National culture and the degree of disclosure by auditors within the extended auditor’s report

(1)

National culture and the degree of disclosure by auditors within the extended auditor’s report

Master thesis, MSc Accountancy & Controlling, track Accountancy University of Groningen, Faculty of Economics and Business

Danny Bos¹

g.a.bos.1@student.rug.nl

Supervisors University:

R.S. Tuinsma prof. dr. D.A. de Waard

January 21^st, 2019

Word count: 9768 (Excluding tables)

1 Student number: 2748959

(2)

1 | Danny Bos

National culture and the degree of disclosure by auditors within the extended auditor’s report

ABSTRACT: The implementation of the extended auditor’s report resulted in inequalities between disclosures by auditors from different countries. This paper responds to this by investigating the influence of culture, as a possible determinant, on the degree of disclosure by auditors. Consequently, this is the first study that relates different theories to the disclosure from an auditor’s perspective in an international context. This study is based on cross-sectional data of 575 firms from 7 countries, and uses the national cultures framework of Hofstede and accounting values of Gray to proxy for culture. Using a multiple linear regression analysis, this study finds that national culture is significantly related to the disclosure of key audit matters.

More specifically, power distance, secrecy and long-term orientation are negatively related to the disclosure of key audit matters, whereas individualism and masculinity are positively related to the disclosure of key audit matters. As the disclosure of key audit matters proxies for the degree of disclosure by auditors, it is concluded that national culture influences the degree of disclosure by auditors.

Keywords: Extended auditor’s report, disclosure by auditors, key audit matters, national culture, Hofstede, Gray, agency theory, stakeholder theory, legitimacy theory

(3)

2 | Danny Bos

In March 2018, the Association of Chartered Certified Accountants (ACCA) reviewed the world-wide implementation of the extended auditor’s report, which was issued by the International Auditing and Assurance Standards Board (IAASB). A major conclusion drawn from this study is that the disclosure by auditors differs between countries. More specifically, they found that auditors from the UK disclose quantitatively more compared to other countries.

However, nothing is mentioned about possible determinants for the differences between the reports. This study will respond to this by examining a possible determinant, being culture, of the degree of disclosure by auditors within the extended auditor’s report.

Whereas the former auditor’s report comprised a standard, two-page document, the extended auditor’s report provides “unprecedented levels of transparency of the work carried out by the auditor” (Smith, 2018). According to Aarnink (2014), one of the dangers of the extended auditor’s report is that it can lead to differences in reporting between countries, which is confirmed by ACCA (2018). This concern arises mainly because of the ISA 701², which requires auditors to communicate key audit matters. Auditors consider these matters as significant and require more investigation during the audit. Therefore, they also require a more detailed explanation towards third parties. The ISA 701 is part of the revised ISA 700³, which addresses “the auditor’s responsibility to form an opinion on the financial statements”. In the end, this opinion is presented in the extended auditor’s report, including the key audit matters.

Information asymmetry, derived from the agency, stakeholder and legitimacy theory, is an important concept in justifying the added value of the extended auditor’s report. Auditors can reduce the information asymmetry between the company and their shareholders and other stakeholders, by providing additional information in the extended auditor’s report. For the financial year 2014 (2013 for the United Kingdom), this extended auditor’s report has become mandatory for listed companies in the Netherlands (Aarnink, 2014). Other countries that apply the ISAs followed in the years thereafter. The fact that the application of the extended auditor’s report is mandatory, might suggest that the disclosure is strictly bounded to these guidelines.

However, the auditors use their professional judgment to decide on the degree of disclosure of

2 See https://www.ifac.org/system/files/publications/files/Proposed%20ISA%20701%20(Revised)-final.pdf

3 See https://www.ifac.org/system/files/publications/files/Proposed%20ISA%20700%20%28Revised%29- final.pdf

(6)

5 | Danny Bos key audit matters (Sirois, Bédard & Bera, 2018). Because auditors can decide whether or not to report an issue, the degree of disclosure of key audit matters is considered to be voluntary. Prior research shows that cultural aspects play an important role in explaining the differences in voluntary disclosure between countries (Luo & Tang, 2016; Orij, 2010; Ullmann, 1985).

Consequently, differences in the assessment and disclosure of key audit matters by auditors can be a result of cultural influences. Hence, this study will focus on the influence of culture on the degree of disclosure by the auditor within the extended auditor’s report. It is emphasized that it is not the goal of this research to examine the quality of the disclosure. Rather, this research attempts to address determinants of the degree of disclosure by auditors.

1.2 Academic contribution

This research paper can be placed in the literature with regard to the agency theory, the stakeholder theory and the legitimacy theory, related to the extended auditor’s report.

The extended auditor’s report is seen as a way to diminish the information asymmetry between the principal and the agent from the agency perspective (An, Davey & Eggleton, 2011).

The degree of disclosure in this report is considered to be voluntary. Voluntary disclosure by a company has been investigated from an agency perspective several times in the current literature (e.g. García-Meca, Parra, Larrán & Martínez, 2005; Shroff, Sun, White & Zhang, 2013). However, the agency view has not been applied to the disclosure by an auditor yet.

Hence, this research will extend the literature on the agency theory by taking the auditor’s perspective as the starting point, instead of a management’s perspective.

The main method for a company to communicate with their stakeholders from an accounting perspective is through the annual report (Van der Laan Smith, Adikhari & Tondkar, 2005). Consequently, the stakeholder theory provides the relevance of the annual report as well as the extended auditor’s report. Van der Laan Smith et al. (2005) refer to multiple corporate failures to highlight the “renewed debates and calls for corporations to report and be held responsible to a broader social mandate beyond their primary fiduciary duties” (p. 124), which comprises the stakeholder theory. Sneller, Bode and Klerkx (2017) investigate IT key audit matters in a stakeholder perspective. However, they solely focus on the Netherlands as their research area. For that reason, the use of an international perspective will extent the existing literature on the stakeholder theory related to disclosure by the auditor. This multi-country perspective will include Australia, France, Germany, Hong Kong, the Netherlands, Spain and the United Kingdom (UK).

(7)

6 | Danny Bos Limperg (1932, p. 11) argued that the auditor is a “confidential agent of the society”.

Consequently, the legitimacy theory is also applicable to this study. The legitimacy theory is mainly used to describe the relation between the organizations and the society (An et al., 2011;

Deegan, 2006). However, it is argued that the organizations’ perspective with regard to the society is comparable to the auditor’s perspective with regard to the society. To clarify, the auditor can signal its legitimacy by means of the extended auditor’s report. As becomes clear, current literature on the legitimacy theory takes the legitimacy of the organization as a starting point (e.g. Deegan, 2006; Deegan & Rankin, 1996). Hence, similar to the first contribution, this study will extend the literature on the legitimacy theory by using the perspective of the auditor, instead of the perspective of the organization.

Next to the theoretical relevance of this study, it is also practically relevant for several stakeholders of the extended auditor’s report. To clarify, when it is found that culture is of influence on the degree of disclosure by auditors, this will be valuable to investors as they can use this information in their analysis of the extended auditor’s report. It will also be relevant for the standard setters, because they can use this information to develop and improve the standards that led to the extended auditor’s report.

1.3 Research question

The goal of this research is to study the effect of culture on the degree of disclosure by the auditor. This fits within the research area developed by Gray (1988), which focuses on the cultural influence on accounting mechanisms. Gray (1988) proposes four hypotheses as a development of the theory, which are based on the widely accepted national cultures framework of Hofstede (1983). Consequently, using the cultural dimensions of Hofstede (1983; 2001;

2010), this paper tries to explicate the effect of culture on the degree of disclosure by the auditor within the extended auditor’s report. This leads to the following research question:

What is the effect of culture on the degree of disclosure by the auditor within the extended auditor’s report?

The remainder of this study will first provide the theoretical framework, which leads to the development of hypotheses. After that, the methodology of the study will be elaborated, which will be followed by the results section. Eventually, the results will be discussed in the conclusion and discussion section.

(8)

7 | Danny Bos

2. THEORY

In this section, the theoretical basis of this research paper will be elaborated. First, the theories underlying the different hypotheses will be described. After that, the hypotheses that will be tested in this research paper are developed by using empirical evidence from the literature.

2.1 Underlying theories

In this subsection, the applicable theories will be explained. First, the ‘Agency Theory’ in relation to the disclosure by the auditor will be examined. After that, the disclosure by the auditor will be linked to the ‘Stakeholder Theory’. Next, the ‘Legitimacy Theory’ related to the disclosure by the auditor is introduced. Lastly, ‘Hofstede’s National Cultures Framework’ will be described.

2.1.1 Agency Theory

The agency theory is founded by Jensen and Meckling in 1976 and primarily emphasizes the consequences of separation between ownership and control in a company (Fama & Jensen, 1983). The theory focuses on the relation between a principal (i.e. the shareholder), who has the ownership, and an agent (i.e. the management), who has the control (Bosse & Phillips, 2016). In the principal-agent relation, the agent has an information advantage due to its daily activities within the company and can therefore pursue its own interests at the costs of the principal (An et al., 2011). This is described as a conflict of interests, which can arise due to information asymmetry. In this sense, the principal will be incentivized to align the interests of the agent with its own interests, these efforts are called ‘agency costs’ (Jensen & Meckling, 1976). An example of agency costs are the costs related to auditing services (Williams, 1988).

The extended auditor’s report will diminish information asymmetry, as it discloses private information about the practices of the agent to the principal. As mentioned before, the degree of disclosure within the extended auditor’s report is voluntary (Sirois et al, 2018). Shroff et al.

(2013) find that voluntary disclosure will lead to a reduction of information asymmetry.

Moreover, Walker (2018) argues that the extended auditor’s report reduces information asymmetry because of a gain in transparency. As a result, the disclosure by auditors will lead to a reduction of information asymmetry.

2.1.2 Stakeholder Theory

The stakeholder theory expands the agency view by stating that “there are other groups to whom the corporation is responsible in addition to stockholders” (Freeman & Reed, 1983, p. 89).

Freeman (1984) defines stakeholders as groups that “can affect or are affected by the

(9)

8 | Danny Bos achievement of the organization’s objectives” (p. 52). According to Van der Laan Smith et al.

(2005), the annual report is the most important way for a company to communicate with its different stakeholders. This study finds that the stakeholder theory can be applied in explaining differences in voluntary disclosure between cultures, which is comparable to the structure of this study. In addition, the extended auditor’s report is not only prepared for and used by the shareholders, other stakeholders also base their (financial) decisions on the report, for example analysts (Mohanram, Saiy & Vyas, 2018; Prasad & Chand, 2017). Hence, next to the agency theory, the stakeholder theory is also applicable to this research.

The stakeholder theory is associated with the ‘accountability’ of a firm. Mulgan (1997) defines this term as the responsibility of one party to fulfil its duty towards another party.

According to An et al. (2011), the disclosure of accounting information is a major way for companies to discharge accountability. The extended auditor’s report is seen as a way to make the discharge more reliable. As a result, the provided assurance within the extended auditor’s report will further reduce the information asymmetry between the company and their stakeholders.

2.1.3 Legitimacy Theory

The legitimacy theory is related to the stakeholder theory, but extends it by positing that organizations are related to the society as a whole. Legitimacy theory states that organizations have to comply with the terms of a social contract between them and the society (Brown &

Deegan, 1998). It is implicated that the boundaries of the social contract are not fixed, as they are subject to changes over time (Deegan, 2006). The implementation of the extended auditor’s report is a clear example of such a change, as the auditor is able to provide more information on the audit, which can influence the expectations of the society towards the auditor (i.e. the social contract). The theory states that organizations can only continue their operations when they comply with the social contract (An et al., 2011). A failure of the organization to not comply with the social contract is referred to as the legitimacy gap (An et al., 2011).

It can be argued that this theory is also applicable to the auditor and its relation with the society, as the auditor is seen as a “confidential agent of the society” (Limperg, 1932, p. 11).

Next to that, the commonly known expectation gap, which entails that the society expects more from the auditor than what is reasonable, can be linked to the legitimacy gap. The expectation gap can be diminished by means of disclosure by the auditor within the extended auditor’s report, because the auditor can explicate how different issues are addressed during the audit.

(10)

9 | Danny Bos This will lead to a better understanding of the practices of an auditor, which will diminish the expectation gap (Cohen, Ding, Lesage & Stolowy, 2017). In this way, the auditor can communicate its legitimacy within the extended auditor’s report by providing more transparency towards the society when disclosing on key audit matters.

2.1.4 Hofstede’s National Cultures Framework

Culture is defined as the phenomenon of “the collective programming of the mind that distinguishes one group of people from another” (Hofstede, 2001, p. 9). Hofstede was the first to come up with an adequate theoretical model that enabled researchers to quantitatively describe a national culture (Taras, Steel & Kirkman, 2012). In 1983, Hofstede’s model consisted of four dimensions: uncertainty avoidance, power distance, individualism and masculinity (Taras et al., 2012). In 1987, Hofstede added long-term orientation as the fifth dimension. Finally, in 2010, indulgence was added as the sixth dimension (Hofstede, Hofstede

& Minkov, 2010). Next to these individual dimensions, Gray (1988) extended Hofstede’s model by introducing secrecy, which is the opposite of transparency and relates to several individual dimensions of Hofstede’s model (Hope, Tang, Thomas & Yoo, 2008). One could argue that disclosure is mainly related to country-specific risks. However, country-specific risks are reflected in the national culture (Chui, Kwok & Zhou, 2016; Kanagaretnam, Lobo, Ma & Zhou, 2016), which is exactly the determinant of disclosure that is investigated in this research.

Uncertainty avoidance

Uncertainty avoidance is defined as “the extent to which the members of a culture feel threatened by ambiguous or unknown situations” (Hofstede et al., 2010, p. 191). It measures to what extent people in a society try to avoid those insecure situations (Kanagaretnam et al., 2016). Societies with low uncertainty avoidance are characterized by people who accept the fact that uncertainty is part of life (Luo & Tang, 2016). As a result, people in these societies will take relatively more risks. Conversely, people in a high uncertainty avoidance society are more risk averse, as they feel uncomfortable in novel situations since these situations may not meet their expectations (Offermann & Hellmann, 1997).

Power distance

This dimension is described as the degree of hierarchy versus equality within a society (Jang, Shen, Allen & Zhang, 2017). Power distance indicates the degree to which members of a society who are not powerful accept that the power is not equally distributed within that society (Hofstede, 1980). Inequalities of power arise because people differ in their physical- as well as intellectual capabilities (Luo & Tang, 2016). Consequently, societies with low power distance

(11)

10 | Danny Bos allow these differences to emerge, whereas societies with high power distance will try to maintain the power inequalities.

Individualism

Gupta, Veliyath and George (2018) define individualism as “the preference for loosely-knit social frameworks with dominant individual rights” (p. 228). According to Jang et al. (2017), this dimension focuses on the emphasis on individuals versus groups within a society. To clarify, people in individualistic societies act independently, whereas people in collectivistic societies prefer to act in a group. An important distinction must be made between institutional collectivism and in-group collectivism (Brewer & Venaik, 2011). In institutional collectivism, individuals and organizations strive to distribute their resources to the society. On the contrary, when the individuals only strive to distribute their resources to the in-group (i.e. an organization), this is defined as in-group collectivism.

Secrecy

Secrecy is considered to be very applicable to this research, as it especially focuses on information disclosure. Secrecy is relatively similar to confidentiality, and is about disclosing as little information as possible to those not closely involved in managing and financing the company (Hope et al., 2008). Therefore, secrecy can also be considered as the opposite of transparency. In this research, secrecy is seen as a characteristic of the auditor towards its disclosure, instead of a characteristic of the management (Gray, 1988). Gray (1988) relates secrecy to three dimensions of Hofstede. Moreover, secrecy is related to professionalism, another construct developed by Gray (1988). To clarify, professionalism is measured using the same dimensions in an opposite direction. Consequently, only secrecy will be used as an independent variable.

Masculinity

Masculinity is the opposite of femininity, and indicates “the degree to which a society differentiates and emphasizes traditional roles between genders” (Gallén & Peraita, 2018, p.

2970). When a society scores low on masculinity, and therefore high on femininity, it is relatively socially-orientated and has a clear stakeholder orientation (Orij, 2010). Bradley, Schipani, Sundaram and Walsh (1999) describe high femininity as communitarianism, where justice and cooperation are important concepts (Zheng, El Ghoul, Guedhami & Kwok, 2012).

Societies that are more masculine have a shareholder orientation (Orij, 2010). To describe this situation, Bradley et al. (1999) introduce the term contractarianism, where values as competition and liberty are important concepts (Gupta et al., 2018).

(12)

11 | Danny Bos Long-term orientation

In 1987, Hofstede added the long-term orientation dimension to indicate the time horizon of societies (Orlova, Rao & Kang, 2017). In the broad definition of a long-term orientation, thrift and perseverance play a central role (Hofstede & Minkov, 2010). Gallén and Peraita (2018) also use a broad definition, and state that it is the degree to which a society orients its efforts towards the future versus the past or the present. Logically, a society with a focus on the future is classified as a society with a long-term orientation, and vice versa. Gupta et al. (2018) refer to long-term orientation as the ability of a society to adapt traditions to changing circumstances.

Societies with a short-term orientation tend to hold on to traditions and focus on achieving results quickly.

Indulgence

According to Beugelsdijk, Maseland & Van Hoorn (2015), indulgence is identified as a result of an analysis of the World Values Survey data from Minkov (2007), which critically analysed the quality of Hofstede’s cultural dimensions at that time. The degree of indulgence relates to the amount of gratification of human desires in a society, which is related to the enjoyment of life (Lu, Pattnaik, Xiao & Voola, 2018). The opposite of indulgent societies are restrained societies, which are characterized by strict social norms (Hofstede et al., 2010). In indulgent societies, there is relatively free gratification of human desires, such as shopping. Also, they consider freedom of speech as an important attribute of the human being and describe themselves as being happy (Lu et al., 2018). On the other hand, people in restrained societies are more pessimistic, and face difficulties in remembering positive experiences (Lu et al., 2018).

2.2 Empirical findings & Hypotheses development

In this subsection, the relevant empirical findings per independent variable will be discussed.

The explication of the empirical literature will lead to the hypotheses, which will be in accordance with the theories discussed and will be presented in a conceptual model. To derive the hypotheses, first the dependent variable, the degree of disclosure by the auditor, will be elaborated.

2.2.1 The degree of disclosure by the auditor

The focus of this study is to investigate the influence of culture on the degree of disclosure by the auditor. As mentioned in the academic contribution, there is a limited amount of literature available on the disclosure by auditors within the extended auditor’s report, due to its novelty.

(13)

12 | Danny Bos To clarify, in 2011 the IAASB suggested⁴ to narrow the information gap by reporting on key audit issues, which would lead to a more efficient market (Fama, 1970). Disclosure can be measured in quantity as well as in quality (Arping & Sautner, 2012). Current research in this topic is mainly focused on the quality of the disclosure. However, this research does not intend to indicate the quality of the disclosure. The degree of disclosure is objective, while assessing the quality is subjective (Van der Laan Smith et al., 2005). It is in accordance with Freedman and Stagliano (1992) who state that to determine quality, one has to focus on the meaning of the disclosure, which is not included in the scope of this research. Therefore, this study will focus on the quantity (i.e. the degree) of disclosure by the auditor.

2.2.2 Uncertainty avoidance and the degree of disclosure by auditors

Uncertainty avoidance related to the disclosure by companies has been extensively investigated (e.g. Orij, 2010; Hope, 2003). According to Gray (1988), societies with high uncertainty avoidance tend to be more secretive. These societies will be more secretive, as they are not willing to disclose information if the costs may exceed the benefits of disclosure. This is confirmed by Williams (1999), who finds that uncertainty avoidance leads to less disclosure in the Asia-Pacific region. Hope’s research (2003) contains a worldwide sample, and also confirms the expectation of Gray (1988) that uncertainty avoidance leads to less disclosure.

Similar findings result from studies of Gray and Vint (1995) and Orij (2010), which leads to the following hypothesis:

Hypothesis 1. Uncertainty avoidance has a negative influence on the degree of disclosure by auditors.

2.2.3 Power distance and the degree of disclosure by auditors

As mentioned before, societies with high power distance will try to preserve the power inequalities. According to Gray (1988), this preservation of power distance may be characterized by the restriction of information disclosure. As a result, individuals possessing little power will not receive specific information that could lead them to narrow the difference in power. This is confirmed by Luo and Tang (2016, p. 21), who conclude that “power distance discourages informational openness and inhibits the free exchange of ideas”. Moreover, Orij (2010) and Zarzeski (1996) put power distance in a broader perspective, and both find that

4 See https://www.ifac.org/system/files/publications/exposure-drafts/CP_Auditor_Reporting-Final.pdf, paragraph 23

(14)

13 | Danny Bos power distance is negatively related to information disclosure. As the current literature is found to be relatively consistent, the following hypothesis is formulated:

Hypothesis 2. Power distance has a negative influence on the degree of disclosure by auditors.

2.2.4 Individualism and the degree of disclosure by auditors

Individualism related to information disclosure has been investigated extensively (e.g. Gallén

& Peraita, 2018; Luo & Tang, 2016). This led to ambiguous findings, as Gallén and Peraita (2018) find a negative influence of individualism on disclosure, whereas Luo and Tang (2016) document a positive relationship. However, these studies do not emphasize the difference between institutional collectivism and in-group collectivism (Brewer & Venaik, 2011). This study will take the in-group collectivism perspective as a starting point, because this perspective is also applied by Gray (1988) in defining his accounting values. According to the in-group collectivism perspective, collectivistic cultures are less concerned with stakeholders and the society as a whole compared to individualistic cultures, because they only focus on their in- group, such as a company (Brewer & Venaik, 2011). This will lead collectivistic societies to disclose less to outside stakeholders, as they tend to be transparent and honest towards their in- group only. This reasoning leads to the following hypothesis:

Hypothesis 3. Individualism has a positive influence on the degree of disclosure by auditors.

2.2.5 Secrecy and the degree of disclosure by auditors

Multiple studies tested the secrecy hypothesis. Gray and Vint (1995) and Zarzeski (1996) both confirm that secrecy is negatively related to accounting disclosures. They split their results into the different effects of the individual dimensions, and find results as hypothesized. Moreover, tests of Gray’s theory (1988) by Salter and Niswander (1995) and Orij (2010) also confirm the secrecy hypothesis. In addition, Yusoff, Othman and Yatim (2014) find that secrecy is negatively related to environmental (i.e. voluntary) disclosure. This is in accordance with the combined results of Hope et al. (2008) and Liu, Kong, San and Tsang (2018). Hope et al. (2008) find that less secretive (i.e. more transparent) societies have higher audit quality, which can be read as better accounting disclosures, because more transparent societies are more committed to high-quality audits. Next to that, Liu et al. (2018) find that audit quality is positively associated with voluntary disclosure by the management. Therefore, it can be argued that transparent societies are positively related to the degree of voluntary disclosure, as both terms

(15)

14 | Danny Bos are related to audit quality. This implies that secretive societies are negatively related to the degree of voluntary disclosure. To summarize, a society with high secrecy will lead to a lower degree of disclosure by auditors, which leads to the following hypothesis:

Hypothesis 4. Secrecy has a negative influence on the degree of disclosure by auditors.

2.2.6 Masculinity and the degree of disclosure by auditors

Gray (1988) argues that there is a weak link between masculinity and transparency, which is supposed to be negative. Williams (1999) tests this expectation of Gray (1988), and confirms that masculinity is negatively related to voluntary disclosure. Van der Laan Smith et al. (2005) argue that feminine societies have a stakeholder orientation, which should lead to more voluntary disclosure in these societies. This hypothesis is confirmed, which implicates that masculine societies experience relatively less disclosure. Maali and Al-Attar (2017) and Orij (2010) also test the influence of masculinity on the degree of disclosure, and both find that masculinity is negatively related to levels of disclosure. The current literature agrees to a large extent on the effect of masculinity on disclosure, which leads to the following hypothesis:

Hypothesis 5. Masculinity has a negative influence on the degree of disclosure by auditors.

2.2.7 Long-term orientation and the degree of disclosure by auditors

According to Orij (2010), long-term orientation in a society is the same as long-term orientation in business. Bradley et al. (1999) argue that a long-term orientation is related to a stakeholder perspective, which emphasizes the needs of the stakeholders and therefore demands more disclosure compared to a shareholder perspective. Luo and Tang (2016) provide statistical evidence for this claim, by linking long-term orientation to voluntary disclosure. Moreover, several studies conclude that long-term orientation will have a positive effect on the degree of disclosure (Garcia-Sanchez, Cuadrado-Ballesteros & Freis-Aceituno, 2016; Halkos &

Skouloudis, 2017). As a result, the next hypothesis is formulated as follows:

Hypothesis 6. Long-term orientation has a positive influence on the degree of disclosure by auditors.

2.2.8 Indulgence and the degree of disclosure by auditors

Due to its novelty, there is little literature available about indulgence related to disclosure.

According to Hofstede (2011), freedom of speech is an important characteristic of an indulgent society. This characteristic leads Maali and Al-Attar (2017) to formulate a positive association

(16)

15 | Danny Bos between indulgence and corporate disclosure. Halkos and Skouloudis (2017) confirm this and find a significant positive relationship between indulgence and voluntary disclosure. Disli, Ng and Askari (2016) dispute this by stating that indulgence will lead to more environmental pollution due to the gratification of human desires, which will, in turn, result in less disclosure.

Gallego-Álvarez and Ortas (2017) agree with this view by predicting a negative relationship.

They argue that an indulgent society is focused on the gratification of human desires, and therefore demands less voluntary disclosure by companies. Because of the contradicting results within the literature, the following hypothesis is formulated:

Hypothesis 7. Indulgence has an influence on the degree of disclosure by auditors.

2.2.9 Conceptual model

Figure I shows the conceptual model, which graphically describes the purpose of this study based on the hypotheses as elaborated before.

Figure I: Conceptual model

(17)

16 | Danny Bos

3. METHODOLOGY

This section will elaborate on the methodology used to execute this research. First, the applicable sample and the way that the data is extracted will be described. After that, the dependent variable will be operationalized. The next subsection discusses and operationalizes all the independent variables, including the control variables. Lastly, the research model along with several methodological aspects are elaborated.

3.1 Sample

As mentioned in the academic contribution, this research has an international focus which comprises the disclosure by auditors from Australia, France, Germany, Hong Kong, the Netherlands, Spain and the UK. The international focus will create a situation where the impact of different cultures can be measured, according to Hofstede (1983; 2001). Moreover, all of these countries apply the ISAs. This is of great importance to conduct this research since it has become mandatory in the ISAs to use the extended auditor’s report along with its key audit matters (Boolaky & Quick, 2016). Taking into consideration that the extended auditor’s report is implemented in Australia, France, Germany, Hong Kong and Spain starting from the financial year 2017, the sample focused on the auditor’s reports from the financial year 2017 in order to optimally compare the reports between the different countries.

For the countries applicable to this research, national indices were applied. From these national indices, only the main indices were included. It was assumed that the companies and the auditors from the main indices of a country reflect the overall national sentiment in that country, because of the size and the significance of the companies and auditors included in the main indices.

A number of adjustments were made during the collection of the data. Especially due to a lack of data availability with regard to the disclosure of key audit matters, several items had to be excluded from the sample. As culture is based on the nationality of an auditor, companies that were audited by auditors from countries outside the sample were excluded from the initial sample. However, several companies were audited by auditors from another country within the sample. These items were moved to the applicable country of the auditor’s nationality, and resulted in an adjustment of +5 items for the UK. All in all, the final sample consisted of 575 companies, divided over the countries as shown in table I.

(18)

17 | Danny Bos

Country Indices Number of

companies

Percentage of total Australia

Excluded & Total

ASX 100 100

5 95

16.5 France

Excluded & Total

CAC 40, CAC next 20, CAC mid 60

120

10 110

19.2 Germany

Excluded & Total

DAX 30 30

0 30

5.2 Hong Kong

Excluded & Total

HSLI 110

8 102

17.7 The Netherlands

Excluded & Total

AEX, AMX, AScX 75

10 65

11.3 Spain

Excluded & Total

IBEX 35, IBEX Medium Cap, IBEX Small Cap

85

17 68

11.8 United Kingdom

Included & Total

FTSE 100 100

5 105

18.3

Total 575 100.0

Table I: Sample selection

3.2 Data gathering

Archival data is the type of data that was used in this study, as it is the most appropriate type when analysing large amounts of data due to its statistical power (Barnes, Dang, Leavitt, Guarana & Uhlmann, 2018). Archival data on the cultural dimensions of Hofstede was collected from the webpage where the model of Hofstede is operationalized (https://www.hofstede- insights.com/country-comparison/). These scores are based on the work of Hofstede and it is in accordance with the method that Orij (2010) used, which both enhance the reliability as well as the external validity of the data. Archival data on the disclosure by auditors was hand-collected from the annual reports of the companies that were selected as described in the sample section.

According to Fogarty (2006), at least one variable should be hand-collected to increase the reliability of the data used in a study. By collecting the data of disclosure of key audit matters manually, the reliability of the data was therefore enhanced.

3.3 Dependent variable

The goal of this study is to research the effect of culture on the degree of disclosure by the auditor. The degree of disclosure by the auditor was based on the disclosure of key audit matters. The Public Company Accounting Oversight Board (PCAOB) uses the term ‘critical audit matter’⁵ to indicate the phenomenon of a key audit matter. This research focuses on countries that apply International Standards on Auditing (ISA), therefore the term ‘key audit

5 See https://pcaobus.org/Rulemaking/Docket034/Release_2013-005_ARM.pdf

(19)

18 | Danny Bos matter’ will be used in this research instead of ‘critical audit matter’. To proxy for the disclosure of key audit matters, the total number of words disclosed on key audit matters was applied, following Gutierrez, Minutti-Meza, Tatum and Vulcheva (2018).

To further enhance the reliability of the data, the data that was hand-collected has undergone a test-retest as adhered by Charles (1995). Several items from the sample were selected, and have been retested to calculate the correlation between the first and second time of data collection. Weir (2005) describes the test-retest in more detail, and discourages the use of Pearson r as it does not detect systematic errors. Consequently, the intraclass correlation coefficient (ICC) was applied, which is a more desirable way to assess reliability (Koo & Li, 2016). According to Nunnally and Bernstein (1994), the measurement error becomes negligible at a level of 0.80 or higher on ICC. Using a two-way mixed-effects model, the ICC was equal to 1.00 at a 95% confidence level, which indicates a high level of reliability.

3.4 Independent variables

This study contains seven independent variables, as shown in figure I. Next to these independent variables, several control variables were applied. Table II summarizes all the variables by presenting the corresponding abbreviations, along with the type of variable and the operationalisation of the variables.

3.4.1 Test variables

Uncertainty avoidance, power distance and individualism

Uncertainty avoidance, power distance and individualism are the first, second and third independent variable in this study, respectively. In accordance with Orij (2010), the scores on these dimensions were derived from the aforementioned webpage. These scores could vary from 0 to 100, where a higher score indicated a higher degree of that particular dimension within a particular country.

Secrecy

Secrecy is the fourth independent test variable that was investigated. Figure II shows that secrecy was measured by the sum of uncertainty avoidance plus power distance, less the score on individualism, which is the same way as Orij (2010) and Hope et al. (2008) operationalize secrecy. To clarify the link between secrecy and the dimensions, Gray (1988) argues that societies with high uncertainty avoidance prefer to restrict the amount of information disclosure. This will avoid conflict and competition and also preserves the security of information, which both diminish uncertainty. Next to that, a society with high power distance will also tend to restrict information disclosure. The restriction of information disclosure will

(20)

19 | Danny Bos preserve the power inequalities through withholding information that would interrupt this power distance. Lastly, Gray (1988) uses the in-group collectivism variant of this dimension.

Following this variant, high collectivism will lead the information to be disclosed only to the internal stakeholders, like an organization, which enhances the secrecy in such a society (Orij, 2010).

Again, the scores on each dimension could range from 0 to 100, where 0 indicated the minimum and 100 the maximum. Hence, the score on secrecy could range between -100 at the minimum and 200 at the maximum.

Gray – Hofstede Uncertainty avoidance Power distance Individualism

Secrecy + + ˗

Professionalism ˗ ˗ +

Figure II: Associations between the values of Gray and the dimensions of Hofstede

Masculinity, long-term orientation and indulgence

Masculinity, long-term orientation and indulgence are the fifth, sixth and seventh independent variables that were investigated, respectively. As mentioned before, the scores on these dimensions were extracted from the webpage where the model of Gerard Hofstede is operationalized (Orij, 2010). Similar to the previous variables, the scores on these dimensions could range from 0 to 100.

3.4.2 Control variables

To control for confounding influences regarding the dependent variable, several control variables were used. These control variables are described below, along with the way how the data on these variables was gathered.

Legal system

The legal system of a country is the first variable that was controlled for, due to its effect on the timing and level of disclosure by firms, and therefore also by auditors (Ball, Kothari & Robin, 2000; Hope, 2003). This variable was measured in accordance with Ball et al. (2000) and Hope (2003) by distinguishing civil law and common law. In accordance with results from Hope (2003), a dummy variable was implemented to classify France, Germany, Netherlands and Spain as civil law countries and Australia, Hong Kong & the UK as common law countries.

(21)

20 | Danny Bos Firm size

The second variable that was controlled for is firm size, as it is often found that it influences the level of disclosure (Cowen, Ferreri & Parker, 1987; Hackston & Milne, 1996). Firm size was measured by the natural logarithm of total asset value, following Liesen, Hoepner, Patten and Figge (2015) and was extracted from the same annual reports as described in the sample section.

Riskiness of the auditee

The riskiness of the auditee is the third control variable that was used in this study. This riskiness was measured by the leverage of each particular company, as this captured the business risk and therefore also the auditor’s risk (O’Keefe, Simunic & Stein, 1994).

Consequently, an auditee with higher business risk was expected to face more disclosure by the auditor. The leverage was measured by total liabilities divided by total assets (O’Keefe et al., 1994), which were extracted from the applicable annual reports.

National wealth

National wealth should always be a control variable when correlating the cultural dimensions with other variables (Hofstede, 2011). For that reason, national wealth was added as the fourth control variable. To clarify, power distance and individualism are significantly correlated with wealth (Gallén and Peraita, 2018; Hofstede, 2011). National wealth was measured by the Gross National Product per capita, following Hofstede (2011), and was extracted from the well- respected International Monetary Fund datacentre.

Type of auditor

Lastly, the type of auditor firm was included as a control variable. A dummy variable was implemented to differentiate between big 4 and non-big 4 accounting firms, because big 4 firms seem to be more independent than non-big 4 accounting firms (DeAngelo, 1981). As a result, big 4 firms are likely to disclose more compared to non-big 4 accounting firms. To conclude, the data on the type of auditor was also extracted from the applicable annual reports.

3.5 Research model

This section will firstly introduce the statistical model. Thereafter, some methodological aspects to increase the reliability of the results are discussed.

3.5.1 Statistical model

This study entailed a multiple linear regression analysis to investigate the effect of national culture on the degree of disclosure by auditors. The next statistical model was used in this study:

(22)

21 | Danny Bos 𝑫𝑺𝑪_{𝒊,𝟐𝟎𝟏𝟕} = 𝜷_𝟎+ 𝜷_𝟏𝑼𝑨𝑽_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟐𝑷𝑫𝑻_{𝒊,𝟐𝟎𝟏𝟕} + 𝜷_𝟑𝑰𝑵𝑫_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟒𝑺𝑬𝑪_{𝒊,𝟐𝟎𝟏𝟕}

+ 𝜷₅𝑴𝑨𝑺_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟔𝑳𝑻𝑶_{𝒊,𝟐𝟎𝟏𝟕} + 𝜷_𝟕𝑰𝑫𝑳_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟖𝑳𝑮𝑳_{𝒊,𝟐𝟎𝟏𝟕}

+ 𝜷_𝟗𝑺𝑰𝒁_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟏𝟎𝑹𝑺𝑲_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟏𝟏𝑾𝑳𝑻_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜷_𝟏𝟐𝑻𝒐𝑨_{𝒊,𝟐𝟎𝟏𝟕}+ 𝜺

In this statistical equation, i refers to the auditor of a particular firm, 2017 refers to the financial year that is applicable and Ɛ represents the error term. As mentioned before, all abbreviations of the betas can be found in table II. The data in this study comprised cross- sectional data, as the analysed data was focused on the same point in time, in this study the financial year 2017. Consequently, time differences did not play a role in the results. This enhanced the internal validity of the results, because of the comparability of the analysed data.

Variable Name Type of

variable

Measurement

Dependent variable Degree of disclosure by the auditor

DSC Interval # of words disclosed on key audit matters Independent variables

Test variables

Uncertainty avoidance UAV Interval Score between 0 and 100 Power distance PDT Interval Score between 0 and 100 Individualism IND Interval Score between 0 and 100

Secrecy SEC Interval Uncertainty avoidance + power distance – individualism

Masculinity MAS Interval Score between 0 and 100

Long-term orientation LTO Interval Score between 0 and 100

Indulgence IDL Interval Score between 0 and 100

Control variables

Legal system LGL Dummy Code law (0) versus common law (1) Firm size SIZ Interval Natural LOG value of total assets Riskiness of the auditee RSK Interval Total liabilities ÷ total assets National wealth WLT Interval Gross National Product per capita

Type of auditor ToA Dummy Big 4 firms (0) versus non-big 4 firms (1)

The values of the variable ‘Firm size’ are winsorized to reduce the effect of outliers. All variables that include monetary amounts are measured in euros. For sample items that use other currencies, the exchange rate from the applicable year-end date is applied.

Table II: An overview of the variables

3.5.2 Robust standard errors

In multiple linear regression analyses, heteroscedasticity is a factor that should be addressed to assess the reliability of the data (Anderson & Schumacker, 2003). When heteroscedasticity is

(23)

22 | Danny Bos present, the standard errors related to the betas from the statistical model are not normally distributed (White, 1980). The use of robust standard errors is a way to deal with heteroscedasticity, as this reduces outliers judiciously (Anderson & Schumacker, 2003). Hence, robust standard errors are applied in the regressions.

3.5.3 Data modifications

As the firm size variable was expected to contain large variances, winsorizing was used to reduce the effect of the outliers. During the process of winsorizing, the maximum and minimum value of a variable were set at the mean of all values plus or less three times the standard deviation, respectively. The outliers that were beyond these borders, were assigned those corresponding values.

(24)

23 | Danny Bos

4. RESULTS

This section deals with the results from the regression as described in the statistical model section. The descriptive statistics will be analysed first. After that, the main results of the hypotheses will be provided. Eventually, the results of several robustness tests will be elaborated.

4.1 Descriptive statistics

Table III shows the descriptive statistics of all the applicable variables within this study. These statistics show that the applied countries in this research provide for cultural differences, as showed by the minimum and maximum values of the independent test variables. Another major observation from the descriptive statistics is that the mean of the type of auditor is equal to 0.10, which means that 90% of the firms is audited by a big 4 firm.

Observations Mean Std. Dev. Min. Max.

Dependent variable

Words on key audit matters 575 1355 743 157 7445

Independent variables

Test variables

Uncertainty avoidance 575 55.97 22.33 29 86

Power distance 575 50.27 15.07 35 68

Individualism 575 67.71 23.27 25 90

Secrecy 575 38.53 43.53 -19 92

Masculinity 575 50.46 15.89 14 66

Long-term orientation 575 53.24 16.52 21 83

Indulgence 575 51.51 19.42 17 71

Control variables

Legal system 575 0.53 0.50 0 1

Firm size 575 23.09 1.99 16.65 28.67

Riskiness of the auditee 575 0.60 0.23 0.0009 1.70

National wealth 575 35652 6249 23454 44815

Type of auditor 575 0.10 0.30 0 1

The values of the variable ‘Firm size’ are winsorized to reduce the effect of outliers. All variables that include monetary amounts are measured in euros. For sample items that use other currencies, the exchange rate from the applicable year-end date is applied.

Table III: Descriptive statistics

To give more insight in the dependent variable, figure III shows the number of words on key audit matters, as well as the average words per key audit matter for every country. From this figure, it becomes clear that differences exist between countries regarding the key audit matters, which is in accordance with the review held by the ACCA (2018) as described in the

(25)

24 | Danny Bos introduction. Despite the fact that this figure does not provide any statistical evidence, it does give a representation of the differences in disclosure between countries (i.e. cultures).

Figure III: Insight in the dependent variable

Multicollinearity occurs when an independent variable is predicted by another independent variable, and should be addressed when analysing and testing quantitative data. A correlation matrix is most often used to get a first impression of possible multicollinearity. This matrix is shown in table IV, where correlations higher than 0.8 are indications of possible multicollinearity (Brooks, 2014). The correlation matrix shows that the correlation between secrecy and power distance may indicate multicollinearity. This correlation was expected in advance, as secrecy consists of power distance, amongst others. Therefore, it is deemed not necessary to exclude secrecy from this research. However, secrecy will be excluded from the model in which all independent variables are tested, to avoid multicollinearity in this model.

Next to this correlation, the correlations between indulgence and power distance and between indulgence and individualism also exceed the critical value of 0.8. These correlations were not expected in advance, as indulgence should not be related to the other dimensions, as elaborated in section 2.1.4. Accordingly, it is decided to exclude indulgence from the regression analyses, as this could lead to multicollinearity. Lastly, the correlation between legal system and uncertainty avoidance exceeds the critical value of 0.8. Because of this, and because the legal system of a country does not seem to have an effect on the disclosure, this control variable is also dropped in further testing as this could lead to multicollinearity.

1246 1279

1927

1019 1115

1418

1802

413 403 542

375 332 435 413

0 500 1000 1500 2000 2500

Australia France Germany Hong Kong The

Netherlands

Spain United

Kingdom

Key audit matters per country

# words on key audit matters # words per key audit matter

(26)

25 | Danny Bos

DSC UAV PDT IND SEC MAS LTO IDL LGL SIZ RSK WLT ToA

DSC 1

UAV 0.0076 1

PDT -0.2127*** 0.3217*** 1

IND 0.1945*** 0.1149*** -0.7566*** 1

SEC -0.1737*** 0.5629*** 0.9156*** -0.7376*** 1

MAS 0.1856*** -0.4314*** -0.1735*** 0.0284 -0.2965*** 1

LTO 0.0221 0.1216*** 0.3735*** -0.3953*** 0.4030*** -0.3617*** 1

IDL 0.1624*** 0.0562 -0.8038*** 0.9690*** -0.7674*** -0.0689 -0.4764*** 1

LGL 0.0112 -0.8466*** -0.2664*** 0.0005 -0.5263*** 0.7239*** 0.5287*** 0.0304 1

SIZ 0.4179*** -0.0524 -0.0701 -0.0940** 0.0476 0.2585*** 0.1392*** -0.1604*** 0.0714* 1

RSK 0.2535*** 0.1200*** 0.0377 -0.0026 0.0760* -0.0390 0.1302*** -0.0207 -0.1426*** 0.3490*** 1

WLT -0.1432*** -0.5463*** -0.3708*** 0.2095*** -0.5206*** 0.1191*** -0.3285*** 0.2051*** 0.4964*** -0.0800* -0.1583*** 1

ToA -0.0872** 0.3848*** 0.3702*** 0.0198 0.3150*** -0.1443*** 0.1932*** -0.0760* -0.3060*** -0.0663 0.0004 -0.1668*** 1

***, ** and * coefficients are statistically significant at a 1%, 5% and 10% confidentiality level, respectively (tested two-sided). See table II for the definitions of the variables.

Table IV: Pearson correlation matrix

(27)

26 | Danny Bos

4.2 Main results

Table V shows the results of all the regressions that are executed to test the developed hypotheses. Each regression is displayed in a separate model. Model A to model F show the results for hypothesis 1 to hypothesis 6, respectively. Model G shows the results when all independent variables are included, except for secrecy to avoid multicollinearity issues. To conclude, all regression are executed using robust standard errors and the VIF values do not exceed the critical value of 10, which indicates that multicollinearity did not influence the results (Hair, Black, Babin & Anderson, 2010).

Model A

Model A shows the results regarding hypothesis 1, which states that uncertainty avoidance has a negative influence on the degree of disclosure by auditors. In accordance with this hypothesis, the results show a negative coefficient. However, this coefficient is not found to be significant (p = 0.624). For that reason, hypothesis 1 cannot be accepted, which means that there is no conclusive evidence that uncertainty avoidance leads to a lower degree of disclosure of key audit matters.

Model B

In model B, hypothesis 2 is processed, which states that power distance is expected to have a negative influence on the degree of disclosure by auditors. The results of this hypothesis show that the coefficient is negative, in accordance with the hypothesis. In addition, this coefficient is significant at a 1% level (p = 0.000). Consequently, hypothesis 2 is accepted, which means that power distance leads to a lower degree of disclosure of key audit matters.

Model C

Model C presents the results regarding hypothesis 3, where it is expected that individualism has a positive influence on the degree of disclosure by auditors. Again, the coefficient is in accordance with the hypothesis, as the sign of the coefficient is positive. Moreover, the coefficient is significant at a 1% level (p = 0.000). Therefore, this hypothesis is accepted, which means that individualism, using the in-group variant, leads to a higher degree of disclosure of key audit matters.

(28)

27 | Danny Bos Model A

H1

Model B H2

Model C H3

Model D H4

Model E H5

Model F H6

Model G

Variable Expected sign Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

Intercept -1405.150*** -476.377 -2035.033*** -694.672* -1473.205*** -1332.567*** -1202.398**

UAV - -0.834 0.290

PDT - -16.450*** -11.204***

IND + 8.589*** 3.839

SEC - -5.848***

MAS - 5.089*** 4.822**

LTO + -3.339** 2.984*

SIZ + 135.603*** 146.528*** 146.165*** 140.160*** 124.157*** 139.587*** 133.514***

RSK + 355.757** 305.835** 292.473** 338.450** 394.689*** 360.624*** 324.985**

WLT - -0.015*** -0.026*** -0.020*** -0.033*** -0.014*** -0.016*** -0.023***

ToA - -182.271*** 64.668 -233.992*** -0.768 -172.102*** -172.727*** -39.104

N 575 575 575 575 575 575 575

R-squared 0.2037 0.2877 0.2716 0.2825 0.2138 0.2080 0.2970

F-statistic 25.88*** 33.59*** 31.65*** 31.06*** 25.07*** 25.16*** 25.80***

Highest VIF 1.66 1.32 1.17 1.48 1.25 1.17 5.12

Mean VIF 1.33 1.20 1.11 1.26 1.13 1.14 2.47

***, ** and * coefficients are statistically significant at a 1%, 5% and 10% confidentiality level, respectively (tested two-sided, using robust standard errors). See table II for the definitions of the variables.

Table V: Regression results

National culture and the degree of disclosure by auditors within the extended auditor’s report

National culture and the degree of disclosure by auditors within the extended auditor’s report

National culture and the degree of disclosure by auditors within the extended auditor’s report

Table of Contents