lei - determinants of cg and the link between cg and performance - evidence from the uk using a corporate governance scorecard

2.3 Endogeneity of Corporate Governance Mechanisms in Firm Valuation Bhagat and Black 2002 find evidence that firms suffering from low profitability respond by increasing the independen

The Determinants of Corporate Governance and the Link between Corporate Governance and Performance: Evidence from the U.K Using a Corporate Governance Scorecard

1 Introduction

1.1 Introduction

Corporate governance practices in the U.K have received increasing attention since the 1990s, with influential reports issued by the Cadbury Committee (1992), Greenbury Committee (1995), Hampel Committee (1998), and Turnbull Committee (2003) and Sir Derek Higgs (2003) These reports resulted in various corporate governance codes and recommendations, the most recent being the Combined Code on Corporate Governance, July 2003 (hereafter U.K Code)

In this study, we use a scorecard developed by Standard & Poor’s to assess the corporate governance of U.K listed companies It provides a comprehensive measure of the extent to which a company has adopted international best practices in corporate governance, as disclosed in their corporate governance disclosures

The evidence on whether there is a link between governance structure and performance remains weak We argue that one possible reason could be due to the research methodology Earlier research has examined subsets of governance mechanisms, usually one or two governance variables only As the firms can choose and modify the structure of their governance system to suit their circumstances, we argue that we should examine a number of governance variables and over a longer time period

1.2 Motivation of Study

Some recent studies have used a broader measure of corporate governance through a composite corporate governance rating, including Gompers et al (2003) for the U.S., Klapper and Love (2004) for fourteen emerging markets, Durnev and Kim (2002) for twenty seven countries, Bauer et al (2003) for the EMU and the U.K These studies generally find a positive relationship between governance standards and firm value

Baure et al (2003) and other studies are based on ratings of one or two years only, assuming that governance ratings should remain constant for a number of years However our data shows otherwise — there is a significant upward trend for the corporate governance scores over the time Without time series data, researchers cannot study how firms adjust their governance structure over time, or analyze the causality between

governance and firm performance found in Black et al (2005) A recent study by Leora, Klapper and Love (2004) find that differences in firm-level contracting environment would affect a firm’s choice of governance mechanisms, in line with arguments put forth

in Himmelberg et al (1999) However because their governance data have no time variation, they are not able to control the fixed effects and to test the causality

Our study can make a potential contribution in this area by analyzing a number of corporate governance mechanisms based on time-varying firm-specific data Using the methodology in Agrawal and Knoeber (1996), we examine the four mechanisms used in controlling agency problems — insider shareholdings, blockholdings, institutional shareholdings and leverage status of the firm In addition, we also include a comprehensive measure of governance using a corporate governance scorecard and measuring governance over a longer time period

Our findings reveal an interesting relationship between governance and performance

It is the change of governance that determines performance rather than the governance level We form an investment strategy that buys firms with greatest improvement in governance and sells firms with largest deterioration in governance It yields 70.4 percent excess returns over the sample period Contrary to the findings in Bauer et al (2003), we find that investors will lose money if they buy firms ranking highest and sell firms ranking lowest

1.3 Objective of Study

This is an empirical study on whether better corporate governance leads to higher valuation through lower expected rate of return We investigate the interdependence of various governance practices, the change of governance structure and the impact on the firm value We look into the boxes of the aggregate governance index in order to find

“key factors” associated with firm performance We conduct a series of tests to differentiate the risk and mispricing explanations to excess returns resulting from governance improvement

1.4 Potential Contributions of Study

This study includes a more complete set of governance mechanisms including the

composite governance scorecard as well as ownership and firm leverage We deal the unobserved firm heterogeneity with a fixed effects estimator and firm endogeneity with the simultaneous equation system The margin effect of governance improvement is also

a new finding in research on corporate governance

2 Literature Review

2.1 The Interaction of Different Governance Mechanisms

Corporate governance comprises many dimensions Based on the U.K Code, it can

be divided broadly into the role of directors, directors’ remuneration, the role of shareholders, and accountability and audit

Some of the structures are complements while others are substitutes to certain extent The previous research has found different governance patterns For example, Peasnell et

al (2001) find evidence of a convex association between the proportion of outside board members and the level of insider ownership in the U.K corporate control process Shivdasani and Yermack (1999) observe, using U.S data, that when the CEO serves on the nominating committee or no nominating committee exists, firms usually appoint fewer independent outside directors and more grey outsiders Similarly, Vafeas (1999) discover that the likelihood of engaging a nominating committee is related to board characteristics such as inside ownership, number and quality of outsider directors for U.S firms

Board structure is an important governance mechanism Kenneth et al (1995) note the substitution effects between outside directors, blockholders, and incentives to insiders

using eighty one U.S bank-holding companies in his study Both Dedman and Elisabeth (2002) and Young (2000) investigate the board structure determinants before and after Cadbury Report They either find managerial entrenchment is reduced or non executive directors are increased following the imposition of new standards of “best practice” regarding board structure

2.2 The Relationship between Governance Mechanism and Firm Performance

Our study builds on Himmelberg et al (1999) who use panel data to show that managerial ownership is explained by key variables in the contracting environment A large fraction of the cross-sectional variation in managerial ownership is explained by unobserved firm heterogeneity Moreover, after controlling for both observed firm characteristics and firm fixed effects, changes in managerial ownership do not affect firm performance statistically

Many other researchers have examined the relationship between variety of governance mechanisms and firm performance However, the results are mixed Some examine only the impact of one governance mechanism on performance as Himmelberg

et al did, while others investigate the influence of several mechanisms together on performance None of them covers a complete set of governance mechanisms Below, we will briefly review some of previous studies on the governance-performance relationship

(1) Board Composition

It is suggested that higher proportion of non-executive directors in the board helps to reduce the agency cost Kee et al (2003) and Hutchinson and Gul (2003) support this view by showing that that higher levels of non-executive directors on the board weaken the negative relationship between the firm’s investment opportunities and firm’s performance However, de Jong et al (2002), Coles et al (2001), and Weir et al (2002) dispute it by stating that there is no significant relationship between non-executive directors’ representation and performance In contrast, in the U.K., Weir and Laing (2000) find a negative relationship between non-executive director representation and performance In addition, Yermack (1996) present that small board has a higher market valuation

Stronger support for the positive impact of non-executive directors comes from event

study analysis The studies by Rosenstein and Wyatt (1990 and 1997) and Shivdasani and Yermack (1999) show that the appointment of non-executive directors increases company value

(2) Leadership Structure

Although U.K Code regards separation of the role of CEO and chairman as a sign of good governance, previous empirical analyses do not support it For example, Coles et al (2001), Weir et al (2002), and Weir and Laing (2000) do not find any significant relationship between CEO duality and performance Brickley et al (1997) observe that costs of separation are larger than benefits for most large U.S firms

(3) Board Ownership

The findings of the primarily U.S based literature suggest that management is aligned at low or possibly high levels of ownership but is entrenched at intermediate ownership levels (e.g., Morck et al., 1988; McConnell and Servaes, 1990) U.K evidence confirms that U.K management becomes entrenched at higher levels of ownership than their U.S counterparts (e.g Faccio et al., 1999; Short and Keasey, 1999) Hutchinson and Gul (2003) report that management share ownership and managers’ remuneration weaken the negative relationship between the firm’s investment opportunities and firm’s performance In contrast, Coles et al (2001) do not find any contribution to performance

by managerial ownership

(4) Institutional Holdings

As the U.K Code encourages institutions to take an active role in governance, we may expect a positive relationship between institutional holdings and firm performance Unfortunately, empirical evidence is not supportive of this recommendation Both Faccio and Lasfer (1999, 2000) fail to find such a significant relationship for U.K firms Besides,

de Jong et al (2002) find that major outside and industrial shareholders negatively influence the firm value

(5) Committee Composition

For U.K companies, Conyon (1997) provides a thorough review of the workings of remuneration committees and shows that firms with remuneration committees pay directors less remuneration Conyon & Mallin (1997) observe that U.K firms have been slow in adopting nominating committees, a symptom of failure of the corporate

governance system By contrast, audit committee use in the U.K has been widespread (e.g Conyon, 1994; Collier, 1993) The results in Forker’s (1992) study suggest that the quality of disclosure is only weakly related with audit committees and non-executive directors

(6) Managers’ Remuneration

The empirical work shows that the role of managers’ remuneration in coordinating managers’ and investors’ interests is limited Hutchinson and Gul (2003) find a positive role for managers’ remuneration, while Coles et al (2001) do not

2.3 Endogeneity of Corporate Governance Mechanisms in Firm Valuation

Bhagat and Black (2002) find evidence that firms suffering from low profitability respond by increasing the independence of their board of directors, but no evidence that firms with more independent boards achieve improved profitability Vafeas (1999) observes that the annual number of board meetings increases following share price declines He further finds that operating performance improves following years of abnormal board activity

Some other studies are in the ownership area None of them provides support to the governance-performance relationship Oyvind and Bernt (2001) discover that qualitative conclusions are sensitive to choice of instruments Demsetz and Villalonga (2001) fail to find significant relationship between ownership and performance What is more, Cho (1998) concludes that investment affects corporate value and in turn corporate value affects ownership but not vice versa

Agrawal and Knoeber (1996) examine the use of seven mechanisms to control agency problems between managers and shareholders These mechanisms are: shareholdings of insiders, institutions, and large blockholders; use of outside directors; debt policy; the managerial labor market; and the market for corporate control The findings are consistent with optimal use of each control mechanism except outside directors Closely following their approach, we construct a simultaneous equation system

to investigate the influence of corporate governance scorecard on firm performance Barnhart and Rosenstein (1998) investigate the combined effect of ownership structure and board composition on corporate performance The results indicate that managerial

ownership, board composition, and Tobin’s Q are jointly determined Vafeas and Theodorou (1998) examine a broad group of board structure variables for U.K firms Contrary to expectations, the results reveal an insignificant relationship between board structure (percentage of non-executive directors, leadership structure, board ownership and committee composition) and firm performance

2.4 Corporate Governance Scorecard in Examining Stock return, Firm value and Performance

Other than focusing on one or two separate variables for corporate control, recently there have been an increasing number of studies that employ corporate governance scorecard as a comprehensive measure to examine the agency problem It has the advantage to implicitly incorporate either the substitutive or complementary effect of variety of governance practices into one study The empirical literature on the relationship between firm value and corporate governance scorecard usually analyzes either inter-country difference or inter-firm variation within a country The most prominent example of studies on inter-country difference is LaPorta et al (2002), who investigate differences in governance standards among twenty seven countries Their evidence shows that firms incorporated in countries with better governance standards tend to have higher valuations Examples of studies investigating inter-firm variation within one country are Drobetz et al (2003) for Germany, Gompers et al (2003) and Marry and Stangeland (2003) for the U.S., Klapper and Love (2004) for fourteen emerging markets, Durnev and Kim (2002) for twenty seven countries, Bauer et al (2003) for the EMU and the U.K., Black et al (2005) for Korea, Black (2001) for Russia, and Callahan et al (2003) for Fortune 1000 firms The results appear to confirm a positive relationship between governance standards and firm value More importantly, the relationship seems to be stronger in countries with less developed standards

To the best of our knowledge, only Klapper and Love (2004), Durnev and Kim (2002) and Black et al (2005) investigate the determinants of corporate governance scorecard Overall Klapper and Love (2004) find that firm-level governance is correlated with firm size, sales growth and assets composition Moreover, they report that good governance is positively correlated with market valuation and operating performance Simliarly, Durnev

and Kim (2002) report higher disclosure level and in turn higher market valuation for firms with greater growth opportunities, greater needs for external financing, and more concentrated cash flow rights Black et al (2005) document size and firm risk as important factors affecting firms’ corporate governance practices Our study differs from theirs in the following ways: Firstly, we use time-varying governance scorecard to control unobserved firm heterogeneity with fixed effects Secondly, we broaden governance measurements with governance scorecard as well as shareholding variables Finally, we explicitly put governance mechanisms into a simultaneous equation system to address the endogeneity problem

Among the inter-firm variation studies, Gompers et al (2003), Marry and Stangeland (2003), Klapper and Love (2004), and Bauer et al (2003) examine the impact of the governance standards on firm performance approximated by profitability ratios as well All of them document a positive relationship between governance scorecard and performance except for Bauer et al (2003) who surprisingly detect a significant negative relationship

When set up a zero investment portfolio, investors can earn abnormal returns by buying firms from higher level corporate governance group and short-selling those from lower level corporate governance group (Gompers et al., 2003; Drobetz et al., 2003; Bauer et al., 2003)

Drobetz et al (2003) and Chen et al (2003) investigate the influence of governance scorecard on cost of equity capital for Germany and nine Asia markets respectively Their findings show that good corporate governance helps to reduce such cost

Creamers et al (2003) find that external and internal governance mechanisms are strong complements in association with long term abnormal returns and accounting measures of profitability Besides, Q’s of firms with both high takeover vulnerability and high public pension fund ownership are high, but lower than the Q’s of firms where only one of the two governance mechanisms is high

3 The Data

The starting point of our sample is the set of firms listed in Index Constituent Rankings FTSE 100 and Index Constituent Rankings FTSE 250 from FTSE European

Monthly Review, January 2001 issue It will be denoted FTSE 350 hereafter We exclude financial firms as they have different financial reporting formats and many of the key variables needed in our study are not available in COMPUSTAT database For each of the remaining industrial and commercial firms, with at least three years of annual reports containing the relevant corporate governance information over the time-period 1999 till

2003, we calculate its corporate governance score using the corporate governance disclosure scorecard provided by the S&P In this study, scorecard will be used interchangeably with score, rating, ranking and/or index

S&P’s corporate governance disclosure scorecard is a methodology based on a synthesis of governance codes and guidelines of global best practices, as well as its own experience in reviewing individual companies With the information extracted from company annual report, we answered 119 questions on governance practices for each company each year It enabled us to construct a time-varying corporate governance scorecard for our study

Ninety three of the questions are binary questions and one point is given for each best practice complied with and zero otherwise or if the company did not disclose whether it had or not complied with such best practice The remaining questions were answered with specific integers such as “number of members in the remuneration committee” Since for a few questions, the answer corresponded to more than one score item, there are altogether 136 best practice items for the scorecard Four out of the 136 items can score up to a maximum two points instead of one Therefore, the maximum possible score for one company with the 136 score items is 140 After filling in the answers to these 119 questions, the formula automatically computes the total score for the company of interests Generally such measure assigns an equal weightage to each disclosed item with the exception of the four items which can attain a maximum score of two These four items carry slightly higher weightage

Scores on the 119 questions are grouped into five categories of corporate governance: Board Matters, Nomination Matters, Remuneration Matters, Audit Matters and Communication We compute the composite governance scorecard by summing up the scores for each group

The financial data employed in our analysis are obtained from COMPUTSTAT

Global Industrial/Commercial file from 1999 to 2003 (as for previous growth, profit

measures, we use the corresponding earlier period data) We obtain the following data

from DataStream: total stock return index for individual firm, FTSE all share total return

index, U.K Treasury Bill interest rates and market capitalization

The 2-digit SIC codes are from COMPUSTAT (global) database Average number of

institutional shareholders for 2-digit SIC industry of a firm is collected from Shareworld

The dummy variable indicating if a firm trades American Depositary Receipts (ADRs) on

a major exchange (NYSE, AMEX, or NASDAQ) in the U.S is identified using the JP

Morgan website: http://www.adr.com The ownership data are collected manually from

annual reports

Our final sample includes 206 firms with 3 to 5 years of data The scores cover

between 105 and 195 firms over the time-period 1999 till 2003 When the scores are

supplemented with ownership and accounting data to perform analysis, the firm-year

observations vary in the range of 837 to 962 with respect to different model

specifications

Table 1: Summary statistics of composite governance scorecard

1999 2000 2001 2002 2003 1999-2000 2000-2001 2001-2002 2002-2003 Mean 56.71 58.18 60.70 63.05 69.60 2.76 2.77 2.19 6.61 Median 57.00 58.00 61.00 63.00 69.00 2.00 2.00 1.00 4.50

St Dev 9.31 8.45 7.96 8.08 11.05 5.57 3.89 4.47 7.74 Min 29.00 33.00 37.00 41.00 45.50 -7.00 -10.00 -11.00 -4.00 Max 79.00 82.00 86.00 85.00 105.00 25.00 21.00 25.00 41.55

No of firms 105 157 185 195 194 100 154 184 193

Table 1 provides summary statistics by year for corporate governance scorecard

This table reports the mean, median, standard deviation, minimum, and maximum of

corporate governance scores for the sample year 1999-2003 In addition, we present the

statistics for the changes of scores each year The last row of Table 1 shows that the

number of firms increases from 105 in 1999 to 194 in 2003 Also, the score is increasing

from 1999 to 2003, changing from a mean (median) of 56.71 (57.00) in 1999 to a mean

(median) of 69.60 (69.00) in 2003 Overall, there exists an upward trend for the

governance scorecard, as can be detected from the average score change that is positive

for all the sample years However some firms’ corporate governance disclosure

deteriorates during the sample period For example, the firm with the most score decrease loses 11 points in 2002 From both the range between minimum and maximum scores and the standard deviation of scores, we can see a certain degree of dispersion for our sample across firms and years The sample is not skewed as the mean and median is very close

Our sample spans many sectors of the economy Table 2 lists the distribution of firms according to their division and major groups SIC includes ten divisions in total Our sample represents nine of them The absentee is the division of Agriculture, Forestry, and Fishing

Table 2: Industry composition

Division/Major Group Number

Percent

of Sample

Mean SCORE

Median SCORE

Mean SCORE Change

Median SCORE Change Mining 12 1.4% 62.47 57.00 3.74 1.00

34: Fabricated Metal Products, Except Machinery And Transportation

36: Electronic And Other Electrical Equipment And Components,

38: Measuring, Analyzing, And Controlling Instruments;

Transportation, Communications, Electric, Gas, And Sanitary

Services 159 19.2% 63.41 63.00 4.16 2.75

41: Local And Suburban Transit And Interurban Highway Passenger

42: Motor Freight Transportation And Warehousing 3 0.4% 54.67 59.00 7.50 7.50

Finance, Insurance, And Real Estate 24 2.9% 60.05 58.50 5.67 4.00

62: Security And Commodity Brokers, Dealers, Exchanges, And

clustered at the new or mature industries As to be mentioned in the later part of the thesis,

our focus is on the changes of corporate governance by controlling firm fixed effects,

industry effects is less important for this purpose

Besides aggregated corporate governance index, we construct index for each of the

five categories, namely Board Matters; Nomination Matters, Remuneration Matters,

Audit Matters, and Communication The results reported in Tables 3 are the subindices

computed by simply summing up all the points for each element within the same category

This construction weighs each element equally

As an alternative, we follow Black et al (2005) to standardize the subindices and

then combine subindices into a standardized overall index Such approach weighs each

subindex equally rather than weighs each element equally Specifically, to compute

subindices, we sum a firm’s score on the elements of a subindex, divide by the number of

elements, and multiply this ratio by 20 Thus, each subindex has a value between 0 and

20 We define the overall standardized score as the sum of all the five subindices As the

correlation between SCORE and standardized SCORE is 0.963 and significant at 0.01,

we employ only SCORE for all of our investigations

Table 3: Summary statistics of subindices

1999 2000 2001 2002 2003 1999-2000 2000-2001 2001-2002 2002-2003 Panel A: Board Matters (43 elements)

Mean 18.10 19.06 19.37 19.98 21.57 1.16 0.48 0.53 1.63 Median 18.00 19.00 19.00 20.00 22.00 0.00 0.00 0.00 1.00

St Dev 4.57 4.10 4.13 4.01 4.68 3.09 2.27 2.08 2.87 Min 6.00 7.00 7.00 8.00 9.00 -5.00 -8.00 -8.00 -4.00 Max 32.00 29.00 31.00 30.00 35.00 12.00 12.00 9.00 13.00

No of firms 105 157 185 195 194 100 154 184 193

Panel B: Nomination Matters (25 elements)

Mean 5.57 5.90 6.03 6.31 7.27 0.55 0.18 0.27 0.94 Median 6.00 6.00 6.00 6.00 7.00 0.00 0.00 0.00 0.00

St Dev 2.38 2.08 1.99 2.07 2.86 1.53 1.00 1.26 2.22 Min 0.00 1.00 0.00 0.00 1.00 -2.00 -2.00 -3.00 -5.00 Max 12.00 13.00 13.00 16.00 20.00 6.00 8.00 8.00 12.00

No of firms 105 157 185 195 194 100 154 184 193

Panel C: Remuneration Matters (34 elements)

Mean 20.99 20.82 20.98 21.46 23.74 0.02 0.12 0.50 2.29 Median 21.00 21.00 21.00 21.00 23.00 0.00 0.00 0.00 2.00

St Dev 2.10 2.10 2.01 2.23 3.03 1.52 0.98 1.53 2.47

Min 15.00 15.00 15.00 16.00 17.00 -4.00 -3.00 -4.00 -3.00 Max 26.00 26.00 27.00 28.00 33.60 7.30 4.00 7.00 10.60

No of firms 105 157 185 195 194 100 154 184 193

Panel D: Audit Matters (29 elements)

Mean 9.41 9.72 11.65 12.57 14.26 0.89 1.99 0.83 1.70 Median 9.00 10.00 12.00 13.00 14.00 0.00 2.00 0.00 1.00

St Dev 3.20 3.04 2.95 2.87 3.28 1.85 1.82 1.75 2.81 Min 2.00 2.00 3.00 5.00 5.00 -2.00 -4.00 -4.00 -3.00 Max 18.00 18.00 18.00 20.00 24.00 9.00 8.00 10.00 14.00

No of firms 105 157 185 195 194 100 154 184 193

Panel E: Communication (5 elements)

Mean 2.64 2.68 2.66 2.73 2.77 0.14 0.01 0.06 0.04 Median 3.00 3.00 3.00 3.00 3.00 0.00 0.00 0.00 0.00

St Dev 0.59 0.59 0.56 0.57 0.59 0.42 0.29 0.32 0.34 Min 2.00 2.00 2.00 2.00 2.00 -1.00 -1.00 -1.00 -1.00 Max 4.00 4.00 4.00 4.00 4.00 2.00 1.00 1.00 1.00

No of firms 105 157 185 195 194 100 154 184 193

Table 3 provides summary statistics for each subindex As our construction is for equal weight of elements, categories with more elements are potentially likely to have high score For example subindex for Communication has only five elements and thus obtains the lowest mean score Board Matters group is the largest group in terms of elements inclusion Although it has more variation in corporate governance than other groups, its mean score is not the highest All of the subindices contribute to the improvement of overall index over time As it can be seen that every subindex has an upward trend for corporate governance as the aggregate index does The average firms have positive score changes for each subindex Year 2003 observes the highest score increase for most of the categories

Table 4 shows a correlation matrix for overall SCORE and each subindex for the entire sample period All correlations are positive and significant Except for Communication, all the other subindices have high correlations with overall score, ranging from 0.684 to 0.801 As the low weights assigned for the subindex of Communication, this group has a lower correlation (0.339) with overall score The correlation between many of the subindices is high Compared with others, the Communication subindex is less correlated with other categories The lowest correlation

(0.145) is between Communication and Audit Matters The significant correlation

between scores of different categories indicates that firms that disclose more in one

category tend to disclose more in other categories

Table 4: Correlation matrix for corporate governance index and subindices

SCORE Board

Matters

Nomination Matters

Remuneration Matters

Audit Matters Communication SCORE 1.00

Sample size is 836 ** Correlation is significant at the 0.01 level (two-tailed).

There are totally 136 corporate governance elements divided into five subindices.:

Board Matters (43 elements); Nomination Matters (25 elements); Remuneration Matters

(34 elements); Audit Matters (29 elements); and Communication (5 elements) Table 5

lists each governance element, and percentage of firm meeting the governance criteria for

our 136 corporate governance elements over time

Table 5: Incidence of individual elements of corporate governance

A14 Does the company’s M&A allow for telephonic or videoconference

A15 Is there disclosure of company’s guidelines of matters that require

A16 Do the guidelines disclose the type of material transactions that

A18

Are the details of training provided to directors disclosed? (like

number of directors sent for training, where did they receive the

training etc)

0.0% 0.6% 0.5% 1.5% 2.6%

A20 Does the orientation program cover the company’s business and

A21 Does the company provide ongoing training on new laws, regulations and changing commercial risks? 12.4% 15.3% 18.4% 24.6% 28.4%

A23 If answer to #15 is yes, is detailed information on each director

A29 Has disclosure been made of the factors and criteria in determining

A30

Are the chairman and CEO positions held by

same(S)/related(R)/unrelated(U) persons? - Score 2 if U, 1 if R

If answer to #24 is yes, do they include matters such as scheduling

board meetings, preparation of agenda for board meetings, control

over information flows between management and the board, and

compliance with company’s guidelines on corporate governance?

3.8% 4.5% 2.7% 3.1% 5.7%

A33 Are all directors required to seek nomination and re-election at

A36 Is the board provided with supporting background or explanatory information for matters brought before the board? 69.5% 78.3% 80.0% 81.0% 85.6% A37 Does the board receive explanation of variances between

A38 Does the board have separate and independent access to the

A40

If answer to #32 is yes, does it include responsibility for ensuring

that board procedures are followed and compliance of applicable

rules and regulations?

20.0% 23.6% 23.8% 25.6% 33.5%

A42 Does the company have an agreed procedure for directors to take

Nomination Matters

B5 Is the chairman of the nominating committee independent? 54.3% 54.8% 60.5% 66.2% 66.0%

B8 If answer to #43 is yes, do these describe responsibilities of the

B9 Does the nominating committee review, at least annually, whether or not a director is independent? 2.9% 2.5% 2.2% 2.6% 4.1%

B10

Does the nominating committee review adequacy of time spent by

directors, who have multiple directorships, on affairs of each

company?

0.0% 0.0% 0.0% 0.0% 3.1%

B11 Is disclosure made of directors’ particulars where their names are

B12 Is disclosure made of individual member's attendance at the

B20 Has the nominating committee established criteria for evaluation of performance of the board? 0.0% 0.0% 0.0% 0.0% 4.6%

B21

Is disclosure made of the process of board evaluation? (e.g

conducted by external party, conducted by NC, by shareholders

etc.)

1.9% 1.9% 1.6% 2.1% 7.2%

B22

Is criteria for evaluating board performance disclosed? e.g

Company’s share price performance over past years; Return on

assets; Return on equity; Return on investment; Economic value

added; Profitability on capital employed

0.0% 0.0% 0.0% 0.5% 2.1%

B24 If the answer to #56 is yes, is criteria for individual director

B25

Is disclosure made of the process of director evaluation? (e.g

conducted by external party, conducted by NC, by shareholders

etc.)

2.9% 3.8% 3.8% 3.6% 5.2%

Remuneration Matters

C6 Is disclosure made of individual member's attendance at the

C14 Does the remuneration committee recommend to the board a framework of remuneration for the board and key executives? 65.7% 67.5% 68.1% 70.3% 69.6%

C15 Does the remuneration committee determine specific remuneration

C17

Does the remuneration committee’s review include all aspects of

remuneration (such as salaries, fees, allowances, bonuses and

options)?

100.0% 100.0% 100.0% 100.0% 100.0%

C18

Is disclosure made of the remuneration committee’s processes (e.g

external compensation specialists hired) to ascertain industry

practices and salary levels for pay and employment conditions?

79.0% 78.3% 78.9% 82.6% 88.7%

C19 Is executive director compensation linked to industry, company

C20 Is the Percentage of performance-related elements of executive

C21 Is compensation of non-executive directors linked to their level of

C22 Were industry experts consulted on the remuneration of

C24 Do service contracts for directors contain onerous removal clauses?

(Score 1 of No and 0 if Yes) 99.0% 99.4% 99.5% 99.5% 99.5%

C25 Did the remuneration committee consider the appropriateness of

C27

Does director remuneration include long term incentives? E.g

bonuses payable after 12 months and/or share option with a vesting

period > 12 months

99.0% 98.7% 98.9% 100.0% 99.5%

C28

Is disclosure made to shareholders of remuneration of executive

directors? (E if in exact amount & B if in bands of $250K) - Score

2 if E, 1 if B and 0 if ND)

99.0% 100.0% 100.0% 100.0% 100.0%

C29

Is disclosure made to shareholders of remuneration of

non-executive directors? (E if in exact amount & B if in bands of

$250K) - Score 2 if E, 1 if B and 0 if ND)

99.0% 100.0% 100.0% 100.0% 100.0%

C30

Is disclosure made to shareholders of remuneration of top 5

executives who are not directors? (E if in exact amount & B if in

bands of $250K) - Score 2 if E, 1 if B and 0 if ND)

1.9% 1.3% 0.5% 0.5% 1.0%

C33

Is disclosure made of remuneration to an employee who is an

immediate family member of a director or the CEO, and whose

own remuneration exceeds $150,000? If there are no such an

employees is this disclosed?

27.6% 27.4% 29.7% 30.3% 32.5%

C34

If the company has any shares/options for employees/directors, are

the details of these disclosed (shares issued to employees or options

granted)? If it does not have such schemes is this fact disclosed?

98.1% 100.0% 100.0% 100.0% 99.5%

Audit Matters

D4 Is the chairman of the audit committee independent? 88.6% 93.0% 96.2% 96.4% 97.9% D5 Is disclosure made of the basis of selection of audit committee

D6

Do at least 2 members of the audit committee have accounting

experience or related financial management expertise or

experience? (this could either be an accounting or financial

qualification or previous work experience in financial or

investment positions)

44.8% 52.2% 54.1% 59.0% 62.9%

D17 Does the audit committee have authority to investigate any matter

D18 Does the audit committee have access to and cooperation of

D19 Does the audit committee meet with external auditors in the

D20 Does the audit committee review scope, results and effectiveness of

D25 Does the annual report include a statement by the board on

D26 Does the internal auditor report primarily to the chairman of the

D27 Does the internal auditor meet standards set by recognized

D28 Does the audit committee review adequacy of internal auditor’s

E3 Are external auditors present at annual general meetings to assist

E4 Are separate resolutions proposed at the AGM for each distinct

E5 Does the company have its annual reports on its website? If the company does not have a website please score zero 100.0% 100.0% 100.0% 100.0% 100.0%

Max Possible Score = 140 (As Q30 & Q96-98 can score max 2 each)

All of sample firms meet the seven corporate governance criteria These criteria are specifically (1) complete list of board members is disclosed; (2) company has a remuneration committee; (3) remuneration committee determines specific remuneration packages for executive directors and the CEO; (4) remuneration committee’s review includes all aspects of remuneration; (5) executive director compensation is linked to industry, company and/or individual performance; (6) list of audit committee members is disclosed; and (7) company has its annual reports on its website

Other widely accepted (90% or more) governance elements include: (1) detailed information on each director is disclosed; (2) detailed information of previous employment on each director is disclosed; (3) other directorships of directors are disclosed; (4) directors’ service contracts for periods are not more than 3 years; (5) list of remuneration committee members is disclosed; (6) service contracts for directors do not contain onerous removal clauses; (7) director remuneration includes long term incentives; (8) disclosure is made to shareholders of remuneration of executive directors; (9) disclosure is made to shareholders of remuneration of non-executive directors; (10) disclosure is made of components of remuneration analyzed by salaries, variable bonuses, options and long-term incentives; (11) full disclosure is made of remuneration of each director by name; (12) If the company has any shares/options for employees/directors, details of these are disclosed If it does not have such schemes, this fact is disclosed; and (13) separate resolutions are proposed at the AGM for each distinct issue

In contrast, some of corporate governance criteria are seldom (less than 10%) met by companies The are list as follows: (1) aggregate board attendance is disclosed; (2) directors are attending over 60% (80%; 100%) of the board meetings; (3) attendance of individual directors at board meetings is disclosed; (4) company’s M&A allows for telephonic or videoconference meetings; (5) details of training are provided to directors disclosed; (6) disclosure has been made of the factors and criteria in determining the size

of the board; (7) chairman’s responsibility includes matters such as scheduling board meetings, preparation of agenda for board meetings, control over information flows between management and the board, and compliance with company’s guidelines on corporate governance; (8) company secretary attends all board meetings; (9) nominating

committee reviews, at least annually, whether or not a director is independent; (10) nominating committee reviews adequacy of time spent by directors; (11) disclosure is made of individual member's attendance at the nomination committee meetings; (12) nomination committee met more than two (four) times in the year; (13) attendance at nomination committee meetings was more than 60% (80%; 100%); (14) nominating committee has established criteria for evaluation of performance of the board; (15) disclosure is made of the process of board evaluation; (16) criteria for evaluating board performance is disclosed; (17) individual performance of board members is evaluated; (18) criteria for individual director performance evaluation is disclosed; (19) disclosure is made of the process of director evaluation; (20) disclosure is made of individual member's attendance at the remuneration committee meetings; (21) attendance at the RC meetings more than 60% (80%; 100%); (22) at least one remuneration committee member is knowledgeable about executive compensation; (23) percentage of performance-related elements of executive directors’ remuneration is greater than 50%; (24) disclosure is made to shareholders of remuneration of top 5 executives who are not directors; (25) disclosure is made of the basis of selection of audit committee members; (26) audit committee met more than four (six; eight) times in the year; (27) attendance at audit committee meetings was more than 60% (80%; 100%); (28) internal auditor reports primarily to the chairman of the audit committee; (29) internal auditor meets standards set by recognized professional bodies; and (30) external auditors are present at annual general meetings to assist responses to shareholders

As for each element, the acceptance rate is increasing over time Rather than a single element is accepted by more firms over time, almost all the governance elements are moving upward together

In order to address the question of whether the mean SCORE increasing are due to entry of new firms, we further report some sample statistics in Tables 6 to 8 Table 6 shows the number of observations by fiscal year The new comers are fifty seven for

2000, thirty one for 2001, eleven for 2002 and one for 2003 Except for year 2002, all the mean scores for new firms are less than those for old firms From this information, corporate governance improvement for our sample is not wholly driven by new entries with higher governance standards Rather, firms adopt governance code longer are more

likely to improve their governance practices

Table 6: Number of observations by fiscal year

No of firms in previous year 105 157 185 195

No of firms in both current and previous years 100 154 184 193

Mean score of new firms 55.71 59.52 65.09 64.00

Mean score of firms in both current and previous

As a complement to Table 6, Table 7 lists the firms with the highest corporate governance disclosure by fiscal year and provides some statistics Column 1 is the firm list Column 2 and 3 report the SCORE and score change for the current year respectively Column 4 indicates whether the firm also lay among the top governance firms last year Column 5 shows whether the firm newly enters into our sample in the current year

Table 7: List of top 10 corporate governance firms by fiscal year

Top 10 in Previous Year

New Firm

MARCONI CORP 79.00 n.a n.a n.a

BT GROUP 78.00 n.a n.a n.a

AGA FOODSERVICE GROUP 73.00 n.a n.a n.a

HILTON GROUP 72.00 n.a n.a n.a

BBA GROUP 71.00 n.a n.a n.a

CADBURY SCHWEPPES 70.00 n.a n.a n.a

BOC GROUP 69.00 n.a n.a n.a

BRITISH AIRWAYS 69.00 n.a n.a n.a

DIXONS GROUP 69.00 n.a n.a n.a

DE LA RUE 69.00 n.a n.a n.a

BBA GROUP 73.00 2.00 Yes No

HILTON GROUP 72.00 0.00 Yes No

VIRIDIAN GROUP 72.00 9.00 No No

MARCONI CORP 80.00 -1.00 Yes No

BBA GROUP 77.00 4.00 Yes No

BOOTS GROUP 72.00 1.00 Yes No

SMITHS GROUP 72.00 1.00 Yes No

MARKS & SPENCER GROUP 72.00 6.00 No No

DIXONS GROUP 72.00 3.00 No No

HILTON GROUP 72.00 0.00 Yes No

INVENSYS 72.00 3.00 No No

VIRIDIAN GROUP 72.00 0.00 Yes No

SMG 92.00 31.00 No No

Consistent with Table 6, the new entries are not a big concern for evaluating governance improvement However, there is some evidence of a mild sticky of firms over corporate governance measure About half of the best-governed firms are also lies among the best-governed firms in the previous year Although firms improve or deteriorate their governance practices autonomously, their relative ranking on the governance disclosure does not change quickly This may result from the relatively small magnitude of the changes The effects of small changes on the absolute level value may not be markedly However, this does not mean the change is not important If one focus on the ranking of the change itself, some interesting results may be found

Table 8 list the firms with the highest score change by fiscal year The structures of Columns 1, Column 2, and Column 3 are the same as Columns 1, Column 3, and Column

2 in Table 7 Column 4 presents whether the firm is among the highest governance improvement group in previous year Columns 5 and Column 6 indicate whether the firm has the highest level of governance in current year and previous year respectively

Table 8: List of top 10 score change firms by fiscal year

Top 10 SCORE_C

in Previous Year

Top 10 SCORE

in Current Year

Top 10 SCORE

in Previous Year

ALLIED DOMECQ 25.00 65.00 n.a No No SAGE GROUP 23.00 56.00 n.a No No SMITHS GROUP 19.00 71.00 n.a Yes No WETHERSPOON (JD) 16.00 54.00 n.a No No DIAGEO 15.00 61.00 n.a No No

ATKINS (WS) 13.00 68.00 No No No

YULE CATTO & CO 25.00 81.00 No Yes No CRODA INTERNATIONAL 18.00 77.00 No Yes No CABLE & WIRELESS 16.00 57.00 No No No SAINSBURY (J) 15.00 76.00 No No No GALEN HOLDINGS 14.00 51.00 No No No

ULTRAFRAME 11.00 67.00 No No No DAIRY CREST GROUP 11.00 56.00 No No No SABMILLER 10.00 77.00 No Yes No TAYLOR NELSON SOFRES 10.00 65.00 No No No SSL INTERNATIONAL 10.00 66.00 No No No BIG FOOD GROUP 10.00 56.00 No No No

VODAFONE GROUP 24.00 91.80 No No No SCOTTISH & SOUTHERN

BIG FOOD GROUP 24.00 80.00 Yes No No

Table 8 reveals that firms with the greatest improvement in governance in current year are less likely to be the firms with the greatest improvement in governance last year

In the other words, firms do not tend to improve their governance practice continually Column 6 of Table 8 shows that only one firm with greatest score change ranks in the highest governance level group in previous year It seems that those firms already obtained the highest ranking in corporate governance disclosure usually do not further increase their governance next year

Governance-improving firms may be ranked as the best-governed firms in the same

year Current year’s governance improvement leads to the highest ranking of governance level in current year rather than the other way round

Table 9 shows the Pearson correlations among the variables employed in our study The variable definitions are in Appendix All variables are measured over the sample period from 1999 to 2003 The second and third columns of the table show the correlations between each of these variables with our absolute governance scorecard and change of the scorecard respectively The scorecard is negatively correlated with both Tobin’s Q and return on assets However, a simple correlation between corporate governance and performance may be masking a more complex functional form for this relationship, a possibility that we later examine in multivariate tests The negative relationships between score and shareholding variable suggest a substitution effect among difference governance mechanisms This correlation analysis shows that without controlling for other variables, the larger firms with higher debt ratio, capable outsiders serving the board, and issuing ADR in major U.S stock exchanges disclose more

While the level of governance is significantly correlated with many other variables, it

is not the case for the change of governance variable Only firm size, outsiders’ quality and the availability of internal funds denoted by cash flow return are positively correlated with the change of governance score

Q is positively correlated with ROA, R&D intensity and firm risk Q is also negatively correlated with size and proportion of fixed assets, consistent with growth firms generally being smaller firms and intangibles being evaluated higher The impact of corporate governance variables as measured by score and ownership on Q is mixed Further investigation will be carried out in multiple regressions later

ROA is positively correlated with market power and negatively correlated with debt ratio and R&D intensity The correlation between ROA and market power and the Correlation between ROA and R&D intensity are expected, as the activity of investing and expensing of R&D will reduce current profitability

The multicollinearity is not a serious problem here Most of the correlations are less than 0.5 The only correlation of concern is between institutional shareholdings and blockholdings We may need to cautiously explain it when examining them simultaneously

SCORE SCORE_C Q ROA INSIDE BLOCK INST FSIZE K/S Y/S R&D/K DTV REG ADR OUTDIRE TENURE NOD RISK RDAI NINSTI CORE_

Table 9: Pearson correlations among study variables

4 Determinants of Corporate Governance: Empirical Evidence

Following Himmelberg et al (1999), we explore the determinants of firm-level governance controlling for firm-specific contracting environment Himmelberg et al (1999) argue that contracting environment influences the optimal level of managerial shareholdings This argument can be easily transferred to our corporate governance scorecard as a comprehensive measure of control

Similar to their setup, we make a comparison among the pooled data without fixed effects, panel with industry fixed effects and panel with firm fixed effects empirically Briefly, estimates from pooled data depend only on the variation in average corporate governance scorecard and performance levels across firms and will not utilize the variation over time in these variables for individual firm It is achieved by pooling all data and implicitly forcing equal intercepts In contrast, estimates from panel with firm fixed effects will depend only on the variation over time in corporate governance scorecard and performance for each individual firm, and will not utilize the variation in variables across firms Similarly, panel with industry fixed effects assumes unobserved heterogeneity is constant for the same industry instead of same firm The measure with industry fixed effects is noisier than the measure with firm fixed effects Including models of panel with industry fixed effects is for the purpose of completion, but our main focus is on pooled data and panel with firm fixed effects

Besides for the purpose of controlling for unobserved firm heterogeneity, we use fixed effects to detect the different impact of within- and between- variation on governance choices and firm performance Specifications using pooled data serve as a base line to examine cross-section differences; whereas, we use fixed effects to investigate the changes of governance in specific firms over time It is interesting to discover whether it is the absolute ranking of the corporate governance or the improvement of corporate governance over time that has more influence on performance Our research differs from Himmelberg et al., in that we substitute corporate governance scorecard as indicated by SCORE for managerial ownership as dependent variable We use the same set of x variables chosen by Himmelberg et al to proxy firm attributes except that we exclude the ratio of advertising expenditures to the stock of Property, Plant and Equipment, for COMPUSTAT does not report data of advertising

expenditure for U.K companies We also exclude the managerial risk aversion measure since it is not relevant to corporate governance scorecard here

As previously described, the corporate governance scorecard incorporates many governance devices, including board structures, managerial compensation, nomination matters, audit matters and communication with shareholders Although it is a rather comprehensive measure, it does not comprise corporate ownership as one of important mechanisms To complement this governance scorecard, we add ownership to one set of our models to explicitly capture the interdependence among various governance mechanisms as well as their impact on performance

This study proceeds in two parts The first part deals with determinants of corporate governance In the second part, we explore the determinants of firm value In the first part,

we regress SCORE on a vector of x variables with and without ownership variables In the second part, we test for a correlation between SCORE and performance measures indicated by Tobin’s Q We add the variables found in part one to be associated with higher governance rankings as controls to filter out their effects on firm performance Following what we have done in part one, we run regressions including and excluding ownership variables in part two Each set of specifications comprises all the three setups, i.e., pooled data, panel with industry fixed effects and panel with firm fixed effects, which will be noted as pooled, SIC effects and firm effects for short hereafter

To summarize, in part one, the model is specified as:

)1(3

/()

/

&

(

)/()

/()

/(

11 10

9

8 2 7

6 2 5

4

it it it

it

it it

it it

it i

it

K I K

D R RDUM

S Y S

K S

K FSIZE

FSIZE SCORE

ηβ

ββ

ββ

ββ

βα

++

++

++

+

=

)3()

/(

8 7

&

()

/()

/()

/(11

10 9

2

2 5

4 3

2 1

it

it it

it it

it it

it i

it

K D R RDUM

S Y S

K S

K

FSIZE FSIZE

INST BLOCK

INSIDE SCORE

ββ

ββ

β

ββ

ββ

βα

++

++

+

++

++

+

=

e respectively

it

where i and t represent the firm and tim

under the pooled setup, αi(i = 1…N) =β0 (N is the number of firms);

under the SIC effects setup, (i = 1

parameters to be estimate (N is the number of firms)

y executive directors and non-executive

rcentage of shares held by institutions

ge of shares held by shareholders who have more than 5% of shares of

e available, and zero otherwise R&

ant It implies a substitutive effect betw

ntified in the first column carry over to the estimates in Col

ownership with firm characteristics as explanatory variables The stimated coefficients are in Columns four to six for the pooled, SIC effects and firm

Table 10: Determ governance: OLS regressions of SCORE on

K/S: Property, Plant and Equipment to sales

Y/S: operating income/sales

RDUM: a dummy variable equal to unity if R&D data ar

D/K: R&D expenditure to Property, Plant and Equipment

I/K: capital expenditure/Property, Plant and Equipment

The empirical analysis of determinants of corporate governance is summarized in Table 10 We begin our exploration of corporate governance determinants with ownership variables only The results are presented in the first three columns The first column reports results from a baseline specification using pooled data for all firm years

It shows that all the shareholdings variables are negatively correlated with SCORE although institutional shareholdings are insignific

een insider/block shareholdings and other corporate governance mechanisms as aggregated by the corporate governance scorecard

The results in second and third columns control for SIC effects and firm effects, respectively The patterns we ide

umn two and three Besides, blockholdings seem to have no influence on SCORE when controls for firm effects

Next, we replace

e

effects, respectively

inants of corporate wner ip and/ r firm ch racteris ics varia les

Nonlinearity exists between firm size and corporate governance scorecard The inclusion of fixed effects changes the estimated coefficients significantly in some cases For example, both the estimated coefficient of the firm size and the coefficient of the R&D expense to fixed assets changed signs when controls for firm fixed effects Consistent with the results of Himmelberg et al (1999), the ratios of operating income to sales are positive while not significant, supportive of the argument that the higher a firm’s free cash flow, ceteris paribus, the higher the desired level of governance Our estimates for the K/S are contradictory to the hypothesis that firms with a greater concentration of

“hard” capital in their inputs will generally have a lower optimal level of governance level Once we control for firm fixed effects, the coefficients become insignificant The coefficient of ratio of capital expenditures to the capital stock for firm fixed effects model

is significantly positive as expected, suggesting higher level of governance is required for

with greater opportunities for discretionary projects To fine tune the proxies for the scope for discretionary spending, following Himmelberg et al., we use the ratio of R&D spending to capital as a measure of “soft capital” We use dummy variable (RDUM) to indicate the availability of R&D Generally, firms report R&D expenses have higher corporate governance disclosure level Among our three setups, R&D intensity appears to have a positive effect on corporate governance scorecard for pooled data, but have a neg

network Broadly speaking, previously observed separate corr

difference from their study the inclusion of ownership variables and discovered significant substitutive and

rket valuation is to regress Tobin’s Q on corporate governance scorecard controlling for those factors affecting firm value The following models are for part two analysis

)4(

5 4

ative effect on corporate governance scorecard if controls for firm effects This result

is concurrent with the finding of Himmelberg et al (1999) but is against the hypothesis Our model specifications in the last three columns of Table 10 include ownership in combination with firm characteristics (the combined model) to address the issue in a contracting environment

elations of governance scorecard with ownership and firm characteristics still hold when combined together

Taking together, this research reveals the differences existing in cross-section and within-variation in determining governance scorecard One proposition is that the unobserved firm characteristics are correlated with the observed characteristics, thus bias the estimated coefficients in the cross-sectional or pooled regression A second possibility

is that the factors driving a firm to improve governance over time are not necessarily the ones shaping the cross-sectional governance ranking Due to availability of our longitude data, we are able to control the unobserved firm specific characteristics As Klapper and Love (2004) study the determinants of corporate governance using only cross-sectional regression, they fail to control the firm fixed effects Another

is

complementary effects among different control mechanisms

5 Corporate Governance and Firm Performance

A common approach to testing whether established good governance is reflected in the firm’s performance and ma

ηβ

ββ

β

=

/

&

()

4 3

/

8 2 7

ββ

ββ

β

=

)/

&

(

)/()

/()

/(

12

11 10

2 9

8 2 7

it it

it it

it it

it

K D R

RDUM S

Y S

K S

K FSIZE

ηβ

ββ

ββ

β

++

++

++

+

Q:

are studies for all the equations as did in part one

Table 11 reports the estimated coefficients with Q as dependent variable The first

characteristics variab

[(No of common shares ×Price of shares at calendar year end) + Book value of Preferred Capital + Book value of total liabilities]/ Book value of total assets

Other variables are the same as previously defined The three setups namely pooled, SIC effects and firm effects

three columns report results from Equation (4) Columns four to six are for regressors of irm c racte ics as iven in quati (5) F ally, th e colu ns pres

esults from Eq ation (6

: Determ nants of irm valu OLS ression of Q on CORE wnersh and/or f

Numbers in parentheses are t statistics Definitions of each variable are given in the Appendix Year dummies are included for all regressions, but not reported Fixed effects at the industry or firm level are included where indicated, but not reported * Means significant at 0.10 level (two-tailed);** Means significant at 0.05 level (two-tailed);*** Means significant at 0.01 level (two-tailed)

The prominent finding of performance analysis is that SCORE has the predicted sign and

ce are of more importance than level of governance

as information content difference in cross-section and time-series disclosures

As our findings suggest that firms improving their corporate governance over time perform better, changes in governan

etermining market valuation This conclusion is useful to investors because they may find that information obtained from longitude corporate governance disclosure in annual report matters more than information gained from cross-sectional absolute governance ranking in performance evaluation

The reason that higher absolute governance ranking may not lead to better performance can be explained as follows Some firms especially FTSE100 firms autonomously build their control mechanisms in line with the regulator’s recommendations and are likely to disclose more on their governance status There is not necessarily a strong link between governance level and performance On the other hand,

as we do not explicitly control endogeneity in pooled model, the inverse relation may go from poorer performance to more voluntary disclosure This is in spirit consistent with Skinner’s (1994) findings, in that large negative earnings surprises are more often preempted by voluntary corporate disclosures In contrast, firms improving their governance over tim

arded when they improve their governance Furthermore, not examined here but elsewhere in this study, we find that with the same degree of increase in governance, firms having lower initial governance rankings show larger performance increase than their counterparts

As for control variables, the inclusion of firm characteristics changes the signs of estimated coefficients of insider shareholdings for all the three setups The coefficient of blockholdings changes sign for pooled data, while it is not affected in firm fixed effects

panel The coefficient of the institutional shareholdings is dominantly negative and statistically significantly different from zero, suggesting that instead of increasing monitoring of managers’ behavior institutions seem to exploit minority shareholders The firm characteristics variables are robust to the inclusion of additional ownership controls

or all Q regression, Q is negatively related to firm size and “hard” capital but positively

ultaneously examine the use of seven mechanisms to control agency problems Variables included in their study are: shareholdings of insiders, institutions, and large blockholders, use of outside directors, debt policy, the managerial labor market, and the market for corporate control

Our research setup differs from Agrawal and Knoeber’s (1996) in a number of respects: First, we include the comprehensive governance scorecard measure Second, we control for unobserved heterogeneity with fixed effects in the simultaneous analysis On the one hand, since fixed effects and instrument variables are alternative approaches to address the endogeneity, using them together further mitigates the endogenous problem

On the other hand, as argued previously, we believe that the changes of governance have more impact on performance than the level of governance itself

iders in the board has already been captured in our governance scorecard, we exclude

it from the equations The managerial labor market is also excluded because it is less related with governance ratings Our focus is on the internal control Therefore, the market for corporate control falls outside this study and is excluded

Altogether, there are five equations relating to governance mechanisms in our investigation The governance mechanisms are corporate governance scorecard (SCORE),

shareholdings of insiders (INSIDE), institutions (INST), and blockholders (BLOCK), and

) All of them are treated as endogenously determined In estimating the

DAI_D, RDAI, NINSTI, and CR They

de ADRs on a major US exchange ero otherwise

yncratic stock price risk, calculated as the stand

g daily data for the period

if data required to estimated RDAI is available, and otherwise equal to zero

average RDA for the 2-digit SIC industry of the firm

CR: operating cash flow return on firm value =

ADR: a dummy variable equal to unity if a firm tra

(NYSE, AMEX, or NASDAQ), z

OUTDIRE: average number of directorships held by non-executive directors in unaffiliated firms

TENURE: average years of directors stayed in board

NOD: total number of directors

RISK_D: a dummy variable equal to unity if RISK data are available, and zero otherwise RISK: the standard deviation of idios

error of the residuals from a CAPM model estimated usin

covered by the annual sample

RDAI_D: a dummy variable equal to unity

capital+ market value of equity

a es – cost of goods sold – selling, general and administrative expense

endogenous variable of DTV is as following defined, while

ables are defined previously

CASHS PFD

V = liabilities-total+ preferred capital+ market value of equity

LTD = Book value of long-term debt

STD = Book value of short-term debt

PFD = preferred capital

CASHS = cash and short-term investments

In order to satisfy the order condition to ensure that the equations in the system are

equation includes four endogenous variables as regressors The specification of Equations

isfy this order condition Nevertheless, as far as possible, we rely on theory or prior research to determ

We estimate the following simultaneous equation systems:

ntified, each equation must exclude at least four of the exog

(7) to (11) below is partially driven by the need to sat

ine the exogenous v riables to be included or excluded in each of the equation

FSIZE REG

DTV INST

BLOCK INSIDE

SCORE= β0 +β1 +β2 +β3 +β4 +β5 +β6

+β7ADR+β8OUTDIRE (7)

FSIZE REG

DTV INST

BLOCK SCORE

INSIDE =β0 +β1 +β2 +β3 +β4 +β5 +β6

RISK D

RISK NOD

INST INSIDE

SCORE BLOCK =β0 +β1 +β2 +β3 +β4 +β5 +β FSIZE

+β7RISK_D+β8RISK+β9RDAI_D+β10RDAI (9)

NINSTI REG

INST BLOCK

INSIDE SCORE

INST =β0 +β1 +β2 +β3 +β4 +β5 +β6

β7ADR (10)

FSIZE REG

INST BLOCK

INSIDE SCORE

DTV =β0 +β1 +β2 +β3 +β4 +β5 +β6

β7UNIQ+β8CR (11) Besides interacting with other mechanisms, SCORE is assumed to be positively

will be captured by

correlated with firm size, ADR dummy and outside directors’ quality measured by average number of directorships held by non-executive directors in unaffiliated firms As larger firms and firms issuing ADR may have more channels for disclosure Outside directors may encourage firms to comply with higher standard of governance Regulated firms may have different disclosure requirement Such difference

REG dummy

Equation for insider shareholdings is similar to that in Agrawal and Knoeber (1996)

We expect insider shareholdings to be positively correlated with average years of service

as directors and number of directors but negatively correlated with firm risk, regulation and firm size We exclude FOUNDER dummy, as the majority of firms in our sample are mature large firms and do not have founders holding key positions

activeness to institutions by Agrawal and Knoeber (1996) How

than 3%

to our assumption We suspect

at it is due to the positive relationship between firm size and board size As INSIDE

ts

I SIDE through NOD Less use of debt by lar nexpected If managers

Equation (9) relates to blockholdings We include the similar set of exogenous variables as inside shareholdings, except for replacing TENURE and NOD with RDAI The directions of our predictions are similar to those for INSIDE Additionally, as argued

by Agrawal and Knoeber that as the industry average R&D to asset ratio rises, technology becomes more firm-specific, making outside monitoring less effective, we expect a negative coefficient for RDAI

The choice of explanatory variables for INST closely follows Agrawal and Knoeber (1996) Following the same line of argument for NYSE listing, we expect larger firms and those issuing ADR in the U.S to be more attractive to institutions NINSTI is used as

an additional measure of attr

ever, we do not expect a positive relation between INST and NINSTI, because our institutional shareholdings are calculated by summing up the shareholdings larger

As the number of institutions in the industry increases, the proportional shareholdings of each institution may decrease Some may be less than 3% and not included in our calculation of INST This may spuriously lead to a negative relationship between INST and NINSTI

The last endogenous variable is use of debt As asserted in Agrawal and Knoeber (1996), DTV should depend positively upon firm size, negatively on REG and CR

Results of the 2SLS estimation for the pooled data without fixed effects are presented

in Table 12 The coefficients on the exogenous variables generally have the predicted sign and most of them are significant at normal level However, the coefficient on NOD for Equation (8) is negative and significant, contradictory

th

decreases when FSIZE increases, it decreases when NOD increases if FSIZE affec

of book val rather tha arket valu atios, DTV ay not be a

good measure for firm m

relationship between debt use and firm not

Table 12: Coefficient es 2SLS regress rol ms

’s long-ter capital structure Consequently, the hypothesized

size may hold

timates from ions of cont mechanis (pooled)

Independent Variables SCORE INSIDE BLOCK INST DTV

(3.53 ) (-2.243183) 6.7396 5*** -0.064243 -0.025215 031131**

(8.6032) (-1.427161) (-1.043182) (-2.380461) 1.2191 9*** -0.048511 -0.009844 -.0351 7***

(-1.681112) 0.013547 -0.028586 (.773088) (-.87 38) -0.133046 3.029593*

(-.29 74) (1.72 )

-.1 (-4.

37 RISK