The Influence of Top Management Team
National Heterogeneity on Firm Financial
Performance
University of Groningen
Faculty of Economics and Business
Ekaterina Myller
Student number: s1579770
Master Thesis in International Business & Management
First supervisor: Dr. H.C van der Blonk
TABLE OF CONTENTS
CHAPTER 1 ...3
INTRODUCTION ...3
1.1 Internationalization and performance...3
1.2 Moderators of organizational outcomes ...5
1.3 The relevance of the study to the international management issues and problem indication...6
1.3.1 Lack in addressing nationality issues in previous top management team research...7
1.3.2 Further testing and extending UE framework...8
1.3.3 Research question and sub questions...9
CHAPTER 2 ...10
THEORETICAL BACKGROUND ...10
2.1 Previous research on nationality: results and gaps ...11
2.2 Organizational outcomes as a reflection of TMT characteristics...13
2.3 Costs of internationalization and explanation based on UE framework why nationally heterogeneous teams are better off in dealing with them...15
2.4 Conceptual model and hypotheses of the study ...21
Figure 1. Conceptual model ...23
CHAPTER 3 ...24 METHODOLOGY ...24 3.1 Type of Study ...24 3.2 Type of data ...24 3.3 Variables...25 Performance...25 Internationalization ratio...26
National Heterogeneity Index of TMT...27
TMT size...27
Company size...29
Debt to equity ratio...29
3.4 Sample ...31
3.5 Regression analysis ...32
CHAPTER 4 ...34
EMPIRICAL RESULTS ...34
4.1 Procedure ...34
4.2 Checking for assumption and evaluation of results...37
4.2.1 Multiple regression assumptions – first and second regression for first hypotheses...37
4.2.2 Evaluating results of the regressions one and two run to test hypothesis one...39
4.2.3 Multiple regression assumptions – third and fourth regression for second and third hypothesis.40 4.2.4 Evaluating results of the regression three and four run to test hypotheses two...43
4.2.5 Evaluating results of regression three and four in run to test hypothesis three...44
5. Discussion and conclusion...45
Chapter 1
Introduction
1.1 Internationalization and performance
In virtually all industrialized countries, multinational expansion has been a major feature of post – war corporate development (Grant 1987) thus it is not surprising that issue of international diversification of the company and its impact on performance has been studied widely for a long period of time. However the empirical results on this relationship vary and different views exist. Some studies have identified that the relationship between internationalization and performance is linear and positive (Delios & Beamish 1999; Doukas & Travlos 1988; Grant R.M 1987), some predicted and found curvilinear relationship: standard U-shape or inverted U-shape relationship (Hitt, Hoskisson, Kim 1997; Lu & Beamish 2001). Furthermore inverted J-shape was found by Gomes and Ramaswamy (1999), S-shape by Riahi - Belkaoui (1998) and Lu & Beamish (2004). Insignificant or equivocal relationship between international diversification and performance was found by Brewer (1981), Morck & Yeung (1991), Buhner (1987), Shaked (1986) and finally negative relationship was uncovered by Denis, Denis & Yost (2002) and Geringer, Tallman & Olsen (2000).
Beamish 1999) and also arbitrage institutional restrictions such as tax code, antitrust provisions and financial limitations ( Doukas & Travlos 1988).
Regardless the resource based view, exploitation of economies of scale & scope; assets available in a foreign country and institutional restrictions arbitrage suggest a positive relationship between internationalization and performance the benefits arising from internationalization advantages can be lower than transaction costs associated with internationalization (Hitt, Hoskisson , Kim 1997; Gomes and Ramaswamy 1999) and costs rising from being new and foreign ( Lu & Beamish 2001 ; Lu & Beamish 2004) . Transaction costs increase further as company starts to operate in more geographically diversified environment and result in inverted U-shape relationship between internationalization and performance (Hitt, Hoskisson, Kim 1997; Gomes & Ramaswamy 1999). Gomes & Ramaswamy (1999) indicated J-shape of the relationship and explained it by introducing an “optimum level” of internationalization beyond which benefits start to decelerate and costs accelerate.
The rational behind standard U-shaped relationship are“ liabilities of newness and foreignness” (Lu & Beamish 2001). It is suggested that as the company expands internationally, to the unknown environments, it faces problems caused by lack of local knowledge. However this holds only when degree of internationalization is low implying that when international expansion of the company goes further these problems vanish because of the gained experience resulting in the positive relationship between performance and internationalization ( Lu & Beamish 2001).
costs attributable to “liabilities of newness and foreignness“ are reduced, transaction costs increase due to the geographical scope growth resulting in drop off in profitability (Lu and Beamish 2004).
1.2 Moderators of organizational outcomes
The variety of diverse and sometimes contradictory empirical results suggests that there is still a space for further investigation of the relationship between firm internationalization and its performance. The existence of mixed result can be due to the fact that the relationship between internationalization and performance is more complex than has been theoretically argued and empirically tested. Another possible explanation for mixed results is the existing gap in current research regarding the measurement of internationalization. Not only the extent of internationalization (e.g. sales generated from overseas operations) but also cultural diversification of the international operations should be considered when studying the relationship between internationalization and performance (Gomez-Mejia & Palich 1997).
Furthermore, mixed bag of findings introduced above can mean that there could be something that moderates the relationship between internationalization and performance, moderator variable that affects the direction and strength of the relation between internationalization and performance. Caves (1996) suggests that one important dimension that can moderate the exploitation of benefits of an internationalization strategy are firms intangible assets e.g. management skills. The study of Hambrick and Mason (1984) introduced Upper Echelon theory which is a framework suggesting that organizational outcomes such as profitability, growth and survival are a reflection of their top management teams.
heterogeneity and firm performance is moderated by the international strategy of the firm. Tihanyi, Ellstrand, Daily and Dalton (2000) found out that the top management team composition plays an important role in the international diversification of firm operations.
Accordingly, Upper Echelon perspective is used as a stepping stone for developing a theoretical framework and studying the moderating effect of top management national heterogeneity on the relationship between firm financial performance and internationalization. The reasoning behind this will be discussed in more details in chapter two.
1.3 The relevance of the study to the international management
issues and problem indication
Internationalization is an actual and an important issue for companies. The effective management of the company in an international setting becomes crucial. A major argument of research in the field of international business is that the management of a company which is highly international, in term of operations, differs in many ways from the companies which business activities are limited to the home market. In order to be successful and being able to compete in the international business environment, managers need to predict and deal with worldwide social, political and market dynamics in a timely flexible and differentiated manner (Kogut., 1985; Prahalad & Doz, 1987)
Tangible and intangible resources of the company are needed to be redistributed in response to changing scale, scope and cost advantages. Also potential problems arising from communication and coordination in culturally dispersed subsidiaries need attention. These problems should be continuously monitored and addressed. (Hofstede , 1980; Roth & O’Donnel, 1996) . Taking into consideration the complexity of the tasks that mangers of international firms face, a question arises: is there any particular top
management characteristics that provide competencies crucial for managing international firms?
national heterogeneity. It will be studied if national heterogeneity has an impact on the performance of international companies. Chapter two will discuss the using upper echelon perspective and social categorization theory to explain why such a relationship is expected to exist. Following chapters 1.3.1 & 1.3.2 formulate the research problem that is reflected by the lack of nationality research within management studies field.
1.3.1 Lack in addressing nationality issues in previous top
management team research
For the past two decades, academic literature has been focusing on how top management teams influence firm performance and strategic behavior (e.g. Finkelstein & Hambrick , 1990; Goodstein, Gautam & Boeker , 1994; Hambrick , Cho & Chen., 1996; Keck, 1997). Despite the number of studies concerning TMT composition and its characteristics, the national context of TMT’s has not been taken into account (Heijtljets, Olie, Glunk 2003; Gong 2003; Hambirck, Davison, Snell & Snow 1998). This is surprising because cultural composition of the group have a great influence on e.g. decision making, problem solving, communication and information processing (Cox, 1991; Watson, Kumar, Michaelsen 1993; McLeod, Lobel & Cox, 1996; Carpenter 2004) that in turn influence organizational outcomes e.g. performance (e.g Elron , 1997).
1.3.2 Further testing and extending UE framework
The objective of this study is to test UE framework and extend previous studies that are concerned with such issues as the relationship between the composition of top management teams, internationalization and firm performance. Testing end extending UE model will make it possible to answer sub and main research questions introduced in chapter 1.4.
Testing of the UE framework will be conducted by studying if company’s financial performance attributable to internationalization is a reflection of its top management teams’ national heterogeneity. Theoretical arguments that make issue of nationality relevant to UE framework and explain why it is worth testing are discussed in the chapter two. Extension of UE framework will take place by taking national heterogeneity variable into account as this variable was not included in the original UE framework. Carpenter (2004) in his review of UE research papers states that race and gender represent relatively recent and needed traits in the characteristics that UE research covers thus research on nationality in the context of UE perspective can be considered as an actual one.
In this study it is argued that the moderating effect of TMT’s national heterogeneity on the relationship between internationalization and performance is not linear but depends not only on the presence of certain characteristics within the TMT but also on the extent of these characteristics. The extent of the particular characteristics in TMT is understood in this thesis as the number of different nationalities within one top management team; in other words heterogeneity of TMT in terms of nationality. In chapter two it will be also discussed why different levels of national heterogeneity within a TMT can have a different impact on the relationship between internationalization and performance.
characteristics influence theoretical constructs such as behaviour and decision making process that in turn result in particular outcomes. This study will give theoretical argumentation on the issue of national diversity and its influence on cognitive bases, knowledge and skills that in turn are recognized to have an effect on e.g. decision quality and information process ability resulting in crucial dealing with increased complexity capability e.g. with costs associated with internationalization.
Testing and extending UE framework will also have implications on the practical level as the result of statistical test will identify if nationally heterogeneous teams have a moderating influence on the relationship between internationalization and performance. If significant relationship will be found between internationalization, performance and national heterogeneity of top management team, it will prove and extend – in terms of national heterogeneity – upper echelon perspective introduced by Hambrick and Mason (1984). If it will be discovered that nationally heterogeneous teams are positively moderating the relationship between internationalization and performance, it will be an important piece of information for the companies. This would imply that one of preconditions to be successful international firm in terms of financial performance is that top management teams should be composed of managers with diverse national backgrounds. Nevertheless, if results turn out to be insignificant it would indicate that national diversity as a characteristic of top management team do not have an influence on the outcomes. The reasons behind possible insignificant findings could be that national heterogeneity has an effect in synthesis with other characteristics or maybe, it is not significant, could have an impact if measured in another way and so on. Possible reasons of insignificant results are discussed in chapter number five.
1.3.3 Research question and sub questions
In order to conduct this study and meet the objective of the thesis the research question is formulated to be following: How TMT’s national heterogeneity influences the
relationship between firm performance and internationalization? To answer main
internationalization and performance? and ‘’At what levels national heterogeneity within top management team will influence negatively the relationship between internationalization and performance? In chapter two, were theoretical background of the study is discussed, three hypotheses reflecting sub-questions are introduced.
Chapter 2
Theoretical background
Chapter two is divided into three parts. In part 2.1 the issue of nationality research is addressed. This part introduces previous researches conducted in that field, their findings and gaps. I will also give arguments why nationality research is considered to be important and worth studying.
In the part 2.2 the upper echelon perspective is introduced. This part will review UE perspective, discuss studies that tested and extended it. Upper echelon part will also discuss the fit of this study into upper echelon model by explaining why and how UE perspective can be used to support the proposition that national heterogeneity influences the relationship between internationalization and performance
2.1 Previous research on nationality: results and gaps
As the objective of this research is to test and extend the statement proposed by upper echelon view that top management team characteristics are influencing the organizational outcomes, national heterogeneity is chosen to test this proposition for several reasons. There have been plenty of research conducted (app1. table 1) contributing to UE framework by studying various types of demographic characteristics and heterogeneity within the top management teams. These researches have been mainly concentrated on such demographical characteristics as age, tenure, educational and functional background, management experience, etc. (e.g. Wiersma & Bantel) but issue of nationality have been ignored (e.g. Hambrick, Cho & Chen, 1996 ). Further, empirical studies that have been carried out in the field of top management teams were focused mainly on U.S based companies (Heiltjes, Olie & Glunk, 2003). As this study uses an “upper echelon perspective’’ as a stepping stone for formulating the conceptual model and theoretical background it is important to mention that extension of this perspective into international context have been narrow (Heiltjes, Olie & Glunk 2003). Even though there were several studies extending UE perspective into international context during last years, the issue of national diversity at the top management level remains unstudied (Heiltjes, Olie & Glunk 2003).
The lack of nationality research on top management team level can take place because a lot of research that have been done focusing on the top management teams and their influence on organizational outcomes is embedded in the ’’upper echelon perspective’’ of Hambrick and Mason (1984). However this framework did not originally include dimension of nationality. Nevertheless nationality is considered to be a relevant demographic variable concerning this framework (Tsui and O’Reilly, 1989; Milliken and Martins, 1996 Harrison, Price & Bell 1998) and its relevance is discussed later in this chapter.
and Esser (1999) reported that between 1995 -1998 the amount of directors on the board of US based companies increased form 39 to 60 percent. European multinational companies are reported to have at least one foreign director on the board in 90 per cent of largest companies by market capitalization. These evidences stress the importance and actuality of the research on nationality (Lublin 2005).
The gap in nationality studies of top management teams was address by Heiltjes et al. by studying national composition of top management teams in Sweden and the Netherlands. They concluded that extent of national diversification of the top management teams in those countries did not reach the internationalization level of the company in general. However this study was descriptive in its nature and national diversity was not linked to other variables of organizational outcomes e.g. performance. The study of Heiljes et al. (2003) gives a stimulus for the future research regarding nationally diversified top management teams and their performance.
Another study that addressed the issue of top management teams’ national diversity is a resent study conducted by Palmer and Varner (2007). They studied national composition of leading multinational companies in US, Asia and Europe and discovered that most international top management boards are in Europe. Palmer and Varner recommend that there should be research done on the issue of top management team internationalization and firm performance. This recommendation serves as one of the motivations for this study.
Moreover, there exists number of researches done where national heterogeneity is not only focus of research but is combined with other demographic attributes and combined with indicators related to strategy , performance and structure. The study of Caligiuri, Lazarova and Zehetbauer (2004) focused on studying whether operations of US based companies worldwide (in terms of number of countries it is active in) and top management teams’ national diversity is associated with other identified indicators of internationalization such as foreign assets ratio, foreign sales ratio, foreign subsidiaries ratio, and foreign employees ratio. This study of Caligiuri et all. (2004) extended UE framework by adding a national diversity variable though it was only tested in relation to strategic choice, not performance.
indicators. Further, even it was not the focus of the study the results indicated that there exists a relationship between national diversity of the top management teams and performance. In particular, national diversity of the top management team was positively related to foreign sales ratio. This relationship motivates the current study which goal is to test whether national diversity of top management team moderates the relationship between internationalization and performance. Study of Caligiuri et all. (2004) gives hints that such a relationship exists.
Elron (1997) studied the relationship between cultural heterogeneity of TMT and group performance as well as organizational performance. In order to carry out the study the respondents were asked to report their nationalities, meaning that nationalities were used as the proxies for the culture. The study sample consisted of the top management teams of MNC subsidiaries. It was indicated that cultural heterogeneity among top management teams was positively associated with both, the level of issue based conflict the team experiences and the top management team performance. Issue based conflict in TMT affected negatively top management team performance but had a positive effect on subsidiary performance. Overall it was concluded that cultural heterogeneity within top management team is important for the functioning of the top management teams and their subsidiaries. The results of Elrons (1997) research stimulate this thesis because the relationship between organizational performance and cultural heterogeneity was established.
2.2 Organizational outcomes as a reflection of TMT
characteristics
equipment newness, backward and forward integration, financial leverage, administrative complexity and response time whereas performance is seen as a consequence of these strategic choices and is measured by such variables as profitability, variations in profitability, growth and survival. Hambrick and Mason (1984) suggest that background characteristics can be used to forecast givens and behaviour of managers. This view is based on the behaviour theory, on empirical studies done and support found in versatile areas of academic research. ’’Givens’’ are stated to reflect the decision maker’s cognitive base and values. In other words demographic characteristics are suggested to be sound proxies for fundamental difference in cognitions, perceptions and values. Carpenter (2004) summarizes the theory underlying UE framework as following ‘’ … faced with such common goals as information overload, ambiguous cues, and competing goals and objectives, executive perceptions of stimuli are filtered and interpreted trough cognitive bases and values. Because this psychological constructs are unobservable, the theory posits that observable managerial characteristics are efficient proxies that provide reliable indicators of the unobservable psychological constructs …’’
The framework of Hambirck and Mason is a theory building and encouragement of academic world to test proposed relationship. As a result about 30 different studies ( app. 1 table 1) were conducted during the period of 1996 - 2003 in order to test the proposed framework. These works brought important insights into executives’ effects on organizational outcomes and enhanced the capability to generalize the UE framework. After reviewing number of studies (app.1 table 1), Carpenter (2004) states that TMT characteristics are clearly linked to the firm strategic profiles as well as to the firm performance. There also exist empirical findings suggesting that UE model is applicable in diverse context including international organizations based worldwide, organizations with different life cycles as well as organizations in different business and strategy arenas. Carpenter (2004) also suggested that because UE framework was tested in international as well as in domestic settings, in recently established and old firms, in profit & non-profit organizations and public agencies, it is increasing the reliability and capability to generalize the model.
states that despite the model having been proved and extended in various ways there still remains a need to show when, why and how TMT demographics influence particular cognitions, socio-cognitions and behaviours. This study will address this issue to a certain extent by discussing it in the following part of this chapter why and how national heterogeneity that is demographic characteristics of TMT is expected to influence TMTs and their behaviour when they are faced with the problems arising from internationalization.
2.3 Costs of internationalization and explanation based on UE
framework why nationally heterogeneous teams are better
off in dealing with them
The stylized version of the original UE model (app.1 figure 2) introduced by Carpenter (2004) on the bases of the synthesis of the research results reported up to 2004 suggests that top management team demographics can serve as the proxies for a number of theoretical constructs (e.g. behavioural propensities, cognitions, access to information, skills, relative status within top management team or across firms) that in turn influence organizational outcomes. At the same time is generally accepted that nationality is stated to influence values, behaviour and cognitive schema (Hambrick, Davison, Snell & Snow, 1998). Consequently it can be deducted that since nationality is a demographic variable, it influences before mentioned constructs and as these constructs influence the performance it can be stated that nationality influences performance through influencing such constructs as behaviours propensities, cognitions and access to information. . Internationalization can be considered here as a condition that is influencing performance by creating advantages but also challenges that company should overcome in order to be successful. Thus, based on the reasoning introduced below, it is expected that national diversity of the TMT can help to overcome the costs of internationalization thus have a positive impact on performance.
(e.g Hitt , Hoskisson & Kim,1997 ; Hambrick , Davison, Snell & Snow, 1998. Elron, 1997) .
The argument, that international diversification of the company creates benefit and costs, is an issue that academic literature seems to agree on. When firms are making foreign investment, firms’ management is faced with the challenges related to new operations. These challenges can include e.g. purchasing and installing facilities, finding stuff, establishing internal management systems and external business networks. The costs associated with these challenges can be classified as problems of the “liabilities of newness and foreignness” ( Lu & Beamish 2001). In particular, company faces problems that result in extra costs because it does not know the new environment and it is not known by that environment itself (customers, suppliers, local authorities).
Resistance or openness of TMT to new information can have an impact on how company will be able to deal with the “liabilities of newness and foreignness“. It is documented that people carry particular stereotypes of different cultures and behave towards members of that culture or in general deal with that culture according to the expectations that reflect these stereotypes (Thomas 2002). New information (that is different from our expectations) about the member of that culture is often discounted as not representative; thereby maintain the stereotype (Thomas 2002). This can slow down the process of overcoming the “liabilities of newness and foreignness“ because the perception of new environments will be shaped by stereotype picture of that environment. It can be expected that if the TMT is nationally heterogeneous itself the stereotype power is not that strong because members could already experience before that the stereotype picture that we carry about specific culture does not always hold. It is argued, studied and proved by Katz (1995) that stereotype bias occurs less when information is rich and where diagnostic information is available to partakers. Adapting Katz to the case of “liabilities of newness and foreignness’’ and nationally heterogeneous top management teams it would imply that more diversified diagnostic information is available in top management teams than in nationally homogeneous. This can take place because when
topmanagement team consists of managers with different national backgrounds it results
in a greater pool of information on different cultural environments, diverse cognitions, values and perceptions. The availability of such resource would decrease stereotype bias that might slow down the process of overcoming ’’liabilities of newness and foreignness’’.
general. Theory of ’’liabilities of newness and foreignness’’ says that as company gets more internationally involved the cost of these liabilities vanish because as company increases its international expansion experiential learning about how to establish new subsidiary efficiently in a host country decreases the level of costs of being new and foreign (Lu and Beamish, 2004). By synthesising theory of “liabilities of newness and foreignness” and findings of Caligiuri et al. (2004) costs of “ liabilities of newness and foreignness“ for the companies with the nationally diversified top management are expected to be lower. This is expected to take place because companies with nationally diversified top management teams are associated to be more internationalized in general than those with nationally homogeneous and as Lu & Beamish suggest highly internationally expanded firms are said to have lower costs attributable to ‘’liabilities of newness and foreignness’’.
of cultural attitudes within top management teams should result in better decision making.
In general proponents of cultural diversity have maintained that nationally diverse workforce can foster a wide range of creative decision making alternatives, effective decision making and high quality decisions (Cox, 1991; McLeod, Lobel & Cox, 1996). Previous research in the field of heterogeneity and performance suggests that national heterogeneity can influence positively number of alternatives considered, quality of ideas, degree of cooperation in complex tasks ( Cox 1991; Mc Leod , Lobel 1992 ; Watson, Kumar, Michaelsen 1993 ) and in general broaden problem-solving capacity (Hambrick , Davison, Snell & Snow, 1998). McLeod and Lobel (1992) found that ethnically diverse groups generated more and higher quality ideas on a brainstorming task. Nationally diverse groups were also found to outperform homogenous groups at identifying problem perspectives and generating solution alternatives (Watson, Kumar, Michaelsen 1993).
Problem identifying and solving capacities, abilities to generate diverse and high quality ideas, ability to cooperate on complex task can be expected to have a positive impact on firm performance since these abilities are important in day to day functions of top management teams. As a firm becomes more international, complexity of the decisions made increases because of more complex environment, e.g. structural and institutional barriers should be taken into consideration when solving problems and making decisions. Increased amount of information arising from versatile environments results as well in increased complexity. Thus the particular characteristics ( that have been discussed above) of nationally heterogeneous groups should enhance overcoming transaction costs associated with coordination, governance, communication, information processing and managing structural and institutional barriers in international environments.
overload that can arise due to internationalization) executives perceptions of stimuli are filtered and interpreted through cognitive base and values (Carpenter 2004). Consequently it means that as this information is processed through the diverse minds of nationally heterogeneous managers the final result can be creative and more accurate solution for a problem and these in turn result in a better performance.
Above introduced theoretical arguments suggest that moderating effect of national heterogeneity on internationalization and performance is positive and linear. However, looking from the different point of view it is suggested that as cultural diversity (within the group there are members representing diverse nationalities) occurs social comparison and categorization process occur, in-groups / out - groups biases and cognitive biases may emerge, creating barriers to social interaction ( Blau, 1977; Tsui & Egan, O’Reilly 1992) that in turn influence performance.
Social categorization theory suggests that people tend to define themselves into social categories using characteristics e.g. race (Richard & Schelor, 2002). As heterogeneity in group reaches moderate levels, the psychological process associated with social identity theory and self categorization processes may be more likely to occur (Tijfel & Turner, 1985). These processes would result in individual behaviours such as solidarity with others in a group with same nationality and discrimination against out-groups ( Tijfel & Turner, 1985). These behaviours in turn can be expected to influence group cohesion, member satisfaction and communication in a negative way (Hambrick et all.,1998 ; Tsui & Egan, O’Reilly 1992). The presence of different belief systems within the TMT, arising from existence of various nationalities within the group, can lower the integration within executives that in turn leads to loss of cohesion, low satisfaction and biases (Cox, 1991). Difficulties associated with the moderate levels of heterogeneity in group’s e.g. top management teams are likely to lead to negative performance (Richard, Barnett, Dwyer, Chadwick 2004). This point of view would suggest that as national heterogeneity grows from low up to high levels the performance of the company would decline.
will be reduced (Alexander, Nucholos, Bloom, Lee, 1995). Based on the suggestion of the theoretical approach of social categorization theory it can be proposed that different levels of national heterogeneity within the top management team can influence in a different way on the relationship between internationalization and performance.
2.4 Conceptual model and hypotheses of the study
Combining theoretical views discussed in parts 2.1, 2.2 and 2.3 the conceptual model is constructed (figure1). This conceptual model reflects variables of the study excluding control variables. Internationalization of the firm measured as internationalization ratio is an independent variable that influences financial performance that is the dependent variable of this study. The relationship between internationalization and performance that was reviewed in the first chapter was an object of number of previous studies and was empirically proved to exist. Aspects that are influencing performance and are attributable to internationalization (transaction costs and “liabilities of newness and foreignness“ ) were also discussed previously in chapter two.
However mixed results of empirical evidences about the nature of relationship between internationalization and performance lead to proposition that there can be something that might influence this relationship. Theoretical discussion based on UE perspective introduced in previous parts of this chapter suggests that the relationship between internationalization and performance can be influenced by national heterogeneity of the top management team. National heterogeneity of top management team is seen as a moderator variable in this study.
have greater knowledge capital and more diversified cognitive bases. These attributes in turn are suggested to improve decision quality and result in better information processing ability. These abilities are important in order to successfully cope with the challenges arising from internationalization of the companies
On another hand the national diversity within top management team can also cause problems. It is proposed by social categorization theory that as the heterogeneity level starts to increase, problems may occur. The result would be decrease in group cohesion, satisfaction of group members and communication. These issues in turn would result in delay in decision making and low commitment to decisions. This situation would lead to a failure in exploiting advantages of nationally heterogeneous teams that were suggested to help in overcoming costs associated with internationalization. Failure in exploitation of advantages would result in situation were costs associated with internationalization has not been be decreased thus negative influence on the relationship between performance and internationalization can appear. However social categorization theory suggests that this would only occur when national heterogeneity within top management team is increasing up to high level. As national heterogeneity within top management team reaches high levels the negative influence vanish because ‘in group’ & ‘out group’ classification will not take place anymore thus negative effects are decreased. Based on social categorization theory two following hypotheses are formulated: H2. The increasing levels of national heterogeneity up to high level will have a negative moderating effect on the financial performance of the company attributable to internationalization and H3. High national heterogeneity within TMT team will have a positive moderating effect on the financial performances of the company attributable to internationalization.
Internationalization of the Firm Financial Performance National Heterogeneity within TMT
performance however this is expected to take only in the sample firms whose TMT’s national heterogeneity is below high levels. High levels of national heterogeneity are expected to influence positively the relationship between internationalization and performance thus it is expected that in the sample firms with high national heterogeneity within TMT the relationship between internationalization and performance will be positively moderated by national heterogeneity. Also main research question: ’’How TMTs cultural heterogeneity influences the relationship between firm performance and internationalization?’’ is answered by interpreting results of the statistical tests. It can be seen from the results of regression analysis (app.3) how different levels of national heterogeneity within top management team influence the relationship between internationalization and performanceFigure 1. Conceptual model
Chapter 3
Methodology
The purpose of the chapter 3 is to introduce methodological issues of this study. At first, in chapter 3.1, the type of this study will be introduced. It will be followed by chapter 3.2 that discusses issues related to the data. In chapter 3.3 variables used in this study are introduced and operationalised. Chapter 3.4 deals with the sample of this study. Finally chapter 3.5 introduces the regression models.
3.1 Type of Study
This study can be labelled to be a research on international diversification, performance and TMTs national heterogeneity that addresses prerequisite of successful management in an international company. It represents quantitative, deductive reserch implying that at first literature have been studied in order to develop hypotheses & conceptual model. Generated hypothese are then tested by using statistical tools. In this study multiple regression analysis were used.
Another classfication that fits to characterize this study is nomothetic research meaning that this research represents a quantitave study and aims to determine general laws and principles that apply to a population (multinational companies worldwide) in general.
3.2 Type of data
2006. Reason for this is the absence of data on TMT members’ nationality on a yearly basis.
3.3 Variables
To test the hypotheses, following data is needed: nationality of TMT members and TMT size, amount of overseas subsidiaries, number of countries in which firm is operating in, return on assets, debt, equity, number of employees of the company and type industry. This data is needed in order to build up variables that are included into the regression equation and models.
Nationality of TMT members and TMT size are needed to calculate national heterogeneity index of each top management team. National heterogeneity index is an independent, moderating variable of this study. TMT size is also used separately as a control variable. Amount of overseas subsidiaries and number of countries in which firm is operating in is used to calculate the internationalization ratio of the company. Internationalization ratio serves as an independent variable. Return on assets is used as a measure of financial performance of the company and thus serves as a dependent variable. Debt and equity figures are needed to calculate the leverage level of the company that is used as a control variable. Number of employees is needed to reflect a size of the company thus number of employees is used to create a size variable. Size of the company serves as a control variable. Membership in particular industry sector, in this case manufacturing and non-manufacturing sector membership is also used as a control variable. Overall nine different numbers and values were needed to construct seven variables. Further each variable of this study is operationalised and argumentations for its inclusion in this study are given.
Performance
relative efficiency of firm effectiveness regarding output and is also suitable for evaluating the achievements of synergy in business operations (Kim, Hwang, Burgers 1989). ROA is considered to be relatively resistant to the financial manipulations of management (Gerhart & Milkovich 1990). This measure is easy to access and it makes it possible to compare results with previous studies since it is a widely applied measure in research focused on the relationship between internationalization and performance (Ruigrok & Wagner, 2003). The logarithm of return on assets ratio, based on numbers of 2006, was used in regression analysis as a dependent variable. Return on assets ratio was transformed to logarithm in order to meet normality requirements.
Internationalization ratio
The internationalization level of the company, the independent variable, is calculated by adapting methods used by Sanders & Carpenter (1998) and Lu & Beamish (2004). It is calculated in following manner: at first, total amount of overseas subsidiaries and number of countries it is operating in is counted. Second, these two figures for each company are combined into a single ratio of internationalization by dividing each count either by the maximum number of subsidiaries or the maximum number of countries the firm is operating in. The last step in formulating ratio of internationalization involves calculating the average of the divided numbers and combining them into a single ratio.
National Heterogeneity Index of TMT
Heterogeneity index of TMT represents the moderator variable of this study. National heterogeneity of the top management team means the variety of different nationalities represented in it. Nationality is considered to be an efficient measure of culture because researchers within the field of international group studies maintain that national culture determines the group members’ behaviour and patterns of interaction (Bochner & Hesketh, 1994; Cox et al., 1991; Earley, 1993; Maznevski & Chudoba, 2000; Tayeb, 1996). This view is also in line with the stylized model of upper echelon perspective (app.1, fig.2 ) elaborated by Sanders et al. (2004) that suggests that top management demographics, such as nationality, can serve as a proxy for theoretical constructs – such as behaviour propensities (Carpenter, Geletkanycz, Sanders 2004).
The national heterogeneity index of TMT is calculated by adapting Blau’s (1977) index, 1-∑─Рi², where P is the proportion of individuals in a category and i the number of categories. In order to access impact of cultural heterogeneity on the relationship between internationalization and performance it was checked when the TMT member got the position. The rationale behind it is following: if one of the members of TMT joined it only at the end of the year her/his effect on the studied relationship can be questionable. Since performance data was collected for the year 2006 it was required that the member of TMT has hold a position at least starting from the beginning of the year 2006. If member in question did not hold that position yet at that time, it was further studied in order to identify who was holding that position before and consequently that member was taken into account. If such information was not available, following company on the list was taken.
TMT size
In the review of studies focusing on UE perspective Carpenter et al. (2004) stress the importance of the controlling for TMT size. It is reported that surprisingly many studies do not take into consideration TMT size when studying effects of TMT heterogeneity. It is stated in the review that it is an important issue to control for because heterogeneity is well-acknowledged being positively associated with the size of the group being studied. Haleblian & Finkelstein (1993) suggest after reviewing number of publications that the size of the group have implications for group effectiveness, conflict and information processing ability in other words for the strategic capacity of the firm. Accordingly, failing to control for TMT size leads to the problem in interpretation of statistical results. It can be problematic to understand whether significant findings should be attributed to the heterogeneity or to the unobserved effect of TMT size. Thus it is important to include the control for TMT size when studying heterogeneity effect.
Based on the review of large pool of studies dealing with the top management teams, a following definition of top management team was constructed: top management team consists of e.g. Chairman, Chief Executive Officer (CEO), Chief Operating Officer (COO), Chief financial Officer (CFO) of the organizations or any other officers who are members of the corporate board. In view of that definition it was decided to include to the TMT only executives that are also part of the board of directors.
In firms with a two-tier governance structure only management board is taken into account. In a two-tier system, the management board and the supervisory board are separate entities (Heijltjes et al., 2003). The management board is responsible for the day-to-day management and thus represents the TMT. The supervisory board includes the non-executive directors and is responsible for monitoring the management board (Maassen and Van den Bosch, 1997). Apparently, in countries with a one-tier governance structure, it would mean only the executive managers who are also member of the board of directors. Generally the UK is considered as a country with one-tier governance structure thus executives that are board members were included into the TMT. However, case by case approach was used implying that at first the governance structure of the company was identified based on the information provided by the company’s web pages and only after that information on TMT members was collected.
justify their choice of such TMT composition stating that executives who are also members of the firm’s board is an objective and recognized indicator of being a component of a firms dominant coalition. They also suggest that the group of managers in the inner circle of an organization has eventual responsibility for setting course of action. Haleblian & Finkelstein (1993) say that board membership serves as a distinction point between top managers and other managers and is similar across industries.
Given the fact that this study focuses on “top management team” as a group – and given the fact that underlying theoretical framework of the study is particularly concerned with advantages and disadvantages of nationally heterogeneous groups – the above introduced definition of “top management team” is considered to be applicable in this study.
Company size
Company size is included as a control variable and is measured as the number of employees that is well recognized measure of firm size and is widely used in the academic world (Hitt et all. 1997, Tihanyi et all 2000 etc.). Company size is generally included as a control variable in upper echelon research (Birkner, 2005). This variable is stated to have direct impact on the firm performance because of economies of scale, scope and market power (Orlando, 2000). Large firms are considered to have more ability to undertake strategic initiatives e.g. large firms are more capable in facilitating entry to new markets due to the existence of personnel and resources needed for that. (Tihany et. al 2000). Expansions of the activities result in gaining economies of scale and scope that in turn influence overall financial performance. In this study logarithm of size was used in order to meet normality requirements.
Debt to equity ratio
thus is used as a control variable in this study. This ratio was calculated by subtracting shareholders equity from assets and dividing by total assets of` the firm.
Industry
Control for industry was included into the regression model. Number of theoretical perspectives (e.g. economists, environmentally based perspectives, strategy researchers) recognizes the effect of industry membership on performance (Short, Ketchen, Palmer, Hult, 2007). It is reported in recent studies that industry effect plays a significant role in shaping firm performance (Chang and Singh, 2000; McGahan and Porter 1997). It has been documented that after controlling for outliers ten percent of the variance in performance is attributable to industry effect (McNamara , Aime and Vaaler, 2005). The theoretical and empirical evidence support the inclusion of control for industry effect into regression equitation as a control variable.
Table 1. Summary of variables
3.4 Sample
.
Study sample consists of a 100 UK based companies across industries. Data is collected from UK based companies because these disclose data needed for this research. Continental European countries keep such information as the nationality of the employees concealeded and US based compnies have already been widely studied.
While selecting UK based companies from Amadeus database, two criteria were used. At first, it was filtered to extract only companies that are ultimate global owners, ‘Global UOS’. This filter provided possibility to extract companies that are not subsidiaries of bigger companies. This was an important criterion because the objective is to study relationship in question on the company – level, not subsidiary. If sample would consist of just random number of companies that could be wholly owned, UK based
Variable Type Measure Name
National heterogeneity Moderator Variable
Cultural heterogeneity
index of the TMT Nat Het International diversification of the company Independent variable Internationalization ratio Int Performance Dependent variable
Log ROA ratio (Return
on assets ) Log Perf Firm size Control
variable
Log of employees
size Financial leverage Control
Variable
Log of Debt – to
equity ratio logdebtoeq TMT size Control variable Number of mangers in TMT TMT size Industry Control Variable Manufacturing vs. Non - manufacturing
companies or a subsidiary of another company that is based in the UK, the reliability of the study would be reduced.
Another criterion used was the company size that was measured by number of employees. Largest companies from the list were selected in respect to employee number. As a result final sample consists of companies that are in top 117 in respect to size. This is because sometimes if there was no information available on the company, the following company from the list was chosen to be included into the sample. Adapting rule of thumb of Green (1991), ( N>50+8×M , where N is sample size, M is number of independent varibales), research sample is 100 companies.
3.5 Regression analysis
As statistical tool hierarchical multiple regression analysis is chosen. In this thesis it is studied if the relationship between a continuous predictor variable, internationalization (independent variable) and continuous outcome variable, performance (dependent variable) is moderated by a continuous variable, national heterogeneity index (moderator variable). This means that effect of first continuous variable (internationalization) is expected to depend upon the level of second continuous variable (national heterogeneity index).
regression equation in order to see if moderating effect of national heterogeneity in top management exists if the relationship between internationalization and performance is curvilinear.
Two different regression equations were used to test all three hypotheses. Each hypothesis was tested including quadric and cubic terms. This generated two different outputs. The moderating effect of national heterogeneity on the relationship between internationalization and performance was tested separately for quadric and cubic regressions. The regression equation used to generate models 1, 2, 3, 4, 5 for each hypothesis (app. 3A, 3B and 3C) is equation number one that includes additionally to linear term also quadric term. The regression equitation that includes cubic term, equation number two is used to generate models 6, 7 and 8 (app. 3A, 3B, 3C).
Quadric and cubic terms are added to regression analysis because the underlying relationship between internationalization and performance on which moderating effect was tested is suggested by previous studies, that were discussed in chapters one and two, to be linear (U-shape or S-shaped) thus taking this into account exponential, non-linear terms were added. Models 5 and 8 from appendix 3A are used to evaluate hypotheses 1. Models 5 and 8 from appendix 3B are used to evaluate hypotheses 2. Models 5 and 8 from appendix 3C are used to evaluate hypotheses 3.
Table.2. Regression equations
Models Regression Equation
1,2,3,4,5 Y = b0 + b1 logdebttoeq 1 + b2 logsize 2 + b3 tmtsize 3 + b4 industry 4
+ b5 int 5 + b6 int 6 ² + b7 nathetin 7 + b8 nathetin * int 8
Chapter 4
Empirical Results
4.1 Procedure
Hypothesis number one was tested on the sample of 100 companies. Because hypotheses two and three are tested for different groups (national heterogeneity index up to high levels and high national heterogeneity within TMT) the sample was divided into such groups. In order to test hypothesis two and three, data sample was split in two groups indicating cases were national heterogeneity index had values up to 0.42 and cases with national heterogeneity index higher than 0.42. This separation of cases resulted in two groups.
In first group that had values up to 0.42, national heterogeneity was considered to be low and moderate. Sample with values equal or higher than 0.42, was considered to indicate cases high in national heterogeneity. Sample was split in two groups by including values within 15% around the mean and below to the first group. Consequently, second group included values that were higher than right side values within 15 % around the mean. This division of the sample resulted in two equal in size groups that included 50 cases each.
In order to test hypothesis such variables as size, performance and debt to equity variables were transformed into logarithms. This was done to avoid problems of positive kurtosis and skewness. This technique is considered as appropriate to use if such problems occur (Pallant, 2005). In order to test the moderating effect of national heterogeneity a moderator variable was constructed by multiplying standardised ratios of internationalization by standardised national heterogeneity indexes.
are below 2, preferable below 1 (Miles , Shevlin 2001). Logarithm of performance, industry, internationalization ratio and national heterogeneity index seem to have kurtosis values greater than 1. For this reason instead of Pearson correlation Spearman correlation is calculated in order to assess correlation between variables.
Table 3. Descriptive Statistics for the whole sample
Dependent variable, logarithm of performance shows missing values. This occurs because of transformation into logarithms. Values that are equal to or below zero can not be transferred into logarithms. Raw values of performance and debt to equity ratios had some negative values and consequently were not converted into logs. It was first tried to add a constant when transforming values to logarithms in order to make them first positive and then covert. However this approach resulted in even higher kurtosis thus only transformation of positive values took place.
Descriptive statistics of the sample show minimum value of -1.20% and maximum value of 4.30%, where mean is 2.0596 and standard deviation is 0.90396 for logarithm of performance. Logarithm of debt to equity also shows missing values, this happens for the same reason as in case of performance. The maximum and minimum values for logarithm of debt to equity ratio are 1.53 and -.74 respectively. Mean is at 0.3332 and standard deviation at .3894. Logarithm of size is valid for all 100 cases with the minimum value of 3.96 and maximum of 5.61. The mean for logarithm of size is 4.4629 and deviation is 1.80177.
Membership of manufacturing or non- manufacturing industry was converted into dummy variables. This resulted in the situation were about 30% of the total cases belong to the manufacturing industry (29 cases) and about 70 % to the non-manufacturing
N Min. Max. Mean Std. Dev. Skewness Kurtosis
VARIABLE Stat. Stat. Stat.
industry (71 cases). Even if the groups seem to be unequal in respect to the size, distribution is approaching normality according to skewness and kurtosis statistics.
Internationalization ratio of the sample firms vary between 0 and 0.6 with the mean of 0.1408 and standard deviation of 0.13202. This variable also shows rather high positive skewness and kurtosis, 1.174 and 1.147 respectively, however, it is notably higher than preferable values of skewness and kurtosis. Transformation for this variable was also not conducted because of the potential mulitcollinearity problems resulting from the transformation.
The national heterogeneity index for the sample varies between 0 and 0.75 with the mean of 0.3584 and standard deviation of 0.25103. This suggests that in general TMTs in UK are rather heterogeneous in terms of nationalities. Size of top management team varied quite a lot, from 2 to 10 persons. Mean of 4.6 is in line with average top management team size in the previous top management team research where only members of the board were considered as top management team members (app.1 table1).
All three hypotheses were tested by using hierarchical multiple regression analysis. It means that at first control variables (log of debt to equity ratio, log size, TMT size and industry) were added to equation. Model 1 (app. 3A, 3B, 3C) shows results when including these variables. Second step was to add independent variable, internationalization variable. Model 2 (app. 3A, 3B, 3C) shows result after including internationalization. Third step was to include either quadric or cubic term of internationalization. The results of addition of quadric term can be seen from the model 3 (app. 3A, 3B, 3C). When cubic term of internationalization is added, the outcome can be seen from model 6 (app. 3A, 3B, 3C). Further national heterogeneity ratio variable is added to regression analysis and as the last one variable that test for moderating effect (standardised national heterogeneity index * standardised internationalization ratio) is added. Outcomes can be seen in models 4 and 5 ( app. 3A, 3B, 3C ) for equation with the quadric term and in models 7 and 8 for cubic term respectively (app. 3A, 3B, 3C).
in order to test the moderating effect of national heterogeneity of TMT on possible S -shaped relationship between internationalization and performance (results in app. 3A). This regression was run also to test hypotheses 1.
Third regression was run including quadric term of internationalization again but output was requested for 2 separate groups in respect to national heterogeneity indexes. This was done to test hypotheses 2 (results in app. 3B) and 3 (results in app. 3C). Fourth regression was run including cubic term instead of quadric and output was requested again for two separate groups to test hypotheses 2 (app. 3B) and 3 (app. 3C). .
4.2 Checking for assumption and evaluation of results
4.2.1 Multiple regression assumptions – first and second regression
for first hypotheses
When testing first hypothesis general assumptions of multiple regressions were checked and only after that significance and predictive power of separate variables and models were evaluated.
Checking for assumptions involved at first an evaluation of the correlation matrix (app.4 A ). This correlations matrix shows the correlations between variables of the whole sample, 100 companies. When testing hypotheses two and three, sample was split in two groups and consequently two separate correlation matrixes were generated (app. 4B,4C). The assumptions for hypotheses two and three will be discussed in more details later in this chapter.
From correlation matrix (app.4A) it can be seen that the assumption of multicollinearity is not violated even though there is a significant correlation between independent variables, it is not above 0.7. It is stated that independent variables that correlate with each other with the values of 0.7 and above should not be included into the sample (Pallant, 2005). In this case it is not needed to delete or transform into different ones any of the variables used in this correlation matrix.
coefficient table indicate if problems of multicollinearity exist. Tolerance value is an indicator of how much variability of the specified independent variable is not explained by other variables in the model. The value of tolerance should not be lower than 0.10. VIF (Variance inflation factor) statistics are 1/ Tolerance and should not be above 10.
For testing hypothesis number one, two separate regressions were run. The first one was run including quadric term and second including cubic term of internationalization. In case of the first regression VIF and Tolerance statistics show values of 10.240 and 0.096 respectively for internationalization variable. Values deviate from what they should be in order to meet multicollinearity assumptions. These, too high and too low values, appear after including a moderator variable into regression. Moderator variable is a crossproduct of internationalization and performance. Because crossproduct is not a pure independent variable these deviating values for internationalization variable that appear after including the moderator into regression equation can be ignored.
It is stated that it is up to the researcher which values to use, VIF or Tolerance, however correlation matrix should be evaluated always (Pallant, 2005). In this case correlation matrix does not show any concern even though VIF and Tolerance values are slightly different from what they are assumed to be when evaluating multicollinerity. The absence of violations in the correlation matrix supports the ignoring of abnormal values for internationalization that are caused by including moderator variable.
VIF and Tolerance values for second regression show no concern. To summarize: multicollinearity assumptions concerning regression one and two that were both run in order to test hypothesis one, have not been violated.
Another assumption to be evaluated when running regression is an assumption of outliers. To check for outliers the inspection of Mahalanobis distances was conducted. This indicated one outlier for each regression, regression one and two. In these cases Mahlanbolis distance had values of 25.660 for the first regression and 25.055 for the second regression. Adapting Pallant (2005) critical value for the regression analysis with six independent variables is 22.46. Accordingly, these values of 25.660 and 25.055 are considered to be outliers.
Distance values. These values should be below one. Cooks distance value is 0.106 for the first regression and 0.108 for the second regression consequently depicting no concern. Outlier cases will not be excluded from the sample because they appear only once in each regression, the difference between critical and observed values is not remarkable and Cooks distance statistics do not show any concern.
Assumption of homoscedasticity that indicates that at each level of predictor variance of residual is constant, was tested by plotting the standardized predicted (ZPRED) values of the dependent variables based on the models against the standardized residuals (ZRESID) ,that are the standardized differences between the observed data and the values predicted by the model. These plots look like a random array of dots evenly dispersed around 0 in both sets of models, for the first and second regression test (app 2 figure 1 and 2).
4.2.2 Evaluating results of the regressions one and two run to test
hypothesis one
Looking at the models of appendix 3A that are the result of two first regressions and were run in order to test first hypotheses, it can be observed that control variable, debt to equity shows a significant relationship (p < 0.05) and TMT size a marginal relationship (p<0.1) with the performance variable, ROA. Logarithm of size, control variable, indicates that size of the company has a marginal relationship (p < 0.1 ) in models 2,3,6,7, and 8 (app. 3A) with the dependent variable . Other control variables do not show statistically significant relationship with the dependent variable.
