Stock market reaction to the engagement partner`s disclosure announcements in the U.S.

(1)

MSc Accountancy & Control, track Accountancy Faculty of Economics and Business, University of Amsterdam

Master thesis:

Stock Market Reaction to the

Engagement Partner`s Disclosure

Announcements in the U.S.

Author: Jeral Davarci Student number: 10680101

Date: 13th_of_{August, 2015} Version: Final version

Study: Accountancy & Control Faculty: Economics and Business University: University of Amsterdam Supervisor: Dr. R. Felleg

(2)

Statement of Originality

This document is written by student Jeral Davarci 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 engagement partner`s disclosure is at the centre of discussion in the United States. According to the Public Company Accounting Oversight Board (PCAOB), the engagement partner`s disclosure increases audit quality. However, results from prior literature are very inconsistent on whether audit quality increases. In this study, the market reaction to the engagement partner`s disclosure is examined by measuring the cumulative abnormal return of firms with equity traded in the U.S. stock market on engagement partner`s disclosure announcements. The results show that investors react positively to the engagement partner announcements. I conclude that this positive reaction to the engagement partner announcements suggests that investors expect audit quality to increase. This result implies that while it is not certain whether audit quality increases, investors do expect an increase, which may have unintended consequences, such as an increase of the audit gap.

Keywords: PCAOB; engagement partner`s disclosure; audit quality; market reaction.

Acknowledgement

I would like to thank my thesis supervisor Dr. Réka Felleg for providing me feedback to guide me through the process of creating this MSc thesis and for her support and willingness to help me at all times.

(4)

Generally, audits in the U.S. only disclose the name of the firm that issued the auditor's report. As a result, information about other firms and persons that participated in the audit is unknown to the users of the auditor's report (“PCAOB Reproposes Amendments”, 2013). On the 11th of October 2011, the Public Company Accounting Oversight Board (PCAOB), the auditing standards of the U.S., voted for public comment proposed amendments to the PCAOB auditing standards that would provide greater transparency into audits of public companies, brokers, and dealers about the engagement partner and certain other participants in the audit. The proposal requires accounting firms to disclose the name of the engagement partner in the audit report as well as the disclosure of other accounting firms and other persons that are not employed by the auditor, but took part in the audit (“PCAOB Proposes Amendments”, 2011). According to James R. Doty, the Chairman of the PCAOB, a requirement for the engagement partner to sign the audit report could also improve audit quality, because it might increase the engagement partner's sense of accountability. In addition, it would increase transparency, allowing the investing public to identify and judge quality, thereby improving the usefulness of the auditor`s report, increasing the investor`s decision making (“PCAOB Reproposes Amendments”, 2013).

However, many representatives of the accounting profession, including Deloitte, KPMG, EY, PwC, BDO, Grant Thornton and some smaller firms, either objected to naming the engagement partner or said naming the engagement partner would not improve audit quality. In the comments submitted in response to the proposal, Big Four firm Deloitte told the audit overseer that the signature requirement would fail to achieve either objective. According to Deloitte (2009, p. 1) “The suggested beneficial effects on accountability and transparency are speculative, and the concept release does not present evidence that an engagement partner signature requirement will enhance audit quality. However, the associated costs and burdens, including unintended consequences, are real and could be significant.” Although the accounting firms are not supporting the engagement partner`s disclosure, because it would not increase audit quality, the existing literature is inconsistent. Carcello and Li (2013) investigated the costs and benefits of requiring an engagement partner signature and the effect on audit quality and audit fees in the United Kingdom. Carcello and Li found that abnormal accruals and the likelihood of meeting an earnings threshold

(6)

decline after the effective date of the partner signature requirement, and that the incidence of qualified audit opinions and earnings informativeness increase. The authors also found a significant increase in audit fees in the United Kingdom. Blay, Notbohm, Schelleman and Valencia (2014) also studied the effects of engagement partner`s signature on audit quality. The authors conducted their study in the Netherlands. Their results support that the engagement partner`s disclosure has no effect on the audit quality. In addition, the authors state that the reason that they did not find an effect on audit quality is because the audit quality is already at a high level, as argued by a study of Francis (Francis, 2005) and in a comment of PricewaterhouseCoopers (PricewaterhouseCoopers, 2009).

Most studies regarding the engagement partner`s disclosure concern whether audit quality increases or not. The PCAOB assumes that the audit quality will increase, while, as mentioned before, there is much inconsistency in the literature regarding the effects of the engagement partner`s disclosure on audit quality. In addition, according to Carcello and Li (2013), arguments in favour of a partner signature requirement implicitly assume that current audit performance is suboptimal, while the highly regulated nature of the audit market may mean that the optimum level of audit quality is currently being delivered. According to Knutson (1994), an audit expectation gap arises from a combination of excessive expectations and insufficient performance. According to Ruhnke and Schmidt (2014), the increase in audit regulation could result in the increase of the audit gap, if the audit regulation does not increase the audit`s usefulness. As a result, the public trust in the audit profession could be disturbed. In addition, emphasising the difficulty of measuring the actual audit quality itself (King et

al., 2012; Bedard et al., 2010; Oliverio, 2007; Watkins et al., 2004) it is important to

take the investors reaction into consideration. At last, the possible drawbacks of the engagement partner`s disclosure become clear in the comment letters of Bierstaker et

al. (2009) and Jones et al. (2012a) of the Auditing Standards Committee of the Auditing

Section of the American Accounting Association (ASC-AAA). Bierstaker et al. and Cole (2014) state that as a consequence of increased partner liability, auditors might engage in defensive auditing, resulting in higher audit costs. Jones et al. also indicate that partners may be subject to more litigation, privacy, and security issues as a result of the disclosures. Emphasizing the importance of not creating unwanted or non-anticipated effects incurring after mandating the engagement partner`s disclosure, such as a gap between the desired/expected audit quality and taking all the drawbacks into

(7)

consideration, it is of importance to investigate whether investors, the users of financial statements, really perceive a higher audit quality and thereby experience more decision usefulness, which is the main goal of the engagement partner`s disclosure.

According to the efficient market hypothesis (Fama et al., 1969), all relevant information is reflected in the stock price. In addition, new and relevant information is reflected in the stock prices immediately. Under the efficient market hypothesis assumption, it is possible to study the investor`s reaction to the engagement partner`s disclosure announcement. By investigating the stock price reaction on the engagement partner`s disclosure announcement it is possible to find out whether the investors actually believe the engagement partner`s disclosure provides new and relevant information regarding the audit quality. This has led to the following research question:

What is the stock market reaction to the Engagement Partner`s Disclosure announcements in the U.S.?

Since there is no decisive answer whether audit quality increases, it is important to understand the insights of the investors and to predict how the investors will react to the movement towards the engagement partner`s disclosure. In order to gain insight on whether investors believe audit quality increases, this study examines the stock market reaction to events associated with the announcements of the amendment to disclose the engagement partner`s signature. The impact on the share prices could give a more conclusive answer on whether the main goal of the engagement partner`s disclosure, the increase of usefulness by giving more information, is achieved, as shown by the study of Sreejith and Nageswara Rao (2014). With the use of an event study, I examine the three-day and seven-day cumulative abnormal return for firms with equity traded in the U.S. stock market, centred on events regarding the announcements of the engagement partner`s disclosure. I find a positive cumulative abnormal return, suggesting that the investors expect audit quality to increase.

This research contributes, first of all, by providing evidence of the reaction of the engagement partner`s disclosure from the market. According to the PCAOB, the engagement partner`s disclosure increases transparency and accountability and thereby increasing audit quality. The increase in transparency would accordingly improve the investor`s decision making, because of the increase in usefulness of the auditor`s report. However, while the evidence on the effect on audit quality is inconsistent, there is little

(8)

research and evidence whether investors share this opinion. Therefore, this research could add to the discussion whether engagement partner`s disclosure increases audit quality and provides more clarity, by investigating the investors perspective. In addition, this study aims to narrow the research gap regarding the stock market’s reaction to the Engagement Partner`s Disclosure announcements in the U.S., by investigating whether investors will experience the possible increase in audit quality, as little research on the market reaction to the engagement partner`s disclosure has been done yet. This study thereby contributes to existing literature from Carcello and Li (2013), Yen, Chang and Chen (2013), King, Davis and Mintchik (2012) and Schelleman and Valencia (2014), by investigating whether investors react positively to the engagement partner`s disclosure, suggesting that investors expect an increase in audit quality as more information becomes known. In addition, this study aims to contribute to the regulatory bodies, such as the PCAOB, by providing additional information on whether the engagement partner`s disclosure increases the usefulness of the financial reports for investors. Last, this paper contributes towards society and audit firms. By investigating whether the perception of the investors is that audit quality increases, this research provides evidence which could be useful in terms of the potential gap between the expectations of investors and intentions of regulatory bodies regarding the mandated engagement partner`s disclosure and whether the engagement partner`s disclosure outweighs the possible drawbacks for auditors. As mentioned earlier, an audit gap could arise when the expectations regarding the engagement partner`s disclosure are excessive. This has consequences for the users of an annual report as well as the auditors and the stock markets. It is important that the engagement partner`s disclosure is not just making it appear that audit quality increases. If investors react positive to the engagement partner`s disclosure, it could mean that the investors expect relevant new information regarding the audit quality, while the PCAOB assumes the audit quality would increase. Therefore, with the insights of the investors on the audit quality, this research could add to the discussion regarding the implementation of the engagement partner`s disclosure.

This paper proceeds as follows. Chapter two discusses the literature that is used to subsequently support and develop the hypothesis. The third chapter starts with the sample selection procedures, followed by the timeline of the engagement partner`s disclosure announcements and the empirical model. In chapter four, the descriptive

(9)

statistics and the results of the event study are discussed, followed by the conclusion in chapter five.

2. Literature Review and Hypothesis Development

2.1 Background

On the 11th of October 2011, the Public Company Accounting Oversight Board (PCAOB), the auditing standards of the U.S., voted for public comment proposed amendments to the PCAOB auditing standards that would provide greater transparency into audits of public companies, brokers, and dealers about the engagement partner and certain other participants in the audit (“PCAOB Proposes Amendments”, 2011). The proposed amendment builds upon a related Concept Release issued by the PCAOB on the 28th of July 2009. Instead of disclosing the name of the firm that issued the auditor's report, the new proposal requires accounting firms to disclose the name of the engagement partner in the audit report as well as the disclosure of other accounting firms and other persons that are not employed by the auditor, but took part in the audit (PCAOB, 2009). The International Auditing and Assurance Standards Board (IAASB) has developed a similar standard with the revised International Standard on Auditing (ISA) 700. In the revised ISA, the IAASB includes the disclosure of the name of the engagement partner to be included in the auditor’s report for audits of complete sets of general purpose financial statements of listed entities. However, the ISA 700 (Revised) has an effective date for audits of financial statements for periods ending on or after the 15th _{of December, 2016 (International Standards on Auditing, 2015).}

According to James R. Doty, PCAOB Chairman, the amendment will increase the usefulness of the auditor's report for investors when making their investment decisions, as well as when voting on the ratification of a company's choice of accounting firm as its auditor. It would enable investors to determine whether the firm is registered with the Board and has been subject to PCAOB inspection, or is located in a country that does not allow PCAOB inspections. Investors would also be able to verify whether a disclosed firm or person has had any publicly available disciplinary history with the PCAOB or other regulators. A requirement for the engagement partner to sign the audit report could also improve audit quality. This improvement is twofold. First, it might increase the engagement partner's sense of accountability to financial

(10)

statement users, which could lead him or her to exercise greater care in performing the audit. The act of signing his or her own name may increase the engagement partner’s sense of responsibility for the quality of the audit. Second, it would increase transparency about who is responsible for performing the audit, which could provide useful information to investors and, in turn, provide an additional incentive to firms to improve the quality of all their engagement partners. In conclusion, the amendment will be of aid for investors by making previous unknown information known or previous difficult accessible information more easy to access. In addition, according to the James R. Doty, the engagement partner`s disclosure may also help the investing public identify and judge quality, leading to better auditing (“PCAOB Reproposes Amendments”, 2013).

Many representatives of the accounting profession, including Deloitte, KPMG, EY, PwC, BDO, Grant Thornton and some smaller firms, either objected to naming the engagement partner or said naming the engagement partner would not improve audit quality. However, in comments submitted in response to the engagement partner`s disclosure proposal, Big Four firm Deloitte told the audit overseer that the signature requirement would fail to achieve either objective. According to Deloitte (2009, p. 1) “The suggested beneficial effects on accountability and transparency are speculative, and the concept release does not present evidence that an engagement partner signature requirement will enhance audit quality. However, the associated costs and burdens, including unintended consequences, are real and could be significant.” Other leading firms took a similar stand. According to Grant Thornton (2009, p. 1) “Partners are held accountable by their own professionalism, supplemented by mechanisms that are in place that allow third parties to hold them accountable.” Because, according to Grant Thornton (2009, p. 2), “A significant portion of the partner`s net worth is invested in the firms in which they are partners. A single audit failure can take down a firm, resulting in a significant personal loss to the engagement partner, and all of his or her partners. We do not believe individually signing a report will promote greater accountability.” Big Four firm PwC also commented on the engagement partner`s disclosure (2009, p. 11). In their comment letter, PwC concludes that the name of the engagement partner will not increase the audit quality and it will not increase the partner`s accountability or transparency. PwC believes that the accountability will not be affected by the new regulation because engagement partners are already strongly motivated to assure high audit quality, for example, to maintain their current reputation

(11)

within the firm. According to PwC (2009, p. 2) “The current practice of signing the audit report in the name of the firm appropriately reflects the reality that the quality of an audit depends not on just the qualifications and competence of the engagement partner, but on the qualifications and competence of many people at the firm with involvement in the audit, as well as on the quality controls established and monitored by the firm.” In addition, EY (EY, 2014), BDO (BDO, 2014) and KPMG (2014) also do not support the engagement partner`s disclosure, because it will result in certain operational challenges, such as legal issues. According to EY, these challenges will consume time and are likely to be costly and complex and will not offset the benefits of the engagement partner`s disclosure.

Even though the accounting firms are not supporting the engagement partner`s disclosure, the existing literature is inconsistent. Carcello and Li investigated the costs and benefits of requiring an engagement partner signature and the effect on audit quality and audit fees in the United Kingdom. Carcello and Li found that abnormal accruals and the likelihood of meeting an earnings threshold decline after the effective date of the partner signature requirement, and that the incidence of qualified audit opinions and earnings informativeness increase. The authors also found a significant increase in audit fees in the United Kingdom. The evidence suggests that the partner signature requirement in the U.K. has benefited investors and other financial statement users, but that these benefits have come at the cost of significantly higher audit fees. Blay, Notbohm, Schelleman and Valencia (2014) also studied the effects of engagement partner`s signature on audit quality. The authors conducted their study in the Netherlands. Their results support that the engagement partner`s disclosure has no effect on the audit quality. In addition, the authors state that the reason that they did not find an effect on audit quality is because the audit quality is already at a high level, as argued by a study of Francis (Francis, 2005) and in a comment of PricewaterhouseCoopers (PricewaterhouseCoopers, 2009). King, Davis and Mintchik (2012) studied the effects of the disclosure of the engagement partner`s identity. The authors distinguished audit quality in appearance and in fact. According to King et al., there is a great inconsistency in the empirical evidence regarding the increase of audit quality, as assumed by the PCAOB. The authors found that audit quality in appearance will most likely increase, but the audit quality in fact is hard to measure and remains unclear.

(12)

2.2 Benefits and Drawbacks of the Engagement Partner`s Disclosure

Carcello and Li (2013) investigated the costs and benefits of requiring an engagement partner signature and the effect on audit quality and audit fees in the United Kingdom. Carcello and Li consider changes in three earnings quality measures, which are also commonly used to proxy for audit quality: abnormal accruals, the propensity to meet an earnings threshold, and earnings informativeness. Carcello and Li found that abnormal accruals and the likelihood of meeting an earnings threshold decline after the effective date of the partner signature requirement, and the incidence of qualified audit opinions and earnings informativeness increase. The authors also found a significant increase in audit fees in the United Kingdom. The evidence suggests that the partner signature requirement in the U.K. has benefited investors and other financial statement users, but that these benefits have come at the cost of significantly higher audit fees.

Blay, Notbohm, Schelleman and Valencia (2014) also studied the effects of engagement partner`s signature on audit quality. The authors conducted their study in the Netherlands before and after the engagement partner`s disclosure requirement and compared the audit quality in the Netherlands with the audit quality in the United Kingdom. The authors used abnormal accruals, the magnitude of accruals relative to cash flows, and benchmark earnings targets. Their results support that the engagement partner`s disclosure has no effect on the audit quality. In addition, the authors state that the reason that they did not find an effect on audit quality is because the audit quality is already at a high level, as argued by a study of Francis (Francis, 2005) and in a comment of PricewaterhouseCoopers (PricewaterhouseCoopers, 2009).

In addition, the possible drawbacks of the engagement partner`s disclosure become clear in the comment letters of Bierstaker et al. (2009) and Jones et al. (2012a) of the Auditing Standards Committee of the Auditing Section of the American Accounting Association (ASC-AAA). In the comment letters on the proposed rule on improving the transparency of disclosure of the engagement partner, the authors suggest that it is possible that the engagement partners could be subject to civil lawsuit with negative consequences. Bierstaker et al. state that as a consequence of increased partner liability, auditors might engage in defensive auditing, resulting in higher audit costs. Jones et al. also indicate that partners may be subject to more litigation, privacy, and security issues as a result of the disclosures. In addition, according to Cole (2014) in which all the concerns of the partner signature or disclosure requirement are listed,

(13)

EY commented the concern naming individual partners in lawsuits could force audit firms to settle more lawsuits rather than to prolong a lawsuit with its negative consequences to the individual partner. In the study of Cole, another listed statement of E&Y is that merely being sued for fraud or negligence could lead to the loss of clients for the individual partner, emotional and personal financial difficulties.

King, Davis and Mintchik (2012) examined the effects of the disclosure of the engagement partner`s identity in the form of a theoretical analysis. According to King

et al., there is a great inconsistency in the empirical evidence regarding the increase of

audit quality, as assumed by the PCAOB. The authors distinguished audit quality in appearance and in fact. To examine the potential effects of the engagement partner`s disclosure, the authors use three different kinds of academic frameworks: source credibility (effect of the transparency of the audit report), accountability (responsibility for actions) and the theory of affordances (effect of partner`s signature on self-identity perception). According to King et al., the source credibility addresses the potential impact of the engagement partner`s disclosure on audit quality in appearance. According to the authors, the source credibility theory suggests that the investors` perception on audit quality in appearance will increase. To comment on the effect of audit quality in fact, the authors used the accountability theory and the theory of affordances. The authors found that it is not clear whether the accountability increases as the research on accountability is limited and lacking on the ability to client pressure or to solve ambiguous accounting or auditing issues or whether accountability has a certain limit and what the current level of accountability is, perceived by the engagement partner`s. The authors therefore conclude that it is not certain whether accountability increases. At last, the authors provide insights on the impact of the engagement partner`s disclosure on the engagement partner`s behaviour using the theory of affordances. The authors conclude that the literature is scarce and the results are inconclusive. Overall, the authors conclude that audit quality in appearance will most likely increase, but the audit quality in fact is hard to measure and remains unclear. In addition to the inconsistency in prior literature, the mandatory disclosure of the engagement partner`s disclosure could have unwanted side effects, for instance a gap between the desired/expected audit quality and the actual audit quality itself.

The inconsistency is also shown by the research of Yen, Chang and Chen (2013). The authors used surveys to investigate the opinions of audit report stakeholders in Taiwan, because of the proposed regulations in 2009 and again in 2011, related to

(14)

the engagement partner`s signature in audit reports. The authors found that 92% of the participants support the regulations as proposed by the PCAOB. Fewer than half of these participants believe that mandating the engagement partner`s signature increases audit quality. This indicates that experts in the field want further empirical testing on whether the audit quality increases. This result is consistent with the results from the studies mentioned above, because with all the prior research, no decisive answer on the effects on audit quality can be given.

In conclusion, there is a lot of inconsistency in the literature on the effect of the engagement partner`s disclosure on audit quality. Carcello and Li found that abnormal accruals and the likelihood of meeting an earnings threshold decline after the effective date of the partner signature requirement. However, Blay, Notbohm, Schelleman and Valencia did not find any relationship between the engagement partner`s disclosure and a change in accruals quality or meeting an earnings threshold. This inconsistency could result from the fact that it is hard to measure audit quality (King et al., 2012; Bedard et

al., 2010; Oliverio, 2007; Watkins et al. 2004). In addition, according to Bierstaker et al. and Jones et al., the consequences of implementing the engagement partner`s

disclosure could lead to possible drawbacks for the auditors, while it is not certain if the auditor`s accountability and the audit quality will increase, as the audit quality in fact is hard to measure and remains unclear (King et al. 2012).

2.3 Prediction on the Investor`s Insights on the Engagement Partner`s Disclosure

The main advantage of disclosing the engagement partner`s name is the information usefulness for the investors. Clinch et al. (2012) investigated the relation between audit quality and information asymmetry between informed and uninformed traders. The authors found that when a Big 4 auditor and an industry specialist is used for an audit, the information asymmetry is lower between traders. So, high-quality auditing increases information quality and lowers information asymmetry between traders. The authors thereby extend prior research, showing that audit quality is valuable to investors. Therefore, more information about the engagement partner`s disclosure would result in less information asymmetry between traders and a higher audit quality. This is consistent with the view of the PCAOB that providing additional information about the audit process, the audit quality and decision usefulness increase. This would result in a positive market reaction on the events discussed earlier. This expectation is

(15)

confirmed with the research of Christensen et al. (2014). In their research in understanding audit quality, they found that investors rely on the individual characteristics of the engagement team performing the audit in assessing audit quality. Gul et al. (2013) investigated the effect of an individual auditor on audit quality. The authors found a significant effect of the individual auditor on the earnings quality, which they used as a proxy of audit quality. This means that the auditor`s individual characteristics have an effect on audit quality. Knechel et al. (2013) also conclude in their research that the engagement partner is essential to the audit quality. They measured audit reporting aggressiveness and conservatism by the type 1 and type 2 errors made, meaning giving a going concern opinion while the firm does not file bankruptcy within one year of the audit opinion and giving a clean opinion while bankruptcy is filed within one year of the audit opinion, respectively. The authors found that the audit reporting aggressiveness and conservatism persists over time, meaning that the auditor has a certain characteristic which increases the chance of the same error happening in the future, also with different engagements. These findings support the positive market reaction at the chosen event dates mentioned earlier, because it can be assumed that audit quality increases if the investor knows who the auditor is and what the past experience of the auditor is.

The primary benefits that the PCAOB expects from adopting either a signature or disclosure requirement are increased investor protection arising from greater partner accountability and audit transparency, which are expected to lead to enhanced audit quality and make it easier to assess the audit quality with regards to the track record of the auditor and the experience. The audit transparency leads to more helpful information on the audit, as well as increasing audit quality, because it could cause firms to improve the skills of the auditors and increase the competition for higher quality auditors (PCAOB, 2009, 2011). If this is really the case, I would expect a positive market reaction. However, there are several opponents of the standard. Auditors are concerned, because of the increased private liability as the U.S. has a more litigious litigation environment then the EU countries for example. However, this does not have to be a drawback for the investors. Francis and Krishnan (2002) underpin this with their study in which they provide evidence that, after the Private Securities Litigation Reform Act of 1995, the auditor`s legal exposure was reduced resulting in a less strict risk-management policy for both Big 4 as non-Big 4 auditors. The authors also provide evidence that because of the less strict risk-management, the Big 4 and

(16)

non-Big 4 had riskier clienteles and auditors became less conservative in their reporting strategies. These results imply that the amount of legal liability can affect the auditor`s incentives and behaviour. This would mean that with the increased liability because of the engagement partner`s disclosure, the auditors will tend to be more conservative and maintain a strict risk-management policy, leading to an increase in audit quality. In addition, it is probable that a higher litigation risk results in an increase in audit fees, as according to Prat and Stice (1994), the litigation risk is reflected in the audit fees. In addition, Higgs and Skantz (2006) find that there is a positive association between the earnings response coefficient, a proxy for earnings quality, and the level of residual fees. This means that the market interprets high audit fees as a sign for high earnings quality. Venkatamaran et al. (2008) found that in a higher-litigation regime, both audit quality and audit fee is higher. Carcello and Li (2013) also found in their study on the costs and benefits of requiring an engagement partner`s signature in the United Kingdom, that both the audit quality and the audit fees increase. In conclusion, the drawbacks concerning the increased private liability of the auditor isn’t a drawback for investors. Based on these arguments, it could be expected that the market reaction would still be positive.

The benefits for the investors mentioned above are aligned with the statement of the Council of Institutional Investors (CII) (2009). The CII states that “… the engagement partner plays a unique role in ensuring the overall quality of an audit. That role includes responsibility for planning the audit, supervising engagement team members, and determining whether the financial statements taken as a whole are fairly stated” (p. 2). The CII also thinks that with the enhanced investor oversight of selection and evaluation of the auditors, coming from the engagement partner`s disclosure, investors will demand skilled auditors and thus the audit quality increases (CII, 2009, p. 2). This is confirmed by the study of Chi et al. (2013) in which they conclude that for public companies a partner`s experience can be used to assess audit quality and that the pre-client experience enhances earnings quality, a proxy of audit quality. However, although there is literature suggesting audit quality to increase, there is also literature suggesting audit quality does not increase. In conclusion, the engagement partner`s disclosure does not have direct negative consequences for investors. Therefore the prediction is that investors either react positive to the announcements, or the investors do not react at all. Therefore the hypothesis is a signed prediction and can be stated as follows:

(17)

H1: The Engagement Partner`s disclosure announcements lead to a positive market

reaction.

3. Sample and Research Methodology

3.1 Sample

To test the hypothesis, a sample of the stock prices of all the listed companies on the NYSE, AMEX and NASDAQ in the United States stock markets during the days of the chosen events is required. The stock prices are collected from the Center for Research in Security Prices (CRSP) daily files of the Wharton Research Data Services. The CRSP data base however has data available till the 31st of December 2014. Since the sixth event took place in 2015, this event will be dropped. To test the hypothesis I will make inferences about the market reaction associated with the remaining five events taken together. Following Armstrong et al. (2010), the data on the stock prices therefore has to be available on all of the five events and on the event window [-1, 1] and [-3, 3], which will be discussed in chapter 3.2.2. The sample selection process is shown in Table 1. First all the data of all companies on every event date is collected. As mentioned, the data of all the companies have to be available on all the events. Due to this criterion, certain companies are dropped, as shown in Table 1. After the collection of the data on the available companies on all events, the data was imported into EVENTUS in order to gain additional data needed to perform statistical tests, which will be discussed in chapter 3.3. However, EVENTUS dropped certain events of which there is no, or not enough, data available, which resulted in the dropping of a few more companies. Eventually 3539 companies remained as the final sample. For the additional analysis, I require additional information, which reduced the sample size by 1471, resulting in a sample of 2068 companies.

(18)

Table 1: Sample selection Event 1: February 16, 2005 Event 2: October 6, 2008 Event 3: July 28, 2009 Event 4: October 11, 2011 Event 5: December 4, 2013 Total sample all events Number of companies 6893 6941 6648 6756 6796 34034

Companies dropped due to

unavailability on all events 3266 3346 3039 3180 3196 16027

Sample for EVENTUS 3627 3595 3609 3576 3600 18007

Security events dropped by

EVENTUS 88 56 70 37 61 312

Total sample from

EVENTUS 3539 3539 3539 3539 3539 17695

Total sample T-Test 3539 3539 3539 3539 3539 17695

3.2 Theoretical Background Event Study Methodology

3.2.1 Efficient Market Hypothesis

According to Fama et al. (1969), a stock market is efficient if all relevant information is accurately reflected by the security prices. This means that there is an instant or at least very rapid market response to all new information, which is called the efficient market hypothesis. The announcement of the PCAOB regarding the engagements partner`s disclosure and the potential benefits and/or drawbacks provides new information to the market. According to the efficient market hypothesis, the new information is shown in the stock prices instantly or at least very rapidly. Malkiel (1989) describes three levels of the Efficient Market Hypothesis (EMH) which can be distinguished. The first level is the weak form of EMH, in which the stock price is reflecting al market information. In the weak form, the information of past price patterns is already reflected in the share price. The semi-strong form of the EMH assumes that the stock price reflects all publicly available information and that stock prices adjust almost immediately to new information. At last, the strong version of the EMH assumes that stock prices reflect both publicly and private information. Event studies build upon the semi-strong form of the EMH, assuming investors cannot make higher than normal returns, because publically made information is reflected in the stock price instantly after the announcement.

(19)

In conclusion, under the assumption of the efficient market hypothesis, the impact of the engagement partner`s disclosure on the market can be measured with an event study, by measuring the stock price changes around the event dates of the PCAOB`s announcements of the engagement partner`s disclosure. In the next paragraph, I will explain in detail how the event study will be used in this research. In chapter 3.3 the empirical model of the event study is discussed.

3.2.2 Event study

An event study tests the impact of a financial, economic, accounting or political event on stock prices. Sreejith and Nageswara Rao (2014) critically examined the event study methodology. The authors conclude that the event study is a powerful technique to ascertain the impact of an event. Also according to Mackinley (1997, pp. 16) the event study is an ideal tool for examining the information content of disclosures.

Reynolds (2008), however, did a study on the accuracy of event studies and financial markets regarding the benefits associated with a new law. Reynolds concludes that event studies are poor predictors of the actual impact of the law on firms. This conclusion is based on two main problems. One of the problems is that investors themselves find it difficult to predict the impact of a regulatory change on a firm. In the case of the engagement partner`s disclosure, however, there has already been a lot of discussion regarding the new regulation and there is a lot of research done already. Therefore, investors will have a clear understanding of the new regulation at the announcements of the new regulations, especially because the discussion is going on for 10 years now. However, this could mean that investors have a different view on the regulation on the first event in comparison with the other events. As stated before, to examine the market reaction to the engagement partner`s disclosure announcements I will conduct an event study, in which I will examine the stock returns of all U.S. firms listed on the Dow Jones, New York Stock Exchange and NASDAQ on the engagement partner`s disclosure announcements of which the stock price data is available on all five events. The five different announcement dates regarding the engagement partner`s disclosure are discussed in the next paragraph.

As explained in chapter 3.2.1, according to the semi-strong version of the EMH, publically made information is reflected in the stock price instantly after the announcement. In addition, according to Mcwilliams and Siegel (1997) short windows

(20)

are less likely to contain other events which could disturb the study. Therefore, I will use a short event window of 3 and 7 days starting 1 and 3 days prior to the event and ending 1 and 3 trading days after the event [-1, 1] [-3, 3], respectively, with the event itself as day 0, as used by Kanas (2005).

The last steps of the event study are calculating the expected return, the actual return, the abnormal return and the cumulative abnormal return (Mackinlay, 1997). These steps will be discussed in chapter 3.3.

3.2.3 Engagement Partner`s Disclosure Announcements

Since the announcement of the engagement partner`s disclosure in the U.S., there has been a lot of discussion regarding the benefits and the drawbacks of the possible adoption. During the process, several concepts and comments were released. These events are summarized in Table 2.

Table 2: The Engagement Partner`s disclosure announcements in the United States

Date

Announcement

February 16, 2005 The Standing Advisory Group discussed whether the audit report signature should include the name of the partner and second partner signing-off on the report or other members of the engagement team. Proponents suggest that naming the

individuals in the auditor`s report might enforce the importance of performing the audit more diligently, only suggesting that it would increase audit quality (Auditor's Reporting Model February 16, 2005 Page 8).

(21)

October 6, 2008 Recommendation of the Advisory Committee on the Auditing Profession to the department of treasury (Panel Discussion – Signing the Auditor's Report October 22-23, 2008). In this final report the Advisory Committee urges the Public Company Accounting Oversight Board (PCAOB) to undertake a standard-setting initiative to consider mandating the engagement partner`s disclosure. In the report it is stated that advocates of the

engagement partner`s disclosure believe that it will lead to greater accountability of the individual auditors, enhanced transparency, and that it could improve audit quality. In addition, they state that the disclosure will not create any additional liability concerns for the engagement partner.

July 28, 2009 Concept release issued by the Public Company Accounting Oversight Board (PCAOB) with the request for public comment on whether the PCAOB should require the auditor with the final responsibility for the audit to sign the report (PCAOB Release No. 2009-005 July 28, 2009). In the concept release the PCAOB argues that the engagement partner`s signature included in the audit report would make the audit report more informative, enhances the engagement partner`s accountability and increases transparency, leading to a higher audit quality as an additional incentive.

October 11, 2011 Amendments after the concept release in 2009. The main change here is that, after considering commenter`s views, the

engagement partner`s name is to be disclosed, but there will be no requirement to include the signature in the audit report. The firm`s signature will be the only signature included in the audit report. This would result in the same benefits of the increased audit quality, but it would reduce concerns that the engagement partner`s signature would minimize the firms role in the audit.

December 4, 2013 PCAOB is re-proposing amendments. These amendments would improve the transparency, by disclosing not only the name of the engagement partner, but also 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. The PCAOB believes that the additional amendment would be useful to investors and other financial statement users.

(22)

January 15, 2015 The International Auditing and Assurance Standards Board (IAASB) revised the International Standard on Auditing (ISA) 700. In the revised ISA, the IAASB includes the disclosure of the name of the engagement partner to be included in the auditor’s report for audits of complete sets of general purpose financial statements of listed entities. The ISA 700 (Revised) is effective for audits of financial statements for periods ending on or after December 15th 2016.

3.3 Empirical Model

To measure the impact on the stock market I will use the cumulative abnormal return (CAR). In order to calculate the CAR, certain calculations must be made first. These are the expected, actual and abnormal returns. With the abnormal returns it is possible to investigate whether shareholders react to a certain event and if the information arising from the event is relevant to the shareholders. The expected return (1) shows the price of stock if the chosen events would not have taken place and is calculated by the following formula:

(1)

Where: E(Rit) is the expected return on the capital asset Rf is the return on an investment with no risk

β is the risk compared to the rest of the market (volatility) Rm is the return of the market

Following the study of Armstrong et al. (2010) and Cheung (2011), the CRSP value-weighted market-adjusted returns are used to calculate the expected return. The CRSP value-weighted index includes the NYSE, AMEX and Nasdaq Market Indices. According to the study of Campbell et al. (1997), by using the market-adjusted returns, the effect of events that occur on the same day as the engagement partner`s disclosure announcements, which could disturb the results of the event study, are mitigated. The estimation window to calculate the expected return is set on 250 trading days, ending 46 days prior to every event. With the settings mentioned above, EVENTUS calculates the beta automatically in order to calculate the cumulative returns. The actual return is

(23)

the current stock price at the day of the announcement. The abnormal return (2) can be calculated by extracting the expected return from the actual return:

(2)

Where: ARit is the abnormal return of the stock Rit is the actual return of the stock E(Rit) is the expected return of the stock

The CAR is the sum of all the abnormal returns. The CAR (3) can then be calculated by the following formula:

(3)

Where: CARit is the cumulative abnormal return of the stock ARit is the abnormal return of the stock

EVENTUS calculates the CAR automatically. In order to determine whether the CAR is significantly different from 0, meaning that H1 is supported, I will use a t-test. The t-test will be performed by STATA. The t-test will be performed for all the events taken together, which is based upon the CAR of all the five events.

4. Results

4.1 Descriptive Statistics

In Table 3 the descriptive statistics for all events are shown for both event windows separately. The mean and median for all events together for both event windows are 0.002 [-1, 1] -0.0002 [-3, 3], 0.002 [-1, 1] and -0.001 [-3, 3], respectively. This means that the average CAR for window [-1, 1] is 0.22% and for window [-3, 3] -0.02%, as shown in Table 5. Overall, the CAR of all events together is slightly higher for window [-1, 1], meaning that the market reacted slightly more positive than normal. For window [-3, 3] the CAR of all events together is negative, although very small, which means that when a larger event window is used, the market reaction goes from slightly positive

(24)

to slightly negative. Although, the t-test will show whether the changes are significant or not, the slight decrease of the CAR are no reason to change the signed prediction of the hypothesis, which is that the investors will either react positive or will not react at all.

Table 3: Descriptive statistics of the cumulative abnormal return of all events All events Window [-1, 1] All events Window [-3, 3] N 3539 3539 Mean 0.002 -0.0002 SD 0.03 0.04 Median 0.002 -0.001 Min -0.12 -0.20 Max 0.63 0.56 SE 0.0004 0.0007

In Table 4 the descriptive statistics for event window [-1, 1] and [-3, 3] per separate event are shown. At first glance, it is notable that the means for the third event show the most positive CAR of 1.06% and 1.62% for event window [-1, 1] and [-3, 3], respectively (Table 5). This could be because the third event is the first event in which a concrete concept version of the engagement partner`s disclosure is presented (Table 2). The first event shows a slightly positive CAR of 0.02% and 0.2% for event window [-1, 1] and [-3, 3], respectively (Table 5). The second event shows a negative CAR of -0.67% and -1.49% (Table 5). This event took place on the 6th of October 2008 (Table 2). Therefore the financial crisis could cause disturbance on the chosen event date. The fourth event shows a slightly positive CAR of 0.57% on event window [-1, 1] and a very small negative CAR of -0.03% on event window [-3, 3] (Table 5), which is therefore again no reason to not assume the signed prediction as discussed before. The fifth event shows a slight increase of 0.1% for event window [-1, 1] and a slight decrease of -0.37% for event window [-3, 3] (Table 5). Table 6 shows the abnormal returns in percentages on the event dates separately and taken together, starting 3 days before the event date (0) till 3 days after the event date. In Table 6 at the event day, the abnormal return is either positive or slightly negative, except for the second event date, which shows a more negative abnormal return. As discussed before, this could be due to effects of the financial crisis as the abnormal returns in the second event show a lot

(25)

of fluctuation, which could be an indication of an instable market. However, it is interesting that at day 1 the abnormal return for every event is positive, even for the second event, which justifies the assumption of the signed positive prediction in the hypothesis development. In Table 5, I also show the CAR for all the events separately and all events together for event window [0, 1] and [0, 3]. Measuring the CAR with these event windows assumes that there was no information leakage before the event date itself at day 0. The results show an overall positive CAR for both event windows, with again the second event as an exception as the second event still shows a negative CAR for both event windows. Furthermore, Table 5 shows that also with the event window at [0, 1] and [0, 3] the third event shows the most positive CAR. However, the positive results remain to be rather small. Before any inferences can be made, these result are tested for statistical significance in the next paragraph.

(26)

Table 4: Descriptive statistics of the cumulative abnormal return per separate event Event 1 Window [-1, 1] Event 1 Window [-3, 3] Event 2 Window [-1, 1] Event 2 Window [-3, 3] N 3539 3539 3539 3539 Mean 0.0002 0.0020 -0.0067 -0.0149 SD 0.03 0.05 0.09 0.15 Median -0.0004 -0.0019 -0.0005 -0.0001 Min -0.22 -0.30 -0.56 -0.95 Max 0.29 0.39 0.39 1.54 SE 0.0006 0.0009 0.0014 0.0026 Event 3 Window [-1, 1] Event 3 Window [-3, 3] Event 4 Window [-1, 1] Event 4 Window [-3, 3] N 3539 3539 3539 3539 Mean 0.0106 0.0162 0.0057 -0.0003 SD 0.08 0.11 0.05 0.07 Median 0.0053 0.0015 0.0018 -0.0046 Min -0.39 3.41 -0.34 -0.40 Max -0.57 3.23 0.83 0.75 SE 0.0014 0.0018 0.0009 0.0012 Event 5 Window [-1, 1] Event 5 Window [-3, 3] N 3539 3539 Mean 0.0010 -0.0037 SD 0.04 0.06 Median 0.0007 -0.0049 Min -0.34 -0.52 Max 1.00 1.44 SE 0.0006 0.0010

Table 5: Cumulative abnormal return in percentages

CAR (%) CAR (%) CAR (%) CAR (%) CAR (%) CAR (%) N=3539 N=3539 N=3539 N=3539 N=3539 N=17695 Event 1 Event 2 Event 3 Event 4 Event 5 All events

Window(-1,1) 0.02 -0.67 1.06 0.57 0.10 0.22

Window(-3,3) 0.20 -1.49 1.62 -0.03 -0.37 -0.02

Window(0,1) 0.25 -0.31 0.63 0.79 0.12 0.29

(27)

Table 6: Abnormal Returns

AR (%) AR (%) AR (%) AR (%) AR (%) AR (%) N=3539 N=3539 N=3539 N=3539 N=3539 N=17695 Event 1 Event 2 Event 3 Event 4 Event 5 All events

Day -3 0.05 0.08 0.05 -0.07 0.34 0.09 Day -2 0.1 0.37 0.11 -0.45 -0.41 -0.05 Day -1 -0.22 -0.36 0.43 -0.22 -0.02 -0.08 Day 0 0.18 -1.2 0.49 0.38 -0.01 -0.03 Day 1 0.07 0.89 0.13 0.41 0.12 0.33 Day 2 -0.13 -1.34 0.20 0.18 -0.14 -0.25 Day 3 0.15 0.07 0.19 -0.25 -0.26 -0.02

4.2 Results of Hypothesis Tests

In Table 7, the one sample t-test is shown for all events on event window 1, 1] and [-3, 3], based upon a 95% confidence interval. The t-value for event window [-1, 1] is 4.8910 and for event window [-3, 3] -0.2324. The p-value of event window [-1, 1] is smaller than .001, which is highly significant on a 1% level. However, the p-value of event window [-3, 3] with 0.592 based on the one sided p-value is not. This means that the first event window shows support for the hypothesis 1 as the mean of 0.22% for the CAR significantly differs from 0, so the null hypothesis can be rejected. The mean of the second event window of -0.02% for the CAR shows no support for H1, which is in line with the predictions that investors either react positively or do not react at all, as the CAR of the second window is slightly negative. As the 2-sided p-value of 0.816 is not significant either, the null hypothesis cannot be rejected.

By performing a one sample t-test, a normal distribution is assumed. However, as Table 7 shows, the data is skewed and the kurtosis is very high, thereby indicating a normal distribution. To maintain the power of the results, I will perform a non-parametric test in addition to the one sample t-test for robustness of the results. Non-parametric tests are more appropriate in situations that deal with data that depart from normality (Siegel, 1957). The non-parametric test will be the Wilcoxon signed-rank test, because with the Wilcoxon signed-rank test it is not necessary to assume that the data is normally distributed. For the Wilcoxon signed-rank test I test the same hypothesis as for the t-test. The results for the Wilcoxon signed-rank test are shown in Table 8. The p-value for the first event window [-1, 1] is again very significant and the

(28)

second event window [-3, 3] is not significant, as we saw also with the t-test results. Overall, the results of the Wilcoxon signed-rank test are consistent with the t-test.

Table 7: One sample t-test all events All events Window [-1, 1] All events Window [-3, 3] N 3539 3539 Mean 0.0022 -0.0002 SD 0.03 0.04 Median 0.0018 -0.0010 Min -0.12 -0.20 Max 0.63 0.56 SE 0.0004 0.0007 Skewness 4.43 1.71 Kurtosis 101.68 24.67 Df 3538 3538 t 4.8910 -0.2324 P-value (2-sided) p < .001*** 0.816 P-value (1-sided) p < .001*** 0.592 *, **, *** Significant at 10%, 5%, 1% levels, respectively

Table 8: Wilcoxon signed-rank test

All events Window [-1, 1] All events Window [-3, 3]

Z-value 5.251 -1.61

One-sided p-value p < .001*** 0.9463 *, **, *** Significant at 10%, 5%, 1% levels, respectively

For additional robustness I also performed t-tests per separate events for both event window [-1, 1] and [-3, 3] to investigate whether the results of the t-test of all the events taken together are aligned with the results per separate event. These results are shown in Table 9. The one-sided p-values for event window [-1, 1], based upon a 95% confidence interval, for event 1 till 5 are 0.3487, bigger than 0.999, smaller than 0.001, smaller than 0.001 and 0.0616, respectively. The first two events are not significant, however, the third till fifth event are significant, the third and fourth event on a 1% level and the fifth event on a 10% level. This means that for these events, the null hypothesis can be rejected as the mean significantly differs from 0. For the first event the null hypothesis cannot be rejected. For the second event, the null hypothesis cannot be rejected either. However, as discussed before, the second event shows very opposite

(29)

results compared with the other events. The second event is significant by the two-sided p-value with a p-value smaller than 0.001 and a mean CAR of -0.67% for event window [-1, 1], meaning that the mean CAR of the second event is significant, but not positive as was expected. The results of the t-test per separate event for the event window [-3, 3] are as follows, event 1 shows a one-sided p-value of 0.0150, for event 2 the p-value is higher than 0.999, for event 3 smaller than 0.001, for event 4 0.5877 and for event 5 0.9998. This means that for event 1 is significant on a 5% level, with a mean cumulative return of 0.2%, showing support for H1, so the null hypothesis can be rejected as the mean CAR significantly differs from 0. Event 2 is again not significant, but is consistent with the event window of [-1, 1] as discussed before. For the event window [-3, 3], event 2 is significant with a two-sided p-value smaller than 0.001 and with a mean cumulative return of -1.49%. The third event is significant again on a 1% level and with a mean cumulative return of 1.62%, supports H1. Event 4 shows no significance. With a mean cumulative return of -0.03% the fourth event also shows no significance on the two-sided p-value, which is consistent with the reasoning that investors will react positively or will not react at all, as the negative cumulative return is very small and negligible as the two-sided p-value is 0.8247, which means that the null hypothesis cannot be rejected and the mean CAR is not significant different from 0. The fifth event does not show a significant result for event window [-3, 3], while event window [-1, 1] for the fifth event supported H1. The mean CAR on the fifth event is -0.37% and with a two-sided p-value of 0.0004, this is significantly different from 0, but not on the positive end, as was expected. As with the t-test for all event together, the normal distribution is also assumed for the t-test for every event separate. As Table 9 shows, the data is skewed and the kurtosis is very high, thereby indicating a non-normal distribution. To maintain the power of the results, I will also perform a non-parametric test for the one sample t-test per separate event with the Wilcoxon signed-rank test. The results for the Wilcoxon signed-rank test are shown in Table 10. At first site, it becomes obvious that the z-values of the Wilcoxon signed-rank test differ a lot, as the overall positive results from the t-test do not recur in the Wilcoxon signed-rank test as the Wilcoxon signed-rank test shows a lot of negative z-values, which represent the total difference of the CAR when compared to 0. This could mean that the overall reaction is in fact negative, which means the null hypothesis cannot be rejected. However, I continue with the signed prediction as has been done with the t-test. The one-sided p-values for event 1 till 5 for event window [-1, 1] are 0.8502, 0.993, smaller than 0.001,

(30)

smaller than 0.001 and 0.396, respectively. Of these p-values, only the p-values for the third and fourth events are significant (at a 1% level). This is consistent with the t-test results for the same event window as it mean that the second and fourth event significantly differ from 0, showing support for H1. Although in the Wilcoxon signed-rank test the fifth event is not significant, where the fifth event was significant in the t-test, however only at a 10% level. For the second event the results are again consistent with the t-test. With a one-sided p-value of 0.993 the second event is almost certain significant on the negative side. The one-sided p-values for event 1 till 5 for event window [-3, 3] are 0.7193, higher than 0.999, smaller than 0.001, 0.5877 and 0.998, respectively. Of these p-values, only the p-value of the third event is significant at a 1% level, supporting H1. However, with the t-test, the first event was also significant (5% level) while the first event at window [-3, 3] does not show significance with the Wilcoxon signed-rank test.

Overall, the results of the Wilcoxon signed-rank test are consistent with the t-test. Although the first event for window [-3, 3] is not significant and the fifth event for window [-1, 1] is not significant either, the outcomes are not very extreme and the significance level with the t-test are at a 5% level for the first event and at a 10% level for the fifth event, which already gives a lower amount of certainty as with the 1% level of significance, of which the results do still hold with the Wilcoxon signed-rank test. Also, the tests for all events together are consistent, so inferences can be made with certainty for all the events taken together. However, the outcomes for the second event are very extreme and could bias the overall result, as the values for the second event are very significant on the negative side, which is not as expected as the investors do not have any downside regarding the engagement partner`s disclosure, as discussed in chapter 2.3. In addition, the announcement of the second event (chapter 3.2.3) does not give any fundamentally different information to expect investors to react in a different manner. Therefore, it could be possible that other events bias the results for the second event. To add more robustness I will perform another t-test, excluding the second event.

(31)

Table 9: One sample t-test per separate event Event 1 Window [-1, 1] Event 1 Window [-3, 3] Event 2 Window [-1, 1] Event 2 Window [-3, 3] N 3539 3539 3539 3539 Mean 0.0002 0.0020 -0.0067 -0.0149 SD 0.03 0.05 0.09 0.15 Median -0.0004 -0.0019 -0.0005 -0.0001 Min -0.22 -0.30 -0.56 -0.95 Max 0.29 0.39 0.39 1.54 SE 0.0006 0.0009 0.0014 0.0026 Skewness 0.68 0.74 -0.49 -0.33 Kurtosis 11.74 9.63 5.79 8.60 Df 3538 3538 3538 3538 t 0.3889 2.1720 -4.6370 -5.7684 P-value (2-sided) 0.6974 0.0299** p < .001*** p < .001*** P-value (1-sided) 0.3487 0,0150** p > .999 p > .999 Event 3 Window [-1, 1] Event 3 Window [-3, 3] Event 4 Window [-1, 1] Event 4 Window [-3, 3] N 3539 3539 3539 3539 Mean 0.0106 0.0162 0.0057 -0.0003 SD 0.08 0.11 0.05 0.07 Median 0.0053 0.0015 0.0018 -0.0046 Min -0.39 3.41 -0.34 -0.40 Max -0.57 3.23 0.83 0.75 SE 0.0014 0.0018 0.0009 0.0012 Skewness 19.67 8.81 2.48 1.54 Kurtosis 750.55 227.14 34.72 15.45 Df 3538 3538 3538 3538 t 7.4800 8.8296 6.2689 -0.2215 P-value (2-sided) p < .001*** p < .001*** p < .001*** 0.8247 P-value (1-sided) p < .001*** p < .001*** p < .001*** 0.5877 Event 5 Window [-1, 1] Event 5 Window [-3, 3] N 3539 3539 Mean 0.0010 -0.0037 SD 0.04 0.06 Median 0.0007 -0.0049 Min -0.34 -0.52 Max 1.00 1.44 SE 0.0006 0.0010 Skewness 5.96 6.12 Kurtosis 158.29 128.23 Df 3538 3538 t 1.5416 -3.5477 P-value (2-sided) 0.1233 0.0004*** P-value (1-sided) 0.0616* 0.9998 *, **, *** Significant at 10%, 5%, 1% levels, respectively

(32)

Table 10: Wilcoxon signed-rank test all events separate Event 1 Window [-1, 1] Event 1 Window [-3, 3] Event 2 Window [-1, 1] Event 2 Window [-3, 3] Z-value -1.037 -0.581 -2.456 -3.307 One-sided p-value 0.8502 0.7193 0.993 p > .999 Event 3 Window [-1, 1] Event 3 Window [-3, 3] Event 4 Window [-1, 1] Event 4 Window [-3, 3] Z-value 8.814 5.808 4.22 -4.097 One-sided p-value p < .001*** p < .001*** p < .001*** p > .999 Event 5 Window [-1, 1] Event 5 Window [-3, 3] Z-value 1.756 -10.903 One-sided p-value 0.0396** p > .999 *, **, *** Significant at 10%, 5%, 1% levels, respectively

In Table 11 the CAR in percentages are shown. With the second event removed, all the event windows show a very positive and stable cumulative return in comparison with the second event still included. The mean CAR for event window [-1, 1], [-3, 3], [0, 1] and [0, 3] are 0.44%, 0.35%, 0.45% and 0.43%, respectively. In Table 12 the t-test for all events, excluding the second event, is shown. The one-sided p-values for both event windows of [-1, 1] and [-3, 3] are smaller than 0.001 which is significant at the 1% level. This means that H1 is supported and that the null hypothesis can be rejected, but moreover, this means that the mean CAR of 0.44% and 0.35% are significantly different from 0. This suggests a slightly positive reaction from the investors to the new information regarding the announcements of the engagement partner`s disclosure.

However, as shown in Table 12, the kurtosis and skewness are still very high, especially for the first event window. Therefore, I performed a Wilcoxon signed-rank test also with the second event excluded. The results of the Wilcoxon signed-rank test are shown in Table 13. The first event window [-1, 1] shows a z-value of 8.845 with a one-sided p-value smaller than 0.001 on a 1% significance level, meaning that the Wilcoxon signed-rank test for the first event window also shows support for H1, which is consistent with the t-test results for the first event window. For the second event window [-3, 3] the Wilcoxon signed-rank test shows a z-value of 1.516 with a one-sided p-value of 0.0648, meaning that the Wilcoxon signed-rank test for the second window also shows support for H1, which is consistent with the t-test results for the second event window. However, the significance for the second event window is at a 10% level,

(33)

which does give less statistical power. Overall, the t-test and Wilcoxon signed-rank test show similar results.

Table 11: Cumulative abnormal return in percentages all events ex. 2nd event CAR (%)

N=17695 All events ex.

2nd event Window(-1,1) 0.44

Window(-3,3) 0.35

Window(0,1) 0.45

Window(0, 3) 0.43

Table 12: One sample t-test all events ex. 2nd event All events ex.

2nd event [-1, 1]

All events ex. 2nd event [-3, 3] N 3539 3539 Mean .0044 .0035 SD .03 .04 Median .0017 -.0008 Min -.11 -.17 Max .82 .81 SE .0005 .0007 Skewness 9.11 3.80 Kurtosis 248.56 59.28 Df 3538 3538 t 9.6478 5.1809 P-value (2-sided) p < .001*** p < .001*** P-value (1-sided) p < .001*** p < .001*** *, **, *** Significant at 10%, 5%, 1% levels, respectively

Table 13: Wilcoxon signed-rank test all events ex. 2nd event All events ex.

2nd event [-1, 1]

All events ex. 2nd event [-3, 3]

Z-value 8.845 1.516

One sided p-value p < .001*** 0.0648* *, **, *** Significant at 10%, 5%, 1% levels, respectively

Stock market reaction to the engagement partner`s disclosure announcements in the U.S.

Master thesis: