cheung et al - 2007 - do investors really value corporate governance - evidence from the hong kong market [hcgi]

Do Investors Really Value Corporate
Governance? Evidence from the
Hong Kong Market
Yan-Leung Cheung
Department of Economics and Finance, City University of Hong Kong, Tat Chee Avenue,
Kowloon Tong, Kowloon, Hong Kong
e-mail:
J. Thomas Connelly
Faculty of Commerce and Accountancy, Chulalongkorn University, Bangkok 10330, Thailand
e-mail:
Piman Limpaphayom
Sasin Graduate Institute of Business Administration, Chulalongkorn University, Sasa Patasala
Building, Phyathai Road, Pathumwan, Bangkok 10330, Thailand
e-mail:
Lynda Zhou
Department of Economics and Finance, City University of Hong Kong, Kowloon Tong,
Kowloon, Hong Kong
e-mail:
Abstract
To examine the relation between corporate governance and firm value, we develop an
instrument to assess the corporate governance practices of listed companies in Hong
Kong. Based on the Revised OECD Principles of Corporate Governance (OECD) and the
Code of Best Practices (HKEx), we construct a corporate governance index (CGI) for
Hong Kong listed companies. Unlike measures used in other studies, the CGI score reflects
the presence of good corporate governance practices as well as variation in the quality of
corporate governance practices. Empirical evidence shows that a company’s market
valuation is positively related to its overall CGI score, a composite measure of a firm’s
corporate governance practices. We also find that the transparency component of the CGI
score drives the relation with market valuation. In summary, this study provides
supporting evidence for the notion that, in Hong Kong, good corporate governance
practices are consistent with value maximization.

The authors are indebted to Herbert Hui, Peter Wong, and the Hong Kong Institute of Directors
for financial support, and Charnchai Charuvastr of the Thai Institute of Directors Association for
providing technical assistance. This project was substantially supported by a grant from City
University of Hong Kong (Project no. 7001912). We also thank Cindy Chen, Sam Lam, Simon
Lam, Justin Liang, Sarah Wan, and Sam Yang for their research support.
Journal of International Financial Management and Accounting 18:2 2007
r Blackwell Publishing Ltd. 2007, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
1. Introduction
The financial crisis of 1997 that swept through most of East Asia
accentuated the need for corporate governance reform in the region.
As part of the clean-up effort, numerous corporate governance reform
initiatives have been launched, including regional (PECC, 2001) and
international efforts (OECD, 2004). These initiatives are not without
controversy. Many, particularly corporate representatives, feel that the
costs of practicing good corporate governance may outweigh the
benefits. They argue that investors do not care about corporate govern-
ance practices when determining stock prices and that good corporate
governance practices do not necessarily lead to value maximization.
Others argue that good corporate governance is related to better firm
performance and that firms with better corporate governance practices
should perform better than those with relatively worse practices. Numer-
ous studies have examined the relation between corporate governance
practices and firm performance, but mostly in developed markets.
Hong Kong provides a unique setting to study. Although in East Asia,
the Hong Kong stock market has many characteristics and practices
observed in developed economies, and international rating agencies rank
Hong Kong as one of the more advanced markets in the region. This may
be due in part to the legacy of an Anglo-Saxon legal system that
influences corporate governance and regulatory practices. Hong Kong
also has a well-developed financial market infrastructure, known for

close and careful regulatory supervision. In addition, local accounting
standards are being harmonized with international standards, allowing
firms to cross-list in other markets.
On the other hand, the Hong Kong market has many characteristics
more commonly observed in developing economies, especially in Asia.
Unlike the more diffuse ownership structures frequently found in devel-
oped economies, Hong Kong firms normally have a closely intertwined
management and ownership structure, with firms quite often being
family owned and managed. It is common to find that the chairman of
the board is also the chief executive officer in Hong Kong listed
companies.
1
A report of the Corporate Governance Working Group of
the Hong Kong Society of Accountants in December 1995 remarked
‘‘almost 90 per cent of listed companies have a major shareholder who by
himself or with family members owns 25 per cent or more of the share
capital.’’ Ownership structures have changed little following the financial
crisis. A corporate governance report published by Standard & Poor’s in
Do Investors Really Value Corporate Governance? 87
r Blackwell Publishing Ltd. 2007.
2002 echoes the previous statement, noting ‘‘the main weaknesses in the
Hong Kong governance environment are less at the legal and regulatory
levels than at the individual firm level. Standard & Poor’s recognized that
family domination of companies’ ownership structure and the limited
(though growing) shareholder activism culture present particular chal-
lenges.’’
This highly concentrated, family dominated ownership structure
means that the classic agency problem stemming from the separation
of ownership (shareholders) and control (managers) is rarely observed in
Hong Kong. However, this type of ownership structure often results in

an opaque rather than transparent shareholding structure, potentially
leading to a unique agency conflict between owners who are also
managers (insiders) and minority shareholders (outsiders). This also
means that market-disciplining mechanisms such as hostile takeovers
cannot function properly in Hong Kong.
2
In sum, Hong Kong is an international financial center that combines
characteristics of both developed and developing economies: concen-
trated ownership of listed companies, family owned and managed firms,
an Anglo-Saxon legal and corporate governance system, strong regula-
tory oversight, and a weak market for corporate control. Thus, Hong
Kong is a good testing ground to examine whether corporate governance
influences firm valuation in a market where family ownership and
concentrated shareholdings predominate.
The objective of this study is to investigate whether there is a relation
between corporate governance practices and firm valuation in Hong
Kong. Are investors concerned with the good corporate governance
practices of Hong Kong firms? Do better-governed firms perform better
than firms that are more poorly governed? A positive relation between
corporate governance practices and firm value will provide supporting
evidence to the potential benefit of good corporate governance. Using
publicly available information from publicly traded companies in 2002,
we rate the corporate governance practices of the 168 largest firms
constituting the four major Hong Kong stock indices. While this sample
is only 15 per cent of the total number of listed companies, the firms
chosen account for almost 90 per cent of total market capitalization and
almost 80 per cent of the total market turnover.
Previous efforts to measure corporate governance practices used
surveys with a binary scale to note the presence or absence of a practice.
This ‘‘box-ticking’’ approach has been criticized for not measuring the

quality of corporate governance practices. The present study uses a newly
88 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
designed survey to measure corporate governance practices. This unique
measurement tool permits assessment of both the quality and quantity of
corporate governance practices. The quality of corporate governance
practices is captured using information provided in source documents.
The addition of qualitative measures leads to a richer understanding of
the relation between corporate governance practices and firm perfor-
mance. Survey questions are derived from the Revised OECD Principles
of Corporate Governance (OECD, 2004) and the Code of Best Practices
(HKEx, 1999a). The survey is comprised of 86 questions addressing the
five categories of the OECD Principles of Corporate Governance: rights
of shareholders; equitable treatment of shareholders; role of stake-
holders; disclosure and transparency; and board responsibilities and
composition. The survey responses and qualitative values are combined
to create a corporate governance index (CGI) score for each surveyed
firm.
The findings show that listed Hong Kong firms exhibit a wide
disparity in the quality of their corporate governance practices. Overall
CGI scores range from a low of 32 (out of 100) to a high of 77 for firms
with good practices. The empirical findings offer compelling evidence
that good corporate governance practices do matter in Hong Kong.
A positive and statistically significant relation is found between the
performance measures and the CGI, even after the inclusion of firm
characteristics as control variables. The results are robust whether
market-to-book ratio (MTBV) or return on equity (ROE) is used as
the performance indicator. The results are also robust to the number of
survey questions used in the index, up to a random reduction of 10 of the
86 questions.

Of the five categories of the OECD Principles of Corporate Govern-
ance, we find that disclosure and transparency are highly correlated with
market valuation. To further examine the effect of corporate disclosure
and transparency on firm value, we construct sub-indices for transpar-
ency and non-transparency and find a positive and statistically significant
relation between the transparency index and market valuation. The
findings provide empirical support for the notion that, for Hong Kong
listed companies, good corporate governance practices are consistent
with value maximization.
The remainder of this paper is organized as follows: Section 2 frames
the study through a literature review; the data sources and study
methodology are described in Section 3; Section 4 presents the results;
and Section 5 concludes the paper.
Do Investors Really Value Corporate Governance? 89
r Blackwell Publishing Ltd. 2007.
2. Literature Review
Numerous studies have examined the relation between corporate govern-
ance practices and firm performance, primarily in developed countries.
These prior studies often concentrate on a narrow aspect of governance
such as ownership structure, board composition, executive compensa-
tion, or disclosure and transparency. These studies, taken together, often
show conflicting results, indicating that the link between good corporate
governance practices and superior firm performance is not clear. Differ-
ent aspects of corporate governance have complex interrelations and may
be complementary. Therefore, the lack of a particular aspect may be
offset by the presence of other aspects.
Looking first at the relation between ownership characteristics and
firm performance, Morck et al. (1988) and McConnell and Servaes (1990)
document a non-linear relation between insider ownership and firm
value. These authors argue that ownership by firm insiders aligns the

interests of insiders with the interests of outside shareholders, increasing
firm value. However, as inside ownership increases, the entrenchment
effects of inside ownership dominate and higher inside ownership
becomes associated with a lower firm value. In contrast, Himmelberg
et al. (1999) suggest that managerial ownership and firm performance are
determined by a common set of characteristics and question the causal
relation between ownership and firm performance implied by Morck
et al. (1988). A series of studies finds that concentrated ownership may
lead to more active monitoring, resulting in better corporate governance
(Hill and Snell, 1988; Weiss and Nikitin, 2004).
3
Bhagat et al. (2004),
however, do not find evidence supporting a positive association between
ownership concentration and firm performance. In East Asia, the
concentration of insider ownership is higher than in other regions of
the world, making ownership concentration a critical factor (Claessens
et al., 2000).
Another major topic in the corporate governance literature is board
composition, especially the impact of outside directors, that is, indepen-
dent non-executive directors (INEDs). In 2002, at the time of this study,
the board of each listed company in Hong Kong is required to have at
least two INEDs. Beginning in 2005, the number of INEDs required
increased to a minimum of three per company. Agency theory suggests
that outsiders are important monitors of management and providers of
relevant expertise central to the effective resolution of agency problems
between managers and shareholders (Fama, 1980; Fama and Jensen,
90 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
1983). Independent directors play major roles in US and UK public
companies. Bhagat and Black (2002) report that most large US public

companies have independent directors making up a high proportion
of the board. Faccio and Lasfer (1999) report that on average,
non-executive directors make up 43 per cent of boards in the United
Kingdom.
In contrast, several researchers have raised concerns that outside
directors lack the necessary time, expertise, and incentives and thus
may not be able to make a meaningful contribution to shareholder
wealth creation (Mace, 1986; Patton and Baker, 1987). Agrawal and
Knoeber (1996) find a negative relation between the proportion of
outside directors and firm performance among US firms. They posit
that a political process within firms influences the selection of outside
directors, and the directors may be less effective as they are beholden
through the selection process. Bhagat and Black (2002) investigate US
firms and find no significant relation between board independence and
long-term firm performance. Klein (1998) finds no association between a
firm’s committee structure and firm value. Neither activism of institu-
tional investors (Carleton et al., 1998) nor ownership by outside
blockholders (Bhagat et al., 2004) is found to have an important effect
on firm value.
Disclosure and transparency practices can also influence the extent
and quality of a firm’s corporate governance practices. Recently, much
effort has been focused on increasing the quality of corporate disclosure
and transparency in the Asia-Pacific region. Bushman et al. (2004) define
corporate transparency as the availability of firm-specific information to
outside investors and stakeholders. This information is needed to make
investment and resource allocation decisions. The levels of corporate
transparency depend on firms’ willingness to disclose relevant informa-
tion to the public. Disclosure and transparency can help protect share-
holders’ rights, helping them to feel confident that the firm is being
managed with their interests in mind.

The divergence of findings in these studies may be due to different
proxies being used to measure corporate governance.
4
The lack of
consistency may also be attributed to the narrow focus of previous
studies. Typically, studies consider one or at most a few components
constituting corporate governance instead of a composite measure.
Studies on the association between overall corporate governance
practice and firm market value are limited, but work by Gompers
et al. (2003), Black (2001), Bebchuk et al. (2004), and Brown and
Do Investors Really Value Corporate Governance? 91
r Blackwell Publishing Ltd. 2007.
Caylor (2006) examine this connection for US firms while Black (2001),
Drobetz et al. (2003), and Black et al. (2006) do the same for non-US
firms.
Gompers et al. (2003) construct a governance index to proxy the level
of shareholder rights with respect to takeovers. Their data are derived
from Investor Responsibility Research Center (IRRC) publications that
track 22 measures such as company charter provisions, bylaw provisions,
and other firm-level rules, plus coverage under takeover laws of six states.
Gompers et al. (2003) identify 24 unique provisions and build an index.
They find evidence that firms with stronger shareholder rights have a
higher firm value, higher profits, and lower capital expenditures. How-
ever, they only consider takeover defense provisions and other provisions
related to shareholder rights. Although their study makes an important
contribution to the literature on takeover defenses in the United States, it
is of limited relevance to Asian markets where hostile takeovers seldom
occur. Bebchuk et al. (2004) construct an entrenchment index based on
six provisions: staggered boards, limits to shareholder bylaw amend-
ments, supermajority requirements for mergers, supermajority require-

ments for charter amendments, poison pills, and golden parachute
arrangements. The first four provisions measure the rights and participa-
tion of shareholders and the last two provisions cover hostile takeovers.
These measures are selected from a total of 24 governance provisions
developed by the IRRC. They find that increases in the level of the
entrenchment index are monotonically associated with significant reduc-
tions in firm valuation. These authors also find higher firm valuation, as
measured by Tobin’s Q, at firms with stronger shareholders’ rights.
Brown and Caylor (2006) create a CGI based on 51 corporate governance
provisions propounded by Institutional Shareholder Services. They find
that better-governed firms are associated with both higher returns on
equity and higher returns on assets.
Compared with US market studies, recent research on emerging
markets generates results that generally support a link between corporate
governance practices and firm valuation. For Russian firms, Black (2001)
finds a positive relation between corporate governance behavior and
market performance; however, his result is based on a small sample of 21
firms. Drobetz et al. (2003) follow the approach of Gompers et al. (2003),
developing a governance index and linking it to the performance of
German firms. As with the original study, the conclusions from this
takeover defense study for German firms throw limited light on Asian
markets. Ho and Wong (2001) compare the effectiveness of different
92 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
corporate disclosure practices in Hong Kong and identify ways to
enhance communication between corporations and investors. Leung
and Horwitz (2004) find that: (1) concentrated board ownership is
correlated with low voluntary disclosure, and this negative relation is
stronger when firm performance is very poor; (2) non-executive directors
enhance voluntary disclosure for firms with low director ownership but

not for firms with concentrated ownership. Black et al. (2006) create a
governance score using a survey conducted by the Korean Stock
Exchange and find that firms with higher scores have a higher market
value. However, their survey relies on companies’ responses that could
generate a selection bias in the data.
There are other studies examining the relation between corporate
governance and firm performance for Asian markets. Durnev and Kim
(2005) use the Credit Lyonnais Securities Asia (CLSA, 2002) governance
index and the S&P disclosure score (Standard and Poor’s, 2002) to
measure corporate governance practices for a sample of 859 large firms in
27 countries. In their theoretical model, they identify three firm attributes
(investment opportunities, external financing, and ownership structure)
that relate to corporate governance and conclude that firms with higher
scores carry higher stock market valuations. Klapper and Love (2003)
use the CLSA governance index and find a positive correlation between
market value and corporate governance for 374 firms in 14 countries.
They also document a positive relation between governance and operat-
ing performance as measured by return on assets (ROA). Mitton (2004)
also uses the CLSA governance index to show that firms (365 firms from
19 countries) with stronger corporate governance have higher dividend
payouts. However, despite its frequent use, the CLSA index has a
fundamental limitation. The index is based on the judgment of security
analysts, who would naturally be subjective, potentially creating bias in
the data.
From the literature review, measuring the quality of corporate
governance practices is clearly an important issue. Previous studies
only identify the presence of certain practices without considering the
variation among those practices. Previous studies also narrowly define
corporate governance and are thus applicable only to some parts of the
world. Our study seeks to resolve these issues by creating a measurement

scale that can detect variation in corporate governance quality. The
measurement scale is also comprehensive, incorporating aspects of
corporate governance practices accepted and recognized by international
organizations.
Do Investors Really Value Corporate Governance? 93
r Blackwell Publishing Ltd. 2007.
3. Data and Methods
Our sample consists of the 168 largest companies
5
constituting the four
major Hong Kong Exchange and Clearing Limited (HKEx) indices:
Hang Seng Index (HSI), Hang Seng Hong Kong Composite Index
(HSHKCI), Hang Seng China-Affiliated Corporate Index (HSCCI),
and Hang Seng China Enterprise Index (HSCEI). The HSI is the most
widely quoted performance indicator of the Hong Kong stock market. It
currently has 33 constituent stocks, which cover about 70 per cent of the
market capitalization of all eligible stocks listed on the Main Board of
the HKEx. Constituents of the HSHKCI include the HSI constituent
stocks plus the largest Hong Kong companies. The HSCCI and HSCEI
constituent stocks are China-related companies; the HSCCI is for red-
chip companies, while the HSCEI is for H-share companies.
6
Both
H-share and red-chip companies are listed in Hong Kong. H-share
companies are incorporated in mainland China while red-chip companies
are incorporated in Hong Kong but are controlled (at least 35 per cent
owned) by state, provincial, or municipal-owned organizations in China.
The constituents of these four indexes are representative of all Hong
Kong shares. Although the firms in the four indices account for 15 per
cent of all listed companies, they account for almost 80 per cent of the

Hong Kong market turnover and almost 90 per cent of the total market
capitalization.
Based on the Revised OECD Principles (OECD, 2004) and the Code
of Best Practices (HKEx, 1999a), we develop a survey composed of 86
questions.
7
The question classification scheme matches the five OECD
Principles: rights of shareholders; equitable treatment of shareholders;
role of stakeholders; disclosure and transparency; and board responsi-
bilities and composition. From the five OECD Principles, the questions
are modified to be consistent with the Code of Best Practices (HKEx,
1999a) and make the questionnaire more applicable to Hong Kong firms.
8
The corporate governance practices of listed companies are examined
from an outside shareholder’s perspective, using publicly available
information an investor would use to make an investment decision.
Our data sources include annual reports, articles of association, memor-
andums of association, notices to call shareholders’ meetings, annual
general meeting minutes, company websites, analyst reports, and other
sources. After gathering the data, we rate each company on the 86 survey
questions. To construct a CGI, each question within a specific survey
category is weighted, as is each category: rights of shareholders (15 per
94 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
cent); equitable treatment of shareholders (20 per cent); role of stake-
holders (5 per cent); disclosure and transparency (30 per cent); and board
responsibilities and composition (30 per cent).
9
Major questions under
each category are equally weighted, and sub-questions under each major

question are equally weighted as well.
10
We combine question scores
within the five category sub-indices, which are then combined into an
overall score. A total CGI value, ranging from 0 to 100, is then calculated
for each company.
The majority of questions (87 per cent) have a binary (yes, no)
response. For the remaining questions, clear standards are used to
evaluate the quality of corporate governance practices. There are three
quality levels based on practices observed: good, matching international
standards or best practice for Hong Kong firms; fair, that the practice
meets but does not exceed the required standard; or poor, that the
practice is below standard or missing. This feature of the questionnaire
represents a major improvement in measuring the quality of corporate
governance practices because it can capture variation in degrees of
corporate governance practices. In addition, each company is assessed
by two different raters to ensure consistency. The final results are cross-
checked and audited by the research team.
Once individual firm scores are complete, we examine the relation
between the overall CGI scores and firm performance. A series of
regression analyses are performed using both accounting performance
and market performance as dependent variables. The regression model is
given by
MTBV ¼ a þ b
1
CGI þ b
2
ROA þ b
3
LnðTAÞþb

4
Current þ b
5
D=E
þ b
6
BOUT þ b
7
BEXC þ b
8
Top5 þ b
9
DummyCEO & Chairman
þ b
10
DummyAudit þ b
11
DummyCompensation þ b
12
DummyADR
þ b
13
DummyMSCI þ b
14
DummyHshares & redchips þ e ð1Þ
The dependent variable is a proxy of market valuation, the MTBV.
This ratio provides an estimate of the total value of a firm (including
intangible assets such as monopoly power, goodwill, high-quality man-
agers, and growth opportunities) and reflects firm performance (Tobin
and Brainard, 1968). MTBV is considered a better measure of firm

performance than accounting measures (e.g., ROE or ROA) because it is
based on market value, not just accounting earnings and is not affected
by earnings management or accounting manipulations. The explanatory
variable of interest is the CGI. The control variables used include
Do Investors Really Value Corporate Governance? 95
r Blackwell Publishing Ltd. 2007.
financial measures such as profitability (ROA ) and firm size (total assets
or TA), and risk factors as described by the current ratio (Current) and
debt to equity ratio (D/E). Other explanatory variables include board
characteristics such as the number of outside (BOUT) and executive
directors (BEXC) on the board.
11
Ownership concentration is captured
by the variable TOP5, the percentage of shares owned by the top five
shareholders. Lastly, a series of corporate governance dummy variables
are included as indicators of whether the CEO is also the Chairman,
whether the firm has audit or compensation committees, whether the
company has American depositary receipts (ADRs) trading outside of
Hong Kong, whether the company is a constituent of a Morgan Stanley
Capital International Index (MSCI) index, and whether the company is
an H-share or Red Chip firm. To check for robustness of the results, the
model is re-run replacing MTBV with firm profitability as measured by
ROE. The descriptions of variables are presented in Table 1.
Endogeneity is always a concern for studies dealing with the relation
between firm value and corporate governance attributes (Black, 2001).
Even though a firm that practices good corporate governance may be
more likely to make a high profit, investors may value the high profit
rather than the corporate governance characteristics. To avoid misinter-
preting investor behavior, we tackle this problem using two approaches.
First, we include a comprehensive set of control variables to mitigate the

omitted-variable bias and the possibility that our results are affected by
endogeneity. Second, we include an instrumental variable to minimize
endogeneity. Each of these approaches will be discussed in the following
sections describing the robustness tests. Furthermore, we examine the
relation between each of the five sub-indices of the CGI and market
valuation to test whether any one of these sub-components is more
significantly driving the relation. To perform the analyses, the overall
CGI scores are replaced in the regression models with the scores from the
five survey sub-indices.
In their study on the association between overall corporate govern-
ance practice and firm market value for US firms, Bebchuk et al. (2004)
construct an entrenchment index based on six governance provisions
out of a larger set of 24. The six chosen provisions are negatively
correlated with firm value during the 1990–2003 sample period and
better explain the relation with firm value than the remaining 18
provisions. In a similar vein, we propose to split the CGI into two
components, a transparency index and a non-transparency index, in
order to test whether transparency characteristics (firms with greater
96 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
disclosure) better explain the relation to firm value than do the remaining
four sub-components of the CGI. To construct these two indices, we
separate the survey questions listed in Appendix A into two groups:
questions related to disclosure, which are used to compile the transpar-
ency index, the remaining questions constituting the non-transparency
index. We then examine the impact of corporate disclosure and trans-
parency on market valuation.
Accounting information and market-related data are obtained from
Datastream and Bloomberg. Accounting data are based on the firms’
fiscal year ends. Firm market values and other performance data are

Do Investors Really Value Corporate Governance? 97
Table 1. Variable Definitions
Variables Descriptions
CGI Corporate Governance Index constructed based on OECD principles
TINDEX Transparency Index constructed based on all disclosure-related
questions in the survey
NONTINDEX Non-transparency Index constructed based on all non-disclosure
related questions in the survey
MTBV Defined as market value of common stock/book value of common
stock. We drop two firms with negative book value of common stock
ROA Return on assets
Ln(TA) Natural log of total assets
Current Current ratio
D/E Debt to equity ratio
BOUT Number of outside directors on the board (including independent
non-executive directors, non-executive directors, and honorable
directors)
BEXC Number of executive directors on the board
TOP5 Percentage of total outstanding shares held by five largest
shareholders
DummyCC Dummy variable, equals 1 if CEO is also the Chairman of the board; 0
otherwise
DummyA Dummy variable, equals 1 if firm has a board audit committee; 0
otherwise
DummyC Dummy variable, equals 1 if firm has a board compensation
committee; 0 otherwise
DummyHR Dummy variable, equals 1 if firm is a red chip or H-share; 0 otherwise
DummyADR Dummy variable, equals 1 if firm has ADRs trading in the United
States; 0 otherwise
DummyH Dummy variable, equals 1 if firm is an H-share; 0 otherwise

DummyMSCI Dummy variable, equals 1 if firm is included in an MSCI index; 0
otherwise
This table provides brief descriptions of the variables used in the regression analyses.
Accounting data and firm performance information are for 2002, drawn from Datastream
and Bloomberg. The firm performance variables are geometric averages of monthly data based
on firms’ fiscal years. Accounting variables are annual values based on firms’ fiscal years.
r Blackwell Publishing Ltd. 2007.
based on monthly data for 2002. We then use geometric averaging to
calculate annual values.
4. Results
4.1 Descriptive Statistics
A major contributio n o f t his s tudy is the creation of a CGI for the largest
168 listed companies in Hong Kong. Panel A in Table 2 contains
descriptive statistics of the CGI and its five sub-indices. On a scale of 0–
100, the CGI ranges from a low of 32.86 to a high of 76.34; the average
score is 48.33. The results of the sub-indices indicate that companies do best
in Section B (equitable t reatment of shareholders) and Section D (dis-
closure and transparency), with mean scores of 82.78 and 74.88, respec-
tively. The results for Section C (role of stakeholders; mean score of 6 9.54)
and Section E (board responsibilities and composition; mean score of
60.70) show that, on average, companies perform well in these areas. The
mean score for Section A (rights of shareholders) was the lowest at 42.96.
Panel B of Table 2 presents CGI values for each of the four stock
market indices. The results show that H-share companies (HSCEI) have
the lowest CGI scores, with a mean score of 44.54. The HSI constituent
stocks (HSI) show the highest mean CGI score of 52.26. The CGI scores
grouped by industry sector are summarized in Panel C of Table 2. Across
industry classifications, there does not seem to be much difference in the
mean scores. Companies in the finance and utilities sectors have the
highest average scores at 52.86 and 51.68, respectively, while companies

in the property sector have the lowest average score at 45.12.
Table 3 shows the descriptive statistics of companies included in our
sample. Statistics for each stock market index grouping are shown as
well. For the full sample, covering only 2002, the average firm size is
HK$20,396 million; the ROA is 4 per cent; the current ratio is 2.05; and
the debt-equity ratio is 1.39. There are some obvious differences across
the four stock index groupings. HSI firms are larger, judging from total
assets; are more profitable, judging from ROA; and have much larger
market values than the other three groupings. Other financial measures,
such as the debt/equity ratio and the current ratio, differ widely across
index groupings as well. Table 3 also includes descriptive statistics for
board composition. Across the whole sample, the mean number of
outside directors is just under five (4.97), while the mean number of
executive (inside) directors is just under six (5.95).
98 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
Turning next to a measure of ownership concentration, we see that on
average, the top five shareholders control more than 50 per cent of
company shares, although the ownership concentration varies across the
Do Investors Really Value Corporate Governance? 99
Table 2. Descriptive Statistics for the Corporate Governance Index (CGI)
Panel A: Descriptive statistics of CGI and sub-indices
Total
CGI
Section A Section B Section C Section D Section E
Rights of
shareholders
Equitable
treatment of
shareholders

Role of
stakeholders
Disclosure
and
transparency
Board
responsibilities
and composition
Mean 48.33 42.96 82.78 69.54 74.88 60.70
Minimum 32.86 27.79 63.33 41.67 64.01 41.67
Maximum 76.34 84.49 93.33 100.00 90.33 81.67
SD 6.98 9.10 6.53 13.03 5.20 7.89
N 168 168 168 168 168 168
Panel B: CGI by listing category
Total CGI HSI HSHKCI HSCCI HSCEI
Mean 48.33 52.26 49.23 49.00 44.54
Minimum 32.86 35.52 35.52 40.56 32.86
Maximum 76.34 76.34 76.34 65.83 59.01
SD 6.98 9.15 7.07 5.68 6.59
N 168 33 110 27 31
Panel C: CGI by industry sector
Total
CGI Industrial Properties Finance Utilities
Consolidated
enterprises
Hotel
and
others
Mean 48.33 46.39 45.12 52.68 51.68 49.66 49.32
Minimum 32.86 32.86 35.52 40.32 36.95 35.34 39.53

Maximum 76.34 61.05 56.87 71.96 76.34 65.83 60.65
SD 6.98 5.41 5.97 7.54 13.86 5.89 10.64
N 168 62 21 19 9 54 3
This table presents the descriptive statistics of the corporate governance index (CGI) for the 168
companies included in the sample in 2002. The CGI ranges from a low of 0 up to 100. Panel A
shows the descriptive statistics of the CGI and the five sub-indices. Panel B shows the descriptive
statistics of the CGI categorized by stock market index. Panel C shows the descriptive statistics
of the CGI by industry.
HSI, Hang Seng Index; 33 best-known Hong Kong companies; HSHKCI, Hang Seng Hong
Kong Composite Index; the largest Hong Kong companies, including HSI; HSCCI, Hang Seng
China Affiliated Corporate Index; ‘‘red chip’’ companies; HSCEI, Hang Seng China Enterprise
Index; H-share companies.
r Blackwell Publishing Ltd. 2007.
100 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
Table 3. Descriptive Statistics for Key Variables
All HSI HSHKCI HSCCI HSCEI
Market Value (in million HK$) Mean 20,395.87 88,227.87 27,870.77 25,754.78 3,920.87
Minimum 575.32 11,767.79 575.32 890.12 810.59
Maximum 832,653.75 832,653.75 832,653.75 416,826.03 27,369.95
SD 76,356.70 156,370.02 90,962.89 80,826.03 5,899.57
ROA (return on assets) Mean 0.04 0.07 0.04 0.05 0.05
Minimum (0.42) (0.15) (0.42) (0.34) 0.00
Maximum 0.62 0.18 0.62 0.28 0.21
SD 0.10 0.07 0.12 0.10 0.04
Ln(TA) (total assets) Mean 16.18 17.88 16.16 16.04 16.63
Minimum 11.57 15.27 11.57 13.61 14.94
Maximum 22.50 22.50 22.50 19.41 20.00
SD 1.66 1.58 1.85 1.26 1.19
Current (current ratio) Mean 2.05 1.61 2.11 1.89 2.03
Minimum 0.32 0.36 0.32 0.37 0.38

Maximum 12.44 3.43 8.02 4.97 12.44
SD 1.68 0.95 1.47 1.06 2.39
D/E (debt/equity) Mean 1.39 1.92 1.61 0.98 0.79
Minimum 0.06 0.15 0.06 0.11 0.07
Maximum 13.60 13.60 13.60 2.93 3.36
SD 2.23 3.59 2.61 0.69 0.79
BOUT Mean 4.97 6.76 5.33 3.74 4.57
(Number of outside directors) Minimum 2.00 2.00 2.00 2.00 2.00
Maximum 14.00 14.00 14.00 7.00 13.00
SD 2.83 3.39 3.01 1.29 2.53
r Blackwell Publishing Ltd. 2007.
Do Investors Really Value Corporate Governance? 101
Table 3. (Continued.)
All HSI HSHKCI HSCCI HSCEI
BEXC Mean 5.95 7.06 5.54 7.67 6.30
(Number of executive directors) Minimum 1.00 3.00 1.00 3.00 2.00
Maximum 15.00 15.00 15.00 14.00 10.00
SD 2.87 3.41 2.92 3.20 2.09
TOP5 Mean 0.53 46.79 49.58 55.15 65.63
(Percentage of shares owned by
the TOP5 shareholders)
Minimum 0.01 3.06 1.06 23.54 41.74
Maximum 0.99 77.72 98.20 75.69 98.97
SD 0.19 19.22 19.15 13.63 16.01
Number of firms:
With dual role for CEO and
Chairman
56 13 43 8 8
With audit committee 134 30 99 24 17
With compensation committee 38 12 30 4 8

With ADR s traded in the
United States
91 28 63 11 22
Belong to MSCI 36 22 36 – –
That are an H-Share or
red-chip company
58 6 6 27 31
HSI, Hang Seng Index; HSHKCI, Hang Seng Hong Kong Composite Index; HSCCI, Hang Seng China Affiliated Corporate Index; HSCEI, Hang Seng
China Enterprise Index.
r Blackwell Publishing Ltd. 2007.
stock index groupings. Some additional statistics, shown at the bottom of
Table 3, help further characterize the sample. Among the 168 firms in the
sample, 56 firms (33 per cent) have the same person serving as both CEO
and board chairman. A majority of firms (134 companies; 80 per cent)
have an audit committee, but only 38 firms (23 per cent) have compensa-
tion committees. More than 90 companies have ADRs currently trading
in the United States, with the largest portion from the HSI (28 out of 33
firms) and HSHKCI (63 out of 110 companies) groupings. A large
portion of the HSI and HSHKCI sample companies are included in the
MSCI. Lastly, of the sample of 168 companies, 58 firms are either
H-shares or red chips.
12
4.2 Regression Analyses
The main research question is to assess whether good corporate govern-
ance practices lead to higher firm values in Hong Kong. Regression
analyses between the CGI measure and explanatory and control variables
should provide an additional insight. Figure 1 provides the first indica-
tion of a possible relation between corporate governance practices and
market valuation. Using the MTBV as a proxy for market value, Figure
1 shows a positive and statistically significant correlation between the

102 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
30 35 40 45 50 55 60 65 70 75 8
0
0
1
2
3
4
5
6
7
8
9
10
Corporate Governance Index
Market Value
Figure 1. The relation between market value and corporate governance
index (CGI). The correlation is statistically significant at the 5 per cent level.
r Blackwell Publishing Ltd. 2007.
CGI and the market value. This implies that companies with higher
corporate governance scores also have higher MTBV s, and vice versa.
Our results provide initial support for the conclusion that good corporate
governance is positively related to market valuation. Table 4 lists the
correlation coefficients between all variables with a statistical significance
at the 5 per cent level or better.
4.2.1 Regression results for market performan ce (market-to-book value).
Table 5 displays ordinary least squares regression results for the CGI and
control variables, excluding 19 financial institutions in the sample as
noted earlier.
13

The first column in Table 5 shows the result of regressing
MTBV on CGI. The CGI coefficient is 0.0330 and is statistically
significant at the 5 per cent level. We progressively add control variables
in models (2) and (3), and obtain similar results: CGI is positively and
significantly associated with market performance, as measured by
MTBV. When all control variables are included in model (3), the
coefficient for CGI is 0.0337, which is still statistically significant at the
5 per cent level. The consistent results in Table 5 show that several
control variables explain the variation in MTBV. The first set of control
variables are financial characteristics that may be related to market
valuation. For example, it is reasonable to expect that investors evaluate
listed companies based on their profitability. Therefore, ROA,an
indicator for firm profitability, is likely to be associated with MTBV.
We find that ROA is positively and significantly related to MTBV in two
regression models as shown in Table 5. In models (2) and (3), the
coefficients are all statistically significant at the 5 per cent level or better,
implying that investors value Hong Kong listed companies based on
profitability. Turning next to firm size, Black et al. (2006) claim that firm
size can plausibly affect both a firm’s market value and its governance
practices. Consistent with prior research (e.g., Lang and Stulz, 1994),
models (2) and (3) in Table 5 show coefficients for firm size that are
negative and statistically significant at the 5 per cent level or better. In
terms of risk factors, both capital structure and leverage can affect a
firm’s MTBV and, potentially, CGI. We include the current ratio and
debt/equity ratio to control for liquidity and financial risk. The coeffi-
cients are not statistically significant in any of the regression results
shown in Table 5.
Both board structure and shareholding structure are important ele-
ments of corporate governance practices. These characteristics are
Do Investors Really Value Corporate Governance? 103

r Blackwell Publishing Ltd. 2007.
104 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
Table 4. Correlation Matrix for Regression Variables
12345678910111213141516
1 Total CGI 1.000
2 MTBV 0.190 1.000
3 Ln(TA) 0.196 À 0.282 1.000
4 ROA 0.084 0.305 À 0.056 1.000
5 Current À 0.062 0.131 À 0.386 0.139 1.000
6 BEXC À 0.151 À 0.248 0.258 0.020 À 0.180 1.000
7 BOUT 0.355 0.043 0.460 0.027 À 0.067 À 0.190 1.000
8 TOP5 À0.116 À 0.093 0.097 0.005 À 0.154 0.067 À 0.190 1.000
9 DummyA 0.370 0.040 0.025À 0.056 0.082 À 0.004 0.147 À 0.118 1.000
10 DummyC 0.556 0.135 0.203 0.138 À 0.091 À 0.203 0.186 À 0.096 0.272 1.000
11 DummyCC À 0.037 0.079 À 0.048 0.016 À 0.025 À 0.130 À 0.102 À 0.016 À
0.052 0.040 1.000
12 DummyMSCI 0.227 0.169 0.392 0.088 À 0.088 À 0.072 0.402 À 0.321 0.010 0.134 0.062 1.000
13 DummyADR 0.107 À 0.096 0.436 0.045 À 0.130 0.025 0.314 0.002 À 0.017 0.126 0.068 0.247 1.000
14 DummyHR À 0.179 À 0.127 0.078 0.073 À 0.043 0.255 À 0.205 0.291 À 0.164 À 0.033 À 0.089 À 0.379 0.040 1.000
15 D/E 0.085 0.302 0.245 À 0.037 À 0.025 À 0.030 0.212 0.078 0.014 0.142 À 0.150 0.066 À 0.026 À 0.049 1.000
16 DummyH À 0.259 0.093 À 0.042 0.091 0.569 À 0.075 0.086 À 0.075 À 0.708 0.087 À 0.130 0.205 À 0.362 0.084 À 0.111 1.000
This table provides correlation coefficients between variables used in the regression analyses. Statistically significant correlations (at 5% level or better)
are shown in bold.
r Blackwell Publishing Ltd. 2007.
Do Investors Really Value Corporate Governance? 105
Table 5. Regression Results for Firm Performance Measures with CGI Scores
MTBV ROE
(1) (2) (3) (4) (5) (6)
CGI 0.0330
nn

0.0475
nnn
0.0337
nn
0.0043
nn
0.0029
n
0.0044
n
(1.98) (2.86) (1.98) (2.48) (1.87) (1.92)
ROA 3.0430
nnn
2.2866
nn
(2.67) (2.25)
Ln(TA) À 0.2043
nn
À 0.3583
nnn
0.0214
n
0.0113
( À 2.56) ( À 3.37) (1.71) (0.83)
Current À 0.0317 0.0098 0.0090 0.0101
( À 0.71) (0.23) (0.76) (0.83)
D/E À 0.0190 0.1423 À 0.0832
nnn
À 0.0832
nnn

( À 0.18) (1.22) ( À 3.65) ( À 3.74)
BOUT À 0.0249 À 0.0193 À 0.0057 À 0.0081
( À 0.51) ( À 0.38) ( À 1.05) (À 1.57)
BEXC À 0.1089
nnn
À 0.0622 0.0008 0.0035
( À 2.97) ( À 1.39) (0.15) (0.51)
TOP5 À 0.0128
nnn
À 0.0011 À 0.0002 0.0004
( À 2.85) ( À 0.17) ( À 0.26) (0.36)
DummyCC 0.1738 À 0.0192
(0.85) ( À 0.50)
DummyA 0.2321 À 0.0606
n
(1.31) ( À 1.72)
DummyC 0.3000 À 0.0085
(0.90) ( À 0.22)
DummyADR À 0.1421 0.0230
( À 0.61) (0.79)
DummyMSCI 1.4939
nnn
0.0601
(3.26) (1.23)
DummyHR 0.3374 À 0.0171
(1.65) ( À 0.34)
Adjusted R
2
0.0258 0.2653 0.3816 0.0208 0.2751 0.2686
F-Statistics 4.87

nn
6.37
nnn
6.24
nnn
4.48
nn
7.45
nnn
4.12
nnn
This table shows the ordinary least squares regression results using (i) market-to-book value
(MTBV) and (ii) return on equity (ROE) as the dependent variables. MTBV is the market value
divided by the book value of common stock. ROE is net income after taxes divided by the book
value of equity. CGI is the Corporate Governance Index; ROA is return on assets, Ln(TA) is the
natural logarithm of total assets, Current is the current ratio, D/E is the debt to equity ratio,
BOUT is the number of outside directors on the board, BEXC is the number of executive
directors on the board, and TOP5 is the percentage of total outstanding shares held by the five
largest shareholders. DummyCC equals 1 if the CEO is also the Chairman of the board and 0
otherwise; DUMMYA equals 1 if the firm has a board audit committee and 0 otherwise;
DUMMYC equals 1 if the firm has a board compensation committee and 0 otherwise;
DummyADR equals 1 if the firm has ADRs trading in the US and 0 otherwise; DummyMSCI
equals 1 if the firm is included in an MSCI index and 0 otherwise; and DummyHR equals 1 if the
firm is a red chip or H-share and 0 otherwise. t-statistics, based on White’s heteroskedasticity-
consistent standard errors and covariance, are reported in parentheses.
n
,
nn
, and
nnn

denote statistical significance at the 10%, 5%, and 1% level (two-tailed test),
respectively.
r Blackwell Publishing Ltd. 2007.
particularly important in Hong Kong because most listed companies are
dominated by a single person or family group.
14
The survey instrument
includes several questions to evaluate board characteristics such as board
size, composition (the numbers of non-executive and independent
directors), and CEO duality. We also include two board composition
measures as control variables, because the board structure can influence
the role of the board and hence the board’s ability to provide effective
oversight.
A high proportion of executive directors on the board may be an
indicator of bad corporate governance practices. Investors may under-
value a firm with a board dominated by inside directors. The regression
results for models (2) and (3) in Table 5 show that the number of
executive (inside) directors (BEXC) is negatively related to firm value
(MTBV) and is statistically significant at the 5 per cent level or better.
Dispersed ownership could be an indicator of good corporate govern-
ance practices, as there may be less chance for a dominant owner(s) to
take advantage of minority shareholders. Similar results are observed for
ownership concentration (TOP5), which is negative and statistically
significant at the 1 per cent level in model (2). This negative and
significant relation bears further study as it implies that less concentrated
ownership increases value. However, for model (3), which includes all
control variables and dummy variables for other corporate governance
practices, neither the ownership concentration nor board structure
variables are significant. The coefficients for the number of independent
directors (BOUT) are not significant in any regression shown in Table 5.

The results imply that the marginal benefits to market valuation coming
from board characteristics are not statistically significant.
The regression models include other dummy variables to gauge the
adoption of best practices in corporate governance espoused by the Hong
Kong Exchange. These practices include separating the board chairman-
ship and top executive officer positions (non-duality), plus establishment
of an audit committee and a compensation committee. Companies
following these recommended but not required best practices could
send a good signal to investors. Regression results show that the relations
between these corporate governance indicator variables (CEO & Chair-
man, Audit, and Compensation dummy variables) and MTBV are not
statistically significant. These variables are not related to MTBV among
Hong Kong companies. Further, the presence of ADR shares is included
as a control variable as an overseas or dual listing could serve as an
indicator of internationally accepted corporate governance practices.
15
106 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
r Blackwell Publishing Ltd. 2007.
These dual-listed firms may need to comply with more stringent
corporate governance rules when listing in a foreign market, which, in
turn, can lead to higher market valuation. The results do not support this
view as the coefficient for the ADR dummy variable is not statistically
significant in model (3), Table 5.
16
Another control variable is an indicator variable representing whether
or not a company is included in the MSCI, a principal international
stock index. The MSCI index includes 36 major Hong Kong firms. The
results from model (3) in Table 5 show a positive relation that is
statistically significant at the 1 per cent level. This result suggests that
investors find inclusion in the MSCI index to be important for market

valuation. Lastly, we include a dummy variable to separate China-related
firms (H-shares and red-chip companies) from Hong Kong listed firms.
The reason for this is that investors may have different criteria for these
firms because their core business is in China or they are controlled by
Chinese state-owned enterprises. The result, however, is not statistically
significant at conventional levels.
4.2.2 Regression results for accounting performance (ROE). The above
analysis uses MTBV, an indicator of market value, as a market-based
measure of firm performance. This section will examine the relation
between accounting performance and corporate governance practices.
ROE is widely used by investors as a way of assessing firm profitability,
and so we replace MTBV by ROE in the regression model described
above and re-run the same sequence of regressions with the same
explanatory and control variables (except for ROA). The simple regres-
sion of ROE on CGI, model (4) in Table 5, shows that ROE is positively
related to CGI, with a coefficient that is significant at the 5 per cent level.
This relation still holds as we add control variables in models (5) and (6).
These results are consistent with those obtained by Gompers et al. (2003),
who find US firms with weaker corporate governance to be less profit-
able. Our study provides empirical evidence that this is also true in Hong
Kong and that companies with good corporate governance practices are
more profitable.
4.2.3 Regression results for sub-components of the CGI. Table 6 presents
the regression results using independent variables derived from the five
sections of the OECD Principles. The independent variables are the five
sub-indices, one for each section of our survey instrument. The results
Do Investors Really Value Corporate Governance? 107
r Blackwell Publishing Ltd. 2007.
108 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
Table 6. Regression Results for Firm Performance Measures with CGI

Sub-Indices
MTBV ROE
(1) (2) (3) (4)
Rights of shareholders À 0.0177 À 0.0148 À 0.0014 À 0.0011
( À 1.55) ( À 1.24) ( À 1.13) ( À 0.78)
Equitable treatment of shareholders 0.0145 0.0133 0.0016 0.0008
(1.13) (1.06) (0.57) (0.29)
Role of stakeholders 0.0114 0.0136 0.0017 0.0016
(1.05) (1.29) (1.16) (1.10)
Disclosure and transparency 0.0287 0.0364
n
0.0046
nn
0.0052
nn
(1.63) (1.96) (2.14) (2.40)
Board responsibilities 0.0377 0.0290 0.0038 0.0037
(1.49) (1.39) (1.44) (1.47)
ROA 2.7280
nn
1.9732
nn
(2.58) (2.02)
Ln(TA) À 0.2006
nn
À 0.3800
nnn
0.0170 0.0079
( À 2.14) ( À 3.75) (1.51) (0.59)
Current À 0.0468 À 0.0204 0.0081 0.0090

( À 0.91) ( À 0.45) (0.74) (0.79)
D/E À 0.0907 0.0062 À 0.0860
nnn
À 0.0822
nnn
( À 0.77) (0.05) ( À 4.73) ( À 4.45)
BOUT À 0.0180 À 0.0375 À 0.0061 À 0.0086
( À 0.33) ( À 0.80) ( À 1.10) ( À 1.64)
BEXC À0.0889
nn
À 0.0605 0.0009 0.0033
( À 2.18) ( À 1.43) (0.14) (0.47)
TOP5 À 0.0098
nn
0.0014 À 0.0001 0.0005
( À 2.03) (0.24) ( À 0.09) (0.41)
DummyCC 0.2170 0.1975 À 0.0151 À 0.0172
(0.86) 0.92 ( À 0.40) ( À 0.44)
DummyA À 0.0137 0.1226 À 0.0805
nn
À 0.0744
nn
( À 0.07) 0.67 ( À 2.31) ( À 2.18)
DummyC 0.0226 0.0380 À 0.0296 À 0.0321
(0.06) 0.11 ( À 0.62) ( À 0.67)
DummyADR À 0.0330 0.0251
( À 0.15) (0.94)
DummyMSCI 1.4801
nnn
0.0615

(3.50) (1.32)
DummyHR 0.3764
n
À 0.0123
(1.80) ( À 0.24)
Adjusted R
2
0.3778 0.4953 0.3659 0.3805
F-Statistics 4.25
nnn
5.56
nnn
4.33
nnn
3.69
nnn
This table shows the OLS regression results using MTBV and ROE as the dependent variables.
CGI is broken into five components: (i) rights of shareholders, (ii) equitable treatment of
shareholders, (iii) role of stakeholders, (iv) disclosure and transparency, and (v) board
responsibilities. t-statistics, based on White’s heteroskedasticity-consistent standard errors
and covariance, are reported in parentheses.
n
,
nn
, and
nnn
denote statistical significance at the 10%, 5%, and 1% level (two-tailed test),
respectively.
r Blackwell Publishing Ltd. 2007.
show that of the five sub-indices, corporate disclosure and transparency

is the only factor showing a statistically significant relation with market
valuation and firm profitability. This finding leads us to investigate
further corporate disclosure and transparency to see the association with
firm valuation and performance. We create a transparency index using
questions in the original survey that reflect the degree of corporate
disclosure and transparency as related to corporate governance practices
(see Appendix A).
Table 7 presents the regression results using these two indices as
independent variables. Control variables are progressively added from
model (1) to model (3). Both the transparency index and the non-
transparency index are positive and significantly related to MTBV in
models (1) and (2). The non-transparency index is not statistically
significant in model (3), which includes the full set of control variables.
The results show that transparency by itself can explain the variation of
MTBV among Hong Kong companies.
4.3 Robustness
To check the robustness of our findings, we perform two additional tests.
First, we address whether the corporate governance score, and subse-
quent ranking of companies in our sample, is affected by any particular
question or questions. In other words, will omitting certain questions
substantially change the findings? The second test examines the validity
of the regression model. We propose using an instrumental variable
approach and a two-stage regression model to correct for endogeneity.
4.3.1 Number of questions. To ensure that no specific criteria dominate
the results, we run a simulation by randomly removing survey questions
from the equation and recalculating the CGI. The new ranking should
not significantly differ from the old ranking. There are three steps in the
analysis. First, we randomly remove one of the 86 questions and compute
a new CGI value for every firm and re-rank the 168 companies based on
the re-calculated CGI. Next, we calculate the Spearman rank correlation

coefficient between the new and the original ranking. Finally, we repeat
the above procedure 10 times. We extend the analysis by removing two,
five, and 10 randomly chosen questions, repeating the re-scoring and re-
ranking procedure each time. As can be seen in Table 8, it is apparent
that the new and original rankings are highly and significantly correlated,
Do Investors Really Value Corporate Governance? 109
r Blackwell Publishing Ltd. 2007.
with most correlations close to 0.9, even with 10 questions removed.
These results provide evidence that the CGI scores and subsequent
ranking of companies are not dominated by any one question or group
of questions.
110 Y L. Cheung, J. Thomas Connelly, P. Limpaphayom and L. Zhou
Table 7. Regression Results for Firm Performance with Transparency Index
MTBV
(1) (2) (3)
TINDEX 0.0371
nn
0.0414
nn
0.0388
nnn
(2.32) (2.45) (2.80)
NONTINDEX 0.0179
n
0.0202
nn
0.0108
(1.81) (2.19) (0.89)
ROA 2.3482
nn

2.8850
nnn
2.1137
nn
(2.30) (2.63) (2.15)
Ln(TA) À 0.3414
nnn
À 0.2419
nnn
À 0.3850
nnn
( À 4.85) ( À 2.72) (À 3.60)
Current À 0.0575 À 0.0496 À 0.0108
( À 1.05) ( À 1.08) (À 0.25)
D/E À 0.0589 À 0.0319 0.1205
( À 0.58) ( À 0.29) (1.08)
BOUT À 0.0277 À 0.0267
( À 0.56) ( À 0.54)
BEXC À 0.1046
nnn
À 0.0558
( À 2.87) ( À 1.27)
TOP5 À 0.0140
nnn
À 0.0017
( À 3.10) ( À 0.27)
DummyCC 0.1935
(0.95)
DummyA 0.2637
(1.51)

DummyC 0.2254
(0.68)
DummyADR À 0.1382
( À 0.61)
DummyMSCI 1.4847
nnn
(3.30)
DummyHR 0.2346
(1.14)
Adjusted R
2
0.1896 0.2734 0.3943
F-Statistics 5.76
nnn
5.97
nnn
6.16
nnn
This table shows the ordinary least squares regression results using market-to-book value
(MTBV) as the dependent variable. TINDEX is a transparency index composed of disclosure-
related questions from the survey; NONTINDEX is a non-disclosure index consisting of the
remaining survey questions. t-statistics, based on White’s heteroskedasticity-consistent standard
errors and covariance, are reported in parentheses.
n
,
nn
, and
nnn
denote statistical significance at the 10%, 5%, and 1% level (two-tailed test),
respectively.

r Blackwell Publishing Ltd. 2007.