Upper echelons, dynamism, and external corporate venturing : predictions of governance mode choice through internal and external perspectives using multinomial logistic regression analysis

(1)

Master’s Thesis

Executive Program in Management Studies – Strategy Track University of Amsterdam

Upper Echelons, Dynamism, and External Corporate Venturing

Predictions of Governance Mode Choice Through Internal and External

Perspectives Using Multinomial Logistic Regression Analysis

Supervisor: Dr. Marten Stienstra Student: Richard Koopman Student number: 11236566

Date: August 31, 2018 Version: Final

(2)

Statement of Originality

This document is written by Richard Koopman who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources

other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for

(3)

Abstract

Acquiring external knowledge and technology has proven vital for firms to cope with the relentless pace of today’s markets. When venturing outside existing boundaries, firms adopt different modes of governance, which allow for either flexibility and adaptability, or control and coordination. Despite its evident importance in strategic decision-making, the role of the executive team has received relatively little attention within the realm of external corporate venturing. Through the framework of upper echelon theory, this study examined the effects of top management team’s composition, in terms of heterogeneity and executive tenure, on governance mode choice. Furthermore, environmental contextual effects on governance mode decisions were captured through the construct of market dynamism, and examined for its main and moderating effects. Four distinct modes of external corporate venturing were within the scope of this study, being venture capital investments, minority interests, joint ventures, and acquisitions. A multinomial logit approach was adopted, recognizing the unique nature of each venturing option.

Results indicate that the degree of heterogeneity within the executive team, as well as the degree of market dynamism, positively influences the likelihood of firms adopting arm’s length modes of governance, over hierarchical modes of governance. A linear ranking of governance modes on the markets-hierarchy continuum, though, was only partially supported by the data. CEO tenure had a similar directional effect, pointing towards an increased likelihood of arm’s length modes of venturing, over hierarchical modes of governance, a sharp contrast to the hypothesized effect. The data showed no evidence of market dynamism interacting with either top management team heterogeneity or executive tenure, in relation to external corporate venturing governance mode choice. Further implications and conclusions are provided, as well as directions for future research.

(5)

1 Introduction

Firms in today’s markets are frequently disrupted due to unpredictable technological, economic or political changes in their environment. As markets become more dynamic and uncertainty is abound, there is at least one certainty for all firms: responding to disruptive events is a requisite to keep pace with the market and ensure firm survival, or, rather, capitalizing on new business opportunities presented by disruptive events. In such setting, firms ability to strategically renew themselves, by refreshing or replacing organizational attributes through, amongst others, the acquisition of knowledge or technology, has been argued to be critical for sustained success and firm survival (Agarwal and Helfat, 2009). It has been much emphasized in academic literature that firms source new knowledge or technology through external corporate venturing (Schildt et al., 2005), and firms use different modes of corporate venturing to learn and adapt.

While many studies have explored different antecedents of governance mode choices in the context of external corporate venturing, this study joins a small group of research (e.g., Schildt et al., 2005; Keil et al., 2008; Van De Vrande et al., 2009; Van De Vrande, 2013) that identifies multiple external corporate venturing options as discrete choices. As such, it is acknowledged that firms have multiple sourcing modes at their disposal, and each mode has its (dis)advantages. Furthermore, this study contributes to external corporate venturing literature, by aiming to provide an unexplored explanation for the variance that is observed in investment vehicles adopted by firms, when sourcing knowledge or technology through external corporate venturing. Firm executives have a pivotal role in external sourcing decisions, resource allocations, and responding to disruptive environmental changes (Maula et al., 2013); such key strategic decision-making lies at the firm’s strategic apex, the domain of the firm’s top management (Mintzberg, 1979). This, then, forms the basis for upper echelons theory; strategic trajectories and subsequent performance outcomes, stem from the idiosyncrasies of the firm’s top executives (Hambrick and Mason, 1984). Therefore, it is hypothesized here that the top management team, and, more specifically, their composition in terms of heterogeneity, shape firms’ strategic renewal initiatives, which are operationalized as multiple external corporate venturing modes (i.e., corporate venture capital, minority interests, joint ventures, and acquisitions).

As discussed by Hambrick (2007) and Nielsen (2010) in their recent reviews of the extent upper echelons literature, most studies have empirically tested, and confirmed, direct influences of top

(6)

management team characteristics on constructs of firm performance. However, the basic logic of upper echelons theory as first introduced by Hambrick and Mason (1984), posits that top management teams make strategic choices, through a filtered and limited perception of the environment, based on their cognitive base and values, which is subsequently followed by performance outcomes. By hypothesizing and testing the asserted relationship between top management team characteristics, and external corporate venturing choices, this study contributes to the academic literature on upper echelons theory, by empirically testing how top management team composition, and their orientation towards either innovation or efficiency, influences the choice of external sourcing initiatives by firms, rather than directly measuring the outcome of such strategic choices. Additionally, next to the top management team’s composition, chief executive tenure is introduced as an important antecedent of such strategic choices, as the top management team’s chief executive is typically its most powerful actor (Mintzberg, 1979). CEO tenure, like top management team composition, has been related to outcome measures such as firm innovation, or financial performance (Miller, 1991; Wu et al., 2005; Simsek, 2007). Yet again, a priori, executives are strategic decision-makers, and strategies are aimed at influencing firm performance.

Research on the relationship between top management teams and external corporate venturing is fragmented, and incomplete. Most studies define a broader concept of organizational outcome that is consequential to top management team composition, for example, as the firm’s strategic innovation orientation (Talke et al., 2010; Talke et al., 2011). In an international venturing context, top management team composition was found to influence a firm’s foreign entry mode (Nielsen and Nielsen, 2011), and overall strategic approach (Nielsen, 2010). Focusing on venture capital investors, certain educational and functional experiences of top management teams from venture capital firms, have been confirmed to influence such firms’ investment decisions, in terms of a focus on early stage ventures, their geographic scope, and overall investment strategy (Dimov et al., 2006; Patzelt et al., 2009). They are, however, tailored specifically to venture capital investments. While, through these studies, several relationships have been established between top management teams composition, and either a broader concept of strategic renewal actions, or limited to a single external corporate venturing option, none of these previous works specifically discriminate between different strategic sourcing options, which are identified here through four distinct governance modes. By adopting a multinomial logit approach, this study recognizes

(7)

multiple external corporate venturing options as a set of unique, mutually exclusive choices, rather than defining a broader, less specific form of strategic orientation, or direction.

Strategic decisions, however, are not made in a vacuum, based on either internal or external contingencies. Rather, strategizing is an iterative process, where firms navigate through highly complex environments, through which multiple sources of information must be considered and weighted, and where executives incrementally decide on their strategic routes (Quinn, 1989). By hypothesizing the moderating effects of market dynamism on the relationship between the top management team’s composition and external corporate venturing modes, this study combines internal perspectives with environmental contingencies. Adopting such a dual approach, exploring internal and external factors simultaneously in relation to strategic renewal actions, has been encouraged by Schmitt et al. (2016) in their recent review of existing strategic renewal literature. Building on upper echelon theory, top management team heterogeneity and chief executive tenure are applied as internal predictors of firm behavior, whereas notions from real options theory and transaction costs perspectives, drawing on previous research by several scholars (e.g., Leiblein, 2003; Santoro and McGill, 2005; Van De Vrande et al., 2006), are adopted to predict the (main and) moderating effects of market dynamism on firm’s external corporate venturing actions.

Following the introduction above, the research question for this study is as follows:

What is the effect of top management team composition on external corporate venturing choices, and how does market dynamism influence this relationship?

Next, the theoretical framework is included in section 2, which discusses how different theoretical constructs relate to this subject, and the main empirical advancements that have been made, for the purpose of building hypotheses.

(8)

2 Theoretical framework

2.1 External corporate venturing modes

While there are many paths firms choose to grow or renew, external access to knowledge and technology has long been recognized as a pivotal means to that end. Accessing external knowledge to develop the business, is what is referred to as external corporate venturing (Schildt et al., 2005; Keil et al., 2008). They are external in nature, provided that this form of strategic renewal action involves firms venturing outside existing boundaries (Flier et al., 2003). Although external corporate venturing can come in many forms, the most common and widespread capital-based modes for external knowledge or technology sourcing that have been discussed by academics, and adopted by firms in practice, include corporate venture capital, minority interests, joint ventures, and acquisitions (Schildt et al., 2005; Keil et al., 2008; Van De Vrande, 2013). This section further defines each of these discrete strategic choices, illuminating important differences between these modes of external corporate venturing. Section 2.3. provides further theoretical frameworks and literature review in relation to factors influencing governance mode decisions, as it relates to the predicted moderating effects of environmental dynamism.

Venture capital investments have been much associated with sourcing innovative knowledge, or technology that is still in its infancy. Investments in start-ups have been much used by large incumbent firms, and extensively studied, as a means to ensure timely attention to potential disruptive technological change (Hellman and Puri, 2002; Benson and Ziedonis, 2009; Maula et al., 2013). Firms tend to adopt venture capital investments as “a window on new technologies” (Benson and Ziedonis, 2009), and these investment vehicles have shown to be particularly useful in drawing top executives attention to disruptive technologies (Maula et al., 2013). Hence, venture capital investments have been largely associated with firm innovation (Keil et al., 2008), and explorative learning outcomes (Schildt et al., 2005), and such strategic objectives have been argued to dominate the choice for investing through venture capital (Basu et al., 2011). Venture capital investments generally take the form of capital contributions or equity stakes, through which the incumbent firm acquires a minority interest, although investments through debt financing are also not uncommon. Thus, external corporate venturing through venture capital is defined here, as an investment by an incumbent firm, in an entrepreneurial firm, for the purpose of exploring

(9)

technological advancements. Within the boundaries of external corporate venturing options that are adopted in this study, this form of external corporate venturing is argued to be the most at arm’s length option within the market-hierarchy continuum.

An equity stake can also be acquired through a minority investment, rather than through a venture capital investment. While both options in their legal form generally represent a minority interest, gaining a measure of control in another entity, often as well within an existing strategic alliance relationship to facilitate coordination (Zollo, Reuer, and Singh, 2002), or adopting minority stakes as a stepping stone towards full acquisition (Van De Vrande, 2009), are important factors for acquiring a minority holding. This is where venture capital investments are fundamentally different; rather than gaining a measure control, a venture capital investment generally has the purpose to gain access to unproven technological developments, or other potential strategic benefits to sustain the firm’s competitive advantage (Chesbrough, 2002) that may, or may not be, adopted or integrated into the focal firm in a later phase. As such, from an organizational design point of view, focal firms often structure their venture capital investments through a separate, autonomous investment unit (Schildt et al., 2005)(examples are Alphabet’s venture capital arm Google Ventures or GV, or Walmart’s Store No_{8). Hence, minority interests generally provide less flexibility, as they}

are more integrated into the focal firm’s core business and structure, and are therefore argued to rank higher on the governance mode continuum.

Ownership or control of one over the other, can generally not be imposed by either party in a joint venture relationship. Within a joint venture relationship, ownership is shared; joint ventures entail the incorporation of a legal entity, in which each firm contributes assets, generally through an equally divided capital contribution. By virtue of each firms’ contribution to the legal entity, the partners share the venture’s ownership (Sampson, 2004), and both parties are its beneficiaries in terms of economic gains or knowledge development that stems from the venture’s activities. Rothaermel and Deeds (2004) further discriminate a partnership in the form of a joint venture, from partnerships as non-equity, or contract alliances; non-equity strategic alliances facilitate exploration in the early stages of product development. Thus, such partnerships mostly focus on innovation. When new products mature, and are ready to be sold to the market, other contingencies enter the arena, and firms tend to shift from exploration to exploitation, and adapt their governance mode accordingly to equity alliances. Such dynamic trade-offs are also discussed by Teece et al. (2016),

(10)

in their discussion of how to manage uncertainty, versus risk, by shifting the firm’s focus from agility, to efficiency. In managing risk, hierarchical modes of organizing allow for improved coordination and integration of knowledge (Leiblein, 2003). Hence, joint ventures, being equity alliances, are argued to point towards an efficiency focus, over an innovation focus; they are erected to, in cooperation with a partner, exploit existing business opportunities.

Finally, acquisitions are at the end of the market-hierarchy continuum; integration into the firm’s boundaries is achieved by acquiring the majority (that is, fifty percent or more) of another firms shares and voting rights (Schildt et al. 2005), and thereby gaining control over its resources. As such, acquisitions tend to have an efficiency focus, as the purpose of an acquisition is to integrate any acquired knowledge or technology within the firm’s existing pool of resources. This reasoning is supported by the evidence Schildt et al. (2005) provide; hierarchical governance modes have shown to facilitate exploitative learning. Likewise, Keil et al. (2008) have shown that acquisitions contribute less to firm-level innovation performance than do less integrated forms of collaboration.

Each of the four distinct modes of external corporate venturing discussed above are examined in this study, by adopting these modes as the outcome of strategic choices made by top management teams, and, as discussed, they are argued to display a linear ranking along the market-hierarchy continuum, by virtue of their degree of integration, flexibility, levels of commitment, reversibility (Van De Vrande et al., 2006), and focus on either innovation or efficiency.

2.2 Top management team composition, as predictor for venture choice

This section proceeds as follows: first, upper echelons theory is discussed, including its key constructs, propositions, and recent advancements in the academic literature. Next, this section advances to relate this theoretical framework to external corporate venturing choices, and the preceding discussion of what previous studies have argued to be (dis)advantages, and unique properties of each external corporate venturing mode. Combining and integrating these discussions, this section ends with a set of hypotheses which are aimed at answering the research question proposed in the introduction.

(11)

Upper echelons and top management team composition

In their seminal article, Hambrick and Mason (1984) popularized the idea that firms are a reflection of the values and cognitive base of the firm’s executive team, and strategic action stems from the idiosyncrasies of top executives. Build on the assumption of bounded rationality, executives face complex environments with multiple stimuli, in which strategic choices need to made. Given that all human beings have cognitive limits and, ex ante, specific values and preferences, those stimuli are filtered down, and strategic choice is based on a selection, interpretation, and perception of all those stimuli. However, rather than rationally weighing all potential outcomes in order to arrive at the economically most efficient option, executives can also discard such iterative processes, and base strategic choices solely on their personal values and preferences. This way, the strategic routes firms take in order to survive, are argued to be largely dependent on managerial background and characteristics. Those strategic choices, in turn, yield specific performance outcomes.

Having argued these propositions, many scholars verified direct effects of top management team (TMT) characteristics on organizational outcomes. Examples of such work is abound; multiple studies have reported relationships between TMT characteristics, and firms’ tendency to globalize (e.g., Sambharaya, 1996; Tihanyi et al., 2000; Carpenter and Frederickson, 2001; Nielsen and Nielsen, 2011), or firm financial performance (Certo et al., 2006; Buyl et al., 2011). Others have considered contextual effects that influence TMT characteristics and their effectiveness, indicating that environmental, organizational, and leadership contingencies, moderate the relationship between TMT characteristics, and organizational outcomes (Carpenter et al., 2004).

A distinct stream of upper echelon research has verified that specific TMT attributes are preferable in the quest for innovation, versus striving for efficiency. Bantel and Jackson (1989) contributed significantly in their study of TMT attributes and innovation; amongst others, they observed that TMT heterogeneity, especially in terms of education and functional experience, positively contributed to firm innovation. More recently, Buyl et al. (2011) echo these findings; TMT (functional) diversity is especially critical in highly dynamic, innovation-oriented environments, while adding that the CEO-TMT relationship can reinforce the TMT diversity-performance relationship of firms in such environments.

(12)

Next to TMT diversity, executive tenure has also received much attention. Per Miller (1991), executives grow “stale in the saddle”; long tenure tends to lead to a reluctance to strategic change. Studies by Wiersema and Bantel (1992), Wu et al. (2005), and Simsek (2007), indicate similar results within CEO tenure and firm innovation relationships, while adding different moderating factors. Henderson et al. (2006) refined these conclusions; the tenure-innovation relationship is likely to simplistic. Henderson et al. (2006) propose an inverted U-shape relationship; new executives need to grow in their position, prior to initiating strategic change, whereas longer tenured executives lose touch with their changing environment. In relation to a firms’ internationalization moves, Herrmann and Datta (2002) found that CEO’s with high levels of tenure were more inclined to adopt hierarchical modes of market entry, versus market entry through contractual agreements, or joint ventures.

In sum, different levels of TMT diversity and chief executive tenure, appear to foster an orientation towards either innovation or efficiency, and subsequent firm-level financial performance, or innovation outcomes.

On the relationship between TMT and external corporate venturing

The pivotal role of the TMT in the more broader perspective of strategic change and renewal has been emphasized by several scholars. Tripsas (2009), for example, has argued that TMT action is needed to instigate identity and technological change, both of which are viewed as sources of organizational inertia, reducing strategic flexibility, and a firm’s ability to adapt. Additionally, several studies have discussed the importance of executives in relation to resource allocation decisions, and a firm’s technology sourcing initiatives (e.g., Rothaermel and Alexandre, 2009; Kwee et al., 2011; Maula et al., 2013). External corporate venturing, as argued previously, is used by many firms to source new technology or knowledge, as a means to drive corporate growth, or enter new markets (Dushnitsky and Birkinshaw, 2016). Given that corporate venturing modes exhibit differences in terms of their effectiveness and applicability (Nicholls-Nixon and Woo, 2003), it is only reasonable to argue that firms, and, more specifically, TMT’s, make a choice for a specific mode taking those unique properties into account. Until date, however, research has yet to empirically verify such assertions.

(13)

Two notable exceptions are mentioned here, as it relates to TMT composition and external corporate venturing. The first is a study by Kwee et al. (2011), whom performed a longitudinal study of Royal Dutch Shell. Here, it was found that the TMT’s corporate governance orientation, being either Rhino or Anglo-Saxon, influenced the firm’s strategic trajectory over time, depending on the TMT’s orientation to explore (Rhino), or to exploit (Anglo-Saxon). The generalizability of these findings, however, is unclear, given that Kwee et al. (2011) focus on a single firm, which, by definition, neglects industry effects on firms’ strategic renewal paths. Nonetheless, the concluding observations from Kwee et al. (2011), provide useful guidance in building hypotheses. Kwee et al. (2011) concluded that TMT’s with an explorative orientation, preferred arm’s length governance modes and internal growth options, while TMT’s with an exploitative orientation, preferred hierarchical and external modes of growth. Although pointing as such to different governance modes, the level of measurement was not on discrete governance mode choices; rather, it was measured as the ratio of explorative renewal actions, against the total of strategic renewal actions. Second, conclusions from two studies by Talke et al. (2010, 2011) indicate similar results, although in a different setting. Here, TMT diversity was adopted as a proxy to explain differences in firm’s strategic innovation-orientation. The empirical results indicated that TMT diversity had a strong positive influence on a firm’s orientation towards exploration, and innovation. Again, however, the studies by Talke et al. (2010, 2011) did not discriminate specifically between different modes of external corporate venturing, while each mode comes which its unique (dis)advantages.

Building on these previous works, as well as the studies providing insights into TMT attributes and innovation, it is postulated here, that TMT’s will have a preference for a specific external corporate venturing mode, provided that, based on previous TMT composition studies, varying levels of TMT heterogeneity yield an orientation towards innovation vis-à-vis efficiency. Likewise, studies on CEO tenure have advocated the tenure-innovation relationship. Thus, both variables are predicted to influence the strategic choices made by firms. Grounded in the discussed literature, innovation-orientation entails low tenure, and high heterogeneity. Contrary, efficiency innovation-orientation entails high tenure, and low heterogeneity. As it relates to external corporate venturing choices, the different modes have been argued to rank along the market-hierarchy continuum, and varying in their focus, being innovation-driven, or efficiency-driven. Hence, it is hypothesized that highly heterogeneous TMT’s will have an innovation-focus, and will, therefore, be more likely to opt for arm’s length modes of external corporate venturing. Likewise, building on findings from tenure-innovation

(14)

studies, low tenure fosters an orientation towards innovation, and, consequently, firms with lower tenured chief executives are expected to prefer modes of external corporate venturing that are at the arm’s length end of the market-hierarchy continuum, over hierarchical forms.

Aggregating the observations regarding TMT’s composition and heterogeneity, and CEO tenure, with the discussion about the unique properties and outcomes of external corporate venturing modes, the following hypotheses were developed:

Hypothesis 1: The degree of TMT heterogeneity increases the likelihood of arm’s length modes of external corporate venturing, over hierarchical modes of external corporate venturing.

Hypothesis 2: The degree of CEO tenure increases the likelihood of hierarchical modes of external corporate venturing, over arm’s length modes of external corporate venturing.

2.3 Market dynamism and governance choice

Market dynamism is conceptualized here, as the extent to which the firm is subject to unpredictability (Baron and Tang, 2011), pursuant to the pace of environmental changes. As discussed, Teece et al. 2016 have argued that under conditions of deep uncertainty, that is, events that cannot be predicted, remaining flexible and agile becomes crucial, whereas firms should shift towards efficiency to manage risk, versus uncertainty. Thus, firms adapt their strategic trajectory, according to multiple internal or external factors, such as market dynamism and the uncertainty it brings to the firm’s current operations, and future growth options. Therefore, while TMT’s may, ex ante, have an orientation towards innovation or efficiency, and are therefore hypothesized to display differences in terms of governance mode preferences, external contingencies should not be neglected. A study by Carpenter and Fredrickson (2001) applied this same logic when empirically testing the moderating effects of environmental uncertainty, on the relationship between TMT demographics, and firms’ posture to internationalize. Likewise, but in relation to CEO tenure, Wu et al. (2005) discuss the moderating effects of technological dynamism on the tenure-innovation relationship. Drawing on the agility-efficiency trade-off as discussed by Teece et al. (2016), market dynamism is expected to moderate the relationship between TMT heterogeneity, and CEO tenure, and their corresponding preference for specific external corporate venturing modes. Furthermore,

(15)

testing market dynamism and uncertainty for moderating effects, versus main effects, follows the suggestion from other scholars (Van De Vrande et al., 2009).

Based on these premises, this section will first discuss the predicted main effects of market dynamism on external corporate venturing modes, drawing on transaction costs perspectives and real options theory, and concludes by hypothesizing under what condition market dynamism may interact with firm-level TMT attributes, and thereby jointly influencing external corporate venturing governance choices.

Governance choice and uncertainty: transaction costs and real options perspectives

Transaction costs perspectives and its relation to governance mode choices, originates from Coase’s (1937) discussion about the existence of firms. Coase’s (1937) initial reasoning was that, in light of the efficient market theorem, firms should not exist: if perfect decentralization and the division of labor maximizes economic welfare, than the market should not give rise to firms. Yet, it may be obvious that firms do exist in the real world. Coase (1937) explained this phenomenon by deviating from the neo-classical assumption that transactions are always costless. By allowing authority to coordinate the division of labor and allocation of resources, costs associated with organizing transactions through the market can be avoided. Which mode of governance is less costly, determines whether transactions are coordinated through either the market, or through the firm.

Williamson (1981) later builds on the observations from Coase, however, adds further insights regarding the conditions that determine the most efficient governance mode. The transaction costs theory as further developed by Williamson (1981), rests on assumptions about human behavior of bounded rationality, and opportunism. Whereas the first holds that economic actors have rational intent, but also cognitive limits, exchange contracts are incomplete as one is not able to capture all possible contingencies in a contract. The latter, opportunism, assumes that (at least some) actors are self-interested, and will aim to exploit the contracting relationship for their own benefit. The choice for an efficient governance mode, then, depends on the potential to reduce the problems created by bounded rationality and opportunism. Conditions under which bounded rationality and opportunism become problematic, per Williamson (1981), are high levels of uncertainty, transaction-specific investments, and frequency of transactions. Exchange transactions may require

(16)

an investment, from one or both parties, that is specific to that transaction relationship. Due to this specificity, those investments have little value outside that relationship, making one prone to opportunistic behavior. Given that external corporate venturing concerns the deliberate investment by a firm in specific knowledge or technology, transaction-specific investments are a requisite, rather than an option. This, combined with increased external uncertainty in dynamic markets, should make coordinating exchanges using authority within a firm the preferred mode of organizing, above coordinating efforts through the price mechanism in the market. In other words, the market will fail in coordinating transactions efficiently, since economic value will be lost pursuant to opportunistic behavior by economic actors.

Studies incorporating transaction costs perspectives, however, and specifically its prediction for hierarchy under the condition of high uncertainty, have yielded no conclusive or consistent evidence (Sutcliffe and Zaheer, 1998; Leiblein, 2003). This is largely due to measuring uncertainty as either internal uncertainty or external uncertainty. While uncertainty within contracting relationships may lead to integration, following transaction costs reasoning, dynamic markets bring external uncertainty into transactions which largely lies beyond the firm’s span of control. This, therefore, suggests the existence of complementing mechanisms for governance choice beyond transactions costs reasoning, and a distinction should be made clearly between internal and external uncertainty. Under the condition of high dynamism, the propositions of transaction costs theory about firm boundaries, become less evident, and its predictions should not be applied uncritically (Williamson, 1991; Van De Vrande et al., 2006). Furthermore, the theory does not incorporate the potential for future growth when investing under (external) uncertainty (Brouthers et al., 2008), and it is somewhat limited in recognizing hybrid forms of governance. To complement the reasoning about the effects of market dynamism on governance choice, other theoretical perspectives should be considered. In uncertain, dynamic markets, real options theory provides a useful framework for deciding on corporate investments, and thereby complements transaction costs perspectives (Folta, 1998).

Real options theory was originally developed in finance literature by Myers (1977), as a means to evaluate and improve capital budgeting decisions, and the corresponding allocation of resources. From a financial perspective, such decisions are commonly based on net present value workings, by discounting the future cash flows of an investment opportunity. Myers (1977), however,

(17)

emphasized that investments may be phased; firms have the option to continue, or increase, their initial investment if the market retains its interest, or retract the allocated resources and deploy them to other opportunities. Traditional financial models did not capture the value of such real options to change. In relation to strategy and governance choice, real options analysis primarily posits that investments can be deferred (Leiblein, 2003); thus, when the future value of investments is uncertain, reversibility of investments may be preferred, allowing for flexibility. Therefore, committing significant funds under high levels of uncertainty, restricts a firms’ ability to adapt if the market is suddenly disrupted by new innovations. Minority interests, and corporate venture capital investments even more so, are characterized by high levels of reversibility, and relatively low levels of commitment (Van De Vrande et al., 2006). Both of these governance modes, therefore, provide a firm with a real option for future growth, while minimizing its exposure to an uncertain environment (Brouthers et al., 2008) where returns cannot be reliably estimated.

To conclude, the main effect of high dynamism is expected to be that firms shift towards less integrated modes of corporate venturing.

Hypothesis 3: The degree of market dynamism increases the likelihood of firms choosing arm’s length modes of external corporate venturing, over hierarchical modes of external corporate venturing.

Moderating effects of market dynamism on TMT composition and external sourcing choices

While the above main effect of high dynamism is hypothesized, each industry has its leading innovators, and its laggards, and therefore the preferred choice of governance mode likely varies not only by industry (Van De Vrande et al., 2009), but depending on firm-specific characteristics as well. This perspective is adopted here through the lens of upper echelon theory: TMT attributes have been discussed to yield an orientation towards innovation or efficiency, depending on their level of heterogeneity, and CEO tenure, subsequently affecting the governance mode choice for external corporate venturing. Thus, in other words, while industry innovators will aim to develop new business and technology through external corporate venturing, and therefore prefer adaptability and arm’s length governance modes, industry laggards will aim to exploit new technologies and knowledge that have grown beyond the start-up phase, through more hierarchical governance modes.

(18)

Following this line of thought, the moderating effects of market dynamism on TMT’s with an innovation focus, that is, low CEO tenure, and high TMT heterogeneity, is predicted to be that under the condition of high dynamism, the preference for arm’s length governance modes is amplified, thus valuing the adaptability that such modes offer (Van De Vrande et al., 2006).

TMT’s with an efficiency focus, that is, high CEO tenure, and low TMT heterogeneity, are argued to follow another strategic route; it is predicted that the moderating effects of market dynamism on TMT’s with an efficiency focus, is contrary to TMT’s with an innovation focus. Thus, if TMT’s are characterized by a focus on efficiency, market dynamism amplifies the preference for hierarchical modes of organizing, allowing for control and coordination of technologies to enable an efficient integration into the firm’s existing knowledge and technological base.

Integrating the predictions from the theoretical perspectives as such, the following hypotheses were developed for the moderating effect of market dynamism on the relationship between TMT heterogeneity and external corporate venturing choice, and CEO tenure and external corporate venturing choice:

Hypothesis 4: The degree of market dynamism moderates the relationship between TMT heterogeneity and the likelihood of arm’s length modes of external corporate venturing, such that these modes are more likely when market dynamism is high, versus when market dynamism is low.

Hypothesis 5: The degree of market dynamism moderates the relationship between CEO tenure and the likelihood of hierarchical modes of external corporate venturing, such that these modes are less likely when market dynamism is high, versus when market dynamism is low.

(19)

2.4 Conceptual model

As a visual aid, the hypothesized relationships from the previous sections are shown below.

Figure 1: conceptual model for the relationship between TMT composition and external corporate venturing choice, moderated by market dynamism

Two primary assumptions are implicit to the model presented above. First, as will be discussed in more detail in the methods section, volatility in a firm’s stock price is chosen as a proxy for market dynamics. Events that have the potential to disrupt future prospects, should be reflected by financial markets in a decreasing (or increasing) stock price. When markets are more frequently disrupted by unpredictable events, thus argued to be more dynamic, the result should be a volatile stock price. Hence, it is assumed that financial markets are, at least to a large extent, efficient; information relevant for future growth opportunities is assumed to be accurately reflected in a firm’s stock price. The second key assumption relates to rationality in decision-making. Such assumptions are implicit in real options theory; deciding for any investment option involves rationally weighing pros and cons. As it relates to upper echelons theory and transactions costs theory, however, it is more nuanced: while executives are assumed to have rational intent in decision-making and organizing transactions, they are limited by cognitive abilities. To synthesize this with efficient market hypothesis, where financial markets are fully rational, a weaker form of this concept is assumed here; less than rational behavior is for sure observed within financial markets, although stock markets are still argued to be highly efficient by scholars (Malkiel, 2003). A semi-strong form of efficient market hypothesis, which limits available information to public sources, and allows for less than fully rational behavior, has consensus of being the accepted paradigm (Jensen, 1978).

External perspective Market dynamism

External corporate venturing Acquisitions

Joint ventures Minority interests Venture capital investments Firm perspective

Degree of TMT heterogeneity (functional, educational, age,

gender) CEO tenure

(20)

3 Data and method

3.1 Type of research

This study is quantitative in nature; a quantitative study is considered to be most appropriate and feasible within the constraints and timeframe of the thesis program. Provided that this study primarily entails capturing data from several sources, the overall research design is archival: data on all variables have been obtained through several databases and publically available sources, as will be discussed below. To further support this approach, it is emphasized that a substantial part of the empirical studies on external corporate venturing activity and governance options, are quantitative in nature, and are designed as archival studies (e.g., Hellman and Puri, 2002; Rothaermel and Deeds, 2004; Schildt et al., 2005; Keil et al., 2008; Benson and Ziedonis, 2009; Van De Vrande et al., 2009; Maula et al., 2013), as are a substantial part of the research on top management team composition (Nielsen, 2010).

3.2 Population vs sample

To test the hypotheses, the sample selected for testing comprised the investment activity of the top-110 U.S. based publically traded firms in the year 2016, whereas the top-top-110 firms were based on total revenues, and it was taken from the U.S. Fortune 500 list for the year 2016. Other studies have considered similar size samples of firms; both Schildt et al. (2005) and Keil et al. (2008) used the top-110 largest public U.S. corporations based on revenues, although focused on four ICT industries. Maula et al. (2013) similarly focused on the largest public U.S. firms in four ICT industries. Van De Vrande (2013) used a sample of 78 independent companies, although all were in the pharmaceutical sector. Based on the preceding, the sample of 110 firms is deemed to be appropriate, and sufficient. To prevent governance choice bias in the sample, investment banks and similar funds that reached this top-110 (for example, Berkshire Hathaway), were excluded from the sample, provided that, due to their nature, such firms do not necessarily invest outside their existing boundaries to gain access to new technology or knowledge, which is essential within the framework of external corporate venturing (Schildt et al., 2005; Keil et al., 2008). As such, the selection of firms continued downward with the next firm on the U.S. Fortune 500 ranking, until the sample of 110 firms was reached. Appendix 1 contains a list of the firms included in the sample.

(21)

The year 2016 was chosen since the preceding years, and continuing thereafter, had seen a significant rise of technologies disrupting multiple markets; social media gaining significant influence, big data analytics developments, bio-engineering advancements, and robotics are some key examples. Furthermore, selecting the year 2016 will ensure that data on external corporate venturing activity has been collected and aggregated by the data providers, versus a more recent year. The year 2010 was also considered, however, many industries were highly affected by the global credit crunch in the period between 2008 and 2012, which may have forced firms into an efficiency, or exploitative mode. To prevent such bias, the year 2016 was chosen since the major stock markets had for certain recovered by then (see for example the MSCI World Index, which was back on pre-crisis levels during 2014). Any earlier years could also be an option but are deemed less interesting, as it would yield too much historically oriented data, while it is generally acknowledged that the pace of today’s markets continues to increase, pursuant to technological advancements.

Provided that this study intends to capture industry effects in the analysis through the construct of market dynamism, narrowing down to specific industries would not be suitable. Selecting a minimum of 110 U.S. based publically traded firms will ensure sufficient heterogeneity in industries, allowing for measurement of market dynamism effects. Kwee et al. (2011) showed that there may be differences in corporate governance orientations at TMT level, which can influence investment decisions. Selecting only U.S. based firms, will support preventing such national or cultural variation in the sample. The focus is on larger firms only, given that larger firms publish significantly more information as a result of regulations; a similar argument was provided by previous studies (Keil et al., 2008; Van De Vrande, 2013). Additionally, larger firms are more likely to engage in external corporate venturing, than are smaller firms (Dushnitsky and Lenox, 2006; Basu et al., 2011).

For the selected firms, a total of 555 external corporate venturing deals was sourced through two separate databases; the data collection process is discussed in more detail in section 3.4.

3.3 Measurement of variables

This section elaborates on how each variable was operationalized, to allow for hypothesis testing, and subsequent analysis.

(22)

Dependent variable: external corporate venturing mode

As discussed, this study ventures to measure the effects of the independent variables on discrete strategic choices, rather than adopting a proxy such as ratios to signify strategic orientations of firms. The dependent variable, therefore, essentially includes four options: corporate venture capital, minority holdings, joint ventures, and acquisitions. Each venturing mode is coded according to their position on the market-hierarchy continuum, provided that these modes vary in their level of integration (Van De Vrande, 2009), their level of reversibility, and required commitment (Van De Vrande, 2006), as well as in their learning outcomes, from exploitative to explorative (Schildt et al., 2005). Thus, following the discussion of each venturing mode from the theoretical framework in section 2.1., acquisitions are coded (1), being the most hierarchical form of governance, followed by joint ventures (2), minority holdings (3), and, finally, corporate venture capital (4), being the most at arm’s length option within the market-hierarchy continuum. The designation of the governance mode of each investment was assigned by the data providers. Deal synopsis and other relevant investment information (such as percentages of shares acquired) as provided by the data providers, did not give reason to deviate from the assigned modes. These were consistent with the definition of the external corporate venturing modes as discussed in section 2.1. Other studies concerning governance mode choice also did not report deviating from the pre-assigned governance modes, and the data sources adopted in this study were largely the same as those used by academics. Furthermore, as will be further elaborated on in section 3.4., by sourcing the external corporate venturing investment activity of the firms in the sample from two separate databases, reliability of the observations of external corporate venturing, and their governance mode designation, is further corroborated.

Independent variables: TMT heterogeneity and CEO tenure

First, top management teams are defined here as all executives above vice president level of the firms in the sample, thus excluding non-executives, provided that executive directors are charged with strategic decision making and managing daily operations. Non-executive directors are usually charged with governance, and monitoring the strategic actions of the executive directors. This level of measurement was furthermore used by many other studies in relation to TMT research (e.g. Hambrick et al. 1996; Gordon et al., 2000; Geletkanyez and Black, 2001; Bertrand and Schoar, 2003).

(23)

CEO tenure is measured by the number of days that each executive has been in the chief executive position (Henderson et al., 2006; Simsek, 2007), until the date of each individual external corporate venturing investment. As such, it provides the most accurate measurement possible of CEO tenure at the level of each individual external corporate venturing deal.

The TMT heterogeneity measure was constructed using the Blau index (1977), which measure is accepted to be the norm for studies involving TMT heterogeneity (Nielsen, 2010). Carpenter and Fredrickson (2001) described in detail the procedure for constructing the measure, while applying the coding method from Wiersema and Bantel (1992). A similar procedure was followed in this study.

Educational backgrounds were coded as arts, sciences, engineering, business and economics, or law. If specialties were absent, MSc and BSc degrees were coded as general MSc or BSc degree. TMT functional background diversity was coded following the same scholars’ example, coding functional background as being predominantly marketing, distribution and/or logistics, sales, R&D, production and/or operations, engineering and/or technical, finance, law, or general management. For TMT age, ten age categories were used ranging from <35 to >75, with intermediary steps of five years, resulting in ten age categories. The ten age categories were defined as such, after the age of each individual TMT-member (TMT n = 1,087; firms n = 110) was already sourced. The data of TMT individual age showed a mean value of 54.19, a standard deviation of 5.87, and ranging from 35 to 85. Based on this mean value and level of spread, the mentioned ten categories for age were considered to be appropriate. While it is recognized that it is usually not preferable to categorize age, for the purpose of measuring heterogeneity it is required since absolute values of age do not provide information about the heterogeneity of age within a TMT. Finally, gender was coded as either male or female, and grouped into either category.

Next, the degree of heterogeneity for each attribute was constructed using the Blau index (1977). The index is calculated as1 − ∑(P ) , where Pi is the percentage of individuals in the ith category (Carpenter and Fredrickson, 2001). To correct for the number of categories in each measure, the

index scores were multiplied by (

( )), whereas k was the number of categories within each

heterogeneity variable. As such, the index figures ranges from 0, being the lowest heterogeneity score, to 1, being the highest heterogeneity score. The final measure for TMT heterogeneity was

(24)

computed as the mean value of the four (category corrected) index scores, such that the heterogeneity measure included the full spectrum of diversity indicators, as previous studies had similarly considered multiple aspects of heterogeneity into a single measure (Nielsen, 2010).

Moderator: market dynamism

Market dynamism is conceptualized here as the extent to which the firm is subject to unpredictable events, which affect its current operations and future growth options. To compute a proxy for market dynamism, a method applied by Folta (1998) was adopted, which largely draws on finance literature. The measure comprises the stock price volatility on a weekly basis, for the 26 weeks preceding the date of the investment. Following the reasoning provided by Folta (1998), the 26-week period was used, as any longer periods cannot be asserted to have influenced strategic actions. As such, market dynamism proxies are computed at the level of each individual external corporate venturing deal.

For each individual investment, the variance of the historical weekly return (R) was computed through ( ) =

( ) ∑( − μ ) , and finally taking the square root of the variance to

arrive at the standard deviation of historical weekly returns, in other words, the stocks’ volatility for the 26-week period preceding the deal date. As such, higher values of the measure indicate higher levels of dynamism, and vice versa.

Control variables

Control variables were defined from a firm perspective, and from a deal perspective. In relation to the first, firm size and financial slack were controlled for. In relation to the latter, geographic scope and business relatedness of the investment were controlled for.

Firm size was defined as each firm’s net revenues. Financial slack intended to capture a firm’s excess cash position by measuring the quick ratio, which is defined as current assets minus inventory, against current liabilities. Both control variables were measured for the year(-end) 2015, hence, the year(-end) preceding the year of measurement of the dependent variable.

Geographic scope was chosen based on previous research by Carpenter and Fredrickson (2001), whom confirmed the relationship between TMT attributes and the firm’s posture to

(25)

internationalize. It was defined here as a dichotomous variable, where each individual investment was coded as being either domestic (U.S.), or foreign. The control for business relatedness was chosen from resource-based perspectives influencing governance choice, since per such perspectives, firms will be more inclined to choose hierarchical forms of governance when the target business lies close(r) to its core business. This control was similarly defined as a dichotomous variable, where each individual investment was classified as being either related to the firm’s existing business, or not related. The classification was based on the focal and acquired firms’ U.S. SIC codes, at a two-digit level. Evaluating business relatedness using SIC codes has been common practice through multiple studies (Pehrsson, 2006). Where SIC codes were not provided by either of the databases (since not all investments were made within the U.S.), the deal synopsis provided by the data providers was scanned to classify the investment as being related, or not related to the firm’s existing business.

3.4 Data collection and reliability

The data on external corporate venturing activity has been sourced through two separate databases: ThomsonONE (which has been used in multiple other studies, i.e. Schildt et al., 2005; Keil et al., 2008; Benson and Ziedonis, 2009; Maula et al., 2013; Van De Vrande, 2013), and Zephyr, the latter being a database that is maintained by Bureau van Dijk, a Moody’s analytical company1_.

Combining the external corporate venturing actions from both databases has been a manual process, since the databases often used slightly different naming conventions for firms.

The decision to source the data from two different databases was driven by both reliability and completeness considerations. Overlap in the deals sourced from both databases improves the reliability of the observations in the sample, including their governance mode designation. In terms of completeness, ThomsonONE has a separate module that tracks venture capital activity (previously known as VenturXpert); sourcing venture capital deals through this separate module using ThomsonONE, therefore, ensured that sufficient observations at the very end of the governance mode spectrum (e.g. investments at the arm’s length-end of the continuum) were included in the sample. Given the nature of this separate ThomsonONE module, overlap was

(26)

primarily observed within the acquisition governance mode, and to a lesser extent within the venture capital governance mode. On an overall basis, 67.6% of the observations included in the sample were present in both databases. When excluding venture capital deals, the overlap between both databases was at 82.5%. The table below shows a summary of the external corporate venturing activity included in the sample, and its distribution across the databases.

Table 1: Distribution of external corporate venturing activity in 2016 by top-110 U.S. firms across the databases

Database Acquisitions Joint

Ventures Minority Interests Venture Capital Total ThomsonONE 30 - 7 114 151 Zephyr 16 3 2 8 29

Sourced from both 245 4 25 101 375

Total 291 7 34 223 555

Data on TMT characteristics has been primarily obtained through each firm’s 10-K filings (which are required to be filed annually at the U.S. Securities Exchange Committee). The 10-K filings generally provided a sufficiently detailed account of each firm’s top executives, including their functional and educational backgrounds, age, and gender, allowing for categorization of the data, and subsequently computing heterogeneity measures. It is furthermore noted that several other studies in the field of upper echelons research also used 10-K filings to source these data (Gordon et al., 2000; Herrmann and Datta, 2002; Certo et al., 2006). Where insufficient information was provided in the 10-K filings, the data was complemented using other publically available sources, being each firm’s corporate website, LinkedIn, and Bloomberg. Provided that the corporate venturing deals were sourced for the year 2016, the TMT composition as reported in the 2015 annual 10-K filings was used as a basis, correcting for any changes in TMT composition up to the date of each individual external corporate venturing deal. As such, this ensures that any strategic actions at the date of measurement of the dependent variable, that is, external corporate venturing choice, is actually the result of strategies developed by those executives. When correcting for any changes in TMT composition, the required data was again sourced through publically available sources such as each firm’s corporate website, LinkedIn, and Bloomberg. While collecting the data through this procedure, it was observed that in a substantial part of the cases, that is, each individual investment, the firms’ TMT composition had little or no changes between the deal dates, or TMT members were replaced by executives with similar backgrounds. This, therefore, implies that the TMT heterogeneity measure may assume an equal value across different values of the dependent

(27)

variable, inflating the data and results, since firms did adopt different governance modes through their investments. This limitation, and directions for future research to prevent this issue, is further discussed in section 5.

Weekly closing stock prices, the base data required to compute the proxy for market dynamism, being the stock price volatility for the previously mentioned 26-week period preceding the investment date, was sourced through Yahoo! Finance. This source is publically available, and it efficiently provides access to historical (closing) stock prices for all firms in the sample on a daily, weekly, and monthly basis.

Data for the control variables – firm size in terms of revenues, firms’ financial slack, geographic scope, and industry relatedness – were sourced through either ThomsonONE or Zephyr. To validate the financial measures reported by the data providers, these were verified against the firms’ 10-K filings for a random number of firms. As no differences other than rounding were observed, the financials reported by the data providers were concluded to be appropriate and reliable for further analysis.

3.5 Validity

Validity was considered from three perspectives; internal, external, and construct validity.

Improving internal validity in order to isolate the predicted effects, was achieved by defining the control variables discussed earlier. These control variables were, similar to the predictor and outcome variables, defined from the firm perspective (firm size, firm slack), as well as at the deal level (geographic scope, industry relatedness) to allow controlling for influences on governance choices that were confirmed through previous research. Selecting only U.S. firms allowed for (indirectly) controlling for national and cultural differences.

While the latter may improve internal validity, it reduces external validity of the analysis; any findings will only be generalizable onto the U.S. market, and only onto large, publically traded firms. In favor of the chosen sample, as it relates to external validity, is the fact that multiple industries and markets are covered. Generalizability of many studies is reduced, as samples are drawn from a single, or a limited number of industries (e.g. Sampson, 2004; Keil et al., 2008; Van De Vrande et al., 2009; Van De Vrande, 2013).

(28)

Construct validity of the measures chosen to test the hypotheses is considered to be appropriate for several reasons. First, the normative application of the Blau index (1977) in studies of TMT heterogeneity (Nielsen, 2010) corroborates the validity of this measure, and the construction of this measure was consistent with previous studies, as discussed earlier. Second, while many studies have adopted ratio measures as a proxy of firms’ strategic orientation, to allow for ordinary least squares regression methods through continuous scale variables, this study captured discrete strategic choices, through multiple modes of external corporate venturing as the outcome variable. Hence, this directly measures strategic action, rather than attempting to approximate it. Third, market dynamism is acknowledged to be an ambiguous concept, for which no clear definition or a normative measurement construct exists. Here, stock price volatility was adopted as a measure of dynamism, following the example from Folta (1998), which essentially builds on the efficient market hypothesis from finance literature. It assumes that a firm’s stock is fairly priced, given that all information of risks and opportunities influencing a firm’s future cash flows, is readily available to all investors (Berk and DeMarzo, 2017). As there appeared to be no consensus in the literature for the “best” measure, a somewhat arbitrary choice had to be made. This proxy of market dynamism was chosen pursuant to its relative ease of data capturing, and was considered to be most feasible within the time constraints of a thesis, as the measure had to be sourced for all (555) individual external corporate venturing deals.

3.6 Statistical analysis

Analysis method and model choice

Provided that the dependent variable comprised a set of discrete choices, logistic regression is the appropriate analysis method for this study. Per the theoretical discussion, the four venturing modes that are within the scope of this thesis can be ranked along the market-hierarchy continuum, based on their level of integration, reversibility, and resource commitments. This, then, suggests an ordinal ranking within these outcomes. Per Long and Freese (2003), the appropriate model for this is the ordinal logistic regression model, also referred to as the proportional odds model. However, Long and Freese (2003) also argue that the assumptions under the ordinal regression model are often violated, and that the avoidance of violation of assumptions outweighs the potential efficiency and data richness loss by switching to nominal outcome models. More specifically, violation of the

(29)

parallel regression assumption often occurs: that is, the slope coefficients are required to be identical for each predicted outcome (Long and Freese, 2003), thus only shifting from one outcome to another, without changing the probability curve. Considering the data sourced for this study, it is highly unlikely that this assumption will be met. Therefore, the multinomial logit model was preferred over the ordinal logit model, which is more flexible in terms of each venturing mode’s position along the theorized continuum. To corroborate the choice for the multinomial logit model, the existing literature on governance choices was reviewed to identify which regression model is most commonly adopted by scholars.

Folta (1998) adopted the multinomial logit model in analyzing the choice between minority investments, joint ventures, or acquisitions, although a ranking was also not theoretically specified. A similar approach was taken by Hagedoorn and Duysters (2002) when discriminating between strategic alliances, and M&A’s. In a study by Van De Vrande et al. (2009) the ordinal structure in the dependent variable is explicitly acknowledged, and argumentation in this study partly draws on this work. Van De Vrande et al. (2009), however, also applied a multinomial logistic regression model, even though it is argued by the authors that the dependent variable, the choice of governance mode, is ordinal. The choice for this model was, indeed, driven by the fact that the assumptions for the ordinal regression model were violated. The only study that has been found in relation to external corporate venturing governance mode decisions that did incorporate an ordinal logit model, was Santoro and McGill (2005), which defined different governance modes along the market-hierarchy continuum, although the study did not discuss whether the parallel regression assumption was met.

Based on the above, it appears that, in practice, most scholars have adopted the multinomial logistic regression model. Therefore, in order to perform the logistic regression analysis, this model was also adopted in this study.

An important property of the multinomial logit model, is the underlying assumption of independence of irrelevant alternatives. In plain English, what this essentially boils down to, is that the relative odds of choosing one alternative over the other, should not be affected by the inclusion or exclusion of a third possible outcome (Kwak and Clayton-Matthews, 2002). Thus, for example, within the boundaries of this study, the preference, or relative odds, for venture capital investments over acquisitions, should not be affected by whether minority interests or joint ventures are

(30)

included or excluded in the model. Within the field of econometrics, multiple specification tests have been developed to confirm that this assumption holds (Cheng and Long, 2007), with the most well-known test being the Hausman and McFadden (1984) test. Until this date, however, this feature is not available in SPSS (IBM, 2016). Cheng and Long (2007), in addition, observe that, in practice, many studies do not report whether the independence of irrelevant alternatives assumption was met. Consequently, no proper exemplary work could be traced that specifies in sufficient detail how such tests should be executed without the aid of statistics programs. Hence, within the boundaries and scope of this thesis, statistically testing whether the independence of irrelevant alternatives assumption is met, was unfortunately not feasible.

Nonetheless, to give some further substance as to whether it can be asserted that this assumption holds, the multinomial model was reduced to binary choices, which procedure was performed to further infer whether a ranking existed within the governance choice options (see section 4.3 for details). As it turned out, the binary choice models resulted in similar relative odds, and significance levels were also similar to those of the model that included the full spectrum of alternatives, that is, the four governance modes adopted in this study. Further argumentation is provided through earlier conclusions from Amemiya (1981: 1517), and Cheng and Long (2007: 598): alternative outcomes should be as much as possible dissimilar, and distinct outcomes should be included in any model, that are no substitutes for one another. Arguments for these statements have been extensively discussed through literature review in section 2.1. Based on these premises, it is therefore asserted that, within the boundaries and scope of this thesis, the independence of irrelevant alternatives assumption that underlies the multinomial logit model, holds.

Pre-analysis steps and hypotheses testing procedures

Prior to performing the statistical analysis, preparatory steps were taken to ensure suitability of the collected data. Variables were standardized since they were partly measured on different scales. The theorized interacting variables were mean-centered following guidance provided by Aiken and West (1991), prior to probing any moderating effects.

Next, some procedural considerations are set forth here, to ensure that it is clear how the analyses align with the hypotheses, and on what basis each hypothesis will be accepted, or rejected. First, it is important to recognize that the multinomial logit regression test will yield, among other statistics,

(31)

odds ratios, being the ratio of the choice probability for two distinct outcomes (Kwak and Clayton-Matthews, 2002). In other words, these represent the odds of choosing one alternative relative to a pre-set default category (Field, 2013). In this case, one of the external corporate venturing governance modes must be set as the reference group, and all other modes of venturing are compared to this reference group. There are no methodological requirements in relation to which outcome is set as the reference group, or whether the analysis is run multiple times utilizing different reference groups, but theoretical or hypothetical considerations could lend themselves to choose for a specific alternative and methodological approach (Field, 2013).

As discussed in the theoretical framework, it is hypothesized that higher values of TMT heterogeneity and market dynamism should increase the likelihood of arm’s length modes of governance, over hierarchical governance modes, and interacting effects should amplify this preference. Acquisitions have been included as the most hierarchical form of governance, and all other venturing modes continue to shift more towards the arm’s length-end of the continuum. Therefore, acquisitions is set as the reference group in the analysis. Given the properties of the odds ratios and hypothesized effects, values of > 1 are expected for any of the (hybrid) venturing modes, when compared to the default category acquisitions. The degree of CEO tenure is hypothesized to have an effect opposite to that of TMT heterogeneity and market dynamism. When retaining acquisitions as the reference group, in order not to overcomplicate the analysis, odds ratios of < 1 are expected for any combination. Since four modes of venturing were adopted in this study, each hypothesis is essentially tested three times. The testing results for all three combinations of governance modes, in terms of their sign and significance levels, will be evaluated in order to reach an overall conclusion in relation to the hypotheses.

(32)

4 Results

4.1 Univariate analysis

To test the hypotheses, data for the variables have been sourced for the top-110 U.S. based publically traded firms in the year 2016. A total of 555 external corporate venturing deals were sourced for the firms in the sample (n = 110), distributed across the identified deal types as follows:

Figure 2: Deal activity of firms in the sample and distribution across deal types (firms n = 110, deals n = 555)

The mean value for the total deals per firm in the sample was at 5.045, with the standard deviation at 8.416 – this rather large spread is mainly due to a number of firms being highly active within the venture capital market. It is observed that the number of joint ventures and minority interests in the sample is relatively small. Other studies (Keil et al., 2008; Van De Vrande et al., 2009), however, similarly reported a relatively low fraction of joint ventures and minority interest compared to the total deal activity that was included in the sample. As such, it appears that these forms of collaboration are simply being less adopted by firms. Hence, external validity is argued not to be reduced by this observation. While the distribution of deal types appears to be consistent with literature, it is unlikely that significant outcomes will be observed for these deal types, especially that of joint ventures. Further analysis is provided in the following sections.

Descriptive statistics for the control and predictor variables are included in table 2. It is noted that these are the mean and standard deviation based on absolute values; these were purposefully provided here such that it is more clear what these measures represent. The variables were standardized within the dataset prior to the analyses since they were partly measured on different scales. Legend n % Acquisitions 291 52.4 Joint Ventures 7 1.3 Minority Interests 34 6.1 Venture Capital 223 40.2 Total 555 100.0

Upper echelons, dynamism, and external corporate venturing : predictions of governance mode choice through internal and external perspectives using multinomial logistic regression analysis

Master’s Thesis