Int. Journal of Business Science and Applied Management, Volume 10, Issue 2, 2015

Listed firm’s level of stakeholder transparency - The comply or

explain evidence from the Danish corporate governance code

Caspar Rose

Copenhagen Business School,

Department of International Economics and Management

Porcelænshaven 24A, 2000 Frederiksberg, Denmark

Phone: +45 20342652

Email: car.int@cbs.dk

Abstract

This article analyzes how Danish listed firms comply with the Danish Corporate Governance Code’s

recommendations regarding the categories: Role of shareholders, role of stakeholders and transparency.

It is shown that the number of recommendations can be explained by six different underlying factors which

account for the vast majority of the variation. The analysis reveals that the official classification of the three

different sections in the Danish corporate governance must be abandoned. It is interesting to note that even

though the “comply or explain” principle assumes that a meaningful explanation is equally good as compliance -

the analysis documents that the vast majority of the firms complies with the recommendations. The article

introduces a new methodology to measure the degree of compliance within these specific areas. This

categorization serves as input for a multivariate analysis that explores how the different recommendations

covariate as well as can be placed into distinct discrete groups. The policy implication is that future code

revisions should rely on a multivariate approach when seeking to classify and structure the different code

sections regarding the firm’s stakeholders. Otherwise, there is a risk that board members may associate a large

number of recommendations as mechanistic “tick the box” exercises, which does not add value.

Keywords: corporate governance, transparency, compliance, comply or explain, factor and cluster analysis

Acknowledgements: I am grateful to Nicolai Søpstand for excellent and careful research assistance.

Int. Journal of Business Science and Applied Management / Business-and-Management.org

1 INTRODUCTION

During the last decade most developed countries have issued their own corporate governance codes that

vary in scope and size accordingly to institutional differences, but transparency is an important common

denominator, see Aguilera and Cuervo-Cazurra (2009) for an overview. The main objective of such

recommendations is to increase investor confidence assuring that executive management and board members

serve the interests of shareholders by providing sufficient of information about a firm’s corporate governance

structure. Transparency is a key ingredient facilitating shareholders to actively participate at the AGM or

alternatively at investor meetings. However, transparency is not only relevant towards the owners, but also in

relation to a number of other key stakeholders see e.g. Freeman (1990) such as creditors, customers, suppliers,

the local community, NGOs, the media etc.

There has been a relatively large attention to the potential value effect of corporate governance compliance,

while there is only little evidence on how the different individual recommendations vary and covariate as well as

can be classified into distinct groups. This article analyzes Danish firm’s corporate governance compliance.

Furthermore, it also presents a novel study of the interrelationship between the individual recommendations in

the first three sections of the Danish corporate governance code.

Effective communication with different stakeholders is vital in order to build trust as well as to create

visibility on the stock exchange. Listed firms, especially smaller firms may find it difficult to attract and

maintain investor’s attention. As a consequence, a firm must not only identify its key stakeholders, but it also

needs to evaluate the different stakeholders relative importance for the firm’s ability to run it’s business, which

may be expressed as their “license to operate” as a publicly listed company.

National corporate governance codes seek to accomplish similar goals but their structure and content e.g.

on transparency varies substantially.

The point is that the notion of transparency in the form of soft law is interpreted differently in different

countries. Moreover, transparency in the Danish code is considered to be broader as it does not rely on a narrow

legal perspective. In the UK, the traditional corporate governance code is even supplemented with a stewardship

code that contains specific recommendations to investors in the listed companies. The purpose is to formulate

good practice that aspire investors. The disclosure by investors will assist companies to understand the approach

and expectations of their major shareholders. The Financial Reporting Council states that the recommendations

“will assist those issuing mandates to asset managers to make a better informed choice, assist managers to

understand the expectations of current and potential clients, and may help investors interested in collective

action to identify like-minded institutions” (Homepage of FRC).

The Danish corporate governance code was introduced in 2001 and during the following years it has

undergone major changes. The code is soft law and builds on the “comply or explain” principle. The formal

work of developing and maintaining the Danish Code is delegated to the Danish Corporate Governance

Committee which consists of representatives from the listed firms, investors as well as advisors. The Committee

emphasizes that “Transparency is essential to ensure that shareholders and other stakeholders are able to

evaluate the performance of publicly traded companies.” (page 3 in the Code recommendation from April

2010).

The Danish Code consists of 9 separate sections dealing with different aspects of corporate governance.

This article focuses on the first three sections i.e. the role of shareholders and their interactions with the

management of the company, the role of stakeholders and their importance to the company’s corporate social

responsibility as well as openness and transparency.

The vast majority of national codes are typically quite comprehensive since they contain a large number of

specific recommendations that the board of directors must adhere to or alternatively give a reasonable

explanation why a firm has decided not to follow “best practice”, see Bauwhede and Willekens (2008). As a

consequence, there is a potential risk that too many detailed recommendations could stimulate a “tick the box”

mindset in which formality gets first priority at the expense of substance. By substance is meant that board

member’s norms are changed in order to stimulate an effective mechanism to discipline a certain type of

behavior, see Fasterling (2012).

The article is organized as follows. A literature overview is presented in the next section which is followed

by a description in section three of the data as well as the methodology to quantify the level of compliance. The

degree of stakeholder compliance of Danish firm’s is presented in section four. The results of the multivariate

analysis are presented in section five. The article ends with a conclusion and discussion.

2 LITERATURE

Fasterling (2012) links norms to companies’ compliance disclosure where he argues that disclosure regimes

may have negative effects if disclosure addresses use disclosed information without questioning it. The author

points to an interesting key point i.e. if there is no legal obligation to “comply or explain”, a company could

decide to send a positive signal to the public by voluntarily stating its compliance with a well-reputed corporate

Caspar Rose

governance code. In essence the author argues that compliance disclosure regimes could provide specific

opportunities for explicit and empirically traceable public discourse on applicable normative standards and

thereby facilitate the identification of adequate norms for regulating business activities.

The notion of soft law in relation to compliance is analyzed in Hooghiemstra and van Ees (2011) who

analyze a sample of 126 listed Dutch firms. They find that the overall compliance rate is quite high arguing that

firms fear about their reputation if they score low. However, they reveal that firms tend to use similar arguments

for non-compliance, hence the authors argue that the uniformity in adopting the standard of good corporate

governance may not be in line with the logic of corporate governance codes, which may casts doubt on the

effectiveness of using soft law.

2.1. Empirical studies

There a several studies that link corporate governance with firm performance, see e.g. Bozec and Dia

(2012), Gutierrez, Isabel and Jordi Surroca (2012), Seidl and Roberts (2013), Sanderson et al. (2010) and

Werder and Talaulicar (2005). However, there a quite few studies that study compliance patterns using

multivariate statistics. An exception is Talaulicar and Werder (2008) who rely on cluster analysis to identify

discrete groups of companies with similar patterns of code compliance. They find eight patterns of compliance

which are characterized by distinct forms of code conformity. Specifically, the authors investigate whether the

form of compliance with the recommendations of the German Corporate Governance Code appears to be

idiosyncratic to a specific company or feature similarities across firms. The authors find that the cluster solution

does not merely reflect the number of rejected code recommendations. Rather, companies with very similar rates

of overall compliance with the German code are assigned to different clusters because they feature different

patterns of conformity.

The issue of measurability of corporate governance compliance is crucial in the analysis. If one does not

quantify the compliance in a coherent and systematic way, any result may offer poor guidance for future policy

recommendations within this important area of research. Tsipouri and Xanthakis (2004) discuss this issue in

their analysis of how Greek companies adhere to the OECD guidelines. They find that Greek companies

demonstrate a fairly satisfactory degree of compliance although areas such as the role of stakeholders and CSR

score relatively low.

According to the authors, the merit of the exercise from a methodological perspective comes in its

approach towards the creation of “collectively subjective” weightings i.e. an effort to discuss the benefits of

separating the rating of the market from the rating of the companies.

2.2 Transparency studies

Bushman et al. (2003) conduct a large study of transparency which the authors define as the availability of

firm specific information to those outside publicly listed firms. The authors rely on a factor model of

transparency measures worldwide in order to analyze the underlying structure. Specifically, they find that their

factor analysis isolates two factors from the array of country-level measures of the firm-specific information

environment. The first factor, interpreted as financial transparency, captures the intensity and timeliness of

financial disclosures, and their interpretation and dissemination by analysts and the media. The second factor,

interpreted as governance transparency, captures the intensity of governance disclosures and, to a lesser extent,

the intensity and timeliness of financial disclosures used by outside investors to hold officers and directors

accountable. Thus, the authors investigate whether these factors vary with countries’ legal/judicial regimes as

well as political economies. Their main statistical result is that the governance transparency factor is primarily

related to a country’s legal/judicial regime, whereas the financial transparency factor is primarily related to

political economy. The authors use six different measures to quantify the corporate reporting environment in

combination with dissemination of information by the media. One may always discuss how transparency is

measured, but the authors results are interesting and highly relevant as the authors are some of the very few

scholars who rely on multivariate analysis of firm transparency within corporate governance.

The link between transparency and financial ratios has been studied by Adiloglu and Vuran (2012) who

study transparency in listed Turkish firms. They argue that high compliance with the corporate governance

standards means more accountable and transparent companies for investors. Specifically, they conduct

MANOVA analysis to examine the relationship between the calculated transparency levels and financial ratios.

The results reveal that transparency level has statistical differences among the group means of return on asset,

total debt/total assets, long-term debt/total assets and corporate governance index variables.

The challenge of how to measure transparency in corporate governance context is also analyzed by

Stefanescu (2014) who study corporate governance disclosure in the EU in relation to how firms comply with

the OECDs corporate governance guidelines. The author develops a disclosure index which consists of sub-

indices within different categories such as owners, boards, executives, committees and stakeholders. Stefanescu

measures how close each code in the sample was to the recommendations for good corporate governance using

Jaccard’s similarity coefficient in order to measure the disclosure level. Their results reveal that those codes

Int. Journal of Business Science and Applied Management / Business-and-Management.org

developed through the collaborations of a wider range of specialist from various economic fields and issued by

special committees set for this purpose appeared to best approximate the ideal model of best practices for

corporate governance transparency/disclosure.

There are several studies that seek to explore the link between financial performance and the level of

corporate governance compliance. However, these studies do not offer a coherent methodology for quantifying

or measuring the degree of compliance. The issue of transparency is the focus of the second category of studies

that rely on quantitative studies. The aim is to obtain a better understanding of the notion of transparency, but

there seems not to be common line of research within this group. This article seeks to fill this research gap.

3 DATA AND METHODOLOGY

This article analyzes the governance practices associated with the Danish corporate governance

recommendations introduced in April 2010. This version of the governance recommendations apply to all listed

companies with fiscal years ending in 2010 or mid 2011 (hereafter; fiscal year 2010). Thus, this article addresses

the fiscal year 2010 (ranging from the calendar year 2010 as well as the period mid 2010/2011).

When evaluating the companies with respect to corporate governance practices the analysis relies on all

relevant publicly available information. The main sources of information are reported governance practices,

relevant material on the companies’ web pages, and annual reports for the fiscal years 2010. The data has been

carefully manually collected.

There are six specific recommendations regarding the role of shareholders, three on the role of stakeholders

and transparency as well, see Recommendations on Corporate Governance (2010).

All companies have been assessed based on how they communicate their governance practices using a

binary scale of 0 or 1 point for each recommendation. If a company has chosen to comply with a given

recommendation, and this can be verified, the company receives a score of 1 (denoted: Complies). If a company

claims to comply but its practice proves otherwise it receives a score of zero (Complies poorly). Companies that

have chosen not to comply with a given recommendation score 1 if the explanation is accompanied by a

reasonable argument (Explains). Finally in case a company does not explain why a recommendation is chosen

not to be followed, or the explanation does not seem justified, a 0 score is given (Explains poorly). Most

recommendations are divided into sub recommendations, but since each of these sub recommendations are

considered equally important to those with only a single recommendation, they have all been treated similar, and

hence they all count as one each. This is to elude a discussion of the relative importance or weighting of the

different recommendations.

The different recommendations vary considerably in nature. Some are easily verified due to being very

specific with respect to disclosure, while others are harder to verify. This means that a company who claims to

be complying with a non-verifiable recommendation cannot be penalized since it cannot be checked if the

compliance statement is in fact true. Companies who have chosen not to comply with such recommendations

can be given the score 0 if their explanation is not satisfying.

As per September 2011 there were a total of 188 companies listed on Nasdaq OMX Copenhagen. The final

sample consisted of 155 companies. There are five main reasons for this: 1) Some companies are listed with two

share classes. 2) Some companies are listed as separate entities, but are in essence organized in a very similar

manner. To illustrate, Formuepleje/FormueEvolution consist of a series of listed companies, where the main

difference is their investment profile. Including all Formue companies would have yielded the same result for all

companies, thus resulting in a total sample score skewed disproportionately towards the score of these

companies. 3) Some companies have been liquidated or taken private before we got to assess their governance

practices, so information were no longer available. 4) Some companies are also listed in other countries, and

have therefore chosen to adhere to other governance frameworks. 5)

To ensure the highest possible validity, all companies have afterwards been contacted via e-mail and

presented the recommendations where they had been given a score of 0. The companies were given the

opportunity to respond, and argue their case, if they believed the evaluation was incorrect. When received the

companies’ feedback the initial scores were reassessed, and changed accordingly if the arguments had any merit.

In order not to favor companies who responded, all other companies who had been given 0 with respect to these

recommendations were also reassessed a second time.

The response rate was less than five percent. In all cases it was observed that only firms with relatively low

corporate governance scores did respond. As such, there is an inherent bias in the respond rate, as high

complying firms would rarely respond. However, only in a very few cases, we had to change the original

assessment.

Caspar Rose

4 DISCUSSION

This section presents the scores given on the individual recommendations. The following figures measure

the percentage of companies on the vertical axis, and the respective recommendations on the horizontal axis.

The reader should note that the vertical axis may be truncated in some of the figures in order to make the figures

easier to interpret. Each column in the following figures incorporates up to four different assessments of the

companies’ attitude towards the respective recommendations, and each column sums to 100%.The list below

describes each of the four categories:

Complies = Companies that comply with the recommendation; 1 point.

Complies poorly = Companies claim to comply, but in fact does not; 0 point.

Explains = Companies who does not follow the recommendation and explains well; 1 point.

Explains poorly = Companies who does not follow the recommendation and explains poorly; 0 point

Figure 1 Classification of compliance (simple average)

The classification of the four possibilities is shown in figure A that displays the distribution. The overall

impression is that the firms comply (87) whereas there is a quite equal distribution regarding the three other

categories.

4.1. The role of shareholders

The total number of recommendations studied is equal to the number of firms multiplied by the number of

recommendations (155 companies x 12 recommendations) i.e. 1380.

Figure 1 Role of the Stakeholders

Int. Journal of Business Science and Applied Management / Business-and-Management.org

Figure 1 displays the recommendations addressing the role of the shareholders which are generally

characterized by a high degree of compliance. Nearly all listed firms have their own IR department or an IR

responsible person who may facilitate an ongoing dialogue between the company and shareholders. This may

help the board to get a better understanding of the investor’s preferences and expectations.

Active ownership by shareholders can only be formally exercised at the AGM. Shareholders have a number

of legal rights such as the right to ask questions, formulate proposals, appoint members of the board, and most

importantly the right to vote. Therefore it is positive that all companies seek to promote shareholders attendance

at the general meeting.

There is a separation between ownership and control in many listed companies due to the existence of

various share structures that allow some shareholders to have more votes than their cash flow stakes. This issue

has been debated for several years, especially in the EU. It is documented that in the vast majority of EU

member states, a large proportion of listed firms have share structures that deviate from the “one share – one

vote” principle, see report by the EU Commission (2007) as well as Rose (2008) for an analysis of the one share

– one vote controversy.

Initially, the first Danish corporate governance code viewed such share deviations as harmful hence firms

should explain why their share structure deviated from the “one share – one vote” principle. However, this was

later considerably modified and the recommendation now states:

“1.2.1. The Committee recommends that the central governing body every year evaluate whether the

company’s capital and share structures continue to be in the interests of the shareholders and the company and

account for this evaluation in the management commentary in the annual report and/or on the company’s

website” (CG 2010, p. 7).

Figure 1 shows that several companies were given a score of 0, despite claiming to comply, which is

surprising given the considerable debate about this issue. The main issue here is whether claiming that both

share as well as capital structure is deemed to be in the interest of the shareholders, or if more substantive

argumentation is warranted. When judging the response to this recommendation it has been decided to follow an

approach where deviations from ‘one share, one vote’ require more substantive arguments for why this is the

best approach. When it comes to capital structure it is the opinion that investors are capable of making qualified

assessments on their own behalf if the current capital structure is reasonable and justifiable.

Most foreign shareholders do not attend the AGM so having the possibility to arrange an AGM

electronically is surely an ideal way to promote active ownership and dialogue with top management. The

Danish Company Act has for several years offered a legal possibility to arrange an AGM electronically, which

has also been enacted by the EU in the Shareholders Rights Directive. However, despite that the legal setup is

already in place there are several technical and practical obstacles that are not easily solved e.g. regarding

language and practical voting issues, including IT safety. As a consequence, the number of electronically

arranged AGMs has been extremely limited so far, but there is reason to believe that this will change in the

coming years as firms score extremely high on this matter.

One of the more controversial issues within the Danish corporate governance debate has been the use of

“blanco proxies” where the supervisory board sends prewritten proxies based on the company’s shareholder

supervisory board authority to vote on all matters in the name of the shareholder. Giving the board “blanco”

proxies provides management with a considerable degree of power, so when shareholders are to consider each

individual item on the agenda, management may find it harder to “push” their own agenda through at the AGM.

Moreover, in a situation of a proxy fight in a hostile takeover this will also weaken target management’s power

if it seeks to deter a bidder’s attempt to acquire the firm. Therefore it is positive that so many firms comply with

the recommendation to issue proxies with individual items. The Danish code also requires that both the

executive board and supervisory board are present at the AGM. Figure 1 shows that this is fully supported by

the firms.

Caspar Rose

4.2. The role of stakeholders

Figure 2 Role of the Stakeholders

Chapter 2 in the Danish Code deals with the role of a firm’s stakeholders. Figure 2 shows that companies

have identified their key stakeholders in accordance with 2.1.1 whereas only 15 percent explains poorly on their

stakeholder policy i.e. 15% of the companies have chosen to not adopt a stakeholder policy (2.1.2), without

explaining convincingly why this approach is chosen.

Regarding the adopting a CSR policy, 21% of the companies have chosen to deviate. Of these companies

two thirds, totally 14% of the companies are not arguing reasonably why this approach is chosen. 5% claims to

have such a policy, but does not communicate their approach. This is a fairly interesting result, as listed

companies are legally required to address CSR. The Danish Parliament has passed a law that requires large

firms to report on how their use of CSR can contribute to support responsible growth giving companies a

competitive advantage from implementing CSR (policies, actions results and actions). The requirement is

codified in the Danish Accounting Act (Årsregnskabsloven) and was enacted January 1. 2009, so companies

have had a few years to adapt to the new rules. However, despite the legal obligations, a relatively large

proportion of the firms do not address CSR sufficiently, as 14 percent explains poorly and 5 percent complies

poorly. This corresponds to the findings by Tsipouri and Xanthakis (2004) mentioned in section 2.1 where the

compliance degree within CSR also was found to be relatively poor.

4.3. Transparency

Figure 3 Transparency

Int. Journal of Business Science and Applied Management / Business-and-Management.org

Figure 3 addresses transparency, particularly with emphasis on written communication to stakeholders.

Most companies have adopted a communication strategy in accordance with 3.1.1.

A large proportion i.e. 38% of the companies has chosen to not communicate information to the market in

both Danish and English. Most firms in this group put forth reasonable arguments for choosing this practice.

Most companies which do not communicate in both languages have opted to communicate only in Danish. A

common argument among this group is that they have a local focus and that targeted shareholders are Danish,

which may be challenged as foreign investors now account for more than half of the market cap on the

Copenhagen Stock Exchange.

Regarding publishing quarterly reports, close to 23% of the companies have chosen to deviate from this.

Around half of these companies have not explained their deviation with reasonable arguments, where typical

claims that are made, are that quarterly reports will not contribute with additional valuable information about the

company’s situation and prospects.

This article has formulated a new approach to quantify the level of corporate governance compliance. It

builds on previous work e.g. by Talaulicar and Werder (2008) and Stafanesco (2014), but it differs in the

categorization process, as it relies on a coherent and more dense approach.

5. STATISTICAL ANALYSIS

This article makes use of multivariate analysis or more specifically, principal component, factor and cluster

analysis. The primary focus of factor analysis is to explain the interrelationships among a number of original

variables. Table 1 shows descriptive statistics for the dataset.

Table 1Variables

Mean

Std Dev

Investor relations activities

Capital/share structure

Physical or electronic AGM

Proxies to individual items

Management present at the AGM

Indentify key stakeholders

Stakeholder policy

CSR

Communication strategy

Danish and English

Quarterly reports

0.980

0.685

0.993

0.980

0.993

0.980

0.852

0.807

0.961

0.878

0.137

0.465

0.080

0.137

0.080

0.137

0.358

0.395

0.192

0.328

156

Table 1 shows that the overall compliance within the three sections on transparency is remarkable high.

The highest value is Physical or electronic AGM whereas the lowest value concern with companies capital and

share structure. Most of the variables are close to one i.e. 100 percent compliance. The standard variation is also

quite high among the firms.

5.1 Principal component analysis

The total number of recommendations within the first three chapters in the Danish Corporate Governance

Code equals 12 hence we have 12 original variables. Since all firms comply with Shareholders attendance at the

AGM, this means that there is no variation and therefore this variable has been excluded from the analysis. In

order to simplify the description of the set of board variables, one may wish to transform the 11 variables into

new uncorrelated variables called principal components, hence the name principal component analysis. This is

an exploratory technique that enables one to reduce the dimensionality of the problem i.e. reduce the number of

variables without losing much of the information, see e.g. Chatfield and Collins (1980) for a description of the

model, or Rose (2006) for a study of board composition. The model is a mathematical technique, where the

researcher does not need to specify any underlying model, such as specifying an “error term”.

The analysis focuses only on the variance, as the mean is normalized to zero. Each principal component is

a linear combination of the original variable, and the measure of the information conveyed by each principal is

its variance. The principals are arranged in order of decreasing variance. Let X

= (X

,…,X

) be a 11

dimensional random variable representing the vector of the twelve board variables in the analysis. The problem

consists of finding a new set of variables denoted Y

,…Y

which are uncorrelated with decreasing variance,

from first to last, where a

is a vector of coefficients. Since each Y

is a linear combination of the X’s, we have.

Caspar Rose

XaXaXaXaY

jjjjj



11122211

...

(1)

The first principal Y

is found choosing a

, so that Y

has the largest possible variance, where the objective

function equals Var(Y

) = a

a

,and  denotes the covariance matrix, subjected to the orthogonal

transformation a

= 1, and so forth with Y

etc. (which also is uncorrelated with Y

Estimating the components turns out to be identical to finding eigenvalues and eigenvectors (the vector of

’s), where the former is the variance of each component (the sum of eigenvalues equals the original variance).

Table 2 displays the results of the principal component analysis generated by the SAS statistics program.

Table 2 Eigenvalues of the Correlation Matrix: Total = 11 Average = 1

Eigenvalue

Difference

Proportion

Cumulative

2.026

0.676

0.184

1.350

0.186

0.122

0.307

1.164

0.041

0.105

0.412

1.122

0.097

0.102

0.515

1.025

0.022

0.093

0.608

1.002

0.142

0.091

0.699

0.859

0.114

0.078

0.777

0.744

0.049

0.067

0.845

0.694

0.056

0.063

0.908

0.638

0.266

0.058

0.966

0.372

0.033

1.000

Table 2 displays the associated eigenvalues, i.e. how much each principal component explains of the

variance in the board data. The first component has a value of 2,026 and this component accounts for 18,4

percent of the variation in the data. Six components (all with a value above 1) seem to explain the interrelation

among the variables. These six components explain nearly 70 percent of the variation in the dataset.

5.2 Factor analysis

Factor analysis has been used in many different studies where the aim is to study the underlying structure

of a set of variables, see Afifi et al. (2004) for a description of factor analysis. Factor analysis represents each of

the variables as a linear combination of a smaller set of common factors plus a unique to each of the response

variables. As a consequence, on may wish to use factor analysis, which attempts to explain the correlation

between a set of variables, in terms of a small number of factors. Contrary to principal component analysis,

factor analysis is not only concerned with explaining the variance, but in particular the covariance structure of

the variables. A major assumption is that it is not possible to observe these factors directly i.e. so-called latent

variables.

Factor analysis is applied when, e.g. there is not a clear distinction between dependent and independent

variables, but when one seeks to explain and identify, which underlying factors that account for the variation

among the variables.

The factor model assumes that there are m underlying factors (less than the number of variables), which are

denoted f

, f

and that each observed variable is a linear function of these factors, together with a residual

unique factor. The analysis specifies that the number of underlying factors is equal to six, based on the previous

principal component analysis, based on the so-called Kaiser’s rule, see Stevens (2002). The model can be

written as

x = f + u (1)

Let x be the vector of the twelve variables and f the vector of the six factors with coefficient matrix 

where 

is called the factor loading i.e. the loading on the j’th variable on the k’th factor. The vector u

describes the residual variation specific to the j’th variable. The six factors are usually denoted the common

factors, while the residuals are called the specific factors.

The model relies on a number of assumptions. First the specific factors are assumed to be independent of

one another and of the common factors. Thus, the common factors are usually assumed to be independent of

each other, although this assumption can be relaxed when the factors are rotated (if an orthogonal rotation is not

applied).

Since the variables have been standardized to have zero mean, the factors also have zero mean and unit

variance although the variances of the individual factors may vary (let the variance of u

be denoted by 

From the above assumptions it is easily shown that the covariance of x, which is denoted as  can be written as

expression (2), where  is the off diagonal terms of  (the co-variances).

Int. Journal of Business Science and Applied Management / Business-and-Management.org

 = 

+  (2)

The above equation is of crucial importance, since it demonstrates that the factors explain the off diagonal

terms of  exactly since  is diagonal. This implies that finding the factor loadings is equivalent to factorizing

the covariance matrix of x (given that the diagonal elements are non-negative).

In essence, the factor model breaks the variance of each variable into two parts. Since x

is standardized, its

variance equals 1 and is composed of the following two parts:

The communality, denoted by h

for variable j, i.e. the variance that is due to the common factors

The specificity denoted s

, i.e. the part of the variance that is due to the unique factor u

As a consequence, the variance of variable x

= 1 = (h

+ s

). Table 3 shows the communality estimates.

Table 3 Final Communality Estimates: Total = 7.691513

Investor

relations

activities

Capital

/share

structure

Physical or

electronic

AGM

Proxies

to indivi

dual

items

Manage

ment

present at

the AGM

Indentify

key

stakeholders

Stake

holder

policy

CSR

Communi

cation

strategy

Danish

and

English

Quarter

reports

0.737

0.564

0.865

0.815

0.805

0.635

0.722

0.688

0.604

0.647

0.605

To illustrate, the six factors explain 74 % of the variation within the IR activities, while they only explain

56 % of the capital/share structure compliance, which is the lowest percentage. Table 3 also shows that the six

factors nearly explain 87 % of the variation regarding the recommendation on whether the general meeting

should be conducted physically or as partly/entirely electronic general meeting. The overall picture is that the

six factors explain a high degree of the total variance in the dataset.

Recall, that the purpose of factor analysis is to identify the underlying factors, which enables the researcher

to more easily interpret the common factors. However, the initial factors are often not well suited for

identification. As a consequence, the researcher may rotate the initial factors, so that some of the factor loadings

are more close to +/- 1, and the rests are closer to zero. Technically, a rotation consists of finding new axes to

represent the factors see e.g. Johnson and Wichern (2002).

There are several different rotation methods, but this analysis relies on the so-called verimax procedure that

consists of an orthogonal rotation, which does not violate the assumption that the factors are uncorrelated (the

communalities are also unchanged when performing verimax rotation).

Table 4 Rotated Factor Pattern (orthogonal verimax)

Factor1

Factor2

Factor3

Factor4

Factor5

Factor6

Investor relations activities

0.020

-0.063

0.855

0.001

0.016

-0.026

Capital/share structure

0.154

0.277

0.242

0.559

0.301

-0.028

Physical or electronic AGM

-0.060

0.006

-0.015

0.008

0.928

0.007

Proxies to individual items

0.029

-0.160

-0.148

-0.078

0.067

0.869

Management present at the AGM

-0.040

-0.124

-0.106

0.875

-0.104

-0.010

Indentify key stakeholders

0.759

-0.086

-0.216

0.068

-0.000

-0.008

Stakeholder policy

0.746

0.389

0.115

0.011

-0.007

0.008

CSR

0.250

0.726

0.026

0.001

0.301

-0.073

Communication strategy

0.640

-0.091

0.426

-0.047

-0.042

-0.019

Danish and English

-0.104

0.777

-0.103

-0.008

-0.145

0.016

Quarterly reports

-0.078

0.350

0.356

0.109

-0.146

0.561

To interpret what variables with high loadings have in common i.e. to name the factor, one needs to rely on

specialist knowledge i.e. in this case knowledge about corporate governance, which inevitably may entail a

degree of subjectivity. Moreover, the important thing here is that any judgment can be motivated within a

corporate governance framework.

Table 4 displays that the first factor loads strongly on Identify of key stakeholders, Stakeholder Policy and

Communication Strategy, issues that may be regarded as stakeholder visibility. The firm must identify its key

stakeholders and communicate its stakeholder policy in order to stay visible.

The second factor loads heavily on Danish or English and CSR, which both are issues that are relevant for

the firm’s annual accounts, since the firms CSR policy must also be presented in the annual accounts. Therefore

this second factor may be interpreted as the Annual account factor.

The third factor loads strongly on Investor Relations Activities and to a lesser extent on Communication

Strategy. A firm’s investor relations department/person is responsible for the communication with its

Caspar Rose

shareholders. A main objective is to make the firm attractive for existing and potential shareholders. Therefore

this third factor may be entitled shareholder visibility factor.

The forth factor loads on Management present at the AGM as well as Capital/share structure. If a firm

decides to change its share capital or capital structure it must be approved by the shareholders at the AGM

where the supervisory board must explain the reasons for changing a firm’s capital or share structure. To

illustrate, according to Danish law, the AGM can delegate the possibility to issue new shares to the board

without asking the shareholders at the AGM.

This delegation of power means that the preemptive rights for the existing shareholders are set aside as

management may have discretion to conduct a rights issue to a new shareholder. Therefore this forth factor may

be interpreted as the AGM delegation of power factor.

The fifth factor may be a bit difficult to interpret, since there is one strong positive factor loading on

Physical or electronic AGM, which is followed by two other loadings both on 0,301 regarding Capital/share

structure as well as CSR. However, the issues of CSR and Capital/share structure are issues that shareholders

must confront its top management with at the AGM. Therefore this factor can be entitled as AGM accountability

factor.

The last factor loads positively on Proxies to individual items as well as Quarterly reports. This factor may

therefore be interpreted as shareholder information factor. Both are issues that are of priority when informing

the owners i.e. the shareholders about the firms realized financial performance as well as future plans.

Table 5 shows the variance explained by each factor. One notices that there is not a single factor that

dominates the variation in the dataset. Instead the six factors are relatively equal in describing the variation since

factor 1 accounts for 1.653 whereas the last factor six accounts for 1.079.

Table 5 Variance Explained by Each Factor

Factor1

Factor2

Factor3

Factor4

Factor5

Factor6

1.653

1.546

1.204

1.104

1.103

1.079

5.3 Cluster analysis

Cluster analysis is a methodology in which one may try to combine variables into groups when group

membership is not known in advance. Cluster analysis may be considered as an alternative to factor analysis

although the output is quite different. When the number of variables is limited one may use a scatter diagram,

but in this case one needs to rely on other ways of illustrating the grouping of variables. Cluster analysis boils

down to the analysis of distance between the variables as variables which are similar are located closer to each

other.

The first step is to determine the number of clusters or groups using either hierarchical or non-hierarchical

methods. The data is originally divided into three sections i.e. the role of shareholders, the role of stakeholders

as well as transparency. Shareholders are by definition a key stakeholder group, since they are the owners of the

firm and exercise their rights at the AGM. However, there are also other important stakeholders, so it may be of

interest to analyze the number of groups based on the firm’s corporate governance compliance data.

This article uses a methodology entitled K-means clustering which starts by dividing the data into K initial

clusters i.e. in this case three clusters. Then the means or centroids of the clusters are calculated and for a given

case, the distance to each centroid is calculated. If the case is closest to the centroid of its own cluster, it is left in

that cluster, otherwise it is resigned to the cluster whose centroid it is closets to and the process is repeated. The

process successively finds that particular variable and the cluster producing the larger variance and splits that

cluster accordingly until K clusters are obtained, see Afifi et al. (2004) for a description. The calculations are

done in SAS where the program assigns each observation to the cluster with the nearest seed. The result of the

analysis is shown in table that presents the cluster summary.

Table 6 Cluster Summary

Cluster

Frequency

RMS Std Deviation

Maximum Distance

from Seed

to Observation

Radius

Exceeded

Nearest Cluster

Distance Between

Cluster Centroids

122

0.178

1.370

1.168

0.295

1.306

1.073

0.338

1.530

1.073

Int. Journal of Business Science and Applied Management / Business-and-Management.org

The first cluster consists of the largest number of observations namely 122, followed by cluster two and three that consist of 21 and 13 respectively. Table 6 also reveal

that the distance between the cluster centroids is quite similar and close to one.

Figure 7 K means clustering with K = 3. Initial Seeds

Cluster

Investor

relations

activities

Capital/share

structure

Physical or

electronic

AGM

Proxies to

individual

items

Management

present at the

AGM

Indentify key

stakeholders

Stakeholder

policy

CSR

Communication

strategy

Danish

and

English

Quarterly

reports

1.0

0.0

1.0

0.0

1.0

0.0

1.0

0.0

1.0

0.0

1.0

0.0

1.0

0.0

1.0

Root-Mean-Square Total-Sample Standard Deviation

0.26

Root-Mean-Square Distance Between Observations

1.23

Criterion Based on Final Seeds =

0.2156

Caspar Rose

Table 7 presents how the different variables/recommendations are grouped together. All observations

naturally belong to the same cluster if there is only one cluster. However, notice when there are three groups.

All the variables: Physical /electronic AGM, Proxies to individual items, Management present at the AGM,

Stakeholder policy, Communication strategy and Danish/English annual account now joins group two. The third

group is joined by the following variables: Investor relation activities, CSR, Quarterly reports but group three is

also joined by Physical/electronic AGM, Proxies to individual items, Management present at the AGM and

Danish/English annual reports. Only the variable Capital/share structure remains in group one. This means that

we end up having seven variables in group three and three variables in group two and a single variable in group

one.

Figure 8 Statistics for Variables

Variable

Total STD

Within STD

R-Square

RSQ/(1-RSQ)

Investor relations activities

0.137

0.136

0.025

0.026

Capital/share structure

0.465

0.390

0.306

0.441

Physical or electronic AGM

0.080

0.078

0.041

0.043

Proxies to individual items

0.137

0.138

0.005

Management present at the AGM

0.080

0.001

Indentify key stakeholders

0.137

0.122

0.215

0.275

Stakeholder policy

0.355

0.222

0.612

1.582

CSR

0.395

0.286

0.480

0.924

Communication strategy

0.192

0.163

0.294

0.417

Danish and English

0.192

0.185

0.089

0.097

Quarterly reports

0.328

0.298

0.182

0.223

OVER-ALL

0.261

0.213

0.343

0.522

Table 8 depicts the cluster statistics based on the grouping of three clusters. Notice that the within standard

deviation is smaller (or equal to in a few cases) to the original standard deviations. The aim is to reduce the

variance or standard deviation and the ability for each variable to do so is captured in the R-square value. Notice

that in the two cases where the standard deviation is not reduced i.e. Proxies to individual items and

Management present at the AGM, the R-square value is nearly zero. The opposite is the case for the variables:

Stakeholder Policy and CSR where the R-square value is much higher since the cluster variation is smaller than

the original variation.

Table 8 also shows the R

values for predicting the variable from the cluster. The ratio of between-cluster

variance to within-cluster variance (R

to 1 - R

) also appears in the last column. The R square is for predicting

the variable for the cluster where the observations: Stakeholder Policy, CSR and Capital/share

Structure have the highest values. The pseudo F-statistics is equal to 39,99 an approximate expected

overall R

. The higher R square, the higher the variable contributes to the cluster formation. The last column

shows the statistic (RSQ/1-RSQ). A high ratio means that the variable is important in differentiating between the

clusters.

The key lesson is that the grouping and categorization in the Danish Corporate governance Codes is not at

all reflected in the statistical cluster analysis. This means that future revisions of the code should consider the

categorization and grouping in the first three sections using the above results.

The articles findings are related to the existing literature. Talaulicar and Werder (2008) also use cluster

analysis even though their scope is slightly different. They find eight discrete groups,whereas three groups are

identified in this article. The reason is that their sample covers all German corporate governance

recommendations (although they do not use the quantification process as in this article). However, the article’s

findings are in in line with Tsipouri and Xanthakis (2004) who find that the role of stakeholders as well as CSR

scores low.

6. DISCUSSION AND CONCLUSION

Transparency is a crucial element in corporate governance if listed firms want to attract capital from

external investors. However, transparency cannot only be seen from the perspective of the shareholders. It must

be considered from a broader perspective in which the board of directors identifies the firm’s key stakeholders.

This is also the main point which is emphasized in the Danish Code.

This article introduces a methodology to assess the level of corporate governance transparency in relation

to the recommendations regarding the firm’s stakeholders. This approach ensures that the classification of the

individual recommendations is not made ad hoc in an unsystematic manner, which may serve as inspiration for

corporate governance codes in other countries. The analysis reveals that the compliance level concerning the

codes recommendations regarding firm’s stakeholders is quite high. It is interesting to note that even though the

comply or explain principle assumes that a meaningful explanation is equally good as compliance the analysis

Int. Journal of Business Science and Applied Management / Business-and-Management.org

documents that the vast majority of the firms simply complies with the recommendation i.e. they do very often

explain why a firm has deviated from “best practice”.

The only exception is the recommendation that a firm must explain not only its capital structure, but also its

share structure e.g. if a firm has an ownership/voting ceiling, shares with dual class voting rights etc. see Rose

(2002) for a description of Danish takeover defenses.

This article combines the insight from institutional knowledge, in this case about Denmark with a sound

statistical analysis, which is quite rare in the literature. This enables one to get a better understanding of the

underlying structure of the firms reported transparency level as well as how different recommendations may be

classified into distinct sections. This approach may guide code drafters in a systematic way thereby improving

the quality of firm’s transparency communication.

This article demonstrates that the current classification of the Danish stakeholder recommendations needs

to be revisited. This article has demonstrated a way in which this can be done in the future. This implies that one

needs to build on a multidisciplinary approach that combines institutional/legal insight with multivariate

statistics. The latter discipline involves a whole range of different methods, but this article has shown that;

principal, factor as well as cluster analysis may be used as building blocks for such an analysis.

Transparency is a necessary precondition for creating trust among outside investors and top management,

as the presence of asymmetric information may create agency costs. When investors get a clearer picture of

firms corporate governance structure, in particular how a firm deals with its key stakeholders, they are more

inclined to believe that management serves the interests of the company. However, too many specific

recommendations may create a false sense of trust, as there is a risk that the board of directors may view the

process as a “tick the box” exercise. The consequence is that stakeholder transparency is not taken seriously

enough but instead appears as “empty words”. As a result, outside investors need to have a clear picture of how

a firm complies with the specific recommendations as well as how the recommendations are implemented in

practice. If corporate governance is to be applied in a sound manner creating added value for all parties, it is

crucial that investors feel that the recommendations are classified in a meaningful way. This article has

presented a systemic methodology for this task which can be generalized to other countries.

The future research implications of the articles findings as well as methodology are twofold. First, it seems

likely that the outlined methodology can be successfully used in order to quantify the degree of corporate

governance compliance. This entails that we will get a more reliable picture of the compliance level and that this

knowledge can be compared across countries. Secondly, in order to better understand the mechanisms and

nature of transparency, this article has shown that cluster analysis may a fruitful methodology, which may be

used more frequently than standard regression models. One the other hand, one should also acknowledge that

there are some research limitations. It may be difficult to compare across countries with different jurisdictions

and institutions. To illustrate, the shareholder value doctrine is well recognized in the US, whereas in continental

Europe, there is a broader acceptance that listed firms should also take into account the interests of other

stakeholders than the owners

The findings of the article also have managerial implications. The knowledge of what is considered best

practice in relation to transparency and stakeholder communication is vital when a firm wants to enter into a

dialogue with all its stakeholders. To illustrate, 15 % of the firms in the article’s sample explain poorly their

stakeholder policy. As a consequence, it will become more difficult to communicate and create relations with

stakeholders due to the absence of a clear and trustworthy stakeholder communication. This is especially the

case if a firm explains poorly in relation to its CSR policy.

REFERENCES

Adiloglu, Burcu and Vuran Bengu, (2012), The relationship between financial ratios and transparency levels of

financial information disclosures within the scope of corporate governance: Evidence from Turkey, Journal

of Applied Business Research, Vol. 28, iss. 4: 543-554

Afifi, Abdelmonem, Virginia A. Clark and Susanne May (2004), Computer-aided multivariate analysis, forth ed.

Chapman and Hall

Aguilera, Ruth V. and Alvaro Cuervo-Cazurra (2009), Codes of Good Governance, Corporate Governance: An

International Review, 17, Vol. 3, 376-387

Bauwhede, Heidi Vander and Marleen and Willekens (2008), Disclosure on Corporate Governance in the

European Union, Corporate Governance: An International Review, 16, 2: 101-115

Bozec, Richard, Mohamed Dia (2012). Convergence of corporate governance practices in the post –Enron

period: Behavioral transformation or box-ticking exercise, Corporate Governance, 2, 234-256

Bushman, Robert, Joseph D. Piotroski and Abbuie J. Smith (2003), What Determines Corporate Transparency?

Journal of Accounting Research, Vol. 42, No. 2 :207-252

Caspar Rose

EU Commission (2007) , Report on the Proportionality Principle in the European Union, joint report by:

Shearman and Sterling, ISS and ECGI, available on:

http://ec.europa.eu/internal_market/company/docs/shareholders/study/final_report_en.pdf

Fasterling, Björn, (2012), Development of Norms Through Compliance Disclosure, Journal of Business Ethics,

106: 73-87

Gutierrez, Isabel and Jordi Surroca (2012), Revisiting the corporate governance through the lens of the Spanish

evidence, Journal of Management and Governance, 08, November

Freeman, R. Edward; Evan, William M. (1990); Corporate Governance: A stakeholder

Interpretation, Journal of Behavioral Economics, v. 19, iss. 4, pp. 337-59

Hooghiemstra, Reggy and Hans van Ees, (2011), Uniformity as response to soft law: Evidence from compliance

and non-compliance with the Dutch corporate code, Regulation and Governance, 5, 480-498

Johnson, Richard A. and Dean W. Wichern (2002), Applied multivariate statistical analysis, fifth ed. Peson

Education

Recommendations on Corporate Governance (2010), Danish Committee on Corporate Governances, see;

http://corporategovernance.dk/file/291826/committee_recommendations_april_2010.pdf

Rose, Caspar (2002), Corporate Financial Performance and the Use of Takeover Defenses, European Journal of

Law and Economics, Vol. 13, No. 2: 91-112

Rose, Caspar (2008), A critical analysis of the one share – one vote” controversy, International Journal of

Disclosure and Governance, Vol. 5, No. 2:126-139

Talaulicar, Till, and Axel v. Werder, (2008), Patterns of Compliance with the German Corporate Governance

Code, Corporate Governance: An International Review, Vol. 16, No. 4: 255-273

Stevens, James P. (2002), Applied multivariate statistics for the social science, forth ed. Lawrence Erlbaum

Associates, Publishers

Tsipouri, Lena and Manolis Xanthakis (2004), Can Corporate Governance be Rated? Ideas based on the Greek

experience, Corporate Governance: An International Review, Vol. 12, No. 1, 16-27

Sanderson, Paul, David Seidl, John Roberts and Bernhard Krieger (2010), Flexible or not? The comply or

explain principle in UK and German Corporate Governance, Working Paper No. 407, University of

Cambridge, June 2010

Werder, v. Axel, Talaulicar Till and Kolot, Georg L. (2005), Compliance with the German Corporate

Governance Code: an empirical analysis of the compliance statements by the German listed companies.

Corporate Governance: An International Review, 13, 178-187

Stefanescu, Cristina A. (2014), Transparency and disclosure requirements – an analysis of corporate governance

codes, Atlantic Economic Journal, 42-113-114

Seidl, David, Paul Sanderson and John Roberts (2013), Applying the “comply-or-explain principle: discursive

legitimacy tactics with regard to codes of corporate governance, Journal of Management and Governance,

17, 791-826