Effects of the Size, Gender and Age Diversity of Top Management Teams on Company Performance and the Moderation Effect of Industry Turbulence

Effects of the Size, Gender and Age Diversity of Top

Management Teams on Company Performance and the

Moderation Effect of Industry Turbulence

Study of Listed Companies within the Central and Eastern

European Transition Countries

MSc International Business & Management 22nd January 2018

Master Thesis

Olga Balaziova Supervisor: dr. Olof Lindahl

Abstract

This master thesis focuses on the link between top management team (TMT) characteristics, namely TMT size, gender diversity and age diversity and company financial performance. Although a significant amount of academic attention has already been directed towards this field, the majority of such studies were conducted in stable developed countries with almost no attention paid to different settings. This paper fills in the research gap under the Upper Echelons theory by focusing on institutionally unique, complex and unstable transition countries of Central and Eastern Europe (CEE), as organizational and environmental fit might be achieved through different TMT characteristics than suggested by previous predominantly western research. Moderation is introduced to investigate effects of industry turbulence on the TMT-performance relationship within the distinct CEE environment, a previously unexplored phenomenon. With the help of multivariable OLS regression, 121 listed company TMTs were analysed leading to surprising findings. The notion of uniqueness of the CEE environment and inapplicability of traditional assumptions and findings from stable western countries have been corroborated. Within CEE countries, smaller TMTs were found to lead to a higher financial performance contradictory to the extant literature suggestions. More gender diverse TMTs drove higher firm performance, with age diversity showing no significant impact. Partial positive moderation of industry turbulence was found for TMT size, in line with the overviewed literature. However, strong negative moderation of industry turbulence for gender diversity and performance relationship presents another unexpected finding. Our results have significant theoretical and practical implications which are discussed at the end of this thesis.

Keywords: top management team (TMT), Central and Eastern European transition countries,

Table of Content

Introduction ... 6

1 Literature review ... 9

1.1 Central and Eastern European countries ... 9

1.1.1 Institutional environment of the CEE region ... 10

1.1.2 Turbulence of the CEE environment ... 11

1.2 Top management teams ... 12

1.3 Upper Echelons Theory ... 13

1.4 Resource-based View ... 13

1.5 Diversity and effects on company performance ... 14

1.6 Hypotheses ... 15

1.6.1 Top Management Team Size ... 16

1.6.2 Gender diversity ... 17

1.6.3 Age diversity ... 19

1.6.4 Environmental determinants and TMT diversity, the moderation effect... 21

1.7 Conceptual model ... 24

2 Methodology ... 25

2.1 Data Sample and Data Collection ... 25

2.2 Variables ... 27 2.3 Data analysis... 30 2.4 Robustness test ... 33 3 Results ... 33 3.1 Descriptive statistics ... 33 3.2 Correlation ... 36 3.3 Regression results ... 38 3.4 Robustness test ... 41 4 Discussion ... 43 5 Concluding remarks ... 47 5.1 Theoretical implications ... 47 5.2 Managerial implications... 48

5.2 Limitations and Future research ... 48

References ... 50

List of Figures

Figure 1: TMT decision-making framework (Cannella et al., 2008) ... 12

Figure 2: Conceptual model... 24

List of Tables

Table 2: Correlation Matrix ... 37

Table 3: Regression results ... 39

Table 4: Robustness test regression results ... 42

List of Abbreviations

FDI Foreign Direct Investments GDP Gross Domestic Product OLS Ordinary Least Squares ROA Return on Assets RBV Resource-Based View TMT Top Management Team UE Upper Echelons

6

Introduction

Does size matter? Is gender relevant? Does youth have anything to offer? These age-old questions that can be applied to a multitude of concepts are especially important for executive teams. Scholars have realized this long ago, focusing their attention on the top management team (TMT) characteristics and their impact on company outcomes. This thesis will continue their legacy and focus on the TMT size, gender diversity and age diversity in relation to firm financial performance. The relationship between TMT size and company performance has been on the periphery of this research area, while the field of executive diversity has been richly explored. However, the main focus of a majority of the TMT diversity literature has been on developed western countries, namely the United States (Acar, 2016), with little to no attention paid to a different institutional setting, specifically transitioning post-communist countries of Central and Eastern Europe (CEE) that this thesis aims to explore.

CEE countries provide a prospective alternative to the increasingly saturated western markets. However, these new growth opportunities for companies worldwide come with their distinct challenges resulting from the nature of the CEE transition environment. Despite the transitioning process from centrally-planned to market-based economies, relaxation of strict policies and reformation of their institutions (Jindra, 2005; Bennett & Brewster, 2009), persisting underdeveloped market mechanisms and institutional voids pose difficulties for business conduct and continue to distinguish CEE countries from developed ones (Bevan et al., 2001). CEE environment also differs from other regions and emerging markets, thanks to the unique transition process, unparalleled by any other world region by its speed, abruptness and institutional starting position. As a result, some of the standard assumptions of commonly accepted theories applicable to the developed economies do not hold for CEE countries (Meyer & Peng, 2005). Moreover, as firm performance depends on the organizational fit with its environment (Wieserma & Bantel, 1993), in institutionally distinct CEE context different TMT characteristics might be required. Altogether, this provides an interesting new ground for testing western TMT diversity literature findings as they cannot be assumed to apply in the CEE unique conditions without verification.

7

environment fit and influencing company performance (Wieserma & Bantel, 1993). Larger and more heterogeneous TMTs provide a higher number of perspectives, ideas and increased decision-making quality (Nielson, 2010) improving team’s information-processing capabilities (Hoffman et at., 1997) which is especially important in unstable environments (Hambrick & Mason, 1984).

This thesis is conducted under the umbrella of Hambrick and Mason’s (1984) Upper Echelons Theory emphasizing the importance of executives’ characteristics influencing company outcomes through their manifestation in strategic decisions. We continue the legacy of studies focusing on TMT structure operationalized as TMT size and composition operationalized as TMT gender and age diversity, chosen thanks to their potential contributions and fit with the CEE environmental characteristics.

Increased TMT size provides the team with more capabilities and perspective thus enhancing problem-solving and decision-making quality, leading to improved company performance (Haleblian & Finkelstein, 1993). The importance of larger TMTs is said to be more pronounced in dynamic environments involving complex tasks (Hambrick & Mason, 1984), thus could play a major role especially in CEE context. The theoretical notion of a positive relationship between TMT size and company performance has been supported by studies such as Cooper and Bruno (1977), Eisenhardt and Schoonhoven (1990), Haleblian and Finkelstein (1993), etc.

When it comes to female senior managers, worldwide underrepresentation is significantly pronounced (Heilman, 2012). Given the specificities of the CEE transition environment, such as institutional voids or network-based relationships (Meyer & Peng, 2005), increased gender executive diversity could ensure a better fit of organizations with environmental requirements. As women tend to possess different qualities than their male counterparts, such as better communication and interpersonal skills (Hillman, 2008), their presence in TMTs could be essential for overcoming these CEE context shortcomings. The notion of female executive presence and higher company performance has been supported by previous academic research, such as Adler (2001), Krishnan and Park (2005), Campbell and Mínguez-Vera (2008), Dezsö and Ross, (2012), etc.

8

environment unmarked by the socialist thinking, while older generation could help to steer this new powerful force, leading to an increased performance. Positive effects of age diversity on company outcomes were found by Kilduff et al. (2000), Richard and Shelor (2002), Ilmakunnas and Ilmakunnas (2011), etc.

However, TMTs do not operate in vacuum. They gather and translate information from the external environment to identify opportunities and threats for their respective companies (Finkelstein & Hambrick, 1990; Wiersema & Bantel, 1992). Empirical research found support for a vast influencing role of these environmental characteristics for the relationship between TMT features and company outcomes. However, as the majority of these studies has been conducted in the developed stable environments, instability and complexity were operationalized only in terms of industry turbulence (i.e. Murray, 1989; Geletkanycz & Hambrick, 1997; Keck, 1997). No research exists on how turbulence of industries influences the link between TMT characteristic and company performance under conditions of overall high dynamism, such as CEE transitioning countries provide. Would industry turbulence still matter in the face of a greatly complex and unstable CEE transition environment?

With this thesis the aim is to explore the unique CEE transition environment and fill in the research void that exists outside of the developed world by finding answers to our main research questions:

Do TMT size, TMT gender diversity and TMT age diversity matter for financial performance of companies operating within CEE transitioning countries? If so, are these relationships

moderated by the industry turbulence within the unique CEE environment?

9

1 Literature review

In the following section, research area will be specified and theoretical concepts used will be described. Furthermore, relevant hypotheses stemming from the literature will be formulated and lastly, the conceptual model will be introduced.

1.1 Central and Eastern European countries

As the developed markets are becoming increasingly more saturated, companies are forced to look for growth opportunities elsewhere. The transitioning economies of CEE countries, namely Bulgaria, Croatia, Czech Republic, Georgia, Hungary, Lithuania, Poland, Romania, Russian Federation and Ukraine present such opportunities. This can be illustrated by their annual percentage GDP growth rate in the era after their liberalization. Compared to the G7 countries in the period from 1995 to 2016, CEE’s growth was on average 3,03% while G7 grew at the pace of 1,65% (World Bank, 2017a). That multinational enterprises realized the potential of these countries and opportunities they provide is clearly visible from the FDI net inflows as a percentage of GDP during the same period. CEE countries received on average 4,9% in comparison to 2,1% for G7 countries (World Bank, 2017b).

Although the CEE countries vary considerably in languages spoken, size or prevailing ethnicities, their pronounced similarities make it possible for researches to use them as a unit for analysis (i.e. Holland & Pain, 1998; Barkema & Drogendijk, 2007). As argued by Shenkar (2001), these commonalities, described in the following section, can be contributed to one of the most important mechanisms of closing cultural distance, geographic proximity. Further convergence of CEE institutional settings can be attributed to the shared socialist regime in their recent history, which influenced their subsequent development in a similar manner (North, 1990).

10

conflicts have also left their irreversible mark (Ekiert & Hanson, 2003). Thus, current CEE institutional setting is a unique mix of historical heritage of these countries with their new liberalizing efforts resulting in significantly different environment compared to western societies. Meyer and Peng (2005) stated that this distinction between developed countries and CEE transitioning economies is so pronounced that the standard assumptions of some theories commonly accepted in the stable western setting do not apply in the CEE context. This environment is unique even when compared to the emerging markets, which in contrast to CEE countries only have to overcome struggles related to the transition from agricultural to industrialized societies.

1.1.1 Institutional environment of the CEE region

Institutional structure of the environment is of profound significance for the operations and performance of economic agents, as they facilitate business transactions and market economy functions (North, 1990; DiMaggio & Powell, 1991). Interactions of organizational strategies at the company level with the economic institutions at the national level are still for the most part not understood due to the lack of research. This is of particular importance for the CEE transition countries as the market mechanisms underlined by the institutional arrangements are often underdeveloped and weak and scarcely studied (Bevan et al., 2001). Primary focus during the transitioning process has been on four institutional areas, namely “privatization, liberalization and establishment of market institutions, legal development” and reformation of financial sector (Bevan et al., 2001, p. 20). However, as Meyer (2001) points out, this transition towards market-based economy resulted in an inconsistent institutional framework characterized by great instability as it is not fully reformed. Companies operating within these highly imperfect markets have to considerably adapt their strategies to avoid and compensate for the institutional voids to lower the costs associated with organizing business in such unstable environment (Meyer, 2001).

11

be explained by the communist legacy, where deep-rooted attitudes are changing only slowly (Coyle & Zhiglinskaya, 2006). The socialist view of women as primary caretakers of households and families in these countries persists (Pollert, 2003) and is also manifested in the position of women in the workforce and on the senior management level. While expatriates view female managers as more effective, this contrasts with the view of locals. Furthermore, female senior managers still earn less than their male counterparts (Bennett & Brewster, 2009; Sányová et al., 2015).

1.1.2 Turbulence of the CEE environment

The CEE environment can be further characterized by high volatility, lack of market-supporting institutions and constant or rapid structural institutional changes accompanied by high levels of uncertainty resulting from the transition process (Meyer & Peng, 2005). Critical lack of available information, diffusion of knowledge due to the weak intellectual property protection, judicial institutional voids and inexperienced and ineffective bureaucracies further illustrate the complicated nature of the institutional arrangement posing additional challenges for businesses (Meyer, 2001). Some authors even called the changing institutional framework of these countries transitioning from centrally-planned to market economy as “unusually unstable” (Meyer, 2001, p. 358). As the rapid institutional changes result in uncertainty and inconsistent requirements between individual institutions, organizations have to adapt appropriately. They need to seek mechanisms quite different from the standard western practices, such as the use of network-based coordination, personal ties or barter to overcome the environmental shortages (Meyer, 2001).

12 1.2 Top management teams

Following Wiersema & Bantel (1992), we define a top management team as a group of individuals that makes strategic decisions about the firm’s future based on the interactions between members. TMTs are in literature considered as the “information-processing centers” of their organizations as they pose a link between company itself and its external environment (Haleblian & Finkelstein, 1993, p. 845). One of the key roles of executives comprising such teams is strategic decision-making, based on translation of information gathered both from external and internal environment and identifying all possible opportunities and threats in combination with their personal characteristics and behavioral features (Finkelstein & Hambrick, 1990; Wiersema & Bantel, 1992). As effectiveness and competitiveness of firms are also determined by the internal processes of strategic decision-making (Hitt et al., 2017), a link between top management teams and company performance can be identified.

Thus, it can be said that one of the most critical roles of executives is the boundary spanning acquisition of environmental information, their translation and resulting insight into the external context, on which their strategic decision-making process can be based (see Figure 1). In the stable, fully institutionally developed settings of western countries, the task is already hard enough. However, in the turbulent, institutionally underdeveloped CEE environment, this function becomes increasingly more difficult. Absence of adequate market-supporting institutions resulting in information problems between buyers and sellers and judicial institutional voids (Bevan et al., 2001; Khanna & Palepu; 1997) call for better information-processing capabilities with multiple perspectives considerations, all of which are typically attributed to more diverse and larger TMTs.

The dynamics in TMTs’ interactions reflect the most important team characteristics: composition, structure and processes (Cannella et al., 2008). This thesis focuses on heterogeneity as part of the team composition and size of the team as part of the structure.

13 1.3 Upper Echelons Theory

The main theoretical concept of this thesis is the Upper Echelons (UE) theory conceived by Hambrick and Mason in 1984, which is contradictory to population economists and neoclassical economics. These scholars and theories suggest that contextual environmental characteristics are the main forces dictating companies’ operations, with managers’ roles being only symbolic (Hodgson, 1992; Augier & Teece, 2009). The UE theory is built upon an argument that strategic decisions, firm’s performance and other company outcomes are not just conditioned by the external environment but are a result of executives’ behaviors and choices which are in turn shaped by their personal characteristics, values and beliefs (Hambrick & Mason, 1984). Other significant influencing factors entering this process are previous experiences and personalities of executives (Hambrick, 2007). Based on the UE concept, we argue that difference in TMTs’ composition could explain variances in firms’ performance outcomes when comparing companies operating within the same institutional environment. Numerous studies are supporting this view (i.e. Lieberson & O'Connor, 1972; Mackey, 2008), with Quigley and Hambrick (2015) even suggesting that the top managers’ influence on company performance is only growing.

1.4 Resource-based View

Another cornerstone theory this thesis is built upon is the Resource-based view (RBV) of the organization, a theoretical perspective widely used in the diversity research area. RBV argues that competitive advantage of a firm depends on the internally held tangible and intangible resources which are responsible for the differentiated performance of companies (Rumelt, 1984, 1991; Wernerfelt, 1984; Barney, 1991). Barney (1991) and Rumelt (1984) suggested that these resources must possess certain characteristics to provide bases for a competitive advantage, namely being rare, valuable, inimitable and non-substitutable. An important category of competitive advantage generating resources is Human capital resources (Barney, 1991; Becker, 1964). Recently, sources of sustainable competitive advantage for increasing number of firms have shifted towards human resources, away from traditionally recognized sources such as economies of scale, technological advancements or protected markets (MacDuffie, 1995). These resources include characteristics of individual managers such as their experience, relationships, judgment or previous training (Barney, 1991).

14

resources are hard to imitate (MacDuffie, 1995). As managerial talent is already scarce in CEE countries due to the extended period of state-owned enterprises and lack of capitalist business environment, managerial human resources are difficult to gather and even more difficult to imitate. TMT diversity, bringing new skills and points of view enhancing team’s information-processing capabilities (Li et al., 2011) could provide companies with a competitive advantage and improve their performance, as TMTs are a part of firm’s human capital, the critical growth resource for organizations (Becker, 1994).

1.5 Diversity and effects on company performance

Demographic diversity can be explained as a level of heterogeneity of a unit regarding demographic attributes. Relevant demographic attributes for this thesis can be classified as permanent characteristics of individual executives that cannot be altered by his or her actions, such as gender or age (Pelled et al., 1999). Within the organizational demography literature, it is commonly assumed that the visible demographic differences signal underlying differences in cognitive processes, therefore serving as proxies for these invisible variations. Thus, demographic diversity is taken as a predictor of cognitive diversity, which in turn affects both team and organizational performance via strategic decision-making of executives (Kilduff et al., 2000).

Effects of TMT homogeneity such as strong internalization of norms, terminology, values and symbols, stability and culture sharing can have a positive impact on company performance by fostering communication, faster problem-solving and consensus-reaching (Glick et al., 1993). Results of TMT heterogeneity such as variety of interpretations, opinions, perspectives and novel ideas lead to higher decision-making quality (Nielsen, 2010), support learning and creative tendencies leading to innovations and therefore enhancing company performance (Glick et al., 1993; Stahl et al., 2010). Furthermore, heterogeneous teams were found to possess increased adaptability to the change of current situation (Katz, 1982) and generate larger productive conflicts, as resolving these can lead to enhanced and innovative solutions regarding environmental adaptability problems (Murray, 1989).

15

1979), Tsui et al. (1992) even finding that with increasing diversity, psychological attachment between team members decreases. Furthermore, heterogeneity naturally generates contrasting opinions leading to prolonged and less effective decision-making (Campbell & Mínguez-Vera, 2008) and slower response to competitor’s actions (Hambrick et al., 1996).

With higher homogeneity, team member comparison might occur, as such comparison to others similar to oneself is more telling and meaningful. However, this could create competition between similar individuals resulting in rivalry, jealousy and hostility (Pelled et al., 1999), deterring effectiveness of group processes and subsequently leading to a decrease in company performance. Glick et al. (1993) further associate homogeneity with shrinking pool of potential perspectives, leading to less innovative solutions. Moreover, the problem of groupthink can arise as a result of a fear of exclusion from a group of highly similar members. Unchallenged agreement and conformity can lead to defective decisions based on descent of judgment, efficacy of psychological processes or reality questioning (Glick et al., 1993).

Furthermore, another line of research exists finding no effects of diversity. According to Kilduff et al. (2000), there is no significant difference in decision-making between diversified and homogeneous teams. Fitza (2016) argues that the weight of CEOs’ characteristic is being overstated and does not have effect on company performance. Instead, variance in performance can be explained by random events, luck or chance. However, we build our hypotheses under the EU theory, hence supporting the view that executives’ characteristics and behaviors influence company performance.

1.6 Hypotheses

16

growth of the CEE region. Furthermore, as markets of the developed world are becoming saturated, CEE countries present new market and growth opportunities for MNEs looking for further expansion. Our findings could provide practical guidance for these MNEs and local firms alike on how to compose their TMTs to contend in their increasingly competitive backyard.

However, we believe that just examining the basic relationship between executive diversity and company performance is not enough when it comes to the CEE countries. One of the main characteristics of the CEE transitioning environment is high instability and complexity, as explained earlier. Therefore, we follow another branch of the TMT diversity literature focusing on the environmental characteristics, namely turbulence, as a moderating factor of the main studied relationship.

Following this summary, the studied concepts include TMT size, gender diversity and age diversity of TMTs driving company performance measured by return on assets, with industry turbulence possibly influencing these relationships. The studied types of demographic diversity were selected based on their relations with certain CEE contextual characteristics, as discussed below.

1.6.1 Top Management Team Size

The causal relationship between executive teams and company performance has been richly explored from many perspectives. However, focus on the TMT size as a driver of performance has been quite limited. Most of the extent literature focuses on TMT size as a control variable (i.e. Marimuthu & Kolandaisamy, 2009). Leading studies from the perspective of TMT size having direct link to the company performance are Hambrick and D’Aveni (1992) and Haleblian and Finkelstein (1993), both found a positive relationship between these concepts. Hambrick and D'Aveni’s (1992) study of bankrupting firms showed that among the four most significant TMT related factors causing these failures was the smaller size of their executive teams compared to the striving companies. Haleblian and Finkelstein (1993) found that companies with larger executive teams achieved higher performance, with the relationship being moderated by turbulent external environment.

17

However, contrasting studies, such as Amason and Sapienza (1997) found that TMT size is positively related to presence of both cognitive and affective conflicts, resulting from enhanced dissimilarities of more diverse opinions, feelings and values, which could potentially lead to negative impact on company performance. Finding no relationship with firm outcomes are Hambrick et al. (1996), Iaquinto and Fredrickson (1997) and Ensley et al. (2002). To the best of our knowledge and supported by Certo et al. (2006), no empirical study found negative relationship between TMT size and company financial performance.

In line with the RBV, it was suggested that larger groups possess more capabilities and resources and are therefore better positioned to solve group tasks, leading to higher quality decisions. This is a result of their enhanced problem-solving propensities, thanks to the increase in number of perspectives, available solutions and judgements resulting in lower error occurrence and boosted information-absorbing abilities (Haleblian & Finkelstein, 1993). The diversity that increased team size naturally brings is especially important for turbulent and dynamic environments with complex tasks (Hambrick & Mason, 1984). As mentioned before, CEE transitioning context represents such volatile environment, therefore requires match of environmental requirements with the information-processing capabilities of their TMTs (Haleblian & Finkelstein, 1993). In line with the theoretical thinking presented, we propose the first hypothesis:

Hypothesis 1: There is a positive relationship between top management team size and company financial performance in CEE countries.

1.6.2 Gender diversity

Notable differences exist between behavior of men and women. According to Hillman (2015), women executives tend to be more ethical, take longer to consider all alternatives before making a decision, take into account implications for more stakeholder groups and are better at predicting negative consequences, while men are faster decision-makers with focus on economic implications with not much consideration of other criteria. These different perspectives enrich the decision-making process and can lead to higher company performance, as mentioned earlier.

18

Fortune 500 companies remaining the same. In Eastern Europe, around 35% of senior management positions are held by women. Although compared to the global average this number is quite high, path to equal power distribution on the executive level is still long (Foy, 2015).

There are several reasons for studying gender diversity specifically in the CEE context. Firstly, institutional environment of these countries is characterized by institutional voids (Meyer & Peng, 2005). As primary function of institutions as “rules of the game” is to facilitate transactions and thus enable business conduct (North, 1990, p. 98), their absence and underdevelopment pose a significant challenge for companies. A way for businesses to bridge these voids is by interpersonal social interactions (Peng et al., 2008). As Hillman (2008) and others claim, female communication and interpersonal skills surpass those of their male counterparts, therefore women executives could be of great advantage particularly in this context.

Meyer and Peng (2005) further argue that in CEE countries, wide-spread use of networks and informal lines of authority are especially important. Agent and firm relationships are typically network-based, and so are their growth strategies. These strategies encompass intangible assets, namely intraorganization interpersonal ties and interorganizational relationships with foreign or domestic partners, creating basis for organizational growth (Meyer & Peng, 2005). Relationship building based on people and communication skills is, as suggested by Olsson and Walker (2003), forte of women, which could foster both informal and formal ties within and in between firms alike. Therefore, in CEE conditions, women executives could prove to be a valuable asset.

One specific CEE institutional void is underdeveloped corporate governance. In the presence of weak formal structures, informal mechanisms of control are needed. Adams and Ferreira’s (2009) found that in these specific conditions, female presence in boards increased monitoring activities and led to higher performance. We propose that increased monitoring would be similarly beneficial for TMTs in CEE countries leading to potential higher performance via improved informal mechanisms of corporate governance.

19

Another feature of CEE context is the lack of critical business resources which are naturally available in developed countries. Therefore, other resources need to be sought out to gain competitive advantage (Meyer & Peng, 2005). According to the RBV, TMTs could provide such an advantage being an important part of firm’s human capital, presenting vital organizational growth resource (Becker, 1994). If executive gender heterogeneity would provide this advantage, it could be inherently more valuable for companies operating in less resource-endowed environments, such as CEE.

Women representation in executive suites can also bring stability to the companies, an essential feature in turbulent environments. According to Hillman (2008), if the workforce is adequately demographically represented by females in executive positions, employees are less likely to leave, while other potential candidates are more attracted to seek job opportunities in such companies.

Supporting the UE perspective, many studies found positive implications of the presence of female executives on company performance. Some of them showing a direct positive link with the company financial performance (Adler, 2001; Krishnan & Park, 2005; Campbell & Mínguez-Vera, 2008) while others focused on non-financial outcomes, such as enhanced entrepreneurial capabilities (Lyngsie & Foss, 2016). There are others, however, showing negative relationship (e.g. Du Rietz & Henrekson, 2000; Laible, 2013) or no relationship (Marimuthu & Kolandaisamy, 2009). In this thesis, however, we follow the positive UE theory concept and propose the following hypothesis:

Hypothesis 2: There is a positive relationship between increased gender diversity in TMTs and company financial performance in CEE countries.

1.6.3 Age diversity

20

People of similar age tend to have similar characteristics, values and perspectives, as they emerged from a similar environment and were brought up under similar circumstances (Murray, 1989). Older working individuals are uniquely valuable due to their industry and organizational knowledge obtained through extensive working experience, contribution of socially complex networks (Li et al., 2011) and their generally more caring and responsible nature compared to younger individuals (Van Yoder, 2002). Furthermore, individuals over 50 were found to be more committed to their tasks, reliable, diligent and loyal, making them a valuable resource difficult to imitate by other firms (Li et al., 2011). Younger executives are associated with organizational growth, novel and unprecedented solutions, higher risk-taking tendencies (Hambrick & Mason, 1984), creativity, flexibility, higher physical and mental stamina, innovativeness and better education (Li et al., 2011). Thus, from the perspective of information and decision-making theory, combination of executives from different age categories would bring their unique characteristics together. This would result in more heterogeneous TMTs producing larger variety of ideas and solutions, leading to better decision-making outcomes and positively affecting firm performance (Richard & Shelor, 2002).

Impact of age diversity on company performance is especially interesting to study within the CEE country context. Individuals tend to derive values from society in which they were brought up (Murray, 1989). Given the enormous transformation that CEE societies underwent in the last 30 to 40 years, these generational differences are even more pronounced. Older generations grew up in an environment celebrating socialist values, while younger generations have already been exposed to liberalism and capitalistic values. As strategic-decision making process is based on information-processing of individuals, which is filtered through the lens of cognition and underlying values, it can be argued that values embedded in these individuals in the process of growing up influence executive strategic choices (Shaw, 1990). Furthermore, younger executives had the chance to obtain higher quality business education often studying abroad, exposing themselves to different ways of thinking and gaining valuable and scarce experience compared to the older generation where such possibilities did not exist.

21

The only companies found to improve their performance were those privatised by sale to foreign investors (Uhlenbruck and De Castro, 2000). Therefore, younger executives could bring new ideas and perspectives the same way as foreign investors did after the liberalization.

However, empirical research provides conflicting evidence regarding effects of age diversity. Processes of categorization based on age dissimilarities can result in hampered communication and cooperation between TMT members leading to more conflicts and thus influencing performance negatively (Richard & Shelor, 2002). Zenger and Lawrence (1989) found that increased age diversity was negatively related to the frequency of technical communication between team members working on group projects. Increase in affective conflict between individuals with diverse demographic attributes (age, gender) was found by Jehn, Chadwick & Thatcher (1997). Positive relationship of TMT diversity with turnover was found by O’Reilly et al. (1989) and Wiersema and Bird (1993). No effects of age heterogeneity on group outcomes have been found by several studies, mostly focusing on innovation (i.e. O'Reilly et al., 1998; Bantel & Jackson, 1989). Negative effects of this type of diversity on company outcomes were found by Zajac et al. (1991), specifically on innovation.

Positive effects of age diversity can also be found within the diversity literature. Pelled (1993) found lower occurrence of affective conflicts in age-diverse teams, proving a negative relationship with age heterogeneity contrary to her expectations. Ilmakunnas and Ilmakunnas (2011) argued that age diversity can lead to complementarity of information, experiences and knowledge within a task performance, enhancing the outcomes. Richard and Shelor (2002) found a significant positive relationship between TMT age diversity and sales growth, with diversity having a curvilinear effect. Kilduff et al. (2000) found only one of their demographic variables, age diversity, positively affecting firm performance. We build on this line of literature and propose the following hypothesis:

Hypothesis 3: There is a positive relationship between increased age diversity in TMTs and company financial performance in CEE countries.

1.6.4 Environmental determinants and TMT diversity, the moderation effect

22

through the process of strategic decision-making, steering companies towards their preferred visions (Finkelstein & Hambrick, 1990; Wiersema & Bantel, 1992). This process of information interpretation is highly complex and uncertain, with quality and timespan of translation determined by the cognition of executives, making the fit of TMT with environment vital (Wiersema & Bantel, 1993). Thus it can be said that TMTs respond to the environmental demands not only through their group processes but also via interaction dynamics reflected in their composition and structure.

Fundamentally important environmental dimensions affecting organizational operations are complexity and instability. Environmental complexity can be defined by the number of factors that influence organizations and have to be considered in the decision-making process within the specific environment. These environments are characterized by numerous constituencies posing conflicting demands upon organizations (Thompson, 1967). Coping with these conditions requires companies to build larger structural variation with different competencies and skills (Cannella et al., 2008). With the higher complexity of the environment, necessity of diversified executive team increases in order to successfully monitor the environmental heterogeneity (Gupta, 1988). Larger and more diverse TMTs tend to generate more questionings and discussions due to the broader skill range and the variety of perspectives, interpretations and opinions that can help companies cope with solving novel and uncertain problems resulting from the environmental complexities (Cannella et al., 2008).

Environmental instability has been described as the rate at which the environmental factors impinging on company operations change (Thompson, 1967). Instability implies unpredictability resulting from unanticipated environmental changes (Mintzberg, 1979). These conditions influence organizational structures and their operations as well as impact the composition, structure and decision-making processes of executive teams (Cannella et al., 2008). Unstable environments increase differentiation and contribute to fragmentation of executive work (Mintzberg, 1979). Furthermore, they increase task uncertainty which calls for greater information-processing capabilities due to the extensive amount of information that needs to be handled (Galbraith, 1973). As larger and more heterogeneous TMTs provide superior information-processing capabilities due to the wider scale of absorbed information, variety of perspectives in processes of problem-solving and larger pool of alternatives, they are better suited for such unstable environments (Cannella et al., 2008).

23

Finkelstein (1993) expanded upon, companies in dynamic environments with unpredictable outcomes could benefit more from larger and more diversified TMT composition in terms of higher firm performance as opposed to companies in stable environments.

Empirical research focusing on the relationship between TMT composition and environmental characteristics is quite scarce. However, the limited number of studies supports the theoretical arguments presented. For example, Sanders and Carpenter (1998) found a positive relationship between environmental complexity resulting from a degree of internalization and a larger scale TMTs. Majority of studies focusing on this relationship operationalized the environmental variable as the industry in which companies in question operate within. For example, Geletkanycz and Hambrick (1997) compared computer industry (turbulent) and branded foods industry (stable), finding support for larger and more heterogeneous TMTs in the turbulent industry. Wiersema and Bantel (1993) found that TMT turnover is determined by the instability and complexity of the external environment, which were also operationalized according to the industry characteristics. Murray (1989) selected for his empirical study two environments, an oil and a food industry, finding that temporal heterogeneity had a more significant effect in the more complex oil industry. Keck (1997) compared TMTs of companies operating within cement (stable) and minicomputer (turbulent) industries finding that heterogeneous teams led to higher company performance in turbulent industry context. Richard and Shelor (2002) found that age diversity within executive teams contributed to a higher financial performance measured by sales growth with the most important role of moderating this relationship being context. They concluded that companies’ competitive advantage can be within complex environments derived from TMT age diversity (Richard & Shelor, 2002).

24

Hypothesis 41_{: In CEE countries, the higher or lower industry turbulence does not moderate}

the positive relationship of TMT size with company financial performance.

Hypothesis 5: In CEE countries, the higher or lower industry turbulence does not moderate the positive relationship of TMT gender diversity with company financial performance.

Hypothesis 6: In CEE countries, the higher or lower industry turbulence does not moderate the positive relationship of TMT age diversity with company financial performance.

1.7 Conceptual model

The conceptual model summarizing the hypotheses can be visualized as follows:

Figure 2: Conceptual model

1 _{Hypotheses 4, 5 and 6 are formulated in a Null Hypothesis manner contrasting with the common Alternative}

25

2 Methodology

This section will firstly introduce the selection process of data sampling and methods of collection. Secondly, variables and their measurements will be described. Lastly, the estimation model used will be presented.

2.1 Data Sample and Data Collection

This research has been conducted on a sample of listed companies operating within 31 sectors in 10 CEE transitioning countries (Bulgaria, Croatia, Czech Republic, Georgia, Hungary, Lithuania, Poland, Romania, Russian Federation and Ukraine). All used data can be categorized as secondary data, therefore collected by a different party and recorded and stored in the form of accessible databases (Blumberg et al., 2014). The opinions on a sufficient sample size differ considerably within the research literature. Bailey (1994) suggests the minimum sample span of thirty to a hundred units for analysis to be reliable, while some others argue for larger samples, preferably around 200 units (Thomas, 2004). Insufficient sample size is also associated with major problems of regression results. A sample size corresponding to 10 individual units per variable ought to mitigate such problems and produce reliable results (Peduzzi et al., 1996). As our study focuses on three independent, one dependent, one moderating and five control variables, the desired sample size should consist of at least 100 companies.

After the exclusion of firm for which the data representing dependent, independent or control variables could not be obtained, final company sample size consists of 121 firms. As this number exceeds the above mentioned 100 units limit, our sample size fulfills the Peduzzi et al. (1996) criteria. The full list of companies can be seen in Appendix 1. The scarcity and obscurity of data for the CEE countries could be felt heavily, with a significant amount of manual data collection required. The relevant financial information was obtained primarily from two databases: the Thomson Reuter Datastream database containing financial time series data (Thomson Reuter, 2017) and from the Orbis database, containing various financial indicators, including Return on Assets, about more than 275 million firms from around the world (Bureau van Dijk, 2017). Missing data was gathered from annual reports and company websites.

26

of missing data, Bloomberg database, LexisNexis database, company sites, annual reports and various other Internet sources were checked to fill in the gaps. The industry turbulence data were obtained from the New York University’s Stern School of Business’ professor Aswath Damodaran, who provides data for industry averages calculated separately for different world regions in areas of valuation metrics and corporate finance (Damodaran, 2017). Control variables, depending on their nature, were gathered from Datastream, Orbis, Boardex, World Bank databases and company web sites respectively.

2.1.1 Time Period

The time span chosen for our analysis is a period of five years, beginning 2011 to 2016. This time length was deemed appropriate as it is the most recent period providing insight into the current situation in the CEE transitioning countries. The bottom time limit of 2011 was chosen due to the prior data unavailability requirements regarding our moderating variable of industry turbulence. A period of five years is also consistent with the extant literature, as executive diversity research spans from cross-sectional studies capturing only a single year to longitudinal studies spanning several decades.

27 2.2 Variables

2.2.1 Independent variables

The independent variables consist of visible TMT features, namely one structural and two compositional variables. Within the composition sphere, this thesis focuses on the demographic diversity characterized by observability and relative accessibility of data.

Top Management Team Size: Although most extant literature used this structural variable as a control, consistent with previous research we measured it as the total number of executive directors within the given top management team (Wiersema & Bantel, 1992).

Gender Diversity: As a measure for gender diversity has been selected the percentage of female executives compared to their male counterparts, calculated as the number of women top managers divided by the number of all members of the TMT.

Age Diversity: The third independent variable was measured as the coefficient of variation calculated as standard deviation divided by the mean (Bantel & Jackson, 1989; Wiersema & Bantel, 1992; Richard & Shelor, 2002).

As it is commonly believed that the translation process of strategic decision-making of TMTs into observable organizational outcomes is delayed by a certain amount of time (Geletkanycz & Hambrick; 1997), several studies employed a time lag of their variables to ensure that these effects are captured accurately (i.e. Keck, 1997; Geletkanycz & Hambrick; 1997). As there is no consensus on the ideal lag period, for the purposes of this analysis we employ a one year lag for all 3 of our independent variables, mathematically transcribed as t – 1.

2.2.2 Dependent variable

Company performance was operationalized for the purposes of this thesis as the financial performance. As there are no absolute ratios universally believed to determine the financial performance of companies, we selected a measure that has bases in the diversity literature explored in the previous section.

28 2.2.3 Moderating variable

Industry turbulence: As discussed in the previous section, majority of the diversity studies operationalized environmental complexity or instability as the turbulence of the industry within which their sample companies conducted business in. As Geletkanycz & Hambrick (1997) stated, the main characteristics of industries according to which their turbulence can be categorized are dynamism and uncertainty. Widely accepted factors contributing to these two dimensions are levels of innovation, instability of demand and supply, technological discontinuity, competition levels and rate of market growth. Therefore as an example, ICT industry can be categorized as turbulent, with rapid technological changes, unstable demand, highly volatile growth and frequent market share position changes, while in sharp contrast stand food industries, which experienced slower changes in these areas with consistent growth, justifying their labelling as stable (Geletkanycz & Hambrick, 1997).

The common way in which extent research operationalized this variable is as follows. Researches chose two industries at the opposite ends of the turbulence spectrum. Next step included statistical analysis, either by using a dummy variable, assigning one to the turbulent industry and zero to the stable industry (i.e. Geletkanycz & Hambrick, 1997) or two or more separate analyses were conducted for each industry respectively (i. e. Keck, 1997). Higher scarcity of the CEE transitioning countries data compared to the western developed countries did not allow for research centred around such small number of sectors. Therefore, our sample consists of companies operating within 31 industries, excluding both methods as inappropriate. Other methods used by studies focusing on multiple industries are, however, also not applicable to our research. Some used quantitative measures to categorize industries by conducting surveys (i.e. Auh & Menguc, 2005), others quantified different environmental characteristics (such as Wiersema and Bantel’s (1992) industry structure based on industry growth, profitability and concentration) or used multiple industries within the same country (such as Wiersema and Bantel’s (1993) study of US manufacturing companies). Obscurity of such data for CEE region that would be consistent across countries and quantitative nature of our research prevented us from using these measures.

29

industry turbulence was in this thesis measured by the industry betas – an investment analysis instrument summarizing uncertainty (Geletkanycz & Hambrick, 1997), calculated for each world region and year by professor Aswath Damodaran (Damodaran, 2017).

2.2.4 Control Variables

Five control variables were selected on the bases of existing literature that could influence the studied relationship.

Company Age: On the firm level, company age is measured as the total number of years from the date of incorporation and is expected to have a positive effect on the company performance. This could be due to more extensive experience and resulting preparedness for variety of situations which younger organizations lack (Keck, 1997; Marimuthu & Kolandaisamy, 2009; Majumdar, 1997; Low et al., 2015).

Company Size: Company size was measured as the amount of firm’s total assets (Marimuthu & Kolandaisamy, 2009). It was suggested that difference in firm resources has an impact on the financial performance, with larger the company pool of resources positively influencing company financial outcomes (Krishnan & Park, 2005; Campbell & Mínguez-Vera, 2008; Vedd & Yassinski, 2015).

TMT Tenure: On the TMT level, executives’ tenure was measured as the average number of years TMT members spent working within the team, in accordance with previous studies (Wiersema et al., 1993; Amason & Sapienza, 1997). Wiersema et al. (1993) showed that increased TMT tenure heterogeneity might lead to lower interaction, cohesiveness and higher power imbalance between members, leading to worse group outcomes, which in turn can negatively affect performance. Murray (1989) also found a significant relationship between this kind of diversity and company performance, especially in turbulent environments, supporting the notion of controlling for tenure effects on performance.

30

GDP growth: As our research focuses on 10 different CEE transitioning countries, it is important to control for effects on the national level as well. The control variable is measured as a percentage GDP growth of any given country and is expected to have a positive effect on company performance.

Consistent with the independent variables, control variables on the TMT level, namely Level of Education and Team tenure, are also lagged by 1 year period. The full list of variables with their distinct measures can be seen in Appendix 2.

2.3 Data analysis

The aim of this paper is to explore the causal relationship between variables, therefore multiple linear regression analysis is applied. More specifically the Ordinary Least Square

(OLS) method adjusted for panel dataset, consistent with previous studies (i.e. Marimuthu &

Kolandaisamy, 2009). Regression model used can therefore be transcribed as indicated by the following regression line:

Y = b0 + b1X1 + b2X2 + b3X3 + b4X4 + b5X5 + b6X6 +B7X7 + B8X8 + e Y = dependent variable

X = independent variable b0 = constant

b = slope coefficient related to independent variables e = error term

We designed and operationalized this regression model to reflect specificities of our data sample and the simple regression equation is as follows:

ROA = Constant + b1*lnTMTsize + b2*PercentageFemale + b3*Age + b4*lnFirmSize + b5*lnFirmAge + b6*lnTMTtenure + b7*lnEducationLevel + b8*GDPgrowth + e

After the inclusion of the interaction terms, the final regression line can be transcribed as:

ROA = Constant + b1*lnTMTsize + b2*PercentageFemale + b3*Age + b4*lnbeta +

b5*(lnTMTsize*lnbeta) + b6*(PercentageFemale*lnbeta) + b7*(Age*lnbeta) +

31

To analyze the data and allow for clear interpretation of results we use Stata statistical software. As the textbook of Hill et al. (2008) presents, there are three basic OLS estimation methods, namely pooled OLS, OLS with random effects and OLS with fixed effects, incorporating distinct assumptions that are appropriate for datasets with different properties.

Pooled OLS method assumes constant beta coefficients across individual units as well as constant error terms. It does not allow for individual differences within the dataset as it pools the observations together, therefore individual observations’ association with a specific time or group is disregarded. As time and cross-sectional variations exist within our panel dataset, pooled OLS can be rejected as an appropriate method.

However, before the appropriate estimation method of random or fixed effects is selected, the precondition of normal distribution of data needs to be checked. Based on the results of the skewness and kurtosis tests, TMT size, TMT tenure, Level of Education, Firm size, Firm age and Industry beta variables were transformed by natural logarithm to meet OLS requirements. ROA and GDP growth variable entered the analysis in their untransformed state as they can reach both positive and negative values, in accordance with the TMT diversity literature (i.e. Marimuthu & Kolandaisamy, 2009).

Both random and fixed effects estimation methods allow for individual heterogeneity, the difference represented in the treatment of such variations. Random effects treats the variance between individuals as unsystematic thanks to the recognition of sampling of the individuals as random, while fixed effects method considers the differences systemic. From the theoretical perspective, random effects seems to fit the nature of our dataset more appropriately. Fixed effects method is not designed to include time-invariant causes and will exclude variables that remain constant across time for a specific individual. As TMTs in our dataset scarcely change every year and executives remain in their positions over extended periods, random effects should produce a better fit. Furthermore, even if structural or compositional changes occur, our independent variables might not capture them and remain constant. For example, TMT size will not change if one executive is replaced by another. In a scenario that male executive is replaced by another male executive, gender diversity will not increase. As mentioned in the literature review section, age similarities between top managers also suggest the possibility of replacement of an executive with a manager of comparable age.

32

coefficients not being systemic cannot be rejected, therefore supporting our theoretical reasoning of preferability of the random effects model.

To further verify fitness of the random effects model, the Breusch and Pagan Lagrangian multiplier test for random effects is conducted. The null hypothesis in this case states that no random effects are existing in our dataset. According to the results (see Appendix 4) we reject the null hypothesis that pooled OLS is appropriate and we accept the alternative hypothesis that random effects are indeed present in our dataset.

Before the actual data analysis can be performed, several more tests need to be implemented. Firstly, to test for autocorrelation, Wooldridge test is employed. As the p-value is above the significance level of 0.05 (see Appendix 5), we cannot reject the null hypothesis stating that no first-order autocorrelation is present in our dataset.

Secondly, to detect the presence of heteroscedasticity, Panel Groupwise Heteroscedasticity Tests for random effects are employed with the null hypothesis stating there is homoscedasticity present in our dataset. The assumption of homoscedasticity implies that for each level of our independent variable we can be equally certain about the interval of values in which our dependent variable might fall with regards to their mean value, with uncertainty being independent from other factors. All the performed tests reject the null hypothesis (see Appendix 6) concluding that heteroscedasticity is present, therefore proceeding with the use of standard robust error terms in our regressions is necessary to account for these findings.

Thirdly, to mitigate problems associated with reverse causality and resulting endogeneity, following the approach of Carpenter et al. (2010) lagged independent variables are used. In this thesis the time lag employed is one year, supporting the notion of delayed effects of TMT actions on observable reality outcomes. Basic check of running the regressions in the opposite direction, independent variables taking the place of dependent one and vice versa, proves with insignificant results one-way relationship and therefore no reverse causality present.

33

correlation. Extremely low levels of VIF with a mean of 1.11 for our variables (see Appendix 7) suggest there is no collinearity present, confirming the nature of our dataset (Hill et al., 2008).

2.4 Robustness test

To check whether the regression results hold and are consistent across various measures, robustness test is conducted. As different financial indicators reflect different aspects of company performance (such as return on assets versus return on equity), possibly skewing the results of robustness analyses, selecting alternative measures of independent variables was deemed more appropriate, as they correspond to the same underlying phenomenon (approach consistent with i.e. Pelled et al., (1999)). The compositional independent variables measured as Female percentage and Coefficient of variance of Age within the TMT are recalculated and replaced with Blau indices, a commonly used diversity measure within the diversity research (i.e. Wiersema & Bantel 1993; Richard et al., 2004). The Blau Index diversity equation (for coding see Appendix 8) can be transcribed as follows:

𝐵𝑙𝑎𝑢 𝐼𝑛𝑑𝑒𝑥 = 1 − . 𝑃₀1 2

034

where K represents number of categories starting from K = 1, … and P stands for the proportion of individual units within a category (Blau, 1977; Richard et al., 2004). Similarly to our original measures, increase in the Blau index indicates increase in the TMT diversity.

3 Results

The following section starts with the descriptive statistics of our data sample. Table 1 presents the overview of 496 observations spread over 121 listed companies from 31 different industries operating within 10 CEE countries in the period between 2011 till 2016.

The positive and negative values of our dependent variable ROA show that companies in our sample encountered both profitable and loss-making years. Majority of firm-year observations are skewed to the lower end of the spectrum as can be seen from the relative position of median and mean values. However, both reach positive values indicating profitability even on the lower levels of the ROA spectrum.

34

largest TMT of our sample with 14 members. The majority of TMTs is larger than the average of the sample (which represents 4 member teams), as the median reached higher value relative to the mean. The second independent variable measured as a percentage of females in the TMT with the median being 0 reflects the fact that women were included in the executive suites only in 136 firm-year observations. Maximum female representation in our sample occurs with 4 women out of 7-member TMT reaching 0.57 percent, a value that according to Krishnan and Park (2005) represents balanced proportion of male and female executives. From both the mean and median values it is apparent that the sample is skewed in favor of males.

The third independent variable measuring age diversity as coefficient of variation reaches minimum values of 0 thanks to the one member TMTs with the maximum variance of 0.43 reflecting a TMT with the youngest member of 30 and oldest member of 63 years of age. The youngest executive in our sample is 292 _{while the oldest is 81}3_{. From the low value of}

standard deviation (0.0678242), we can conclude that there is not much age heterogeneity present in our sample, which is, however, not unexpected and in accordance with the theoretical literature presented. The lower number of observations for this variable reflects the unavailability and resulting missing age data for 118 firm-year observations. This means that all regression models including variables Age or BlauAge will analyze only observations for which data for these variables is available.

The moderating variable of industry turbulence operationalized as the average beta of industries and transformed by natural logarithm reaches both positive and negative values, with the extremes being -0.52 representing the most stable industry and 0.7 the most turbulent one.

All control variables are transformed by natural logarithm due to the normal distribution requirements of OLS regression except for GDP growth, therefore their log versions are included in Table 1. Firm size measured by Total Assets in thousands of US dollars spanned from 320184 _{to 5.483e+08}5_{, representing the vast scale differences between sample companies.}

After the transformation, its log value extremes reach only 10.37405 and 20.12228 respectively. From the relative position of median to mean values, only Firm Age is skewed towards the lower values, indicating majority of the companies being comparatively younger than the rest. This could be explained by the only recent liberalization of CEE countries and transition from centrally-planned to market-based economies.

2_{executive from Croatia} 3_{executive from Hungary}

35

Lastly, measured with the help of Blau index, are the independent variables of age and gender diversity with the highest diversity ratio of 0.5 for women representation and 0.67 for age heterogeneity, zero values reflecting complete homogeneity. As mean, standard deviation and median indicate, there is a higher age diversity present in our sample compared to the gender diversity.

Table 1: Descriptive statistics

Variable # of observ.

Mean SD Median Max Min

36 3.2 Correlation

In Table 2 below we present the Correlation matrix. There is an expected and high correlation present between the two alternative measures of diversity, especially between the female percentage measure and the Blau index for gender diversity. However, as these variables do not enter the regression models simultaneously but are substitutes for the robustness test, it does not present a problem for the analysis. There is also relatively high correlation present between Firm size and the Turbulence of the industry, which we can interpret for our sample as a tendency of larger companies to operate in the more turbulent industries within the CEE region. There is also a low to medium correlation between the Blau indices and TMT size, reflecting a natural predisposition of larger groups for higher diversity.

38 3.3 Regression results

In order to find answers to our research questions, OLS regression method is implemented with the results presented in Table 3 below. Eight different models are used to test the proposed hypotheses with gradual introduction of the variables. As model (8) is the most complex one containing all variables and including the interaction terms, we will take it as the primary model considering it should produce the most accurate results.

Model (1) tests the predictive power for ROA of the control variables only. Coefficients for all control variables are negative and insignificant. The only significant result provides Level of education at the highest 0.01% level, which is unexpected given opposing findings of previous studies (i.e. Bantel & Jackson, 1989).

Models (2) (3) and (4) introduce each of the independent variables separately. Model (2) shows that TMT size is a highly significant predictor of company ROA (p<0.01) as expected, but contrary to our predictions with a negative coefficient of -1.831, therefore surprisingly rejecting Hypothesis 1. In line with our expectations, model (3) shows positive relationship between Female presence in TMTs and company performance. However, this effect is insignificant, therefore support for the Hypothesis 2 cannot be accepted just yet. Results of model (4) suggest a nonsignificant negative relationship between Age diversity, contrary to the predictions of Hypothesis 3. However, results of these three models have to be interpreted cautiously, as the leading model is the last model (8).

In the model (5) the direct effect of the moderating variable is tested. With a high significance at 0.01% level and a negative coefficient of -4.720 the model suggests that companies within the CEE transition countries perform better in the stable industries than in highly turbulent ones.

Model (6) introduces all three independent variables together, supporting results of models (2) (3) and (4). Values of coefficients of Female percentage and Age diversity slightly increased but signs remained unchanged. Higher gender diversity is positively associated with company financial performance while age diversity shows a negative relationship, both remained insignificant. This model confirms the significant negative relationship between TMT size and ROA of model (2), although showing a slight decrease in significance (from p<0.01 to p<0.05), further supporting the rejection of Hypothesis 1.

39

Table 3: Regression results