Board Structure and Agency Costs

M. Ameziane Lasfer *

City University Business School, Barbican Centre, London EC2Y 8HB, UK

Abstract.

The purpose of the paper is to test the hypothesis that board structure and its impact on value is afunction of firm’s growth opportunities. Consistent with this hypothesis, the results show that, whilelow growth firms are less likely to have an independent board, i.e., to split the roles of thechairman and CEO, to have a high proportion of non-executive directors and to appoint a non-executive as a chairman, their value is positively related to these board structure variables. Incontrast, for high growth firms, the relationship between board structure and firm value is weak,suggesting that board structure does not always mitigate the agency conflicts. The results suggestthat imposing the same board structure for all companies independently of their specificcharacteristics is likely to reduce the value of firms that may be forced to depart from optimalcorporate governance structures which have been successful.

Key words: Corporate Governance; Growth opportunities, Non-executive directors; Split ofroles of chairman and CEO

JEL Classification: G30; G32

This draft: 27 May, 2002

* Tel: (+44)(0) 2070408634; Fax: (+44)(0) 2070408881; email: m.a.lasfer@city.ac.uk.I would like to thank John McConnell (the editor) and an anonymous referee for helpful comments. Iam also grateful to Mara Faccio for her suggestions and contribution to an earlier version of thepaper and to Yakov Amihud, Lorenzo Caprio, Diane Dennis, Michael Hammerslag, Larry Lang,Gulnur Muradoglu, Geof Stapledon and seminar participants at Bilkent University, HacettepeUniversity (Turkey), the 1999 EFMA meeting in Paris, and the 2000 FMA Meeting in Seattle forvaluable insights. Aslihan Ersoy supplied the latest UK data used in this paper. The usual disclaimerapplies.

2

Board Structure and Agency Costs

Abstract.

The purpose of the paper is to test the hypothesis that board structure and its impact on value is a

function of firm’s growth opportunities. Consistent with this hypothesis, the results show that, while

low growth firms are less likely to have an independent board, i.e., to split the roles of the

chairman and CEO, to have a high proportion of non-executive directors and to appoint a non-

executive as a chairman, their value is positively related to these board structure variables. In

contrast, for high growth firms, the relationship between board structure and firm value is weak,

suggesting that board structure does not always mitigate the agency conflicts. The results suggest

that imposing the same board structure for all companies independently of their specific

characteristics is likely to reduce the value of firms that may be forced to depart from optimal

corporate governance structures which have been successful.

Key words: Corporate Governance; Growth opportunities, Non-executive directors; Split of roles ofchairman and CEO

JEL Classification: G30; G32

This draft: 27 May, 2002

3

I. Introduction

Research on corporate governance has identified a number of mechanisms intended to

insure that management teams act in the best interests of shareholders.1 These include external

mechanisms such as institutional ownership, large creditors, long-term relationships, debt financing,

and the market for managerial labour, and internal mechanisms which include managerial ownership,

executive compensation and the board of directors.

In this paper I focus on the role of the board of directors in controlling the agency conflicts

between managers and shareholders. This issue is controversial. On the one hand, organisations

such as investment funds, stock exchanges, professional bodies and corporate governance task

forces, try to impose a “one size fits all” board with majority seats held by non-executive directors

and the roles of the chief executive officer and chairman are split.2 These proposals aim at providing

adequate safeguards for investors’ capital and reflect the concern about the way in which

remuneration packages for senior executives have been determined, the role of the CEO in

directors’ appointments, the spectacular collapse of a number of large companies and by the

fraudulent use of funds such as the use of pension fund of Mirror Group Newspapers to finance

an illegal scheme for supporting the share price of Maxwell Communications Corporation. These

proposals are also based on academic studies that show that outside directors play a significant role

in protecting shareholders’ interests when a good decision control, such as management turnover or

the adoption of poison pill, is required (e.g., Weisbach, 1988, Brickley et al., 1994, and Cotter et al,

1997) and by the proposition that outside directors have incentives to make decisions that signal their

abilities as efficient decision-makers (e.g., Borokhovich et al. 1996, Weisbach, 1988).3

On the other hand, a number of academic studies show that board composition varies with

both firm specific factors and the institutional environment the firm faces (e.g., Brickley and James,

1987, Denis and Denis, 1994, and Hermalin and Weisbach, 1988), suggesting that imposing a

homogeneous board may be optimal for some firms but not for others. In addition, there is evidence

that it is not the board of directors per se that reduces the agency conflicts and creates value, but

rather its size, number of meeting and its power. For example, consistent with Jensen’s (1993)

argument that a value-relevant attribute of corporate boards is their size, Yermack (1996) shows that

companies with smaller boards have high value. Vafeas (1999) reports a negative relationship

between firm value and the number of board meetings. Jensen (1993), and Hermalin and Weisbach

(1991) argue that the CEO often ends up controlling the composition of the board and lessening its

monitoring role. In this perspective, boards evolve over time as a function of the bargaining power of

the CEO, and managers tend to reduce the power of boards as their equity ownership increases

resulting in a weak relationship between board structure and firm value (e.g., Denis and Sarin, 1999,

Hermalin and Weisbach, 1991, Mikkelson and Partch, 1997, Weisbach, 1988). Franks et al., (1998)

4

show that managers dominate the board and that high managerial turnover rates are confined only to

the worse performing firms.

These arguments imply that boards are not always effective and imposing a single board

model for all companies is likely to increase costs and result in a reduction of shareholder wealth.

The present study adds to this line of literature by suggesting that the effectiveness of the board of

directors in reducing the agency costs depends on the firm’s growth characteristics. Following

McConnell and Servaes (1995), I expect agency conflicts to be dependent on firm’s growth

opportunities. In particular, if agency problems are associated with conflicts over the use of free

cash flow, then governance mechanisms are expected to play a minor role within high growth firms

but a significant role within low growth firms which have usually substantial free cash flow and have

a tendency to overinvest by accepting marginal investment projects with negative net present values

(Jensen, 1986). However, if, as argued by Smith and Watts (1992), agency problems are associated

with greater information asymmetry, then board structure is expected to play a significant role only in

high growth firms. In both information asymmetry and free cash flow cases, I expect board structure

to be dependent on firm’s growth opportunities. If the board of directors acts as a monitoring device

in reducing the free cash flow (information asymmetry) problem, low (high) growth firms are

expected to have small and independent boards with a high proportion of non-executive directors and

where the roles of the chairman and CEO are split, and such board structure will be significantly

related to firm value.

I also test the stability of this relationship in the pre- and post Cadbury periods. In 1992, the

Cadbury Committee issued the Code of Best Practice which recommends that the offices of the

chairman and the chief executive officer should be separated to prevent excessive concentration of

power in boardrooms and that companies should appoint independent non-executive directors with

high caliber so that their views will carry weight in board discussions. The main reservation of the

code centers on the issue of compliance and enforcement as the corporate governance system in the

UK has traditionally stressed the importance of internal controls and financial reporting and

accountability as opposed to a large amount of external legislation. In this spirit, the Code of Best

Practice is entirely voluntary but as a continuing obligation of listing, the London Stock Exchange

requires all companies registered in the UK to state, after June 1993, whether they are complying

with the code and to give reasons for any areas of non-compliance. Therefore, if the code is adopted

by companies to conform to the London Stock exchange listing requirement, then we would expect

all firms, independently of their level of agency costs, to comply with the recommendations and the

relationship between the board structure and agency conflicts will be weak. In contract, if companies

chose to adopt the code, then only companies that suffer from the agency costs are expected to do

so.

5

I analyse the board structure of 627 UK non-financial listed companies in the pre-Cadbury

period (1990-1991) and 1171 companies in the post-Cadbury period (1996-1997). As expected,

Cadbury had a substantial impact on the board structure of UK companies as, for example, the

proportion of companies that split the roles of the chairman and CEO and appoint a minimum of

three non-executive directors (i.e., adopt the Cadbury recommendations) increased significantly from

27% to 60%. However, in both periods, high growth firms are more likely to have such board

structure than low growth firms. For example, in the pre-Cadbury period 32% of high growth

companies had already adopted the recommendations, compared to 22% in the low growth sample

(p = 0.00). In the post-Cadbury period, the respective proportions are 63% and 58% (p = 0.06).

Further analysis reveals that high growth firms are more likely to have an independent board with a

larger number and proportion of non-executive directors than low growth firms. The split of the

samples into growth quintiles shows a monotonic relationship between growth levels and the

proportion of companies that adopted the code, the proportion of non-executive directors and the

appointment of a non-executive director as a chairman. The results also show that the adoption

probability is significantly related to firms’ growth opportunities. In particular, over the two periods,

30% of low growth firms always adopted the recommendations, compared to 17% for high growth

firms. These results suggest that board structure is not homogeneously distributed across firms’

growth potentials.

The analysis of the relationship between firm value and board structure shows that, despite

their high adoption propensity, high growth companies do not fully benefit from the Cadbury

recommendations. While in the pre-Cadbury period, the relationship between firm value and board

structure is weak for both low and high growth firms, in the post-Cadbury period, board structure

affects only low growth firms’ values. In particular, for low growth firms, there is a strong and

positive relationship between firm value and the adoption of the Cadbury recommendations, the

proportion of non-executive directors and the appointment of a non-executive director as a chairman.

The responsiveness of firm value to board structure is steeper in the post- compared to the pre-

Cadbury period, suggesting that the economic significance is also higher and that the quality of the

board is valued more in the post-Cadbury period. In contrast, for high growth firms, the relationship

between firm value and the adoption of the Cadbury recommendations is actually negative and

significant. This negative relationship holds also for the two components of the recommendations,

i.e., the split of the roles of the chairman and CEO which is negative but not significant and the

appointment of three or more non-executive directors which is negative and statistically significant.

These results suggest that, for high growth companies, the adoption of the Cadbury

recommendations could lead to value destruction. The proportion of non-executive directors and the

appointment of a non-executive as a chairman are positive but not significant. Consistent with

6

previous evidence (e.g., Yermack, 1996) the relationship between firm value and number of

directors is negative and significant for both sets of firms, suggesting that larger corporate boards

result in poorer communication and decision making. However, unlike Yermack (1996), the

relationship between firm value and number of executive directors is also negative and significant for

both low and high growth firms. The results simulated for various measures of value and for a

sample of 744 UK firms in 1998/99 are qualitatively similar. The results hold also for Australian

companies which are faced with similar institutional framework as their UK counterparts but are not

required to adopt a corporate governance code similar to Cadbury.

In sum, the results suggest that, unlike low growth firms, high growth firms do not benefit

from the Code of Best Practice, implying that a “one size fits all” board with majority seats held by

non-executive directors and the roles of the chief executive officer and chairman are split does not

necessarily lead to reduction in agency costs and value creation for all companies. Instead, the

results suggest that low growth firms are easy to monitor because their value is predominantly

determined by their assets-in-place while the behaviour of managers of high growth firms is difficult

to observe and to monitor given their discretionary investment opportunities and the complexity in

their decision-making. In this case, the results are consistent with Smith and Watts (1992) who argue

that high growth firms will dispose of other internal and/or external mechanisms to control the

agency conflicts leaving little role for board structure. One such mechanism is managerial ownership

which is found to be statistically significant for high growth but not for low growth firms. Thus, the

results suggest that the high propensity of high growth firms to adopt the Cadbury recommendations

is not likely to mitigate their agency conflicts but to comply with the London Stock Exchange listings,

probably to provide an ‘image’ and to access external markets.

The rest of the paper is structured as follows. The next Section presents the theoretical

background. Section III describes the data and the methodology. Section IV presents the results and

the conclusions are in Section V.

II Theoretical Background

Previous studies find that the size and the composition of the board are correlated with the

quality of the board’s decisions of CEO replacement, acquisitions, poison pills, and executive

compensation.4 For example, Borokhovich et al. (1996) find a positive monotonic relation between

the proportion of outside directors and the likelihood that an outside director is appointed as CEO and

such an appointment benefits shareholders. Rosenstein and Wyatt (1990) find that the appointment

of outside directors is followed by a significant share price increase, even when outside directors

already dominate the board. In another study, Rosenstein and Wyatt (1997), find that the stock price

reaction to the announcement of the appointment of a new inside director depends on the proportion

7

of shares held by the same director. In particular, the reaction is negative when the director owns

less than 5%, positive when the director's ownership ranges between 5% and 25%, and it is not

significant when the director's ownership exceeds 25%. This suggests that the benefits of the

appointment of an inside director exceed the costs of managerial entrenchment only when the

interest of managers and outside directors are closely aligned. Brickley, Coles and Terry (1994) find

that the adoption of poison pills has a negative effect on share prices only when the board is

dominated by insiders, while stock prices react positively when outsiders dominate the board.

Weisbach (1988) finds a strong relationship between managerial turnover and performance within

firms with outsider-dominated boards. These studies recognise the importance and monitoring

effectiveness of outside directors and are consistent with Fama and Jensen (1983) arguments that

reputation and the threat of legal action motivate outside directors to act in the best interest of

shareholders.

However, Jensen (1993) and Hermalin and Weisbach (1991) argue that CEOs often end up

controlling the composition of the board and lessening its monitoring role. They do this by being

themselves chairmen and adopting large boards to increase communication problems among board

members. Consistent with these arguments, Yermack (1996) and Eisenberg et al., (1998) report a

negative relationship between board size and firm value, for large and small firms, respectively.

Other studies report that boards appear to evolve over time as a function of the bargaining power of

the CEO relative to the existing directors and that managers tend to reduce the monitoring role of

boards as soon as their equity ownership increases. Denis and Sarin (1999), Hermalin and Weisbach

(1991), and Weisbach (1988) find that insider ownership is inversely correlated to the proportion of

outside directors. Mikkelson and Partch (1997) report an inverse relationship between the stake of

directors and officers in the firm and the turnover of directors and top-officers. Similarly, Yermack

(1996) reports a negative relationship between CEO stock ownership and the probability that the

CEO is replaced. Denis and Denis (1994) find that the rate of top-managers' turnover is about a half

in majority-owned, as opposed to widely held, firms.

Bhagat and Black (1998), in their review paper, present a different view on the

effectiveness of board composition. They argue that previous studies do not tell us how board

composition affects overall firm performance as companies with independent boards could perform

better on particular tasks, yet worse on other unstudied tasks, leading to no net advantage in overall

performance. In addition, they show that previous results (e.g., Weisbach, 1988) have marginal

effects on firm value, while others (e.g., Yermack, 1996) are not strong to the choice of

performance measures. Instead, they analyse the relationship between board independence and

long-term performance of large firms in the US to find that firms with the most independent boards

8

perform worst that firms with more balanced boards. The relationship between board structure and

firm value is also dependent on the testing methodology used. For example, Agrawal and Knoeber

(1996) examine the inter-relationships among seven “control mechanisms” using a six-equation

simultaneous model to find that the proportion of outsiders on the board is significantly negative

determinant of firm value but with the three-stage least square method the relationship is

insignificant.

The purpose of the paper is to test the hypothesis that board structure and its impact on

value is a function of the firm’s growth opportunities. I expect the agency conflicts between

managers and shareholders, and thus the monitoring role of the board of directors, to be a function of

firm’s growth opportunities. However, the direction of this relationship depends on whether the

agency conflicts are created by the free cash flow problem or information asymmetry. Jensen

(1986) argues that agency conflicts are prevalent in low growth firms because they have substantial

free cash flow and have a tendency to overinvest by accepting marginal investment projects with

negative net present values. This suggests that board structure is an important monitoring device for

low growth firms. In contrast, high growth firms are not likely to suffer from the free cash flow

problem as they are usually short of cash and need to recur to external financing to cover their

financing needs. Instead, as argued by Smith and Watts (1992), agency problems of high growth

firms are associated with greater information asymmetry. However, since these high growth firms

are already subject to monitoring when they raise external finance, the role of the board of directors

as a monitoring device is likely to be minor. Moreover, since managers’ behaviour is difficult to

observe given their discretionary investment opportunities, high growth firms are likely to adopt

alternative governance mechanisms to board monitoring.

I use various variables to measure the composition of the board including the split of the

roles of the chairman and CEO, the number of directors on the board, the proportion of non-

executive directors and the appointment of a non-executive as a chairman. If the board of directors

acts as a monitoring device in reducing the agency conflicts that arise from the free cash flow

problem, then low growth companies should have a more ‘efficient’ and independent board which

comprises a small number of directors, a high proportion of non-executive directors and where the

roles of the chairman and CEO are more likely to be split. In addition, I expect the value of low

growth firms to be negatively related to the number of directors but positively associated with the

proportion of non-executive directors, the split of the roles of CEO and chairman and the

appointment of a non-executive director as a chairman. In contrast, since high growth companies are

less prone to agency conflicts that result from the free cash flow problem, they are expected to have

a small proportion of non-executive directors and a low probability of appointing a non-executive

9

director as a chairman, and the composition of their board should not be an important determinant of

their value. If, on the other hand, agency conflicts are driven by the information asymmetry between

managers and shareholders, then high growth companies are expected to have a more efficient and

independent board and a positive relationship between their value and their board structure is

expected.

III Data and methodology

I first select all companies quoted on the London Stock Exchange with year ends spanning

over the period June 1996 to June 1997. I exclude financial companies, because of the specific

characteristics of their financial ratios. This initial sample includes 1,650 non-financial companies.

The 1996-97 period is likely to be affected by the Cadbury recommendations which had a significant

effect on the board structure (e.g., Dahya et al 2002). In order to account for this effect, I select a

sample of 1,090 firms in the pre-Cadbury period of 1990-91. The data on managerial ownership and

board structure (number of directors, number of non-executive directors, appointment of a non-

executive as a chairman and the split of the roles of chairman and chief executive officer) is

collected from companies’ financial statements and Extel Financial.5 The data is collected by hand

as it is not in available in machine-readable form. I split the companies into high and low growth sub-

samples using earnings-to-price (EP) ratio. Companies with an EP ratio below the median are

included in the high growth group and companies with an EP ratio equal or above the median are in

the low growth group. In order to avoid outliers and discrepancies in the statistics throughout the

text, I exclude firms with no EP data and companies with missing data.6 The final samples include

1,171 listed firms in the post-Cadbury period (1996-97) and 627 firms in the pre-Cadbury period

(1990-91).7

I use four main variables to measure the composition of the board: (i) a dummy variable

equal to one if the company has adopted the Cadbury recommendations, i.e., split the roles of the

chairman and CEO and has three or more non-executive directors on the board (Adopt),8 (ii)

number of directors on the board (#DIR), (iii) the ratio of non-executive directors to the total number

of directors in the board (%NED); and (iv) a dummy variable, (NeChair), equal to one if the

chairman was not previously an executive director, 0 otherwise.9 I also analyse separately the two

components of the Adopt variable, i.e., a dummy variable equal to one if the company has split the

roles of the chairman and CEO (Split) and another dummy variable equal to one if the number of

non-executive directors is three or more (NED3), and include number of executive directors (#ED)

in the analysis.

I use a number of control variables defined in the previous literature to account for any

potential effects of external factors on the analysis. First, I control for managerial ownership by

10

using the proportion of shares held by directors.10 I expect companies with high managerial

ownership to be subject to lower agency conflicts and as such the need for alternative governance

controls such as board monitoring is reduced. I also use the squared value of managerial ownership

to account for the non-linearity relationship between firm value and managerial ownership, as shown

by Morck, Shleifer and Vishny (1988) and McConnell and Servaes (1995). Second, I control for

block-ownership using the proportion of shares held by all block holders, Block , as a proxy for the

incentive of large shareholders to monitor.11 Third, I use book value of leverage, Blev, defined as the

ratio of total debt over the sum of total debt and book value of equity to assess the monitoring role of

debt holders. I simulate the results using market value of leverage defined as total debt over the sum

of total debt and market value of equity. Fourth, I use the log of market value of equity, LN(ME), to

control for size, which, as Smith and Watts (1992) suggest, is positively related to various types of

corporate governance controls such as dividends and managerial compensation. I simulate the results

using log of total assets and log of sales. Finally, I use Tobin’s Q defined as market value of equity

plus total debt over total assets as a proxy for firm value. The results are simulated using industry-

adjusted Q (QADJ) defined as Q less industry median Q, and market-to-sales (M/S) defined as

market value of equity plus total debt over sales.

IV. Empirical results.

A. Characteristics of board structure of high and low growth companies

Table 1 provides a summary statistic of the board structure variables and the control

variables for the pre- and post-Cadbury periods and for high and low growth companies. The last

two columns show the p-value of differences in means and medians between high and low growth

firms. I also compare the means and medians of the pre and post-Cadbury periods and the asterisk

(*) indicates that the differences in means or medians between the two periods are significant at

0.01 level. As the pre- and post-Cadbury period samples are driven by data availability rather than

the matching principles, the differences in firm values, Q, are statistically significant, as shown in the

last two rows of Table 1. Thus, the comparison between the two periods is not likely to provide a

complete picture of the evolution of board structure from the early to the mid-1990s. 12

The first row of Table 1 shows that, for the post-Cadbury period, there are significantly

more high growth companies that adopted the Cadbury recommendations in both pre- and post-

Cadbury periods. On average 63% of high growth companies adopted the recommendations

compared to 58% low growth companies (p = 0.06). The next two rows indicate that the difference

in the adoption rate between the high growth and low growth companies is mainly driven by the

higher proportion of companies that split the roles of the chairman and CEO (89% compared to 84%,

11

p = 0.01) rather than the proportion of companies that had at least three non-executive directors on

the board (68% compared to 65%, p = 0.37). In the pre-Cadbury period, there are also a number of

companies that adopted the recommendations and there are more high growth companies that did so

(32% for high compared to 22% for low growth firms, p = 0.00). However, unlike the post-Cadbury

period, high growth companies are more likely to have three or more non-executive directors (46%

compared to 33%, p = 0.00) rather than split the roles of chairman and CEO (62% compared to

67%, p = 0.23). As expected, compared to the pre-Cadbury period, the proportion of companies that

adopted the recommendations has roughly doubled in the post-Cadbury period for both low and high

growth firms.

The next four rows indicate that, in the post-Cadbury period, companies had a larger number

of directors and a higher number of non-executive and executive directors on the board. However,

the proportion of non-executive directors has remained relatively constant over the two sub-periods.

The comparison between high and low growth firms indicates, however, that high growth companies

have substantially higher proportion of non-executive directors in both sub-periods. In particular, in

the post-Cadbury period, while both high and low growth firms have the same total number of

directors of 7, high growth companies have substantially lower number of executive directors but a

higher number of non-executives.13 As a result, high growth firms have an average of 46.4% non-

executive directors (NED %) compared to 43.7% for low growth firms. The differences in means

and medians in the number of executive directors and the proportion of non-executive directors are

statistically significant, suggesting that, on average, low growth firms have more executives on the

board than high growth firms. Finally, row 8 indicates that, in both sub-periods, high growth firms are

not more likely to appoint a non-executive as a chairman as the differences in means and medians

between high and low growth firms are not statistically significant, and the proportion of companies

that appoint a non-executive as a chairman (NeChair) has more than doubled in the post-Cadbury

period for both high and low growth firms.

The rest of the results in Table 1 show that in both periods, high growth companies have the

same managerial ownership and block ownership as low growth companies. The average managerial

ownership of about 14% and the median of about 6% in the post-Cadbury period are comparable to

the 13.3% and 11.5% reported by Short and Keasey (1999) for a sample of 225 UK listed

companies in 1988 and in 1992, respectively.14 In both sub-samples, the average block holders’

interest is statistically higher than that of managerial ownership. Finally, in the pre-Cadbury period,

high growth companies are larger and have a higher leverage than low growth companies, and, as

expected, high growth companies have significantly higher Q than low growth firms. Similar results

are obtained when size is measured with total asset and sales and value measured with market-to-

12

sales and industry adjusted Q.

[Insert Table 1 here]

In sum Table 1 indicates that board structure and the adoption of the Cadbury

recommendations depend significantly on the firm’s growth potential. I test further this relationship

by splitting companies in the pre- and post-Cadbury periods into five equal groups according to the

level of growth opportunities. I expect the relationship between board structure and growth to hold

across all growth quintiles. The results, reported in Table 2, indicate that as companies move down

from one growth quintile into another, the probability of adopting the Cadbury recommendations

decreases. In particular, the proportion of companies that adopted the recommendations decreases

from 66% in the highest quintile to 54% for the lowest quintile (p = 0.01). As in Table 1, the adoption

rate is not driven by the recommendation of a minimum of three non-executive directors (NED3),

but by the split of the chairman and CEO. The second row of Table 2 indicates that the proportion of

companies that have at least three non-executives is relatively constant across growth quintiles. In

contrast, the proportion of companies that split the roles of the chairman and CEO decreased

monotonically from 90% for the highest growth quintile to 81% for the lowest quintile (p = 0.00). In

the pre-Cadbury period the proportion of very high growth companies (quintile 1) that adopted the

Cadbury recommendations is 30% compared to 24% for the lowest growth quintile. However, the

difference in means is not statistically significant (p = 0.25). There are more very high growth

companies that have three or more non-executive directors (47% compared to 32%, p = 0.01). The

probability of splitting the roles of the chairman and the CEO is not monotonically distributed across

growth quintiles.

The remaining rows of Table 2 show that there is no statistical difference in the size of the

board (#DIR) across growth quintiles in both sub-periods. However, the split of the total number of

directors into executive and non-executive directors shows significant differences across growth

quintiles, but the distributions are not monotonically distributed. The highest numbers of executive

directors in both the pre- and post-Cadbury periods are in quintiles 2 to 5. Nevertheless, the highest

growth companies have significantly lower number of executives than the lowest quintile in both sub-

periods. The distribution of the number of non-executive directors across growth quintiles is also not

monotonic, but in the pre-Cadbury period the highest growth quintile has a larger number of non-

executives than the lowest quintile. As a result, the proportion of non-executive directors is not

monotonically distributed across growth quintiles but the proportion of non-executive directors of the

highest growth quintile is significantly larger than that of the lowest growth quintile in both sub-

periods. Table 2 also reports that in the post-Cadbury period, the relationship between the growth

quintiles and the probability of appointing a non-executive as a chairman is monotonic. As

13

companies’ growth options decrease from the highest to the lowest growth quintile, the probability of

appointing a non-executive as a chairman decreases from 61% to 53% (p = 0.05). This wasn’t the

case in the pre-Cadbury period. Finally, it is interesting to note that in the post-Cadbury period all the

board composition variables have changed relative to the pre-Cadbury period with the exception of

the proportion of non-executive directors that remained relatively constant. As in Dahya et al (2002),

these results suggest that Cadbury (1992) had a significant impact on the board structure of UK

companies, but the question still remains as to why only about 62% of UK companies complied with

the recommendations and why high growth companies are more likely to comply than low growth

companies.

[Insert Table 2 here]

The statistical difference in the board composition between high and low growth firms

documented above may be due to other factors that might affect the decision to opt for a particular

board. I correct for these potential effects by running a set of logit regressions designed to highlight

the board structure differences between the high growth and low growth companies after accounting

for size, leverage, ownership, value and industry factors. The results are reported in Table 3. The

dependent variable is a dummy variable equal to one if the company is in the high growth sub-

sample, zero otherwise. In the pre-Cadbury period, the board structure of high growth companies is

relatively similar to that of low growth companies, with the exception of the number of non-executive

directors that are higher for high growth companies (coefficient of 0.12, p =0.04). The coefficient of

the Adopt variable of 0.29 is not significant (p = 0.18). However, when this variable is divided into

its two components, the results (not reported) indicate that high growth companies are not more

likely to split the roles of the chairman and CEO (coefficient of Split dummy is -0.19, p = 0.30) but

they are more likely to have three or more non-executive directors (coefficient of NED3 is 0.10, p =

0.04). The results also indicate that high growth firms do not have higher managerial ownership,

block ownership, leverage or market value of equity than low growth firms.

In contrast, in the post-Cadbury period, there are significant differences between high and

low growth companies in all board structure variables, with the exception of number of directors and

the appointment of a non-executive as a chairman. As shown in the univariate analysis above, high

growth firms are more likely to adopt the Cadbury recommendations than low growth firms

(coefficient of Adopt is 0.29, p = 0.05). The coefficient of the Split dummy (not reported) is also

positive and significant (coefficient = 0.42, p = 0.04), suggesting that high growth firms are more

likely to separate the roles of CEO and chairman. However, the coefficient of the dummy variable

for three or more non-executive directors (not reported) is not significant (0.03, p = 0.32). Thus, as

shown above, in the post-Cadbury period, high growth companies are, on average, more likely to

14

adopt the Cadbury recommendations by splitting the roles of CEO and chairman but they are not

more likely to have the minimum number of non-executive directors than low growth firms. The

results also indicate that high growth firms are more likely to have a larger number (#NED) and

proportion (%NED) of non-executive directors and a lower number of executive directors (#ED)

than low growth firms (Equations 8 to 10).

The remaining results in Table 3 indicate that high and low growth firms have relatively the

same managerial and block ownership. However, high growth companies have larger leverage and

Tobin's Q but are smaller than low growth firms as the coefficients of Q and Blev are all positive

and significant while that of ln(ME) is negative and significant. These results can, however, be due

to the relatively high correlation between log of market value of equity and the remaining variables.

To overcome this multicolinearity problem, I use log of total assets and log of sales, which are not

highly correlated with the remaining explanatory variables, as a proxy for size. The results, not

reported, indicate a negative and significant relationship between the growth dummy and size,

suggesting that high growth firms are smaller than low growth firms. I also obtain similar results

when market leverage defined as total debt over total debt plus market value of equity is used

instead of Blev. These results do not provide support for Smith and Watts (1992) and Gaver and

Gaver (1993) who show that high growth firms have low leverage.

[Insert Table 3 here]

B. Board structure and firm value

Table 4 reports the regression results of firm value, Q, on lagged values of board structure,

managerial and block ownership, leverage, size, and industry factors. I use lagged values of the

explanatory variables to account for possible endogeneity problem. I hypothesize that the board

structure at the beginning of year t will have an effect on the agency conflict during year t and this

will be valued accordingly at the beginning of subsequent year t+1. Thus, the correspondent firm

value (Q) in the pre-Cadbury period (1990/91) is measured in 1991/92 and in the post-Cadbury

period (1996/97) Q is measured in 1997/98 period.15 The coefficients of industry dummies are not

reported. The t-statistics are based on standard errors that are heteroskedastic-consistent (White

1982). The results based on one explanatory variable at time and industry dummies to control for

possible multicolinearity problem are qualitatively similar.

Table 4, Panel A, reports the results for the pre-Cadbury period. Equations (1) and (6)

show that firm value is not statistically related to the adoption rate for both high and low growth

firms. I also run but not report the regressions with dummies for the split of the roles of the

chairman and CEO and the three or more non-executive directors, the two constituents of the

15

Adopt variable. The coefficients of the split dummies are not significant for both high (0.09, p =

0.17) and for low (0.024, p = 0.68) growth firms. Similarly the coefficients of the three or more

non-executive directors are not significant for high (-0.134, p = 0.11) and low growth (0.055, p =

0.46) firms. Consistent with Yermack (1996), Equations (2) and (7) show that the number of

directors (#DIR) is negatively related to firm value but it is only significant for high growth

companies (p = 0.04). In contrast, Equations (3) and (8) show that the number of executive

directors (#ED) is negatively related to firm value but it is only significant for low growth firms (p =

0.00). Finally, the proportion of non-executive directors and the appointment of a non-executive

as a chairman are not significantly related to firm value for both high and low growth firms. Overall,

the relationship between firm value and board structure in the pre-Cadbury period is not strong.

Table 4, Panel B, reports the results for the post-Cadbury period. For high growth

companies, the relationship between firm value and the adoption rate is negative and significant

(Equation 1), suggesting that high growth companies that split the roles of the chairman and CEO

and have a minimum of three non-executives on the board generate lower value than companies

that do not adopt the recommendations. I also run, but not report, the regressions using the Split

and a minimum of three non-executives (NED3) dummies. The results indicate that, for high

growth firms, NED3 is negative and significant (-0.18, p = 0.07) while Split is negative but not

significant (-0.21, p = 0.19), suggesting that the negative relationship between firm value and

Adopt for high growth firms is driven by the minimum of three non-executives recommendation.

These results imply that, since high growth companies are likely to make complex and timely

decisions, the appointment of a large number of non-executive directors may only result in high

wages and co-ordination costs and delay in decision making. In contrast, for low growth firms, the

relationship is positive and significant (Equation 6), suggesting that low growth firm benefit from

adopting the Cadbury recommendations. For these firms the coefficient of Split of 0.10 is

significant (p = 0.07) but that of NED3 is not (coefficient of 0.05, p = 0.34), suggesting that the

partition of the offices of the chairman and CEO is likely to decrease the agency costs that might

result from the free cash flow problem.

Equations (2) and (7) show that firm value decreases with the board size (#DIR). These

results are consistent with previous evidence (e.g., Yermack, 1996, Eisenberg et al, 1998) and

suggest that large boards do not create value because their size exacerbates the free riding

16

problem among directors vis-à-vis the monitoring of management. Firm value also decreases with

the number of executive directors on the board (Equations 3 and 8). However, the proportion of

non-executive directors is only positive and significant for the case of low growth firms (Equation

9). For high growth firms the relationship is negative but not significant (Equation 4). Similar results

are obtained when the number of non-executive directors is used. There are two implications from

these results. First, unlike previous studies (e.g., Yermack, 1996), the negative relationship

between firm value and number of directors is likely to apply to only executive directors. Second,

the monitoring role of non-executive directors differs between the two sets of firms. Low growth

firms gain by having a higher proportion of non-executive directors on the board. In contrast, the

value of high growth companies is not affected by the presence of non-executive directors. Thus,

the results of Hermalin and Weisbach (1991) and Bhagat and Black (2000) that there is no

noticeable relation between the proportion of outside directors and firm value appear to apply only

to high growth firms.

The relationship between firm value and the appointment of a non-executive as a chairman

(NeChair) also depends on firm’s growth prospects. Equation (5) indicates that the appointment

of a non-executive director as a chairman does not affect the value of high growth companies. In

contrast, Equation (10) shows that, for low growth firms, the relationship between firm value and

the appointment of a non-executive as a chairman is positive and statistically significant. The results

suggest that low growth firms that appoint a non-executive director as a chairman suffer less from

the agency conflicts that result from the free cash flow problem and, as a result they have a higher

value than other companies.

The comparison of the pre- and post-Cadbury periods (Panel A and Panel B) indicates

that the coefficients of the board structure variables are not only insignificant in the pre-Cadbury

period but they are lower than those of the post-Cadbury period. For example, for high growth

firms, the coefficient of the proportion of non-executive directors (%NED) in the pre-Cadbury

period is 0.0022 compared to 0.33 in the post-Cadbury period. This suggests that, in the post-

Cadbury period, the relationship between firm value and board structure is not only significant but

it is also stronger economically.

Table 4 also reports the results of the relationship between firm value and the control

variables. For high growth companies, the relationship between firm value and managerial

17

ownership is quadratic and significant in both the pre- and post-Cadbury periods. For example,

Equation (1) indicates that, for the pre-Cadbury period, the coefficient of managerial ownership of

0.014 is significant (p = 0.04) and that of its squared value of -0.15E-03 is also significant (p =

0.09). The correspondent coefficients for the post-Cadbury period are 0.02 (p = 0.03) and -

0.24E-03 (p = 0.07). The results suggest that firm value is optimised when managerial ownership

reaches 47% in the pre-Cadbury period and 42% in the post-Cadbury period.16 These finding are

consistent with previous US and UK studies (e.g., McConnell and Servaes, 1990 and Short and

Keasey, 1997). In contrast, for low growth firms the coefficients of managerial ownership variable

and its squared value are not significant at any confidence level in both sub-periods. The results

suggest that high growth firms substitute board structure for managerial ownership as a way of

controlling agency costs that may arise from information asymmetry problem.

In terms of external corporate governance mechanisms, Table 4 shows that, in the pre-

Cadbury period, the relationship between block ownership and firm value is weak for both high and

low growth firms. In contrast, in the post Cadbury period (Panel B), the relationship is negative and

significant for high growth firms but weak for low growth firms. To the extent that these block

holders are pension funds, the largest investor category in the UK, these results are consistent with

Faccio and Lasfer (2000) who show that pension fund investment is negatively related to firm value.

The results also show that leverage is negatively associated with value for high growth firms but it is

not significant for low growth companies in both pre- and post-Cadbury periods. When market

leverage is used (total debt over total debt plus market value of equity), the relationship is negative

and significant for both high and low growth firms. These results are not consistent with McConnell

and Servaes (1995) who find that leverage is positively related to value for low growth firms, but

negatively related to value for high growth companies. The results suggest that the monitoring role of

block holders and bondholders is not effective.

[Insert Table 4 here]

C. Adoption of the Cadbury Recommendations and Firm Value

The results presented above indicate that the internal control mechanisms depend

significantly on firm’s growth opportunities. In particular, despite the fact that high growth firms are

more likely to adopt the Cadbury recommendations, they rely more on managerial ownership as a

way of mitigating the potential agency problems than on their board structure. In contrast, low

growth firms are slow in adopting the Cadbury recommendations but their value is significantly

related to their board structure. In the pre-Cadbury period, there is no relationship between firm

18

value and board structure. These results could be due to differences in sample size or to market

perceptions of the importance of the board structure between the pre- and post-Cadbury periods.

To address this question I select companies that had data in both the pre- and post-Cadbury

period. In addition, I followed these companies from 1990 to 1997 and exclude any company that

changed more than once its growth classification. This condition is necessary to classify companies

into high growth in both the pre- and post-Cadbury period (HG Pre-HG Post), low growth in both

periods (LG Pre-LG Post), low growth in the pre-Cadbury period and high growth in the post-

Cadbury period (LG Pre-HG Post) and, finally, into high growth in the pre-Cadbury and low-growth

in the post-Cadbury period (HG Pre-LG Post). These restrictions resulted in 374 firms.17 Then, as in

Dahya et al (2002), I analyse separately firms that always complied with the Cadbury

recommendations (Always), those that never complied (Never), those that complied in the post-

Cadbury but not in the pre-Cadbury (Post-Adopt) and, finally, those that complied in the pre-

Cadbury but not in the post-Cadbury period (Pre-Adopt). The results are reported in Table 5. The

last two columns provide the p-value of differences in means between high growth (HG Pre-HG

Post) and low growth (LG Pre-LG Post) companies. The Pre-Adopt results are not reported as

there are only 12 firms in the sample that adopted the recommendations in the pre-Cadbury period

but not in the post-Cadbury period.18

For the sample as a whole the first column indicates that 25.4% of companies (95 firms)

always adopted the recommendations, 31.3% (117 firms) never adopted and 40.1% (150 firms)

adopted the recommendations in the post-Cadbury period. The differences in mean rates are all

statistically significant as reported in Panel B, suggesting that, as expected, there are significantly

more companies that adopted the recommendations in the post-Cadbury period (p = 0.00 for Always

vs. Post-Adopt and p = 0.01 for Never vs. Post-Adopt) but that there are more companies that

never adopted the recommendations compared to those that have always adopted the

recommendations (p = 0.07). These results are different from Dahya et al (2002) who show that

only 5% of the 460 companies (22 firms) never complied with the recommendations while 33% (150

firms) always complied.19

Table 5 shows that high growth companies appear to have significantly higher adoption rates

than low growth companies. There are 29.5% of companies in the HG Pre-HG Post sample that

always adopted the recommendations compared to 17.3% in the LG pre-LG Post sample (p = 0.04).

There are relatively similar number of companies that adopted in the post-Cadbury period for both

sub-samples (40.4% compared to 42.9%, p = 0.72). As a result, there are more low growth than

high growth companies (37.5% compared to 26.7%) that never adopted the recommendations (p =

0.09). The results also show that for high growth firms the proportion of companies in Always

19

(29.5%) is similar to that of Never (26.7%). In contrast, for low growth firms, there are significantly

more companies in the Never sample (37.5%) compared to that of Always sample (17.3%) but the

Never and Post-Adopt are similar (p = 0.67). The differences in the adoption rates in the two

remaining sub-samples are not statistically significant, except that there are more LG Pre-HG Post

compared to HG Pre-LG Post companies that adopted in the post-Cadbury period (p =0.09).

To assess the market perception of the adoption of Cadbury recommendations, I first

compute the average cumulative annual abnormal returns for each company in the sample from 1990

to 1996. The abnormal returns are computed as the difference between the actual return on the

share and the percentage return available over the same period from an investment in a diversified

portfolio with the same beta. The data is extracted from the London Business School Risk

Measurement Service. Table 5 shows that, for the sample as a whole, the average annual abnormal

return is -0.99%. Companies that have always adopted the Cadbury code generate 0.06% (median

0.00%) compared to -1.71% (median -0.93%) for those that have never adopted the code.

However, the differences in means reported in Panel B (and median not reported) are not

statistically significant (p = 0.23). HG Pre-HG Post companies that have never adopted the code

generate higher abnormal returns of 2.03% per year compared to 1.49% for companies that have

always adopted the code, but the differences in means (and medians) are not significant (p = 0.88).

It is interesting to notice that companies that adopted in the post-Cadbury period generate negative

abnormal returns of -3.97% per year, significantly lower than those of the Never or Always samples

(p = 0.04). Unfortunately, data on the exact date of the adoption is not available, thus the negative

abnormal returns of the Post-Adopt sample may not be necessarily due to the adoption of the

Cadbury recommendations.

For the LG Pre-LG Post sample the average annual cumulative abnormal returns of

companies that have always adopted the code is 1.13 compared to 0.00% for companies that have

never adopted the code (p = 0.08). The returns of companies that adopted the code in the post-

Cadbury period are also higher than those of the Never sample (0.58% compared to 0.00%) but the

difference in means is not statistically significant (0.74). The remaining two sub-samples also provide

interesting results. In particular, HG Pre-LG Post companies appear to be distressed as they

generate negative abnormal returns of -3.99% per year. This negative performance is much more

pronounced for companies that have never adopted the code (-7.04) rather than for companies that

have always adopted the code (-1.91%), (p = 0.07). These results are not driven by outliers as the

median returns, not reported, also show that firms that have never adopted the code generate -

4.69% compared to -0.09% for those that have always adopted the code. These results appear to

provide support for the findings in the previous section and suggest that there are only low growth

companies that benefit from the adoption of the Cadbury recommendations.

20

[Insert Table 5 here]

As a robustness check, I analysed the cumulative abnormal returns for 1992 and 1997, i.e.,

one year ahead of the original sample periods. I find, but not report, that in 1992, HG Pre-HG Post

companies that always adopted the code generate a median CARS of 0.0% and those that never

adopted had 1.0% (p = 0.41). The CARs of LG Pre-LG Post companies that always adopted the

code are also 0.0%, but the CARs of those that never adopted the Code are -1.91%. However, the

differences between the Always and the Never sub-samples are not significant (p = 0.38). In

contrast, in 1997, the corresponding CARs for HG Pre-HG Post companies are -7.74% and -9.0%

(p = 0.76) while for LG Pre-LG Post companies they are -0.33% and -12.40% (p = 0.08). These

results suggest that, in the post-Cadbury period, the market is valuing more low growth companies

that have adopted the Code.

These results are also simulated using Qt+1 to conform to the results presented in the

previous section. I find, but not report, that, for the sample as a whole, the average Q in 1997 of

companies that have always adopted the recommendations of 1.38 is significantly higher than the

1.06 of those that never adopted (p = 0.07) or than the 1.10 of those that adopted in the post-

Cadbury period (p = 0.00). However, for high growth companies the relationship between firm value

and the adoption rate is relatively homogeneous: The average Q of companies that have always

adopted the recommendations is 1.56 compared to 1.46 for companies that have never adopted (p =

0.75). In contrast, for low growth companies, Q of companies that have always adopted the

recommendations is 1.19 compared to 0.82 for companies that have never adopted (p = 0.07).

Similarly, the average Q of HG Pre-LG Post firms that have always adopted the recommendations

of 1.33 is significantly higher than that of the Never and Post-Adopt samples. This trend is not

observed in 1992. For example, the average Q of LG Pre-LG Post companies that have always

adopted is 1.04, compared to 1.05 for those that have never adopted (p = 0.90). These findings are

also simulated using changes in firm value from 1990 to 1996 as a measure of performance. The

results, not reported, indicate that for the HG Pre-HG Post sample, the average increase in Q for

the Always, Never and Post-Adopt companies are 12.8%, 12.9% and 33.0%, respectively. The

differences in means are not significant. In contrast, for LG Pre-LG Post companies, the average

increase in Q of the Never adopted companies is 0.73% compared to 21.4% for the Always sample

and 24.9% for the Post-Adopt companies. The differences in means between the Always and

Never and Post-Adopt and Never are significant (p = 0.02 and p = 0.03, respectively). In sum,

although there are significantly more high growth companies that adopted the recommendations,

such adoption, on average, does not necessarily lead to value creation, while, for low growth

companies, the adoption of the recommendations is critical in value creation in the post-Cadbury

period.

21

D. Robustness

In this section, I briefly describe the results of some of the robustness checks of the findings.

I check whether the results are sensitive to alternative proxies for firm value and test whether they

are sample period and country dependent.

D.1. Are the results sensitive to the definition of firm value?

To test whether the regression results in Table 3 are shaped by the choice of the proxy

variable for firm value, Q, I re-estimate the results using Q adjusted for industry median, QADJ and

market-to-sales, M/S.20 As in Table 4, five regressions are run for both high growth and low growth

firms. Table 6 reports the coefficients of each board structure variable obtained from each separate

regression. The results of the control variables are not reported as most are qualitatively similar to

those in Table 4.

Panel A reports the results using industry adjusted Q as the dependent variable. The results

mimic those reported in Table 4. In the pre-Cadbury period, there are only the coefficients of board

size (#DIR) and that of the number of executive directors (#ED) that are negative and significant for

high growth and low growth firms, respectively. In the post-Cadbury period, the adoption dummy is

negative and significant for high growth firms but positive and significant for low growth firms,

implying that only low growth firms benefit from the Cadbury Code. Board size (#DIR) and the

number of executive directors (#ED) are all negative and significant for both high and low growth

firms, suggesting that larger boards and, in particular, higher number of executive directors reduce

value. The remaining board structure variables are not significant for high growth companies. In

contrast, for low growth firms, the proportion of non-executive directors, the split dummy and the

appointment of a non-executive director as a chairman are all positive and statistically significant,

suggesting that non-executive directors and the separation of the roles of the chairman and CEO

play a significant role in mitigating the agency conflicts between managers and shareholders. The

results based on market-to-sales (M/S) as a measure of firm value, reported in Panel B, are

qualitatively similar, with two exceptions. First, in the pre-Cadbury period, the number of directors

and executive directors are only negative and significant for high growth firms. Second, in the post-

Cadbury period, none of the board structure variables of the high growth firms is significant, and, for

low growth firms, the coefficient of the split dummy is negative but not significant. This suggests that

the results are not too sensitive to the definition of the dependent variable, and, as above, they

suggest that board structure is more important in reducing the agency conflict of low growth firms

but does not affect the value of high growth firms.

D.2. Is the impact more pronounced in extreme growth groups?

22

I test the hypothesis that if the board structure effect on low growth firms in the post-

Cadbury period is true, then I would expect this effect to be more pronounced in the extreme groups

The results are reported in Table 6, Panel C. The samples include very high growth (Quintile 1) and

very low growth (Quintile 5) firms, as analysed in Table 2 and each quintile includes about 108 firms

in the pre-Cadbury period and 220 firms in the post-Cadbury period. The dependent variable is also

Market-to-Sales for ease of comparison with Panel B, but the results based on Q and industry

adjusted Q are all similar. The results show that in the pre-Cadbury period, none of the board

structure variables is significant at any confidence level. In the post-Cadbury period, the coefficients

of the board structure variables are, as in Panel B, all negative but not significant. The levels of the

coefficients are all much higher than those reported in Panel B. For low growth firms, the

coefficients are all significant and signed as expected with the exception of the split dummy which is

not significant. These results are similar to those reported in Panel B, except that the magnitude of

the coefficients is much higher in Panel C. For example, the coefficient of %NED increases from

0.86 for all low growth firms to 1.68 for the very low growth firms. Similarly, the coefficient of

Adopt increases from 0.18 to 0.30 and that of executive directors from -0.11 to -0.23.

[Insert Table 6 here]

The results are also simulated using changes in firm value Q from 1990-91 to 1991-92 in the

pre-Cadbury period and from 1996-97 to 1997-98 in the post-Cadbury period. The results, not

reported are qualitatively similar. For example, the adoption dummy is not significant in the pre-

Cadbury period and for high growth firms in the post-Cadbury period. However, for low growth

firms in the post-Cadbury period, the coefficient of Adopt is 0.071 (p = 0.04). Similarly, for the very

low growth firms (Quintile 5), the coefficient of the Adopt dummy amounts to 0.15 (p = 0.04) while

for the very high growth firms (Quintile 1) and in the pre-Cadbury period none of the board structure

variables is significant.

D.3. Are the results sample and country-dependent?

The above sample is based on only 1990/91 and 1996/97 annual reports because the time

series detailed data on board characteristics, managerial ownership and block ownership is not

available in machine readable form. However, the results could be sample dependent. They could

also be country-dependent as they are only based on UK firms. To overcome these potential

problems, I first collected data on board structure, managerial ownership, block ownership and the

remaining financial data, as above, from Extel Financial for 744 firms with 1998/99 year-ends. Then

I chose Australia, for which the relevant data is available on Extel and collected by hand data on

board structure, managerial and block ownership for 258 companies.21 Australia is very close

23

institutionally to the UK but does not have a Code similar to Cadbury.22 Thus, it presents a unique

natural experiment on the relationship between board structure and firm value. Then, I replicated the

results in Table 1 to 4 using these new data sets. The results, not reported in full for space reasons,

are summarised below.

Table 7 provides a comparative analysis of board structure in UK and Australia. Since

Australian companies are not subject to the Cadbury recommendations, the Adopt and NED3

variables are not analysed.23 The last columns of Table 7 report the differences in means and

medians between high and low growth firms and the asterisk (*) indicates that the differences in

mean or median between Australian and UK companies are significant at 0.01 level.

Table 7 shows that both UK and Australian high growth companies have significantly higher

number of directors, number of non-executive directors and proportion of non-executive directors

than low growth firms. The results also indicate that, in Australia, high growth companies are more

likely than low growth firms to split the roles of the chairman and CEO (79.1% compared to 65%, p

= 0.00) and to appoint a non-executive director as a chairman (45% compared to 31%, p = 0.02).

For the UK, the average (but not the median) split dummy is higher for high growth firms (p = 0.06)

but there is no statistical difference between high and low growth firms in the appointment of a non-

executive as a chairman (p = 0.85). For the control variables, there is no strong evidence that high

growth companies in both UK and Australia have lower average managerial ownership and

leverage. However, high growth firms are larger than low growth firms in both UK and Australia. In

order to assess whether these differences between high and low growth firms in board structure are

driven by size and other factors, I run logit regressions with dummy equal to 1 if the firm belongs to

high growth group. The results, not tabulated, show that, as in Table 3, UK high growth companies

are more likely to adopt the Cadbury recommendations (coefficient of 0.36, p = 0.05), split the roles

of the chairman and CEO (coefficient = 0.44, p = 0.00), have a higher number of directors

(coefficient 0.10, p = 0.03), higher number of non-executive directors (coefficient = 0.19, p = 0.00).

As in Table 3, the coefficients of managerial ownership and block ownership are not significant

while Q is positive and significant. However, unlike Table 3, leverage (Blev) and size (ln(ME)) are

not significant. The average fraction of correct predictions is 65%. For Australian companies, the

coefficients of number of directors is 0.03 (p = 0.79), number of executive directors -0.11 (p = 0.30),

%NED 0.08 (p = 0.07), split 0.59 (p = 0.09) and non-executive chairman 0.56 (p = 0.08). For the

remaining variables only size and Q are positive and significant. The average proportion of correct

predictions is 76%. Overall, these results provide some further evidence that board structure is a

function of firm’s growth opportunities in both UK and Australia.

Table 7 also reports interesting differences between UK and Australian companies. For the

sample as a whole, the first two columns of Table 7 show that UK companies have larger boards

24

with a higher number of executive and non-executive directors than Australian firms. UK companies

are more likely to split the roles of chairman and CEO (87% in UK compared to 72% in Australia)

and to appoint a non-executive as a chairman (51% in UK compared to 38% in Australia).

However, Australian companies have significantly higher proportion of non-executive directors

(%NED) than their UK counterparts. The median %NED of Australian companies is 66.7%. These

results are consistent with Arthur (2001) who report a median proportion of non-executive directors

of 67%. In contrast, in the UK, the median %NED of 42.9% is significantly lower than in Australia

(p = 0.00). There is also evidence that block ownership is significantly higher in Australia and the

average Q in both countries is the same. The differences in managerial ownership, leverage and size

are not strong.

[Insert Table 7 here]

The results of Table 4 are summarised in Table 8. The results for UK companies reported in

Panel A. mimic those of Table 4, Panel B. In particular, most board structure variables of high

growth firms are negative but not significant. The coefficient of Adopt variable is negative and just

about significant at 0.10 level. In contrast, for low growth companies, the majority of board structure

variables are significant and signed as expected, as in Table 4. Board size and number of executive

directors are negative and significant, suggesting that large board with many executive directors do

not create value. The proportion of non-executive directors, the split dummy and the adoption

dummy are all positive and significant, suggesting that low growth firms benefit by having an

independent board. The appointment of a non-executive as a chairman and the minimum of 3 non-

executive directors dummy are positive but not significant. This suggests that the results reported

above are not sample dependent.

Panel B reports the results based on Australian data. The results show that, for high growth

firms, with the exception of the negative relationship between firm value and the number of

directors, none of the remaining board structure variables is significant. As for the UK, the

coefficients of %NED, Split and NeChair are negative but not significant. In contrast, for low

growth firms, the coefficient of board size is negative and significant, in line with previous US

evidence (e.g., Yermack, 1996), suggesting that larger boards reduce firm value. As in Yermack

(1996), the coefficient of executive directors is negative but not significant. In line with the UK

results, the coefficients of the proportion of non-executive directors and the split dummy are all

positive and significant; suggesting that board structure is an effective monitoring device for low

growth firms. The remaining results of Table 8, Panel B. relate to the control variables based on the

regression of firm value and board size (#DIR). It is also interesting to note that, in line with the

results in Table 4, the relationship between high growth firms’ value and managerial ownership is

curve linear optimised at about 25%. In contrast, for low growth firms, the relationship between firm

25

value and managerial ownership is weak. The results suggest that the findings reported above are

not country-specific and provide further evidence that the monitoring role of the board structure is

not homogeneous across firms of different growth options.

[Insert Table 8 here]

V. Discussion and Conclusions

The purpose of the paper is to analyse the relationship between board structure and firm

value in the UK. I expect firms with high agency costs to adopt the Cadbury recommendations, i.e.,

to split the roles of the chairman and CEO and the board should include three or more non-executive

directors. These firms are also expected to have a large proportion of non-executive directors and to

have a non-executive as a chairman. In addition, I consider the agency costs to result from either the

free cash flow problems or the information asymmetries. Thus, I analyse separately the board

structure of high and low growth firms. If agency coats result from the free cash flow problem then,

low growth firms are expected to have a small and independent board, while if agency costs reflect

the level of information asymmetry, then high growth firms should have an efficient board.

Using data for 627 UK non-financial companies in the pre-Cadbury period (1990/91), I find

no strong relationship between firm value and board structure. In contrast, the analysis of 1171 UK

non-financial companies in the post-Cadbury period (1996/97) shows significant differences in the

board structure and in the relationship between firm value and board structure between high growth

and low growth firms. In particular, high growth firms are more likely to adopt the Cadbury

recommendations (i.e., to split the roles of the chairman and CEO and to have three or more non-

executive directors) and they have a higher proportion of non-executive directors and a lower

number of executive directors than low growth firms. However, despite this high propensity to adopt

the Cadbury recommendations, the relationship between these high growth firms’ value and board

structure is weak or negative. In contrast, for low growth firms, I find a positive relationship between

firm value and the adoption of the Cadbury recommendations, the proportion of non-executive

directors and the appointment of a non-executive director as a chairman. These results hold even

after accounting for size and other monitoring and incentive mechanisms and appear to provide

support for the monitoring role of the board of directors only for the case of low growth firms. These

results are not sample-period dependent and/or country specific as they apply also for a sample of

Australian companies.

The difference between the high and low growth firms is also observed in the relationship

between firm value and managerial ownership. The results show that, for low growth firms, the

relationship between managerial ownership and firm value is weak, implying that low growth

26

companies benefit from board structure as an internal corporate governance mechanism which aims

at reducing the agency costs resulting form the free cash flow problems. In contrast, for high growth

firms, there is a strong non-linear relationship between firm value and managerial ownership,

optimised at about 47% in the pre-Cadbury period and 42% in the post-Cadbury period, suggesting

that high growth firms rely more on managerial ownership to mitigate these problems. These results

are consistent with Hermalin and Weisbach (1991) who show that there is no relationship between

board composition and performance, while there is a strong relationship between managerial

ownership structure and performance. However, these conclusions apply only to high growth firms.

High growth companies do not seem to adopt the Cadbury recommendations to minimise their

agency costs, but rather to reflect the need to provide a good corporate governance image so that

they can raise additional finance in the market to overcome their severe information asymmetries

and capital constraints. The overall results suggest that imposing the same board structure for all

companies independently of their specific characteristics and needs is likely to reduce the value of

firms that may be forced to depart from corporate governance structures which have been

successful.

Although the results highlighted the importance of growth options on the design of internal

corporate governance system, further work is required to fully understand some puzzling issues

reported in this paper. First, given that Cadbury had a dramatic effect on the board of UK

companies it is surprising to see that it is only in the post-Cadbury period that the relationship

between board structure and firm value is apparent. If the board structure reduces agency conflicts,

the relationship between firm value and board variables should have been stronger in the pre-

Cadbury period when companies choose, without any obligation, to have a board that includes a large

proportion of non-executive directors and to split the roles of the chairman and the CEO. One

possible explanation is that in the pre-Cadbury period the board is not likely to be scrutinised by the

market, thus allowing companies to recruit non-independent directors. In contrast, in the post-

Cadbury period, the market is expected to pay more attention to the quality of the non-executives.

The analysis of stock returns of companies with continuous data over the two sample periods

provides support for this argument and suggests that the Cadbury has improved the quality of non-

executive directors.24 However, this argument is not fully supported by the results of Australian

companies which are not subject to Cadbury code. Thus, further research is required to assess the

true level of independence of the non-executive directors in the pre- and post-Cadbury periods

through, for example, the analysis of cross-directorships.

Second, the results point to a space for a further improvement in the structure of boards of

high growth firms as the weak relationship between board structure and value could imply a high

27

demand for more capable managers, instead of a relatively simple restructuring of the board. High

growth firms are likely to have more complex decisions to make than low growth firms. Thus, they

require non-executive of high calibre who are not likely to be available in the market. This suggests

that the non-executive directors appointed on the boards of high growth firms are not capable to fulfil

their role in monitoring management, and thus, do not increase firm’s value. On the other hand, the

weak relationship between the value of high growth firms and board structure could suggest that

non-executive directors are not willing to monitor managers because, unlike their US counterparts,

they are not likely to be sued if things go wrong. In this case, the results provide support for the

recent criticism of the role of non-executive directors in the UK who are considered to be the

‘missing link’ in the chain of good corporate governance. They are called upon to reshape their role

and to meet once a year with a company’s top five or six shareholders without the presence of the

company’s CEO.25 The use of data on, say, the qualification of the non-executive directors will

provide more insights into this issue.

Third, it is also important to understand further whether board structure of high growth firms

reduces the information asymmetry, especially at a time of raising external financing. In this case we

would expect fund raising of high growth firms that adopted the Cadbury recommendations to be

much easier and cheaper than high growth companies that have not adopted the Code. Finally, the

paper did not deal with the endogeneity issue and with the question of whether boards are dominated

by managers, whether the appointment of non-executive directors is determined by the CEO or

whether board structure is a part of a package of responses to mitigate the agency conflicts. The

extent to which these additional issues will strengthen or alter the results of this paper is the subject

of further research.

28

References

Agrawal, A. and C. R. Knoeber. "Firm Performance And Mechanisms To Control Agency

Problems Between Managers And Shareholders." Journal of Financial and Quantitative

Analysis, 31 (1996), 377-397.

Ang, J. S., R. A. Cole and J. W. Lin. “Agency costs and ownership structure.” Journal of

Finance, 55 (2000), 81-106

Arthur, N. “Board composition as the outcome of an internal bargaining process: Empirical

evidence,” Journal of Corporate Finance, 7 (2001), 307-340.

Bhagat, S., and B. S. Black. “The uncertain relationship between board composition and firm

performance.” In: K. Hopt, M. Roe and E. Wymeersch, eds, Corporate Governance: The State

of the Art and Emerging Research. Oxford University Press, Oxford, (1998).

Bhagat, S., and B. S. Black. “Board independence and long-term firm performance.” Working

paper, University of Colorado (2000).

Borokhovich, K. A., R. Parrino and T. Trapani. “Outside directors and CEO selection.” Journal of

Financial and Quantitative Analysis, 31 (1996) 337-355.

Brickley, J. A., J. L. Coles and R. L. Terry. “Outside directors and the adoption of poison pills.”

Journal of Financial Economics, 35 (1994), 371-390.

Brickley, J. A. and C. M. James. “The takeover market, corporate board composition, and

ownership structure: The case of banking.” Journal of Law and Economics, 30 (1987), 161-

180.

Cadbury, A. “Report of the Committee on the Financial Aspects of Corporate Governance.” Gee

Publishing, London (1992)

California Public Employees’ Retirement System. “CalPERS US Corporate governance core

principles and guidelines” (1998)

Cho, M. H. “Ownership structure, investment, and the corporate value: an empirical analysis.”

Journal of Financial Economics, 47 (1998), 103-121.

Cotter, J.F., A.S and M. Zenner. “Do independent directors enhance target shareholder wealth

during tender offers?” Journal of Financial Economics, 43 (1997), 195-218.

Dahya, J., J. J. McConnell, and N. G. Travlos. “The Cadbury Committee, Corporate Performance

and Top Management Turnover.” Journal of Finance, 57 (2002), 461-484

Denis, D. J., and A. Sarin. “Ownership and Board Structures in Publicly Traded Corporations.”

Journal of Financial Economics, 52 (1999), 187-223

Denis, D. J., and D. K. Denis. “Majority owner-managers and organizational efficiency.” Journal

of Corporate Finance, 1 (1994), 91-118

Denis, D. J., and T. A. Kruse. “Managerial discipline and corporate restructuring following

performance decline.” Working Paper, Purdue University (1999).

29

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

firms.” Journal of Financial Economics, 48 (1998), 35-54.

Faccio, M. and M.A. Lasfer. “Do occupational pension funds monitor companies in which they hold

large stakes.” Journal of Corporate Finance 6, (2000), 71-110.

Fama, E. F. “Agency problems and the theory of the firm.” Journal of Political Economy 88

(1980), 288-307.

Fama, E. F. and M. C. Jensen. “Separation of ownership and control.” Journal of Law and

Economics, 26 (1983), 301-326.

Franks, J. R., C. Mayer and L. Renneboog. “Who disciplines bad managers?” Working Paper,

London Business School (1998).

Gaver, J.J. and K.M. Gaver. “Additional evidence on the association between the investment

opportunity set and corporate financing, dividend, and compensation policies.” Journal of

Accounting and Economics, 16 (1993), 125-160.

Hermalin, B. E. and M. S. Weisbach. “The Determinants of Board Composition.” Rand Journal of

Economics, 19 (1988), 589-606.

Hermalin, B. E. and M. S. Weisbach. “The effects of board composition and direct incentives on

firm performance.” Financial Management, 20 (1991), 101-112.

Hermalin, B. E. and M. S. Weisbach. “Board of directors as an endogenously determined institution:

A survey of the economic literature.” Federal Reserve Bank of New York Economic Policy

review (Forthcoming, 2002).

Holderness, C. G., R.S. Kroszner and D. P. Sheehan. “Were the good old days that good? Changes

in managerial stock ownership since the depression.” Journal of Finance, 54 (1999), 435-470.

Jensen, M. C. and J. B. Warner. “The distribution of power among corporate managers,

shareholders, and directors.” Journal of Financial Economics, 20 (1988), 3-24.

Jensen, M. C. “Agency costs and free cash flows, corporate finance and takeovers.” American

Economic Review, 76 (1986), 323-329.

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

The Journal of Finance, 48 (1993), 831-880.

John, K. and L. W. Senbet. “Corporate governance and board effectiveness.” Journal of Banking

and Finance, 22 (1998), 371-403.

Lin, L. “The effectiveness of outside directors as a corporate governance mechanism: Theory and

evidence.” Northwestern University Law Review, 90 (1996), 898-976.

McConnell, J. J. and H. Servaes. “Additional evidence on equity ownership and corporate value.”

Journal of Financial Economics, 27 (1990), 595-612.

McConnell, J. J. and H. Servaes. “Equity ownership and the two faces of debt.” Journal of

Financial Economics, 39 (1995), 131-157.

30

Mikkelson, W. H. and M. M. Partch. “The decline of takeovers and disciplinary managerial

turnover.” Journal of Financial Economics, 44 (1997), 205-228.

Monks, R.A. and N. Minow. “Corporate Governance.” Blackwell, Cambridge, MA. (1995).

Morck, R., A. Shleifer and R. W. Vishny. “Management ownership and market valuation. An

empirical analysis.” Journal of Financial Economics, 20 (1988), 293-315.

Palia, Darius. "The Endogeneity Of Managerial Compensation In Firm Valution: A Solution," Review

of Financial Studies, 14 (2001), 735-764.

Rosenstein, S. and J. C. Wyatt. “Outside directors, board effectiveness and shareholder wealth.”

Journal of Financial Economics, 26 (1990), 175-191.

Rosenstein, S. and J. C. Wyatt. “Inside directors, board effectiveness and shareholder wealth.”

Journal of Financial Economics, 44 (1997), 229-250.

Shleifer, A. and R. W. Vishny. “A survey of corporate governance.” The Journal of Finance, 52

(1997), 737-784.

Short, H. and K. Keasey. “Managerial ownership and the performance of firms: evidence from the

UK. Journal of Corporate Finance, 5 (1999), 79-101.

Smith, C.W. and R. L. Watts. “The investment opportunity set and corporate financing, dividend, and

compensation policies.” Journal of Financial Economics, 32 (1992), 263-292.

Stapledon, G. P., 1996. Institutional Shareholders and Corporate Governance. Oxford:Clarendon Press.

Vafeas, N. “Board meeting frequency and firm performance” Journal of Financial Economics, 53

(1999), 113-142.

Weisbach, M. S. “Outside directors and CEO turnover.” Journal of Financial Economics, 20

(1988), 431-460.

White, H. “Maximum likelihood estimation of misspecified models” Econometrica, 50 (1982), 1-25

Yermack, D. “Higher market valuation of companies with a small board of directors.” Journal of

Financial Economics, 40 (1996), 185-211.

31

Table 1: Summary statistics.The sample includes 627 UK non-financial companies in the pre-Cadbury period (Pre) and 1171 in the post-Cadbury period (Post). High (Low) Growth companies are companies with Earning Price ratio below (above)the industry median. Adopt is a dummy variable equal to 1 if the number of non-executive directors is higherthan 3 (NED3) and the roles of the chairman and CEO are separated (Split); #DIR is number of directors in theboard; #ED (#NED) is the number of (non-) executive directors in the board; %NED is the proportion of non-executive directors in the board; NeChair is a dummy equal to 1 if a non-executive director is a chairman. MGTis the proportion of equity held by managers; Block is the proportion of shares held by large shareholdersother than directors; Blev is the ratio of total debt over total debt plus shareholders’ funds; ME is marketvalue of equity at year end; Q is the ratio of the sum of market value of equity and book value of long andshort-term debt over total assets. The p-values of differences in means and medians (using Mann Whitneytest) between high and low growth companies are in last two columns.

High Growth Companies Low Growth Companies p of differences in

Variables Period Mean Median Min Max Mean Median Min Max Mean Median

Adopt PrePost

0.32*

0.630.00*

1.000.000.00

1.001.00

0.22*

0.580.00*

1.000.000.00

1.001.00

0.000.06

0.000.06

NED3 PrePost

0.46*

0.680.00*

1.000.000.00

1.001.00

0.33*

0.650.00*

1.000.000.00

1.001.00

0.000.37

0.000.46

Split PrePost

0.62*

0.891.00*

1.000.000.00

1.001.00

0.67*

0.841.00*

1.000.000.00

1.001.00

0.230.01

0.320.01

#DIR PrePost

5.50*

7.155.00*

7.001.002.00

17.0020.00

4.86*

7.124.00*

7.001.002.00

15.0015.00

0.000.85

0.000.70

#ED PrePost

2.93*

3.793.00*

4.001.001.00

12.0015.00

2.81*

3.973.00*

4.001.001.00

12.0012.00

0.410.07

0.450.04

#NED PrePost

2.57*

3.352.00*

3.000.000.00

9.0015.00

2.05*

3.152.00*

3.000.000.00

9.009.00

0.000.03

0.000.11

%NED PrePost

45.246.4

50.0050.00

0.000.00

87.786.0

40.3*

43.742.9042.90

0.000.00

88.086.0

0.000.00

0.040.00

NeChair PrePost

0.22*

0.580.00*

1.000.000.00

1.001.00

0.25*

0.540.00*

1.000.000.00

1.001.00

0.470.19

0.590.26

MGT % PrePost

11.2*

14.272.28*

6.250.000.00

82.887.2

11.5*

14.73.65*

6.060.000.00

72.584.3

0.830.68

0.170.45

Block % PrePost

31.2*

34.5228.30*

33.500.000.00

96.788.6

32.1*

36.1730.66*

36.000.000.00

97.298.0

0.590.17

0.310.20

Blev % PrePost

31.7630.06

32.00*

27.780.000.00

96.099.5

27.2727.23

27.5027.75

0.000.00

97.398.4

0.070.02

0.000.14

ME (£m) PrePost

723.4571.0

85.4*

58.51.001.00

27,90033,600

258*

527.539.3950.20

1.271.00

8,80339,600

0.000.74

0.000.38

Q PrePost

1.06*

1.590.88*

1.320.260.23

7.154.98

0.78*

1.120.75*

1.020.150.26

3.264.45

0.000.00

0.000.00

The (*) indicates that the differences in mean or median between the pre- and post-Cadbury periods are

32

significant at 0.01 level.

33

Table 2: Distribution of Mean Board Structure Variables by Growth Quintiles.

The sample includes 627 UK non-financial companies in the pre-Cadbury period 1990/91 (Pre)and 1171 in the post-Cadbury period 1996/97 (Post). Companies are split into growth quintilesaccording to their Earning Price ratio. Adopt is a dummy variable equal to 1 if the number of non-executive directors is higher than 3 and the roles of CEO and chairman are split; #DIR is numberof directors in the board; #ED (#NED) is the number of (non-) executive directors in the board;%NED is the proportion of non-executive directors in the board; Split is dummy equals to 1 if theroles of chairman and CEO are split; NeChair is a dummy equal to 1 if a non-executive directoris a chairman. The last tow columns provide the p-values of differences in means and medians(using Mann Whitney test) between quintile 1 (very high growth) and quintile 5 (very low growth)companies.

Growth Quintiles

High 2 3 4 Low p-differenceHigh-Low

Adopt PrePost

0.30*

0.660.35*

0.610.30*

0.630.20*

0.580.24*

0.540.250.01

NED3 PrePost

0.47*

0.650.51*

0.670.40*

0.730.36*

0.650.32*

0.610.010.33

Split PrePost

0.61*

0.900.65*

0.880.62*

0.870.59*

0.860.71*

0.810.580.00

#DIR PrePost

5.13*

6.725.94*

7.115.40*

7.855.22*

7.094.69*

6.840.130.60

#ED PrePost

2.58*

3.513.16*

3.713.04*

4.393.07*

4.002.68*

3.750.010.08

#NED PrePost

2.56*

3.222.78*

3.402.36*

3.452.15*

3.082.00*

3.080.010.38

%NED PrePost

47.947.0

46.347.5

41.343.5

38.443.2

41.244.2

0.020.05

NeChair PrePost

0.23*

0.610.23*

0.590.22*

0.520.22*

0.540.26*

0.530.580.05

The (*) indicates that the differences in mean or median between the pre- and post-Cadbury periods aresignificant at 0.01 level.

35

Table 3: Logit regressionsThe sample includes and 627 UK non-financial companies in the pre-Cadbury period (1990/91) and 1171 companies in the post-Cadbury period (1996/97). The dependent variable is adummy variable equal to 1 if the company is a high growth firm (Earning Price ratio below the median), zero otherwise. The board structure variables are: Adopt is a dummy variable equalto 1 if the number of non-executive directors is higher than 3 and the roles of CEO and chairman are split; #DIR is number of directors in the board; #ED (#NED) is the number of (non-)executive directors in the board; %NED is the proportion of non-executive directors in the board; NeChair is a dummy equal to 1 if a non-executive director is a chairman. The controlvariables are: MGT, the proportion of equity held by managers; Block, the proportion of shares held by shareholders other than directors; Blev the ratio of total debt over total debt plusshareholders’ funds; ME, market value of equity at year end; Q, the ratio of the sum of market value of equity and book value of long and short-term debt over total assets. All regressionsinclude industry dummies and the t-statistics are based on standard errors computed from analytic second derivative. R2 is the Kullback-Leibler R-squared which measures the goodness offit relative to a model with just a constant term. % Predict is the fraction of correct predictions. The p-values are in parentheses.

Pre-Cadbury Period Post-Cadbury Period

Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Constant

#DIR

#NED

#ED

% NED

Adopt

NeChair

MGT

Block

Blev

Ln(ME)

-1.47(0.19)0.06

(0.16)-

-

-

-

-

0.89E-02(0.16)

0.54E-02(0.32)0.36

(0.34)0.05

(0.47)

-1.32(0.25)

-

0.12(0.04)

-

-

-

-

0.01(0.12)0.005(0.36)0.31

(0.40)0.04

(0.54)

-1.63(0.15)

-

-

-0.01(0.78)

-

-

-

0.88E-02(0.17)

0.57E-02(0.29)0.37

(0.31)0.11

(0.12)

-1.72(0.13)

-

-

-

0.61E-02(0.10)

-

-

0.98E-02(0.13)

0.53E-02(0.33)0.33

(0.37)0.09

(0.19)

-1.51(0.18)

-

-

-

-

0.29(0.18)

-

0.92E-02(0.15)

0.50E-02(0.35)0.36

(0.34)0.08

(0.23)

-1.57(0.16)

-

-

-

-

-

-0.11(0.58)

0.82E-02(0.21)

0.57E-02(0.29)0.38

(0.31)0.10

(0.13)

-20.4(0.99)

0.49E-02(0.90)

-

-

-

-

-

-0.62E-02(0.15)

-0.52E-02(0.16)1.48

(0.00)-0.11(0.05)

-19.8(0.99)

-

0.12(0.03)

-

-

-

-

-0.59E-02(0.17)

-0.61E-02(0.11)1.43

(0.00)-0.17(0.00)

-20.8(0.99)

-

-

-0.09(0.06)

-

-

-

-0.54E-02(0.21)

-0.57E-02(0.13)1.46

(0.00)-0.07(0.14)

-20.8(0.99)

-

-

-

1.34(0.00)

-

-

-0.47E-02(0.27)

-0.66E-02(0.08)1.41

(0.00)-0.12(0.01)

-20.2(0.99)

-

-

-

-

0.29(0.05)

-

-0.54E-02(0.21)

-0.59E-02(0.11)1.47

(0.00)-0.14(0.00)

-20.6(0.99)

-

-

-

-

-

0.13(0.34)

-0.54E-02(0.21)

-0.55E-02(0.14)1.47

(0.00)-0.11(0.02)

36

Q

R2

% Predict

0.89(0.00)0.10266.2

0.88(0.00)0.10565.9

0.86(0.00)0.10765.2

0.86(0.00)0.10366.4

0.85(0.00)0.10266.0

0.86(0.00)0.10465.1

0.98(0.00)0.13464.7

1.01(0.00)0.13765.1

0.97(0.00)0.13665.0

0.99(0.00)0.14065.4

0.99(0.00)0.13765.0

0.97(0.00)0.13564.7

37

Table 4: Impact of board structure on firm valueThe Table presents the results of regression of Q, against lagged dependent variables. The sample includes 627 UK non-financial firms in the pre-Cadbury and 1171 inpost-Cadbury period. High (Low) Growth Companies are companies with Earning Price ratio below (above) the industry median. #DIR is number of directors; #ED isthe number of executive directors, %NED is the proportion of non-executive directors, Split is a dummy variable equal to one if company splits the roles of CEO andchairman, zero otherwise; NeChair is a dummy variable equal to one if companies appointed a non-executive director as a chairman. MGT the proportion of equity heldby managers; Block, the proportion of shares above 3% held by largest shareholders other than directors, Blev the ratio of total debt over total debt plus shareholders’funds and ME, market value of equity at year end. All regressions include industry dummies and the t-statistics are based on standard errors that are heteroskedastic-consistent (White 1982). The p-values are in parentheses.

High Growth Companies Low Growth Companies

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Panel A. Pre-Cadbury Period

Constant

Adopt

#DIR

#ED

%NED

NeChair

MGT

MGT2

Block

Blev

Ln(ME)

-0.75(0.08)

0.19E-02(0.98)

-

-

-

-

0.014(0.04)

-0.15E-03(0.09)

0.26E-02(0.36)-0.70(0.04)0.14

-0.72(0.08)

-

-0.04(0.04)

-

-

-

0.014(0.03)

-0.17E-03(0.05)

0.27E-02(0.34)-0.69(0.05)0.17

-0.59(0.14)

-

-

-0.03(0.16)

-

-

0.015(0.02)

-0.18E-03(0.04)

0.25E-02(0.36)-0.73(0.04)0.15

-0.75(0.08)

-

-

-

0.70E-03(0.66)

-

0.014(0.03)

-0.16E-03(0.07)

0.25E-02(0.35)-0.71(0.04)0.14

-0.77(0.06)

-

-

-

-

0.10(0.28)0.015(0.03)

-0.17E-03(0.07)

0.26E-02(0.34)-0.69(0.04)0.14

-0.34(0.44)0.14

(0.12)-

-

-

-

0.93E-02(0.16)

-0.11E-03(0.25)

0.15E-02(0.40)-0.03(0.83)0.09

-0.45(0.30)

-

-0.01(0.31)

-

-

-

0.092E-02(0.18)

-0.11E-03(0.27)

0.23E-02(0.18)-0.01(0.92)0.11

-0.37(0.37)

-

-

-0.05(0.00)

-

-

0.01(0.11)

-0.13E-03(0.19)

0.23E-02(0.17)-0.02(0.90)0.11

-0.42(0.31)

-

-

-

0.22E-02(0.11)

-

0.99E-02(0.13)

-0.12E-03(0.24)

0.22E-02(0.20)-0.02(0.88)0.09

-0.50(0.22)

-

-

-

-

0.01(0.12)0.01

(0.12)-0.12E-03

(0.23)0.22E-02

(0.20)-0.02(0.87)0.10

38

2R(0.00)

0.38

(0.00)

0.39

(0.00)

0.38

(0.00)

0.38

(0.00)

0.38

(0.00)

0.14

(0.00)

0.13

(0.00)

0.15

(0.00)

0.14

(0.00)

0.13Table 4 Cont.

High Growth Companies Low Growth Companies

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Panel B. Post-Cadbury Period

Constant

Adopt

#DIR

#ED

%NED

NeChair

MGT

MGT2

Block

Blev

Ln(ME)

2R

-0.40(0.54)-0.20(0.04)

-

-

-

-

0.02(0.03)

-0.24E-3(0.07)

-0.41E-02(0.11)-1.09(0.00)0.13

(0.00)

0.20

-0.86(0.20)

-

-0.07(0.00)

-

-

-

0.02(0.01)

-0.26E-3(0.05)

-0.45E-02(0.08)-1.06(0.00)0.17

(0.00)

0.21

-0.57(0.38)

-

-

-0.08(0.00)

-

-

0.02(0.00)

-0.30E-03(0.03)

-0.54E-02(0.04)-1.09(0.00)0.15

(0.00)

0.21

-0.23(0.73)

-

-

-

-0.05(0.87)

-

0.02(0.02)

-0.27E-03(0.05)

-0.47E-02(0.07)-1.09(0.00)0.11

(0.00)

0.20

-0.35(0.60)

-

-

-

-

0.09(0.23)0.02

(0.01)-0.28E-03

(0.04)-0.49E-02

(0.06)-1.08(0.00)0.12

(0.00)

0.20

-1.13(0.03)0.08

(0.09)-

-

-

-

0.18E-02(0.64)

0.74E-05(0.91)

0.28E-02(0.18)0.02

(0.86)0.11

(0.00)

0.18

-1.43(0.00)

-

-0.03(0.08)

-

-

-

0.19E-02(0.61)

0.36E-05(0.96)

0.31E-02(0.14)0.04

(0.77)0.14

(0.00)

0.18

-1.40(0.00)

-

-

-0.04(0.06)

-

-

0.28E-02(0.45)

0.85E-05(0.89)

0.28E-02(0.18)0.03

(0.85)0.14

(0.00)

0.19

-1.31(001)

-

-

-

0.33(0.05)

-

0.24E-02(0.53)

0.22E-05(0.97)

0.27E-02(0.21)

0.88E-02(0.95)0.11

(0.00)

0.19

-1.30(0.01)

-

-

-

-

0.08(0.08)

0.28E-02(0.46)

0.64E-05(0.92)

0.28E-02(0.18)0.01

(0.92)0.12

(0.00)

0.18

39

40

Table 5 Adoption Rate of Cadbury Recommendations and Firm ValueThe table reports the proportion of companies that adopted the Cadbury recommendations and the average annual cumulative abnormal returns from1990 to 1996. The sample includes 374 companies with data in both the pre- (1990-91) and post-Cadbury (1996-97) periods. The sample firms areclassified first into four categories according to whether they were high growth in both pre- and post-Cadbury periods (HG Pre – HG Post), low growthin both periods (LG Pre – LG Post) or whether they were high growth in the pre- and low growth in the post-Cadbury period (HG Pre – LG Post), orconversely, low growth in the pre- and high growth in the post-Cadbury period (LG Pre – HG Post). Then companies are sorted into four categoriesaccording to whether they were (a) Always in compliance with the Cadbury recommendations, (b) Never in compliance with the recommendations, and(c) they complied only in the Post-Cadbury period. The results of companies that complied only in the Pre-Cadbury period are not reported as there areonly 9 cases. The last two columns report the differences in means between companies that were classified as high growth in both the pre- and post-Cadbury periods (HG Pre-HG Post) and those that were always low growth (LG Pre-LG Post.

All 374companies (1)

HG Pre – HG Post(2)

LG Pre – HG Post(3)

HG Pre – LG Post(4)

LG Pre – LG Post(5)

p-value (2)–(5)

Rate CAR90-97 Rate CAR90-97 Rate CAR90-97 Rate CAR90-97 Rate CAR90-97 Rate CAR90-97

Panel A. Rates and Average Firm Value

Always

Never

Post-Adopt

All (N/Average)

25.4

31.3

40.1

374

0.06

-1.71

-1.21

-0.99

29.5

26.7

42.9

105

1.49

2.03

-3.97

-0.49

24.1

28.9

44.6

82

-0.34

-3.28

1.86

-0.10

31.8

31.7

31.7

83

-1.91

-7.04

-2.67

-3.99

17.3

37.5

40.4

104

1.13

0.00

0.58

0.17

0.04

0.09

0.72

-

0.17

0.01

0.02

0.00

Panel B. p-Value of Differences in Means

Always vs. Never

Always vs. Post-Adopt

Never vs. Post-Adopt

0.07

0.00

0.01

0.23

0.39

0.74

0.65

0.05

0.01

0.88

0.04

0.04

0.48

0.00

0.04

0.30

0.44

0.09

0.98

0.98

0.99

0.07

0.70

0.04

0.00

0.00

0.67

0.08

0.68

0.74Table 6: Robustness CheckThe table shows the regressions coefficients of the board structure variables using Equations (1) to (10) in Table 3 with lagged control variables and industry dummies.The coefficients of the control variables are similar to Table 3, thus not reported. Low (High) Growth Companies are companies with Earning Price ratio above (below)

41

the industry median. #DIR is number of directors; #ED is the number of executive directors, %NED is the proportion of non-executive directors, Split is a dummyvariable equal to one if firm splits the roles of CEO and chairman, zero otherwise; NeChair is a dummy variable equal to one if firms appointed a non-executive director asa chairman. In Panel A. the dependent variable is QADJ, Q ratio (the ratio of the sum of market value of equity and book value of long and short-term debt over totalassets) less industry median. In Panels B the dependent variable is Market-to-Sales, the ratio of the sum of market value of equity and book value of long and short-termdebt over sales. In Panel C the dependent variable is Market –to-Sales but the sample includes very high growth (Quintile 1) and very low growth (Quintile 5) firms. Thet-statistics are based on standard errors that are heteroskedastic-consistent (White 1982).

Pre-Cadbury Period Post-Cadbury Period

High Growth Companies Low Growth Companies High Growth Companies Low Growth Companies

Coefficient p-value 2R Coefficient p-value 2R Coefficient p-value 2R Coefficient p-value 2R

Panel A. Dependent Variable QADJ

Adopt#DIR#ED%NEDSplitNeChair

0.5E-02-0.03-0.03

0.9E-030.100.11

0.960.050.140.580.160.25

0.210.220.220.210.220.22

0.14-0.9E-02

-0.040.2E-02

0.010.11

0.120.510.000.130.840.18

0.010.020.020.010.020.01

-0.20-0.07-0.08-0.05-0.210.09

0.040.000.000.870.190.23

0.100.100.100.090.090.09

0.08-0.02-0.040.300.090.08

0.090.080.060.090.090.08

0.090.090.090.090.080.09

Panel B. Dependent Variable Market-to-sales

Adopt#DIR#ED%NEDSplitNeChair

0.14-0.05-0.06

0.2E-020.160.27

0.540.090.060.510.230.18

0.460.460.460.460.460.46

-0.15-0.04-0.03

0.3E-02-0.110.14

0.200.180.180.160.220.21

0.260.260.260.260.260.26

-0.71-0.14-0.11-2.08-0.33-1.14

0.340.270.640.500.730.11

0.130.130.130.130.130.13

0.18-0.07-0.110.86

-0.010.17

0.050.030.000.010.890.06

0.200.200.210.210.190.20

42

Table 6 Cont.

Panel C. Dependent Variable Market-to-sales Highest Growth Quintile vs. Lowest Growth Quintile

Pre-Cadbury Period Post-Cadbury Period

High Growth Companies Low Growth Companies High Growth Companies Low Growth Companies

Coefficient p-value 2R Coefficient p-value 2R Coefficient p-value 2R Coefficient p-value 2R

Adopt#DIR#ED%NEDSplitNeChair

-0.050.030.07

-0.5E-020.290.24

0.920.790530.610.380.61

0.600.600.600.600.600.60

-0.09-0.03-0.01

-0.1E-02-0.08-0.02

0.490.220.640.400.390.84

0.310.320.310.320.320.31

-2.70-0.310.13

-7.05-1.28-2.59

0.150.500.820.270.560.19

0.110.110.110.110.110.11

0.30-0.12-0.231.680.040.28

0.070.040.000.030.820.06

0.110.110.150.120.090.11

43

Table 7. Comparative analysis of board structures in UK and AustraliaThe sample includes 744 non-financial companies in the UK and 256 Australian companies. High (Low)Growth companies are companies with Earning Price ratio below (above) the industry median. #DIR isnumber of directors in the board; #ED (#NED) is the number of (non-) executive directors in the board;%NED is the proportion of non-executive directors in the board; Split is a dummy variable equal to 1 if theroles of the chairman and CEO are held by two different individuals, NeChair is a dummy equal to 1 if anon-executive director is a chairman. MGT is the proportion of equity held by managers; Block is theproportion of shares held by large shareholders other than directors; Blev is the ratio of total debt overtotal debt plus shareholders’ funds; ME is market value of equity at year end converted for Australiancompanies into Sterling using the exchange rate in 2000; Q is the ratio of the sum of market value of equityand book value of long and short-term debt over total assets. The p-values of differences in means andmedians (using Mann Whitney test) between high and low growth companies are in last two columns.

All Companies High Growth Low Growth p-valueVariables Country

Mean Median Mean Median Mean Median Mean Median

#DIR AustraliaUK

4.36*

7.254.00*

7.004.84*

7.574.0*

7.03.88*

6.934.00*

7.000.000.00

0.000.01

#ED AustraliaUK

1.86*

4.031.00*

4.001.88*

4.111.0*

4.01.83*

3.951.0*

4.00.770.19

0.520.70

#NED AustraliaUK

2.50*

3.222.00*

3.002.95*

3.463.0*

3.02.05*

2.982.0*

3.00.000.00

0.000.00

%NED AustraliaUK

53.0*

44.066.7*

42.957.8*

45.066.7*

45.847.442.7

50.0*

42.80.000.04

0.000.03

Split AustraliaUK

72.0*

87.0100*

10079.1*

90.0100*

10065.0*

85.2100*

1000.010.06

0.050.28

NeChair AustraliaUK

38.0*

51.00.00*

10045.052.0

0.00100

31.0*

51.00.00*

1000.020.85

0.050.87

MGT AustraliaUK

14.013.3

6.4*

3.812.513.2

3.3*

2.315.613.4

7.6*

4.90.170.91

0.090.18

Block AustraliaUK

52.6*

32.051.0*

29.453.3*

30.254.0*

26.951.8*

34.150.0*

31.10.560.02

0.500.00

Blev AustraliaUK

31.928.0

30.0*

14.330.839.0

35.0*

17.233.0*

17.225.0*

12.10.770.19

0.060.02

ME (£m) AustraliaUK

4871,104

26.0*

58.09061,930

65.087.0

67.0*

277.010.0*

43.00.020.00

0.000.00

Q AustraliaUK

1.461.27

0.940.89

1.841.64

1.251.06

1.080.89

0.710.75

0.000.00

0.000.00

The (*) indicates that the differences in mean or median between Australian and UK companies aresignificant at 0.01 level.

44

Table 8. Firm value and board structure in UK and AustraliaThe sample includes 744 UK firms and 258 Australian companies. Low (High) Growth Companies arecompanies with Earning Price ratio above (below) the industry median. #DIR is number of directors; #ED isthe number of executive directors, %NED is the proportion of non-executive directors, Split is a dummyvariable equal to one if firm splits the roles of CEO and chairman, zero otherwise, NeChair is a dummyvariable equal to 1 if the chairman is a non-executive director. The dependent variable is Q, the ratio of thesum of market value of equity and book value of long and short-term debt over total assets. MGT theproportion of equity held by managers; Block, the proportion of shares held by largest shareholders otherthan directors, Blev the ratio of total debt over total debt plus shareholders’ funds and ME, market valueof equity at year end. All regressions include industry dummies and the t-statistics are based on standarderrors that are heteroskedastic-consistent (White 1982). The coefficients of the control variables are onlyreported for the regression with #DIR as an explanatory variable.

High Growth Companies Low Growth Companies

Coefficient p-value 2R Coefficient p-value 2R %

Panel A. UK Companies

#DIR#ED%NEDSplitNeChairAdoptNED3

-0.010.64E-03

-0.15-0.08-0.06-0.34-0.25

0.840.990.810.840.110.100.24

0.240.240.240.240.240.240.24

-0.03-0.050.500.110.040.120.06

0.030.010.000.040.430.040.29

0.250.250.280.240.240.250.24

Panel B. Australian Companies

#DIR#ED%NEDSplitNech

ControlVariables:MGTMGT2

BlockBlevLn(ME)

-0.34-0.04

-0.64E-03-0.08-0.26

0.05-0.94E-03-0.25E-02

-1.350.46

0.010.620.120.740.37

0.000.000.720.140.01

0.220.120.120.120.12

-0.16-0.150.110.18

-0.10

-0.030.22E-03

-0.41E-02-0.470.26

0.040.860.080.050.68

0.140.450.430.000.06

0.260.230.250.250.23

45

Notes: 1 See Shleifer and Vishny (1997) for an extensive survey on corporate governance.2 See Cadbury (1992) for recommendations in the UK and CalPERS (1998) for the US.3 For example, Borokhovich, Parrino and Trapani (1996) and Weisbach (1988) show that

decisions made by outside dominated boards are more favourably received by the equity

markets compared to the same decision made by firms with a lesser proportion of outside

directors.4 See Bhagat and Black (1998), Hermalin and Weisbach (2002), John and Senbet (1998) and

Lin (1996) for surveys of the literature on the monitoring role of the board.5 Extel Financial is a database that reports all the information contained in the financial

statements and stock market data of all UK companies.6 There are a number of proxies for growth options, including Tobin’s Q, market-to-sales and

R&D over total assets. However, Gaver and Gaver (1993) show that investment opportunities

set are significantly correlated. I use Tobin’s Q and market-to-sales to proxy for firm value.7 The sample is comparable to previous similar studies. For example, the results of Ang et al.

(2000) are based on cross-sectional analysis of 1,708 unquoted companies in 1992. Short

and Keasey (1999) analysed 225 UK companies. Morck et al (1988) selected 371 large US

companies from the Fortune 500. Holderness et al (1999) include in their sample 1,464

NYSE companies in 1995 and 651 in 1935.8 The Cadbury (1992) states that companies should appoint independent non-executive

directors with high caliber so that their views will carry weight in board discussions. These

non-executive directors are to be in a majority on the nominating committee which is

responsible for making recommendations for board membership, they should be the sole or

majority members of the remuneration committee which makes recommendations to the board

on the pay of executive directors, and of the audit committee whose function is to advise on

the appointment of auditors, to insure the integrity of the firm’s financial statements and to

discuss with the auditors any problems arising during the course of the audit. Unfortunately,

data on the independence criteria and on the composition of the various committees is not

available. Thus, I assume that companies that adopted the recommendations are those that

46

split the roles of the chairman and CEO and those that have three or more non-executive

directors on the board.9 This variable requires that the roles of chairman and CEO are split and that the chairman is a

non-executive director.10 UK quoted companies are required to disclose in their financial statements the names of all

the board members, and the proportion of shares held directly and indirectly (beneficial and

non-beneficial) by executive and non-executive directors, even if the ownership stake is zero

(Companies Act 1985). The officers who are not members of the board are only subject to

the ordinary disclosure rules of 3% or above. This legal disclosure requirement meant that I

had to define managerial ownership as ownership by members of the board of directors.

Although this definition is consistent with that of Morck et al (1988) and Short and Keasey

(1999), it differs from that of McConnell and Servaes (1990) and Holderness et al (1999) as

I do not include shares owned by corporate officers not members of the board. I tried to split

managerial ownership variable into ownership of executive and non-executive directors. I find

that non-executive directors’ ownership is very small (less than 1%). I assume that the

inclusion of this holding is not going to affect the analysis.11 I define a block holder as a shareholder, other than directors, that individually holds at least

3% of a company's ordinary shares. This level is set by disclosure rules (Company Act 1995,

Sections 198 and 199). The threshold was 5 per cent from 1985 to 1989. The variable Block

represents the sum of all the stakes held by block holders.12 However, the differences in Q over the two periods are likely to be driven by the relatively

low market values in the 1990-91 recession period. For the sample of all UK non-financial

firms the average Q in 1990-91peiod is 0.994 compared to 1.696 for the 1996-97 period (p

= 0.00). Thus, the intrinsic characteristics of the samples, other than market values, are likely

to be similar.13 This is lower than the 12 reported by Yermack (1996) and 13 by Hermalin and Weisbach

(1988) in the US. However, these two studies analysed mainly large firms.14 However, this proportion is different from that reported in US studies. For Fortune 500, the

average holding is between 10.6% and 12.4% (e.g., Jensen and Warner, 1988, Morck, et al,

47

1988, and Cho, 1998). For a sample of US middle-size firms, Denis and Kruse (1999) find

that officers and directors hold on average (median) 20% (11.3%) of shares. For all listed

firms in the US, Holderness et al (1999) find an average managerial holding of 21% in 1995,

but for NYSE firms, the average holding is 12%.15The results using contemporaneous dependent and independent variables may be subject to

endogeneity problem of board composition and firm value. Hermalin and Weisbach (1998)

argue that poor performance leads to increases in board independence. In cross-sectional

analysis this effect is likely to lead to a negative relation between firm performance and the

proportion of non-executive directors. There may also be endogeneity between board

structure and other factors that may serve to mitigate the agency problem such as managerial

ownership, block ownership and debt. In this case, a system of equations such as those of

Agrawal and Knoeber (1996) or the instrumental variables method as those used by Palia

(2001) will overcome the endogeneity problem. However, these methods require specific data

on the members of the board and managers, such as the tenure, age and founder dummy.

Moreover, as argued by Palia (2001), the variables that are selected to be used as instruments

should affect the second equation, in my case the board structure, but not the firm value. These

specific variables, such as tenure and age, are not available in the UK. Thus, I recognise that

the endogeneity issue is not directly accounted for in this paper, but I expect the use of lagged

independent variables and the split of companies in the sample into high and low growth firms

to mitigate this problem.16 For example, the inflexion points for the pre-Cadbury period are found as a solution to the

following equation: 200015.0014.0 MGTMGTQ −= . I differentiate Q with respect to MGT,

MGTMGT

Q00030.0014.0 −=

δδ

I let MGT

Qδ

δ = 0 and I solve for MGT.

17 This sample is bound to be subject to survivorship bias. However, data unavailability and

the low market values in 1990-91 recession period made it difficult to use control samples

based say on size and market-to-book. These 374 firms are significantly larger (mean

£1,020m compared to £549m, p = 0.00) but have lower Q (1.16 compared to 1.30, p =

0.00) than the 1171 sample firms used in the previous section.

48

18 There is one company in the HG Pre-HG Post group, 2 in the LG Pre-HG Post group, 4

in the HG Pre-LG Post group and 5 in the LG Pre-LG Post group.19 It is possible that some of my companies have changed from compliance to non-compliance

over the 1991/95 period which I am unable to capture as the data is not available. However,

this is an unlikely case as it may be costly for companies to do so in a relatively short-time

period. An alternative explanation could be related to the fact that Dahya et al (2002) results

appear to be based on small companies as the average book value of assets is £157m compared

to £1,027m for my sample. However, a simple regression shows that the adoption dummy is

positively related to size as measured by ln(ME) (t = 14.0), suggesting that larger companies are

more likely to adopt the recommendations.

20 Results based on stock returns are not reported here as they require different explanatory

variables such as beta and book-to-market21 The sample size is driven by data availability. The 258 firms in the sample cover 28 industries

and their average market value of A$964m is higher the A$521m for all companies (p =

0.04), implying that the sample represents mainly large Australian firms.22 In Australia there is no equivalent to the Cadbury Code. There are various guidelines which

recommend certain board structures, the most prominent ones are issued by Institutional

Investor and another by a “Working Party” representing the Institute of Directors, accountants

and lawyers. These guidelines do not have the same significance as the UK’s Cadbury code as

companies are not required to disclose whether or not they comply with the guidelines. See

Stapledon (1996) for a comparative analysis of the institutional framework in UK and

Australia.23 In line with the results presented in Table 1, the average proportion of companies that adopted

the Cadbury recommendations in 1998/99 is 63% for high growth firm and 53% for low growth

firms (p = 0.00), and there are 67% high growth companies that have three or more non-

executive directors compared to 53% for low growth firms (p = 0.0). The remaining board

structure variables of UK companies are also similar to those presented in Table 1. The analysis

of growth quintiles is in line with the results reported in Table 2.24 These results are consistent with Dahya et al (2002) who find that, following Cadbury

adoption, the increase in outside board members has resulted in an increase in the sensitivity of

49

management turnover to corporate performance, thus suggesting that the Cadbury

recommendations have improved the quality of the board oversight in the UK.25 See, for example, Financial Times 18 February 2002.