Disclosure of audit partner identification : effects of the disclosure and industry specialization on the audit quality and audit fee

(1)

1

MSc Accountancy & Control, variant Accountancy

Faculty of Economic and Business, University of Amsterdam

DISCLOSURE OF AUDIT PARTNER

IDENTIFICATION

EFFECTS OF THE DISCLOSURE AND INDUSTRY

SPECIALIZATION ON THE AUDIT QUALITY AND AUDIT

FEE

Final Version

22

th

of June 2015

Zhi Kwan Cheung (10095039)

Word count: 19355

First supervisor:

Dr. Qi Yang

Second supervisor:

(2)

2 Statement of Originality

This document is written by student Zhi Kwan Cheung, 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 Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

Purpose

The purpose of this study is to provide empirical evidence regarding the relationship between the mandatory disclosure of audit partner identification and the audit quality and audit fee in the United Kingdom. Additionally, the relationship will also be examined with the presence of auditor industry expertise (as a moderating variable). The impact of the mandatory disclosure is yet unclear as there is mixed evidence whether the audit quality in fact would improve. Moreover, there is some evidence that public disclosure of auditor’s identity through signature requirements enables partner-specific measures of audit quality (Blay, Notbohm and Valencia, 2014; Carcello and Li, 2013). No study has investigated the effectiveness of audit identification disclosure with the presence of the industry specialization characteristics of auditor.

Methodology / approach

This is an empirical archival approach, using United Kingdom companies as a sample for the period 2007 through 2011. To measure the audit quality the following measures are used: abnormal accruals, accrual estimation errors, the propensity to issue qualified audit opinion, and the likelihood to meet/beat earnings benchmark. In addition, audit fees are measured to examine the costs of the implementation. Finally, the independent variable auditor specialization is included for all measures to examine the effect of this presence.

Findings

This study documents a negative association of the mandatory disclosure with abnormal accruals and accrual estimation errors. The mandatory disclosure has a positive relation with positively associated with the propensity to issue a qualified opinion. However there is no evidence of the relationship with the likelihood of meeting or beating earnings and the audit fee. This suggests that audit quality improved after the implementation of the mandatory disclosure of auditor partner identification generally. However, it does not pose significant higher audit fees. Moreover, I find no relation between the mandatory disclosure and the audit quality provided by the auditor industry specialists for the majority of the models.

(4)

4 Originality

This paper contributes in several ways. First, it contributes to the limited literature around the mandatory disclosure of audit partner identification by providing additional empirical evidence on the impact of the regulatory rule on the audit quality and audit fees. Second, this study extends the increasing literature on the audit industry specialization and is among the first to provide insight to the relation between the audit fees, audit quality and audit industry specialization. Finally, this adds additional insights into the debate of whether the disclosures of audit partner identification are feasible and should be mandated (PCAOB 2009; PCAOB 2011; PCAOB 2013).

Research limitations

The results are subject to country specific characteristics and therefore the appliance in another country may result in other conclusions. Also, there could be measurement errors in the measure for industry expertise as it is not directly observable. Finally, the current proposal of PCAOB is to include the name instead of the signature which affects the benefits and costs differently.

Key words

Audit quality, auditor industry expertise, auditor partner identification disclosure, audit fee, accounting conservatism, accountability, transparency

(5)

5

Investors usually rely extensively on the audit report for information. Consequently, this report must be useful for their decision-making. The Public Company Accounting Oversight Board oversees audits of companies to project the interests of investors and those of the public. However, the current audit report does not provide information about the participants in the audit (PCAOB, 2011). As PCAOB (2011, p.2) states: ‘The audit report typically contains no information about who served in the role of engagement partner, or whether the firm issuing the report actually performed all of the work.’ This issue is also stimulated by the recommendation of the U.S. Treasury Department’s Advisory Committee on the Audit Profession and the increased international developments (Cole, 2014). In the European Union the partner signature requirement was already implemented in 2006: “[w]here an audit firm carries out the statutory audit, the audit report shall be signed by at least the statutory auditor(s) carrying out the statutory audit on behalf of the audit firm” (Cole, 2014).

Therefore PCAOB considers mandating the disclosure of the identity of the engagement partner’s name and the names of other persons and independent public accounting firms that took part in the audit to enhance the audit quality in the US (2011). This would increase transparency into and accountability for the preparation and the issuance of audit reports by providing investors with information regarding certain key participants in the audit process. In turn, it creates incentives for the firms to improve the quality of their engagement partners further (PCAOB, 2011). While PCAOB (2009; 2011) and investors believe that this will provide useful information for a wider range of users, many auditors believe that it would not provide additional information and would minimize the role of firms quality control system. For instance, EY stated in the comment letter that “’identifying engagement partner would result in operational challenges, as a result of legal requirements in connection with public offerings that will, of necessity, increase the costs, complexity and amount of time required for a company to access the capital markets, but will not provide meaningful additional information to investors that will offset such costs and challenges. We also believe that this proposal will not improve audit quality and will likely have potentially negative effects on the profession. The execution of an effective audit is a collective effort that can involve many individuals and depends on a variety of factors’’ (EY, 2014, p.1).

More recently, the International Auditing and Assurance Standards Board (IAASB) proposed to disclose the name of the engagement partner in the auditor’s report of a listed

(7)

7 entity, which will become the norm in those jurisdictions that follow IAASB standards (PCAOB, 2013). Nevertheless, it is not clear whether this would improve audit quality in practice and/or even have unintended negative consequences.

Prior research focused solely on the potential benefits and the unintended consequences in theory and could not make strong conclusions (Bailey et al., 2010; Cole, 2014; King et al., 2012). Cole (2014) discussed the rationales of the disclosure, the unintended consequences, the major concerns and the most relevant research to the related issues. Bailey et al. (2010) summarized the comments of the responders for the release of 2009 and compared this to provisions contained in the SOX of 2002. They conclude that audit partner identification would not enhance audit quality and it would only increase costs of additional audit and/or audit control. King et al. (2012) examined the factors that could affect audit quality in appearance and audit quality in fact, and applied three frameworks for this. They found that audit quality in appearance would increase, however the impact on the audit quality in fact remains unclear. They encouraged PCAOB to take the limited evidence into account and suggest PCAOB to focus on studies on the mechanisms of the requirement’s impact.

Due to mixed theories, empirical research was needed. There are in the current studies mixed empirical findings about the improvement of the audit quality in fact. Audit fee is also examined because the proposal does not only provide benefits but also creates costs. Carcello and Li (2013, p. 1511) investigated the effects on audit quality and audit fees of requiring the engagement partner to sign the audit report in the UK. Audit fees were used as an independent variable to measure the effort that was required for the higher quality audits. They found that the audit quality improved after the signature requirement was adopted. Blay et al. (2014) did a similar research in the Netherlands and found that mandating partner signatures did not improve audit quality. Lee (2014) studied on the effect of identifying the engagement partner on audit quality and the relation with the individual partner’s incentives and the partnership’s choice of internal quality control. He found that audit quality would be lower if the external inspection is too low to motivate partnership to maintain sufficient internal quality.

There is some evidence that public disclosure of auditor’s identity through signature requirements enables partner-specific measures of audit quality (Blay et al., 2014; Carcello and Li, 2013). Gul et al. (2013) provided some evidence that individual auditor effects on audit quality can be partially explained by auditor characteristics, like auditor industry expertise. An increasing literature supports the view that industry specialism can enhance

(8)

8 audit quality (Craswell et al., 1995; Dunn and Mayhew, 2004; Reichelt and Wang, 2009). Reichelt and Wang (2009) for example found that auditors who are industry specialist have clients with lowest abnormal accruals, suggesting that they provide higher audit quality. Some studies have also found that higher audit fees are paid for auditors with industry expertise than those without (Craswell et al. 1995; Ferguson, Francis, and Stokes, 2003; Francis, Reichelt, and Wang, 2005).

No study has investigated the effectiveness of audit identification disclosure with the presence of the industry specialization characteristics of auditor. Thus, it is interesting to find out whether the regulatory rule improves the audit quality in this presence and what the consequences are for the audit fees. Therefore the research question will be: Do disclosures of audit partner identification impact audit quality in presence of auditor industry specialization and how does the disclosure affect audit fees?

This is executed by examining the relation between the disclosure and audit quality and at the relation between the disclosure and audit quality in presence of auditor industry specialization. The study is measured through the output of audit quality; abnormal accruals with cross-sectional modified Jones (1991) approach, the accrual quality with a modified Dechow – Dichev (2002) model (LaFond and Schipper, 2005; Srinidhi and Gul, 2006), the propensity to issue a qualified audit opinion (Carcello and Li, 2013), and small earnings forecasts (Ashbaugh et al., 2003; Reichelt and Wang, 2009; Carcello and Li, 2013). Moreover, audit fee is also measured to see the effects of the disclosure (Deis and Giroux, 1996; Carcello et al.2002; Carcello and Li, 2013). Several control variables are included that relates to and affects the partner name, auditor industry expertise and audit quality.

The sample is collected from DataStream, consisting United Kingdom companies representing years 2007 through 2011. In general, the audit quality improved after the implementation of the mandatory disclosure of auditor partner identification. I find that the mandatory disclosure is negatively associated with abnormal accruals and accrual estimation errors and positively associated with the propensity to issue a qualified opinion. However there is no evidence of the relation between the mandatory disclosure and the likelihood of meeting or beating earnings and between the mandatory disclosure and the audit fee. For the second hypothesis, I find no relation between the mandatory disclosure and the audit quality provided by the auditor industry specialists for the majority of the models. Although the audit fees did increase after the mandatory disclosure, this should be taken with caution as auditor industry experts generally pose higher audit fees then other auditors.

(9)

9 This paper contributes in several ways. First, it contributes to the limited literature around the mandatory disclosure of audit partner identification by providing additional empirical evidence on the impact of the regulatory rule on the audit quality and audit fees. It also extends the general research literatures about the impact of a regulatory rule on audit fees and audit quality. Second, this study extends the increasing literature on the audit industry specialization and is among the first to provide insight to the relation between the audit fees, audit quality and audit industry specialization. Much of the prior literature investigated the effectiveness of the mandatory disclosure requirement on the audit quality alone, but did not look further at other factors that could affect the relation. This study focuses on the expertise of the auditor. Therefore, the study provides more in-depth explanation for the differential levels of audit quality by focussing on the individual of an auditor (as a moderating factor). Finally, this adds additional insight into the debate of whether the disclosures of audit partner identification are feasible and should be mandated (PCAOB 2009; PCAOB 2011; PCAOB 2013).

The remainder of the paper is organized as follows. Section II provides the background of the PCAOB, the relevant literature and the development of the hypotheses. In section III gives a explanation of the sample. Descriptive statistics and results are presented in section IV, followed by the conclusion in the last section.

2 Background information and hypotheses development:

2.1 Background PCAOB proposal and the releases.

In this section, the releases and rule filings will be explained from the concept till the current situation to clarify the understanding of the issue and the concerns.

2.1.1 ‘2009 Release’

Following the EU who issued the Eight Company Law Directive, which required the state members to adopt the requirement of engagement partner to sign the audit report, the PCAOB was encouraged by the U.S. Department of Treasury to “undertake a standard-setting

(10)

10 initiative to consider mandating the engagement partner’s signature on the auditor’s report’’ in order to increase the quality of the audited reports (PCAOB, 2009). This would not impose the signing partner any duties, obligations or liability that are greater than those imposed on such person as a member of an auditing firm. Since the requirement is similar to Section 302 certification requirements of SOX 2002, it should give similar benefits (King et al., 2012).

PCAOB responded by publishing a concept release in July 28, 2009, which required feedback and comments for improvement (PCAOB, 2009). The concept required auditors to sign and to disclose their own name in the audited report providing more useful information for investors, in terms of enhancing the accountability and the transparency and creating incentives to increase the audit quality (PCAOB, 2009).

Some feedback and / or comments were received on the release (PCAOB, 2011). Investors find that the proposal increases the investors protection by enhancing the accountability and transparency. Accounting professors had mixed opinions, some studies showed enhancement of personal accountability could lead to benefits, but the signature requirement itself could have negative effect if it diminishes the firm’s accountability. Auditors think that the focus on the role of firm’s quality control system, expertise and skill of firm as a whole would be minimized which is most important for them (besides the skill and expertise of the individual engagement partner). Readers might misunderstand the signature requirement of the audit report, which should reflect significant changes in audit procedures, or a shift in responsibility for the audit from the audit firm to the engagement partner. Some were also concerned about the unintended consequences of the signature requirement, which could increase audit costs. This include engagement partners supporting defensive auditing which increase costs of audit, avoiding riskier clients to maintain audit quality profile (portfolio), leaving of talented individuals, or refusing to enter the profession. (Bierstaker et al., 2009; PCAOB, 2011).

2.1.2. ‘2011 Release’

After receiving the required feedbacks and comments for improvement, PCAOB issued proposal in October 11, 2011 (PCAOB, 2011). This time PCAOB (2011) required disclosing the name of the partner (without a signature) in the auditor’s report and certain other participants in the audit. Moreover, registered firms need to disclose the name of the

(11)

11 engagement partner for each audit report and are required to be reported on the Board’s Annual Report Form. This would need slightest changes to the audit report, but (the same assumptions still applies and) could increase transparency by providing more useful information to investors about the certain key participants in the audit process.

The comments on the release 2011 remained the same as those of 2009 (PCAOB, 2013). Investors favoured more transparency, together with some audit committee members, corporate officials, and European auditors. Accounting firms opposed to disclose the name and signature of the engagement partner in the auditor’s report, because they were concerned that it would confuse readers of the auditor’s report or would lead to unintended consequences.

2.1.3 ‘2013 Release’

PCAOB reproposed the rule in December 4, 2013 based on comments and experiences in the other countries (PCAOB, 2013). They required (1) the disclosure in the auditor's report of the name of the engagement partner and (2) disclosure in the auditor's report of the names, locations, and extent of participation of other independent public accounting firms that took part in the audit and the locations and extent of participation of other persons not employed by the auditor that took part in the audit. This would be more useful to investors because the disclosure would add to a mix of information like, individual restatements or going-concern opinions that investors hold for their decision-making. Moreover, disclosing all participants who entered the process of auditing a particular report discourages practices that were outsider without the expertise or qualifications play significant roles in the audit of issuers. In the present situation, the investors would not know whether the audit report is audited by other participations than the signing firm. As the PCAOB (2013, p.18) mentioned: “In many situations, the signing firm uses another firm in a foreign country to audit the financial statements of a subsidiary in that foreign country. These arrangements can be affective and cost-efficient way to audit today’s multinational corporations. At the same time the quality of the audit is dependent, to some degree, on the competence and integrity of the participating accounting firms. Especially when the signing firm has not reviewed all the work done by the other firm.”

(12)

12 This section has explained the contents of the releases and the relating benefits, concerns and issues of each release. The biggest concern is whether the audit quality will increase in reality / fact if the disclosure is implemented. In the next section, audit quality will be defined and hypothesis will be developed to examine the audit quality in fact.

2.2 Hypothesis development

The main point of the released proposals is the usefulness of the disclosure of aspects like identity of engagement partner and other firms associated with audit to investors and other financial statement users. This is consistent with the Board’s mission to “further the public interest in the preparation of informative, accurate and independent audit reports” (PCAOB, 2013).

A primary justification of PCAOB is that the mandatory disclosure of identity partner engagement would enhance the transparency and accountability and therefore complementing the audit quality (Carcello and Li, 2013, p. 1512; King et al, 2012, p. 534). To elaborate this assumption, audit quality, audit effort and auditor industry expertise will be explained in sections in order to develop the hypotheses.

2.2.1 Audit Quality

This section will handle the proposed aim audit quality. This will be supplemented with the related frameworks transparency and accountability. Together with all these information, the relations that PCAOB proposed will be built as hypotheses.

Audit quality is the probability that an auditor will discover irregularities or accounting misstatement(s) in the financial accounting report and will report those irregularities or accounting misstatement(s) (DeAngelo, 1981). Quality will be higher when the audit provides a reasonable assurance that the audited financial statements and related disclosures are presented in accordance with generally accepted accounting principles (GAAP) and are not materially misstated due to errors or fraud (Bedard, Johnstone and Smith, 2010).

Audit effort is required in order to acquire more evidence, which is as result more effective in detecting earnings management activities and detecting conditions that are not

(13)

13 according the qualification of the audit opinion, therefore improving the audit quality (Bedard et al., 2010; Carcello and Li, 2013). King et al. (2012) stated that the audit quality in fact is affected by the changes in the behaviour of the engagement partner and the audit firm. Note that the focus of audit quality lies on the behaviour of an auditor and not on how this appear through the perception of users (investors) in terms of their uncertainty. Also, the improvement of audit quality in fact is based on the assumption that the audit quality level is not at its optimum (Carcello and Li, 2013).

Audit quality in fact is the actual, but ‘unobservable’ audit quality that is delivered by the auditors which is dependent on the accountability of the engaging partner (King et al., 2012). This means that the partner is being held responsible to others for one’s judgements on predefined standards of performance (Carcello and Li, 2013; Blay et al., 2014).

Implementation of the partner identity disclosure would enhance the accountability of the auditors, which influences the information processing as well as decisions and judgements. Auditors will have greater realization of their behaviour reflected in their efforts, which is observable by a larger audience after the appliance of the disclosure (Blay et al., 2014; King et al, 2012). This creates an incentive for auditors to avoid negative consequences associated with a perceived audit failure (like criticism and embarrassment) and also with a decreased reputation of human capital (Carcello and Li, 2013). Greater accountability often results in better performance, increased personal responsibility, more careful evaluation of alternatives, reduction of bias and overall greater audit effort (Blay et al., 2014; King et al, 2012).

To be more specific, a several partner behaviour changes due to greater accountability can increase the audit effort and therefore increase the audit quality (Hoitash, Markelevich and Barragato, 2007; Carcello and Li, 2013). First, it would induce the partner and audit team to perform more work by extending the procedures. A greater accountability can also change the nature of the audit procedures performed in terms of more evidence and better substantive evidence. Likewise, it also may lead to greater diligence in the performance of the partner and the audit team. Finally, more conservative auditor reporting will be provided due to enhanced accountability.

When irregularities or accounting misstatements in financial statements are more likely to be detected caused by greater professional care of the partner and audit team, then the mandatory disclosure of partner identification would lead to higher audit quality (DeAngelo, 1981; PCAOB, 2013). Taken together, when the more audit effort is provided,

(14)

14 and/or the nature of audit procedures are changed, and/or greater diligence in performance is provided, or more conservative audit reporting is provided, then audit quality will increase.

In the literature, this is reflected in a lower level of earnings management in the outputs like lower level of abnormal accruals, lower accrual estimation errors, lower meeting an earnings benchmark, and the greater likelihood of issuing a qualified audit opinion (Dechow and Dichev, 2002; Watts, 2003; Carcello and Li, 2013; Francis and Yu, 2009; Bedard et al., 2010; Ball and Shivakumar, 2006; Defond et al. 1999; Bartov et al., 2002; Becker et al., 1998; Francis et al., 1999; Francis, 2004; Craswell et al., 2002’Francis and Krishnan, 1999; Defond et al., 1999).

In prior literature concerning the abnormal/discretionary accruals, accounting choices of managers with the intention to bias accounting information are examined (often with Jones model 1991). Higher audit quality is delivered due to ability to constrain aggressive and potentially opportunistic reporting of accruals (Becker et al., 1998; DeFond et al., 1998; Francis et al., 1999; Francis, 2004). This suggests that higher audit quality is delivered when the level of abnormal accruals is kept low.

Dechow and Dichev (2002) introduced a new measure of the quality of working capital accruals and earnings. While abnormal accruals focus on earnings management goals to mislead users of financial statements, the accrual estimation errors focus on the managerial intent in terms of the incidence and magnitude. The concept is that the quality of accruals is higher when the estimates of the future cash flow contain less error meaning that the adjusted numbers reflect the firm performance better. Investors rely on the auditor to detect material misstatements caused by less reliable accruals, thus higher audit quality is delivered when the level of accrual estimation errors is low (DeAngelo, 1981; Dechow and Dichev, 2002;

Hoitash et al., 2007).

In the research using the issuance of qualified audit opinion, the underlying assumption here is that independent auditors are more likely to issue qualified audit opinions providing higher audit quality (Craswell et al., 2002). Francis and Krishnan (1999) and Defond et al. (1999) find that Big 4 auditors are more likely to issue modified audit reports, which indicates greater reporting conservatism for a given set of client characteristics. This suggests that higher audit quality is delivered when the auditors are more likely to issue qualified audit opinion.

Finally, another research relating to the earnings benchmarks, Carey and Simnet (2006) and Reichelt and Wang (2009) find that high audit quality is provided when clients

(15)

15 engage less in beating earnings benchmarks. Those benchmarks are important for managers because investors and other stakeholders see this as important indicator for their decision-making. Therefore, there is higher chance of earnings management when managers engage in beating earnings benchmarks (Bartov et al., 2002). This suggests that higher audit quality is delivered when auditors are less tolerant of aggressive earnings management. Their clients are then less likely to meet or beat analysts’ earnings forecasts within one penny of earnings per share.

Two studies examined the impact of the mandatory disclosure on audit quality in fact (Carcello and Li, 2013; Blay et al., 2014). Carcello and Li (2013) investigated the effects on audit quality and audit fees of requiring the engagement partner to sign the audit report in the United Kingdom. They find a significant decline in abnormal accruals and the propensity to meet an earnings threshold. Moreover, they also find a significant increase in the incidence of qualified audit reports and in earnings informativeness. Overall, their results suggested that the audit quality is improved in United Kingdom firms after the signature requirement is adopted. In contrast, Blay et al. (2014) examined the audit quality effects of a partner signature mandate by comparing multiple measures of audit quality in years after adoption of an audit partner signature mandate in the Netherlands to the audit quality in years prior to adoption, as well as to the United Kingdom. No substantial change in audit quality using multiple measures and sample selection criteria was found and thus they believe that mandatory partner signature will not be effective. The level of audit quality could explain this. The highly regulated nature of the audit market may mean that the level of audit quality is already at its optimum (Dunn and Mayhew, 2004; Carcello and Li, 2013).

As the mixed results on the feasibility of the PCAOB’s statement to enhance audit quality in fact are present, this will be examined. Hence, the prediction is:

H1a: The mandatory disclosure of audit partner identification has a positive relation with the audit quality.

By requiring disclosure of the partner identity means that PCAOB acknowledges a substantial work by highly skilled individual auditors. The engagement partner therefore plays essential role in the audit quality (King et al., 2012). As King et al. (2012) stated, an auditor need to have enough task-related expertise and resources, including cognitive abilities, in order to complete a task successfully. Mandating a disclosure means more assurance needs to be

(16)

16 provided (Carcello and Li, 2013). This leads to more audit work and more audit evidence caused by the disclosure of audit partner identity and additional costs will likely be charged (Carcello et al., 2002, p. 369; Hoitash et al., 2007; Carcello and Li, 2013).

In general, auditors will charge audit fee for the effort provided depending on the client’s agency-driven demand for external auditing and the supply-side factors, such as audit firm size, auditor expertise, auditor-perceived risk factors like litigation costs, complexity and other characteristics of the audited entity (Simunic and Stein, 1996; Zerni, 2012). If more work is required due to the disclosure of audit partner identity, audit fees would likely increase caused by higher audit effort (Simunic, 1980; O’Keefe et al., 1994; King, et al., 2012; Carcello and Li, 2013). This is indicated in the prior research: Bedard and Johnstone (2004) find that auditors plan increased effort and billing rates for clients with earnings manipulation risk. Carcello et al. (2002, p.381) find that auditor’s cost of providing high quality audit services are incorporated into audit fee. Hoitash et al. (2007) find that there is an positive association between audit fees and audit quality using two measures. In summary, providing high audit quality due to the increased transparency and accountability, more audit effort is required, which is reflected in higher audit fees charged. Hence, the prediction is:

H1b: The mandatory disclosure of audit partner identification has a positive relation with the audit fees.

2.2.2 Presence of auditor industry expertise

Gul et al. (2013) provided some evidence that individual auditor effects on audit quality can be partially explained by auditor characteristics, such as educational background, Big 4 audit firm experience, rank in the audit firm, and political affiliation. Therefore, this study will focus on the auditor expertise.

Academic literatures commonly suggest that industry specialist generally have in-depth knowledge of types of clients produce higher audit quality on the audited financial statements (Defond, Francis and Wong, 2000; Balsam et al., 2003; Zerni, 2012). A change in the requirements of the identities of individual partners means that the regulators and standard-setters acknowledge that a public company audit involves work of high skilled auditors exercising their own professional judgment. Therefore, it is important to take

(17)

17 industry specialization of the auditors in account (Zerni, 2012). There is some evidence that public disclosure of auditor’s identity through signature requirements enables partner-specific measures of audit quality (Blay et al., 2014; Carcello and Li, 2013)

Audit firms develop specializations to meet the clients’ needs to create benefits in audit firms’ competition (Dunn and Mayhew, 2004). Industry expertise is a result from intensive practice and the reputation of similar clients (Zerni, 2012). They use differentiation strategy to service a relatively large group of clients with similar characteristics. Clients that use auditor’s expertise can benefit from enhanced audit quality and disclosure advice. Moreover, clients can signal investors that they provide enhanced disclosures by using auditor industry expertise (Dunn and Mayhew, 2004).

Industry-specialist auditors increase audit quality and enhance financial statement credibility due to their greater industry specific knowledge that they have collected by serving other clients in the same industry and learning and sharing best practices across the industry (Dunn and Mayhew, 2004). Therefore, they can provide more effective audits, identify misstatements more effectively, and help clients creating more enhanced disclosure.

Moreover, industry-specialist auditors have a reputation for their expertise and therefore have the incentive to protect their reputation (in order to earn audit fee premium) (Reichelt and Wang, 2009). Reichelt and Wang (2009) mentioned that they would provide higher audit quality by resisting client pressure for greater discretion and by imposing stricter standards on clients.

This is shown in the prior literature that examines the impact of the auditor industry expertise on the audit quality. In general, they found that the expertise would increase the audit quality and therefore enhance the credibility of financial statements (Balsam et al., 2003; Krishnan, 2003; Reichelt and Wang, 2009; Grambling et al., 2000; Lim and Tan, 2008; Griffin and Lont., 2009; Knechelt et al. 2007; Romanus et al., 2008; Carcello and Nagy, 2004 ). Gramling et al. (2000) for example provide evidence that industry specialists are strongly related with current earnings and subsequent cash flows. Several other studies find lower levels of accruals earnings management for auditors with industry expertise (Balsam et al., 2003; Krishnan, 2003; Reichelt and Wang, 2010; Chi et al., 2011) and smaller income-increasing and income-decreasing abnormal accruals magnitude is found (Reichelt and Wang, 2009). Gul et al. (2009) find evidence suggesting that auditor industry specialization is negatively related with the shorter auditor tenure and lower earnings quality. Auditor industry specialization is associated with lower likelihood of client just meeting analyst earnings

(18)

18 forecasts (Reichelt and Wang, 2009). Also industry specialists are more likely to issue a going-concern audit opinion (Lim and Tan, 2008; Griffin and Lont, 2009; Reichelt and Wang, 2009). Dunn and Mayhew (2004) examined the relation between the use of industry specialist audit firm and the quality of firm’s disclosure’s using rank fraction of the overall disclosure quality score. They find that industry-specialist audit firms (only) in unregulated industries provide value added services to their clients in the form of improved disclosure quality, and the choice of an industry-specialist auditor is a signal of enhanced disclosure quality. Knechelt et al. (2007) find that firms with specialized auditor receive higher valuations. Carcello and Nagy (2004) find lower involvement in SEC enforcement actions when auditors with industry expertise are present. In addition, Romanus et al. (2008) find a negative relation between auditor industry expertise and probability of restatements.

Taken together, prior literature overall suggests that industry-specialist auditors increase their audit effort in order to provide higher audit quality by benefitting from their greater industry specific knowledge and their incentives to protect their reputation (Dunn and Mayhew, 2004; Francis and Yu, 2009; Reichelt and Wang, 2009; Zerni, 2012). Based on prior results, it is likely under the regime with mandatory disclosure that audit quality will be higher. Hence, the prediction is:

H2a: The mandatory disclosure of audit partner identification has a positive relation with audit quality, in presence of auditor industry expertise.

As already mentioned, the increase of audit fees is caused by the auditor skills (Reichelt and Wang, 2009; King et al. 2012). Prior literature suggests industry specialist generally have in-depth knowledge of types of clients, hence the higher probability to detect irregularities or accounting misstatements. Consequently, they charge higher audit fee for the audited financial statements then the regular auditors (Craswell, Francis and Taylor, 1995; Defond, Francis and Wong, 2000; Balsam et al., 2003; Zerni 2012). “These costs are in addition to any additional fees charged by industry specialists’’(Dunn and Mayhew, 2004, p.40). These engagement fee differentials reflect the differences in audit quality where Big 4 firms and industry leaders claim additional premiums due to their greater audit effort (hours) or expertise (billing rates) in order to enhance credibility of the financial statements and essentially equate audit quality with the investors’ perception (King et al., 2012).

(19)

19 Prior literature confirms the relation between the higher audit fees charged and the auditor industry expertise. Craswell et al. (1995) find results that support the notion that additional investments in expertise will increase audit fees. Defond et al. (2000) find higher audit fees are paid for the Big 4 due to their brand name and to industry specialization. Both Ferguson et al. (2003) and Francis, Reichelt and Wang (2005) investigated the pricing of national and city-specific reputations for industry expertise and find that those specialists charge significantly higher fees than other auditors.

In summary, in the presence of audit industry expertise, higher audit fees are charged due to their greater likelihood to detect material misstatements. Mandatory disclosure would likely increase the audit fees additionally due to the costs of complying with the mandatory disclosure of the audit partner identification. Hence, the prediction is:

H2b: The mandatory disclosure of audit partner identification has a positive relation with the audit fees, in presence of auditor industry expertise.

3 Research methodology

3.1 Sample and data

U.K. adopted the Eighth Directive through the Companies Act of 2006, but the requirement for engagement partner signature was not effective until audits of financial statements ending in April 2009 or later. Therefore U.K. is the appropriate country to examine the audit quality after the implementation of the disclosure requirement.

The data will be collected for United Kingdom companies from database DataStream and Compustat Global by combining several samples FBRIT, LTOTMKUK, WSCOPEUK and those of Compustat Global in order to create a large sample that covers most UK companies. The observations of the U.K. companies are from 2007 through 2011, as data is needed before and after the disclosure requirement. Moreover, the effects of financial crisis and time to adapt to new regulations by audit firms needed to be considered hence the requirement of 2 years prior and after the implementation of the disclosure.

(20)

20 DataStream provided 24,240 firm-year observations for year 2007 through 2011. Five subsamples are created for the analysis: abnormal accruals analysis, accrual estimation errors analysis, qualified audit opinion analysis, small earnings forecasts analysis and audit fee analysis. Abnormal accrual analysis consists of 6,795 observations after deducting missing values to compute abnormal accrual and 6,147 observations after deducting missing values to compute control variables. To be able to observe changes before and after the implementation in a balanced way, firms need to exist both 2 year prior the implementation and after the implementation. This result in a sample of 3,328 firm-year observations. Similar procedure is done for all other analysis which is summarized in Table 1.

TABLE 1 Sample Selection Panel A: Sample for Abnormal Accruals Analysis

UK companies for the period 2007 to 2011 from DataStream. 24,240 Delete: Firms without necessary financial data to compute expertise and

abnormal accruals (17,445)

Delete: Firms without necessary financial data to compute control variables (648) Delete: Firms that do not exist both 2 years before and 2 years after the (2,819)

signature requirement.

Final Sample in Abnormal Accrual Analysis 3,328

Panel B: Sample for Accrual Estimation Errors

UK companies for the period 2007 to 2011 from DataStream. 24,240 Delete: Firms without necessary financial data to compute expertise and

estimation errors. (16,683) Delete: Firms without necessary financial data to compute control variables. (2,154) Delete: Firms that do not exist both 2 years before and 2 years after (3,731)

the signature requirement.

Final Sample in Accrual Estimation Errors 1,672

Panel C: Sample for Qualified Audit Opinion Analysis

UK companies for the period 2007 to 2011 from DataStream. 24,240 Delete: Firms without audit opinion and expertise data. (14,894) Delete: Firms without necessary financial data to compute control variables. (1,124) Delete: Firms that do not exist both 2 years before and 2 years after (2,074)

(21)

21

Final Sample in Audit Opinion Analysis 6,148

Panel D: Sample for Small Earnings Forecasts Analysis.

UK companies for the period 2007 to 2011 from DataStream. 24,240 Delete: Firms with missing prior or current years’ earnings data and

expertise data. (13,116)

Delete: Firms without necessary financial data to compute control variables (824) Delete: Firms that do not exist both 2 years before and 2 years after (4,236)

Final Sample in meeting or beating analysts’ earnings forecasts 6,064

Panel E: Sample for Audit Fee Analysis

UK companies for the period 2007 to 2011 from DataStream. 24,240 Delete: Firms without expertise and audit fee data (13,838) Delete: Firms without necessary financial data to compute control variables (590) Delete: Firms that do not exist both 2 years before and 2 years after (3,964)

Final Sample in Audit Fee Analysis 5,848

3.2 Research design

The justification of PCAOB is that mandating disclosure of identity partner engagement would enhance the transparency and accountability and therefore increases the audit quality (Carcello and Li, 2013, p. 1512; King et al, 2012, p. 534; PCAOB, 2011). Based on the studies (Carcello and Li, 2013; Blay et al., 2014; Balsam et al., 2003; Reichelt en Wang, 2009), the following variables will be used in order to examine the dependent variable audit quality:

AUDITORNAME, which is an indicator / dummy variable coded 1 if it is the effective year of the identity disclosure requirement, and 0 if it is before the identity disclosure requirement. This serves as an independent variable in the relation with the dependent variable audit quality for the first hypothesis.

For the second hypothesis, SPECIALIST act as moderating variable in relation with the dependent variable audit quality and is coded 1 if the auditor is an industry specialist in the effective year, and 0 otherwise. Balsam et al. (2003) examined audit industry expertise using different measures as a base, like largest supplier in each industry, industry dominance, continuous market share based upon client sales, number of clients in the industry, number of

(22)

22 clients and number of clients audited by the auditor. Following earlier studies of Balsam et al. (2003) and Reichelt and Wang (2009), the proportion of two-digit SIC industry sales audited by each audit firm is used as a proxy for auditor industry specialization. The findings of Reichelt and Wang (2009) and Balsam et al.(2003) concluded that for national and city level expertise both would enhance the audit quality. Therefore both levels are taken together in this study as the focus is auditor industry expertise in general.

Auditor industry expertise is defined as expertise that increase with the size of market share and a sufficient large market share in conjunction with the incentives to deliver higher audit quality (Dunn and Mayhew, 2004; Balsam et al., 2003; Reichelt and Wang, 2009). These auditors has a market share greater than 30% in a two-digit SIC category covering for both levels of expertise (Reichelt and Wang, 2009).

Panel A of Table 2 presents the descriptive statistics of observations, mean and the standard deviation of the industry specialist. The mean of 0.149 and standard deviation of 0.357 are in line with the results from Reichelt and Wang (2009). Panel B shows the number of industry specialists by auditor firm and by year. Variable definitions are reported in Table 3.

TABLE 2

Descriptive statistics of auditor industry expertise Panel A: Descriptive statistics of auditor industry specialist

Obs Mean SD

Industry Specialist 24,240 0.149 0.357

Panel B: Industry specialists by Auditor and Year.

Auditor/year 2007 2008 2009 2010 2011 Total

BDO International 1 3 1 1 1 7

Baker Tilly Interna.. 3 3 1 1 1 9

Deloitte Touche Toh.. 39 103 94 98 141 475

Ernst & Young 56 54 46 60 51 267

Grant Thornton 6 5 4 4 6 25

KPMG 134 250 258 248 235 1,125

Not disclosed 1 1 3 3 1 9

PriceWaterhouseCoop.. 390 374 302 289 338 1,693

(23)

23 TABLE 3

Variable Definitions Dependent variables

|ACCt| The absolute value of discretionary accruals for year t. |ESTMERRORt| The absolute value of accrual estimation errors for year t.

BEATt An indicator variable coded 1 if earnings exactly meet or beat the latest

analysts’ earnings forecast by one cent per share (for year t), and 0 otherwise.

QOPINIONt An indicator variable coded 1 if the auditor issued a qualified audit opinion for year t, and 0 otherwise.

LNFEEt Natural logarithm of total audit fees in year t.

Independent variables

AUDITORNAME An indicator variable coded 1 if it is the effective year of the identity disclosure requirement, and 0 if it is before the identity disclosure requirement.

SPECIALIST An indicator variable coded 1 if the auditor is an industry specialist at national and/or city levels in effective year and 0 otherwise.

Control variables

SIZE Logarithm of total assets.

PYACCRUAL Last year’s total current accruals for year t. Calculated by summing net income before extraordinary items and depreciation & amortization minus operating cash flow for year t and then dividing it by lagged total assets for year t

ROA Return on assets.

Calculating by income before extraordinary items in year t by total assets for year t.

LEVERAGE Total debt for year t divided by total assets for year t.

LOSS An indicator variable coded 1 if net income < 0, and 0 otherwise. MB Market to Book value for year t. Calculated as follows: market value at

the end of year t divided by book value at the end of year t. OPC Natural logarithm of operating cycle. Calculated as follows:

[360/(Sales/Average AR) + 360/(Cost of goods sold)/(Average Inventory)].

CFO Cash flow from operations divided by total assets for year t.

LITIGATE An indicator variable coded 1 if the company operates in a high litigation industry (SIC codes of 2833-2836, 3570 – 3577, 3600-3674, 5200 – 5961, and 7370-7370), and 0 otherwise.

BIG 4 An indicator variable coded 1 if the firm is audited by a Big 4 firm in year t, and 0 otherwise.

Industry Dummies A dummy variable coded 1 if manufacturing firms and trade firms (SIC codes of 2000-3999, 5000-5999), 0 otherwise in case of service-oriented firms.

(24)

24 3.2.1 Measures of audit quality

Audit quality is normally unobservable (Bedard et al., 2010). However the indicators like abnormal accruals, accrual estimation errors of Dechow and Dichev model (2002), propensity to issue qualified opinion, model, and patterns around meeting earnings thresholds are accepted in academic research as indicative of whether earnings are being managed. So changes in those indicators suggesting the diminishing presence of earnings management are linked to the improved audit quality. Definitions of all variables that will be mentioned in the following sections are explained in table 3.

Abnormal Accrual

The first measure of audit quality is the cross-sectional modified Jones (1991) approach in order to estimate abnormal accruals. This model is used to examine the relation between abnormal accruals and the audit partner identification requirement as this is indicated as the best measure of discretionary portion of total accruals (Balsam, Krishnan and Yang, 2003; Bartov et al., 2000). Using cross-sectional modified Jones makes the sample size less biased and allows for examining firms with less existing years (Bartov et al., 2000). The model is estimated for every industry group with at least 15 firms in a given year. Industry groups are defined by matching all firms based on year and 2-digit SIC codes (Francis et al., 2008; Carcello and Li, 2013). Continuous variables and extreme values are winsorized to the 1 and 99 percentiles.

Using the model of Kothari et al. (2005, p. 173) which adjusts the change of receivables:

𝑁𝐷𝐴𝑖𝑡 = 𝛽1 (_{𝐴𝑖𝑡−1}1 ) + 𝛽2 (Δ𝑅𝐸𝑉𝑖𝑡_{𝐴𝑖𝑡−1} − Δ𝑅𝐸𝐶𝑖𝑡_{𝐴𝑖𝑡−1}) + 𝛽3(𝑃𝑃𝐸𝑖𝑡_{𝐴𝑖𝑡−1}) (1) Where:

NDAt nondiscretionary accruals in year t scaled by lagged total assets (relates to the

changes in the firm’s economic condition/normal operating activities) ∆REVt revenues/sales in year t less revenue in year t -1

PPEt gross property plant and equipment in year

A t-1 total asset at t-1

β1, β2,β3 firm-specific parameters

∆RECt net receivables in year t less receivables in year t -1

The parameters are generated using the following model in estimation period:

𝑇𝐴𝑖𝑡 𝐴𝑖𝑡−1= 𝛼1 ( 1 𝐴𝑖𝑡−1) + 𝛼2 (𝛥𝑅𝐸𝑉𝑖𝑡) 𝐴𝑖𝑡−1 + 𝛼3 ( 𝑃𝑃𝐸𝑖𝑡 𝐴𝑖𝑡−1) + 𝜀𝑖𝑡 (2) Where:

(25)

25 TAit total accruals for the year t

Final step is:

𝐴𝐶𝐶𝑅𝑡 = 𝑇𝐴𝑖𝑡 − 𝑁𝐷𝐴𝑖𝑡 (3)

Where:

ACCRt total discretionary accruals for year t

The expectation is that AUDITORNAME has a negative coefficient with abnormal accruals and this relation is even more magnified in the presence of the independent variable

SPECIALIST (due to their specific knowledge and reputational incentive). Thus, higher audit quality is delivered due to ability to constrain aggressive and potentially opportunistic

reporting of accruals (Becker et al., 1998; DeFond et al., 1998; Francis et al., 1999; Francis, 2004). Specifically, the following regressions are estimated:

H1: |𝐴𝐶𝐶𝑅𝑡| = 𝑏0 + 𝑏1𝐴𝑈𝐷𝐼𝑇𝑂𝑅𝑁𝐴𝑀𝐸 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑃𝑌𝐴𝐶𝐶𝑅𝑈𝐴𝐿 + 𝑏4𝑅𝑂𝐴𝑡 + 𝑏5𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏6𝐿𝑂𝑆𝑆𝑡 + 𝑏7𝑀𝐵𝑡 + 𝑏8𝐶𝐹𝑂𝑡 + 𝑏9𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏10𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (4) H2: |𝐴𝐶𝐶𝑅𝑡 | = 𝑏0 + 𝑏1𝑆𝑃𝐸𝐶𝐼𝐴𝐿𝐼𝑆𝑇 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑃𝑌𝐴𝐶𝐶𝑅𝑈𝐴𝐿 + 𝑏4𝑅𝑂𝐴𝑡 + 𝑏5𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏6𝐿𝑂𝑆𝑆𝑡 + 𝑏7𝑀𝐵𝑡 + 𝑏8𝐶𝐹𝑂𝑡 + 𝑏9𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏10𝐵𝐼𝐺4𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 " (5)

Accrual Estimation Errors

The second measure is the accrual quality based on Dechow – Dichev (2002) and Francis, LaFond and Schipper (2005). This is a regression model of changes in working capital on past, present and future operating cash flows. The residuals of this model are the accrual estimation errors indicating the level of accrual quality. The model is also estimated for every industry group with at least 15 firms in a given year. Industry groups are defined by matching all firms on year and 2-digit SIC codes (Francis et al., 2008; Carcello and Li, 2013).

Following Ashbaugh-Skaife et al. (2008), period of 3-5 years is used. Continuous variables and extreme values are winsorized to the 1 and 99 percentiles.

Following Francis, LaFond and Schipper (2005) and Srinidhi and Gul (2006), a modified version of Dechow – Dichev (2002) model is used as it takes the expectation of current accruals into account (Srinidhi and Gul, 2006):

(26)

26 ∆𝑊𝐶𝑖𝑡 = 𝛼0 + 𝛼1𝐶𝐹𝑂𝑖𝑡 − 1 + 𝛼2𝐶𝐹𝑂𝑖𝑡 + 𝛼3𝐶𝐹𝑂𝑖𝑡 + 1 + 𝛼4∆𝑆𝐴𝐿𝑖𝑡 + 𝛼5𝑃𝑃𝐸𝑖𝑡 + 𝜀𝑖𝑡

(6) Where:

ΔWC change in working capital between year t-1 and t = (ΔCA − ΔCL - ΔCash − ΔSTDEBT)

ΔCA change in current assets between year t-1 and t ΔCL change in current liabilities between year t -1 and t ΔCash change in cash balance between year t -1 and t

ΔSTDebt change in short term debt included in current liabilities between year t -1 and t

All variables are scaled by the average of total assets between year t-1 and t. The residuals reflect parts of accruals that do not map into cash flows and the standard deviation of the residuals measures the extent of estimation errors.

The expectation is that AUDITORNAME has a negative coefficient with accrual estimation errors and this relation is even more magnified in the presence of the independent variable SPECIALIST (due to their specific knowledge and reputational incentive). Thus, the quality of accruals is higher when the estimates of the future cash flow contain less error meaning that the adjusted numbers reflect the firm performance better (DeAngelo, 1981; Dechow and Dichev, 2002; Hoitash et al., 2007). Specifically, the following regressions are estimated: H1: |𝐸𝑆𝑇𝑀𝐸𝑅𝑅𝑂𝑅| = 𝑏0 + 𝑏1𝐴𝑈𝐷𝐼𝑇𝑂𝑅𝑁𝐴𝑀𝐸 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑂𝑃𝐶 + 𝑏4𝑅𝑂𝐴𝑡 + 𝑏5𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏6𝐿𝑂𝑆𝑆𝑡 + 𝑏7𝑀𝐵𝑡 + 𝑏8𝐶𝐹𝑂𝑡 + 𝑏9𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏10𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (7) H2: |𝐸𝑆𝑇𝑀𝐸𝑅𝑅𝑂𝑅| = 𝑏0 + 𝑏1𝑆𝑃𝐸𝐶𝐼𝐴𝐿𝐼𝑆𝑇 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑂𝑃𝐶 + 𝑏4𝑅𝑂𝐴𝑡 + 𝑏5𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏6𝐿𝑂𝑆𝑆𝑡 + 𝑏7𝑀𝐵𝑡 + 𝑏8𝐶𝐹𝑂𝑡 + 𝑏9𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏10𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (8)

Qualified Audit Opinion

An additional measure is auditor’s propensity to issue a qualified audit opinion (Carcello and Li, 2013). The expectation is that AUDITORNAME has a positive coefficient with the issuance of a qualified audit opinion and this relation is even more magnified in the presence of the independent variable SPECIALIST (due to their specific knowledge and reputational

(27)

27 incentive). As independent auditors are more likely to issue qualified audit opinions providing higher audit quality (Craswell et al., 2002). Specifically, the following logistic regressions based on Craswell et al. (2002) are estimated:

H1: 𝑄𝑂𝑃𝐼𝑁𝐼𝑂𝑁𝑡 = 𝑏0 + 𝑏1𝐴𝑈𝐷𝐼𝑇𝑂𝑅𝑁𝐴𝑀𝐸 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑅𝑂𝐴𝑡 + 𝑏4𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏5𝐿𝑂𝑆𝑆𝑡 + 𝑏6𝑀𝐵𝑡 + 𝑏7𝐶𝐹𝑂𝑡 + 𝑏8𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏9𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (9) H2: 𝑄𝑂𝑃𝐼𝑁𝐼𝑂𝑁𝑡 = 𝑏0 + 𝑏1𝑆𝑃𝐸𝐶𝐼𝐴𝐿𝐼𝑆𝑇 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑅𝑂𝐴𝑡 + 𝑏4𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏5𝐿𝑂𝑆𝑆𝑡 + 𝑏6𝑀𝐵𝑡 + 𝑏7𝐶𝐹𝑂𝑡 + 𝑏8𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏9𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (10)

Small Earnings Forecasts

The last measure is the likelihood that firms will report a small earnings increase, in other words earnings benchmark (Blay et al., 2014; Carcello and Li, 2013). Following Ashbaugh et al., 2003, Reichelt and Wang (2009) and Carcello and Li (2013) a logistic regression model is used to test whether the firms meet or beat the earnings with by one cent per share. This is calculated by taking the difference between a firm’s income before extraordinary items in years t and t-1, scaled by the market value of equity at the end of year t-1.

The expectation is that AUDITORNAME has a negative coefficient with the likelihood of reporting small earnings increase and this relation is even more magnified in the presence of the independent variable SPECIALIST (due to their specific knowledge and reputational incentive). So higher audit quality is delivered when auditors are less tolerant of aggressive earnings management. Their clients are then less likely to meet or beat analysts’ earnings forecasts within one penny. Specifically, the following regressions are estimated: H1: 𝐵𝐸𝐴𝑇𝑡 = 𝑏0 + 𝑏1𝐴𝑈𝐷𝐼𝑇𝑂𝑅𝑁𝐴𝑀𝐸 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑅𝑂𝐴𝑡 + 𝑏4𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏5𝐿𝑂𝑆𝑆𝑡 + 𝑏6𝑀𝐵𝑡 + 𝑏7𝐶𝐹𝑂𝑡 + 𝑏8𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏9𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (11) H2: 𝐵𝐸𝐴𝑇𝑡 = 𝑏0 + 𝑏1𝑆𝑃𝐸𝐶𝐼𝐴𝐿𝐼𝑆𝑇 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑅𝑂𝐴𝑡 + 𝑏4𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏5𝐿𝑂𝑆𝑆𝑡 + 𝑏6𝑀𝐵𝑡 + 𝑏7𝐶𝐹𝑂𝑡 + 𝑏8𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏9𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (12) 3.2.2 Audit Fee

(28)

28 In order to examine the relation of audit fee with the mandatory disclosure, the natural

logarithm of total audit fees (LNFEE), which is based on earlier studies (Carcello et al.2002; Carcello and Li, 2013; Deis and Giroux, 1996). This measure looks whether the mandatory disclosure is related to higher audit fees, as auditors would charge more audit fee for the effort provided through the requirement of the mandatory disclosure (Simunic and Stein, 1996; Zerni, 2012).

The expectation is that AUDITORNAME has a positive coefficient with audit fee and this relation is even more magnified in the presence of the independent variable SPECIALIST (due to their higher likelihood to detect material misstatements). By providing high audit quality more audit effort is required, which is reflected in higher audit fees. Continuous variables and extreme values are winsorized to the 1 and 99 percentiles. Specifically, the following regressions are estimated:

H1: 𝐿𝑁𝐹𝐸𝐸𝑡 = 𝑏0 + 𝑏1𝐴𝑈𝐷𝐼𝑇𝑂𝑅𝑁𝐴𝑀𝐸 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑅𝑂𝐴𝑡 + 𝑏4𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏5𝐿𝑂𝑆𝑆𝑡 + 𝑏6𝑀𝐵𝑡 + 𝑏7𝐶𝐹𝑂𝑡 + 𝑏8𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏9𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (13) H2: 𝐿𝑁𝐹𝐸𝐸𝑡 = 𝑏0 + 𝑏1𝑆𝑃𝐸𝐶𝐼𝐴𝐿𝐼𝑆𝑇 + 𝑏2𝑆𝐼𝑍𝐸𝑡 + 𝑏3𝑅𝑂𝐴𝑡 + 𝑏4𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑡 + 𝑏5𝐿𝑂𝑆𝑆𝑡 + 𝑏6𝑀𝐵𝑡 + 𝑏7𝐶𝐹𝑂𝑡 + 𝑏8𝐿𝐼𝑇𝐼𝐺𝐴𝑇𝐸𝑡 + 𝑏9𝐵𝐼𝐺6𝑡 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 (14) 3.2.3 Control variables

Various control variables are used in this study to control for factors that could affect audit quality, auditor name and auditor industry specialization.

The first control variable is firm size, as conservative accounting increase with the size of firm. Moreover, more timely and less asymmetric information is present with the size of the firm (LaFond and Watts, 2008). It controls the heteroskedasity by using the natural logarithm of total assets for year t.

The second control variable is last year’s current accrual (income before

extraordinary items + depreciation and amortization – operating cash flow, divided by lagged total assets), following Carcello and Li (2013).

The next control variable is firm performance (ROA), controlling for firm performance. This is calculated by dividing income before extraordinary items in year t by total assets for year t (Carcello and Li, 2014).

(29)

29 The fourth control variable is leverage (debt-equity ratio), because firms with higher leverage are more likely to engage in earnings management when they cannot produce enough assets to cover the promised payments to the debt holders (Watts, 2003).

The fifth control variable is loss. Managers have incentives to meet the expectations of investors and therefore would engage in earnings management when losses are made (Bradshaw et al., 2001). This is a dummy variable, which is coded 1 if net income is lower than zero, and 0 otherwise.

The sixth control variable is market to book ratio (market value divided by book value), used to control for firm growth opportunity and market pressures regarding firm’s performance (Carcello and Li, 2013).

The following control variable is the natural log of operating cycle [360/(Sales/Average AR) + 360/(Cost of goods sold)/(Average Inventory)], following Dechow and Dichev (2002); Francis, LaFond and Schipper (2005) and Srinidhi and Gul, (2006).

The eighth control variable is cash flow from operations, following Carcello and Li (2013), Dechow and Dichev (2002), Reichelt and Wang (2009), Balsam et al. (2003).

The ninth control variable is high litigation risk, as firms with their main operations in high-litigation industry (biotechnology, computers, electronics, and retail industries) are more likely to overstate their earnings net assets than firms that understate their net assets (Watts, 2003). Firms with high litigation risk are firms with SIC codes of 2833-2836, 3570 – 3577, 3600-3674, 5200 – 5961, and 7370-7370 and are coded with 1, otherwise 0 (Reichelt and Wang, 2009).

The tenth control variable is Big 4 in order to control for Big N audit firms that would provide more quality than non-Big N audit firms and to control brand name (Balsem et al., 2003; Reichelt and Wang, 2009; Carcello and Li, 2013). This is a dummy variable, which is coded 1 if the firm is audited by a Big 6 firm, and 0 otherwise.

The last control variable is industry. This is a dummies variable, which is 1 for manufacturing forms and trade firms and 0 for service-oriented firms. Different industries can have effect on the earnings management (Vemala et al., 2014).

(30)

30 4 Empirical results

4.1 Univariate results

Industry distribution is shown in Table 4 based on observations and auditor industry expertise. Auditor industry specialists cover 15% of the observations in this sample.

TABLE 4

Industry Distribution

Industry SIC Number of

Observations

Specialists

Agriculture, Forestry and Fishing 0100-0999 0 0

Agricultural, Production-Livestock and Animal Specialties

1000-1999 2765 339

Manufacturing 2000-3999 5385 1,063

Transportation, Communication, Electric, Gas and Sanitary

4000-4999 1435 321

Wholesale and retail estate 5000-5999 2230 317

Financing firms 6000-6999 6650 1,025

Services 7000-8999 5750 530

Public administration 9000-9999 25 15

Total 24,240 3610

Descriptive statistics for the auditor name and expertise are reported in Table 5. Each panel is separated in three categories: before the mandatory disclosure (pre-MD), after the mandatory disclosure and after the mandatory disclosure (post-MD) with the presence of auditor industry expertise.

Panel A of Table 5 provides the mean, standard deviation and t-values for abnormal accruals. Comparing the mean of pre-MD and post-MD, a decrease in mean from 0.119 to 0.098 is reported. This is consistent with H1a and the results of Carcello and Li (2013) and Balsam, Krishnan and Yang (2003). In the presence of auditor industry expertise in post-MD the mean is even lower which is line with H2a Balsam, Krishnan and Yang(2003). Thus, there is significant univariate evidence that firms have lower abnormal accruals after the disclosure requirement. The mean values of control variables show that in post-MD, firms are greater in size and in cash flow from operations. For post-MD with presence of auditor industry expertise, firms are greater in size, have lower returns, higher leverage and are greater in cash flow. Moreover, they operate in high-litigate industries and are audited by a big 4 firm. Similar results are reported for accrual estimation errors in Panel B (except for the