Non-audit services and perceived audit quality:
The impact of the audit committee
Final version
Maaike Boot (10105743) June 2015
Words count: 15655
Supervisor: Prof. Dr. V.S. Maas
MSc Accountancy & Control, variant Accountancy Amsterdam Business School
Statement of Originality
This document is written by student Maaike Boot 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.
Abstract
This study focuses on whether audit committee characteristics moderate the relationship between non-audit services and perceived audit quality. Prior literature does not agree on whether non-audit services impair perceived audit quality. According to the Sarbanes Oxley Act (SOX) of 2002, the main purpose of the audit committee is to oversee the audit of its financial statements. I test whether the independence, financial expertise and the frequency of meetings of the audit committee increases the value relevance of earnings information, as measured by the earnings response coefficient (ERC). I expect firms with effective audit committees to report earnings information that is more precise and of higher accuracy. Therefore, I expect the ERC to be higher when firms have effective audit committee
characteristics. The results suggest that financial expertise of the board indeed moderates the negative relationship between non-audit services and perceived audit quality. However, the independence of the audit committee and the number of meetings deteriorates the negative relationship between non-audit services and perceived audit quality.
Key words: Audit committee, non-audit services, perceived audit quality, earnings response coefficient
Table of contents
1. Introduction……….. 1
2. Literature review………. 3
2.1 The provision of non-audit services and perceived audit quality……… 3
2.2 The audit committee and perceived audit quality……… 6
2.3 Hypotheses development………. 9
3. Data and research design……… 14
3.1 Data collection………. 14
3.2 Sample selection……….. 15
3.3 Research design………... 16
3.3.1 Measuring perceived audit quality: the ERC……… 16
3.3.2 Measurement of non-audit services……….. 18
3.3.3 Audit committee variables………18
3.3.4 Control variables………... 20
3.3.5 The research model……….. 20
4. Results……….. 23
4.1 Descriptive statistics……… 23
4.2 Empirical results……….. 26
4.2.1 Non-audit services and perceived audit quality……… 26
4.2.2 Audit committee characteristics and perceived audit quality………... 28
4.2.3 The effect of the audit committee on the relationship between non-audit services and perceived audit quality………. 29
4.3 Additional tests……… 35 4.3.1 Market-to-book value………... 35 4.3.2 Industry-specific results……… 36 5. Discussion………..38 6. Conclusion……….40 7. References………. 43
1 1. Introduction
This study focuses on whether the relationship between non-audit services and perceived audit quality is conditional upon the independence, the financial expertise and the activity of the audit committee.
Krishnan et al (2005) find a negative association between the purchase of non-audit services and the earnings response coefficient (ERC), which implies that investors did perceive non-audit services as impairing audit quality. However, this research does not provide us with an explanation about potential moderating factors that have influenced the association between the purchase of non-audit fees and perceived audit quality.
DeFond and Zhang (2014) define higher audit quality as greater assurance that the financial statements faithfully reflect the firm’s underlying economics, conditioned on its financial reporting system and innate characteristics. I expect the audit committee to be part of the innate characteristics of the firm and thus to affect audit quality.
This research studies whether the effect of non-audit services on perceived audit quality is dependent upon the audit committee. I test whether the association between the provision of non-audit services and the perceived impairment of audit quality is moderated by several audit committee characteristics.
The results in prior literature on the effects of the provision of non-audit services on audit quality are mixed, mainly dependent upon the proxy of quality used. Literature on the relationship between non-audit services and audit quality is subdivided into two main areas: studies that focus on the effects on actual quality and studies that focus on perceived audit quality. On the one hand, the provision of non-audit services creates economic bonds that weaken auditor independence and therefore perceived auditor quality (Lim and Tan, 2008). On the other hand, research indicates that reputation concerns, knowledge spillovers and litigation exposure serve to counter incentives for auditors to remain independent. Some governance failures in the beginning of this century, such as the Enron bankruptcy in 2001 and the related collapse of Arthur Andersen, made that the provision of non-audit services, the independence of the auditor and related audit quality has been questioned. There were policy debates in early 2002 about the possibility of proscribing or limiting the provision of non-audit services by accounting firms. The Sarbanes- Oxley Act (hereafter mentioned as SOX) of 2002 is the most important legislation affecting corporate financial reporting enacted in the US since the 1930s. The purpose of SOX is to improve accuracy and reliability of accounting information that is reported to investors (Li et al, 2008).
2
One of the main elements of the SOX Act is the focus on the role of the audit committee and the prohibition of some non-audit services. According to the Security and Exchange
Commission (SEC), the audit committees play a critical role in the financial reporting system by overseeing and monitoring management’s and the independent auditor’s participation in the financial reporting process. They believe that the audit committee will promote investor confidence in the integrity of the financial reporting process. Because of the expected positive influence on perceived audit quality, I expect the audit committee to moderate the negative relationship found by Krishnan et al (2005) between non-audit services and perceived audit quality. The research question I will answer is as follows: Does the audit committee
moderates the association between non-audit services and perceived audit quality?
To answer the research question, a final sample of 239 firms of the S&P500 is studied in the year 2011. Data is collected from Datastream, Compustat, AuditAnalytics and the EDGAR database of the SEC, where data on the audit committee characteristics is hand collected from the proxy statements. As a proxy for perceived audit quality, the earnings response coefficient is used. In the regressions, I control for the firms’ systematic risk, loss, firm size and growth prospects.
This study contributes to prior literature in the following ways: First, it contributes to the ongoing debate whether non-audit services impair perceived auditor quality. This study focuses on the year 2011, which is ten years later than the time period in the study of
Krishnan et al (2005). Therefore, I can focus on whether their results still hold ten years later. Second, this study focuses on a moderating factor that has not been studied yet before, namely audit committee characteristics. I study whether the audit committee is able to increase
perceived audit quality, especially when non-audit services are provided. Finally, it also studies which audit committee characteristics are more important in affecting the association between the provision of non-audit services and perceived audit quality. I therefore look at the three most mentioned audit committee characteristics in prior literature, namely the
independence, the financial expertise and the number of meetings of the audit committee. The results of this study could be interesting for regulators to see whether the rules imposed by, for example the Security and Exchange Commission and the SOX Act in 2002 on enhancing audit quality by strengthening the requirements of the audit committee have improved the perceived audit quality. If this is indeed the case, then it might not be necessary to consider prohibiting non-audit services, because perceived audit quality can indirectly be improved by the audit committee. In this way, the positive effects of the provision of non-audit services such as knowledge spillovers can remain, whilst the negative side effects of
3 non-audit services can be limited.
I find that perceived audit quality is indeed lower when non-audit services are provided. When the audit committee is independent or includes a financial expert, perceived audit quality is higher. However, when the audit committee is independent or the frequency of meetings increases, the negative association between non-audit services and perceived audit quality deteriorates. The financial expertise of the audit committee is the only moderating characteristic of the audit committee found that improves perceived audit quality when non-audit services are provided. In other words, the financial expertise of the non-audit committee increases the value relevance of information and therefore the investors respond more strongly to the reported earnings information.
The structure of the rest of the paper is organized as follows. The second section provides results from prior literature on the relationship between non-audit services and perceived audit quality, the relationship between the audit committee and perceived audit quality and finally develops the hypotheses. The third section describes the data selection process, introduces the variables and explains the research models. The fourth section contains the empirical findings and the fifth and last section describes the discussion of the results and the final conclusion.
2. Literature review
2.1 The provision of non-audit services and perceived audit quality
DeAngelo (1981) defines audit quality as the joint probability that an auditor will detect and report a financial misstatement. DeFond and Zhang (2014) believe that this characterization understates the benefits of high audit quality, which extend well beyond the simple detection and reporting of GAAP violations to assuring financial reporting quality. They expect high quality auditors to consider not only whether the client’s accounting choices are in technical compliance with GAAP, but also how faithfully the financial statements reflect the firm’s underlying economics. DeFond and Zhang (2014) define audit quality as the continuous construct that assures financial reporting quality, with high quality auditing providing greater assurance of high quality financial reporting. The supply of audit quality is a function of the auditor’s independence and competency, where independence arises from reputation and litigation incentives, and competency refers to the ability to deliver high audit quality (DeFond and Zhang, 2014). If auditors provide non-audit services to their clients, then this
4
may lead to decreased auditor independence because the non-audit services create economic bonds that make the auditor more dependent on the firm. Decreased auditor independence could potentially mean that the likelihood of detecting and reporting a financial misstatement decreases, and could thus potentially lead to lower audit quality. Lower audit quality might decrease financial reporting quality by decreasing the credibility of the financial reports (DeFond and Zhang, 2014). This could in turn have implications for the effective capital allocation in the economy, which is seen as the ultimate goal of financial reporting regulation. There might, however, also be a positive association between non-audit services and audit quality, for example when reputation concerns, knowledge spillovers and litigation exposure serve to counter incentives for auditors to remain independent, even when they provide non-audit services. In this case, it is likely that the non-audit services provided will have little impact on the auditor’s ability to detect a material misstatement. However, prior literature did not find strong evidence that reputation and litigation risk improves audit quality. At the same time, this does not mean that the provision of non-audit services may not be perceived as reducing the auditor’s willingness to report a material misstatement (Abbott et al. 2003), and thus reducing perceived auditor quality. Perceived auditor quality, which is studied in this research, might be as important as actual audit quality. In the end, it is the opinion of investors about audit quality that is important, since they are the ones who decide on allocating their capital. If they perceive the auditor’s quality to be impaired, they may decide not to allocate their capital because information asymmetries between the firm and the investors are still present as the disclosed information might not be reliable and therefore not useful for their decisions.
Due to some large governance failures in the past, such as the Enron Debacle, the provision of non-audit services has been debated extensively over the last years. The Enron case highlighted many of the characteristics that had previously raised concerns about auditors’ declining independence, objectivity and professional scepticism. In the years following the Enron debacle several regulations tried to impose rules that would improve the quality of the audit and investors’ perceptions of it. For example, the SOX Act of 2002 mainly focuses on the issue of auditor independence. Public confidence in the integrity of financial statements of publicly traded companies is based on the belief in the independence of the auditor from the audit client. Title II of the SOX Act is devoted to auditor independence. Among its provisions are strengthened rules on prohibited activities, codifying and expanding on existing auditor independence, audit committee pre-approval of audit and non-audit
5
policies and practices to be used, alternative treatments discussed with management and other material written communications and a cooling off period of one year before the auditor can serve in a key management position at an audit client. In other sections of the SOX legislation also appear provisions that deal with auditor independence. For example, audit committees are made responsible for appointing and paying the auditor and overseeing the conduct of the audit, including resolving any conflicts with management. SOX also strengthens
independence and financial expertise requirements for audit committees.
Notwithstanding the concerns about the effect of the provision of non-audit services on audit quality, from prior literature there is no clear consensus whether non-audit services impair audit quality. Therefore, it seems that there are other characteristics of the firm that investors take into account when assessing the quality of the audit.
Ashbaugh et al (2003) investigate whether non-audit fees are positively associated with measures of biased financial reporting. They find no evidence that the market reacts to the magnitude of non-audit fees relative to total fees, in other words, that auditors violate their independence as a result of clients purchasing relatively more non-audit services. On the other hand, Quick et al (2005) found that shareholders, bank loan officers and journalists perceive a negative effect on auditor independence if non-audit services are provided, but that the type of non-audit services influences the degree to which auditor independence is perceived to be impaired. They conclude that independence rules related to non-audit services should be differentiated. In line with Quick et al (2005), Jenkins and Krawczyk (2011) consider three different non-audit services and find that only one of these services, namely legal consulting, evoked negative perceptions of auditor independence from financial statement users, while the other two services failed to produce any significant response. Solomon et al (2005) also find that participants have less confidence in financial information audited by firms that simultaneously provide significant non-audit services, suggesting that auditors are perceived to be less independent when they provide additional non-audit services. Studies that examine the overall effects of SOX find ambiguous results. This studies find that banning non-audit services does not seem to affect audit quality, and tax-related non-audit services actually improves it (DeFond and Zhang, 2014).
The study of Krishnan et al (2005) is used as the starting point for this study, mainly because this is one of the few studies that focus on the relationship between the provision of non-audit services and perceived audit quality using the Earnings Response Coefficient as a proxy for perceived audit quality. They find that the investors’ perception of audit quality is impaired when firms provide non-audit services. Taking into account the results from this
6
study, I attempt to find whether certain audit committee characteristics moderate the negative relationship between the purchase of non-audit services and perceived audit quality. The next paragraph examines how the audit committee is expected to influence audit quality and which results are found in prior literature.
2.2 The audit committee and perceived audit quality
With an increasing number of investors entering the markets and with changes in the way and speed with which investors receive information, it is vitally important for investors to remain confident that they are receiving the highest quality financial reporting. However, companies might be under increasing pressure to meet earnings expectations which make regulators concerned about inappropriate earnings management, the practice of distorting the true financial performance of the company. These changes in the markets and the increasing pressure on companies to maintain positive earnings trends have highlighted the importance of strong and effective audit committees.
Based on the recommendations of the Blue Ribbon Committee on Improving the Effectiveness of Corporate Audit Committees, with an effective date of January 31, 2000, the Security and Exchange Commission adopted new rules and amendments to its current rules to require that companies’ independent auditors review the companies’ financial information and to require that companies include in their proxy statements certain disclosures about their audit committees and reports from their audit committee containing certain disclosures. These rules are designed to improve disclosure related to the functioning of corporate audit
committees and to enhance the reliability and credibility of public companies. According to the SEC, audit committees play a critical role in the financial reporting system by overseeing and monitoring managements’ and the independent auditors’ participation in the financial reporting process. They believe that additional disclosures about the audit committee and its interaction with the company’s auditors and management will promote investor confidence in the integrity of the financial reporting process. In addition, increasing level of scrutiny by independent auditors of companies’ quarterly financial statements should lead to fewer year-end adjustments, and, therefore, more reliable information about companies throughout the year. Subsequently, the new rules require that the company must include a report of their audit committee in their proxy statements, where the audit committee must state whether the audit committee has: 1) reviewed and discussed the audited financial statements with management; 2) discussed with the independent auditors the matters required to be discussed; 3) received
7
from the auditors disclosures regarding the auditor’s independence. Also, the rules require that companies with listings on the Nasdaq, the American Stock Exchange (AMEX) or the NYSE, disclose in their proxy statements whether the audit committee members are
independent and disclose certain information regarding any director on the audit committee who is not independent (Security and Exchange Commission, 2000).
Even before the final rules of the SEC, the NYSE and NASDAQ already modified their requirements for audit committees. In 1999 the new standards required firms to maintain an audit committee with at least three directors, all of whom have no relationship to the company that they may interfere with the exercise of their independent from management and the company. These requirements were a response to the Security and Exchange
Commission’s call for improving the effectiveness of the corporate audit committees in overseeing the financial reporting process (Klein, 2002).
After the Enron Debacle in 2001, the SOX Act of 2002 also focused on improving audit quality and repairing auditor independence. SOX states that the purpose of the audit committee is to oversee the financial reporting processes of the company as well as the audit of its financial statements (Bédard et al, 2004). SOX requires that at least one audit committee member have financial expertise, that all the members be independent from firms’
management and that the audit committee oversees the accounting and financial reporting processes as well as the audits of financial statements (Bédard et al, 2004). Specific guidelines are for example that the SOX Act requires audit committees to pre-approve all NAS fees greater than five percent of the total fees paid to the incumbent auditor in the prior year (Abbott et al, 2003).
The primary audit committee duties are selecting and retaining the auditor and
ensuring the independence of the external auditor, which suggest that certain audit committee characteristics critically affect the execution of audit committee duties (Abbott et al, 2003). The audit committee duties can be classified into three categories: oversight of financial statements, of the external audit and of the internal control system (Bédard et al, 2004) Prior research suggests that audit committees possessing certain characteristics are important participants in the process of managing the client-auditor relationship (Abbott et al, 2003). Abbott et al (2003) study whether independent and active audit committees are
associated with lower ratios of non-audit service fees and find that audit committee members perceive a high level of audit fees in a negative light and take actions to decrease the non-audit fee ratio.
8
DeFond and Zhang (2014) defined higher audit quality as “greater assurance that the financial statement faithfully reflect the firm’s underlying economics, conditioned on it financial reporting system and innate characteristics”. They mention that the firm’s financial reporting system and innate characteristics affect the quality of the pre-audited statements, which constrain the achievable level of financial reporting quality. I consider audit committee characteristics to be part of the innate characteristics of the firm and therefore to affect audit quality.
Prior literature studied the effect of audit committee characteristics on audit quality. For example, Abbott et al (2004) find that the independence and activity level of the audit committee exhibit a significant negative association with the occurrence of restatements, which is used as a proxy for audit quality. They also find a negative association between an audit committee that includes at least one member with financial expertise and restatements. In an earlier study Abbott et al (2000) find that firms with audit committees which are composed of independent auditors and which meet at least twice per year are less likely to be sanctioned for fraudulent or misleading reporting.
Bédard et al (2004) investigate whether the expertise, independence and activities of a firm’s audit committee have an effect on the quality of its publicly released financial
information. They use earnings management as a proxy for audit quality and find a negative association between aggressive earnings management and the financial and governance expertise of audit committee members, with indicators of independence and a clear mandate defining the responsibilities of the committee. However, the frequency of meetings do not seem to affect the likelihood of aggressive earnings management.
The above studies mainly focused on the effect of audit committee characteristics on actual audit quality, while literature on audit quality makes a distinction between actual audit quality and perceived audit quality. Actual audit quality studies proxy audit quality by using measures for earnings management, earnings conservatism, restatements, going concern opinions, etcetera.
As far as my knowledge reaches, there are no studies yet that study the relationship between certain audit committee characteristics and perceived audit quality, as measured by the earnings response coefficient. Therefore, related to the assumption that perceived audit quality might be as important as actual audit quality, this study focuses on the audit
committee characteristics that influence perceived auditor perceptions when firms purchase non-audit services. Since firms are required to include information about the audit committee in their proxy statements, investors are likely to respond to these disclosures. The level of
9
response will be measured by the earnings response coefficient, which is used as a proxy for perceived audit quality.
The next chapter states the hypotheses tested in this study and the predicted effects of the provision of non-audit services and audit committee characteristics on the earnings response coefficient.
2.3 Hypotheses development
Figure 1 summarizes the conceptual model, where the audit committee is shown as the moderating variable in the relationship between non-audit services and perceived audit quality.
As mentioned before, the study of Krishnan et al (2005) is used as a starting point for this study and I test whether their results that the provision of non-audit services leads to lower perceived audit quality still holds ten years later in 2011. Therefore, by testing the following hypothesis, I contribute to prior literature and their mixed results on the impact of non-audit services on perceived audit quality. Assuming that the provision of non-audit services might decrease the quality of the audit and therefore the accuracy of the information, I expect the ERC to be lower when firms provide non-audit services. This is tested with the following hypothesis:
H1: The provision of non-audit services is associated with lower levels of perceived audit
quality
Following from prior literature, it seems that the characteristics of the audit committee that deserve the main focus are the independency of the audit committee, the financial expertise and the level of activity of the audit committee, i.e. the frequency of meetings. I also expect that these three characteristics mainly influence the perceived auditor quality of investors, because of their impressive existence in prior literature and because these characteristics are also recommended by the Blue Ribbon Committee (BCR) on Improving the Effectiveness of Corporate Audit Committees, documented in 1999. Their purpose was to develop
recommendations aimed at improving financial reporting by strengthening the audit
committees’ role as a financial monitor (Abbott et al, 2004). Although these recommendations might be a bit outdated, the characteristics are still used in recent research to study the
10
-
(+) Independence (H2a) (+) Financial Expertise (H2b) (+) Frequency of meetings (H2c) Figure 1 – Conceptual model
As mentioned before, DeFond and Zhang (2014) state that high audit quality increases the credibility of the financial reports. Audit quality, however, is not the only component of financial reporting quality. Financial reporting quality is also dependent upon the quality of the pre-audited financial statements, which are an input to the audit process. The quality of pre-audited statements is further determined by the firm’s financial reporting system, which maps its underlying economics into the financial reports; and the firms innate characteristics, which determine its underlying economics (DeFond and Zhang, 2014). I expect the audit committee to be one of the innate characteristics that influence the pre-audited statements and might therefore affect audit quality.
According to the SEC, the audit committee plays an important role in the financial reporting system. It’s interaction with the company’s auditor and management will promote investor confidence in the integrity of the financial reporting progress. Teoh et al (1993) expect that an earnings report has a greater effect on investor’s valuation of the firm when the reported numbers more accurately reflect true economic value. The model of Holthausen and Verrecchia (1988) predicts that the magnitude of the stock price response increases with the precision of the information. Therefore, I assume that the earnings response coefficient is
Non-audit Services (H1) Perceived Audit Quality Audit Committee Characteristics (+) Independence (H3a) (+) Financial Expertise (H3b) (+) Frequency of meetings (H3c)
11
higher when firms that have an audit committee with certain characteristics buy non-audit services than firms without certain audit committee characteristics, because information provided by firms with more effective audit committees is more accurate, informative and more useful for decisions. This assumption is taken into account when discussing the expectations of this study.
The BCR first recommends an independent audit committee, based upon the idea that independent directors are better able to objectively evaluate the propriety of management’s accounting, internal control and reporting practices. The BRC defines independence as the exclusion of current and former employees, relatives of management and persons receiving compensation form the company (Abbott et al, 2004).
There are several reasons why the independence of audit committee members might lead to higher audit quality. Audit committee independence can strengthen a firm’s internal control structure and increase the effectiveness of the external audit, thereby reducing the likelihood of misstatements and increasing the audit quality. Therefore, the first hypothesis is stated as follows:
H2a: An audit committee with independent directors is associated with higher perceived
audit quality
The second recommendation from the BRC concerns the financial expertise of the audit committee. They note that the complexity of the accounting and financial matters facing audit committees merits the commitment of considerable director resources. Therefore, at least one of the audit committee directors needs to have accounting or related financial management expertise, which is also required by the SOX Act of 2002. At the same time, the audit committee’s oversight role may be discounted by the external auditor if the auditor believes the audit committee does not possess the knowledge necessary to understand the technical accounting and financial reporting matters (Abbott et al, 2004). For example, an audit committee director with financial expertise might be in a better position to understand the audit issues and risks, as well as the audit procedures proposed to address and/or detect these issues and risks. Financial expertise in the audit committee also increases the likelihood that detected material misstatements will be communicated to the audit committee and corrected in a timely fashion (Abbott et al, 2004).
12
Because of the reasons mentioned above, I expect financial expertise in the audit committee to be positively associated with perceived audit quality, resulting in the following hypothesis:
H2b: An audit committee with at least one director possessing financial expertise is
associated with higher perceived audit quality
The final recommendation of the BRC used in this study is the frequency of meetings of the audit committee. According to Bédard et al (2004), expertise and independence will not result in effectiveness unless the committee is active. There are three aspects related to the activity of the audit committee, namely the duties it has to perform, the frequency of its meetings and its size. This study mainly focuses on the frequency of its meetings, because an audit
committee eager to carry out its functions of control must maintain a constant level of activity and best practices suggest three or four meetings a year (Bédard et al, 2004). In line with Bédard et al (2004), Abbott et al (2004) mention that the only way an effective audit
committee can be apprised of recent audit developments and issues is to meet frequently with both the internal and external auditors. By meeting frequently with the internal auditor, the audit committee will remain informed and knowledgeable about accounting and audit issues. By meeting frequently with the external auditor, the audit committee can proactively direct additional external audit resources toward a particular auditing issue in a timely fashion. This can reduce year-end audit time pressures that can potentially compromise external audit quality and can thereby increase audit quality. Based on this information, the last hypothesis regarding the relationship between audit committee characteristics and perceived audit quality is stated as follows:
H2c: An audit committee that meets frequently is associated with higher perceived audit
quality
As a third hypothesis, I test whether the audit committee characteristics weaken or strengthen the relationship between the provision of non-audit fees and perceived audit quality. In other words, I test whether the audit committee is a moderating variable in the relationship between non-audit services and perceived audit quality. I start explaining this relationship between these three variables by providing the definition of perceived audit quality as the joint
probability that an auditor will detect and report a financial misstatement. The ultimate goal of the audit committee is to increase this probability by monitoring the financial reporting
13
process. This assumes that the audit committee influences the audit quality directly through decreasing the probability of a misstatement and thus increases the perceived audit quality, i.e. the trust of the investors in the quality of the reported numbers. This potential positive effect on perceived audit quality might also influence the relationship between the purchase of non-audit services and perceived audit quality. I expect that there may be a difference in the response of investors between firms with an audit committee which purchase non-audit services and firms who purchase non-audit services but do not have an audit committee. Investors may respond differently to those two cases and accept purchases of non-audit services from a firm with an audit committee more than from a firm without an audit committee. To illustrate this, an investor who believes that the audit committee of a firm is effective may have less problems with the firms purchasing non-audit services than if the investor does not believe that the audit committee is effective. Therefore, investors might believe that the reported earnings of the firm with the audit committee are more accurate and might also believe that this effect is larger than the potential negative effect of non-audit services on the accuracy of the financial report. In the end, this may lead to a higher earnings response coefficient. I look at the different characteristics of the audit committee and their influence on the relationship between the purchase of non-audit services and perceived audit quality. I expect all three characteristics, independence, financial expertise and frequency of meetings to weaken the expected negative relationship between non-audit services and perceived audit quality. This brings me to the following three hypotheses:
H3a: An audit committee with independent directors weakens the negative relationship
between the provision of non-audit services and perceived audit quality
H3b: An audit committee with at least one director possessing financial expertise weakens the
negative relationship between the provision of non-audit services and perceived audit quality
H3c: An audit committee that meets frequently weakens the negative relationship between the
14 3. Data and research design
3.1 Data collection
Data is collected for the year 2011, to see whether the results of Krishnan et al (2005) still hold 10 years later. This is a ten-year period in which the independence of auditors and the related quality was of high concern to regulators and a period in which several rules are made to increase the quality of the audit. Therefore, it is interesting to study this period and
compare it later on. There is no other specific reason for choosing 2011 besides for the above mentioned reason.
First, Datastream is used to obtain the list of listed companies on the S&P500 in 2011. The S&P500 is an American Stock Market Index based on the market capitalizations of 500 large companies having common stock listed on the NYSE or NASDAQ. These companies are chosen to have a comparable dataset and because the S&P500 is mainly chosen as the research object in prior literature. Besides, due to the data on audit committee characteristics that has to be hand collected from the firm’s proxy statements in 2011 and due to a limited study period, a number of 500 firms is the maximum amount that can be studied accurately. After collecting the names of the firms and the cusip firm identifiers, their market betas, used as a control variable to proxy for firm risk, is also collected from Datastream.
Secondly, I use Wharton Research Data Services (WRDS) to obtain data from Compustat and Audit Analytics. Compustat (North-American Version) is a database which contains fundamental and market information on active and publicly held companies in the U.S. and Canada. It mainly provides annual and quarterly income statements, balance sheets and statement of cash flows. From Compustat, the following variables are collected from the annual fundamentals database: total assets, liabilities, common shares outstanding, common shares used to calculate earnings per share, dividends, earnings per share including and excluding extraordinary items, income before extraordinary items, net income, the standard industry classification code and the closing price of the fiscal year are collected. From the security daily database, I obtain the daily opening price and the daily current earnings per share.
The second database used is Audit Analytics, which provides detailed research on over 150.000 active audits and more than 10.000 accounting firms. The data collected from this database is the level of non-audit fees, audit fees and subsequently the total fees.
15
Central Index Key (CIK) identifier. This code is an unique, public number that is assigned to each entity that submits fillings to the SIC.
Data on audit committee variables is collected from proxy statements, since the SEC adopted disclosure requirements on audit committee’s composition, charter and activities in proxy statements (Abbott et al, 2003). These proxy statements are found under filing number 14A in the EDGAR database from the Security and Exchange Commission.
3.2 Sample selection
Firstly, I obtain 501 firms listed on the S&P500 in 2011 from the Datastream database. In the Compustat annual fundamentals, information for only 446 firms is provided. In the daily stock file, 444 firms are found. After merging the two Compustat databases, 179 firms are removed because they do not provide information on the essential variables. The dataset with the remaining 324 firms are merged with the Audit Analytics dataset to obtain non-audit and audit fees. Five observations are deleted because they do not provide us with information on the non-audit and audit fees. After merging the three datasets into a final data set, 319 observation remain.
Some variables, such as the return over the year, the earnings per share, the market to book value and the different non-audit fee metrics cannot directly be obtained from the databases and need to be measured separately. The next paragraph explains the measurement of these variables. After the creation of these variables, 77 observations are deleted due to missing values. A sample of 244 firms remain. Three more observations are removed from the sample because they do not have proxy statements filed under their current name in 2011. Because this is such a small number and I do not expect that removing these three firms will have a significant impact on the results, I have decided to not keep the observations. After checking on outliers and extreme values, I remove firms from the dataset with a return variable higher than 1, because this variable is measured in percentages and returns higher than 1 would have indicated a stock priced that increased with more than 100%. Because I do not believe that this return is solely due to earnings information, the provision of non-audit services or audit committee characteristics, I decided to delete two more observations. As explained later on, the average market to book value is 2.98. However, some firms have market to book values of over 18 or under -12, which means that the market value is 18 times as high as the book value, or the market value 12 times as small as the book value. Because I am studying the largest 500 stock-listed firms in the U.S., I assume that these ratios may not
16
be uncommon or extreme and therefore I do not remove them from the dataset. However, in the additional analysis section, I will study whether removing the firms with high market-to-book values from the sample would have influenced the results. In the end, the final sample consists of 239 firms.
3.3 Research design
3.3.1 Measuring perceived audit quality: the ERC
As a proxy for perceived audit quality, I use the earnings response coefficient. The earnings response coefficient measures the relation between earnings and stock returns, which has occupied much of the accounting literature during the last decades. In prior literature, there is no clear consensus on how to measure the ERC and a rich variety of metrics are used. For example, metrics used are the returns and level of earnings, normally scaled by prices, and returns and the unexpected portion of earnings, quantified using expectation models for
earnings, such as time-series models, analysts’ forecasts, and for returns, such as size-adjusted and market-model-predicted-returns (Bartov et al, 2001). Before explaining how I decided to measure the earnings response coefficient, the conceptual idea behind the earnings response coefficient is described.
In prior literature, the ERC is described as a measure of the extent to which new earnings information is capitalized in the stock price (Teoh and Wong, 1993). High perceived audit quality means that investors believe that the audited earnings information is relevant and faithfully represented. A firm’s stock price potentially shows the perceived audit quality when earnings information is provided. The stock price is a function of expected future dividends, where the expectations of future dividends are dependent upon the reported earnings. When these reported earnings are different from investors’ expectations, expectations about future dividends will be revised, leading to a change in stock price. Therefore, studying the relation between earnings and stock returns informs us how investors’ respond to the received
information about the firm’s earnings in the annual report.
However, investors’ might not always respond in the same way to the earnings information provided. Teoh and Wong (1993) expect the earnings report to have a greater effect on the valuation of the firm when the reported numbers more accurately reflect the true economic value of the firm. According to the model of Holthausen and Verrecchia (1988), the magnitude of the stock price response increases with the precision of information. Therefore,
17
when perceived audit quality is higher, investors will respond strongly to the reported earnings. This study will therefore contribute to the question whether the audit committee, established to increase the quality of the audit, increases perceived audit quality, as measured by the increase in the magnitude of the stock price.
According to Mahjoubi et al (2015), the ERC is studied in prior literature according to two different approaches: the event studies which depict the reaction of the stock price to the earnings announcement and the association studies that investigate the relationship between stock price and earnings over a relatively long period. The advantage of event studies lies in their ability to identify over time the net effect of earnings announcements on the stock price or on its return, which avoids confusion with other effects that may occur during the study period. However, an important weakness of this method is the choice of the optimal
observation window. This problem does not arise in association studies as the study period is so long that it may exceed 1 year, even if this leads to an increase in the risk of confusion with other effects (Mahjoubi, 2015). This study focuses on the year 2011 and falls therefore in the category of an association study.
Mainly because I do not study a particular earnings announcement, I decided to study the association between earnings over the year and returns over the year instead of studying the association between unexpected earnings and unexpected returns. This method allows to test whether the earnings information provided in the year 2011 affected the stock price of the firms. In other words, I test whether the earnings information is value relevant to the investors and how much of the earnings explain the change in returns over the year.
The return over the year is measured by subtracting the stock price of the end of the year, including dividends, by the stock price at the beginning of the year. Distributed dividends over the year are included because these dividends will not influence investors’ perception of future dividends. This return is scaled by the price at the beginning of the year, as mentioned by Bartov et al (2001). The earnings over the year 2011 are measured as the net income divided by number of shares outstanding. This earnings per share variable is also scaled by the price at the beginning of the year, to be able to compare it to the returns. According to DeFond and Zhang (2014), perception-based measures such as the ERC have several unique advantages over other audit quality measures. First, they capture audit quality more comprehensively than actual output measures. Second, they are continuous and thus capture both egregious failures as well as more subtle variations in audit quality. Third, they capture the net benefits or costs associated with audit quality. Finally, changes in client market share can be viewed as uniquely capturing audit committee’s perception of audit
18
quality. The main disadvantage of the perception-based measures is that they are relatively indirect, because financial reporting quality usually only has a second order effect on firm value. Another disadvantage is that these measures have more error in measuring it.
3.3.2 Measurement of non-audit services
To capture the economic bonding between the client and the auditor through the provision of non-audit services, three different non-audit fee measures are used in this study. These proxies are similar to the proxies used by Krishnan et al (2005).
First, I use the fee ratio, which is the proportion of non-audit fees to total fees and which captures the relative significance of non-audit fee to the total fees revenues received by the auditor. Second, I use the natural logarithm of non-audit fees, which captures the
economic bonding resulting from the purchase of non-audit services. Third, the natural log of total fees is used as a measure for non-audit services, which captures the total economic bonding of the client to the auditor created by the provision of both non-audit and audit services (Lim and Tan, 2008). However, related to the third measure, Ashbaugh et al (2003) point out that this ratio may not matter to the auditor if the client’s total fee is an
inconsequential part of the auditor’s client portfolio.
3.3.3 Audit committee variables
The audit committee characteristics of main interest are the independence of the committee members, the financial expertise of, at least one, audit committee members and the frequency of meetings of the audit committee.
Klein (2002) states that independence means that the three required directors have no relationship to the company that may interfere with the exercise of their independence from management and the company. Excluded from the audit committee are directors who are current employees, former employees within the last three years, have cross compensation committee links or are immediate family members of an executive officer.
According to Klein (2002) independence can be measured in three different ways, namely, firstly, by measuring the percentage of outside directors on the audit committee, secondly, by only assuming independence if all members are outside directors and thirdly, when only a majority of the members is independent, so more than 50%. In contradiction to Klein (2002), Bédard et al (2004) do not find a significant effect on earnings management
19
when the audit committee is composed of 50-99% independent members, but only a significant reduction when 100% percent of the members is independent. However, since listed firms are required to have only independent members in their audit committee, the sample only consists of firms that have an audit committee with only 100% independent members. Therefore, I am not able to create a dummy variable that differentiates between an audit committee that completely consists of independent members and an audit committee with less than 100% independent members. The variable INDEPENDENCE is therefore measured as the absolute number of independent directors, which is equal to the total number of directors on the audit committee for every firm in the sample.
The SEC also requires that listed firms include at least one financial expert on the audit committee. Because all the firms in the sample are listed, every firm confirms to the rules of the SEC and includes at least one financial expert in their audit committee. Therefore, also in this case, it enables us to create a dummy variable for this requirement, as is usually done in prior literature. It is, however, not an option to remove this variable from the analysis since it is assumed as one of the most important factors in affecting the quality of the audit committee and thus the perceived audit quality. I therefore look at the number of financial experts on the audit committee as a percentage of the total audit committee members. The variable EXPERT, i.e. the number of financial experts divided by the total number of audit committee members, is created.
Finally, there are no strict rules on the frequency of meetings and rules used in prior literature differ. For example, Bédard et al (2004) defines the minimum of meetings at three times per year, while Abbot et al (2004) states that the audit committee should meet at least four times a year. All of the firms in the sample meet at least four times a year, which is probably due to the fact that the S&P500 is chosen to be studied, which contains the largest stock-listed firms of the United States. It is therefore highly probable that these large firms meet more than four times a year. Also in this case, it is difficult to create the same dummy variables as is done in prior literature, where the dummy is coded 1 if the audit committee meets at least quarterly, and 0 otherwise. To be able to make conclusions about the effect of the number of meetings on the perceived audit quality, I do create the variable MEETINGS, where the value is the absolute number of meetings.
20 3.3.4 Control variables
Krishnan et al (2005) use different control variables which could potentially affect the earnings response coefficient. These variables are also suggested by previous studies as explanatory factors for the ERC, such as in the study of Teoh et al (1993). Therefore, the same control variables are also used in this study.
The first control variable is the firm’s market beta, which proxies for firm risk. The market beta is a measure of a stock’s volatility in relation to the market. By definition, the market has a beta of 1, and individual stocks are ranked according to how much they deviate from the market. Beta is used by investors to get a sense of stocks’ risk profiles. I expect that firms with a higher risk are having a smaller stock price reaction to unexpected earnings announcement. Therefore, the ERC is negatively associated with firm risk, such that if the firm risk increases, the ERC decreases.
A second control variable is the firm’s growth prospect, which is measured by the ratio of market value to book value of equity. This ratio is an indication of how much shareholders are paying for the net assets of the company, which provides them a way to compare market value to a conservative measure of the value of the firm. Growth prospects are positively related to the ERC.
The third control variable is firm size, which is measured by the natural logarithm of the market value. This variable proxies for the degree of uncertainty before an earnings announcement. Easton and Zmijewski (1989) find firm size to be not related to the ERC, while Lipe (1990) finds firm size to have a significant impact on the ERC. Due to this different findings in prior literature, no sign is predicted.
Finally, the last control variable is firm loss. Hayn (1995) found that the ERC is much lower for firms reporting losses, because losses contain less information. A negative sign on the ERC is predicted.
3.3.5 The research model
To test the hypotheses, three different models are used. First, I test whether non-audit services are associated with lower perceived audit quality. To be able to make these results comparable to the results of Krishnan et al (2005), a regression is performed on model (1). Secondly, model (2) is used to test whether the audit committee characteristics are associated with higher perceived audit quality. Finally, model (1) and (2) are combined into one final model,
21
model (3). This last model tests whether different audit committee characteristics moderate the relationship found between non-audit services and perceived audit quality. The definitions of the variables in the three models are explained in table 1.
Model 1: = + + + ∗ + ∗ + ∗ + ∗ + ∗ ! "# + $ Model 2: = + + ∗ % + ∗ % + ∗ + ∗ + ∗ + ∗ ! "# + $ Model 3: = + + + % + ∗ + ∗ % + ∗ % + ∗ % ∗ + & ∗ + ' ∗ + ∗ ( + ∗ ! "# + $
In models (1) and (2), is the normal earnings response coefficient, while is the earnings response coefficient of main interest and measures the incremental effect of the purchase of non-audit services and the presence of an effective audit committee on the earnings response coefficient.
In model (3), is again the normal earnings response coefficient, while measures the incremental effect of the purchase of non-audit services on the earnings response
coefficient, measures the incremental effect of the audit committee on the ERC and finally, measures the effect of the audit committee on the relationship between non-audit services and the ERC.
I thought about only using the complete model (3) for testing the relationship between non-audit services and audit committee characteristics on the one hand and perceived audit quality on the other hand, however I do think that the results on in models (1) and (3) and
and respectively in model (3) are too different to only study model (3), especially because it might be interesting to know that the results are different depending on the model used.
22
Table 1 – Variables
Variable Definition
RET Return over the year, as measured by )+ * + − )- , with ) = Price at the
end of the fiscal year; * + = dividends over the fiscal year; )- = Price at the
beginning of the fiscal year
EPS Earnings per share, measured by dividing net income over the year by common shares used to calculate earnings per share. The EPS variable is divided by the stock price in the beginning of the year, to make it comparable to the return variable;
NAF Non-audit fees, as measured by the 1) fee ratio, calculated by dividing non-audit fees by the total fees; 2) the natural log of non-non-audit fees and 3) the natural log of total audit fees
AC Audit committee characteristics, as measured by 1) the independence of the audit committee members; 2) financial expertise of the audit committee members and 3) the number of meetings of the audit committee
EPS * NAF Interaction variable that measures the effect of non-audit service fees and
earnings information together on the return variable
EPS * AC Interaction variable that measures the combined effect of audit committee characteristics and earnings on the return variable
NAF * AC Interaction variable that measures the combined effect of non-audit fees and audit committee characteristics on the return variable
EPS * AC * NAF
Interaction variable that measures the combined effect of audit committee characteristics, non-audit fees and earnings on the return variable
Loss A dummy variable coded 1 if net income is smaller than zero and coded 0 otherwise
MB The market to book value of the firm, where the total market value is the price at the end of the fiscal year times the common shares outstanding and the total book value is the total assets minus the total liabilities.
Firmsize The size of the firm, measured by the natural logarithm of the total market value of the firm
23 4. Results
4.1 Descriptive statistics
Table 2 reports the number of observations, mean, standard deviation and the minimum and maximum observation of the sample. Panel A shows the descriptive statistics for all the other variables used in the model except for the audit and non-audit fees and the audit committee characteristics, which are shown in panel B and C. Panel A shows that the average beta of the sample is 1.04, that the average return in 2011 is $0.012 and the earnings per share at the end of 2011 divided by the price per share at the beginning of the year is on average $0.06. The book value of the 239 firms, as measured by the assets minus the liabilities, is on average $10,632 and the market value of the firms, as measured by the common shares outstanding times the closing price of the fiscal year, is on average $25,291. The average market to book ratio is 2.69, which means that the market value of the firms is on average 2.69 as high as the book value. Panel B shows that the average audit fees paid to the auditor is $7,745,071 and that the average non audit fee is $2,349,511, which is lower than the audit fees. The average total fees is approximately $10,1 million, with a non-audit to audit fee ratio of approximately 0.19, which means that around 20% of the total fees paid to the auditor consists of non-audit fees. The above results are comparable to the results found by Krishnan et al (2005), although in this study the average return is higher and the audit fees are higher, while the non-audit fees are lower. Krishnan et al (2005) study the cumulative abnormal returns, which might explain why their returns found are lower. The lower non-audit fees can be explained by the fact that non-audit services have been under increasing pressure in the last ten years, especially since 2001. This decrease in non-audit fees might be compensated by an increase in audit fees. Panel C shows that the average number of independent directors on the board is 4.4, the average number of financial expertise members on the board is 2.5 and the average number of meetings of the audit committee is 8.2. The SOX Act and the SEC both require at least three directors in the audit committee, all of whom who should be independent, including at least one financial expert, but do not have rules for the number of meetings on the audit committee, but a minimum number of four meetings is usually recommended. All the firms in the sample confirm to these rules, which is probably due to the fact that the sample consists of the largest listed firms in the United States, who are most likely to have an audit committee that confirms to the rules of the SEC.
24
Table 2 - Descriptive statistics
Panel A
Variable Mean Std. Dev. Min Max
Return 0.0129 0.2588 -0.7440 0.9875 EPS/./-0 0.0608 0.0466 -0.2536 0.3132 Beta 1.0390 0.3992 0.243 1.861 MB-ratio 2.9821 3.8656 -12.4374 18.0392 Firm size 9.4767 1.0257 7.6647 12.2587 Loss 0.0334 0.1802 0 1 Panel B
Audit fees 7745071 8972390 523453 8.71e+07
Non audit fees 2349511 4064243 2149 3.84e+07
Total fees 1.01e+07 1.24e+07 525602 1.13e+08
Fee ratio 0.1891 0.1239 0.0017 0.6496
Log non-audit fees 13.700 1.5778 7.6728 17.4636
Log total fees 15.6576 0.9489 13.1723 18.5447
Panel C
Independence 4.4017 1.0951 1 8
Expert 2.5105 1.5226 0 8
25
Table 3 presents the correlations between the variables. This matrix tells us something about the strength of the relation between the dependent and the independent variables but does not tell us anything about the direction of the relationship. Krishnan et al (2005) mention the correlations which exceed 0.4. If I assume that a correlation of more than 0.4 is high, high correlations are found between the audit fees and the non-audit fees and the loss of the firm and the earnings per share. Finally, I also find an high correlation between the size of the firm and the level of audit and non-audit fees. However, I do not expect these correlations to be of such a high level that they influence the results and that I cannot exclude these variables in the model due to potential multicollinearity, because these variables are too important in
explaining the dependent variable.
Table 3 – Correlation Matrix
Return EPS MB Firm
Size
Loss Fee ratio
Log NAF
Log TF FE Meetings Independence
Return 1.0000 EPS 0.2187 1.0000 MB 0.2292 -0.0408 1.0000 Firm Size 0.2278 0.2212 0.1260 1.0000 Loss -0.1658 -0.5671 0.0048 -0.1972 1.0000 Fee ratio 0.0091 0.0405 0.0371 0.2003 -0.1333 1.0000 Log NAF -0.0450 0.1752 -0.0150 0.4928 -0.0997 0.7243 1.0000 Log AF -0.0407 0.2272 -0.0237 0.6095 -0.0803 0.3825 0.8548 1.0000 FE -0.0145 0.0535 -0.0653 0.0071 -0.0166 -0.0226 0.1280 0.2040 1.0000 Meetings 0.0345 -0.0098 -0.0467 0.1700 0.1001 0.1033 0.2139 0.2252 0.0687 1.0000 Independence -0.0151 0.1290 -0.0375 0.1865 -0.0471 -0.0381 0.1759 0.2978 0.3528 -0.0169 1.0000
26 4.2 Empirical Results
4.2.1 Non-audit services and perceived audit quality
The first hypothesis I test is whether perceived audit quality decreases when non-audit services are provided. Table 4 shows the results of the first regression, where the three different measures of non-audit fees are regressed against the return over the year 2011. The coefficient on EPS shows the normal ERC and the coefficient on the interaction variable EPS * NAF shows the ERC of interest, namely the earnings response coefficient when firms buy non-audit services. If the non-audit services are measured as the fee ratio, a significant positive ERC of 4.30 (p-value: 0.057) is found. Measuring the non-audit services as the ratio of audit fees and total fees (Panel A), no significant negative relationship between non-audit services and perceived non-audit quality is found, which is stated in the first hypothesis. However, when the non-audit services are measured as the log of non-audit fees and the log of audit fees (Panel B and C respectively), a negative coefficient, 0.16 (pvalue: 0.495) and -0.68 (p-value: 0.080) respectively, on the interaction variables of interest are found, which is in accordance with the first hypothesis that the provision of non-audit services leads to lower perceived audit quality. This result is only significant when I measure non-audit services as the log of total audit fees, but not when audit services are measured as the log of non-audit fees. Therefore, this result supports the first hypothesis.
As expected, a positive association is found between the growth prospects of the firm and the ERC, a negative relationship between the risk of the firm and the ERC. I did not predict a sign for the relationship between the firm size and the ERC, but I do find a positive relationship. The adjusted is between approximately 15% and 18%, depending on the measure of non-audit services used. This means that the explanatory variables explain between 15% and 18% of our dependent variable. This adjusted is in accordance to prior literature on the effects of non-audit services on perceived audit quality. The F-statistic is highly significant, meaning that our overall model is significant.
27
Table 4 – Non-audit services and perceived audit quality Panel A Coefficient t-statistic
(p-value) Adjusted 12 F-statistic Intercept -0.0869 -1.60(0.112) EPS 4.3036* 1.91 (0.057) Non-audit fees -0.0555 -0.22 (0.824) EPS * NAF 0.0741 0.02 (0.984) EPS * MB 0.0755 1.39 (0.166)
EPS * Firm size 0.0286 0.14 (0.890)
EPS * Loss 0.3317 0.33(0.742)
EPS * Beta -2.9456*** -5.35(0.000)
0.1464 6.83***
Panel B Coefficient t-statistic (p-value) Adjusted 32 F-statistic Intercept 0.1418 0.67 (0.506) EPS 4.0725 1.28 (0.203) Non-audit fees -0.0183 -1.18 (0.241) EPS * NAF -0.1585* -0.68 (0.495) EPS * MB 0.0887 1.65 (0.100)
EPS * Firm size 0.2917 1.23 (0.220)
EPS * Loss 0.1883 0.19 (0.847)
EPS * Beta -2.8825*** -5.29 (0.000)
1.1677 7.85***
Panel C Coefficient t-statistic (p-value) Adjusted 32 F-statistic Intercept 0.1882 0.46 (0.649) EPS 10.1473* 1.87 (0.063) Non-audit fees -0.0200 -0.76 (0.450) EPS * NAF -0.6785* -1.76 (0.080) EPS * MB 0.0944* 1.77 (0.078) EPS * Firmsize 0.5438* 2.10 (0.037) EPS * Loss -0.1518 -0.16 (0.877) EPS * Beta -2.6369*** -4.78 (0.000) 0.1807 8.50***
Panel A: NAF measured by the fee ratio (non-audit fees divided by the total fees) Panel B: NAF measured by the log of non-audit fees
Panel C: NAF measured by the log of total fees
28
4.2.2 Audit committee characteristics and perceived audit quality
The second hypothesis I test is whether audit committee characteristics affect perceived audit quality. I use three different audit committee characteristics, namely the independence of the audit committee members, the number of financial experts on the audit committee and finally the number of meetings of the audit committee. To see whether these characteristics influence the perceived audit quality, I am interested in the coefficients on the interaction variables EPS * Independence, EPS * Expert and EPS * Meetings, which measure the incremental effect of the audit committee characteristic on the earnings response coefficient. I do expect to find that these audit committee characteristics lead to higher perceived audit quality and therefore I expect to find a positive coefficient on these variables. Independent of the audit committee characteristic, a positive ERC is found. However, the normal ERC is only significant when the number of meetings is included in the model, with an ERC of 4.62 (p-value: 0.040).
Panel A shows that the earnings response coefficient is positive when the audit
committee members are independent. Return and thus perceived audit quality increases, when the number of independent directors on the audit committee increases. Panel B shows the effect from financial experts on the perceived audit quality. As the SEC states that the audit committee should at least have one financial expert, I expect this to positively influence the earnings response coefficient. I find an insignificant ERC of 0.82 (p-value: 0.452), which is positive and thus positively influences return for an increase in financial experts on the audit committee. Finally, I also expect a positive ERC on the number of meetings the audit
committee has. However, I find an insignificant, negative ERC of -0.06 (p-value: 0.592), which would assume that returns are not positively associated with the number of meetings of the audit committee.
The expectations regarding the control variables also hold, namely that firm risk is negatively associated with the ERC and the growth prospects positively associated with the ERC. This model also shows that firm size is positively related to the ERC. The adjusted of all three regressions with the different audit committee characteristics is approximately 15%, which means that 15% of the variation in the return is explained by the included
independent variables. I also find a F-statistic of approximately 7, which is highly significant at a 1% level and tells me that the overall model is significant. However, I do not find any significant earnings response coefficients on the variables of interest, which does not provide enough evidence to state that there is a positive relationship between the audit committee characteristics and perceived audit quality.
