Volume 7, issue 2, Winter 2009, Continued - 3.pdf

Loretta Baryeh, Peter DaDalt, Varda

Yaari study a sample of 233 firms that conducted SEOs in the 1987-2004 period and either their directors and/or their senior officers traded in the firm's shares. We find that 15% have insider trading by directors only, and 85% by both directors and senior officers.

The market discounts the insider trading at the issuance date (the discount increases in the volume of insiders sales), but it treats insider trading by directors as a favorable signal that reduces the discount. Our study then identifies two ways directors monitor opportunistic insider trading before SEO. One is to ban it, as evident by the fact that under our selection criteria, 791 firms conducted SEOs in the 1987-2004 period. The other is to trade too as a positive signal to the market.

Johan de Beer narrates the introduction of single stock futures to a market allows for a per company impact-assessment of futures trading activity. Thirty-eight South African companies were evaluated in terms of a possible price and volume effect due to the initial trading of their respective single stock futures contracts. An event study revealed that SSF trading had little impact on the underlying share prices while a normalised volume comparison pre to post SSF trading showed a general increase in spot market trading volumes.

Tom Mortimer considers the traditional approach to the 'state' Models of corporate governance, namely shareholder Model and stakeholder Model. It then considers the extent to which developments in a recent accession EU country, Poland, reflects either of these Models or adopts a hybrid approach. It then offers proposals for the future development of corporate governance within Poland.

I. Introduction

It is possible to identify three levels of determinants of firms' performance. The first, relates to external factors that are beyond the control of the firms, and are generally economy-wide. Second, are factors that are internal and under the direct purview of the firms. These factors, which include managerial efficiency, governance structure, ownership structure etc affects the ability of the firms to cope with external factors. Finally, there are other factors such as size, leverage, and nature of the industry that affect firms' performance. However, corporate governance is considered to involve a set of complex indicators, which face substantial measurement error due to the complex nature of the interaction between governance variables (such as board size, board composition, return on assets etc) and firm performance indicators. Nevertheless, previous empirical studies have provided the nexus between corporate governance and firm performance. However, despite the volume of the empirical work, there is no consensus on the impact of corporate governance on firm performance. Consequently, this lack of consensus has produced a variety of ideas (or mechanisms) on how corporate governance influence firm performance.

In the case of Nigeria, poor management and weak internal control systems account for some of the lapses in the operation of some corporate organizations. In addition, technical mismanagement involving inadequate polices, lack of standard practices, poor lending, mismatching of assets and liabilities, weak and ineffective internal control systems as well as poor and lack of strategic planning has bee prevalent in the Nigerian corporate industry. Thus, the significance of this study is very high in an environment like Nigeria, which is characterized by growing calls for effective corporate governance, particularly for public limited liability companies. This call is understandable in view of the importance of effective governance at both microeconomic and macroeconomic levels. Understanding the effect of internal and external mechanisms and pattern of corporate governance among the Nigerian corporate firms will proved invaluable information to top policy makers and assist the government on the restructuring of the Nigerian corporate sector.

In the present changing economic environment, the corporate sector must brace up to the challenges of globalization where firms that cannot adapt to modern business culture may not survive. It is therefore important for firms to find out the best corporate practices in other parts of the world and how they can integrate these into their business culture to enhance their performance. Consequently, this study investigates the relationship between internal and external mechanisms on corporate governance and firm performance. It provides an analysis of the governance structure of Nigerian firms and their managerial characteristics and the extent to which the governance structure and internal/external monitors influence their performance. There is no doubt that the structure of ownership of a firm and its internal/external effect has important impact on the capability of the firm to respond to external factors impinging on its performance. In addition, despite the renewed interest in issues of corporate governance in the African continent, relevant empirical studies are still few and far in-between in Nigeria (except for the studies of Oyejide and Soibo, 2001; Sanda et al; 2005; Kajola;2008). However, these studies suffers from the weakness of excluding important mechanisms for addressing the corporate governance and firm performance relationship. This has limited the depth of our understanding of corporate governance. Thus, we believe that this study should improve our understanding of the relationship between corporate governance and firm performance in Nigeria.

In addition, the studies are usually generalized for all the sectors of the economy. This may lead to making sweeping generalization that did not cut across the sectors. This study focuses on the manufacturing sectors that are quoted on the Nigerian Stock Exchange (NSE) to enhance the reliability of information obtained since they are required by law to publish their annual report and accounts. 1 The scope of the study shall cover a period of five years. This is between 2002 and 2006. The choice of the period and the firms included in the analysis were guided by data availability considerations. The rest of this study is divided into four sections. Section II discusses the background to the study while section III presents a brief review of related studies. The analytical framework is discussed in section IV while section V presents and discusses the empirical findings. Section V summarizes and concludes.

II. Structure and Performance of the Corporate Sector in Nigeria

The Nigerian stock exchange (NSE), the apex body on the Nigeria capital market was established in 1960 as the Lagos Stock Exchange. It later became the Nigeria Stock Exchange in 1977. At present there are six branches with each having a trading floor. The branch in Lagos was opened in 1961, Kaduna 1978, Port Harcourt 1980, Kano 1989, Onitsha February 1990; and Ibadan August 1990. The exchange which started with only 19 securities traded on its floors in 1961 has about 257 securities as at 2002 with a total market capitalization of approximately N763.9 billion. The total value of reading transaction on the exchange rose from N13.6 billion in 1998 to N59.0 1 However, additional information was obtained from interviews conducted with the companies, the from Nigeria Stock Exchange (NSE Fact books. Banks and Insurance companies were excluded because the recapitalization policy effect have not really been captured in the Nigerian banking system. Also the debt structure of banks and insurance institutions are not comparable to the firms in other sectors. This is consistent with some other studies in the literature (Adenikinju, 2005). billion in 2002. As at 2003, 180 companies were listed on the first tier market of the stock exchange and there were 19 listed on the second tier securities market (Adenikinju, 2005). There is an increase in all the parameter use to capture the performance summary of the Nigerian stock exchange from the year 2003-2006. However, the Nigerian stock market is small illiquid and volatile. Although the number of listed securities is increasing, trading activity is still very thin due to the observed reluctance of institutional and individual investors to trade in the secondary market. Table 1 provides summary information on the performance of the capital market in Nigeria between 1998 and 2006.

The regulatory body of the Nigeria capital market is the security and Exchange Commission (SEC) which was established in 1979, almost 2 decades after the NSE was established. Furthermore, the first comprehensive legal document providing rules sand regulations for the conduct of operations in the stock exchange, the Securities and Investment Act No. 45 was promulgated in 1999. the implication of this is very clear -the stock exchange operated for almost two decades with a regulatory organ and for another two decades with a regulatory organ weakened by the absence of a comprehensive legal document to assist in the discharge of its regulatory system‖ (Sanda, et al 2003). Whilst, the SEC supervises the activities of the various entities that operate on the capital market, the statutory body that deals with incorporation is the Corporate Affairs Commission (CAS) through the provisions of the Corporate and Allied Matters Act No. 1 of 1990. The board picture of shareholding in Nigeria across the various sectors of the capital market is presented in Table2. The information in Table 2 shoes that concentration varies across the sectors. But on the average it may be safe to conclude that concentration ratio is quite high. Most of the sectors have a concentration ratio that is above 50 percent. CR is highest in Airline (99%), Agriculture (88.%), Automobile (89.6%),Commercial Services (85.7%) and Chemical and Paints (72%). It is however low in Footwear (25%) and Printing and Office Equipment (49%). The mean value of CR for the entire corporate sector is 63.5 per cent. The concentration of shareholding could have positive or negative effect on firm performance. While it can help reduces agency problem, it can also lead to poor governance as a small group can exercise control over a firm and pursue the objectives of the insider at the cost of the outsiders, or small shareholder (Claessens et al, 1999).

Another feature of ownership structure shown in Table 2 is that foreign institutions were more prominent than individuals in the foreign shareholdings. In other words, foreign institutions held the bulk of the shares by foreigners. Thus, in many cases, foreigners hold block minority shares that would have give them the leverage to control the firm. In addition, on the aggregate, institutions account for 43 per cent of shareholdings compared to 38.9 per cent for individuals. In other words, institutions are ahead of private individuals in term of total shareholdings. However, when we compare foreign institutions with foreign private individual, it is obvious from the table that foreign institutions' holding (26.4%) is several multiple of foreign private individuals. One explanation for this is that in a weak property rights environment, institutions are more able to exact protection over their investment compared to private individuals.

The above pattern is however different for Nigeria investors. Private individuals in Nigeria control more share than their local institutions counterpart. The former has 37 per cent compared to 17 per cent for domestic institutions. Perhaps, it should be mentioned the fact that in most cases foreigners (institutions) are the single largest shareholders accounting for 40 per cent of shareholding in many instances. This implies that by and large the bulk of Nigerian shareholder own minority shares in their own companies. It also suggests that foreign institutions come close to outright ownership of most of these companies. Again this could be explained by the fact that most of these companies have foreign origin. They stated out as subsidiaries of parent companies located in the western countries.

Internal Mechanisms for Good Governance

A well governed corporation needs to balance the roles of three groups of players: shareholders (and employees, if they have a governance role), boards of directors, and manages, while meeting all of its financial commitments and other obligations to a broad array of stakeholders. Shareholders provide (risk) capital in return for the opportunity to benefit from profits and increases in corporate value. Shareholder may have a range of rights and powers under law and regulation that can include the right to elect and remove directors and auditors and to appoint and approve or disapprove fundamental changes, such as mergers or changes in capital structure. The shareholders' interest is generally in maximizing the value of the firm's equity and distributions relative to risk over time.

The board of directors represents the interests of shareholders and may have obligation to other stakeholders under various statutory and voluntary provisions. An independent board of directors, the core internal governance mechanism, is the bridge between management and owners, other stakeholders, and the outside world. The board need to be independent particularly of management, and its members should be well-versed in the firm's line of business or in general business areas such as business law, accounting, marketing, finance, or production. The board should also be of reasonable size, and the terms of its directors should be fixed. Making the board more effective is at the centre of the corporate governance debate. Internal mechanisms of corporate governance work to check and balance the power of mangers, shareholders, directors and stakeholders. But while internal incentives are necessary for efficiency, they are not sufficient for good governance. In addition to these internal factors, corporations in market economics are also disciplined externally.

External mechanisms for good Governance

Formal legal and regulatory obligations are part of the external incentive structure designed to ensure that competing companies abide by common standards of fairness, transparency, accountability, and responsibility to protect shareholders, consumers, workers, the environment, and even competitors from abusive practices. A good legal and regulatory framework efficiently addresses the entry, operations, and exists of firms. Other external elements are developed by national and international bodies on best practices (quality of disclosure, accounting and auditing standards, labour rules, environment standards, industrial product standards, listing requirements) and other areas of practices that are qualitative them in law can lead to overregulation and can curb entrepreneurial spirit.

Both equity and debt markets impose substantial discipline on management. Equity markets continuously monitor and place an objective value on corporations and, by extension, on their management. The day-to day performance of a company's shares on a stock exchange is a transparent reminder to managers and Owners of the company's perceived viability and value. This assessment permits shareholders to assess management performance and gives managers an incentive to minimize the costs of equity, since failure to do so will make them vulnerable to takeover. An active market for corporate control, fluctuations in stock prices and the influence of shareholders keep managers focused on efficiency and commercial success.

III Review of Related Studies

The literature identifies several channels through which corporate governance effects growth and development. The first is the increased access to external financing by firms. This in turn can lead to larger investments, higher growth and greater employment creation. The second channel is a lowering of the cost of capital and associated higher firm valuation. This makes more investments attractive to investors, also leading to growth and more employment. The third channel is better operational performance though better allocation of resources and better management. This creates wealth more generally. Fourth, good corporate governance can be associated with a reduced risk of financial crises. This is particularly important, as financial crises. This is particularly important, as financial crises can have large economic and social costs. Fifth, good corporate governance can mean generally better relationships with all shareholders. This helps to improve social and labour relationships and aspects such as environmental protection. All these channels matter for growth, employment, poverty reduction and well being more generally. Empirical evidence, using various techniques, has documented these relationships at the level of the country, the sector and the individual firm and from the investor perspectives (Claessens, 2003).

There are two basic principles of corporate governance (Charkam, 1994). The first is that management must be bale to drive the enterprise forward from undue constraint caused by government interference, fear of litigation, or fear of displacement. Second, is that this freedomto use managerial power or patronagemust be exercised within a framework of effective accountability. Essentially, corporate governance failures may come about for two broad reasons. One, management may operate the firm inefficiently, resulting in an overall decrease in firm profits, compared to the potential profitability of the firm. Two, while management may operate the firm efficiently and generate maximum profits, they may divert a proportion of those profits from shareholders via the consumption of excess perquisites, for example, through remuneration which is not related to performance. Hence, a system of corporate governance needs to consider both efficiency and stewardship dimensions of corporate management.

Stewardship emphasizes issues concerning, for example, the misappropriation of funds by non-owner manager. Equally important, however, is the issues of how the structure and process of governance motivate entrepreneurial activities which increase the wealth of business (Short et al, 1998).

The relationship between corporate governance and performance is based on the principal agent approach. The agency relationship is defined as a contract under which one or more persons (the principal(s) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent. The separation of ownership and control, which occurs as a result of the introduction of external investors, bring to fore the agency problem: managers are expected to represent the interests of the external owners of the enterprise; however, it is difficult for owners to ensure that managers do so. Shleifer and Vishny (1996) argue that managers and equity investors should be capable of entering into a binding contract, which would ensure that investors' interest are fully represented.

However, it is unlikely that it will be possible to specify contracts ex-ante that accommodates all possible future contingencies. If unforeseen circumstances arise, managers assume contingent control rights that proved them with the potential to operate against investors' best interests, by, for example, expropriating investors' funds or engaging in assets stripping. The discretionary control rights of managers are further increased by the existence of asymmetric information between themselves and external investors. Although it is precisely this insider knowledge that encourages investors to permit mangers to operate as their agent, this allows managers the freedom to conceal information from external investors.

Such action serves to increase the costs of monitoring and therefore enables managers to pursue their own goals rather than those of the shareholders, by entrenching their position or engaging in behaviour that is sub-optimal for the shareholder. The possibility of higher monitoring costs is particularly strong if there are a large number of dispersed external investors, because a free-rider problem emerges if there are large costs to monitoring which the benefits accruing to each individual are relatively small.

Metrick and Ishii (2002) identify four dimensions of corporate governance at the level of the firm that can help to minimize the agency probable: board of directors, ownership structure, executive incentive contracts, charter and bye law provisions.

(a) Board of Directors: This is often considered to be one of the major sources of monitoring firm's conducts and performance. It is responsible for hiring and firing executives, setting executive compensations and making key decisions in the life of the firm. The board of directs should in principle be one of the major checks one the management. It is directly elected by the shareholders to act on their behalf. A high level of independence is important for it to perform its monitoring duties more effectively.

The standard view is that the board of directors is more independent as the number of outside directors' increases. Executive directors are not likely to self monitor effectively the performance of the CEO because their career is closely tied to the incumbent CEO (Jensen 1999). Several studies show that board membership is related to the degree of agency problems at a firm (see for example, Borokhovich, Parrino and Trans (1996), Weisbach (1988) and Rosenstein and Wyatt (1990). The larger the percentage of outside directors, the more likelihood of (i) an outside executive being appointed chief executive officer (CEO) (ii) a non-performing, CEO to be dismissed and (iii) significant positive share reactions. With respect to the size of the board and performance, Yermack (1996) provides evidence of a negative relationship between the size of the board and firm value. However, Hermalin and Weisbach (1999) found no significant relationship between board composition and performance while Yermack also shows that the percentage of outside directors does not significantly affect firm.

Jensen (1993) and Hermalin and Weisbach (1991) argue that CEOs often control the composition of the board and lessen its monitoring role. This is especially possible when a person combine the position of chairman and CEO, and the use of exclusively large boards which increases communication problems among board members. Yermack (1996) report a negative relationship between board size and firm value for large and small firm respectively.

(b) Ownership Structure: This is another method of mitigating agency problems. The free-rider problem is minimize and internal constrains on managerial discretion can probably be imposed if ownership is concentrated in the hands of a large block of shareholders irrespective of whether they are individuals, organizations or investment funds. In this event, the returns to monitoring will increase monitoring activity, which may also be subject to economies of scale.

Moreover, large shareholders will be more likely to be able to utilize their voting power to influence managerial behaviour, although, as Shleifer and Vishny (1996) note, this does require shareholding voting rights. This leads to the proposition that large shareholders will exercise more effective corporate governance; a finding that has been supported by a host of studies on developed market economies. For example, Franks and Mayer (1994), in a study of German Private Enterprises find that concentrated share ownership is associated with high rates of turnover of directors. In the study of Japan, Kaplan and Milton (1994) find that the existence of large shareholders raises the probability that managers of poorly performing firms will be replaced. La Porta et al (1999) posit that high concentration could minimize agency costs since it could serve as a substitute for legal protection. -Even without strong legal institutions, large investors have the means and the incentives to monitor managers, large investors have the means and the incentives to monitor managers, though they bear the cost of undiversified risk‖. However, the cost here is that large shareholders may use their control rights to expropriate minority interests.

Nevertheless, there is no consensus yet as to the impact of ownership concentration on performance. In some countries, such as Austria, the Netherlands and Spain, companies with dispersed ownership perform inadequately than those with concentrated shareholdings, while in others the reverse seems to be true (Gugler 2001). On the contrary, Holderness and Sheehan (1988) find little evidence that high ownership concentration directly affects performance. The composition of ownership may also matter for performance. Institutional investors have been very active in the firm level corporate governance.

Frydman et al (1997) examined the impact of private ownership on corporate performance in the transition economies. The study reports that private ownership dramatically improves the most essential aspects of corporate performance in the countries undergoing post-communist transition. Furthermore, the study also reports that outsider-owned firms perform better than insider-owned firms on most performance measures. Jensen and Meckling (1970) suggest that agency costs can be reduced through the concentration of ownership and control within one single owner-manager. However the possibility of interplay between incentive alignment effect and entrenchment effect suggest a non-monotonic relationship between managerial stock ownership and firm value.

(c) A third mechanism through which shareholders can induce managers to behave efficiently is incentive contracts which tile managers' compensations to measures of corporate performance. This can be accomplished through performance related bonuses, stock grants and stock options. However, executive incentives pay has been criticized as being manipulated or controlled by the executive themselves. Jensen and Murphy (1990) examine the link between pay and performance for CEOs in the U.S. They argue that the conflict of interests between the shareholders and CEO represent a classic example of principal-agency problem. Agency theory predicts that compensation policy will be designed to give manager incentives to select and implement actions that increase shareholders wealth‖.

(d) Finally corporate charter and bye law provisions are an important source of governance. Federal and State laws containing provisions that establish firm level rules for a variety of areas such as shareholders voting, managers and directors liability and takeovers. State laws that provide takeover protection may increase agency costs (Bertrand and Mullainathan 1999).

Klapper and Love (2002) examine corporate governance and performance in a sample of firms in 14 countries, most of which are developing economies. They find that better corporate governance is associated with better performance in the form of Tobin's q and Returns on Asset (ROA) and that good governance seems to matter more when the legal environment of a county provides investors with weaker protections. John and Senbet (1998) provide a comprehensive review of the Stakeholders theory of corporate governance. The main issue raised in the theory is the presence of many parties with competing interests in the affairs of the firm. They also emphasized the role of non-market mechanisms such as the size of the board, committee structure as important to firm performance. Jensen (2001) critique the stakeholders theory for assuming a single-valued objective. He proposed an extension of the theory called an enlightened stakeholder theory. However, problems relating to empirical testing of the extension have limited its relevance (Sanda et al 2003).

Although the empirical works in the general areas of corporate governance have grown considerably, not much has been documented on the being industry in Nigeria. Lack of consensus on how toe resolve the agency problem has produced a variety of mechanisms on how to deal with it, such mechanisms include: striking a balance between outside and inside directions; promotion of insider (i.e. mangers and directors) shareholding; keeping the size of the board low; and encouraging ownership concentration.

Studies on corporate governance in Nigeria include Adenikinju and Ayorinde (2001), Oyejide and Soyibo (2001), Alshi (2002) and Sanda et al (2003), Sobodu and Akiode (1998) examined managerial efficiency in the banking industry. The study focused on the managerial efficiency of the banks using Data development Analysis (DEA) approach. Managerial efficiency is measured in operating expense to total assets. There are several problems associated with the measurement of banks' operational efficiency. First, there is the problem of identifying banks' inputs and outputs. Second, though not peculiar to banking, is the existence of several heterogeneous inputs and outputs that cannot be easily compared. Besides, rates of return, instead of operational efficiency, are most often used by investors to appraise the performance of their investments.

In the study by Adenikinju and Ayorinde (2001), the implication of ownership structure and control (governance) on the performance of publicly listed companies (excluding banks) in Nigeria was investigated. Banks were excluded because that is regarded as income in the banking sector is a liability in other sectors and vice versa. Also, the study by Oyejide and Soyibo (2001) reviews and analyses the practice and the standard of corporate governance in Nigeria. Their review of the legislations on corporate governance and the analysis of its standard between 1995 and 1998 show clearly that the institutions and the legal framework for effective corporate governance exists in Nigeria. However, compliance and/ or enforcement appear to be weak of non-the other hand, investigated the efficiency of corporate governance mechanism as a means of increasing firm financial performance between 1996 and 1999, in its analysis of 93 firms quoted on the Nigerian stock exchange, the study sampled 10 banks, a figure not reprehensive of the banking industry. Its findings show that small boards perform better than large boards, although it does not state what an optimal board size should be.

The conclusion of the literature reviewed that corporate governance has been variously defined by different authors, and the relationship between corporate governance and performance is board membership is related to the degree of agency problem at a firm. Good corporate governance can mean generally better relationships with all stakeholders. This helps to improve social and labour relationships and aspects such as environmental protection.

IV Analytical Framework

Corporate governance, as a concept, can be viewed from a narrow and broad perspective. The narrow view perceives corporate governance in terms of issues relating to shareholder protection, management control and the popular principal-agency problems of economic theory. In contrast, the broad perspective notes that issues of institutional, legal and capacity building and the rule of law are important to corporate governance. The theoretical framework upon which this study is based is the stakeholder theory, a modified version of the agency theory, which posits that in the presence of information asymmetry, the agent (in this case, the directors and managers) is likely to pursue interests that may hurt the principal or shareholder and other stakeholders (Ross, 1973;Fana, 1980;John and Senbet, 1998). Thus, corporate governance is a means by which various stakeholder exert control over a corporation by exercising certain rights as established in the existing legal and regulatory frameworks as well as corporate byelaws.

John and Senbet (1998) classify agency problems on the basis of conflicts among particular parties to the firm, such as conflicts between stockholders (principals) and management (agent) (managerial agency), between stockholders (agents) and bondholders (debt agency), between the private sector (agent) and the public sector (social agency), and even between the agents of the public sector (e.g regulations) and the rest of the society or taxpayers (political agency). They noted that agency problems detract from efficient operation of an enterprise. Departures from efficient investment strategies are detrimental to financial environment that fosters efficient corporate governance and efficient contracting among parties with diverse interests, promotes efficient allocation of resources, and hence ultimately economic development.

The existence of agency problem is potentially harmful to the owners of the firm and may lead to inefficiency and wealth destruction in an economy. It is in the best interests of owners to resort to control mechanisms that move the operation of the firm toward full efficiency of the Fisherian principle. This approach that attempts to align the interest of managers and all stakeholders it known as the stakeholder theory. The stakeholder theory, as discussed by John and Senbet (1998), emphasizes the role of non-market mechanisms, citing as an example, the need to determine an optimal size of the board of directors especially in view of the tendency for board size to exhibit a negative correlation with firm performance. Other non-market mechanisms include the need to design a committee structure in a way that allows the setting up of specialized committees with different membership on separate critical areas of operations of the firm. Such a structure would allow, for example, the setting up of productivity-oriented committees and monitoring-oriented ones.

In an extension of the stakeholder theory, Jensen (2001) also recognizes the multiplicity of stakeholders. He agreed with John and Senbet that certain actions of management might have conflicting effects on various classes of stakeholders. This implies that the managers have multiplicity of objective functions to optimize, something that Jensen sees as an important weakness of the stakeholder theory -because it violates the proposition that a single-valued objective is a prerequisite for purposeful or rational behaviour by any organization‖ (Jensen 2001). In search of a single valued objective function that conforms to rationality, Jensen suggests a refinement of the stakeholder theory-the enlightened stakeholder theory. For him, the enlightened stakeholder theory offers at least two advantages.

First, unlike the earlier version with multiple objectives, the modified form of the theory proposes only one objective that managers should pursue: the maximization of long run value of the firm. If the interest of any major stakeholder were not protected the objective of long run value maximization would not be achieved. A second, related appeal of the enlightened stakeholder theory is that it offers a simple criterion to enable managers decide whether they are protecting the interests of all stakeholders: invest a dollar of the firm's resources as long as that will increase by, at least, one dollar the long term value of the firm. There is an important caveat, however -Jensen himself cautions that the criterion may be weakened by the presence of monopoly situation or externalities. Despite its appeal, Sanda et al (2003) note that the stakeholder theory of the variety proposed by Jensen has not been subjected to much empirical evaluation. At least two factors might have contributed to the gap between theory and evidence. The first concerns the prevalence of externalities and monopoly situation. The second is the problem of measurement. Jensen himself offers no clue on how to obtain an accurate measure of the long-term value of the firm, let alone offer an indication of how to assess the possible impact of an investment on that long term value.

Model specification

Following the studies carried out by Miyajima et al (2005), and based on the method of data collection and analysis, we discovered that there is similarity between their works and this study. Consequently, this forms the basis of the adoption of the analytical framework for this study. We therefore have: TQ = γ 1 bs + γ 2 out + γ 3 drs + γ 4 blk + γ 5 aud + γ 6 debt + γ 7 size + u t -----------------------------------(1) ROA = γ 1 bs + γ 2 out + γ 3 drs + γ 4 blk + γ 5 aud + γ 6 debt + γ 7 size + u t -----------------------------------(2) The variable definitions are given in Table A1 in the appendix. Equations (1) and (2) are specified to capture the industrial fixed-effect and random effect while the time fixed-effect is ignored for two reasons.

V Empirical Analysis

We start by examining the effects of internal control mechanisms (director shareholding, board size, ownership concentration, and outside directors, block holders, independence of audit, leverage and firm size) on firm performance. The results are presented in Table3, the table shown the results obtained by regressing the governance mechanisms on an important measure of firm performance, ROA. Both director shareholding and board size show no significant relationship with return on assets. Table 3, there is a positive and significant relationship relationship between board size, block shareholders, leverage and firm size and the dependent variable Tobin's Q. For example, A 1% increase in B.S will lead to 0.14% increase in TQ while a 1% increase in BLK will lead to 0.05% increase in TQ. However, the empirical result in Table 3 reveals an inverse relationship between director's shareholdings, size, independence of the audit committee and the numbers of outside directors on board. A similar result however emerged when the return on asset (ROA) was used as the dependent variable as presented in Table 3. The result supports the existence of the positive relationship between board size, block holders, leverage and return on asset. However, there was a negative relationship between the number of outside directors on board, director's shareholdings, independence of the audit committee and the return on asset.

However, our result did not support Adenikinju and Ayorinde (2001), who found no significant relationship between firm performance and insider ownership in Nigeria. Perhaps, the conflicting results could be due to the differences in the methods used in measuring some of the variables as well as the sample size. For example, in computing directors shareholdings, we included only the shareholding of directors while they included those of directors and all other staff of the firms. Inadequate data did allow us to do this. The results of the random effects model is presented in Table 3. The result in Table 5 reveals a positive relationship between board size, number of outside directors, block holders, leverage and the Tobin's Q measure of firm performance. However, there is a negative effect of director's shareholding and independence of the audit committee and the Tobin's Q. However, using the return on asset as a measure of firm performance, we found a positive relationship between board size and block holders and the dependent variable. In addition, a negative relationship was reported in the case of number of outside directors, director's shareholding, independence of the audit committee, leverage and firm size and the return on asset.

Taking a synopsis of this result, both the fixed effect and random effect models reveals that there is a mixed result with result to the performance of some of the governance variables. Contrary to studies by Jensen (1993), Lipton and Lorsch (1992), Yemarck (1996), this study show that the larger the size of the board (BS), the better the Tobin's q. This explains the view that larger boards have better corporate performance because members have a range of expertise to help make better decisions, and that it is difficult for the chief executive officer (CEO) to influence the decision of the board. The board size is highly significant in explaining Tobin's q for firms in Nigeria.

Similar to the board size, the board composition (OUT) has a negative relationship with Tobin's q implying that when there are more external board members, performance of the firm tends to be worse. This contradicts the empirical study of Brickley and James (1987) that outside directors support the beneficial monitoring and advisory functions to firm shareholders. However, this is consistent with findings by Agrawal and Knoeber (1996) who suggest that boards expanded for political reasons often result in to many outsiders on the board, which does not help performance. It must rather be indicated that this variable is not significant.

Our results also show that leverage has significant positive influence on firm performance, indicating the tendency for firms with higher levels of debt as a proportion of equity to perform better, a finding that is consistent with the literature. We also found that the concentration ratio has negative impact on performance, therefore the directors shareholding can improve on their effort in order to exert positive impact on performance. These seemingly contradictory finding seems to suggest that concentration ratio is positively related to performance up to a point beyond which it has a negative impact. In other words, excessively high concentration may lead few shareholders to use their positions to benefit only themselves. The policy implication flowing from this finding is self evident. Firms must motivate their chief executive officers (CEOs) in order to encourage them to deliver good returns on the shareholders' investments. It is also imperative that the salary and other perks attached to the position of the CEO if tied to performance indices will be a useful tool in the hands of shareholders/stakeholders in ensuring greater overall company performance.

VI. Summary and Conclusion

There has been a renewed interest within academic circles as well as amongst policy makers in both government and industry on the need to strengthen mechanisms to ensure that managers and directors take measures to protect the interest of a firm's stakeholders. The events at Enron and other cases of spectacular failure have helped to bring to the limelight the important role that the strengthening of governance mechanisms could play to improve firm performance. This study made use of panel data regression analysis between 2002 and 2006 for a sample of 62 firms listed on the Nigerian Stock Exchange to examine the relationship between internal governance mechanisms and firm financial performance. The results have the implication that regulatory agencies should encourage firms to achieve a reasonable board size since overly large boards may be detrimental to the firm.

The results of the study point to the need for a reasonable number of individuals and/or corporate bodies with more than a typical share of equity of the firm as this will encourage them to undertake the monitoring process. Unlike the findings in developed countries, our results show no significant evidence to support the idea that outside directors help promote firm performance. This suggests the need for the regulatory authorities to reassess the procedures for the appointment of outside directors in order to remove the influence of CEOs from the appointment process. In addition, the study found that the measure of performance matter for analysis of corporate governance studies. We found in some cases different results from the use of Returns on Assets (ROA) and Tobin's Q as measures of firm performance. Furthermore, we also found the type of governance environment a firm operates also has implications for its overall performance, as well as on the directional and quantitative impact of managerial characteristics on firm performance.

In spite of the findings in this study, there are many issues in corporate governance in Nigeria that remain unresolved. The dearth and poor quality of data continue to be a major constraint in a comprehensive study of corporate governance in Nigeria. It is common knowledge that there are some margins of error in Nigeria statistics. However, it should be noted that the main general trends and findings that have been disclosed in this study are so pronounced that potential data biases have to be very large indeed to reverse them. Nevertheless, data on some specific variables that would have made the study more interesting were unavailable. A number of points could be classified by further work, which could give greater specificity to policy guidelines. The sample itself was determined by data availability, not by a probability criterion.

References

Introduction

Financial markets play a crucial role in facilitating the intermediation between savers and investors, thereby helping translate savings into investments. The more efficient this process is, the less the cost of investing, and subsequently, the higher the rate of investment/saving. The development of stock exchanges is crucial to achieve economic growth for developing economics. The increasing globalisation of financial markets has heightened interest in emerging markets. However, much of the research in accounting and finance has focused on developed markets, in particular, the US and European markets. The assumptions which underpin the models employed in developed markets provide a challenge when examining emerging markets such as Jordan. In an uncertain economic environment, characterized by informational asymmetry, the announcement of changes in corporate finance variables (e.g., dividends) are often regarded as signals emitted by company management which have to be interpreted by outside investors in developed countries. Since the documented evidence suggests that these announcements convey private information to the market, numerous signalling models have been developed to explain the accompanying share price reaction to such announcements.

Since its establishment in 1978, Amman Stock Exchange has experienced a remarkable development and it has become one of the most active and organized market among the emerging markets. It plays an important role in financing development in Jordan. Jordan's capital market was increasingly viewed as a critical component in the economic development plans of the country. The Jordanian government adopted a comprehensive capital market reforming policy, which aimed at building on the previous 20 years' experience, boosting the private sector, expanding and diversifying the national economy, and improving regulation of the securities market to international standards. Among the most important features of the new orientation were institutional changes in the capital market, use of international electronic trading, settlement and clearance systems, elimination of obstacles to investment, and strengthening capital market supervision to reach optimum transparency and safe trading in securities. Although many studies have been conducted on dividend signalling and information content of dividend in developed markets, there is no such comprehensive study found about the effectiveness of the dividend announcement as a signalling device to influence the security prices of an emerging market. Therefore, the existing published evidence is of limited relevance in investigating the market reaction to dividend announcement and in identifying the appropriate dividend policy and behaviour in the emerging market, and still these issues of market reaction to dividend announcement and dividend policy and behaviour of an emerging market remain unresolved.

This study will contribute to the literature by providing empirical evidence of the market's reaction to dividend announcements in Jordan. Information content studies provide the opportunity to understand the markets' assessment of dividend payments, and consequently, to help for a better understanding of the dividend policies of Jordanian firms. This is important for investors, regulators, and management. The objective of this study is to investigate the market reaction to dividend change announcements on the Amman Stock Exchange. Specifically, how the market reacts when the change in dividend payment from the previous year is positive, negative, and unchanged. This objective will be achieved by reviewing the literature of the signalling hypothesis in Jordan, applying the event study methodology to measure the market's reaction to the change in dividend payments on the announcement date.

The remainder of this study is organised as follows. Section 2 presents literature review. Research methodology is described in section 3, while section 4 and 5 present hypotheses development and sample selection, respectively. Section 6 discusses the empirical results and section 7 provides the conclusion, while section 8 presents suggestions for further research.

In this study, we examine the association between insiders' trading before seasoned equity offerings (SEO) by directors and senior officers and the proceeds of the SEO, given that firms inflate earnings. The seasoned equity offerings (SEO) event is an important event in the life of the firm. Some firms conduct SEOs to finance working capital and to prolong their survival, while others use the infusion of capital to finance expansion. Naturally, the interests of the company and its incumbent shareholders are to maximize the proceeds of the SEO. For this reason, firms try to time the SEO to occur when the stock is overvalued, 2 obtain a higher valuation. 3 To the extent that good corporate governance requires -the board and management to pursue objectives that are in the interests of the company and its shareholders‖ (OECD p. 3), insider selling seems to be a poor governance practice, since such sales signal to the less-informed but suspicious market that the stock is indeed overvalued. 4 What is more puzzling is that some of the insiders are directors, whose role in the corporate governance is to monitor management and to watch for the interests of shareholders. If they so wish, the boards can ban insider trading before the SEO. We examine earnings management and insider trading around seasoned equity offerings (SEOs) by a sample of 233 firms out of a population of 791 firms that conducted SEOs in the 1987-2005 period at least once. We first examine the pattern of insider trading and earnings management, arguing that firms who inflate earnings induce their insiders to sell shares and profit. We then check the market's response around the issuance of the shares given the observable insider trading signal and the earnings management activity of the firm, assuming that as a rational player, it discounts the insiders' sales and the managed earnings.

The majority of firms do not have insider trading. An examination of the firms with insider trading by directors, senior officers, and/or blockholders, shows a different pattern of earnings management inter-temporally: Firms whose insiders sell manage earnings more aggressively than firms whose insiders purchase shares and hence, the former (latter) have negative (positive) abnormal accruals at the year the firm conducts the SEO and in the following year. This pattern is consistent with the incentives of the traders to move the price in the direction that increases their wealth. Sellers would like to inflate earnings to increase the price they sell the firm's stock, hence they -borrow‖ reported earnings from future periods more aggressively. Buyers prefer the price before the SEO to be as low as possible to increase their wealth when they sell shares in the future. They also manage earnings upwards to inflate the stock price of the SEO, but not that aggressively.

Our major findings are as follows. We find that insider selling is negatively associated with the cumulative abnormal returns around the issuance of the SEO. When we consider insider trading by both directors and senior officers (in 85% of the subsample of 233 firms), we find that the market's discount is lower. That is, the market views insider trading by directors as a favorable signal that the trading is innocuous in that it is not driven by a collusion between directors and managers to manager earnings and make profitable trading gains at the expense on investors.

This study makes several contributions. First, we augment scholarship by examining empirically the theoretical papers by Ronen, Tzur, and Yaari (2006,2007). These two studies advance the notion that directors who wish to make insider trading gains, do not take steps to curb the misleading earnings management by privately-informed managers. Our findings provide some support to this theory since we find that the market does not discount the earnings management by firms whose insiders sell stock despite the fact that these firms manage earnings more aggressively. This finding is consistent with the disincentives of the directors to expose the true magnitude of earnings management, and explain why insider trading is profitable.

Second, we contribute to the governance literature by focusing on the directors' role in the insider trading around the SEO event. Hillier and Marshall's, 2002, finding that directors' trades is not always informative raises the question of why do they trade then. One explanation in the US context is that the trading is not voluntary. That is, the insider trading laws that are enforced by the Security and Exchange Commission; induce insiders to make trading plans wherein they commit to sales' volume on a quarterly basis as a means to avoid the charge of illegal trades.

Our results provide another explanation.

Directors can fulfill their role in governance either by banning trades, or by participating it to signal the market that the trade is not driven by opportunistic motivation.

This study proceeds as follows: Section 2 presents our hypotheses. Section 3 presents the sample selection and the methodology. Section 4 presents the results, which are concluded in Section 5.

Corporate governance has been much debated in many arenas in recent years. This has been brought back into the limelight by the current global economic crisis. The purpose of this article is to review and evaluate the current position with regard to the development of corporate governance in a recent accession member state to the EU -Poland.

This article is divided into two parts: the first presents a brief overview of the main theories and models of corporate governance. Traditionally, these are categorised into two main camps: the shareholder or stakeholder. It is not the purpose of this article to review whether this static approach to corporate governance is appropriate, effective or realistic (for example, see Griffin and Mahon, 1997; Prabhaker, 1998;Friedman and Miles, 2002). Rather, the position in Poland will be examined to see which of these 'traditional' approaches, if either, is being adopted within Poland.

It will be posited that Poland is not 're-inventing the wheel'. That it is, in fact, drawing from these existing approaches. Conclusions will be offered as to the extent to whether this is necessarily the best way forward for a developing country in a relatively young, free market economy. That one cannot ignore corporate governance regimes internationally is noted by Detomasi, (2006, p.225) who states: "It is difficult to avoid the topic of corporate governance". This is also the case domestically in order to understand processes of regime change and transformation (Roe, 2003;Gourevitch and Shinn, 2005). But this is especially the case in emerging markets, where national systems of corporate governance are not as well institutionalised and where the costs of corporate governance failure are very high. This can be seen in the economic consequences of the East Asian financial crisis of the late 1990s which resulted in the five most heavily affected countries losing more than 60 per cent of their combined gross domestic product (Schwab, 2003).

For a country such as Poland a new accession state in the European Union, the pressure to converge in financial regulation and corporate governance so that it can compete for investors and capital with established markets is significant. According to Reed (2002, p.223) corporate governance reforms in developing countries "occur in a larger context that is primarily defined by previous attempts at promoting 'development' and recent processes of economic globalisation".

One would assume for a topic which has been thoroughly debated and examined in many fora for many years that the concept of corporate governance would be clear. It still seems to be though a concept discussed in terms of generality as we shall see below. Nonetheless, it is accepted that corporate governance is not simply for the benefit of companies themselves but also for the whole market and society (Mueller, 2006, .p 207).

In light of this, the next section will consider the 'traditional' models from which to choose.

Traditionally, it is assumed that corporate governance is based either on the shareholder model or the stakeholder model.

The shareholder model is most common in 'liberal market economies' such as the USA and the UK. The stakeholder model a feature of more 'coordinated market economies' like Japan and Germany (Hall and Soskice, 2001). This split is also referred to as:

'stock market capitalism' versus 'welfare capitalism' (Dore, 2000) and even 'Anglo-Saxon' versus 'Rhineland' capitalism (Albert, 2003). An interesting recent development in this debate has been to describe this split as being 'market-centred' or 'bank-centred' (Allen and Gale, 2000).

This approach has been criticised as being 'too simplistic' (La Porta et al, 2000) and a more appropriate way of explaining this distinction being to consider the extent to which investors enjoy legal protection (La Porta et al, 1999).

A further recent debate has struck at the very essence of the assumptions above. That is, that whilst traditionally the UK and USA are seen as one 'Anglo-American' Model there are now fundamental differences being identified within this Model, (Aguilera et al, 2006;Toms and Wright, 2005). It is suggested that whilst there are differences in the regulatory methods and approaches adopted by each country this is more to do with the enactment of legislation in the wake of recent US corporate scandals. But, whilst this distinction is not significant in terms of the 'social purpose' of the two traditional Models it does represent an interesting distinction particularly for emerging markets undergoing corporate governance reform. We will return to this discussion subsequently.

Literature Review: The Signalling Model

The assumptions of Modigliani and Millers' theory (1959) stated that the current value of the firm is independent of the method of financing, and the firm's main objective is to maximise shareholders value. Thus, the market value of a firm is not affected by its dividend policy. They assumed perfect capital markets, rational investors' behaviour and no tax discrimination between sources of income.

Miller and Modigliani (1961) argued that in a world without any market imperfections like taxes, transactions costs or asymmetric information, a firm's dividend policy should have no effect on its market value. An important assumption to this argument is the independence of firm's investment policy from its dividend policy. Hence, the irrelevance argument holds only if investments decisions are not influenced by management's insistence on maintaining or raising the firm's dividend.

Accordingly, the market imperfection of asymmetric information is the basis to explain corporate dividend policy. The mitigation of the information asymmetries between managers and owners via unexpected changes in dividend policy is the cornerstone of dividend-signalling models. Ross of signalling theories believe that corporate dividend policy can be used as a means of putting the message of quality. Dividends have a lower cost than other alternatives. The use of dividends as signals implies that alternative methods of signalling are not perfect substitutes.

One of the important implications of this signalling argument is that it suggests the possibility of optimal dividend policy. The signalling benefits from paying dividends may be traded off against the tax disadvantages in order to achieve an optimal payout. Bhattacharya (1979) suggested that, if stockholders have imperfect information about firms' profitability, and if there is a tax rate differential between capital gains and dividends, then dividends will be a surrogate for a signal of expected cash flows. John and Williams (1985) developed Bhattacharya's signalling model in the context of a tax penalty on dividends over capital gains. Corporate insiders with more valuable private information optimally distribute larger dividends and receive higher prices for their stock whenever the demand for cash by both their firm and its current stockholders exceeds its internal supply of cash.

Ofer and Thakor (1987), Ambarish, John and Williams (1987), Bar-Yosef and Huffman (1988), John and Lang (1991), Hausch and Seward (1993) and Noe and Rebello (1996) present further analysis of dividend signalling models. These models extend earlier approaches by focusing on firms that signal simultaneously with dividends, investment or, equivalently, dividends and stock repurchase. These multiple signal models recognise that the change in the firm's dividend policy cannot be evaluated independently of other management decisions or firm characteristics. Although these signalling models provide important insights into why firms signal via dividends, and consequently why the market reaction to an increase in dividends is positive and to dividend decrease is negative, they rely on restrictive assumptions (for example, John and Williams model assumes all equity financing and homogeneous expectations by investors). However, these models do not resolve the central issue of what information management signal (or hope to signal) to the market via dividend announcements. So, the question is why the dividend increase is always treated by the market as good news and the dividend decrease as a bad news.

One of the primary literatures that investigated the stock price reaction to dividend announcements was conducted by Pettit (1972). Therefore, number of empirical studies followed the theoretical work of signalling models mentioned by Bhattacharya (2001)). These debates based on the idea that insiders of a firm have better information regarding the firm's earnings potential than do market participants, and that to maximize share holder wealth, insiders reveal this information to the market by taking some observable action. The market subsequently uses this new information to update its expectations regarding the firm's earnings prospects, and as a result of this process a new stock price is determined. The existence of asymmetric information in the market makes dividends -relevant‖, and changes in dividends are therefore a source of information about a firm's earnings potential.

Two studies examine the effect of dividend changes announcements on emerging markets. Mollah (2001) investigated whether dividend announcements conveyed information to the market or whether investors considered the announcement of dividends as a signal of firms' future prospects in Bangladesh. He applied the event study methodology for 153 firms over 1988-1997, and his final sample was consisting of 380 cash dividend announcements amongst 213 dividend increasing announcements, 84 dividend decreasing announcements, and 83 dividend maintaining announcements. He did not find a significant impact of dividend announcement on the security prices and further no evidence to support the dividend signalling hypothesis. Furthermore, he found a similarity among his samples when he found that security prices is decreasing after increasing dividends, decreasing dividends, and maintaining dividends, which he indicated it as an ineffectiveness of the announcements of dividends in Bangladesh. He mentioned that insiders are holding higher proportion of stocks, so, usually insiders start to buy back the shares before the general assembly meeting for higher voting rights, moreover, insiders off load shares after the general assembly meeting start to sell their shares and that causes higher supply of shares and consequently returns fall. Moreover, Mollah (2001) studied the determination of dividend policy in Bangladesh. He found that leverage, size, insider ownership, and collateralizable assets are the major determinants of dividend payout ratio in Bangladesh. While supporting the agency cost hypothesis and the transaction cost hypothesis, his analysis of the determination of dividend policy did not support the signalling hypothesis, the tax clientele hypothesis, the residual hypothesis, and the pecking order hypothesis.

Research Methodology

One of the quantitative methods that have been used widely in conducting the dividend change effect is event methodology and it is one of the most frequently used statistical techniques in the applied financial research. Initial event studies emerged in the United Stated three decades ago (Ball and Brown, 1968; Beaver, 1968; Fama, Fisher, Jensen, and Roll 1969; and Kaplan and Roll, 1972), and have continually been extended (Brown andWarner, 1980, 1985;Blume and Stambaugh, 1983;Dyckman et al, 1984;Morse, 1984;Schipper and Smith, 1986;and Fama and French, 1992). Event studies are used to measure the impact of an economic event on firm value. Assuming that the event will be reflected in traded asset prices, these studies focus on how asset prices respond to information releases during a public announcement of the event.

The objective of an event study is to measure whether there are any abnormal or excess returns earned by security holders accompanying special events (e.g., dividend announcements). The results of event studies not only can have serious implications in forming accounting or government policy, but also provide an important source of market information to both individual and institutional investors (Beaver, 1988). On the other hand, event studies can also be used to study market efficiency on the semi-strong form by analysing the market reactions to these ‗events' (Firth, 1979). The event study methodology has become popular because it obviates the need to analyze accounting-based measures of profit, which have been criticized because they are often not very good indicators of the true performance of firms. For example, managers can manipulate accounting profits because they can select accounting procedures (Benston, 1982). Stock prices on the other hand, are not as subject to manipulation by insiders. Stock prices are supposed to reflect the true value of firms, because they are assumed to reflect the discounted value of all future cash flows and incorporate all relevant information. Therefore, event studies, which are based on stock price changes, should measure the financial impact of a change in corporate policy, leadership, or ownership more effectively than a methodology based on accounting returns. Furthermore, the event study method is relatively easy to implement, because the only data necessary are the names of publicly traded firms, event dates, and stock prices. The major elements in conducting event study methodology is, specifying the event date, specifying the estimation and event period, and calculating the daily returns. To achieve the research question, we will adapt the event study methodology for the above reasons and for the following: (1) this method calculates abnormal return by differentiating the actual returns and expected returns calculate based on previous performance (2) Abnormal returns are the best reflection for the announcements (3) As the Amman stock exchange is an emerging market, it is important to conduct event study on a longer period, which this method helps to do. This study explains the Event Study Methodology which includes specifying the event date, the estimation period, the event period, and the equations and test statistics which will be applied.

Specifying of Event Date, Estimation period and Event period

The event date is often recognised as the release date of new information to market participants through the financial press or through corporate release. The identification of event date is of prime important because the correct event date can provide a powerful test of the information content of announcements. To investigate the market reaction to dividend change announcements on Amman Stock Exchange, we will use the general assembly meeting date to consider the event date. According to the Jordanian company law 1997, article 190 A and B, which reveals that the right entitling shareholder to obtain their shares of the company's profits accrues on the date of adopting a resolution by the general assembly regarding distribution of dividends and the right to receive the profit against the company for the shareholders shall be on the date of the meeting of the general assembly on which it decides to distribute its profits. According to this, we see that the most important date in which the dividends are officially approved and the public know about the dividend decisions are notified on this day. Furthermore, on the general assembly meeting, the shares are not allowed to be traded on the market according to the listing securities regulation on Amman Stock Exchange (Article 26-a of the Securities Law no. 23 of 1997). So, the following day of the GAM day is the announcement day.

The estimation period is used for estimating the parameters of the benchmark expected return. This allows predicted abnormal returns to be calculated within the test period. Most of event studies establish the estimation period prior to the event period. This is mainly based on an assumption that the normal or predicted returns over the estimation period will be independent with the release of the events if there is no leakage of the information before the announcements.

Brown and Warner (1980,1985) and Peterson (1989) mentioned that daily returns are more powerful than monthly returns. Morse (1984) supported the use of daily return data to estimate information effects, with the possible exception of cases in which there is uncertainty about the date of the information release. Morse (1984) mentioned that even with this uncertainty, however, daily returns may still be preferred to monthly returns. His analytical results were consistent with some empirical results in Brown and Warner (1980,1983) and Dyckman, Philbrick, and Stephan (1982). Because most of the previous researches employed daily data and to allow comparing the results with the previous ones, this study will employ the daily rather than the monthly data. This study will use 130 trading daily observations as the estimation period from day T=-21 to Time 1=-150 before the event period.

Selection of the length of the event period is also subjective. Typical lengths of the event period range from 21 to 121 days for daily studies and from 25 months to 121 months for monthly studies (Peterson, 1989). However, the calculated abnormal returns over a narrow event period around announcements might understate the usefulness of information, if the period fails to capture information-induced price revisions beyond the period. On the other hand, abnormal returns calculated by using wide event period, might overstate the information contribution of announcements, as price changes within the period probably reflect investors' reaction to other timely information published (Lev, 1989). This study will use 41 trading days, covering the period T+1=-20 to T+m=+20.

Calculating Daily Returns and Return Generating Models

The return is computed as the natural logarithms of the stock price relatives given by the equation below.

The daily stock returns for security i at day t. As a consequence of non-synchronise trading, OLS estimates of the market model parameters will be biased and inconsistent resulting in biased estimates of abnormal returns and consequently mis-specified test statistics in event studies. To take the account of this problem, Scholes and Williams (1977), Dimson (1979), Flower and Rorke (1983) and Cohen et al (1983), provided methods to remove a greater deal of bias from beta. The Scholes-Williams and the Fowler-Rorke methods are the only two thin trading adjustment methods that will be employed in this study.

Abnormal Return (AR) and Cumulative Abnormal Return (CAR)

The abnormal return is the difference between the firm's actual and estimated returns. Any significant abnormal returns in the test period may be attributed to the information content of dividend change announcement. Theses abnormal returns are used to investigate whether the event being considered has resulted in significant excess return. Mathematically, abnormal returns were calculated as follows: The mean abnormal return used to investigate if there is any excess return on individual days, and this will enable us to measure the market reaction to dividend changes announcements when they have been released to the market.

The cumulative abnormal return (CAR) is a sum of an individual period abnormal return over a number of periods. The cumulative abnormal returns over holding periods, from day m1 to day m2, are calculated as follows:

MAR CAR

As noted by Brown and Warner (1985) and Strong (1992), where there is uncertainty regarding the event day and even if the event date is known with certainty, it is not always clear when the information content of the event becomes available. So, cumulative abnormal return comes to resolve this problem by taking a sum of individual abnormal returns.

Parametric and Non-parametric Tests

The parametric tests proposed in the literature rely on the important assumption that individual firm's abnormal returns are normally distributed. We will employ two parametric tests in this study namely, the t-test, and Coutt's ZD test. Hamill, Opong and McGregor (2002) provide details of the ZD test which accounts fully for the increased variance of prediction errors outside of the estimation period and for the cumulating of these errors across different event windows. It also takes account of the fact that market model residuals are typically serially correlated, heteroscedastic and non-normal. The ZD-test will be used in this study to investigate the price performance in different event windows around the announcements of dividend changes.

Also, this study will employ Corrado's nonparametric rank test. Under the assumption of no cross-sectional correlation, the statistic is distributed unit normal. Cowen and Sergeant (1996) showed that if the return variance is unlikely to increase, then Corrado's rank test provides better specification and power than parametric tests. Seiler (2000) justify using Corrado's test is that Corrado (1989) developed a nonparametric rank test which is correctly specified independent of the degree of skewness in the crosssectional distribution of abnormal returns. Further, this nonparametric rank test is found to be less affected by event-induced variance. Maynes and Rumsey (1993) pointed that Corrado's test shows that the test statistic derived from the ranks is superior for testing for presence of abnormal return and the success of that test arises because the distribution of ranks is uniform.

Sample Selection

This study expands over seven years from 1996-2002 covering all the industrial firms. This leads to identify 360 observations distributed on the whole sample period from 1996-2002.

Stage One: 56 observations will be excluded from the whole sample. The observation will be excluded if: (21) observations from the sample. 3. Any observation for a firm that was suspended or reintroduced, because of reporting problems or financial losses will be eliminated from the study, these firms are unlikely to have a consistent run of daily stock price data in order to generate returns and hence will be eliminate from the sample. This resulted in (9) observation to be eliminated. Stage Two: Some information content studies set a criterion to select only firms with high trading activities like Rippinton and Taffler (1995). Following Al-Ghamidi (1998), observations should have had at least 60% trading days. The justification is that non-trading days result in zero returns, which in turn bias the results. This resulted to eliminate 107 observations from the sample.

As our aim is to investigate the market reaction to the change on dividends and if they reflect information content when they have been released. Four groups have been mentioned on developing the hypothesis depending on the dividend per share for each observation for time t comparing it with the previous dividend per share t-1. Each group will represent a sample for the purpose of this study, as follows:

1. If

DPS DPS

: then any observation meet this equation will consider as dividend no change. Furthermore, dividend no change could be with dividends and without dividends. According to that, this sample will be divided to two sub-samples; the first dividend no change sample, and the second no dividend no change sample.

The final sample is 197 observations. After investigating the changes in dividend and applying the above methodology, we observed 38 observations of those relating to the dividend increase, 34 of those to the dividend decrease, 24 observations related to dividend no change, and 101 related to no change no dividend (see Table 5.2).

Insert Table 2 Daily stock prices information over the period 1996 till 2002 were collected from Jordan Security Commission (JSC), Amman Stock Exchange (ASE), and Security Depository Centre (SDE). Daily stock prices and value-weighted price indexes were collected for the industrial firms all over the period of study, and then it was computerised. From the same source, dividends and their ex-dates over the period of study were collected. Other information like stock split, stock dividend, merger, acquisition, capital increase and decrease were collected for the purpose of not including the observations which announce any of theses events around the announcement date. Other sources like company reports and the stock market publications like monthly statistical bulletin and annual report, companies guide, and the daily official list were used to check the information collected for accuracy.

The justification of using the naïve model in this study is that, no available information in Amman stock exchange about the dividend forecasts from the market or from the analysts to compare the actual dividend with these expectations. Also, to compare our results with that obtained by Mollah (2001) who used the naïve model in his study. Previous literature like Aharony and Swary (1980), Eddy and Seifert (1992), and Opong (1997) applied the naïve model in their study. Table 6.2 reports the proprieties of the market value and it reports summary statistics for the four samples: the mean, standard deviation, skewness, kurtosis, minimum, median, maximum, normality test, and number of observations. Insert Table 4 Table 6.2 shows that the highest mean market value was for the dividend no change sample. It is clear from this table that the lowest mean market value was observed for the no dividend no change sample and it observed the lowest mean market value comparing with those samples which have dividends. The skewness statistics, which assess the symmetry of the distribution, is positive, which indicate generally a highly skewed distribution. The kurtosis coefficient is greater than zero, which indicates that the distribution inhabit some large observations. The kurtosis coefficient measures the relative peakness or flatness of the distribution, compared with the normal distribution. The kurtosis coefficient is positive which indicates that the data is peaked. A negative kurtosis value would indicate a relatively flat distribution.

Empirical Results: Mean Abnormal Return and Cumulative Abnormal Return

Descriptive Statistics

The Anderson-Darling normality test was also performed. The null hypothesis for this test is that the market value for the sample of dividends is normally distributed. The results demonstrate that the market value for the selected sample exhibited a high degree of non-normality which means that we accept the null hypothesis of the normality test. Table 6.3 shows the actual average dividend per share for each sample in our study. It is clear that the highest average dividend per share was observed for the dividend no change sample with 18%, then for the dividend increase sample with 16.8%, and then for the dividend decrease sample with 9%. Insert Table 5 6

.2 Market Reaction to Dividend Announcements

Numerous studies for developed markets reported that changes in dividends and earnings conveys specific information to the market (e.g., Aharony and Swary (2001)). In general positive (negative) dividend change announcements produce positive (negative) security prices reaction. The few studies of emerging markets provide alternative results. Mollah (2001) concluded that the market reacts negatively for both dividend positive and negative announcements, clarifying that a decrease on the security share prices occurred for dividend increase and decrease samples. A summary of the major empirical studies on the security price reaction to the announcement of dividends along with the data set they used, research methodology, and their findings are presented in the table 6.4.

Insert Table 6

After reviewing the previous empirical studies and their findings on table 5.4, table 5.5 summarises our expectations for the mean abnormal return and the cumulative abnormal return for the dividend increase, dividend decrease, dividend no change, and no dividend no change samples. A positive market reaction is expected to occur when dividend payments are increased comparing with the previous period, and accordingly, a positive mean abnormal return (MAR) is expected on the dividend announcement day (day 0). The expectation for the cumulative abnormal return on the pre-event period is zero, which would indicate that no information regarding dividends have been released before the announcement day. Therefore, if a positive market reaction occurred before the announcement day, this will be an indication to information leakage. On the other hand, a negative market reaction is expected when dividend payments are decreased compared with the previous period, and accordingly, a negative mean abnormal return will be observed on the dividend announcement day (day 0). The expectation for the cumulative abnormal return on the pre-event period is zero, which indicate that no information regarding dividends have been released before the announcement day. Therefore, if a negative market reaction occurred before the announcement day, this will be also an indication to information leakage. For both dividend no change sample and no dividend no change sample, no market reaction is expected to occur when dividend payments are not changing comparing with the previous period, and accordingly, no abnormal return will be observed on the dividend announcement day (day 0). Furthermore, the cumulative abnormal return on the post-event period is expected to be zero. The expectations for the cumulative abnormal return on the pre-event period are to be also zero, which means that no new information regarding dividends have been released to the market on the announcement day. Therefore, if a positive or negative market reaction occurred before the announcement day for both dividend no change and no dividend no change samples, this will be an indication of misleading information or rumours.

Insert table 7 6.3 Discussion of the Event Study Results

A summary of the event study results have been reported on table 6.6, table 6.7, and on table 6.8. Also, figure 6.1 and 6.2 illustrate graphically the results for dividend increase, dividend decrease, dividend no change, and no dividend no change samples. Table 6.6 shows the mean abnormal return for 10 days (±5) around the dividend announcements applying the market model adjusted with Scholes and Williams. The mean abnormal return is -1.32, -1.50, -2.33, and -.066 for the dividend increase, dividend decrease, dividend no change, and no dividend no change samples, respectively. Furthermore, the mean abnormal return is significant negative on the announcement day (day 0) at 1 percent level applying Scholes and Williams parametric test, and significant at 1 percent level applying the non-parametric Corrado's test except for the no dividend no change sample which is significant at 5 percent level. Moreover, the results did not change when other models were applied to generate the parameters. The market model, the mean adjusted model, the market adjusted model, the robust market model, and the market model adjusted with Flower and Rorke results were consistent with those results observed by applying Scholes and Williams model appeared on table 6.6. The parametric test shows that the market reaction was significant for both days 0 and 1, and for dividend increase and dividend no change samples, the market still reacts significantly on day 2. The empirical results show that security return decreases after the announcement of dividend increase, dividend decrease, dividend no change, and no dividend no change in Amman Stock Exchange. Even though, the empirical results of dividend decrease sample support the previous empirical studies, the empirical results of the dividend increase sample, dividend no change sample, and no dividend no change sample completely disagree with the previous empirical studies except for Mollah (2001). So, this is an indication of the ineffectiveness of the announcements of dividends in ASE. Our results are consistent with Mollahs' results that a negative market reaction occurred when dividends are announced. Mollah (2001) concluded that: "… security prices is decreasing after the announcement of good news (increasing dividends), bad news (decreasing dividends), and no news (maintaining dividends). So, this is the indication of the ineffectiveness of the announcements of dividends in emerging markets."

Moreover, the consistent between our study and Mollahs' study comes from applying data from the emerging markets while applying data from Jordan, Mollah applied data from Bangladesh, and may be the characteristics underlying the emerging markets are different from their counterparts on the developed markets. Table 6.7 summarises the cumulative abnormal return for 25 holding periods around the announcement day of dividends, and ZD test was used to report the results. This test correct the misspecifications which is related to the market model; non-normality, heteroscedasticity, and serially correlation. The cumulative abnormal returns were based on the market model adjusted with Scholes and Williams. The results show that the highest cumulative abnormal return was observed for the period from 0 to 2 days after the dividend announcements for the dividend increase, dividend decrease, and no dividend no change sample, and from 0 to 1 for the dividend decrease sample. Nevertheless, the most significant reaction occurred after the announcements and remains for 1 to 2 days after it. The pre-event period (-20, -1) did not observe a significant reaction for the four samples, which means that there is not information leakage before the announcements. The post-event period observed a significant negative market reaction except for the no dividend no change. Table 6.8 panel A and Panel B summaries the expected and actual mean and cumulative abnormal return for the dividend increase, dividend decrease, dividend no change, and no dividend no change samples. This table shows the result of investigate the market reaction to dividend announcements. On the mean abnormal return, the only sample which support our argument is the dividend decrease sample when it shows a negative market reaction realised after announcing dividend decreases, this is supported the previous studies on Table 6.4. Moreover, the cumulative abnormal return for the dividend decrease sample support our argument that after decreasing the dividends the security prices fall down, for the no dividend no change sample, no market reaction was observed for the post event period from 0 to 20 days after the announcement date. This conclusion supports the previous studies summarized on Table 6.4. The interpretation for the cumulative abnormal return of the no dividend no change sample that there is no change in security prices is that, the investors do not expect any dividend from those firms so they did not buy and hold these securities to get the right of getting cash dividend. Figure 6.1 and 6.2 support our finding and compare the four samples together. Even though, the dividend no change sample observed negative reaction than those dividend increases and dividend decreases, but the difference still insignificant. Furthermore, the figures show some differences when the market reacts to dividend announcements for the four samples, the dividend increase graph seems to be postponing the reaction until the first and second day after the announcements but in fact the reaction for the dividend increase sample begin on the announcement day (table 6.6 shows that the significant reaction begins on day 0). Moreover, the dividend no change sample is the one in which the market reaction of it begins before the announcement day, this is related to the sample size which is 24 observations, and then looking to the market reaction from the non-parametric point of view will give us the interpretation. Accordingly, the market reaction for the four samples is significant negative on the announcement day (day 0), which means that no differences graphically between the four samples.

Insert Table 8, 9 and 10

Insert Figure 1 and 2

Suggestions for Future Research

The issue of the information content of dividends is far from been solved. As Black (1976, p. 5) comments: -What should corporations do about dividend policy? We don't know‖. Very recently, about the dividend subject, Chu and Partington (2005, p. 2) state that -(…) this remains a controversial issue‖. Thus, the research in this domain of corporate finance is still not over. The phenomenon of an inverse relationship between dividend changes and market reaction was not satisfactory explained with this study.

We would like to understand the reasons behind failing to document a positive reaction with dividend increase and a negative reaction with dividend decrease in Jordan. In these situations, we wonder if the adverse relation between dividend change announcements and the market reaction could be endorsed to the failure of the naïve dividend changes model rather than to a real adverse reaction to dividend changes. Consequently, and also for robustness reasons, we will try to consider, in spite of the dividend changes, the dividend forecasts, when computing unexpected dividend changes, and the dividend yield ratio, in order to see if the main conclusions are unchanged. Furthermore, a possible path of future research might be the attempt to understand the reasons why the market reaction of dividend change announcements is negative, addressing this question by analysing the firmspecific variables that can influence the dividend change applying a panel data (longer period) rather than event study which analyses the reaction on the announcement day (short period).

Appendices

CAR =0

There is not a significant change on the cumulative abnormal return for the no dividend no change sample.

Hypotheses Development

In this section, we present our hypotheses regarding the relationship between earnings management, insider trading, and the market's response. The common wisdom is that firms manage earnings to inflate the issuance price of the SEO The stronger the reported performance before the SEO, the higher the firm's valuation, and the corresponding issuance price (see e.g., Kim and Park, 2005). Being rational, firms have incentives then, to inflate earnings in order to enhance their perceived performance and increase the proceeds. 6 This dynamics implies that firms hoard reported earnings for the SEO year before the event, present high earnings in the event year and then report low earnings after the SEO event since accruals must reverse. 7 5 Gordon formula states that V= E(x)/(r-g), where V = value of the firm, x = (permanent) earnings, r= discount rate, and g = growth of earnings. 6 While the Sarbanes-Oxley Act reduced the scope of earnings management, it did not eliminate it completely (Cohen, Dey, and Lys, (2005a). 7 The initial interest in earnings management in SEOs was motivated by an attempt to explain the underperformance of SEOs firms.

Loughran and Ritter (1995) examined companies that issued stock during the 1970-1990 periods. They found that investors obtained only 7% return for SEO.

Corporate Ownership & Control / Volume 7, Issue 2, Winter 2009 -Continued -3 359

The argument that firms manage earnings to inflate the price implicitly assumes that firms manage earnings overtly and hence can fool the market. For example, during the recent market bubble's, firms who managed earnings the most experienced higher returns than firms that managed earnings the least (Huddart and Louis, 2006).

We propose the following hypothesis: H1: The market price around on the SEO's issuance date does not discount earnings management.

There is an extensive literature that shows that insiders' trades are informative. For example, they are profitable contrarian traders: they sell when performance is strong while other investors buy and they buy when performance is poor while other investors sell ( The US environment is litigious. That is, if insiders sell their shares and the price dropped precipitously, they might be sued by investor for illegal insider information (Karpoff and Lee, 1991; Gombola, Lee, and Liu, 1997; Jones and Weingram, 1999). The awareness that the firm is subject to scrutiny by their investors following the SEO event, puts pressure to perform well and present high earnings after the SEO event. This motivation is exacerbated when insiders plan to sell shares after the SEO events, where they have incentives to manage earnings upwards in order to inflate the price (Huddart and Louis, 2006). As discussed above, the market is rational. Hence, it uses the insider trading as a signal (John and Mishra, 1990; Ching et al.

2006). If insiders sell, their trading indicates that the

If investors had instead of investing in these issuers invested the same amount in a non-issuing firm which was equal in size they would have received returns of 15% per year. 8 In the discussion of H1, we focused on the SEO as the motivation for earnings management. The association could be that insiders who plan to sell shares before the SEO attempt to manage earnings to increase their trading profits (Beneish and Vargus, 2002; Park and Park, 2004; Ronen, Tzur, and Yaari, 2006, 2007). 9 There is a debate in the literature regarding the timing of selling, because postponing the sale to after the SEO event can spare the insiders from costly litigation for illegal insider trading (Gombola et al. 1997), but selling before the SEO might be more profitable, because there is usually a price run-up before the announcement of the event.

price is overvalued and discounts it. We therefore propose the following hypotheses:

H2: Firms whose insiders sell shares before the SEO, manage earnings less aggressively. H3: Insiders' selling reduces the proceeds of the SEO.

The law defines insiders as blockholders, management, and directors.

The directors are expected to mitigate the agency conflict between shareholders and managers (Jensen and Meckling, 1976). So far, we focused on motivation for the insider trading to make profits by all groups, including the directors (Ronen, Tzur, and Yaari, 2006, 2007). However, as Hillier and Marshall, 2002 have shown, trade by directors is not fully explained by greed. Hence, we look for another explanation for insider trading.

The agency theory of insider trading that focuses on the impact of allowing the managers to trade on the ability of shareholders to align the interests of the managers with their interests, offers a new perspective.

Shareholders are willing to allow managers to trade because they can benefit from it. Such trades reveal information that is valuable for contracting with the managers (Dye, 1984), and for inducing the manager to make decisions that maximize shareholders' value when their attitude towards risk differs from that of shareholders. But because shareholders do not have full control over the manager's insider trading and he trades to maximize his personal wealth, this policy might also have some costly repercussions on shareholders' wealth (Bagnoli and Khanna, 1992, Bebchuk and Fershtman, 1991, 1993, 1994; Elitzur and Yaari, 1995). Since directors represent shareholders, directors' trade can construe a signal to shareholders that they do not to have to worry that managers do not act in their best interests, since shareholders can observe this trade but not the decisions done behind the closed doors of the boardroom. This discussion lends the following hypotheses:

H4: The market discount of insider selling is lower when the directors trade too.

Sample selection and methodology 3.1. Sample selection

The initial sample contains 10,787 firms issuing seasoned equity offerings between 1985 to 2004, from the Thomson SDC Platinum new issues database. The cut-off of 1985 coincides with calculating accruals from the statement of cash flows. We deleted firms under the following filters: we take the first SEO if the firm conducts multiple SEO (2) We delete financial institutions (SIC codes 6000-6999), and regulated industries (SIC codes 4900-4999), since their accounting is different from other industries.

After deleting firms with missing data on CRSP or COMPUSTAT, the sample includes 791 firms that are divided between 233 firms with insider trading and 558 firms without insider trading.

Insert Table 1 about here

For each SEO, we identified all non-issuing firms sharing the same three-digit SIC code as the issuing firm in the year prior to the SEO to derive our measures of earnings management as detailed below. 10 Insert Table 2

about here

The data for insider trading is obtained from the Thompson Financial (TFN insider Filing Data), which contains information on all publicly traded U.S. companies. We use their insider trading definition and define corporate insiders broadly to include those that have -access to non-public, material, insider information.‖

The results show insignificant association between market returns and earnings management, which confirm H1. Since, in untabulated results, we 11 A numerical example illustrates this issue. Consider a firm that announces an SEO on December, where the announcement price should be 5 dollars a share. In one scenario, the firm's insiders did not sell shares before that, and hence, the price before the announcement was 10. In another scenario, the firms insiders sold shares before the announcement and the price dropped to 9. Since Eventus uses the data during the trading period to calculate the betas for the CARs around the issuance date, it will yield higher abnormal returns for the firms whose insiders sell shares and these trades depress the price out of the window of the announcement period. The abnormal returns in the first scenario are (5-10=)-5, and in the second scenario, they are (5-9=)-4 >-5.

find that firms manage earnings upwards in the same fashion as described in Teoh, Welch, and Wong, 1998, this result is consistent with firms managing earnings to appear stronger performers than they really are.

For all cohorts, we find significant negative association between CAR and IPR. The coefficients for managers' cohort, directors' cohort, blockholders' cohort, and all officers' cohort are -0.38, -0.49, -0.67, and -0.39, respectively with the associated t-statistics of -1.59, -2.10, -2.37, and -1,75, respectively. These findings confirm H3. The market regards insiders selling as a signal that the stock is overvalued and discounts the price accordingly.

To test H2, we divide the sample of firms with inside trading into quintiles. Quintiles 1 and 2 include SEO firms with high IPR ratios, which represent the majority sales group; quintiles 4 and 5 represent firms with low IPR ratios, which represent concentration of purchases, quintile 3 is neutral. In each quintile, the firm manages earnings upwards, consistent with prior studies cited in section 2. As postulated by H2, firms with insiders' sales are more aggressive prior to the SEO in that their abnormal accruals are negative subsequently (DCA in period t-the year the SEO is done-and t+1, in quintile 1 are -0.00026 and -0.00017, respectively, while firms with insiders purchases manage earnings less aggressively. The abnormal accruals, in period t and t+1, as a fraction of lagged assets in quintile 5 are 0.00103 and 0.00141, respectively.

Methodology

We measure earnings management using the crosssectional variant of Jones (1991) methodology developed in Teoh, Welch, and Wong (1998) and Kothari, Leone, and Wasley (2005). These approaches separate accruals into two components; normal, or non-discretionary, accruals that results as a natural consequence of business structure and operations common to the industry (i.e. credit policy, business conditions, etc…) and abnormal, or discretionary, accruals that arise from earnings management. We identify abnormal accruals (the proxy for earnings management) using a two-step process. Following Hribar and Collins (2002), Total Accruals are the difference between Net Income and Cash flow from operations (Compustat items # 172 -#308). We define Current Accruals, CA, as: Current Accruals = Total Accruals +Depreciation expense (#196) + loss/gain on Sale of Property Plant and Equipment (#213).

(

We decompose current accruals into its discretionary and non-discretionary components in a two-stage procedure as follows. In the first stage we regress accruals on a model that links normal accruals to change in cash (Sales change less change in accounts receivables) and to lagged return on assets 10 While many prior studies match on 2 digit SIC codes (e.g. Teoh et. al, 1998), this results in SEO firms being matched with firms in widely varying industries. Using 4 digit SIC codes provides a closer match, but shrinks our sample size considerably. We therefore employ 3 digit SIC codes as a compromise between increased accuracy and sample size. Table 2 however aggregates our sample using their 2 digit SIC code for presentation purposes.

(proposed by Kothari et al. to account for the nonlinear relationship between accruals and performance). To alleviate heteroskedasticity, we scale all variables by lagged total assets (Compustat item #6), A t-1 . For each SEO firm, we estimate a regression, using all non-SEO firms in the same 3digit SIC code as the SEO firm in the year prior to the issuance of the SEO. In the second stage of the estimation, we use the coefficients from the first regression in the first equation to calculate discretionary current accruals (DCA) as follows:

In equation (2), discretionary current accruals deflated by lagged total assets (henceforth referred to as DCA) are defined as the difference between total current accruals and -non-discretionary‖ or -normal‖ accruals (the bracketed term on the right hand side of the equation). They represent the -abnormal‖ or managed component of current accruals and is used as our proxy for earnings management.

To analyze the pattern of insider trading of issuers of seasoned equity offerings, we adopt the insider purchase ratio used by Piotroski and Roulstone, (2005) and Sawicki, (2005) that measures insider trading behavior. We calculate the insider purchase ratio (IPR) as follows; To test for the impact of insider trading on the SEO's proceeds, we measure the cumulative abnormal returns around the SEO issuance date. A matching of firms with insiders trading to firms without insider trading--the matching is based on firms conducting an SEO in the same year and within the same industry--, does not yield a meaningful sample. The benefit of using CARs is that they capture the change in returns relative to the time when insiders traded. To sharpen this point, observe that insider trading takes place out of the issuance window, so that the information of the trade is already compounded in the price. This will impact the direction of the expected signs of the coefficients in our regression models, as discussed below.

Results

To test H1 and H3, we segregate insiders according to their cohorts where top management is made up of Chairman, Chief executive Officer (CEO), Chief Operating Officer (COO) and President.

Top financial officers are the Chief financial Officer (CFO) Controller and Treasurer. The category of all officers are the corporate officers, top management, principal financial officer, principal accounting officer, vice presidents in charge of principal business units, divisions or functions and other persons who perform a policy making function. All directors' category includes members of the company's board. Block holders are beneficial owners of 10% or more of the company's outstanding equity. Insiders were segregated according to cohorts to find out if the different types of insiders had different trading patterns which in turn influenced market returns differently. Also, since not all the insider groups are equally knowledgeable about earnings manipulation, segregation according to cohorts would also bring to light differences in trading patterns which might affect market returns. Specifically, we run the following model:

where CAR is the eleven-day cumulative abnormal returns and the independent variables are the insider purchase ratio IPR, discretionary accruals for the year before the SEO, DCA, multiplied by 100, and the standard controls for size: total assets (Compustat # 6) and market value (Compustat item #24 times Compustat item #25) divided by 100. By H1, we expect that the sign on DCA to be non-significant, and the signs on both control variables to be positive. The sign on IPR is a bit tricky. We measure insider trading for trades that take place one year before the issuance. We wish to focus on the impact of insiders sales on the negative announcement effect on the market price. Since by the time the firm makes the issuance, the information that the sale is overvalued is already filtered into the market price for no other reason than that insiders' trades are informative and public. Hence, if indeed sales have a negative impact on the proceeds of the SEO, the sign on IPR should be negative. 11 Insert Table 3

Insert Table 4 about here

To test for H4, we run the following model: Given that the total sample of firms with and without insider trading is 791 firms, this result shows that insiders control pernicious insider trading by officers in two ways: one is banning it, and the other is to trade themselves and signal that the trade may be driven from reasons other than greed.

Insert Table 5 about here

Summary and Conclusions

Since firms conduct seasoned equity offerings to raise much needed capital, insider selling seems a selfdefeating practice because it conveys to the market that the price of the firm's shares is overvalued. In this study, we examine insider trading a year preceding the issuance of stock at a SEO event and the proceeds of the SEO; the link between insider trading and earnings management, and the role of directors in the occurrence of insider trading before the SEO. Our main findings are that firms whose insiders sell shares manage earnings more aggressively, but that this information is ignored by the market. The market takes into account the information content of insiders' sales, but directors' sales play a mitigating role, in that the discount is lower for firms with both managers and directors' trades. Our results then indicate that directors can control the unfavorable impact of insider trading on the proceeds of an SEO in two ways: one is by banning it, and the other is by trading themselves and conveying to the market that not all trades are motivated by opportunistic greed. (Peat and McCorry 1997) and the majority of studies done on the impact of single stock futures trading originated from the Australian market.

The trading of a SSF-contract, the price of which is derived from an underlying equity share, may conceivably impact on the underlying spot price as a result of price discovery and the setting of a future spot price. Similarly, trading volume in the futures market may either generate (equivalent trades to cover positions) or curtail (substituting one market for another) spot activity. Many past studies, mainly on share indices, investigated the impact of derivatives trading on the underlying with regards to a possible price and volume effect. The introduction of single stock futures presents an opportunity to revisit this subject from an individual company perspective. The following papers provide an overview of the results on price and volume effects experienced with initial futures trading.

2

Literature review Robbani and Bhuyan (2005), using a two-sample (pair-wise) t-test and the Wilcoxon signed-rank test, found evidence that the average daily rate of return on the thirty underlying component shares of the Dow Jones Industrial Average (DJIA) decreased significantly following the introduction of futures and options. They also noted a significant increase in the daily trading volume for the majority of index constituents (23 from 30 shares). An event study (market-adjusted model) conducted by Aitken and Segara (2005), likewise, reported the introduction of warrants on individual equity shares in Australia to be associated with a negative price effect. Their finding was corroborated by Clarke, Gannon and Vinning (2007) in an event study (mean reversion and market models) showing generally negative abnormal returns around the date of warrant issuance. These two studies contradicted an earlier event study (market model) by Faff and Hillier (2003) that revealed positive abnormal returns by newly optioned shares (in the United Kingdom) even though no discernable trend in the magnitude of these returns over time were recorded. Concerning any volume effect, Faff and Hillier (2003) witnessed an increased trading volume in the ten-day period immediately subsequent to options introductions (dummy variable regression), while Aitken and Segara (2005) established that the relative trading volume (trading volume divided by total number of securities outstanding) in the underlying share following warrant listing was significantly greater than in the pre-warrant listing period (Wilcoxon ranksum test). Clarke et al (2007) in this instance disagreed, providing evidence that when adjusting for the inherent upward trend in volume, warrant introduction generally causes a decrease in trading volume. Peat and McCorry (1997) pioneered research on the listing-effect of single stock futures (SSF) and found no significant change in the underlying price level. This market-adjusted-model event study examined the impact on ten individual equity shares trading in Australia. A significant increase in trading volume was evidenced from a t-test for change in mean performed on the ten individual underlying shares. This was confirmed by Lee and Tong (1998) with an equal means and equal variances t-test and rank sum tests, accounting for the (G)ARCH effect and using a control group to rule out confounding events. Their evidence suggested that volume distributions have significantly changed after the inception of single stock futures and that post-futures volumes have higher means but smaller variation.

The only reported study in South Africa on the volume effect of futures trading was done by Swart (1998) Research methodology

An event study was conducted in an attempt to detect a possible price effect with the introduction of futures trading (i.e., the event) to the South African equity market. A market model was used to generate the abnormal returns required to analyse the impact. The average normalised trading volume pre and post SSFintroduction was evaluated using a dummy variable with trend coefficient regression to uncover any volume effectthat is, significantly changed trading volumes in the respective underlying shares.

Price effect

The effect of a financial event on the value of a listed company can be measured using financial market data in an event study (Campbell, Lo & MacKinlay 1997:149). The effect of a firm-specific event (e.g., introduction of a single stock futures contract on a share) should reflect as an abnormal or unexpected change (positive or negative) in the firm's share price. Event study methodology encompasses the econometric techniques used to estimate and draw inferences from the impact of an event or multiple identical events in a particular period. A viable and effective event study requires the isolation of the event to the greatest degree possible, independence of individual company returns, and the assumption of constant systematic risk as represented by the beta coefficient used to determine the -normal‖ return. Wells (2004:66-67) states that samples should be from different industries, with each sample security having a different event day, and a large sample size.

Horizon length has a big influence on event study properties and largely determines the specification level, power, and sensitivity of test statistic specification. These possible problems do not apply to short-horizon event studies (less than 12 months) as these methods are relatively straightforward and trouble-free. Short-horizon event methods are powerful if the abnormal returns are concentrated in the event window. Also, the specification of the test statistic is not highly sensitive to the benchmark model. A problem shared by both short-and long-horizon studies is the possible increase in the variance of the security's abnormal returns conditional on the event. As a result, test statistics can be miss-specified and the null hypothesis rejected too often (Kothari & Warner 2006:15-18, 50).

A market model approach was used to generate company betas and calculate normal and subsequently abnormal returns (i.e., actual minus normal). The market model is a risk-adjusted statistical model that relates the return of any given security to the return of the market portfolio. A one-factor OLS regression analysis generates the intercept or alpha ( i α ), and

The market model

The concept of abnormal returns is the central element of event studies and the benchmark or model generating normal returns is consequently central to conducting an event study. The chosen model for this study is the market model, specified as follows:

For any security i the market model is:

Measuring and analysing abnormal returns

The key focus of an event study is to measure the sample securities' average and cumulative average abnormal returns around the time of an event (Kothari & Warner 2006:7). Time is redefined relative to the day of the event and the average security pricemovement for the sample securities is examined during specific days around the event month.

0 The residual iη ε from the market model corresponding to day  is the estimator of the abnormal return for security i during event day . This, according to Binder (1998:112-113), removes the effect of economy-wide factors on the return of the security and retains the portion of the return attributable to firm-specific information.

The design of the time line (timing sequence) eliminates any overlap between the estimation window and event window, ensuring that the estimated parameters of the normal return model are uninfluenced by the returns around the event. The exclusion of the event window when measuring the normal returns upholds the assumption that the abnormal returns will capture the impact of an event.

The estimation framework will include the event window if the null hypothesis is expanded to accommodate any changes in risk (variance) of a firm due to the event. The post-event window data may also be included with the estimation period to estimate the normal return model (MacKinlay 1997:20).

Estimation of the market model

The relationship between a security's returns and returns on the market is estimated by ordinary least squares (OLS) regression and this relationship is used to estimate expected returns, given returns on the market. Beta is calculated as the covariance between the market and the security during the estimation period, divided by the variance of the market in that period. The intercept term (alpha) is the difference between the average return () on the security and the estimated return on the security as determined by the market return and calculated beta.

Statistical properties of abnormal returns (AR)

The

Aggregation of abnormal returns (CAR)

In an event study the focus is on the mean and the cumulative mean of the dispersion of abnormal returns. The abnormal return observations must be aggregated across observations of the event in order to draw overall inferences for the event of interest. Aggregation may occur along two dimensionsthrough time and/or across securities. This study focuses on the impact of a single event on several different firms (i.e., aggregation across securities). The event windows of the included securities should not overlap (no clustering assumption). The absence of any overlap and the distributional assumptions imply that the abnormal returns and the cumulative abnormal returns are independent across securities. The individual securities' abnormal returns are aggregated and averaged (8).

The cross-sectional average abnormal returns (residuals) in common event time:

For a large estimation window, 1 L , the variance is: The average abnormal returns per event day are summed across days to measure the average cumulative effect of an event on the sample securities from day 1 η to day 2 η . Cumulating these periodic average residuals over a particular time interval (number of days in the event window) allows for inferences concerning the general impact of the event.

The aggregated average abnormal returns over the event window: T<η £ η £ T

Tests for significance

A test statistic is calculated and compared to the assumed distribution under the null hypothesis that the event has no impact on the behaviour of returns (i.e., the mean abnormal return equals zero). The null hypothesis is rejected if the test statistic exceeds a critical value corresponding to the specified test level or size of the test. The standard test statistic is obtained by dividing the average or cumulative average abnormal return by the relevant standard deviation (13). Inferences about the cumulative average abnormal returns are drawn using:   (13) A possible modification of the basic approach that may lead to more powerful tests (i.e., the ability to detect non-zero abnormal returns) is to standardise each abnormal return using an estimator of its standard deviation (MacKinlay 1997:24).

The purpose, according to Serra (2002:5) is to ensure that each abnormal return has the same variance. By dividing an abnormal residual by its standard deviation, each residual has an estimated variance of one. An amended test statistic of the hypothesis that the average standardised residual of a firm is equal to zero is calculated (not shown).

A two-sided test of the null hypothesis on the cumulative abnormal return based statistic 1 θ from (13) is used. A confidence-interval approach as shown in (14) is used and values lying in this interval are plausible under H 0 with 100(1-)% confidence. Hence, do not reject the null hypothesis if the value lies within this region. Outside the interval it may be rejected.

The null distribution is standard normal for a two-sided test size of α and the null hypothesis will be rejected if 1 θ lies in the following critical region:

Φx is the standard normal cumulative distribution function (CDF)]

Volume effect

Share trading volume is a highly volatile factor, according to Clarke et al (2007:30), often resulting in large variances, generally non-normal distributions and many outliers. An exponential smoothing process was applied to the data to normalise the volume and these normalised volume figures were used in the analysis. An exponentially weighted moving average (EWMA) process assigns exponentially decreasing weights to older data.

The single exponential smoothing method is appropriate for a series that moves randomly above and below a constant mean with no trend or seasonal patterns. A double smoothing method is appropriate for a series with a linear trend. The smoothed series is calculated recursively, by evaluating the formula presented in (15). The forecasted value is a weighted average of the past values of the series where the weights decline exponentially with time. The smaller the damping or smoothing factor, the more smoothed the eventual forecasted series (NIST 2006).

Single exponential smoothing formula:

s= αy + 1-α s = s + α y -s To determine whether the event caused a permanent change in volume, the average normalised volume in a specified number of days prior to the event was compared to the average normalised volume in the period subsequent to the event. Trading volume generally tends to increase over time and a dummy variable regression (16) considering this trend was used, as this is not captured by a t-test for change in mean.

Equation used to estimate volume: The dummy variable takes the value of zero for the pre-event period and one for the post-event period. The coefficient is interpreted as a change in trading volume after considering any underlying trend which may bias the results of the dummy variable. The level of significance is indicated by the relevant p-value of the statistical output.

Data and statistical analysis

The South African market saw three-hundred and fifty-seven (357) first-time introductions (available for trade) of physically-settled SSF contracts from 1999 to 2007. Ninety-nine (99) cash-settled SSF contracts (dual issues) have been introduced since February 2007. Two (2) inward listed contracts (i.e., based on foreign reference assets or issued by foreign entities and listed on the JSE) were made available for trade in this period. These potential candidates for inclusion in the study were subjected to the following selection criteria:  No corporate actions or events that may have affected the shareholder's entitlement to benefits (i.e., share splits, capitalisation/rights issues, unbundling, mergers, or takeovers).  No direct or indirect prior introductions of SSF contracts.  Trading activityavailable for trade and the actual trading of contracts occurring within a two week period.  No overlapping of the 11-day (including day zero) event periodsno clustering assumption  250 days (excluding the event period) of spot trading before and after the event. Thirty-eight (38) companies matched all the selection criteria, representing seven (7) different industries which in turn translate to twelve (12) supersectors, eighteen (18) sectors, and twenty-five (25) subsectors (refer to Appendix A). This satisfies the event study requirement that samples are from different industries and not focused on a specific industry.

The company returns were regressed on the returns of the market (All Share Index -ALSI), thereby determining the company's beta (slope coefficientsensitivity to return of the market) in order to establish the -normal‖ daily returns of a company during the event period. The difference between this normal or anticipated return (beta times the market return) and the actual return of the company, represents the abnormal return on a specified day. The 250 daily returns of the company and the ALSI preceding the event period (ten days excluding the event date) were used in the market model regression (while there are 365 days in a year, only approximately 250 of them are trading days, thus representing a one-year period of trading data before and after an event or event window). The abnormal returns (actual minus normal) five days before (pre-SSF period), on the actual first trading day (day 0), and after (post-SSF period) the event were calculated and assigned to either a one-, five-or ten-percent level of significance, or as non-significant (too small relative to the standard deviation). These daily abnormal returns for each company were averaged (average abnormal return -AAR), cumulated (cumulative average abnormal return -CAAR) and evaluated for statistical significance.

The selected model (market model), generating individual company betas, was not assessed in terms of -goodness-of-fit‖ (R-squared) in each instance, but simply used to establish a normal return as determined by the market, to be compared with the actual return from the movement in individual share prices. The discrepancies between the relative and actual company returns during the event period are presented and attributed in the following table (table 1). Inferences regarding the statistical validity of each abnormal return and the conclusion reached on the impact of initial SSF trading on the underlying share price follow the statistical output.

In general, most daily abnormal returns proved to be non-significant, not exceeding the 1,68 critical value cut-off for a 10%-level of significance (90% confidence level). SSF trading, according to this event study, had no effect (no significant abnormal returns during the event period) on the share prices of eighteen (out of thirty-eight) companies included in this study.

INSERT TABLE 1 ABOUT HERE

The following fourteen companies showed a statistically significant abnormal return on a single day during the event period under investigation -Allied Technologies EOH Holdings (14 -EOH) and Paragon Holdings (29 -PCN) each exhibited only two days of statistically significant abnormal returns, providing virtually no evidence that SSF trading had influenced their share prices. Similarly, only three days of sufficiently sized abnormal returns reported by Shoprite Holdings (33 -SHP), Super Group (36 -SPG) and Winhold (38 -WNH) confirmed that SSF trading had little effect on the returns of these companies.

Hudaco Industries (19 -HDC) was the only company to reveal some share-price impact caused by initial SSF trading. Showing abnormal returns on six days (including day zero), Hudaco Industries mainly experienced abnormal share-price activity in the fiveday period leading up to the availability of SSF contracts on its equity shares. Only three companies displayed statistically significant abnormal returns on trading-day zero, namely Hudaco Industries, Pick and Pay Holdings, and Winhold.

On an individual company-by-company basis it is clear that the introduction (trading) of single stock futures had little or no impact on the underlying companies' share prices and the event study presented no conclusive evidence to establish either a positive or a negative price effect due to SSF trading.

However, in an event study the focus is on the mean and the cumulative mean of the dispersion of abnormal returns. The individual securities' abnormal returns are aggregated and averaged. These average abnormal returns per event day are summed across days to measure the average cumulative effect of the event on the sample securities for the whole event period or a variety of periods within the event window. Cumulating these periodic average residuals over a particular time interval (number of days in the event window) allows for meaningful inferences concerning the general impact of the event. If the initial SSF trading caused a price effect, significant abnormal returns on day zero and possible significant abnormal returns on day -1 and day +1 should be uncovered. Significance should be lower as one moves further from the event date and for longer periods or periods not including the actual event. Table 2 shows that the only significant average abnormal return was recorded on day +3 of the event period. Periods (-5 to -3), (-5 to -2), (-5 to -1) and (-5 to 0) all showed significant cumulative average abnormal returns. Results indicate a 10% significance level for a period that includes the event day (-5 to 0), but for shorter periods inclusive of day zero [(-3 to 0), (0 to +3), (-1 to 0) and (0 to +1)] no significance is evidenced. Shorter time periods (-5 to -2) and (-5 to -3) do show increased significance, but do not include the actual event date. The most promising result, therefore, is the significant positive CAAR of 2,17% during the six-day (including the event day) pre-SSF period. Positive , but non-significant, average abnormal returns on day -1 and 0, as well as positive, but nonsignificant, cumulative average abnormal returns in periods (-3 to 0) and (-1 to 0) tend to confirm the favourable impact of SSF trading on the underlying share price in the period immediately preceding the event and on the event day itself, as shown by the significant (-5 to 0) subperiod.

INSERT TABLE 2 ABOUT HERE

Diminishing significance for shorter periods closer to the event implies that no clear evidence to suggest any price effect (positive or negative) on the underlying due to the introduction of single stock futures resulted from this study.

The effect of SSF trading on the spot market volume of the underlying company before and after the first futures market transaction was tested by comparing the average normalised trading volume pre-and post-SSF with a dummy variable and trend coefficient (a time series variable that checks for a trend) regression to determine whether the volume significantly changed after accounting for the tendency of the volume to increase (trend) over time. Figure 2 depicts the normalised daily trading volume (red) of AECI Holdings (1 -AFE), used as an example, after an exponential smoothing process was performed on the actual data (blue). The forecasted value is a weighted average of the past values of the series where the weights decline exponentially with time (higher weight allocated to more recent data).

Trading volume generally tends to increase over time and a dummy variable regression considering this trending nature of volume is used, as this is not captured by a t-test for change in mean. The dummy variable regression is augmented with a trend (day) coefficient to isolate the size of the increase/decrease in trading volume witnessed after initial SSF trading. The dummy variable takes the value of zero for the pre-SSF period and one for the post-SSF period. The coefficient is interpreted as a change in trading volume after considering any underlying trend which may bias the results of the dummy variable.

INSERT TABLE 3 ABOUT HERE

In line with the dummy variable regression (with trend), the number of companies showing a significant increase in average normalised volume is nineteen (three non-significant increases). Fifteen companies exhibited a significant decrease in average normalised trading volume (one non-significant decrease). A small majority of companies (19 vs. 15), therefore, experienced a significant increase in trading volume following the onset of single stock futures trading.

A t-test for change in mean (not accounting for any trend) confirmed that the majority of companies (27 of 37 significant results) experienced highly significant increases in normalised trading volume after the introduction of SSF-trading (De Beer 2008:78).

Summary of results

A pre-SSF (estimation period) regression analysis generated beta coefficients for each of the thirty-eight individual companies and established normal daily returns via the market return during the event period. The difference between this normal return and the actual return of the company represented the abnormal return (AR) on a specified day. Average abnormal returns (AAR) and cumulative average abnormal returns (CAAR) for the sample were also calculated. The abnormal return in excess of the relevant standard deviation (acceptable divergence) determined statistical significance. Fourteen companies (37%) showed a statistically significant abnormal return (AR) on a single day during the event period under investigation. Two companies (5%) exhibited only two days of statistically significant abnormal returns. Three days of sufficiently sized abnormal returns were reported in three instances (8%). Only one company (3%) revealed some share-price impact caused by initial SSF trading, showing abnormal returns on six days (including day zero), and mainly in the five-day period leading up to the availability of SSF contracts on its equity shares. Only three companies (5%) displayed statistically significant abnormal returns on the day of the event itself. The only significant average abnormal return (AAR) was recorded three days after the event. Significant cumulative average abnormal returns (CAAR) were recorded for longer periods and periods further away from and before the event. The six day pre-event period that included the event day showed statistical significance, but no shorter periods inclusive of day zero displayed any significance.

374

No-trend Trend

Increase Decrease Non-significant Figure 3 illustrates the result from the t-test (notrend) compared to that of the dummy regression with trend coefficient. A t-test for change in mean showed that the introduction of SSF trading resulted in a highly significant increase in normalised trading volume in the majority (71%) of cases. A smaller majority (56%) of statistically significant outcomes from the dummy variable with trend coefficient evaluation indicated an increase in trading volume following the onset of single stock futures trading.

Note

This article is based on a study done on the impact of single stock futures on the South African equity market that also reported on any changes in the level and structure of volatility post initial single stock futures trading. (z-stat)*** -1% significance (z-stat)** -5% significance (z-stat)* -10% significance

Models of corporate governance

Stakeholder and Shareholder Models

Stakeholder Model

Essentially, this is based on the notion that private ownership results in a fundamental desire of social order and an efficient economy. This can be seen in relation to a company in that the right to incorporate is a right to own property and therefore corporation should be seen is a legal extension of their owners (Allen, 1992). Since shareholders are the owners of the company, the company has legitimate obligations and the managers have a fiduciary duty to act in the interest of the shareholders (Barker, 1958;Mayson et al 1994). This is the Chicago School of Law and Economics. Under this theory, assets of the company are the property of the shareholders. Directors and managers, as agents of the shareholders, have no legal obligation to any other stakeholders (Allen, 1992;Blair, 1995). This approach is supported by neoclassical economists such as Hayek and Friedman. For Hayek (1969) this approach of pursuing self interest is the most efficient way to manage economic activities -thus, the company must use shareholders capital to maximise profits in order to enhance shareholder value. Any 'social purpose' beyond the shareholders interest could be viewed as an abuse of power as it will not lead to efficient use of corporate resources.

This view is developed by Friedman (1962Friedman ( , 1970 who asserts that other stakeholder interests are looked after by contracts or government regulation. These are not the remit of corporate governance.

This approach has been defended recently by Sternberg (1998Sternberg ( , 2000 who asserts that considering interests beyond the shareholders undermines private property, agency duty and value-creating capabilities of a business. To address the problem of potential 'abuse' she suggests internal monitoring through nonexecutive directors, voting rights and information disclosure to shareholders. This method of protection is also posited by Malegam (2008).

So, this Model regards the company as an extension of its owners and that only market forces can achieve efficiency (West, 2006). To resolve any potential conflict between the owners and managers rewards are linked to corporate performance (Letza et al, 2004).

This Model is directly at odds with the notion of inherent property rights. It regards the company not as a private association united by individual property rights but rather as a public association constituted through political and legal processes and as a social entity for pursuing collective goals with public obligations (Gamble and Kelly, 2001 p.115).

This approach is summed up by Sullivan and Conlon: "The standard of a corporations usefulness is not whether it creates individual wealth but whether it helps society gain a greater sense of the meaning of community by honouring individual dignity and promoting overall welfare" (1997, p. 713).

Thus, the corporation is a 'social entity'. It is responsible to and accountable to a broader set of actors than its owners (Wieland, 2005). These actors include employees, suppliers, local communitiesindeed, anyone affected by the behaviour of the company (West, 2006).

In addition to the considerable literature on the characteristics of these Models (Donaldson and Preston, 1995;Letza et al, 2004) the assumptions made by these Models about the nature and purpose of a company and to whom it is ultimately responsible constitute key issues for the direction of corporate governance development in any developing jurisdiction.

Corporate Governance in Poland

Corporate governance was introduced to Poland by the July 13 th 1990 Privatisation of State Owned Enterprises Decree. This also transformed the system from a centrally managed one with a planned economy to a free market one. The whole Polish adventure with corporate governance started with large companies, as these were expected to be at the forefront of implementing corporate policies. The initial step after the transformation was to create supervisory boards. One might ask why the building of corporate governance started in this way. The answer is that at that time all companies were national and had no private share. Many of them were massively privatised, therefore some supervisory body was essential to oversee and look after stateowned entities (SOE). A key task for the government was to find new and stable owners who would effectively run enterprises embraced by the privatisation process. Quite frequently these companies required considerable financial input in order to restore their full capacity or to improve obsolete technologies of production. Therefore, in order to assure success for the newborn Polish economy, the government had to search for financially sound investors with long-term goals. According to Koldakiewicz (2001) the nineties provided enough time to build the basis (emphasis added) of corporate governance, but much still remained to be done.

The whole process of privatisation not only in Poland but in all of Eastern and Central Europe is a great example of building different forms of corporate governance policies (Grosfeld and Hashi, 2007). Although these countries had similar economies and were culturally convergent, still their experiences soon after transformation were very different. The first completely new thing that post-socialistic countries had to face in their free markets was the separation of ownership from control, something well known to developed countries (Koldakiewicz, 2001) and discussed above.

There was a lack of professional and experienced managers, so initially there was chaos in the markets. Governments launched mass privatisation programmes that ran through societies, which dispersed shares amongst a huge number of individuals, something that was totally new. In Poland alone, out of 29 million people entitled to participate in the privatisation programme nearly 26 million took part. The experience was very unusual and among certain groups was strongly criticised for its artificiality. However, there was a grain of truth in the criticism. The quick process of privatisation did not give law-makers and companies enough time to adapt themselves to the new conditions. There was insufficient time to build a firm basis of corporate governance. This led to a large group of opponents forming the view that the Polish form of corporate governance failed in the nineties (Grosfeld and Hashi, 2007). To this day, politicians from the parliamentary lectern accuse the then government, with its prime minister at the vanguard, for the inept process of privatisation that cost the nation an unbelievable amount of money. Among those widely criticised was the then viceprime minister and also the Secretary of the Treasury, who also became the subsequent president of the National Bank of Poland. He introduced a package of 11 new acts that were to transform the economy. The truth is that Balcerowicz had to act quickly. At that time inflation in Poland had reached an astonishing 650 per cent of the gross domestic product. There was not time to hesitate (www.prawo.uni.wroc.pl, 2008).

Before drawing any conclusions on how far Poland managed to get in terms of developing and implementing corporate governance, especially in comparison to the UK and the US, it is essential to consider all of the stages from the very beginning of the Polish free market. As mentioned previously, the process began in the early nineties with the reforms initiated by Leslaw Balcerowicz, and the subsequent creation of the financial system and capital market. The next phase was the transformation of state-owned companies into 'sole-shareholder companies of the State Treasury'.

This was also the time that shareholders had their first general meetings and selected their supervisory boards. So far, however, the only shareholder was the State Treasury, so members were selected by the Secretary of the Treasury. On average, supervisory boards consisted of five to six members, out of which, one third were selected by employees (Koldakiewicz, 2001).

In Poland, initially only 512 state-owned companies were embraced by the privatisation programme. When compared to the significantly smaller Czech Republic which had 1700 companies on the list, this does not appear to be a strong response. However, the two economies went in different ways. In relation to Poland, 60 per cent of the equity stakes of 512 companies that were undergoing the privatisation process were given to 25 specially set up National Investment Funds (NFI). Exactly 33 per cent went to one particular NFI, and the remaining 27 per cent were proportionally split between the 14 remaining NFIs giving them slightly less than 2 per cent of the equity share. In this regard the NFI's share simultaneously made these separate 15 entities major shareholders. The remaining 40 per cent were split between the Treasury and employees, 25% and 15% respectively. The managerial functions over the new 15 NFIs were entrusted to commercial institutions, such as 'investment banks' and 'consulting firms', both Polish and foreign. The selection of managerial bodies was conducted 'through international tender offers'. The artificial split of shares between the NFIs was intended to prevent a high concentration of shares being held by one entity. However, neither in Poland nor the Czech Republic did this strategy succeed. By the end of 2000, through acquisitions and mergers, many single shareholders already had over 50 per cent of the equity share of particular companies. In effect, they could easily control companies from their portfolios and accordingly affect the managers' decisions. With regard to Poland, shareholders received one 'certificate' equal to 'one share' in each of the 15 Funds.

As a result, they became 'indirect shareholders'. The NFIs were listed on the Warsaw Stock Market and certificates were exchanged to shares of NFIs in 1998 (Grosfeld and Hashi, 2007).

It is worth highlighting that these NFIs were simply conducting their business activities as investment funds, and fulfilled all the requirements to be classified as capital groups. Unlike in Western Europe or America, they came into existence immediately. In the UK or the US the process of creation of powerful capital groups lasts for many years. In the case of Poland this happened immediately and was a very new procedure, not only for Poles, but for all economies (Szczepkowska, 2003).

Poland made a huge effort from the very beginning in terms of implementing effective laws to protect the market and investors. This especially applies when Poland is compared to other Eastern and Central European countries. As Grosfeld and Hashi point out: "NFI managers and the stock exchange listing requirements were carefully designed to ensure the transparency of the process and to avoid expropriation of minority investors". The major concern of Polish authorities was on the one hand not to allow too high a dispersion of shares in order to keep some strategic investors as watchdogs, but on the other hand not to concentrate the ownership too much, so as to avoid "the potential danger of private" incentives with no regard to minority shareholders. Therefore, government implemented "the limit of 33 per cent on the lead fund's holding in each" privatised company (2007 p.522).

In the early stages of the new Polish economy, a major problem concerned the lack of independent institutions that could monitor the activities of companies in the market. This situation changed when the Securities Commission issued regulations in 1991. This provided regulations to assure fair competition and equal access to verified information in relation to the securities market. Also, more importantly, the new regulations touched upon minority shareholders' rights. Soon after, in April 1991, the Warsaw Stock Exchange (WSE) was established and still remains the major proponent of best business practices and corporate governance. What is more, the WSE imposes several obligations on companies floated on the stock market, e.g. submission of quarterly and annual reports, with nonconformity penalised by fines.

Further developments of Polish corporate governance came in 1993 with the Act on the Financial Restructuring of Enterprises and Banks and Act on National Investment Funds and Their Privatisation. Both of these brought new and more efficient supervisory functions. Furthermore, they dealt with the problem of bad debts. At that time, banks had serious problems in judging whether to grant loans to companies to facilitate their further development. The new regulations of 1993 transformed the role of banks from 'a lender to an owner' with debts exchanged into shares of capital stakes. There were high hopes in relation to the new provisions. Regrettably, these expectations were not achieved in reality. There were no major changes in the ownership structure observed. Although the provisions were intended to encourage banks to take more risk and grant loans to finance more challenging projects, this did not happen to any great extent. Banks very rarely exchanged their debts for shares of their debtors. Inexperienced bank managers presumed that if a particular company could not pay off its liabilities, in the case of transfer of debts into shares. Bad debts would be exchanged for 'bad shares' (Koldakiewicz, 2001).

A major problem for Poland in the nineties concerned minority shareholders. Indeed, this is still a problematic issue. It is common that majority shareholders and minorities have divergent goals; however, this is not always the case. As already discussed, there can be very different institutional investors in the market. Some of them may want to have large numbers of shares in order to monitor the managers in control, some may want to diversify their portfolios in order to minimize risks, others may want to influence managers' decisions and pursue their interests (Grosfield and Hashi, 2001). Whatever the case, effective mechanisms providing protection for minority shareholders are a must. However, this issue is not the sole problem for Poland or other emerging markets, but all markets worldwide. All developed economies have to or have had to deal with this major problem. After all, minority shareholders are the group most vulnerable to abuse. With regard to the Polish market there are several examples of unfair transfer pricing, unjustified investment projects or excessive licence fees etc. However, it is very difficult to assess the scale of such behaviour in the market (www.pfcg.org.pl, 2008). What is even more difficult is to prove that motives were not genuine and the policies were not crried out with the best intentions.

As time went by, Polish law continuously altered and tried to accommodate itself to the needs of the market. A substantial number of these adaptations concerned effective supervisory mechanisms. In effect, several duties were imposed on supervisory boards, inter alia, control of balance sheets, compatibility of accounting books, assessment of management boards' reports and dealing with proposals for distribution of profits. An additional duty imposed on supervisory boards was to issue annual reports for shareholder's general meetings in respect of their work and conducted assessments. On top of the above, some judicial responsibilities included suspension of individual members and/or the whole board of management for important reasons and delegation of members of the board who are incapable of fulfilling such functions. Put simply, obligations that supervisory boards owe shareholders include: An information function: submission of quarterly reports about the company A review function: issuing opinions regarding the activities of the board of directors A reporting function: summaries of the activities conducted by the supervisory board itself (Koldakiewicz, 2001). The report issued by The World Bank in 2005 on Polish compliance with standards and codes of corporate governance includes an assessment of Polish supervisory board institutions. Following The World Bank's guidelines, Polish law grants supervisory boards extensive powers. Boards play an important part in the selection of CEOs and also have the final say on companies' strategies. Also, they effectively monitor the flow of information within and outside companies. However, concern has grown about the level of professionalism and of independence of the SOE supervisory board members. It is very difficult to resolve this issue due to the fact that the state still owns a considerable stake in many companies. Therefore, politicians have an adverse influence on a company's performance and on the composure of its board.

The World Bank acknowledged that the Polish government had made a huge effort in terms of implementation of accurate corporate policies. Similarly, the authors of the guidelines noticed that ownership concentration, securities market regulation, levels of foreign investment and general patterns of corporate organisation are moving towards Continental European norms. So far, however, most probably due to the high state share in state-owned companies, the government plays the key role in the Polish scene of corporate governance. This situation may change with the growing powers of pension funds that currently correspond to 10 per cent of the capitalization of the market.

Since Poland joined the EU, market capitalization has nearly doubled year on year. By the end of 2004 it equalled $71billion, while by the end of 2007 it was already $330 billion. Approximately 55 per cent of this relates to foreign companies and foreign capital. These figures are enough to keep Poland at the forefront of the countries that acceded to the EU in 2004. Although capitalization of the market has accelerated enormously in the recent past, it should be noted that there are a few large companies that themselves compose a significant stake in the market, for example, Unicredito, with a capitalization of approximately $90 billion (www.skarbiec.biz, 2008). Recently however, there have been changes. Due to the worldwide crisis, the capitalization of the market in Poland has dropped dramatically from $330 billion to approximately $150 billion. Bearing in mind that this happened in less than one year, this has had a dramatic effect on corporate governance development in Poland (www.gpw.com.pl, 2008).

Thus, the early nineties gave Poland the foundations of a capital market with basic institutions. Also, the freshly introduced concept of corporate governance had a chance to find its feet on the new ground. The second half of the 1990s brought further advances, especially thanks to the new legislation that came as the answer to the needs of that time, namely the Act of the Commercialisation and Privatisation of State Owned Enterprises, which came into effect from 30 th August, 1996. This Act allowed employees of companies to supervise and control enterprises they worked in to a higher degree than ever before. Since then it has become the rule that employees select two out of the five members of supervisory boards. Also, the nineties set new goals for Poland due to the country's accession to the OECD and the resulting commitments made by the Polish government. The most influential from the standpoint of corporate governance were the introduction of the Act on Investment Funds on 28 th August, 1997 and the Act on the Organisation and Functioning of Retirement Pension Funds. The above Acts introduced a very new group of institutional investors. Their major task involved effective investment of 'publicly collected' funds; therefore they became institutional shareowners rather than institutional shareholders. The latter Act also introduced retirement pension funds to the market. Their activity began on 1 st January, 1999. The oversight functions over all the above investors were performed by the newly established Retirement Pension Fund Supervisory Authority and the Investment Fund Association incorporated as a joint stock company (Koldakiewicz, 2001).

Despite the Polish achievements of the last twenty years in relation to corporate governance, many issues still need improvement. The high dispersion of shares at the beginning of the privatisation process was quickly reduced and governmental restrictions did not prevent excessive acquisitions (ROSC, 2005). Although supervisory boards have wide powers, their efficiency needs to be enhanced. According to Koldakiewicz the Polish supervisory system of the nineties was very similar to the one in Germany, especially with regard to Treasury owned companies. Also, there were similar solutions in relation to the participation of employees in the process of privatisation. However, there were remarkable differences in banks' attitudes and their activities. In Poland, as mentioned previously, banks tended to invest very passively, restricting themselves to lending funds necessary for investment and development.

Managers had no experience in investment banking and there was no such tradition, contrary to the situation in the German market. Additionally, certain banks that had considerable amounts of shares of other companies were not even privatised themselves. As a result, they had little interest in managing other companies.

The process of privatisation has been ongoing since the beginning of the 1990s. From this date until the end of August 2008, 5894 state-owned enterprises have been through the privatisation process. So far, 1688 companies have been fully commercialized and 2291 companies out of 5894 have been directly privatised as a result of the act of mass privatisation of state-owned companies. 2204 companies ceased to exist due to enhanced competition, ineffective management and so on (www.prywatyacja.msp.gov.pl 2008). In 1994, the state owned all of the 5894 companies and by the end of 2000, the government reduced the state's share in 99 companies to zero per cent. In the remaining entities the national share was reduced to 20 per cent (Grosfeld and Hashi, 2007).

Bearing in mind the above, currently there are over 1500 companies in which the Polish State Treasury still has a considerable share of around 20 per cent, which is one of the highest in Europe. Given this, Poland still has some way to go. The political party Platforma Obywatelska (PO) announced it will do its best to reduce the ratio to 10 per cent or even below (www.dziennil.pl, 2008). All the above has influenced the Polish corporate governance system. The direct connections between the world of politics and the economy are too strong. Many of the state controlled companies are in crucial sectors of the economy, e.g. finance, energy etc. What is more, according to the current regulations, managers of such companies which have a state share, cannot earn more than six times the average wage. The exact figure depends on whether the company is controlled directly or indirectly by the State Treasury. These restrictions may lead to more problems than benefits in the form of savings on managerial salaries. Representatives of PO say the wages of top management must be increased in order to attract highly skilled managers. They say this is part of the reason that these companies are struggling and cannot compete in the market. Therefore, Platforma Obywatelska is pursuing new legislation in order to change the situation (www.bankier.pl, 2008). Even international independent bodies highlight the fact that developing markets have to attract highly skilled managers and offer salaries comparable to those they can achieve abroad (Koldakiewicz, 2001).