The effect of board compensation on the degree of
internationalization of listed firms in Singapore
Name Jitteke van ‘t Hek Supervisor Dr. I. Haxhi Student number 10000205 2nd reader Dr. N. Pisani
2
3
Abstract
This thesis focuses on the relationship between compensation and internationalization of listed firms in Singapore. Internationalization is a hot topic but the relationship with compensation is not studied a lot. This study researches the relationship between short-term CEO compensation, TMT compensation and its compensation gap towards Degree of Internationalization (DOI) in Singapore. Results indicate a negative relationship between TMT compensation and DOI as well as a negatively moderating factor of firm performance on the relationship between CEO compensation and the DOI. Contrary to obtained literature no relationship between CEO compensation and internationalization is found. Contrary to obtained literature no relationship between CEO compensation, CEO-TMT compensation gap and internationalization was found. Culture has high impact on compensation and so research cannot easily be generalized, the contribution of this study to literature that it is conducted in Singapore which is not done before. For managers it gives a better understanding of the impact of compensation on decision-making hence internationalization.
5
Table of Content
1. Introduction ... 8
2. Literature review and hypotheses ... 12
2.1 Internationalization... 13
2.1.2 ‘Depth’ and ‘breadth’ of internationalization ... 14
2.2 CEO and TMT compensation ... 17
2.3 CEO and TMT compensation gap ... 21
2.4 Firm performance ... 23 2.5 Hypotheses ... 25 2.6 Conceptual model ... 29 3. Methodology... 30 3.1 Data collection ... 30 3.2 Variables ... 31 3.2.2 Dependent variables ... 31 3.2.3 Independent variables ... 32 3.2.4 Moderating variables ... 33 3.2.5 Control variables ... 33 3.3 Data analysis ... 34 4. Results ... 36 4.1 Descriptive analysis ... 36 4.2 Statistical analysis ... 41 4.2.1 Correlation analysis ... 41 4.2.2 Regression analysis ... 43
6
4.2.2 Multicollinearity check ... 45
5. Discussion ... 48
5.1 Discussion and implications ... 48
5.2 Limitations and Future Research ... 54
6. Conclusion ... 55
7. References ... 58
8. Appendices ... 69
Appendix 1: List of companies... 69
7
List of Figures
Figure 1: Rhythm of internationalization ... 15
Figure 2: Multinationality and performance: A three-phase model... 16
Figure 3: Development of hypothesis 1 ... 26
Figure 4: Conceptual model of hypotheses regarding the Degree of Internationalization ... 29
Figure 5: Conceptual model with the results of the study ... 53
List of Tables Table 1: Regression models ... 35
Table 2: Correlation and Descriptive 2004-2012 ... 38
Table 3: Correlation and Descriptive 2004-2007 ... 39
Table 4: Correlation and Descriptive 2008-2012 ... 40
Table 5: Collinarity statistics DOI 2004-2012... 46
Table 6: Collinarity statistics DOI 2004-2007... 47
Table 7: Collinarity statistics DOI 2008-2012... 48
Table 8: Regression on DOI period 2004-2012 ... 72
Table 9: Regression on DOI period 2004-2007 ... 73
Table 10: Regression on DOI period 2008-2012 ... 74
8
1. Introduction
During the past few decades the world became more globalized. As the world is getting smaller, emphasis has shifted from domestic focus to international focus. From 2000 forward the payment of executives increased immensely (The Economist, 2001) There are many reasons for firms to go abroad, some of which are tax incentives, low labor-costs and the exploration of new markets (Harris et. Al., 1993; Hymer, 1960). As internationalization is such an important topic, lot of research is done; most research is about entry modes (Brothers and Hennart, 2007). Barkema and Vermeulen (2002) have explored the different aspects of internationalization: pace, rhythm and scope, and their effect on firm performance. In the past a lot of studies were aimed at firm performance and internationalization, focusing on where and when to go abroad, and not so much on the pace, rhythm and scope of internationalization (Lin and Cheng, 2013).
Doing business abroad entails complexity (Carpenter and Sanders, 2004; Sanders and Carpenter, 1998) i.e. the different cultural environment, corporate governance, and many other factors (Hofstede, 1980; Prahalad, 1990). When a firm operates in more than one country, its decision-making process becomes more complex (Carpenter and Sanders, 2004; Kim and Mauborgne, 1991). One reason is that the communication within firms becomes more difficult. Chief Executive Officer (CEO) and Top Management Team (TMT) need to make strategic decisions of which to decide whether or not to go abroad. As stated before, the more international a firm becomes, the more complex the information, complicating matters for managers. This is in line with the information-processing theory. With this complex information, knowledgeable managers are an absolute requirement. However, they are hard to come by, because not that many people have the ability for this
9
information processing, these employees are valuable and scarce (Carpenter and Sanders, 2004). Henderson and Fredrickson (1996) link the information processing theory to managerial payment. The CEO and TMT are paid in accordance with the complexity of the information they have to deal with. This is explained in the compensation-complexity theory. The more complex the matter, the more the executive receives in compensation (Henderson and Fredrickson, 1996). However, the payment of the executive may also influence his/her attitude. It may influence his/her decisions and willingness to take risks (Jensen and Murphy, 1990; Devers et al, 2007). Worth mentioning is the fact that Carpenter and Sanders (2002) showed that there are compensation differences between CEO and TMT. Most researched is confined at CEO level but a firm’s strategies are not solely shaped by the CEO but the entire TMT shapes the strategy (Hambrick and Mason, 1984; Hambrick et. Al, 1996; Kauer et al, 2007)
The research field of internationalization has mainly been aimed at the relation between firm performance and entry mode (Werner, 2002). Besides, the link between executive payment and firm performance (Murphy, 1999) is studied extensively, which is not the case as regards the relationship between compensation and decision-making (Lin and Cheng, 2013). Their study discusses Taiwanese companies and their internationalization process. A limitation of this research is that it was only carried out in Taiwan and not in a cross-cultural context. Culture influences the decision making process (Kim and Mauborgne, 1991) and it may differ between cultures (Werner, 2002). People’s perception of fair compensation depends on national cultures (Gelfand et al, 2002; Chiang and Birtch, 2006). Lin and Cheng (2013) intend to do further research in other newly industrialized countries such as Singapore. Singapore is a country that is highly involved in
10
international trade, searching for cross-border mergers and acquisitions, and has rapidly growing outflow of Foreign Direct Investment (FDI). In 2010 Singapore was in the top-20 of FDI inflow and outflow (UNCTAD, 2010, 2011) which makes it an interesting country to study.
Payment of executives, decision-making process and internationalization are linked (Henderson and Fredrickson, 1996; Sanders and Carpenter, 1998). To understand the internationalization process of firms, the payment of executives must be understood which the aim of this thesis is: to understand the relation between executive payment and the internationalization process of firms.
This thesis aims to investigate the relationship between board compensation and internationalization, taken the 2008 crisis into account as well. At Q4 of 2007 the economy of Singapore dropped heavily. Before the drop the economic conditions were to the roof where after the crisis firm’s had less financial stability to make a move abroad (Mah-hui & Maru, 2010). The thesis is not solely about CEO compensation but TMT compensation and the compensation gap between CEO and TMT are taken into account. Strategic decision making is not solely the responsibility of the CEO but the whole TMT is responsible. Tournament theory and fairness theory (Henderson and Fredrickson, 2001; Lin and Cheng, 2013) show that compensation gap between CEO and TMT can influence decision-making, risk taking and goals. Last the influence of firm performance on the relationship is researched. High performance is related to good strategic decision making (Carpenter and Sanders, 2004). Lin and Cheng (2013) state that when there is high firm performance a CEO is more likely to stick to the internationalization strategy instead of changing it. Also they state that high performance influences the incentive to go abroad.
11
The core argument underlying the hypotheses is that compensation is an important factor in strategic decision making regarding goals, alignments and risk taking (Devers et al., 2007). Together with the information-processing theory, in order to understand the internationalization of firms one must understand the effects of compensation (Carpenter and Sanders, 1998; Carpenter and Sanders, 2004).
The research gap found is threefold. First the linkage between compensation and internationalization has not been research a lot, second compensation and theories that are linked to compensation are cultural dependent. Generalization of those studies is therefore not possible and research in multiple cultures is needed to get a clear understanding of the relationship and third if compensation is researched it is done solely on CEO compensation and not TMT compensation as well. Therefore, the research question of this thesis is: “What is the influence of board compensation on the internationalization process of public firms in Singapore?” The question will not only focus on CEO and TMT compensation but also on the CEO-TMT compensation gap.
This study will have multiple contributions. First, theoretically, considering the limited numbers of studies on the effect of TMT compensation on the internationalization, our research reiterates the context-depend aspects of decision-making process and compensation. More particularly, this relationship is dependent on the cultural context, indicating the low level of generalizability in previous literature. Second, practically, companies have been criticized for their compensation arrangements. In order to reciprocate, it is important to know why the compensation is fairly high, and what the (positive) consequences of this compensation are on the globalization of firms. This research may help managers to come to grips with the influence of compensation on their
12
decision making process in terms of internationalization. Becoming an international player requires a sound knowledge of this process and its influence on internationalization to be able to face competition.
The remainder of this thesis is structured as follows: chapter one will provide an in depth literature review on internationalization, CEO and TMT compensation and the compensation gap, chapter two addresses the methodology of the research followed by the results in chapter three. The final sections contain discussion, limitations and the conclusion of the thesis.
2. Literature review and hypotheses
This section offers the literature about the most important topics concerning the research. Section one offers literature about internationalization, after which, in section two CEO and TMT compensation is discussed. Section three provides literature on the compensation gap between CEO en TMT compensation and section four offers literature on firm performance. Section five provides the research gap and hypotheses development.
The various sections pay attention to the link between the different subjects. There are many theories about the link between performance and internationalization. The complexity-information (Carpenter and Sanders, 2004; Sanders and Carpenter, 1998) theory and the information-processing theory (Carpenter, Sanders and Gregersen, 2001; Carpenter and Sanders, 2004; Henderson and Fredrickson, 1996) are highlighted as well.
13
2.1 Internationalization
Internationalization is one of the most important business phenomena of the last decades (Sapienza et al., 2006). There are many reasons for a company to take part in Foreign Direct Investment (FDI). Internationalization can lead to monopolistic advantages (Hymer, 1960). Other advantages are tax benefits, due to a lower tax burden. A multinational can spread its income over the companies to reduce its total tax burden (Harris et. Al, 1993), at the same time enhancing operational flexibility and stability, volume economies, intelligence gathering, product improvement, and organizational advantages (Mitchell et. Al., 1993). Another form of internationalization is moving operations to low-cost countries, which leads to cost reductions (Hennart, 1982). Many studies have been done about the relationship between internationalization and firm performance (Lu and Beamish, 2004; Barkema and Vermeulen, 2002).
According to Stopford (1992), a firm is considered a multinational if it has production in at least three foreign countries. The process of internationalization is being researched extensively. Important models are the Uppsala model (Johanson and Vahlne, 1977), which explains how firms gradually expand to foreign countries. Other theories are the eclectic paradigm (Dunning, 2001), the transaction-cost theory (Hennart, 1982) and the internalization theory (Buckley and Casson, 1976). Vermeulen and Barkema (2002) distinguish three elements of internationalization: the pace, scope and rhythm of the internationalization process. In the next section the theory of Vermeulen and Barkema (2002) about the ‘scope’ and ‘breadth’ of internationalization will be discussed.
14
2.1.2 ‘Depth’ and ‘breadth’ of internationalization
In their paper they discuss the link between these elements and the performance of firms. Performance is measured as Return of asset (ROA). The pace of internationalization is described as the speed of international expansion. So this is a time-based measurement indicating how long it takes before a certain level of performance is achieved (Jones and Covillo, 2005).
Barkema and Vermeulen (2002) state that if the speed of expansion is too high, it moderates the relationship with performance in a negative way. The scope of internationalization is the dispersion of the internationalization process across different geographic or product markets (Vermeulen and Barkema, 2002). The rhythm of internationalization is described as the regularity of the expansion pattern. This is measured by the kurtosis. In figure 1 you see the different rhythms that can occur in the expansion speed of a company. The two diagrams on the left show a regular pattern with each time the same step in expansion. This shows a regular rhythm and is associated with performance and profitability. Vermeulen and Barkema (2002) state that a regular pace indicates that you can cope with the new information and you do not get an overload, which is better for the performance of a company. Wang et al (2012) point out that the benefits of overspills, knowledge for example, in regular foreign expansion is higher than in irregular expansion. Thus, regular expansion is better than irregular expansion. The degree of multi-nationality is described as the depth (mentioned above as pace) and the breadth (mentioned above as scope) of internationalization (Kundu and Hsu, 2003; Lin and Cheng, 2013). Carpenter and Sanders (2004) made this distinction as well. To understand
15
internationalization they use these dimensions to understand the Degree of Internationalization (DOI) of a company.
Figure 1: Rhythm of internationalization
Source: Barkema and Vermeulen (2002)
Internationalization of a firm is a complex phenomenon. Barkema and Vermeulen (2002) state that when a firm expands abroad, the firm has to learn to cope with new and/or unknown cultural and institutional factors. The capacity of a firm to expand and absorb is limited, learning how to operate in a foreign setting cannot be compressed by time indefinitely. They conclude the depth and breadth of internationalization influences performance, ROA. Firms can really profit from doing business abroad (Hennart, 1982; Harris et. Al, 1993; Hymer, 1960; Mitchell et. Al, 1993) but dispersion of the expansion process into foreign countries and the irregular kurtosis negatively moderate these profits (Barkema and Vermeulen, 2002). Lu and Beamish (2004) identify three phases in the relationship between internationalization and firm performance. In phase 1 the firm has to cope with liability of foreignness (LoF) and newness (LoN). The firm can overcome the LoF and LoN by learning how to operate in a foreign country. This learning process involves
16
costs and contributes to a negative ROA (performance). The second phase involves increasing scope which is associated with higher returns. Due to increasing scope the firm gains knowledge through experiential learning, decreasing the costs associated with LoF and LoN. However, this growing network of subsidiaries and the increased complexity of the firm may inflate the costs of coordination and governance. These costs can rise so high that they may surpass the benefits of diversification and scope, leading to diminished performance, marked as phase three. The different phases are shown in figure 2. Lu and Beamish (2004) identify an S-shaped curve between internationalization and firm performance.
Figure 2: Multinationality and performance: A three-phase model
The company has to deal with different and unknown cultures, governance and other institutional factors (Barkema and Vermeulen, 2002; Lu and Beamish, 2004). The more global a firm is, the more complex strategic decisions become (Kim and Mauborgne, 1991). Traditionally, the compensation research focuses most on the CEO. The theory of CEO compensation is than mostly generalized for the Top Management Team (TMT) payment
17
arrangements (Carpenter and Sanders, 2004). But research shows that there are payment differences between CEO and TMT (Carpenter and Sanders, 2002; Henderson and Fredrickson, 2001). Carpenter and Sanders (2004) argue that there is a theoretical and practical need for the recognition of the importance of TMT compensation in the executive-compensation research. The tasks and responsibilities differ between them so they receive different compensation (Henderson and Fredrickson, 2001).
Henderson and Fredrickson (1996) link the information-processing theory and managerial payment. The information-processing theory is a model that helps describe and explain information processes in human beings. The CEO is paid according to the level and complexity of the information they have to process. As the task of decision making also lies with the TMT, they also receive a higher compensation when a firm has a more diversified strategy (Henderson and Fredrickson, 1996). The next section goes into the relationship between CEO-TMT compensation and the internationalization process.
2.2 CEO and TMT compensation
During the past decades numerous studies have been done about CEO and TMT compensation (Carpenter and Sanders, 2004; Goergen and Renneboog, 2011; Lin, Liu and Cheng, 2011; Lin and Cheng, 2013). Compensation is seen as the total amount of compensation combined (stock options, salary, etc.) that an executive receives in contingent, long-term forms (Carpenter and Sanders, 2002). Lin and Cheng (2013) describe the relationship between compensation and its internationalization process. Most research is done about CEO and TMT compensation in the relationship between compensation and performance (Carpenter and Sanders, 2002; 2004; Drucker, 1984;
18
Hartzell and Starks, 2003; Jensen and Murphy, 1995; Mehran, 1995 and Siegel and Hambrick 2005), the relationship between TMT characteristics (age, education, tenure etc.) and strategic decision making (Hambrick, Cho and Chen, 1996; Hermann and Datta, 2005). Also the relationship between compensation and risk management is thoroughly discussed (Eisenhardt, 1989). CEO payment is influenced by several factors, these include risk factors, human capital, power, skill and managerial discretion (Hengartner, 2007).The structure and characteristics of top management can influence the firm’s strategic decisions such as internationalization (Tushman and Nadler, 1978). Managerial discretion which is the latitude of action in making strategic choices is one of the factors influencing CEO compensation. When there is high discretion the risk taking is higher and thus the CEO compensation (Haleblian and Finkelstein, 1993). Other factors to consider are human capital. When managing an international firm requires skills that go beyond purely managing a domestic firm (Dailey at al, 2000). Not only is the CEO responsible for the strategic decisions, the same goes for the entire TMT (Hambrick and Mason, 1984). Devers et. Al. (2007) believe that TMT compensation is an important factor regarding the decision making process of strategic choices. The CEO is seen has significant impact on the strategic decision making process within the firm (Bliss & Rosen, 2001; Datta et al, 2001; Sanders, 2001). According to Sanders and Carpenter (1998) the degree of internationalization has significant effect on its choice of corporate governance arrangements and the degree of internationalization is positively associated with CEO payment. Consequently, the higher the internationalization level, the higher the CEO payment. Behavioral theory shed lights on the information-processing theory (Cyert and March, 1963). Henderson and Fredrickson (1996) link the processing theory and managerial payment. The
information-19
processing theory is a model that helps describe and explain information processes in human beings. According to Murphy (1999) payment levers and structures are changing which reflects the increasing (global) market for managerial talent. This is important because management of international firms need human capital that can process the information (Hengartner, 2007).
The CEO is paid according to the level and complexity of the information he or she has to process. Complexity is explained as a matter that is difficult to understand or explain because of many different parts and components of the task. This complexity can come from structure, multicultural workforce etc. (Rosen, 1982). When managers need to work in a more complex environment the compensation is higher as well (Rosen, 1982). Executive talents receive higher managerial return when the complexity is higher and the compensation will be higher as well (Hengartner, 2007).
As the task of decision making also lies with the TMT, they also received a higher compensation when a firm had a more diversified strategy (Henderson and Fredrickson, 1996). According to Oxelheim and Randoy (2003) CEO compensation does not only reflect the work they does but reflects also on risk as well as a risk premium for losing the job.
The link between compensation and internationalization can been seen in various theories, most important of which are the complexity-compensation theory (Henderson and Fredrickson, 1996; Carpenter and Sanders, 2004) and the information-processing theory (Egelhoff, 1982; Carpenter and Sanders, 1998; Weick and Van Orden, 1990). First, it is important to understand the complexity that arises when a firms goes international. Secondly, the link between complexity and compensation as well as the information-processing theory need to be understood.
20
The complexity of internationalization has also been discussed in previous sections. A purely domestic firm has a less complex decision making process than a multinational firm (Carpenter and Sanders, 2004; Sanders and Carpenter, 1998). When a firm operates in different countries there are different cultures, governance, competitors etc. with which they have to cope (Prahalad, 1990; Hofstede, 1980). When the environment becomes more complex it also becomes more difficult to process this information and understand it (Black and Farias, 1997).
The information-processing theory needs to be understood as well. To cope with this complex environment, managers who can handle this information are a requirement. In a multinational firm these needs are higher than in purely domestic firms (Egelhoff, 182; Carpenter and Sanders, 1998). Therefore, you need a high level of managerial human capital in the right places (Carpenter and Sanders, 2004). These managers are rare, valuable, and therefore associated with higher compensation (Carpenter, Sanders and Gregersen, 2001; Henderson and Fredrickson, 1996). If a lower level executive held a strategically sensitive position, they would earn more than a lower level executive who did not have that position (Carpenter and Wade, 2002). So the payment an executive receives is influenced by the position in the firm as well as the human capabilities of the person. However, the payment of executives can also influence the executive. It can influence his behavior as regards risk-taking (Jensen and Murphy, 1990). Group dynamics and behavioral factors determine the level of information process and the strategic decision making associated with it (Haleblian and Finkelstein, 1993). As stated before, firms work in a globalized environment and there is an increased global competition. To outperform competitors you need to attract suitable employees, who can cope with the complex
21
information. Appropriate compensation systems can attract and retain those outstanding managers into the organization and can have a significant effect on the strategic goals of the company (Philips and Fox, 2003).
2.3 CEO and TMT compensation gap
In most research CEO and TMT are treated in the same way. According to Carpenter and Wade (2002) this is a misconception. CEO and TMT possess different characteristics and different roles in the firm which comes with responsibilities (Henderson and Fredrickson, 2011). Also Lin and Cheng (2013) discovered a compensation gap. The CEO presents unique characteristics within the firm. The job context and characteristics of the CEO also greatly differ from other TMT executives (Henderson and Fredrickson, 2001; Jaw and Lin, 2009). Jaw and Lin (2009) conclude that in research there should be a division between CEO and TMT (Carpenter and Wade, 2002; Jaw and Lin, 2009). The reason that there should be a shift is that currently most research on CEO payment is generalized to TMT payment. But as seen in literature there are great differences between CEO and TMT within a company, which probably also affects its payment levels. According to Philips and Fox (2003) internationalization makes the job of the CEO even more difficult to handle, due to the complex information-processing, and you need more top quality managers to be able to make these complex decisions, which is why they should be compensated accordingly. The pattern of the internationalization process influences the information-processing of TMT (Barkema and Vermeulen, 2002). The irregular the pattern of internationalization is, the more complex the information and the decision making of CEOs and TMTs. This results in higher payment of CEO and TMT (Lin and Cheng, 2013). There are two theories about
22
the CEO-TMT compensation gap that contradict each other. Those are the fairness theory and the tournament theory (Henderson and Fredrickson, 2001; Lin and Cheng, 2013). The fairness theory argues that large gaps in compensation results in a feeling of unfairness for the TMT and psychological distress (Adams, 1963; Lin and Cheng, 2013). Henderson and Fredrickson (2001) state that large compensation gaps between CEO and TMT may promote unproductive behavior from TMT toward the CEO. This can result that the CEO does not have the proper information to make the correct decision which can influence the decisions on internationalization as well as the firm performance. Research is done regarding the fairness theory and the psychological distress that can result from unfair compensation. Bloom and Michel (2002) found that a large CEO TMT compensation gap leads to higher managerial turnover and Siegel and Hambrick (2005) found a relationship between dispersion and poorer performance.
On the other hand there is the tournament theory. This theory is stated within the Agency theory. Agency theory assumes that individuals mainly interested in pursuing self-interest, it explains the principal-agent relationship (Jensen & Meckling, 1976). Tournament theory states that that due to payment differences one will work hard to promote themselves with higher compensation (Lazear and Rosen, 1982). This theory states that large gaps will motivate members of TMT to put in more effort solely aimed at winning, which comes with a payment that is the same as the ‘winner’ or in this case the CEO (Lambert, Larcker and Weigelt, 1993; Main, O’Reilly and Wade, 1993). This theory suggests that one is willing to do more in order to win. The theory is highly supported within literature (Main et al, 1993; Eriksson, 1991 and Conyon et al, 2001). According to Henderson and Fredrickson (2001) this dispersion in payment leads to greater
23
performance and make the pattern of internationalization more regular. Thus, it influences the degree of internationalization. These two theories are opposites and suggest that a gap is only beneficial when it motivates the TMT members (Lin and Cheng, 2013).
The gap between CEO and TMT must reach a certain level before it will motivate TMT to work more effectively. Conyon et al. (2001) state that if there are only small prices to win, TMT slacks off to win more prices in the future, so the gap should not be too small. Only when the gap is too large TMT members feel unfairness and work less effectively and oversea information is preceded the less.
After the start of the crisis in 2008 there was a sharp drop, globally, in trade and production. For countries as Singapore, which is an export-oriented market, it hit the economy hard. Although advanced markets (7,5% decline) were hit more than emerging markets (4% decline), the crisis had a deep impact on the economy of Singapore (Sharma, 2013). So the crisis had major impact on FDI and wages in Singapore were cut because of extensive pressure on them. Both are important variables for the research it is decided to investigate the research also pre- and post -crisis.
2.4 Firm performance
As seen in the previous sections it is clear that the decision making process within the firm is a difficult concept. In the end the decisions that CEO and TMT make are meant to increase firm performance, ROA (Barkema and Vermeulen, 2002). High performance is associated with appropriate strategy and good decision making (Carpenter and Sanders, 2004). When the firm is performing well, the TMT and CEO are usually eager to maintain the current strategy as well as the internationalization strategy. On the other hand, when a
24
firm is not performing well, one of the key changes a firm should make is adopt a different strategy and combined with that there is a fair chance that internationalization strategy is changed (Lin and Cheng, 2013). CEO and TMT compensation combined with behavioral factors are an important indication of how well the decision making is going and what the performance of the firm will be (Haleblian and Finkelstein, 1993). When the company is in a flow of high firm performance, the CEO is likely to get a higher compensation and is more likely to stick with their initial foreign expansion strategy. Following this statement the stimulating effect of compensation is greater under low firm performance (Lin and Cheng, 2013). High performance may strengthen the incentive for CEO and TMT to go abroad.
As mentioned earlier there is a compensation gap between CEO and TMT. Here the tournament theory is linked. That states that when there is a compensation gap, TMT become highly to perform well than when this gap is not there. In line with this theory, TMT will be more eager to understand and develop an internationalization pattern that is considered to be better for the firm performance, according to Barkema and Vermeulen (2002).
Zhang et al (2008) state that low performance put TMT in a bad and weak position and therefore their wealth is decreasing. Lant et al (1992) state that when a firm is not performing well, TMT of that firm is less likely to go abroad because poor performance may cause psychological distress. When a firm is performing poorly, the CEO will change its strategy and internationalization pattern. On the other hand when a firm is performing well the CEO is more likely to stay with the same strategy.
25
2.5 Hypotheses
This section offers the hypotheses development in order to answer the research question of the thesis, what is the influence of board compensation on the internationalization process of listed firms in Singapore?
The first hypothesis examines the effect of CEO compensation on the degree of internationalization. As mentioned in previous sections CEO compensation is influenced by many factors such as risk, skill, power, human capital and managerial discretion (Hengartner, 2007). CEO compensation is significantly related to strategic decision making and one the decisions made is firm’s internationalization strategy (Sanders, 2001; Datta et al, 2001; Bliss and Rosen, 2001). Decision-making process has links to two important theories, information- processing theory and complexity-compensation theory. The information –processing theory states that that international environment it is more difficult to cope with information and employees that can cope with this information are scarce and valuable (Sanders and Carpenter, 1998, Sanders and Carpenter, 2004). To hire these employees compensation needs to be up to it (Carpenter, Sanders and Gregersen, 2001; Henderson and Fredrickson, 1996). The complexity-compensation theory explains that when a task is more complex the compensation should be higher (Rose, 1982). International activity makes the environment to work in complex and hence compensation to deal with it should be higher.
So compensation influences the decision making process and hence internationalization of the firm. The relation is made clearer in figure 3.
26
Figure 3: Development of hypothesis 1
Combining the arguments of information-processing theory and the complexity-compensation theory it is expected that high level of CEO complexity-compensation will enhance internationalization compared to a situation with lower CEO compensation arrangements.
Hypothesis 1: There is a positive relationship between CEO compensation and internationalization.
The second hypothesis examines the effect of TMT compensation on internationalization. Most literature focuses solely on CEOs when researching compensation (Carpenter and Sanders, 2004; Goergen and Renneboog, 2011; Lin, Liu and Cheng, 2011; Lin and Cheng, 2013). But not only CEOs are responsible for the decision making in the firm, the entire TMT is responsible (Hambrick and Mason, 1984; Hambrick
27
et. Al., 1996; Kauer et al, 2007). So that is why the second hypothesis focuses on the influence of TMT compensation on internationalization.
Here information-processing and the complexity-compensation theory are linked to TMT compensation and together with the fact that strategy is shaped by the entire TMT it is expected that higher TMT compensation will enhance internationalization than when there are lower compensation arrangements for TMT.
Hypothesis 2: There is a positive relationship between TMT compensation and internationalization
The third hypothesis focuses on the compensation gap between CEO and TMT. Most research focus solely on CEO but strategic decision making lies with the whole TMT (Herngartner, 2007). Furthermore the payment differences between CEO and TMT are the highest among all hierarchical levels (Rosen, 1986; Leonard, 1990). Two theories are important, fairness theory and tournament theory. On one hand fairness theory states that TMT members can preserve stress when payment differences are too high (Siegel and Hambrick, 2005) and tournament theory states that compensation gap leads to motivation to make better decision in order to win and make sure they get the same rewards as CEO (Lazear and Rosen, 1981). Henderson and Frederickson (2001) state that when payment differences are too high the tournament theory is not applicable and it will be counterproductive. Together these theories conclude that when there is a small gap in compensation TMT will be motivated to make better decisions and counter wise when the
28
gap is too large it will backfire and it will negatively influence the decision making hence internationalization.
This concludes in that the gap between CEO compensation and TMT compensation will negatively influence the relationship with internationalization. Larger gaps are associated with lower internationalization and smaller gaps are associated with higher internationalization.
Hypothesis 3: There is a negative relationship between the CEO-TMT compensation gap and internationalization. Larger gaps will be associated with lower internationalization whilst smaller gaps will be associated with higher internationalization.
The fourth and last hypothesis focuses on firm performance as moderating factor on the relationship between CEO compensation and internationalization. Firm performance may influence the decision making process because performance indicates whether a firm should stick to the current path or make different decisions (Markides, 1995). When a firm is performing well it has more assets and is more likely to go abroad than when the firm is not performing well (Lant et al, 1992). Together the expectation is that firm performance will negatively moderate the relationship between CEO compensation and internationalization.
Hypothesis 4: There is a negative influence of firm performance on the relationship between CEO compensation and internationalization.
29
2.6 Conceptual model
Previous section summarizes the core arguments and hypotheses of the study. In figure 4 the conceptual model of the hypotheses is constructed. The dependent variable is DOI with the independent variables of CEO compensation, TMT compensation and the CEO-TMT compensation gap. The following hypotheses are shown in the model:
H1: There is a positive relationship between CEO compensation and internationalization. H2: There is a positive relationship between TMT compensation and internationalization H3: There is a negative relationship between the CEO-TMT compensation gap and internationalization. Larger gaps will be associated with lower internationalization whilst smaller gaps will be associated with higher internationalization.
H4: There is a negative influence of firm performance on the relationship between CEO compensation and internationalization.
30
3. Methodology
This section will describe the methods used to test the different hypothesis. The first section describes the sample and data collection. The second section describes the different variables, starting with the control variables and ending with control variables and moderating variable. The third section explains the regression analysis used.
3.1 Data collection
The data sample for this research contains information on 104 listed firms in Singapore from 2004 till 2012, a list of all the firms incorporated in this research can be found in appendix one. One condition was their multinational character. Stopford's (1992) definition of a multinational firm is that a firm is a multinational when its production takes place in at least three foreign countries. All the firms are listed on the Singapore Exchange and have a subsidiary in at least three or more foreign countries.
The data was obtained in different stages. First data Orbis was collected where the data of 452 companies registered in Singapore were obtained provided that there was sufficient financial information of the company in the years 2004-2012. The second step was to determine the number of multinationals in the dataset, according to the definition of Stopford (1992). Other information from Orbis including Return on Assets (ROA), number of subsidiaries, number of countries present, number of years since incorporation and type of industry. The third step was to gather information from DataStream, where the following variables were found: foreign assets, total assets, foreign sales and total sales. For 33 companies this data was not available, this is why the firms were excluded from the dataset. The last variables needed for this study were CEO compensation and TMT
31
compensation. Relevant information was obtained from all annual reports of the companies between the years 2004-2012. Some companies did not describe the compensation variable clearly enough and were consequently excluded from the study. This resulted in a final number of 104 firms with a total of 931 observations were included in the dataset; in appendix 1 the full list of companies can be found. The study prior to the economic crisis had 415 observations and prior to the crisis 517 observations. This difference is due to the interval of the measured data of the two periods. (Respectively four years and five years).
3.2 Variables
3.2.2 Dependent variables
The dependent variable used in investigating the hypothesis is internationalization. Internationalization cannot be captured using one dimension; there are many factors to reconsider when studying this matter. In line with Carpenter and Sanders (2004) the distinction between ‘depth’ and ‘breadth’ of internationalization was made. To capture both dimensions Sullivan’s (1994) measurement of internationalization was used, the Degree of Internationalization. To look at more than one different variable the measurement of internationalization is more secure (Sullivan, 1994). There are many factors that one can reconsider using. Sullivan (1994) concludes that, even if a relationship is established, that there still is chance that the measurement is not correct. To be as correct as possible the following dimensions are used to determine the Degree of Internationalization (Lin, et al, 2011)
32
The first dimension is the ratio of foreign sales to total sales. Second dimension is the foreign production. This is measured by foreign assets as a percentage of total assets. The third dimension is related to scope, in this case geographical dispersion which is measured by the number of countries in which a firm has a subsidiary, divided by the highest number of countries represented in the sample. The first two measurements represent the ‘depth’ of internationalization and the third dimension captures the ‘breadth’. The sum of these dimensions is the measure for the firm’s degree of internationalization (DOI).
3.2.3 Independent variables
Three different measurements for compensation were taken CEO compensation, TMT compensation and the CEO-TMT compensation gap.
Lin and Cheng (2013) used as CEO compensation the amount of US$ paid to the CEO in a calendar year. This includes all elements of payment such as salary, bonuses etc. the total CEO compensation was arrived by summing all identifiable elements in the CEO pay. Long-term incentives were not taken into account.
TMT compensation is measured following Carpenter and Sanders (2004) the compensation per year is measured as the natural logarithm of the average of all compensation (salary, bonuses etc.) in a calendar year that the top four non-CEO executives receive. The compensation was arrived by summing all identifiable elements in the top four non CEO pay divided by four.
The CEO-TMT compensation is the third measurement of compensation. The measurement of this gap is done by the difference of total compensation (salary, bonuses etc.) in a calendar year of the CEO, compared with the average compensation of the top four
33
non-CEO members (Carpenter and Sanders, 2004; Henderson and Fredrickson, 2001; Lin and Cheng, 2013). *Note that long-term incentives are not taken into account.
3.2.4 Moderating variables
The moderating variable considered in this study is firm performance. The push–and-pull factors of a company to go abroad are related to the decision making process within the firm, hence related to the compensation of TMT and CEO. (Carpenter & Sanders, 2002). Return on Asset (ROA) is used to measure firm performance, which is a commonly used variable to measure firm performance (Barkema and Vermeulen, 2002; Lin and Cheng, 2013)
3.2.5 Control variables
A number of control variables have been used in this study. In this section the three control variables used in the study are explained.
Firm size is known to be related to the decision-making process, and complexity of the information processing in a firm, as well as the compensation structure within a firm (Carpenter and Sanders, 2004; Henderson and Fredrickson, 1996; Lin and Cheng, 2013). Firm size was measured as the logarithm of the assets in a given year (Lin and Cheng, 2013).
Firm age is the second control variable used in the study, firm age is well known to
positively relate to broad international activity. Yip et al (2000) conclude that older firms have more international experience than younger firms. Firm age is measured by the number of years since incorporation.
34
Third control variable is industry type. The type of industry in which a firm is active can have influence on compensation schemes (Bhohraj et al. 2003) the type of industry was obtained through Orbis; in the study industry type was used as dummy variable. A distinction was made between heavy industry (chemicals, plastic, rubbers etc.), high-technology industries (machinery, electronics, and computers) and light industries (food, textiles, tourism) (Lin and Cheng (2013).
3.3 Data analysis
To test the validity of the hypotheses regression analysis is used. Regression analysis determines the relationship among variables in a research model. It explains how the dependent variable will change when one or more independent variables change. This method then plots an optimally fitting line through a set of points which represents observations. The equation for the regression method is:
Y = α + ß1 * X1 + ß2 * X2 + ßi * Xi + ε
In this equation Y is the dependent variable; Degree of Internationalization, X is the dependent variable; CEO compensation, TMT compensation and the CEO-TMT compensation gap. The α is the constant where the best fitting line intersects with the Y-axis and ε is the residual error term. This gives for this research the following equation: Model 1
YDegree of internationalization = α + ß* XCEO compensation + ß* XTMT compensation + ß* XTMT compensation + ß*
35
The analysis of the hypothesized relationships gave away to construction of seven regression models. The first model includes only the control variables. The independent variables were individually included. Hence, the last model consisted of all relevant variables. Table 1 gives an overview of all seven regression models. Because there in the timeframe of my research data (2004-2012) there was an economic crisis the decision was made to not only conduct information about the whole timeframe but compare this to the data before the crisis (2004-2007) and after the crisis (2008-2012). The reason is that while the data was conducted there is a huge difference in the data of compensation. In the period before the crisis the data was not constructed very clearly and most companies did not report all compensation, mostly they’d reported that compensation was $500.000 and above whilst after the crises the amount of compensation was adequately reported in the annual reports.
Table 1: Regression models
Model # Control Variables Independent Variables
Firm
size Industry Type Firm age CEO compensation TMT Compensation CEO- TMT compensation gap ROA*CEO compensation Model 1 X X X Model 2 X X X X Model 3 X X X X Model 4 X X X X Model 5 X X X X X X Model 6 X X X X X Model 7 X X X X X X X
For the analysis it was necessary to rule out multicollinearity by checking for Tolerance and Variance Inflation Factor (VIF) scores.
36
4. Results
This section describes the results obtained from the different regression analyses (model 1-7). Section 1 provides a descriptive analysis of the different variables, section 2 offers a correlation analysis, section 3 the regression analyses are explained and in section 4 check for tolerance and VIF can be found.
4.1 Descriptive analysis
Table 2 provides the descriptive statistics and correlations on all variables taken into account the different models for the entire period. The total sample includes 931 observations over nine years (2004-2012). In the period before the crises the sample includes 415 observations and the period after the crisis 517 observations. The smaller sample size of the first period is due to the obligation of reporting financial data.
As seen in table 3 the average age of a firm in the sample was almost 26 years and the average firm size was 2,5. Industry, in this study, was diversified in three categories, light, heavy and high-tech industries. In this sample firms in the light industry were highly represented, accounting for 58% of the firms in the sample. Other firms were fairly evenly distributed over the sample between heavy and high-tech industries (15% and 16%, respectively). The average return on asset for the sample was almost 11.
Average compensation earned by a CEO in this sample was €1.210.966 with a standard deviation of €1.841.114 for TMT (logarithm) the average compensation in this sample was 84,8. Furthermore the average compensation gap between CEO and TMT in this sample was € 614.886.
37
Table 3 and 4 provide descriptive statistics and correlations on all variables for period 1 (2004-2007) and period 2 (2008-2012) separately to provide extra information about the influence of the crisis on the study. The average firm size was in period 1 somewhat smaller and period 2 a bit larger (2,38 and 2,61, respectively) compared to the overall sample. The return on asset before the crisis is extensively higher than in the period after the crisis. Before the crisis the return on assets is 19,87 and down to 3,34 after the crisis. This difference can be explained by the difficulties firms were facing due to the crisis.
Strangely the average CEO compensation was in the period after the crisis higher than in the overall data sample (respectively over €1.3 million and over €1,2 million). The period before the crisis the average CEO compensation was lower in comparison with the overall sample. The TMT compensation in the period before the crisis was substantially higher than after the crisis (respectively 153 and 29,34). The compensation gap between CEO and TMT was higher after the crisis than before the crisis (respectively €515.998 and €693.997).
The last variable described in the sample is the Degree of Internationalizing. The average of the overall sample is 87,4. In period 1 this was higher and in period 2 this was lower than the overall sample (respectively 88,93 and 86,19).
Table 2: Correlation and Descriptive 2004-2012
Mean Std. Dev. 1 2 3 4 5 6 7 8 9 10 11
1. Age of the firm 26,56 18,914 1
2. Logarithm of assets 2,5104 ,81741 .270** 1 3. Light industry ,58 ,494 .154** -.121** 1 4. Heavy industry ,15 ,361 ,022 .086** -.498** 1 5. Tech industry ,16 ,370 -.219** -.081* -.516** -.188** 1 6. CEO compensation 1210966 1841114 .115** .441** -.097** -,007 ,048 1 7. TMT compensation 84,8041 253,98058 -.076* -,045 -.073* ,031 ,034 -.087** 1 8. CEO-TMT compensation gap 614886,66 1768626,110 .110** .241** -,057 ,005 ,054 .834** -,058 1 9. ROA*CEO compensation 7126897,24 17734106,133 ,032 .289** -.066* ,049 -,007 .761** -,056 .671 * * 1 10. ROA 10,6961 80,44080 -,042 -,053 ,013 ,018 -,014 -,016 -,020 -,003 ,059 1 11. DOI 87,4067 57,80851 -.184** -.069* -.109** ,039 .068* -,057 -.074* -,053 -.075* -.078* 1
**Correlation is significant at the 0.01 level (2-tailed). *Correlation is significant at the 0.05 level (2-tailed).
39
Table 3: Correlation and Descriptive 2004-2007
Mean Std. Dev. 1 2 3 4 5 6 7 8 9 10 11
1. Age of the firm 24,00 18,618 1
2. Logarithm of assets 2,3820 ,80146 ,272** 1 3. Light industry ,58 ,495 ,159** -,102* 1 4. Heavy industry ,15 ,361 ,018 ,069 ,498**- 1 5. Tech industry ,16 ,370 -,225** -,081 -,516** -,188** 1 6. CEO compensation 1.012.566 1.687.405 ,112* ,411** -,051 -,052 -,003 1 7. TMT compensation 153,8790 343,98100 -,065 ,010 ,126**- ,056 ,063 -,092 1
8. CEO-TMT compensation gap 515.998 1.539.412 ,094 ,248** -,011 -,031 -,025 ,888** -,064 1
9. ROA*CEO compensation 7273891,00 17294001,000 -,006 ,315** -,080 ,037 -,025 ,830** -,081 ,759** 1 10. ROA 19,8700 119,294 -,050 -,088 ,018 ,018 -,009 -,025 -,049 -,011 ,005 1 11. DOI 88,9310 56,14700 -,137** -,033 -,060 ,059 -,048 ,010 -,154** -,002 -,040 ,113-* 1
**Correlation is significant at the 0.01 level (2-tailed). *Correlation is significant at the 0.05 level (2-tailed).
40
Table 4: Correlation and Descriptive 2008-2012
Mean Std. Dev. 1 2 3 4 5 6 7 8 9 10 11
1. Age of the firm 28,63 18,915 1
2. Logarithm of assets 2,6130 ,81638 ,247** 1 3. Light industry ,58 ,495 ,151** -,138** 1 4. Heavy industry ,15 ,361 ,025 ,100* -,498** 1 5. Tech industry ,16 ,370 -,218** -,083 -,516** -,188** 1 6. CEO compensation 1.369.865 1.942.453 ,101* ,450** -,130** ,024 ,083 1 7. TMT compensation 29,5440 121,30100 -,030 -,061 ,011 -,010 -,013 -,046 1 8. CEO-TMT compensation gap 693.997 1.930.353 ,114** ,232**
-,088* ,027 ,105* ,806** -,043 1 9. ROA*CEO compensation 7009884,00 18062561,000 ,065 ,276** -,056 ,058 ,007 ,723** -,040 ,624** 1 10. ROA 3,3750 13,10900 ,027 ,175** ,007 ,063 -,087* ,094* -,023 ,083 ,488** 1 11. DOI 86,1850 59,13200 -,217** -,090* -,147** ,023 ,156** -,098* ,037 -,082 -,102* -,091* 1
**Correlation is significant at the 0.01 level (2-tailed). *Correlation is significant at the 0.05 level (2-tailed).
4.2 Statistical analysis 4.2.1 Correlation analysis
Table 2 provides descriptive statistics and correlations on all variables for the entire period. To assess the level of correlation between the different variables, a correlation analysis was performed.], this in order to identify the patterns of correlation between variables in the model. This section reports on correlations. Obviously, the different industry types form mutually exclusive categories and therefore the negative correlations between them are logical. Furthermore a number of positive as well as negative correlations can be found.
First of all CEO compensation is significantly and positively correlated with firm age. This was to be expected. Between CEO compensation and TMT-CEO compensation gap there is also a significant correlation with a significant level of 0.01. This is not taken into account because both variables contain some of the same data and hence it is logical that there is a significant correlation. CEO compensation is significantly and positively correlated to the light industry as well as the size of the firm. These variables are included as control variables in the model. CEO compensation is also significantly positively correlated to the moderator in the model, ROA*CEOcompensation. Last, which is interesting, CEO compensation and TMT compensation are significantly and negatively correlated which shows that both compensation are linked and probably the gap between the two is an interesting variable as proposed.
TMT compensation is significantly and negatively correlated to firm age as well as light industry. Both variables are included as control variables.
42
Degree on internationalization is significantly and negatively correlated with firm age, size and light industry (respectively p<0.01, p<0.05 and p< 0.01). Degree of internationalization is significantly and negatively correlated to TMT compensation. This is an interesting effect because the opposite effect was expected. Degree of internationalization is also significantly and negatively correlated to the compensation gap between CEO and TMT is with (p<0.05) as expected before the research. Between moderator ROA*CEO compensation and degree of internationalization a negative relationship can be found. No correlations between degree on internationalization and CEO compensation is found.
Table 3 and 4 provide correlations separately for data before and after the crisis (before and after 2007). The correlation between CEO and the CEO-TMT compensation gap is found in all periods, which makes sense as explained before. Another correlation that can be found in all periods is the correlation between degree on internationalization and the moderator in the model, ROA* CEOcompensation. Also correlation between CEO compensation and TMT compensation can be found in all periods. The effect of industry and firm size can be found in all periods. The effect of age of the firm is much stronger in period 2. This can be explained with the fact that firms are overall older in the second period than in the first period. CEO compensation is in period 1 as well as period 2 positively correlated with the moderator ROA*CEOcompensation. Interesting is that only in period 2, CEO compensation is positively correlated to return on assets and negatively correlated to the degree on internationalization. In both periods there is a significant relationship between degree in internationalization and return on assets. In period 1 no
43
significant relationship between degree on internationalization and ROA*CEOcompensation exists.
4.2.2 Regression analysis
Using regression analysis, the effect of the independent variables CEO compensation, TMT compensation, the TMT-CEO compensation and the moderating effect of the variable ROA have been tested on the dependent variable of DOI. In this study the theoretical regression model is described as:
Y
Degree of internationalization= α + ß
0* X
CEO compensation+ ß
1* X
TMT compensation+ ß
2*
X
TMT compensation+ ß
3* X
CEO-TMT compensation gap+ ß
4* X
Firm Size+ ß
5* X
Firm age+
ß
6* I
ndustry type+ ß
7* (X
ROA*X
CEO compensation)+ε
R2 provides information about the fit of the regression model. The R2 is the
correlation between the real values and the values predicted by the regression model. High values of R2 predict a large correlation between the real and the predicted values and an R2
+1 presents a model where the prediction of the values is consistent with the real values. The R2 is the amount of variation in the dependent variables and thus results in an R2 close
to the one that can be explained by the regression model (Field, 2009).
Tables 8 to 10 (see appendix 2) present the results of the regression analyses for the three different time periods. As mentioned, for all periods 7 models were constructed. The first model only consists of the control variables, the second model tests CEO compensation, the third model deals with TMT compensation and in the fourth model the CEO-TMT compensation gap is elaborated. In the fifth model all independent variables have
44
been added. Model six tests the regression analysis with moderator ROA on the dependent variable DOI. Finally, model seven comprises all variables. As seen in tables 8 to 10, the control variable firm size was in all three periods and in all models statistically significant (p<0,01). The control variable light industry was statistically significant in the first and second period. In the first period with p<0,05 and in the second period with p<0,10. Heavy industry was not statistically significant and the high-tech industry was only statistically significant in the second period (p<0,05). The adjusted R2 of model 5 and model 7 explain
the variance of the model of the proposed variables. In the first period this is 4,6% for model 5 and 5,3% for model 7. In the second period it is 4,7% and 7,2% and for the last model it is 6,1% and 6,2%.
In the overall model CEO compensation is negative and not statistically significant. In the period 2004-2007 the relationship is positive, but not statistically significant either. In the final period 2008-2012 the relationship is negative but statistically significant. As for the first hypothesis, that there is a positive relationship between CEO compensation and DOI which is not supported. The second independent variable is TMT compensation. In the overall period, as well as period 1 there is a negative statistically significant relationship. The former p<0,05, the latter p<0,01. In the period 2008-2012 there is a positive relationship which is not statistically significant. This implies that in the second hypothesis, a positive relationship between TMT compensation and DOI is not supported but seems to contradict existing literature. The third and last independent variable is the CEO-TMT compensation gap. In both periods (2004-2012 and 2008-2012) there is a negative relationship which is not statistically significant, whereas in period 1 there is a positive relationship, but not statistically significant. This means that the third hypothesis provides
45
a negative relationship between the TMT-CEO compensation gap and DOI. Not being statistically significant, the hypothesis is not supported. Finally, the study focuses on ROA as moderator of the model. In periods 2004-2012 and 2004-2007 this is a negative relationship which is statistically significant, with p<0,10 and p<0,01 respectively. In the period 2008-2012 there is a negative relationship which is not statistically significant. Therefore, the fourth hypothesis stating that ROA will negatively influence the relationship between CEO compensation and DOI is supported. This is true for the periods 2004-2012 and 2004-2007.
To sum up there is evidence of a negative relationship between TMT compensation and DOI and that ROA will negatively influence the relationship between CEO and DOI.
4.2.2 Multicollinearity check
Regression analyses take some assumptions into account. This section describes the check for multicollinearity among all variables. These checks are done in order to draw, as good as possible, conclusions in this thesis. First there is check for tolerance and second for Variance Inflation Factor (VIF).
Table 2 provides the correlation analysis and table 5-7 offers and the tolerance and VIF analysis. There is a need to control for tolerance and VIF. Bowerman and O’Connell (1990) and Myers (1990) state that a VIF greater than 10 results in cause for concern. On the other hand tolerance between 0.1 indicates a problem and tolerance below 0.2 indicates a potential problem (Menard, 1995). Table 2-4 shows a negative correlation between the dummy variable of industry type. Because these variables are mutually exclusive categories these can be disregarded.
46
Table 5-7 offers the collinearity statistics. The study does not show a VIF close to 10, so this does not indicate a problem; in this case it is true for all periods. In all periods CEO compensation has a tolerance below 0.2 so that can indicate a potential problem. In period 1 the CEO-TMT compensation gap has a tolerance of 0.188. These variables tend to be most responsible for the variance in the model. All other tolerance statistics are above 0.2 so no problem is indicated.
Table 5: Collinarity statistics DOI 2004-2012 Standardize d Coefficients Beta t Collinearity Statistics Toleranc e VIF Age of firm -,176 -5,113*** ,860 1,163 Logarithm of Assets -,028 -,723 ,675 1,481 Light industry -,123 -2,242** ,341 2,936 Heavy industry -,013 -,283 ,464 2,153
High Tech industry -,036 -,746 ,444 2,254
CEO compensation ,013 ,173 ,189 5,293 TMT compensation -,101 -3,128** ,981 1,020 TMT – CEO compensation gap ,011 ,176 ,286 3,491 ROA -,071 -2,210** ,982 1,018 ROA*CEOCOMP -,088 -1,744* ,403 2,481
Dependent variable: Degree of Internationalization *** significant at the level of 0.01
** significant at the level of 0.05 * significant at the level of 0.10
47
Table 6: Collinarity statistics DOI 2004-2007 Standardized
Coefficients
t
Collinearity Statistics
Beta Tolerance VIF
Age of firm -,185 -3,571*** ,841 1,190
Logarithm of Assets -,030 -,531 ,696 1,437
Light industry -,155 -1,909* ,342 2,920
Heavy industry -,009 -,127 ,459 2,180
High Tech industry -,163 -2,298* ,448 2,230
CEO compensation ,228 1,686* ,124 8,066 TMT compensation -,181 -3,730*** ,963 1,038 TMT – CEO compensation gap -,012 -,111 ,188 5,317 ROA -,108 -2,271** ,990 1,011 ROA*CEOCOMP -,242 -2,736*** ,289 3,464
Dependent variable: Degree of Internationalization *** significant at the level of 0.01
** significant at the level of 0.05 * significant at the level of 0.10
