Effects of the new auditor’s report on the decision usefulness of the annual report

(1)

Effects of the New Auditor’s Report on the

Decision Usefulness of the Annual Report

Name: Metin Topalak Student number: 11423110

Thesis supervisor: Dr. E.E.O. Roos Lindgreen Date: June 25, 2018

Word count: 16,742

MSc Accountancy & Control, specialization Accountancy Amsterdam Business School

(2)

Statement of Originality

This document is written by student Metin Topalak 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)

Abstract

The International Auditing and Assurance Standards Board (IAASB) has introduced new

standards for audit reporting. The reason for new standards is to increase the decision usefulness of the reports for users. Experimental studies prior to the introduction of new standards indicate that the extended auditor’s report has a positive effect on the decision usefulness. The studies on the decision usefulness of the extended auditor’s report after the UK Financial Reporting Council (FRC) introduced new standards in the United Kingdom and the Republic of Ireland prior to the global implementation find mixed results while focusing on the same sample (United Kingdom). The event study methodology is used to research whether the extended auditor’s report results in improvement of the decision usefulness. The effects of the new standards on the decision

usefulness are examined using a difference-in-differences research design, with firms listed on STOXX Europe 600 index categorized into non-large cap and large cap firms. The results indicate that the introduction of the new standards has not led to an increase in decision usefulness for users. The price reactions indicate no effect at all for Treatment firms and no difference between large cap and non-large cap firms, while the volume reactions show some effects under certain circumstances for Treatment firms. However, these effects disappear when using the bootstrap method. There are no differences in volume reactions between large cap and non-large cap firms. Altogether, the results indicate that the new standards do not improve the decision usefulness of the auditor’s report. Moreover, the results show no differences between large and non-large cap firms, indicating that there is no proof for the voluntary disclosure theory.

(4)

1.Introduction

After the accounting scandals at the beginning of this century, the accounting profession has experienced a lot of criticism and pressure to improve the quality of their work. In order to improve the quality of the work and the image of the profession, regulators and governments around the world have implemented new laws and rules and have increased the oversight. In the U.S., for example, one of the most important steps was the introduction of the Sarbanes-Oxley Act (SOX) in 2002. The International Auditing and Assurance Standards Board (IAASB), the American Accounting Association (AAA) and the Auditing Standards Board (ASB) of the American Institute of Certified Public Accountants (AICPA) started a joint research initiative in 2006 in for a better understanding of users’ perceptions of the auditor’s report on financial statements and issued a request for proposal for research on unqualified auditor’s report communications (American Institute of Certified Public Accountants, American Accounting Association, International Auditing and Assurance Standards Board, 2006). The research reports completed in 2009. As these researches were conducted, the financial crisis exhibited the

weaknesses of the accounting profession and showed that there was still a lot to do to regain the trust of the stakeholders. The financial crisis has played a catalyzing role in the development of new standards for financial accounting and audit requirements. These new requirements include important changes to disclosure and reporting regulation.

In this thesis, the focus is on the role of the auditor’s report. During the joint project of the IAASB and AICPA, the users of the audit reports have indicated that the standard auditor’s report did not communicate valuable firm specific information and it did not have much value for the users. Academic research confirms these findings as Lennox (2005) finds that almost all companies receive unqualified opinions. 23,611 (83.45%) audit opinions of a sample of 28, from 1995 to 1998 are unqualified and unmodified. 4,635 (16.38%) opinions are unqualified but modified. Only 46 opinions are not qualified. Humphrey et al. (2009) argue that the auditor’s report is not helpful because it contains a lot of standardized statements and little specific

information. In another study, Gray et al. (2011) find that is not clear what the auditor’s report is intended to communicate to the users, preparers and auditors. Moreover, they find that investors perceive the auditor’s reports as boilerplate and therefore not informative. It is also important to point out to the concerns expressed by users during this period. The CFA Society of the UK

(6)

6

(2013) was enthusiastic about the proposed standards but expressed concern that there is risk that the auditor may not disclose something because it is wary of sensitive information becoming public. Furthermore, one respondent in the study of Vanstraelen et al. (2012) expressed that the auditor is limited by what management wants to show and tell the auditor.

As a result of the demand for extended auditor reporting, standard-setters have

considered new standards for auditor’s reports’ for improving the audit quality and the decision usefulness. The IAASB explains the reason for changes in the auditor report as calls from users of financial statements for more informative auditor’s reports; the users expect the auditors to provide more relevant information based on the audit that was performed (IAASB, 2015). According to IAASB, the overall objective of the IAASB’s auditor reporting project has been to enhance the communicative value of the auditor’s report, in the public interest. The IAASB intends several benefits for its new and revised auditor reporting standards. The most important benefit is an increased confidence in the audit and the financial statements as a result of

increased transparency and enhanced informational value. Another intended benefit is enhanced communication between investors and the auditor and between the auditor and those charged with governance. Moreover, IAASB expects an increased attention by management and those charged with governance to the disclosures in the financial statements to which reference is made in the auditor’s report. Finally, the IAASB expects a renewed focus of the auditor on matters to be communicated in the auditor’s report, which could indirectly result in an increase in

professional skepticism (IAASB, 2015).

The IAASB has revised standards ISA 700, ISA 705, ISA 706, ISA 720, ISA 570, ISA 260 and adopted a new standard ISA 701 regarding the auditor’s report. Standard ISA 700 deals with forming an opinion and reporting on Financial Statements. ISA 705 relates to modifications to the opinion in the independent auditor's report and ISA 706 relates to emphasis of matter paragraphs and other matter paragraphs in the independent auditor's report. The revised ISA 720 aims to clarify and increase the auditor’s involvement with “other information”. The revised ISA 570 deals with the auditor’s responsibilities in an audit of financial statements relating to going concern and the implications for the auditor’s report. ISA 260 deals with the auditor’s

responsibility to communicate with those charged with governance in an audit of financial statements. However, the most important change is the adoption of ISA 701, which deals with the auditor’s responsibility to communicate key audit matters (KAM) in the auditor’s report. Key

(7)

7

audit matters are also called critical audit matters (CAM). The IAASB defines key audit matters as those matters that, in the auditor’s professional judgment, were of most significance in the audit of the financial statements of the current period. Key audit matters are selected from matters communicated with those charged with governance (IAASB, 2015). The professional body for accountants in the Netherlands (NBA) responded by adopting Standard 702N for fiscal years ending on or after 15 December 2014 for the financial statements of public interest entities while the IAASB standards only became mandatory for financial statements for period ending on or after 15 December 2016. In addition to the standards of IAASB, the European Commission (EC) implemented Directive 2014/56/EU on statutory audits of annual accounts and consolidated accounts and Regulation (EU) No 537/2014 on specific requirements regarding statutory audit of public-interest entities on 17 June 2016 for all European Union member states. The new

standards became mandatory for annual reports with year ending on or after 30 June 2017. Next to the new audit report, the new EC regulations include an additional report to the audit

committee and other reporting requirements such as reporting irregularities, reports to

supervisors of public interest entities (PIEs) and transparency reporting (European Commission Reform of the EU Statutory Audit Market, 2016).

The new standard regarding the extended auditor’s report made The Netherlands one of the first countries to introduce such standards. Before the introduction in The Netherlands, the UK Financial Reporting Council (FRC) has introduced new standards for fiscal years ending on or after 30 September 2013 in the United Kingdom and the Republic of Ireland. EU member states have voluntarily adopted the ISAs at a national level and require extended auditor’s reports for fiscal year 2016 except Germany, France and Portugal, which follow the EC adoption and introduce the new report for annual reports for fiscal years starting on or after 16 June 2016. In the United States, the Public Company Accounting Oversight Board (PCAOB) set the effective date for all provisions other than those related to critical audit matters for audits for fiscal years ending on or after December 15, 2017. Provisions for large accelerated filers related to critical audit matters will take effect for audits for fiscal years ending on or after June 30, 2019, while the provisions will be effective for all other companies to which the requirements apply, for fiscal years ending on or after December 15, 2020.

Although there are some minor differences between the standards of the NBA, FRC, IAASB, EU and PCAOB, the key changes of the new auditor’s report are the structure, which

(8)

8

places the opinion to the beginning of the report instead of the end, the inclusion of key/critical audit matters, additional information regarding going concern, the qualitative and quantitative aspects of materiality and the scope of the group audit.

Prior to the introduction of the new standards, there have been experimental studies regarding the information value of an extended report. Boolaky and Quick (2016) have done a survey-based study on the impact of expanded audit reports, on bank director perceptions of the quality of the financial statements, the audit and the audit report, and their credit approval

decisions in Germany. The findings suggest that disclosing the assurance level has a significantly positive impact, while there is not a material effect of expanding the audit report to include the materiality level or KAM. Christensen et al. (2014) conduct an experiment before the

introduction of the new audit report regarding the influence of Critical Audit Matter (CAM) Paragraphs on Nonprofessional Investors' decisions to invest. They find an information effect (investors who received a CAM paragraph are more likely to change their investment decision than investors who received the standard audit report) and a source credibility effect (investors who received a CAM paragraph are more likely to change their investment decision than investors who received the same CAM paragraph information in management’s footnotes). In another experimental study, Doxey (2014) finds that auditor-provided estimate disclosures are value-relevant for users’ investment decisions. Moreover, he finds that in case of an unqualified opinion, users view auditors as more (less) independent when auditors agree (disagree) with management and management as less (more) credible when estimates are incentive consistent (inconsistent). Sirois et al. (2016) conduct an experimental eye-tracking study on the inclusion of key audit matters in the auditor’s report and conclude that the additional information has

attention directing value; users’ perception of the audit seems to be negatively affected. The communication of additional information is associated with lower perceived audit quality.

As the first introducers of the standards, NBA and FRC have both studied the effects of the extended auditor’s report and concluded that the reports included more information than before (Inzicht in de uitgebreide controleverklaring, 2015) and that the auditors had met the requirements of the new standard (Extended auditor's reports: A review of experience in the first year, 2015). Besides the research of these regulatory bodies, there have been academic studies on the consequences of this extended auditor’s report. Most of these researches are based on the effects of the new standards in the United Kingdom. Reid et al. (2018) study the effects of the

(9)

9

new standards on the audit quality and find that the financial reporting quality as proxied by absolute abnormal accruals, the propensity to just meet or beat analyst forecasts, and earnings response coefficients, has improved after the new standard. Gutierrez et al. (2018) also study the effects in the UK but do not find any differences in the decision usefulness, audit quality and audit fees related to the extended report. Lennox et al. (2017) study the effects on the information value (decision usefulness) of the new report in the UK and conclude that investors do not find these disclosures incrementally informative. They indicate that the disclosures reliably capture the uncertainty in accounting measurements but lack incremental information content because most of the risks had already been disclosed previously by management (in earning’s

announcements, conference calls, or previous year’s annual reports). So, investors were already informed about the risks before these were disclosed by auditors in the expanded audit reports. It is remarkable that these studies find mixed results while both focusing on the same sample (United Kingdom). Li et al. (2018) investigate the impact of audit reporting changes on audit quality and audit fees in the New Zealand context and find an improvement in audit quality as proxied by a reduction in absolute abnormal accruals upon the adoption of the new audit reporting requirements.

Besides the studies on the effects of the new report on the audit quality (including decision usefulness) and audit fees, there have been several experimental studies on the effects on the litigation effects of the extended report on auditors. Basel et al. (2016) find evidence that CAM disclosures can reduce auditor liability judgments in some cases. They find that disclosure of any related or unrelated CAM provides litigation protection instead of stating that there were no CAMs indicates that auditors might be incentivized to expand CAM disclosures which could result in developing boilerplate CAM disclosures. Backof et al. (2017) find that clarifying the meaning of reasonable assurance mitigates the increase in auditors’ liability exposure by

reducing the expectations of auditors. They find that without this clarification, jurors perceive as more negligent when the audit report includes a related CAM report compared to reports

without. Kachelmeier et al. (2018) investigate whether CAM disclosures protect auditors by forewarning users of misstatement risk. They (2001, p. 2) find that “CAM disclosures lower user

confidence in CAM-related financial statement areas prior to any knowledge of a misstatement and lower assessments of auditor responsibility for misstatements subsequently revealed in CAM-related areas”.

(10)

10

Because of the mixed results regarding the audit quality and information value in the United Kingdom, and the availability of more data after the international adoption of the standards for fiscal year 2016, it will be a valuable contribution to take the effects of the new standards in Europe into consideration and study the effects of the new standards on decision usefulness for large and non-large cap firms in Europe. This could be valuable in four ways. First, it can help eliminate country specific characteristics by using data from more countries. Second, because there is no clear answer whether the introduction of this new standard has led to improvements in decision usefulness, it can provide more clarity regarding the effects of the new standards. Third, there has not been a study examining the effects of the new standards on large and non-large cap firms and whether there are differences between these two groups. Fourth, the auditors have had the opportunity to learn how to apply the new standards in the United

Kingdom, Ireland and The Netherlands and have gained experience and knowledge which can be transferred to other settings (countries). Therefore, the auditors are not expected to face the same difficulties as they did in the early adopter countries.

In this paper, the following research question is examined: has the new auditor’s report improved the decision usefulness of the annual report? This research provides insights about the effects of the new auditor’s report in Europe for large and non-large cap firms.

Section 2 reviews the literature. Section 3 presents the research hypotheses. Section 4 outlines the research method and design. Section 5 provides the empirical results. Section 6 concludes with limitations and suggestions for future research.

2.Literature

Three important theories form the basis of this paper. The agency theory of Jensen and Meckling (1976), the voluntary disclosure theory and the decision usefulness approach of financial reporting. In section 1, information is provided regarding the agency theory. The voluntary disclosure theory is discussed in section 2. The decision usefulness approach is explained in section 3. Finally, section 4 presents the event study methodology.

2.1 Agency theory

Wallace (2004) recognizes the agency theory as one of the most important features of the audit in free and regulated markets. Jensen and Meckling (1976) notice that due to the separation

(11)

11

of ownership and control of the firm, frictions can arise due to asymmetric information and imperfectly aligned incentives between the principal and the agent, which can result in moral hazard (hidden action) and adverse selection (hidden information). The resulting costs are called agency costs. The principal is the market participant (owner) who provides capital to the agent (who is in control) and in return, the agent provides financial and other information to the principal. The auditor plays an important role between the principal and the agent. On the one hand, the auditor is hired by the agent and on the other hand, the task of the auditor is to narrow the information asymmetry and align incentives by auditing the financial statements and

contemporaneously the sustainability figures of firms and issuing reports about these.

Historically, aligning the incentives (stewardship) and reducing information asymmetry (decision usefulness) were the two major objectives of financial reporting (Pelger, 2016). The decision usefulness objective concerns the dissemination of (private) information and the emphasis is on relevance and timeliness. This helps the investor (principal) in choosing among alternative investments. The stewardship objective is about measuring performance for

monitoring purposes and it emphasizes on reliability and conservatism. It is about the capital allocation within the firm. Over the years, the decision usefulness objective has gained importance and became the most important objective of financial reporting, while the

stewardship function has become less important. The annual report is an important tool to narrow the information gap between the principal and the agent. For the users of the annual report it is important that it is audited and that the auditor provides reasonable assurance about whether the financial statements as a whole are free from material misstatement. Due to the numerous accounting scandals and deterioration of confidence of the society in the profession of the accountant and auditor, the principals require that the reports provide high quality information. The auditor’s report is an important part of the annual report, although until a couple of years, this report was very standardized and did not provide much information. Church et al. (2008) and Carcello (2012) conclude in their papers prior to the introduction of new standards that the audit report does not communicate useful information to the user of financial reports.

2.2 Voluntary disclosure theory

Miller and Skinner (2015) argue that the disclosure literature has emerged as one of the most important areas in accounting research over the past two decades, because researchers

(12)

12

realized that the way managers communicate information about their firms have important effects on capital market outcomes. They study how technological change and the emergence of new forms of media impact manager’s disclosure practices and find after talking to CFOs and other practitioners that managers spend significant time thinking about ways to manage their firms’ disclosures and that managers believe that their disclosure decisions have important value implications. Milgrom (1981) theorizes that managers will timely disclose good news and that news that is not disclosed will be interpreted as bad news. As a result, managers will have to disclose all news. The studies by Verrecchia (1983), Dye (1985), and Lang and Lundholm (1993) form the foundation for the voluntary disclosure theory. Verrecchia (1983) shows in his model how costs related to disclosures offer an explanation for the reasons of the management to use their discretion in disclosing information. In the model of Verrecchia, costs related to

disclosures are not only costs of preparing and disseminating information, but these costs also include proprietary costs.

Proprietary costs of disclosure are costs involved with the dissemination of good

information which has negative consequences for the firm (e.g., competition) and result in lower profits. If proprietary costs exist but there is no disclosure, investors are not sure whether

information is withheld because (i) the information is bad news, or (ii) the information is good news but not sufficient to pass the threshold. Therefore, investors are unable to interpret the withheld information and react less strongly. Dye (1985) identifies three explanations for management’s failure to disclose information. First, investors’ knowledge management’s

information is not complete, which allows managers to successfully suppress bad news. Second, managers possess a vast array of private information and some of this may be proprietary. Nonproprietary information may not be disclosed if it is part of such an array. Third, the release of information would be costly either directly or indirectly. Dye (2001) argues that the voluntary disclosure theory is most interesting in the way it demonstrates how to interpret less than full disclosure. According to Dye (2001, p. 184): “any entity contemplating making a disclosure will

disclose information that is favorable to the entity and will not disclose information

unfavorable to the entity. Moreover, in order to interpret sensibly the remarks of the entity making - or not making - a disclosure, one should anticipate the entity’s incentives to behave in the preceding fashion”. Healy and Palepu (2001) identify several costs and benefits as

(13)

13

The benefits relate to the reduction of information asymmetry such as the reduction of estimation risk and undervaluation and managerial talent signaling. These determinants have consequences on liquidity, cost of capital, information intermediation, litigation and

volatility. Skinner (1994) argues that there are at least two reasons for managers to disclose negative earnings surprises. First, there is litigation risk if they choose not to disclose. So, managers will voluntarily disclose bad news in order to preempt such lawsuits and reduce the damages of these. Second, there might be reputational costs for the management if they fail to disclose.

King et al. (1990) hypothesize that disclosure is positively correlated with firm size, as for larger firms the gains from having foreknowledge of earnings are greater. Lang and Lundholm (1993) expect a relation between disclosure and firm size if disclosure cost is decreasing in firm size. They argue that the consideration of firm size by the FASB and SEC in mandatory disclosure requirements suggests that there may be a fixed part of disclosure costs, which results in a decreasing cost per unit of size. Consistent with their expectations, they find that disclosures are correlated with firm size. It is costlier for large firms not to disclose; therefore, larger firms disclose more information.

2.3 The decision usefulness approach

Staubus gives in his book The decision-usefulness theory of accounting: A limited history (2000) a description of the increasing importance that the decision usefulness approach has gained throughout the years and explains that it has become the primary objective of financial reporting.

The decision usefulness approach was first developed by Staubus as “a theory of accounting to investors” in his dissertation in Chicago (University of Chicago). After reviewing the historical literature, Zeff (2013) finds that previously all authors have identified a separate role for

stewardship in their framework until the International Accounting Standards Board (IASB) and the Financial Accounting Standards Board (FASB), in their joint Conceptual Framework

published in 2010, dismissed it from separate standing. In their joint Conceptual Framework for Financial Reporting, published in 2010, the Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) defined the objective of general purpose

(14)

14

financial reporting, which is to provide financial information about the reporting entity that is useful to existing and potential investors, lenders, and other creditors in making decisions about providing resources to the entity. Those decisions involve buying, selling, or holding equity and debt instruments and providing or settling loans and other forms of credit (OB2). In the

Conceptual Framework, stewardship is defined as to assess an entity’s prospects for future net cash inflows, existing and potential investors, lenders, and other creditors need information about the resources of the entity, claims against the entity, and how efficiently and effectively the entity’s management and governing board have discharged their responsibilities to use the

entity’s resources (OB4). Pelger (2016) describes how decision usefulness has gained importance and became the most important objective of financial reporting, whereas the stewardship

function has become less important over the years. Pelger discusses how the elimination of stewardship was driven by the focus on one particular body of knowledge, the decision usefulness programme which was based on the US neoliberal government rationality of the Chicago School (Foucault, 2008), which had been introduced into the US standard-setting field by academics in the 1970s. Stewardship was historically one of the main objectives of financial reporting. One of the objectives of Financial Reporting was, according to the Statement of Financial Accounting Concepts No. 1 of the FASB, to provide information about an enterprise’s financial performance during a period and about how management of an enterprise has

discharged its stewardship responsibility to owners (FASB, 1978). However, stewardship was downgraded because of a lack of a competing rationality as the neoliberal rationality which has driven the decision usefulness approach.

The expansion of the decision usefulness approach has not been limited to financial accounting, it has also expanded into the field of auditing. In the early years of the profession, the core function of the profession was rooted in the confidence that society places in the

effectiveness of the audit and in the opinion of the auditor (Limperg, 1932). Limperg argues that confidence is the key and that the auditor should perform his work in a manner which would not betray the confidence of the society. Moreover, Limperg argues that the auditor should not raise expectations beyond the work performed and the justifiable expertise. In an environment of confidence, it was sufficient for the auditor to just sign off. However, throughout the years, the confidence of the society in the auditor and the auditing profession has deteriorated because of the numerous cases of fraud, corruption and accounting scandals. While it is not the topic of this

(15)

15

study, there are studies indicating that the numerous scandals and fraud cases are not a cause for the expansion of neoliberalism, but an effect of neoliberalism itself. Wiegratz and Whyte (2016) notice that there is an assumption in academic literature that corruption and fraud will diminish with the expansion of neoliberalism as global markets become more open, competitive and efficient. In contrast, they argue that the rise in a range of business and consumer frauds in many countries can be explained as a direct consequence of the liberalization of market, the

reregulation of particular economic sectors and the restructuring of the state. Guénin-Paracini et al. (2014) argue that the resilience of neoliberalism partly ensues from the tendency to foster spontaneous and widespread processes of scapegoating in times of turmoil. Not the system itself, but particular actors are made accountable when things go wrong, which protects the regime from systemic questioning. Next to the erosion of confidence, an expectation gap has emerged between the auditor and the principle. Mock et al. (2013) indicate that there is a significant expectations gap between the expectations of the public regarding what an audit delivers and what the audit profession believes what it provides. The erosion of confidence and a widening expectation gap, taken together with the expansion of the decision usefulness approach, has led to a demand for more information by the users (society).

2.4 Event study

Event studies are extensively used in accounting and finance research to measure the effects of an economic event on the value of firms. Kothari and Warner (2004) find 565 papers reporting event study results for the years 1974 through 2000, despite many academic and practitioner-oriented journals are excluded. An event study measures the impact of a specific event on the value of a firm using financial market data (MacKinley, 1997).

The use of an event study is related to the assumption of the existence of an efficient market. In an efficient market, all information regarding a firm is immediately reflected in the security price (Ball and Brown, 1968; Fama, 1970). In the case of an event study, the information released to the market is related to an event such as an earnings announcement, issuance of debt etc. In this study, the event is the publication of the auditor's report. Two commonly used models for the normal return, according to MacKinley (1997), are the constant mean return model and the market model. While the constant mean return model assumes that the mean return of given security is constant over time, the market model assumes a linear relation between the market

(16)

16

return and the security return. MacKinley (1997) identifies James Dolley (1933), who examines the price effects of stock splits, as one of the first published studies. While there have been publications after this study, the methodology used today stems from the studies by Ball and Brown (1968) and Fama et al. (1969). Corrado (2011) argues that there are a number of factors that contributed to the success of these studies. These factors are their use of the 'market model', their use of data from the Center for Research in Security Prices (CRSP) at the University of Chicago, and the expanding access to computer systems with statistical software. Fama (1970) reports that the "market model" was originally suggested by Markowitz (1952). The event study methodology is used to research whether the extended auditor’s report results in improvement of the decision usefulness.

3.Hypothesis development

There have been several experimental studies regarding the information value of the extended report. Boolaky and Quick (2016) find that disclosing the assurance level has a significantly positive impact, while there is no material effect of including the materiality level or KAM. Christensen et al. (2014) find an information effect and a source credibility effect. The former indicates that investors who received a CAM paragraph are more likely to change their investment decision than investors who received the standard audit report, the latter indicates that investors who received a CAM paragraph are more likely to change their investment decision than investors who received the same CAM paragraph information in management’s footnotes. Doxey (2014) finds that auditor-provided estimate disclosures are value-relevant for users’ investment decisions and that users view auditors as more (less) independent when auditors agree (disagree) with management and management as less (more) credible when estimates are incentive consistent (inconsistent). Sirois et al. (2016) conclude that the additional information in the report has attention directing value; users’ perception of the audit seems to be negatively affected. The communication of additional information is associated with lower perceived audit quality.

Gutierrez et al. (2018) do not find any differences in the decision usefulness, audit quality and audit fees related to the extended report. Lennox et al. (2017) conclude that the disclosures reliably capture the uncertainty in accounting measurements but lack incremental information content because most of the risks had already been disclosed previously by management. So,

(17)

17

investors were already informed about the risks before these were disclosed by auditors in the expanded audit reports. Reid et al. (2018) find that the financial reporting quality has improved after the new standard. It is remarkable that these studies find mixed results while both focusing on the same sample (United Kingdom). Li et al. (2018) find an improvement in audit quality as proxied by a reduction in absolute abnormal accruals upon the adoption of the new audit reporting requirements in New Zealand. Besides the studies on the effects of the new report on the audit quality, information value (decision usefulness) and audit fees, there have been several experimental studies on the effects on the litigation effects of the extended report on auditors. Basel et al. (2016) find that disclosure of any related or unrelated CAM provides litigation protection compared to stating that there were no CAMs, which indicates that auditors might be incentivized to expand CAM disclosures which could result in developing boilerplate CAM disclosures. Backof et al. (2017) find that clarifying the meaning of reasonable assurance mitigates the increase in auditors’ liability exposure by reducing the expectations of auditors. They find that without this clarification, jurors perceive as more negligent when the audit report includes a related CAM report compared to reports without. Kachelmeier et al. (2018) find that CAM disclosures lower user confidence in CAM-related financial statement areas prior to any knowledge of a misstatement and lower assessments of auditor responsibility for misstatements subsequently revealed in those areas. While disclosing material information can decrease auditors’ liability exposure, the auditors may not disclose something because it is wary of sensitive information becoming public (The CFA Society of the UK, 2013). Furthermore, the auditor is limited by what management wants to show and tell the auditor (Vanstraelen et al., 2012).

According to the agency theory, the additional information provided by the auditor will narrow the information gap between the principal and the agent and therefore will improve the decision usefulness of the annual report. However, the voluntary disclosure theory suggests that the disclosure of material information by the auditor might result in costs for the firm, which the firm might want to prevent by voluntarily disclosing the information prior to the release of the auditor’s report. Lang and Lundholm (1993) expect and find a relation between disclosure and firm size if disclosure cost is decreasing in firm size; disclosures are correlated with firm size. Therefore, the information disclosed by the auditor about large capitalization firms will not be new information, but information that was previously disclosed by the firm, because of the high

(18)

18

costs for the firm not to disclose these costs. The costs of not disclosing for smaller firms such as mid capitalization and small capitalization firms are lower. So, these firms are more likely to withhold information which can be disclosed by the auditor. As a consequence, the auditors are expected to disclose valuable information regarding these firms.

These theoretical arguments lead to the following hypothesis:

H1: The new auditor’s report does only improve the decision usefulness of the annual report in Europe for non-large cap firms.

The costs for not disclosing material information will be higher for large cap firms compared to mid-cap and small cap firms. Especially, the managers of these firms will try to avoid damage to their reputation and want to signal good behavior by disclosing the material information voluntarily before the auditor’s report. For smaller firms, the costs for not disclosing material information is lower. Hence, for managers of small and mid-cap firms, the threshold to

voluntarily disclose is higher. Therefore, this study expects to find differences between large cap, non-large cap firms due to the differences in costs for these firms:

H1a: The new auditor’s report does not improve the decision usefulness of the annual report for large cap firms in Europe.

H1b: The new auditor’s report does improve the decision usefulness of the annual report for non-large cap firms in Europe.

4.Research method and design

The decision usefulness approach is related to the perception of the users, the demand side of audit. Therefore, measures relating to investor reactions are used to evaluate the decision usefulness. An event study measures the impact of a specific event on the value of a firm using financial market data (MacKinley, 1997) The impact of the new standards on the decision usefulness of the audit report is measured based on the market reactions. The market reactions are measured by two proxies. The first proxy is the cumulative abnormal returns (CAR) around

(19)

19

the public distribution of the annual report, the second proxy is the abnormal trading volume (ABVOL) around the filing date of the report. In order to evaluate the impact, audit report publication year 2016 (pre-adoption period) is compared to publication year 2017 (post-adoption period).

Figure 1. Time line for an event study (MacKinley, 1997).

An event study consists of three phases as depicted in Figure 1 (MacKinley, 1997). The first phase is the estimation window, the second phase is the event window and the last phase is the post-event window. The event date is defined as τ = 0, τ = T₁ + 1 to τ = T₂ represents the event window, and τ = T₀ + 1 to τ = T₁ is the estimation window. The length of the estimation window (L₁) is L₁ = T₁ - T₀ and the length of the event window (L₂) is L₂ = T₂ - T₁. The post-event window is from τ = T₂ + 1 to τ = T₃ and the length (L₃) is L₃ = T₃ - T₂. MacKinley (1997) states that even if the event is an announcement on given date, it is typical to set the event window length to be larger than one. Furthermore, he states that including the event window in the estimation of the normal model parameters could lead to the event returns having a large influence on the normal return measure, and this would be problematic because the methodology is built around the assumption that the event impact is captured by the abnormal returns.

Therefore, the estimation window and the event do not overlap in this study. The post-event window is only used occasionally to estimate the normal return model. In this study, there is no application of the post event window.

The actual returns (Rᵢₜ) in both the estimation window and the event window are

calculated by comparing the current price (Pᵢₜ) of a stock (i) on day (t) to the price of the previous day (Pᵢₜ₋₁). Ordinary least squares (OLS) is a consistent estimation procedure for the market model parameters (MacKinley, 1997). For any security i the market model is (MacKinley, 1997):

estimation window event window post-event window τ 𝑇0 𝑇1 0 𝑇2 𝑇3

(20)

20

𝑅_𝑖𝜏 = 𝛼_𝑖+ 𝛽_𝑖𝑅_𝑚𝜏 + 𝜀_𝑖𝜏

Where 𝑅_𝑖𝜏 and 𝑅_𝑚𝜏 are the period t returns on security i and the market portfolio and 𝜀_𝑖𝜏 is the zero mean disturbance term. The firm-specific return 𝜀𝑖𝜏 is unrelated to the overall market and has an expected value of zero (Corrado, 2011). An advantage of the market model is that it can lead to increased ability to detect event effects by reducing the variance of the abnormal return. The benefit of the market model depends on the 𝑅2 of the market model regression. A higher 𝑅2 leads to greater variance reduction of the abnormal return. The abnormal returns are calculated after obtaining the market model parameter estimates. The formula to calculate the abnormal returns is (MacKinley, 1997):

𝐴𝑅_𝑖𝜏 = 𝑅_𝑖𝜏− 𝛼_𝑖 − 𝛽_𝑖𝑅_𝑚𝜏

𝐴𝑅_𝑖𝜏 is the sample of L₂ abnormal returns for firm i in the event window. The cumulative abnormal return (CAR) is the sum of the abnormal returns for firm i in the event window (L₂).

Bamber and Cheon (1995) find a positive relation between the magnitudes of price and volume reactions in general. However, in nearly a quarter of the announcements they study, they find a difference in relative magnitudes between price and volume reactions. Therefore, the relation between the two measures is closer to independence than to a strong positive relation. Furthermore, Kandel and Pearson (1995) find that there are significant positive abnormal volumes associated with announcements even when the prices do not react to the

announcements. While absolute abnormal returns capture the average change in investors’ belief in response to an event (Gutierrez et al., 2018), abnormal trading volume proxies for opinion divergence among investors (Garfinkel and Sokobin, 2006). Beaver (1968) indicates that trading volume reflects a lack of consensus among investors regarding the price of a firm’s shares and concludes that volume reactions may be better suited to test the usefulness of disclosures than price reactions. Bamber et al. (2011) argue that abnormal return reactions have the potential to yield insights regarding information asymmetry and investor disagreement. Moreover, Cready and Hurtt (2002) find evidence that volume-based metrics provide more powerful tests of

(21)

21

with volume-based metrics increases the power of tests to detect investor response, especially when power is critical due to small sample sizes or small anticipated investor response. Hence, both measures are used in this study to detect investors’ reactions to the new report. 𝐴𝐵𝑉𝑂𝐿_𝑖𝜏 is measured as the natural logarithm of the firm’s average event-period volume divided by the firm’s mean estimation-period volume (Gutierrez et al., 2018). The event window in this study consists of days -1 to +1 relative to the annual report (containing the auditor’s report) filing day (day 0) as used by Gutierrez et al. (2018). The estimation window consists of days -45 to -10 relative to the filing date as used in the study of Ball et al. (2012). The use of the control variables natural logarithm of total market value (LOGMKT), profitability (ROA, LOSS and CHNI), market-to-book ratio (MTB), leverage (LEV), use of a Big4 auditor (BIG4) and industry fixed effects follow prior literature (Reichelt and Wang, 2010; Reid et al., 2018; Gutierrez et al., 2018).

No major financial accounting standards are introduced in the study period. Important IFRS projects as IFRS 9 Financial instruments (effective date 1 January 2018), IFRS 15 Revenue from contracts with customers (effective date 1 January 2018) and IFRS 16 Leases (effective date 1 January 2019) are introduced after this period (IFRS). During the period from January 2016 to December 2017 there were time-related trends in the STOXX Europe 600 index. Figure

2 illustrates the closing prices for this period. The figure shows a clear difference in closing

prices between the years 2016 and 2017. Therefore, following Gutierrez et al. (2018) this study implements a difference-in-differences research design (DD) to mitigate the effect of time-related trends with the adoption of the new standards, inlcuding economic, political and other factors. Gutierrez et al. (2018) state that the new standards in general constitute an unexpected and transparent source of variation in the auditor’s report content.

The investor reactions are measured by two proxies. The first proxy is the three-day cumulative abnormal returns (CAR) around the public distribution of the annual report, the second proxy is the sum of three-day abnormal trading volume (ABVOL) around the filing date of the report. The following model is used to estimate the cumulative abnormal returns (CAR):

(22)

22

𝐶𝐴𝑅_𝑖,𝜏 = 𝛽₀+ 𝛽_𝟏𝑷𝑶𝑺𝑻_𝒊,𝝉+ 𝜷₂𝑻𝑹𝑬𝑨𝑻_𝒊,𝝉+ 𝜷_𝟑𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬_𝒊,𝝉+ 𝜷_𝟒𝑷𝑶𝑺𝑻 ∗ 𝑻𝑹𝑬𝑨𝑻_𝒊,𝝉 + 𝜷_𝟓𝑷𝑶𝑺𝑻 ∗ 𝑻𝑹𝑬𝑨𝑻 ∗ 𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬_𝒊,𝝉+ 𝛽₇𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏+ 𝛽₈𝑅𝑂𝐴_𝑖,𝜏 + 𝛽₉𝐿𝑂𝑆𝑆_𝑖,𝜏+ 𝛽₁₀𝐵𝐼𝐺4_𝑖,𝑡 + 𝛽₁₁𝑀𝑇𝐵_𝑖,𝜏+ 𝛽₁₂𝐿𝐸𝑉_𝑖,𝜏+ 𝛽₁₃𝐶𝐻𝑁𝐼_𝑖,𝜏 + 𝛽₁₄𝐴𝐵𝑅𝐸𝑇_{𝐸𝐴𝑅𝑖,𝜏}+ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐹𝐸 + 𝜀_𝑖,𝜏

Figure 2 STOXX Europe 600 Closing Prices 2016-2017

Similarly, the following model is used to estimate the abnormal trading volume (ABVOL): 𝐴𝐵𝑉𝑂𝐿𝑖,𝜏 = 𝛽0 + 𝛽𝟏𝑷𝑶𝑺𝑻𝒊,𝝉 + 𝜷2𝑻𝑹𝑬𝑨𝑻𝒊,𝝉+ 𝜷𝟑𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬𝒊,𝝉+ 𝜷𝟒𝑷𝑶𝑺𝑻 ∗ 𝑻𝑹𝑬𝑨𝑻𝒊,𝝉

+ 𝜷_𝟓𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬_𝒊,𝝉+ 𝜷_𝟔𝑷𝑶𝑺𝑻 ∗ 𝑻𝑹𝑬𝑨𝑻 ∗ 𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬_𝒊,𝝉+ 𝛽₇𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏 + 𝛽8𝑅𝑂𝐴𝑖,𝜏+ 𝛽9𝐿𝑂𝑆𝑆𝑖,𝜏+ 𝛽10𝐵𝐼𝐺4𝑖,𝑡+ 𝛽11𝑀𝑇𝐵𝑖,𝜏+ 𝛽12𝐿𝐸𝑉𝑖,𝜏+ 𝛽13𝐶𝐻𝑁𝐼𝑖,𝜏 + 𝛽₁₄𝐴𝐵𝑅𝐸𝑇_𝐸𝐴𝑅_𝑖,𝜏+ 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐹𝐸 + 𝜀_𝑖,𝜏

(23)

23

where for firm i in year t, 𝑷𝑶𝑺𝑻_𝒊,𝝉 is a time-period indicator that equals one for the period following the new standards, and zero otherwise; 𝑻𝑹𝑬𝑨𝑻_𝒊,𝝉 is an indicator variable for the treatment and control firms that is equal to one for firms from non-early adopting countries and zero for firms from early adopting countries; 𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬_𝒊,𝝉 is an indicator for the capitalization of the firms and is equal to one for large capitalization firms and zero otherwise; 𝑷𝑶𝑺𝑻 ∗

𝑨𝑫𝑶𝑷𝑻_𝒊,𝝉 is an interaction of the treatment and time indicator variable; 𝑷𝑶𝑺𝑻 ∗ 𝑨𝑫𝑶𝑷𝑻 ∗ 𝑵𝑶𝑵𝑳𝑨𝑹𝑮𝑬_𝒊,𝝉 is an interaction of the treatment and time indicator variable and the non-large indicator; 𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏 is the natural logarithm of total market value; 𝑅𝑂𝐴_𝑖,𝜏 is net income before extraordinary items divided by total assets; 𝐿𝑂𝑆𝑆_𝑖,𝜏is an indicator variable which is equal to one if 𝑅𝑂𝐴_𝑖,𝜏 is negative and zero otherwise; 𝑀𝑇𝐵_𝑖,𝜏 is the market-to-book ratio; 𝐿𝐸𝑉_𝑖,𝜏 is the long term debt divided by total assets; 𝐶𝐻𝑁𝐼_𝑖,𝜏 is (𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑏𝑒𝑓𝑜𝑟𝑒 𝑒𝑥𝑡𝑟𝑎𝑜𝑟𝑑𝑖𝑛𝑎𝑟𝑦 𝑖𝑡𝑒𝑚𝑠_𝑡 - 𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑏𝑒𝑓𝑜𝑟𝑒 𝑒𝑥𝑡𝑟𝑎𝑜𝑟𝑑𝑖𝑛𝑎𝑟𝑦 𝑖𝑡𝑒𝑚𝑠_𝑡−1)/𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠_𝑡−1; 𝐴𝐵𝑅𝐸𝑇_𝐸𝐴𝑅_𝑖,𝜏 is the sum of the three-day absolute abnormal returns around the annual report release date by company; and 𝐵𝐼𝐺4_𝑖,𝜏 is equal to 1 if the company is audited by a Big-4 auditor, and 0 otherwise; and

𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐹𝐸 are industry fixed effects. The pre-adoption data is compared to the post-adoption data. The unit of analysis is year (“it”), because annual reports are issued once per firm-year.

4.1 Sample

This study focuses on the introduction of the extended auditors report in Europe, implemented by the IAASB for financial statements for period ending on or after 15 December 2016 and the European Union for annual reports with year ending on or after 30 June 2017. The sample consists of firms listed on the STOXX Europe 600 index on 1 January 2016. The sample covers the period 1 January 2016 to 31 December 2017. The index consists of large, mid and small cap stocks from the STOXX Europe Large 200, STOXX Europe Mid 200 and STOXX Europe Small 200 indices. In 2016, firms listed on the London Stock Exchange, Irish Stock Exchange and Euronext Amsterdam were already applying the new auditor’s report as FRC and NBA were early adopters of the standards. This group of firms is used as a control group in this study. On 1 January 2016 the STOXX Europe 600 index consisted of 599 equities and 1

(24)

24 December 2017.

Accounting data for European companies are obtained through the Compustat Global and Datastream databases, the constituents of the STOXX Europe 600 index are obtained through the Compustat Global and Datastream databases and the publication dates of the annual reports are hand collected. Company announcements of annual report filings are searched using the websites of Luxembourg Stock Exchange; Irish Stock Exchange; Dutch Autoriteit Financiële Markten (AFM); Belgian Financial Services and Markets Authority (FSMA); Spanish Comisión Nacional de Mercado de Valores (CNMV); French Autorité des Marchés Financiers (AMF);

German Unternehmensregister and Dgap; Italian 1info, eMarket STORAGE and Borsa Italiana; London Stock Exchange Regulatory News Service (RNS) and Morningstar National Storage Mechanism; NASDAQ Nordic); and Oslo Stock Exchange. The annual report filings of firms

Figure 1 Example of search of annual report filing dates

from Austria, Czech Republic and Switzerland are obtained through company websites. The annual report publication data could not be found for four firms. The majority of German, French and Portuguese firms have not reported according to the new auditor’s reporting in 2017. After elimination of these firms, the sample consists of 400 firms which applied the new auditor’s report in 2017. The sample includes firms from The Netherlands, United Kingdom and Ireland which are classified as the control group.

Table 1 exhibits the primary exchanges of the sample. The sample consists of firms from

(25)

25

consists of 153 firms listed on the LSE. Table 2 illustrates the auditors of the sample. The majority of the sample firms are audited by BIG4 firms. Only eight audits are performed by non-BIG4 auditors, while PriceWaterhouseCoopers has performed 260 audits.

Table 1 Stock Exchange listings

Stock Exchange Number of firms

Irish Stock Exchange 8

Euronext Amsterdam 25

Euronext Brussels 13

Euronext Paris 3

OMX Helsinki 15

London Stock Exchange 153

BME Spain 28

Borsa Italiana 24

OMX Copenhagen 18

Oslo Stock Exchange 10

Prague Stock Exchange 2

SIX Swiss 45

OMX Stockholm 38

Vienna Stock Exchange 4

XETRA Germany 14

Total 400

Table 2 Auditors

Auditor Audits

BDO International 4

Deloitte Touche Tohmatsu 189

Ernst & Young 171

Grant Thornton 4

KPMG 172

PriceWaterhouseCoopers 260

Total 800

5.Empirical Results

After dropping observations for missing data for the variables, the continuous variables are winsorized at 1 and 99 percent, except for 𝐶𝐴𝑅𝑖,𝜏 and 𝐴𝐵𝑉𝑂𝐿𝑖,𝜏 (Ball et al., 2012; Gutierrez et al., 2018) to ensure that the results are not driven by outliers. To solve for the

(26)

26

errors that are additionally corrected for unwanted correlation. The standard errors are clustered based at the firm-level, with gvkey as the firm identifying variable (Veenman, 2013) in order to control for company fixed effects. Furthermore, standard errors have been calculated using the bootstrap method. The bootstrap method is one of several nonparametric methods for estimating standard errors and performs best among the genuinely nonparametric methods as the bootstrap estimate is the nonparametric maximum likelihood estimate of the standard error (Efron, 1981). According to Efron and Tibshirani (1997) bootstrap procedures can reduce the variability of error rate predictions substantially. DiCiccio and Efron (1996) discuss bootstrap methods for constructing highly accurate approximate confidence intervals and find that bootstrap intervals are more accurate than standard intervals. Bootstrap method involves taking multiple samples (1,000 replications in this study) from the same sample to estimate the accuracy of the estimates regarding the population.

5.1 CAR Analysis

Table 3 shows the descriptive statistics and a test of differences in means for cumulative

abnormal returns before and after the global adoption of the new rules (𝑃𝑂𝑆𝑇𝑖,𝜏=1 and 𝑃𝑂𝑆𝑇𝑖,𝜏=0), treatment (𝑇𝑅𝐸𝐴𝑇𝑖,𝜏=1) and control firms (𝑇𝑅𝐸𝐴𝑇𝑖,𝜏=0) and non-large cap

(𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸𝑖,𝜏=1) and large cap firms (𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸𝑖,𝜏=0). The 𝐶𝐴𝑅𝑖,𝜏 analysis consists of 367 firms with a total of 734 firm year observations. 468 of these observations relate to non-large cap firms, while 266 firm year observations are available for large cap firms. Table 3B illustrates the overall differences between pre-post years. The 𝐶𝐴𝑅_𝑖,𝜏 does not show a statistically significant difference between the pre-adoption year and post-adoption year. The control variables 𝑅𝑂𝐴_𝑖,𝜏 and 𝐶𝐻𝑁𝐼_𝑖,𝜏 show an increase with a significance level of 10% and 5%, respectively. There is a significant decrease in 𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏 at the 1% level. Table 3C illustrates the difference between the pre-adoption and post-adoption period for Treatment firms. 𝐶𝐴𝑅_𝑖,𝜏 is not significantly different in the post-period. The only significant differences relate to 𝐿𝑂𝑆𝑆_𝑖,𝜏 (10% level) and 𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏 (1% level). While there are no significant differences for Treatment firms, Table 3D exhibits a statistically significant difference in means for 𝐶𝐴𝑅_𝑖,𝜏 in 2017. Moreover, 𝑅𝑂𝐴_𝑖,𝜏 and 𝐶𝐻𝑁𝐼_𝑖,𝜏 have changed significantly. Table 3E shows no difference for non-large cap firms

(27)

27

between the pre-adoption period and post-adoption period. The only statistically significant difference concerns the decrease of the variable 𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏 at the 1% level.

Table 3A Descriptive Statistics 𝐶𝐴𝑅_𝑖,𝜏 analysis

NONLARGEi,τ = 1 NONLARGEi,τ = 0

Variables Mean Median S. D. Mean Median S. D. Mean Median S. D. CARi,τ -0.0013 -0.0004 0.0339 -0.0007 -0.0012 0.0364 -0.0022 0.0001 0.0289 POSTi,τ 0.5 0.5 0.5003 0.5 0.5 0.5005 0.5 0.5 0.5009 TREATi,τ 0.5450 1 0.4983 0.5427 1 0.4987 0.5489 1 0.4985 NONLARGEi,τ 0.6376 1 0.4810 1 1 0 0 0 0 ROAi,τ 0.0542 0.0457 0.0585 0.0565 0.0462 0.0608 0.0501 0.0450 0.0542 LOSSi,τ 0.0749 0 0.2635 0.0726 0 0.2598 0.0789 0 0.2702 BIG4i,τ 0.9891 1 0.1039 0.9829 1 0.1298 1 1 0.0000 MTBi,τ 2.5885 2.0881 2.9933 2.4008 2.0712 3.1849 2.9188 2.1793 2.5953 LEVi,τ 0.2098 0.1903 0.1505 0.2124 0.1902 0.1629 0.2051 0.1935 0.1258 CHNIi,τ 0.0070 0.0046 0.0521 0.0064 0.0055 0.0560 0.0079 0.0035 0.0444 ABRET_EARi,τ -0.0013 -0.0004 0.0318 -0.0008 -0.0012 0.0337 -0.0021 0.0001 0.0283 LOGMKTi,τ 23.1050 22.8778 1.1325 22.5332 22.3885 0.8098 24.1109 23.9658 0.8938 N. Obs. 734 468 266

Table 3B Descriptive Statistics CAR_i,τ Analysis POST

POSTi,τ = 0 POSTi,τ = 1 Diff. in Means

Variables Mean Median S. D. Mean Median S. D.

CARi,τ -0.0026 -0.0012 0.0392 0.0001 -0.0002 0.0275 0.0026 POSTi,τ 0 0 0 1 1 0 1 TREATi,τ 0.5450 1 0.4987 0.5450 1 0.4987 0 NONLARGEi,τ 0.6376 1 0.4813 0.6376 1 0.4813 0 ROAi,τ 0.0501 0.0428 0.0580 0.0583 0.0503 0.0588 0.0082 * LOSSi,τ 0.0845 0 0.2785 0.0654 0 0.2476 -0.01907 BIG4i,τ 0.9891 1 0.1040 0.9891 1 0.1040 0 MTBi,τ 2.6453 2.1874 3.0723 2.5318 2.0217 2.9152 -0.1136 LEVi,τ 0.2108 0.1977 0.1491 0.2087 0.1860 0.1520 -0.0021 CHNIi,τ 0.0027 0.0029 0.0515 0.0113 0.0058 0.0524 0.0086 ** ABRET_EARi,τ -0.0023 -0.0012 0.0366 -0.0002 -0.0002 0.0261 0.0020 LOGMKTi,τ 23.2445 23.0487 1.2426 22.9655 22.7848 0.9928 -0.2790 *** N. Obs. 367 367 734

(28)

28

Table 3C Descriptive Statistics CAR_i,τ Analysis Treatment Firms

POSTi,τ = 0 TREATi,τ = 1 POSTi,τ = 1 TREATi,τ = 1

Variables Mean Median S.D. Mean Median S.D. Diff. in Means CARi,τ 0.0010 0.0012 0.0409 0.0005 0.0004 0.0274 -0.0005 POSTi,τ 0 0 0 1 1 0 1 TREATi,τ 1 1 0 1 1 0 0 NONLARGEi,τ 0.6350 1 0.4826 0.6350 1 0.4826 0 ROAi,τ 0.0506 0.0426 0.0588 0.0565 0.0451 0.0563 0.0059 LOSSi,τ 0.0850 0 0.2796 0.0400 0 0.1965 -0.0450 * BIG4i,τ 0.9950 1 0.0707 0.9950 1 0.0707 0 MTBi,τ 2.5585 2.1333 2.7631 2.4227 1.9967 2.7782 -0.1359 LEVi,τ 0.1970 0.1749 0.1424 0.1961 0.1723 0.1482 -0.0009 CHNIi,τ 0.0070 0.0027 0.0440 0.0083 0.0044 0.0434 0.0013 ABRET_EARi,τ 0.0011 0.0012 0.0378 0.0002 0.0004 0.0257 -0.0009 LOGMKTi,τ 23.6897 23.5710 1.2542 23.0153 22.8824 0.9671 -0.6744 *** N. Obs. 200 200 400

***, **, * indicate statistical significance at the 1%, 5% and 10% levels.

Table 3D Descriptive Statistics CAR_i,τ Analysis Control Firms

POSTi,τ = 0 TREATi,τ = 0 POSTi,τ = 1 TREATi,τ = 0

Variables Mean Median S.D. Mean Median S.D. Diff. in Means CARi,τ -0.0069 -0.0034 0.0367 -0.0005 -0.0005 0.0277 0.0064 * POSTi,τ 0 0 0 1 1 0 1 TREATi,τ 0 0 0 0 0 0 0 NONLARGEi,τ 0.6407 1 0.4812 0.6407 1 0.4812 0 ROAi,τ 0.0495 0.0440 0.0572 0.0605 0.0533 0.0617 0.0110 * LOSSi,τ 0.0838 0 0.2780 0.0958 0 0.2952 0.0120 BIG4i,τ 0.9820 1 0.1332 0.9820 1 0.1332 0 MTBi,τ 2.7493 2.1987 3.4115 2.6624 2.0573 3.0747 -0.0869 LEVi,τ 0.2274 0.2146 0.1556 0.2238 0.2084 0.1556 -0.0036 CHNIi,τ -0.0025 0.0049 0.0589 0.0148 0.0098 0.0614 0.0173 *** ABRET_EARi,τ -0.0063 -0.0034 0.0348 -0.0008 -0.0005 0.0266 0.0056 LOGMKTi,τ 22.7112 22.4179 0.9969 22.9058 22.6757 1.0224 0.1946 * N. Obs 167 167 334

(29)

29

Table 3E Descriptive Statistics CAR_i,τ Analysis Non-large cap firms

NONLARGEi,τ = 1 POSTi,τ = 0 NONLARGEi,τ = 1 POSTi,τ = 1

Variables Mean Median S. D. Mean Median S. D. Diff. in Means CARi,τ -0.0027 -0.0027 0.0423 0.0012 0.0007 0.0293 0.0039 POSTi,τ 0 0 0 1 1 0 1 TREATi,τ 0.5427 1 0.4992 0.5427 1 0.4992 0 NONLARGEi,τ 1 1 0 1 1 0 0 ROAi,τ 0.0537 0.0451 0.0608 0.0593 0.0492 0.0607 0.0056 LOSSi,τ 0.0812 0 0.2737 0.0641 0 0.2455 -0.0171 BIG4i,τ 0.9829 1 0.1299 0.9829 1 0.1299 0 MTBi,τ 2.4496 2.1684 3.2657 2.3520 2.0266 3.1082 -0.0976 LEVi,τ 0.2136 0.1944 0.1612 0.2112 0.1774 0.1649 -0.0025 CHNIi,τ 0.0037 0.0059 0.0551 0.0092 0.0054 0.0569 0.0055 ABRET_EARi,τ -0.0023 -0.0027 0.0390 0.0007 0.0007 0.0273 0.0031 LOGMKTi,τ 22.6956 22.3837 0.9980 22.3707 22.3919 0.5152 -0.3249 *** N. Obs. 234 234 468

Table 3F Descriptive Statistics CAR_i,τ Analysis Non-large cap Treatment firms

TREATi,τ = 1 POSTi,τ = 0 TREATi,τ = 1 POSTi,τ = 1

Variables Mean Median S. D. Mean Median S.D. Diff. in Means CARi,τ 0.0025 0.0007 0.0461 0.0011 -0.0002 0.0277 -0.0013 POSTi,τ 0 0 0 1 1 0 1 TREATi,τ 1 1 0 1 1 0 0 NONLARGEi,τ 1 1 0 1 1 0 0 ROAi,τ 0.0532 0.0446 0.0593 0.0592 0.0489 0.0558 0.0059 LOSSi,τ 0.079 0 0.2704 0.0157 0 0.1250 -0.0630 ** BIG4i,τ 0.992 1 0.0887 0.9921 1 0.0887 0 MTBi,τ 2.3741 2.2798 2.7410 2.2856 2.0385 2.9649 -0.0884 LEVi,τ 0.1966 0.1689 0.1546 0.1958 0.1667 0.1606 -0.0008 CHNIi,τ 0.0113 0.0055 0.0466 0.0088 0.0043 0.0465 -0.0025 ABRET_EARi,τ 0.0027 0.0007 0.0417 0.0006 -0.0002 0.0251 -0.0021 LOGMKTi,τ 23.1752 22.8870 1.0545 22.4518 22.4741 0.5431 -0.7234 *** N. Obs. 127 127 254

(30)

30

Table 3G Descriptive Statistics CAR_i,τ Analysis Large cap firms

NONLARGEi,τ = 0 POSTi,τ = 0 NONLARGEi,τ = 0 POSTi,τ=1

Variables Mean Median S. D. Mean Median S. D. Diff. in Means CARi,τ -0.0024 0.0016 0.0331 -0.0020 -0.0008 0.0240 0.0004 POSTi,τ 0 0 0 1 1 0 1 TREATi,τ 0.5489 1 0.4995 0.5489 1 0.4995 0 NONLARGEi,τ 0 0 0 0 0 0 0 ROAi,τ 0.0438 0.0395 0.0522 0.0565 0.0524 0.0556 0.0128 * LOSSi,τ 0.0902 0 0.2876 0.0677 0 0.2521 -0.0226 BIG4i,τ 1 1 0 1 1 0 0 MTBi,τ 2.9897 2.2854 2.6756 2.8480 1.9986 2.5205 -0.1417 LEVi,τ 0.2059 0.1998 0.1254 0.2043 0.1869 0.1266 -0.0015 CHNIi,τ 0.0009 0.0016 0.0445 0.0149 0.0069 0.0433 0.0140 *** ABRET_EARi,τ -0.0022 0.0016 0.0322 -0.0020 -0.0008 0.0240 0.0002 LOGMKTi,τ 24.2100 24.0497 1.0248 24.0118 23.8929 0.7307 -0.1982 * N. Obs. 133 133 266

Table 3H Descriptive Statistics 𝐶𝐴𝑅_𝑖,𝜏 Analysis Large cap Treatment firms

𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 = 1 𝑃𝑂𝑆𝑇𝑖,𝜏 = 0 𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 = 1 𝑃𝑂𝑆𝑇𝑖,𝜏 = 1

Variables Mean Median S. D. Mean Median S. D. Diff. in Means 𝐶𝐴𝑅𝑖,𝜏 -0.002 0.002 0.030 0.000 0.001 0.027 0.001 𝑃𝑂𝑆𝑇𝑖,𝜏 0 0 0 1 1 0 1 𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 1 1 0 1 1 0 0 𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸𝑖,𝜏 0 0 0 0 0 0 0 𝑅𝑂𝐴𝑖,𝜏 0.046 0.030 0.058 0.052 0.043 0.057 0.006 𝐿𝑂𝑆𝑆𝑖,𝜏 0.096 0 0.2965 0.0822 0 0.2766 -0.0137 𝐵𝐼𝐺4𝑖,𝜏 1 1 0 1 1 0 0 𝑀𝑇𝐵𝑖,𝜏 2.8795 2.0214 2.7910 2.6611 1.9307 2.4203 -0.2184 𝐿𝐸𝑉𝑖,𝜏 0.1978 0.1837 0.1195 0.1966 0.1750 0.1245 -0.0012 𝐶𝐻𝑁𝐼𝑖,𝜏 -0.0005 0.0014 0.0384 0.0073 0.0045 0.0378 0.0078 𝐴𝐵𝑅𝐸𝑇_𝐸𝐴𝑅𝑖,𝜏 -0.0015 0.0017 0.0301 -0.0005 0.0014 0.0271 0.0010 𝐿𝑂𝐺𝑀𝐾𝑇𝑖,𝜏 24.5850 24.4491 1.0594 23.9957 23.8161 0.7308 -0.5893 *** N. Obs. 73 73 146

(31)

31

Table 4 Difference-in differences and pre-post analysis

Dep. Var. = 𝐶𝐴𝑅_𝑖,𝜏 Variables (1) (2) (3) (4) (5) 𝑃𝑂𝑆𝑇_𝑖,𝜏 0.000419 0.000419 0.000516 0.000713 0.000713 (0.000353) (0.000360) (0.000377) (0.000947) (0.000781) 𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 -0.000254 -0.000254 (0.000744) (0.000856) 𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸𝑖,𝜏 0.000266 0.000266 0.000169 0.000426 0.000426 (0.000297) (0.000437) (0.000398) (0.000328) (0.000434) 𝑃𝑂𝑆𝑇 ∗ 𝑇𝑅𝐸𝐴𝑇_𝑖,𝜏 -0.000169 -0.000169 -0.000303 (0.000632) (0.000737) (0.000356) 𝑃𝑂𝑆𝑇 ∗ 𝑇𝑅𝐸𝐴𝑇 0.000573 0.000573 0.000579 ∗ 𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸_𝑖,𝜏 (0.000834) (0.000669) (0.000828)

𝐿𝑂𝐺𝑀𝐾𝑇𝑖,𝜏 6.72e-05 6.72e-05 9.48e-05 0.000112 0.000112

(0.000168) (0.000354) (0.000107) (0.000259) (0.000504)

𝑅𝑂𝐴_𝑖,𝜏 0.00312 0.00312 -0.000294 0.00326 0.00326

(0.00286) (0.00273) (0.00236) (0.00428) (0.00471)

𝐿𝑂𝑆𝑆_𝑖,𝜏 0.00160* 0.00160* 0.00140 0.00296 0.00296

(0.000954) (0.000960) (0.000919) (0.00191) (0.00189)

𝑀𝑇𝐵_𝑖,𝜏 -4.08e-06 -4.08e-06 -1.18e-05 3.19e-05 3.19e-05

(3.48e-05) (3.30e-05) (3.44e-05) (5.37e-05) (5.98e-05)

𝐿𝐸𝑉𝑖,𝜏 -0.00104 -0.00104 -0.000677 -0.00138 -0.00138 (0.000948) (0.000936) (0.000969) (0.00184) (0.00175) 𝐶𝐻𝑁𝐼𝑖,𝜏 -0.000480 -0.000480 0.000362 0.00206 0.00206 (0.00268) (0.00268) (0.00288) (0.00311) (0.00400) 𝐴𝐵𝑅𝐸𝑇_𝐸𝐴𝑅_𝑖,𝜏 1.054*** 1.054*** 1.053*** 1.062*** 1.062*** (0.0231) (0.0175) (0.0225) (0.0386) (0.0287) 𝐵𝐼𝐺4_𝑖,𝜏 0.00250 0.00250 0.00266 -0.00138* -0.00138 (0.00322) (0.00382) (0.00321) (0.000809) (0.00186) Constant -0.00459 -0.00459 -0.00506 -0.00223 -0.00223 (0.00453) (0.00883) (0.00450) (0.00564) (0.0115)

Industry FE Included Included Included Included

Bootstrap S.E. Yes Yes

N. Obs. 734 734 734 400 400

R-squared 0.981 0.981 0.981 0.973 0.973

Column (1) shows the results including industry fixed effects, Column (2) the results including industry fixed effects and calculating standard errors using the bootstrap method with 1,000 replications, Column (3) the results including company fixed effects and excluding the 𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 intercept and industry fixed effects. Column (4) shows the results for the pre-post model for Treatment firms only, including industry fixed effects and Column (5) shows the results for the pre-post model for Treatment firms only, including industry fixed effects and calculating

standard errors using the bootstrap method with 1,000 replications. ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively.

(32)

32

As illustrated in Table 3F, the only difference between the means for non-large cap treatment firms in post-adoption period compared to pre-adoption relates to a significant decrease in the variable 𝐿𝑂𝑆𝑆_𝑖,𝜏 at the 5% level. Table 3G provides descriptive statistics for Large-cap firms. Again, there is no difference in 𝐶𝐴𝑅_𝑖,𝜏 between 2016 and 2017. There is an increase in 𝑅𝑂𝐴_𝑖,𝜏 (10% level) and 𝐶𝐻𝑁𝐼_𝑖,𝜏 (1% level) and a decrease in 𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏 (10% level) for large cap firms. Table 3H illustrates descriptive statistics for large cap treatment firms. There is no significant difference in 𝐶𝐴𝑅_𝑖,𝜏, while the only significant change relates to 𝐿𝑂𝐺𝑀𝐾𝑇_𝑖,𝜏, at the 1% level.

Table 4 illustrates the results using cumulative abnormal returns 𝐶𝐴𝑅_𝑖,𝜏 as the dependent variable. Column (1) through (3) show the coefficients and t-statistics for the difference-in-differences (DD) model. Column (1) exhibits the results including industry fixed effects, Column (2) the results including industry fixed effects and calculating standard errors using the bootstrap method, Column (3) the results including company fixed effects but excluding the 𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 intercept and industry fixed effects. The regression exhibits high 𝑅2 values in all models. MacKinley (1997) argues that this can lead to increased ability to detect event effects. The DD effect for 𝑃𝑂𝑆𝑇 ∗ 𝑇𝑅𝐸𝐴𝑇𝑖,𝜏 in column (1) = -0.000169 (p-value > 0.10) and DD effect for 𝑃𝑂𝑆𝑇 ∗ 𝑇𝑅𝐸𝐴𝑇 ∗ 𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸𝑖,𝜏 in column (1) = 0.000573 (p-value > 0.10) are not significant and are robust to computing standard errors using the bootstrap method in Column (2) and company fixed effects in Column (3). Column (4) and Column (5) exclude control firms and illustrate the pre-post model and capitalization effects of treatment firms. Column (4) shows the results including industry fixed effects and Column (5) shows the results including industry fixed effects and calculating standard errors using the bootstrap method with 1,000 replications. These Columns exhibit no statistically significant differences for 𝑃𝑂𝑆𝑇_𝑖,𝜏= 0.000713 (p-value > 0.10) and 𝑁𝑂𝑁𝐿𝐴𝑅𝐺𝐸_𝑖,𝜏= 0.000426 (p-value > 0.10) as indicated in the descriptive statistics. These results indicate that the extended auditor’s report does neither have an impact on the 𝐶𝐴𝑅_𝑖,𝜏 variable for Treatment firms nor there is a difference between large cap and non-large cap firms.

5.2 ABVOL Analysis

The 𝐴𝐵𝑉𝑂𝐿_𝑖,𝜏 consists of 338 firms with a total of 676 firm year observations. 428 of these observations relate to non-large cap firms, while 248 firm year observations are available

Effects of the new auditor’s report on the decision usefulness of the annual report