MSc in International Business and Management – International
Financial Management
(Double Joined Degree)
MASTER THESIS:
“The Relationship between Corporate
Governance and Firm Performance: Evidence
from the Athens Stock Exchange”
Author:
Katsigianni Vasiliki
s1752308@student.rug.nl
Supervisor:
Dr. Niels Hermes
ACKNOWLEDGMENT
ABSTRACT
In this paper I investigate whether variations of corporate governance practices across firms can explain variations in firm performance. Using data from 124 firms listed in the Athens Stock Exchange over the period 2004-2007 I construct an index so as to proxy for firm-specific corporate governance. This index is based primarily on the Greek corporate governance code and is composed by five sub-indices. The empirical findings suggest that better governed firms are associated with higher performance as measured by firm value (Tobin’s Q) and operating performance (ROA). Two-stage least squares analysis is conducted in order to account for endogeneity, and the results remain same. The paper finds also evidence that not all the sub-indices matter to performance rather a subset of them does.
TABLE OF CONTENTS
I. Introduction ... 5
II. The Evolution of Corporate Governance in Greece ... 7
III. Theoretical Framework... 9
A. Agency Theory...10
B. Corporate Governance...11
B.1 Internal Mechanisms ...12
B.1.1 Board of Directors ...12
B.1.2 Executive Compensation Schemes...16
B.1.3 Transparency and Disclosure ...17
B.1.4. Ownership Structure...17
B.1.5 Shareholder Rights ...18
B.2 External Mechanisms ...19
B.2.1 Leverage ...19
B.2.2 Market for Corporate Control ...19
IV. Literature Review... 20
V. Data and Methodology: ... 24
A. General Model...24
B. Variable Definition...25
B.1 Dependent Variables ...25
B.2 Independent Variable ...26
B.3 Control Variables ...28
C. Taking Endogeneity into account...30
D. Sample Selection and Data Sources...32
VI. Empirical Results and Analysis ... 34
A. Descriptive Statistics:...34
B. Correlation:...36
C. Regression Analysis: ...37
C.1 Composite Governance and performance ...37
C.2 Governance Sub-Indices and Performance ...38
C.3 Durbin-Wu-Hausman Test and Two-Stage Least Squares ...39
VII. Conclusion ... 41
References ... 43
I. Introduction
During the last decade the global economy has been rocked to its foundations due to a number of financial crises and multinational corporations’ scandals such as the collapse of Enron and WorldCom in the United States, Marconi in the United Kingdom and more recently Royal Ahold in the Netherlands. In both cases, corporate governance inefficiencies were depicted as the source of all evil bringing the issue of corporate governance at the center of attention. Specifically, a plethora of debates among academics, governmental and institutional bodies took place about the effectiveness of the existing corporate governance mechanisms and the need for new ones.
These debates brought upon a number of initiatives. Such initiatives consist of regulations (e.g. Sarbanes Oxley) and recommendations issued by several international (e.g. OECD) and/or national supervisory authorities of securities markets (e.g. Hellenic Capital Market Commission). The theoretical underpinning behind those movements was that implementation of effective corporate governance mechanisms help in the alleviation of agency problems in modern corporations by mitigating the despotic managerial behavior and increasing the shareholder voice in the management of the public companies.
fact ensures variation in my observations1. Moreover, even though the code is based on international standards (mainly OECD principals) it is adjusted to the Greek reality; therefore, it constitutes the most appropriate guide for the construction of my corporate governance index. Overall, I conclude to 27 provisions that are categorized to five sub-indices: (1) board of directors, (2) board compensation, (3) audit committee, (4) shareholder rights, and (5) disclosure and transparency.
My paper makes several contributions to the existing literature. First, I construct my own index for Greek companies by carefully selecting the suitable corporate governance provisions to be included in. Moreover, I gather my data for the index through publicly available information, deviating from the majority of previous papers that use governance measures provided by rating agencies (Gompers et al., 2003; Brown and Caylor, 2004) and those that construct their own indices using questionnaires (Drobetz, Schillhofer, and Zimmerman, 2003; Karathanassis and Toudas, 2007). In this way I overcome the drawbacks of such approaches which are: the former measures are subjective as the company usually pays to be rated while the latter, are subject to potential self-selection bias 2 and self-report bias3.
Second, I focus on the Greek market. It is expected that each economy, with its own unique characteristics provides a different research framework. For example a series of studies of La Porta et al. (1997, 1998, and 2002) have shown that countries with common law tradition have generally higher corporate governance standards and relatively better shareholder protection comparing to civil law countries. The majority of prior studies concern the Anglo-Saxon countries and particularly the U.S. At a later stage, emerging countries attracted the attention of the researchers. Yet, Greece is an interesting case since it does not fit in any of the aforementioned categories. In particular, Greece is a civil law country with a financial system recently liberalized and more bank-based. Moreover, it involves a small capital market consisting of Greek Sociététes Anonymes (apart from one) where most of them are family-owned and with one-tier board system. In addition, corporate governance is in an evolutionary stage. The paper of Karathanasis and Toudas (2007) is the only study
1
That is, corporate governance mechanisms that are imposed by law are followed by almost 100 percent of the companies. This non-variation in firm-specific corporate governance quality will generate statistical results that do not represent the true relationship between corporate governance and performance.
2
That is, companies with poor governance practices do not usually respond to the questionnaire. 3
concerning Greece. However, the authors use a completely different methodology. They construct their index based on questionnaires and they only include questions related to anti-takeover provision measures. However, takeovers are not really common in the Greek market.
Finally, although it is well recognized in the literature that endogeneity plagues studies on corporate governance, only few researchers apply suitable econometric techniques in their paper so as to control for endogeneity. In this paper, I deal with endogeneity in several ways. In particular, I use a comprehensive set of control variables, Durbin-Wu-Hausman test and a two-stage least squares (2SLS) estimator.
The remainder of the paper is organized as follows. Section II presents the evolution of corporate governance in Greece. Section III provides the theoretical foundation upon which the empirical investigation has been based. In particular, it exhibits a detailed discussion of the conceptual notions that justify the relationship between corporate governance and firm performance. Section IV quotes some of the relevant previous papers on that subject. In section V I analyze the data and methodology that I follow to answer my research questions. Section VI presents the empirical evidence on the relationship between corporate governance and performance. Finally, in section VII I summarize the main findings of my paper.
II. The Evolution of Corporate Governance in Greece
Traditionally Greek companies were, and most of them still remain, family owned (Tsipouri, Xanthakis and Spanos, 2005). The distinguishing characteristic of family-owned firms is that the owner (family) is usually involved in the management (LaPorta et al., 1998). Moreover, the capital market was underdeveloped and the companies were fueling their operations through either their own capital or bank funding. During the mid-1990’s and since Greece was about to enter the European Monetary Union (EMU) the Greek government attempted to give the capital market a boost. Thus, in the late 1990’s a significant number of those companies raised capital by issuing equity through an initial public offering (IPO) (Spanos, 2005)4. As a result, those companies transformed from private family owned to public listed companies
4
and the first signs that the previous obsolete operating methods should be reconsidered have emerged. In particular, the debate on corporate governance in Greece was brought to the forefront for the first time in 1998 through an introductory paper published by the Athens Stock Exchange (ASE).
Along with the massive entrance of companies an increasing number of individual and institutional investors entered the capital market mostly through placements on small and medium capitalization stocks (Spanos, 2005). The inexperience and the short-term speculative placements of these investors led to the increase of stock prices way beyond the equilibrium levels. When the first “bubbles” started to burst the Greek capital market entered a cycle of self-fulfilling expectations and in the end of 1999 suffered an immense crisis. That period of time, many companies and their executives exploited the situation earning great amounts of money while the majority of the individual investors suffered huge losses. Consequently, the investors’ confidence and faith in the market was seriously damaged.
The Hellenic Capital Market Commission (HCMC) in order to reestablish the investor confidence set up a Committee on Corporate Governance. This committee issued a voluntary Code of Conduct for the companies whose shares are listed on the ASE, entitled “Principles of Corporate Governance in Greece (1999)”5. This code aimed at the protection of investors against market abuse, the improvement of the transparency of the market and the establishment of appropriate business ethics (Mertzanis, 2001). It includes 44 principals divided into seven main categories: (a) the rights and obligations of shareholders, (b) the equitable treatment of shareholders, (c) the role of stakeholders in corporate governance, (d) transparency, disclosure of information and auditing, (e) the board of directors, (f) the non-executive members of the board of directors, and (g) executive management.
As it is mentioned above, this code intended to operate as a “soft law” without any legally binding force. However, empirical evidence showed that the level of compliance of Greek public firms was very low6. The main explanation for the
5
The principles and best practice rules incorporated were closely modeled according to OECD Principles on Corporate Governance (OECD, 1999).
6
disobedience of the Greek companies is that they are rather agnostic to non-legally enforceable principles, and also the Greek legal culture is not familiar with “soft law” instruments (Staikouras, 2008).
As a consequence, three years later the Greek government expressed its intention to amend the corporate law so as to incorporate additional corporate governance elements which would become mandatory. This triggered the reaction of the Federation of Greek industries which supported that public firms should be self-regulated as the potential market premium to corporations that voluntarily comply is a sufficient incentive. The Federation, also, created its own code of conduct for its members (Florou and Galarniotis, 2007).
Finally, in May 2002 the Ministry of Economy issued a law (No. 3016/2002) which for the first time mandates Greek listed companies to implement a set of governance guidelines (Florou and Galarniotis, 2007). This law incorporates a subset of the principles contained in the HCMC code. Greek legislation in now fully harmonized with the directives of European Union. However, even though the last decade in Greece has occurred significant improvements in corporate governance, still they are mostly limited to large listed companies which are more in line with the international corporate standards (Tsipouri, Xanthakis and Spanos, 2008).
III. Theoretical Framework
In the late eighteenth century industrial revolution took place. During that period of time many large scale businesses were developed. The finance of those ventures called for huge investments that could not be afforded by one person or family. Hence, joint-venture companies emerged where people from different sections of the society provided the necessary fund. In that kind of widely-held companies the providers of finance and those entrusted to manage the firm were different parties.
of the separation of ownership and control as a prevalent characteristic of modern corporations. Few years later, Jensen and Meckling (1976) pioneered the use of agency theory in explaining why the managers in modern corporations deviate from the shareholder wealth maximization principle postulated by the traditional theory of the firm.
A. Agency Theory
Agency theory attempts to describe agency relationships (Eisenhardt, 1989). According to Jensen and Meckling (1976) a firm is a nexus of such agency relationships which occur when one party (principal) appoints another party (agent) to perform some service on her/his behalf delegating certain decision-making authority. In modern corporations, as described by Berle and Means (1932), the primary agency relationship is that between shareholders and managers. Shareholders finance a company; however, they need the managers’ specialized human capital so as to generate returns on their investment (Shleifer and Vishny, 1997). Thus, shareholders (principals) hire managers (agents) to control the company on their behalf.
The principal-agent relationship is not necessarily harmonious. This is attributed mainly to the assumption that there will be incongruence of interests among individuals involved in an agency relationship (Jensen and Meckling, 1976). This assumption is derived by the “model of the man” that the agency theory is structured upon. Specifically, as in neoclassical economics, the “man” is considered as an opportunistic actor who rationally seeks to maximize her/his own utility (Jensen and Meckling, 1976). Thus, if both shareholders and managers are utility maximizers there is a great possibility the latter not to act in the best interest of the former (Jensen and Meckling, 1976). Rather, there is a constant temptation for managers to take decisions that maximize their own interests and not the shareholder wealth. Evidence of opportunistic managerial behavior includes cases of managers absconding with shareholders’ money, excessive management compensation, consumption of some corporate resources in the form of perks, empire building, and managers that adopt risk averse attitude with respect to corporate investment decision7.
The ability of managers to expropriate shareholders is rooted in the information asymmetry and the free riding problem. Managers have advantage in the access of
7
information while shareholders cannot perfectly and costlessly monitor and acquire information available or possessed by the managers. Moreover, in highly dispersed companies the stakes owned by individual shareholders are not enough to justify the time and money needed to monitor managers.
The combination of the incongruence of owners-managers incentives and the inability of principals to determine if the manager has behaved appropriately constitutes the agency problem (Eisenhardt, 1989). Two aspects of the agency problem discussed in the literature are that of moral hazard and adverse selection (Eisenhardt, 1989). Arrow (1985) equates these two terms with hidden action and hidden information respectively. Moral hazard arises when agent’s actions are hidden from the principal or are costly to observe. In that case manager has the tendency or incentives to behave inappropriately from the view point of the shareholder and not to put forth the maximum effort shirking his responsibilities (Eisenhardt, 1989). Adverse selection arises when the agent intentionally withholds information about himself and his abilities from the principal that is impossible or expensive to verify. Therefore, shareholder may be deceived into believing that the manager has the appropriate skills so as to perform his job without being able to confirm it, not only before but also after the manager’s appointment (Eisenhardt, 1989).
Taking into account the agency problem the question is how to persuade the agent to act in the best interest of the principal. Jensen and Meckling (1976) argue that it is impossible for the principal at zero cost to ensure that the agent will make optimal decision from the principal’s viewpoint. Therefore, they claim that the principal should bear the cost of monitoring the agent’s actions (monitoring cost) and bonding agents to her/his interests (bonding cost) by reward them for achieving her/his goals or else penalize them. However, even after the adoption of effective monitoring and bonding mechanisms there will still be some divergence between the agent’s decisions and those decisions which would maximize the welfare of the principal (residual loss). The sum of monitoring cost, bonding cost and residual loss is defined by Jensen and Meckling (1976) as agency cost.
B. Corporate Governance
magnitude of that cost and its impact on firm performance. The consequence was the emergence of corporate governance. Shleifer and Vishny (1997) define corporate governance as the “ways in which suppliers of finance to corporations assure themselves of getting a return on their investment”. In other words, corporate governance is a set of mechanisms which firms operate so as to reduce the divergence of agents’ interests from those of principals, and, thus mitigate agency costs (Bhuiyan, Hamid and Biswas, 2008). In particular, it is a hybrid of internal and external governance mechanisms. Internal mechanisms concern the interactions between or among firm insiders and specifically between firm management and the board of directors. External mechanisms concern interactions between external stakeholders and firm’s managers and directors.
B.1 Internal Mechanisms
B.1.1 Board of Directors
8Fama and Jensen (1983) suggest the employment of directors as an effective internal mechanism to control abuses of managerial power. The board of directors is the bridge between shareholders and managers. They are elected by the shareholders of the company in order to monitor managers and to ensure that they take decisions that maximize the shareholder wealth. Particularly, their responsibilities include hiring, firing, compensating and advising top management (Denis, 2003). Hence, the board can mitigate the conflicts of interests between shareholders and managers and, therefore, reduce the agency costs and improve the performance within a firm.
The board consists of both executive and non-executive directors. Executive directors are those who are engaged in the daily management of the company. They participate in the board because they have intimate knowledge of the firm’s activities without which the board cannot perform its monitoring role (Pallathitta, 2005). On the other hand, non-executive directors are those that beyond their participation in the board they are neither members of the management team nor are employed in any other position in the company. Non-executive directors are further divided into those that have close ties with the management, the controlling shareholders, and/or the company itself (dependent) and those that their only tie with the company is their
8
directorship (independent)9. Non-executive directors owe their position in the board due to their expertise in a field that is valuable to the firm10 (Pallathitta, 2005) and their role is to ensure that the executive directors are pursuing policies consistent with shareholders’ interests (Fama, 1980).
While in theory the board is an effective corporate governance mechanism, in practice its value is not that clear. This led to the increase of literature that tries to define the determinants of board structure that contribute to its efficient supervisory role. The most significant board characteristics discussed in the previous relevant literature are analyzed below.
B.1.1.2 Board Composition
Many researchers argue that the efficient oversight function of a board depends on its composition. There are two assumptions about the relationship between board composition and its effectiveness. One assumption is that a board dominated by non-executives is more effective (Fama, 1980). Non-executive directors due to their concern about maintaining their reputation in the director’s labor market (Fama and Jensen, 1983), and, in the case of independent directors, their independence (Cadbury, 1992), have more incentives to be active and unbiased in their monitoring role. Jensen (1993) goes even further maintaining that the only executive participating in the board of directors should be the CEO. He explains that as the executives report directly to the CEO it is almost impossible to confront with her/him. The opposite assumption is that the board where executives outnumber non-executives is more efficient. The rationale is that executive directors have privileged access to information on firm’s operations which afterwards they provide to the other members of the board. This type of information is very important as it facilitates the decision-making and the monitoring role of the board and it alleviates information asymmetries as well (Fama
9
According to the Greek Code a director is considered to be independent when:
1. S/he is not a member of executive management or of a Board of Directors of a corporation, directly or indirectly connected with the corporation, presently or in the past year.
2. S/he is not related to other executive members of the Board.
3. S/he is not simultaneously a member of the group forming the majority of shareholders of the corporation, has not been elected as a candidate by that group nor is s/he involved in any transactions with the group.
4. S/he has no other relationship with the corporation, which by its nature may affect his/her independent judgment; more specifically, s/he is not a supplier of goods or services to the
corporation, nor a member of a corporation that provides consulting services to the corporation. Any negotiations with the corporation should be confined only to compensation matters.
10
and Jensen, 1983). Moreover, the directorship demands a lot of time to perform a satisfactory job. However, non-executives usually participate in multiple boards thus they cannot devote the necessary time for the job (Monks and Minow, 2001). At the same time, their compensation usually is uneven to their worth while they hold no or insignificant shares in the company. Consequently, there are worries about their incentives to evaluate and monitor successfully the top management despite some concerns for their reputation (Jensen, 1989).
B.1.1.3 CEO Duality
Board effectiveness depends also on the separation of the CEO and Chairman of the board roles. The CEO is appointed so as to manage the overall operations of the company and to implement the strategies and policies that are mapped out by the board of directors. The role of the Chairman is to head the board and to assess the top management’s, including CEO’s, performance. Agency theory suggests that CEO duality hinders board’s effectiveness in monitoring top management (Fama and Jensen, 1983). This is because, in that case CEO has too much power within the decision making process that allow her/him to take decisions according to her/his own interests at the expense of shareholders. In this regard, Jensen (1993) adds that CEO duality favors the underperforming CEO as it is difficult for the board to remove him. Therefore, it is proposed that the separation of CEO and Chairman position would improve firm performance. On the other hand, a potential advantage of having the same person occupying both positions is that she/he exhibits greater level of understanding and knowledge of the company being able to make quick decisions in a changing environment (Weir, Laing, and Knight, 2002). Moreover, CEO duality entails alignment of interests between the two roles and as such elimination of the potential agency problems resulting in higher firm performance.
B.1.1.4 Board Committees
time to be dedicated and greater specialization by the board members (Pallathitta, 2005). Hence, by dividing the work of the board into smaller groups allow directors to accomplish more in their limited time and make maximum use of their expertise. However, the existence of board committees does not deprive the full board of its responsibilities in the matters concerned. In contrast, board committees should report regularly to the board (Mallin, 2007).
Another reason for the set up of board committees is to limit the influence of executive directors and particularly that of the CEO in conflicting issues such as the overview of corporate accounts, the executive compensation determination and the nomination of new directors (Mallin, 2007). That is, executive directors should not participate in committees whose tasks are to monitor management (audit), appoint those who do the monitoring (nomination), and decide upon management compensation (remuneration) (Carter and Lorsch, 2004). Thus, it is essential that these committees should be comprised by non-executive directors.
Audit Committee:
The main role of the audit committee is to oversee the effectiveness of the financial reporting and auditing process. This typically involves ensuring the integrity of the firm’s financial statements, monitoring the firm’s internal audit function, reviewing the firm’s internal control and risk management, and making recommendations to the board in relation to the selection, removal, and compensation of the external auditors. Moreover, it is also responsible for monitoring and reviewing the independence and objectivity of the external auditors (Clarke, 2007). By performing these functions, the audit committee is expected to ensure the timely release of high quality accounting information by managers to shareholders resulting in the reduction of information asymmetries and hence to the increase of firm performance (Klein, 1998).
Remuneration Committee:
only to hold down the level of the executive payment but also to design and implement the appropriate reward structures that better align the interests between senior managers and shareholders. By compensating executives in congruence with shareholders interests helps in alleviating the agency problems and therefore increasing firm performance (Klein, 1998).
Nomination Committee:
Nomination committees are charged with managing the board’s internal affairs. Specifically, they should identify and select the prospective board nominees. Further, they are responsible for reviewing the performance of each current director to be certain that she/he meets the board’s criteria for renomination, and also for evaluating the effectiveness of the whole board and its committees (Carter and Losch, 2004). Nomination committees are essential in achieving good governance since the board’s monitoring effectiveness depends on the quality of appointed directors. In turn, the more effective the nomination committee is the better the board quality is (Vafeas, 1999). The precondition for the effectiveness of such a committee is to be dominated by non-executive directors. In particular, Vafeas (1999) argues that director nominations made by executive directors are more likely to be made in self-interest and against shareholder interests. Likewise, Jensen (1993) maintains that when the nomination process is dominated by a powerful CEO she/he is more inclined to select directors under her/his influence in order to contain the intensity of board monitoring. In contrast, empirical evidence indicates that when non-executives sit on the nomination committee it is more likely to nominate directors who will challenge the decisions of CEO (Shivdasani and Yermack, 1998). Concluding, nominating committees composed of a majority of non-executives can strengthen the board monitoring quality by nominating directors who will act as shareholder advocates resulting in the increase of firm performance.
B.1.2 Executive Compensation Schemes
performance (Jensen and Murphy, 1990). The underlying logic is that these bonding mechanisms tie manager’s compensation directly to firm performance and, hence, managers have fewer incentives to take actions that will have negative effect in their personal wealth as well.
B.1.3 Transparency and Disclosure
One major element of good corporate governance is transparency. Transparency is related to the full, timely and detailed disclosure of information to the firm’s stakeholders (especially shareholders) so as to enable them to make decisions (Patel and Dallas, 2002). According to OECD (1998a), this information should cover the overall firm performance, the board and management structure, the ownership structure, the strategy and the future prospects, the risk management, the remuneration policy, and the shareholder rights. Moreover, lately it is suggested the companies to disclose information about their social and environmental awareness and their codes of ethics. Sufficient disclosure of information leads to the reduction of information asymmetries between the company’s management and its shareholders, mitigating the agency costs and, hence, increasing firm performance (Chen, Chung, Lee, and Liao, 2007).
B.1.4. Ownership Structure
B.1.4.1 Large (outside) Shareholding
Shleifer and Vishny (1997) suggest as a solution to the agency problem deriving from the separation of ownership and control the ownership concentration. Specifically, they maintain that large shareholders have more incentives to supervise management and can do so more effectively than an average shareholder. This is because the latter holds a small fraction of a firm’s equity stakes and, therefore, is not inclined to spend money and time to exert monitoring behavior. On the other hand, shareholders with substantial stakes in a company have a general interest in profit maximization and enough control over the assets of the firm to have their interests’ respected (Shleifer and Vishny, 1986). As a result, the ownership concentration mitigates the free-riding problems of monitoring effort and, hence, increases firm performance.
shareholders and managers into one involving large and minority shareholders: Large shareholders, as agents, want to maximize their own benefits. Specifically, Denis (2001) states “…blockholders seek both to increase firm value (share benefits of control) and to enjoy benefits that are not available to other shareholders (private benefits of control11)”. Problems arise when they try to maximize these private benefits at the expense of minority shareholders. The ultimate effect of blockholder ownership on firm performance depends on the trade-off between the shared and the private benefits of blockholders (Denis and Conell, 2003).
B.1.4.2 Managerial Shareholding
Jensen and Meckling (1976), based on convergence hypothesis, argue that managerial ownership can be a possible solution to the agency problem deriving from the separation of ownership and control. Specifically, they maintain that the interests of shareholders and managers are more aligned as the proportion of shares owned by managers in a firm increases. This is because in such cases managers are more inclined to maximize firm value as the cost incurred by managerial opportunistic behaviour will have an impact on their personal wealth as well. Yet, later researchers show that insider ownership can lead to managerial entrenchment (entrenchment hypothesis) (Morck, Shleifer, and Vishny, 1988). In particular, at high levels of equity ownership, managers gain so much power within the firm that are able to pursue their own interests (which don’t necessarily equate with value maximization) without the fear of reprisal; i.e. they can entrench themselves (Shleifer and Vishny, 1986). Thus, the ultimate effect of managerial ownership on firm performance depends on the trade-off between the convergence and entrenchment effects.
B.1.5 Shareholder Rights
As I mentioned above, it is easier for large shareholders to be heard and to monitor the management. In contrast, minority shareholders may not be able to protect themselves from expropriation of their rights by managers or large shareholders. La Porta et al. (2002) evaluate the influence of shareholder protection on firm value. They conclude that better shareholder protection is associated with higher firm valuation. They justify their results by saying that when shareholders rights are better
11
protected by the law, shareholders are more willing to finance firms. In practice shareholder rights are expressed by the extent to which the regulatory system of the firm is designed to motivate and allow the easy participation of shareholders in corporate governance, i.e. proxy voting, one-share-one-vote, and cumulative voting.
B.2 External Mechanisms
B.2.1 Leverage
It is widely acknowledged that debt financing helps to the reduction of agency problems relating to information asymmetry and free cash flow (Florackis, 2008) and, thus, to the increase of firm performance. In particular, Shleifer and Vishny (1997) argue that large creditors can assume the role of active monitors. This is because, like large shareholders, they have large investments in a firm and, therefore, incentives to monitor managers. Moreover, Jensen (1986) maintains that debt reduces free cash flow and so limits managerial discretion. Rather than spending any excess funds on lavish perks or futile negative net present value investments, debt forces managers to keep these funds to service company’s debt.
B.2.2 Market for Corporate Control
interests, although this is how they present them. As a result, establishment of anti-takeover defenses may have a negative impact on firm performance.
IV. Literature Review
This paper examines the relationship between corporate governance and performance. Corporate governance is a wide and multifaceted notion covering subjects such as the ownership structure, the board composition, the executive remuneration, the investor protection and so on. There is already a bulk of literature which empirically investigates the relationships between the aforementioned corporate governance mechanisms and performance. However, the results are mixed and inconclusive. One potential explanation that has gained increasing acceptance lately is that, taking into account only one aspect of corporate governance makes it difficult to capture the true relationship unless that specific aspect is controlled for other aspects of governance (Boehren and Oedegaard, 2004). Based on this criticism, many researchers lately started studying the impact of the overall corporate governance on performance formulating indices that incorporate several aspects of corporate governance. In my paper I embrace this point of view and thus I review only the respective strand of literature.
The first12 and most oft-cited related study was conducted by Gompers et al. (2003). They compute a governance index (G-index) for 1.500 large U.S. companies during 1990’s consisting of 24 provisions related to takeover defenses and shareholder rights derived from Investor Responsibility Research Center (IRRC). They construct two portfolios separating firms into those with strong and weak shareholder rights and they assign them the terms democracies and dictatorships respectively. The authors assert that an investment strategy that purchases shares in Democracies and sells shares in Dictatorships earns average annualized abnormal returns of 8.5% during the sample period. Moreover, they conclude that democracies have higher firm value but not better operating performance.
This last point was met with skepticism by Core, Guay and Rusticus (2004) who suggested that Gompers’ et al. (2003) results are time-period specific. Thus, they extend Gompers et al. (2003) paper studying the following 4 years (2000-2003) and conclude that not only weak governance does not cause poor stock returns but also
12
document that dictatorships outperform democracies during that period of time13. Bebchuk, Cohen, and Ferrell (2005) investigate the same 24 IRRC provisions and identify six that, according to their results, are most likely to help managers to entrench themselves by making a hostile takeover almost impossible. The authors aggregate these six provisions into one index that they call “entrenchment “index. They find evidence that their index is significantly correlated with firm valuation as well as with stock returns, thus, they conclude that these six provisions fully drive the Gompers’ et al. (2003) results.
One of the main critiques with regard to the paper of Gompers et al. (2003) is that G-index is focusing only on the external mechanisms being reflective of an anti-takeover protection index, rather than a corporate governance index. Brown and Caylor (2004) add to the literature by constructing a more comprehensive index (Gov-score) which encompasses 51 provisions covering both internal and external mechanisms. In order to construct their index they use governance information provided by the Institutional Shareholder Services (ISS) for 2003 and find that better-governed firms are relatively more profitable, more valuable and pay out more cash to their shareholders. Moreover, they narrow their index down to seven provisions that are associated more often with good performance and form the Gov-7 index. As Bebchuk et al. (2005), they conclude that these provisions are the “key drivers» of their results. Cremers and Nair (2005) also find that effective corporate governance requires both internal and external mechanisms. Specifically, using shareholder activism as their proxy for internal governance and G-index as a measure of external governance, they conclude that neither governance mechanism alone affects performance, rather both of them do.
Black (2001) raised the issue of the country under examination. That is, all the research up till then has been focusing in the US market. However, Black (2001) maintained that one of the primary reasons for the non-consistent previous results is that in U.S. the variation in firm governance is really small due to the very high standards been set by laws and norms. In contrast, in Russia due to the weak regulatory environment these differences are much larger and could have measurable effects on firm performance. Hence, he finds a strong positive correlation between
13
governance and share prices for the 17 Russian firms of his sample. However, his results are questionable because of the small size of the sample. Corroborating evidence is provided by Aggarwal and Williamson (2006) that use ISS governance attributes so as to create their index. They distinguish between mandated regulations and recommendations and provide evidence that for the U.S. market only the latter has a significant impact on firm performance.
Klapper and Love (2004) and Durnev and Kim (2003) find consistent evidence with Black (2001) indicating that firm-level corporate governance matters more in countries with weak legal environments. Specifically, they note that a strong institutional setting might act as a substitute for firm-level corporate governance. Klapper and Love (2004) use the Credit Lyonnaise Securities Asia (CLSA) governance index and find that better corporate governance is associated with better operating performance and higher market valuation for 374 firms in 14 emerging countries. Similarly, Durnev and Kim (2003) use CLSA governance index complemented with the Standard and Poor's (S&P) disclosure index to measure corporate governance practices for a sample of 859 firms in 27countries. Their empirical tests indicate that firms that score higher in governance and disclosure rankings have higher stock market valuation.
In the context of Europe, Bauer et al. (2004) provide insights about the nature of the relationship between corporate governance and firm performance. Specifically, they use the Deminor Corporate Governance Ratings for companies included in the FTSE Eurotop 300 index in 2000 and 2001 and find almost no significant relationship between governance scores and firm valuation. Interestingly, their findings suggest that there is a negative correlation between governance and operating performance14.
Using a broad sample of 515 firms Black et al. (2006) study the case of Korea. In this paper, the authors discuss and control for endogeneity, an issue that plagues all the studies in the field. They formulate an index using information provided by a survey based on questionnaires conducted by the Korea Stock Exchange from 1998 to 2003 and find a strong positive relation between corporate governance and market valuation. They also find that better-governed firms pay higher dividends but they are not more profitable than other firms. To control for endogeneity, the authors conduct
14
two-stage and three-stage least squares analyses but they conclude that it has no effect in their results. In this context, Renders and Gaeremynck (2006) reexamine the paper of Bauer et al. (2004) finding analogous results as they document insignificant and even negative relationship of corporate governance with performance under OLS. However, after controlling for sample selection bias (through inverse Mills ratio) and endogeneity (through two-stage least square analysis) they document a positive and highly significant relationship. As such, they conclude that controlling for both sample selection bias and endogeneity is important, as the coefficient of corporate governance is severely underestimated when these problems are not taken into account.
Another aspect of concern is the development of the index. All the aforementioned papers with the exception of Black et al. (2006) use governance measures provided by agencies so as to create their index. Larcker, Richardson, and Tuna (2007) question the effectiveness of those indices. Specifically, they maintain that these typical “structural” indicators of corporate governance provided by or based on the institutional rating services have very limited ability in explaining variations in firm performance. Thus, using principal component analysis they come up with 14 governance indicators and find that these are associated with future operating performance and stock returns. However, they find no evidence for the impact of corporate governance on the abnormal accruals and accounting restatements.
In the same vein, Drobetz et al. 2003) self-construct their index based on data obtained from questionnaires sent to German public companies and document a significant positive relationship between corporate governance and performance. Likewise, Aman and Nguyen (2006) construct their own governance index for Japan and use it to form 5 portfolios whose monthly returns are analyzed over the period 2000–2005. However, in contrast to Drobetz et al. (2003), they find that poorly governed firms significantly outperform better-governed firms even after adjusting for industries and checking for different sample period15.
Finally, concerning Greece, there is just one paper studying the governance-performance relationship which is conducted by Karathanassis and Toudas (2007). They create their own index using questionnaires which comprises 6 questions
15
concerning mainly anti-takeover protection measures. The questionnaires were sent to the executives of quoted firms on the ASE and they form a dataset of 262 firms. They conclude that firms with few shareholder rights are accompanied by poor performance both in terms of firm value and stock returns.
V. Data and Methodology:
A. General Model
The purpose of my paper is to determine whether variation of corporate governance practices is associated with changes in performance. I therefore estimate a model in which performance is regressed on firm-level governance scores. In particular, I run the following multivariate regression using ordinary least squares (OLS):
PERFit = α + βGovit + γXit + εit (1)
where i refers to the firm, and t is time. PERFit denotes the two performance variables
applied in my empirical analysis: Tobin’s (Q), and Return on Assets (ROA). Govit is
the key variable in this study as it represents the firm-level governance scores, thus, β is the parameter of primary interest. Xit is a vector of control variables that in prior
literature have shown that affect performance, i.e. firm size, firm age, growth prospects, leverage, and ownership concentration. In addition, I control for industry and ownership type with the use of dummy variables. εit represents the composite
error term.
Moreover, following the methodology of Bebchuk et al. (2005) I regress performance against each one of the five sub-indices so as to check whether any of them is more value-relevant than the others:
PERFit = α + βSubGovit + γXit + εit (2)
where SubGovit denotes the five sub-indices incorporates in my overall index.
B. Variable Definition
In this section I provide a detailed definition of the variables I use in my empirical analysis. Table 1 summarizes the calculation methods of these variables.
B.1 Dependent Variables
Following the approach used in prior research (i.e. Gompers et al., 2003; Klapper and Love, 2004; Brown and Caylor, 2004), I measure firm performance using two separate sets of metrics: (a) firm valuation, and (b) operating performance.
Valuation Variable
I employ Tobin’s Q as a proxy for firm value because it has been extensively used in prior corporate governance research “since the work of Demsetz and Lehn (1985) and Morck, Shleifer, and Vishny (1988)” (Gompers et al., 2003). This variable is computed as the market value of assets (book value of assets - book value of equity + market value of equity – deferred taxes) divided by the book value of assets and represents the investors’ evaluation about the future profitability of the firm.
According to the agency theory, I expect a positive relation between corporate governance and firm valuation for two reasons. First, good corporate governance increases investors’ trust as they consider less possible their funds to be expropriated by the managers who control the firm. Hence, the cost of capital for well-governed firms decreases, leading to higher firm value. Second, investors value higher well-governed firms because they realize that most of the firm’s profits will flow back to them as interests or dividends (La Porta t al., 2002).
Operating Performance Variable
The theoretical model of agency theory predicts that in well-governed firms managers are more likely to invest in positive net present value projects (Shleifer and Vishny, 1997) and, thus, a higher return from these investments is expected resulting in better operating performance.
B.2 Independent Variable
Corporate Governance Measure
In line with previous research of this nature, I construct an index so as to assess the state of corporate governance quality at a firm level. As can be implied by the literature review, there is no one generally accepted model of corporate governance that works for all countries and all companies. In particular, there are different codes of “best practices” that take into account the specific legal, political, and economic environment of each country (Zheka, 2006). Nevertheless, there are some global standards that can be applied across a range of jurisdictions and have been formulated by international organizations such as the OECD (i.e. fairness, transparency, accountability, and responsibility).
With these in mind, I developed my index based on the theory discussed in section III, the OECD principals, and the Greek voluntary code. From the constellation of question that I came up with, I eventually choose and include in my index only those that could be objectively answered through access to publicly available information such as annual and periodic reports, web-sites, minutes of the boards of directors and shareholders’ meetings and so on. This is because, I preferred to answer these questions directly rather than send questionnaires to companies’ executives since I wanted my paper to suffer less from self-selection and self-reported biases and to have as large sample as possible. Hence, I ended up with 27 questions pertaining to different corporate governance provisions. Then, I grouped these questions into five sub-indices, namely: (1) board of directors, (2) board compensation, (3) audit committee, (4) shareholder rights, and (5) disclosure and transparency.
by non-executives, the CEO-Chairman roles are separated, and there is a nomination committee chaired by a non-executive director. Moreover, it is advisable the Chairman of the board to be non-executive director, as it is expected to create additional monitoring benefit on incumbent management (Larcker et al., 2007). The excessive representation of the controlling family on the board relative to independent non-executive directors also has a negative impact on board effectiveness. This is because the disproportionate voice in the decision making gives the opportunity to family members to serve the family interests at the expense of those of the other shareholders. In the same vein, the occupation of the CEO post by a member of the controlling family further compromises board effectiveness.
The second sub-index refers to Board Compensation (COMP) issues. Based on the agency theory, the implementation of long-term incentive plans is considered as a good mechanism to mitigate the managerial discretion aligning the interests of executives with those of shareholders. Likewise, it is argued that CEO should have an ownership stake in the firm because in that case her/his personal wealth depends on firm value. However, it is recommended the establishment of a remuneration committee to determine the compensation of executives comprised mainly and chaired by non-executive directors.
Audit committee (AUDIT) sub-index focuses on two main issues. Firstly, it
concerns the existence of an audit committee. According to the theory, audit committee helps in the more effective financial monitoring and, therefore, provides better quality of financial reporting. Secondly, it measures the degree to which an audit committee is efficient (i.e. its members are accounting/financial literate) and whether or not is comprised solely by non-executive directors.
Agency theory suggests that better transparency and disclosure quality diminishes the information asymmetries between managers and shareholders. The purpose of the final sub-index is to evaluate whether companies have adequate
Transparency and Disclosure (DISC) standards in place. Hence, in this category I
include questions about whether or not a company provides information about the audit committee structure, the board members resumes and remuneration, the ownership structure, and the corporate governance practices. According to OECD principals and the Greek Code such elements allow current and potential investors to have a clear picture about the company’s governance structure.
I use a rather straightforward way to answer the questions included in my index. Specifically, I code the 27 binary questions, assigning one for every provision that is met in the firm and zero otherwise. Then, I equally weight all the provisions within a sub-index to compute an average score. Similarly, I equally weight each sub-index to obtain the governance score of the whole index. I understand that I do not differentiate for the relative importance of each provision or sub-index but without prior theory uneven weighting is subjective (Black et al., 2006). Moreover, this approach has the advantage of being more transparent and allowing the easy interpretation of results. Finally, the overall corporate governance index is normalized to have a value between 0 and 100, with better-governed firms having higher scores. Table 2 presents details about the index.
B.3 Control Variables
The rationale for the control variable is that firm performance may be affected by factors besides that of governance. Hence, these factors need to be controlled for so as to isolate the effect of governance on performance. Specifically, I use the following control variables in my model:
Firm size
Age
I proxy for firm age with AGE, defined as the natural logarithm of the number of the years a company has been listed on ASE. A number of researchers suggest that older firms are more experienced with well-established reputation and usually have relatively lower cost of capital as compared to younger firms. Thus, they normally enjoy superior performance. However, another stream of research suggests that with age may also come rigidity and inertia. That means that older firms do not adapt fast to changing circumstances and, therefore, lose their advantage against the younger more flexible companies.
Growth Prospects
I also include a measure of the growth of the company. Previous studies have shown that a firm’s growth prospects have a positive influence on performance (Klapper and Love 2004; Black et al., 2006). The underlying notion is that growing firms represent a good investment opportunity and this is reflected on firm’s price. Lacking a better measure of growth, I follow Klapper and Love (2004) and use the average annual growth of sales denoted as GROWTH.
Leverage
Leverage (DEBT) is calculated as the book value of debt to the book value of assets at the end of the financial year. The impact of leverage on performance can be either positive or negative. First, due to the tax-shield benefits that the trade-off theory of capital structure suggests. The second reason stems from the work of Jensen (1986) who argues that debt helps in decreasing the agency costs as it reduces the free cash flows which executives are likely to squander. Consequently, more indebted firms perform better. On the contrary, investing in a highly leveraged firm is often considered as a risky decision, since high external financing increases the possibility the firm not to be able to pay off its debt.
Ownership Concentration
concentration is associated with both costs and benefits. In particular, Shleifer ad Vishny (1997) and Jensen and Meckling (1976) argue that concentrated ownership is beneficial to corporate performance since large shareholders are better at monitoring management. However, Morck et al. (1988) suggest that beyond a certain level of concentration, the relationship might be negative.
Ownership Type
It is not clear whether measures of corporate governance affect performance in the same way when ownership is not in general widely dispersed, and in particular when ownership is concentrated in the hands of families (Klein, Shapiro, and Young, 2005). Therefore, I include a set of dummy variables that proxy for the identity of the controlling owner: (a) foreign ownership (FOR): control in the hands of foreign investors, (b) family ownership (FAM): control in the hands of one family or one single investor, (c) state ownership (GOV): control exercised by the government, and (d) other type of ownership (OTHER): control in the hands of other types of controlling shareholders not included into the aforentioned categories (e.g. national companies).
Industry Dummy Variables
Finally, I take into account the sector effects by including 14 binary dummy variables in my model, each representing the industry under which a company operates16.
C. Taking Endogeneity into account
Endogeneity occurs when the independent variable (Govit) is correlated with the error
term (εit) in a regression model. In this case, the estimated OLS coefficient will be
biased and may suggest a causal relationship that does not exist. The two main endogeneity problems in corporate governance research refer to omitted variable bias and reverse causality. First, omitted variable bias occurs when firm’s characteristics that are not included in the regression can simultaneously influence the independent and dependent variables. Second, firms with higher performance choose better governance structures, as they believe that doing so will further raise their performance. In this case, the direction of causality is the reverse from the one denoted in equation (1).
16
I have already tried to dissipate the omitted variable problem including an extensive set of control variables in equation (1). In order to adequately address the potential reverse causality problem, I use instrumental variables. These variables should satisfy two main conditions in order to be appropriate: strong correlation with the endogenous variable (Govit) and null correlation with the error term from the
equation (1). In other words, the instruments should predict the dependent variable (PERFit) only indirectly, through its effect on the endogenous variable, not directly.
Given these restrictions it would be difficult to identify an ideal instrument unless some restrictive assumptions are made or an exogenous condition on corporate governance practices is imposed on firms (Lefort and Walker, 2005). For example, Black et al. (2006) to construct their instrument (i.e. size dummy indicator) use a peculiarity of Korean legislation that requires companies with assets over 2 trillion Won to apply different governance requirements; this requirements is by definition an exogenous factor. Since in Greece there are not such purely exogenous phenomena I follow the methodology of several prior researchers (Durnev and Kim, 2005) and use “imperfect” instruments. Specifically, in line with Larcker et al. (2007) I assume that firm size and industry constitute the two most appropriate instrument variables in cases of lack of exogenous phenomena.
As previously, I calculate firm size as the natural logarithm of total assets and I expect that it will have a positive impact on corporate governance. The underlying logic is that usually bigger companies are subject to greater public scrutinity and have more resources to undertake additional governance initiatives. At the same time, different industries experience different market conditions. These sectoral differences may be relevant in explaining variations in corporate governance practices (Durnev and, Kim, 2005). Thus, I use the industry corporate governance average as a proxy for industry affiliation.
Having selected SIZE and INDav as my instrument variables, I conduct the Durbin-Wu-Hausman test and run two-stage-least squares (2SLS) analysis. The Durbin-Wu-Hausman test is a two stage procedure that is used to identify whether my model suffers from endogeneity problems. In the first stage, Govit is regressed on the
instrument variables and the set of control variables. Specifically, I use the following equation:
where, IVit represents the instrument variables: SIZE and INDav, Φit denotes the set of
control variables: the same as in equation (1) excluding firm size that now is used as instrument, Govit is the firm-level governance score, and εitis the error component.
The residual term represents the variation in corporate governance that cannot be explained by the instrument and control variables.
In the second stage, I regress PERFit on Govit, the control variables used in
equation (3), and the residual term from the first stage regression:
PERFit = α + βResidit + γGovit + δΦit + εit (4)
If the coefficient of the first-stage residual (Residit) is significant, this is a sign
of endogeneity. This is because in that case other factors than corporate governance and control variables can explain variation in performance.
After having checked whether my model suffers from endogeneity or not, I conduct two-stage least square (2SLS) analysis. The 2SLS procedure is similar to Durbin-Wu-Hausman test but is used to control for endogeneity. Specifically, while the first stage is common (i.e. equation 3), in the second stage PERFit is now
regressed on the residual term from the first stage regression (Residit), instead of the
Govit itself, and the control variables:
PERFit = α + βResidit + γΦit + εit (5)
Note, that in all cases I control for potential heteroscedasticity by employing the White (1980) standard errors test.
D. Sample Selection and Data Sources
non-financial sectors. Obviously, my firm sample covers all the non-financial sectors and is large enough (48.2%) to be considered as representative of the Greek listed companies17. Table 4 reveals the differences between the means of my sample firms and the rest of the population in terms of age and size for the year 2007. Apparently, in both cases the differences are not significant confirming that my dataset does not suffer from sample bias.
As I mentioned above, I estimate the firm-specific corporate governance scores using data provided by secondary information such as annual and periodic reports, web-sites, minutes of the boards of directors and shareholders’ meetings. Greek listed companies in their annual reports and websites provide explicit information about the classification of board members into executives, non-executives, and independent non-executives, the tenure of directors, and who are the Chairman and the CEO of the company. In their annual reports they publish also information about the existence or not of relationships among the board members within the second degree of kinship. In case of lack of such information I distinguish the controlling family members that participate in the board through their last names18. In addition, Greek listed companies are obliged by law (No. 3371/2005) to incorporate in their annual reports a “Board of Directors management” report. In this report, they have to release information on the number of outstanding shares, the types of shares, the shareholders who directly or indirectly hold more than 5% of the company’s shares, and the agreements that the company has with its shareholders and its board members that come into force after a takeover bid. Additional information such as the existence of board committees, the board members’ resumes, their remuneration and their stake in the company are upon company’s discretion to disclose it.
In line with previous researchers (Bauer et al, 2003), I assume that corporate governance remains relatively constant over time. Thus, I answer the questions included in my index based solely on the 2007 releases (annual and periodic reports and minutes from meetings) and the firm’s website. To check the validity of this assumption I estimate the correlation between the corporate governance scores of 25 of the sample companies for which I have time series data for the 2004-2007 period.
17
My sample does not have any company from the Telecommunication industry which, however, includes only two firms.
18
The correlations range between 0.89 and 0.97 and are all significant at the 1% level, corroborating the argument that corporate governance is relatively persistent over short period of time.
Financial data used to calculate the dependent and control variables are collected from the 2004-2007 annual reports. Information of the year end stock prices so as to measure firms’ market capitalization is acquired from DataStream. Finally, I gather share ownership data from the ASE which provides a database with annual information on the shareholders of the listed companies that hold directly or indirectly more than 5 % of the company’s shares.
Since I have firm-specific corporate governance observations only for one year, the rest of the variables are calculated as averages of their 2004-2007 annual values. The approach of using averages when comparing to the panel data approach has the drawback of missing the time variation of the observations. However, in the current case the corporate governance variable is considered as constant over the years; thus, the fact that there is no time variation to the independent variable suggests that the panel data analysis would not make any substantial difference. Nevertheless, I also conduct a robustness check under panel data analysis employing the 2007 scores backwards until 2004.
In order to reduce the effect of outliers on my regressions all the performance variables are winsorized by 1 and 99%. Table 5 presents the descriptive statistics for the performance measures and the control variables. The mean Tobin’s q is 1.4 and the ROA is 5%. Greek firms are highly concentrated with more than 60 percent of the average firm’s shares being held by major shareholders. From the total 124 firms, 88 are controlled by families while the rest 36 are controlled by foreign companies, the government and other types of controlling shareholders (18, 3, and 14 respectively).
