Amsterdam Business School
Audit Specialization: Beyond Industries
Name: Taeke Willemsen
Student number: 10092587
Date: 19 June 2015
Word count: 12.203
Thesis supervisor: Dr. Jeroen van Raak
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
2
Statement of originality
This document is written by student Taeke Willemsen, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
This study examines the relationship between account specialization on the one hand and respectively audit quality and audit pricing on the other hand. Account specialization is a form of audit specialization in which auditors specialize themselves in specific financial statement accounts. This is a relative new field of study which has not been studied before in contrast to other forms of audit specialization like industry specialization. The independent variable account specialization is measured on the basis of three balance sheet accounts (goodwill, inventory and accounts receivable). The dependent variables audit quality and audit pricing are measured on the basis of restatements and the natural logarithm of the total audit fees respectively. The sample size consists of 18.938 U.S. ‘company-years’, in a time range of ten years from 2004 until 2013. The following conclusions can be drawn from this study. First, this study find no evidence for a certain relationship between account specialization and audit quality. Secondly, this study concludes that the deployment of audit account specialists results in lower audit fees.
4
Table of Contents
1. Introduction and Motivation p. 5
1.1. Introduction p. 5
1.2. Motivation p. 6
2. Literature Review and Hypotheses p. 8
2.1. Audit Quality p. 8
2.1.1. What is Audit Quality? p. 8
2.1.2. Importance of High Audit Quality p. 9
2.2. Specialization p. 10
2.2.1. Concept of Audit Specialization p. 10
2.2.2. Industry Specialization p. 10
2.2.3. Account Specialization p. 11
2.3. Impact of Account Specialization & Hypotheses p. 12 2.3.1. Hypothesis 1: Impact of Account Specialization on Audit Quality p. 12 2.3.2. Hypothesis 2: Impact of Account Specialization on Audit Pricing p. 13
3. Research Design p. 16
3.1. Account Specialization Measurement p. 16
3.2. Models regarding Audit Quality and Audit Pricing p. 17
3.3. Sample Selection Process p. 21
4. Findings p. 23
4.1. Descriptive Statistics p. 23
4.2. Correlations p. 24
4.3. Regression Models p. 27
5. Conclusion and Discussion p. 32
5.1. Conclusion p. 32
5.2. Discussion p. 33
5
1. Introduction and Motivation
This first introduction chapter is divided in two parts. The first part provides a first impression about the subject matters, and the second part serves as an explanation about the motivation for this study.
1.1 Introduction
The auditor is the link between the management of an organization and their stakeholders (Pong and Whittington, 1994). Auditors are verifying the financial reporting data provided by the management of an organization on criteria like reliability, correctness and completeness. The work performed by the auditors is commissioned by the shareholders of an organization, because they want to know to which extent they can rely on the financial reporting data provided by the management of an organization. In this way, the auditor operates as a ‘gatekeeper’, to prevent information asymmetry between the management and the shareholders (Ghosh and Moon, 2005). The work performed by the auditors is a solution for the problems resulting from the agency theory. According to this theory, shareholders of an organization are identified as the principles in the organization and the management is identified as the agent in the organization. The agents are driven by a twofold interest; because they want to maximize both their own interests and the interests of the shareholders. In this situation where possibly conflicting interests are met, auditors have to ensure that the shareholders can rely on the reported financial statements by the management (Wallace, 2004).
For the sake of all the involved parties, it is important that the audit performed by the auditor is a high quality audit. This means that the audit contains as little as possible mistakes to prevent situations wherein users take the wrong decisions based on incorrect information. DeAngelo (1981) explains this principle in a very comprehensive way. In the second chapter, this study continues in an extensive manner on this principle. It is also important for the preparers of financial information and for the auditors themselves that the audit is a high quality audit, as Asthana et al. (2010) shows regarding the case of audit firm Arthur Andersen and their audit client Enron. This study elaborates on Asthana et al. (2010) in the second chapter.
In short, it is not daring to state that audit quality is very important and that high audit quality is very desirable. In that context, it is also not surprising that research has been done to factors which could increase the quality of auditing. Francis (2004) designates different factors which influence the quality of auditing. The most important factors which Francis (2004) appoints are firm size, industry specialization and office characteristics. This study focuses on the
6
specialization factor. Most of the published studies which have written about the specialization factor within audit quality focuses on the relationship between industry specialization on the one hand and audit quality or audit pricing on the other hand. However, it is also possible to study for audit specialization on other measures, like audit specialization in intangibles valuation. In other words, some kind of ‘non-industry audit specialization’ or ‘account specialization’. This study uses the ‘account specialization’ term to mention this from now on. Bamber and Ramsay (1997) is one of the papers which studies for audit specialization beyond industrialism; they show that review level on specific kinds of errors is a much more effective form of audit reviewing than all-encompassing reviews. This study elaborates on Bamber and Ramsey (1997) in the next chapter.
As mentioned above, this study analyses the effects of account specialization on audit quality and audit pricing. This is based on prior studies, which study the effects of industry specialization on audit pricing and audit quality. The main difference between prior studies and this study is that this study replaces industry specialization by account specialization; this study looks beyond industry specialization.
This study contains four more chapters. The second chapter discusses related literature and presents the hypotheses. The third chapter describes the quantitative research method of this study. Chapter four presents the findings of the research. The last chapter of this study contains a conclusion and a discussion.
1.2 Motivation
As shown in the prior paragraph, lots of research has been done to industry specialization. In contrast to industry specialization, there is little published regarding account specialization. This is quite strange, because account specialization is also a form of audit specialization which could have an impact on the quality and pricing of audits. However, the exact impact of account specialization on audit quality and audit pricing is still unknown. Because of this, it is very interesting to study this subject because in this way it is filling a gap in the audit literature.
This study showed the academic relevance of this study already above. Furthermore, this study discusses in this part some points regarding the relevance of this paper from a societal point of view. The relevance from a societal point of view is a little bit more complicated to explain, because the society itself does not profit directly from this research since this paper only studies for a possible factor which could improve audit quality. However, several parties could profit from this paper, like audit clients, audit firms and share- and stakeholders from audit clients. All these parties benefit from higher audit quality, so they should profit from studies
7
which are studying for factors which could increase audit quality, like this study. Because some of these parties represent the society, society could profit in an indirectly manner from this research. It is possible to explain this in much more detail. This paper namely provides audit firms guidance how they could improve quality. Let’s presume that this paper generates enough evidence to state that account specialization impacts audit quality in a positive way. If audit firms make use of this information and apply this in a well way, this could improve their audit quality. Arguably higher audit quality increases the reputation of audit organizations in society, and society (represented by stakeholders) can make better decisions based on more verified information because of the increased audit quality. In this way, society could profit from this study. Besides that, audit clients and all their related parties like shareholders and stockholders, know that the audit quality is expected to be higher if audit specialists are involved in the audit process. Possibly this is also an explanation for higher audit pricing, if audit specialists are involved in the process.
8
2. Literature Review and Hypotheses
This chapter consists of a comprehensive literature review and the presentation of the hypotheses of this study. This literature review explains in first instance the concept of audit quality, including an explanation regarding the relevance of high audit quality reporting. After that, this study expands on the concept of specialization. This study explains the notion of specialization in general and this is subsequently applied on auditing, which results in the introduction of the terms ‘industry specialization’ and ‘account specialization’. The extensive explanation of those two notions is recorded in the last two part sections of the second paragraph in this chapter. The third paragraph provides an explanation regarding the way in which the impact of audit specialization is measured. As last, this study presents two hypotheses about the expected relationships of this research between account specialization on the one hand and audit quality and audit pricing on the other hand.
2.1. Audit Quality
In the first paragraph this study provides an overview of audit quality. Especially the meaning of audit quality and the relevance of high audit quality is explained.
2.1.1. What is Audit Quality?
Francis (2004) reviews all the published empirical literature regarding audit quality over the past 25 years, in order to determine what is currently known about audit quality. Therefore, Francis (2004) is very appropriate to determine the exact significance of audit quality. Often audit quality is related to the number of audit failures. Audit failures are measured on the basis of sanctions, litigation rates, business failures and earnings restatements. Supposed is that the lower the number of audit failures are, the higher the quality of audit is. This paper mainly focuses on the parts of Francis (2004) which discusses the determinants of audit quality. Francis (2004, p. 352) states that “it is difficult to assess audit quality ex ante because the only observable outcome of the audit is the audit report”.
However, Francis (2004) hypothesizes that there are some ‘differences’ in audit quality, which can be detected by comparing different audit groups and classes. The most important ‘differentiation waves’ are already shortly mentioned in the introduction part of this study. This study would only like to pay some extra attention to the last determinant mentioned by Francis (2004), the industry expertise determinant. Several studies with different arguments state
9
unanimous that industry expertise has a positive impact on audit quality. All those arguments are discussed extensively further on in paragraph 2.2.3.
2.1.2. Importance of High Audit Quality
For a complete understanding of this study and to be able to interpret this study in a wider perspective, it is important to get knowledge about the importance of high audit quality. DeAngelo (1981) mentions the importance of auditing in a very convincing way. DeAngelo (1981) argues that the society has to trust on the role of the auditors. Although it is difficult to provide an exact definition of society, it is justifiable to state that society is represented by all the involved stakeholders of an organization. The financial statements of organizations like banks, insurance companies and non-governmental organizations are audited by audit firms. Several stakeholders use the audit information to make financial decisions related to the organization. For example, consumers use the audit information when deciding on buying products or making use of services. If auditors present an audit report without an unqualified opinion, which means that the presented financial statements by the management do not reflect reality by whatever reason, there is a chance that the reputation of the organization among stakeholders decreases which can result in, referring to the example above, reduced sales to consumers. So it is very important that all the audits are from high quality to prevent incorrect stakeholder reactions. Stakeholders expect high quality audit statements composed by auditors on which they can rely.
Next to that, audit quality is also important for the reputation of the auditors. Audit scandals, like possibly purposely wrongly provided audit statements, decrease the reputation of audit partners and their employers, the audit firms. This is very dangerous, as the situation with audit firm Arthur Andersen and their audit client Enron has proven. After that scandal, the reputation of audit firm Arthur Andersen decreased so much that most of their audit clients moved away from them because of a lack of confidence in the audit practices of Audit Andersen. In the end, this resulted in the bankruptcy of audit firm Arthur Andersen (Asthana et al., 2010). On the other side, Asthana et al. (2010) shows that the confidence of share- and stakeholders increase if the financial statements of organizations are audited by auditors who have a reliable reputation. Concluding, the statement that audit quality is important for the auditor reputation is quite legitimate.
10
2.2. Specialization
This part of the literature review provides an explanation about the role of audit specialization. First this study shows the general concept of specialization. After that, the concept of specialization is applied on auditors and audit firms. This results in the introduction of the notions industry specialization and account specialization.
2.2.1. Concept of Audit Specialization
Specialization is the process in which someone changes his or her knowledge from general knowledge in a wide range of subjects to extensive knowledge in a specific subject. Regarding industry specialization, “industry specialists are auditors whose training and experience are largely concentrated in a particular industry” (Fleming et al., 2008, p. 391). Solomon et al. (1999, p. 192) defines industry specialists “as auditors who are designated by their firms and whose training and practice experience largely are in a particular industry”, largely in line with the definition provided by Fleming et al. (2008, p. 391) presented above.
Let's take a look on audit specialization at the audit firm level. Audit firms can make the decision to become an industry specialist by only accepting audit clients within a specific sector, or to specialize themselves in certain balance sheet items like goodwill valuation or specific profit and loss statement items like depreciation issues within sectors where items like this have a big influence on the total value of the balance sheet of income- and loss statement. So it is possible to split up the audit specialization up into an industry specialization part and into an account specialization part, as an overarching term for the two examples just mentioned regarding goodwill valuation and depreciation issues. In the following parts this study expands on those two concepts.
2.2.2. Industry Specialization
Solomon et al. (1999) states that the concept of industry specialization is now widely embraced by audit firms to increase audit quality. Recent changes in audit practice have led to more substantive roles for specialized industry knowledge. Most of the audit firms are now structured based on industry lines, and most of their auditors are now industry-specified and specialized in specific audit industries. Auditor industry knowledge can be collected via specialized indirect experiences like trainings and via specialized direct experiences like only working on audit engagements in a specified industry. The concept behind this is that the more clients the audit firm audits in a specific industry, the more opportunities there are for auditors to gather industry
11
expertise within the audit firm from other auditors based on their experiences and gathered knowledge. This should result in more accurate audit judgments.
Fung et al. (2009) continues on the just described industry specialization concept by Solomon et al. (1999). Fung et al. (2009) state among others that client-specific knowledge is necessary to conduct high quality audits. Getting knowledge about the industry wherein a client operates is one of the most important aspects of client-specific knowledge which is necessary to conduct high quality audits. Inexperienced auditors have to rely more on information provided by the client which makes the audit quite more vulnerable. Moreover, Fung et al. (2009) argues that auditors who have much specific industry knowledge are more likely to understand the business of the client. As a consequence of this, auditors with specific industry knowledge are better able to detect misrepresentations and irregularities.
2.2.3. Account Specialization
Most of the published studies focus on the relationship between industry specialization on the one hand and audit quality or audit pricing on the other hand. But it is also possible to study for audit specialization on other measures. Bamber and Ramsay (1997) is one of the papers which studies for audit specialization beyond industries. They show that review level on specific kinds of errors is a much more effective form of audit reviewing than all-encompassing reviews. In many audit firms it is customary that audits are reviewed by higher-level auditors, based on the idea that audit reviews increase the quality of the audit itself. There are lots of possibilities how to review an audit. Bamber and Ramsey (1997) argues that focusing on specific kinds of errors, like errors which are frequently made or errors which are difficult to detect, is in this way much more effective than all-encompassing reviews. This is a good example of audit specialization beyond industries which increases the audit quality.
Nevertheless, much more research has been done to industry specialization in comparison to non-industry specialization like account specialization, mentioned before in this study. The aim of this paper is to look for a relationship between account specialization on the one hand, and audit quality and audit pricing on the other hand. It is interesting to study for this, because little research has been done until now regarding this field of study, and to look for specialization beyond industries in this way.
12
2.3. Impact of Account Specialization & Hypotheses
The third paragraph of the literature review discusses the impact of account specialization on the one hand on audit quality and audit pricing on the other hand. This paragraph includes an explanation of why this study measures the effects of account specialization on audit quality and audit pricing. This study presents also two hypotheses about the expected results of the research in this paragraph.
2.3.1. Hypothesis 1: Impact of Account Specialization on Audit Quality
In this part this study discusses in first instance the published literature regarding the impact of industry specialism on audit quality to develop a hypothesis for this study. For a lack of existing literature regarding the impact of account specialization on respectively audit quality and audit pricing, this study focuses in first instance on industry specialization and the relationship with audit quality. Subsequently, this study tries to translate the results of the presented literature review into expectations regarding account specialization.
Davies et al. (2007) focuses on the relationship between industry specialization on the one hand and audit quality and audit pricing on the other hand. With regard to this paper, this is quite unique because most of the papers do not connect audit quality and audit pricing in the same study. Davies et al. (2007) identifies a positive association between industry specialization and audit quality, but does not identify a certain relationship between industry specialization and audit pricing. Nevertheless, thanks to the unique setting of this research regarding the two dependent variables (audit quality and audit pricing), it is appropriate to use the study from Davies et al. (2007) as an example for this study. In this study, industry specialization is replaced in this study by account specialization.
Regarding the dependent variables, let’s first focus on the dependent variable ‘audit quality’. Balsam et al. (2003) examines the relationship between industry specialization and audit quality on the basis of earnings quality measures. Balsam et al. (2003) make a comparison between the extent of discretionary accruals and earnings response coefficients audited by auditors who are identified as industry specialists and the extent of discretionary accruals and earnings response coefficients audited by auditors who are not recognized as industry specialists. In line with their expectations, they come to the conclusion that the level of discretionary accruals is lower and the earnings response coefficient is higher if audit clients are audited by industry specialists in comparison with audit clients which are audited by auditors who are not identified as industry specialists.
13
Several studies support and complement the conclusions of Balsam et al. (2003). Elder and Zhou (2002) concludes the same as Balsam et al. (2003); clients who are audited by auditors who are identified as industry specialists show smaller abnormal accruals which results in constrained earnings management. This results in much more credible financial statements from audit clients. Krishnan (2004) expands above mentioned literature by stating that auditors who are recognized as industry specialists are better capable to publish bad news regarding the earnings of audit clients in time than auditors who are not recognized as industry specialists.
In this paragraph, it is also appropriate to refer to Solomon et al. (1999) and Fung et al. (2009), already mentioned earlier in this chapter. In contrast to the studies mentioned in the foregoing lines, those two studies do not really come up with strong empirical evidence, but their reasoning is still very interesting to keep in mind while developing the hypotheses. Solomon et al. (1999) state that more accurate audit judgments are a result of industry specialization, because in such a scenario there are more opportunities to gather industry expertise within the audit firm by making use of the knowledge and experiences of their audit colleagues. Moreover, Fung et al. (2009) argue that inexperienced auditors have to rely more on information provided by the audit client which makes the audit less credible. They have to do this because of a lack of own knowledge and experience in a specific sector.
Based on the literature presented above, this study presents the following hypothesis regarding account specialization and audit quality:
H1: Account specialization results in higher audit quality
The approach of this hypothesis is that account specialization results in higher audit quality. This is based on the results from other studies which concludes that industry specialization results in higher audit quality, as showed in the foregoing paragraphs. The first intuition is that this would not be different for account specialization. If auditors are always focusing on the same balance sheet- or profit and loss accounts they create specialized knowledge. This knowledge could result in higher audit quality, just as in the case of industry specialism.
2.3.2. Hypothesis 2: Impact of Account Specialization on Audit Pricing
There are many studies published which show details regarding the relationship between audit specialization and audit fees. Francis et al. (2005) gathers evidence for the theorem that higher audit fees are provided for audit engagements performed by audit firms which are leader in the related industry. This implies that the audit quality is higher, because the fee paid by the client is
14
higher. This statement hypothesizes the positive relationship between high audit fees, high audit quality and industry leadership. Supportive for the relationship just mentioned is that Francis et al. (2005, p. 354) states that an industry leader in the United States has a fee premium relative to other auditors participating in the same industry. DeFond et al. (2002) and Ferguson et al. (2003) confirm this impression. They state that the industry leaders in respectively Australia and Hong Kong earn a fee premium relative to other audit firms.
Furthermore, Ferguson and Stokes (2002) gathers evidence for the theorem that the Big N audit firms are better capable to earn audit fee premiums in contrast to non-Big N audit firms, due to the employability of audit specialists. With regard to this paper it is possible to conclude that audit specialization results in higher audit quality which creates a situation where it is possible to earn audit fee premiums, as showed in other papers like Francis et al. (2005).
Mayhew and Wilkins (2003) also studies the relationship between audit firm industry market share and audit pricing. However, they study this relationship from another perspective. They hypothesize that an increasing industry market share should result in efficiency improvements, which should be a condition for competitive advantages regarding costs and services. They conclude that if the audit firm industry market share increases without a differentiation in the market share as a whole, the audit fee decreases. This implies that the client profits from the cost savings of the audit firm. This is a direct result of the fact that the audit firm is not capable to differentiate itself from its competitors, which have a smaller audit firm industry market share. However, a new situation arises if an audit firm has a considerably higher industry market share than their competitors. In that case, the audit firm is capable to earn fee premiums. This suggests that audit firms are only able to earn fee premiums, if their industry market share is considerably higher than the industry market share of their competitors. Summarizing, Mayhew and Wilkins (2003) provide an interesting theory about the proposed relationship between audit pricing and industry specialization: an industry specialized audit firm is only capable to demand higher audit fees if their industry market share is high relative to their competitors. Otherwise, the fact that an audit firm is an industry specialist has no effect on the audit fees up to Mayhew and Wilkins (2003).
Based on the literature studied and presented above, this study formulates the following hypothesis regarding the impact of audit specialization on audit pricing:
15
The approach of this hypothesis is quite the same as the first hypothesis; account specialization results in higher audit pricing. This is based on the results from other studies which concludes that industry specialization results in higher audit pricing, as showed in for example Francis et al. (2005). Although there are also studies published which show other results regarding audit specialization and audit pricing like Mayhew and Wilkins (2003), for this study the results and arguments in line with the paper of Francis et al. (2005) are more convincing. Mayhew and Wilkins (2003) argues that audit firms are only capable to earn fee premiums if their industry market share is considerable higher than the industry market share of their competitors. However, an audit firm can only be considered as an industry specialist if their industry market share is actually considerable higher than the industry market share of their competitors. Otherwise, their industry market share is unsufficient to be distinguished as an industry specialist. This argumentation attenuates the reasoning of Mayhew and Wilkins (2003). Therefore, this study prefers to follow the results of studies in line with Francis et al. (2005) while determining the hypotheses regarding audit pricing.
The first intuition is that this would not be different for account specialization, because the way of argumentation can also be applied on audit pricing. Francis et al. (2005) gathers evidence that higher audit fees are provided for audit engagements which are performed by audit firms which are leader in the related industry. It is interesting to study if there are also higher audit fees provided if there are auditors engaged in an audit which are for instance specialized in goodwill valuation and who are deployed at an audit client in which the goodwill valuation is a major part of the valuation of the whole audit client firm. If that is the case, it is justifiable to state that account specialization results in higher audit fees.
16
3. Research Design
In this chapter this study provides an overview of the research design of this study. The first paragraph explains in particular the selection process of three financial statement items for the account specialization measurement. Subsequently, the second paragraph introduces the two research model formulas regarding audit quality and audit pricing. Finally, the last paragraph of this chapter provides an explanation about the applied sample size in the research.
3.1. Account Specialization Measurement
The research method of this thesis is a quantitative archival database research. This study prefers a quantitative archival database research because of the high availability of data. Moreover, many comparable studies make also use of a quantitative archival database research, which increases the comparability of the results of this study with other studies. This study tests the hypotheses presented in the prior chapter on the basis of the quantitative data collected using a regression model. This study relies on prior research to empirically measure the theoretical constructs in this paper in an appropriate way. On the basis of that, this study empirically measures the following terms: ‘account specialization', 'audit quality' and 'audit pricing'. Account specialization is the independent variable in this study. Audit quality and audit pricing are the dependent variables in this study. In this paragraph this study elaborates on the measurement of the independent variable ‘account specialization’. Based on prior literature, this study selects three major balance sheet accounts to measure the ‘account specialization’ variable, namely goodwill, inventory and accounts receivable.
Mian and Smith (1992) states that accounts receivable are a substantial fraction of all the corporate assets in the United States. For this reason, accounts receivable is a relevant measure for account specialization, because it covers a considerably part of the total corporate assets in the United States. Furthermore, Mian and Smith (1992) observes a large variety in the accounts receivable management policies of companies. This could be an indication of desirable specific knowledge of auditors regarding accounts receivable, because of the large variety in management policies. Quite the same arguments are applicable to the inventory account. Arcelus and Trenholm (1989) state that inventory contains a comprehensive part of the total assets in most firms. Moreover, they ascertain a considerable difference in inventory valuation methods on the one hand and the guidance provided by standard setters on the other hand. Both facts offer enough evidence to consider inventory as one of the measures of account specialization, looking to the stated impact of inventory on the total assets and the possibly complications for auditors with regard to inventory auditing due to the differences in inventory valuation methods and the
17
guidance provided by the standard setters. Regarding goodwill, Dagwell et al. (2007) states that goodwill accounting is still a controversial issue. The accounting treatment for goodwill is still very challenging for preparers of financial statements. Therefore it is likely that the audit of goodwill is also very challenging for auditors and could be an outstanding example of account specialization in this way.
This study determines account specialization per Micro- or Metropolitan Statistical Area in the United States (Francis et al., 2005). This study selects data from the fiscal years start date 1 January 2004 until end date 31 December 2013. The total value of respectively accounts receivable, inventory and goodwill of all listed US companies in the several Micro- or Metropolitan Statistical Areas is determined in US Dollar during a period of ten years. This study subsequently determines which percentage of accounts receivable, inventory and goodwill is audited by the same audit firm. To determine if an audit firm can be identified as a specialist, this study refers to Neal and Riley (2004). According to Neal and Riley (2004), an audit firm can be recognized as an industry specialist if the audit firm has a market share of more than 20% in a specific industry. To measure account specialization, this study applies the same percentage of 20% as showed by Neal and Riley (2004). This study acknowledges that is maybe not the best way to just translate this model to account specialization and assuming the percentage of 20% to determine accounts specialists. However, given the lack of guidelines regarding specific account specialization measurements, this study argues that it is a legitimate option to measure account specialization on this way because the percentage is a directive for specialization which can be applied on both industry specialization and account specialization. If an audit firm audits more than 20% of the total value of one of the three balance sheet items in a Micro- or Metropolitan Statistical Area during ten years, this study allocates a dummy variable value of 1 to the ‘company-year’ which is audited by an audit firm specialist. Otherwise, if a ‘company-year’ is audited by an audit firm which is not identified as a specialist in the way described above, this study allocates a dummy variable of 0 to that ‘company-year’.
3.2. Models regarding Audit Quality and Audit Pricing
In this paragraph this study shows the relationship between the dependent variables ‘audit quality’ and ‘audit pricing’ on the one hand and the above discussed independent variable ‘account specialization’ on the other hand.
18
The final model regarding ‘audit quality’ is partly inspired by prior studies which studied for the relationship between audit quality and industry specialization, like Davies et al. (2007) and Fleming et al. (2008). In the model of this study, industry specialization is replaced by account specialization. Following Fleming et al. (2008), audit quality is determined on the basis of audit restatements. Financial restatements are an indication of poor audit quality, because a restatement is necessary to create financial statements which give a true and fair view of the economic situation of the company (Fleming et al., 2008). Moreover, Davies et al. (2007) and Fleming et al. (2008) provide some specific control variables which this study also records in the model. This results in the following model:
Moving on to the second model with the dependent variable ‘audit pricing’. Just as in the formula presented above, the ‘account specialization’ factor is the independent variable which influences the dependent variable ‘audit pricing’ together with a set of control variables. The formula is partly inspired by Casterella et al. (2004) and Craswell et al. (1995), which studies for the relationship between audit pricing on the one hand and industry specialization on the other hand. Both studies uses the natural logarithm of audit fees as representative of the dependent variable ‘audit pricing’. This study replaces industry specialization by account specialization in the model of this study. Furthermore, both studies provide some specific control variables which this study also records in the model. This results in the following formula:
19
In the last part of this paragraph this study provides an overview of the exact meaning and definition of the different variables in the two model formulas.
Dependent variables: Audit Quality and Audit Pricing
REST: Restatement
- dummy variable: 0 if there is a financial restatement in a ‘company-year’; 1 otherwise
LN_FEES: Natural logarithm of the total audit fees paid by the clients to the audit firms
Independent variables: Account Specialization
GWILL_SPEC: The audit firm is identified as goodwill specialist in a Micro- or
Metropolitan Statistical Area in the way as described earlier in this chapter
- dummy variable: 1 if the audit firm of a ‘company-year’ is identified as goodwill specialist; 0 otherwise
INV_SPEC: The audit firm is identified as inventory specialist in a Micro- or
Metropolitan Statistical Area in the way as described earlier in this chapter
- dummy variable: 1 if the audit firm of a ‘company-year’ is identified as inventory specialist; 0 otherwise
REC_SPEC: The audit firm is identified as accounts receivable specialist in a Micro- or
Metropolitan Statistical Area in the way as described earlier in this chapter
- dummy variable: 1 if the audit firm of a ‘company-year’ is identified as accounts receivable specialist; 0 otherwise
LN_TOT_GWILL: Natural logarithm of the total value of goodwill of all
‘company-years’
LN_TOT_INV: Natural logarithm of the total value of inventories of all
‘company-years’
LN_TOT_REC: Natural logarithm of the total value of accounts receivable of all
‘company-years’
INT_GWILL: Interaction variable
- variable to verify the interaction and relationship between the deployment of goodwill specialists and the natural logarithm of the total value of goodwill of all ‘company-years’
20
INT_INV: Interaction variable
- variable to verify the interaction and relationship between the deployment of inventory specialists and the natural logarithm of the total value of inventories of all ‘company-years’
INT_REC: Interaction variable
- variable to verify the interaction and relationship between the deployment of accounts receivable specialists and the natural logarithm of the total value of the accounts receivable of all ‘company-years’
Control variables:
INDUS_SPEC: Audit firm is identified as an industry specialist
- dummy variable: 1 if a ‘company-year’ is audited by an audit firm which is identified as an industry specialist; 0 otherwise
- the audit firm is recognized as an industry specialist if the audit firms audits more than 20% (Neal and Riley, 2004) of the total assets of all ‘company-years’ in a certain industry based on the Global Industry Classification Standards (GICS) for industry categorization
BIG-4: Audit firm is one of the Big-4 audit firms
- dummy variable: 1 if a ‘company-year’ is audited by an audit firm which is one of the Big-4 audit firms; 0 otherwise
AUD_TENURE: The total tenure in years wherein an audit firm audits a ‘company-year’
INDEP: Degree of independence of the audit firm relating to a ‘company-year’
- formula: total non-audit fees / (total audit fees + total audit related fees)
AUD_OPINION: The auditor opinion of the auditor regarding a ‘company-year’
- dummy variable: 0 if the auditor provides a ‘qualified opinion’ to a ‘company-year’, 1 if the auditor provides an ‘unqualified opinion with additional language’ to a ‘company-year’
LN_TOT_ASSETS: Natural logarithm of the total assets of all ‘company-years’
ROA: Return on Assets of all ‘company-years’
- formula: total net income / total assets
LEV_RATIO: Leverage Ratio of all ‘company-years’
21
LOSS: ‘Company-year’ reported a loss in the income statement
- dummy variable: 1 if a ‘company-year’ reported a loss in the income statement; 0 otherwise
MSA: Micro- or Metropolitan Statistical Area in which a company is registered
- dummy variable: 1 if the company is registered in a certain Micro- or Metropolitan Statistical Area; 0 otherwise
INDUSTRY: Industry in which a company is classified following the Global Industry
Classification Standards (GICS) for industry categorization
- dummy variable: 1 if the company is registered in a certain industry; 0 otherwise
YEAR: Related fiscal year to a ‘company-year’
- dummy variable: 1 if the company is registered in a certain fiscal year; 0 otherwise
3.3. Sample Selection
This study uses data from the fiscal years start date 1 January 2004 until end date 31 December 2013. This study does not select the most recent fiscal year 2014, because possibly not all data are already known regarding the fiscal year 2014. In particular regarding the audit quality, there is a considerable chance that possibly necessary restatements regarding the financial statements of 2014 are not observed yet. This observation is also applicable to a lesser extent for the last financial years in the time period of this study. However, it is not possible to control for that, because a possibly less number of restatement could also be a result of for example better audit quality.
A period of ten years is a quite long time which provides much information and data. This increases the usefulness of the results of this study. This study collects all data which this study needs for performing the research via WRDS. This study makes use of the Compustat- and Audit Analytics databases in WRDS. This study selects all listed companies in the United States in first instance. This results in 59.786 available ‘company-years’ from Compustat and 40.818 available ‘company-years’ from Audit Analytics. The term ‘company-year’ is a combination of the company name and the related fiscal year. This study merged the data from both datasets in Microsoft Excel. This results in 40.818 ‘company-years’ which this study implement in SPSS. This study ascertains 17.589 missing values in the different variables of the ‘company-years’ which this study needs to perform the research. This study removes all those ‘company-years’ from the sample. Furthermore, this study removes 1.545 outliers regarding the scale variables. For removing the outliers, this study determines first all the Z-values of the scale
22
variables of all ‘company-years’. After that, this study removes all ‘company-year’ Z-values below -3 or above the 3 (Comrey, 1985).
Subsequently, this study removes another 133 ‘company-years’; companies which are at the moment of publishing this study inactive and ‘company-years’ which were unaudited or did not receive any audit opinion. This study removes also 1 ‘company-year’ which received a qualified opinion. In this way it is possible to create a dummy variable of the variable Auditor Opinion. Furthermore, this study removes another 1.092 ‘company-years’ because that companies are not registered in a micro- or metropolitan statistical area in the United States. All the micro- or metropolitan statistical areas in the United States are determined on the basis of the information of the United States Government Census (Francis et al., 2005). There are two options why a company is not classified in an United States Micro- or Metropolitan Statistical Area: the U.S. company is not officially located in the United States or the location of the company in the United States is not part of a micro- or metropolitan area. As last, this study removes 1.519 ‘duplicate company-years’ which were audited by two or more audit firms, because in that case it is very complicated to determine the exact portion of the three balance sheet items which is audited by an audit firm. This results in a total remaining sample size of 18.939 ‘company-years’.
23
4. Findings
In this chapter this study presents the results and the findings of the research. The first paragraph presents the descriptive statistics. The second paragraph discusses the different correlations between the variables. The last paragraph discusses the results of the two regression models regarding audit quality and audit pricing. Every paragraph has the same design; first a presentation of the statistics followed by an interpretation of the consequences of the results for the study subject.
4.1. Descriptive Statistics
On the next page, Table 1 presents the results of the descriptive statistics. In this part, this study discusses some interesting and important results regarding the descriptive statistics. Looking to the minima and the maxima, the results of the Return On Assets variable and Leverage variable are quite notable. A negative Return on Assets is an indication of a negative net result. The maximum Return on Assets is notably high. However, this is possible in case of a high profit and a low value of total assets. Also the maximum of the leverage ratio is strikingly high. Regarding the means, the mean of the restatement variable is 0.890. Since the 1 dummy value indicates that there has been no restatement made to the financial statements, there are much more ‘company-years’ without a restatement than with a restatement. Furthermore, the means regarding the specialist variables (goodwill, inventories, receivable and industry) are all around the 50%. This means that approximately half of the auditors involved in the audit of ‘company-years’ is recognized as a specialist. Regarding the three balance sheet measures for the account specialization variables, this study concludes that the mean total value of the receivables is considerably higher than those of the goodwill and inventories. About 75% of the ‘company-years’ is audited by a Big-4 audit firm. All the audit firms are quite independent, because the proportion of the total non-audit fees in comparison with the total audit fees is only about 20%. This should have positive effects on the audit quality, because such a low independence ratio means that the audit firms are not suggestible and manipulative from a financial point of view (Ashbaugh et al., 2003). The Return On Assets ratio is negative which means that there are considerably negative results reported by the ‘company-years’ since there are more profits reported than losses (looking to the loss variable; a 0 dummy value indicates a profit) by the ‘company-years’ and the Leverage ratio is also higher than 1 which means that the total value of debt is higher than the total value of assets. Regarding the standard deviations, those of the dummy variables are not relevant. Moreover, the standard deviation of the Audit Tenure in Years, the Return on Assets and the Leverage ratio are quite high.
24
Table 1: Descriptive Statistics (N = 18.938)
Variable Minimum Maximum Mean Standard
Deviation
Restatement 0 1 0.890 0.316
Ln Total Audit Fees 7.540 16.780 13.664 1.256
Goodwill Specialist 0 1 0.540 Inventories Specialist 0 1 0.470 Receivables Specialist 0 1 0.550 Ln Total Goodwill 0 14.400 6.372 5.208 Ln Total Inventories 0 13.860 6.487 4.631 Ln Total Receivables 0 15.170 9.240 2.890
Interaction Variable Goodwill 0 14.400 4.013 5.208
Interaction Variable Inventories 0 13.860 3.489 4.870
Interaction Variable Receivables 0 15.170 5.367 5.280
Industry Specialist 0 1 0.510
Big-4 Audit Firm 0 1 0.740
Audit Tenure in Years 0 116 14.080 15.941
Independence 0 6.590 0.191 0.238 Auditor Opinion 0 1 0.330 Ln Total Assets 0 17.630 11.886 2.041 Return on Assets -273.230 1587.740 -0.008 14.818 Leverage Ratio 0 884.610 1.537 13.633 Loss Reported 0 1 0.330 4.2. Correlations
On the next page, table 2 shows the Pearson Correlation Matrix. The correlation coefficients in the table represents an indication of the underlying relationship between two variables. The higher the correlation coefficient, the higher the power of the relationship between two variables. However, if the correlation is higher than 0.8, there is possibly a problem with multicollinearity. Multicollinearity gives an indication of a possible direct relationship between two variables, which means that one variable could be predicted straightly from the other variable (Kumari, 2008). Another attention point is that a correlation coefficient is only significant if the p-value is below the 0.05 level. Otherwise, the correlation coefficient is not significant and thereby useless.
In particular, this study focuses on the relationship between the two dependent variables (restatements and the natural logarithm of the total audit fees) and the independent variables. This study concludes that there are hardly significant relationships between the dependent variable restatements and the independent variables. Moreover, the only two significant relationships (between the dependent variable restatements on the one hand and the independent variables independence and auditor opinion on the other hand) are properly weak.
Table 2: Pearson Correlation Matrix (N = 18.938)
Variable Rest Ln TAF GS IS RS Ln TG Ln TI Ln TR IG II IR IS Big-4 ATY Indep AO Ln TA ROA Lev Loss Restatement 1 -0.010 0.009 0.014 0.013 -0.007 -0.009 -0.005 0.005 0.000 0.008 -0.004 0.007 0.004 0.034¹ -0.022¹ 0.008 0.003 0.003 0.000 Ln Total Audit Fees -0.010 1 0.332¹ 0.272¹ 0.310¹ 0.469¹ 0.276¹ 0.353¹ 0.404¹ 0.295¹ 0.340¹ 0.371¹ 0.582¹ 0.367¹ 0.007 0.145¹ 0.325¹ 0.006 -0.026¹ -0.234¹ Goodwill Specialist 0.009 0.332¹ 1 0.445¹ 0.650¹ 0.227¹ 0.102¹ 0.183¹ 0.711¹ 0.351¹ 0.598¹ 0.424¹ 0.560¹ 0.186¹ 0.020¹ 0.076¹ 0.227¹ 0.011 -0.003 -0.099¹ Inventory Specialist 0.014 0.272¹ 0.445¹ 1 0.571¹ 0.165¹ 0.203¹ 0.173¹ 0.339¹ 0.768¹ 0.540¹ 0.388¹ 0.458¹ 0.221¹ 0.031¹ 0.068¹ 0.209¹ 0.018² -0.012 -0.097¹ Receivable Specialist 0.013 0.310¹ 0.650¹ 0.571¹ 1 0.176¹ 0.106¹ 0.207¹ 0.449¹ 0.440¹ 0.922¹ 0.396¹ 0.560¹ 0.220¹ -0.006 0.065¹ 0.249¹ 0.010 -0.001 -0.088¹ Ln Total Goodwill -0.007 0.469¹ 0.227¹ 0.165¹ 0.176¹ 1 0.274¹ 0.473¹ 0.670¹ 0.244¹ 0.296¹ 0.210¹ 0.270¹ 0.189¹ 0.059¹ 0.050¹ 0.415¹ 0.004 -0.005 -0.277¹ Ln Total Inventory -0.009 0.276¹ 0.102¹ 0.203¹ 0.106¹ 0.274¹ 1 0.388¹ 0.217¹ 0.588¹ 0.209¹ 0.097¹ 0.132¹ 0.163¹ 0.022¹ 0.024¹ 0.286¹ 0.001 -0.015² -0.211¹ Ln Total Receivables -0.005 0.353¹ 0.183¹ 0.173¹ 0.207¹ 0.473¹ 0.388¹ 1 0.364¹ 0.299¹ 0.465¹ 0.174¹ 0.218¹ 0.123¹ 0.041¹ -0.006 0.671¹ 0.017² -0.011 -0.339¹ Interaction Goodwill 0.005 0.404¹ 0.711¹ 0.339¹ 0.449¹ 0.670¹ 0.217¹ 0.364¹ 1 0.360¹ 0.528¹ 0.347¹ 0.419¹ 0.198¹ 0.059¹ 0.073¹ 0.359¹ 0.003 0.004 -0.214¹ Interaction Inventory 0.000 0.295¹ 0.351¹ 0.768¹ 0.440¹ 0.244¹ 0.588¹ 0.299¹ 0.360¹ 1 0.497¹ 0.306¹ 0.353¹ 0.241¹ 0.044¹ 0.059¹ 0.286¹ 0.002 -0.007 -0.161¹ Interaction Receivables 0.008 0.340¹ 0.598¹ 0.540¹ 0.922¹ 0.296¹ 0.209¹ 0.465¹ 0.528¹ 0.497¹ 1 0.377¹ 0.514¹ 0.217 0.017² 0.057¹ 0.412¹ 0.011 0.004 -0.169¹ Industry Specialist -0.004 0.371¹ 0.424¹ 0.388¹ 0.396¹ 0.210¹ 0.097¹ 0.174¹ 0.347¹ 0.306¹ 0.377¹ 1 0.600¹ 0.219¹ 0.040¹ 0.069¹ 0.255¹ 0.007 -0.016² -0.084¹ Big-4 Audit Firm 0.007 0.582¹ 0.560¹ 0.458¹ 0.560¹ 0.270¹ 0.132¹ 0.218¹ 0.419¹ 0.353¹ 0.514¹ 0.600¹ 1 0,322¹ 0,051¹ 0,116¹ 0,302¹ 0,022¹ -0,012 -0,151¹ Audit Tenure Years 0.004 0.367¹ 0.186¹ 0.221¹ 0.220¹ 0.189¹ 0.163¹ 0.123¹ 0.198¹ 0.241¹ 0.217 0.219¹ 0,322¹ 1 0,037¹ 0,071¹ 0,094¹ 0,018² -0,011 -0,197¹ Independence 0.034¹ 0.007 0.020¹ 0.031¹ -0.006 0.059¹ 0.022¹ 0.041¹ 0.059¹ 0.044¹ 0.017² 0.040¹ 0,051¹ 0,037¹ 1 -0,012 0,042¹ 0,014² 0,000 -0,070¹ Auditor Opinion -0.022¹ 0.145¹ 0.076¹ 0.068¹ 0.065¹ 0.050¹ 0.024¹ -0.006 0.073¹ 0.059¹ 0.057¹ 0.069¹ 0,116¹ 0,071¹ -0,012 1 -0,018² 0,008 0,010 0,027¹ Ln Total Assets 0.008 0.325¹ 0.227¹ 0.209¹ 0.249¹ 0.415¹ 0.286¹ 0.671¹ 0.359¹ 0.286¹ 0.412¹ 0.255¹ 0,302¹ 0,094¹ 0,042¹ -0,018² 1 -0,001 -0,158¹ -0,233¹ Return on Assets 0.003 0.006 0.011 0.018² 0.010 0.004 0.001 0.017² 0.003 0.002 0.011 0.007 0,022¹ 0,018² 0,014² 0,008 -0,001 1 -0,013 -0,052¹ Leverage Ratio 0.003 -0.026¹ -0.003 -0.012 -0.001 -0.005 -0.015² -0.011 0.004 -0.007 0.004 -0.016² -0,012 -0,011 0,000 0,010 -0,158¹ -0,013 1 0,010 Loss Reported 0.000 -0.234¹ -0.099¹ -0.097¹ -0.088¹ -0.277¹ -0.211¹ -0.339¹ -0.214¹ -0.161¹ -0.169¹ -0.084¹ -0,151¹ -0,197¹ -0,070¹ 0,027¹ -0,233¹ -0,052¹ 0,010 1
1) correlation is significant at the 0.01 level (2-tailed) 2) correlation is significant at the 0.05 level (2-tailed)
The second dependent variable, the natural logarithm of the total audit fees, shows totally other relationships with the independent variables regarding significance and power. All the relationship between the natural logarithm of the total audit fees and the independent variables are significant, except for the independent variables Independence and Return on Assets. Regarding the power of the relationship between the natural logarithm of the total audit fees and the independent variables, all relationships are relatively strong. The dependent variable audit fees has the most strong relationships with the natural logarithm of the total goodwill of all ‘company-years’ and with the Big-4 dummy variable. However, also the relationships between the dependent variable audit fees and the interaction variables of goodwill, inventories and receivables are significant and quite strong. This relationship is in particular interesting because the interaction variables indicates whether specialists are actually deployed in relation to the total value of the balance sheet items. Concluding, it is justifiable to state that there exist many strong relationships between the dependent variable of the audit fees and the independent variables, which makes this model at the first sight much more convincing and interesting than the model regarding the restatements which has much less significant and powerful relationships.
As last, this study discusses the earlier mentioned term of multicollinearity. In the Pearson Correlation Matrix, there exists one correlation coefficient which exceeds the 0.8. This coefficient represent the relationship within the interaction variable of accounts receivable. This is quite explicable, because the interaction variable represents a multiplication of other variables within the model. There also some other strong coefficient (however, below the 0.8). Most of those relate to the interaction variables. However, there also exist some strong correlation coefficients outside the interaction variables. One of those is the relationship between goodwill specialists and receivable specialists. This is quite explainable, because it is conceivable that an audit firm has more different specialists within their audit firm. The relationship between the 4 audit firms variable and the industry specialist variable is also quite explicable, because Big-4 own more resources to attract and develop specialists (Francis and Yu, 2009). As last, this study cannot imagine a certain reason for the strong relationship between the independent variables industry specialist and auditor tenure.
Moreover, this study performs a test on Variance Inflation Factors (VIF). This factor examines the degree of multicollinearity of an independent variable with other independent variables in a regression model (O’Brien, 2007). If this factor is higher than 10, issues regarding multicollinearity could arise. The test is only performed on the audit pricing model, because it is not possible to compute Variance Inflation Factors (VIF) in a logistic regression model setting (the audit quality model) in SPSS. However, this does not seems to be a problem looking to the
27
very low correlation coefficient- and significance levels in the Pearson Correlation Matrix (Table 2). Table 3 presents the results of the test on Variance Inflation Factors (VIF).
Table 3: Test on Multicollinearity – Variance Inflation Factors (VIF)
Variable VIF Tolerance (1/VIF)
Goodwill Specialist 4.161 0.240 Inventories Specialist 3.846 0.260 Receivables Specialist 15.737 0.064 Ln Total Goodwill 2.876 0.348 Ln Total Inventories 2.204 0.454 Ln Total Receivables 3.477 0.288
Interaction Variable Goodwill 5.198 0.192
Interaction Variable Inventories 4.936 0.203
Interaction Variable Receivables 16.814 0.059
Industry Specialist 1.637 0.611
Big-4 Audit Firm 2.280 0.439
Audit Tenure in Years 1.183 0.846
Independence 1.014 0.986 Auditor Opinion 1.024 0.976 Ln Total Assets 2.061 0.485 Return on Assets 1.005 0.995 Leverage Ratio 1.048 0.954 Loss Reported 1.196 0.836
Table 3 shows possibly multicollinearity issues with the ‘Receivables Specialist’ and the ‘Interaction Variable Receivables’. The Variance Inflation Factors (VIF) for the other interaction variables are also quite high. However, Robinson and Schumacker (2009) argue that the results regarding Variance Inflation Factors (VIF) of interaction variables are difficult to interpret. So the most important result of the test on Variance Inflation Factors (VIF) is the possibly issue with multicollinearity regarding the variable ‘Receivable Specialist’.
4.3. Regression Models
In this paragraph this study presents and interprets the results of the two regressions models regarding audit quality and audit pricing. First this study discusses the model regarding audit quality. This study performed a logistic regression, because the dependent variable Restatement representing the audit quality is a dummy variable. The R-square of this model is only 0.6%. This means that this model lacks explanatory power, because it indicates that the variables explain only a very small part of the model. This result was already somewhat expectable looking to the results of the Pearson Correlation Matrix. There are no issues with the significance of this model, because the p-value is 0.000 (p-value < 0.1 as threshold).
28
Table 4: Logistic Regression Analysis ( R² = 0.006 | p = 0.000 )
Variable Beta (β) Wald Sig.
Goodwill Specialist -0.063 0.427 0.513 Inventories Specialist 0.222 5.578 0.018 Receivables Specialist 0.296 2.471 0.116 Ln Total Goodwill -0.013 2.936 0.087 Ln Total Inventories 0.002 0.045 0.832 Ln Total Receivables -0.004 0.079 0.779
Interaction Variable Goodwill 0.013 1.702 0.192
Interaction Variable Inventories -0.017 2.636 0.104
Interaction Variable Receivables -0.025 1.937 0.164
Industry Specialist -0.123 4.219 0.040
Big-4 Audit Firm 0.021 0.069 0.792
Audit Tenure in Years 0.001 0.193 0.660
Independence 0.536 21.661 0.000 Auditor Opinion -0.148 9.208 0.002 Ln Total Assets 0.035 4.515 0.034 Return on Assets 0.001 0.112 0.738 Leverage Ratio 0.002 0.653 0.419 Loss Reported -0.001 0.001 0.982
Based on Table 4, this study concludes that only a few variables are significant with p-value < 0.1, namely the Inventory Specialist variable, the variable referring to the natural logarithm of the Total Goodwill, the Industry Specialist variable, the Independence variable, the Auditor Opinion variable and as last the variable referring to the natural logarithm of the Total Assets. Unfortunately, most of the variables memorized above are control variables. Therefore, it is not really relevant to discuss the possible relationships and influences on the dependent variable audit quality measured by restatements. The only two variables deriving from the account specialization measurement which are significant are the Inventory Specialist variable and the variable referring to the natural logarithm of the Total Goodwill. By the lack of significant results regarding all the other variables, this study concludes that this model provides unsufficient evidence to identify a relationship in this model. Due to this, this study concludes that on the basis of this research it is not possible to state that there is a certain relationship between the dependent variable audit quality (measured on the basis of restatements) and the independent variables regarding account specialization measurement. Therefore, this study rejects the first hypothesis. It is on the basis of this research impossible to state that account specialization results in higher audit quality. This result of this model implies that it is not possible to translate
29
the industry specialization literature directly to account specialization literature, because the result of this model contradicts all studied existing industry specialization literature which encountered a positive relationship between the deployment of audit industry specialists and the audit quality level. However, this study would like to mention again that the studied literature regarding industry specialism is something different than account specialism. This means that it is difficult to translate the results from industry specialism studies directly to an account specialism study like this.
Table 5 presents the results of the second model regarding the audit pricing. This study performs a linear regression, because the dependent variable total audit fees representing the audit pricing is a scale variable. The R-square of this model is almost 50%. This means that this model provides reasonable explanatory power, because it indicates that the variables explain a moderate part of the model. There are no issues with the significance of this model, because the p-value is 0.000 (p-value < 0.1 as threshold).
Table 5: Linear Regression Analysis (R² = 0.499 | p = 0.000 )
Variable Beta (β) T Sig.
Goodwill Specialist 0.043 1.637 0.102 Inventories Specialist -0.044 -1.719 0.086 Receivables Specialist 0.314 6.078 0.000 Ln Total Goodwill 0.070 32.348 0.000 Ln Total Inventories 0.028 13.497 0.000 Ln Total Receivables 0.061 14.547 0.000
Interaction Variable Goodwill -0.010 -3.628 0.000
Interaction Variable Inventories -0.005 -1.767 0.077
Interaction Variable Receivables -0.044 -8.769 0.000
Industry Specialist 0.032 1.956 0.050
Big-4 Audit Firm 1.271 57.262 0.000
Audit Tenure in Years 0.012 27.044 0.000
Independence -0.209 -7.665 0.000 Auditor Opinion 0.195 14.058 0.000 Ln Total Assets -0.004 -0.903 0.366 Return on Assets -0.001 -2.193 0.028 Leverage Ratio -0.001 -2.874 0.004 Loss Reported -0.073 -4.867 0.000
30
This study concludes that most of the variables in Table 5 are significant, expect for the Goodwill Specialist variable and the natural logarithm of the Total Assets (p-value < 0.1). This study focuses not on the relationship between the control variables (industry specialist, Big-4 audit firm, audit tenure in years, independence, auditor opinion, return on assets, leverage ratio and loss reported) and the dependent variable audit fees, because it is not the aim of this research to study for those relationships. However, it is very interesting to interpret the betas of the three interaction variables to measure the possible impact of account specialization on audit fees.
For all the three interaction variables (goodwill, inventories and receivables) the beta is negative. This means that the deployment of audit account specialists has a limited decreasing effect on the audit fees. Due to the significant results of this research, this study gathers enough evidence to reject the hypothesis of this study regarding audit pricing. On the basis of this research it is possible to conclude that account specialization results in lower audit pricing, in contrast to the hypothesis which argues that account specialization results in higher audit pricing. However, this result is in line with Mayhem and Wilkins (2003), which argues that audit firms are only able to earn fee premiums if their industry market share is considerable higher than the industry market share of their competitors. So a possible explanation for the result of this research is that the audit firms are not capable to earn fee premiums because their account market share is not considerable higher than the account market share of their competitors. Besides, the result of this study contradicts the results of Francis et al. (2005) which gathers evidence for the positive relationship between audit pricing and industry-leading audit firms. Again, this study would like to pay attention to the differences between industry- and account specialization.
Moreover, this study performs a robustness check to test the result presented above. On the next page, Table 6 presents the results of the robustness model regarding the audit pricing model. This study concludes that a few p-values has slightly changed, like the goodwill specialist-, inventories specialist- and inventory interaction variables; however no variables changed from significant to insignificant or the other way around with a threshold regarding the p-value of 0.1. This means that the robustness test does not lead to a change of the results; in other words the robustness test has no significant impact on the results of this study.
This study does not perform a robustness test for the audit quality model. According to this study such a test is not relevant, mainly due to two reasons. First, the R-square of this model is only 0.6%. This means that this model lacks explanatory power, because this indicates that the variables explain only a very small part of the model. Secondly, only a few variables are
31
significant in the audit quality model. A robustness test does not reduce these two weaknesses in the model. Therefore, a robustness check on the audit quality model is not necessary.
Table 6: Linear Robustness Regression Analysis (R² = 0.499 | p = 0.000 )
Variable Beta (β) T Sig.
Goodwill Specialist 0.043 1.542 0.123 Inventories Specialist -0.044 -1.697 0.091 Receivables Specialist 0.314 5.394 0.000 Ln Total Goodwill 0.070 34.349 0.000 Ln Total Inventories 0.028 14.553 0.000 Ln Total Receivables 0.061 15.412 0.000
Interaction Variable Goodwill -0.010 -3.796 0.000
Interaction Variable Inventories -0.005 -1.905 0.058
Interaction Variable Receivables -0.044 -8.388 0.000
Industry Specialist 0.032 1.858 0.065
Big-4 Audit Firm 1.271 54.182 0.000
Audit Tenure in Years 0.012 22.715 0.000
Independence -0.209 -6.670 0.000 Auditor Opinion 0.195 13.717 0.000 Ln Total Assets -0.004 -0.834 0.407 Return on Assets -0.001 -1.810 0.071 Leverage Ratio -0.001 -3.321 0.001 Loss Reported -0.073 -4.926 0.000
