arcay and vázquez - 2005 - corporate characteristics, governance rules and the extent of voluntary disclosure in spain

CORPORATE CHARACTERISTICS,
GOVERNANCE RULES AND THE
EXTENT OF VOLUNTARY
DISCLOSURE IN SPAIN
M. Rosario Babı
´
o Arcay and
M. Flora Muin
˜
oVa
´
zquez
ABSTRACT
This study examines the relationships among corporate characteristics,
the governance structure of the firm, and its disclosure policy. Empirical
evidence supporting this investigation has been gathered from a sample of
Spanish firms listed on the Madrid Stock Exchange. This setting is of
interest because of its low level of investor protection, high ownership
concentration, and poorly developed capital market. Our results show that
a firm’s size, along with some mechanisms of corporate governance such
as the proportion of independents on the board, the appointment of an
audit committee, and directors’ shareholdi ngs and stock option plans, are
positively related to voluntary disclosure. We have also observed that
these governance practices are significantly influenced by cross-listings
and by the ownership structure of the firm.
Advances in Accounting
Advances in Accounting, Volume 21, 299–331
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 0882-6110/doi:10.1016/S0882-6110(05)21013-1
299

1. INTRODUCTION
Corporate disclosure is of critical importance for the efficient functioning of
capital markets. The Special Committee on Financial Reporting of the
American Institute of Certified Public Accountants (AICPA) stated:
Few areas are more central to the national economic interest than the role of business
reporting in promoting an effective process of capital allocation (AICPA, 1994, p. 2).
Although regulators have enforced legislation to ensure that companies
provide at least a minimum set of infor mation to third parties,
1
legal re-
quirements do not always satisfy stakeholder demands.
2
Not surprisingly,
there is considerable variation among companie s in the disclosure of infor-
mation that is not legally required.
Research on the determinants of voluntary disclosure initially focused on
corporate characteristics.
3
The basic assumption is that corporate disclosure
policy is determined by a trade-off between the costs and benefits associated
with disclosure,
4
for which corporate characteristics such as company size
(Depoers, 2000), listing status (Meek, Roberts & Gray, 1995), and a firm’s
performance (Singhvi & Desai, 1971) may serve as useful proxies. On the
other hand, recent research suggests that factors other than cost-benefit
analysis may determine a firm’s disclosure policy. In particular, corporate
governance mechanisms may exert some control over a manager’s actions
(Core, 2001, p. 444), and such mechanisms may help to fulfil the informa-
tional demands of stakeholders. For example, Watts and Zimmerman

(1990), drawing on the insights of the theory of the firm, suggest that man-
agers make acco unting choices that can be considered to be efficient when
they maximize the value of the firm, or opportunistic if they enhance the
manager’s welfare at the expense of other contracting parties. Therefor e, the
argument of efficiency assumes an alignment between organizational and
managerial goals.
The board of directors is regarded as a relevant mechanism in the over-
sight of managerial actions
5
(Fama & Jensen, 1983). Researchers have ex-
amined the role played by certain practices aimed at enhancing the
monitoring role of the board on the provision of voluntary information
(Chen & Jaggi, 2000; Ho & Wong, 2001; Nagar, Nanda & Wysocki, 2003).
Comparisons among these studies provide some interesting insights. For
example, Chen and Jaggi (2000) found that the appointment of non-
executive directors was significantly related to disclosure, whereas Ho and
Wong (2001) did not find it to be a significant variable in explaining cor-
porate disclosure. Moreover, previous research focused on a particular
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ300
corporate governance mechanism (i.e. the board of directors), but as
Bushman and Smith (2001, p. 286) stated, a complete understanding of the
role played by corporate governance requires an explicit recognition of in-
teractions across different mechanisms.
In this paper, we seek to extend previous research in three different ways:

(1) by exploring the role played by a number of good governance practices
in enhancing corporate disclosure, (2) by examining the role of majority
shareholders in the adoption of practices of good governance, and (3) by
analysing the direct and indirect effects of various corporate characteristics
on voluntary disclosure.
First, we analyse the role played by the board of directors in enhancing
corporate disclosure by examining a wide range of practices of good gov-
ernance: the appointment of independent directors, the formation of an
audit committee, the separation of the functions of CEO and chairman of
the board, the participation of board members in the capital of the com-
pany, the establishment of stock option plans as a means of directors’ re-
muneration, and the size of the board. Nonetheless, the individual analysis
of a particular practice of good governance does not allow for a full as-
sessment of its role in promoting transparency, inasmuch as complement-
arities and substitutability between them make the whole worthier than the
mere aggregation of its individual constituents.
6
In this study, we have em-
ployed confirmatory factor analysis to reduce governance practices to a
single factor for the assessment of their global effect on corporate disclosure.
Second, our analysis takes into account the role played by majority
shareholders in corporate governance, as well as interactions between the
adoption of good governance practices and the ownership structure of the
company. Evidence is gathered from Spa in, a country especially suited to
the development of this analysis. In contrast with common law countries,
the Spanish institutional setting has in common with other European Con-
tinental countries a relatively low number of listed companies,
7
an illiquid
capital market, the existence of a large number of inter-corporate share-

holdings and, above all, a high level of concentration in corporate share-
holdings
8
(Leech & Manjo
´
n, 2002, p. 169). Research has shown that unlike
civil-law nations, common-law countries enact rules and legislation to shield
the informational rights of investors. As noted by La Porta, Lopez-de
Silanes, Shleifer and Vishny (1998), civil-law countries enforce loose legis-
lation because of the determinant role of ownership concentration
9
and
ownership structure determines the governance practices followed by their
companies. Wymeersch (2002) argues that compliance with the recommen-
dations of codes of good governance in European countries is more difficult
Corporate Characteristics, Governance Rules 301
than in the U.K. because of the greater presence of majority shareholders in
Europe – investors who do not wish to see their influence reduced by the
appointment of independent directors. Moreover, non-compliance with rec-
ommendations such as the separation of the functions of the CEO and the
chairman of the board may not be as negatively assessed in civil-law coun-
tries as it is in common-law countries, which have been the focus of more
studies. For example, the Spanish code of good governance recognizes that
the combining of CEO and chairman positions is a common practice in
Spain, because it is thought to sup port leadership both internally and ex-
ternally (Olivencia Committee, 1998).
Finally, our analysis of the determinants of voluntary disclosure focuses
on such general company characteristics as size, cross-listing status, and the
industry in which it operates. We employed structural equation modeling to
analyse both the direct and indirect effects of these characteristics on cor-

porate disclosure. In contrast with previous research, which focused on the
direct effect attributed to the balance between benefits and costs linked to
the provision of infor mation, we also recognize the effect that these char-
acteristics exert on the firms’ practi ces of good governance (Denis & Sarin,
1999; Doige, 2004), which in turn would affect corporate disclosure policy.
Corporate governance in Spain has relied heavily upon the role played by
majority shareholders who were usually involved in the management of the
company. Nevertheless, during the 1990s, free floating capital started to
represent a significant proportion of equity in some listed companies, giving
rise to greater concern about corporate governance and the protection of
investors’ interests. Hence, Spain provides a suitable environment in which
to test for the existence of interactions among corporate characteristics,
good governance rules, and corporate disclosure.
The remainder of the paper is structured as follows: Section 2 presents a
review of the literature and the development of hypotheses to be tested;
Section 3 provides a description of the data and methodology used in this
study; Section 4 contains the empirical results of the study; and in Section 5
we discuss the results and conclusions.
2. CORPORATE DISCLOSURE AND
FIRM CHARACTERISTICS
Agency costs arising from the separation of ownership and control in mod-
ern corporate forms may be reduced by strengthening the monitoring and
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ302
overseeing functions of the board of directors.

10
In large corporations, share-
holders are not involved in the management and control of the corporation,
but delegate such responsibilities to the board of directors to ensure goal
congruence between shareholders’ interests and management actions. The
corporate board’s role of overseeing is particularly relevant in protecting the
interests of powerless minority shareholders.
11
In fact, commissions charged
with the preparation of the so-called codes of good governance focused their
attention on the composition and functioning of the board of directors.
Practices such as the appointment of non-executive directors, the separation
of the functions of chairman of the board and CEO, or the creation of
specialized committees inside the board, such as the audit committee and the
compensation committee, were considered as essential mechanisms to im-
prove the monitoring of management.
12
They reduce the manager’s latitude
to act opportunistically and contribute to the alignment of the internal and
external interests of the organization (Denis, 2001, p. 195).
Independent non-executive directors. Codes of good governance include a
number of recommendations, one of them being the appointment of non-
executive directors, an inclusion designed to reduce agency conflicts between
managers and shareholders ( Gregory & Simmelkjaer, 2002, p. 53 ). Arguably,
boards controlled by management may result in practices of collusion, among
them the expropriation of shareholders’ wealth. Under such rationality, the
inclusion of outsiders on the board should reduce the likelihood of collusive
arrangements, inasmuch as non-executive directors are disciplined by the mar-
ket, which a ssesses and r ewards performance (Fama, 1980, pp. 2993–2294).
The extant literature provides empirical evidence supporting the role of

non-executive directors in promoting higher transparency and better dis-
closure policies. Chen and Jaggi (2000) found a positive association between
the proporti on of independent directors on the board and the extent of a
firm’s disclosure. The proportion of outside directors on corporate boards
was also negatively associated with indicators that measured the (poor)
quality of the information disclosed, such as the publication of fraudulent or
defective financial statements (Beasley, 1996; Peasnell, Pope, & Young,
2001), as well as measures of earnings management (Peasnell, Pope, &
Young, 2000). These results suggest that compan ies with a higher propor-
tion of independents on the board show a greater concern for disclosure.
Hence, we hypo thesize that:
H1a. voluntary disclosure is positively related to the proportion of in-
dependent directors on the board.
Corporate Characteristics, Governance Rules 303
Audit committee. The audit committee operates as a monitoring mech-
anism to improve the quality of information conveyed to external parties
(Pincus, Rusbarsky, & Wong, 1989) and oversees the preparation and com-
munication of financial information to third parties to ensure that such data
fulfils the requisites of clarity and the completeness of disclosure (Smith
Report, 2003, p. 12). Empirical evidence indicates that voluntary disclosure
is positively related to the functioning of an audit committee (Ho & Wong,
2001). Furthermore, Dechow, Sloan and Sweeney (1996) and Peasnell,
Pope, & Young (2001) observed that audit committees help to reduce the
likelihood of accounting fraud. These factors lead us to hypothesize:
H1b. voluntary disclosure is positively related to the existence of an audit
committee.
Chairman of the board and chief executive officer being the same person.
Separating the positions of CEO and chairman of the board arguably helps
to improve the monitoring function of the board. Jensen (1993) argues that
conflicts of interests and difficulties in performing the monitoring function

over management arise when the same individual holds both positions. This
dual-role situation is quite common in some European countries (U.K.,
France, Spain, and Italy), but it may require a balance. The Olivencia Code
of 1998 states, for example, that these dual roles may support leadership
within the organization and towards external parties, but recommends the
appointment of independent directors in order to balance the dominance of
the CEO/chairman of the board. A number of studies have identified the
combining of these two positions with poor disclosure practices. For ex-
ample, Forker (1992) found a significant negative relationship between the
combination of the two roles and the extent of disclosure. Furthermore, Ho
and Wong (2001) observed a negative relationship, albei t a non-significant
one, between corporate disclosure and the presence of a dominant person-
ality on the firm’s board. Therefore, we are led to hypothesize:
H1c. voluntary disclosure is positively related to the separation of the
functions of CEO and chairman.
Board participation in the capital of the company. Directors’ shareholdings
constitute a relevant vehicle for monitoring the management, as it tends to
restrain managerial incentives to divert resources that may ultimately jeop-
ardize the attainment of shareholder value maximization (Jensen &
Meckling, 1976). Furthermore, directors’ shareholdings help to align goals
and financial incentives of board members with those of outside sharehold-
ers (Bushman, Chen, Enge l, & Smith, 2004, p. 177). Shivdasani (1993)
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ304
observed that, relative to a control sample, outside directors in firms that are

targets of hostile takeovers have lower ownership stakes. Bhagat, Carey and
Elson (1999) found a positive correlation between the directors’ stock own-
ership and firm’s performance, as well as a positive correlation between di-
rector’s personal equity holdings and the likelihood of a disciplinary-type
CEO succession in poorly performing companies. On the other hand, the
examination of insider shareholdings has shown the existence of a non-linear
relationship between ownership and a firm’s value – as ownership increases,
firm value rises up to a point, and then decreases (McConnell & Servaes,
1990; Short & Keasey, 1999; Faccio & Lasfer, 1999). The authors of these
studies considered the non-linearity between a firm’s value and ownership to
be a consequence of the entrenchment effect associated with high levels of
managerial ownership. In cases of low levels of director ownership, therefore,
the monitoring role of the board is strengthened, which has a positive effect
on voluntary corporate disclosure. Hence, we hypothesize that:
H1d. voluntary disclosure is positively related to board participation in
the capital of the company.
Stock option plans as directors’ pay. Stock option plans serve the purpose
of compensating board members by aligning their interests with the firm’s
performance; increases in share prices lead to greater compensation for
board members (Jensen & Murphy, 1990; Tosi & Gomez-Mejia, 1989).
Gutie
´
rrez, Llore
´
ns and Arago
´
n (2000, p. 426) suggest that the linkage of
management compensation to performance results in a transfer of risk to
management and acts as a deterrent to opportunistic behaviour. In this
respect, empirical evidence shows that stock option plans for outside direc-

tors improves a firm’s value (Fich & Shivdasani, 2004) and increases the
monitoring role played by the board (Perry, 2000). Moreover, a number of
studies examines the relationship between stock options and disclosure
practices. Nagar, Nanda and Wysocki (2003, p. 287) argue that general
stock-priced-based incentives represent an effective means of encouraging
both good and bad news disclosures; stock price appreciation promotes the
release of good news, whereas withholding bad news may lead to litigation
costs and a decrease in stock price, because investors may interpret non-
disclosure as ‘‘worse’’ ne ws (Verrechia, 1983). In this respect, Miller and
Piotroski (2000) and Nagar, Nanda and Wysocki (2003) report a positive
association between corporate disclosure and the proportion of CEO com-
pensation affected by stock price. Although these studies refer to the com-
pensation of managers rather than the compensation of directors, we expect
Corporate Characteristics, Governance Rules 305
to observe the same type of incentives to disclosure when directors benefit
from stock option plans. This leads us to hypothesize:
H1e. voluntary disclosure is positively related to the establishment of
stock option plans as directors’ remuneration.
Board size. In addition to the foregoing mechanisms for aligning man-
agement and shareholder interests, codes of good governance usually rec-
ommend limitations to the size of a board.
13
By restricting the number of
directors, it is believed that the exchange of ideas between board members
will be enhanced, as well as flexibility in the decision-making process. Jensen
(1993, p. 865) argues that small boards are more effective in monitoring the
CEO and are tougher for the CEO or the chairman to manipulate. In a
similar vein, Yermack (1996) has shown that firms with smaller boards are
valued more highly by the market than are their counterparts with larger
boards; wher eas Vafeas (2000) has observed that investors place higher

value on earnings information when provided by firms with smaller boards.
Investigations have also examined the quality of disclosure and its relation-
ship with board size. For example, Peasnell, Pope and Young (2001) found a
tendency, albeit statistically non-significant, for mean board size to be
higher for firms reporting defective financial statements than for those in-
cluded in their control sample. The strength of these arguments, however,
leads us to hypothesize:
H1f. voluntary disclosure is negatively related to the size of the board.
The adoption of any one of the above-mentioned practices of good gov-
ernance does not exclude the assumption of others. In contrast, the effect of
some practices may be strengthened when other rules are observed. As an
example, Peasnell, Pope and Young (2000) observed that the role played by
independent directors was enhanced by the functioning of a boa rd audit
committee. Moreover, non-compliance with some rules of good governance
may be partially offset by the adoption of others. Gul and Leung (2004)
have shown that the presence of highly experienced non-executive directors
on the board tends to moderate the negative effect of combining the po-
sitions of CEO and chairman. As a consequence, we expect that the greater
the degree of compliance with codes of good governance, the greater the
improvement in corporate disclosure. Hence, we formulate the following
general hypothesis:
H1. voluntary disclosure is positively related to the adoption of rules of
good governance.
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ306

The ownership structure of a firm may be a possible determinant of or-
ganizational disclosure (Raffournier, 1995). At one extreme, high levels of
concentration of capital may be accompanied by the owner’s considerable
involvement in the firm’s management, which, in turn, may lead to unre-
stricted access to information. Under these circumstances, the demand for
information would be very low, or even absent, particularly if the manager
owns all the firm’s shares. On the other hand, in cases of dispersion of
ownership, investors lack first-hand access to information, and this may lead
to increased demands for organizational information that can be used to
monitor management (Gelb, 2000, p. 169). In this respect, McKinnon and
Dalimunthe (1993, p. 37) suggest that voluntary disclosure may be helpful in
reducing conflicts between managers and shareholders that arise when a
firm’s shares are widely held. Furthermore, ownership dispersion may in-
fluence the supply of information. For example, Craswell and Taylor (1992,
p. 299) argue that increases in the separation of ownership and control are
likely to be accompanied by additional disclosures of information to third
parties. By ov ercoming owners’ perceived asymmetry of information, man-
agement expects to reduce the discount implicit in its compensation pack-
age. Additionally, the ownership structure may have a significant impact on
the adoption of rules of good governance which, in turn, will affect cor-
porate disclosure. As suggested by Wymeersch (2002), compliance with the
recommendations of codes of good governance is more difficult when a
significant proportion of a firm’s equity is held by a majority shareholder.
Therefore, we are led to hypothesize:
H2. voluntary disclosure is negatively related to the level of ownership
concentration.
Corporate size was commonly used as an explanatory factor for voluntary
disclosure. Ball and Foster’s (1982) review shows that firm size has been
regarded as an adequate proxy for the costs and benefits linked to the
provision of information. The cost of gathering and preparing detailed in-

formation an d the risk of creating a competitive disadvantage through dis-
closure will be lower for larger companies that prepare the information for
internal use and invest large amounts of resources in fixed assets and in-
novation processes. This large company advantage, it is argued, acts not
only as an entry barrier towards smaller companies (Depoers, 2000, p. 251),
but it also enables larger firms to benefit from the salutary effects of better
disclosure because the provision of information facilitates access to capital
markets (Singhvi & Desai, 1971, p. 131). Additionally, corporate disclosure
leads to the allaying of public criticism or government intervention (Watts &
Corporate Characteristics, Governance Rules 307
Zimmerman, 1986). Therefore, the balance between benefits and costs linked
to the provision of information is expected to be more favourable for large
than for small companies.
Furthermore, codes of good governance may also lead large companies to
provide more voluntary information than their smaller counterparts do.
Research has shown the existence of a positive association between a firm’s
size and the adoption of some practices of good governance, such as the
appointment of independent directors or the separation of the functions of
CEO and chairman (Denis & Sarin, 1999 ; Dehaene, De Vuyst, & Ooghe,
2001). Thus, the size of the company may exert an indirect influence on
disclosure. Therefore, we hypothesize:
H3. voluntary disclosure is positively related to the size of the company.
A number of studies regard listing status as a determinant of disclosure
variability (Firth, 1979; Cooke, 1991; Meek, Roberts & Gra y, 1995). Listed
companies must comply with stock market regulations that require far more
information disclosure than applies to unlisted companies. Additionally, listed
firms voluntarily disclose information in order to garner investors’ trust and to
obtain better financing conditions (Raffournier, 1995, p. 263). Empirical stud-
ies have shown that fi rms that more readily practice voluntary information
disclosure enjoy such beneficial effects a s increased stock p rices (Healy, Hutto n,

& Palepu, 19 99; Lang & Lundholm, 2000), higher a nalyst following (Lang
& L undholm, 1996), improvements in stock liquidity (Welker, 1995), and a
reduction in t he cost of cap ital (Botosan, 1997; Botosan & Plumlee, 2 002 ).
Our study focuses on listed companies, but we attempt to examine dif-
ferences in disclosure between firms that are only traded domestically and
firms that are both traded domestically and cross-listed internationally.
Meek, Roberts and Gray (1995) suggest that listed companies face addi-
tional capital market pressures for the provision of information. Further-
more, such pressures will arguably increase with the efficiency of the
markets in which the firms are traded, as may be the c ase for firms trading in
foreign stock markets as opposed to those trading solely on the Madrid
Stock Exchange.
14
Therefore, once a firm discloses information volunta rily
to foreign stock markets, it would incur only a marginal cost increase if it
also reported such information in its domestic market. Empirical research
has shown evidence of greater information disclosure for companies that are
listed both domestically and in foreign stock exchanges (Cooke, 1991). In a
similar vein, Khanna, Palepu, and Srinivasan (2004) found a positive as-
sociation between disclosure and a U.S. listing for a sample of European
and Asian-Pacific companies.
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ308
The positive association between cross-listing and disclosure can be ex-
plained not only by the trade off between benefits and costs, but also by the

improvement in corporate governance practices that follow cross-listing.
Wo
´
jcik, Clark, and Bauer (2004) found, for instance, that European com-
panies that are cross-listed on U.S. exchanges have higher corporate gov-
ernance ratings than firms without such U.S. cross- listing. For non-U.S.
firms, Doidge (2004) observed that cross- listing on U.S. exchanges enhances
the protection afforded to minority investors and reduces the private ben-
efits of control. These arguments and evidence lead us to test the following
hypothesis:
H4. voluntary discl osure is positively related to listing on foreign stock
exchanges.
Industry sector may explain disclosure differences among companies.
For example, financial entities may be encouraged to provide voluntary
information in order to garner client trust. In a similar vein, firms operating
in regulated markets are exposed to thorough scrutiny from stakeholders,
who may advise them to provide external parties with additional informa-
tion about their activities and outcomes. Finally, the strategic importance of
some industries (e.g. oil) may attract disproportionate surveillance measures
by governmental agencies and interest groups, and this surveillance
may lead these companies to favour high levels of reporting to third
parties (Watts & Zimmerman, 1986, p. 239). Furthermore, disclosure of
organizational information involves hidden costs, such as those related to
revealing crucial informat ion to competitors, which, in turn, may lead to
competitive disadvantages. Research in this area demonstrates that the
effects of such disclosure vary across industries (Meek, Roberts &
Gray, 1995).
Additionally, operating in a regulated industry might have an indirect
effect on disclosure through its influence on the governance rules of firms.
Companies operating in regulated industries have greater visibility and may

be more easily encouraged to adopt practices of good governance than
would firms operating in unregulated sectors. In fact, the report of the
governing body of the Madrid Stock Exchange – the Comisio
´
n Nacional del
Mercado de Valores – on compliance with the Good Governance Code
(CNMV, 2000), shows that the highest levels of compliance correspond to
companies operating in regulated sectors. Thus, we hypothesize that:
H5. voluntary disclosure is positively related to operati ng in regulated
industries.
Corporate Characteristics, Governance Rules 309
3. DATA AND METHODOLOGY
3.1. Sample
Our sample consists of 117 firms that were indexed in the Actualidad
Econo
´
mica
15
Index in 1999,
16
all of which operate in the continuous (elec-
tronic) market of the Madrid Stock Exchange. We obtained information on
corporate characteristics and governance practices from the CNMV. For
firms that did not submit this informat ion to the CNMV,
17
we requested
their data directly. Overall, we collected data on 91 of the 117 firms listed on
the Actualidad Eco
´
nomica Index. These companies constituted the sample

for our study.
As we missed data from some firms included in the Actualidad Econo
´
mica
Index, we tested for sample bias. Argua bly, firms that did not submit their
data to the CNMV could be reluctant to disclose voluntary information.
Comparison of mean indexes through the Student t-test for companies in-
cluded in our study and those omitted from it revealed significant differ-
ences (p ¼ 0:06). The lowest mean corresponded to firms excluded from our
study (Table 1), which led us to control for firm size, measured by the
natural logarithm of total assets. The results indicate that firms in our study
are significantly larger than those from which we did not gather data
(po0.001), which in turn indicates that our findings cannot be extrapolated
to small firms. Arguably, there seems to be less concern about the provision
of voluntary information to external parties in small firms than in their
larger counterparts.
3.2. Operationalization of Variables
3.2.1. Disclosure Index
Actualidad Econo
´
mica prepares a disclosure index that consists of 18 indi-
cators. The index, however, embraces issues that have little to do with
Table 1. Selection Bias.
Variable Companies N Mean S.D. t Sig. (bilateral)
INDEX Excluded 26 42.385 14.458 –1.882 0.062
Analysed 91 48.242 13.861
LASSETS Excluded 26 4364 2629 –5.072 0.000
Analysed 91 6806 2018
M. ROSARIO BABI
´

O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ310
voluntary disclosure: auditor’s opinion (Indicator #2), clarity in the pres-
entation of data (Indicator #9), design of the report (Indicator #10), and use
of additional venues to make the annual report visible to the public opinion
(Indicator #18). To measure the amount of information provided by the
company, we set these indicators aside.
18
Although some of the items refer
to compulsory information (i.e. information on shareholders, subsidiaries,
remuneration to board members, and analysis of operations), the degree of
detail captured by the Actualidad Econo
´
mica Index is much greater than that
required by the Spanish legislation. Therefore, the resulting index consti-
tutes a measure of voluntary disclosure.
3.2.2. General Corporate Characteristics
Company size was measured in this study by total assets (ASSETS), as has
been done in other studies on voluntary disclosure (Depoers, 2000; Ho &
Wong, 2001). We used dichotomous variables to account for the firm’s
cross-listing status (LISTING) (Khanna, Palepu & Srinivasan, 2004) and
industry (INDUSTRY) (Eng & Mak, 2003), giving variables a value of one
(1) if the firm was listed on foreign stock exchanges or operated in a reg-
ulated industry, and zero (0) otherwise. The Herfindahl index was used as a
measure of stock dispersion (OWNERSHIP). Following Santerre and Neun
(1986), our use of the Herfindahl index relied on the following formula:
OWNERSHIP ¼

X
n
i¼1
ðP
i
=0:51Þ
2
,
where P
i
is the percentage of the firm’s stock that is held by the ith largest
stockholder so that the sum of the P
i
is equal to 51% (majority control). The
index takes the value of one (1) when a single shareholder owns the majority
of the shares of the company and it approaches zero (0) when corporate
stock is widely dispersed – that is, when there is a large number of share-
holders, each owning a small proportion of the firm’s stock.
3.2.3. Governance Rules
We used the definition of independent directors included in the Olivencia
Code (1998, par.2.1) – those who are not related to the management team or
to the hard core of shareholders – measured through the ratio of independ-
ent directors to total number of directors (Peasnell, Pope & Young, 2001;
Eng & Mak, 2003). We used dummy variables to account for the existence
of an audit committee (AUDITCOM) (Ho & Wong, 2001), the separation
of the functions of chairman of the board and the CEO or president
Corporate Characteristics, Governance Rules 311
(CHAIRM) (Beasley, 1996), and the establishment of stock option plans as
a means of directors’ remuneration (OPTION) (Fich & Shivdasani, 2004).
These variables assumed a value of one (1) if the good governance practice

was adopted, and a zero (0) otherwise.
Equity held by board members is usually measured through the total
number of shares owned by directors, scaled by the total number of out-
standing shares (O’ Sullivan, 2000; Coles, McWilliams, & Sen, 2001). Nev-
ertheless, in order to measure directors’ ownership that would help to
strengthen the monitoring function of the board, we departed from this
common measure by using a dummy variable (BEQUITY), which assumed
a value of one (1) if the directors’ participat ion in the equity exists but is
lower than 3%,
19
and zero (0) otherwise. In this manner, we measured for
the directors’ participation in the firm’s capital in situations in which they do
not have control of the firm or a significant influence on its developments.
20
Finally, the total number of directors is usually employed as a measure of
board size (Vafeas, 2000; Peasnell, Pope & Young, 2001). Nevertheless, due
to the extremely small board size of some Spanish companies, we used a
dummy variable (SIZEB), with a value of one (1) if the size of the board fits
within the recommendations of the Olivencia Code (i.e. the number of board
members should range from 5 to 15), and zero (0) otherwise.
3.3. Methods
We employed univariate and multivariate techniques for data analysis in
this study, using a one-factor ANOVA as well as the Kruskal–Wallis non-
parametric test to examine the association between each of the independent
variables and the disclosure index. We then applied a technique of struc-
tural-equations modelling, path analysis, to test simultaneously for existing
relationships among the variables included in our study (Hoskisson,
Johnson & Moesel, 1994; Bisbe & Otley, 2004). Structural equations, which
have been widely employed in areas such as marketing or psychology, are
increasingly being used in accounting studies (Hunton, Wier, & Stone, 2000;

Baines & Langfield-Smith, 2003; Widener, 2004). In this study, structural
equations are particularly suitable for at least two reasons: (1) A model of
structural equations allows for the simultaneous an alysis of a series of mul-
tiple regression equations and is particularly useful when the dependent
variable in one equation becomes an independent variable in subsequent
ones. (2) Path analysis allows for confirmatory factor analys is, facilitating
the introduction of non-observed concepts (latent constructs), accounting
for the measurement error in the estimation process.
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ312
4. EMPIRICAL RESULTS
Table 2 shows the variables included in our study as well as the descriptive
statistics.
Descriptive statistics show that companies in our study are widely dis-
tributed regarding the provision of voluntary infor mation. There are also
large differences in corporate size, measured by total assets, ranging from 30
to 287,155 milli on Euros, and we employed the natural logarithm to account
for this difference. The Herfindahl index of stock dispersion also exhibits
substantial cross-sectional variation, ranging from zero (0) in cases of widely
dispersed companies, to one (1) for firms in which a single shareholder owns
the majority of equity. We observed a value of 1 for 26 of the 91 companies
included in our sample, which in turn indicates a high level of ownership
concentration in firms listed on the Madrid Stock Exchange. Descriptive
statistics also show that compliance with recommendations of the Olivenci a
Code for appointing an audit committee and for limiting the size of the

board are common practices in firms included in our sample (i.e. 75% have
an audit committee, and the total number of directors in 76% of the firms is
between 5 and 15). Finally, we observed high variability in the proportion of
independent directors on the board. The mean value for this variable is
36%, but 11 companies in our study did not have any independent directors.
We conducted a series of one-way ANOVAs
21
using the index as the de-
pendent variable and each of the governance rules considered in our study as
afactor.
22
The Kruskal–Wallis test and ANOVA results (Table 3)showthat
the mean disclosure index is significantly higher (po0.01) for companies with
a higher proportion of independent directors on the board, providing support
for Hypothesis 1a. The disclosure index is also significantly higher (po0.01)
for firms that have appointed an audit committee than for those in which such
committee was not formed. These results support Hypothesis 1b. We found
significant differences in disclosure (po0.05) between companies in which
directors have a participation in their capital lower than 3% relative to their
counterparts with no director participation or with director participation that
is higher than 3%. These results provide support for Hypothesis 1d. Table 3
also shows that the mean disclosure index is significantly higher (po0.01) for
those companies that have established a stock option plan as a means of
director remuneration, thereby providing support for Hypothesis 1e.
Comparisons of mean disclosure indexes are non-significant between
firms that separate the functions of CEO and chairman and their counter-
parts that combine both posts. Furthermore, the relationship between
the index and the independent variable is opposite to the direction
Corporate Characteristics, Governance Rules 313
Table 2. Descriptive Statistics.

N Mean S.D. Minimum Q1 Median Q3 Maximum Skewness Kurtosis
INDEX 91 48.242 13.861 13 39 47 58 82 –0.026 –0.025
ASSETS 91 9436 35338 30 224 607 3589 287155 6.421 45.974
LASSETS 91 6.806 2.018 3.40 5.41 6.41 8.186 12.568 0.640 0.014
OWNERSHIP 91 0.494 0.396 0 0.104 0.417 1 1 0.200 –1.680
LOWNERSHIP 91 –1.319 1.385 –5.81 –2.263 –0.875 0 0 –0.962 0.303
LISTING 91 0.308
INDUSTRY 91 0.264
OUTSIDE 91 35.758 23.464 0 18.18 33.33 50 100 0.369 –0.293
BEQUITY 91 0.538
AUDITCOM 91 0.747
OPTION 91 0.297
CHAIRMAN 91 0.593
SIZEBOARD 91 0.758
INDEX Company’s aggregate score.
ASSETS Total assets (millions of Euros)
LASSETS Natural logarithm of total assets
OWNERSHIP Herfindahl index of stock dispersion
LOWNERSHIP Natural logarithm of the Herfindahl index of stock dispersion
LISTING Binary variable: 1 if the company is cross-listed, 0 otherwise
INDUSTRY Binary variable: 1 if regulated industry, 0 otherwise
OUTSIDE Proportion of independent directors on the board
BEQUITY Binary variable: 1 if the board participates in the equity of the company and its participation is lower than 3%, 0 otherwise
AUDITCOM Binary variable: 1 if audit committee exists, 0 otherwise
OPTION Binary variable: 1 if stock option plan exists, 0 otherwise
CHAIRM Binary variable: 1 if the chairman and the CEO are not the same person, 0 otherwise
SIZEBOARD Binary variable: 1 if 5 p number of the members on the board p 15, 0 otherwise
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN

˜
OVA
´
ZQUEZ314
expected – that is, the highest mean index corresponds to firms in which
both functions are combined. Therefore, Hypothesis 1c is not supported.
Finally, there were no significant differences in disclosure between com-
panies that complied versus those that did not comply with the Olivencia
Code recommendations regarding the size of the board. Results contradict
our prediction because the highest mean corresponds to firms with larger
boards Therefore, Hypothesis 1f is not supported.
We conducted the same type of analysis using each one of the general
corporate characteristics analysed as a factor, recoding the continuous var-
iables (i.e. total assets and ownership concentration) into discrete variables
by using percentiles 33.33 and 66.67%.
23
Results of the ANOVA and the
Kruskal–Wallis tests are shown in Table 4.
Results show the existence of significant differences in disclosure as a
consequence of the ownership structure of the firm. Hypothesis 2 stated that
voluntary disclosure is negatively related to the level of ownership concen-
tration. Our findings show that the highest mean disclosure index corre-
sponds to Group 1 – that is, to firms with widely dispersed ownership
(Table 4). Nevertheless, the lowest mean disclosure index does not corre-
spond to Group 3 (those entities with the highest level of ownership con-
centration), but to Group 2. The Bonferroni test shows that the mean
disclosure index for Group 1 is significantly higher (po0.05) than that cor-
responding to Group 2.
Table 3. Relationship Between the Disclosure Index and Corporate
Governance Rules.

Group N Mean S.D. F Sig. w2 Sig.
Dependent variable: INDEX
INDEP 0 44 43.386 12.787 11.695 0.001 9.505 0.002
1 47 52.787 13.395
AUDITCOM 0 23 39.087 11.774 15.604 0.000 13.190 0.000
1 68 51.338 13.194
CHAIRMAN 0 37 50.865 14.174 2.264 0.136 2.549 0.110
1 54 46.444 13.48
OPTION 0 64 45.281 13.273 10.924 0.001 9.469 0.002
1 27 55.259 12.865
BEQUITY 0 42 44.62 14.368 5.600 0.020 5.921 0.015
1 49 51.35 12.752
SIZEBOARD 0 22 49.909 13.55 0.417 0.520 0.131 0.718
1 69 47.71 14.015
Corporate Characteristics, Governance Rules 315
We also found significant differences in disclosure between large and
small firms. The highest mean corresponds to the largest companies in our
study (Group 3). The post-hoc Bonferroni test revealed significant differ-
ences between Groups 1 and 3, after performing pairwise comparisons.
Thus, we found significant differences between the largest a nd the smallest
companies in our sample, providing support for Hypo thesis 3. The mean
disclosure index is also higher for cross-listed entities than for firms listed
solely on the Madrid Stock Exchange. Both the ANOVA and the non-
parametric tests show that this difference is highly significant (po0.001),
providing support for Hypothesis 4. Finally, both the ANOVA results and
the non-parametric test show that the mean disclosure index is significan tly
higher (po0.01) for companies operating in regulated sectors than for those
in unregulated industries, providing support for Hypothesis 5.
In addition to the univariate analyses, we deployed path analysis to test
simultaneously for the relationships among all variables included in our

study. Path analysis requires the design of a path graphic representing all the
relationships among variables that are expected by the researcher. Based on
our hypotheses and on the results of the univaria te analysis, we plotted the
model shown in Fig. 1, showing a positive association between the disclo-
sure index and the size of the company (LASSETS), its listing status (LIST-
ING), the industry in which it operates (INDUSTR Y), and its ownership
structure (OWNERSHIP). We also found that the disclosure index was
significantly associated with the adoption of some practices of good
governance. Because of the high correlation coefficients observed among
these variables (Table 5), we have included a non-observed variable in our
Table 4. Relationship between the Disclosure Index and the
Characteristics of the Firm.
Group N Mean S.D. F Sig. w2 Sig.
Dependent variable: INDEX
LOWNERSHIP 1 30 53.530 15.213 3.717 0.028 6.529 0.038
2 31 44.420 14.773
3 30 46.900 9.625
LASSETS 1 30 42.733 14.893 5.863 0.004 10.109 0.006
2 31 47.677 12.448
3 30 54.333 12.004
LISTING 0 63 44.937 13.464 13.223 0.000 12.110 0.001
1 28 55.679 11.889
INDUSTRY 0 67 45.925 13.682 7.616 0.007 7.211 0.007
1 24 54.708 12.467
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´

ZQUEZ316
analysis (GOVERNANCE), aimed at representing the governance practices
followed by the company. We have also included relationships in the path
diagram in order to measure the influence of the ownership structure, listing
status, industry, and size of the company on the governance factor. Finally,
we have added correlations between the independent variables: LASSETS,
LISTING, INDUSTRY, and OWNERSHIP.
The model was estimated (maximum likelihood estimation) by using
AMOS 4.0 through the SPSS 10.0 and a bootstrapping technique run to
avoid problems derived from non-fulfilment of the normality conditions of
the variables. The estimates were non-significant for the following regres-
sions
24
:
GOVERNANCE - CHAIRMAN
GOVERNANCE - SIZEB
INDUSTRY - GOVERNANCE
INDUSTRY - INDEX
LASSETS - GOVERNANCE
LISTING - INDEX
LOWNERSHIP - INDEX
LOWNERSHIP 2 LISTING
LOWNERSHIP 2 LASSETS
GOVERNANCE
OUTSIDE
e
1
1
1
AUDITCOM

e
2
1
OPTION
e
3
1
BEQUITY
e
4
1
CHAIRMAN
e
5
1
SIZEB
e
6
1
LISTING
INDUSTRY
LASSETS
INDEX
z
1
1
z
2
1
LOWNERSHIP

Fig. 1. PATH analysis (I)
Corporate Characteristics, Governance Rules 317
Table 5. Correlation Matrix (Spearman Correlation Coefficients).
INDEX OUTSIDE BEQUITY AUDITCOM OPTION CHAIRMAN SIZEBOARD LASSETS LOWNERSHIP LISTING INDUSTRY
INDEX 1.000 0.306
ÃÃ
0.256
Ã
0.383
ÃÃ
0.324
ÃÃ
–0.168 –0.038 0.395
ÃÃ
–0.217
Ã
0.367
ÃÃ
0.283
ÃÃ
OUTSIDE 0.306
ÃÃ
1.000 0.023 0.353
ÃÃ
0.263
Ã
–0.088 –0.082 0.078 –0.350
ÃÃ
0.130 0.058
BEQUITY 0.256

Ã
0.023 1.000 0.172 0.215
Ã
0.086 –0.111 0.343
ÃÃ
0.14 0.283
ÃÃ
0.204
AUDITCOM 0.383
ÃÃ
0.353
ÃÃ
0.172 1.000 0.322
ÃÃ
–0.018 0.085 0.167 –0.144 0.223
Ã
0.061
OPTION 0.324
ÃÃ
0.263
ÃÃ
0.215
Ã
0.322
ÃÃ
1.000 –0.099 –0.139 0.253
Ã
–0.182 0.401
ÃÃ
0.048

CHAIRMAN –0.168 –0.088 0.086 –0.18 –0.099 1.000 –0.049 0.032 0.195 –0.175 0.140
SIZEBOARD –0.038 –0.082 –0.111 0.085 –0.139 –0.049 1.000 –0.340
ÃÃ
0.019 –0.346
ÃÃ
–0.361
ÃÃ
LASSETS 0.395
ÃÃ
0.078 0.343
ÃÃ
0.167 0.253
Ã
0.032 –0.340
ÃÃ
1.000 –0.035 0.450
ÃÃ
0.546
ÃÃ
LOWNERSHIP –0.217
Ã
-0.350
ÃÃ
0.014 0.144 –0.182 0.195 0.019 –0.035 1.000 –0.115 –0.175
LISTING 0.367
ÃÃ
0.130 0.283
ÃÃ
0.223
Ã

0.401
ÃÃ
–0.175 –0.346
ÃÃ
0.450
ÃÃ
–0.115 1.000 0.303
ÃÃ
INDUSTRY 0.283
ÃÃ
0.058 0.204 0.061 0.048 0.140 –0.361
ÃÃ
0.546
ÃÃ
–0.175 0.303
ÃÃ
1.000
Ã
Significant at 0.01 level.
ÃÃ
Significant at 0.05 level.
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ318
Drawing on these results, we developed a model, as shown in Fig. 2.
The Goodness of Fit test (Table 6) shows that the model is fitted to the

data.
25
The level of significance of the w
2
test of overall model fit
(w
2
¼ 22:945; df ¼ 19; p ¼ 0:240) suggests the existence of an acceptable
fit between the estimated and the actual data matrix.
26
The sensitivity of the
w
2
test to sample size led us to complement it with other descriptive measures
of global fit: the Goodness of Fit Index (GFI) and the Root Mean Square
Error of Approximation (RMSEA). Two measures of incremental fit were
also examined: the Adjusted Goodness of Fit Index (AGFI) and the Tuck-
er–Lewis Index (TLI). Additionally, we analysed two measures of parsimo-
ny: the Comparative Fit Index (CFI) and the Akaike Information Criterion
GOVERNANCE
1
OUTSIDE
e
1
1
AUDITCOM
e
2
1
OPTION

e
3
1
BEQUITY
e
4
1
LISTING
LASSETS
INDEX
z
1
1
z
2
1
LOWNERSHIP
Fig. 2. PATH analysis (II)
Table 6. Goodness of Fit Tests.
w
2
independence test
22.945 (df ¼ 19; p ¼ 0:240)
Goodness of fit test (GFI) 0.942
Root mean square error of approximation (RMSEA) 0.048
Adjusted goodness of fit index (AGFI) 0.891
Tucker–Lewis index (TLI) 0.944
Comparative fit index (CFI) 0.962
Akaike information criterion (AIC) 56.945 (saturated model 72.000)
Corporate Characteristics, Governance Rules 319

(AIC). Values observed for these indexes indicate that the model fits to
the data.
27
Table 7 shows the estimated regressions. Standardized regression weights
and standard errors are shown in Table 8 .
The governance factor obtained from the model exerted some influence
on corporate disclosure. As shown in Fig. 2, however, not all governance
factors were included in our model – for example, the size of the board and
the combination of the functions of chairman and CEO were not included.
These results conform to those obtained from the univariate analysis and,
hence, Hypotheses 1c and 1f must be rejected. The other four practices of
good governance remain in the model and have significant coefficients
(Table 8), a result that could be inferred from the ANOVA and from
examination of the correlation matrix. We reduced these practices to
a governance factor that was positively associated with disclosure, and sug-
gest that the appointment of independent directors, the formation of an
audit committee, the participation of the board in the equity of the com-
pany, and the establishment of stock option plans as a means of directors’
Table 7. Estimated Equations.
OUTSIDE ¼ b
1
GOVERNANCE+e
1
AUDITCOM ¼ b
2
GOVERNANCE+e
2
BEQUITY ¼ b
3
GOVERNANCE+e

2
OPTION ¼ b
4
GOVERNANCE+e
3
GOVERNANCE ¼ b
5
LISTING+ b
6
OWNERSHIP+Z
2
INDEX ¼ b
7
LASSETS+b
8
GOVERNANCE+Z
1
Table 8. Standardized (Beta) Weights.
Standard
Regression
Weights
S.E. Critical
Ratios
Regression: LASSETS - INDEX 0.285 0.095 3.000
Regression: GOVERNANCE - INDEX 0.481 0.133 3.616
Regression: LOWNERSHIP - GOVERNANCE –0.364 0.140 2.600
Regression: LISTING - GOVERNANCE 0.453 0.144 3.146
Correlation: LISTING 2 ASSETS 0.482 0.089 5.415
Factorial: GOVERNANCE - OUTSIDE 0.487 0.150 3.246
Factorial: GOVERNANCE - AUDITCOM 0.541 0.100 5.410

Factorial: GOVERNANCE - BEQUITY 0.297 0.148 2.006
Factorial: GOVERNANCE - OPTION 0.600 0.107 5.607
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN
˜
OVA
´
ZQUEZ320
remuneration are positively related to the provision of voluntary informa-
tion. These findings endorse the findings shown above for Hypotheses 1a,
1b, 1d and 1e. Therefore our general Hypothesis 1, stating that voluntary
disclosure is positively related to the adoption of good governance practices
is partially supported.
Path analysis shows that company size is a significant variable in ex-
plaining corporate disclosure variability (po0.01), a result that could be
inferred from the ANOVA. The sign of the coefficient indicates that, as
hypothesized, a positive association exists between corporate size and dis-
closure, thereby adding support for Hypothesis 3.
Results also show that both listing status and ownership concentration
exert a significan t influence on the governance factor, accounting for 38% of
its variability (Table 9). Considering that the governance factor influences
the disclosure policy of firms, listing status and ownership structure have an
effect on disclosure, although this impact is mediated by the governance
practices followed by the company. This finding provides supports for the
results provided by the univariate analysis for Hypotheses 3 and 4. Finally,
despite the results of the univariate analysis, the industry variable was non-
significant in explaining disclosure variability and was therefore eliminated
from the model. Therefore, Hypothesis 5 must be rejected.
Square multiple correlation coefficients (Table 9) show that, taken to-

gether, the size of the entity and the factor representing its governance rules
account for a significant proportion of the variability in the disclosure index
(39.4%). Results are similar using the unadjusted Actualidad Econo
´
mica
Index, and indicate that the size of the company, along with its governance
structure, are important factors in explaining not only the amount of vol-
untary information provided but also the disclosure strategy of the co mpany
regarding the design and presentation of reports.
Table 9. Squared Multiple Correlations.
Variable Squared Multiple Correlations
INDEX 39.4
GOVERNANCE 37.8
OUTSIDE 25.9
AUDITCOM 30.3
BEQUITY 11.0
OPTION 37.1
Corporate Characteristics, Governance Rules 321
5. DISCUSSION AND CONCLUSIONS
The results of this study indicate that corporate decisions regarding the
provision of voluntary information are complex processes affected by a
number of interrela ted factors: the governance rules followed by the firm,
corporate size, cross-listing status, and the ownership structure of the firm.
Our results show that the adoption of a number of good governance prac-
tices such as the appointment of independent directors, the formation of
audit committees, participation of the board in the capital of the company,
and establishment of stock option plans as a means of director remuneration
exert a significant influence on corporat e voluntary disclosure. As expected,
independent directors and audit committees strengthen the monitoring
function of the board, so that firms become more responsive to stakehold-

ers’ demands for information. As for directors’ participation in the capital
of the company and stock option plans, our results suggest that both
mechanisms contribute to the alignmen t of managers’ and shareholders’
interests, inasmuch as they reduce managem ent reluctance to disclose vol-
untary information.
Contrary to expectations, neither the separation of the functions of CEO
and chairman, nor compliance with the recommendation of the Olivencia
Code (1998) regarding the size of the board is significantly associated with the
provision of voluntary information. Two corporate features stated by the
Olivencia Code may explain this finding. First, a high ownership concentra-
tion in Spanish companies, along with a low separation between ownership
and management, especially in family-owned firms, has led to the combined
CEO-chairman role in a number of firms. Second, the appointment of outside
directors who do not act as substitutes for existing directors, but fill additional
positions is a possible reason for the large boards in our study.
Our findings clarify prior inconclusive evidence regarding the effect of the
adoption of practices such as the appointment of outsi de directors or the
formation of an audit committee on the provision of voluntary information.
Empirical evidence supporting this study was gathered in Spain, a country
where corporate compliance with good governance rules is much lower
than that observed in common-law countries (Deminor Rating, 2003).
However, our findings reveal that practices of good governance have a
significant influence on voluntary disclosure. Furthermore, our results
indicate that, as consequence of complementarities and substitutability
between practices of good governance, their beneficial effects on disclosure
cannot be fully assessed unless we examine the global set of practices
followed by the entity.
M. ROSARIO BABI
´
O ARCAY AND M. FLORA MUIN

˜
OVA
´
ZQUEZ322
Our findings reveal that the ownership structure of firms and cross-listings
have a significant influence on governance practices, and thereby exert an
indirect effect on corporate disclosure. As expected, ownership concentra-
tion is negatively associated with the adoption of practices of good gov-
ernance, suggesting that firms with a majority shareholder do not achieve
the same levels of compliance with recommendations of the good govern-
ance code as do their widely dispersed counterparts. Therefore, the positive
association between corporate ownership structure and disclosure is ex-
plained by the influence exercised by the dispersion of ownership on the
adoption of rules of good governance that, in the end, strengthens the dis-
closure policy of the company.
Cross-listing in foreign stock markets also exerts a significant influence on
the adoption of practices of good governance. Spanish companies listed on
foreign stock markets seem to be willing to achieve a system of corporate
governance in accordance with usual standards in use in foreign stock ex-
changes. Moreover, the necessity of attracting investors’ confidence and of
avoiding any mistrust relative to the protection of their interests may explain
the eagerness of cross-listed companies to adopt recommendations aimed at
protecting shareholders’ interests. Our results extend those obtained in pre-
vious studies (Cooke, 1991; Khanna, Palepu & Srinivasan, 2004), and sug-
gest that it is not the cost-benefit trade-off that explains the positive
association between cross-listing and disclosure, but the improvement in
corporate governance that is realized after cross-listing occurs.
Additionally, our findings reveal that corporate size is a significant de-
terminant of corporate disclosure. However, and contrary to our prediction,
corporate size does not have a significant impact on the adoption of rules of

good governance. These resul ts are consistent with those obtained in pre-
vious resear ch (Ahmed & Courtis, 1999) and suggest that large companies,
as opposed to small entities, face a more favourable balance between costs
and benefits when disclosing information.
In contrast with the results we obtained for our univariate analysis (i.e.
significant differences were observed between companies operating in reg-
ulated and unregulated sectors), path analysis shows that operating in a
regulated industry does not have a significant effect on the provision of
voluntary information. A high correlation between this variable and com-
pany size may explain its absence from the final model (Table 5). In the
Spanish economy, companies operating in regulated industries such as fi-
nance, gas, power utility, oil, and telecommunications are the largest firms in
this country. Therefore, the indust ry variable may not be adding significant
information to that contained in the firm’s size.
Corporate Characteristics, Governance Rules 323