinct views exist within team composition research: the diversity and faultlines perspective. While diversity research mostly looks at the benefits of a diverse team composition, as its greater variability in team members stimulates the greater inclusion of data and viewpoints, which ultimately creates higher creativity and improves objectivity in decision- making (Carpenter et al. 2004; Makri et al. 2010 Siciliano, 1996), faultlines research looks at the drawbacks of a multifarious team (Lau & Murnighan, 1998; Meyer et al., 2014).
This study takes on the faultline perspective, and argues that team member diversity can negatively influence group performance. Because it decreases communication and increases relational conflicts, which both hamper the productivity of the group and ability to reach goals (De Clercq et al., 2009).
FLS quantifies the likelihood of a group being spilt into homogeneous sub-groups (Meyer et al. 2014). FLS increases with greater diversity1. In teams, the lines represent socio- demographic differences between team members. A faultline marks a potential for the reformation of a group into sub-groups, this increases the likelihood of conflicts between sub- group (Meyer et al. 2014; Lau & Murnighan, 1998). For example, when a group or board is more diverse in terms of age, this may result in subgroup formation around the basis of age- groups. FLS increases even more when other socio-demographic characteristics are also aligned within the sub-group (Lau & Murninghan, 1998; Shaw, 2004).
These sub-groups support each other internally, creating a bias towards own group members in decision making. This internal view creates a ‘we versus them’ culture that may end in an emotional attachment to sub-groups members, which increase the chance of relational conflicts between sub-groups (Jehn, 1995). Relationship conflicts harm the innovation process since the associated negative emotions distract members’ attention from task and decision making (De Clercq, 2009; Jehn & Chatman, 2000). These conflicts may lead to poorer decision making in the boardroom, which may reduce the firm’s innovation performance (Carpenter, Geletkaycz & Sanders, 2004).
1 In theory, faultlines can only exist in teams with some degree of diversity. As teams with no diversity will form one cohesive group because they will have no attributes to base sub-groups on, and because with perfect diversity (everyone is different) no subgroups will be created. Such extreme cases are, however, uncommon in reality, because no group is the same or perfectly different (Lau & Murnighan, 1998). Even when a team is socio- demographically perfectly diverse in terms of age or gender, some other non-demographic characteristics will likely create possible faultlines (Vandebeek et al. 2016).
9
Most studies test the effect of FLS on firm performance, as proxied by financial performance (e.g. Ndofor et al., 2015; Vandebeek et al., 2016; Carter et al., 2010). Only recently studies started focusing on the influence of FLS on innovation performance (e.g. Coad, Amoroso & Grassano, 2017; Midavaine, Dolfsma & Aalbers, 2016; Sperber & Linder, 2018).
To measure commercial innovation performance, this study uses an input-output measure. The input relates to the firm’s intensity to invest in knowledge creation and innovation (as measured by R&D intensity). And the output of the innovation process is measured with the firm’s market valuation (Tobin’s Q). The ration between the two indicates the commercial innovation performance. A further explanation for the calculation of this ratio can be found in paragraph 3.3.2.
This study hypothesizes a negative effect of FLS on CIP. FLS seems to be particularly disruptive for the performance of teams striving to achieve efficiency (Hom, Manz, & Millikin, 1998; Jackson, 1992), as it hampers productivity and communication. Strong demographic faultlines undermine cohesion in groups. They trigger a variety of interpersonal process which hinder the performance of the group, as the differences between social identities become more salient (Jackson & Joshi, 2004). Demographic dissimilarities within teams have been associated with lower perceived performance and greater role conflict and ambiguity (Tsui & O’Reilly, 1989).
Boards with greater FLS tend to communicate less and the communication interactions that take place are of lower quality. Van Knippenberg, De Dreu and Homan (2004) found that within sub-groups members are more open for others, compared to communication with members who are not part of the sub-group, which has a negative effect on internal communication. FLS thus hinders the exchange and integration of information within a team.
As this process is critical to the performance of boards dealing with non-routine situations such as the innovation process (van Knippenberg et al., 2004), this study expects FLS to have a negative effect on CIP.
H1: Faultline strength negatively influences commercial innovation performance.
2.2.3. Moderating effect of contextual factors
Contextual influences may influence the FLS-CIP relationship because demographic identities are constructed by the social context team members operate in (Lau & Murnighan, 1998; Wharton, 1992). This study hypothesizes that board size and board independence moderate this relationship.
10 Board size
Previous studies tend to be inconclusive on the effect of board size on firm performance.
The agency problem is the most widely spread reason for negative effect of board size on firm performance (Yermack, 1996; Eisenberg, Sundgren, and Wells, 1998). The agency problem involves coordination and communication disabilities, due to a difference in ownership and execution (Bathala & Rao, 1995). For example, shareholders (the principal) and board members (the agents) may have another understanding of the risk they take, which results in actions which may not comply with the wishes of the shareholders. So, dysfunctional norms and actions in the boardroom can create the agency problem, and these problems are exacerbated with a greater board size.
The agency problem becomes more severe when the board size increases, such that it may outweigh the benefits that a larger board (e.g., more knowledge) can bring. With increasing board size, members may become more hesitant to criticize the policies of other members due to the lack of available time to change policies and act on misattributions. This may strengthen the negative effect of FLS on CIP. With larger boards, the negative consequences of faultlines may become more apparent, as monitoring becomes more time consuming (Lipton & Lorsch, 1992), making communication between subgroups even more time consuming and ineffective (Sah & Stiglitz, 1991; Moscovici and Zavalloni, 1969).
Also from a transaction cost perspective to the agency problem, it seems that a larger group involves more communication and coordination and has higher costs to reach to a decision (Jensen, 1993). Thus, with equal levels of FLS, a larger group would be more strongly harmed in its CIP. Hence:
H2: Board size moderates strengthens the negative relationship between faultline strength and commercial innovation performance.
Board independence
A board is elected to act in the best interest of the shareholders (Goergen et al., 2008;
Rhomsen & Conyon, 2012). Independent board members are external members who are appointed by shareholders, to protect their interests in the firms’ decision making.
Van Knippenberg et al. (2007) conducted a social categorization analysis within the measures of FLS to identify factors that attenuates the negative influence of faultlines. They concluded that perspective alignment alleviates the negative effect of faultlines on firm
11
performance. For example, where the independent members enable a better alignment between the perspectives of the board and the shareholders.
This highlights the importance of board independency. Board independence, determined as the ratio of external and independent board members compared to the number of internal board members. High independence attenuates the negative relationship between FLS and CIP.
In boards, which are composed of internal and external members, independent members and are often included to represent shareholders and reduce agency problems. The involvement of independent directors may thus alleviate problems caused by FLS, and positively influence firm performance (Fuzi, Halim & Julizaerma, 2016). Lefort & Uzúa (2008) argue that adding outside directors to boards with a high chance on agency problems and sub-group formation (i.e., high FLS) lowers the effect of the problems that may arise. Hence:
H3: Board independence attenuates the negative relationship between faultline strength and commercial innovation performance.
3. Methodology
This research uses a database to find the answer on the research question: “What is the effect of FLS on CIP, and how is this relation affected by contextual factors?”. Sufficient data is available to reconstruct the influence of FLS on CIP and to add contextual factors to this equation. The goal is to develop a dataset which can be used for a regression analysis. The setup of this data-study is stated below. Starting with the research setting, followed by the research method and sample. Lastly, all the variables, their collection and calculation is presented.
3.1. Research setting
This study focuses on boards who are in the phase just after an M&A (post-M&A).
M&As are a type of external expansion or sourcing investment that grows a business overnight through corporate combinations, instead of growing on its own (Kalra, 2013). Acquisitions refer to the purchases of another company, in which the bought company stops to exist from a legal point of view. Mergers relate to the decision of two firms going forward as one by forming a new entity.
Continuously changing markets and technologies force firms to actively maintain their innovative ability (e.g. M&A’s). The idea behind an M&A is that two firms together are more valuable than the sum of the two. This is, however, often not the case (Bleeke & Ernst, 1991;
Chang, Chan, & Lai, 2006; Soda, 2011; Kale & Singh, 2009). The pharmaceutical industry is chosen as a research context, as innovation plays a critical role in this a high-tech industry, and
12
M&A’s are a popular strategy to maintain the innovative ability through gaining external resources from partners (Hagedoorn and Duysters, 2002).
As M&As are rarely successful (more than 70% fails (Bleeke & Ernst, 1991; Chang, Chan, & Lai, 2006; Soda, 2011; Kale & Singh, 2009)), it is interesting to assess what role faultlines may play. M&A’s are extremely difficult to manage as it inherently requires company re-structuring. This re-structuring also takes place in the boardroom. Hence, it will show whether the communicative difficulties that accompany strong faultlines are detrimental for post-M&A innovation performance.
This paper is structured as a theory-testing study, where the theory on FLS is tested on real-life data. In order to test this, three hypotheses are developed based on the findings of previous studies on FLS.
3.2. Research method and sample
The research of Lau and Murninghan (1998 & 2005) has been the most salient in recent research on faultlines. However, research has seldomly focused on the effect on commercial innovation performance combined with the upper echelon perspective. This research uses the dataset of a previous study on FLS of El Sawy (2014), who focused on the effect of FLS on film stock prices.
The faultline data was collected on two levels. The board composition was retrieved using the BoardEx database. And the socio-demographic characteristics of the 1,246 board members were retrieved from the LexisNexis database and originated all from FY2014 reports.
Tobin’s Q and R&D intensity and the corresponding control variables will be measured on FY2014, the data on these variables are retrieved from the CompuStat database.
The data for board size and board independence originate from the BoardEx database.
Where the ‘member title’ is used to determine the independence of the board.
This research selects boards (after the establishment of an M&A in the year 2014) that are active in the pharmaceutical industry (SIC-code 283). The sample of firms originates from a Thomson SDC database, which contained a list of 174 pharmaceutical firms. A total of 52 firms were excluded due to missing values, resulting in a sample of 122 firms (i.e., boards). A further explanation on the data collection and measurement can be found below.
13 3.3. Measures
3.3.1. Independent variable - Faultline strength
The independent variable, FLS, is the result of diversity in the composition of a board.
This value is computed using the calculation of Shaw (2004). It measures the demographic characteristics and the spread of the demographic backgrounds of a team, which divides the team in subgroups.
Measuring FLS implies combining the calculations of the FLS of every socio- demographic factor to end up with one FLS for a team. The idea of faultlines originates from the research of Lau & Murnighan (1998), who presented a simplified version of the faultline theory used in their report. They measured FLS on an intuitive scale from non-existent to existent. Shaw (2004) further developed this measure to deliver a percentage of the existence of faultlines within a team.
Not all socio-demographic faultlines are suited as a measure. For example, education, by the time a manager becomes part of the board, he or she will have gained a lot of experience.
Their formal education, which typically took place decades before and buried under the knowledge which is created by the experiences, is no longer a good proxy for differences in cognitive characteristics (Barkema & Shvyrkov, 2007). Also, a few adaptions took place on the measurement scales from Lau and Murnigham based on the findings of El Sawy (2014).
Because of the difficulty to identify race from the datasets, it was dropped. Hence, this study selects the following socio-demographic characteristics to measure FLS: age, gender and experience. Since they are found to be the most salient indicators for faultlines (Van Knippenberg et al., 2009). Van Knippenberg et al. (2009) found gender to be an important determinant within the faultline methodology since gender is associated with widely shared stereotypic beliefs. Also, managers tend to favor same-gender team members (Jackson & Joshi, 2004). That affects the perspective members have on the faultline created by gender, as members subjectively see gender faultlines as more meaningful than other determinants. So, stereotypes (age, gender and experience) determinants are more salient because of the subjective meaning members give to them.
Using statistical software, SAS, the dataset on socio-demographic was used to calculate the overall FLS per firm, as well as the FLS per attribute. Lastly, FLS was matched with the corresponding firm based on GVKEY and were given an additional company code. The additional code was used because of the inability to integrate the GVKEY in SAS. FLS was computed from a sample of 154 firms. An in-depth explanation of its calculation is explained in the appendix A. A summarized explanation of the calculation is discussed below.
14 FLS calculation
There are two ways to calculate FLS, the method of Thatcher (2009) and Shaw (2004).
The method of Shaw (2004) will be used, the differences between them will be explained with the use of an example.
As an explanation of the FLS calculation an example is provided. So, if you would like know the FLS within a group, Lau & Murnighan (1998) argue that faultlines are based around one base attribute (e.g. age, gender or experience). One should then calculate the internal alignment (IA) and cross-group alignment (CG) of all attributes available to know what sub- groups can be formed based on the existing faultlines.
For IA consider the following example: a faultline is based around one attribute (e.g.
gender). In that case this results in two sub-groups: males and females. This would be the first faultline, based on gender. To get a precise look on faultlines, one needs to consider the similarity (IA) based on other attributes within the sub-group. The gender faultline strengthens when the inside-members of the sub-group are very similar regarding the other attributes. We look at similarity in sub-groups which results an IA. Up to this point, are Thatcher’s and Shaw’s methods are approximately equally useful.
However, faultlines do not always exist on the obvious places. Therefore, we must calculate all possible IAs based on every attribute across sub-groups, to fully capture the structure of a group. In other words, to check of there are similarities across groups. The previous calculation of IA must be repeated three times, each time with another attribute as faultline base. This results in the CG. If the CG is high, the FLS will be lower since groups look alike based on other attributes apart from the basis attribute. For example, when the females are between 50-59 and males are also between 50-59. This results in a high CG, this reduces the FLS. Another example where FLS is increased by CG, when the females are between 50-59 and males are all 67>. This case there is a low CG, resulting in a higher FLS.
This is where the method of Shaw is superior to Thatcher’s. Thatcher only considers the strong initial faultlines resulting from the IA, whereas Shaw considers all possible splits by looking at CG, which results is in a more precise valuation of FLS.
Shaw (2004) describes a five-step calculation to calculate the FLS. A summary of the steps is described below in Table 2. A more extensive explanation from El Sawy (2014) can be found in Appendix A.
15
Step Action Operationalization
1 Determining attributes Gender, age and experience;
2 Categorization of attributes, values need to be coded into numeric values.
Gender Age Experience
1 Male <50 1-3y
2 Female 50-59 4-7
3 60-67 8-11
4 67> 11>
3 Calculation IA and CG
𝐼𝐴𝑏𝑎𝑠𝑒/𝑥 =∑(0 − 𝐸)2 𝐸
4 Calculation overall faultline strength 𝐹𝐿𝑆 = 𝐼𝐴 ∗ (1 − 𝐶𝐺)
5 Effect of FLS on firm performance. In our case Tobin’s Q and R&D intensity.
Hypotheses 1 & 2
Table 1 Five step FLS calculation Shaw (2004)
FLS1
To conclude, this study chooses Shaw’s method instead of the more commonly used method of Thatcher, because (a) Shaw’s method allows for group sizes above six (83% of the boards in the dataset contain seven or more members), (b) allows for more than two sub-groups, (c) and enables the measurement of cross-group alignment. Resulting in a more precise and better suited measure.
3.3.2. Dependent variable - Commercial innovation performance
Only recently studies have started to study the effect of FLS on innovation performance.
As innovation performance is often a reason to merge or acquire, weconsider it to be an important measure for the success of an M&A.
Innovation performance can be measured in different ways. To get an objective observation, innovation performance is inherently more difficult than for example financial performance to measure, due to the nature of the variety of the performance indicators themselves (Hagedoorn & Cloodt, 2003).
Three drawbacks exist when using patents as innovation performance measure. First, patents are a formal means to protect property, but not every firm uses patenting to protect its innovation property. The use of patents is context dependent (Laursen & Salter, 2014). An alternative for patenting: secrecy is a (unmeasurable) frequently used method. Or some industries, like the service industries, innovate but not often use protection mechanisms. Even in the industries where patents are most common, e.g. pharmaceutical industry, the percentage of patented inventions barley rise above 50% (Kale & Singh, 2009).
16
Second, patents are sometimes used for non-innovative purposes. Blocking patents are patents that are used to block or deter developments of competing firms by patenting vital parts of their competing design (Giuri et al., 2007). The patent holder may not commercialize these inventions in its own products.
Third, patents do not necessarily lead to commercial innovation. Only when firms commercialize the invention, it is considered to be innovation (Giuri et al., 2007). An invention does not directly lead to a product or process innovation. To conclude, patents do not provide a transparent measure for innovation performance, as they are not directly linked to all aspects of innovation.
This study introduces a new method of measuring innovation performance that overcomes some of the shortcomings of using patents as indicator of innovation performance.
This new method looks at firm’s R&D intensity and Tobin’s Q as an in- and output process, where the relation between the two represents firms’ commercial innovative performance. R&D intensity (equation 1) represents the input; Tobin’s Q (equation 2) represents the output of the innovation process (Xue et al., 2012). Tobin’s Q marks the market value of the firm, compared to the actual value. For example a Tobin’s Q of 1.2, this means that 20% extra market value is created by firm’s activities. For example when a high-tech firm invents a new form of WiFi, which is superior to the previous version. This might have cost the firm €1M, but the market values this at €10M. This creates a Tobin’s Q higher than 1.
This ratio was firstly introduced by Xue et al. (2012), where it was used to indicate the effectiveness of R&D expenditure. As an increase in R&D intensity should result in new products and patents, the latter two are market-valued in the Tobin’s Q. The measure in this paper looks at the ratio between R&D intensity and Tobin’s Q (equation 3), to see how the innovation activities relate to the created value for the firm. For example, when a firm invests heavily in innovation (e.g. a high R&D intensity) but does not create additional firm value (e.g.
a low Tobin’s Q) has a low CIP.
This measure provides a transparent and benchmarked measure for the commercialization of inventions. Their model also included patent count, which would mimic the middle state in the innovation process. However, they did not combine the variables into one measure, but kept all variables as separate measures of innovation efficiency. Therewith, they do not capture the mutual cohesion between the variables, which would provide the researcher with a valuation of innovative activities.
𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 = 𝑅&𝐷 𝑒𝑥𝑝𝑒𝑛𝑑𝑖𝑡𝑢𝑟𝑒
𝑇𝑜𝑡𝑎𝑙 𝑠𝑎𝑙𝑒𝑠 (Equation 1)
17 𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄 = 𝑀𝑉𝐸+𝑃𝑆+𝐷𝐸𝐵𝑇
𝑇𝐴 (Equation 2)
MVE = (Closing price of share at the end of the financial year) × (Number of common shares outstanding);
PS = Liquidating value of the firm's outstanding preferred stock;
DEBT = (Current liabilities - Current assets) + (Book value of inventories) + (Long term debt);
TA = Book value of total assets.
𝐶𝑜𝑚𝑚𝑒𝑟𝑐𝑖𝑎𝑙 𝑖𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 = 𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄
𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 (Equation 3)
3.3.3. Moderators
Contextual factors (e.g. board size and board independence) influence the composition of a board, and thus they influence how a team operates (Lau & Murninghan, 1998; Wharton, 1992). By analogy, we conjecture that contextual factors have a moderating effect on the FLS- CIP relationship. That is why this study conducts further research on the influential effect of contextual factors on FLS. The measurements for board size and board independence are discussed below.
Calculating board size, the sum of the total number of active members at the moment of the publishing date of the annual report (Wang, 2012).
Board independence is a representation of the percentage of independent directors within a board (equation 4) (Lefort & Uzúa, 2008).
𝐵𝑜𝑎𝑟𝑑 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑚𝑒𝑚𝑏𝑒𝑟𝑠
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑚𝑒𝑚𝑏𝑒𝑟𝑠 (Equation 4)
3.3.4. Control variables
As used by previous scholars, who used R&D intensity and Tobin’s Q as dependent variables in the context of innovation, the control variables selected for CIP are: firm size (equation 5), invested capital, earnings before interest and tax (EBIT) and leverage (equation 6) (Bharadwaj et al., 1999; Midavaine et al., 2016; Xue et al., 2012).
𝐹𝑖𝑟𝑚 𝑠𝑖𝑧𝑒 = log (𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠) (Equation 5)
𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 = 𝐿𝑜𝑛𝑔 𝑡𝑒𝑟𝑚 𝑑𝑒𝑏𝑡+𝐷𝑒𝑏𝑡 𝑖𝑛 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠
𝑆𝑡𝑜𝑐𝑘ℎ𝑜𝑙𝑑𝑒𝑟𝑠′𝑒𝑞𝑢𝑖𝑡𝑦 (Equation 6)
18
Variable Variable type Scale
type
Operationalization Sources
Faultline
strength Independent Ratio The probability that a faultline
will be activated
Shaw (2004)
Commercial innovation performance
Dependent Interval
The ratio to which R&D is transformed into market value in (Tobin’s Q).
Tobin’s Q: Xue et al.
(2012), Bharadwaj et al.
(1999); R&D intensity:
Barker & Mueller (2002), Midavaine et al. (2016), Chen & Hsu (2009);
Board size Moderator Ratio Number of active board
members.
Coles et al. (2008)
Board
independence Moderator Interval
Percentage of independent board members, as compared to the total number of members.
To see how
independent/objective the board could operate.
Dah et al. (2018), Lefort
& Urzúa (2007), Chen &
Hsu (2009);
Firm size Control Ratio An indication of the size of the
firm.
Chen & Hsu (2009);
Barker & Mueller (2002);
Invested capital Control Interval
The total amount of money raised by a firm by issuing shares to shareholders, as compared to the worth of the firm.
Bharadwaj et al. (1999);
Midavaine et al. (2016);
Xue et al. (2012);
EBIT Control Ratio Earnings before interest and tax. Bharadwaj et al. (1999);
Leverage Control Interval
The use of borrowed capital to fund investments to expand the firm's asset base and returns.
Chen & Hsu (2009);
Bharadwaj et al. (1999);
Xue et al. (2012); Barker
& Mueller (2002);
Table 2 Overview of variables
3.4. Preparing dataset
The dataset was prepared to enable a trustful and robust analysis. First, possible outliers were be detected. Second, correlation analysis and multicollinearity checks were performed.
Third, the effects of missing values were investigated by doing replacement and exclusion.
Lastly, a normality test on the data distribution, as well as a reliability analysis were conducted.
3.4.1. Outliers
An outlier is an observation within the dataset that is distant from other observations.
An outlier occur due to variability in the measurement or measuring errors. True errors must be removed from the dataset, because an outlier can cause problems in the analyses. On the other hand can extreme values be really important for the analysis. Tests can be conducted (e.g. Z- test or boxplots) to test if an outlier is a true error or just an extreme value.
Outliers are found using boxplots, Z-tests and Winsorize method (Dixon, 1980). The boundaries, for determining what observations should be marked as outliers, are set on a maximum standard deviation of -3 and +3.
19
The results of the boxplots and Z-test found the same outliers, which resulted in the replacement of 74 of the 1,098 observations. An overview of the outliers can be found in appendix B. The Winsorize method helps to assess the robustness of the findings regarding outliers. Rather than excluding outliers, it changes the original value by the nearest value of an observation which is not an outlier. This method allows for a more stable analysis than a trimmed analysis would produce (Dixon, 1980). In Appendix B, you can find a table with the variable and the number of replaced observations. The plots before and after replacing the observations can be found in appendix C. The high number of outliers, in R&D intensity and EBIT, is caused by missing values or faulty observations from CompuStat.
3.4.2. Correlation analysis
Table 6 displays the correlation analysis of all involved variables. CIP correlates negatively with board independence, and positively with firm size and EBIT; FLS correlates positively with board size and firm size, and negatively with board independence; Board size correlates with board independence, firm size and leverage; Board independence correlates positively with firm size; Firm size correlates positively with invested capital and EBIT; and invested capital correlates negatively with leverage.
The correlation analysis on the relation between R&D intensity and Tobin’s Q, found a significantly relation (r= 0.200, p= 0.027). The new variable looks at the ratio between Tobin’s Q and R&D intensity, and will be referred to as commercial innovation performance. A further explanation of the calculation of CIP can be found in paragraph 3.3.2.
3.4.3. Multicollinearity
Based on the coefficients output, a multicollinearity test is conducted to check if variables are approximately linearly related to the dependent variable. This can harm the value of the results, as they increase the chance on a type two measurement error (failing to reject a false null hypothesis).
If the VIF value exceed 2.5 between two variables, one of both variables needs to be deleted or the variables need to be combined. Results indicate that multicollinearity is not a serious issue since no variable exceeds the 2.5 VIF limit (Hair et al., 1998). The output of the multicollinearity test can be found in Appendix E.
3.4.4. Missing values
Because of the results from the outlier analysis. An extra step is included to the data preparation: searching for missing values. This to make sure that the regression analysis later
20
is not affected by missing values. After checking for missing values, debt-equity ratio had 55 missing values and R&D intensity had eight missing values. The missing values are replaced by the sample mean value. Debt-equity ratio is replaced by leverage, due to the high number of missing values compared to the total number of observations. Otherwise this would not be a trustful control variable, since 45% of the observations is replaced by the mean value.
3.4.5. Normality check
After checking for outliers, a check of normality is conducted using the Shapiro-Wilk test (Razali & Wah, 2011). After the test was conducted, Tobin’s Q, R&D intensity, board size, board independence, firm size, invested capital, EBIT and leverage; were found to be non- normally distributed. The distributions before and after the logarithmic scale and inverse distribution function can be found in Appendix D.
Shapiro-Wilk test
Statistic df Sig.
FLS .984 122 .165
Tobin’s Q .871 122 .000
R&D intensity .715 122 .000
Board size .955 122 .000
Board independence .768 122 .000
Firm size .963 122 .002
Invested capital .903 122 .000
EBIT .831 122 .000
Leverage .924 122 .000
Table 3 Shapiro-Wilk test
To solve this problem, the variables are replaced by a logarithmic scale version. For CIP and EBIT an alternative means for transforming the variables in a logarithmic scale was used, since they both measures contained negative observations.
𝐿𝑛(𝑣𝑎𝑙𝑢𝑒) = log (𝑣𝑎𝑙𝑢𝑒 + 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑣𝑎𝑙𝑢𝑒 + 1)
After this transformation, R&D intensity, board size, board independence and EBIT remained non-normal. The reason for this non-normality lies in its extreme values. When using the Winsorize method, all outliers get replaced with the nearest normal value. Since R&D intensity and EBIT both contained more than 20 outliers, their normality is affected by the replacement values. When excluding these observations, the results show a normal distribution. Board independence suffers from non-normality due to a large number of board without independent members. The remaining non-normal variables are computed with the use of the inverse distribution function, as stated below.
𝑉𝑎𝑙𝑢𝑒 𝑁𝑜𝑟𝑚𝑎𝑙 = 𝐼𝐷𝐹. 𝑁𝑂𝑅𝑀𝐴𝐿(𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝑎𝑙 𝑟𝑎𝑛𝑘 𝑣𝑎𝑙𝑢𝑒, 𝑚𝑒𝑎𝑛, 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑖𝑜𝑛)
21 3.4.6. Reliability analysis
To assess the reliability of the FLS scale, a reliability analysis was performed to see if the components correlate. The Cronbach alpha of the combination of gender, age and experience (α = .690) is higher than the minimum 0.6 (see Table 4). The Cronbach alpha (α = .746) increases in reliability when age would be excluded. An additional regression will be conducted with a FLS variables containing gender and experience (see Equation 8). The measure for FLS will be computed as follows:
𝐹𝑎𝑢𝑙𝑡𝑙𝑖𝑛𝑒 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ = 𝐹𝐿𝑆𝐺𝑒𝑛𝑑𝑒𝑟+𝐹𝐿𝑆𝐴𝑔𝑒+𝐹𝐿𝑆𝐸𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒
3 (Equation 7)
𝐹𝑎𝑢𝑙𝑡𝑙𝑖𝑛𝑒 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ𝐺𝑒𝑛.𝐸𝑥𝑝.= 𝐹𝐿𝑆𝐺𝑒𝑛𝑑𝑒𝑟+𝐹𝐿𝑆𝐸𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒
2 (Equation 8)
Item total statistics Scale mean if
item deleted
Scale variance if item deleted
Corrected item- total correlation
Cronbach’s alpha if item
deleted
Gender .397 .042 .391 .677
Age .337 .044 .285 .746
Experience .340 .039 .588 .556
Table 4 Cronbach alpha FLS
4. Results
The sample in the regression analysis consists of 122 observations. Table 5 present a descriptive table.
Descriptive statistics and correlation
Mean Std.
Deviation
(1) (2) (3) (4) (5) (6) (7) (8)
(1)
Commercial innovation performance (CIP)
3.536 22.070 1 -.085 .085 -.226* .358** .141 -.320 .370**
(2) Faultline
strength (FLS) .117 .059 1 .564** -.187* .308** -.065 .145 -.002
(3) Board size 8.230 1.974 1 -.198* .542** -.080 .228* .059
(4) Board
independence 77.517% 28.985% 1 -.364** -.113 .024 -.167
(5) Firm size 2.701 .955 1 .277** .082 .296**
(6) Invested
capital .718 .166 1 -.662** .137
(7) Leverage 62.458 172.490 1 -.050
(8) EBIT .503 .338 1
* Correlation is significant at the 0.05 level
** Correlation is significant at the 0.01 level Table 5 Descriptive statistics and correlations
22
Table 5 shows that the mean of the dependent variable: R&D intensity and other variables are exceeded by the standard deviation. These differences suggest that the data is over-dispersed, and therefore the use of a negative binominal regression is appropriate (Field, 2013). But since this dataset does not contain count variables, a normal multiple-regression analysis will suffice.
A multiple linear regression using SPSS was used to assess the hypothesized relationship between FLS and CIP, as well as the moderating effect of board size and board independence. The overall model, including all control-, dependent-, independent- and moderating variables explain 35% (R2 = .348) of the variation found in the dataset. Model 1 demonstrates the regression results using the control variables only. Model 2 includes the independent variable, FLS, to test hypotheses 1. Model 3 adds the moderating variable of board size (as well as a direct effect), to test hypotheses 2. Model 4 excludes board size and analyzes the moderating and direct effect of board independence, to test hypothesis 3. Model 5 shows the results when introducing both moderators at once.
Regression table
1 2 3 4 5
Firm size .936* (.082) 1.452** (0.011) 2.184*** (.001) 1.326** (.023) 2.026*** (.002)
Invested capital .303 (0.830) .143 (.877) .086 (.940) .154 (.868) .052 (.955)
EBIT 1.045*** (.000) .955*** (.001) .846*** (.002) .927*** (.001) .831*** (.002)
Leverage -.176 (.517) -.130 (.624) -.119 (.648) -.120 (.635) -.104 (.690)
FLS (H1) -.715** (.013) -2.662** (.042) -1.014 (.312) -2.233** (.044)
Board size (BS) .453 (.779) 2.234 (.338)
Board independence (BI) -.182 (.754) 1.694* (.065)
BS × FLS (H2) .999* (.091) -.997 (.203)
BI × FLS (H3) .057 (.773) -.299 (.302)
R2 .252 .293 .337 .300 .348
Adjusted R2 .225 .261 .294 .255 .292
F 9.341 (.000) 9.120 (.013) 7.838 (.032) 6.619 (.577) 6.278 (.418)
∆ R2 .252 (.000) .041 (.000) .044 (.000) 0.007 (.000) .055 (.000)
* p < .10, ** p < .05, *** p < .01 Table 2 Regression output
As suggested by Hypothesis 1, this study finds that FLS, indeed, has a negative effect on CIP in almost all models (all p’s < .05). FLS was found to have no significant effect on CIP in model 4, where the direct and moderating effect of BI is added. Hypotheses 2 predicted a strengthening effect of board size on the relationship between FLS and CIP. The results reveal that – at the less restrictive 10% significance level – board size increases the negative effect of
23
FLS on CIP at low levels of FLS. But opposite of what was expedited when FLS increases, we find that smaller boards experience a stronger negative effect of FLS on CIP.
The results for Hypothesis 3 do not provide any proof for the attenuating effect of board independence on the negative relationship between FLS and CIP. Surprisingly, the regression analysis found a positive direct effect of board independence on CIP (B = 1.694, p = .065).
Which suggests that board independence does increase CIP, but does not decrease the effect of FLS on CIP. This effect was only found in Model 5, where both contextual factors are added.
Robustness checks
The robustness checks provide a stronger test for our hypotheses. Six checks were performed to see if the results hold when altering measurements and using smaller samples.
The first check was to see what would happen if the regression was done on a dataset which is not prepared. The preparation of the dataset included a check of outliers and normality.
This led to no significant results in the regression analysis, suggesting that a type 1 error would occur if the preparation was not conducted.
The second check tested the effect of the individual FLS attributes on CIP. Each individual attribute is tested by performing a regression analysis. To assess the impact of the individual components of FLS, we tested the FLS of each attribute separately using similar regression analyses. Results show no significance for the individual FLS attributes, age, gender and experience, R2 = .323, F (3,72) = 4.906, B = (gender = .071; age = -.292; experience = - .050), t (gender = .364; age = -1.231; experience = -.208), p = (gender = .717; age = .222;
experience = .836). So, their effect can only found if we look at all variables at once, as suggested by Shaw’s approach.
Additionally, a FLS measure was computed based on gender and experience. This decision was based on three findings. First previous scholars which suggests that some attributes are more salient than others. Where gender and age are found to be most salient of the three (van Knippenberg et al., 2007; Midavaine et al., 2016). Second, since the dataset
-12 -10 -8 -6 -4 -2 0
Low FLS High FLS
CIP
Low BS
High BS
Figure 1 Interaction plot board size
24
contained a high percentage from a specific age group (57% > 60, 91% > 50), which would in theory attenuate FLS. And third, the results from the Chronbach Alpha also suggest that age should be excluded.
Results (B = -.339, t = -1.806, p = .074) showed that age does contribute to the FLS as the negative regression coefficient decreases (∆B = +.376) when age is excluded. Suggesting that faultlines based on age might not need large differences in age to be activated, as 57% of the members are no more than 10 years apart.
To check if the insignificant results on the moderating effect of BI, a mean spilt (77%) was made. To see if including solely high (or low) independence levels would affect the outcomes of the regression analysis. Results showed no difference in significance. An additional division on 50% is analyzed, which also resulted in an insignificant relationship.
To check the insignificant results of the moderating effect of board independency vary across small vs. large boards, the dataset was split into two groups: small boards (<=7 members) and large boards (7> members) (Jensen, 1993). A regression analysis was conducted on both groups. No significance was found in the larger board size group (size > 7, N = 73), but surprisingly, when looking at smaller board sizes (size <=7, N = 49) board independence does have a significant moderating effect. The positive beta of the interaction term (B = .684, t = 1.985, p = .055) suggests that board independence indeed has an attenuating effect on the FLS–
CIP relationship as showed In figure 2.
5. Discussion
In this final section, the theoretical implications of the previously mentioned findings will be discussed. Followed by the managerial implications. Finally, the main limitations as well as the opportunities for future research based on the results of this study.
-20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0
Low FLS High FLS
CIP
Low BI
High BI
Figure 2 Interaction plot board independence
25 5.1 Theoretical implications
In this study, the impact of FLS in boards of directors on post-M&A CIP is investigated.
Extant research on faultline theory mostly focused on the consequences of FLS on financial performance, but much less on innovation performance (Dolfsma & Van der Eijk, 2016;
Jackson & Joshi, 2004; Midavaine et al., 2016; Ndofor et al., 2016). From the limited set of papers that focus on innovation performance, most bear a great uncertainty in measurement accuracy by relying on patent data. This study contributes to the measurement method for innovation performance by creating a new measure for CIP that aims to analyze commercial innovation performance in a more accurate way, as it looks at the relation between the input (R&D intensity) and output (Tobin’s Q) of the innovation process.
The regression analysis confirms the negative effect of FLS on CIP of earlier research (Lau & Murninghan, 1998; van Knippenberg, 2005). A regression analysis on individual faultline attributes showed no significance. This would mean that especially IA is important for the effect of FLS and that faultlines are not solely based on the base attribute, but on the alignment of multiple attributes. Even though the teams from this dataset contained a high percentage from a specific age group (57% > 60, 91% > 50), which would in theory attenuate FLS. This would suggest that gender and experience would create the lion’s share of the FLS.
A regression on the effect of FLS based on a combined variable of gender and experience found no significant relation between FLSgender/age and CIP. Future research should focus on investigating the effect of individual attributes on FLS within the context of innovation, by adding more socio-demographic and non-demographic attributes to the equation. As this would improve the measure for FLS.
By building on the faultline theory, this study contributes by providing a better understanding on the effect of context on FLS. We gain a better understanding of the impact of FLS on CIP under varying conditions. Yermack (1996) and Eisenberg et al. (1998) suggest that board size aggravates the effect of FLS on CIP, because of increased agency problems. More members would mean more possible sub-groups which increases FLS (Lau & Murninghan, 1998). The correlation analysis surprisingly found that a large board size increases the negative effect of FLS, at low levels of FLS. Resulting in an increasing negative effect on CIP, harming the process as the agency problems increase (Bathala & Rao, 1995).
But opposite of what was expedited when FLS increases, we find that smaller boards experience a stronger negative effect of FLS on CIP. The existence of extra members seems to attenuate the effect of FLS on CIP. Disconfirming literature that suggests that FLS increases with an increase in team size. The extra members seem to have an positive effect. Previous
26
scholars found that this effect can be created by extra idea generation and problem-solving. The heterogeneous background of the board members creates a variety in expertise, knowledge and skills. This enriches the supply of alternative, which positively effects the novelty of idea generation (Akron, Feinblit, Hareli & Tzafrir, 2016; De Clercq el al, 2009). This diversity in background results in a cross-functional discussion and a broader set of perspectives on the current situation (Bantel and Jackson, 1989; Carpenter et al. 2004), which results in the development of more creative and dynamic development process (Carpenter et al. 2004; De Clercq et al. 2009). These effects eventually lead to a higher innovation performance.
Van Knippenberg et al. (2011) found that perspective alignment attenuates the negative effect of faultlines on firm performance. To ensure that the board acts in the best interest of the firm. Shareholders assign independent members to a board to make sure that their perspectives are represented. A high percentage of independent members would thus result in a decline of the opposing objectives created by faultlines, as everyone acts in the best interest of the firm.
This identity creates a team-spirit and shared objective among the team members, which would decrease the negative effect of faultlines. Unfortunately, this relationship was not found in the regression analysis. However, Jensen (1993) found that a team should not contain more than seven members. Based on that assumption, another regression was done on both large and small board. There we found a significant attenuating effect of board independence on FLS within small boards. This means that the effectiveness of attracting independent members is encounters the same contextual limit as Jensen (1993) suggested. A board should be limited by size to decrease the agency problem (Yermack, 1996; Eisenberg et al., 1998). Including independent members in small boards is found to be decreasing the effect of FLS on CIP, by improving the objective alignment among the team members and shareholders (van Knippenberg et al., 2007).
Further research needs to be done on the finding that the context matters in relation with FLS, to find contextual factors which attenuate the FLS effect.
All in all, the outcomes of hypotheses 1-3 have theoretical implications for faultline research in an M&A context. As it finds that stronger faultlines harm the chance of gaining a post-M&A benefit in terms of commercial innovation performance. But strong faultlines can be weakened by attracting more members (board size) or independent members (board independence. So, before entering into an M&A, a firm should look at their board of directors’
FLS when composing a new board, as it damages the benefits that an M&A could offer.
