Contrarian Profits and the Overreaction Hypothesis: the Case of the Athens Stock Exchange

European Financial Management, Vol. 12, No. 5, 2006, 763–787

Corporate Governance andInformation Efficiency inSecurity Markets

Charlie X. Cai, Kevin Keasey and Helen ShortLeeds University Business School, Maurice Keyworth Building, The University of Leeds, Leeds,LS2 9JT, UKe-mail: [email protected]

Abstract

This paper investigates a neglected topic in corporate governance research;namely, do governance characteristics affect the market reaction to news? Thetopic is important given the emphasis by governance regulations and codes of bestpractice on the need for greater transparency of corporate activities. For the firsttime in the corporate governance literature, we show that corporate governancecharacteristics (particularly the presence of founding family directors and genderdiverse boards) affect the market reaction to company specific news. The resultsof the paper point to the analysis of the impact of governance characteristics onthe market reaction to news being a new and complementary research agendawithin corporate governance.

Keywords: corporate governance; market reaction to news

JEL Classification: G34, G14

1. Introduction

The purpose of this paper is to analyse whether governance characteristics affect themarket reaction to news. The topic is important given the emphasis by governanceregulations (e.g., Sarbannes-Oxley, 2002) and codes of best practice (for example,the Combined Code, 1998) on the need for corporations to show greater levels oftransparency.

The results show, for the first time, that a small group of corporate governancecharacteristics (particularly the presence of founding family directors and gender diverse

We thank an anonymous referee, David Hillier and John Doukas, the Editor, for very helpfulsuggestions, as well as participants at the European Financial Managenent (EFM) CorporateGovernance Symposium held at Leeds in the spring of 2005. The usual disclaimer applies.Correspondence: Kevin Keasey.

C© 2006 The AuthorsJournal compilation C© 2006 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA02148, USA.

764 Charlie X. Cai, Kevin Keasey and Helen Short

boards) explain the market reaction to company specific news. If the results of this paperare shown to be robust by further work, there are clear policy implications. To date, stockexchange rules regarding the announcement of news and good governance practiceshave been largely kept separate but robust results from this new research agenda wouldsuggest the need for a more comprehensive and joined up approach to the regulation ofmarket activities.

The structure of the paper is as follows. Section two reviews the relevant parts of thecorporate governance literature and develops hypotheses. Section three follows with adescription of data and variables (especially the construction of the information measuresused to capture the market reaction to news), while section four provides results andsection five offers conclusions and suggestions for further research.

2. Hypotheses Development

Given this paper analyses whether corporate governance characteristics affect the marketreaction to news, hypotheses are developed from the corporate governance and disclosureliteratures.

In terms of disclosure, accounting research has investigated why companies voluntar-ily disclose information over and above that which is mandatory. Here voluntary disclo-sures are defined as disclosures in excess of requirements, representing free choices onthe part of company managements to provide accounting and other information deemedrelevant to the decision needs of users of their annual reports (Meek et al., 1995,p. 555).

An argument for higher levels of corporate disclosure is provided by agency theory(Jensen and Meckling, 1976) where disclosure presents managers with an opportunityto reduce the information asymmetry between themselves and owners. The incentivefor managers to disclose is that they can reduce investor concerns as to whether themanagement is acting in the owners’ interests. That is, disclosure is an opportunity formanagement to convince owners they are optimising shareholder wealth.

There have been a number of empirical studies which support the above type ofreasoning. Verrechia (1983) and Darrough and Stoughton (1990) both found thatcompetition among companies affected disclosure levels. Similarly Choi (1973), Foster(1986), Diamond and Verrecchia (1991) and Lev (1992) found that voluntary disclosurereduces the cost of capital, and Botosan (1991) and Sengupta (1998) found disclosureto affect the cost of equity and debt capital, respectively. Finally, Welker (1995), Healyet al. (1999) and Heflin et al. (2001) show that relative bid ask spreads are negativelyrelated to overall disclosure quality.

We now consider how various measures of the market reaction to news might berelated to corporate governance and company characteristics. More specifically, boardcharacteristics are likely to impact on how disclosure activities are managed andinterpreted, and ownership characteristics have the potential to influence how the marketreacts to disclosure activities.

2.1. Boards

2.1.1. Board independence. Board independence is defined here as the percentage ofthe non-executive (fully independent) directors on the board. The Combined Code(1998) defines independent directors as having no significant financial or personal

C© 2006 The AuthorsJournal compilation C© 2006 Blackwell Publishing Ltd, 2006

Corporate Governance and Information Efficiency 765

ties to the management apart from their fees and shareholdings, and no relationships orcircumstances which would affect the exercise of their independent judgment.

The contribution of non-executive (independent) directors to the board are based ontwo theories, agency theory and resource dependence theory; the latter of which dealswith the link between firms and their external environment due to outside non-executivedirectors’ experience and expertise (Haniffa and Cooke, 2000). One perspective is that ifthere are more non-executive directors on a board it will become more effective becausenon-executive directors are more capable of checking and monitoring managers andreducing agency problems (Fama and Jensen, 1983; Weisbach, 1988; Mak, 1996; Haniffaand Cooke, 2000); they are also influential due to their expertise, experience and successin providing access to information and ideas from outside (Kesner and Johnson, 1990;Wiersema and Bantel, 1992; Goodstein et al., 1994). In contrast, non-executive directorsmay lack company specific knowledge and excessive monitoring may take place as asubstitute for such knowledge (Baysinger and Butler, 1985; Haniffa and Cooke, 2000).

Fama and Jensen (1983), Forker (1992) and Haniffa and Cooke (2000) argue that themore independent non-executive directors on the board, the more power they will have toforce the management to release more information, enhance the information disclosedand reduce the benefits of withholding information. In support of these arguments,Chen and Jaggi (2002) found that a company with a higher proportion of independentdirectors on the board has a higher level of corporate disclosure.

The above arguments are summarised by the following null hypothesis:

• Hn1: There is no relationship between the proportion of independent directors on theboard and information disclosure.

2.1.2. Board size. When considering board independence, board size should be takeninto account as well since the number of directors can influence board activity and,potentially, disclosure activities. Because boards may be considered as a burden oncorporate resources, the ‘small is beautiful’ perspective has gained increasing popularity.It mainly deals with two issues – the direct costs of keeping large boards of directors(financial rewards and perquisites) and the indirect costs of greater bureaucracy, lessopenness, slower decision making and less entrepreneurial risk-taking, etc. Yermack(1996) studies a sample of 452 large US firms between 1984 and 1991 and finds thatthere is an inverse relationship between board size and firm value measured by Tobin’sQ. The result is robust to numerous controls for company size, industry membership,inside stock ownership, growth opportunities, and alternative corporate governancecharacteristics. He concludes that small boards of directors are more effective. Jensen(1993) supports the idea by suggesting that larger boards lead to less candid discussionof managerial performance and greater control by the CEO. Thus, larger board size canreduce the board’s ability to resist CEO control. So, a large board in a firm can makecoordination, communication, and decision making more cumbersome than is the caseof smaller boards.

However, Eisenberg et al. (1998) argue that larger boards can consist of moreoutsiders, who foster more careful decision-making because they bear external reputationcosts if projects fail or the firm encounters financial difficulties and their share ofany upside gain is limited. In addition, according to resource dependency theory byincreasing in size, boards help to link the organisation to its external environment andsecure critical resources, including prestige and legitimacy (Pearce and Zahra, 1992).



While the above arguments and evidence are not directly tied to disclosure, theysuggest that board size can have both positive and negative effects in terms of theactions of boards and hence the second null hypothesis to be explored here is:

• Hn2: There is no association between board size and information disclosure.

2.1.3. Founders. There is increasing interest (Anderson and Reeb, 2004) in the rolefounders have on board actions and corporate performance. McConaughy et al. (2001)found that founder family directors have more incentives to maximise firm values thannon-founder family ones. While a founder family dominated board may make decisionsin favour of the founder family shareholders at the expense of public investors, there aretwo potential influences on disclosure activities. First, since founder family directorsmay well have substantial shareholdings, disclosure may not be needed to influencethe buying behaviour of the shareholder base, essentially they have access to insiderinformation and will care less about public disclosure (Shleifer and Vishny, 1997).Second, as substantial shareholders, they have every incentive to ensure that ‘good’information is passed to the market and ‘bad’ information is kept back. In other words,it is possible that the proportion of founder family directors on the board has an impacton information disclosure:

• Hn3: There is no relationship between the proportion of founder family directors onthe board and information disclosure.

2.1.4. CEO/chairman split. A further aspect of the independence of boards is the splitresponsibility of the Chief Executive Officer (CEO) and the Chairman of the board. In theUK, due to the unitary board structure, independent board leadership is recommendedwith a clear division of roles between the Chairman and the CEO; with the Chairmanbeing in charge of running the board and the CEO being responsible for running thebusiness.

The arguments against the CEO/Chairman roles being vested in a single individualare grounded in agency theory (Jensen and Meckling, 1976). If the roles of CEO andChairman are separated, it will provide essential checks on management performanceand prevent powers from being concentrated on one potentially dominant person onthe board (Rechner and Dalton, 1991; Forker, 1992). The arguments for CEO/Chairmanroles being merged are based on the notion that managers are seen as acting in the interestof shareholders, so there will be no problem if the two roles are merged (Eisenhardt,1989; Dahya et al., 1996; Rechner and Dalton, 1991; Donaldson and Davis, 1991). Fora review of the arguments both for and against the splitting of the CEO and Chair roles,see Dahya and Travlos (2000).

In the case of information disclosure, the separation of the CEO and the Chairmanwill benefit the monitoring of management performance and increase the benefits fromdisclosing information (Forker, 1992). In fact, where the two roles are held by a singleindividual, there is evidence of a lower level of corporate disclosure (Forker, 1992; Guland Leung, 2002). The null hypothesis is formed as follows:

• Hn4: There is no relationship between CEO/Chairman split and informationdisclosure.



2.2. Director characteristics

2.2.1. Gender. In recent years, there has been pressure put on companies in both theUSA and UK to increase the diversity of the membership of corporate boards. Forexample, in the UK, the Higgs Report (2003) recommended that companies look at newways to identify and recruit non-executive directors so as to broaden the pool of talentavailable, in the belief that diversity will provide for a more effective board.

The gender mix of the board is one of the key measures of diversity used in thegovernance literature. However, research into the effect of gender diversity on corporateperformance has produced mixed results. Carter et al. (2003) found that the percentageof females on the board was positively and significantly associated with firm value,whilst Erhardt et al. (2003) found a significant and positive relationship between thepercentage of females on the board and accounting measures of performance. Adams andFerreira (2004) reported that firms which have more variability in their stock returnshave fewer women on their boards, suggesting that when uncertainty is high, grouphomogeneity is preferred. Farrell and Hersch (2005) report evidence to suggest that,rather than gender diversity resulting in improved performance, women tend to serveon the boards of better performing companies.

In the context of the present paper, we test the notion that board gender diversity mayimpact on the actions of the board, including disclosure activities. Adams and Ferreira(2004) provide evidence that firms with more diverse boards provide their directorswith more pay-performance incentives and hold more board meetings. Hence, our fifthhypothesis is as follows:

• Hn5: There is no relationship between the board diversity and information disclosure.

2.2.2. Professional background. The Cadbury Report (1992) suggested the professionaltraining of directors is an important issue and it is necessary for directors to undertakesome forms of external training (also see Smerdon, 1998). In addition, Haniffa andCooke (2000) argue that educational background will be an important explanatoryelement in information disclosure of the company. If there are more directors on the boardwith accounting or professional business backgrounds, they will tend to disclose moreinformation to improve the image of the firms and the credibility of the managementteam. Particularly, the accounting background of a financial director will be crucial forthe information released in the company’s annual report. Ahmed and Nicholls (1994)found that qualified directors receiving more professional training may disclose moreinformation. Therefore, the null hypothesis is:

• Hn6: There is no relationship between the proportion of directors holding professionalqualifications on the board and information disclosure.

2.3. Ownership characteristics

2.3.1. Shareholding of the board of directors. Verrecchia (2001) states that if thedirectors have incentives, such as beneficial shareholdings, they may tend to biasinformation disclosure to make it more or less favourable than truly expected. Dedman



(2004) argues that if managers’ remuneration includes the shares of the company, theymay tend to release news when their companies are inaccurately valued by the marketin order to increase their personal wealth. Similarly, Gray et al. (1984) point out thatdirectors may disclose information voluntarily if it is in their own interest. If directorsown the shares of the company, their interest will coincide with that of the shareholdersand they will pay more attention to the share price of the company, which fully reflectsthe performance and prospects of the corporation. In other words, the shareholdings ofthe directors may have an impact on the information disclosure by the board of directors.Therefore, the null hypothesis is:

• Hn7: There is no relationship between the shareholdings of board of directors andinformation disclosure.

2.3.2. Dominant shareholdings. While there has been a wealth of literature on block-holding and institutional shareholdings (see Short and Keasey, 2005), in terms of currentpurposes there is a complementary argument to that concerning directors’ shareholdings.Rather than being sensitive to disclosing information, there is in this situation likely tobe a heightened sensitivity to receiving and being receptive to information disclosuresbecause of the potential effects on wealth and the null hypothesis is as follows:

• Hn8: There is no relationship between dominant shareholdings and informationdisclosure.

2.4. Control variables

Prior literature suggests a range of control variables including share price, returnvolatility, firm size, profitability, gearing and industry classification (e.g. see Hanleyet al., 1993; Welker, 1995; Brockman and Chung, 1999; Sarin et al., 2000; Hedge andMcDermott, 2004). These control variables are defined in the next section along withdependent and independent variables.

3. Data, Variables and Descriptive Statistics

3.1. Data

As the majority of the approximately 2,000 companies listed on the London StockExchange (LSE) are not large, the sample chosen for this exploratory study consists of114 randomly selected non financial companies outside of the FTSE 350 for the years1999–2003 inclusive. While it has been the tendency of finance research to focus on thelargest companies because of data availability and quality, we feel the current samplewill be more representative of the vast body of listed companies than the usual sampleof very large companies. The sample of 114 firms was eventually chosen because ofthe amount of work involved in manually collecting the data and missing data. Thecorporate governance and financial accounting data are taken from the annual accountsof the sample companies.

In terms of the information measurement, intraday data are used in this study. Theraw intra day data (transaction, price, volume, etc.) are acquired from the LSE and news



announcements for each firm are collected via the Regulatory News Services (RNS)1

from the LSE website. The time period of the market and news data are closely matchedto the calendar time period of each firm’s financial year.

3.2. Variables

The description and definition of the full list of variables are described in Appendix A.While most of the variable definitions are self explanatory, the nature of the corporategovernance variables and the forms of the dependent variables need further explanation.

3.2.1. Corporate governance variables. Corporate governance first came to the fore inthe UK with the publication of the Cadbury Report on the Financial Aspects of CorporateGovernance in 1992. Since its publication, the corporate governance structures andpractice within UK companies have undergone significant changes in response to therecommendations of various committees and reports (Greenbury, 1995; Hampel, 1998,Combined Code, 1998, 2003). The Combined Code 1998 was in operation for the timeframe of the sample used in this paper. The key recommendations of the CombinedCode are set out in Appendix B.

Corporate governance in the UK, as embodied in the Combined Code is based onself-regulation and a ‘comply or explain’ approach; that is, companies are requiredby the Stock Exchange Listing Rules either to comply with the governance code or toexplain why they do not comply. It is then a matter for the shareholders to judge whetherthe reasons for non-compliance are satisfactory. For example, the Code advises that theroles of Chair and CEO should be split, but companies are not compelled to split theseroles. If the roles are combined, the company must set out in the annual report theirreasons for combining the roles.

The governance variables set out in Appendix A are collected from the annual reportsof the companies. In the annual report, all directors are named, their roles identified(e.g., CEO, Chair, etc.) and their independence or non-independence is stated. Profes-sional qualifications are provided in the biography of each director. The status of a direc-tor as a member of the firm’s founding family is gauged either from their biography orfrom a search of the company’s history.2 Shareholder data is taken from the annual report.Firms are required to state all external shareholdings equal to or greater than 3% of totalshareholdings. The shareholdings of directors are given in full in the annual report.

3.2.2. Adverse selection costs in trading (AS). The adverse selection component of thebid-ask spread has received increasing attention as a measure of informational efficiency.The market microstructure literature (see Copeland and Galai, 1983) argues that bid-askspreads have an information component due to the adverse selection faced by dealerswhen they trade with better informed traders. Dealers pass on potential losses incurredthrough dealing with informed traders via increased bid ask spreads. Key papers on

1The Regulatory News Service (RNS) is the system by which firms listed on the LondonStock Exchange make any announcement to the Stock Market which may have an effecton their share price and it is a condition of Listing that firms make announcements via theRNS.2The search of a company’s history is carried out via the company’s website or via on-linedatabases.



empirical measures of bid-ask spreads and its component parts are: Glosten and Harris(1988), Stoll (1989), George et al. (1991), Madhavan and Smidt (1991), Hasbrouck(1991), Foster and Viswanathan (1993), Lin et al. (1995), Madhavan et al. (1997) andHuang and Stoll (1997).

In terms of the purposes of this paper, Brennan and Subrahmanyman (1995), usingNYSE data, find a negative relationship between the adverse selection component ofa spread and the number of analysts following a share. The number of analysts isused as a proxy for the number of informed traders and information production. As aword of caution, however, when Van Ness et al. (2001) compared the adverse selectioncomponent of the bid ask spread to other measures of asymmetric information (marketto book, volatility and institutional ownership) they found mixed results. For example,analyst forecast errors and market to book were found not to be related to the adverseselection component of bid-ask spreads and proxies for informed traders provided mixedresults.

The adverse selection cost is estimated using high-frequency data. Three maindifferent approaches, which are developed in the market microstructure literature,are used in the current study to characterise components of spreads; namely, theGlosten and Harris (1988), the Huang and Stoll (1997) and the Lin et al. (1995)models.

Glosten and Harris (1988) develop a price formation model where order flow isuncorrelated through time. �pt can be written as:

�pt = c0 (Qt − Qt−1) + c1

(Qt V − Qt−1t Vt−1

) + z0 Qt + z1 Qt Vt + εt

Where �pt is the transaction price change, Vt is number of shares traded, Qt is tradeindicator variable which equals 1 (–1) if the trade at time t was a public buy (sell). Theadverse selection cost of a transaction is Z t = (z0 + z1Vt). The order processing andinventory holding components is Ct = (c0 + c1Vt). and ε t is a zero mean disturbanceterm that reflects price changes due to the arrival of public information.

The three different methods of calculating the adverse selection component [AS(1)is calculated in accordance with Glosten and Harris (1988), AS(2) in accordance withHuang and Stoll (1997), and AS(3) in accordance with Lin et al. (1995)] are utilised totest the robustness of the results.

The average adverse selection component is calculated using the average transactionvolume. The bid-ask spread in Glosten and Harris (1988) is the sum of the adverseselection, order processing and inventory holding components. The proportion which isthe adverse selection components is calculated as follows:

Z = z0 + zV

(z0 + z1V ) + (c0 + c1V )

Another measure of adverse selection costs is obtained from Huang and Stoll’s (1997)portfolio trading pressure model. The decomposition model is as follows:

�pt = S

2(Qt − Qt−1) + α

S

2Qt−1 + β

S

2Q∗

t−1 + εt

where �pt is the transaction price change, Qt is the trade indicator variable which equals1 (–1) if the trade at time t was a public buy (sell), Qt

∗ is the market wide trading pressureindicator which equals 1 (0, or –1) when the sum of all sample stocks of Qt is positive(zero, or negative). β is the adverse selection component expressed as the percentage ofthe spread.



Finally, according to the indicator model developed by Lin et al. (1995), the adverseselection component can be regressed by the following equation:

Qt+1 − Qt = λzt + et+1;

�Qt+1 = λzt + et+1

where Qt is the quote midpoint at time t, zt is the half effective spread at time t, λ is thespread’s adverse selection component and et is a normal distributed error term.

3.2.3. Probability of information-based trades (PIN). In their analysis of the informa-tional role of financial analysts, Easley et al. (1998a) develop a new approach to lookingat information. More specifically, they estimate the risk of information-based tradingfor a sample of NYSE stocks that differ in analyst cover. They use a model developedby Easley et al. (1996) in which the probability of informed trading can be directlyestimated using the daily number of buys and sell orders. The idea of the model is thaton days without private information a given buy and sell order flow is realised, whileadditional buy (sell) orders occur on days with positive (negative) private informationabout a stock’s fundamental value. In this model, the daily number of buys and sells aresufficient to estimate the probability of informed trading (PIN).

The PIN approach to informed trading has been used by a number of authors (Easleyet al. 1998b, considered the informed trader’s choice between stock and option markets,Easley and O’Hara, 2004, for stock splits, Grammig et al. 2001, for informed tradingon parallel markets and Easley et al. 2002, for the impact on asset pricing). None theless, it needs to be recognised that the PIN model has a potential weakness in that everyextraordinary trading activity is attributed to trading on private information. Recentevidence has shown that there is cross sectional correlation in order flow (Hasbrouckand Seppi, 2001; Chordia et al. 2002) and, therefore, other factors not related to security-specific private information impact trading activity.

The Easley et al. (2002) model is used to estimate PIN. The likelihood function is asfollows:

L (θ | B, S,) = (1 − α) e−εb (εb)Bi

Bi !e−εs (εs )Si

Si !

+ αδe−εb (εb)Bi

Bi !e−(μ+εs ) (μ+εs )Si

Si !

+ α(1 − δ)e−(μ+εb) (μ+εb)Bi

Bi !e−ε (εs )Si

Si !

Where Bi and Si are the number of buy and sell trades on day i respectively. θ{α, δ, ε, μ}are parameters to be estimated. The probability of information-based trade is calculatedas

PIN = αμ

αμ + εb + εs

3.2.4. The market impact of news (MIN). Different methodologies have been developedin finance research in order to quantify the market impact of news. The most widelyused methodology is the event study method. It measures the market impact of news(events) by the abnormal return in the event window. It requires a benchmark periodwhich is normal trading days without news release. The traditional event study method isconstrained to study non-frequent and single isolated events for a sample of firms. Giventhe purpose of this paper is to analyse the relationship between corporate governanceand information efficiency in general, the market impact of news cannot be constrained



to only a few corporate news releases. To quantify the MIN measure in general, theaccumulated impulse response (AIR) of market variables to all firm specific news areused. The AIR are obtained from a system of vector autoregressive (VAR) models (seeHamilton, 1994, chap. 11 for detailed discussions of VAR) with the number of newsreleases as an exogenous variable.

Generally speaking, impulse response coefficients are impact multipliers. Theymeasure the impact of a one-unit change in one variable (number of news release)on the other variable (market variables such as volatility and volume) over time. Thecumulative effects of unit impulses can be obtained by the summation of the coefficientsof the impulse response functions over a period of time. For example, the 5-day marketimpact of news can be measured by the summation of impulse response coefficientsover five days.

The cumulative impulse-response of volatility and volume to news is used as a measureof the market impact of news. The impulse-response functions are estimated from thefollowing vector autoregressive regression (VAR) model with the number of regulatorynews announcement (RNS) releases as the exogenous variable. The model is estimatedusing daily data.

The following form of VAR model is estimated:

yt = a +5∑

i−1

Ai yt−i + bxt + et

where

yt = (ALRt; VOLt), ALRt is the absolute log returns which is the measure of volatilityand VOLt is the log volume of trades at day t;

xt = RNSt, which is the number of RNS releases at day t;a = an (2 x 1) vector of intercept terms;Ai = (2 x 2) matrices of coefficients;b = an (2 x 1) vector of coefficients;et = an (2 x 1) vector of error terms.

The impulse response coefficients fi is an (2 x 1) vector which measures the effectof one unit RNSt–1 shock on ALRt and VOLt. These coefficients are impact multipliers.The accumulated effects of unit impulse in RNSt are obtained by the summation of thecoefficients of the impulse response functions:

gn =n∑

i=0

fi

The 5-day cumulative impulse response coefficients are used to measure the impactof RNS release on volatility and volume response. In order to make these news impactmeasurement comparable cross firms, the cumulative impulse response coefficients aredivided by volatility of daily return and average daily volume of the firms respectivelyto obtain standardised measures. Up to 12 days cumulative impulse response measureshave been use to check the robustness of the results and no material differences werenoted.

In order to further analyse market responses to news, the current research splits theannouncements into ‘good’ and ‘bad’ news, based on whether the 3-day post-event



abnormal return increases or decreases by 1% per day;3 events with abnormal returnswithin this 1% band are considered as ‘no news’ and dropped from the analysis. HenceALRG and VOLG represent the ‘good’ news variables and ALRB and VOLB representthe ‘bad’ news variables.4

3.3. Descriptive statistics

Descriptive statistics are provided in Table 1 and a correlation matrix in Table 2. Thenumber of observations for each variable varies due to missing values and issues ofestimation.

3.3.1. Dependent variables. In terms of the dependent variables, Table 1 shows that thevarious measures have average percentage values ranging from 6.8% for the bid-askadverse selection measure AS(3) (that is, 6.8% of the overall spread is a function ofadverse selection) to 23% for the probability of an informed trade.

3.3.2. Independent variables. Table 1 shows that on average, companies in the samplehad boards where 40% of the directors are independent, a size of 7 directors, 4% of thedirectors were from founding families and 90% had split the roles of the Chairman andCEO.5

In terms of director characteristics, only 3% of the directors in the sample were femalebut 31% of the directors had professional qualifications.

The average shareholdings of directors other than founders were 5% and in terms ofdominant shareholdings, the variable NumInstShare shows that on average companieshad 4 institutional shareholders.

Finally, the average size of the companies in sales revenue is £256 million and inmarket capitalisation £156 million, the companies had an average return of capital

3A regression base event-study method is proposed here to estimate abnormal returns:

ri,t = αi + βi rm,t ++10∑

τ=−5

γ i,τ Dτ,t + εi,t

Where,ri,t is the return of stock i at time t,rm,t is the return of FTSE All Share at time t,Dτ,t are 16 dummy variables which equal to one when t is in the event window (–5to +10 days around the event), and equal to zero otherwise.

The abnormal return during the event windows are then calculated as:

ARi,τ = r i,τ − (α̂i + β̂i rm,τ ),

where τ = −5, . . . 0, . . . ,+10.4Of course, it needs to be recognised that specific news announcements may either, dependentupon the exact context of the company and the market, increase or decrease volatility andvolume. Future research will examine this topic.5The referee of this paper made the interesting observation that the mere presence offounding families, female directors and institutional shareholders may be more importantthan that captured by percentage measures. Alternative specifications using dummy variablesproduced similar results to those reported here.



Table 1

Descriptive statistics of variables

This table reports descriptive statistics for dependent and independent variables (Panel A) and news

measurement categories (Panel B). Panel A reports the mean, minimum (Min), median, maximum

(Max) and number of firm-year observations (N) of the dependent and independent variables. A

list of variable descriptions can be found in Appendix A. Panel B reports the yearly mean, median

minimum (Min), maximum (Max), total number (Sum) of news announcements and number of firm-

year observations (N) for all news, good news and bad news categories.

Panel A. Descriptive statistics of variables

Variables Mean Min Median Max N

Dependent Variables

AS(1) 0.101 0.008 0.089 0.441 209AS(2) 0.105 0.009 0.095 0.459 192AS(3) 0.068 0.000 0.056 0.473 200PIN 0.238 0.045 0.234 0.556 238ALRG 0.046 −0.043 0.039 0.254 419VOLG 0.032 −0.147 0.031 0.222 419ALRB 0.033 −0.044 0.027 0.200 419VOLB 0.015 −0.216 0.016 0.177 419Independent VariablesBoard CharacteristicsIND 0.408 0.000 0.400 0.769 566BoardSize 7.095 4.000 7.000 13.000 566FD 0.043 0.000 0.000 0.500 566Split 0.898 0.000 1.000 1.000 566Director CharacteristicsFem 0.031 0.000 0.000 0.429 566Prof 0.313 0.000 0.250 1.000 566Ownership StructureDirShare 0.096 0.000 0.031 0.715 566NFDShare 0.054 0.000 0.017 0.554 566NumInstShare 4.387 0.000 4.000 13.000 566Control VariablesSales 255.639 1.471 153.189 1751.855 566MarketCap 155.733 2.170 110.586 2739.128 556ROCE 10.470 −103.140 10.015 142.090 560Gearing 27.318 0.000 23.850 175.600 566Vol 0.096 0.000 0.101 0.142 541STD 0.023 0.005 0.021 0.065 541Manufacturing 0.491 0.000 0.000 1.000 566Retail 0.237 0.000 0.000 1.000 566

Panel B. Yearly News Measurement Summary

Mean Median Min Max Sum N

newsall 21.91 19 4 95 9288 424GoodNews 4.63 4 1 18 1963 424BadNews 4.02 3 1 23 1703 424



employed of 10.5%, a gearing level of 27%, and a split between manufacturing, retailand other sectors in the ratio of 49%, 24% and 27%, respectively.

The above descriptive statistics indicate that while the sample is comprised ofsmaller cap companies, they are significant plc’s with large sales revenues and marketcapitalisations. However, even this size of company is likely to differ in its governancecharacteristics from the very largest plc’s (those in the FTSE 100 and FTSE 350) becauseof their need to have high degrees of transparency and alignment with the governancecodes of practice. The key differences between the present sample and a sample ofthe very largest plcs are likely to consist of the following: because of their smallersize, lower market liquidity and reduced news flow/coverage, the current companies arelikely to show greater reaction to an item of news; the boards are likely to be smaller,have fewer independent directors, have a greater percentage of founder directors, havea lower percentage split between Chairman and CEO roles, a lower degree of qualifieddirectors, a higher percentage of director ownership and a lower number of institutionalshareholders.

Given the above differences, we would expect the relationship between market reactionvariables and governance variables to be stronger for the current sample than for thevery largest plc’s listed on the LSE. However, it needs to be borne in mind that thecurrent sample of companies is more representative of the majority of companies listedon the LSE than is the case for the very largest companies.

4. Empirical Results

This section reports empirical results for each of the three separate information measuresof the market response to news announcements.

Table 2

Correlation matrix

This table reports the Pearson correlation matrices for the dependent and independent variables in

Panels A and B respectively. ∗, ∗∗ and ∗∗∗ denote significance at 10%, 5% and 1% levels, respectively.

Panel A. Dependent variables

AS(1) AS(2) AS(3) PIN ALRG VOLG ALRB VOLB

AS(1) 1

AS(2) 0.9764 1

∗∗∗AS(3) 0.4363 0.4422 1

∗∗∗ ∗∗∗PIN −0.0218 −0.0772 0.1588 1

ALRG 0.0219 −0.0379 0.0314 0.2013 1

∗∗VOLG 0.1684 0.175 0.018 0.0955 0.3145 1

∗∗ ∗∗ ∗∗∗ALRB 0.2435 0.238 0.1222 −0.0171 −0.0033 0.1038 1

∗∗∗ ∗∗∗ ∗∗VOLB 0.0799 0.0779 0.0659 0.0315 −0.0452 0.0379 0.2701 1

∗∗∗



Tab

le2

Conti

nued

.

Pane

lB

Cor

rela

tion

sof

inde

pend

ent

vari

able

s

LIND

LBoardsize

LFD

Split

LFem

LProf

LNFDShare

LNumInstShare

LSales

ROCE

LGearing

Vol

STD

Manufacturing

LB

oar

dsi

ze0.0

56

91

LF

D−0

.05

80.1

34

61

∗∗∗

Sp

lit

0.0

74

60.0

28

3−0

.07

61

∗∗

LF

em−0

.03

90.2

34

90.1

83

3−0

.02

51

∗∗∗

∗∗∗

LP

rof

−0.0

55

−0.0

1−0

.04

9−0

.04

2−0

.04

11

LD

irS

har

e−0

.17

90.1

48

80.4

93

−0.1

36

0.1

39

3−0

.02

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

LN

FD

Sh

are

−0.2

11

0.1

79

2−0

.04

2−0

.10

40.1

41

20.0

16

91

∗∗∗

∗∗∗

∗∗∗∗

∗L

Nu

mIn

stS

har

e0.1

03

7−0

.11

6−0

.24

40.1

75

9−0

.00

60.0

87

5−0

.13

61

∗∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

LS

ales

0.0

99

10.2

85

6−0

.14

70.0

71

40.0

37

−0.0

15

−0.0

99

−0.0

71

1∗∗

∗∗∗

∗∗∗

∗∗∗

∗R

OC

E−0

.00

50.0

26

10.0

56

8−0

.05

80.0

29

20.1

06

30.2

80

3−0

.00

80.0

90

31

∗∗∗∗

∗∗∗

LG

eari

ng

−0.0

22

0.0

39

7−0

.15

5−0

.00

5−0

.07

−0.1

08

−0.0

56

0.0

67

70. 2

68

8−0

.12

1∗∗

∗∗

∗∗∗∗

∗∗∗

∗V

ol

0.0

74

7−0

.06

2−0

.10

70.1

22

40.0

59

8−0

.08

−0.2

62

−0.0

32

0.2

03

4−0

.21

90.0

17

91

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗S

TD

−0.0

11

−0.0

58

0.0

35

70.0

08

0.0

35

8−0

.11

7−0

.09

2−0

.15

7−0

.14

3−0

.33

−0.1

69

0.4

69

51

∗∗∗

∗∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗M

anu

fact

uri

ng

0.0

86

9−0

.00

30.0

51

3−0

.01

8−0

.04

20.0

60

50.0

15

40.0

62

60.0

10

7−0

.03

40.0

09

5−0

.01

5−0

.02

21

∗∗R

etai

l0.0

38

0.0

36

−0.0

81

0.1

19

70.0

46

9−0

.08

−0.0

01

−0.0

21

0.2

10

60.0

97

10.0

92

1−0

.04

8−0

.09

1−0

.54

7∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗∗

∗∗∗



The relationship between the information measures and the independent variables isestimated via the following general form using OLS.

�i = βxi + εi

Where �i are the information measures (AS(1), AS(2), AS(3), PIN, ALR, or VOL) offirm i; xi is a vector of independent variables {LIND, LBoardSize, LFD, Split, LFEM,LProf, LNFDShare, LNumInstshare LSales, ROCE, LGearing, Vol, STD, Manufactory,Retail}; β is a vector of coefficients and ε i is the error term.

We take natural logarithms of a number of the independent variables to mitigateheteroscedasticity and to improve the normality of the continuous variables.6

4.1. The adverse selection measure (AS)

The OLS estimates shown in Table 3 indicate that only a few of the governance variableshave an effect on the adverse selection component of the bid-ask spread. Specifically,the number of founding family directors (LFD) has a positive effect on all measures ofAS and the number of female members of the board (LFem) has a negative effect on allmeasures of AS. The significant, positive coefficient on the percentage of directorswhich are members of the founding family (LFD) suggests that boards are givenless credence where there are more founding family directors. The adverse selectioncomponents are larger when there are higher levels of family board representation. Thisresult supports the argument that founding directors have every incentive to maximisethe value of the firm and hence may have a tendency to promote good news and holdbad news back.

The negative and significant coefficient on the LFem variable suggests that boardsare given more credence when the board has higher proportions of female members.Although such a finding appears to support the recommendation of the Higgs Report(2003) that diversity will provide for a more effective board, such a finding mayalso suggest that women tend to serve on boards that have lower adverse selectioncomponents.

The positive and significant (for a 1 tailed test) coefficient for the number ofinstitutional shareholders (LNumInstShare) in the AS(1) model is consistent withthe result for founding families in that where there is a larger number of influentialinstitutional shareholders, there may be a tendency to maximise firm value through newsmanagement and this is allowed for by the market. This result, however, contrasts withVan Ness et al. (2001) finding. Finally the negative effect of board size (LBoardSize)in the AS(3) model suggests that smaller boards are associated with lower levels ofinformation asymmetry, a finding which supports Yermack’s (1996) contention thatsmaller boards are more efficient.

The effect of the control variables is mixed. For the AS(1) and AS(2) models,profitability (ROCE) has a positive effect on the adverse selection component, whilst thestandard deviation of the daily returns (STD) and the industry dummies have a negativeeffect. However, for the AS(3) model, only the size variable (LSales) is significant.

Overall, these first set of results provide strong support that the presence of a foundingfamily member on the board and the presence of female directors influencing the

6To deal with the issue of zero values, log variables are constructed as the logarithm of the(original value plus 0.0001).



Table 3

Adverse selection regression

This table reports the OLS estimates of the following equation:

�i = βxi + εi

Where �i are the adverse selection measures (AS(1), AS(2) and AS(3)) of firm i; xi is a vector of

independent variables {LIND, LBoardSize, LFD, Split, LFEM, LProf, LNFDShare, LNumInstShare

LSales, ROCE, LGearing, Vol, STD, Manufacturing, Retail}; β is a vector of coefficients and ε i is

the error term. t-values are showed in the brackets. Rsqd and adjRsqd report the r2 and adjusted r2,

respectively. ∗, ∗∗ and ∗∗∗ denote significance at 10%, 5% and 1% level, respectively.

Parameter AS(1) AS(2) AS(3)

Intercept 0.1443 ∗∗∗ 0.1551 ∗∗∗ −0.0349(3.03) (3.00) (−0.64)

Board characteristicsLIND 0.0005 −0.0001 −0.0038

(0.19) (−0.02) (−0.82)LBoardSize −0.0053 −0.0143 −0.0362 ∗

(−0.26) (−0.65) (−1.73)LFD 0.0040 ∗∗∗ 0.0037 ∗∗ 0.0038 ∗∗

(2.74) (2.48) (2.48)Split 0.0093 0.0122 −0.0033

(0.85) (1.11) (−0.23)Director characteristicsLFem −0.0029 ∗ −0.0030 ∗∗ −0.0058 ∗∗∗

(−1.85) (−2.04) (−3.08)LProf −0.0017 −0.0010 −0.0017

(−0.98) (−0.60) (−1.20)Ownership structureLNFDShare 0.0006 0.0008 0.0047

(0.24) (0.27) (1.33)LNumInstshare 0.0041 ∗∗ 0.0040 0.0007

(2.13) (1.65) (0.20)Control variablesLSales 0.0035 0.0045 0.0129 ∗∗∗

(0.86) (1.02) (2.87)ROCE 0.0004 ∗∗ 0.0003 ∗∗ 0.0002

(2.25) (1.99) (0.89)Gearing −0.0002 −0.0005 0.0008

(−0.17) (−0.38) (0.42)Vol −0.4177 −0.5022 −0.0475

(−1.36) (−1.58) (−0.17)StD −1.1969 ∗∗ −0.9924 ∗ 0.6223

(−2.51) (−1.81) (1.07)Manufacturing −0.0281 ∗∗∗ −0.0275 ∗∗ −0.0072

(−2.66) (−2.49) (−0.63)Retail −0.0241 ∗∗ −0.0285 ∗∗ −0.0089

(−1.97) (−2.19) (−0.63)Rsqd 0.3043 0.2889 0.171adjRsqd 0.2382 0.2148 0.0897



information risk in security markets. Thus while governance variables have been foundto have an impact on the performance of companies, the results reported here alsoindicate that governance variables influence the information efficiency of their stock.

4.2. The probability of information-based trade measure (PIN)

The results for the probability of informed trading measure are shown in Table 4. Becausethe PIN measure focuses on the likelihood of private-informed trading, it provides adifferent insight into information efficiency. However, as discussed in sub-section 3.2.3,it needs to be recognised this measure has some potential weaknesses.

The governance variables which show significance are the split between the chairmanand CEO (Split) and the number of institutional shareholders (LNumInstShare). Thenegative relationship between PIN and Split indicates that splitting the role of leadinga company between the chairman and CEO improves the information diffusion processto the general public and this reduces the opportunity for private informed trades.Whereas, the positive relationship between the number of institutional shareholdersand PIN suggests that when there are more institutional investors in a company, thereis a higher probability of private informed trades; with private trades following the leadof better informed institutional shareholders.

The other variables concerned with the independence of the board and ownershipstructure have little significant impact on the probability of informed trading. In general,the results for the PIN measure show little significance and the adjusted R squaredmeasure has a value of 0.06.

4.3. The market impact of news measure (MIN)

The final measure of the market reaction to news is the cumulative 5-day response ofvolatility and volume to news. While the PIN measure in the last section focused onprivate information, the MIN measures concentrate on the impact of public announce-ments. As stated in section 3, these measures are split into ‘good’ and ‘bad’ news, withALRG and VOLG measuring the cumulative 5-day response of volatility and volumefor ‘good’ news and ALGB and VOLB measuring the cumulative 5-day response ofvolatility and volume for ‘bad’ news.

The results shown in panel A of Table 5 indicate that in the case of ‘good’ news,the variable LProf (the percentage of directors holding professional qualifications) ispositively related to volatility, whilst the percentage of independent directors on the board(LIND) is positively related to volume and the number of institutional shareholders isnegatively related to volume. Overall, these results suggest that for ‘good news, a moreindependent board with more directors with a professional background enhances themarket reaction to good news, whilst the number of institutional shareholders reducesthe market reaction to good news.

The results of panel B of Table 5 indicate that the governance variables have littleimpact on the market reaction to ‘bad’ news. The only governance variable to showstatistical significance is the percentage shareholding of non-founding family directors,which is positively related to the volume measure.

An interesting observation is the role of directors with professional qualificationswhich has a significant and positive effect on the market reaction to ‘good’ news, butno effect on ‘bad’ news. Given that firms will attempt to manage news in such a waythat ‘good’ news is transmitted to the market quickly and ‘bad’ news transmitted less



Table 4

PIN regression


�i = βxi + εi

Where �i are the probabilities of informed trades (PIN) of firm i; xi is a vector of independent

variables {LIND, LBoardSize, LFD, Split, LFEM, LProf, LNFDShare, LNumInstShare LSales, ROCE,

LGearing, Vol, STD, Manufacturing, Retail}; β is a vector of coefficients and ε i is the error term.

t-values are showed in the brackets. Rsqd and adjRsqd report the r2 and adjusted r2, respectively. ∗, ∗∗

and ∗∗∗ denote significance at 10%, 5% and 1% level, respectively.

Parameter PIN

Intercept 0.2462 ∗∗∗(3.37)

Board characteristicsLIND −0.0011

(−0.33)LBoardSize 0.0259

(1.16)LFD 0.0001

(0.06)Split −0.0392 ∗∗∗

(−2.79)Director characteristicsLFem −0.0019

(−1.13)LProf 0.0021

(1.34)Ownership structureLNFDShare 0.0044

(1.07)LNumInstshare 0.0048 ∗∗

(1.98)Control variablesLSales −0.0092

(−1.47)ROCE 0.0001

(0.46)Gearing 0.0001

(0.08)Vol 0.5153 ∗∗

(2.30)StD 1.9404 ∗∗∗

(2.92)Manufacturing −0.0068

(−0.48)Retail 0.0249 ∗

(1.72)Rsqd 0.129adjRsqd 0.0665



Table 5

Market impact of good/bad news regressions


�i = βxi + εi

Where �i are the market impacts of good/bad news measures (ALRG, VOLG, ALRB, VOLB) of

firm i; ALRG and VOLG are the volatility and volume impacts of good news and ALRB and VOLB

are the volatility and volume impacts of bad news. xi is a vector of independent variables {LIND,

LBoardSize, LFD, Split, LFEM, LProf, LNFDShare, LNumInstShare LSales, ROCE, LGearing, Vol,

STD, Manufacturing, Retail}; β is a vector of coefficients and ε i is the error term. t-values are showed

in the brackets. Rsqd and adjRsqd report the r2 and adjusted r2, respectively. ∗, ∗∗ and ∗∗∗ denote

significance at 10%, 5% and 1% level, respectively.

Panel A: Good news Panel B: Bad news

Parameter ALRG VOLG ALRB VOLBIntercept 0.0769 ∗∗∗ 0.0319 0.0773 ∗∗∗ 0.0233

(2.62) (1.19) (3.30) (0.71)Board characteristicsLIND 0.0000 0.0024 ∗∗∗ 0.0008 0.0010

(0.02) (3.06) (1.04) (1.00)LBoardSize −0.0037 −0.0013 0.0076 −0.0092

(−0.41) (−0.13) (1.04) (−0.97)LFD 0.0000 −0.0008 0.0006 −0.0003

(0.06) (−1.06) (0.78) (−0.31)Split 0.0014 0.0091 0.0007 −0.0042

(0.18) (1.52) (0.11) (−0.76)Director characteristicsLFem 0.0005 −0.0001 0.0001 −0.0003

(0.68) (−0.12) (0.19) (−0.44)LProf 0.0011 ∗∗ 0.0003 −0.0006 −0.0002

(2.23) (0.53) (−0.94) (−0.23)Ownership structureLNFDShare −0.0004 0.0002 0.0000 0.0022 ∗

(−0.22) (0.15) (0.00) (1.68)LNumInstshare −0.0005 −0.0016 ∗ −0.0003 0.0011

(−0.33) (−1.75) (−0.20) (1.04)Control variablesLSales −0.0020 0.0016 −0.0023 0.0016

(−1.17) (0.84) (−1.38) (0.89)ROCE 0.0002 ∗ 0.0000 0.0000 −0.0002 ∗∗

(1.68) (0.12) (−0.57) (−2.11)Gearing −0.0001 −0.0005 0.0003 −0.0010 ∗∗

(−0.16) (−1.00) (0.82) (−2.37)Vol −0.2018 −0.0169 −0.1774 0.1136

(−1.52) (−0.10) (−1.13) (0.88)StD 0.6901 ∗∗∗ −0.5350 ∗∗∗ −0.3832 ∗ −0.4040 ∗∗

(2.81) (−2.77) (−1.80) (−2.23)Manufacturing 0.0107 ∗∗ −0.0094 ∗∗ −0.0017 0.0024

(2.31) (−2.04) (−0.37) (0.53)Retail 0.0078 −0.0204 ∗∗∗ 0.0009 0.0017

(1.64) (−4.21) (0.16) (0.29)Rsqd 0.0622 0.0742 0.0506 0.0293adjRsqd 0.0261 0.0386 0.0141 −0.0079



quickly, this finding suggests that firms with more professional directors are able to dothis better than firms with fewer professional directors.

5. Conclusions

The results of this paper show that a small group of governance characteristics affectsinformation efficiency in a security market. Whilst the results are not strong for theprobability of information-based trades and the market impact of news, the results ofthe adverse selection costs regressions provide some strong support that the larger theproportion of the board from the founding family, the greater the information asymmetryin the market, and the more diverse the board in terms of the presence of female directors,the less the information asymmetry in the market.

Given the positive results of this exploratory study in terms of establishing arelationship between information efficiency and governance structures, future researchmight address how governance characteristics impact on the market reaction to newswhen news quality is further categorised, different types of news are isolated and largerfirms are considered.

References

Adams, R. B. and Ferreira, D., ‘Gender diversity in the boardroom’, ECGI – Finance Working PaperNo. 57/2004 (2004).

Ahmed, K. and Nicholls, D., ‘The impact of non-financial company characteristics on mandatory

disclosure compliance in developing countries: the case of Bangladesh’, The International Journalof Accounting, Vol. 29, 1994, pp. 62–77.

Andres, P. A., Azofra, V. and Lopez, F., ‘Corporate boards in some OECD countries: size, composition,

committee structure and effectiveness’, Corporate Governance, Vol. 13, 2005, pp. 197–210.

Anderson, R. C. and Reeb, D., ‘Board composition: balancing family influence in S&P 500 firms’,

Administrative Science Quarterly, Vol. 49, 2004, pp. 209–37.

Balachandran, J. and Bliss, M., ‘Board independence and voluntary disclosures: evidence from

Malaysia’, Working Paper (University of Hong Kong, 2002).

Baysinger, B. D. and Butler, H. N., ‘Corporate governance and the board of directors: performance

effects of changes in board composition’, Journal of Law, Economics and Organisations, Vol. 1,

1985, pp. 101–24.

Botosan, C., ‘The effect of disclosure level on cost of equity’, Accounting Review, Vol. 72, 1997,

pp. 323–50.

Brennan, M. and Subrahmanyam, A., ‘Investment analysis and price formation in securities markets’,

Journal of Financial Economics, Vol. 38, 1995, pp. 361–81.

Brockman, P. and Chung, D. Y., ‘An analysis of depth behavior in an electronic, order-driven

environment’, Journal of Banking and Finance, Vol. 23, 1999, pp. 1861–86.

Cadbury, A., Report of the Committee on the Financial Aspects of Corporate Governance (London:

Gee, 1992,).

Carter, D. A., Simkins, B. J. and Simpson, W. G., ‘Corporate governance, board diversity, and firm

value’, The Financial Review, Vol. 38, 2003, pp. 33–53.

Chen, C. J. P., and Jaggi, B., ‘Association between independent non-executive directors, family control

and financial disclosures in Hong Kong’, Journal of Accounting and Public Policy, Vol. 19, 2000,

pp. 285–310.

Choi, F., ‘Financial disclosure and entry to the European capital market’, Journal of AccountingResearch, Vol. 11, 1973, pp. 159–75.



Chordia, T., Roll, R. and Subrahmanyam, A., ‘Order imbalance, liquidity and stock returns’, Journalof Financial Economics, Vol. 65, 2002, pp. 111–30.

Combined Code, The Combined Code (London: Gee, 1998).

Combined Code, The Combined Code on Corporate Governance (London: Financial Reporting

Council, 2003).

Copeland, T. and Galai, D., ‘Information effects on the bid-ask spread’, Journal of Finance, Vol. 31,

1983, pp. 1457–69.

Dahya, J., Lonie, A. A. and Power, D. M., ‘The case for separating the roles of chairman and CEO: an

analysis of stock market and accounting data’, Corporate Governance, Vol. 4, 1996, pp. 71–77.

Dahya, J. and Travlos, N. G., ‘Does the one man show pay? Theory and evidence on the dual CEO

revisited’, European Financial Management, Vol. 6, 1, 2000, pp. 85–98.

Dalton, D. R., Daily, C. M., Ellstrand, A. E. and Johnson, J. L., ‘Meta-analytic reviews of board

composition, leadership structure, and financial performance’, Strategic Management Journal,Vol. 19, 1998, pp. 269–90.

Darrough, M. and Stoughton, N., ‘Financial disclosure policy in an entry game’, Journal of Accountingand Economics, Vol. 12, 1990, pp. 219–43.

Dedman, E., ‘Discussion of reactions of the London Stock Exchange to company trading statement

announcements’, Journal of Business Finance & Accounting, Vol. 31, 2004, pp. 37–47.

Diamond, D. W. and Verrecchia, R. E., ‘Disclosure, liquidity, and the cost of capital’, Journal ofFinance, Vol. 46, 1991, pp. 1325–59.

Donaldson, L. and Davis, J. H., ‘Stewardship theory or agency theory: CEO governance and shareholder

returns’, Australian Journal of Management, Vol. 16, 1991, pp. 49–63.

Easley, D., Hvidkjaer, S.and O.Hara, M., ‘Is information risk a determinant of asset returns’, Journalof Finance, Vol. 57, 2002, pp. 2185–2221.

Easley, D., Kiefer, N. M., O.Hara, M. and Paperman, J. B., ‘Liquidity, information, and infrequently

traded stocks’, Journal of Finance, Vol. 51, 1996, pp. 1405–36.

Easley, D. and O’Hara, M., ‘Information and the cost of capital’, Journal of Finance, Vol. 59, 2004,

pp. 1553–83.

Easley, D., O’Hara, M. and Paperman, J. B., ‘Financial analysts and information-based trade’, Journalof Financial Markets, Vol. 2, 1998a, pp. 175–201.

Easley, D., O’Hara, M. and Srinivas, P. S., ‘Option volume and stock price changes, on where informed

traders trade’, Journal of Finance, Vol. 53, 1998b, pp. 431–65.

Easley, D., O’Hara, M. and Saar, G., ‘How stock splits affect trading: a microstructure approach’,

Journal of Financial and Quantitative Analysis, Vol. 36, 2001, pp. 25–52.

Eisenberg, T., Sundgren, S. and Wells, M. T., ‘Larger board size and decreasing firm value in small

firms’, Journal of Financial Economics, Vol. 48, 1998, pp. 35–54.

Eisenhardt, K., ‘Making fast strategic decisions in high-velocity environments’, Academy of Manage-ment Journal, Vol. 32, 1989, p. 543.

Erhardt, N. L., Werbel, J. D. and Shrader, C. B., ‘Board of director diversity and firm financial

performance’, Corporate Governance, Vol. 11, 2003, pp. 102–11.

Fama, E. F. and Jensen, M. C., ‘Separation of ownership and control’, The Journal of Law andEconomics, Vol. 25, 1983, pp. 301–25.

Farrell, K. A. and Hersch, P. L., ‘Additions to corporate boards: the effect of gender’, Journal ofCorporate Finance, Vol. 11, 2005, pp. 85–106.

Forker, J. J., ‘Corporate governance and disclosure quality’, Accounting and Business Research,

Vol. 22, 1992, pp. 111–24.

Foster, F. D. and Viswanathan, S., ‘The effect of public information and competition on trading volume

and price volatility’, Review of Financial Studies, Vol. 6, 1993, pp. 23–56.

Foster, G., Financial Statement Analysis (Engle-wood Cliffs: Prentice-Hall, 1986).

George, T. H., Kaul, G. and Nimalendran, M., ‘Estimation of the bid-ask spread and its components:

a new approach’, Review of Financial Studies, Vol. 4, 1991, pp. 623–56.

Glosten, L. R. and Harris, L. E., ‘Estimating the components of the bid-ask spread’, Journalof Financial Economics, Vol. 21, 1988, pp. 123–42.



Goodstein, J., Gartum, K. and Boeker, W., ‘The effects of board size and diversity on strategic change’,

Strategic Management Journal, Vol. 15, 1994, pp. 241–50.

Gramming, J., Schiereck, D. and Theissen, E., ‘Knowing me, know you: trader anonymity and informed

trading in parallel markets’, Journal of Financial Markets, Vol. 4, 2001, pp. 385–412.

Gray, S., Shaw, J. and McSweeney, L., Information Disclosure and the Multinational Corporation(John Wiley and Sons, 1984.).

Greenbury, R., Director’s Remuneration: Report of a Study Group Chaired by Sir Richard Greenbury(London: Gee, 1995).

Gul, F. A. and Leung, S., ‘Board leadership and voluntary corporate disclosures’, Working Paper (City

University of Hong Kong, 2002).

Hambrick, D. C. and Mason, P. A., ‘Upper echelons: the organisation as a reflection of its top managers’,

Academy of Management Review, Vol. 9, 1984, pp. 193–206.

Hamilton, J. D., Time Series Analysis (Princeton: Princeton University Press, 1994).

Hampel, R., Committee on Corporate Governance: Final Report (London: Gee, 1998).

Haniffa, R. and Cooke, T., ‘Culture, corporate governance and disclosure in Malaysian corporations’,

Working Paper (Exeter University, 2000).

Hanley, K. W., Kumar, A. A. and Seguin, P. J., ‘Price stabilization in the market for new issues’,

Journal of Financial Economics, Vol. 34, 1993, pp. 177–97.

Hasbrouck, J., ‘Measuring the information-content of stock trades’, Journal of Finance, Vol. 46, 1991,

pp. 179–207.

Hasbrouck, J. and Seppi, D. J., ‘Common factors in prices, order flows, and liquidity’, Journal ofFinancial Economics, Vol. 59, 2001, pp. 383–411.

Healy, P., Hutton, A. and Palepu, K., ‘Stock performance and intermediation changes surrounding

sustained increases in disclosure’, Contemporary Accounting Research, Vol. 16, 1999, pp. 485–

520.

Heflin, F., Shaw, K. W. and Wild, J. J., ‘Disclosure quality and market liquidity’, SSRN Working PaperSeries (2001).

Hedge, S. P. and McDermott, J. B., ‘Firm characteristics as cross-sectional determinants of adverse

selection’, Journal of Business Finance & Accounting, Vol. 31, 2004, pp. 1097–1124.

Higgs, D., Review of the Role and Effectiveness of Non-Executive Directors (London: Department of

Trade and Industry, 2003).

Huang, R. D. and Stoll, H. R., ‘The components of the bid-ask spread: a general approach’, Review ofFinancial Studies, Vol. 10, 1997, pp. 995–1034.

Jensen, M. C., ‘The modern industrial revolution, exit, and the failure of internal control systems’,

Journal of Finance, Vol. 48, 1993, pp. 831–80.

Jensen, M. C. and Meckling, W. H., ‘Theory of the firm: managerial behavior, agency costs and

ownership structure’, Journal of Financial Economics, Vol. 3, 1976, pp. 305–60.

Kesner, I. F. and Johnson, R. B., ‘An investigation of the relationship between board composition and

shareholder suits’, Strategic Management Journal, Vol. 11, 1990, pp. 327–36.

Lev, B., ‘Information disclosure strategy’, California Management Review, 1992, pp. 9–32.

Lin, J. C., Sanger, G. and Booth, G., ‘Trade size and components of the bid–ask spread’, Review ofFinancial Studies, Vol. 8, 1995, pp. 1153–83.

Madhavan, A., Richardson, M. and Roomans, M., ‘Why do security prices change? A transaction-level

analysis of NYSE stocks’, Review of Financial Studies, Vol. 10, 1997, pp. 1035–64.

Madhavan, A., and Smidt, S., ‘A Bayesian model of intraday specialist pricing’, Journal of FinancialEconomics, Vol. 30, 1991, pp. 99–134.

Mak, Y. T., ‘The voluntary use of outside directors by initial public offering firms’, CorporateGovernance, Vol. 4, 1996, pp. 94–106.

McConaughy, D. L., Matthews, C. H. and Fialko, A. S., ‘Founding family controlled firms: perfor-

mance, risk, and value’, Journal of Small Business Management, Vol. 39, 2001, pp. 31–49.

Meek, G., Gray, S. and Roberts, C., ‘Factors influencing voluntary annual report disclosures by US, UK

and continental European multinational corporations’, Journal of International Business Studies,

Vol. 26, 1995, pp. 555–72.



Pearce, J. H. and Zahra, S. A., ‘Board composition from a strategic contingency perspective’, Journalof Management Studies, Vol. 29, 1992, pp. 411–38.

Rechner, P. L. and Dalton, D. R., ‘CEO duality and organizational performance: a longitudinal analysis’,

Strategic Management Journal, Vol. 12, 1991, pp. 155–60.

Sarbannes-Oxley Act, US Congress website, at H.R. 3763. ENR (2002).

Sarin, A., Shastri, K. and Shastri, K., ‘Ownership structure and stock market liquidity’, Working Paper(University of Pittsburg, 2000).

Sengupta, P., ‘Corporate disclosure quality and the cost of debt’, Accounting Review, Vol. 73, 1998,

pp. 459–74.

Shleifer, A. and Vishny, R., ‘A survey of corporation governance’, Journal of Finance, Vol. 52, 1997,

pp. 737–83.

Short, H., ‘Non-executive directors, corporate governance and the Cadbury report: a review of the

issues and evidence’, Corporate Governance, Vol. 4, 1996, pp. 123–31.

Short, H. and Keasey, K., ‘Institutional shareholders and corporate governance in the UK’, in K.

Keasey, S. Thompson, and M. Wright, eds, Corporate Governance, Accountability, Enterprise andInternational Comparisons (Chichester: John Wiley & Sons Ltd, 2005).

Smerdon, R., A Practical Guide to Corporate Governance (London: Sweet & Maxwell, 1998).

Stoll, H. R., ‘Inferring the components of the bid/ask spread: theory and empirical tests’, Journalof Finance, Vol. 44, 1989, pp. 115–34.

Van Ness, B. F., Van Ness, R. A. and Warr, R. S., ‘How well do adverse selection components measure

adverse selection?’ Financial Management, Vol. 30, 2001, pp. 77–98.

Verrecchia, R., ‘Discretionary disclosure’, Journal of Accounting and Economics, Vol. 5, 1983,

pp. 179–94.

Verrecchia, R., ‘Essays on disclosure’, Journal of Accounting and Economics, Vol. 32, 2001,

pp. 97–180.

Weisbach, M. S., ‘Outside directors and CEO turnover’, Journal of Financial Economics. Vol. 20,

1988, pp. 431–60.

Welker, M., ‘Disclosure policy, information asymmetry, and liquidity in equity markets’, ContemporaryAccounting Research, Vol. 11, 1995, pp. 801–827.

Wiersema, B. D. and Bantel, K. A., ‘Top management team demography and corporate strategic

change’, Academy of Management Journal, 35, 1992, pp. 91–121.

Yermack, D., ‘Higher market valuation of companies with a small board of directors’, Journal ofFinancial Economics, Vol. 40, 1996, pp. 185–211.

Appendix A: Variables Description

This appendix reports the description of the variables used in the analyses. The firstcolumn gives the category and brief description of the variable. The second column givesthe variable names used in the analyses. The third column gives detail descriptions ofthe variables.

Dependent variables

Adverse selection Adverse selection component of bid-ask spreadestimated using:

AS(1) Glosten and Harris (1988) modelAS(2) Huang and Stoll (1997) market pressure model.AS(3) Lin et al. (1995) model

Probability ofinformed trades

PIN Probability of informed trades estimated usingEasley et al. (1998) model.



Continued.

Dependent variables

Market impact ofnews

ALRG, VOLGALRBVOLB

The 5-day cumulative impulse response ofvolatility and volume to news estimated from theVAR model split into good (G) and bad (B) newsbased on the post event return being 1% per day

Independent variables

Board characteristicsBoard independence LIND Log percentage of independent directors on board.Board size LBoardSize Log total number of directors on board.Founder LFD Log percentage of number of founding family

directors on board.CEO/chairman split Split Dummy variable which equals one for companies

where CEO and Chairman positions are not heldby the same person, and zero otherwise.

Directorcharacteristics

Gender LFem Log percentage of female directors on board.Professional

backgroundLProf Log percentage of directors holding professional

qualification such as (ACA, ACCA, CIMA,etc.).

Ownership structureShareholding of the

board of directorsLDirShare Log percentage of shareholding of all directors.

LNFDShare Log percentage of shareholding of non-founderfamily directors.

Dominantshareholding

LNumInstShare Log number of institutional share holders whoseholding is larger than 3% in the firm

Control variablesSize LSales Log of sales in million pounds

LMarketCap Log of market capitalisation in million poundsProfitability ROCE Return on capital employedRisk Gearing Debt divided by debt plus equityVolume Vol Market turnover which is measured as the daily

volume of trades over market capitalisationVolatility STD Standard deviation of daily returnsIndustry dummies Manufacturing Equal to one if a company is in the manufacturing

industry (UK-SIC code between 15 to 45)Retailing Equal to one if a company is in the retail industry

(UK-SIC code between 50 to 63)



Appendix B: The Combined Code 1998

This appendix gives a brief summary of the Combined Code 1998.

Directors

• The combination of the roles of CEO and chair should be publicly explained.• A senior independent NED should be identified in the annual report (regardless of

whether CEO and chair are combined).• There should be a minimum of three NEDs and NEDs should comprise of not less

than one third of the board. The majority of NED should be independent.• Independent NEDs are those who are ‘independent of management and free from

any business or other relationships which could materially interfere with theexercise of their independent judgement’.

• Nomination committees should be established (unless the board is small)comprising of a majority of NEDs, chaired by the board chair or NED and allmembers should be identified in the annual report.

• All directors should be required to submit themselves for re-election at regularintervals and at least every three years.

Directors’ remuneration

• The remuneration committee should consist of independent NEDs.• The remuneration report should be in the name of the board, rather than of the

remuneration committee.• Full disclosure is required of all elements of the remuneration package (including

share options and pension entitlements) of each named director.• Remuneration report should provide an explanation of the company’s policy on

the setting of executive remuneration.• Executive directors’ contracts exceeding one year should be disclosed and

explained.• Shareholders’ approval is required for the adoption of long-term incentive plans• Awards in one large block rather than phased in should be explained and justified.

Shareholders

• Companies should indicate level of proxy votes logged on each resolutionproposed at the AGM and the balance for and against each.

• Companies should propose a separate resolution at the AGM on each substantiallyseparate issue.

• The chairmen of the audit, remuneration and nomination committees should beavailable to answer questions at the AGM.

• The Notice of the AGM and related papers should be sent to shareholders at least20 working days before the AGM.

Accountability and audit

• Audit committees should consist of at least three NEDs, the majority of whomshould be independent and all members are to be named in the annual report

• The board should, at least annually, conduct a review of the effectiveness of thegroup’s system of internal controls and should report to shareholders that theyhave done so.

• The directors should report that the business is a going concern.


Contrarian Profits and the Overreaction Hypothesis: the Case of the Athens Stock Exchange

Documents

Transcript of Contrarian Profits and the Overreaction Hypothesis: the Case of the Athens Stock Exchange