The influence of diversity on single firm versus
interorganizational technological development project
performance: A multilevel approach
Master thesis MSC Business Administration
Strategy and Innovation Management
June 2017
Supervisor: I. Estrada Vaquero
Co-supervisor: N. Balogh
Co-assessor: A.A. Oleksiak
Stefany M. Engelhardt
S279018
Word count: 14616
ABSTRACT
Technological development projects are increasingly used as a way of improving performance, enabled by the introductions of new technologies and management practices. Even though these new introductions provide improvement opportunities, they also come with a lot of complexity, ambiguity, and uncertainty. In order to deal with these complexities and uncertainties, these projects require effective collaboration through integrated teams. In other words, diversity within projects teams is argued to be an enabler of performance improvement, however it is not that straight forward as there are different types of diversity and the different effects are not fully clear. This study addresses this gap by using a more integrative approach to examine the effects of two types of diversity on project performance. Moreover, a comparison is also made between single firm and interorganizational projects as these may have different implications for managers. A dataset of 85 wind farm projects in the wind energy industry was created using data from Companies House and Orbis databases. The findings suggest that there is not always an inverted U-shape relationship between board level diversity and technological development project performance, as the opposite was found. Moreover, other first indications were found arguing that interorganizational projects have a more difficult time to improve performance when dealing with diversity and partner level diversity is interrelated to board level diversity. In sum, this study concludes that the relationship between board level diversity and technological development project performance is context-related and also requires the adjustment of management processes and practices.
INTRODUCTION
Opportunities for technological development projects are increasing due to a fast pace of change in the environment that introduces new technologies and management practices, which is changing project management (Fellows & Liu, 2016). These new technologies and management practices provide the means to drastically improve performance, however it also brings with it a lot of complexity, ambiguity, and uncertainty (Fellows & Liu, 2016). Within these technological developments projects, performance and as a consequence profitability is truly impacted by including the right people that perform well (Morris, 1998). Moreover, these projects require collaboration through effective teamwork in order to reach goals, leading to an increase of integrated project teams (Morris, 1998). In other words, diversity in technological development projects are argued to have great impact on project performance.
The wind energy industry is increasing in popularity due to its social relevance and innovative technological development projects. As the wind energy industry is dynamic in nature, there are many risks involved such as high costs, capital requirements, complexity (Dedecca, Hakvoort, & Ortt, 2016). Moreover, Dedecca et al. (2016) find that the wind energy industry is multi-disciplinary, as different types of experts are required throughout the project of developing wind farms (e.g. project developers, wind turbine manufacturers, engineering companies). As a result, many firms tend to use integrated teams of diverse disciplines in order to cope with the complexity of the industry. The wind energy industry shows two different type of technological development projects. The first type of projects are single-firm projects that use integrated teams to develop windfarms, whereas the second type of projects are interorganizational projects in which multiple firms collaborate. The main difference between the two types relates to the interorganizational projects having a higher need to spread risks and obtain the required disciplines (Dedecca et al., 2016).
Prior research has focused on two (inter)organizational levels in which the effects of the two types of diversity on project performance are examined. On the one hand, prior research has focused on board level diversity within single firm projects. Out of Parkhe’s (1991) 5 different dimensions of cultural and operational incompatibility, culture and management practices have been researched most often. For cultural diversity, Fellows & Liu (2016) argue that individuals have different ways of sensemaking due to their distinct believes and values which influences information acquisition and processing and therefore decision-making. For management practices, functional background within top management teams has been the focus of prior research. The importance of studying this type of diversity is emphasized as functional background is argued to be of large influence on processes and outcomes (Buyl, Boone, Hendriks, & Matthyssens, 2011; Cannella, Park, & Lee, 2008). Nevertheless, research conducted so far provide mixed results. First, a positive association between of top management team intrapersonal functional diversity on firm performance is found (Buyl et al., 2011; Cannella et al., 2008). On the contrary, Gong, Shenkar, Luo & Nyaw (2005) found a negative association between functional diversity on JV performance.
On the other hand, a different focus of prior research has been on partner level diversity within interorganizational projects. The importance of this type of diversity, in other words resource complementarity, is emphasized as the access to these types of resources is one of the primary reasons why collaboration between firms happens, for instance through strategic alliances or joint ventures (Choi & Beamish, 2013; Ireland, Hitt, Vaidyanath, 2002). Moreover, Chung, Singh, & Lee (2000) actually find empirical evidence that firms with complementary resources are more likely to partner in an alliance. Finally, collaborations with firms that have complementary resources has become an important way to increase competitive advantages and/or performance (Harrisson, Hitt, Hoskisson, & Ireland, 2001). The core logic of collaborating with a diverse partner is based on the premise that combining resources and competences with a partner provides opportunities to achieve objectives that are cannot be achieved by one partner alone (Mohr & Puck, 2005). Petrovic, Kakabadse & Kakabadse (2006) provide a comprehensive review of the research available on the governance of international joint ventures and also study the influence of board level diversity on JV effectiveness and performance. Again, research conducted so far provide mixed results. First, several studies found empirical support for a positive association between resource complementarity and performance (Gong et al., 2005; Choi & Beamish, 2013). Contrary to these results, several authors found empirical support for a negative relationship between culture & management difference as two factors of interfirm diversity and performance (Lin & Chen, 2002; Mohr & Puck, 2005).
diversity together as they argue that performance of interorganizational projects is contingent upon both types of diversity. Finally, comparison between single firm- and interorganizational technological development projects is lacking. They are only studied separately, even though a comparison could result in alternative decision-making based on a broad overview of the advantages and disadvantages of the two types of projects. It is important to address these gaps in the literature for several reasons. First, providing an integrative research of the two types of diversity provides a more complete explanation on how performance is truly influenced and also in which direction. Furthermore, a comparison of the different types of projects provides a better understanding of the context in which either of the projects is more beneficial. From an academic perspective addressing this gap provides a definite conclusion on the effects. On the other hand, managers’ understanding of the implications of their decisions increases, making them better equipped to make decisions.
The purpose of this study is to refine the understanding of the effects of the two types of diversity on technological development project performance for both single firm- as well as interorganizational projects and providing a more integrative perspective of the two types of diversity together. After all, collaboration through integrative teams in projects has important implications for performance, whether applied within a single firm project or an interorganizational project (Morris, 1998). Therefore, this study seeks to bridge the gap by researching the effect of board level diversity on technological development project performance for both single firm- as well as interorganizational projects. Afterwards, a comparison between the two (inter)organizational levels of projects will be researched. Finally, this study will zoom in only on interorganizational projects and assess the interaction of board level diversity and partner level diversity. Hence, the following research questions have been developed:
1. What are the effects of board level diversity on technological development project performance? 2. How does partner level diversity moderate the effect of board level diversity on technological development project performance for interorganizational projects?
In order to answer research question, a sample of 85 single firm- and interorganizational projects from the wind energy industry was collected in the years 2010 and 2011. Secondary data regarding technological development projects in the wind energy industry was collected from Companies House and Orbis. Several statistical analysis was performed using this secondary data. The logic for this approach stems from the rather mature literature field regarding diversity within projects.
This study contributes to the existing diversity literature in the following ways. First, this study shows that the relationship between board level diversity and technological development project performance does not always show the widely accepted inverted U-shape. Surprisingly, the results showed a U-shaped curve, indicating that context (e.g. technology-driven projects) needs to be taken into consideration as it can influence the direction of the curve. However, the results should be taken with caution as there is an adjusted R2 of only 5,6%. Second, this study uses a more integrative approach
regression analysis were not significant, the results of the cross tabulation provide a first indication that the two are indeed interrelated and influence each other. Finally, the integrative approach also includes comparing single firm projects and interorganizational projects as they are argued to have different experiences dealing with diversity. Again, the results of the regression analysis were not significant, but the results of the cross tabulation show a first indication that interorganizational projects have more difficulty in improving performance as diversity is more difficult to manage. Finally, this paper contributes to the diversity literature by applying the developed theory in another industry setting to show whether the proposed relationships still hold. As mentioned above, the hypothesized inverted U-shape was not found, therefore opening up opportunities for more research to indicate in which contexts this relationship does hold.
The paper will proceed as follows. First, a literature review is provided to give an overview of the literature that cover board level diversity and partner level diversity. It is finalized with the conceptual model and formulated hypotheses. Subsequently, the methodological approach is discussed in detail. Afterwards, statistical analyses and hypotheses’ results are presented. Finally, discussion and conclusions are provided, including theoretical-, managerial implications, limitations and further research.
THEORY AND HYPOTHESES
In this section, a review is of the literature on diversity is provided. This review is structured as follows; first board level diversity is reviewed for both types of diversity, after which partner level diversity is reviewed for both types of diversity. Finally, hypotheses and conceptual models are introduced. Diversity literature
(inter)organizational levels, respectively board level diversity and partner level diversity. The first organizational level is board level diversity, which includes both types of diversity and is relevant for both single firm- as well as interorganizational projects. The second interorganizational level is partner level diversity, which also includes both types of diversity, but is only relevant for interorganizational projects. The first level is measured through functional diversity, whereas the second level is measured through industry diversity to view whether there are complementary resources.
Board level diversity
Resource complementarity. Project management is changing due to introductions of new technologies and management practices, therefore new approaches are required to deal with the complexity, ambiguity, and uncertainty that come along with these introductions (Fellows & Liu, 2016). In order to benefit from the opportunities by improving performance to use these new technologies and management practices, collaboration is required through effective teamwork (Morris, 1998). More specifically, integrated teams are becoming more common place to drive performance (Morris, 1998). Project teams can especially benefit from a composition of individuals with broad functional backgrounds, as these have experience with different functional backgrounds and are therefore less influenced by functional biases and stereotypes (Bunderson & Sutcliffe, 2002). Moreover, intrapersonal functional diversity positively influences team process and consequently performance, as Bunderson & Sutcliffe (2002) found evidence for the influence of intrapersonal functional diversity on increased information sharing and therefore better performance. Other benefits associated with functional diversity include an increase of resources which means increased levels of information (Certo, Lester, Dalton & Dalton, 2006). Especially when decision-making is based on ill-defined or non-routine problems, multiple perspectives through functional diversity are required to operate effectively (Certo et al., 2006). Furthermore, Knight, Pearce, Smith, Olian, Sims, Smith, & Flood (1999) argue that functional diversity leads to increased creativity and innovation as teams are able to generate greater variance in decision-making. Moreover, other benefits associated with diversity in functional backgrounds is the ability to consider a wide range of perspectives and skills (Buyl et al., 2011) and an increase in the breadth of knowledge, which includes a wide range of experiences and capabilities to solve complex problems (Cannella et al., 2008). Some of the cognitive consequences of functional diversity are: increase in ability to process information, deal with stimuli, and make decisions, which leads to higher quality decisions and higher creativity (Miliken & Martins, 1996).
gender, age, values, and skills/knowledge. Diversity of skills and knowledge refers to educational background, functional background and range of industry experiences (Miliken & Martins, 1996). According to Cannella et al. (2008), differences in TMT will affect the functioning and success of the project but also provides a larger array of strategic choices and performance outcomes.
There have been two streams of logic when researching functional diversity. On the one hand, researchers argue that functional diversity increases the range of expertise and therefore team effectiveness. On the other hand, researchers argue that it inhibits team process and effectiveness as diversity comes with different perspectives. Therefore, functional diversity is argued to be “a double-edged sword, in that it has positive implications in some contexts and for some process or performance variables but negative implications in other contexts for other process or performance” (Bunderson & Sutcliffe, 2002, p. 875). Risks associated with dominant functional diversity are slow speed of decision-making, communication breakdowns, interpersonal conflict (Cannella et al., 2008), and team fragmentation (Buyl et al., 2011). Some of the benefits of increased information and multiple perspectives can also be a downfall for project teams if they are unable to deal with this effectively, which then leads to the risks mentioned above.
Partner level diversity
Resource complementarity. Nowadays more than ever, it is crucial to take into consideration the contingencies that affect the success of interorganizational projects as a paradox has been established (Kale & Sigh, 2009). This paradox shows that interorganizational projects are still increasingly established while at the same time, they are experiencing failure rates of 30 to 70%. From a theoretical perspective, both the resource-based view and relational view argue that some critical resources may extend beyond firm boundaries, creating a need to combine resources especially with distinct but complementary partners to create a synergistic effect (Barney, 1991; Dyer & Singh, 1998). Therefore, the strategic need for resources (Das & Teng, 2000) leads to better results when cooperating with partners that are more diverse in their resources and capabilities.
formation (Rothaermel & Boeker, 2008). Finally, another study compares when either commitment, complementarity and financial pay-off is more important for managers when selecting partners (Shah & Swaminathan, 2008). On the other hand, there has been some research on resource complementarity and project performance. Sarkar et al. (2001) examined the impact of several partner characteristics (e.g. complementarity) on project performance. Another study examines specific international resources and their influence on project performance (Choi & Beamish, 2013). Finally, the study of Hill & Hellriegel (1994) has an integrative approach of examining different aspects of project formation and their influence on performance.
Therefore, collaborating with firms that have complementary resources provides opportunities for economies of scope, creating synergies (Rothaermel & Boeker, 2008), economic benefits (Wang & Zajac, 2007), and developing new resources (Ireland, Hitt, & Vaidyanath, 2002). Pooling complementary resources is a common reason for the formation of interorganizational projects as it provides opportunities to produce a product or service faster and cheaper than either firm could achieve by itself (Harrison et al., 2001), but also increases the feasibility of some projects that could not be pursued alone (Chung et al., 2000). This pooling of resource is sometimes done by collaborating with partners from different industries or markets, hence industry diversity. Another reason resource complementarity is so important for collaborating is the required diverse disciplines for new product development, therefore needing a complementary partner (Deeds & Hill, 1996). Moreover, resource complementarity in interorganizational projects offers a mix of resources available to the project to achieve goals (Sarkar et al., 2001). Finally, collaborations with firms that have complementary resources has become an important way to increase competitive advantages and/or performance (Harrisson et al., 2001). The core logic of collaborating with a partner from other industries or markets is based on the premise that combining resources and competences with a partner provides opportunities to achieve objectives that are cannot be achieved by one partner alone (Mohr & Puck, 2005).
who examine the management of cultural diversity at different levels and identifies different tensions that arise.
Partner level diversity also experience functional diversity as they have two subsystems that interact with each other, respectively relational interface and within-interorganizational interface (Gong et al., 2005). Therefore, resource complementarity is only valuable if partners also have organizational complementarity. This includes similar cultures, decision making processes, and systems (Deeds & Hill, 1996; Dyer & Singh, 1998). Kale & Singh (2009) add that a partner needs to be compatible, so complementarity alone is insufficient. This translates to the need for cultural and operational compatibility, as Kim & Parkhe (2009) argued that performance is contingent upon both types of diversity. As interorganizational projects have to deal with these two subsystems simultaneously, the dynamics in a project become increasingly more complex due to having to collaborate with employees from the other partner. It is already difficult with diversity in functional backgrounds and skills when only one partner firm operates in a project, but adding another partner to the project worsens this tension by increasing the difficulty of integrating and coordination costs (Miliken & Martins, 1996). This tension is not only related to the functional backgrounds but also the corporate cultures the partners have. Therefore, especially interorganizational projects generate issues of governance, integration and commitment as various diverse members of the partners are put together (Fellows & Lui, 2016). Furthermore, the incompatibilities for single firms (e.g. slow decision-making, communication breakdowns, interpersonal conflict, and team fragmentation) affect interorganizational projects even more as these dynamics are more complicated (Cannella et al., 2008; Buyl et al., 2001). In the end, interorganizational projects have a to deal with more complex dynamics that influence the process, functioning and success of a project than single firm projects (Li, et al.,1999; Cannella et al., 2008). Hypotheses development
shows an increase in breadth of knowledge to help solve complex problems (Cannella et al., 2008). Furthermore, it also increases creativity and innovation, leading to a greater variance in decision-making (Knight et al., 1999). Finally, diversity on skill-based dimensions such as education and functional background show benefits for short-term cognitive consequences such as the ability to process information and make decisions (Miliken & Martins, 1996). They argue that skill-based diversity dimensions show divergent points of view and knowledge bases, which stimulates creativity and therefore increase decision-making and problem-solving.
On the other hand, from a cultural and operational incompatibility perspective, a negative effect of board level diversity on technological development project performance is expected. Cannella et al. (2008) argue that differences in the top management team affect the functioning and success of the project. Miliken & Martins (1996) argue that teams with diverse backgrounds and skills have integration problems, being unable to deal with a variety of perspectives and expertise, increasing coordination costs to effectively work together. Moreover, group diversity leads to task conflict which will lead to slower decision-making and breakdowns in communication and information processing (Miliken & Martins, 1996; Cannella et al., 2008). Moreover, it could even go as far as a team experiencing interpersonal conflict (Cannella et al., 2008) and team fragmentation (Buyl et al., 2011). Furthermore, some of the benefits (e.g. increased information and multiple perspectives) could also become a hazard if a team is unable to deal with this effectively, which then leads to a negative effect.
The prior studies presented thus far provide evidence that board level diversity tends to have both positive as well as negative consequences on technological development project performance. Therefore, this study expects an inverted U-shape effect of board level diversity on technological project performance for both single firm projects as well as interorganizational projects. Thus, the following hypothesis is developed:
H1: The relationship between board level diversity and on technological development project performance follows an inverted U-shape curve.
First of all, interorganizational projects have a more difficult time to achieve high performance change due to several negative board level diversity influences that they deal with. According to Parkhe (1991), all five dimensions (social culture, national context, corporate culture, strategic direction, management practices and organization) of board level diversity will have a negative influence on interorganizational project performance change, therefore flattening the inverted U-shaped relationship between board level diversity and technological development project performance. Moreover, interorganizational projects can have diverse national backgrounds, again implying negative effects on performance as these partners have many personal differences (e.g. values) which takes time to get over (Miliken & Martens, 1996). These differences can be found in both national cultures that have different social institutions as well as internal cultures that have different management practices (Hambrick, Li, Xin, & Tsui, 2001). Other problems that are already present within diverse projects of one partner but become increasingly more complex when adding another partner, is the speed of decision-making, communication breakdowns, interpersonal conflict (Cannella et al., 2008), and team fragmentation (Buyl et al., 2011). In the end, interorganizational projects experience more difficulty during the process, functioning and success of a project than single firm projects (Li, et al.,1999; Cannella et al., 2008). These differences do not only affect the level of performance change an interorganizational project can achieve, but also when they experience their peak. Tensions worsen due to an extra partner, making it more difficult to integrate and coordinate activities, which increases complexity (Miliken & Martins, 1996). Therefore, interorganizational projects experience these difficulties earlier on than single firm projects, shifting the inverted U-shaped relationship between board level diversity and technological development project performance to the left.
Prior research also finds empirical evidence for a more negative effect of board level diversity in the context of interorganizational projects. For example, studying strategic alliances formed by US firms, Lin & Chen (2002) find a negative relationship between diversity and interorganizational performance. Specifically, diversity on cultural and management dimensions seem to have a negative and significant impact on interorganizational performance (Lin & Chen, 2002). This shows that interorganizational projects’ performance worsens at lower levels of diversity and therefore experience their peak earlier. Likewise, studying German-Chinese JVs, Mohr & Puck (2005) find that functional diversity has a negative influence on interorganizational performance. They argue that diversity leads to inefficient task performance which leads to misunderstandings and conflicts but they also recognize the need to find ways to manage diversity (Mohr & Puck, 2005). Finally, Sarkar et al. (2001) found insignificant or negative influences of operational compatibility on project- and strategic performance. These final few studies prove that interorganizational projects experience lower performance change, therefore a flatter inverted U-shaped curve.
U-shaped relationship between board level diversity and technological development project performance that peaks earlier in comparison to single firm projects. Thus, the following hypothesis is developed:
H2: The relationship between board level diversity and technological development project performance is negatively moderated by the presence of interorganizational projects in such a way that the inverted U-shaped relationship will be flatter and peaks earlier.
Interaction between board level diversity and partner level diversity in interorganizational projects. Apart from the separate influences of board level diversity and partner level diversity on technological development project performance, Kim & Parkhe (2009) argue that interorganizational performance is contingent upon both types of diversity. Therefore, prior research lacks an integrative approach researching the interaction effect of both board level diversity and partner level diversity on technological development project performance in interorganizational projects. These interorganizational projects are perceived to be multilevel in nature, involving both relationships between the project and each partner as well as between partners (Ren, Gray, & Kim, 2009). Likewise, Robson, Leonidou & Katsikeas (2002) propose a framework that shows both types of diversity to be part of interpartner fit and that they influence each other, therefore determining whether the partners are able to efficiently work together without too much conflict. According to Chung et al. (2000), resource complementarity is the motive for forming interorganizational projects especially in high-growth industries as maintaining a competitive advantage becomes increasingly more challenging. This is in line with Das & Teng (2000), who argue that firms have a strategic need to collaborate in order to pool resources as resources are heterogeneous and not perfectly mobile.
Finally, Gong et al. (2005) study the effects of interactions between the subsystems of interorganizational projects focusing on human resources. The two subsystems being the system within the project and the second system being the relationship between the partners and the project. Gong et al. (2005) find that the effect of the second system (resource complementarity) on the first system (cultural and operational compatibility) is positive as the systems are interrelated. In other words, the benefits that complementary resources bring offer opportunities to achieve higher performance change. Therefore, the inverted U-shaped curve is expected to become steeper when board level diversity and partner level diversity interact.
The prior studies presented thus far provide evidence that especially from a resource complementarity perspective, partner level diversity tends to have a more positive interaction with board level diversity. This interaction provides the opportunity to integrate and create skills and knowledge that leads to higher performance change, therefore steepening the inverted U-shaped relationship between board level diversity and technological development project performance. Therefore, this study expects the interaction between board level diversity and partner level diversity steepens the inverted U-shaped curve between board level diversity and technological development project performance. Thus, the following hypothesis is developed:
H3: The relationship between board level diversity and technological development project performance for interorganizational projects is positively moderated by partner level diversity in such a way that the inverted U-shaped relationship will be steeper.
Figure 1. Conceptual model
Technological development project performance Control variables Single firm vs interorganizational projects Firm Size Firm age Project size Project Age Board Size Board Age Board level diversity
Moderators:
Single firm vs interorganizational projects (H2: -) Partner level diversity (H3: +)
METHODOLOGY
This study follows a theory testing approach as the business phenomenon of diversity is already conceptualized theoretically. However, there is still no conclusive evidence reporting how diversity influences performance, which is why theory testing is the appropriate research process according to Van Aken, Berends, and van der Bij (2012).
Research setting1
Both single firm and interorganizational projects are pursued within the wind energy industry. Nevertheless, nowadays collaborations occur more frequently, which are often formed between firms to deal with the dynamic nature of the industry, concerning high costs, capital requirements and complexity (Dedecca et al., 2016; Jacobsson & Karltorp, 2013). Wind power offers a clean and sustainable way of generating electricity, which is important to reduce further damage to the environment while keeping up with the increasing electricity demands (Jahanpour, Ko, & Nof, 2016). However, there is much uncertainty due to the importance of wind direction, wind speed, and weather conditions but also failures and maintenance, showing that there is limited controllability and much variability involved in the operability of a windfarm (Jahanpour et al., 2016). The reasoning for diversity within these interorganizational projects occur frequently due to the multi-disciplinary nature of the industry, requiring different types of experts throughout the project of developing wind farms (e.g. project developers, wind turbine manufacturers, engineering companies) (Dedecca et al., 2016). Both the initiatives taken by the European Union to use more low-carbon power systems to stabilize climate change (Jacobsson & Karltorp, 2013) and the Paris agreement adopted in December 2015 to hold the global average temperature below 2 degrees (Rogelj, den Elzen, Höhne, Fransen, Fekete, Winkler, Schaeffer, Sha, Riahi, & Meinshausen, 2016), have increased investments in wind energy solutions. Therefore, this study focuses on the wind energy industry that involve both board level diversity and partner level diversity within projects.
The UK is one of the prominent leaders in the wind energy industry, establishing important policies to foster the development of wind farms (International Energy Agency, 2016). Moreover, this market offers their data regarding wind farm projects online, free to access for all (Companies House, 2017). Therefore, a choice was made to collect data regarding wind farms in the UK. Some of this available data is shown in figure 2, which clearly shows how the amount of installed capacity and energy generated grows at a faster pace from 2010 on (Renewable Energy Foundation, 2016). In the years before 2010 a steady but slow growth is shown. However, there are more reasons 2010 and 2011 were so important in the wind energy industry. One of the most important policies established was the National Renewable Energy Action Plan (NREAP) in 2010, which was an obligation from the EU
Directive to be handed in to the European Commission to outline the plans to make the 2020 renewable energy targets (International Energy Agency, 2016). In this plan, The UK planned to have 15% share of energy generated from renewable energy by 2020, in which most measures taken will entail regulatory by encouraging planning authorities to support the development of renewable energy sources and financial in providing funds for wind farm development projects by for example banks (European Commission, 2017). This plan is still in force as of today and includes many more measures to encourage renewable energy and also specifically wind energy development. Hence, based on the policies established in 2010 and the growth rates around 2010 and 2011, this study focused on collecting data from the UK for the years 2010 and 2011.
Figure 2: Installed capacity and energy generated between 2002 and 2016 (Source: Renewable Energy foundation).
Data collection
The first database used was the Companies House database to identify projects formed in the wind energy industry in the period from 2010 to 2011. This database is appropriate as it consists of all information about particular wind farms (e.g. partner names, board of directors, financial data) and any changes made. This database provided the opportunity to collect rich data not usually available in this industry. The data collection process started by identifying wind farm projects in the database based on the SIC code 35110, which stands for production of electricity. The collection process was done
manually, which required opening multiple files for each project for most data but also reporting on any changes within the project. The data collected about the wind farm projects included date of incorporation, date of dissolution, the partners, changes in partners, changes in board of directors, financial performance, occupation of board of directors, project size, board size and finally board age. Initially, 200 wind farms were considered for the period of 2010-2011. As a result of being limited to available data, the sample excluded wind farm projects that did not have any available data on financial performance or were missing data for other variables necessary for the dataset.
Since the Companies House database did not contain all necessary information on the wind farms projects, a second database was used to complement the data from the sample. This second database was Orbis in order to find partner-related information such as firm size, firm age, industry codes, date of incorporation. This database is appropriate as it offers company information over years. When the required data from Orbis was collected, the dataset was complete. Finally, the data collected was hand coded in Excel in order to build a comprehensive dataset that could be used in SPSS to run analyses. The data collection process was time-consuming due to the manual data collection, but also very rich as the database provided data that is often not publicly available. The measures part below will explain how each variable in the dataset was transformed to the appropriate value. Different measures were tried for the control variable firm size, but the decision for total assets came down to the most available data in Orbis for the observations.
Sample size
This study initially identified 200 wind farm projects for period of 2010 – 2011. As the goal of this study is to see the effects of board level diversity on project performance, data from 2010 was not sufficient as the data collected occur in the same year. Therefore, data from both 2010 and 2011 was collected, resulting in a sample of 400 observations. This way the relationship between board level diversity as dependent variable and project performance as independent variable could be tested with a year lag. Unfortunately, due to a high number of missing data in the observations, the sample size became too small. This happened because any observations without financial performance data in 2011 had to be excluded. Due to the rich data collection process and coding of data, the sample size had to be sacrificed. Meaning, there was no time to collect more data to increase the sample size. Therefore, the dataset collected was combined with the dataset of a colleague (Toufic Elcure Alvarez). This colleague collected some similar data in the wind energy industry, however had a different research setting, requiring extra data collection to supplement the missing data for this study. Hence, the final sample size is 85 wind farm project observations.
Measurements
has influence on the performance, a year lag was applied by using the performance data of 2011 to show the change in performance. This data has been used to calculate the change in performance to see if diversity actually caused change. Therefore, the measure for technological development project performance is performance change. In order to do so, the following formula was used: Performance change = Performance 2011Performance 2010− 1.
Independent variable. The primary focus of this study is the effect of board level diversity on technological development project performance. The Blau Index has been applied in order to transform the collected data into useful values, following the approach of Oerlemans, Knoben, & Pretorius (2013), Bunderson & Sutcliffe (2002), and Vlaisavljevic, Cabello-Medina, & Pérez-Luño (2015). Board level diversity. Similar to other studies, functional diversity, in other words the occupations of the board of directors in the projects, are applied for the measure board level diversity (Mohr & Puck, 2005; Buyl et al., 2011). This shows whether the directors from different functional backgrounds are able to work together effectively. For the analyses, all these functional occupations were categorized the following way: company director, finance, operations, other directors, managers, and other. The category company director included company directors and directors, which are the same functional background however both used in the documentation of wind farm project files. The next category finance included functions such as finance controller, fund manager, accountants, and group treasurer. The following category operations included head of project and operations, engineers, and head of wind operations. Following, the category other directors included legal directors, project directors, and human resource director. The category managers included commercial manager, investment manager, asset manager, and business manager. Finally, the category other included functions unexpected in the wind energy industry, such as cheesemaker, handyman, farmer, architect, and district nurse. Based on these categorizations, the Blau Index was calculated following this formula: 1- Pi2 (Bunderson & Sutcliffe, 2002). First, to calculate the portions for each category, the amount of times a category appears was divided by the total amount of directors in the wind farm project. Then these portions were squared and finally all deducted from 1. This lead to the Blau Index values for each wind farm project.
Moderators. Single firm vs Interorganizational projects. The number of partners active in any wind farm project in 2010 were applied as a measure to decide whether a single firm or multiple firms were involved in the project. In order to categorize these results, a ‘1’ was given if there were multiple partners in 2010 (interorganizational project). Otherwise, a ‘0’ was given for the variable (single firm project).
transformed to the wind farm project. As mentioned above, the Blau Index was used, to find values between 0 (low diversity) and 1 (high diversity). The formula used to calculate these values again was: 1- Pi2 (Bunderson & Sutcliffe, 2002). A first step in this calculation entailed dividing the amount of times a particular industry appeared in one wind farm project by the total amount of partners in that project. All these values are then squared and finally deducted from 1 to get the Blau Index value.
Control variables. This study controlled for various firm- and project-specific factors that might influence technological development project performance. In order to measure firm size, similar to prior research the value of total assets of a partner has been used (Lavie & Rosenkopf, 2006). Several studies found that size is interrelated with both performance and resource endowments (Oerlemans et al., 2013). Moreover, size offers economies of scale, which is a clear advantage large firms have over SME’s (Umans, 2013). As the data for total assets was collected for partners involved in the project, an average was taken from all partners involved in a wind farm project to find the final value of total assets of the wind farm project. For the analyses, the variable was transformed using a natural logarithm to improve the non-normal distribution of total assets. Another control variable use by prior research is firm age, for which the measure date of incorporation of the partners has been used. Older firms are more established and mature, providing opportunities to benefit more than SME’s (Lavie & Rosenkopf, 2006). The data collected for date of incorporation was transformed using a natural logarithm improving the non-normal distribution. The following control variable is project size, measured by the wind farm size in MW. For the analyses, the variable was transformed using a natural logarithm to improve the non-normal distribution. The next control variable is project age, for which the measure date of incorporation of the project has been used. The data collected for date of incorporation was transformed using a natural logarithm improving the non-normal distribution. Another control variable is board size, similar to prior research it has been measured by the number of individuals in the board. Several studies found board size to have important implications on the processes and outcomes within a project (Bunderson & Sutcliffe, 2002; Umans, 2013). Moreover, information sharing and decision-making within larger groups becomes increasingly more challenging (Bunderson & Sutcliffe, 2002). The final control variable is board age, measured by the ages of the directors in the board. As almost all projects had more than one director, an average of all these directors’ ages was taken for the wind farm project. Statistical analysis
relationship between board level diversity and technological development project performance is for only interorganizational projects. For the final hypothesis, a moderation analysis was performed by constructing a new model including the variable partner level diversity. As the sample of interorganizational projects is very low, additional analyses were performed to conclude reliable results for the hypothesis. A comparison of the means of board level diversity was performed between either low or high partner level diversity (also comparing young and old projects). Finally, cross tabulation was performed to compare high and low partner level diversity levels and the differences in the levels of board level diversity and the levels of technological development project performance. This study uses unstandardized coefficients (b) as they provide meaningful information and no comparison needs to take place between the coefficients. If comparison was needed, then unit of measurement becomes important and using standardized coefficients is more logical although it still complicates the interpretation process (King, 1986). Behind the coefficients between brackets, the standard errors are presented.
RESULTS
This study examines the effect of board level diversity on technological development project performance and how single firm projects vs interorganizational projects and partner level diversity moderate this relationship. First, the descriptive statistics and correlations of the variables will be discussed as shown in Table 1. Afterwards, the results of the regression analyses and additional analyses (comparison of means and cross tabulation) will be discussed, whose results are shown in table 2 till 9. Descriptive Statistics
The results presented in table 1 do not show extreme correlation values to indicate multicollinearity. To conclude that there is no presence of multicollinearity in the sample, a collinearity test was performed. None of the values exceeded the variance inflation factors (VIF) standard of 10 (Mason & Perreault Jr., 1991). Board size has the highest VIF value (2.218) and board age has the lowest VIF value (1.144). The independent variable functional diversity has a VIF value of 1.590. Table 1. Descriptive statistics and correlations
Hypothesis testing
Table 2 shows the results of the regression analysis for the hypothesis testing of the independent variable functional diversity and the dependent variable project performance. Model 1 presents the results of the control variables. The variable single firm vs interorganizational projects is used as a control variable for hypothesis 1, but later is introduced as a moderator. Only the coefficients for single firm vs interorganizational projects are significant, negatively impacting project performance (b = -.182, p-value < .05). Model 2 adds the independent variable functional diversity and its squared counterpart to test whether a non-linear relationship exists between functional diversity and project performance (H1). Model 3 introduces the interaction between functional diversity and single firm vs interorganizational projects and its squared counterpart. Model 4 shows the coefficients for the control variables of hypothesis 3. Then, the independent and moderator variables are added in model 5 to test the relationship between functional diversity and project performance for interorganizational projects. Finally, model 6 shows the results of the interaction effect between functional diversity and industry diversity and their effect on project performance.
Hypothesis 1 suggests that board level diversity has an inverted U-shaped relationship with technological development project performance. Contrary to the hypothesized effect, the results find a U-shaped relationship between board level diversity and technological development project performance. Looking at model 2, functional diversity shows an initial significant negative coefficient (b = -.588, p-value < .10) followed by a significant positive coefficient for its squared counterpart (b = 1.023, p-value < .05). This unexpected finding will be further discussed in the discussion section. However, the results should be taken with caution as the adjusted R2 is only 5,6%, indicating that the
hypothesize model only explains 5,6% of the variance of board level diversity and technological development project performance. Figure 3 shows that there is an actual U-shaped relationship. As the results did not find an inverted U-shape relationship between board level diversity and technological development project performance, hypothesis 1 is not supported.
Table 2. Regression results
Figure 3. The relationship between board level diversity and technological development project performance.
The second hypothesis of this study suggests that there is a negative moderating effect on the relationship between board level diversity and project performance when interorganizational projects are present in such a way that the inverted U-shaped relationship will be flatter and peaks earlier (H2). Haans et al. (2016) argues that a flattening effect would occur for inverted U-shaped relationships if b4 is significant and positive. However, as the results of the regression analyses show a U-shaped relationship and were insignificant, any calculations to determine a flattening effect and earlier peak would be inaccurate. Therefore, no conclusions can be made on the elasticity or shifting of the curve. As the distribution of the moderation variable is not even, additional analyses have been performed to improve the comparison quality by creating sub-samples of the same size. Table 3 shows the comparison of the relationship between board level diversity and technological development project performance of single firm projects and interorganizational projects. Model 1 of table 3 shows the coefficients of the control variables of the single firm projects. Model 2 shows the results of the relationship between board level diversity and technological development project performance for the single firm projects. Model 3 reports coefficients of the control variables for interorganizational projects and finally model 4 reports the results of the relationship between board level diversity and technological development project performance for interorganizational projects. Model 1 shows significance for a positive effect of projects size (b = .718, p < .010). Unfortunately, both the results for single firm projects in model 2 as well as the results for interorganizational projects show insignificant coefficients so no conclusions can be drawn. Possible explanations for the insignificance is the low sample size (N = 18) and the low adjusted R2.
0
0.3
0.6
0.9
Low Board level diversity
High Board level diversity
Table 3. Regression results comparison single firm projects with interorganizational projects (H2)
Another analysis performed is a comparison of the means between single firm projects and
interorganizational projects, a similar analysis to Li (2013) who compare the means of two samples of the independent variable. Based on the results of table 4, single firm projects seem to experience more functional diversity within their projects than interorganizational projects. However, the results do not differ that much from each other. Table 5 shows that there is a difference in the amount of diversity experienced between older single firm- and interorganizational projects especially. For young projects, this difference is not visible. The final analysis performed for hypothesis 2 is a cross tabulation, comparing how low or high functional diversity result in either low, medium or high performance change between single firm projects and interorganizational projects. The direct
measures of the cross tabulation of table 6 shows a significant positive association of .669 (p = .004), indicating that this study improves its guess of performance change by 66,9% when functional diversity is known (results found in Appendix A). Table 6 shows that single firm projects experiencing high functional diversity show higher performance change (75% out of total) than interorganizational projects. The latter firstly shows more balance between low and high functional diversity, concluding that there are also quite some interorganizational projects that experience low performance change. Moreover, the high performance change for interorganizational projects is only 37,5% of the total. Therefore, a first indication is found that interorganizational projects do flatten the
Variables Model 1 Model 2 Model 3 Model 4
U-shaped relationship between board level diversity and technological project performance as achieving high performance change happens less often than for single firm projects. However, as the sample size is too small no final interpretations are able to be made at this time. Therefore, hypothesis 2 is not supported.
Table 4. Comparison of matching samples on independent variable (H2)
Table 5. Comparison of young and old projects (moderator H2) on independent variable
Table 6. Cross tabulation of matching samples on independent and dependent variables (H2)
Mean (overall sample)
Mean (Single firm projects) Mean (Interorganizational projects) Sample Size 106 18 18 Functional Diversity .3930 .4727 .4397 (Functional Diversity)2 .2266 .2734 .2476 Mean (overall sample) Mean (Young single firm projects) Mean (Young interorganizational projects) Mean (Old single firm projects) Mean (Old interorganizational projects) Sample Size 106 5 13 10 9 Functional Diversity .3930 .4722 .4727 .3677 .5198 1 Low Performance Change Medium Performance Change High Performance Change Total Low Functional Diversity Count % of Total 0 0,0% 0 0,0% 1 6,3% 1 6,3% High Functional Diversity Count % of Total 1 6,3% 2 12,5% 12 75,0% 15 93,8% Total Count % of Total 1 6,3% 2 12,5% 13 81,3% 16 100,0% 2 Low Performance Change Medium Performance Change High Performance Change Total Low Functional Diversity Count % of Total 4 25,0% 2 12,5% 0 0,0% 6 37,5% High Functional Diversity Count % of Total 1 6,3% 3 18,8% 6 37,5% 10 62,5% Total Count % of Total 5 31,3% 5 31,3% 6 37,5% 16 100,0%
The final hypothesis suggests that there is a positive moderating effect of partner level diversity on the relationship between board level diversity and project performance in such a way that the inverted U-shaped relationship will become steeper (H3). Haans et al. (2016) argues that a steepening effect would occur for inverted U-shaped relationships if b4 is significant and negative. However, neither of the coefficients of the interaction effect between functional diversity and industry diversity and its squared counterpart show any significance, so no conclusions can be drawn from these results regarding the of the curve in general and its elasticity. As the sample for this analysis was so low, additional analyses have been performed. The first additional analysis is a comparison of the means of functional diversity between low- and high industry diversity. Table 7 shows a comparison of the means, in which high partner level diversity, otherwise when the partners bring in complementary resources, correlates with high functional diversity. Moreover, table 8 shows a huge difference in young projects experiencing low or high partner level diversity. Specifically, young projects with high partner level diversity experience a lot more functional diversity. For old projects this difference is a lot smaller. The direct measures of the cross tabulation of table 9 shows a significant positive association of .767 (p = .000), indicating that this study improves its guess of performance change by 76,7% when functional diversity is known (results found in Appendix B). For the final analysis, Table 9 shows that projects experiencing high industry diversity and high functional diversity show more potential for medium to high performance change than project experiencing low industry diversity. Therefore, a first indication is found that high industry diversity has a steepening effect on the relationship between board level diversity and technological development project performance. However, as the sample size is too small no final interpretations are able to be made at this time. Thus, hypothesis 3 is not supported.
Table 7. Comparison of matching samples (partner level diversity) on independent variable (H3)
Table 8. Comparison of young and old projects (moderator H3) on independent variable Mean (overall
sample)
Mean (Low Industry Diversity)
Table 9. Cross tabulation of matching samples (partner level diversity) on independent and dependent variables (H3)
Robustness checks
In order to check whether the results of the analyses are robust, additional analyses were performed. For the first hypothesis, after running a regression analysis, a surprising U-shaped relationship was found between board level diversity and technological development project performance. In order to confirm whether there is an actual U-shaped curve, some calculations were made according to Haans et al. (2016) three-step procedure. The following formula was assumed: 𝑌 = b-+ b/𝑋 + b1𝑋1. The
first condition consists of a significant b2 and of the hypothesized sign, which were found in the
regression results, showing a positive and significant b2. According to Haans et al. (2016), this
indicates a U-shaped relationship. The second condition is that there has to be a sufficiently steep slope for both sides of the curve, which is calculated according to the following two formulas: 𝑋234 =
𝑏/+ 2𝑏1𝑋7 and 𝑋9:;9 = 𝑏/+ 2𝑏1𝑋<. A low X would be to the left of the tipping point and a high X would be to the right of the tipping point, so low X was set to 0.1 and high X was set to 0.7 (as the maximum value for functional diversity was 0.8). The following slopes were found, 𝑋7= -0.3834 and 𝑋<= 0.8442. These are steep enough slopes at both ends of the curve and also show the expected signs
of a U-shaped curve. The last condition is that the tipping point needs to be within the data range. The formula for the tipping point is − =>
1=? (Bianchi, Croce, Dell’Era, Di Benedetto, & Frattini, 2016) and
found a tipping point at X = 0.29. This point is well within the data range. These analyses confirm the U-shaped curve found in the first hypothesis.
1 Low Performance Change Medium Performance Change High Performance Change Total Low Functional Diversity Count % of Total 4 57,1% 0 0,0% 0 0,0% 4 57,1% High Functional Diversity Count % of Total 0 0,0% 0 0,0% 3 42,9% 3 42,9% Total Count % of Total 4 57,1% 0 0,0% 3 42,9% 7 100,0% 2 Low Performance Change Medium Performance Change High Performance Change Total Low Functional Diversity Count % of Total 0 0,0% 2 22,2% 0 0,0% 2 22,2% High Functional Diversity Count % of Total 1 11,1% 3 33,3% 3 33,3% 7 77,8% Total Count % of Total 1 11,1% 5 55,6% 3 33,3% 9 100,0%
Additional analyses have been performed to exclude the possibility of a linear relationship. For the first hypothesis, the table in appendix C shows a linear regression, in which model 1 shows the control variables and model 2 shows the linear relationship between board level diversity and technological development project performance. Model 2 shows a positive coefficient indicating an increasing linear line, however this result shows no significance. Moreover, the adjusted R2 of 1,9% is worse than in the regression analysis run for the curvilinear relationship. Both hypotheses 2 and 3 do not provide any other insights as the results of the interactions were all found insignificant when testing a linear relationship. As mentioned in the methodology section, another variable was also considered as a conceptualization of board level diversity, specifically age diversity. Therefore, some additional analyses were performed using age diversity instead of functional diversity to see whether the U-shaped relationship is also found when using a different variable. The table in appendix D shows a curvilinear relationship, in which model 1 shows the control variables, model 2 the linear and squared terms of age diversity, and model 3 shows the moderation analysis by the presence of
interorganizational projects (H2). Model 2 finds a similar relationship between board level diversity and technology development project performance when using age diversity as the conceptualization of board level diversity, however the results are not significant. Finally, model 3 shows again a similar curve when age diversity is moderated by the presence of interorganizational projects. Therefore, the U-shaped relationship between board level diversity and technological development project performance seem to be robust when using other variables and a linear relationship is excluded from consideration due to the insignificant results.
DISCUSSION
Part of this section are the theoretical- and managerial implications which will discuss the results of the statistical analyses in more depth by giving possible explanations for the unexpected results. Finally, the limitations and future research of this study are given.
Theoretical implications
of interorganizational projects on this relationship and within this focus researching the interaction effect between the both types of diversity on technological development project performance.
This study contributes to the literature on diversity by identifying the need for an integrative approach of the two types of diversity and their influences on each other. By looking further than the prior studies researching linear relationships between diversity and performance for either single firms or interorganizational projects, this study has identified a curvilinear relationship and also the comparison of single firm versus interorganizational projects. The results are however unexpected. Contrary to the hypothesized inverted U-shaped relationship, this study finds a U-shaped relationship between board level diversity and technological development project performance. However, these results should be taken with caution as the adjusted R2 is only 5,6%. These results imply that as board
decision-making, Parkhe (1991) argues that interorganizational projects may need to set up unitary management processes and structures to deal with the negative influence of board level diversity on the process and actually proposes it as a moderator. Mohr & Puck (2005) also argues and finds support for a negative impact of board level diversity as it reduces efficient operations and finds that adaptation is a significant moderator, reducing the negative effects of board level diversity. Parkhe & Kim (2004) also argue that the role of adaptation has a moderating effect on the relationship between board level diversity and project performance. Thus, based on these prior studies it can be argued that there is an initial decrease in performance as diversity hurts the effectiveness of a project’s process. Moreover, a project is better able to deal with the negative consequences on a projects’ process and improve performance when they adapt their processes and practices to increase the effectiveness of the team, hence the U-shaped relationship.
The other two hypotheses contribute to the diversity literature by introducing the need to compare the differences between single firm- and interorganizational projects and later on the interaction effect of board level diversity with partner level diversity within interorganizational projects. Neither of the final two hypotheses found significance. However, the results of cross tabulation show an initial indication that interorganizational projects more negatively moderate the relationship between board level diversity and technological development project performance in such a way that the curve flattens. The results imply that interorganizational projects have a more difficult time improving performance when experiencing high functional diversity in comparison to single firm projects. Partners collaborating together may have different management practices and styles (Mohr & Puck, 2005) that make it more difficult to work effectively. The final hypothesis contributes to the diversity literature by emphasizing the integrative approach of the two types of diversity. Based on a cross tabulation, the results imply that projects experiencing high industry diversity show more potential for medium to high performance change than project experiencing low industry diversity. This means that a first indication is found that high industry diversity indeed has a positive effect on the relationship between board level diversity and technological development project performance in such a way that the curve steepens. However, as the sample for both hypotheses was too low, future research is required to conclude whether these first indications hold.
Managerial implications
should consider whether they are willing to adapt processes to be able to operate efficiently as diversity can cause great process losses otherwise (Miliken & Martins, 1996). Finally, managers need to take into consideration whether they want to pursue a project by themselves or with a partner. Collaborating with another partner might seem like the preferred choice, however it seriously complicates operations. This study found a first indication that seems to show that interorganizational project have a more difficult time to benefit from board level diversity as a more negative relationship ensues. One of the reasons for the more negative relationship is the partner level diversity that adds complexity as the different partners may differ in management practices and styles (Mohr & Puck, 2005), which may complicate operations to achieve efficiency while working together.
Limitations and further research
There are several limitations for this study that need to be taken into consideration. First, the study was affected by the choice to research one specific industry. As the industry is relatively young, the availability of data is somewhat limited. One of the reasons for the limited available data is the fact that Companies house is dependent on whether all relevant data of the wind farm projects are reported by the partners. Moreover, additional data extracted from Orbis was also not complete. This lead to a small main sample size (N = 85), but even smaller for the following hypotheses (interorganizational projects N = 19, partner level diversity N = 9). Unfortunately, due to the rich and time-consuming manual data collection process and coding of data, there was no time to increase the size by collecting additional data from more wind farm projects. Future research could cope with this limitation by examining these relationships in a different industry that has more available data to facilitate building a comprehensive sample size.
the variables. Furthermore, future research should take into account more long-term effects of diversity on performance by researching multiple performance years.
Finally, this study only used single measures to test the variables, consequently resulting in low R2. Future research could cope with this limitation by combining multiple measures to operationalize a
specific variable, increasing the variance of the model as the variables are better explained. Moreover, this study did not include cultural compatibility as part of partner level diversity as most wind farms were UK based. However, future research could improve the explanation of both board level- and partner level diversity by including this aspect and examining the relationship in a more international setting.
CONCLUSION
The purpose of this study is to examine the relationship between board level diversity and technological development project performance, using a more integrative approach of two types of diversity and comparison single firm and interorganizational projects. Using a sample of 85 wind farm projects in the wind energy industry over the years 2010 and 2011, several quantitative analyses have led to the following results. This study provides evidence that the relationship between board level diversity and technological development project performance is non-linear. However, contrary to the hypothesized inverted U-shaped curve, the results show a U-shaped curve. This contributes to the diversity literature about the influence a context can have on the dynamics of the relationship. However, the results should be taken with caution with an adjusted R2 of only 5,6%. Furthermore, although the other two hypotheses
